1 2017-09-21 Richard Biener <rguenther@suse.de>
3 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4 Restore valid IL after code generation errors.
5 * graphite.c (graphite_transform_loops): Diagnose code
6 generation issues as MSG_MISSED_OPTIMIZATION and continue
9 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10 Alan Hayward <alan.hayward@arm.com>
11 David Sherwood <david.sherwood@arm.com>
13 * calls.c (compute_argument_addresses): Use simplify_gen_binary
14 rather than choosing between plus_constant and gen_rtx_<CODE>.
15 * expr.c (emit_push_insn): Likewise.
16 (expand_expr_real_2): Likewise.
18 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
19 Alan Hayward <alan.hayward@arm.com>
20 David Sherwood <david.sherwood@arm.com>
22 * loop-unroll.c (split_iv): Call copy_rtx on the step.
24 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
25 Alan Hayward <alan.hayward@arm.com>
26 David Sherwood <david.sherwood@arm.com>
28 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
31 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
32 Alan Hayward <alan.hayward@arm.com>
33 David Sherwood <david.sherwood@arm.com>
35 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
36 INTEGER_CST rather than a negative test for ADDR_EXPR.
38 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
39 Alan Hayward <alan.hayward@arm.com>
40 David Sherwood <david.sherwood@arm.com>
42 * tree-vrp.c (extract_range_from_binary_expr_1): Check
43 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
45 2017-09-21 Richard Biener <rguenther@suse.de>
47 PR tree-optimization/71351
48 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
49 graphite_create_new_loop_guard): Remove, fold remaining parts
51 (translate_isl_ast_node_for): ... here and simplify.
53 2017-09-21 Jakub Jelinek <jakub@redhat.com>
56 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
57 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
58 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
59 alternative always use QI mode, for -Os imov (=R,R) alternative
60 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
63 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
64 Jeff Law <law@redhat.com>
66 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
67 (allocate_stack_space): New function, partially extracted from
69 (s390_emit_prologue): Track offset to most recent stack probe.
70 Code to allocate space moved into allocate_stack_space.
71 Dump actions when no stack is allocated.
72 (s390_prologue_plus_offset): New function.
73 (s390_emit_stack_probe): Likewise.
75 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
77 * common.opt (Wa, Wl, Wp, g, gz=): Add
79 (gno-column-info): Remove.
80 (gcolumn-info): Drop RejectNegative.
82 (gno-record-gcc-switches): Remove.
83 (grecord-gcc-switches): Drop RejectNegative.
84 (gno-split-dwarf): Remove.
85 (gsplit-dwarf): Drop RejectNegative.
86 (gno-strict-dwarf): Remove.
87 (gstrict-dwarf): Drop RejectNegative.
88 * config/darwin.opt (gfull, gused): Add RejectNegative.
89 * dwarf2out.c (gen_producer_string): Drop
90 gno-record-gcc-switches handler.
91 * optc-gen.awk: Add g to prefixes with negative forms.
92 * opts-common.c (remapping_prefix_p): New.
94 (generate_canonical_option): Test g prefix.
95 (option_map): Add -gno- mapping.
96 (add_misspelling_candidates): Check remapping_prefix_p.
98 2017-09-20 Jeff Law <law@redhat.com>
100 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
101 thinko in stack clash protection support.
103 * explow.c (compute_stack_clash_protection_loop_data): Use
104 CONST_INT_P instead of explicit test. Verify object is a
105 CONST_INT_P before looking at INTVAL.
106 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
107 instead of explicit test.
109 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
112 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
113 address instead of to r1 and r11.
115 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
117 * config.gcc: Support "knm".
118 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
119 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
121 * config/i386/i386.c (m_KNM): Define.
122 (processor_target_table): Add "knm".
124 (ix86_option_override_internal): Add "knm".
125 (ix86_issue_rate): Add PROCESSOR_KNM.
126 (ix86_adjust_cost): Ditto.
127 (ia32_multipass_dfa_lookahead): Ditto.
128 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
129 (fold_builtin_cpu): Add M_INTEL_KNM.
130 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
131 (processor_type): Add PROCESSOR_KNM.
132 * config/i386/x86-tune.def: Add m_KNM.
133 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
135 2017-09-20 Richard Biener <rguenther@suse.de>
137 PR tree-optimization/80213
138 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
139 are allowed in empty BBs as well.
140 (canonicalize_loop_closed_ssa): Also look for other complex
142 (scop_detection::get_sese): Include the loop-closed PHI block
144 (scop_detection::merge_sese): Remove code adding extra blocks.
145 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
146 (build_scops): Assert the final returned scop is invalid.
148 2017-09-20 Richard Biener <rguenther@suse.de>
150 PR tree-optimization/82264
151 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
153 (vn_phi_lookup): Likewise.
154 (vn_phi_insert): Likewise.
156 2017-09-20 Jakub Jelinek <jakub@redhat.com>
158 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
159 that fits into uhwi or shwi, add DW_AT_const_value regardless
160 of early_dwarf without going through RTL, using add_AT_unsigned
163 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
164 (DEBUG_LTO_ABBREV_SECTION): Likewise.
165 (DEBUG_LTO_MACINFO_SECTION): Likewise.
166 (DEBUG_MACRO_SECTION): Likewise.
167 (DEBUG_LTO_MACRO_SECTION): Likewise.
168 (DEBUG_STR_DWO_SECTION): Likewise.
169 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
170 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
171 (DEBUG_LTO_DWO_LINE_SECTION): Define.
172 (DEBUG_LTO_LINE_STR_SECTION): Define.
173 (init_sections_and_labels): Initialize debug_line_str_section
174 variable. Initialize debug_loc_section for -gdwarf-5 to
175 DEBUG_LOCLISTS_SECTION. Formatting fixes.
177 2017-09-20 Richard Biener <rguenther@suse.de>
179 * graphite-sese-to-poly.c (extract_affine): Properly handle
180 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
182 2017-09-20 Richard Biener <rguenther@suse.de>
184 PR tree-optimization/81373
185 * graphite-scop-detection.c (build_cross_bb_scalars_def):
186 Force SESE live-out defs to be handled even if they are
189 2017-09-19 Jeff Law <law@redhat.com>
191 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
192 nothing for stack adjustments with REG_STACK_CHECK.
193 * sched-deps.c (parse_add_or_inc): Reject insns with
194 REG_STACK_CHECK from dependency breaking.
195 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
196 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
197 * reg-notes.def (STACK_CHECK): New note.
199 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
200 (ix86_expand_prologue): Dump stack clash info as needed.
201 Call ix86_adjust_stack_and_probe_stack_clash as needed.
203 * function.c (dump_stack_clash_frame_info): New function.
204 * function.h (dump_stack_clash_frame_info): Prototype.
205 (enum stack_clash_probes): New enum.
207 * config/alpha/alpha.c (alpha_expand_prologue): Also check
208 flag_stack_clash_protection.
209 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
210 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
211 (arm_frame_pointer_required): Likewise.
212 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
213 (ia64_expand_prologue): Likewise.
214 * config/mips/mips.c (mips_expand_prologue): Likewise.
215 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
216 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
217 (sparc_flat_expand_prologue): Likewise.
218 * config/spu/spu.c (spu_expand_prologue): Likewise.
220 * explow.c: Include "params.h".
221 (anti_adjust_stack_and_probe_stack_clash): New function.
222 (get_stack_check_protect): Likewise.
223 (compute_stack_clash_protection_loop_data): Likewise.
224 (emit_stack_clash_protection_loop_start): Likewise.
225 (emit_stack_clash_protection_loop_end): Likewise.
226 (allocate_dynamic_stack_space): Use get_stack_check_protect.
227 Use anti_adjust_stack_and_probe_stack_clash.
228 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
229 (emit_stack_clash_protection_loop_start): Likewise.
230 (emit_stack_clash_protection_loop_end): Likewise.
231 * rtl.h (get_stack_check_protect): Prototype.
232 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
233 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
234 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
236 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
238 * doc/tm.texi: Rebuilt.
239 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
240 get_stack_check_protect.
241 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
242 * config/arm/arm.c (arm_expand_prologue): Likewise.
243 (arm_frame_pointer_required): Likewise.
244 * config/i386/i386.c (ix86_expand_prologue): Likewise.
245 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
246 * config/mips/mips.c (mips_expand_prologue): Likewise.
247 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
248 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
249 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
250 (sparc_flat_expand_prologue): Likewise.
252 * common.opt (-fstack-clash-protection): New option.
253 * flag-types.h (enum stack_check_type): Note difference between
254 -fstack-check= and -fstack-clash-protection.
255 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
256 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
257 * toplev.c (process_options): Issue warnings/errors for cases
258 not handled with -fstack-clash-protection.
259 * doc/invoke.texi (-fstack-clash-protection): Document new option.
260 (-fstack-check): Note additional problem with -fstack-check=generic.
261 Note that -fstack-check is primarily for Ada and refer users
262 to -fstack-clash-protection for stack-clash-protection.
263 Document new params for stack clash protection.
265 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
267 * config/i386/i386.c (ix86_split_long_move): Do not handle
268 address used for LEA in a special way.
270 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
272 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
274 2017-09-19 Martin Sebor <msebor@redhat.com>
277 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
278 of incompatible types.
280 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
282 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
283 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
284 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
285 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
287 2017-09-19 Richard Biener <rguenther@suse.de>
289 PR tree-optimization/82244
290 * tree-vrp.c (remove_range_assertions): Do not propagate
291 a constant to abnormals but replace the assert with a copy.
293 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
295 PR rtl-optimization/57878
296 PR rtl-optimization/68988
297 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
298 avoidance test involving non_reload_pseudos. Move frequency test
299 below the general fragmentation avoidance test.
301 2017-09-19 Richard Biener <rguenther@suse.de>
303 PR tree-optimization/69728
304 * graphite-sese-to-poly.c (schedule_error): New global.
305 (add_loop_schedule): Handle empty domain by failing the
307 (build_original_schedule): Handle schedule_error.
309 2017-09-19 Richard Biener <rguenther@suse.de>
311 * graphite-scop-detection.c (scop_detection::can_represent_loop):
312 Do not iterate to sibling loops but only to siblings of inner
315 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
318 * config/m68k/m68k.md (moveq feeding equality comparison): Check
319 that the registers are different.
321 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
323 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
324 to processor_model and "amdfam17h" to arch_names_table.
325 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
327 2017-09-18 Jakub Jelinek <jakub@redhat.com>
330 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
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-vectorizer.h (vect_slp_analyze_operations): Replace parameters
338 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
340 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
341 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
342 vect_schedule_slp_instance.
343 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
344 Update call to vect_slp_analyze_node_operations. Simplify return
346 (vect_slp_analyze_bb_1): Update call accordingly.
347 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
348 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
349 (vect_schedule_slp): Update call accordingly.
351 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
352 Alan Hayward <alan.hayward@arm.com>
353 David Sherwood <david.sherwood@arm.com>
355 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
356 with types that aren't in fact scalar.
358 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
360 * tree-vect-slp.c (vect_record_max_nunits): New function,
362 (vect_build_slp_tree_1): ...here.
363 (vect_build_slp_tree_2): Call it for phis too.
365 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
367 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
368 to vect_get_vec_def_for_operand when getting the mask operand.
370 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
371 Alan Hayward <alan.hayward@arm.com>
372 David Sherwood <david.sherwood@arm.com>
374 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
377 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
378 Alan Hayward <alan.hayward@arm.com>
379 David Sherwood <david.sherwood@arm.com>
381 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
382 calculation for vector booleans.
384 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
385 Alan Hayward <alan.hayward@arm.com>
386 David Sherwood <david.sherwood@arm.com>
388 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
390 (vect_transform_stmt): ...here.
391 (vect_analyze_stmt): Use it instead of calling
392 vectorizable_live_operation directly.
394 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
396 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
397 non-SIMT targets in acc vector loops.
399 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
401 * configure.ac: Add arc and check if assembler supports gdwarf2.
402 * configure: Regenerate.
404 2017-09-18 Richard Biener <rguenther@suse.de>
406 PR tree-optimization/82220
407 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
408 epilogue niters from the min_profitable_iters compute.
410 2017-09-18 Jakub Jelinek <jakub@redhat.com>
413 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
414 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
415 (ix86_init_pic_reg): Revert 2017-09-01 changes.
417 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
420 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
421 switching to a new text section.
423 2017-09-18 Richard Biener <rguenther@suse.de>
425 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
427 (build_alias_set): Reject aliases with no access function.
429 2017-09-18 Richard Biener <rguenther@suse.de>
431 PR tree-optimization/79622
432 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
434 (build_cross_bb_scalars_use): Likewise.
436 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
438 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
440 2017-09-18 Alan Modra <amodra@gmail.com>
443 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
444 stack_pointer_rtx for count 0. Update comments. Break up
445 large rtl expression.
447 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
449 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
450 Increase to 20 bytes.
451 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
452 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
453 or avx version of the stub.
455 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
458 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
459 compute the minimum stack alignment. Also update preferred stack
460 boundary for leaf functions.
462 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
464 PR tree-optimization/82228
465 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
468 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
470 * common/config/nds32/nds32-common.c
471 (nds32_option_optimization_table): Refine formatting.
472 (nds32_option_optimization_table): Use -fsched-pressure and
473 -fomit-frame-pointer for specific optimization level.
475 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
477 * config/nds32/nds32.c: Refine formatting and comments.
478 * config/nds32/nds32.h: Likewise.
479 * config/nds32/nds32.md: Likewise.
480 * config/nds32/nds32-cost.c: Likewise.
481 * config/nds32/nds32-isr.c: Likewise.
482 * config/nds32/nds32-md-auxiliary.c: Likewise.
483 * config/nds32/nds32-multiple.md: Likewise.
484 * config/nds32/nds32-predicates.c: Likewise.
486 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
487 Jakub Jelinek <jakub@redhat.com>
489 Add support for -std=c++2a.
490 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
492 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
494 2017-09-15 Steve Ellcey <sellcey@cavium.com>
497 * doc/extend.texi (Common Function Attributes): Add
498 references to ARM, AArch64, and S/390 specific attributes.
499 (Function Specific Option Pragmas): Add AArch64 and S/390
500 to list of back ends that support the target pragma.
502 2017-09-15 Nathan Sidwell <nathan@acm.org>
504 * doc/standards.texi: Fix C++17 description. Update URLs for
507 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
509 * common.opt (Wcast-align=strict): New warning option.
510 * doc/invoke.texi: Document -Wcast-align=strict.
512 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
514 * cgraph.h (cgraph_thunk_info): Add comments.
515 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
516 assert for VIRTUAL_* arguments stricter.
518 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
520 PR tree-optimization/71026
521 * match.pd: Move RDIV patterns from fold-const.c
522 * fold-const.c (distribute_real_division): Removed.
523 (fold_binary_loc): Remove calls to distribute_real_divison.
525 2017-09-15 Jakub Jelinek <jakub@redhat.com>
527 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
528 c++1z and gnu++1z as deprecated. Change other references to
529 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
530 Change -Wc++1z-compat to -Wc++17-compat.
531 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
532 * dwarf2out.c (highest_c_language): Handle C++17.
533 (gen_compile_unit_die): Likewise.
535 2017-09-15 Jakub Jelinek <jakub@redhat.com>
537 PR rtl-optimization/82192
538 * combine.c (make_extraction): Don't look through non-paradoxical
539 SUBREGs or TRUNCATE if pos + len is or might be bigger than
542 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
543 Alan Hayward <alan.hayward@arm.com>
544 David Sherwood <david.sherwood@arm.com>
546 * target.def (function_arg_offset): New hook.
547 * targhooks.h (default_function_arg_offset): Declare.
548 * targhooks.c (default_function_arg_offset): New function.
549 * function.c (locate_and_pad_parm): Use
550 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
551 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
552 (TARGET_FUNCTION_ARG_OFFSET): ...this.
553 * doc/tm.texi: Regenerate.
554 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
555 * config/spu/spu.c (spu_function_arg_offset): New function.
556 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
557 * system.h (FUNCTION_ARG_OFFSET): Poison.
559 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
560 Alan Hayard <alan.hayward@arm.com>
561 David Sherwood <david.sherwood@arm.com>
563 * target.def (truly_noop_truncation): New hook.
564 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
565 than TRULY_NOOP_TRUNCATION.
566 * hooks.h (hook_bool_uint_uint_true): Declare.
567 * hooks.c (hook_bool_uint_uint_true): New function.
568 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
569 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
570 * doc/tm.texi: Regenerate.
571 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
572 rather than TRULY_NOOP_TRUNCATION in comments.
573 (simplify_comparison): Likewise.
574 (record_truncated_value): Likewise.
575 * expmed.c (extract_bit_field_1): Likewise.
576 (extract_split_bit_field): Likewise.
577 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
578 instead of TRULY_NOOP_TRUNCATION.
579 * function.c (assign_parm_setup_block): Likewise.
580 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
581 * rtlhooks.c: Include target.h.
582 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
583 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
584 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
585 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
586 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
587 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
588 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
589 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
590 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
591 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
592 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
593 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
594 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
595 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
596 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
597 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
598 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
599 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
600 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
601 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
602 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
603 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
604 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
605 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
606 * config/mips/mips.c (mips_truly_noop_truncation): New function.
607 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
608 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
609 rather than TRULY_NOOP_TRUNCATION in comments.
610 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
611 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
612 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
613 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
614 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
615 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
616 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
617 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
618 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
619 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
620 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
621 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
622 rather than TRULY_NOOP_TRUNCATION in comments.
623 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
624 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
625 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
626 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
627 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
628 TRULY_NOOP_TRUNCATION condition.
629 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
630 (TRULY_NOOP_TRUNCATION): Delete.
631 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
632 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
633 * config/spu/spu.c (spu_truly_noop_truncation): New function.
634 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
635 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
636 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
637 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
638 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
639 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
640 rather than TRULY_NOOP_TRUNCATION in comments.
641 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
642 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
643 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
644 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
645 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
646 * system.h (TRULY_NOOP_TRUNCATION): Poison.
648 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
651 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
652 (*cmp_ior): Likewise.
653 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
654 (*ior_scc_scc_cmp): Likewise.
655 (*and_scc_scc): Likewise.
656 (*and_scc_scc_cmp): Likewise.
658 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
659 Alan Hayard <alan.hayward@arm.com>
660 David Sherwood <david.sherwood@arm.com>
662 * target.def (can_change_mode_class): New hook.
663 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
664 (hard_regno_nregs): Likewise.
665 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
666 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
667 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
668 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
669 (LOAD_EXTEND_OP): Update accordingly.
670 * doc/tm.texi: Regenerate.
671 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
672 CANNOT_CHANGE_MODE_CLASS.
673 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
674 (REG_CAN_CHANGE_MODE_P): ...this new macro.
675 * combine.c (simplify_set): Update accordingly.
676 * emit-rtl.c (validate_subreg): Likewise.
677 * recog.c (general_operand): Likewise.
678 * regcprop.c (mode_change_ok): Likewise.
679 * reload1.c (choose_reload_regs): Likewise.
680 (inherit_piecemeal_p): Likewise.
681 * rtlanal.c (simplify_subreg_regno): Likewise.
682 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
683 instead of CANNOT_CHANGE_MODE_CLASS.
684 (reload_cse_simplify_operands): Likewise.
685 * reload.c (push_reload): Use targetm.can_change_mode_class
686 instead of CANNOT_CHANGE_MODE_CLASS.
687 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
688 REG_CANNOT_CHANGE_MODE_P.
689 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
690 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
691 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
692 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
693 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
694 (arm_can_change_mode_class): New function.
695 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
696 than CANNOT_CHANGE_MODE_CLASS in comments.
697 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
698 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
699 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
700 (ix86_can_change_mode_class): ...this new function, inverting the
701 sense of the return value.
702 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
703 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
704 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
705 (ia64_can_change_mode_class): New function.
706 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
707 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
708 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
709 (m32c_can_change_mode_class): ...this new function, inverting the
710 sense of the return value.
711 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
712 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
713 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
714 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
715 (mips_can_change_mode_class): ...this new function, inverting the
716 sense of the return value.
717 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
718 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
719 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
720 (msp430_can_change_mode_class): New function.
721 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
722 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
723 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
724 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
725 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
726 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
727 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
728 (pa_cannot_change_mode_class): Replace with...
729 (pa_can_change_mode_class): ...this new function, inverting the
730 sense of the return value.
731 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
732 than CANNOT_CHANGE_MODE_CLASS in comments.
733 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
734 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
735 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
736 (pdp11_cannot_change_mode_class): Replace with...
737 (pdp11_can_change_mode_class): ...this new function, inverting the
738 sense of the return value.
739 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
740 * config/powerpcspe/powerpcspe-protos.h
741 (rs6000_cannot_change_mode_class_ptr): Delete.
742 * config/powerpcspe/powerpcspe.c
743 (rs6000_cannot_change_mode_class_ptr): Delete.
744 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
745 (rs6000_option_override_internal): Assign to
746 targetm.can_change_mode_class instead of
747 rs6000_cannot_change_mode_class_ptr.
748 (rs6000_cannot_change_mode_class): Replace with...
749 (rs6000_can_change_mode_class): ...this new function, inverting the
750 sense of the return value.
751 (rs6000_debug_cannot_change_mode_class): Replace with...
752 (rs6000_debug_can_change_mode_class): ...this new function.
753 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
754 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
755 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
756 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
757 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
759 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
760 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
761 (rs6000_option_override_internal): Assign to
762 targetm.can_change_mode_class instead of
763 rs6000_cannot_change_mode_class_ptr.
764 (rs6000_cannot_change_mode_class): Replace with...
765 (rs6000_can_change_mode_class): ...this new function, inverting the
766 sense of the return value.
767 (rs6000_debug_cannot_change_mode_class): Replace with...
768 (rs6000_debug_can_change_mode_class): ...this new function.
769 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
770 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
771 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
772 (s390_can_change_mode_class): ...this new function, inverting the
773 sense of the return value.
774 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
775 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
776 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
777 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
778 (sh_cannot_change_mode_class): Replace with...
779 (sh_can_change_mode_class): ...this new function, inverting the
780 sense of the return value.
781 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
782 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
783 (sparc_can_change_mode_class): New function.
784 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
785 * config/spu/spu.c (spu_can_change_mode_class): New function.
786 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
787 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
788 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
789 (visium_can_change_mode_class): New function.
790 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
792 2017-09-15 Richard Biener <rguenther@suse.de>
794 PR tree-optimization/82217
795 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
796 but not undefined case.
798 2017-09-15 Jakub Jelinek <jakub@redhat.com>
801 * postreload.c (reload_cse_simplify_operands): Skip
802 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
804 2017-09-15 Richard Biener <rguenther@suse.de>
806 PR tree-optimization/68823
807 * graphite-scop-detection.c (build_alias_set): If we have a
808 possible dependence check whether we can handle them by just
809 looking at the DRs DR_ACCESS_FNs.
810 (build_scops): If build_alias_set fails, fail the SCOP.
812 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
814 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
815 to support float128 built-in functions that require the ISA 3.0
817 (BU_FLOAT128_3_HW): Likewise.
818 (SQRTF128): Add support for the IEEE 128-bit square root and fma
822 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
823 support for built-in functions that need the ISA 3.0 IEEE 128-bit
824 floating point instructions.
825 (rs6000_invalid_builtin): Likewise.
826 (rs6000_builtin_mask_names): Likewise.
827 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
828 (RS6000_BTM_FLOAT128_HW): Likewise.
829 (RS6000_BTM_COMMON): Likewise.
830 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
832 * doc/extend.texi (RS/6000 built-in functions): Document the
833 IEEE 128-bit floating point square root and fused multiply-add
836 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
838 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
839 reg (r2) isn't in the set of registers defined in the prologue.
841 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
842 Alan Hayward <alan.hayward@arm.com>
843 David Sherwood <david.sherwood@arm.com>
845 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
846 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
847 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
848 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
849 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
851 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
852 max_vectorization_factor.
853 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
855 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
856 Alan Hayward <alan.hayward@arm.com>
857 David Sherwood <david.sherwood@arm.com>
859 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
860 (vect_worthwhile_without_simd_p): Declare.
861 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
862 (vectorizable_reduction): Use it.
863 * tree-vect-stmts.c (vectorizable_shift): Likewise.
864 (vectorizable_operation): Likewise.
866 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
867 Alan Hayward <alan.hayward@arm.com>
868 David Sherwood <david.sherwood@arm.com>
870 * tree-vectorizer.h (vect_get_num_copies): New function.
871 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
872 * tree-vect-loop.c (vectorizable_reduction): Likewise.
873 (vectorizable_induction): Likewise.
874 (vectorizable_live_operation): Likewise.
875 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
876 (vectorizable_bswap): Likewise.
877 (vectorizable_call): Likewise.
878 (vectorizable_conversion): Likewise.
879 (vectorizable_assignment): Likewise.
880 (vectorizable_shift): Likewise.
881 (vectorizable_operation): Likewise.
882 (vectorizable_store): Likewise.
883 (vectorizable_load): Likewise.
884 (vectorizable_condition): Likewise.
885 (vectorizable_comparison): Likewise.
886 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
888 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
889 Alan Hayward <alan.hayward@arm.com>
890 David Sherwood <david.sherwood@arm.com>
892 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
895 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
896 Alan Hayward <alan.hayward@arm.com>
897 David Sherwood <david.sherwood@arm.com>
899 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
900 an inline wrapper that provides a location.
901 (gimple_build_vector): Likewise.
902 * gimple-fold.c (gimple_build_vector_from_val): New function.
903 (gimple_build_vector): Likewise.
904 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
905 functions to build the initial value. Always return a gimple value.
906 (get_initial_defs_for_reduction): Likewise. Only compute
908 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
909 vect_init_vector on the results from get_initial_def(s)_for_reduction.
910 (vectorizable_induction): Use gimple_build_vector rather than
913 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
914 Alan Hayward <alan.hayward@arm.com>
915 David Sherwood <david.sherwood@arm.com>
917 * target.h (vec_perm_indices): New typedef.
918 (auto_vec_perm_indices): Likewise.
919 * optabs-query.h: Include target.h
920 (can_vec_perm_p): Take a vec_perm_indices *.
921 * optabs-query.c (can_vec_perm_p): Likewise.
922 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
923 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
924 * tree-vect-generic.c (lower_vec_perm): Likewise.
925 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
926 (vect_grouped_load_supported): Likewise.
927 (vect_shift_permute_load_chain): Likewise.
928 (vect_permute_store_chain): Use auto_vec_perm_indices.
929 (vect_permute_load_chain): Likewise.
930 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
931 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
932 Update uses of can_vec_perm_p.
933 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
934 mode with a number of elements. Take a vec_perm_indices *.
935 (vect_create_epilog_for_reduction): Update accordingly.
936 Use auto_vec_perm_indices.
937 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
938 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
939 (vect_transform_slp_perm_load): Likewise.
940 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
941 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
942 (vect_gen_perm_mask_checked): Likewise.
943 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
944 (vect_gen_perm_mask_checked): Likewise.
945 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
946 (vectorizable_store): Likewise.
947 (vectorizable_load): Likewise.
948 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
949 (vectorizable_bswap): Likewise.
951 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
952 Alan Hayward <alan.hayward@arm.com>
953 David Sherwood <david.sherwood@arm.com>
955 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
956 * tree.c (build_vector): Likewise.
957 (build_vector_from_ctor): Update accordingly.
958 (build_vector_from_val): Likewise.
959 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
960 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
961 * tree-vect-generic.c (add_rshift): Likewise.
962 (expand_vector_divmod): Likewise.
963 (optimize_vector_constructor): Likewise.
964 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
965 (vect_transform_slp_perm_load): Likewise.
966 (vect_schedule_slp_instance): Likewise.
967 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
968 (vectorizable_call): Likewise.
969 (vect_gen_perm_mask_any): Likewise. Add elements in order.
970 * expmed.c (make_tree): Likewise.
971 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
972 a vector passed to build_vector.
973 (fold_convert_const): Likewise.
974 (exact_inverse): Likewise.
975 (fold_ternary_loc): Likewise.
976 (fold_relational_const): Likewise.
977 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
978 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
979 (const_unop): Likewise. Store the reduction accumulator in a
980 variable rather than an array.
981 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
982 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
983 the new vector, rather than constructing it after the input arrays.
984 (native_interpret_vector): Use auto_vec<tree> when building
985 a vector passed to build_vector. Add elements in order.
986 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
987 auto_vec<tree> when building a vector passed to build_vector.
988 (vect_create_epilog_for_reduction): Likewise.
989 (vectorizable_induction): Likewise.
990 (get_initial_def_for_reduction): Likewise. Fix indentation of
992 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
994 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
995 passed to build_vector.
997 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
998 Alan Hayward <alan.hayward@arm.com>
999 David Sherwood <david.sherwood@arm.com>
1001 * tree-core.h (tree_base::u): Add an "nelts" field.
1002 (tree_vector): Use VECTOR_CST_NELTS as the length.
1003 * tree.c (tree_size): Likewise.
1004 (make_vector): Initialize VECTOR_CST_NELTS.
1005 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
1006 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
1007 TYPE_VECTOR_SUBPARTS.
1008 * expr.c (const_vector_mask_from_tree): Consistently use "units"
1009 as the number of units, setting it from VECTOR_CST_NELTS.
1010 (const_vector_from_tree): Likewise.
1011 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
1012 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
1013 (fold_negate_expr_1): Likewise.
1014 (fold_convert_const): Likewise.
1015 (const_binop): Likewise. Differentiate the number of output and
1017 (const_unop): Likewise.
1018 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
1019 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
1020 in cases that did the opposite.
1022 2017-09-14 Richard Biener <rguenther@suse.de>
1024 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
1027 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
1029 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
1031 2017-09-14 Jakub Jelinek <jakub@redhat.com>
1034 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
1035 if and where to split a bb, except for splitting before debug insn
1036 sequences followed by non-label real insn. Delete debug insns
1037 in between basic blocks.
1039 * combine.c (make_compound_operation_int): Formatting fixes.
1041 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
1042 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1043 * config/netbsd.h (LINK_EH_SPEC): Likewise.
1044 * config/sol2.h (LINK_EH_SPEC): Likewise.
1045 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1046 * config/s390/linux.h (LINK_SPEC): Likewise.
1047 * config/freebsd.h (LINK_EH_SPEC): Likewise.
1048 * config/openbsd.h (LINK_EH_SPEC): Likewise.
1049 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1050 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
1051 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
1052 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1053 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
1054 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
1056 2017-09-13 Jakub Jelinek <jakub@redhat.com>
1058 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
1060 (ENDFILE_LINUX_SPEC): Likewise.
1061 (LINK_EH_SPEC): Likewise.
1062 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
1063 (LINK_OS_LINUX_SPEC32): Likewise.
1064 (LINK_OS_LINUX_SPEC64): Likewise.
1065 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
1066 (LINK_OS_LINUX_SPEC): Likewise.
1068 2017-09-13 Martin Liska <mliska@suse.cz>
1071 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
1072 CASE_HIGH is NULL_TREE.
1074 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
1075 Alan Hayward <alan.hayward@arm.com>
1076 David Sherwood <david.sherwood@arm.com>
1078 * target.def (secondary_memory_needed): New hook.
1079 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
1080 instead of SECONDARY_MEMORY_NEEDED.
1081 (secondary_memory_needed_mode): Likewise.
1082 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
1083 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
1084 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
1085 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
1086 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
1087 * doc/tm.texi: Regenerate.
1088 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
1089 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
1090 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1091 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
1092 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
1093 * config/i386/i386.c (inline_secondary_memory_needed): Put the
1094 mode argument first and change the reg_class arguments to reg_class_t.
1095 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
1096 Make static. Update the call to inline_secondary_memory_needed.
1097 (ix86_register_move_cost): Update the call to
1098 inline_secondary_memory_needed.
1099 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1100 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
1102 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
1103 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
1105 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
1106 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
1107 * config/mips/mips.c (mips_secondary_memory_needed): Make static
1108 and match hook interface. Add comment from mips.h.
1109 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1110 * config/mmix/mmix.md (truncdfsf2): Refer to
1111 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
1113 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
1114 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
1115 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1116 (pa_secondary_memory_needed): New function.
1117 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
1118 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
1119 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1120 (pdp11_secondary_memory_needed): Make static and match hook interface.
1121 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
1122 * config/powerpcspe/powerpcspe-protos.h
1123 (rs6000_secondary_memory_needed_ptr): Delete.
1124 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
1126 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1127 (rs6000_option_override_internal): Assign to
1128 targetm.secondary_memory_needed rather than
1129 rs6000_secondary_memory_needed_ptr.
1130 (rs6000_secondary_memory_needed): Match hook interface.
1131 (rs6000_debug_secondary_memory_needed): Likewise.
1132 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
1133 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
1134 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
1135 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1136 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
1137 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
1139 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
1140 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1141 (rs6000_option_override_internal): Assign to
1142 targetm.secondary_memory_needed rather than
1143 rs6000_secondary_memory_needed_ptr.
1144 (rs6000_secondary_memory_needed): Match hook interface.
1145 (rs6000_debug_secondary_memory_needed): Likewise.
1146 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
1147 * config/s390/s390.c (s390_secondary_memory_needed): New function.
1148 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1149 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
1150 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1151 (sparc_secondary_memory_needed): New function.
1152 * lra-constraints.c (check_and_process_move): Refer to
1153 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
1155 (curr_insn_transform): Likewise.
1156 (process_alt_operands): Use targetm.secondary_memory_needed
1157 instead of TARGET_SECONDARY_MEMORY_NEEDED.
1158 (check_secondary_memory_needed_p): Likewise.
1159 (choose_split_class): Likewise.
1160 * reload.c: Unconditionally include code that was previously
1161 conditional on SECONDARY_MEMORY_NEEDED.
1162 (push_secondary_reload): Use targetm.secondary_memory_needed
1163 instead of TARGET_SECONDARY_MEMORY_NEEDED.
1164 (push_reload): Likewise.
1165 * reload1.c: Unconditionally include code that was previously
1166 conditional on SECONDARY_MEMORY_NEEDED.
1167 (choose_reload_regs): Use targetm.secondary_memory_needed
1168 instead of TARGET_SECONDARY_MEMORY_NEEDED.
1169 (gen_reload): Likewise.
1170 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
1172 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
1173 Alan Hayward <alan.hayward@arm.com>
1174 David Sherwood <david.sherwood@arm.com>
1176 * target.def (secondary_memory_needed_mode): New hook:
1177 * targhooks.c (default_secondary_memory_needed_mode): Declare.
1178 * targhooks.h (default_secondary_memory_needed_mode): New function.
1179 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
1180 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
1181 * doc/tm.texi: Regenerate.
1182 * lra-constraints.c (check_and_process_move): Use
1183 targetm.secondary_memory_needed_mode instead of
1184 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
1185 (curr_insn_transform): Likewise.
1186 * reload.c (get_secondary_mem): Likewise.
1187 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1188 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
1190 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1191 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1192 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
1193 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1194 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
1196 * config/powerpcspe/powerpcspe-protos.h
1197 (rs6000_secondary_memory_needed_mode): Delete.
1198 * config/powerpcspe/powerpcspe.c
1199 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1200 (rs6000_secondary_memory_needed_mode): Make static.
1201 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1202 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
1204 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
1206 (rs6000_secondary_memory_needed_mode): Make static.
1207 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1208 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
1209 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1210 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1211 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
1213 (sparc_secondary_memory_needed_mode): New function.
1214 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
1216 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
1218 * config/aarch64/constraints.md (Umq): New constraint.
1219 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
1221 (mov<mode>): Update condition.
1223 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1225 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
1226 when bitposition is the same.
1228 2017-09-13 Richard Biener <rguenther@suse.de>
1230 * dwarf2out.c (output_die_symbol): Remove.
1231 (output_die): Do not output a DIEs symbol.
1233 2017-09-13 Richard Biener <rguenther@suse.de>
1236 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
1237 default-def to avoid breaking iterator update with the weird
1238 interaction with cgraph_update_edges_for_call_stmt_node.
1240 2017-09-13 Richard Biener <rguenther@suse.de>
1242 * tree-cfg.c (verify_gimple_assign_binary): Add verification
1243 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
1244 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
1245 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
1247 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
1249 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
1250 Disable pc relative literal load irrespective of
1251 TARGET_FIX_ERR_A53_84341 for default.
1253 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
1255 * config/sparc/sparc.c (output_return): Output the source location of
1256 the insn in the delay slot, if any.
1257 (output_sibcall): Likewise.
1259 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
1262 * common.opt (-static-pie): New alias.
1263 (shared): Negate static-pie.
1264 (-no-pie): Update help text.
1266 (static-pie): New option.
1267 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
1268 -static-pie support.
1269 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
1270 (LINK_EH_SPEC): Likewise.
1271 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1272 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
1273 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
1274 * gcc.c (LINK_COMMAND_SPEC): Likewise.
1275 (init_gcc_specs): Likewise.
1276 (init_spec): Likewise.
1277 (display_help): Update help message for -pie.
1278 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
1281 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
1283 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
1284 (movdi_aarch64): Likewise.
1285 (movti_aarch64): Likewise.
1287 2017-09-12 Simon Wright <simon@pushface.org>
1290 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
1291 calculation of the minor version, always output as 0.
1293 2017-09-12 Jakub Jelinek <jakub@redhat.com>
1296 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
1297 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
1298 on it early, rather than manual conversion late. For
1299 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
1300 instead of performing manual conversion.
1302 2017-09-12 Carl Love <cel@us.ibm.com>
1304 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
1305 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
1307 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
1308 VMULOSW): Add definitions.
1309 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1310 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
1311 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
1312 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
1313 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
1315 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
1317 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
1319 (movti_aarch64): Likewise.
1320 (movdf_aarch64): Likewise.
1321 (movtf_aarch64): Likewise.
1322 (load_pairdi): Likewise.
1323 (store_pairdi): Likewise.
1324 (load_pairdf): Likewise.
1325 (store_pairdf): Likewise.
1326 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
1327 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
1328 (ldr_got_small_<mode>): Likewise.
1329 (ldr_got_small_28k_<mode>): Likewise.
1330 (ldr_got_tiny): Likewise.
1331 * config/aarch64/iterators.md (ldst_sz): New.
1332 (ldpstp_sz): Likewise.
1333 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
1335 (thunderx_load): Split load_8 to load_16.
1336 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
1338 (thunderx2t99_storepair_basic): Split store_8 to store_16.
1339 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
1340 (xgene1_store_pair): Split store_8 to store_16.
1341 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
1342 (falkor_st_0_st_sd): Split store_8 to store_16.
1344 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
1346 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
1347 and store1/2/3/4 to store_4/8/12/16.
1348 * config/aarch64/aarch64.md: Update for rename.
1349 * config/arm/arm.md: Likewise.: Likewise.
1350 * config/arm/arm.c: Likewise.
1351 * config/arm/thumb1.md: Likewise.
1352 * config/arm/thumb2.md: Likewise.
1353 * config/arm/vfp.md: Likewise.
1354 * config/arm/arm-generic.md: Likewise.
1355 * config/arm/arm1020e.md: Likewise.
1356 * config/arm/arm1026ejs.md: Likewise.
1357 * config/arm/arm1136jfs.md: Likewise.
1358 * config/arm/arm926ejs.md: Likewise.
1359 * config/arm/cortex-a15.md: Likewise.
1360 * config/arm/cortex-a17.md: Likewise.
1361 * config/arm/cortex-a5.md: Likewise.
1362 * config/arm/cortex-a53.md: Likewise.
1363 * config/arm/cortex-a57.md: Likewise.
1364 * config/arm/cortex-a7.md: Likewise.
1365 * config/arm/cortex-a8.md: Likewise.
1366 * config/arm/cortex-a9.md: Likewise.
1367 * config/arm/cortex-m4.md: Likewise.
1368 * config/arm/cortex-m7.md: Likewise.
1369 * config/arm/cortex-r4.md: Likewise.
1370 * config/arm/exynos-m1.md: Likewise.
1371 * config/arm/fa526.md: Likewise.
1372 * config/arm/fa606te.md: Likewise.
1373 * config/arm/fa626te.md: Likewise.
1374 * config/arm/fa726te.md: Likewise.
1375 * config/arm/fmp626.md: Likewise.
1376 * config/arm/iwmmxt.md: Likewise.
1377 * config/arm/ldmstm.md: Likewise.
1378 * config/arm/marvell-pj4.md: Likewise.
1379 * config/arm/xgene1.md: Likewise.
1380 * config/aarch64/thunderx.md: Likewise.
1381 * config/aarch64/thunderx2t99.md: Likewise.
1382 * config/aarch64/falkor.md: Likewise.
1384 2017-09-12 Martin Liska <mliska@suse.cz>
1386 * attribs.c (private_lookup_attribute): New function.
1387 * attribs.h (private_lookup_attribute): Declared here.
1388 (lookup_attribute): Called from this place.
1390 2017-09-12 Richard Biener <rguenther@suse.de>
1392 PR tree-optimization/82157
1393 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
1394 stmts with side-effects.
1396 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1397 Alan Hayward <alan.hayward@arm.com>
1398 David Sherwood <david.sherwood@arm.com>
1400 * target.def (hard_regno_nregs): New hook.
1401 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
1402 * targhooks.h (default_hard_regno_nregs): Declare.
1403 * targhooks.c (default_hard_regno_nregs): New function.
1404 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
1405 (TARGET_HARD_REGNO_NREGS): ...this hook.
1406 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
1407 (CLASS_MAX_NREGS): Likewise.
1408 * doc/tm.texi: Regenerate.
1409 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
1410 instead of HARD_REGNO_NREGS.
1411 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
1412 HARD_REGNO_NREGS in the comment.
1413 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
1414 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
1415 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
1416 Return an unsigned int.
1417 (TARGET_HARD_REGNO_NREGS): Redefine.
1418 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
1419 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
1420 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
1421 (arc_hard_regno_nregs): New function.
1422 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
1423 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
1424 (arm_hard_regno_nregs): New function.
1425 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
1426 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
1427 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
1428 (TARGET_HARD_REGNO_NREGS): Redefine.
1429 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
1430 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
1431 (HARD_REGNO_NREGS): Delete.
1432 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
1433 (cr16_hard_regno_nregs): New function.
1434 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
1435 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
1436 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
1437 (cris_hard_regno_nregs): New function.
1438 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
1439 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
1440 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
1441 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
1442 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
1443 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
1444 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
1445 (frv_hard_regno_nregs): Make static. Take and return an
1447 (frv_class_max_nregs): Remove outdated copy of documentation.
1448 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
1449 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
1450 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
1451 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
1452 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
1453 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
1454 (TARGET_HARD_REGNO_NREGS): Redefine.
1455 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
1456 (CLASS_MAX_NREGS): Update comment.
1457 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
1458 (ia64_hard_regno_nregs): New function.
1459 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
1460 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
1461 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
1462 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
1463 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
1465 (m32c_hard_regno_nregs): Likewise. Make static.
1466 (TARGET_HARD_REGNO_NREGS): Redefine.
1467 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
1468 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
1469 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
1470 (m68k_hard_regno_nregs): New function.
1471 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
1472 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
1473 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
1474 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
1475 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
1476 Take and return an unsigned int.
1477 (TARGET_HARD_REGNO_NREGS): Redefine.
1478 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
1479 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
1480 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
1481 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
1482 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
1483 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
1484 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
1485 (msp430_hard_regno_nregs): Make static. Take and return an
1487 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
1488 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
1489 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
1490 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
1491 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
1492 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
1493 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
1494 (TARGET_HARD_REGNO_NREGS): Redefine.
1495 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
1496 (PA_HARD_REGNO_NREGS): ...this.
1497 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
1498 (PA_HARD_REGNO_NREGS): ...this.
1499 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
1500 (pa_hard_regno_nregs): New function.
1501 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
1502 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
1503 (pdp11_hard_regno_nregs): New function.
1504 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
1505 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
1506 (rs6000_hard_regno_nregs_hook): New function.
1507 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
1508 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1509 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
1510 Take and return an unsigned int. Move earlier in file.
1511 (TARGET_HARD_REGNO_NREGS): Redefine.
1512 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
1513 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
1514 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
1515 (rl78_hard_regno_nregs): Make static. Take and return an
1517 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
1518 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
1519 (rs6000_hard_regno_nregs_hook): New function.
1520 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
1521 * config/rx/rx.c (rx_hard_regno_nregs): New function.
1522 (TARGET_HARD_REGNO_NREGS): Redefine.
1523 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
1524 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
1525 instead of HARD_REGNO_NREGS.
1526 (s390_hard_regno_nregs): New function.
1527 (s390_hard_regno_mode_ok): Add comment from s390.h.
1528 (TARGET_HARD_REGNO_NREGS): Redefine.
1529 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
1530 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
1531 (sh_hard_regno_nregs): New function.
1532 (sh_pass_in_reg_p): Use it.
1533 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
1534 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
1535 (sparc_hard_regno_nregs): New function.
1536 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
1537 * config/spu/spu.c (spu_hard_regno_nregs): New function.
1538 (spu_function_arg_advance): Use it, supplying a valid register number.
1539 (TARGET_HARD_REGNO_NREGS): Redefine.
1540 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
1541 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
1542 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
1543 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
1544 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
1545 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
1546 (CLASS_MAX_NREGS): Remove copy of old documentation.
1547 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
1548 (visium_hard_regno_nregs): New function.
1549 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
1550 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
1551 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
1552 xtensa_hard_regno_nregs): New function.
1553 * system.h (HARD_REGNO_NREGS): Poison.
1555 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1557 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
1558 hard_regno_nregs instead of HARD_REGNO_NREGS.
1559 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
1560 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
1561 (c6x_expand_epilogue): Likewise.
1562 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
1563 (frv_read_iacc_argument): Likewise.
1564 * config/sh/sh.c: Include regs.h.
1565 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
1566 (regs_used): Likewise.
1567 (output_stack_adjust): Likewise.
1568 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
1569 * expmed.c: Include regs.h.
1570 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
1571 * ree.c: Include regs.h.
1572 (combine_reaching_defs): Use hard_regno_nregs instead of
1574 (add_removable_extension): Likewise.
1576 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1578 * regs.h (hard_regno_nregs): Turn into a function.
1579 (end_hard_regno): Update accordingly.
1580 * caller-save.c (setup_save_areas): Likewise.
1581 (save_call_clobbered_regs): Likewise.
1582 (replace_reg_with_saved_mem): Likewise.
1583 (insert_restore): Likewise.
1584 (insert_save): Likewise.
1585 * combine.c (can_change_dest_mode): Likewise.
1586 (move_deaths): Likewise.
1587 (distribute_notes): Likewise.
1588 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
1589 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
1590 (rs6000_split_multireg_move): Likewise.
1591 (rs6000_register_move_cost): Likewise.
1592 (rs6000_memory_move_cost): Likewise.
1593 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
1594 (rs6000_split_multireg_move): Likewise.
1595 (rs6000_register_move_cost): Likewise.
1596 (rs6000_memory_move_cost): Likewise.
1597 * cselib.c (cselib_reset_table): Likewise.
1598 (cselib_lookup_1): Likewise.
1599 * emit-rtl.c (set_mode_and_regno): Likewise.
1600 * function.c (aggregate_value_p): Likewise.
1601 * ira-color.c (setup_profitable_hard_regs): Likewise.
1602 (check_hard_reg_p): Likewise.
1603 (calculate_saved_nregs): Likewise.
1604 (assign_hard_reg): Likewise.
1605 (improve_allocation): Likewise.
1606 (calculate_spill_cost): Likewise.
1607 * ira-emit.c (modify_move_list): Likewise.
1608 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
1609 (ira_hard_reg_in_set_p): Likewise.
1610 * ira.c (setup_reg_mode_hard_regset): Likewise.
1611 (clarify_prohibited_class_mode_regs): Likewise.
1612 (check_allocation): Likewise.
1613 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1614 (lra_setup_reg_renumber): Likewise.
1615 (setup_try_hard_regno_pseudos): Likewise.
1616 (spill_for): Likewise.
1617 (assign_hard_regno): Likewise.
1618 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1619 * lra-constraints.c (in_class_p): Likewise.
1620 (lra_constraint_offset): Likewise.
1621 (simplify_operand_subreg): Likewise.
1622 (lra_constraints): Likewise.
1623 (split_reg): Likewise.
1624 (split_if_necessary): Likewise.
1625 (invariant_p): Likewise.
1626 (inherit_in_ebb): Likewise.
1627 * lra-lives.c (process_bb_lives): Likewise.
1628 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
1629 (get_hard_regs): Likewise.
1630 (do_remat): Likewise.
1631 * lra-spills.c (assign_spill_hard_regs): Likewise.
1632 * mode-switching.c (create_pre_exit): Likewise.
1633 * postreload.c (reload_combine_recognize_pattern): Likewise.
1634 * recog.c (peep2_find_free_register): Likewise.
1635 * regcprop.c (kill_value_regno): Likewise.
1636 (set_value_regno): Likewise.
1637 (copy_value): Likewise.
1638 (maybe_mode_change): Likewise.
1639 (find_oldest_value_reg): Likewise.
1640 (copyprop_hardreg_forward_1): Likewise.
1641 * regrename.c (check_new_reg_p): Likewise.
1642 (regrename_do_replace): Likewise.
1643 * reload.c (push_reload): Likewise.
1644 (combine_reloads): Likewise.
1645 (find_dummy_reload): Likewise.
1646 (operands_match_p): Likewise.
1647 (find_reloads): Likewise.
1648 (find_equiv_reg): Likewise.
1649 (reload_adjust_reg_for_mode): Likewise.
1650 * reload1.c (count_pseudo): Likewise.
1651 (count_spilled_pseudo): Likewise.
1652 (find_reg): Likewise.
1653 (clear_reload_reg_in_use): Likewise.
1654 (free_for_value_p): Likewise.
1655 (allocate_reload_reg): Likewise.
1656 (choose_reload_regs): Likewise.
1657 (reload_adjust_reg_for_temp): Likewise.
1658 (emit_reload_insns): Likewise.
1659 (delete_output_reload): Likewise.
1660 * rtlanal.c (subreg_get_info): Likewise.
1661 * sched-deps.c (sched_analyze_reg): Likewise.
1662 * sel-sched.c (init_regs_for_mode): Likewise.
1663 (mark_unavailable_hard_regs): Likewise.
1664 (choose_best_reg_1): Likewise.
1665 (verify_target_availability): Likewise.
1666 * valtrack.c (dead_debug_insert_temp): Likewise.
1667 * var-tracking.c (track_loc_p): Likewise.
1668 (emit_note_insn_var_location): Likewise.
1669 * varasm.c (make_decl_rtl): Likewise.
1670 * reginfo.c (choose_hard_reg_mode): Likewise.
1671 (init_reg_modes_target): Refer directly to
1672 this_target_regs->x_hard_regno_nregs.
1674 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1676 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
1677 instead of hard_regno_nregs.
1679 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1681 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
1682 end_hard_regno instead of hard_regno_nregs.
1683 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
1684 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
1685 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
1686 * ira-color.c (improve_allocation): Likewise.
1687 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1688 * lra-lives.c (mark_regno_live): Likewise.
1689 (mark_regno_dead): Likewise.
1690 * lra-remat.c (operand_to_remat): Likewise.
1691 * lra.c (collect_non_operand_hard_regs): Likewise.
1692 * postreload.c (reload_combine_note_store): Likewise.
1693 (move2add_valid_value_p): Likewise.
1694 * reload.c (regno_clobbered_p): Likewise.
1696 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1698 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
1700 * config/v850/v850.c (v850_reorg): Likewise.
1701 * reload.c (refers_to_regno_for_reload_p): Likewise.
1702 (find_equiv_reg): Likewise.
1703 * reload1.c (reload_reg_reaches_end_p): Likewise.
1705 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1707 * caller-save.c (add_used_regs): Use REG_NREGS instead of
1709 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
1710 * config/arm/arm.c (output_move_neon): Likewise.
1711 (arm_attr_length_move_neon): Likewise.
1712 (neon_split_vcombine): Likewise.
1713 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
1714 (c6x_mark_reg_written): Likewise.
1715 (c6x_dwarf_register_span): Likewise.
1716 * config/i386/i386.c (ix86_save_reg): Likewise.
1717 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
1718 (rws_access_reg): Likewise.
1719 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
1720 * mode-switching.c (create_pre_exit): Likewise.
1721 * ree.c (combine_reaching_defs): Likewise.
1722 (add_removable_extension): Likewise.
1723 * regcprop.c (find_oldest_value_reg): Likewise.
1724 (copyprop_hardreg_forward_1): Likewise.
1725 * reload.c (reload_inner_reg_of_subreg): Likewise.
1726 (push_reload): Likewise.
1727 (combine_reloads): Likewise.
1728 (find_dummy_reload): Likewise.
1729 (reload_adjust_reg_for_mode): Likewise.
1730 * reload1.c (find_reload_regs): Likewise.
1731 (forget_old_reloads_1): Likewise.
1732 (reload_reg_free_for_value_p): Likewise.
1733 (reload_adjust_reg_for_temp): Likewise.
1734 (emit_reload_insns): Likewise.
1735 (delete_output_reload): Likewise.
1736 * sel-sched.c (choose_best_reg_1): Likewise.
1737 (choose_best_pseudo_reg): Likewise.
1739 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1740 Alan Hayward <alan.hayward@arm.com>
1741 David Sherwood <david.sherwood@arm.com>
1743 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
1744 * target.def (slow_unaligned_access): New hook.
1745 * targhooks.h (default_slow_unaligned_access): Declare.
1746 * targhooks.c (default_slow_unaligned_access): New function.
1747 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
1748 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
1749 * doc/tm.texi: Regenerate.
1750 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
1751 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
1752 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
1754 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
1755 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
1757 (rs6000_slow_unaligned_access): New function.
1758 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
1759 (expand_block_compare): Likewise.
1760 (expand_strn_compare): Likewise.
1761 (rs6000_rtx_costs): Likewise.
1762 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
1763 (riscv_slow_unaligned_access): Likewise.
1764 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
1765 (riscv_slow_unaligned_access_p): ...this and make static.
1766 (riscv_option_override): Update accordingly.
1767 (riscv_slow_unaligned_access): New function.
1768 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
1769 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
1770 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
1771 (rs6000_slow_unaligned_access): New function.
1772 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
1773 (rs6000_rtx_costs): Likewise.
1774 * config/rs6000/rs6000-string.c (expand_block_compare)
1775 (expand_strn_compare): Use targetm.slow_unaligned_access instead
1776 of SLOW_UNALIGNED_ACCESS.
1777 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
1778 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
1779 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
1780 of SLOW_UNALIGNED_ACCESS.
1781 * expmed.c (simple_mem_bitfield_p): Likewise.
1782 * expr.c (alignment_for_piecewise_move): Likewise.
1783 (emit_group_load_1): Likewise.
1784 (emit_group_store): Likewise.
1785 (copy_blkmode_from_reg): Likewise.
1786 (emit_push_insn): Likewise.
1787 (expand_assignment): Likewise.
1788 (store_field): Likewise.
1789 (expand_expr_real_1): Likewise.
1790 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1791 * lra-constraints.c (simplify_operand_subreg): Likewise.
1792 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
1793 * gimple-ssa-store-merging.c: Likewise in block comment at start
1795 * tree-ssa-strlen.c: Include target.h.
1796 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
1797 of SLOW_UNALIGNED_ACCESS.
1798 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
1800 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1802 PR rtl-optimization/82185
1803 * expmed.c (emit_store_flag_int): Only test tem if it has been
1806 2017-09-12 Richard Biener <rguenther@suse.de>
1809 * match.pd ((FTYPE) N CMP CST): Fix typo.
1811 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
1813 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
1815 (mips_near_type_p): Add 'short_call' attribute as a synonym
1817 * doc/extend.texi (short_call): Document new function attribute.
1819 2017-09-12 Jakub Jelinek <jakub@redhat.com>
1822 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
1823 assertion check that in the condition.
1824 (get_atomic_generic_size): Likewise. Before testing if parameter
1825 has pointer type, if it has array type, call for C++
1826 default_conversion to perform array-to-pointer conversion.
1828 2017-09-12 Richard Biener <rguenther@suse.de>
1830 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
1831 for operations we cannot scalarize.
1833 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
1835 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
1836 vectors heap vectors. Clean up comments.
1837 Make visited_bbs a reference.
1838 (profitable_jump_thread_path): Make GC
1839 vectors heap vectors. Clean up comments.
1841 (convert_and_register_jump_thread_path): Make GC vectors heap
1843 (check_subpath_and_update_thread_path): Same. Clean up comments.
1844 Make visited_bbs a reference.
1845 (handle_phi): Abstract common code to to
1846 register_jump_thread_path_if_profitable.
1847 Rename VAR_BB to DEF_BB.
1849 Make GC vectors heap vectors.
1850 Make visited_bbs a reference.
1851 (handle_assignment): Same.
1852 (register_jump_thread_path_if_profitable): New.
1853 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
1855 Make GC vectors heap vectors. Clean up comments.
1856 Make visited_bbs a reference.
1857 (find_jump_threads_backwards): Make visited_bbs live in the stack.
1858 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
1861 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
1864 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
1865 words of E_DImode object are reachable by xtensa_uimm8x4 access.
1867 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
1869 Revert r251800 and r251799.
1871 2017-09-11 Martin Jambor <mjambor@suse.cz>
1874 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
1875 arguments in advance.
1876 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
1877 use it to find predecessor edges.
1878 (naive_outof_ssa): Collect vector of predecessors.
1880 2017-09-08 Jason Merrill <jason@redhat.com>
1882 PR c++/70029 - ICE with ref-qualifier and -flto
1883 * langhooks.h (struct lang_hooks_for_types): Add
1884 copy_lang_qualifiers.
1885 * attribs.c (build_type_attribute_qual_variant): Use it.
1886 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
1888 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
1889 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
1891 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
1894 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
1895 (*mulsi3_sp64): New instruction.
1896 (mulsi3): New expander.
1898 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
1900 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
1902 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1904 * sancov.c: Include memmodel.h.
1906 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
1909 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
1912 2017-09-07 Carl Love <cel@us.ibm.com>
1914 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
1915 the sldi instruction.
1917 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
1919 * sancov.c: Include tm_p.h.
1921 2017-09-07 Jakub Jelinek <jakub@redhat.com>
1924 * output.h (switch_to_other_text_partition): New declaration.
1925 * varasm.c (switch_to_other_text_partition): New function.
1926 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
1927 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
1928 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
1929 to the other text partition before emitting LCL label and switch back
1930 after emitting the word after it.
1932 2017-09-07 Richard Biener <rguenther@suse.de>
1934 * passes.def (pass_split_crit_edges): Remove instance before PRE.
1935 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
1936 critical edges here, after loop init.
1937 (pass_data_pre): Remove PROP_no_crit_edges flags.
1938 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
1939 for valueization of call args to avoid leaking VN_TOP.
1940 (visit_use): Assert we do not visit default defs.
1941 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
1942 Use error_mark_node to more easily detect leaking VN_TOP.
1943 All default-defs are varying, not VN_TOP. Mark them visited.
1944 (run_scc_vn): Make code match comment.
1946 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
1948 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
1949 OPTION_MASK_FLOAT128_KEYWORD.
1950 (POWERPC_MASKS): Likewise.
1951 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
1952 support for the -mfloat128-type option, and make -mfloat128
1953 default on PowerPC Linux systems. Define or undefine
1954 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
1955 Define __float128 to be __ieee128 if IEEE 128-bit support is
1956 enabled, or undefine it.
1957 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
1958 Delete defining __FLOAT128_TYPE__.
1959 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
1960 -mfloat128-type option and make -mfloat128 default on PowerPC
1962 (TARGET_FLOAT128_TYPE): Likewise.
1963 (-mfloat128-type): Likewise.
1964 * config/rs6000/rs6000.c (rs6000_option_override_internal):
1965 Delete the -mfloat128-type option and make -mfloat128 default on
1966 PowerPC Linux systems. Always use __ieee128 to be the keyword for
1967 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
1968 128-bit floating point is enabled. Change tests from using
1969 -mfloat128-type to -mfloat128.
1970 (rs6000_mangle_type): Use the correct mangling for the __float128
1971 type even if normal long double is restricted to 64-bits.
1972 (floatn_mode): Enable the _Float128 type by default on VSX Linux
1974 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
1975 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
1976 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
1978 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
1979 documentation for -mfloat128.
1981 2017-09-06 Olivier Hainque <hainque@adacore.com>
1983 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
1985 2017-09-06 Wish Wu <wishwu007@gmail.com>
1986 Jakub Jelinek <jakub@redhat.com>
1988 * asan.c (initialize_sanitizer_builtins): Add
1989 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
1990 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
1991 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
1992 BT_FN_VOID_UINT64_PTR variables.
1993 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
1994 (BT_FN_VOID_UINT16_UINT16): Likewise.
1995 (BT_FN_VOID_UINT32_UINT32): Likewise.
1996 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
1997 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
1998 (BT_FN_VOID_UINT64_PTR): Likewise.
1999 * common.opt (flag_sanitize_coverage): New variable.
2000 (fsanitize-coverage=trace-pc): Remove.
2001 (fsanitize-coverage=): Add.
2002 * flag-types.h (enum sanitize_coverage_code): New enum.
2003 * fold-const.c (fold_range_test): Disable non-short-circuit
2004 optimization if flag_sanitize_coverage.
2005 (fold_truth_andor): Likewise.
2006 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
2007 * opts.c (COVERAGE_SANITIZER_OPT): Define.
2008 (coverage_sanitizer_opts): New array.
2009 (get_closest_sanitizer_option): Add OPTS argument, handle also
2010 OPT_fsanitize_coverage_.
2011 (parse_sanitizer_options): Adjusted to also handle
2012 OPT_fsanitize_coverage_.
2013 (common_handle_option): Add OPT_fsanitize_coverage_.
2014 * sancov.c (instrument_comparison, instrument_switch): New function.
2015 (sancov_pass): Add trace-cmp support.
2016 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
2017 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
2018 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
2019 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
2020 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
2021 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
2022 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
2023 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
2024 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
2025 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
2027 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
2029 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
2030 error. Only quit immediately if parsing is complete.
2031 (BEGIN): Initialize fatal_err and parse_done.
2032 (begin fpu, end fpu): Check number of arguments.
2033 (begin arch, end arch): Likewise.
2034 (begin cpu, end cpu): Likewise.
2035 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
2036 (optalias): Likewise.
2038 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
2040 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
2041 * config/arm/arm-isa.h: Delete. Move definitions to ...
2042 * arm-cpus.in: ... here. Use new feature and fgroup values.
2043 * config/arm/arm.c (arm_option_override): Use lower case for feature
2045 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
2046 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
2047 * config/arm/parsecpu.awk (END): Add new command 'isa'.
2049 (print_isa_bits_for): New function.
2050 (gen_isa): New function.
2051 (gen_comm_data): Use print_isa_bits_for.
2052 (define feature): New keyword.
2053 (define fgroup): New keyword.
2054 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
2055 (arm-isa.h): Add rule to generate file.
2056 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
2057 case for feature bit names.
2059 2017-09-06 Richard Biener <rguenther@suse.de>
2061 * tree-ssa-pre.c (NECESSARY): Remove.
2062 (create_expression_by_pieces): Do not touch pass-local flags.
2063 (insert_into_preds_of_block): Likewise.
2064 (do_pre_regular_insertion): Likewise.
2065 (eliminate_insert): Likewise.
2066 (eliminate_dom_walker::before_dom_children): Likewise.
2067 (fini_eliminate): Do not look at inserted_exprs.
2068 (mark_operand_necessary): Remove.
2069 (remove_dead_inserted_code): Replace with simple work-list
2070 algorithm based on inserted_exprs and SSA uses.
2071 (pass_pre::execute): Re-order fini_eliminate and
2072 remove_dead_inserted_code.
2074 2017-09-06 Olivier Hainque <hainque@adacore.com>
2076 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
2077 for VxWorks 7. Adjust surrounding comments.
2079 2017-09-06 Richard Biener <rguenther@suse.de>
2081 * gimple-ssa-strength-reduction.c
2082 (find_candidates_dom_walker::before_dom_children): Also allow
2085 2017-09-06 Richard Biener <rguenther@suse.de>
2087 PR tree-optimization/82108
2088 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
2089 for gap in the non-permutation SLP case.
2091 2017-09-06 Martin Jambor <mjambor@suse.cz>
2093 PR tree-optimization/82078
2094 * tree-sra.c (sort_and_splice_var_accesses): Move call to
2095 add_access_to_work_queue...
2096 (build_accesses_from_assign): ...here.
2097 (propagate_all_subaccesses): Make sure racc is the group
2098 representative, if there is one.
2100 2017-09-06 Jakub Jelinek <jakub@redhat.com>
2103 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
2106 2017-09-06 Richard Biener <rguenther@suse.de>
2108 * gimple-ssa-strength-reduction.c
2109 (find_candidates_dom_walker::before_doom_children): Use a
2110 type and not a mode check.
2112 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
2115 * config/arm/predicates.md (arm_general_adddi_operand): Create new
2117 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
2119 2017-09-05 Jeff Law <law@redhat.com>
2121 PR tree-optimization/64910
2122 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
2123 cases where we have 3 or more operands.
2125 2017-09-05 Jakub Jelinek <jakub@redhat.com>
2128 * omp-low.c (lower_omp_for): Recompute tree invariant if
2129 gimple_omp_for_initial/final is ADDR_EXPR.
2132 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
2133 into gimple val before gimplification fo the COND_EXPR.
2135 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
2137 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
2138 REGION_COPY argument.
2139 (thread_through_all_blocks): Remove unused argument to
2140 duplicate_thread_path.
2142 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2143 Alan Hayward <alan.hayward@arm.com>
2144 David Sherwood <david.sherwood@arm.com>
2146 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
2147 Take a scalar_mode rather than a machine_mode.
2148 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2149 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
2150 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2151 (aarch64_gen_adjusted_ldpstp): Likewise.
2152 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
2154 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2155 Alan Hayward <alan.hayward@arm.com>
2156 David Sherwood <david.sherwood@arm.com>
2158 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
2159 Take a scalar_int_mode instead of a machine_mode.
2160 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
2161 (aarch64_output_scalar_simd_mov_immediate): Likewise.
2162 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
2163 (aarch64_simd_attr_length_rglist): Delete.
2164 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
2165 a scalar_int_mode instead of a machine_mode.
2166 (aarch64_add_offset): Likewise.
2167 (aarch64_internal_mov_immediate): Likewise
2168 (aarch64_add_constant_internal): Likewise.
2169 (aarch64_add_constant): Likewise.
2170 (aarch64_movw_imm): Likewise.
2171 (aarch64_rtx_arith_op_extract_p): Likewise.
2172 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
2173 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
2174 Remove assert that the mode isn't a vector.
2175 (aarch64_output_scalar_simd_mov_immediate): Likewise.
2176 (aarch64_expand_mov_immediate): Update calls after above changes.
2177 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
2178 (aarch64_and_bitmask_imm): Check for scalar integer modes.
2179 (aarch64_move_imm): Likewise.
2180 (aarch64_can_const_movi_rtx_p): Likewise.
2181 (aarch64_strip_extend): Likewise.
2182 (aarch64_extr_rtx_p): Likewise.
2183 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
2184 a CONST_INT when the mode parameter is VOIDmode.
2185 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
2187 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2189 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
2190 * stor-layout.c (bitwise_mode_for_mode): Likewise.
2191 (bitwise_type_for_mode): Update accordingly.
2193 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2195 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
2196 * stor-layout.c (mode_for_size_tree): Likewise.
2197 (mode_for_array): Update accordingly.
2198 (layout_decl): Likewise.
2199 (compute_record_mode): Likewise. Only set the mode once.
2201 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2203 * target.def (get_mask_mode): Change return type to opt_mode.
2205 * doc/tm.texi: Regenerate.
2206 * targhooks.h (default_get_mask_mode): Return an opt_mode.
2207 * targhooks.c (default_get_mask_mode): Likewise.
2208 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
2209 * optabs-query.c (can_vec_mask_load_store_p): Update use of
2210 targetm.get_mask_mode.
2211 * tree.c (build_truth_vector_type): Likewise.
2213 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2215 * machmode.h (mode_for_vector): Return an opt_mode.
2216 * stor-layout.c (mode_for_vector): Likewise.
2217 (mode_for_int_vector): Update accordingly.
2218 (layout_type): Likewise.
2219 * config/i386/i386.c (emit_memmov): Likewise.
2220 (ix86_expand_set_or_movmem): Likewise.
2221 (ix86_expand_vector_init): Likewise.
2222 (ix86_get_mask_mode): Likewise.
2223 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
2225 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
2226 * expmed.c (extract_bit_field_1): Likewise.
2227 * expr.c (expand_expr_real_2): Likewise.
2228 * optabs-query.c (can_vec_perm_p): Likewise.
2229 (can_vec_mask_load_store_p): Likewise.
2230 * optabs.c (expand_vec_perm): Likewise.
2231 * targhooks.c (default_get_mask_mode): Likewise.
2232 * tree-vect-stmts.c (vectorizable_store): Likewise.
2233 (vectorizable_load): Likewise.
2234 (get_vectype_for_scalar_type_and_size): Likewise.
2236 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2238 * machmode.h (mode_for_int_vector): New function.
2239 * stor-layout.c (mode_for_int_vector): Likewise.
2240 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
2241 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
2242 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
2243 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
2244 (s390_expand_vcond): Likewise.
2246 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2248 * machmode.h (opt_machine_mode): New type.
2249 (opt_mode<T>): Allow construction from anything that can be
2251 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
2252 (mode_for_size): Return an opt_machine_mode.
2253 * stor-layout.c (mode_for_size): Likewise.
2254 (mode_for_size_tree): Update call accordingly.
2255 (bitwise_mode_for_mode): Likewise.
2256 (make_fract_type): Likewise.
2257 (make_accum_type): Likewise.
2258 * caller-save.c (replace_reg_with_saved_mem): Update call
2260 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2261 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2262 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2263 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2264 * expmed.c (extract_bit_field_1): Likewise.
2265 * reload.c (get_secondary_mem): Likewise.
2266 * varasm.c (assemble_integer): Likewise.
2267 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
2270 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2272 * machmode.h (decimal_float_mode_for_size): New function.
2273 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
2274 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
2275 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
2276 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
2277 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
2278 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
2280 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2282 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
2283 (get_builtin_sync_mode): Likewise.
2284 (expand_ifn_atomic_compare_exchange): Likewise.
2285 (expand_builtin_atomic_clear): Likewise.
2286 (expand_builtin_atomic_test_and_set): Likewise.
2287 (fold_builtin_atomic_always_lock_free): Likewise.
2288 * calls.c (compute_argument_addresses): Likewise.
2289 (emit_library_call_value_1): Likewise.
2290 (store_one_arg): Likewise.
2291 * combine.c (combine_instructions): Likewise.
2292 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
2293 * config/arm/arm.c (arm_function_value): Likewise.
2294 (aapcs_allocate_return_reg): Likewise.
2295 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
2296 * config/i386/i386.c (construct_container): Likewise.
2297 (ix86_gimplify_va_arg): Likewise.
2298 (ix86_expand_sse_cmp): Likewise.
2299 (emit_memmov): Likewise.
2300 (emit_memset): Likewise.
2301 (expand_small_movmem_or_setmem): Likewise.
2302 (ix86_expand_pextr): Likewise.
2303 (ix86_expand_pinsr): Likewise.
2304 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
2305 * config/microblaze/microblaze.c (microblaze_block_move_straight):
2307 * config/mips/mips.c (mips_function_value_1) Likewise.
2308 (mips_block_move_straight): Likewise.
2309 (mips_expand_ins_as_unaligned_store): Likewise.
2310 * config/powerpcspe/powerpcspe.c
2311 (rs6000_darwin64_record_arg_advance_flush): Likewise.
2312 (rs6000_darwin64_record_arg_flush): Likewise.
2313 * config/rs6000/rs6000.c
2314 (rs6000_darwin64_record_arg_advance_flush): Likewise.
2315 (rs6000_darwin64_record_arg_flush): Likewise.
2316 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
2317 (sparc_function_value_1): Likewise.
2318 * config/spu/spu.c (adjust_operand): Likewise.
2319 (spu_emit_branch_or_set): Likewise.
2320 (arith_immediate_p): Likewise.
2321 * emit-rtl.c (gen_lowpart_common): Likewise.
2322 * expr.c (expand_expr_real_1): Likewise.
2323 * function.c (assign_parm_setup_block): Likewise.
2324 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
2325 * reload1.c (alter_reg): Likewise.
2326 * stor-layout.c (mode_for_vector): Likewise.
2327 (layout_type): Likewise.
2329 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2331 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
2332 (spu_convert_move): Likewise.
2333 * lower-subreg.c (resolve_simple_move): Likewise.
2335 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2338 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
2339 define_insn to a define_expand.
2340 (altivec_vsum2sws_direct): New define_insn.
2341 (altivec_vsumsws): Convert from a define_insn to a define_expand.
2343 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
2345 * config/arm/arm.c (arm_option_params_internal): Improve setting of
2348 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
2353 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
2354 realign stack if stack alignment needed is less than incoming
2357 2017-09-05 Marek Polacek <polacek@redhat.com>
2360 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
2363 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
2365 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
2367 2017-09-05 Richard Biener <rguenther@suse.de>
2369 PR tree-optimization/82084
2370 * fold-const.c (can_native_encode_string_p): Handle wide characters.
2372 2017-09-05 Richard Biener <rguenther@suse.de>
2374 PR tree-optimization/82102
2375 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
2377 2017-09-05 Martin Liska <mliska@suse.cz>
2379 PR tree-optimization/82032
2380 * tree-cfg.c (generate_range_test): New function.
2381 * tree-cfg.h (generate_range_test): Declared here.
2382 * tree-cfgcleanup.c (convert_single_case_switch): New function.
2383 (cleanup_control_expr_graph): Use it.
2384 * tree-switch-conversion.c (try_switch_expansion): Remove
2386 (emit_case_nodes): Use generate_range_test.
2388 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
2391 * config/i386/i386.md (*<btsc><mode>_mask): Add
2392 TARGET_USE_BT to insn constraint.
2393 (*btr<mode>_mask): Ditto.
2395 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
2397 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
2398 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
2400 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2403 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
2404 TARGET_NEON and TARGET_IWMMXT.
2405 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
2406 TARGET_NEON and TARGET_IWMMXT.
2407 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
2409 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
2411 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
2412 (ix86_rewrite_tls_address): Ditto.
2413 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
2414 (ix86_rewrite_tls_address_1): Ditto.
2415 (ix86_rewrite_tls_address): Ditto.
2416 * config/i386/predicates.md (tls_address_pattern): New predicate.
2417 * config/i386/i386.md (TLS address splitter): New splitter.
2419 2017-09-04 Richard Biener <rguenther@suse.de>
2421 PR tree-optimization/82084
2422 * fold-const.h (can_native_encode_string_p): Declare.
2423 * fold-const.c (can_native_encode_string_p): Factor out from ...
2424 (native_encode_string): ... here.
2425 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
2426 vectorizing stores from constants we later cannot handle.
2428 2017-09-04 Marek Polacek <polacek@redhat.com>
2431 * doc/invoke.texi: Update -Wtautological-compare documentation.
2433 2017-09-04 Jeff Law <law@redhat.com>
2435 PR tree-optimization/64910
2436 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
2437 swap the first and last operand if the last is a constant.
2439 2017-09-04 Marek Polacek <polacek@redhat.com>
2442 * convert.c (do_narrow): When sanitizing signed integer overflows,
2443 bail out for signed types.
2444 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
2446 2017-09-04 Richard Biener <rguenther@suse.de>
2448 PR tree-optimization/82060
2449 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
2450 Move devirtualization after stmt folding and before EH/AB/noreturn
2451 cleanup to get the stmt refs canonicalized. Use a bool instead
2452 of gimple_modified_p since that doesn't work for NOPs. Schedule
2453 NOPs generated by folding for removal.
2455 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2456 Alan Hayward <alan.hayward@arm.com>
2457 David Sherwood <david.sherwood@arm.com>
2459 * coretypes.h (pad_direction): New enum.
2460 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
2461 (FUNCTION_ARG_PADDING): Likewise.
2462 * target.def (function_arg_padding): New hook.
2463 * targhooks.h (default_function_arg_padding): Declare.
2464 * targhooks.c (default_function_arg_padding): New function.
2465 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
2466 (TARGET_FUNCTION_ARG_PADDING): ...this.
2467 * doc/tm.texi: Regenerate.
2468 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
2469 instead of direction.
2470 (compute_argument_addresses): Likewise.
2471 (load_register_parameters): Likewise.
2472 (emit_library_call_value_1): Likewise.
2473 (store_one_arg): Use targetm.calls.function_arg_padding instead
2474 of FUNCTION_ARG_PADDING.
2475 (must_pass_in_stack_var_size_or_pad): Likewise.
2476 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
2477 (emit_group_store): Likewise.
2478 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
2479 instead of FUNCTION_ARG_PADDING.
2480 (emit_push_insn): Likewise, and propagate enum change throughout
2482 * function.h (direction): Delete.
2483 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
2485 * function.c (assign_parm_find_stack_rtl): Likewise.
2486 (assign_parm_setup_block_p): Likewise.
2487 (assign_parm_setup_block): Likewise.
2488 (gimplify_parameters): Likewise.
2489 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
2490 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
2492 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
2493 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2494 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
2495 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
2496 (aarch64_function_arg_padding): ...this new function.
2497 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
2498 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2499 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
2500 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2501 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
2502 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2503 (arm_pad_arg_upward): Replace with...
2504 (arm_function_arg_padding): ...this new function.
2505 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
2507 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
2508 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
2509 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2510 (ia64_hpux_function_arg_padding): Replace with...
2511 (ia64_function_arg_padding): ...this new function. Use pad_direction
2512 instead of direction. Check for TARGET_HPUX.
2513 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
2514 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2515 (iq2000_function_arg_padding): New function.
2516 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
2517 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
2518 (mips_function_arg_padding): ...this new function.
2519 (mips_pad_reg_upward): Update accordingly.
2520 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2521 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
2522 targetm.calls.function_arg_padding.
2523 (FUNCTION_ARG_PADDING): Delete.
2524 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2525 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
2526 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2527 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
2528 (nios2_block_reg_padding): Return pad_direction instead of direction.
2529 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
2530 instead of direction.
2531 (nios2_function_arg_padding): Likewise. Make static.
2532 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2533 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
2534 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
2535 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
2536 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2537 (pa_function_arg_padding): Make static. Return pad_direction instead
2539 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
2540 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2541 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
2542 instead of direction. Use targetm.calls.function_arg_padding.
2543 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
2544 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
2545 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
2546 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
2547 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
2549 (function_arg_padding): Rename to...
2550 (rs6000_function_arg_padding): ...this. Make static. Return
2551 pad_direction instead of direction.
2552 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
2553 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
2554 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2555 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
2556 instead of direction. Use targetm.calls.function_arg_padding.
2557 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
2558 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
2559 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
2560 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
2561 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2562 (function_arg_padding): Rename to...
2563 (rs6000_function_arg_padding): ...this. Make static. Return
2564 pad_direction instead of direction.
2565 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
2566 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
2567 * config/s390/s390.c (s390_function_arg_padding): New function.
2568 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2569 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
2570 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
2571 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2572 (function_arg_padding): Rename to...
2573 (sparc_function_arg_padding): ...this. Make static. Return
2574 pad_direction instead of direction.
2575 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
2576 * config/spu/spu.c (spu_function_arg_padding): New function.
2577 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2578 * system.h (FUNCTION_ARG_PADDING): Poison.
2580 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2581 Alan Hayward <alan.hayward@arm.com>
2582 David Sherwood <david.sherwood@arm.com>
2584 * target.def (modes_tieable_p): New hook.
2585 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
2586 (TARGET_MODES_TIEABLE_P): ...this.
2587 * doc/tm.texi.in: Regenerate.
2588 * hooks.h (hook_bool_mode_mode_true): Declare.
2589 * hooks.c (hook_bool_mode_mode_true): New function.
2590 * combine.c (subst): Use targetm.modes_tieable_p instead of
2592 * dse.c (find_shift_sequence): Likewise.
2593 * expmed.c (extract_low_bits): Likewise.
2594 * lower-subreg.c: Include target.h.
2595 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
2597 * rtlanal.c (rtx_cost): Likewise.
2598 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
2599 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
2600 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
2601 (TARGET_MODES_TIEABLE_P): Redefine.
2602 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
2603 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
2604 (TARGET_MODES_TIEABLE_P): Redefine.
2605 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
2606 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
2607 (arc_modes_tieable_p): New function.
2608 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
2609 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
2610 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
2611 (arm_modes_tieable_p): Make static.
2612 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
2613 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
2614 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
2615 (TARGET_MODES_TIEABLE_P): Redefine.
2616 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
2617 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
2618 (TARGET_MODES_TIEABLE_P): Redefine.
2619 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
2620 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
2621 (cr16_modes_tieable_p): New function.
2622 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
2623 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
2624 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
2625 (TRULY_NOOP_TRUNCATION): Update comment.
2626 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
2627 (TRULY_NOOP_TRUNCATION): Update comment.
2628 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
2629 (frv_modes_tieable_p): New function.
2630 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
2631 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
2632 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
2633 (TARGET_MODES_TIEABLE_P): Redefine.
2634 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
2635 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
2636 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
2637 (TARGET_MODES_TIEABLE_P): Redefine.
2638 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
2639 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
2640 (ia64_modes_tieable_p): New function.
2641 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
2642 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
2643 (iq2000_modes_tieable_p): New function.
2644 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
2645 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
2646 (lm32_modes_tieable_p): New function.
2647 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
2648 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
2649 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
2650 (TARGET_MODES_TIEABLE_P): Redefine.
2651 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
2652 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
2653 (m32r_modes_tieable_p): New function.
2654 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
2655 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
2656 (m68k_modes_tieable_p): New function.
2657 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
2658 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
2659 (mcore_modes_tieable_p): New function.
2660 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
2661 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
2663 (TARGET_MODES_TIEABLE_P): Redefine.
2664 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
2665 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
2666 * config/mips/mips.c (mips_modes_tieable_p): Make static.
2667 (TARGET_MODES_TIEABLE_P): Redefine.
2668 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
2669 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
2670 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
2671 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
2672 (mn10300_modes_tieable_p): ...this and make static.
2673 (TARGET_MODES_TIEABLE_P): Redefine.
2674 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
2675 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
2676 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
2677 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
2678 (msp430_modes_tieable_p): Make static.
2679 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
2680 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
2681 (TARGET_MODES_TIEABLE_P): Redefine.
2682 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
2683 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
2684 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
2685 (TARGET_MODES_TIEABLE_P): Redefine.
2686 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
2687 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
2688 * config/pa/pa.c (pa_modes_tieable_p): Make static.
2689 (TARGET_MODES_TIEABLE_P): Redefine.
2690 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
2691 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
2692 (pdp11_modes_tieable_p): New function.
2693 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
2694 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
2695 (rs6000_modes_tieable_p): New function.
2696 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
2697 * config/powerpcspe/powerpcspe.md: Update comment.
2698 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
2699 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
2700 (TARGET_MODES_TIEABLE_P): Redefine.
2701 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
2702 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
2703 (rl78_modes_tieable_p): New function.
2704 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
2705 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
2706 (rs6000_modes_tieable_p): New function.
2707 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
2708 * config/rs6000/rs6000.md: Update comment.
2709 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
2710 * config/rx/rx.c (rx_modes_tieable_p): New function.
2711 (TARGET_MODES_TIEABLE_P): Redefine.
2712 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
2713 * config/s390/s390.c (s390_modes_tieable_p): New function.
2714 (TARGET_MODES_TIEABLE_P): Redefine.
2715 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
2716 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
2717 (sh_modes_tieable_p): New function.
2718 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
2719 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
2720 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
2721 (sparc_modes_tieable_p): Make static.
2722 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
2723 * config/spu/spu.c (spu_modes_tieable_p): New function.
2724 (TARGET_MODES_TIEABLE_P): Redefine.
2725 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
2726 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
2727 (TARGET_MODES_TIEABLE_P): Redefine.
2728 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
2729 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
2730 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
2731 * config/v850/v850.c (v850_modes_tieable_p): New function.
2732 (TARGET_MODES_TIEABLE_P): Redefine.
2733 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
2734 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
2735 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
2736 (visium_modes_tieable_p): New function.
2737 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
2738 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
2739 (xtensa_modes_tieable_p): New function.
2740 * system.h (MODES_TIEABLE_P): Poison.
2742 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2743 Alan Hayward <alan.hayward@arm.com>
2744 David Sherwood <david.sherwood@arm.com>
2746 * target.def (hard_regno_mode_ok): New hook.
2747 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
2748 (TARGET_HARD_REGNO_MODE_OK): ...this.
2749 * doc/tm.texi.in: Regenerate.
2750 * hooks.h (hook_bool_uint_mode_true): Declare.
2751 * hooks.c (hook_bool_uint_mode_true): New function.
2752 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
2754 * genpreds.c (write_insn_preds_c): Add an include of target.h.
2755 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
2756 instead of HARD_REGNO_MODE_OK.
2757 * caller-save.c: Include target.h.
2758 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
2760 * combine.c (can_combine_p): Likewise.
2761 (combinable_i3pat): Likewise.
2762 (can_change_dest_mode): Likewise.
2763 * expr.c (init_expr_target): Likewise.
2764 (convert_move): Likewise.
2765 (convert_modes): Likewise.
2766 * ira.c (setup_prohibited_class_mode_regs): Likewise.
2767 (setup_prohibited_mode_move_regs): Likewise.
2768 * ira.h (target_ira): Likewise.
2769 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2770 * lra-constraints.c (process_alt_operands): Likewise.
2771 (split_reg): Likewise.
2772 * recog.c (peep2_find_free_register): Likewise.
2773 * ree.c (combine_reaching_defs): Likewise.
2774 * regcprop.c (maybe_mode_change): Likewise.
2775 * reginfo.c (init_reg_sets_1): Likewise.
2776 (choose_hard_reg_mode): Likewise.
2777 (simplifiable_subregs): Likewise.
2778 * regrename.c (check_new_reg_p): Likewise.
2779 * reload.c (find_valid_class): Likewise.
2780 (find_valid_class_1): Likewise.
2781 (reload_inner_reg_of_subreg): Likewise.
2782 (push_reload): Likewise.
2783 (combine_reloads): Likewise.
2784 (find_dummy_reload): Likewise.
2785 (find_reloads): Likewise.
2786 * reload1.c (find_reg): Likewise.
2787 (set_reload_reg): Likewise.
2788 (allocate_reload_reg): Likewise.
2789 (choose_reload_regs): Likewise.
2790 (reload_adjust_reg_for_temp): Likewise.
2791 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
2792 (simplify_subreg_regno): Likewise.
2793 * sel-sched.c (init_regs_for_mode): Likewise.
2794 * varasm.c (make_decl_rtl): Likewise.
2795 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
2796 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
2798 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
2799 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
2800 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2801 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
2802 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
2803 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2804 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
2805 (arc_mode_class): Delete.
2806 (HARD_REGNO_MODE_OK): Delete.
2807 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2808 (arc_hard_regno_mode_ok): Rename old array to...
2809 (arc_hard_regno_mode_ok_modes): ...this.
2810 (arc_conditional_register_usage): Update accordingly.
2811 (arc_mode_class): Make static.
2812 (arc_hard_regno_mode_ok): New function.
2813 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
2814 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
2815 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2816 (arm_hard_regno_mode_ok): Make static.
2817 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
2819 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
2820 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
2821 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
2823 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2824 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
2825 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
2826 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
2827 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
2828 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2829 * config/bfin/predicates.md (valid_reg_operand): Use
2830 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
2831 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
2832 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
2833 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2834 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
2835 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
2836 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2837 (cr16_hard_regno_mode_ok): Make static and return a bool.
2838 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
2839 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2840 (cris_hard_regno_mode_ok): New function.
2841 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
2842 (epiphany_mode_class): Delete.
2843 (HARD_REGNO_MODE_OK): Delete.
2844 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
2845 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2846 (hard_regno_mode_ok): Rename to...
2847 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
2848 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
2849 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
2851 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
2852 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
2853 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2854 (frv_hard_regno_mode_ok): Make static and return a bool.
2855 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
2857 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
2858 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
2859 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
2860 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
2862 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2863 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
2864 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
2865 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
2867 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2868 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
2869 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2870 (ia64_hard_regno_mode_ok): New function.
2871 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
2872 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2873 (iq2000_hard_regno_mode_ok): New function.
2874 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
2875 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2876 (lm32_hard_regno_mode_ok): New function.
2877 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
2878 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
2879 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
2880 instead of HARD_REGNO_MODE_OK.
2881 (m32c_hard_regno_ok): Rename to...
2882 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
2883 (m32c_cannot_change_mode_class): Update accordingly.
2884 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2885 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
2886 (m32r_mode_class): Delete.
2887 (HARD_REGNO_MODE_OK): Delete.
2888 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2889 (m32r_hard_regno_mode_ok): Rename to...
2890 (m32r_hard_regno_modes): ...this.
2891 (m32r_mode_class): Make static.
2892 (m32r_hard_regno_mode_ok): New function.
2893 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
2894 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
2895 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2896 (m68k_hard_regno_mode_ok): Make static.
2897 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
2898 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2899 (mcore_hard_regno_mode_ok): New function.
2900 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
2901 (HARD_REGNO_MODE_OK): Delete.
2902 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
2904 (microblaze_hard_regno_mode_ok_p): ...this and make static.
2905 (microblaze_hard_regno_mode_ok): New function.
2906 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2907 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
2908 (mips_hard_regno_mode_ok): Delete.
2909 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
2910 (mips_hard_regno_mode_ok_p): ...this and make static.
2911 (mips_hard_regno_mode_ok_p): Rename to...
2912 (mips_hard_regno_mode_ok_uncached): ...this.
2913 (mips_hard_regno_mode_ok): New function.
2914 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
2915 of HARD_REGNO_MODE_OK.
2916 (mips_option_override): Update after above name changes.
2917 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2918 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
2919 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
2920 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
2921 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
2922 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2923 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
2924 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
2925 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
2926 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2927 (msp430_hard_regno_mode_ok): Make static and return a bool.
2928 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
2929 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
2930 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
2932 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2933 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
2934 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
2935 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
2936 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
2937 (PA_HARD_REGNO_MODE_OK): ...this
2938 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
2939 (PA_HARD_REGNO_MODE_OK): ...this.
2940 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2941 (pa_hard_regno_mode_ok): New function.
2942 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
2943 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2944 (pdp11_hard_regno_mode_ok): New function.
2945 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
2946 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
2948 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
2950 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2951 (rs6000_hard_regno_mode_ok): Rename to...
2952 (rs6000_hard_regno_mode_ok_uncached): ...this.
2953 (rs6000_init_hard_regno_mode_ok): Update accordingly.
2954 (rs6000_hard_regno_mode_ok): New function.
2955 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
2956 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
2957 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
2958 (riscv_hard_regno_mode_ok): ...this and make static.
2959 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2960 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
2961 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
2962 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2963 (rl78_hard_regno_mode_ok): Make static and return bool.
2964 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
2965 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
2967 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
2968 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2969 (rs6000_hard_regno_mode_ok): Rename to...
2970 (rs6000_hard_regno_mode_ok_uncached): ...this.
2971 (rs6000_init_hard_regno_mode_ok): Update accordingly.
2972 (rs6000_hard_regno_mode_ok): New function.
2973 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
2974 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
2975 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2976 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
2977 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
2978 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
2979 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2980 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
2981 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
2982 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2983 (sh_hard_regno_mode_ok): Make static.
2984 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
2985 instead of HARD_REGNO_MODE_OK.
2986 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
2987 (sparc_mode_class): Delete.
2988 (HARD_REGNO_MODE_OK): Delete.
2989 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2990 (hard_regno_mode_classes): Make static.
2991 (sparc_mode_class): Likewise.
2992 (sparc_hard_regno_mode_ok): New function.
2993 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
2994 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
2995 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
2997 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2998 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
2999 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
3000 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
3001 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
3002 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3003 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
3004 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
3005 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3006 (visium_hard_regno_mode_ok): New function.
3007 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
3008 instead of HARD_REGNO_MODE_OK.
3009 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
3010 (HARD_REGNO_MODE_OK): Delete.
3011 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
3012 (xtensa_hard_regno_mode_ok_p): ...this and make static.
3013 (xtensa_option_override): Update accordingly.
3014 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3015 (xtensa_hard_regno_mode_ok): New function.
3016 * system.h (HARD_REGNO_MODE_OK): Poison.
3018 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3019 Alan Hayward <alan.hayward@arm.com>
3020 David Sherwood <david.sherwood@arm.com>
3022 * target.def (hard_regno_call_part_clobbered): New hook.
3023 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
3024 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
3025 * doc/tm.texi: Regenerate.
3026 * hooks.h (hook_bool_uint_mode_false): Declare.
3027 * hooks.c (hook_bool_uint_mode_false): New function.
3028 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3029 * cselib.c (cselib_process_insn): Use
3030 targetm.hard_regno_call_part_clobbered instead of
3031 HARD_REGNO_CALL_PART_CLOBBERED.
3032 * ira-conflicts.c (ira_build_conflicts): Likewise.
3033 * ira-costs.c (ira_tune_allocno_costs): Likewise.
3034 * lra-constraints.c (need_for_call_save_p): Likewise.
3035 * lra-lives.c: Include target.h.
3036 (check_pseudos_live_through_calls): Use
3037 targetm.hard_regno_call_part_clobbered instead of
3038 HARD_REGNO_CALL_PART_CLOBBERED.
3039 * regcprop.c: Include target.h.
3040 (copyprop_hardreg_forward_1): Use
3041 targetm.hard_regno_call_part_clobbered instead of
3042 HARD_REGNO_CALL_PART_CLOBBERED.
3043 * reginfo.c (choose_hard_reg_mode): Likewise.
3044 * regrename.c (check_new_reg_p): Likewise.
3045 * reload.c (find_equiv_reg): Likewise.
3046 * reload1.c (emit_reload_insns): Likewise.
3047 * sched-deps.c (deps_analyze_insn): Likewise.
3048 * sel-sched.c (init_regs_for_mode): Likewise.
3049 (mark_unavailable_hard_regs): Likewise.
3050 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
3051 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3052 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
3054 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3055 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3056 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
3058 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
3060 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3061 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3062 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
3064 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3065 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3066 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
3068 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3069 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
3071 * config/powerpcspe/powerpcspe.c
3072 (rs6000_hard_regno_call_part_clobbered): New function.
3073 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3074 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3075 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
3077 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3078 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3079 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
3081 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3082 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3083 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
3085 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3086 Alan Hayward <alan.hayward@arm.com>
3087 David Sherwood <david.sherwood@arm.com>
3089 * rtl.h (subreg_memory_offset): Declare.
3090 * emit-rtl.c (subreg_memory_offset): New function.
3091 * expmed.c (store_bit_field_1): Use it.
3092 * expr.c (undefined_operand_subword_p): Likewise.
3093 * simplify-rtx.c (simplify_subreg): Likewise.
3095 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
3097 PR rtl-optimization/57448
3100 * optabs.c (expand_atomic_load): Place compiler memory barriers if
3101 using atomic_load pattern.
3102 (expand_atomic_store): Likewise.
3104 2017-09-04 Jakub Jelinek <jakub@redhat.com>
3107 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
3108 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
3109 handling. Use replace_call_with_value with NULL instead of
3110 gsi_replace, unlink_stmt_vdef and release_defs.
3112 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
3115 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
3117 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3120 * rtl.h (emit_library_call_value_1): Declare.
3121 (emit_library_call): Replace declaration with a series of overloads.
3122 Remove the parameter count argument.
3123 (emit_library_call_value): Likewise.
3124 * calls.c (emit_library_call_value_1): Make global. Replace varargs
3125 with an "rtx_mode_t *".
3126 (emit_library_call_value): Delete.
3127 (emit_library_call): Likewise.
3128 * asan.c (asan_emit_stack_protection): Update calls accordingly.
3129 (asan_emit_allocas_unpoison): Likewise.
3130 * builtins.c (expand_builtin_powi): Likewise.
3131 (expand_asan_emit_allocas_unpoison): Likewise.
3132 * cfgexpand.c (expand_main_function): Likewise.
3133 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
3134 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
3135 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
3136 * config/arm/arm.c (arm_trampoline_init): Likewise.
3137 (arm_call_tls_get_addr): Likewise.
3138 (arm_expand_divmod_libfunc): Likewise.
3139 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
3140 (smulsi3_highpart): Likewise.
3141 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
3142 (c6x_expand_compare): Likewise.
3143 (c6x_expand_movmem): Likewise.
3144 * config/frv/frv.c (frv_trampoline_init): Likewise.
3145 * config/i386/i386.c (ix86_trampoline_init): Likewise.
3146 (ix86_expand_divmod_libfunc): Likewise.
3147 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
3148 (ia64_expand_compare): Likewise.
3149 (ia64_profile_hook): Likewise.
3150 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
3151 (nonlocal_goto): Likewise.
3152 (restore_stack_nonlocal): Likewise.
3153 * config/m32r/m32r.c (block_move_call): Likewise.
3154 (m32r_trampoline_init): Likewise.
3155 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
3156 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
3157 (m68k_call_m68k_read_tp): Likewise.
3158 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
3159 (microblaze_expand_divide): Likewise.
3160 * config/mips/mips.h (mips_args): Likewise.
3161 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
3162 (MIPS_ICACHE_SYNC): Likewise.
3163 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
3164 (nios2_trampoline_init): Likewise.
3165 * config/pa/pa.c (hppa_tls_call): Likewise.
3166 (pa_trampoline_init): Likewise.
3167 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
3168 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
3169 (expand_strn_compare): Likewise.
3170 (rs6000_generate_compare): Likewise.
3171 (rs6000_expand_float128_convert): Likewise.
3172 (output_profile_hook): Likewise.
3173 (rs6000_trampoline_init): Likewise.
3174 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
3175 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
3176 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
3177 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
3178 (rs6000_generate_compare): Likewise.
3179 (rs6000_expand_float128_convert): Likewise.
3180 (output_profile_hook): Likewise.
3181 (rs6000_trampoline_init): Likewise.
3182 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
3183 * config/sh/sh.c (sh_trampoline_init): Likewise.
3184 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
3185 (sparc_emit_float_lib_cmp): Likewise.
3186 (sparc32_initialize_trampoline): Likewise.
3187 (sparc64_initialize_trampoline): Likewise.
3188 (sparc_profile_hook): Likewise.
3189 * config/spu/spu.c (ea_load_store): Likewise.
3190 * config/spu/spu.md (floatunssidf2): Likewise.
3191 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
3192 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
3193 * config/visium/visium.c (expand_block_move_4): Likewise.
3194 (expand_block_move_2): Likewise.
3195 (expand_block_move_1): Likewise.
3196 (expand_block_set_4): Likewise.
3197 (expand_block_set_2): Likewise.
3198 (expand_block_set_1): Likewise.
3199 (visium_trampoline_init): Likewise.
3200 (visium_profile_hook): Likewise.
3201 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
3202 (xtensa_setup_frame_addresses): Likewise.
3203 (xtensa_trampoline_init): Likewise.
3204 * except.c (sjlj_emit_function_enter): Likewise.
3205 (sjlj_emit_function_exit): Likewise.
3206 * explow.c (allocate_dynamic_stack_space): Likewise.
3207 (probe_stack_range): Likewise.
3208 * expr.c (convert_mode_scalar): Likewise.
3209 * optabs.c (expand_binop): Likewise.
3210 (expand_twoval_binop_libfunc): Likewise.
3211 (expand_unop): Likewise.
3212 (prepare_cmp_insn): Likewise.
3213 (prepare_float_lib_cmp): Likewise.
3214 (expand_float): Likewise.
3215 (expand_fix): Likewise.
3216 (expand_fixed_convert): Likewise.
3217 (maybe_emit_sync_lock_test_and_set): Likewise.
3218 (expand_atomic_compare_and_swap): Likewise.
3219 (expand_mem_thread_fence): Likewise.
3220 (expand_atomic_fetch_op): Likewise.
3222 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
3224 * doc/generic.texi (OpenACC): Adjust URL.
3225 * doc/invoke.texi (C Dialect Options): Ditto.
3227 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
3229 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
3230 predicate for operand 1. Add (m,<S>) constraint.
3231 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
3232 Prevent memory operand 1 with register operand 2.
3234 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
3236 PR rtl-optimization/82024
3237 * combine.c (try_combine): If the combination result is a PARALLEL,
3238 and we only need to retain the SET in there that would be placed
3239 at I2, check that we can place that at I3 instead, before doing so.
3241 2017-09-01 Jakub Jelinek <jakub@redhat.com>
3244 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
3246 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
3247 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
3250 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
3251 Jeff Law <law@redhat.com>
3253 * varasm.c (bss_initializer_p): Do not put constants into .bss
3254 (categorize_decl_for_section): Handle bss_initializer_p returning
3255 false when DECL_INITIAL is NULL.
3257 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3260 * config/s390/s390.c (s390_can_inline_p): New function.
3262 2017-09-01 Jeff Law <law@redhat.com>
3264 PR tree-optimization/82052
3265 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
3266 Always initialize the returned slot after a hash table miss
3267 when INSERT is true.
3269 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
3271 * config/s390/s390.md (mem_signal_fence): Remove.
3272 * doc/md.texi (mem_signal_fence): Remove.
3273 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
3275 * target-insns.def (mem_signal_fence): Remove.
3277 2017-09-01 Jakub Jelinek <jakub@redhat.com>
3280 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
3283 * asan.c (create_odr_indicator): Strip name encoding from assembler
3284 name before appending it after __odr_asan_.
3286 2017-09-01 Martin Liska <mliska@suse.cz>
3288 PR tree-optimization/82059
3289 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
3290 frequency only when an edge is redirected.
3292 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3294 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
3295 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
3296 (arc_conditional_register_usage): Remove ARC600 lp_count
3298 (arc_file_start): Emit Tag_ARC_CPU_variation.
3299 (arc_can_use_doloop_p): New conditions to use ZOLs.
3300 (hwloop_fail): New function.
3301 (hwloop_optimize): Likewise.
3302 (hwloop_pattern_reg): Likewise.
3303 (arc_doloop_hooks): New struct, to be used with reorg_loops.
3304 (arc_reorg_loops): New function, calls reorg_loops.
3305 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
3306 (arc600_corereg_hazard): Remove ZOL checking, case handled by
3308 (arc_loop_hazard): Remove function, functionality moved into
3310 (arc_hazard): Remove arc_loop_hazard call.
3311 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
3312 into hwloop_optimize.
3313 (arc_label_align): Remove ZOL handling.
3314 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
3315 * config/arc/arc.md (doloop_begin): Remove pattern.
3316 (doloop_begin_i): Likewise.
3317 (doloop_end_i): Likewise.
3318 (doloop_fallback): Likewise.
3319 (doloop_fallback_m): Likewise.
3320 (doloop_end): Reimplement expand.
3321 (arc_lp): New pattern for LP instruction.
3322 (loop_end): New pattern.
3323 (loop_fail): Likewise.
3324 (decrement_and_branch_until_zero): Likewise.
3325 * config/arc/arc.opt (mlpc-width): New option.
3326 * doc/invoke.texi (mlpc-width): Document option.
3328 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3330 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
3331 (arc_ccfsm_advance): Fix checking for delay slots.
3332 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
3334 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3336 * config/arc/arc.md (movqi_insn): Add stores to save constant long
3338 (movhi_insn): Update store instruction constraint which are saving
3339 6-bit short immediates.
3340 (movsi_insn): Consider also short scaled load operations.
3341 (zero_extendhisi2_i): Use Usd constraint instead of T.
3342 (extendhisi2_i): Add q constraint.
3343 (arc_clzsi2): Add type and length attributes.
3344 (arc_ctzsi2): Likewise.
3345 * config/arc/constraints.md (Usc): Update constraint, the
3346 assembler can parse two relocations for a single instruction.
3348 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3350 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
3351 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
3353 2017-08-31 Olivier Hainque <hainque@adacore.com>
3355 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
3356 match as powerpc-wrs-vxworks*.
3358 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
3360 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
3361 register constraint for by-element operand.
3362 (aarch64_mls_elt_merge<mode>): Likewise.
3364 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
3366 * config/arc/arc.c (arc_can_follow_jump): Check for short
3369 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
3371 * config.gcc: Use g.opt for arc.
3372 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
3373 functionality moved to ...
3374 (legitimate_scaled_address_p): New function, ...here.
3375 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
3376 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
3377 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
3379 (arc_override_options): Handle G option.
3380 (arc_output_pic_addr_const): Correct function definition.
3381 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
3382 (arc_decl_anon_ns_mem_p): Delete.
3383 (arc_in_small_data_p): Overhaul this function to take into
3384 consideration the value given via G option.
3385 (arc_rewrite_small_data_1): Renamed and corrected old
3386 arc_rewrite_small_data function.
3387 (arc_rewrite_small_data): New function.
3388 (small_data_pattern): Don't use pic_offset_table_rtx.
3389 * config/arc/arc.h (CC1_SPEC): Recognize G option.
3390 * config/arc/simdext.md (movmisalignv2hi): Use
3391 prepare_move_operands function.
3393 (movmisalign*): Likewise.
3394 * doc/invoke.texi (ARC options): Document -G option.
3396 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
3398 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
3400 * config/arc/arc.c (arc_print_operand): Output scalled address for
3401 sdata whenever is possible.
3402 (arc_in_small_data_p): Allow sdata for 64bit datum when double
3403 load/stores are available.
3404 (compact_sda_memory_operand): Check for the alignment required by
3405 code density instructions.
3406 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
3408 * config/arc/constraints.md (Usd): Update constraint.
3409 (Us0): New constraint.
3410 (Usc): Update constraint.
3412 2017-08-31 Richard Biener <rguenther@suse.de>
3415 * dwarf2out.c (dwarf2out_early_global_decl): Process each
3418 2017-08-31 Tamar Christina <tamar.christina@arm.com>
3420 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
3421 Resize type_signature.
3423 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3424 Alan Hayward <alan.hayward@arm.com>
3425 David Sherwood <david.sherwood@arm.com>
3427 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
3428 subregs whose inner modes can be stored in GPRs.
3429 (aarch64_classify_index): Likewise.
3431 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3432 Alan Hayward <alan.hayward@arm.com>
3433 David Sherwood <david.sherwood@arm.com>
3435 * config/aarch64/iterators.md (V_cmp_result): Rename to...
3436 (V_INT_EQUIV): ...this.
3437 (v_cmp_result): Rename to...
3438 (v_int_equiv): ...this.
3439 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
3440 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
3441 (copysign<mode>3): Likewise.
3442 (aarch64_simd_bsl<mode>_internal): Likewise.
3443 (aarch64_simd_bsl<mode>): Likewise.
3444 (vec_cmp<mode><mode>): Likewise.
3445 (vcond<mode><mode>): Likewise.
3446 (vcond<v_cmp_mixed><mode>): Likewise.
3447 (vcondu<mode><v_cmp_mixed>): Likewise.
3448 (aarch64_cm<optab><mode>): Likewise.
3449 (aarch64_cmtst<mode>): Likewise.
3450 (aarch64_fac<optab><mode>): Likewise.
3451 (vec_perm_const<mode>): Likewise.
3452 (vcond_mask_<mode><v_cmp_result>): Rename to...
3453 (vcond_mask_<mode><v_int_equiv>): ...this.
3454 (vec_cmp<mode><v_cmp_result>): Rename to...
3455 (vec_cmp<mode><v_int_equiv>): ...this.
3457 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3458 Alan Hayward <alan.hayward@arm.com>
3459 David Sherwood <david.sherwood@arm.com>
3461 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
3463 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
3464 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
3465 (UNSPEC_LD4_DREG): New unspecs.
3466 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
3467 (aarch64_ld2<mode>_dreg_be): Replace with...
3468 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
3470 (aarch64_ld3<mode>_dreg_le)
3471 (aarch64_ld3<mode>_dreg_be): Replace with...
3472 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
3474 (aarch64_ld4<mode>_dreg_le)
3475 (aarch64_ld4<mode>_dreg_be): Replace with...
3476 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
3479 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3481 PR tree-optimization/81987
3482 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
3483 insert an initializer in a location not dominated by the stride
3486 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
3488 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
3489 on the entire header of the finally block in the fallthru case.
3491 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
3493 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
3495 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
3497 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
3498 rs6000_emit_move_from_cr and call renamed function.
3499 (rs6000_emit_prologue): Call renamed functions.
3500 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
3501 movesi_from_cr, remove volatile CRs.
3503 2017-08-30 Jon Beniston <jon@beniston.com>
3504 Richard Biener <rguenther@suse.de>
3506 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
3507 of VECTOR_MODE_P check.
3508 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
3509 element vector types.
3511 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3513 * df.h (df_read_modify_subreg_p): Remove in favor of...
3514 * rtl.h (read_modify_subreg_p): ...this new function. Take a
3515 const_rtx instead of an rtx.
3516 * cprop.c (local_cprop_find_used_regs): Update accordingly.
3517 * df-problems.c (df_word_lr_mark_ref): Likewise.
3518 * ira-lives.c (mark_pseudo_reg_live): Likewise.
3519 (mark_pseudo_reg_dead): Likewise.
3520 (mark_ref_dead): Likewise.
3521 * reginfo.c (init_subregs_of_mode): Likewise.
3522 * sched-deps.c (sched_analyze_1): Likewise.
3523 * df-scan.c (df_def_record_1): Likewise.
3524 (df_uses_record): Likewise.
3525 (df_read_modify_subreg_p): Remove in favor of...
3526 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
3527 const_rtx instead of an rtx.
3529 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3530 Alan Hayward <alan.hayward@arm.com>
3531 David Sherwood <david.sherwood@arm.com>
3533 * rtl.h (partial_subreg_p): New function.
3534 * caller-save.c (save_call_clobbered_regs): Use it.
3535 * calls.c (expand_call): Likewise.
3536 * combine.c (combinable_i3pat): Likewise.
3537 (simplify_set): Likewise.
3538 (make_extraction): Likewise.
3539 (make_compound_operation_int): Likewise.
3540 (gen_lowpart_or_truncate): Likewise.
3541 (force_to_mode): Likewise.
3542 (make_field_assignment): Likewise.
3543 (reg_truncated_to_mode): Likewise.
3544 (record_truncated_value): Likewise.
3545 (move_deaths): Likewise.
3546 * cse.c (record_jump_cond): Likewise.
3547 (cse_insn): Likewise.
3548 * cselib.c (cselib_lookup_1): Likewise.
3549 * expmed.c (extract_bit_field_using_extv): Likewise.
3550 * function.c (assign_parm_setup_reg): Likewise.
3551 * ifcvt.c (noce_convert_multiple_sets): Likewise.
3552 * ira-build.c (create_insn_allocnos): Likewise.
3553 * lra-coalesce.c (merge_pseudos): Likewise.
3554 * lra-constraints.c (match_reload): Likewise.
3555 (simplify_operand_subreg): Likewise.
3556 (curr_insn_transform): Likewise.
3557 * lra-lives.c (process_bb_lives): Likewise.
3558 * lra.c (new_insn_reg): Likewise.
3559 (lra_substitute_pseudo): Likewise.
3560 * regcprop.c (mode_change_ok): Likewise.
3561 (maybe_mode_change): Likewise.
3562 (copyprop_hardreg_forward_1): Likewise.
3563 * reload.c (push_reload): Likewise.
3564 (find_reloads): Likewise.
3565 (find_reloads_subreg_address): Likewise.
3566 * reload1.c (alter_reg): Likewise.
3567 (eliminate_regs_1): Likewise.
3568 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3570 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
3572 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
3573 back to if statements, including unpack.
3575 2017-08-30 Martin Liska <mliska@suse.cz>
3578 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
3580 (func_checker::compare_asm_inputs_outputs): ... this function.
3581 (func_checker::compare_gimple_asm): Use the function to compare
3582 also ASM constrains.
3583 * ipa-icf-gimple.h: Rename the function.
3585 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3586 Alan Hayward <alan.hayward@arm.com>
3587 David Sherwood <david.sherwood@arm.com>
3589 * coretypes.h (complex_mode): New type.
3590 * gdbhooks.py (build_pretty_printer): Handle it.
3591 * machmode.h (complex_mode): New class.
3592 (complex_mode::includes_p): New function.
3593 (is_complex_int_mode): Likewise.
3594 (is_complex_float_mode): Likewise.
3595 * genmodes.c (get_mode_class): Handle complex mode classes.
3596 * function.c (expand_function_end): Use is_complex_int_mode.
3598 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3599 Alan Hayward <alan.hayward@arm.com>
3600 David Sherwood <david.sherwood@arm.com>
3602 * coretypes.h (scalar_mode_pod): New typedef.
3603 * gdbhooks.py (build_pretty_printer): Handle it.
3604 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
3605 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
3606 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
3607 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
3610 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3611 Alan Hayward <alan.hayward@arm.com>
3612 David Sherwood <david.sherwood@arm.com>
3614 * machmode.h (mode_for_vector): Take a scalar_mode instead
3616 * stor-layout.c (mode_for_vector): Likewise.
3617 * explow.c (promote_mode): Use as_a <scalar_mode>.
3618 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
3620 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3621 Alan Hayward <alan.hayward@arm.com>
3622 David Sherwood <david.sherwood@arm.com>
3624 * target.def (preferred_simd_mode): Take a scalar_mode
3625 instead of a machine_mode.
3626 * targhooks.h (default_preferred_simd_mode): Likewise.
3627 * targhooks.c (default_preferred_simd_mode): Likewise.
3628 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
3629 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
3630 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
3631 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
3632 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
3633 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
3634 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
3635 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
3637 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
3638 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
3639 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
3640 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
3641 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
3642 * doc/tm.texi: Regenerate.
3643 * optabs-query.c (can_vec_mask_load_store_p): Return false for
3646 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3647 Alan Hayward <alan.hayward@arm.com>
3648 David Sherwood <david.sherwood@arm.com>
3650 * target.def (scalar_mode_supported_p): Take a scalar_mode
3651 instead of a machine_mode.
3652 * targhooks.h (default_scalar_mode_supported_p): Likewise.
3653 * targhooks.c (default_scalar_mode_supported_p): Likewise.
3654 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
3655 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
3656 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
3657 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
3658 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
3659 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
3660 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
3661 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
3662 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
3663 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
3664 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
3665 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
3667 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
3668 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
3669 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
3670 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
3671 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
3673 * doc/tm.texi: Regenerate.
3675 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3676 Alan Hayward <alan.hayward@arm.com>
3677 David Sherwood <david.sherwood@arm.com>
3679 * coretypes.h (opt_scalar_mode): New typedef.
3680 * gdbhooks.py (build_pretty_printers): Handle it.
3681 * machmode.h (mode_iterator::get_2xwider): Add overload for
3683 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
3685 * expr.c (convert_mode_scalar): Likewise.
3686 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3687 * optabs.c (expand_float): Likewise.
3688 (expand_fix): Likewise.
3689 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3691 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3692 Alan Hayward <alan.hayward@arm.com>
3693 David Sherwood <david.sherwood@arm.com>
3695 * optabs.c (expand_float): Explicitly check for scalars before
3696 using a branching expansion.
3697 (expand_fix): Likewise.
3699 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3700 Alan Hayward <alan.hayward@arm.com>
3701 David Sherwood <david.sherwood@arm.com>
3703 * expr.c (convert_mode): Split scalar handling out into...
3704 (convert_mode_scalar): ...this new function. Treat the modes
3707 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3708 Alan Hayward <alan.hayward@arm.com>
3709 David Sherwood <david.sherwood@arm.com>
3711 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
3713 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
3715 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3716 Alan Hayward <alan.hayward@arm.com>
3717 David Sherwood <david.sherwood@arm.com>
3719 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
3720 rather than a machine_mode.
3721 (fixed_from_string): Likewise.
3722 (fixed_convert): Likewise.
3723 (fixed_convert_from_int): Likewise.
3724 (fixed_convert_from_real): Likewise.
3725 (real_convert_from_fixed): Likewise.
3726 * fixed-value.c (fixed_from_double_int): Likewise.
3727 (fixed_from_string): Likewise.
3728 (fixed_convert): Likewise.
3729 (fixed_convert_from_int): Likewise.
3730 (fixed_convert_from_real): Likewise.
3731 (real_convert_from_fixed): Likewise.
3732 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
3734 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3735 Alan Hayward <alan.hayward@arm.com>
3736 David Sherwood <david.sherwood@arm.com>
3738 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
3739 of separate mode class checks. Do not allow vector modes here.
3740 (immed_wide_int_const): Use as_a <scalar_mode>.
3741 * explow.c (trunc_int_for_mode): Likewise.
3742 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
3743 (wi::shwi): Likewise.
3744 (wi::min_value): Likewise.
3745 (wi::max_value): Likewise.
3746 * dwarf2out.c (loc_descriptor): Likewise.
3747 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
3750 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3751 Alan Hayward <alan.hayward@arm.com>
3752 David Sherwood <david.sherwood@arm.com>
3754 * tree.h (SCALAR_TYPE_MODE): New macro.
3755 * expr.c (expand_expr_addr_expr_1): Use it.
3756 (expand_expr_real_2): Likewise.
3757 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
3758 (fold_convert_const_fixed_from_int): Likewise.
3759 (fold_convert_const_fixed_from_real): Likewise.
3760 (native_encode_fixed): Likewise
3761 (native_encode_complex): Likewise
3762 (native_encode_vector): Likewise.
3763 (native_interpret_fixed): Likewise.
3764 (native_interpret_real): Likewise.
3765 (native_interpret_complex): Likewise.
3766 (native_interpret_vector): Likewise.
3767 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
3768 (simd_clone_adjust_argument_types): Likewise.
3769 (simd_clone_init_simd_arrays): Likewise.
3770 (simd_clone_adjust): Likewise.
3771 * stor-layout.c (layout_type): Likewise.
3772 * tree.c (build_minus_one_cst): Likewise.
3773 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
3774 * tree-inline.c (estimate_move_cost): Likewise.
3775 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
3776 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
3777 (vectorizable_reduction): Likewise.
3778 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
3779 (vect_recog_mixed_size_cond_pattern): Likewise.
3780 (check_bool_pattern): Likewise.
3781 (adjust_bool_pattern): Likewise.
3782 (search_type_for_mask_1): Likewise.
3783 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
3784 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3785 (vectorizable_load): Likewise.
3786 (vectorizable_store): Likewise.
3787 * ubsan.c (ubsan_encode_value): Likewise.
3788 * varasm.c (output_constant): Likewise.
3790 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3791 Alan Hayward <alan.hayward@arm.com>
3792 David Sherwood <david.sherwood@arm.com>
3794 * coretypes.h (scalar_mode): New class.
3795 * machmode.h (scalar_mode): Likewise.
3796 (scalar_mode::includes_p): New function.
3797 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
3798 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
3799 * genmodes.c (get_mode_class): Handle remaining scalar modes.
3800 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
3801 * expmed.c (store_bit_field_1): Likewise.
3802 (extract_bit_field_1): Likewise.
3803 * expr.c (write_complex_part): Likewise.
3804 (read_complex_part): Likewise.
3805 (emit_move_complex_push): Likewise.
3806 (expand_expr_real_2): Likewise.
3807 * function.c (assign_parm_setup_reg): Likewise.
3808 (assign_parms_unsplit_complex): Likewise.
3809 * optabs.c (expand_binop): Likewise.
3810 * rtlanal.c (subreg_get_info): Likewise.
3811 * simplify-rtx.c (simplify_immed_subreg): Likewise.
3812 * varasm.c (output_constant_pool_2): Likewise.
3814 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3815 Alan Hayward <alan.hayward@arm.com>
3816 David Sherwood <david.sherwood@arm.com>
3818 * expmed.c (extract_high_half): Use scalar_int_mode and remove
3820 (expmed_mult_highpart_optab): Likewise.
3821 (expmed_mult_highpart): Likewise.
3823 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3824 Alan Hayward <alan.hayward@arm.com>
3825 David Sherwood <david.sherwood@arm.com>
3827 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
3828 instead of a machine_mode.
3829 (builtin_memset_read_str): Likewise.
3830 * builtins.c (c_readstr): Likewise.
3831 (builtin_memcpy_read_str): Likewise.
3832 (builtin_strncpy_read_str): Likewise.
3833 (builtin_memset_read_str): Likewise.
3834 (builtin_memset_gen_str): Likewise.
3835 (expand_builtin_signbit): Use scalar_int_mode for local variables.
3836 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
3837 instead of a machine_mode.
3838 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
3840 (make_extraction): Likewise.
3841 (try_widen_shift_mode): Take and return scalar_int_modes instead
3843 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
3844 a scalar_int_mode instead of a machine_mode.
3845 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
3846 (avr_addr_space_pointer_mode): Likewise.
3847 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
3848 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
3849 (msp430_unwind_word_mode): Likewise.
3850 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
3851 (spu_addr_space_pointer_mode): Likewise.
3852 (spu_addr_space_address_mode): Likewise.
3853 (spu_libgcc_cmp_return_mode): Likewise.
3854 (spu_libgcc_shift_count_mode): Likewise.
3855 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
3856 (rl78_addr_space_pointer_mode): Likewise.
3857 (fl78_unwind_word_mode): Likewise.
3858 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
3860 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
3861 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
3862 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
3863 (mips_valid_pointer_mode): Likewise.
3864 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
3865 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
3866 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
3868 (ft32_addr_space_address_mode): Likewise.
3869 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
3870 scalar_int_mode instead of a machine_mode.
3871 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
3873 (m32c_addr_space_address_mode): Likewise.
3874 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
3875 (rs6000_eh_return_filter_mode): Likewise.
3876 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
3877 (rs6000_eh_return_filter_mode): Likewise.
3878 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
3879 (s390_libgcc_shift_count_mode): Likewise.
3880 (s390_unwind_word_mode): Likewise.
3881 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
3883 * target.def (mode_rep_extended): Likewise.
3884 (valid_pointer_mode): Likewise.
3885 (addr_space.valid_pointer_mode): Likewise.
3886 (eh_return_filter_mode): Return a scalar_int_mode rather than
3888 (libgcc_cmp_return_mode): Likewise.
3889 (libgcc_shift_count_mode): Likewise.
3890 (unwind_word_mode): Likewise.
3891 (addr_space.pointer_mode): Likewise.
3892 (addr_space.address_mode): Likewise.
3893 * doc/tm.texi: Regenerate.
3894 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
3896 (do_jump): Use scalar_int_mode for local variables.
3897 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
3898 rather than a machine_mode.
3899 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
3900 (scompare_loc_descriptor_wide): Likewise.
3901 (scompare_loc_descriptor_narrow): Likewise.
3902 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
3904 * except.c (sjlj_emit_dispatch_table): Likewise.
3905 (expand_builtin_eh_copy_values): Likewise.
3906 * explow.c (convert_memory_address_addr_space_1): Likewise.
3907 Take a scalar_int_mode rather than a machine_mode.
3908 (convert_memory_address_addr_space): Take a scalar_int_mode rather
3909 than a machine_mode.
3910 (memory_address_addr_space): Use scalar_int_mode for local variables.
3911 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
3912 rather than a machine_mode.
3913 * expmed.c (mask_rtx): Likewise.
3914 (init_expmed_one_conv): Likewise.
3915 (expand_mult_highpart_adjust): Likewise.
3916 (extract_high_half): Likewise.
3917 (expmed_mult_highpart_optab): Likewise.
3918 (expmed_mult_highpart): Likewise.
3919 (expand_smod_pow2): Likewise.
3920 (expand_sdiv_pow2): Likewise.
3921 (emit_store_flag_int): Likewise.
3922 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
3924 (extract_low_bits): Likewise.
3925 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
3927 * expr.c (pieces_addr::adjust): Likewise.
3928 (can_store_by_pieces): Likewise.
3929 (store_by_pieces): Likewise.
3930 (clear_by_pieces_1): Likewise.
3931 (expand_expr_addr_expr_1): Likewise.
3932 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
3933 (expand_expr_real_1): Likewise.
3934 (try_casesi): Likewise.
3935 * final.c (shorten_branches): Likewise.
3936 * fold-const.c (fold_convert_const_int_from_fixed): Change the
3937 type of "mode" to machine_mode.
3938 * internal-fn.c (expand_arith_overflow_result_store): Take a
3939 scalar_int_mode rather than a machine_mode.
3940 (expand_mul_overflow): Use scalar_int_mode for local variables.
3941 * loop-doloop.c (doloop_modify): Likewise.
3942 (doloop_optimize): Likewise.
3943 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
3944 than a machine_mode.
3945 (expand_doubleword_shift_condmove): Likewise.
3946 (expand_doubleword_shift): Likewise.
3947 (expand_doubleword_clz): Likewise.
3948 (expand_doubleword_popcount): Likewise.
3949 (expand_doubleword_parity): Likewise.
3950 (expand_absneg_bit): Use scalar_int_mode for local variables.
3951 (prepare_float_lib_cmp): Likewise.
3952 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
3953 rather than a machine_mode.
3954 (convert_memory_address_addr_space): Likewise.
3955 (get_mode_bounds): Likewise.
3956 (get_address_mode): Return a scalar_int_mode rather than a
3958 * rtlanal.c (get_address_mode): Likewise.
3959 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
3960 than a machine_mode.
3961 * targhooks.c (default_mode_rep_extended): Likewise.
3962 (default_valid_pointer_mode): Likewise.
3963 (default_addr_space_valid_pointer_mode): Likewise.
3964 (default_eh_return_filter_mode): Return a scalar_int_mode rather
3965 than a machine_mode.
3966 (default_libgcc_cmp_return_mode): Likewise.
3967 (default_libgcc_shift_count_mode): Likewise.
3968 (default_unwind_word_mode): Likewise.
3969 (default_addr_space_pointer_mode): Likewise.
3970 (default_addr_space_address_mode): Likewise.
3971 * targhooks.h (default_eh_return_filter_mode): Likewise.
3972 (default_libgcc_cmp_return_mode): Likewise.
3973 (default_libgcc_shift_count_mode): Likewise.
3974 (default_unwind_word_mode): Likewise.
3975 (default_addr_space_pointer_mode): Likewise.
3976 (default_addr_space_address_mode): Likewise.
3977 (default_mode_rep_extended): Take a scalar_int_mode rather than
3979 (default_valid_pointer_mode): Likewise.
3980 (default_addr_space_valid_pointer_mode): Likewise.
3981 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
3983 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
3984 rather than a machine_mode.
3985 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
3986 for local variables.
3987 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3988 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
3989 than a machine_mode.
3991 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3992 Alan Hayward <alan.hayward@arm.com>
3993 David Sherwood <david.sherwood@arm.com>
3995 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
3996 the mode argument to scalar_int_mode.
3997 (do_jump_by_parts_zero_rtx): Likewise.
3998 (do_jump_by_parts_equality_rtx): Likewise.
3999 (do_jump_by_parts_greater): Take a mode argument.
4000 (do_jump_by_parts_equality): Likewise.
4001 (do_jump_1): Update calls accordingly.
4003 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4004 Alan Hayward <alan.hayward@arm.com>
4005 David Sherwood <david.sherwood@arm.com>
4007 * is-a.h (safe_dyn_cast): New function.
4008 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
4009 (jump_table_for_label): Likewise.
4010 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
4011 instead of an rtx_insn *.
4012 (shorten_branches): Use dyn_cast instead of LABEL_P and
4013 JUMP_TABLE_DATA_P. Use jump_table_for_label and
4014 rtx_jump_table_data::get_data_mode.
4015 (final_scan_insn): Likewise.
4017 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4018 Alan Hayward <alan.hayward@arm.com>
4019 David Sherwood <david.sherwood@arm.com>
4021 * combine.c (try_combine): Use is_a <scalar_int_mode> when
4022 trying to combine a full-register integer set with a subreg
4025 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4026 Alan Hayward <alan.hayward@arm.com>
4027 David Sherwood <david.sherwood@arm.com>
4029 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
4030 that is always either address_mode or pointer_mode.
4032 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4033 Alan Hayward <alan.hayward@arm.com>
4034 David Sherwood <david.sherwood@arm.com>
4036 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
4037 when the two are known to be equal.
4039 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4040 Alan Hayward <alan.hayward@arm.com>
4041 David Sherwood <david.sherwood@arm.com>
4043 * simplify-rtx.c (simplify_const_unary_operation): Use
4044 is_a <scalar_int_mode> instead of checking for a nonzero
4045 precision. Forcibly convert op_mode to a scalar_int_mode
4046 in that case. More clearly differentiate the operand and
4047 result modes and use the former when deciding what the value
4048 of a count-bits operation should be. Use is_int_mode instead
4049 of checking for a MODE_INT. Remove redundant check for whether
4050 this mode has a zero precision.
4052 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4053 Alan Hayward <alan.hayward@arm.com>
4054 David Sherwood <david.sherwood@arm.com>
4056 * optabs.c (widen_leading): Change the type of the mode argument
4057 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
4058 (widen_bswap): Likewise.
4059 (expand_parity): Likewise.
4060 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
4061 (expand_ffs): Likewise.
4062 (epand_unop): Check for scalar integer modes before calling the
4065 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4066 Alan Hayward <alan.hayward@arm.com>
4067 David Sherwood <david.sherwood@arm.com>
4069 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
4071 (expand_expr_real_1): Update call accordingly.
4073 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4074 Alan Hayward <alan.hayward@arm.com>
4075 David Sherwood <david.sherwood@arm.com>
4077 * expmed.c (store_bit_field_using_insv): Add op0_mode and
4078 value_mode arguments. Use scalar_int_mode internally.
4079 (store_bit_field_1): Rename the new integer mode from imode
4080 to op0_mode and use it instead of GET_MODE (op0). Update calls
4081 to store_split_bit_field, store_bit_field_using_insv and
4082 store_fixed_bit_field.
4083 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
4084 Use scalar_int_mode internally. Use a bit count rather than a mode
4085 when calculating the largest bit size for get_best_mode.
4086 Update calls to store_split_bit_field and store_fixed_bit_field_1.
4087 (store_fixed_bit_field_1): Add mode and value_mode arguments.
4088 Remove assertion that OP0 has a scalar integer mode.
4089 (store_split_bit_field): Add op0_mode and value_mode arguments.
4090 Update calls to extract_fixed_bit_field.
4091 (extract_bit_field_using_extv): Add an op0_mode argument.
4092 Use scalar_int_mode internally.
4093 (extract_bit_field_1): Rename the new integer mode from imode to
4094 op0_mode and use it instead of GET_MODE (op0). Update calls to
4095 extract_split_bit_field, extract_bit_field_using_extv and
4096 extract_fixed_bit_field.
4097 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
4098 to extract_split_bit_field and extract_fixed_bit_field_1.
4099 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
4100 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
4102 (extract_split_bit_field): Add an op0_mode argument. Update call
4103 to extract_fixed_bit_field.
4105 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4106 Alan Hayward <alan.hayward@arm.com>
4107 David Sherwood <david.sherwood@arm.com>
4109 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
4110 * explow.c (hard_function_value): Likewise.
4111 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
4112 convert_to_mode call outside the loop.
4113 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
4114 for the mode iterator. Require the mode specified by max_pieces
4116 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
4118 (copy_blkmode_to_reg): Likewise.
4119 (set_storage_via_setmem): Likewise.
4120 * optabs.c (prepare_cmp_insn): Likewise.
4121 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4122 * stor-layout.c (finish_bitfield_representative): Likewise.
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 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
4129 * expr.c (convert_move): Use them.
4130 (convert_modes): Likewise.
4131 (store_expr_with_bounds): Likewise.
4133 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4134 Alan Hayward <alan.hayward@arm.com>
4135 David Sherwood <david.sherwood@arm.com>
4137 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
4138 parameter for the mode of "x". Remove the "known_x", "known_mode"
4139 and "known_ret" arguments. Change the type of the mode argument
4141 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
4142 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
4143 (reg_num_sign_bit_copies_for_combine): Likewise.
4144 * rtlanal.c (nonzero_bits1): Likewise.
4145 (num_sign_bit_copies1): Likewise.
4146 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
4147 (reg_num_sign_bit_copies_general): Likewise.
4148 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
4149 (reg_nonzero_bits_general): Likewise.
4151 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4152 Alan Hayward <alan.hayward@arm.com>
4153 David Sherwood <david.sherwood@arm.com>
4155 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
4156 than in subroutines. Return 1 for non-integer modes.
4157 (cached_num_sign_bit_copies): Change the type of the mode parameter
4159 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
4160 classes. Handle CONST_INT_P first and then check whether X also
4161 has a scalar integer mode. Check the same thing for inner registers
4162 of a SUBREG and for values that are being extended or truncated.
4164 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4165 Alan Hayward <alan.hayward@arm.com>
4166 David Sherwood <david.sherwood@arm.com>
4168 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
4169 in subroutines. Return the mode mask for non-integer modes.
4170 (cached_nonzero_bits): Change the type of the mode parameter
4172 (nonzero_bits1): Likewise. Remove early exit for other mode
4173 classes. Handle CONST_INT_P first and then check whether X
4174 also has a scalar integer mode.
4176 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4177 Alan Hayward <alan.hayward@arm.com>
4178 David Sherwood <david.sherwood@arm.com>
4180 * expr.c (widest_int_mode_for_size): Make the comment match the code.
4181 Return a scalar_int_mode and assert that the size is greater than
4183 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
4184 (op_by_pieces_d::op_by_pieces_d): Likewise.
4185 (op_by_pieces_d::run): Likewise.
4186 (can_store_by_pieces): Likewise.
4188 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4189 Alan Hayward <alan.hayward@arm.com>
4190 David Sherwood <david.sherwood@arm.com>
4192 * combine.c (extract_left_shift): Add a mode argument and update
4194 (make_compound_operation_int): Change the type of the mode parameter
4195 to scalar_int_mode and update the call to extract_left_shift.
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 * combine.c (simplify_and_const_int): Change the type of the mode
4202 parameter to scalar_int_mode.
4203 (simplify_and_const_int_1): Likewise. Update recursive call.
4205 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4206 Alan Hayward <alan.hayward@arm.com>
4207 David Sherwood <david.sherwood@arm.com>
4209 * combine.c (simplify_compare_const): Check that the mode is a
4210 scalar_int_mode (rather than VOIDmode) before testing its
4212 (simplify_comparison): Move COMPARISON_P handling out of the
4213 loop and restrict the latter part of the loop to scalar_int_modes.
4214 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
4215 and when considering SUBREG_REGs. Use is_int_mode instead of
4216 checking GET_MODE_CLASS against MODE_INT.
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 * combine.c (try_widen_shift_mode): Move check for equal modes to...
4223 (simplify_shift_const_1): ...here. Use scalar_int_mode for
4224 shift_unit_mode and for modes involved in scalar shifts.
4226 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4227 Alan Hayward <alan.hayward@arm.com>
4228 David Sherwood <david.sherwood@arm.com>
4230 * combine.c (force_int_to_mode): New function, split out from...
4231 (force_to_mode): ...here. Keep xmode up-to-date and use it
4232 instead of GET_MODE (x).
4234 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4235 Alan Hayward <alan.hayward@arm.com>
4236 David Sherwood <david.sherwood@arm.com>
4238 * optabs-query.h (extraction_insn::struct_mode): Change type to
4239 opt_scalar_int_mode and update comment.
4240 (extraction_insn::field_mode): Change type to scalar_int_mode.
4241 (extraction_insn::pos_mode): Likewise.
4242 * combine.c (make_extraction): Update accordingly.
4243 * optabs-query.c (get_traditional_extraction_insn): Likewise.
4244 (get_optab_extraction_insn): Likewise.
4245 * recog.c (simplify_while_replacing): Likewise.
4246 * expmed.c (narrow_bit_field_mem): Change the type of the mode
4247 parameter to opt_scalar_int_mode.
4249 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4250 Alan Hayward <alan.hayward@arm.com>
4251 David Sherwood <david.sherwood@arm.com>
4253 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
4254 to a scalar_int_mode instead of a machine_mode.
4255 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
4256 (get_best_mode): Return a boolean and use a pointer argument to store
4257 the selected mode. Replace the limit mode parameter with a bit limit.
4258 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
4259 for the values returned by bit_field_mode_iterator::next_mode.
4260 (store_bit_field): Update call to get_best_mode.
4261 (store_fixed_bit_field): Likewise.
4262 (extract_fixed_bit_field): Likewise.
4263 * expr.c (optimize_bitfield_assignment_op): Likewise.
4264 * fold-const.c (optimize_bit_field_compare): Likewise.
4265 (fold_truth_andor_1): Likewise.
4266 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
4267 Update for new type of m_mode.
4268 (get_best_mode): As above.
4270 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4271 Alan Hayward <alan.hayward@arm.com>
4272 David Sherwood <david.sherwood@arm.com>
4274 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
4275 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
4276 (store_bit_field): Check is_a <scalar_int_mode> before calling
4277 strict_volatile_bitfield_p.
4278 (extract_bit_field): Likewise.
4280 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4281 Alan Hayward <alan.hayward@arm.com>
4282 David Sherwood <david.sherwood@arm.com>
4284 * target.def (cstore_mode): Return a scalar_int_mode.
4285 * doc/tm.texi: Regenerate.
4286 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
4287 * targhooks.h (default_cstore_mode): Likewise.
4288 * targhooks.c (default_cstore_mode): Likewise, using a forced
4290 * expmed.c (emit_cstore): Expect the target of the cstore to be
4293 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4294 Alan Hayward <alan.hayward@arm.com>
4295 David Sherwood <david.sherwood@arm.com>
4297 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
4299 (niter_desc): Likewise mode.
4300 (iv_analyze): Add a mode parameter.
4302 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
4303 and change its type to scalar_int_mode.
4304 * loop-iv.c: Update commentary at head of file.
4305 (iv_constant): Pass the mode paraeter before the rtx it describes
4306 and change its type to scalar_int_mode. Remove VOIDmode handling.
4307 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
4308 (iv_extend): Likewise.
4309 (shorten_into_mode): Likewise.
4310 (iv_add): Use scalar_int_mode.
4311 (iv_mult): Likewise.
4312 (iv_shift): Likewise.
4313 (canonicalize_iv_subregs): Likewise.
4314 (get_biv_step_1): Pass the outer_mode parameter before the rtx
4315 it describes and change its mode to scalar_int_mode. Also change
4316 the type of the returned inner_mode to scalar_int_mode.
4317 (get_biv_step): Likewise, turning outer_mode from a pointer
4318 into a direct parameter. Update call to get_biv_step_1.
4319 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
4320 iv_constant and get_biv_step.
4321 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
4322 and change its type to scalar_int_mode. Don't initialise iv->mode
4323 to VOIDmode and remove later checks for its still being VOIDmode.
4324 Update calls to iv_analyze_op and iv_analyze_expr. Check
4325 is_a <scalar_int_mode> when changing the mode under consideration.
4326 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
4327 Update call to iv_analyze_expr.
4328 (iv_analyze_op): Add a mode parameter. Reject subregs whose
4329 inner register is not also a scalar_int_mode. Update call to
4331 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
4332 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
4333 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
4334 separate mode class checks. Update calls to iv_analyze. Remove
4335 fix-up of VOIDmodes after iv_analyze_biv.
4336 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
4337 don't have a scalar_int_mode. Update call to biv_p.
4339 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4340 Alan Hayward <alan.hayward@arm.com>
4341 David Sherwood <david.sherwood@arm.com>
4343 * cfgexpand.c (convert_debug_memory_address): Use
4344 as_a <scalar_int_mode>.
4345 * combine.c (expand_compound_operation): Likewise.
4346 (make_extraction): Likewise.
4347 (change_zero_ext): Likewise.
4348 (simplify_comparison): Likewise.
4349 * cse.c (cse_insn): Likewise.
4350 * dwarf2out.c (minmax_loc_descriptor): Likewise.
4351 (mem_loc_descriptor): Likewise.
4352 (loc_descriptor): Likewise.
4353 * expmed.c (init_expmed_one_mode): Likewise.
4354 (synth_mult): Likewise.
4355 (emit_store_flag_1): Likewise.
4356 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
4357 of a comparison with size.
4358 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
4359 (reduce_to_bit_field_precision): Likewise.
4360 * function.c (expand_function_end): Likewise.
4361 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
4362 * loop-doloop.c (doloop_modify): Likewise.
4363 * optabs.c (expand_binop): Likewise.
4364 (expand_unop): Likewise.
4365 (expand_copysign_absneg): Likewise.
4366 (prepare_cmp_insn): Likewise.
4367 (maybe_legitimize_operand): Likewise.
4368 * recog.c (const_scalar_int_operand): Likewise.
4369 * rtlanal.c (get_address_mode): Likewise.
4370 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4371 (simplify_cond_clz_ctz): Likewise.
4372 * tree-nested.c (get_nl_goto_field): Likewise.
4373 * tree.c (build_vector_type_for_mode): Likewise.
4374 * var-tracking.c (use_narrower_mode): Likewise.
4376 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4377 Alan Hayward <alan.hayward@arm.com>
4378 David Sherwood <david.sherwood@arm.com>
4380 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
4381 * builtins.c (expand_builtin_signbit): Use it.
4382 * cfgexpand.c (expand_debug_expr): Likewise.
4383 * dojump.c (do_jump): Likewise.
4384 (do_compare_and_jump): Likewise.
4385 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
4386 * expmed.c (make_tree): Likewise.
4387 * expr.c (expand_expr_real_2): Likewise.
4388 (expand_expr_real_1): Likewise.
4389 (try_casesi): Likewise.
4390 * fold-const-call.c (fold_const_call_ss): Likewise.
4391 * fold-const.c (unextend): Likewise.
4392 (extract_muldiv_1): Likewise.
4393 (fold_single_bit_test): Likewise.
4394 (native_encode_int): Likewise.
4395 (native_encode_string): Likewise.
4396 (native_interpret_int): Likewise.
4397 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
4398 * internal-fn.c (expand_addsub_overflow): Likewise.
4399 (expand_neg_overflow): Likewise.
4400 (expand_mul_overflow): Likewise.
4401 (expand_arith_overflow): Likewise.
4402 * match.pd: Likewise.
4403 * stor-layout.c (layout_type): Likewise.
4404 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
4405 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
4406 * tree-ssanames.c (get_range_info): Likewise.
4407 * tree-switch-conversion.c (array_value_type) Likewise.
4408 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
4409 (vect_recog_divmod_pattern): Likewise.
4410 (vect_recog_mixed_size_cond_pattern): Likewise.
4411 * tree-vrp.c (extract_range_basic): Likewise.
4412 (simplify_float_conversion_using_ranges): Likewise.
4413 * tree.c (int_fits_type_p): Likewise.
4414 * ubsan.c (instrument_bool_enum_load): Likewise.
4415 * varasm.c (mergeable_string_section): Likewise.
4416 (narrowing_initializer_constant_valid_p): Likewise.
4417 (output_constant): Likewise.
4419 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4420 Alan Hayward <alan.hayward@arm.com>
4421 David Sherwood <david.sherwood@arm.com>
4423 * machmode.h (NARROWEST_INT_MODE): New macro.
4424 * expr.c (alignment_for_piecewise_move): Use it instead of
4425 GET_CLASS_NARROWEST_MODE (MODE_INT).
4426 (push_block): Likewise.
4427 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
4429 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4431 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4432 Alan Hayward <alan.hayward@arm.com>
4433 David Sherwood <david.sherwood@arm.com>
4435 * postreload.c (move2add_valid_value_p): Change the type of the
4436 mode parameter to scalar_int_mode.
4437 (move2add_use_add2_insn): Add a mode parameter and use it instead
4439 (move2add_use_add3_insn): Likewise.
4440 (reload_cse_move2add): Update accordingly.
4442 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4443 Alan Hayward <alan.hayward@arm.com>
4444 David Sherwood <david.sherwood@arm.com>
4446 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
4448 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
4449 * optabs.c (expand_unop): Likewise.
4451 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4452 Alan Hayward <alan.hayward@arm.com>
4453 David Sherwood <david.sherwood@arm.com>
4455 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
4456 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
4457 (popcount_loc_descriptor): Likewise.
4458 (bswap_loc_descriptor): Likewise.
4459 (rotate_loc_descriptor): Likewise.
4460 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
4461 calling the functions above.
4463 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4464 Alan Hayward <alan.hayward@arm.com>
4465 David Sherwood <david.sherwood@arm.com>
4467 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
4469 (try_combine): Likewise.
4470 (simplify_if_then_else): Likewise.
4471 * cse.c (cse_insn): Likewise.
4472 * dwarf2out.c (mem_loc_descriptor): Likewise.
4473 * emit-rtl.c (gen_lowpart_common): Likewise.
4474 * simplify-rtx.c (simplify_truncation): Likewise.
4475 (simplify_binary_operation_1): Likewise.
4476 (simplify_const_relational_operation): Likewise.
4477 (simplify_ternary_operation): Likewise.
4478 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
4480 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4481 Alan Hayward <alan.hayward@arm.com>
4482 David Sherwood <david.sherwood@arm.com>
4484 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
4485 * reload.c (push_reload): Likewise.
4486 (find_reloads): Likewise.
4488 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4489 Alan Hayward <alan.hayward@arm.com>
4490 David Sherwood <david.sherwood@arm.com>
4492 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
4493 (make_compound_operation_int): Likewise.
4494 (change_zero_ext): Likewise.
4495 * expr.c (convert_move): Likewise.
4496 (convert_modes): Likewise.
4497 * fwprop.c (forward_propagate_subreg): Likewise.
4498 * loop-iv.c (get_biv_step_1): Likewise.
4499 * optabs.c (widen_operand): Likewise.
4500 * postreload.c (move2add_valid_value_p): Likewise.
4501 * recog.c (simplify_while_replacing): Likewise.
4502 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4503 (simplify_binary_operation_1): Likewise. Remove redundant
4504 mode equality check.
4506 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4507 Alan Hayward <alan.hayward@arm.com>
4508 David Sherwood <david.sherwood@arm.com>
4510 * combine.c (combine_simplify_rtx): Add checks for
4511 is_a <scalar_int_mode>.
4512 (simplify_if_then_else): Likewise.
4513 (make_field_assignment): Likewise.
4514 (simplify_comparison): Likewise.
4515 * ifcvt.c (noce_try_bitop): Likewise.
4516 * loop-invariant.c (canonicalize_address_mult): Likewise.
4517 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4519 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4520 Alan Hayward <alan.hayward@arm.com>
4521 David Sherwood <david.sherwood@arm.com>
4523 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
4524 is_a <scalar_int_mode> instead of != BLKmode.
4526 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4527 Alan Hayward <alan.hayward@arm.com>
4528 David Sherwood <david.sherwood@arm.com>
4530 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
4531 instead of != VOIDmode.
4532 * combine.c (if_then_else_cond): Likewise.
4533 (change_zero_ext): Likewise.
4534 * dwarf2out.c (mem_loc_descriptor): Likewise.
4535 (loc_descriptor): Likewise.
4536 * rtlanal.c (canonicalize_condition): Likewise.
4537 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
4539 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4540 Alan Hayward <alan.hayward@arm.com>
4541 David Sherwood <david.sherwood@arm.com>
4543 * simplify-rtx.c (simplify_binary_operation_1): Use
4544 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
4546 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4547 Alan Hayward <alan.hayward@arm.com>
4548 David Sherwood <david.sherwood@arm.com>
4550 * wide-int.h (int_traits<unsigned char>) New class.
4551 (int_traits<unsigned short>) Likewise.
4552 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
4553 Use GET_MODE_UNIT_PRECISION and remove redundant test for
4555 * combine.c (set_nonzero_bits_and_sign_copies): Use
4556 is_a <scalar_int_mode>.
4557 (find_split_point): Likewise.
4558 (combine_simplify_rtx): Likewise.
4559 (simplify_logical): Likewise.
4560 (expand_compound_operation): Likewise.
4561 (expand_field_assignment): Likewise.
4562 (make_compound_operation): Likewise.
4563 (extended_count): Likewise.
4564 (change_zero_ext): Likewise.
4565 (simplify_comparison): Likewise.
4566 * dwarf2out.c (scompare_loc_descriptor): Likewise.
4567 (ucompare_loc_descriptor): Likewise.
4568 (minmax_loc_descriptor): Likewise.
4569 (mem_loc_descriptor): Likewise.
4570 (loc_descriptor): Likewise.
4571 * expmed.c (init_expmed_one_mode): Likewise.
4572 * lra-constraints.c (lra_constraint_offset): Likewise.
4573 * optabs.c (prepare_libcall_arg): Likewise.
4574 * postreload.c (move2add_note_store): Likewise.
4575 * reload.c (operands_match_p): Likewise.
4576 * rtl.h (load_extend_op): Likewise.
4577 * rtlhooks.c (gen_lowpart_general): Likewise.
4578 * simplify-rtx.c (simplify_truncation): Likewise.
4579 (simplify_unary_operation_1): Likewise.
4580 (simplify_binary_operation_1): Likewise.
4581 (simplify_const_binary_operation): Likewise.
4582 (simplify_const_relational_operation): Likewise.
4583 (simplify_subreg): Likewise.
4584 * stor-layout.c (bitwise_mode_for_mode): Likewise.
4585 * var-tracking.c (adjust_mems): Likewise.
4586 (prepare_call_arguments): Likewise.
4588 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4589 Alan Hayward <alan.hayward@arm.com>
4590 David Sherwood <david.sherwood@arm.com>
4592 * machmode.h (is_int_mode): New fuction.
4593 * combine.c (find_split_point): Use it.
4594 (combine_simplify_rtx): Likewise.
4595 (simplify_if_then_else): Likewise.
4596 (simplify_set): Likewise.
4597 (simplify_shift_const_1): Likewise.
4598 (simplify_comparison): Likewise.
4599 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
4600 * cse.c (notreg_cost): Likewise.
4601 (cse_insn): Likewise.
4602 * cselib.c (cselib_lookup_1): Likewise.
4603 * dojump.c (do_jump_1): Likewise.
4604 (do_compare_rtx_and_jump): Likewise.
4605 * dse.c (get_call_args): Likewise.
4606 * dwarf2out.c (rtl_for_decl_init): Likewise.
4607 (native_encode_initializer): Likewise.
4608 * expmed.c (emit_store_flag_1): Likewise.
4609 (emit_store_flag): Likewise.
4610 * expr.c (convert_modes): Likewise.
4611 (store_field): Likewise.
4612 (expand_expr_real_1): Likewise.
4613 * fold-const.c (fold_read_from_constant_string): Likewise.
4614 * gimple-ssa-sprintf.c (get_format_string): Likewise.
4615 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
4616 * optabs.c (expand_binop): Likewise.
4617 (expand_unop): Likewise.
4618 (expand_abs_nojump): Likewise.
4619 (expand_one_cmpl_abs_nojump): Likewise.
4620 * simplify-rtx.c (mode_signbit_p): Likewise.
4621 (val_signbit_p): Likewise.
4622 (val_signbit_known_set_p): Likewise.
4623 (val_signbit_known_clear_p): Likewise.
4624 (simplify_relational_operation_1): Likewise.
4625 * tree.c (vector_type_mode): Likewise.
4627 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4628 Alan Hayward <alan.hayward@arm.com>
4629 David Sherwood <david.sherwood@arm.com>
4631 * machmode.h (smallest_mode_for_size): Fix formatting.
4632 (smallest_int_mode_for_size): New function.
4633 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
4634 instead of smallest_mode_for_size.
4635 * combine.c (make_extraction): Likewise.
4636 * config/arc/arc.c (arc_expand_movmem): Likewise.
4637 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
4638 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
4639 * config/s390/s390.c (s390_expand_insv): Likewise.
4640 * config/sparc/sparc.c (assign_int_registers): Likewise.
4641 * config/spu/spu.c (spu_function_value): Likewise.
4642 (spu_function_arg): Likewise.
4643 * coverage.c (get_gcov_type): Likewise.
4644 (get_gcov_unsigned_t): Likewise.
4645 * dse.c (find_shift_sequence): Likewise.
4646 * expmed.c (store_bit_field_1): Likewise.
4647 * expr.c (convert_move): Likewise.
4648 (store_field): Likewise.
4649 * internal-fn.c (expand_arith_overflow): Likewise.
4650 * optabs-query.c (get_best_extraction_insn): Likewise.
4651 * optabs.c (expand_twoval_binop_libfunc): Likewise.
4652 * stor-layout.c (layout_type): Likewise.
4653 (initialize_sizetypes): Likewise.
4654 * targhooks.c (default_get_mask_mode): Likewise.
4655 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
4657 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4658 Alan Hayward <alan.hayward@arm.com>
4659 David Sherwood <david.sherwood@arm.com>
4661 * machmode.h (opt_mode::else_blk): New function.
4662 (int_mode_for_mode): Declare.
4663 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
4664 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
4666 * cfgexpand.c (expand_debug_expr): Likewise.
4667 * combine.c (gen_lowpart_or_truncate): Likewise.
4668 (gen_lowpart_for_combine): Likewise.
4669 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
4670 * config/avr/avr.c (avr_to_int_mode): Likewise.
4671 (avr_out_plus_1): Likewise.
4672 (avr_out_plus): Likewise.
4673 (avr_out_round): Likewise.
4674 * config/i386/i386.c (ix86_split_to_parts): Likewise.
4675 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
4676 (s390_expand_vcond): Likewise.
4677 * config/spu/spu.c (spu_split_immediate): Likewise.
4678 (spu_expand_mov): Likewise.
4679 * dse.c (get_stored_val): Likewise.
4680 * expmed.c (store_bit_field_1): Likewise.
4681 (convert_extracted_bit_field): Use int_mode_for_mode instead of
4683 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
4684 (extract_low_bits): Likewise.
4685 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
4686 handling rather than repeating the check.
4687 (emit_group_store): Likewise.
4688 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
4689 * optabs.c (expand_absneg_bit): Likewise.
4690 (expand_copysign_absneg): Likewise.
4691 (expand_copysign_bit): Likewise.
4692 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
4693 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
4694 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
4695 * var-tracking.c (prepare_call_arguments): Likewise.
4696 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
4697 int_mode_for_mode instead of mode_for_size.
4698 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
4700 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4701 Alan Hayward <alan.hayward@arm.com>
4702 David Sherwood <david.sherwood@arm.com>
4704 * machmode.h (int_mode_for_size): New function.
4705 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
4706 instead of mode_for_size.
4707 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
4709 * combine.c (expand_field_assignment): Use int_mode_for_size
4710 instead of mode_for_size.
4711 (make_extraction): Likewise.
4712 (simplify_shift_const_1): Likewise.
4713 (simplify_comparison): Likewise.
4714 * dojump.c (do_jump): Likewise.
4715 * dwarf2out.c (mem_loc_descriptor): Likewise.
4716 * emit-rtl.c (init_derived_machine_modes): Likewise.
4717 * expmed.c (flip_storage_order): Likewise.
4718 (convert_extracted_bit_field): Likewise.
4719 * expr.c (copy_blkmode_from_reg): Likewise.
4720 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
4721 * internal-fn.c (expand_mul_overflow): Likewise.
4722 * lower-subreg.c (simple_move): Likewise.
4723 * optabs-libfuncs.c (init_optabs): Likewise.
4724 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4725 * tree.c (vector_type_mode): Likewise.
4726 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
4727 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
4728 * tree-vect-generic.c (expand_vector_parallel): Likewise.
4729 * tree-vect-stmts.c (vectorizable_load): Likewise.
4730 (vectorizable_store): Likewise.
4732 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4733 Alan Hayward <alan.hayward@arm.com>
4734 David Sherwood <david.sherwood@arm.com>
4736 * coretypes.h (pod_mode): New type.
4737 (scalar_int_mode_pod): New typedef.
4738 * machmode.h (pod_mode): New class.
4739 (int_n_data_t::m): Change type to scalar_int_mode_pod.
4740 * genmodes.c (emit_mode_int_n): Update accordingly.
4741 * lower-subreg.h (target_lower_subreg): Change type to
4742 scalar_int_mode_pod.
4743 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
4744 scalar_int_mode_pod.
4746 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4747 Alan Hayward <alan.hayward@arm.com>
4748 David Sherwood <david.sherwood@arm.com>
4750 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
4751 machine_mode to scalar_int_mode.
4752 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
4753 (rs6000_option_override_internal): Remove cast to int.
4754 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
4755 machine_mode to scalar_int_mode.
4756 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
4757 (rs6000_option_override_internal): Remove cast to int.
4758 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
4759 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
4761 * config/s390/s390.c (s390_expand_builtin): Likewise.
4762 * coretypes.h (scalar_int_mode): New type.
4763 (opt_scalar_int_mode): New typedef.
4764 * machmode.h (scalar_int_mode): New class.
4765 (scalar_int_mode::includes_p): New function.
4766 (byte_mode): Change type to scalar_int_mode.
4767 (word_mode): Likewise.
4768 (ptr_mode): Likewise.
4769 * emit-rtl.c (byte_mode): Likewise.
4770 (word_mode): Likewise.
4771 (ptr_mode): Likewise.
4772 (init_derived_machine_modes): Update accordingly.
4773 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
4774 and MODE_PARTIAL_INT.
4775 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
4776 opt_scalar_int_mode.
4778 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4779 Alan Hayward <alan.hayward@arm.com>
4780 David Sherwood <david.sherwood@arm.com>
4782 * target.def (libgcc_floating_mode_supported_p): Take a
4784 * doc/tm.texi: Regenerate.
4785 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
4787 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
4788 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
4791 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4792 Alan Hayward <alan.hayward@arm.com>
4793 David Sherwood <david.sherwood@arm.com>
4795 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
4796 * doc/tm.texi: Regenerate.
4797 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
4798 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
4799 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
4800 * targhooks.h (default_floatn_mode): Likewise.
4801 * targhooks.c (default_floatn_mode): Likewise.
4802 * tree.c (build_common_tree_nodes): Update accordingly.
4804 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4805 Alan Hayward <alan.hayward@arm.com>
4806 David Sherwood <david.sherwood@arm.com>
4808 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
4809 (mode_iterator::iterate_p): Likewise.
4810 (mode_iterator::get_wider): Likewise.
4811 * expr.c (init_expr_target): Use opt_scalar_float_mode.
4813 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4814 Alan Hayward <alan.hayward@arm.com>
4815 David Sherwood <david.sherwood@arm.com>
4817 * coretypes.h (opt_scalar_float_mode): New typedef.
4818 * machmode.h (float_mode_for_size): New function.
4819 * emit-rtl.c (double_mode): Delete.
4820 (init_emit_once): Use float_mode_for_size.
4821 * stor-layout.c (layout_type): Likewise.
4822 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
4824 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4825 Alan Hayward <alan.hayward@arm.com>
4826 David Sherwood <david.sherwood@arm.com>
4828 * output.h (assemble_real): Take a scalar_float_mode.
4829 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
4830 * config/arm/arm.md (consttable_4): Likewise.
4831 (consttable_8): Likewise.
4832 (consttable_16): Likewise.
4833 * config/mips/mips.md (consttable_float): Likewise.
4834 * config/s390/s390.c (s390_output_pool_entry): Likewise.
4835 * varasm.c (assemble_real): Take a scalar_float_mode.
4836 (output_constant_pool_2): Update accordingly.
4837 (output_constant): Likewise.
4839 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4840 Alan Hayward <alan.hayward@arm.com>
4841 David Sherwood <david.sherwood@arm.com>
4843 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
4844 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
4845 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
4846 (native_encode_real): Likewise.
4847 (native_interpret_real): Likewise.
4848 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
4849 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4851 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4852 Alan Hayward <alan.hayward@arm.com>
4853 David Sherwood <david.sherwood@arm.com>
4855 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
4856 <scalar_float_mode>. Simplify.
4857 (gen_extend_conv_libfunc): Likewise.
4859 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4860 Alan Hayward <alan.hayward@arm.com>
4861 David Sherwood <david.sherwood@arm.com>
4863 * coretypes.h (scalar_float_mode): New type.
4864 * machmode.h (mode_traits::from_int): Use machine_mode if
4865 USE_ENUM_MODES is defined.
4866 (is_a): New function.
4868 (dyn_cast): Likewise.
4869 (scalar_float_mode): New class.
4870 (scalar_float_mode::includes_p): New function.
4871 (is_float_mode): Likewise.
4872 * gdbhooks.py (MachineModePrinter): New class.
4873 (build_pretty_printer): Use it for scalar_float_mode.
4874 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
4875 (format_helper::format_helper): Turn into a template.
4876 * genmodes.c (get_mode_class): New function.
4877 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
4878 or machine_mode if none.
4879 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
4880 as_a <scalar_float_mode>.
4881 * dwarf2out.c (mem_loc_descriptor): Likewise.
4882 (insert_float): Likewise.
4883 (add_const_value_attribute): Likewise.
4884 * simplify-rtx.c (simplify_immed_subreg): Likewise.
4885 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
4886 (expand_unop): Update accordingly.
4887 (expand_abs_nojump): Likewise.
4888 (expand_copysign_absneg): Take a scalar_float_mode.
4889 (expand_copysign_bit): Likewise.
4890 (expand_copysign): Update accordingly.
4892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4893 Alan Hayward <alan.hayward@arm.com>
4894 David Sherwood <david.sherwood@arm.com>
4896 * coretypes.h (opt_mode): New class.
4897 * machmode.h (opt_mode): Likewise.
4898 (opt_mode::else_void): New function.
4899 (opt_mode::require): Likewise.
4900 (opt_mode::exists): Likewise.
4901 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
4902 (GET_MODE_2XWIDER_MODE): Likewise.
4903 (mode_iterator::get_wider): Update accordingly.
4904 (mode_iterator::get_2xwider): Likewise.
4905 (mode_iterator::get_known_wider): Likewise, turning into a template.
4906 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
4907 forcing a wider mode to exist.
4908 * config/cr16/cr16.h (LONG_REG_P): Likewise.
4909 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4910 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
4911 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
4912 * lower-subreg.c (init_lower_subreg): Likewise.
4913 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
4914 on the final iteration.
4915 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
4916 a wider mode exists before asking for a move pattern.
4917 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
4918 forcing a wider mode to exist.
4919 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
4920 returning false if no such mode exists.
4921 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
4922 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
4923 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
4924 Avoid checking for a MODE_INT if we already know the mode is not a
4926 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
4927 forcing a wider mode to exist.
4928 (expmed_mult_highpart_optab): Likewise.
4929 (expmed_mult_highpart): Likewise.
4930 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
4932 * lto-streamer-in.c (lto_input_mode_table): Likewise.
4933 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
4934 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
4935 * internal-fn.c (expand_mul_overflow): Update use of
4936 GET_MODE_2XWIDER_MODE.
4937 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4938 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
4939 GET_MODE_WIDER_MODE.
4940 (convert_plusminus_to_widen): Likewise.
4941 * tree-switch-conversion.c (array_value_type): Likewise.
4942 * var-tracking.c (emit_note_insn_var_location): Likewise.
4943 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4944 Return false inside rather than outside the loop if no wider mode
4946 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
4947 and GET_MODE_2XWIDER_MODE
4948 (can_compare_p): Use else_void.
4949 * gdbhooks.py (OptMachineModePrinter): New class.
4950 (build_pretty_printer): Use it for opt_mode.
4952 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4953 Alan Hayward <alan.hayward@arm.com>
4954 David Sherwood <david.sherwood@arm.com>
4956 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
4957 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
4959 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4960 Alan Hayward <alan.hayward@arm.com>
4961 David Sherwood <david.sherwood@arm.com>
4963 * machmode.h (mode_traits): New structure.
4964 (get_narrowest_mode): New function.
4965 (mode_iterator::start): Likewise.
4966 (mode_iterator::iterate_p): Likewise.
4967 (mode_iterator::get_wider): Likewise.
4968 (mode_iterator::get_known_wider): Likewise.
4969 (mode_iterator::get_2xwider): Likewise.
4970 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
4971 (FOR_EACH_MODE): Likewise.
4972 (FOR_EACH_MODE_FROM): Likewise.
4973 (FOR_EACH_MODE_UNTIL): Likewise.
4974 (FOR_EACH_WIDER_MODE): Likewise.
4975 (FOR_EACH_2XWIDER_MODE): Likewise.
4976 * builtins.c (expand_builtin_strlen): Use new mode iterators.
4977 * combine.c (simplify_comparison): Likewise
4978 * config/i386/i386.c (type_natural_mode): Likewise.
4979 * cse.c (cse_insn): Likewise.
4980 * dse.c (find_shift_sequence): Likewise.
4981 * emit-rtl.c (init_derived_machine_modes): Likewise.
4982 (init_emit_once): Likewise.
4983 * explow.c (hard_function_value): Likewise.
4984 * expmed.c (extract_fixed_bit_field_1): Likewise.
4985 (extract_bit_field_1): Likewise.
4986 (expand_divmod): Likewise.
4987 (emit_store_flag_1): Likewise.
4988 * expr.c (init_expr_target): Likewise.
4989 (convert_move): Likewise.
4990 (alignment_for_piecewise_move): Likewise.
4991 (widest_int_mode_for_size): Likewise.
4992 (emit_block_move_via_movmem): Likewise.
4993 (copy_blkmode_to_reg): Likewise.
4994 (set_storage_via_setmem): Likewise.
4995 (compress_float_constant): Likewise.
4996 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4997 * optabs-query.c (get_best_extraction_insn): Likewise.
4998 * optabs.c (expand_binop): Likewise.
4999 (expand_twoval_unop): Likewise.
5000 (expand_twoval_binop): Likewise.
5001 (widen_leading): Likewise.
5002 (widen_bswap): Likewise.
5003 (expand_parity): Likewise.
5004 (expand_unop): Likewise.
5005 (prepare_cmp_insn): Likewise.
5006 (prepare_float_lib_cmp): Likewise.
5007 (expand_float): Likewise.
5008 (expand_fix): Likewise.
5009 (expand_sfix_optab): Likewise.
5010 * postreload.c (move2add_use_add2_insn): Likewise.
5011 * reg-stack.c (reg_to_stack): Likewise.
5012 * reginfo.c (choose_hard_reg_mode): Likewise.
5013 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
5014 * stor-layout.c (mode_for_size): Likewise.
5015 (smallest_mode_for_size): Likewise.
5016 (mode_for_vector): Likewise.
5017 (finish_bitfield_representative): Likewise.
5018 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
5019 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
5020 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5021 * var-tracking.c (prepare_call_arguments): Likewise.
5023 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5024 Alan Hayward <alan.hayward@arm.com>
5025 David Sherwood <david.sherwood@arm.com>
5027 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
5028 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
5029 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
5030 * machmode.h (mode_size): Move earlier in file.
5031 (mode_precision): Likewise.
5032 (mode_inner): Likewise.
5033 (mode_nunits): Likewise.
5034 (mode_unit_size): Likewise.
5035 (unit_unit_precision): Likewise.
5036 (mode_wider): Likewise.
5037 (mode_2xwider): Likewise.
5038 (machine_mode): New class.
5039 (mode_to_bytes): New function.
5040 (mode_to_bits): Likewise.
5041 (mode_to_precision): Likewise.
5042 (mode_to_inner): Likewise.
5043 (mode_to_unit_size): Likewise.
5044 (mode_to_unit_precision): Likewise.
5045 (mode_to_nunits): Likewise.
5046 (GET_MODE_SIZE): Use mode_to_bytes.
5047 (GET_MODE_BITSIZE): Use mode_to_bits.
5048 (GET_MODE_PRECISION): Use mode_to_precision.
5049 (GET_MODE_INNER): Use mode_to_inner.
5050 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
5051 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
5052 (GET_MODE_NUNITS): Use mode_to_nunits.
5053 * system.h (ALWAYS_INLINE): New macro.
5054 * config/powerpcspe/powerpcspe-c.c
5055 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
5056 int for arg1_mode and arg2_mode.
5058 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5059 Alan Hayward <alan.hayward@arm.com>
5060 David Sherwood <david.sherwood@arm.com>
5062 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
5063 Prefix mode names with E_ in case statements.
5064 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5065 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
5066 (aarch64_split_simd_move): Likewise.
5067 (aarch64_gen_storewb_pair): Likewise.
5068 (aarch64_gen_loadwb_pair): Likewise.
5069 (aarch64_gen_store_pair): Likewise.
5070 (aarch64_gen_load_pair): Likewise.
5071 (aarch64_get_condition_code_1): Likewise.
5072 (aarch64_constant_pool_reload_icode): Likewise.
5073 (get_rsqrte_type): Likewise.
5074 (get_rsqrts_type): Likewise.
5075 (get_recpe_type): Likewise.
5076 (get_recps_type): Likewise.
5077 (aarch64_gimplify_va_arg_expr): Likewise.
5078 (aarch64_simd_container_mode): Likewise.
5079 (aarch64_emit_load_exclusive): Likewise.
5080 (aarch64_emit_store_exclusive): Likewise.
5081 (aarch64_expand_compare_and_swap): Likewise.
5082 (aarch64_gen_atomic_cas): Likewise.
5083 (aarch64_emit_bic): Likewise.
5084 (aarch64_emit_atomic_swap): Likewise.
5085 (aarch64_emit_atomic_load_op): Likewise.
5086 (aarch64_evpc_trn): Likewise.
5087 (aarch64_evpc_uzp): Likewise.
5088 (aarch64_evpc_zip): Likewise.
5089 (aarch64_evpc_ext): Likewise.
5090 (aarch64_evpc_rev): Likewise.
5091 (aarch64_evpc_dup): Likewise.
5092 (aarch64_gen_ccmp_first): Likewise.
5093 (aarch64_gen_ccmp_next): Likewise.
5094 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
5095 (alpha_emit_xfloating_libcall): Likewise.
5096 (emit_insxl): Likewise.
5097 (alpha_arg_type): Likewise.
5098 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
5099 (arc_preferred_simd_mode): Likewise.
5100 (arc_secondary_reload): Likewise.
5101 (get_arc_condition_code): Likewise.
5102 (arc_print_operand): Likewise.
5103 (arc_legitimate_constant_p): Likewise.
5104 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5105 * config/arc/arc.md (casesi_load): Likewise.
5106 (casesi_compact_jump): Likewise.
5107 * config/arc/predicates.md (proper_comparison_operator): Likewise.
5108 (cc_use_register): Likewise.
5109 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5110 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
5111 (arm_init_iwmmxt_builtins): Likewise.
5112 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
5113 (neon_expand_vector_init): Likewise.
5114 (arm_attr_length_move_neon): Likewise.
5115 (maybe_get_arm_condition_code): Likewise.
5116 (arm_emit_vector_const): Likewise.
5117 (arm_preferred_simd_mode): Likewise.
5118 (arm_output_iwmmxt_tinsr): Likewise.
5119 (thumb1_output_casesi): Likewise.
5120 (thumb2_output_casesi): Likewise.
5121 (arm_emit_load_exclusive): Likewise.
5122 (arm_emit_store_exclusive): Likewise.
5123 (arm_expand_compare_and_swap): Likewise.
5124 (arm_evpc_neon_vuzp): Likewise.
5125 (arm_evpc_neon_vzip): Likewise.
5126 (arm_evpc_neon_vrev): Likewise.
5127 (arm_evpc_neon_vtrn): Likewise.
5128 (arm_evpc_neon_vext): Likewise.
5129 (arm_validize_comparison): Likewise.
5130 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
5131 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
5132 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
5133 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
5134 (c6x_preferred_simd_mode): Likewise.
5135 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
5136 (epiphany_rtx_costs): Likewise.
5137 * config/epiphany/predicates.md (proper_comparison_operator):
5139 * config/frv/frv.c (condexec_memory_operand): Likewise.
5140 (frv_emit_move): Likewise.
5141 (output_move_single): Likewise.
5142 (output_condmove_single): Likewise.
5143 (frv_hard_regno_mode_ok): Likewise.
5144 (frv_matching_accg_mode): Likewise.
5145 * config/h8300/h8300.c (split_adds_subs): Likewise.
5146 (h8300_rtx_costs): Likewise.
5147 (h8300_print_operand): Likewise.
5148 (compute_mov_length): Likewise.
5149 (output_logical_op): Likewise.
5150 (compute_logical_op_length): Likewise.
5151 (compute_logical_op_cc): Likewise.
5152 (h8300_shift_needs_scratch_p): Likewise.
5153 (output_a_shift): Likewise.
5154 (compute_a_shift_length): Likewise.
5155 (compute_a_shift_cc): Likewise.
5156 (expand_a_rotate): Likewise.
5157 (output_a_rotate): Likewise.
5158 * config/i386/i386.c (classify_argument): Likewise.
5159 (function_arg_advance_32): Likewise.
5160 (function_arg_32): Likewise.
5161 (function_arg_64): Likewise.
5162 (function_value_64): Likewise.
5163 (ix86_gimplify_va_arg): Likewise.
5164 (ix86_legitimate_constant_p): Likewise.
5165 (put_condition_code): Likewise.
5166 (split_double_mode): Likewise.
5167 (ix86_avx256_split_vector_move_misalign): Likewise.
5168 (ix86_expand_vector_logical_operator): Likewise.
5169 (ix86_split_idivmod): Likewise.
5170 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
5171 (ix86_build_const_vector): Likewise.
5172 (ix86_build_signbit_mask): Likewise.
5173 (ix86_match_ccmode): Likewise.
5174 (ix86_cc_modes_compatible): Likewise.
5175 (ix86_expand_branch): Likewise.
5176 (ix86_expand_sse_cmp): Likewise.
5177 (ix86_expand_sse_movcc): Likewise.
5178 (ix86_expand_int_sse_cmp): Likewise.
5179 (ix86_expand_vec_perm_vpermi2): Likewise.
5180 (ix86_expand_vec_perm): Likewise.
5181 (ix86_expand_sse_unpack): Likewise.
5182 (ix86_expand_int_addcc): Likewise.
5183 (ix86_split_to_parts): Likewise.
5184 (ix86_vectorize_builtin_gather): Likewise.
5185 (ix86_vectorize_builtin_scatter): Likewise.
5186 (avx_vpermilp_parallel): Likewise.
5187 (inline_memory_move_cost): Likewise.
5188 (ix86_tieable_integer_mode_p): Likewise.
5189 (x86_maybe_negate_const_int): Likewise.
5190 (ix86_expand_vector_init_duplicate): Likewise.
5191 (ix86_expand_vector_init_one_nonzero): Likewise.
5192 (ix86_expand_vector_init_one_var): Likewise.
5193 (ix86_expand_vector_init_concat): Likewise.
5194 (ix86_expand_vector_init_interleave): Likewise.
5195 (ix86_expand_vector_init_general): Likewise.
5196 (ix86_expand_vector_set): Likewise.
5197 (ix86_expand_vector_extract): Likewise.
5198 (emit_reduc_half): Likewise.
5199 (ix86_emit_i387_round): Likewise.
5200 (ix86_mangle_type): Likewise.
5201 (ix86_expand_round_sse4): Likewise.
5202 (expand_vec_perm_blend): Likewise.
5203 (canonicalize_vector_int_perm): Likewise.
5204 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
5205 (expand_vec_perm_1): Likewise.
5206 (expand_vec_perm_interleave3): Likewise.
5207 (expand_vec_perm_even_odd_pack): Likewise.
5208 (expand_vec_perm_even_odd_1): Likewise.
5209 (expand_vec_perm_broadcast_1): Likewise.
5210 (ix86_vectorize_vec_perm_const_ok): Likewise.
5211 (ix86_expand_vecop_qihi): Likewise.
5212 (ix86_expand_mul_widen_hilo): Likewise.
5213 (ix86_expand_sse2_abs): Likewise.
5214 (ix86_expand_pextr): Likewise.
5215 (ix86_expand_pinsr): Likewise.
5216 (ix86_preferred_simd_mode): Likewise.
5217 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
5218 * config/i386/sse.md (*andnot<mode>3): Likewise.
5219 (<mask_codefor><code><mode>3<mask_name>): Likewise.
5220 (*<code><mode>3): Likewise.
5221 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
5222 (ia64_expand_atomic_op): Likewise.
5223 (ia64_arg_type): Likewise.
5224 (ia64_mode_to_int): Likewise.
5225 (ia64_scalar_mode_supported_p): Likewise.
5226 (ia64_vector_mode_supported_p): Likewise.
5227 (expand_vec_perm_broadcast): Likewise.
5228 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
5229 (iq2000_function_arg_advance): Likewise.
5230 (iq2000_function_arg): Likewise.
5231 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
5232 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
5233 (m68k_libcall_value): Likewise.
5234 (m68k_function_value): Likewise.
5235 (sched_attr_op_type): Likewise.
5236 * config/mcore/mcore.c (mcore_output_move): Likewise.
5237 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
5239 (microblaze_function_arg): Likewise.
5240 * config/mips/mips.c (mips16_build_call_stub): Likewise.
5241 (mips_print_operand): Likewise.
5242 (mips_mode_ok_for_mov_fmt_p): Likewise.
5243 (mips_vector_mode_supported_p): Likewise.
5244 (mips_preferred_simd_mode): Likewise.
5245 (mips_expand_vpc_loongson_even_odd): Likewise.
5246 (mips_expand_vec_unpack): Likewise.
5247 (mips_expand_vi_broadcast): Likewise.
5248 (mips_expand_vector_init): Likewise.
5249 (mips_expand_vec_reduc): Likewise.
5250 (mips_expand_msa_cmp): Likewise.
5251 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
5252 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
5253 (cc_flags_for_mode): Likewise.
5254 * config/msp430/msp430.c (msp430_print_operand): Likewise.
5255 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
5256 (nds32_output_casesi_pc_relative): Likewise.
5257 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5258 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
5259 (nvptx_gen_unpack): Likewise.
5260 (nvptx_gen_pack): Likewise.
5261 (nvptx_gen_shuffle): Likewise.
5262 (nvptx_gen_wcast): Likewise.
5263 (nvptx_preferred_simd_mode): Likewise.
5264 * config/pa/pa.c (pa_secondary_reload): Likewise.
5265 * config/pa/predicates.md (base14_operand): Likewise.
5266 * config/powerpcspe/powerpcspe-c.c
5267 (altivec_resolve_overloaded_builtin): Likewise.
5268 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
5270 (rs6000_preferred_simd_mode): Likewise.
5271 (output_vec_const_move): Likewise.
5272 (rs6000_expand_vector_extract): Likewise.
5273 (rs6000_split_vec_extract_var): Likewise.
5274 (reg_offset_addressing_ok_p): Likewise.
5275 (rs6000_legitimate_offset_address_p): Likewise.
5276 (rs6000_legitimize_address): Likewise.
5277 (rs6000_emit_set_const): Likewise.
5278 (rs6000_const_vec): Likewise.
5279 (rs6000_emit_move): Likewise.
5280 (spe_build_register_parallel): Likewise.
5281 (rs6000_darwin64_record_arg_recurse): Likewise.
5282 (swap_selector_for_mode): Likewise.
5283 (spe_init_builtins): Likewise.
5284 (paired_init_builtins): Likewise.
5285 (altivec_init_builtins): Likewise.
5286 (do_load_for_compare): Likewise.
5287 (rs6000_generate_compare): Likewise.
5288 (rs6000_expand_float128_convert): Likewise.
5289 (emit_load_locked): Likewise.
5290 (emit_store_conditional): Likewise.
5291 (rs6000_output_function_epilogue): Likewise.
5292 (rs6000_handle_altivec_attribute): Likewise.
5293 (rs6000_function_value): Likewise.
5294 (emit_fusion_gpr_load): Likewise.
5295 (emit_fusion_p9_load): Likewise.
5296 (emit_fusion_p9_store): Likewise.
5297 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
5298 (fusion_gpr_mem_load): Likewise.
5299 (fusion_addis_mem_combo_load): Likewise.
5300 (fusion_addis_mem_combo_store): Likewise.
5301 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
5302 (fusion_gpr_mem_load): Likewise.
5303 (fusion_addis_mem_combo_load): Likewise.
5304 (fusion_addis_mem_combo_store): Likewise.
5305 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5307 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
5308 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
5309 (rs6000_preferred_simd_mode): Likewise.
5310 (output_vec_const_move): Likewise.
5311 (rs6000_expand_vector_extract): Likewise.
5312 (rs6000_split_vec_extract_var): Likewise.
5313 (reg_offset_addressing_ok_p): Likewise.
5314 (rs6000_legitimate_offset_address_p): Likewise.
5315 (rs6000_legitimize_address): Likewise.
5316 (rs6000_emit_set_const): Likewise.
5317 (rs6000_const_vec): Likewise.
5318 (rs6000_emit_move): Likewise.
5319 (rs6000_darwin64_record_arg_recurse): Likewise.
5320 (swap_selector_for_mode): Likewise.
5321 (paired_init_builtins): Likewise.
5322 (altivec_init_builtins): Likewise.
5323 (rs6000_expand_float128_convert): Likewise.
5324 (emit_load_locked): Likewise.
5325 (emit_store_conditional): Likewise.
5326 (rs6000_output_function_epilogue): Likewise.
5327 (rs6000_handle_altivec_attribute): Likewise.
5328 (rs6000_function_value): Likewise.
5329 (emit_fusion_gpr_load): Likewise.
5330 (emit_fusion_p9_load): Likewise.
5331 (emit_fusion_p9_store): Likewise.
5332 * config/rx/rx.c (rx_gen_move_template): Likewise.
5333 (flags_from_mode): Likewise.
5334 * config/s390/predicates.md (s390_alc_comparison): Likewise.
5335 (s390_slb_comparison): Likewise.
5336 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
5337 (s390_vector_mode_supported_p): Likewise.
5338 (s390_cc_modes_compatible): Likewise.
5339 (s390_match_ccmode_set): Likewise.
5340 (s390_canonicalize_comparison): Likewise.
5341 (s390_emit_compare_and_swap): Likewise.
5342 (s390_branch_condition_mask): Likewise.
5343 (s390_rtx_costs): Likewise.
5344 (s390_secondary_reload): Likewise.
5345 (__SECONDARY_RELOAD_CASE): Likewise.
5346 (s390_expand_cs): Likewise.
5347 (s390_preferred_simd_mode): Likewise.
5348 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
5349 * config/sh/sh.c (sh_print_operand): Likewise.
5350 (dump_table): Likewise.
5351 (sh_secondary_reload): Likewise.
5352 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5353 * config/sh/sh.md (casesi_worker_1): Likewise.
5354 (casesi_worker_2): Likewise.
5355 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
5356 (fcc_comparison_operator): Likewise.
5357 * config/sparc/sparc.c (sparc_expand_move): Likewise.
5358 (emit_soft_tfmode_cvt): Likewise.
5359 (sparc_preferred_simd_mode): Likewise.
5360 (output_cbranch): Likewise.
5361 (sparc_print_operand): Likewise.
5362 (sparc_expand_vec_perm_bmask): Likewise.
5363 (vector_init_bshuffle): Likewise.
5364 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
5365 (spu_vector_mode_supported_p): Likewise.
5366 (spu_expand_insv): Likewise.
5367 (spu_emit_branch_or_set): Likewise.
5368 (spu_handle_vector_attribute): Likewise.
5369 (spu_builtin_splats): Likewise.
5370 (spu_builtin_extract): Likewise.
5371 (spu_builtin_promote): Likewise.
5372 (spu_expand_sign_extend): Likewise.
5373 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
5374 (tilegx_simd_int): Likewise.
5375 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
5376 (tilepro_simd_int): Likewise.
5377 * config/v850/v850.c (const_double_split): Likewise.
5378 (v850_print_operand): Likewise.
5379 (ep_memory_offset): Likewise.
5380 * config/vax/vax.c (vax_rtx_costs): Likewise.
5381 (vax_output_int_move): Likewise.
5382 (vax_output_int_add): Likewise.
5383 (vax_output_int_subtract): Likewise.
5384 * config/visium/predicates.md (visium_branch_operator): Likewise.
5385 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
5386 (visium_print_operand_address): Likewise.
5387 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5388 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
5389 (xtensa_expand_conditional_branch): Likewise.
5390 (xtensa_copy_incoming_a7): Likewise.
5391 (xtensa_output_literal): Likewise.
5392 * dfp.c (decimal_real_maxval): Likewise.
5393 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
5395 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5396 Alan Hayward <alan.hayward@arm.com>
5397 David Sherwood <david.sherwood@arm.com>
5399 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
5400 (mode_nunits_inline): Likewise.
5401 (mode_inner_inline): Likewise.
5402 (mode_unit_size_inline): Likewise.
5403 (mode_unit_precision_inline): Likewise.
5404 (emit_insn_modes_h): Likewise. Also emit a #define of the
5406 (emit_mode_wider): Add an E_ prefix to mode names.
5407 (emit_mode_complex): Likewise.
5408 (emit_mode_inner): Likewise.
5409 (emit_mode_adjustments): Likewise.
5410 (emit_mode_int_n): Likewise.
5411 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
5412 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
5413 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
5414 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
5415 (CRC32_BUILTIN, ENTRY): Likewise.
5416 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
5417 (aarch64_pop_regs): Likewise.
5418 (aarch64_process_components): Likewise.
5419 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
5420 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
5421 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
5422 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
5423 * config/arm/arm.c (arm_init_libfuncs): Likewise.
5424 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
5426 * config/i386/i386-builtin.def (pcmpestr): Likewise.
5427 (pcmpistr): Likewise.
5428 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
5429 * config/mmix/mmix.c (mmix_output_condition): Likewise.
5430 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
5432 * config/rl78/rl78.c (mduc_regs): Likewise.
5433 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
5434 (htm_expand_builtin): Likewise.
5435 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
5436 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
5437 * config/xtensa/xtensa.c (print_operand): Likewise.
5438 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
5439 (NUM_MODE_VECTOR_INT): Likewise.
5440 * genoutput.c (null_operand): Likewise.
5441 (output_operand_data): Likewise.
5442 * genrecog.c (print_parameter_value): Likewise.
5443 * lra.c (debug_operand_data): Likewise.
5445 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5447 * dwarf2out.c (scompare_loc_descriptor_wide)
5448 (scompare_loc_descriptor_narrow): New functions, split out from...
5449 (scompare_loc_descriptor): ...here.
5450 * expmed.c (emit_store_flag_int): New function, split out from...
5451 (emit_store_flag): ...here.
5453 2017-08-30 Richard Biener <rguenther@suse.de>
5455 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
5456 (dwarf2out_early_finish): Move setting of AT_pubnames from
5457 early debug output to early finish.
5459 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5461 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
5462 and -mdata-region to the assembler.
5464 2017-08-30 Richard Biener <rguenther@suse.de>
5466 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
5468 (gen_subprogram_die): Add DW_AT_object_pointer only early.
5469 (dwarf2out_early_global_decl): Only generate a DIE for the
5470 abstract origin if it doesn't already exist or is a declaration DIE.
5471 (resolve_addr): Do not add the linkage name twice when
5472 generating a stub DIE for the DW_TAG_GNU_call_site target.
5474 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5476 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5477 Use machine_mode rather than int for arg1_mode.
5479 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
5482 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
5483 that the second argument of the built-in functions to unpack
5484 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
5485 switch statement instead a lot of if statements.
5486 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
5487 Allow 64-bit values to be in Altivec registers as well as
5488 traditional floating point registers.
5489 (pack<mode>, FMOVE128_VSX iterator): Likewise.
5491 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
5493 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
5494 MAX_REGS_PER_ADDRESS == 1.
5496 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
5498 * config/i386/i386.opt (flag_fentry): Do not init to -1.
5499 * config/i386/i386.c (ix86_option_override_internal): Simplify
5500 setting of opts->x_flag_entry.
5502 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5503 Jakub Jelinek <jakub@redhat.com>
5504 Richard Biener <rguenther@suse.de>
5506 PR tree-optimization/81503
5507 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
5508 folded constant fits in the target type; reorder tests for clarity.
5510 2017-08-29 Martin Liska <mliska@suse.cz>
5512 * passes.def: Include pass_lower_switch.
5513 * stmt.c (dump_case_nodes): Remove and move to
5514 tree-switch-conversion.
5515 (case_values_threshold): Likewise.
5516 (expand_switch_as_decision_tree_p): Likewise.
5517 (emit_case_decision_tree): Likewise.
5518 (expand_case): Likewise.
5519 (balance_case_nodes): Likewise.
5520 (node_has_low_bound): Likewise.
5521 (node_has_high_bound): Likewise.
5522 (node_is_bounded): Likewise.
5523 (emit_case_nodes): Likewise.
5524 (struct simple_case_node): New struct.
5525 (add_case_node): Remove.
5526 (emit_case_dispatch_table): Use vector instead of case_list.
5527 (reset_out_edges_aux): Remove.
5528 (compute_cases_per_edge): Likewise.
5529 (expand_case): Build list of simple_case_node.
5530 (expand_sjlj_dispatch_table): Use it.
5531 * tree-switch-conversion.c (struct case_node): Moved from
5532 stmt.c and adjusted.
5533 (emit_case_nodes): Likewise.
5534 (node_has_low_bound): Likewise.
5535 (node_has_high_bound): Likewise.
5536 (node_is_bounded): Likewise.
5537 (case_values_threshold): Likewise.
5538 (reset_out_edges_aux): Likewise.
5539 (compute_cases_per_edge): Likewise.
5540 (add_case_node): Likewise.
5541 (dump_case_nodes): Likewise.
5542 (balance_case_nodes): Likewise.
5543 (expand_switch_as_decision_tree_p): Likewise.
5544 (emit_jump): Likewise.
5545 (emit_case_decision_tree): Likewise.
5546 (try_switch_expansion): Likewise.
5547 (do_jump_if_equal): Likewise.
5548 (emit_cmp_and_jump_insns): Likewise.
5549 (fix_phi_operands_for_edge): New function.
5550 (record_phi_operand_mapping): Likewise.
5551 (class pass_lower_switch): New pass.
5552 (pass_lower_switch::execute): New function.
5553 (make_pass_lower_switch): Likewise.
5554 (conditional_probability):
5555 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
5556 * tree-pass.h: Add make_pass_lower_switch.
5558 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5561 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
5564 2017-08-29 Richard Biener <rguenther@suse.de>
5566 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
5567 we do not add a DW_AT_inline attribute twice.
5568 (gen_subprogram_die): Remove code setting DW_AT_inline on
5569 DECL_ABSTRACT_P nodes.
5571 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
5573 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
5574 calls to internal functions.
5575 (gimplify_modify_expr): Likewise.
5576 * tree-call-cdce.c (use_internal_fn): Likewise.
5577 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
5578 (convert_to_divmod): Set the nothrow flag.
5579 * tree-if-conv.c (predicate_mem_writes): Likewise.
5580 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
5581 (vectorizable_call): Likewise.
5582 (vectorizable_store): Likewise.
5583 (vectorizable_load): Likewise.
5584 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
5585 (vect_recog_mask_conversion_pattern): Likewise.
5587 2017-08-29 Martin Liska <mliska@suse.cz>
5590 * gcc.c (driver_handle_option): Add new argument.
5591 * opts-common.c (handle_option): Pass
5592 target_option_override_hook.
5593 * opts-global.c (lang_handle_option): Add new option.
5594 (set_default_handlers): Add new argument.
5595 (decode_options): Likewise.
5596 * opts.c (target_handle_option): Likewise.
5597 (common_handle_option): Call target_option_override_hook.
5598 * opts.h (struct cl_option_handler_func): Add hook for
5599 target option override.
5600 (struct cl_option_handlers): Likewise.
5601 (set_default_handlers): Add new argument.
5602 (decode_options): Likewise.
5603 (common_handle_option): Likewise.
5604 (target_handle_option): Likewise.
5605 * toplev.c (toplev::main): Pass targetm.target_option.override
5608 2017-08-29 Richard Biener <rguenther@suse.de>
5609 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5611 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
5612 life to the active subtree.
5614 2017-08-28 Jeff Law <law@redhat.com>
5616 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
5617 derive_equivalences.
5618 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
5620 (edge_info::derive_equivalences): New private member function
5622 * tree-ssa-dom.c (class edge_info): Changed from a struct
5623 to a class. Add ctor/dtor, methods and data members.
5624 (edge_info::edge_info): Renamed from allocate_edge_info.
5625 Initialize additional members.
5626 (edge_info::~edge_info): New.
5627 (free_dom_edge_info): Delete the edge info.
5628 (record_edge_info): Use new class & associated member functions.
5629 Tighten forms for testing for edge equivalences.
5630 (record_temporary_equivalences): Iterate over the simple
5631 equivalences rather than assuming there's only one per edge.
5632 (cprop_into_successor_phis): Iterate over the simple
5633 equivalences rather than assuming there's only one per edge.
5634 (optimize_stmt): Use operand_equal_p rather than pointer
5635 equality for mini-DSE code.
5637 2017-08-28 Nathan Sidwell <nathan@acm.org>
5639 * gcc.c (execute): Fold SIGPIPE handling into switch
5640 statement. Adjust internal error message.
5642 2017-08-28 Richard Biener <rguenther@suse.de>
5645 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
5646 Do nothing for removed DIEs.
5648 2017-08-28 Richard Biener <rguenther@suse.de>
5650 PR tree-optimization/81977
5651 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
5654 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
5657 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
5658 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
5659 using targetm.gen_mem_thread_fence.
5661 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
5664 * config/i386/i386.md (*<btsc><mode>): Change operand 2
5665 predicate to register_operand. Reorder operands.
5666 (*btr<mode>): Ditto.
5667 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
5668 (*btr<mode>_mask): Ditto.
5670 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
5672 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
5673 * config/rs6000/xmmintrin.h: New file.
5674 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
5676 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5679 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
5680 parameter and_insn and return it.
5681 (recombine_lvx_pattern): Insert a copy to ensure availability of
5682 the base register of the copied masking operation at the point of
5683 the instruction replacement.
5684 (recombine_stvx_pattern): Likewise.
5686 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
5688 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
5689 undocumented switches.
5690 (-mpower9-dform-vector): Likewise.
5691 (-mpower9-dform): Likewise.
5692 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
5693 comments to delete references to -mpower9-dform* switches.
5694 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
5695 Delete reference to -mpower9-dform* switches, test for
5696 -mpower9-vector instead.
5697 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
5698 (OTHER_P9_VECTOR_MASKS): Likewise.
5699 (POWERPC_MASKS): Likewise.
5700 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
5701 tests against -mpower9-dform* to -mpower9-vector. Delete code
5702 that checked for -mpower9-dform* consistancy with other options.
5703 Add test for -mpower9-misc to enable other power9 switches.
5704 (rs6000_init_hard_regno_mode_ok): Likewise.
5705 (rs6000_option_override_internal): Likewise.
5706 (rs6000_emit_prologue): Likewise.
5707 (rs6000_emit_epilogue): Likewise.
5708 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
5709 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
5710 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
5712 (emit_fusion_p9_store): Likewise.
5713 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
5714 resetting these macros if the assembler does not support ISA 3.0
5716 (TARGET_P9_DFORM_VECTOR): Likewise.
5717 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
5718 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
5720 2017-08-25 Alan Modra <amodra@gmail.com>
5722 PR rtl-optimization/81747
5723 * cse.c (cse_extended_basic_block): Don't attempt to record
5724 equivalences for degenerate conditional branches that branch
5725 to their fall-through.
5727 2017-08-24 Martin Sebor <msebor@redhat.com>
5730 * gimple-fold.c (size_must_be_zero_p): New function.
5731 (gimple_fold_builtin_memory_op): Call it.
5733 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
5735 * config/rs6000/mm_malloc.h: New file.
5737 2017-08-24 Bin Cheng <bin.cheng@arm.com>
5739 PR tree-optimization/81913
5740 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
5741 analysis when either IVs in condition can wrap.
5743 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
5745 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
5746 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
5748 2017-08-24 Richard Biener <rguenther@suse.de>
5751 * targhooks.c (default_target_can_inline_p): Properly
5752 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
5753 is present and always compare.
5754 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
5755 infer -mfpmath=sse from TARGET_SSE_P.
5756 (ix86_can_inline_p): Properly use target_option_default_node when
5757 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
5759 2017-08-24 Richard Biener <rguenther@suse.de>
5762 * dwarf2out.c (output_die): Handle flag_generate_offload like
5764 (output_comp_unit): Likewise.
5765 (gen_array_type_die): Likewise.
5766 (dwarf2out_early_finish): Likewise.
5767 (note_variable_value_in_expr): Likewise.
5768 (dwarf2out_finish): Likewise. Adjust assert.
5769 * cgraphunit.c (symbol_table::compile): Move setting of
5770 flag_generate_offload earlier ...
5771 (symbol_table::finalize_compilation_unit): ... here, before
5772 early debug finalization.
5774 2017-08-24 Richard Biener <rguenther@suse.de>
5776 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
5777 and ipa-fnsummary.h.
5778 (ix86_can_inline_p): When ix86_fpmath flags do not match
5779 check whether the callee uses FP math at all.
5781 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
5784 * tree-ssanames.c (get_nonzero_bits): Use element_precision
5785 instead of TYPE_PRECISION.
5787 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
5788 Alan Hayward <alan.hayward@arm.com>
5789 David Sherwood <david.sherwood@arm.com>
5791 * combine.c (make_extraction): Use subreg_offset_from_lsb.
5793 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
5795 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
5797 (ix86_frame::stack_realign_allocate): New field.
5798 (struct machine_frame_state): Modify comments.
5799 (machine_frame_state::sp_realigned_fp_end): New field.
5800 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
5802 (sp_valid_at): Add assertion to assure no attempt to access invalid
5803 offset of a realigned stack.
5804 (fp_valid_at): Likewise.
5805 (choose_baseaddr): Modify comments.
5806 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
5807 ix86_expand_prologue.
5808 (ix86_expand_prologue): Modify stack realignment and allocation.
5809 (ix86_expand_epilogue): Modify comments.
5810 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
5811 avx2_runtime, avx512f, and avx512f_runtime.
5813 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
5815 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
5816 (mstackrealign): Do not init to -1.
5817 * config/i386/i386.c (ix86_option_override_internal):
5818 Check opts_set, not opts when setting default value of
5819 opts->x_ix86_force_align_arg_pointer.
5821 2017-08-23 Richard Biener <rguenther@suse.de>
5823 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
5824 lang_hooks.decl_printable_name.
5825 * print-rtl-function.c (print_rtx_function): Likewise.
5826 * tree-pretty-print.c (dump_function_header): Likewise.
5828 2017-08-23 Richard Biener <rguenther@suse.de>
5831 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
5832 -g0 at compile-time.
5834 2017-08-23 Tamar Christina <tamar.christina@arm.com>
5837 * doc/sourcebuild.texi (Other hardware attributes):
5840 2017-08-23 Tamar Christina <tamar.christina@arm.com>
5843 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
5844 Fix single-use check.
5846 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
5848 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
5850 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
5852 * doc/install.texi: Modify to add more details on running selected
5855 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
5857 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
5858 is combined with -mabi=ms.
5859 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
5862 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5864 PR tree-optimization/81488
5865 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
5866 and cached_basis fields.
5867 (MAX_SPREAD): New constant.
5868 (alloc_cand_and_find_basis): Initialize new fields.
5869 (clear_visited): New function.
5870 (create_phi_basis_1): Rename from create_phi_basis, set visited
5871 and cached_basis fields.
5872 (create_phi_basis): New wrapper function.
5873 (phi_add_costs_1): Rename from phi_add_costs, add spread
5874 parameter, set visited field, short-circuit when limits reached.
5875 (phi_add_costs): New wrapper function.
5876 (record_phi_increments_1): Rename from record_phi_increments, set
5878 (record_phi_increments): New wrapper function.
5879 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
5880 (phi_incr_cost): New wrapper function.
5881 (all_phi_incrs_profitable_1): Rename from
5882 all_phi_incrs_profitable, set visited field.
5883 (all_phi_incrs_profitable): New wrapper function.
5885 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
5886 Alan Hayward <alan.hayward@arm.com>
5887 David Sherwood <david.sherwood@arm.com>
5889 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
5890 that takes the outer and inner modes.
5891 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
5892 comparison as the canonical test for a paradoxical subreg.
5893 * combine.c (simplify_set): Use paradoxical_subreg_p.
5894 (make_extraction): Likewise.
5895 (force_to_mode): Likewise.
5896 (rtx_equal_for_field_assignment_p): Likewise.
5897 (gen_lowpart_for_combine): Likewise.
5898 (simplify_comparison): Likewise.
5899 * cse.c (equiv_constant): Likewise.
5900 * expmed.c (store_bit_field_1): Likewise.
5901 * final.c (alter_subreg): Likewise.
5902 * fwprop.c (propagate_rtx): Likewise.
5903 (forward_propagate_subreg): Likewise.
5904 * ira-conflicts.c (ira_build_conflicts): Likewise.
5905 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
5906 * lra-constraints.c (curr_insn_transform): Likewise.
5907 (split_reg): Likewise.
5908 * lra-eliminations.c (move_plus_up): Likewise.
5909 (lra_eliminate_regs_1): Likewise.
5910 * recog.c (general_operand): Likewise.
5911 * ree.c (combine_reaching_defs): Likewise.
5912 * reload.c (push_reload): Likewise.
5913 (find_reloads): Likewise.
5914 * reload1.c (elimination_effects): Likewise.
5915 (compute_reload_subreg_offset): Likewise.
5916 (choose_reload_regs): Likewise.
5917 * rtlanal.c (subreg_lsb_1): Likewise.
5918 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5919 (simplify_subreg): Likewise.
5920 * var-tracking.c (track_loc_p): Likewise.
5921 * emit-rtl.c (byte_lowpart_offset): Likewise.
5922 (paradoxical_subreg_p): Delete out-of-line definition.
5924 2017-08-22 Jeff Law <law@redhat.com>
5926 PR tree-optimization/81741
5927 PR tree-optimization/71947
5928 * tree-ssa-dom.c: Include tree-inline.h.
5929 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
5930 equivalences if one is more expensive to compute than the other.
5931 * tree-ssa-scopedtables.h (class const_or_copies): Make
5932 record_const_or_copy_raw method private.
5933 (class avail_exprs_stack): New method simplify_binary_operation.
5934 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
5935 avail_exprs_stack::simplify_binary_operation as needed.
5936 (avail_exprs_stack::simplify_binary_operation): New function.
5938 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
5940 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
5941 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
5942 (DOT_SYMBOLS): Likewise.
5943 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
5944 (RELOCATABLE_NEEDS_FIXUP): Likewise.
5945 (RS6000_ABI_NAME): Likewise.
5946 (TARGET_CMODEL): Likewise.
5947 (TOC_SECTION_ASM_OP): Likewise.
5948 (SET_CMODEL): New macro.
5949 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
5951 2017-08-22 Richard Biener <rguenther@suse.de>
5953 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
5954 to allow for free-lang-data replacements similar to verify_type_variant.
5956 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
5957 Alan Hayward <alan.hayward@arm.com>
5958 David Sherwood <david.sherwood@arm.com>
5960 * config/aarch64/aarch64.md (casesi): Use DImode rather than
5961 VOIDmode for the LABEL_REF.
5963 2017-08-22 Richard Biener <rguenther@suse.de>
5965 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
5967 2017-08-22 Richard Biener <rguenther@suse.de>
5969 * common.opt (feliminate-dwarf2-dups): Ignore.
5970 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
5971 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
5972 same_die_p_wrap, compute_section_prefix,
5973 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
5974 (comdat_symbol_id, comdat_symbol_number): Likewise.
5975 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
5977 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
5978 (output_die): Mark unreachable path unreachable.
5979 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
5980 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
5981 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
5982 (dwarf2out_early_finish): Likewise.
5984 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
5986 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
5988 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
5991 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
5992 not VAR_P. Filter attribute warnings with OPT_Wattributes.
5993 (avr_attribute_table) <io, io_low, address>: Initialize
5994 .decl_required with true.
5996 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
5998 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
5999 undocumented debugging options.
6000 (-mvsx-scalar-double): Likewise.
6001 (-mallow-df-permute): Likewise.
6002 (-mvectorize-builtins): Likewise.
6003 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
6004 (rs6000_builtin_vectorized_function): Likewise.
6005 (rs6000_builtin_md_vectorized_function): Likewise.
6006 (rs6000_opt_vars): Likewise.
6008 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
6011 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
6012 (*btrq_imm): Rename from *btrq.
6013 (*btcq_imm): Rename from *btcq.
6014 (btsc): New code attribute.
6015 (*<btsc><mode>): New insn pattern.
6016 (*btr<mode>): Ditto.
6017 (*<btsc><mode>_mask): New insn_and_split pattern.
6018 (*btr<mode>_mask): Ditto.
6020 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6021 Alan Hayward <alan.hayward@arm.com>
6022 David Sherwood <david.sherwood@arm.com>
6024 * function.c (pad_below): Simplify padding calculation.
6026 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6027 Alan Hayward <alan.hayward@arm.com>
6028 David Sherwood <david.sherwood@arm.com>
6030 * target.def (function_prologue): Remove frame size argument.
6031 (function_epilogue): Likewise.
6032 * doc/tm.texi: Regenerate.
6033 * final.c (final_start_function): Update call to function_prologue.
6034 (final_end_function): Update call to function_epilogue.
6035 (default_function_pro_epilogue): Remove frame size argument.
6036 * output.h (default_function_pro_epilogue): Likewise.
6037 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
6038 (arm_output_function_prologue): Likewise.
6039 * config/frv/frv.c (frv_function_prologue): Likewise.
6040 (frv_function_epilogue): Likewise.
6041 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
6042 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
6043 (ia64_output_function_epilogue): Likewise.
6044 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
6045 (m32r_output_function_epilogue): Likewise.
6046 * config/microblaze/microblaze.c (microblaze_function_prologue)
6047 (microblaze_function_epilogue): Likewise.
6048 * config/mips/mips.c (mips_output_function_prologue): Likewise.
6049 (mips_output_function_epilogue): Likewise.
6050 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
6051 (mmix_target_asm_function_epilogue): Likewise.
6052 * config/msp430/msp430.c (msp430_start_function): Likewise.
6053 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
6054 (nds32_asm_function_epilogue): Likewise.
6055 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
6056 * config/pa/pa.c (pa_output_function_prologue): Likewise.
6057 (pa_output_function_epilogue): Likewise.
6058 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
6059 (rs6000_output_function_epilogue): Likewise.
6060 * config/rl78/rl78.c (rl78_start_function): Likewise.
6061 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
6062 (rs6000_output_function_epilogue): Likewise.
6063 * config/rx/rx.c (rx_output_function_prologue): Likewise.
6064 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
6065 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
6066 (sparc_asm_function_epilogue): Likewise.
6068 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6070 * tree.h (type_has_mode_precision_p): New function.
6071 * convert.c (convert_to_integer_1): Use it.
6072 * expr.c (expand_expr_real_2): Likewise.
6073 (expand_expr_real_1): Likewise.
6074 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
6075 * match.pd: Likewise.
6076 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
6077 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
6078 * tree-tailcall.c (process_assignment): Likewise.
6079 * tree-vect-loop.c (vectorizable_reduction): Likewise.
6080 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
6081 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
6082 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6083 (vectorizable_assignment): Likewise.
6084 (vectorizable_shift): Likewise.
6085 (vectorizable_operation): Likewise.
6086 * tree-vrp.c (register_edge_assert_for_2): Likewise.
6088 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
6090 * match.pd: Add pow (C, x) simplification.
6092 2017-08-21 Richard Biener <rguenther@suse.de>
6094 PR tree-optimization/81900
6095 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
6096 for blocks with abnormal predecessors.
6097 (compute_antic): Do not set visited flag prematurely.
6099 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
6102 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
6104 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6106 * stor-layout.h (vector_type_mode): Move to...
6107 * tree.h (vector_type_mode): ...here.
6108 * stor-layout.c (vector_type_mode): Move to...
6109 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
6111 2017-08-21 Richard Biener <rguenther@suse.de>
6113 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
6114 register_external_die hooks.
6115 (debug_false_tree_charstarstar_uhwistar): Declare.
6116 (debug_nothing_tree_charstar_uhwi): Likewise.
6117 * debug.c (do_nothing_debug_hooks): Adjust.
6118 (debug_false_tree_charstarstar_uhwistar): New do nothing.
6119 (debug_nothing_tree_charstar_uhwi): Likewise.
6120 * dbxout.c (dbx_debug_hooks): Adjust.
6121 (xcoff_debug_hooks): Likewise.
6122 * sdbout.c (sdb_debug_hooks): Likewise.
6123 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
6124 * dwarf2out.c (macinfo_label_base): New global.
6125 (dwarf2out_register_external_die): New function for the
6126 register_external_die hook.
6127 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
6128 (dwarf2_debug_hooks): Use them.
6129 (dwarf2_lineno_debug_hooks): Adjust.
6130 (struct die_struct): Add with_offset flag.
6131 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
6132 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
6133 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
6134 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
6135 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
6136 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
6137 defining section names for the early LTO debug variants.
6138 (reset_indirect_string): New helper.
6139 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
6140 (print_dw_val): Add support for offsetted symbol references.
6141 (get_ultimate_context): Split out from is_cxx.
6142 (is_cxx): Use get_ultimate_context.
6143 (is_fortran): Add decl overload.
6144 (compute_comp_unit_symbol): Split out worker from
6145 compute_section_prefix.
6146 (compute_section_prefix): Call compute_comp_unit_symbol and
6147 set comdat_type_p here.
6148 (output_die): Skip DIE symbol output for the LTO added one.
6149 Handle DIE symbol references with offset.
6150 (output_comp_unit): Guard section name mangling properly.
6151 For LTO debug sections emit a symbol at the section beginning
6152 which we use to refer to its DIEs.
6153 (add_abstract_origin_attribute): For DIEs registered via
6154 dwarf2out_register_external_die directly refer to the early
6155 DIE rather than indirectly through the shadow one we created.
6156 Remove obsolete call to dwarf2out_abstract_function for
6157 non-function/block origins.
6158 (gen_array_type_die): When generating early LTO debug do
6159 not emit DW_AT_string_length.
6160 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
6161 late when in LTO. As suggested place a gcc_unreachable for
6162 the DECL_ABSTRACT_P case.
6163 (gen_subprogram_die): Avoid another specification DIE
6164 for early built declarations/definitions for the late LTO case.
6165 (gen_variable_die): Add type references for late duplicated VLA dies
6167 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
6168 we have the abstract instance already.
6169 (process_scope_var): Adjust decl DIE contexts in LTO which
6170 first puts them in limbo.
6171 (gen_decl_die): Do not generate type DIEs late apart from
6172 types for VLAs or for decls we do not yet have a DIE. Do not
6173 call dwarf2out_abstract_function late.
6174 (dwarf2out_early_global_decl): Make sure to create DIEs
6175 for abstract instances of a decl first.
6176 (dwarf2out_late_global_decl): Adjust comment.
6177 (output_macinfo_op): With multiple macro sections use
6178 macinfo_label_base to distinguish labels.
6179 (output_macinfo): Likewise. Update macinfo_label_base.
6180 Pass in the line info label.
6181 (note_variable_value_in_expr): When generating LTO resolve
6182 all variable values here by generating DIEs as needed.
6183 (init_sections_and_labels): Add early LTO debug flag parameter
6184 and generate different sections and names if set. Add generation
6185 counter for the labels so we can have multiple of them.
6186 (reset_dies): Helper to allow DIEs to be output multiple times.
6187 (dwarf2out_finish): When outputting DIEs to the fat part of an
6188 LTO object first reset DIEs.
6189 (dwarf2out_early_finish): Output early DIEs when generating LTO.
6190 (modified_type_die): Check for decl_ultimate_origin being self
6192 (gen_type_die_with_usage): Likewise.
6193 (gen_typedef_die): Allow decl_ultimate_origin being self.
6194 (set_decl_abstract_flags): Remove.
6195 (set_block_abstract_flags): Likewise.
6196 (dwarf2out_abstract_function): Treat the early generated DIEs
6197 as the abstract copy and only add DW_AT_inline and
6198 DW_AT_artificial here and call set_decl_origin_self.
6199 If the DIE has an abstract origin don't do anything.
6200 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
6201 if we have none yet (Go fails to build one, PR78628).
6202 (variably_modified_type_p): Prevent endless recursion for Ada
6203 cyclic pointer types.
6204 * lto-streamer-in.c: Include debug.h.
6205 (dref_queue): New global.
6206 (lto_read_tree_1): Stream in DIE references.
6207 (lto_input_tree): Register DIE references.
6208 (input_function): Stream DECL_DEBUG_ARGS.
6209 * lto-streamer-out.c: Include debug.h.
6210 (lto_write_tree_1): Output DIE references.
6211 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
6212 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
6213 (output_function): Stream DECL_DEBUG_ARGS.
6214 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
6215 Stream DECL_ABSTRACT_ORIGIN.
6216 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
6217 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
6218 DECL_CONTEXT for file-scope decls.
6219 * lto-streamer.h (struct dref_entry): Declare.
6220 (dref_queue): Likewise.
6221 * cfgexpand.c (pass_expand::execute): Do not call the
6222 outlining_inline_function hook here.
6223 * lto-wrapper.c (debug_obj): New global.
6224 (tool_cleanup): Unlink it if required.
6225 (debug_objcopy): New function.
6226 (run_gcc): Handle early debug sections in the IL files by
6227 extracting them to separate files, partially linkin them and
6228 feeding the result back as result to the linker.
6229 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
6230 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
6231 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
6232 sections into a separate segment.
6233 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
6235 (darwin_asm_dwarf_section): Likewise.
6236 (darwin_asm_output_dwarf_offset): Likewise.
6237 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
6239 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6240 Alan Hayward <alan.hayward@arm.com>
6241 David Sherwood <david.sherwood@arm.com>
6243 * read-md.h (md_reader::record_potential_iterator_use): Replace
6244 pointer argument with an rtx and an index.
6245 * read-rtl.c (iterator_group::apply_iterator): Likewise.
6246 (apply_mode_iterator): Likewise.
6247 (apply_code_iterator): Likewise.
6248 (apply_int_iterator): Likewise.
6249 (apply_subst_iterator): Likewise.
6250 (record_iterator_use): Likewise.
6251 (record_attribute_use): Likewise.
6252 (md_reader::record_potential_iterator_use): Likewise. Update calls
6253 to record_iterator_use and apply_iterator.
6254 (iterator_use): Replace ptr with x and index.
6255 (attribute_use): Likewise.
6256 (apply_attribute_uses): Update calls to apply_iterator.
6257 (apply_iterators): Likewise. Update initialization of iterator_use.
6258 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
6259 and record_potential_iterator_use.
6260 (rtx_reader::read_rtx_operand): Likewise.
6262 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6263 Alan Hayward <alan.hayward@arm.com>
6264 David Sherwood <david.sherwood@arm.com>
6266 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
6269 2017-08-21 Richard Biener <rguenther@suse.de>
6272 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
6273 at struct end conservatively when comparing common bases.
6275 2017-08-21 Richard Biener <rguenther@suse.de>
6277 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
6278 (mem_ref_in_stmt): Remove.
6279 (determine_max_movement): Use ref index to get at the reference.
6280 (invariantness_dom_walker::before_dom_children): Deal with
6281 lim data already initialized.
6282 (gather_mem_refs_stmt): Initialize lim data and record ref index.
6284 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
6286 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
6287 (TARGET_ISA_ROUND): Ditto.
6288 (TARGET_ROUND): Ditto.
6289 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
6290 * config/i386/i386.md: Ditto.
6291 * config/i386/sse.md: Ditto.
6292 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
6293 with OPTION_MASK_ISA_SSE4_1.
6295 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
6298 * doc/extend.texi (x86 Built-in Functions): Correct the name of
6299 __builtin_ia32_lzcnt_u16.
6301 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
6304 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
6305 (rs6000_set_current_function): Rewrite function to use it.
6307 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
6310 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
6311 and TYPE_WARN_IF_NOT_ALIGN.
6312 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
6313 (handle_warn_if_not_align): New.
6314 (place_union_field): Call handle_warn_if_not_align.
6315 (place_field): Call handle_warn_if_not_align.
6316 Copy TYPE_WARN_IF_NOT_ALIGN.
6317 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
6318 (layout_type): Likewise.
6319 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
6321 (tree_decl_common): Add warn_if_not_align.
6322 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
6323 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
6324 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
6325 (DECL_WARN_IF_NOT_ALIGN): Likewise.
6326 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
6327 * doc/extend.texi: Document warn_if_not_aligned attribute.
6328 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
6330 2017-08-17 Martin Liska <mliska@suse.cz>
6333 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
6334 (get_data_dependence): Use it as pointer type.
6335 (distribute_loop): Likewise.
6337 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6339 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
6340 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
6341 (p8_vmrgow_<mode>_direct): New define_insn.
6342 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
6343 handle endianness for vmrgew and vmrgow permute patterns.
6345 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
6347 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
6348 * config/rs6000/rs6000-cpus.def: Remove comment.
6349 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
6350 (POWERPC_MASKS): Likewise.
6351 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
6352 use of TARGET_VSX_TIMODE.
6353 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
6354 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
6355 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
6356 (rs6000_option_override_internal): Remove dead code.
6357 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
6358 (rs6000_legitimize_reload_address): Likewise.
6359 (rs6000_legitimate_address_p): Likewise.
6360 (rs6000_opt_masks): Delete "vsx-timode".
6361 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
6362 from function comment.
6363 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
6364 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
6365 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
6367 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
6368 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
6369 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
6372 (define_peephole2): Likewise.
6374 2017-08-17 Martin Sebor <msebor@redhat.com>
6377 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
6378 past the end of an array.
6379 (test_pp_format): Add test cases.
6381 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
6383 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
6384 missing ECF_NOTHROW flags.
6386 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
6389 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
6390 operands residing in integer registers.
6391 (*vsx_le_perm_load_<mode>): Likewise.
6392 (*vsx_le_perm_store_<mode>): Likewise.
6393 (define_peephole2): Add peepholes to optimize the above.
6395 2017-08-17 Marek Polacek <polacek@redhat.com>
6398 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
6399 to mimic what shorten_compare did. Change the return type to bool.
6400 (fold_cond_expr_with_comparison): Update call to
6401 operand_equal_for_comparison_p.
6402 (fold_ternary_loc): Likewise.
6404 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
6406 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
6408 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
6410 2017-08-17 Richard Biener <rguenther@suse.de>
6412 * tree-ssa-structalias.c (solve_graph): When propagating
6413 to successors update the graphs succ edges and avoid duplicate work.
6415 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
6418 * config/i386/i386.c (ix86_option_override_internal): Save target
6419 specific options after ix86_stack_protector_guard_reg was changed.
6421 2017-08-17 Richard Biener <rguenther@suse.de>
6423 PR tree-optimization/81827
6424 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
6425 (new_var_info): Initialize it conservatively.
6426 (get_call_vi): Mark register vars.
6427 (new_scalar_tmp_constraint_exp): Likewise.
6428 (handle_rhs_call): Likewise.
6429 (handle_const_call): Likewise.
6430 (create_function_info_for): Likewise.
6431 (solve_constraints): Sort varinfos to separate register from
6432 non-register vars to pack points-to solution bitmaps during
6435 2017-08-17 Marek Polacek <polacek@redhat.com>
6437 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
6439 2017-08-17 Richard Biener <rguenther@suse.de>
6441 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
6442 to true when overflow is undefined and we saturated the result.
6444 2017-08-17 Alan Modra <amodra@gmail.com>
6447 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
6448 Don't use store multiple if only one reg needs saving.
6449 (interesting_frame_related_regno): New function.
6450 (rs6000_frame_related): Don't emit frame info for regs that
6452 (rs6000_emit_epilogue): Likewise.
6454 2017-08-16 Nathan Sidwell <nathan@acm.org>
6456 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
6457 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
6458 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
6459 * tree.c (free_lang_data_in_type): Use else-if chain. Always
6460 clear TYPE_LANG_1. Remove obsolete member-function stripping.
6461 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
6462 (verify_type): Adjust for TYPE_BINFO move.
6463 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
6464 process TYPE_BINFO directly.
6465 (hash_tree): Likewise.
6466 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
6468 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
6471 2017-08-16 David Malcolm <dmalcolm@redhat.com>
6473 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
6475 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
6478 * config/i386/i386.md (*anddi_1_btr): Change predicates of
6479 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
6480 Add ix86_binary_operator_ok to insn constraint.
6481 (*iordi_1_bts): Ditto.
6482 (*xordi_1_btc): Ditto.
6483 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
6484 Update corresponding peephole2 pattern.
6488 2017-08-16 Bin Cheng <bin.cheng@arm.com>
6490 PR tree-optimization/81832
6491 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
6492 copy loop header which has IFN_LOOP_DIST_ALIAS call.
6494 2017-08-16 Marek Polacek <polacek@redhat.com>
6497 * fold-const.c (fold_indirect_ref_1): Restore original behavior
6498 regarding size_zero_node.
6500 2017-08-16 Martin Liska <mliska@suse.cz>
6503 * config.gcc: Respect previously set extra_objs in case
6506 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
6508 PR tree-optimization/81835
6509 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
6510 the phi SSA_NAME. Check that the condition in a COND_EXPR does
6511 not depend on the phi.
6513 2017-08-16 Alan Modra <amodra@gmail.com>
6515 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
6518 2017-08-16 Alan Modra <amodra@gmail.com>
6520 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
6521 (save_reg_p): ..into this. Update all callers.
6522 (first_reg_to_save): Simplify.
6524 2017-08-16 Alan Modra <amodra@gmail.com>
6526 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
6529 2017-08-15 Joseph Myers <joseph@codesourcery.com>
6533 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
6534 constant count if that count is less than 32.
6536 2017-08-15 Nathan Sidwell <nathan@acm.org>
6538 * gcc.c (execute): Emit friendlier message if inferior is killed
6539 by an external cause.
6541 2017-08-15 Richard Biener <rguenther@suse.de>
6543 PR tree-optimization/81790
6544 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
6545 CONSTRUCTORs from simplifying and VN.
6547 2017-08-14 Martin Sebor <msebor@redhat.com>
6549 * builtin-attrs.def: Add comments.
6551 2017-08-14 Martin Sebor <msebor@redhat.com>
6554 * doc/extend.texi (attribute nonstring): Document new attribute.
6556 2017-08-14 Martin Sebor <msebor@redhat.com>
6559 * tree-diagnostic.c (default_tree_printer): Handle %G.
6560 * gimple-pretty-print.h (percent_G_format): Declare new function.
6561 * gimple-pretty-print.c (percent_G_format): Define.
6562 * tree-pretty-print.c (percent_K_format): Add argument.
6564 2017-08-14 Martin Sebor <msebor@redhat.com>
6566 PR translation/79998
6567 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
6568 Remove a stray space.
6570 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
6573 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
6574 (*iordi_1_bts): Ditto.
6575 (*xordi_1_btc): Ditto.
6577 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6580 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
6581 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6583 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
6584 quoted strings, and make more translator-friendly.
6585 (darwin_rs6000_override_options): Likewise.
6586 (rs6000_option_override_internal): Likewise.
6587 (rs6000_return_in_memory): Fix overlong line.
6588 (init_cmulative_args): Use quoted strings, and make more
6589 translator-friendly.
6590 (rs6000_pass_by_reference): Fix overlong line.
6591 (def_builtin): Use quoted strings.
6592 (altivec_expand_predicate_builtin): Use quoted strings, and make
6593 more translator-friendly.
6594 (htm_expand_builtin): Use quoted strings.
6595 (cpu_expand_builtin): Use quoted strings, and make more
6596 translator-friendly.
6597 (altivec_expand_builtin): Likewise.
6598 (paired_expand_predicate_builtin): Likewise.
6599 (rs6000_invalid_builtin): Likewise.
6600 (builtin_function_type): Use quoted strings.
6601 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
6602 more translator-friendly.
6603 (rs6000_trampoline_init): Likewise.
6604 (rs6000_handle_altivec_attribute): Likewise.
6605 (rs6000_inner_target_options): Use quoted strings.
6606 (rs6000_disable_incompatible_switches): Likewise.
6607 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
6608 strings, and make more translator-friendly.
6609 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
6611 2017-08-14 Bin Cheng <bin.cheng@arm.com>
6613 PR tree-optimization/81799
6614 * tree-loop-distribution.c (version_loop_by_alias_check): Force
6615 cond_expr to simple gimple operand.
6617 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
6620 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
6622 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
6626 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
6627 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
6628 TARGET_GASISR_PROLOGUES.
6629 * config/avr/avr.c (avr_option_override): Same.
6630 (avr_pass_pre_proep::execute): Same.
6632 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
6635 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
6636 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
6638 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
6640 * config/i386/i386.md (*load_tp_<mode>): Redefine as
6641 define_insn_and_split. Split to a memory load from 0 in
6642 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
6643 using PTR mode iterator.
6644 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
6645 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
6646 (*add_tp_<mode>): Redefine as define_insn_and_split.
6647 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
6648 address space. Merge with *add_tp_x32 using PTR mode iterator.
6649 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
6650 Split to an add with a memory load from 0 in
6651 DEFAULT_TLS_SEG_REG address space.
6653 2017-08-12 Andrew Pinski <apinski@cavium.com>
6655 * config/aarch64/aarch64-option-extensions.def (rdma):
6656 Fix feature string to what Linux prints out in /proc/cpuinfo.
6658 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
6661 * dwarf2out.c (modified_type_die): For C typedef types that have
6662 an ultimate origin, process the ultimate origin instead of the
6664 (gen_typedef_die): Assert that input DECLs have no ultimate
6666 (gen_type_die_with_usage): For typedef variants that have an
6667 ultimate origin, just call gen_decl_die on the original DECL.
6668 (process_scope_var): Avoid creating DIEs for local typedefs and
6669 concrete static variables.
6671 2017-08-12 Alan Modra <amodra@gmail.com>
6675 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
6676 match gnu-user.h startfile.
6677 (ENDFILE_LINUX_SPEC): Similarly.
6679 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
6682 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
6684 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
6686 2017-08-11 Tamar Christina <tamar.christina@arm.com>
6687 * config/aarch64/aarch64.md (mov<mode>): Change.
6688 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
6689 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
6690 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
6692 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
6694 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
6695 for storage order barriers.
6697 2017-08-11 Martin Liska <mliska@suse.cz>
6700 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
6701 variables of void type.
6703 2017-08-11 Martin Liska <mliska@suse.cz>
6705 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
6706 TARGET_SUPPORTS_ALIASES.
6707 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
6708 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
6709 (optimize_weakref): Likewise.
6710 * symtab.c (symtab_node::noninterposable_alias): Likewise.
6711 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
6712 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
6714 2017-08-11 Martin Liska <mliska@suse.cz>
6717 * config/i386/i386.c (make_resolver_func): Do complete
6718 refactoring of the function.
6720 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
6723 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
6724 * config/i386/i386.c (ix86_stack_protect_guard): Use
6725 ix86_stack_protect_guard_symbol_str to generate varible declaration.
6726 * doc/invoke.texi (x86 Options): Document
6727 -mstack-protector-guard-symbol= option.
6729 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
6731 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
6732 * config/i386/i386.c (ix86_split_stack_guard): New function.
6733 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
6734 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
6735 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
6736 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
6737 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
6738 (split_stack_space_check): Call ix86_split_stack_guard.
6740 2017-08-10 Martin Sebor <msebor@redhat.com>
6742 * print-tree.c (print_node): Print location using the established
6744 Replace spaces with colons.
6745 (debug_raw, debug): Ditto.
6747 2017-08-10 Martin Sebor <msebor@redhat.com>
6750 * pretty-print.c (pp_format): Correct the handling of %s precision.
6752 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
6755 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
6757 (ix86_finalize_stack_frame_flags): This. Also clear
6758 frame_pointer_needed if -fno-omit-frame-pointer is used without
6760 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
6761 with ix86_finalize_stack_frame_flags.
6762 (ix86_expand_epilogue): Likewise.
6763 (ix86_expand_split_stack_prologue): Likewise.
6764 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
6766 2017-08-10 Martin Liska <mliska@suse.cz>
6769 * c-attribs.c (handle_target_attribute):
6770 Report warning for an empty string argument of target attribute.
6772 2017-08-09 Jakub Jelinek <jakub@redhat.com>
6775 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
6777 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
6778 or DECL_NONLOCAL labels.
6779 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
6780 or DECL_NONLOCAL labels here.
6782 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
6784 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
6785 to indicate when early gimple folding has been disabled.
6786 (rs6000_gimple_fold_builtin): Add debug content.
6787 (rs6000_invalid_builtin): Fix whitespace.
6788 (rs6000_expand_builtin): Fix whitespace.
6789 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
6791 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
6794 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
6795 SAVE_MULTIPLE if not all the registers that saves, should be saved.
6797 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
6799 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
6800 (qdf24xx): Likewise.
6801 * config/aarch64/aarch64.md: Include falkor.md.
6802 * config/aarch64/falkor.md: New.
6804 2017-08-09 Marek Polacek <polacek@redhat.com>
6807 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
6808 * diagnostic.c (emit_diagnostic): Add a comment.
6809 (emit_diagnostic_valist): New function.
6811 2017-08-09 Marek Polacek <polacek@redhat.com>
6814 * input.c (make_location): New overload.
6815 * input.h (make_location): Declare.
6817 2017-08-08 Alan Modra <amodra@gmail.com>
6818 H.J. Lu <hongjiu.lu@intel.com>
6821 * gcc.c (NO_PIE_SPEC): Delete.
6822 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
6824 (LINK_PIE_SPEC): ..to here.
6825 (LINK_COMMAND_SPEC): Support -no-pie.
6826 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
6827 chain of crtbegin*.o selection, update for PIE_SPEC changes and
6829 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
6830 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
6831 (ENDFILE_CRTEND_SPEC): Similarly.
6833 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
6836 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
6837 (mstack-protector-guard-offset=): Ditto.
6838 * config/i386/i386.c (ix86_option_override): Handle
6839 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
6841 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
6842 ix86_stack_protect_guard_offset variables.
6843 (TARGET_STACK_PROTECT_GUARD): Always define.
6844 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
6845 and -mstack-protector-guard-offset= options.
6847 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6849 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
6850 boundary case for the last candidate.
6852 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6854 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
6855 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
6857 2017-08-08 Tamar Christina <tamar.christina@arm.com>
6860 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
6861 * config/aarch64/aarch64-builtins.c
6862 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
6863 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
6864 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
6866 2017-08-08 Tamar Christina <tamar.christina@arm.com>
6867 Andrew Pinski <pinskia@gmail.com>
6870 * internal-fn.def (XORSIGN): New.
6871 * optabs.def (xorsign_optab): New.
6872 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
6873 (convert_expand_mult_copysign): New.
6874 (pass_optimize_widening_mul::execute): Call
6875 convert_expand_mult_copysign.
6877 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6879 PR tree-optimization/81354
6880 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
6881 Insert on edges rather than explicitly creating landing pads.
6882 (analyze_candidates_and_replace): Commit edge inserts.
6884 2017-08-08 Richard Biener <rguenther@suse.de>
6887 * tree-ssa-loop-niter.c: Include tree-dfa.h.
6888 (expand_simple_operations): Also look through ADDR_EXPRs with
6889 MEM_REF bases treating them as POINTER_PLUS_EXPR.
6891 2017-08-08 Richard Biener <rguenther@suse.de>
6893 PR tree-optimization/81723
6894 * tree-vect-slp.c (struct bst_traits): New hash traits.
6895 (bst_fail): New global.
6896 (vect_build_slp_tree_2): New worker, split out from ...
6897 (vect_build_slp_tree): ... this now wrapping it with using
6898 bst_fail set to cache SLP tree build fails. Properly handle
6900 (vect_analyze_slp_instance): Allocate and free bst_fail.
6902 2017-08-08 Martin Liska <mliska@suse.cz>
6905 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
6906 LABEL_DECLs that can be from a different function.
6908 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6910 PR tree-optimization/81744
6911 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
6912 loop's number of iterations.
6914 2017-08-08 Martin Liska <mliska@suse.cz>
6916 * asan.c: Include header files.
6917 * attribs.c (build_decl_attribute_variant): New function moved
6919 (build_type_attribute_qual_variant): Likewise.
6920 (cmp_attrib_identifiers): Likewise.
6921 (simple_cst_list_equal): Likewise.
6922 (omp_declare_simd_clauses_equal): Likewise.
6923 (attribute_value_equal): Likewise.
6924 (comp_type_attributes): Likewise.
6925 (build_type_attribute_variant): Likewise.
6926 (lookup_ident_attribute): Likewise.
6927 (remove_attribute): Likewise.
6928 (merge_attributes): Likewise.
6929 (merge_type_attributes): Likewise.
6930 (merge_decl_attributes): Likewise.
6931 (merge_dllimport_decl_attributes): Likewise.
6932 (handle_dll_attribute): Likewise.
6933 (attribute_list_equal): Likewise.
6934 (attribute_list_contained): Likewise.
6935 * attribs.h (lookup_attribute): New function moved from tree.[ch].
6936 (lookup_attribute_by_prefix): Likewise.
6937 * bb-reorder.c: Include header files.
6938 * builtins.c: Likewise.
6939 * calls.c: Likewise.
6940 * cfgexpand.c: Likewise.
6941 * cgraph.c: Likewise.
6942 * cgraphunit.c: Likewise.
6943 * convert.c: Likewise.
6944 * dwarf2out.c: Likewise.
6945 * final.c: Likewise.
6946 * fold-const.c: Likewise.
6947 * function.c: Likewise.
6948 * gimple-expr.c: Likewise.
6949 * gimple-fold.c: Likewise.
6950 * gimple-pretty-print.c: Likewise.
6951 * gimple.c: Likewise.
6952 * gimplify.c: Likewise.
6953 * hsa-common.c: Likewise.
6954 * hsa-gen.c: Likewise.
6955 * internal-fn.c: Likewise.
6956 * ipa-chkp.c: Likewise.
6957 * ipa-cp.c: Likewise.
6958 * ipa-devirt.c: Likewise.
6959 * ipa-fnsummary.c: Likewise.
6960 * ipa-inline.c: Likewise.
6961 * ipa-visibility.c: Likewise.
6963 * lto-cgraph.c: Likewise.
6964 * omp-expand.c: Likewise.
6965 * omp-general.c: Likewise.
6966 * omp-low.c: Likewise.
6967 * omp-offload.c: Likewise.
6968 * omp-simd-clone.c: Likewise.
6969 * opts-global.c: Likewise.
6970 * passes.c: Likewise.
6971 * predict.c: Likewise.
6972 * sancov.c: Likewise.
6973 * sanopt.c: Likewise.
6974 * symtab.c: Likewise.
6975 * toplev.c: Likewise.
6976 * trans-mem.c: Likewise.
6977 * tree-chkp.c: Likewise.
6978 * tree-eh.c: Likewise.
6979 * tree-into-ssa.c: Likewise.
6980 * tree-object-size.c: Likewise.
6981 * tree-parloops.c: Likewise.
6982 * tree-profile.c: Likewise.
6983 * tree-ssa-ccp.c: Likewise.
6984 * tree-ssa-live.c: Likewise.
6985 * tree-ssa-loop.c: Likewise.
6986 * tree-ssa-sccvn.c: Likewise.
6987 * tree-ssa-structalias.c: Likewise.
6988 * tree-ssa.c: Likewise.
6989 * tree-streamer-in.c: Likewise.
6990 * tree-vectorizer.c: Likewise.
6991 * tree-vrp.c: Likewise.
6993 * ubsan.c: Likewise.
6994 * varasm.c: Likewise.
6995 * varpool.c: Likewise.
6996 * tree.c: Remove functions moved to attribs.[ch].
6998 * config/aarch64/aarch64.c: Add attrs.h header file.
6999 * config/alpha/alpha.c: Likewise.
7000 * config/arc/arc.c: Likewise.
7001 * config/arm/arm.c: Likewise.
7002 * config/avr/avr.c: Likewise.
7003 * config/bfin/bfin.c: Likewise.
7004 * config/c6x/c6x.c: Likewise.
7005 * config/cr16/cr16.c: Likewise.
7006 * config/cris/cris.c: Likewise.
7007 * config/darwin.c: Likewise.
7008 * config/epiphany/epiphany.c: Likewise.
7009 * config/fr30/fr30.c: Likewise.
7010 * config/frv/frv.c: Likewise.
7011 * config/ft32/ft32.c: Likewise.
7012 * config/h8300/h8300.c: Likewise.
7013 * config/i386/winnt.c: Likewise.
7014 * config/ia64/ia64.c: Likewise.
7015 * config/iq2000/iq2000.c: Likewise.
7016 * config/lm32/lm32.c: Likewise.
7017 * config/m32c/m32c.c: Likewise.
7018 * config/m32r/m32r.c: Likewise.
7019 * config/m68k/m68k.c: Likewise.
7020 * config/mcore/mcore.c: Likewise.
7021 * config/microblaze/microblaze.c: Likewise.
7022 * config/mips/mips.c: Likewise.
7023 * config/mmix/mmix.c: Likewise.
7024 * config/mn10300/mn10300.c: Likewise.
7025 * config/moxie/moxie.c: Likewise.
7026 * config/msp430/msp430.c: Likewise.
7027 * config/nds32/nds32-isr.c: Likewise.
7028 * config/nds32/nds32.c: Likewise.
7029 * config/nios2/nios2.c: Likewise.
7030 * config/nvptx/nvptx.c: Likewise.
7031 * config/pa/pa.c: Likewise.
7032 * config/pdp11/pdp11.c: Likewise.
7033 * config/powerpcspe/powerpcspe.c: Likewise.
7034 * config/riscv/riscv.c: Likewise.
7035 * config/rl78/rl78.c: Likewise.
7036 * config/rx/rx.c: Likewise.
7037 * config/s390/s390.c: Likewise.
7038 * config/sh/sh.c: Likewise.
7039 * config/sol2.c: Likewise.
7040 * config/sparc/sparc.c: Likewise.
7041 * config/spu/spu.c: Likewise.
7042 * config/stormy16/stormy16.c: Likewise.
7043 * config/tilegx/tilegx.c: Likewise.
7044 * config/tilepro/tilepro.c: Likewise.
7045 * config/v850/v850.c: Likewise.
7046 * config/vax/vax.c: Likewise.
7047 * config/visium/visium.c: Likewise.
7048 * config/xtensa/xtensa.c: Likewise.
7050 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
7053 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
7054 constraints since the -mupper-regs-* switches have been
7056 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
7057 into a vector from a double word element that was extracted from
7058 another vector, and eliminate extra XXPERMDI instructions.
7059 (vsx_concat_<mode>_2): Likewise.
7060 (vsx_concat_<mode>_3): Likewise.
7061 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
7062 concat to allow optimizing inserts from previous extracts.
7064 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
7066 * config/i386/i386.c (ix86_stack_protect_guard): Generate
7067 memory reference to a SSP offset in TLS address space.
7068 (ix86_print_operand) <case '@'>: Remove.
7069 (ix86_print_operand_punct_valid_p): Remove '@' code.
7070 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
7072 (stack_tls_protect_set_<mode>): Remove.
7073 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
7074 (stack_tls_protect_test_<mode>): Remove.
7075 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
7077 2017-08-07 Olivier Hainque <hainque@adacore.com>
7080 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
7082 2017-08-07 Douglas Rupp <rupp@adacore.com>
7084 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
7085 variable was referenced as multidir in command.
7087 2017-08-07 Jakub Jelinek <jakub@redhat.com>
7090 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
7093 2017-08-07 Martin Liska <mliska@suse.cz>
7095 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
7096 * config/rl78/rl78.c: Add include of attribs.h.
7097 * config/sh/sh.c: Likewise.
7098 * config/v850/v850.c: Likewise.
7100 2017-08-07 Tom de Vries <tom@codesourcery.com>
7103 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
7105 2017-08-07 Martin Liska <mliska@suse.cz>
7107 * config/mips/mips.c: Include attribs.h.
7109 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
7112 * doc/invoke.texi: Document change in behvaior for -Ofast for
7115 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
7117 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
7119 (aarch64_pop_regs): Likewise.
7120 (aarch64_gen_load_pair): Likewise.
7121 (aarch64_save_callee_saves): Likewise.
7122 (aarch64_restore_callee_saves): Likewise.
7124 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
7126 * config/i386/i386.c: Revert the last change.
7128 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
7131 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
7133 (ix86_finalize_stack_frame_flags): This. Also clear
7134 frame_pointer_needed if -fno-omit-frame-pointer is used without
7136 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
7137 with ix86_finalize_stack_frame_flags.
7138 (ix86_expand_epilogue): Likewise.
7139 (ix86_expand_split_stack_prologue): Likewise.
7141 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
7144 * config/i386/i386.c (get_builtin_code_for_version): Set priority
7145 to P_AES for Westmere.
7147 2017-08-07 Jonathan Yong <10walls@gmail.com>
7149 * config/i386/mingw.opt (fset-stack-executable): Removed.
7150 * config/i386/cygming.opt (fset-stack-executable): Moved
7152 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
7154 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
7156 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
7158 2017-08-07 Marek Polacek <polacek@redhat.com>
7161 * fold-const.c (fold_indirect_ref_1): Check type_domain.
7163 2017-08-07 Martin Liska <mliska@suse.cz>
7165 * attribs.h (canonicalize_attr_name): New function.
7166 (cmp_attribs): Move from c-format.c and adjusted.
7167 (is_attribute_p): Moved from tree.h.
7168 * tree-inline.c: Add new includes.
7169 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
7170 (private_is_attribute_p): Remove.
7171 (private_lookup_attribute): Likewise.
7172 (private_lookup_attribute_by_prefix): Simplify.
7173 (remove_attribute): Use is_attribute_p.
7174 * tree.h: Remove removed declarations.
7176 2017-08-07 Jakub Jelinek <jakub@redhat.com>
7179 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
7180 instead of computing it in the function. Formatting fix.
7181 (expand_case): Don't rely on default_edge being the first edge,
7182 clear it if removing it, pass default_edge to
7183 emit_case_dispatch_table.
7184 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
7187 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
7189 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
7190 insn in the function, emit NOP after the insn.
7192 2017-08-06 Tom de Vries <tom@codesourcery.com>
7194 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
7197 2017-08-06 Tom de Vries <tom@codesourcery.com>
7199 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
7202 2017-08-04 Yury Gribov <tetra2005@gmail.com>
7204 PR tree-optimization/57371
7205 * match.pd: New pattern.
7207 2017-08-04 Marek Polacek <polacek@redhat.com>
7210 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
7211 perform the computation in offset_int.
7213 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7215 PR tree-optimization/81136
7216 * tree-vectorizer.h: Include tree-hash-traits.h.
7217 (vec_base_alignments): New typedef.
7218 (vec_info): Add a base_alignments field.
7219 (vect_record_base_alignments): Declare.
7220 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
7222 (DR_IS_CONDITIONAL_IN_STMT): New macro.
7223 (create_data_ref): Add an is_conditional_in_stmt argument.
7224 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
7225 the is_conditional_in_stmt field.
7226 (data_ref_loc): Add an is_conditional_in_stmt field.
7227 (get_references_in_stmt): Set the is_conditional_in_stmt field.
7228 (find_data_references_in_stmt): Update call to create_data_ref.
7229 (graphite_find_data_references_in_stmt): Likewise.
7230 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
7231 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
7232 (vect_record_base_alignment): New function.
7233 (vect_record_base_alignments): Likewise.
7234 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
7235 for nested statements even if we fail to compute a misalignment.
7236 Use pooled base alignments for unconditional references.
7237 (vect_find_same_alignment_drs): Compare base addresses instead
7239 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
7240 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
7242 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7244 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
7245 Add an explicit name for the enum. Use auto_vec for slp_instances
7247 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
7249 (_bb_vec_info): Add a constructor and destructor.
7250 (vinfo_for_stmt): Return NULL for uids of -1 as well.
7251 (destroy_loop_vec_info): Delete.
7252 (vect_destroy_datarefs): Likewise.
7253 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
7254 (vec_info::vec_info): New function.
7255 (vec_info::~vec_info): Likewise.
7256 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
7257 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
7258 destroy_loop_vec_info.
7259 * tree-vect-loop.c (new_loop_vec_info): Replace with...
7260 (_loop_vec_info::_loop_vec_info): ...this.
7261 (destroy_loop_vec_info): Replace with...
7262 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
7263 the stmt_vec_infos. Leave handling of vec_info information to its
7264 destructor. Remove explicit vector releases.
7265 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
7266 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
7267 * tree-vect-slp.c (new_bb_vec_info): Replace with...
7268 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
7269 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
7270 (destroy_bb_vec_info): Replace with...
7271 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
7272 information to its destructor.
7273 (vect_slp_analyze_bb_1): Use new and delete instead of
7274 new_bb_vec_info and destroy_bb_vec_info.
7275 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
7278 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7280 * tree-data-ref.h (subscript): Add access_fn field.
7281 (data_dependence_relation): Add could_be_independent_p.
7282 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
7283 (same_access_functions): Move to tree-data-ref.c.
7284 * tree-data-ref.c (ref_contains_union_access_p): New function.
7285 (access_fn_component_p): Likewise.
7286 (access_fn_components_comparable_p): Likewise.
7287 (dr_analyze_indices): Add a reference to access_fn_component_p.
7288 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
7290 (constant_access_functions): Likewise.
7291 (add_other_self_distances): Likewise.
7292 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
7293 (initialize_data_dependence_relation): Use XCNEW and remove
7294 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
7295 of access functions that have the same type. Allow the
7296 subsequence to end with different bases in some circumstances.
7297 Record the chosen access functions in SUB_ACCESS_FN.
7298 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
7299 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
7300 (subscript_dependence_tester_1): Likewise dra and drb.
7301 (build_classic_dist_vector): Update calls accordingly.
7302 (subscript_dependence_tester): Likewise.
7303 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
7304 DDR_COULD_BE_INDEPENDENT_P.
7305 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
7306 comp_alias_ddrs instead of may_alias_ddrs.
7307 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
7309 (vect_analyze_data_ref_dependence): Use it if
7310 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
7311 distance vectors if that fails.
7312 (dependence_distance_ge_vf): New function.
7313 (vect_prune_runtime_alias_test_list): Use it. Don't clear
7314 LOOP_VINFO_MAY_ALIAS_DDRS.
7316 2017-08-04 Richard Biener <rguenther@suse.de>
7319 * fold-const.c (fold_binary_loc): Properly restrict
7320 minus_var0 && minus_var1 case when associating undefined overflow
7323 2017-08-04 Tom de Vries <tom@codesourcery.com>
7325 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
7327 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7329 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7330 Don't start diagnostic messages with a capital letter.
7331 * config/rs6000/rs6000.c (rs6000_option_override_internal):
7333 (rs6000_invalid_builtin): Likewise.
7334 (rs6000_trampoline_init): Likewise.
7336 2017-08-03 Jakub Jelinek <jakub@redhat.com>
7339 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
7340 after setting changeable df flags.
7342 2017-08-03 Richard Biener <rguenther@suse.de>
7344 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
7345 up if the use is in USE - X.
7347 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
7349 * toplev.c (dumpfile.h): New include.
7350 (internal_error_reentered): New static function. Use it...
7351 (internal_error_function): ...here to handle reentered internal_error.
7353 2017-08-03 Richard Biener <rguenther@suse.de>
7356 * fold-const.c (split_tree): Add minus_var and minus_con
7357 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
7358 here but always use minus_*.
7359 (associate_trees): Assert we never associate with MINUS_EXPR
7360 and NULL first operand. Do not recurse for PLUS_EXPR operands
7361 when associating as MINUS_EXPR either.
7362 (fold_binary_loc): Track minus_var and minus_con.
7364 2017-08-03 Tom de Vries <tom@codesourcery.com>
7367 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
7368 ACCEL_COMPILER, apply finish_options on
7369 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
7371 2017-08-03 Tom de Vries <tom@codesourcery.com>
7374 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
7375 function_entry_patch_area_size > 0.
7377 2017-08-03 Jakub Jelinek <jakub@redhat.com>
7380 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
7381 instead of 10??LU, perform unit multiplication in wide_int,
7382 don't change alloc_object_size_limit if the limit is larger
7385 PR tree-optimization/81655
7386 PR tree-optimization/81588
7387 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
7388 the case when ranges[i].low and high are 1 for unsigned type with
7392 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
7393 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
7395 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7397 * tree-vrp.h: Add include guard.
7399 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
7402 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
7403 (ud2): New insn pattern.
7404 * config/i386/i386.c (ix86_expand_epilogue):
7405 For naked functions, generate ud2 instead of trap insn.
7407 2017-08-02 Marek Polacek <polacek@redhat.com>
7410 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
7412 2017-08-02 Tom de Vries <tom@codesourcery.com>
7413 Cesar Philippidis <cesar@codesourcery.com>
7415 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
7416 Add missing edge probabilities.
7418 2017-08-02 Tamar Christina <tamar.christina@arm.com>
7420 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
7423 2017-08-02 Jakub Jelinek <jakub@redhat.com>
7426 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
7427 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
7428 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
7429 prologue_seq sequences - if any.
7431 2017-08-02 Richard Biener <rguenther@suse.de>
7433 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
7434 via vectors if supported, integer extracts via punning if supported
7435 or otherwise vector extracts.
7437 2017-08-02 Richard Biener <rguenther@suse.de>
7439 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
7441 (bitmap_insert_into_set): ... this.
7443 2017-08-02 Richard Biener <rguenther@suse.de>
7445 PR tree-optimization/81633
7447 2015-08-17 Alan Hayward <alan.hayward@arm.com>
7449 PR tree-optimization/71752
7450 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
7452 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
7454 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
7455 (machine_function::call_ms2sysv_pad_out): Remove field.
7456 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
7457 (ix86_compute_frame_layout): Likewise.
7459 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
7462 * config/i386/i386.c (ix86_set_func_type): Disallow naked
7463 attribute with interrupt attribute.
7465 2017-08-01 Andrew Pinski <apinski@cavium.com>
7467 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
7468 BIT_INSERT_EXPR's operand 1
7469 to see if the types precision matches.
7471 2017-08-01 Martin Liska <mliska@suse.cz>
7474 * builtins.c (expand_builtin_memcpy_args): Remove.
7475 (expand_builtin_memcpy): Call newly added function
7476 expand_builtin_memory_copy_args.
7477 (expand_builtin_memcpy_with_bounds): Likewise.
7478 (expand_builtin_mempcpy): Remove last argument.
7479 (expand_builtin_mempcpy_with_bounds): Likewise.
7480 (expand_builtin_memory_copy_args): New function created from
7481 expand_builtin_mempcpy_args with small modifications.
7482 (expand_builtin_mempcpy_args): Remove.
7483 (expand_builtin_stpcpy): Remove unused argument.
7484 (expand_builtin): Likewise.
7485 (expand_builtin_with_bounds): Likewise.
7487 2017-08-01 Martin Liska <mliska@suse.cz>
7490 Make mempcpy more optimal (PR middle-end/70140).
7492 2017-08-01 Jakub Jelinek <jakub@redhat.com>
7495 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
7496 __builtin_vec_cmpne verify both arguments are compatible vectors
7497 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
7498 verify arg1_type is a pointer or array type. For __builtin_vec_st,
7499 move computation of aligned to after checking the argument types.
7503 * config/rs6000/vsx.md (vextract_fp_from_shorth,
7504 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
7507 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
7508 Doug Rupp <rupp@adacore.com>
7509 Olivier Hainque <hainque@adacore.com>
7511 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
7512 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
7514 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
7515 for TARGET_OS_CPP_BUILTIN.
7516 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
7517 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
7519 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
7520 passing required abi options to the assembler for EABI configurations.
7521 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
7522 of .text.hot and .text.unlikely sections for kernel modules when
7523 using ARM style exceptions.
7524 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
7525 options. Add EXTRA_CC1_SPEC.
7526 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
7528 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
7530 (ARM_TARGET2_DWARF_FORMAT): Define.
7531 * config/arm/t-vxworks: Adjust multilib control to removal of the
7532 Diab command line options.
7534 2017-08-01 Martin Liska <mliska@suse.cz>
7536 PR gcov-profile/81561
7537 * gcov.c (unblock): Make unblocking safe as we need to preserve
7538 index correspondence of blocks and block_lists.
7540 2017-08-01 Richard Biener <rguenther@suse.de>
7542 PR tree-optimization/81181
7543 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7544 (compute_antic): ... end of iteration here.
7546 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
7548 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
7549 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
7550 (ftree-slp-vectorize): Likewise.
7551 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
7552 can no longer be set independent of flag_tree_loop_vectorize.
7553 * omp-general.c (emp_max_vf): Likewise.
7554 * opts.c (enable_fdo_optimizations): Remove references to
7555 flag_tree_vectorize, these are now implicit.
7556 (common_handle_option): Remove handling for OPT_ftree_vectorize,
7557 and leave it for the options machinery.
7559 2017-08-01 Martin Liska <mliska@suse.cz>
7562 * builtins.c (expand_builtin_memcpy_args): Remove.
7563 (expand_builtin_memcpy): Call newly added function
7564 expand_builtin_memory_copy_args.
7565 (expand_builtin_memcpy_with_bounds): Likewise.
7566 (expand_builtin_mempcpy): Remove last argument.
7567 (expand_builtin_mempcpy_with_bounds): Likewise.
7568 (expand_builtin_memory_copy_args): New function created from
7569 expand_builtin_mempcpy_args with small modifications.
7570 (expand_builtin_mempcpy_args): Remove.
7571 (expand_builtin_stpcpy): Remove unused argument.
7572 (expand_builtin): Likewise.
7573 (expand_builtin_with_bounds): Likewise.
7575 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
7578 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
7579 print "ds:" only for immediates in generic address space.
7581 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
7584 * config/i386/i386.c (ix86_funciton_naked): New prototype.
7585 (ix86_function_ok_for_sibcall): Return false for naked functions.
7587 2017-08-01 Richard Biener <rguenther@suse.de>
7589 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
7590 (compute_antic): Seed worklist with exit block predecessors.
7591 * cfganal.c (dfs_find_deadend): For a cycle return the source
7592 of the edge closing it.
7594 2017-08-01 Tamar Christina <tamar.christina@arm.com>
7596 * config/aarch64/aarch64.c
7597 (aarch64_can_const_movi_rtx_p): Move 0 check.
7599 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7601 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
7602 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
7604 * match.pd: Ditto in address comparison pattern.
7606 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7608 PR tree-optimization/81627
7609 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
7610 closed ssa form for store-store chain.
7612 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7614 PR tree-optimization/81620
7615 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
7616 for store-store chain.
7618 2017-08-01 Jakub Jelinek <jakub@redhat.com>
7620 PR tree-optimization/81588
7621 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
7622 ranges[i].in_p, invert comparison code ccode. For >/>=,
7623 swap rhs1 and rhs2 and comparison code unconditionally,
7624 for </<= don't do that. Don't swap rhs1/rhs2 again if
7625 ranges[i].in_p, instead invert comparison code ccode if
7626 opcode or oe->rank is BIT_IOR_EXPR.
7629 * optabs.def (vec_extract_optab, vec_init_optab): Change from
7630 a direct optab to conversion optab.
7631 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
7632 with GET_MODE_INNER as last argument instead of optab_handler.
7633 * expmed.c (extract_bit_field_1): Likewise. Use vector from
7634 vector extraction if possible and optab is available.
7635 * expr.c (store_constructor): Use convert_optab_handler instead
7636 of optab_handler. Use vector initialization from smaller
7637 vectors if possible and optab is available.
7638 * tree-vect-stmts.c (vectorizable_load): Likewise.
7639 * doc/md.texi (vec_extract, vec_init): Document that the optabs
7641 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
7642 of vec_init from half-sized vectors with the same element mode.
7643 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
7644 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
7645 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
7646 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
7647 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
7648 after mode in gen_vec_extract* calls.
7649 (vec_extract<mode>): Renamed to ...
7650 (vec_extract<mode><ssescalarmodelower>): ... this.
7651 (vec_extract<mode><ssehalfvecmodelower>): New expander.
7652 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
7653 element mode after mode in gen_vec_init* calls.
7654 (VEC_INIT_HALF_MODE): New mode iterator.
7655 (vec_init<mode>): Renamed to ...
7656 (vec_init<mode><ssescalarmodelower>): ... this.
7657 (vec_init<mode><ssehalfvecmodelower>): New expander.
7658 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
7659 (vec_extractv2sfsf): ... this.
7660 (vec_initv2sf): Renamed to ...
7661 (vec_initv2sfsf): ... this.
7662 (vec_extractv2si): Renamed to ...
7663 (vec_extractv2sisi): ... this.
7664 (vec_initv2si): Renamed to ...
7665 (vec_initv2sisi): ... this.
7666 (vec_extractv4hi): Renamed to ...
7667 (vec_extractv4hihi): ... this.
7668 (vec_initv4hi): Renamed to ...
7669 (vec_initv4hihi): ... this.
7670 (vec_extractv8qi): Renamed to ...
7671 (vec_extractv8qiqi): ... this.
7672 (vec_initv8qi): Renamed to ...
7673 (vec_initv8qiqi): ... this.
7674 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
7675 (vec_init<mode>): Renamed to ...
7676 (vec_init<mode><VEC_base_l>): ... this.
7677 (vec_extract<mode>): Renamed to ...
7678 (vec_extract<mode><VEC_base_l>): ... this.
7679 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
7680 (vec_initv2sfsf): ... this.
7681 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
7682 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
7683 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
7684 element mode after mode in gen_vec_init* calls.
7685 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
7686 (vec_init<mode><Vel>): ... this.
7687 (vec_extract<mode>): Renamed to ...
7688 (vec_extract<mode><Vel>): ... this.
7689 * config/aarch64/iterators.md (Vel): New mode attribute.
7690 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
7691 Add element mode after mode in gen_vec_extract* calls.
7692 * config/s390/vector.md (non_vec_l): New mode attribute.
7693 (vec_extract<mode>): Renamed to ...
7694 (vec_extract<mode><non_vec_l>): ... this.
7695 (vec_init<mode>): Renamed to ...
7696 (vec_init<mode><non_vec_l>): ... this.
7697 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
7698 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
7700 * config/arm/iterators.md (V_elem_l): New mode attribute.
7701 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
7702 (vec_extract<mode><V_elem_l>): ... this.
7703 (vec_extractv2di): Renamed to ...
7704 (vec_extractv2didi): ... this.
7705 (vec_init<mode>): Renamed to ...
7706 (vec_init<mode><V_elem_l>): ... this.
7707 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
7708 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
7709 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
7710 Add element mode after gen_vec_extract* calls.
7711 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
7712 (vec_init<mode><unitmode>): ... this.
7713 (vec_extract<mode>): Renamed to ...
7714 (vec_extract<mode><unitmode>): ... this.
7715 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
7716 (vec_init<mode><unitmode>): ... this.
7717 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
7718 (vec_initv2sfsf): ... this.
7719 (vec_extractv2sf): Renamed to ...
7720 (vec_extractv2sfsf): ... this.
7721 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
7722 Add element mode after gen_vec_extract* calls.
7723 * config/mips/mips.md (unitmode): New mode iterator.
7724 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
7725 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
7726 * config/spu/spu.md (inner_l): New mode attribute.
7727 (vec_init<mode>): Renamed to ...
7728 (vec_init<mode><inner_l>): ... this.
7729 (vec_extract<mode>): Renamed to ...
7730 (vec_extract<mode><inner_l>): ... this.
7731 * config/sparc/sparc.md (veltmode): New mode iterator.
7732 (vec_init<VMALL:mode>): Renamed to ...
7733 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
7734 * config/ia64/vect.md (vec_initv2si): Renamed to ...
7735 (vec_initv2sisi): ... this.
7736 (vec_initv2sf): Renamed to ...
7737 (vec_initv2sfsf): ... this.
7738 (vec_extractv2sf): Renamed to ...
7739 (vec_extractv2sfsf): ... this.
7740 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
7741 (vec_init<mode>): Renamed to ...
7742 (vec_init<mode><VEC_base_l>): ... this.
7743 (vec_extract<mode>): Renamed to ...
7744 (vec_extract<mode><VEC_base_l>): ... this.
7745 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
7746 (vec_initv2sfsf): ... this.
7747 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
7748 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
7749 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
7750 gen_vec_init* calls.
7752 2017-08-01 Richard Biener <rguenther@suse.de>
7754 PR tree-optimization/81297
7755 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
7756 TREE_OVERFLOW from INTEGER_CSTs.
7758 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
7760 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
7762 2017-07-31 Carl Love <cel@us.ibm.com>
7764 * config/rs6000/rs6000-c: Add support for built-in functions
7765 vector signed char vec_xl_be (signed long long, signed char *);
7766 vector unsigned char vec_xl_be (signed long long, unsigned char *);
7767 vector signed int vec_xl_be (signed long long, signed int *);
7768 vector unsigned int vec_xl_be (signed long long, unsigned int *);
7769 vector signed long long vec_xl_be (signed long long, signed long long *);
7770 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
7771 vector signed short vec_xl_be (signed long long, signed short *);
7772 vector unsigned short vec_xl_be (signed long long, unsigned short *);
7773 vector double vec_xl_be (signed long long, double *);
7774 vector float vec_xl_be (signed long long, float *);
7775 * config/rs6000/altivec.h (vec_xl_be): Add #define.
7776 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
7777 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
7779 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
7780 (altivec_expand_builtin): Add switch statement to call
7781 altivec_expand_xl_be for each builtin.
7782 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
7783 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
7784 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
7785 __builtin_vsx_le_be_v16qi.
7786 * doc/extend.texi: Update the built-in documentation file for the
7787 new built-in functions.
7789 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
7792 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
7794 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
7796 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7798 * config.gcc: Add z14.
7799 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
7800 CPU model numbers for z13s and z14.
7801 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
7803 * config/s390/s390-opts.h (enum processor_type): Rename
7804 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
7805 * config/s390/s390.c (processor_table): Add field for CPU name to
7806 be passed to Binutils.
7807 (s390_asm_output_machine_for_arch): Use the new field in
7808 processor_table for Binutils.
7809 (s390_expand_builtin): Replace arch12 with z14.
7810 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
7811 (s390_get_sched_attrmask): Likewise.
7812 (s390_get_unit_mask): Likewise.
7813 * config/s390/s390.opt: Add z14 to processor_type enum.
7815 2017-07-31 Martin Jambor <mjambor@suse.cz>
7818 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
7819 regardless of optimization level.
7821 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
7822 Martin Liska <mliska@suse.cz>
7824 * predict.def: Remove old comment and adjust probability.
7825 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
7828 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
7831 * config/i386/i386.c (ix86_function_naked): New function.
7832 (ix86_can_use_return_insn_p): Return false for naked functions.
7833 (ix86_expand_prologue): Skip prologue for naked functions.
7834 (ix86_expand_epilogue): Skip epilogue for naked functions
7835 and emit trap instruction.
7836 (ix86_warn_func_return): New function.
7837 (ix86_attribute_table): Add "naked" attribute specification.
7838 (TARGET_WARN_FUNC_RETURN): Define.
7839 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
7841 2017-07-31 Martin Liska <mliska@suse.cz>
7843 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
7844 (dump_gimple_bb_header): Always dump BB info.
7845 (pp_cfg_jump): Do not append info about BB when dumping a jump.
7847 2017-07-31 Martin Liska <mliska@suse.cz>
7850 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
7851 also with current_function_decl non-null equality.
7853 2017-07-31 Jakub Jelinek <jakub@redhat.com>
7856 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
7857 change type to the element type, instead add eltype variable and
7858 use it where we are interested in the element type.
7860 PR tree-optimization/81603
7861 * ipa-polymorphic-call.c
7862 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
7863 offset arithmetic in offset_int, bail out if the resulting bit offset
7864 doesn't fit into shwi.
7866 2017-07-31 Martin Liska <mliska@suse.cz>
7868 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
7869 (gimplify_save_expr): Fix comment.
7871 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
7874 * config/i386/i386.c (ix86_function_arg): Update arguments for
7876 (ix86_compute_frame_layout): Set the initial stack offset to
7877 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
7878 INCOMING_FRAME_SP_OFFSET.
7879 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
7880 stack before exception handler returns.
7881 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
7882 the 'ERROR_CODE' for exception handler.
7884 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
7886 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
7887 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
7888 (ASM_OUTPUT_REG_POP): Ditto.
7889 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
7890 instead of asm_fprintf to output pure string.
7892 2017-07-29 Jakub Jelinek <jakub@redhat.com>
7894 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
7895 to imported_module_or_decl hook.
7896 (debug_nothing_tree_tree_tree_bool): Remove.
7897 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
7898 * debug.c (do_nothing_debug_hooks): Use
7899 debug_nothing_tree_tree_tree_bool_bool instead of
7900 debug_nothing_tree_tree_tree_bool.
7901 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
7902 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
7903 * sdbout.c (sdb_debug_hooks): Likewise.
7904 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
7905 (gen_namespace_die): Add DW_AT_export_symbols attribute if
7907 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
7908 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
7909 attribute, don't add anything.
7911 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7913 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
7914 (fold_build2_stat_loc): Likewise.
7915 (fold_build3_stat_loc): Likewise.
7916 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
7917 (fold_build1_loc): Remove macro.
7918 (fold_build2_loc): Likewise.
7919 (fold_build3_loc): Likewise.
7921 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7923 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
7924 (gimple_build_debug_bind_source_stat): Likewise.
7925 * gimple.h (gimple_build_debug_bind): Remove macro.
7926 (gimple_build_debug_bind_source): Likewise.
7928 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7930 * bitmap.c (bitmap_alloc): Adjust.
7931 (bitmap_gc_alloc): Likewise.
7932 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
7934 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7936 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
7937 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
7938 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
7939 (bitmap_gc_alloc_stat): Likewise.
7940 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
7942 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7944 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
7945 * rtl.h (shallow_copy_rtx): Remove macro.
7947 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7949 * emit-rtl.c (gen_raw_REG): Adjust.
7950 * gengenrtl.c (gendef): Likewise.
7951 * rtl.c (rtx_alloc_stat): Remove _stat from name.
7952 * rtl.h (rtx_alloc): Remove macro.
7954 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7956 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
7957 (build_tree_list_stat): Likewise.
7958 * tree.h (build_tree_list): Remove macro.
7959 (build_tree_list_vec): Likewise.
7961 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7963 * tree.c (make_vector_stat): Remove _stat from name.
7964 (build_vector_stat): Likewise.
7965 * tree.h (make_vector_stat): Remove macro.
7966 (build_vector_stat): Likewise.
7968 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7970 * tree.h (build_var_debug_value): Remove prototype.
7972 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7974 * tree.c (tree_cons_stat): Remove _stat from name.
7975 * tree.h (tree_cons): Remove macro.
7977 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7979 * tree.c (build_vl_exp_stat): Remove _stat from name.
7980 * tree.h (build_vl_exp): Remove macro.
7982 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7984 * tree.c (build_decl_stat): Remove _stat from name.
7985 * tree.h (build_decl): Remove macro.
7987 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7989 * gimple.c (gimple_build_with_ops_stat): Adjust.
7990 (gimple_alloc_stat): Remove _stat from name.
7991 * gimple.h (gimple_alloc): Remove macro.
7993 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7995 * tree.c (make_tree_vec_stat): Remove _stat from name.
7996 (grow_tree_vec_stat): Likewise.
7997 * tree.h (make_tree_vec_stat): Adjust prototype.
7998 (grow_tree_vec_stat): Likewise.
7999 (make_tree_vec): Remove macro.
8000 (grow_tree_vec): Likewise.
8002 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8004 * fold-const.c (fold_build1_stat_loc): Adjust.
8005 (fold_build2_stat_loc): Likewise.
8006 (fold_build3_stat_loc): Likewise.
8007 * tree.c (build0_stat): Remove _stat from name.
8008 (build1_stat): Likewise.
8009 (build2_stat): Likewise.
8010 (build3_stat): Likewise.
8011 (build4_stat): Likewise.
8012 (build5_stat): Likewise.
8013 * tree.h (build1_loc): Remove macro, and rename _stat function
8015 (build2_loc): Likewise.
8016 (build3_loc): Likewise.
8017 (build4_loc): Likewise.
8018 (build5_loc): Likewise.
8020 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8022 * tree.c (make_int_cst_stat): Remove _stat from name.
8023 * tree.h (make_int_cst_stat): Adjust prototype.
8024 (make_int_cst): Remove macro.
8026 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8028 * tree.c (make_tre_binfo_stat): Remove _stat from name.
8029 * tree.h (make_tree_binfo_stat): Adjust prototype.
8030 (make_tree_binfo): Remove.
8032 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8034 * tree.c (copy_node_stat): Rename to copy_node.
8035 (build_distinct_type_copy): Adjust.
8036 * tree.h (copy_node_stat): Adjust prototype.
8037 (copy_node): Remove macro.
8039 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8041 * tree.c (make_node_stat): rename to make_node.
8042 (build_tree_list_stat): Adjust.
8043 (build0_stat): Likewise.
8044 (build2_stat): Likewise.
8045 (build3_stat): Likewise.
8046 (build4_stat): Likewise.
8047 (build5_stat): Likewise.
8048 (build_decl_stat): Likewise.
8049 * tree.h (make_node_stat): Adjust prototype.
8050 (make_node): remove macro.
8052 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
8054 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
8055 (PPC_FEATURE2_SCV): Likewise.
8056 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
8058 2017-07-28 Tamar Christina <tamar.christina@arm.com>
8060 * config/aarch64/aarch64.c
8061 (aarch64_internal_mov_immediate): Add new special pattern.
8062 * config/aarch64/aarch64.md (*movdi_aarch64):
8063 Add reg/32bit const mov case.
8065 2017-07-28 Tamar Christina <tamar.christina@arm.com>
8066 Richard Sandiford <richard.sandiford@linaro.org>
8068 * config/aarch64/aarch64.md (mov<mode>): Generalize.
8069 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
8070 Add integer and movi cases.
8071 (movi-split-hf-df-sf split, fp16): New.
8072 (enabled): Added TARGET_FP_F16INST.
8073 * config/aarch64/iterators.md (GPF_HF): New.
8074 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
8076 2017-07-28 Tamar Christina <tamar.christina@arm.com>
8078 * config/aarch64/aarch64.c
8079 (aarch64_simd_container_mode): Add prototype.
8080 (aarch64_expand_mov_immediate): Add HI support.
8081 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
8082 (aarch64_can_const_movi_rtx_p): New.
8083 (aarch64_preferred_reload_class):
8084 Remove restrictions of using FP registers for certain SIMD operations.
8085 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
8086 (aarch64_valid_floating_const): Add integer move validation.
8087 (aarch64_simd_imm_scalar_p): Remove.
8088 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
8089 (aarch64_legitimate_constant_p): Expand list of supported cases.
8090 * config/aarch64/aarch64-protos.h
8091 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
8092 (aarch64_reinterpret_float_as_int): New.
8093 (aarch64_simd_imm_scalar_p): Remove.
8094 * config/aarch64/constraints.md (Uvi): New.
8095 (Dd): Split into Ds and new Dd.
8096 * config/aarch64/aarch64.md (*movsi_aarch64):
8098 (*movdi_aarch64): Add SIMD mov case.
8100 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8102 * tree-predcom.c: (struct chain): Handle store-store chain in which
8103 stores for elimination only store loop invariant values.
8104 (execute_pred_commoning_chain): Ditto.
8105 (prepare_initializers_chain_store_elim): Ditto.
8106 (prepare_finalizers): Ditto.
8107 (is_inv_store_elimination_chain): New function.
8108 (initialize_root_vars_store_elim_1): New function.
8110 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8112 * tree-predcom.c: Revise general description of the pass.
8113 (enum chain_type): New enum type for store elimination.
8114 (struct chain): New field supporting store elimination.
8115 (struct component): Ditto.
8116 (dump_chain): Dump store-stores chain.
8117 (release_chain): Release resources.
8118 (split_data_refs_to_components): Compute and create component
8119 contains only stores for elimination.
8120 (get_chain_last_ref_at): New function.
8121 (make_invariant_chain): Initialization.
8122 (make_rooted_chain): Specify chain type in parameter and record it.
8123 (add_looparound_copies): Skip for store-stores chain.
8124 (determine_roots_comp): Compute type of chain and pass it to
8126 (initialize_root_vars_store_elim_2): New function.
8127 (finalize_eliminated_stores): New function.
8128 (remove_stmt): Handle store for elimination.
8129 (execute_pred_commoning_chain): Execute predictive commoning on
8131 (determine_unroll_factor): Skip unroll for store-stores chain.
8132 (prepare_initializers_chain_store_elim): New function.
8133 (prepare_initializers_chain): Hanlde store-store chain.
8134 (prepare_finalizers_chain, prepare_finalizers): New function.
8135 (tree_predictive_commoning_loop): Return integer value indicating
8136 if loop is unrolled or lcssa form is corrupted.
8137 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
8139 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8141 * tree-predcom.c (initialize_root): Delete.
8142 (execute_pred_commoning_chain): Initialize root vars and replace
8143 reference of non-combined chain directly, rather than call above
8146 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8148 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
8149 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
8151 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8153 * tree-predcom.c (struct chain): New field init_seq.
8154 (release_chain): Release init_seq.
8155 (prepare_initializers_chain): Record intialization stmts in above
8157 (insert_init_seqs): New function.
8158 (tree_predictive_commoning_loop): Call insert_init_seqs.
8160 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8162 * tree-predcom.c (determine_roots_comp): Skip trivial components.
8164 2017-07-28 Richard Biener <rguenther@suse.de>
8166 * match.pd: Remove superfluous :c.
8167 * genmatch.c (simplify::id): Add member.
8168 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
8170 (current_id): New global.
8171 (dt_node::parent): Move from ...
8172 (dt_operand::parent): ... here. Add for_id member.
8173 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
8174 (decision_tree::find_node): Relax order requirement when
8175 merging DT_TRUE nodes to ones inbetween the current simplify
8176 and the one we try to merge with. Add diagnostic whenever
8177 we need to enforce pattern order by not merging.
8178 (decision_tree::insert): Set current_id.
8179 (decision_tree::print_node): Dump parent node and for_id.
8180 (parser::last_id): Add member.
8181 (parser::push_simplify): Assign unique id.
8182 (parser::parser): Initialize last_id.
8184 2017-07-28 Martin Liska <mliska@suse.cz>
8187 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
8188 gimple_build_debug_bind.
8190 2017-07-28 Richard Biener <rguenther@suse.de>
8192 PR tree-optimization/81502
8193 * match.pd: Add pattern combining BIT_INSERT_EXPR with
8195 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
8197 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
8198 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
8199 for BIT_FIELD_REF args.
8200 * fold-const.c (make_bit_field_ref): Likewise.
8201 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8203 2017-07-28 Jakub Jelinek <jakub@redhat.com>
8206 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
8207 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
8208 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
8209 Or it into SANITIZER_UNDEFINED.
8210 * ubsan.c: Include gimple-fold.h and varasm.h.
8211 (ubsan_expand_ptr_ifn): New function.
8212 (instrument_pointer_overflow): New function.
8213 (maybe_instrument_pointer_overflow): New function.
8214 (instrument_object_size): Formatting fix.
8215 (pass_ubsan::execute): Call instrument_pointer_overflow
8216 and maybe_instrument_pointer_overflow.
8217 * internal-fn.c (expand_UBSAN_PTR): New function.
8218 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
8219 * sanitizer.def (__ubsan_handle_pointer_overflow,
8220 __ubsan_handle_pointer_overflow_abort): New builtins.
8221 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
8222 * internal-fn.def (UBSAN_PTR): New internal function.
8223 * opts.c (sanitizer_opts): Add pointer-overflow.
8224 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
8225 * fold-const.c (build_range_check): Compute pointer range check in
8226 integral type if pointer arithmetics would be needed. Formatting
8229 2017-07-28 Martin Liska <mliska@suse.cz>
8232 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
8233 parameters that are of a variable-length.
8235 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
8237 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
8239 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
8240 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
8241 (CRT_CALL_STATIC_FUNCTION): Likewise.
8242 (ASM_DEFAULT_SPEC): New define.
8243 (ASM_SPEC32): Likewise.
8244 (ASM_SPEC64): Likewise.
8245 (ASM_SPEC_COMMON): Likewise.
8246 (ASM_SPEC): Likewise.
8247 (INVALID_64BIT): Likewise.
8248 (LINK_OS_DEFAULT_SPEC): Likewise.
8249 (LINK_OS_SPEC32): Likewise.
8250 (LINK_OS_SPEC64): Likewise.
8251 (POWERPC_LINUX): Likewise.
8252 (PTRDIFF_TYPE): Likewise.
8253 (RESTORE_FP_PREFIX): Likewise.
8254 (RESTORE_FP_SUFFIX): Likewise.
8255 (SAVE_FP_PREFIX): Likewise.
8256 (SAVE_FP_SUFFIX): Likewise.
8257 (SIZE_TYPE): Likewise.
8258 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
8259 (TARGET_64BIT): Likewise.
8260 (TARGET_64BIT): Likewise.
8261 (TARGET_AIX): Likewise.
8262 (WCHAR_TYPE_SIZE): Likewise.
8263 (WCHAR_TYPE): Undefine.
8264 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
8265 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
8266 (CPP_OS_RTEMS_SPEC): Delete.
8267 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
8268 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
8270 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
8272 2017-07-28 Jakub Jelinek <jakub@redhat.com>
8274 PR tree-optimization/81578
8275 * tree-parloops.c (build_new_reduction): Bail out if
8276 reduction_code isn't one of the standard OpenMP reductions.
8277 Move the details printing after that decision.
8279 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8281 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
8282 related to reload_in_progress.
8283 (splat_input_operand): Likewise.
8284 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
8286 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
8288 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
8289 (TARGET_INSTANTIATE_DECLS): Likewise.
8290 (legitimate_indexed_address_p): Delete reload_in_progress code.
8291 (rs6000_debug_legitimate_address_p): Likewise.
8292 (rs6000_eliminate_indexed_memrefs): Likewise.
8293 (rs6000_emit_le_vsx_store): Likewise.
8294 (rs6000_emit_move_si_sf_subreg): Likewise.
8295 (rs6000_emit_move): Likewise.
8296 (register_to_reg_type): Likewise.
8297 (rs6000_pre_atomic_barrier): Likewise.
8298 (rs6000_machopic_legitimize_pic_address): Likewise.
8299 (rs6000_allocate_stack_temp): Likewise.
8300 (rs6000_address_for_fpconvert): Likewise.
8301 (rs6000_address_for_altivec): Likewise.
8302 (rs6000_secondary_memory_needed_rtx): Delete function.
8303 (rs6000_check_sdmode): Likewise.
8304 (rs6000_alloc_sdmode_stack_slot): Likewise.
8305 (rs6000_instantiate_decls): Likewise.
8306 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
8307 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
8308 Delete reload_in_progress.
8309 (*vec_reload_and_plus_<mptrsize>): Likewise.
8310 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
8311 (vsx_div_v2di): Likewise.
8312 (vsx_udiv_v2di): Likewise.
8314 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8316 * config/rs6000/rs6000.opt (mlra): Replace with stub.
8317 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
8318 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
8319 (rs6000_debug_reg_global): Delete print of LRA status.
8320 (rs6000_option_override_internal): Delete dead LRA related code.
8321 (rs6000_lra_p): Delete function.
8322 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
8324 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
8326 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
8327 * config/riscv/rtems.h: New file.
8329 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8330 Sudakshina Das <sudi.das@arm.com>
8332 * config/aarch64/aarch64.md
8333 (define_split for and<mode>3nr_compare): Move
8334 non aarch64_logical_operand to a register.
8335 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
8336 register immediate operand to a register.
8337 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
8339 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8342 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
8344 2017-07-27 Richard Biener <rguenther@suse.de>
8346 PR tree-optimization/81573
8347 PR tree-optimization/81494
8348 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
8349 multi defuse cycle case.
8351 2017-07-27 Richard Biener <rguenther@suse.de>
8353 PR tree-optimization/81571
8354 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
8357 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
8359 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
8360 earlier and only if MASK_FPU is set. Adjust formatting.
8362 2017-07-27 Martin Liska <mliska@suse.cz>
8364 * opt-functions.awk: Add validation of value of Init.
8365 * optc-gen.awk: Pass new argument.
8367 2017-07-27 Martin Liska <mliska@suse.cz>
8369 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
8370 Fix wrong condition.
8372 2017-07-27 Martin Liska <mliska@suse.cz>
8374 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
8375 BBs and edges seen by autoFDO.
8377 2017-07-27 Richard Biener <rguenther@suse.de>
8379 PR tree-optimization/81502
8380 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
8381 with incompatible but same sized type.
8382 (execute_update_addresses_taken): Likewise.
8384 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
8386 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
8387 flag_tree_loop_vectorize rather than flag_tree_vectorize.
8389 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8392 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
8393 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
8394 Change s_operand to memory_operand.
8396 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
8398 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
8399 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
8400 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
8401 Emit instructions rather than returning an expression. Handle TFmode
8402 and KFmode by casting to TImode.
8403 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
8404 (rs6000_emit_le_vsx_store): Likewise.
8405 * config/rs6000/vsx.md (VSX_TI): New iterator.
8406 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
8407 (*vsx_le_undo_permute_<mode>): Likewise.
8408 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
8409 emit the split sequence.
8410 (*vsx_le_perm_store_<mode>): Likewise.
8412 2017-07-27 Jakub Jelinek <jakub@redhat.com>
8414 PR tree-optimization/81555
8415 PR tree-optimization/81556
8416 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
8417 if true, force CHANGED for the recursive invocation.
8418 (reassociate_bb): Remember original length of ops array, pass
8419 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
8421 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
8422 attributes for noipa attribute. For naked attribute use
8423 lookup_attribute first before lookup_attribute_spec.
8424 * final.c (rest_of_handle_final): Disable IPA RA for functions with
8426 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
8427 for functions with noipa attribute.
8428 (cgraph_externally_visible_p): Return true for functions with noipa
8430 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
8431 for functions with noipa attribute.
8432 * doc/extend.texi: Document noipa function attribute.
8433 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
8434 also for functions with noipa attribute.
8435 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
8437 2017-07-26 Andrew Pinski <apinski@cavium.com>
8439 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
8440 vec_unalign_load_cost and vec_unalign_store_cost.
8442 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8444 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
8445 -mvsx-small-integer option.
8446 (ISA_3_0_MASKS_IEEE): Likewise.
8447 (OTHER_VSX_VECTOR_MASKS): Likewise.
8448 (POWERPC_MASKS): Likewise.
8449 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
8450 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
8451 code, only testing for DImode being allowed in non-VSX floating
8453 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
8454 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
8456 (rs6000_option_override_internal): Delete -mvsx-small-integer.
8457 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
8458 TARGET_P8_VECTOR test.
8459 (rs6000_secondary_reload_simple_move): Likewise.
8460 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
8461 since TARGET_P9_VECTOR was already tested.
8462 (rs6000_opt_masks): Remove -mvsx-small-integer.
8463 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
8464 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8466 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
8467 test for TARGET_VEXTRACTUB was used, and that uses
8469 (p9 extract splitter): Likewise.
8470 (vsx_extract_<mode>_di_p9): Likewise.
8471 (vsx_extract_<mode>_store_p9): Likewise.
8472 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
8473 for TARGET_P9_VECTOR was used. Delete code that is now dead with
8474 the elimination of TARGET_VSX_SMALL_INTEGER.
8475 (vsx_extract_<mode>_p8): Likewise.
8476 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
8477 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
8478 (vsx_set_<mode>_p9): Likewise.
8479 (vsx_set_v4sf_p9): Likewise.
8480 (vsx_set_v4sf_p9_zero): Likewise.
8481 (vsx_insert_extract_v4sf_p9): Likewise.
8482 (vsx_insert_extract_v4sf_p9_2): Likewise.
8483 * config/rs6000/rs6000.md (sign extend splitter): Change
8484 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
8485 (floatsi<mode>2_lfiwax_mem): Likewise.
8486 (floatunssi<mode>2_lfiwzx_mem): Likewise.
8487 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
8488 since a test for TARGET_P9_VECTOR was used.
8489 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8490 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
8491 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8492 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
8493 TARGET_P8_VECTOR test.
8494 (fix_trunc<mode>si2_stfiwx): Likewise.
8495 (fix_trunc<mode>si2_internal): Likewise.
8496 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
8497 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8499 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
8500 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
8501 TARGET_P8_VECTOR test.
8502 (fixuns_trunc<mode>si2_stfiwx): Likewise.
8503 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
8504 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8506 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
8507 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
8508 since a test for TARGET_P9_VECTOR was used.
8509 (splitter for loading small constants): Likewise.
8511 2017-07-26 Andrew Pinski <apinski@cavium.com>
8513 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
8516 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
8519 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
8520 (fp_valid_at): Likewise.
8522 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
8524 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
8525 (qdf24xx_addrcost_table): Likewise.
8526 (cortexa57_tunings): Update to use generic_branch_cost.
8527 (cortexa72_tunings): Likewise.
8528 (cortexa73_tunings): Likewise.
8529 (qdf24xx_tunings): Likewise.
8531 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
8533 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
8534 (thunderx2t99_branch_cost): Likewise.
8535 (cortexa35_tunings): Update to use generic_branch_cost.
8536 (cortexa53_tunings): Likewise.
8537 (cortexa57_tunings): Likewise.
8538 (cortexa72_tunings): Likewise.
8539 (cortexa73_tunings): Likewise.
8540 (thunderx2t99_tunings): Likewise.
8542 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
8544 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
8545 (sparc_option_override): Honour MASK_FSMULD.
8546 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
8547 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
8548 * config/sparc/sparc.opt (mfsmuld): New option.
8549 * doc/invoke.texi (mfsmuld): Document option.
8551 2017-07-26 Marek Polacek <polacek@redhat.com>
8554 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
8556 2017-07-26 Richard Biener <rguenther@suse.de>
8558 * gimple-match-head.c (do_valueize): Return OP if valueize
8560 (get_def): New helper to get at the def stmt of a SSA name
8562 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
8563 do_valueize to get at the def stmt.
8564 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
8566 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
8569 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
8571 2017-07-26 Martin Liska <mliska@suse.cz>
8574 * function.c (expand_function_start): Make expansion of
8575 nonlocal_goto_save_area after parm_birth_insn.
8577 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
8579 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
8580 from all CPU target flags enable members.
8582 2017-07-26 Richard Biener <rguenther@suse.de>
8584 * genmatch.c (dt_simplify::gen): Make iterator vars const.
8585 (decision_tree::gen): Make 'type' const.
8586 (write_predicate): Likewise.
8588 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
8590 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
8591 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
8592 (rs6000_option_override_internal): Likewise.
8593 (rs6000_expand_vector_set): Likewise.
8594 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
8595 (TARGET_UPPER_REGS_SF): Likewise.
8596 (TARGET_UPPER_REGS_DI): Likewise.
8597 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
8598 (TARGET_DIRECT_MOVE_64BIT): Likewise.
8599 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
8600 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8601 (Splitters for DI constants in Altivec registers): Likewise.
8602 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
8603 (vsx_set_v4sf_p9): Likewise.
8604 (vsx_set_v4sf_p9_zero): Likewise.
8605 (vsx_insert_extract_v4sf_p9): Likewise.
8606 (vsx_insert_extract_v4sf_p9_2): Likewise.
8608 2017-07-25 Carl Love <cel@us.ibm.com>
8610 * doc/extend.texi: Update the built-in documentation file for the
8611 existing built-in functions
8612 vector signed char vec_cnttz (vector signed char);
8613 vector unsigned char vec_cnttz (vector unsigned char);
8614 vector signed short vec_cnttz (vector signed short);
8615 vector unsigned short vec_cnttz (vector unsigned short);
8616 vector signed int vec_cnttz (vector signed int);
8617 vector unsigned int vec_cnttz (vector unsigned int);
8618 vector signed long long vec_cnttz (vector signed long long);
8619 vector unsigned long long vec_cnttz (vector unsigned long long);
8621 2017-07-25 Andrew Pinski <apinski@cavium.com>
8623 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
8624 accesses where the use is for the first operand of a BIT_INSERT.
8626 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
8629 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
8630 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
8632 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
8634 * config/i386/gstabs.h: Delete.
8635 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
8637 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
8639 * config/i386/i386.c (ix86_decompose_address): Do not check for
8640 register RTX when looking at index_reg or base_reg.
8641 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
8643 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
8645 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
8646 to update EH info here.
8648 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8650 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
8652 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8654 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
8656 2017-07-25 Torsten Duwe <duwe@suse.de>
8658 * common.opt: Introduce -fpatchable-function-entry
8659 command line option, and its variables function_entry_patch_area_size
8660 and function_entry_patch_area_start.
8661 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
8662 including a two-value parser.
8663 * target.def (print_patchable_function_entry): New target hook.
8664 * targhooks.h (default_print_patchable_function_entry): New function.
8665 * targhooks.c (default_print_patchable_function_entry): Likewise.
8666 * toplev.c (process_options): Switch off IPA-RA if
8667 patchable function entries are being generated.
8668 * varasm.c (assemble_start_function): Look at the
8669 patchable-function-entry command line switch and current
8670 function attributes and maybe generate NOP instructions by
8671 calling the print_patchable_function_entry hook.
8672 * doc/extend.texi: Document patchable_function_entry attribute.
8673 * doc/invoke.texi: Document -fpatchable_function_entry
8674 command line option.
8675 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
8677 * doc/tm.texi: Re-generate.
8679 2017-07-25 Jakub Jelinek <jakub@redhat.com>
8682 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
8683 TARGET_AVX512DQ rather than TARGET_AVX512BW.
8685 2017-07-25 Tamar Christina <tamar.christina@arm.com>
8687 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
8689 2017-07-25 Richard Biener <rguenther@suse.de>
8691 PR tree-optimization/81455
8692 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
8693 not walk in cycles when looking for guards.
8695 2017-07-25 Richard Biener <rguenther@suse.de>
8697 PR tree-optimization/81529
8698 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
8699 when optimizing backedge uses.
8701 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
8703 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
8705 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
8706 to dl_section_ref. On AIX, append an expression to subtract
8707 the size of the section length to dl_section_ref.
8709 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
8711 * configure.ac: If any of the config.* scripts fail, exit 1.
8712 * configure: Regenerate.
8714 2017-07-25 Richard Biener <rguenther@suse.de>
8717 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
8718 of immediate uses, be more verbose on errors.
8720 2017-07-25 Richard Biener <rguenther@suse.de>
8722 PR tree-optimization/81510
8723 * tree-vect-loop.c (vect_is_simple_reduction): When the
8724 reduction stmt is not inside the loop bail out.
8726 2017-07-25 Richard Biener <rguenther@suse.de>
8728 PR tree-optimization/81303
8729 * tree-vect-loop-manip.c (vect_loop_versioning): Build
8730 profitability check against LOOP_VINFO_NITERSM1.
8732 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8734 * domwalk.c (cmp_bb_postorder): Simplify.
8735 (sort_bbs_postorder): New function. Use it...
8736 (dom_walker::walk): ...here to optimize common cases.
8738 2017-07-25 Martin Liska <mliska@suse.cz>
8741 * ipa-visibility.c (function_and_variable_visibility): Make the
8742 redirection just on target that supports aliasing.
8743 Fix GNU coding style.
8745 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
8748 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
8750 * config/arm/rtems.h: Likewise.
8751 * config/bfin/rtems.h: Likewise.
8752 * config/i386/rtemself.h: Likewise.
8753 * config/lm32/rtems.h: Likewise.
8754 * config/m32c/rtems.h: Likewise.
8755 * config/m68k/rtemself.h: Likewise.
8756 * config/microblaze/rtems.h: Likewise.
8757 * config/mips/rtems.h: Likewise.
8758 * config/moxie/rtems.h: Likewise.
8759 * config/nios2/rtems.h: Likewise.
8760 * config/powerpcspe/rtems.h: Likewise.
8761 * config/rs6000/rtems.h: Likewise.
8762 * config/rtems.h: Likewise.
8763 * config/sh/rtems.h: Likewise.
8764 * config/sh/rtemself.h: Likewise.
8765 * config/sparc/rtemself.h: Likewise.
8767 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
8770 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
8771 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
8772 * tree-ssa-structalias.c (alias_get_name): Same.
8774 2017-07-25 Bin Cheng <bin.cheng@arm.com>
8777 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
8778 instructions if no du chain is found.
8780 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
8782 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
8784 2017-07-25 Richard Biener <rguenther@suse.de>
8787 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
8790 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
8792 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
8794 (ISA_2_7_MASKS_SERVER): Likewise.
8795 (ISA_3_0_MASKS_IEEE): Likewise.
8796 (OTHER_P8_VECTOR_MASKS): Likewise.
8797 (OTHER_VSX_VECTOR_MASKS): Likewise.
8798 (POWERPC_MASKS): Likewise.
8799 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
8800 duplicate list of options.
8801 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
8802 explicit -mupper-regs options.
8803 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
8804 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
8805 alias for -mupper-regs-df.
8806 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
8807 (rs6000_init_hard_regno_mode_ok): Likewise.
8808 (rs6000_option_override_internal): Likewise.
8809 (rs6000_opt_masks): Likewise.
8810 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
8811 options in terms of whether -mvsx or -mpower8-vector was used.
8812 (TARGET_UPPER_REGS_DI): Likewise.
8813 (TARGET_UPPER_REGS_SF): Likewise.
8814 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
8815 -mupper-regs-* options.
8817 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
8819 * passes.c (emergency_dump_function): Print some empty lines and a
8820 header before the RTL dump.
8822 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
8824 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
8826 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
8829 * config/aarch64/aarch64.c (aarch64_classify_symbol):
8830 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
8832 2017-07-24 Carl Love <cel@us.ibm.com>
8834 * config/rs6000/rs6000-c.c: Add support for built-in functions
8835 vector float vec_extract_fp32_from_shorth (vector unsigned short);
8836 vector float vec_extract_fp32_from_shortl (vector unsigned short);
8837 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
8838 vec_extract_fp_from_shortl): Add defines for the two builtins.
8839 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
8840 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
8842 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
8843 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
8844 * doc/extend.texi: Update the built-in documentation file for the
8845 new built-in function.
8847 2017-07-24 Jakub Jelinek <jakub@redhat.com>
8850 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
8852 * doc/generic.texi: Likewise.
8853 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
8854 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
8856 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
8858 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
8859 (aarch64_mls_elt_merge<mode>): Likewise.
8861 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
8863 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
8864 having __cxa_atexit.
8866 2017-07-23 Michael Collison <michael.collison@arm.com>
8868 * config/arm/arm.c (arm_option_override): Deprecate
8869 use of -mstructure-size-boundary.
8870 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
8871 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
8873 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8876 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
8877 Reduce cost estimate for direct moves.
8879 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
8882 * config/i386/i386.c (ix86_option_override_internal): Disable
8883 BMI, BMI2 and TBM instructions for -m16.
8885 2017-07-21 Carl Love <cel@us.ibm.com>
8887 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8888 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8889 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8890 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8891 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8892 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8893 VMULOSW): New enum "unspec" values.
8894 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8895 altivec_vmulosw): New patterns.
8896 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8897 VMULOSW): Add definitions.
8899 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
8901 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
8902 (qdf24xx): Likewise.
8903 * config/aarch64/aarch64-options-extensions.def (rdma); New.
8904 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
8905 (AARCH64_FL_V8_1): Renumber.
8906 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
8907 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
8908 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
8909 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
8910 rdma to feature modifiers list.
8912 2017-07-21 Yury Gribov <tetra2005@gmail.com>
8915 * ipa-visibility (function_and_variable_visibility): Convert
8916 recursive PLT call to direct call if appropriate.
8918 2017-07-21 Andrew Pinski <apinski@cavium.com>
8920 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
8921 operand 1 to see if the types precision matches.
8922 * fold-const.c (operand_equal_p): Likewise.
8924 2017-07-21 Richard Biener <rguenther@suse.de>
8926 PR tree-optimization/81303
8927 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
8928 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
8929 (vect_peeling_hash_get_lowest_cost): Adjust.
8930 (vect_enhance_data_refs_alignment): Likewise. Use
8931 vect_get_peeling_costs_all_drs to compute the penalty for no
8932 peeling to match up costs.
8934 2017-07-21 Richard Biener <rguenther@suse.de>
8936 PR tree-optimization/81500
8937 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
8938 we didn't identify a reduction path.
8940 2017-07-21 Tom de Vries <tom@codesourcery.com>
8941 Cesar Philippidis <cesar@codesourcery.com>
8943 PR gcov-profile/81442
8944 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
8947 2017-07-21 Tom de Vries <tom@codesourcery.com>
8950 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
8952 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
8953 nvptx_override_options_after_change.
8955 2017-07-21 Ulrich Drepper <drepper@redhat.com>
8957 * dwarf2out.c (output_file_names): Avoid double testing for
8960 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
8962 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
8964 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
8966 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
8968 (try_crossjump_to_edge): Do not punt on partitioned functions.
8970 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
8972 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
8973 Put all BBs reachable only via paths crossing cold region to cold
8975 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
8977 2016-07-21 Richard Biener <rguenther@suse.de>
8979 PR tree-optimization/81303
8980 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
8981 into account prologue and epilogue iterations when raising
8982 min_profitable_iters to sth at least covering one vector iteration.
8984 2017-07-21 Tamar Christina <tamar.christina@arm.com>
8986 * config/arm/arm.c (arm_test_cpu_arch_dat):
8989 2017-07-20 Nathan Sidwell <nathan@acm.org>
8991 Remove TYPE_METHODS.
8992 * tree.h (TYPE_METHODS): Delete.
8993 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
8994 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
8995 (dbxout_type_methods): Scan TYPE_FIELDS.
8996 (dbxout_type): Don't check TYPE_METHODS here.
8997 * function.c (use_register_for_decl): Always ignore register for
8998 class types when not optimizing.
8999 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
9000 * tree.c (free_lang_data_in_type): Stitch out member functions and
9001 templates from TYPE_FIELDS.
9002 (build_distinct_type_copy, verify_type_variant,
9003 verify_type): Member fns are on TYPE_FIELDS.
9004 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
9005 * tree-pretty-print.c (dump_generic_node): Likewise.
9007 2017-07-20 Jakub Jelinek <jakub@redhat.com>
9010 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
9011 V2TImode and V4TImode.
9012 (ix86_expand_vector_extract): Likewise.
9013 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
9014 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
9015 (ssescalarmode): Handle V4TImode and V2TImode.
9016 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
9017 (*vec_extractv2ti, *vec_extractv4ti): New insns.
9018 (VEXTRACTI128_MODE): New mode iterator.
9019 (splitter for *vec_extractv?ti first element): New.
9020 (VEC_INIT_MODE): New mode iterator.
9021 (vec_init<mode>): Consolidate 3 expanders into one using
9022 VEC_INIT_MODE mode iterator.
9024 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
9026 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
9027 non_spilled_static_chain_regno_p.
9029 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
9031 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
9033 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
9035 * bb-reorder.c (connect_traces): Allow copying of blocks within
9038 2017-07-20 Richard Biener <rguenther@suse.de>
9040 * gimple.h (gimple_phi_result): Add gphi * overload.
9041 (gimple_phi_result_ptr): Likewise.
9042 (gimple_phi_arg): Likewise. Adjust index assert to only
9043 allow actual argument accesses rather than all slots available
9045 (gimple_phi_arg_def): Add gphi * overload.
9046 * tree-phinodes.c (make_phi_node): Initialize only actual
9048 (resize_phi_node): Clear memory not covered by old node,
9049 do not initialize excess argument slots.
9050 (reserve_phi_args_for_new_edge): Initialize new argument slot
9053 2017-07-20 Bin Cheng <bin.cheng@arm.com>
9055 PR tree-optimization/81388
9057 2016-07-21 Bin Cheng <bin.cheng@arm.com>
9059 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
9060 by removing computation of may_be_zero.
9062 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
9063 Tom de Vries <tom@codesourcery.com>
9066 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
9067 when gimple level profile disagrees with what RTL expander did.
9069 2017-07-20 Richard Biener <rguenther@suse.de>
9071 PR tree-optimization/61171
9072 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
9073 (vect_analyze_stmt): Add slp instance parameter.
9074 (vectorizable_reduction): Likewise.
9075 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
9076 (vect_is_simple_reduction): Deal with chains not detected
9077 as SLP reduction chain, specifically not properly associated
9078 chains containing a mix of plus/minus.
9079 (get_reduction_op): Remove.
9080 (get_initial_defs_for_reduction): Simplify, pass in whether
9081 this is a reduction chain, pass in the SLP node for the PHIs.
9082 (vect_create_epilog_for_reduction): Get the SLP instance as
9084 (vectorizable_reduction): Get the SLP instance as arg.
9085 During analysis remember the SLP node with the PHIs in the
9086 instance. Simplify getting at the vectorized reduction PHIs.
9087 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
9088 through SLP instance.
9089 (vect_slp_analyze_operations): Likewise.
9090 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
9091 (vect_transform_stmt): Likewise.
9093 2017-07-20 Tom de Vries <tom@codesourcery.com>
9095 PR tree-optimization/81489
9096 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
9097 read of phi arg location to before loop that modifies phi.
9099 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9101 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
9104 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
9107 * except.c (execute): Fix ordering issue.
9109 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
9111 PR rtl-optimization/81423
9112 * combine.c (make_compound_operation_int): Don't try to optimize
9113 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
9115 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
9117 PR rtl-optimization/81423
9118 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
9119 with a constant that is -1 in the truncated to mode.
9121 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
9123 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
9124 (determine_unlikely_bbs): ... here.
9125 * predict.h (propagate_unlikely_bbs_forward): Declare.
9126 * cfgexpand.c (pass_expand::execute): Use it.
9127 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
9129 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
9130 propagate_unlikely_bbs_forward.
9132 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
9135 * except.c (maybe_add_nop_after_section_switch): New function.
9138 2017-07-19 Tom de Vries <tom@codesourcery.com>
9140 * gimple.h (gimple_phi_set_arg): Make assert more strict.
9142 2017-07-19 Tom de Vries <tom@codesourcery.com>
9144 * gimple.h (gimple_phi_arg): Make assert more strict.
9146 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
9148 * config.gcc (powerpc*-*-*): Add mmintrin.h.
9149 * config/rs6000/mmintrin.h: New file.
9150 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
9152 2017-07-19 Jakub Jelinek <jakub@redhat.com>
9154 PR tree-optimization/81346
9155 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
9157 2017-07-19 Tom de Vries <tom@codesourcery.com>
9159 * config/nvptx/nvptx.md (VECIM): Add V2DI.
9161 2017-07-19 Tom de Vries <tom@codesourcery.com>
9163 * config/nvptx/nvptx-modes.def: Add V2DImode.
9164 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
9165 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
9166 (nvptx_output_mov_insn): Handle lack of mov.b128.
9167 (nvptx_print_operand): Handle 'H' and 'L' codes.
9168 (nvptx_vector_mode_supported): Allow V2DImode.
9169 (nvptx_preferred_simd_mode): New function.
9170 (nvptx_data_alignment): New function.
9171 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
9172 nvptx_preferred_simd_mode.
9173 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
9175 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
9177 2017-07-19 Tom de Vries <tom@codesourcery.com>
9179 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
9180 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
9181 (nvptx_vector_mode_supported): New function. Allow V2SImode.
9182 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
9183 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
9184 (mov<VECIM>_insn): New define_insn.
9185 (define_expand "mov<VECIM>): New define_expand.
9187 2017-07-19 Tom de Vries <tom@codesourcery.com>
9189 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
9191 2017-07-19 Jakub Jelinek <jakub@redhat.com>
9193 PR tree-optimization/81346
9194 * fold-const.h (fold_div_compare, range_check_type): Declare.
9195 * fold-const.c (range_check_type): New function.
9196 (build_range_check): Use range_check_type.
9197 (fold_div_compare): No longer static, rewritten into
9198 a match.pd helper function.
9199 (fold_comparison): Don't call fold_div_compare here.
9200 * match.pd (X / C1 op C2): New optimization using fold_div_compare
9203 2017-07-19 Nathan Sidwell <nathan@acm.org>
9205 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
9206 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
9207 * tree.c (find_decls_types_r, verify_type): Use
9208 TYPE_{MIN,MAX}_VALUE_RAW.
9209 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
9210 (hash_tree): Likewise.
9211 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
9213 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
9216 2017-07-18 Tom de Vries <tom@codesourcery.com>
9219 * omp-expand.c (expand_omp_for_static_chunk): Handle
9220 equal-argument loop exit phi.
9222 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
9225 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
9226 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
9227 operand 2 predicate.
9228 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
9229 operand 2 predicate.
9230 (ror,rol -> rorx splitters): Use const_int_operand as
9231 operand 2 predicate.
9233 2017-06-18 Richard Biener <rguenther@suse.de>
9235 PR tree-optimization/81410
9236 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
9237 the gap in the ! slp_perm SLP case after each group.
9239 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
9242 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
9245 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
9248 * predict.c (set_even_probabilities): Cleanup; do not affect
9249 probabilities that are already known.
9250 (combine_predictions_for_bb): Call even when count is set.
9252 2017-07-18 Nathan Sidwell <nathan@acm.org>
9254 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
9257 2017-07-18 Bin Cheng <bin.cheng@arm.com>
9260 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
9261 optimization for loop niter analysis.
9263 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
9266 * config/avr/avr.c (avr_optimize_casesi): Don't use
9267 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
9269 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
9271 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
9272 body_cost_vec from _vect_peel_extended_info.
9273 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
9274 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
9277 2017-07-18 Bin Cheng <bin.cheng@arm.com>
9279 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
9281 2017-07-18 Richard Biener <rguenther@suse.de>
9283 PR tree-optimization/80620
9284 PR tree-optimization/81403
9285 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
9286 info when re-using a VN table entry.
9288 2017-07-18 Richard Biener <rguenther@suse.de>
9290 PR tree-optimization/81418
9291 * tree-vect-loop.c (vectorizable_reduction): Properly compute
9292 vectype_in. Verify that with lane-reducing reduction operations
9293 we have a single def-use cycle.
9295 2017-07-17 Carl Love <cel@us.ibm.com>
9297 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
9299 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9300 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9301 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9302 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9303 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9304 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
9305 VMULOSW): New enum "unspec" values.
9306 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
9307 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
9308 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
9309 altivec_vmulosw): New patterns.
9310 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9311 VMULOSW): Add definitions.
9312 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
9314 * config/alpha/alpha.c: Include predict.h.
9316 2017-07-17 Yury Gribov <tetra2005@gmail.com>
9318 * tree-vrp.c (compare_assert_loc): Fix comparison function
9319 to return predictable results.
9321 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9323 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
9326 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
9327 * doc/invoke.texi (mexpand-adddi): Update text.
9329 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9331 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
9332 that also clobbers the CC register. The old expand code is moved
9334 (*arc_clzsi2): ... here.
9335 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
9336 the CC register. The old expand code is moved to ...
9337 (arc_ctzsi2): ... here.
9339 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9341 * config/arc/arc.opt (mindexed-loads): Use initial value
9342 TARGET_INDEXED_LOADS_DEFAULT.
9343 (mauto-modify-reg): Use initial value
9344 TARGET_AUTO_MODIFY_REG_DEFAULT.
9345 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
9346 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
9347 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
9348 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
9350 2017-07-17 Martin Liska <mliska@suse.cz>
9353 * opts.c (finish_options): Do not allow -fgnu-tm
9354 w/ -fsanitize={kernel-,}address. Say sorry.
9356 2017-07-17 Bin Cheng <bin.cheng@arm.com>
9359 * tree-loop-distribution.c (classify_partition): Only assert on
9360 numer of iterations.
9361 (merge_dep_scc_partitions): Delete prameter. Update function call.
9362 (distribute_loop): Remove code handling loop with unknown niters.
9363 (pass_loop_distribution::execute): Skip loop with unknown niters.
9365 2017-07-17 Bin Cheng <bin.cheng@arm.com>
9368 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
9369 function sort_partitions_by_post_order.
9371 2017-07-17 Bin Cheng <bin.cheng@arm.com>
9373 PR tree-optimization/81374
9374 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
9375 the max index of basic blocks, rather than number of basic blocks.
9377 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9379 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
9381 (arc_legitimate_pic_operand_p): Likewise.
9382 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
9384 (arc_needs_pcl_p): Likewise.
9385 (arc_legitimate_pc_offset_p): Likewise.
9386 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
9387 function is also used in constrains.md.
9388 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
9389 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
9390 PLUS. Only return true/false in known cases, otherwise assert.
9391 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
9392 is already called in arc_legitimate_constant_p.
9393 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
9395 (LEGITIMATE_PIC_OPERAND_P): Use
9396 arc_raw_symbolic_reference_mentioned_p function.
9397 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
9402 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9403 Andrew Burgess <andrew.burgess@embecosm.com>
9405 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
9406 (arc_return_address_register): New function.
9407 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
9408 (arc_handle_fndecl_attribute): Add naked attribute.
9409 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
9410 (TARGET_WARN_FUNC_RETURN): Likewise.
9411 (arc_allocate_stack_slots_for_args): New function.
9412 (arc_warn_func_return): Likewise.
9413 (machine_function): Change type fn_type.
9414 (arc_compute_function_type): Consider new naked function type,
9415 change function return type.
9416 (arc_must_save_register): Adapt to handle new
9417 arc_compute_function_type's return type.
9418 (arc_expand_prologue): Likewise.
9419 (arc_expand_epilogue): Likewise.
9420 (arc_return_address_regs): Delete.
9421 (arc_return_address_register): New function.
9422 (arc_epilogue_uses): Use above function.
9423 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
9424 (arc_function_type): Change encoding, add naked type.
9425 (ARC_INTERRUPT_P): Change to handle the new encoding.
9426 (ARC_FAST_INTERRUPT_P): Likewise.
9427 (ARC_NORMAL_P): Define.
9428 (ARC_NAKED_P): Likewise.
9429 (arc_compute_function_type): Delete prototype.
9430 * config/arc/arc.md (in_ret_delay_slot): Use
9431 arc_return_address_register function.
9432 (simple_return): Likewise.
9433 (p_return_i): Likewise.
9435 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9437 PR tree-optimization/81428
9438 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
9439 can't be built for those types.
9441 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
9443 Remove stuff dead since r239246.
9445 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
9446 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
9447 (avr_inform_devices): Remove dead stuff.
9449 2017-07-17 Tamar Christina <tamar.christina@arm.com>
9451 * config/arm/arm_neon.h: Fix softp typo.
9453 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9455 PR tree-optimization/81365
9456 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
9457 aggregate moves onto bb predecessor edges, make sure there are no
9458 loads that could alias the lhs in between the start of bb and the
9461 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
9464 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
9465 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
9466 [LSHIFTRT, outer_code = TRUNCATE]: Same.
9468 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9470 PR tree-optimization/81396
9471 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
9472 (init_symbolic_number): Initialize it to 1.
9473 (perform_symbolic_merge): Add n_ops from both operands into the new
9475 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
9476 without base_addr as useless if they need more than one operation.
9477 (bswap_replace): Handle !bswap case for NULL base_addr.
9479 2017-07-17 Tom de Vries <tom@codesourcery.com>
9482 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
9485 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
9487 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
9488 conditional builtin define __FIX_LEON3FT_B2BST.
9490 2017-07-17 Daniel Cederman <cederman@gaisler.com>
9492 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
9493 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
9496 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
9498 PR rtl-optimization/81424
9499 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
9500 to remove potential trapping from operands if -fnon-call-exceptions.
9502 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9504 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
9505 profile_proability for scalling.
9506 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
9508 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9510 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
9512 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9514 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
9515 fixpoint arithmetics.
9517 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9519 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
9520 fixpoint arithmetics.
9522 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9524 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
9525 fixpoint arithmetics.
9527 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9529 * profile-count.h (profile_probability::from_reg_br_prob_note,
9530 profile_probability::to_reg_br_prob_note): New functions.
9531 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
9532 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
9533 * predict.c (probability_reliable_p): Update.
9534 (edge_probability_reliable_p): Update.
9535 (br_prob_note_reliable_p): Update.
9536 (invert_br_probabilities): Update.
9537 (add_reg_br_prob_note): New function.
9538 (combine_predictions_for_insn): Update.
9539 * asan.c (asan_clear_shadow): Update.
9540 * cfgbuild.c (compute_outgoing_frequencies): Update.
9541 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
9542 (update_br_prob_note): Update.
9543 (rtl_verify_edges): Update.
9544 (purge_dead_edges): Update.
9545 (fixup_reorder_chain): Update.
9546 * emit-rtl.c (try_split): Update.
9547 * ifcvt.c (cond_exec_process_insns): Update.
9548 (cond_exec_process_if_block): Update.
9549 (dead_or_predicable): Update.
9550 * internal-fn.c (expand_addsub_overflow): Update.
9551 (expand_neg_overflow): Update.
9552 (expand_mul_overflow): Update.
9553 * loop-doloop.c (doloop_modify): Update.
9554 * loop-unroll.c (compare_and_jump_seq): Update.
9555 * optabs.c (emit_cmp_and_jump_insn_1): Update.
9556 * predict.h: Update.
9557 * reorg.c (mostly_true_jump): Update.
9559 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
9560 * config/alpha/alpha.c (emit_unlikely_jump): Update.
9561 * config/arc/arc.c: (emit_unlikely_jump): Update.
9562 * config/arm/arm.c: (emit_unlikely_jump): Update.
9563 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
9564 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
9565 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
9566 (ix86_print_operand): Update.
9567 (ix86_split_fp_branch): Update.
9568 (predict_jump): Update.
9569 * config/ia64/ia64.c (ia64_print_operand): Update.
9570 * config/mmix/mmix.c (mmix_print_operand): Update.
9571 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
9572 (rs6000_expand_split_stack_prologue): Update.
9573 * config/rs6000/rs6000.c: Update.
9574 * config/s390/s390.c (s390_expand_vec_strlen): Update.
9575 (s390_expand_vec_movstr): Update.
9576 (s390_expand_cs_tdsi): Update.
9577 (s390_expand_split_stack_prologue): Update.
9578 * config/sh/sh.c (sh_print_operand): Update.
9579 (expand_cbranchsi4): Update.
9580 (expand_cbranchdi4): Update.
9581 * config/sparc/sparc.c (output_v9branch): Update.
9582 * config/spu/spu.c (get_branch_target): Update.
9583 (ea_load_store_inline): Update.
9584 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
9585 * config/tilepro/tilepro.c: Update.
9587 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
9589 * gimplify.c (mostly_copy_tree_r): Revert latest change.
9590 (gimplify_save_expr): Likewise.
9592 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9594 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
9596 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9598 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
9600 * timevar.def (TV_IPA_FNSUMMARY): Define.
9602 2017-07-16 Daniel Cederman <cederman@gaisler.com>
9604 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
9605 to back store errata sensitive sequence from being generated.
9606 (sqrtdf2_fix): Likewise.
9608 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9610 * tree-ssa-threadupdate.c (compute_path_counts,
9611 update_joiner_offpath_counts): Use profile_probability.
9613 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
9616 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9618 * config/arm/arm-c.c (arm_cpu_builtins): Define
9619 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
9621 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
9623 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9624 array entries to represent __ieee128 versions of the
9625 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
9626 scalar_extract_sig, and scalar_insert_exp built-in functions.
9627 (altivec_resolve_overloaded_builtin): Add special case handling
9628 for the __builtin_scalar_insert_exp function, as represented by
9629 the P9V_BUILTIN_VEC_VSIEDP constant.
9630 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
9631 exponent support for __ieee128 argument.
9632 (VSESQP): Add scalar extract signature support for __ieee128
9634 (VSTDCNQP): Add scalar test negative support for __ieee128
9636 (VSIEQP): Add scalar insert exponent support for __int128 argument
9637 with __ieee128 result.
9638 (VSIEQPF): Add scalar insert exponent support for __ieee128
9639 argument with __ieee128 result.
9640 (VSTDCQP): Add scalar test data class support for __ieee128
9642 (VSTDCNQP): Add overload support for scalar test negative with
9644 (VSTDCQP): Add overload support for scalar test data class
9646 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
9648 (UNSPEC_VSX_SIEXPQP): New constant.
9649 (xsxexpqp): New insn for VSX scalar extract exponent quad
9651 (xsxsigqp): New insn for VSX scalar extract significand quad
9653 (xsiexpqpf): New insn for VSX scalar insert exponent quad
9654 precision with floating point argument.
9655 (xststdcqp): New expand for VSX scalar test data class quad
9657 (xststdcnegqp): New expand for VSX scalar test negative quad
9659 (xststdcqp): New insn to match expansions for VSX scalar test data
9660 class quad precision and VSX scalar test negative quad precision.
9661 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
9662 special case operand checking to enforce that second operand of
9663 VSX scalar test data class with quad precision argument is a 7-bit
9665 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
9666 prototypes and descriptions of __ieee128 versions of
9667 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
9668 scalar_test_data_class, and scalar_test_neg built-in functions.
9670 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9672 PR tree-optimization/81162
9673 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
9674 replace a negate with an add.
9676 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
9678 * doc/invoke.texi (arm/-mcpu): Document +crypto.
9680 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9682 * config/arm/arm-c.c (arm_cpu_builtins): Define
9683 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
9685 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9687 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
9688 (armv8-r): Set ARM Cortex-R52 as default CPU.
9689 * config/arm/arm-tables.opt: Regenerate.
9690 * config/arm/arm-tune.md: Regenerate.
9691 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
9693 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
9694 extension for -mcpu=cortex-r52.
9696 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9698 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
9699 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
9700 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
9701 (fp-armv8): Define it as FP_ARMv8 only.
9702 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
9703 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
9705 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
9706 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
9707 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
9708 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
9709 than TARGET_FPU_ARMV8.
9710 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
9711 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
9712 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
9714 * config/arm/neon.md (neon_vrint): Likewise.
9715 (neon_vcvt): Likewise.
9716 (neon_<fmaxmin_op><mode>): Likewise.
9717 (<fmaxmin><mode>3): Likewise.
9718 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
9719 * config/arm/predicates.md (arm_cond_move_operator): Check against
9720 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
9722 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
9724 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
9727 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9729 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
9730 loop in comment with memset.
9732 2017-07-14 Martin Liska <mliska@suse.cz>
9734 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
9735 * dwarf2out.c (is_java): Remove the function.
9736 (output_pubname): Remove usage of the function.
9737 (lower_bound_default): Remove usage of DW_LANG_Java.
9738 (gen_compile_unit_die): Likewise.
9739 * gcc.c: Remove compiler defaults for .java and .zip files.
9740 * gimple-expr.c (remove_suffix): Change as there's no longer
9741 extension than 4-letter one.
9742 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
9743 (gimplify_save_expr): Likewise.
9744 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
9745 as it's possible even for other languages than Java.
9746 * langhooks.h (struct lang_hooks): Remove Java from a comment.
9747 * lto-opts.c (lto_write_options): Remove reference to Java.
9748 * opts.c (strip_off_ending): Update file extension handling.
9749 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
9750 * tree-eh.c (lower_resx): Likewise.
9751 * tree.c (free_lang_data_in_type): Remove dead code.
9752 (find_decls_types_r): Likewise.
9753 (build_common_builtin_nodes): Remove Java from a comment.
9754 (verify_type): Remove dead code.
9755 * varasm.c (assemble_external): Remove Java from a comment.
9757 2017-07-14 Martin Liska <mliska@suse.cz>
9759 * opts.c (finish_options): Add quotes.
9760 (common_handle_option): Likewise.
9762 2017-07-14 Martin Liska <mliska@suse.cz>
9764 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
9765 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
9767 * dwarf2out.c (lower_bound_default): Do not handle
9769 (gen_compile_unit_die): Likewise.
9770 * gcc.c: Remove default extension binding for GNU Pascal.
9771 * stmt.c: Remove Pascal language from a comment.
9772 * xcoffout.c: Likewise.
9774 2017-07-13 David Malcolm <dmalcolm@redhat.com>
9777 * diagnostic-show-locus.c (fixit_cmp): New function.
9778 (layout::layout): Sort m_fixit_hints.
9779 (column_range::column_range): Assert that the values are valid.
9780 (struct char_span): New struct.
9781 (correction::overwrite): New method.
9782 (struct source_line): New struct.
9783 (line_corrections::add_hint): Add assertions. Reimplement memcpy
9784 calls in terms of classes source_line and char_span, and
9785 correction::overwrite.
9786 (selftest::test_overlapped_fixit_printing_2): New function.
9787 (selftest::diagnostic_show_locus_c_tests): Call it.
9789 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
9791 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
9792 early if there is no lhs.
9794 2017-07-13 Martin Liska <mliska@suse.cz>
9796 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
9797 (gen_reference_type_die): Likewise.
9798 * stor-layout.c: Remove Pascal-related comment.
9800 2017-07-13 Martin Liska <mliska@suse.cz>
9802 * opts.c (finish_options): Add quotes to error messages.
9803 (parse_sanitizer_options): Likewise.
9805 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
9807 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
9809 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
9811 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
9813 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
9815 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
9817 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
9819 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
9822 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
9823 provides the hardware capability bits, define the macro
9824 __BUILTIN_CPU_SUPPORTS__.
9825 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
9826 if GLIBC does not provide the hardware capability bits. Add a
9827 gcc_unreachable call if the built-in cpu function is neither
9828 __builtin_cpu_is nor __builtin_cpu_supports.
9829 (rs6000_get_function_versions_dispatcher): Change the warning
9830 that an old GLIBC is used which does not export the capability
9831 bits to be an error.
9832 * doc/extend.texi (target_clones attribute): Document the
9833 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
9834 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
9835 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
9836 the macros defined by GCC if the newer GLIBC is available.
9838 2017-07-12 Jeff Law <law@redhat.com>
9840 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
9841 remaining includes slightly.
9842 * config/riscv/riscv-builtins.c: Include profile-count.h.
9844 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
9847 * config/avr/avr.c (avr_set_current_function): In diagnostic
9848 messages: Quote keywords and (parts of) identifiers.
9849 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
9852 2017-07-12 Carl Love <cel@us.ibm.com>
9854 * config/rs6000/rs6000-c.c: Add support for built-in functions
9855 vector bool char vec_revb (vector bool char);
9856 vector bool short vec_revb (vector short char);
9857 vector bool int vec_revb (vector bool int);
9858 vector bool long long vec_revb (vector bool long long);
9859 * doc/extend.texi: Update the built-in documentation file for the
9860 new built-in functions.
9862 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9864 * config/s390/s390.md: Remove movcc splitter.
9866 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9868 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
9869 load/store on condition.
9871 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
9874 * config/avr/avr.c (avr_encode_section_info)
9875 [progmem && !TREE_READONLY]: Error if progmem object needs
9878 2017-07-11 Michael Collison <michael.collison@arm.com>
9880 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
9883 2017-07-11 Carl Love <cel@us.ibm.com>
9885 * config/rs6000/rs6000-c.c: Add support for builtins
9886 vector unsigned int vec_parity_lsbb (vector signed int);
9887 vector unsigned int vec_parity_lsbb (vector unsigned int);
9888 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
9889 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
9890 vector unsigned long long vec_parity_lsbb (vector signed long long);
9891 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
9892 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
9893 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
9894 * doc/extend.texi: Update the built-in documentation file for the
9895 new built-in functions.
9897 2017-07-11 David Malcolm <dmalcolm@redhat.com>
9899 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
9900 (layout::m_primary_loc): New field.
9901 (layout::layout): Initialize new field. Move location filtering
9902 logic from here to...
9903 (layout::maybe_add_location_range): ...this new method. Add
9904 support for filtering to just the lines already specified by other
9906 (layout::will_show_line_p): New method.
9907 (gcc_rich_location::add_location_if_nearby): New method.
9908 (selftest::test_add_location_if_nearby): New test function.
9909 (selftest::diagnostic_show_locus_c_tests): Call it.
9910 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
9913 2017-07-11 Tom de Vries <tom@codesourcery.com>
9915 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
9916 (bb_first_real_insn): New function.
9917 (nvptx_single): Add extra initialization of broadcasted condition
9920 2017-07-11 Nathan Sidwell <nathan@acm.org>
9922 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
9924 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
9926 * doc/extend.texi (AVR Function Attributes): Remove weblink to
9927 Binutils doc as TEXI will mess them up.
9928 * doc/invoke.texi (AVR Options): Same here.
9930 2017-07-11 Daniel Cederman <cederman@gaisler.com>
9932 * config/sparc/sparc.opt (mfix-ut700): New option.
9933 (mfix-gr712rc): Likewise.
9934 (sparc_fix_b2bst): New variable.
9935 * doc/invoke.texi (SPARC options): Document them.
9936 (ARM options): Fix warnings.
9937 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
9938 instructions to prevent sequences that can trigger the store-store
9939 errata for certain LEON3FT processors.
9940 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
9941 (sparc_option_override): Set sparc_fix_b2bst appropriately.
9942 * config/sparc/sparc.md (fix_b2bst): New attribute.
9943 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
9945 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
9948 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
9950 (*rsqrtsf2_sse): Ditto.
9952 (div<mode>3): Macroize insn from divdf3 and divsf3
9953 using MODEF mode iterator.
9955 2017-07-10 Martin Sebor <msebor@redhat.com>
9957 PR tree-optimization/80397
9958 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
9959 instead of testing for equality to INTEGER_TYPE.
9961 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
9963 * config.gcc: Remove uclibc from arc target spec.
9965 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
9967 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
9969 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9972 * lto-wrapper.c (remove_option): New function.
9973 (merge_and_complain): Merge PIC/PIE options more realistically.
9975 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
9977 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
9981 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
9982 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
9983 * config.in: Regenerate.
9984 * configure: Regenerate.
9985 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
9986 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
9987 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
9988 (TARGET_GASISR_PROLOGUES): ...target mask.
9989 * common/config/avr/avr-common.c
9990 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
9991 Set -mgas-isr-prologues.
9992 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
9993 INSERT_PASS_BEFORE for it.
9994 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
9995 * config/avr/avr.c (avr_option_override)
9996 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
9997 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
9998 (avr_attribute_table) <no_gccisr>: Add new function attribute.
9999 (avr_set_current_function) <is_no_gccisr>: Init machine field.
10000 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
10002 (make_avr_pass_pre_proep): New function.
10003 (emit_push_sfr) <treg>: Add argument to function and use it
10004 instead of TMP_REG.
10005 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
10006 and set machine->gasisr.yes.
10007 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
10008 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
10009 __gcc_isr.n_pushed to .L__stack_usage.
10010 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
10011 (avr_asm_final_postscan_insn): ...this new static function.
10012 * config/avr/avr.h (machine_function)
10013 <is_no_gccisr, use_L__stack_usage>: New fields.
10014 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
10015 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
10016 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
10017 (gasisr, *gasisr): New expander and insn.
10018 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
10019 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
10020 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
10022 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
10024 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
10027 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
10029 Move jump-tables out of .text again.
10032 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
10033 (ASM_OUTPUT_ADDR_VEC): New function.
10034 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
10035 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
10036 INSN_ADDRESSes as asm comment.
10037 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
10038 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
10039 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
10040 * config/avr/avr.md (*tablejump): Adjust comment.
10041 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
10042 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
10044 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
10045 (avr_output_addr_vec): New proto.
10046 (avr_log_t) <insn_addresses>: New field.
10048 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
10051 * config/i386/i386.c (ix86_function_arg_advance): Set
10052 outgoing_args_on_stack to true if there are outgoing arguments
10054 (ix86_function_arg): Likewise.
10055 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
10056 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
10057 * config/i386/i386.h (machine_function): Add
10058 outgoing_args_on_stack.
10060 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
10062 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
10063 supporting pthreds.
10064 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
10066 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
10068 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
10069 (REAL_H): Remove $(MACHMODE_H).
10070 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
10072 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
10073 $(MACHMODE_H) and double-int.h.
10074 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
10076 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
10079 2017-07-07 Andrew Pinski <apinski@cavium.com>
10081 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
10082 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
10084 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10086 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
10087 Add warning if GCC was not configured to link against a GLIBC that
10088 exports the hardware capability bits.
10089 (make_resolver_func): Make resolver function private and not a
10090 COMDAT function. Create the name with clone_function_name instead
10091 of make_unique_name.
10094 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
10095 correct operand in doing the split.
10097 2017-07-07 Carl Love <cel@us.ibm.com>
10099 * config/rs6000/rs6000-c: Add support for built-in function
10100 vector unsigned short vec_pack_to_short_fp32 (vector float,
10102 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
10103 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
10104 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
10105 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
10106 (convert_4f32_8i16): Add define_expand.
10107 * doc/extend.texi: Update the built-in documentation file for the
10108 new built-in function.
10110 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10112 * config/sparc/m8.md: New file.
10113 * config/sparc/sparc.md: Include m8.md.
10115 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10117 * config/sparc/sparc.opt: New option -mvis4b.
10118 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
10119 (sparc_option_override): Handle VIS4B.
10120 (enum sparc_builtins): Define
10121 SPARC_BUILTIN_DICTUNPACK{8,16,32},
10122 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
10123 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
10124 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
10125 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
10126 (check_constant_argument): New function.
10127 (sparc_vis_init_builtins): Define builtins
10128 __builtin_vis_dictunpack{8,16,32},
10129 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
10130 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
10131 __builtin_vis_fpcmpde{8,16,32}shl and
10132 __builtin_vis_fpcmpur{8,16,32}shl.
10133 (sparc_expand_builtin): Check that the constant operands to
10134 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
10135 constant and in range.
10136 * config/sparc/sparc-c.c (sparc_target_macros): Handle
10138 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
10139 (SPARC_IMM5_P): Likewise.
10140 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
10141 (enabled): Handle vis4b.
10142 (UNSPEC_DICTUNPACK): New unspec.
10143 (UNSPEC_FPCMPSHL): Likewise.
10144 (UNSPEC_FPUCMPSHL): Likewise.
10145 (UNSPEC_FPCMPDESHL): Likewise.
10146 (UNSPEC_FPCMPURSHL): Likewise.
10147 (cpu_feature): New CPU feature `vis4b'.
10148 (dictunpack{8,16,32}): New insns.
10149 (FPCSMODE): New mode iterator.
10150 (fpcscond): New code iterator.
10151 (fpcsucond): Likewise.
10152 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
10153 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
10154 (fpcmpde{8,16,32}{si,di}shl): Likewise.
10155 (fpcmpur{8,16,32}{si,di}shl): Likewise.
10156 * config/sparc/constraints.md: Define constraints `q' for unsigned
10157 2-bit integer constants and `t' for unsigned 5-bit integer
10159 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
10161 (imm5_operand_dictunpack16): Likewise.
10162 (imm5_operand_dictunpack32): Likewise.
10163 (imm2_operand): Likewise.
10164 * doc/invoke.texi (SPARC Options): Document -mvis4b.
10165 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
10166 ditunpack* and fpcmp*shl builtins.
10168 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10170 * config.gcc: Handle m8 in --with-{cpu,tune} options.
10171 * config.in: Add HAVE_AS_SPARC6 define.
10172 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
10174 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
10176 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
10177 (CPP_CPU_SPEC): Handle m8.
10178 (ASM_CPU_SPEC): Likewise.
10179 * config/sparc/sparc-opts.h (enum processor_type): Add
10181 * config/sparc/sparc.c (m8_costs): New struct.
10182 (sparc_option_override): Handle TARGET_CPU_m8.
10183 (sparc32_initialize_trampoline): Likewise.
10184 (sparc64_initialize_trampoline): Likewise.
10185 (sparc_issue_rate): Likewise.
10186 (sparc_register_move_cost): Likewise.
10187 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
10188 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
10189 (ASM_CPU64_DEFAULT_SPEC): Likewise.
10190 (CPP_CPU_SPEC): Handle M8.
10191 (ASM_CPU_SPEC): Likewise.
10192 (AS_M8_FLAG): Define.
10193 * config/sparc/sparc.md: Add m8 to the cpu attribute.
10194 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
10195 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
10197 * configure: Regenerate.
10198 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
10201 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10203 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
10205 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
10206 ("*movdi_insn_sp32"): Do not set v3pipe.
10207 ("*movsi_insn"): Likewise.
10208 ("*movdi_insn_sp64"): Likewise.
10209 ("*movsf_insn"): Likewise.
10210 ("*movdf_insn_sp32"): Likewise.
10211 ("*movdf_insn_sp64"): Likewise.
10212 ("*zero_extendsidi2_insn_sp64"): Likewise.
10213 ("*sign_extendsidi2_insn"): Likewise.
10214 ("*mov<VM32:mode>_insn"): Likewise.
10215 ("*mov<VM64:mode>_insn_sp64"): Likewise.
10216 ("*mov<VM64:mode>_insn_sp32"): Likewise.
10217 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
10218 ("<vlop:code><VL:mode>3"): Likewise.
10219 ("*not_<vlop:code><VL:mode>3"): Likewise.
10220 ("*nand<VL:mode>_vis"): Likewise.
10221 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
10222 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
10223 ("one_cmpl<VL:mode>2"): Likewise.
10224 ("faligndata<VM64:mode>_vis"): Likewise.
10225 ("alignaddrsi_vis"): Likewise.
10226 ("alignaddrdi_vis"): Likweise.
10227 ("alignaddrlsi_vis"): Likewise.
10228 ("alignaddrldi_vis"): Likewise.
10229 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
10230 ("bmaskdi_vis"): Likewise.
10231 ("bmasksi_vis"): Likewise.
10232 ("bshuffle<VM64:mode>_vis"): Likewise.
10233 ("cmask8<P:mode>_vis"): Likewise.
10234 ("cmask16<P:mode>_vis"): Likewise.
10235 ("cmask32<P:mode>_vis"): Likewise.
10236 ("pdistn<P:mode>_vis"): Likewise.
10237 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
10239 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10241 * config/sparc/sparc.md ("subtype"): New insn attribute.
10242 ("*wrgsr_sp64"): Set insn subtype.
10243 ("*rdgsr_sp64"): Likewise.
10244 ("alignaddrsi_vis"): Likewise.
10245 ("alignaddrdi_vis"): Likewise.
10246 ("alignaddrlsi_vis"): Likewise.
10247 ("alignaddrldi_vis"): Likewise.
10248 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
10249 ("fexpand_vis"): Likewise.
10250 ("fpmerge_vis"): Likewise.
10251 ("faligndata<VM64:mode>_vis"): Likewise.
10252 ("bshuffle<VM64:mode>_vis"): Likewise.
10253 ("cmask8<P:mode>_vis"): Likewise.
10254 ("cmask16<P:mode>_vis"): Likewise.
10255 ("cmask32<P:mode>_vis"): Likewise.
10256 ("fchksm16_vis"): Likewise.
10257 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
10258 ("fmean16_vis"): Likewise.
10259 ("fp<plusminus_insn>64_vis"): Likewise.
10260 ("<plusminus_insn>v8qi3"): Likewise.
10261 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
10262 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
10263 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
10264 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
10265 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
10266 ("*movqi_insn"): Likewise.
10267 ("*movhi_insn"): Likewise.
10268 ("*movsi_insn"): Likewise.
10269 ("movsi_pic_gotdata_op"): Likewise.
10270 ("*movdi_insn_sp32"): Likewise.
10271 ("*movdi_insn_sp64"): Likewise.
10272 ("movdi_pic_gotdata_op"): Likewise.
10273 ("*movsf_insn"): Likewise.
10274 ("*movdf_insn_sp32"): Likewise.
10275 ("*movdf_insn_sp64"): Likewise.
10276 ("*zero_extendhisi2_insn"): Likewise.
10277 ("*zero_extendqihi2_insn"): Likewise.
10278 ("*zero_extendqisi2_insn"): Likewise.
10279 ("*zero_extendqidi2_insn"): Likewise.
10280 ("*zero_extendhidi2_insn"): Likewise.
10281 ("*zero_extendsidi2_insn_sp64"): Likewise.
10282 ("ldfsr"): Likewise.
10283 ("prefetch_64"): Likewise.
10284 ("prefetch_32"): Likewise.
10285 ("tie_ld32"): Likewise.
10286 ("tie_ld64"): Likewise.
10287 ("*tldo_ldub_sp32"): Likewise.
10288 ("*tldo_ldub1_sp32"): Likewise.
10289 ("*tldo_ldub2_sp32"): Likewise.
10290 ("*tldo_ldub_sp64"): Likewise.
10291 ("*tldo_ldub1_sp64"): Likewise.
10292 ("*tldo_ldub2_sp64"): Likewise.
10293 ("*tldo_ldub3_sp64"): Likewise.
10294 ("*tldo_lduh_sp32"): Likewise.
10295 ("*tldo_lduh1_sp32"): Likewise.
10296 ("*tldo_lduh_sp64"): Likewise.
10297 ("*tldo_lduh1_sp64"): Likewise.
10298 ("*tldo_lduh2_sp64"): Likewise.
10299 ("*tldo_lduw_sp32"): Likewise.
10300 ("*tldo_lduw_sp64"): Likewise.
10301 ("*tldo_lduw1_sp64"): Likewise.
10302 ("*tldo_ldx_sp64"): Likewise.
10303 ("*mov<VM32:mode>_insn"): Likewise.
10304 ("*mov<VM64:mode>_insn_sp64"): Likewise.
10305 ("*mov<VM64:mode>_insn_sp32"): Likewise.
10307 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10309 * config/sparc/sparc.md ("type"): New insn type viscmp.
10310 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
10312 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
10313 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
10314 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
10315 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
10317 ("n7_vis_logical_11cycle"): Likewise.
10318 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
10319 * config/sparc/niagara2.md ("niag3_vis": Likewise.
10320 * config/sparc/niagara.md ("niag_vis"): Likewise.
10321 * config/sparc/ultra3.md ("us3_fga"): Likewise.
10322 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
10324 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10326 * config/sparc/sparc.md: New instruction type `bmask'.
10327 (bmaskdi_vis): Use the `bmask' type.
10328 (bmasksi_vis): Likewise.
10329 * config/sparc/ultra3.md (us3_array): Likewise.
10330 * config/sparc/niagara7.md (n7_array): Likewise.
10331 * config/sparc/niagara4.md (n4_array): Likewise.
10332 * config/sparc/niagara2.md (niag2_vis): Likewise.
10333 (niag3_vis): Likewise.
10334 * config/sparc/niagara.md (niag_vis): Likewise.
10336 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10338 * ipa-comdats.c: Remove optimize check from gate.
10339 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
10340 for functions not optimized.
10341 (ipa_fn_summary_read): Skip optimize check.
10342 (ipa_fn_summary_write): Likewise.
10343 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
10345 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
10347 (can_inline_edge_p): Check flag_pcc_struct_return for match.
10348 (check_callers): Give up on caller which is not optimized.
10349 (inline_small_functions): Likewise.
10350 (ipa_inline): Do not give up when not optimizing.
10351 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
10352 away unoptimizes cdtors.
10353 (whole_program_function_and_variable_visibility): Do
10354 ipa_discover_readonly_nonaddressable_vars in LTO mode.
10355 * ipa.c (process_references): Do not check optimize.
10356 (symbol_table::remove_unreachable_nodes): Update optimize check.
10357 (set_writeonly_bit): Update optimize check.
10358 (pass_ipa_cdtor_merge::gate): Do not check optimize.
10359 (pass_ipa_single_use::gate): Remove.
10361 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10363 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
10364 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
10365 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
10366 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
10367 permute_load, permute_store, adjust_extract, adjust_splat,
10368 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
10369 replace_swap_with_copy, dump_swap_insn_table,
10370 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
10371 recombine_lvx_pattern, recombine_stvx_pattern,
10372 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
10373 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
10374 to file rs6000-p8swap.c.
10375 * config/rs6000/rs6000-p8swap.c: New file.
10376 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
10377 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
10378 and rs6000*-*-* targets.
10380 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10382 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
10384 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10386 * lto-wrapper.c (merge_and_complain): Do not merge
10387 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
10388 fsigned_zeros, ftrapping_math, fwrapv.
10389 (append_compiler_options): Do not track these options.
10390 (append_linker_options): Likewie
10392 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10394 * cgraphunit.c (cgraph_node::finalize_function): When
10395 !flag_toplevel_reorde set no_reorder flag.
10396 (varpool_node::finalize_decl): Likewise.
10397 (symbol_table::compile): Drop no toplevel reorder path.
10399 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10401 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
10402 edges; zero probability is not better than uninitialized.
10404 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
10406 * asan.h (asan_sanitize_allocas_p): Declare.
10407 * asan.c (asan_sanitize_allocas_p): New function.
10408 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
10409 (handle_builtin_alloca): Likewise.
10410 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
10411 if !asan_sanitize_allocas_p.
10412 * params.def (asan-instrument-allocas): Add new option.
10413 * params.h (ASAN_PROTECT_ALLOCAS): Define.
10414 * opts.c (common_handle_option): Disable allocas sanitization for
10417 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
10419 * asan.c: Include gimple-fold.h.
10420 (get_last_alloca_addr): New function.
10421 (handle_builtin_stackrestore): Likewise.
10422 (handle_builtin_alloca): Likewise.
10423 (asan_emit_allocas_unpoison): Likewise.
10424 (get_mem_refs_of_builtin_call): Add new parameter, remove const
10425 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
10426 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
10427 (instrument_builtin_call): Pass gimple iterator to
10428 get_mem_refs_of_builtin_call.
10429 (last_alloca_addr): New global.
10430 * asan.h (asan_emit_allocas_unpoison): Declare.
10431 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
10432 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
10433 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
10434 if function calls alloca.
10435 * gimple-fold.c (replace_call_with_value): Remove static keyword.
10436 * gimple-fold.h (replace_call_with_value): Declare.
10437 * internal-fn.c: Include asan.h.
10438 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
10439 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
10441 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10443 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
10444 (C_SELFTEST_FLAGS): New.
10445 (CPP_SELFTEST_FLAGS): New.
10446 (SELFTEST_DEPS): New, from deps of s-selftest.
10447 (C_SELFTEST_DEPS): New, from deps of s-selftest.
10448 (CPP_SELFTEST_DEPS): New.
10449 (selftest): Add dependency on s-selftest-c++.
10450 (s-selftest): Rename to...
10451 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
10452 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
10453 than SELFTEST_FLAGS.
10454 (selftest-gdb): Rename to...
10455 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
10457 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
10458 (selftest-valgrind): Rename to...
10459 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
10461 (selftest-valgrind): Reintroduce as an alias for
10462 selftest-c-valgrind.
10463 (s-selftest-c++): New.
10464 (selftest-c++-gdb): New.
10465 (selftest-c++-valgrind): New.
10467 2017-07-06 Olivier Hainque <hainque@adacore.com>
10469 * gcc.c (process_command): When deciding if undefined variables
10470 should be ignored when processing specs, accept "gcc -v" as well.
10472 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10474 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
10475 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
10477 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
10479 * config/arm/arm-cpus.in (armv8-r): Add new entry.
10480 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
10481 * config/arm/arm-tables.opt: Regenerate.
10482 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
10484 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
10486 2017-07-06 Carl Love <cel@us.ibm.com>
10488 * ChangeLog: Clean up from mid air collision
10490 2017-07-06 Carl Love <cel@us.ibm.com>
10492 * config/rs6000/rs6000-c.c: Add support for built-in functions
10493 vector signed int vec_subc (vector signed int, vector signed int);
10494 vector signed __int128 vec_subc (vector signed __int128,
10495 vector signed __int128);
10496 vector unsigned __int128 vec_subc (vector unsigned __int128,
10497 vector unsigned __int128);
10498 vector signed int vec_sube (vector signed int, vector signed int,
10499 vector signed int);
10500 vector unsigned int vec_sube (vector unsigned int,
10501 vector unsigned int,
10502 vector unsigned int);
10503 vector signed __int128 vec_sube (vector signed __int128,
10504 vector signed __int128,
10505 vector signed__int128);
10506 vector unsigned __int128 vec_sube (vector unsigned __int128,
10507 vector unsigned __int128,
10508 vector unsigned __int128);
10509 vector signed int vec_subec (vector signed int, vector signed int,
10510 vector signed int);
10511 vector unsigned int vec_subec (vector unsigned int,
10512 vector unsigned int,
10513 vector unsigned int);
10514 vector signed __int128 vec_subec (vector signed __int128,
10515 vector signed __int128,
10516 vector signed__int128);
10517 vector unsigned __int128 vec_subec (vector unsigned __int128,
10518 vector unsigned __int128,
10519 vector unsigned __int128);
10520 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
10521 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
10522 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
10523 BU_ALTIVEC_OVERLOAD_X definitions.
10524 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
10525 * doc/extend.texi: Update the built-in documentation file for the new
10526 built-in functions.
10528 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10531 * diagnostic-show-locus.c (layout::layout): Use start and finish
10532 spelling location for the start and finish of each range.
10533 * genmatch.c (linemap_client_expand_location_to_spelling_point):
10534 Add unused aspect param.
10535 * input.c (expand_location_1): Add "aspect" param, and use it
10536 to access the correct part of the location.
10537 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
10539 (expand_location_to_spelling_point): Likewise.
10540 (linemap_client_expand_location_to_spelling_point): Add "aspect"
10541 param, and pass it to expand_location_1.
10543 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
10545 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
10546 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
10547 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
10548 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
10549 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
10550 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
10551 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
10552 _mm_maskz_getmant_ss): New intrinsics.
10553 (__builtin_ia32_getexpss128_mask): Changed to ...
10554 __builtin_ia32_getexpss128_round ... this.
10555 (__builtin_ia32_getexpsd128_mask): Changed to ...
10556 __builtin_ia32_getexpsd128_round ... this.
10557 * config/i386/i386-builtin-types.def
10558 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
10559 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
10560 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
10561 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
10562 __builtin_ia32_getmantss_mask_round): New builtins.
10563 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
10564 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
10565 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
10566 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
10567 * config/i386/sse.md
10568 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
10569 avx512f_sgetexp<mode><mask_scalar_name>
10570 <round_saeonly_scalar_name> ... this.
10571 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
10572 %0, %1, %2<round_saeonly_op3>}): Changed to ...
10573 vgetexp<ssescalarmodesuffix>
10574 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10575 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
10576 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
10577 avx512f_vgetmant<mode><mask_scalar_name>
10578 <round_saeonly_scalar_name> ... this.
10579 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
10580 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
10581 vgetmant<ssescalarmodesuffix>
10582 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
10583 %0<mask_scalar_operand4>, %1, %2
10584 <round_saeonly_scalar_mask_op4>, %3} ... this.
10585 * config/i386/subst.md (mask_scalar_operand4,
10586 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
10587 round_saeonly_scalar_nimm_predicate): New subst attributes.
10589 2017-07-06 Julia Koval <julia.koval@intel.com>
10591 * config/i386/i386.c (ix86_erase_embedded_rounding):
10592 Remove code for old rounding pattern.
10594 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
10596 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
10598 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
10600 * doc/sourcebuild.texi (Test Directives, Variants of
10601 dg-require-support): Add documentation for dg-require-stack-check.
10603 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
10605 * config/i386/subst.md (mask_scalar, round_scalar,
10606 round_saeonly_scalar): New meta-templates.
10607 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
10608 round_scalar_mask_operand3, round_scalar_mask_op3,
10609 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
10610 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
10611 round_saeonly_scalar_constraint,
10612 round_saeonly_scalar_prefix): New subst attribute.
10613 * config/i386/sse.md
10614 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
10615 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
10616 <round_scalar_name> ... this.
10617 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
10618 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
10619 <round_scalar_name> ... this.
10620 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
10621 <sse>_vm<code><mode>3<mask_scalar_name>
10622 <round_saeonly_scalar_name> ... this.
10623 (v<plusminus_mnemonic><ssescalarmodesuffix>
10624 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
10625 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
10626 v<plusminus_mnemonic><ssescalarmodesuffix>
10627 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10628 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
10629 (v<multdiv_mnemonic><ssescalarmodesuffix>
10630 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
10631 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
10632 v<multdiv_mnemonic><ssescalarmodesuffix>
10633 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10634 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
10635 (v<maxmin_float><ssescalarmodesuffix>
10636 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
10637 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
10638 v<maxmin_float><ssescalarmodesuffix>
10639 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10640 %0<mask_scalar_operand3>, %1, %<iptr>2
10641 <round_saeonly_scalar_mask_op3>} ... this.
10643 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
10645 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
10646 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
10648 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
10649 Alan Hayward <alan.hayward@arm.com>
10650 David Sherwood <david.sherwood@arm.com>
10652 * combine.c (simplify_if_then_else): Remove "enum" before
10654 * compare-elim.c (can_eliminate_compare): Likewise.
10655 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
10657 (aarch64_lookup_simd_builtin_type): Likewise.
10658 (aarch64_simd_builtin_type): Likewise.
10659 (aarch64_init_simd_builtin_types): Likewise.
10660 (aarch64_simd_expand_args): Likewise.
10661 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
10663 (aarch64_reverse_mask): Likewise.
10664 (aarch64_simd_emit_reg_reg_move): Likewise.
10665 (aarch64_gen_adjusted_ldpstp): Likewise.
10666 (aarch64_ccmp_mode_to_code): Likewise.
10667 (aarch64_operands_ok_for_ldpstp): Likewise.
10668 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10669 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
10671 (aarch64_min_divisions_for_recip_mul): Likewise.
10672 (aarch64_reassociation_width): Likewise.
10673 (aarch64_get_condition_code_1): Likewise.
10674 (aarch64_simd_emit_reg_reg_move): Likewise.
10675 (aarch64_simd_attr_length_rglist): Likewise.
10676 (aarch64_reverse_mask): Likewise.
10677 (aarch64_operands_ok_for_ldpstp): Likewise.
10678 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10679 (aarch64_gen_adjusted_ldpstp): Likewise.
10680 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
10682 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
10683 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
10684 (arm_lookup_simd_builtin_type): Likewise.
10685 (arm_simd_builtin_type): Likewise.
10686 (arm_init_simd_builtin_types): Likewise.
10687 (arm_expand_builtin_args): Likewise.
10688 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
10689 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
10690 (ft32_setup_incoming_varargs): Likewise.
10691 (ft32_function_arg): Likewise.
10692 (ft32_function_arg_advance): Likewise.
10693 (ft32_pass_by_reference): Likewise.
10694 (ft32_arg_partial_bytes): Likewise.
10695 (ft32_valid_pointer_mode): Likewise.
10696 (ft32_addr_space_pointer_mode): Likewise.
10697 (ft32_addr_space_legitimate_address_p): Likewise.
10698 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
10700 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
10701 (ix86_emit_outlined_ms2sysv_restore): Likewise.
10702 (iamcu_alignment): Likewise.
10703 (canonicalize_vector_int_perm): Likewise.
10704 (ix86_noce_conversion_profitable_p): Likewise.
10705 (ix86_mpx_bound_mode): Likewise.
10706 (ix86_operands_ok_for_move_multiple): Likewise.
10707 * config/microblaze/microblaze-protos.h
10708 (microblaze_expand_conditional_branch_reg): Likewise.
10709 * config/microblaze/microblaze.c
10710 (microblaze_expand_conditional_branch_reg): Likewise.
10711 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
10713 (rs6000_reassociation_width): Likewise.
10714 (rs6000_invalid_binary_op): Likewise.
10715 (fusion_p9_p): Likewise.
10716 (emit_fusion_p9_load): Likewise.
10717 (emit_fusion_p9_store): Likewise.
10718 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
10720 (riscv_hard_regno_mode_ok_p): Likewise.
10721 (riscv_address_insns): Likewise.
10722 (riscv_split_symbol): Likewise.
10723 (riscv_legitimize_move): Likewise.
10724 (riscv_function_value): Likewise.
10725 (riscv_hard_regno_nregs): Likewise.
10726 (riscv_expand_builtin): Likewise.
10727 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
10728 (riscv_build_integer): Likewise.
10729 (riscv_split_integer): Likewise.
10730 (riscv_legitimate_constant_p): Likewise.
10731 (riscv_cannot_force_const_mem): Likewise.
10732 (riscv_regno_mode_ok_for_base_p): Likewise.
10733 (riscv_valid_base_register_p): Likewise.
10734 (riscv_valid_offset_p): Likewise.
10735 (riscv_valid_lo_sum_p): Likewise.
10736 (riscv_classify_address): Likewise.
10737 (riscv_legitimate_address_p): Likewise.
10738 (riscv_address_insns): Likewise.
10739 (riscv_load_store_insns): Likewise.
10740 (riscv_force_binary): Likewise.
10741 (riscv_split_symbol): Likewise.
10742 (riscv_force_address): Likewise.
10743 (riscv_legitimize_address): Likewise.
10744 (riscv_move_integer): Likewise.
10745 (riscv_legitimize_const_move): Likewise.
10746 (riscv_legitimize_move): Likewise.
10747 (riscv_address_cost): Likewise.
10748 (riscv_subword): Likewise.
10749 (riscv_output_move): Likewise.
10750 (riscv_canonicalize_int_order_test): Likewise.
10751 (riscv_emit_int_order_test): Likewise.
10752 (riscv_function_arg_boundary): Likewise.
10753 (riscv_pass_mode_in_fpr_p): Likewise.
10754 (riscv_pass_fpr_single): Likewise.
10755 (riscv_pass_fpr_pair): Likewise.
10756 (riscv_get_arg_info): Likewise.
10757 (riscv_function_arg): Likewise.
10758 (riscv_function_arg_advance): Likewise.
10759 (riscv_arg_partial_bytes): Likewise.
10760 (riscv_function_value): Likewise.
10761 (riscv_pass_by_reference): Likewise.
10762 (riscv_setup_incoming_varargs): Likewise.
10763 (riscv_print_operand): Likewise.
10764 (riscv_elf_select_rtx_section): Likewise.
10765 (riscv_save_restore_reg): Likewise.
10766 (riscv_for_each_saved_reg): Likewise.
10767 (riscv_register_move_cost): Likewise.
10768 (riscv_hard_regno_mode_ok_p): Likewise.
10769 (riscv_hard_regno_nregs): Likewise.
10770 (riscv_class_max_nregs): Likewise.
10771 (riscv_memory_move_cost): Likewise.
10772 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
10773 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
10774 (rl78_addr_space_address_mode): Likewise.
10775 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10777 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
10778 (rs6000_reassociation_width): Likewise.
10779 (rs6000_invalid_binary_op): Likewise.
10780 (fusion_p9_p): Likewise.
10781 (emit_fusion_p9_load): Likewise.
10782 (emit_fusion_p9_store): Likewise.
10783 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
10784 (ok_for_simple_move_operands): Likewise.
10785 (ok_for_simple_move_strict_operands): Likewise.
10786 (ok_for_simple_arith_logic_operands): Likewise.
10787 (visium_legitimize_reload_address): Likewise.
10788 (visium_select_cc_mode): Likewise.
10789 (output_cbranch): Likewise.
10790 (visium_split_double_move): Likewise.
10791 (visium_expand_copysign): Likewise.
10792 (visium_expand_int_cstore): Likewise.
10793 (visium_expand_fp_cstore): Likewise.
10794 * config/visium/visium.c (visium_pass_by_reference): Likewise.
10795 (visium_function_arg): Likewise.
10796 (visium_function_arg_advance): Likewise.
10797 (visium_libcall_value): Likewise.
10798 (visium_setup_incoming_varargs): Likewise.
10799 (visium_legitimate_constant_p): Likewise.
10800 (visium_legitimate_address_p): Likewise.
10801 (visium_legitimize_address): Likewise.
10802 (visium_secondary_reload): Likewise.
10803 (visium_register_move_cost): Likewise.
10804 (visium_memory_move_cost): Likewise.
10805 (prepare_move_operands): Likewise.
10806 (ok_for_simple_move_operands): Likewise.
10807 (ok_for_simple_move_strict_operands): Likewise.
10808 (ok_for_simple_arith_logic_operands): Likewise.
10809 (visium_function_value_1): Likewise.
10810 (rtx_ok_for_offset_p): Likewise.
10811 (visium_legitimize_reload_address): Likewise.
10812 (visium_split_double_move): Likewise.
10813 (visium_expand_copysign): Likewise.
10814 (visium_expand_int_cstore): Likewise.
10815 (visium_expand_fp_cstore): Likewise.
10816 (visium_split_cstore): Likewise.
10817 (visium_select_cc_mode): Likewise.
10818 (visium_split_cbranch): Likewise.
10819 (output_cbranch): Likewise.
10820 (visium_print_operand_address): Likewise.
10821 * expmed.c (flip_storage_order): Likewise.
10822 * expmed.h (emit_cstore): Likewise.
10823 (flip_storage_order): Likewise.
10824 * genrecog.c (validate_pattern): Likewise.
10825 * hsa-gen.c (gen_hsa_addr): Likewise.
10826 * internal-fn.c (expand_arith_overflow): Likewise.
10827 * ira-color.c (allocno_copy_cost_saving): Likewise.
10828 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10829 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
10830 (process_invariant_for_inheritance): Likewise.
10831 * lra-eliminations.c (move_plus_up): Likewise.
10832 * omp-low.c (lower_oacc_reductions): Likewise.
10833 * simplify-rtx.c (simplify_subreg): Likewise.
10834 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
10835 (TARGET_CHKP_BOUND_MODE): Likewise..
10836 * targhooks.c (default_chkp_bound_mode): Likewise.
10837 (default_setup_incoming_vararg_bounds): Likewise.
10838 * targhooks.h (default_chkp_bound_mode): Likewise.
10839 (default_setup_incoming_vararg_bounds): Likewise.
10840 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
10841 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
10842 (have_whole_vector_shift): Likewise.
10843 * tree-vect-stmts.c (vectorizable_load): Likewise.
10844 * doc/tm.texi: Regenerate.
10846 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
10848 Graceful degrade if Binutils PR21472 is not available.
10851 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
10852 .rodata in flash test fails.
10853 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
10854 * confgure: Regenerate.
10855 * config.in: Regenerate.
10856 * config/avr/avr.c (avr_asm_named_section)
10857 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
10858 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
10859 (avr_asm_init_sections): Same.
10861 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10863 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
10864 (fma<VH:mode>4_intrinsic): Likewise.
10865 (*fmsub<VCVTF:mode>4): Likewise.
10866 (*fmsub<VH:mode>4_intrinsic): Likewise.
10868 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
10871 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
10872 Don't depend on "optimize > 0".
10873 (out_movhi_r_mr, out_movqi_mr_r): Same.
10874 (out_movhi_mr_r, out_movqi_r_mr): Same.
10875 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
10876 io_address_operand on "optimize > 0".
10878 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10880 * tree-loop-distribution.c: Add general explanantion on the pass.
10881 (generate_loops_for_partition): Mark distributed loop.
10882 (pg_add_dependence_edges): New parameter. Handle alias data
10883 dependence specially and record it in the parameter if asked.
10884 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
10885 (init_partition_graph_vertices, add_partition_graph_edge): New.
10886 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
10887 (free_partition_graph_vdata, build_partition_graph): New.
10888 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
10889 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
10890 (data_ref_segment_size, latch_dominated_by_data_ref): New.
10891 (compute_alias_check_pairs, version_loop_by_alias_check): New.
10892 (version_for_distribution_p, finalize_partitions): New.
10893 (distribute_loop): Handle alias data dependence specially. Factor
10894 out loop fusion code as functions and call these functions.
10896 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10898 * tree-loop-distribution.c (classify_partition): New parameter and
10899 better handle reduction statement.
10900 (rdg_build_partitions): Revise comment.
10901 (distribute_loop): Compute statements in all partitions and pass it
10902 to classify_partition.
10904 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10906 * tree-loop-distribution.c (enum partition_type): New.
10907 (struct partition): New field type.
10908 (partition_merge_into): Add parameter. Update partition type.
10909 (data_dep_in_cycle_p, update_type_for_merge): New functions.
10910 (build_rdg_partition_for_vertex): Compute partition type.
10911 (rdg_build_partitions): Dump partition type.
10912 (distribute_loop): Update calls to partition_merge_into.
10914 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10916 * tree-loop-distribution.c (struct ddr_hasher): New.
10917 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
10919 (classify_partition): Call get_data_dependence.
10920 (pg_add_dependence_edges): Ditto.
10921 (distribute_loop): Release data dependence hash table.
10923 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10925 * tree-loop-distribution.c (ref_base_address): Delete.
10926 (similar_memory_accesses): Rename ...
10927 (share_memory_accesses): ... to this. Check if partitions access
10928 the same memory reference.
10929 (distribute_loop): Call share_memory_accesses.
10931 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10933 * tree-loop-distribution.c (struct partition): New field recording
10934 its data reference.
10935 (partition_alloc, partition_free): Init and release data refs.
10936 (partition_merge_into): Merge data refs.
10937 (build_rdg_partition_for_vertex): Collect data refs for partition.
10938 (pg_add_dependence_edges): Change parameters from vector to bitmap.
10940 (distribute_loop): Remve data refs from vertice data of partition
10943 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10945 * tree-loop-distribution.c (params.h): Include header file.
10946 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
10947 (datarefs_vec): New global var.
10948 (create_rdg_vertices): Use datarefs_vec directly.
10949 (free_rdg): Don't free data references.
10950 (build_rdg): Update use. Don't free data references.
10951 (distribute_loop): Compute global variable for data references.
10952 Bail out if there are too many data references.
10954 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10956 * tree-loop-distribution.c (loop_nest): New global var.
10957 (build_rdg): Use loop directly, rather than loop nest.
10958 (pg_add_dependence_edges): Remove loop nest parameter. Use global
10960 (distribute_loop): Compute global variable loop nest. Update use.
10962 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10964 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
10965 (partition_merge_into): New parameter. Dump reason for fusion.
10966 (distribute_loop): Update use of partition_merge_into.
10968 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10970 * tree-loop-distribution.c (bb_top_order_index): New.
10971 (bb_top_order_index_size, bb_top_order_cmp): New.
10972 (stmts_from_loop): Use topological order.
10973 (pass_loop_distribution::execute): Compute and release topological
10974 order for basic blocks.
10976 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10978 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
10981 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10983 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
10984 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
10985 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
10986 * internal-fn.def (LOOP_DIST_ALIAS): New.
10987 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
10988 (fold_loop_internal_call): ... this.
10989 (vect_loop_dist_alias_call): New function.
10990 (set_uid_loop_bbs): Call fold_loop_internal_call.
10991 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
10994 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
10997 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
10998 Require dead FLAGS_REG at the beginning of a peephole.
11000 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
11003 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
11004 arguments in the call to __builtin_ia32_sbb_u32.
11005 (_subborrow_u64): Swap _X and _Y arguments in the call to
11006 __builtin_ia32_sbb_u64.
11008 2017-07-04 Jakub Jelinek <jakub@redhat.com>
11011 * tree-vrp.c (compare_assert_loc): Turn into a function template
11012 with stable template parameter. Only test if a->e is NULL,
11013 !a->e == !b->e has been verified already. Use e == NULL or
11014 e != NULL instead of e or ! e tests. If stable is true, don't use
11015 iterative_hash_expr, on the other side allow a or b or both NULL
11016 and sort the NULLs last.
11017 (process_assert_insertions): Sort using compare_assert_loc<false>
11018 instead of compare_assert_loc, later sort using
11019 compare_assert_loc<true> before calling process_assert_insertions_for
11020 in a loop. Use break instead of continue once seen NULL pointer.
11022 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
11024 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
11025 Cortex-R7 and Cortex-R8 processors.
11027 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
11029 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
11030 uninitialized while src is not.
11032 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
11034 * common/config/arm/arm-common.c: Adjust include path for
11036 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
11037 (arm-cpu.h): Create in build directory. Adjust dependency rules.
11038 (arm-cpu-data.h): Likewise.
11039 (arm-cpu-cdata.h): Likewise.
11040 * config/arm/arm-cpu.h: Delete.
11041 * config/arm/arm-cpu-cdata.h: Delete.
11042 * config/arm/arm-cpu-data.h: Delete.
11044 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
11046 * config/arm/arm-cpus.in (cortex-a55): New.
11047 (cortex-a75): Likewise.
11048 (cortex-a75.cortex-a55): Likewise.
11049 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
11051 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
11052 * config/arm/arm-cpu-cdata.h: Regenerate.
11053 * config/arm/arm-cpu-data.h: Regenerate.
11054 * config/arm/arm-cpu.h: Regenerate.
11055 * config/arm/arm-tables.opt: Regenerate.
11056 * config/arm/arm-tune.md: Regenerate.
11058 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
11060 * haifa-sched.c (sched_create_recovery_edges): Update profile.
11062 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
11064 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
11067 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
11069 PR tree-optimization/81292
11070 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
11071 full_string_p, also call adjust_related_strinfos if the adjustment
11072 is simple, otherwise invalidate related strinfos.
11074 2017-07-04 Martin Liska <mliska@suse.cz>
11077 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
11078 newly created variable as DECL_IGNORED_P.
11080 2017-07-04 Martin Liska <mliska@suse.cz>
11083 * ipa-inline.c (inline_small_functions):
11084 Use xstrdup_for_dump.
11086 2017-07-04 Tom de Vries <tom@codesourcery.com>
11088 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
11090 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
11093 * config/darwin.c (darwin_function_switched_text_sections):
11096 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
11098 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
11100 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
11102 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
11104 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11106 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
11107 min_profitable_iters, and th as inclusive lower bounds.
11108 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
11109 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
11110 for min_profitable_iters and min_profitable_estimate.
11111 (vect_transform_loop): Treat th as an inclusive lower bound.
11112 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
11114 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
11117 * config/darwin.c (darwin_function_switched_text_sections):
11118 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
11119 in two pieces, and suppress the use of buf.
11121 2017-07-03 Nathan Sidwell <nathan@acm.org>
11123 * hash-table.h (hash_table_mod1): Fix indentation.
11125 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11127 PR middle-end/81290
11128 * predict.c (force_edge_cold): Be more careful about propagation
11130 * profile-count.h (profile_probability::guessed,
11131 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
11133 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
11135 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
11137 * doc/invoke.texi (rcpc architecture extension): Document it.
11139 2017-07-03 Richard Biener <rguenther@suse.de>
11141 PR tree-optimization/60510
11142 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
11143 the scalar reduction PHI and use it.
11144 (vectorizable_reduction): Properly guard the single_defuse_cycle
11145 path for non-SLP reduction chains where we cannot use it.
11146 Rework reduc_def/index and vector type deduction. Rework
11147 vector operand gathering during reduction op code-gen.
11148 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
11149 chains dissolve the chain and leave it to non-SLP reduction
11152 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11154 * tree-data-ref.h (dr_alignment): Declare.
11155 * tree-data-ref.c (dr_alignment): New function.
11156 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
11157 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
11159 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
11161 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11163 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
11164 and base_misalignment fields.
11165 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
11166 * tree-data-ref.c: Include builtins.h.
11167 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
11168 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
11169 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
11170 * tree-vect-data-refs.c: Include tree-cfg.h.
11171 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
11172 fields instead of calculating an alignment here.
11173 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
11174 innermost_loop_behavior fields.
11176 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11178 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
11180 (DR_STEP_ALIGNMENT): New macro.
11181 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
11182 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
11183 (create_data_ref): Print it.
11184 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
11185 to tell whether the step preserves vector (mis)alignment.
11186 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
11187 Move the check for an integer step and generalise to all INTEGER_CST.
11188 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
11189 Print the outer step alignment.
11191 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11193 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
11194 with offset_alignment.
11195 (DR_ALIGNED_TO): Delete.
11196 (DR_OFFSET_ALIGNMENT): New macro.
11197 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
11198 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
11199 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
11200 (create_data_ref): Likewise.
11201 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
11202 (vect_analyze_data_refs): Likewise.
11203 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
11204 creating dummy innermost behavior.
11206 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11208 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
11209 with a "innermost_loop_behavior *" and refeence tree.
11210 * tree-data-ref.c (dr_analyze_innermost): Likewise.
11211 (create_data_ref): Update call accordingly.
11212 * tree-predcom.c (find_looparound_phi): Likewise.
11214 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11216 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
11217 fields with dr_wrt_vec_loop.
11218 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
11219 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
11220 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
11221 (vect_dr_behavior): New function.
11222 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
11223 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
11224 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
11225 track whether the step preserves the misalignment.
11226 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
11227 Use vect_dr_behavior.
11228 (vect_setup_realignment): Update call accordingly.
11229 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
11230 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
11231 call to vect_create_addr_base_for_vector_ref.
11232 (vect_create_cond_for_align_checks): Likewise.
11233 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
11234 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
11235 (vect_recog_mask_conversion_pattern): Likewise.
11236 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
11237 (new_stmt_vec_info): Remove redundant zeroing.
11239 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
11241 * common/config/arm/arm-common.c (arm_be8_option): New function.
11242 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
11243 (ISA_ARMv6): Add isa_bit_be8.
11244 * config/arm/arm.h (arm_be8_option): Add prototype.
11245 (BE8_SPEC_FUNCTION): New define.
11246 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
11247 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
11248 (mlittle-endian): Similarly.
11249 (mbe8, mbe32): New options.
11250 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
11251 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
11253 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11255 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
11257 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11259 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
11260 (cleanup_tree_cfg_bb): Use it.
11261 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
11263 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
11265 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11268 * loop-doloop.c (add_test): Update profile.
11270 2017-07-03 Martin Liska <mliska@suse.cz>
11273 * sanopt.c (rewrite_usage_of_param): New function.
11274 (sanitize_rewrite_addressable_params): Likewise.
11275 (pass_sanopt::execute): Call rewrite_usage_of_param.
11277 2017-07-03 Richard Biener <rguenther@suse.de>
11279 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
11280 back to using VIEW_CONVERT_EXPR.
11282 2017-07-03 Martin Liska <mliska@suse.cz>
11285 * doc/extend.texi: Document when a resolver function is
11286 generated for target_clones.
11288 2017-07-03 Martin Liska <mliska@suse.cz>
11290 * asan.c (asan_emit_stack_protection): Unpoison just red zones
11291 and shadow memory of auto variables which are subject of
11292 use-after-scope sanitization.
11293 (asan_expand_mark_ifn): Add do set only when is_poison.
11295 2016-07-03 Richard Biener <rguenther@suse.de>
11297 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
11299 (vect_force_simple_reduction): Record reduction def -> phi mapping.
11300 (vectorizable_reduction): Perform reduction PHI creation when
11301 visiting a reduction PHI and adjust and simplify code generation
11302 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
11303 (vect_transform_loop): Visit reduction PHIs.
11304 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
11305 defs into the SLP tree.
11306 (vect_build_slp_tree): Reduction defs terminate the recursion.
11307 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
11309 (vect_get_vec_defs_for_stmt_copy): Export.
11310 (vect_get_vec_defs): Likewise.
11311 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
11313 (vect_get_vec_defs_for_stmt_copy): Declare.
11314 (vect_get_vec_defs): Likewise.
11316 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11318 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
11319 parameter with a "loop" parameter and use it instead of the
11320 loop containing DR_STMT. Don't check simple_iv when doing
11321 BB analysis. Describe the two analysis modes in the comment.
11323 2017-07-03 Tom de Vries <tom@codesourcery.com>
11325 PR tree-optimization/69468
11326 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
11327 (find_same_succ_bb): Handle ignore_edge_flags.
11329 2017-07-03 Tom de Vries <tom@codesourcery.com>
11331 PR tree-optimization/81192
11332 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
11334 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
11336 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
11338 2017-07-03 Tom de Vries <tom@codesourcery.com>
11340 PR tree-optimization/81192
11341 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
11342 BB_SAME_SUCC (bb) == NULL.
11344 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11346 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
11349 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11351 * dumpfile.c: Include profile-count.h
11352 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
11354 (insert_cond_bb): Update profile.
11355 * tree-cfg.h (insert_cond_bb): Update prototype.
11356 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
11357 * tree-dump.c: Do not include tree-cfg.
11359 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11361 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
11363 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11365 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
11368 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11370 * tree-complex.c (expand_complex_div_wide): update profile.
11372 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11373 Alan Hayward <alan.hayward@arm.com>
11374 David Sherwood <david.sherwood@arm.com>
11376 * Makefile.in (MACHMODE_H): Remove insn-modes.h
11377 (CORETYPES_H): New define.
11378 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
11379 (insn-modes-inline.h, s-modes-inline-h): New rules.
11380 (generated_files): Add insn-modes-inline.h.
11381 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
11382 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
11383 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
11384 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
11385 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
11386 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
11387 (build/gencodes.o, build/genconditions.o): Likewise.
11388 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
11389 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
11390 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
11391 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
11392 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
11393 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
11394 * coretypes.h: Include everything up to real.h for generators.
11395 Include insn-modes.h first. Include wide-int-print.h after
11396 wide-int.h. Include insn-modes-inline.h and then machmode.h.
11397 * machmode.h: Don't include insn-modes.h here.
11398 * function-tests.c: Remove includes of signop.h, machmode.h,
11399 double-int.h and wide-int.h.
11401 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
11403 * optc-save-gen.awk: Likewise.
11404 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
11405 * godump.c: Remove include of wide-int-print.h.
11406 * pretty-print.h: Likewise.
11407 * wide-int-print.cc: Likewise.
11408 * wide-int.cc: Likewise.
11409 * hash-map-tests.c: Remove include of signop.h.
11410 * hash-set-tests.c: Likewise.
11411 * rtl-tests.c: Likewise.
11412 * mkconfig.sh: Remove include of machmode.h.
11413 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
11415 (emit_insn_modes_inline_h): ...this new function. Emit the code
11416 into an insn-modes-inline.h header file, adding appropriate
11417 include guards and end comments.
11418 (emit_insn_modes_c_header): Remove include of machmode.h.
11419 (emit_min_insn_modes_c_header): Include coretypes.h rather than
11421 (main): Handle -i flag and call emit_insn_modes_inline_h when
11424 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11426 * tree-ssa-strlen.c (strinfo): Rename the length field to
11427 nonzero_chars. Add a full_string_p field.
11428 (compare_nonzero_chars, zero_length_string_p): New functions.
11429 (get_addr_stridx): Add an offset_out parameter.
11430 Use compare_nonzero_chars.
11431 (get_stridx): Update accordingly. Use compare_nonzero_chars.
11432 (new_strinfo): Update after above changes to strinfo.
11433 (set_endptr_and_length): Set full_string_p.
11434 (get_string_length): Update after above changes to strinfo.
11435 (unshare_strinfo): Update call to new_strinfo.
11436 (maybe_invalidate): Likewise.
11437 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
11438 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
11439 as a uhwi instead of an shwi. Update after above changes to
11440 strinfo and new_strinfo.
11441 (zero_length_string): Assert that chainsi contains full strings.
11442 Use zero_length_string_p. Update call to new_strinfo.
11443 (adjust_related_strinfos): Update after above changes to strinfo.
11444 Copy full_string_p from origsi.
11445 (adjust_last_stmt): Use zero_length_string_p.
11446 (handle_builtin_strlen): Update after above changes to strinfo and
11447 new_strinfo. Install the lhs as the string length if the previous
11448 entry didn't describe a full string.
11449 (handle_builtin_strchr): Update after above changes to strinfo
11451 (handle_builtin_strcpy): Likewise.
11452 (handle_builtin_strcat): Likewise.
11453 (handle_builtin_malloc): Likewise.
11454 (handle_pointer_plus): Likewise.
11455 (handle_builtin_memcpy): Likewise. Track nonzero characters
11456 that aren't necessarily followed by a nul terminator.
11457 (handle_char_store): Likewise.
11459 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11461 PR tree-optimization/80769
11462 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
11463 for malloc and calloc. Document the new invariant that all related
11464 strinfos have delayed lengths or none do.
11465 (verify_related_strinfos): Move earlier in file.
11466 (set_endptr_and_length): New function, split out from...
11467 (get_string_length): ...here. Also set the lengths of related
11469 (zero_length_string): Assert that chainsi has known (rather than
11471 (adjust_related_strinfos): Likewise.
11473 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11475 PR tree-optimization/81136
11476 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
11477 assert that two references with the same misalignment have the same
11478 compile-time misalignment if those compile-time misalignments
11481 2017-07-01 Andi Kleen <ak@linux.intel.com>
11483 * print-tree.c (print_node): Print all attributes.
11485 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11487 * cfg.c (scale_bbs_frequencies): New function.
11488 * cfg.h (scale_bbs_frequencies): Declare it.
11489 * cfgloopanal.c (single_likely_exit): Cleanup.
11490 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
11492 (scale_loop_profile): Likewise.
11493 (loop_version): Likewise.
11494 (create_empty_loop_on_edge): Update.
11495 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
11496 scale_loop_frequencies, scale_loop_profile, loopify,
11497 loop_version): Update prototypes.
11498 * modulo-sched.c (sms_schedule): Update.
11499 * predict.c (unlikely_executed_edge_p): Also check probability.
11500 (probably_never_executed_edge_p): Fix typo.
11501 * tree-if-conv.c (version_loop_for_if_conversion): Update.
11502 * tree-parloops.c (gen_parallel_loop): Update.
11503 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
11504 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11505 * tree-ssa-loop-split.c (split_loop): Update.
11506 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
11507 * tree-vect-loop-manip.c (vect_do_peeling): Update.
11508 (vect_loop_versioning): Update.
11509 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11511 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11513 * trans-mem.c (split_bb_make_tm_edge): Update profile.
11515 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11517 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
11518 to keep profile consistent.
11520 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11522 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
11523 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
11524 * profile-count.h (max_safe_multiplier): Make unsigned.
11525 (profile_count::guessed_zero): New.
11527 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11529 * bb-reorder.c (fix_up_crossing_landing_pad,
11530 fix_crossing_conditional_branches): Use make_single_succ_edge
11531 to keep profile consistent.
11533 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11535 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
11538 2017-07-01 Jakub Jelinek <jakub@redhat.com>
11541 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
11542 the right scopes, make sure cond_jump isn't preserved between multiple
11543 iterations. Search for fallthru edge whenever there are 3+ edges and
11544 use find_fallthru_edge for it.
11546 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11548 Patch by Alexander Monakov <amonakov@ispras.ru>
11549 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
11550 probabilities consistently.
11552 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11554 * pa.c (pa_expand_compare_and_swap_loop): Update call of
11555 emit_cmp_and_jump_insns.
11557 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11560 * tree-inline.c (expand_call_inline): Combine profile statuses.
11562 2017-06-30 Andrew Pinski <apinski@cavium.com>
11564 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
11565 fold_stmt returned true.
11567 2017-06-30 Nathan Sidwell <nathan@acm.org>
11569 * ggc.h (empty_string): Delete.
11570 * cfgexpand.c (expand_asm_stmt): Use plain "".
11571 * optabs.c (expand_asm_memory_barrier): Likewise.
11572 * stringpool.c (empty_string): Delete.
11573 (digit_vector, digit_string): Delete.
11574 (ggc_alloc_string): Use plain "", don't optimize single digit
11575 strings. Use ggc_alloc_atomic.
11577 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
11579 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
11580 comparison set and one other set, use the cost of the non-comparison
11583 2017-06-30 Nathan Sidwell <nathan@acm.org>
11585 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
11588 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
11590 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
11591 loops. Remove now unneeded calls to gimple_switch_set_label() that
11592 just set removed labels to NULL_TREE.
11594 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
11596 * tree-ssanames.c (set_range_info_raw): Abstract from ...
11597 (set_range_info): ...here. Only call set_range_info_raw if domain
11599 (set_nonzero_bits): Call set_range_info_raw.
11600 * tree-ssanames.h (set_range_info_raw): New.
11602 2017-06-30 Jakub Jelinek <jakub@redhat.com>
11605 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
11606 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
11607 of nonimmediate_operand and <store_mask_constraint> instead of m
11608 for the input operand. For V8FI iterator, always split if input
11609 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
11610 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
11611 <store_mask_predicate> instead of register_operand and
11612 <store_mask_constraint> instead of v for the input operand. Make
11613 sure both operands aren't MEMs for if not <mask_applied>.
11615 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
11617 * lto-wrapper.c (copy_file) Close both file descriptors before
11620 2017-06-30 Martin Liska <mliska@suse.cz>
11623 * multiple_target.c (create_dispatcher_calls): Make ifunc
11624 also for function that don't have calls or are not referenced.
11626 2017-06-30 Richard Biener <rguenther@suse.de>
11628 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
11629 analyze the first scalar stmt. Move vector type computation
11630 for the BB case here from ...
11631 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
11632 live operation processing in the SLP case properly.
11634 2017-06-30 Richard Biener <rguenther@suse.de>
11636 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
11638 2017-06-30 Martin Liska <mliska@suse.cz>
11641 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
11642 before BUILT_IN_UNWIND_RESUME when ASAN is used.
11644 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
11646 * doc/invoke.texi (AArch64): Add missing options and remove redundant
11649 2017-06-30 Richard Biener <rguenther@suse.de>
11651 PR tree-optimization/81249
11652 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
11653 condition reduction result to original scalar type.
11655 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11657 * profile-count.h (enum profile_quality): Fix typos and whitespace
11660 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11662 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
11663 type for branch probabilities.
11665 2017-06-29 Julian Brown <julian@codesourcery.com>
11666 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11668 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
11669 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
11670 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
11671 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
11673 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11675 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
11676 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
11677 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
11678 CC usage from generic code to here.
11679 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
11680 CC usage into the target macros.
11682 2017-06-29 Maya Rashish <coypu@sdf.org>
11684 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
11687 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11689 * arm/arm-builtins.c: Include profile-count.h
11690 * except.c (sjlj_emit_function_enter): Use
11691 profile_probability::unlikely.
11693 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11695 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
11696 and tocrel_offset be pointer args rather than implicitly using
11698 (legitimate_constant_pool_address_p, rs6000_emit_move,
11699 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
11700 tocrel_offset and use in toc_relative_expr_p call.
11701 (print_operand, print_operand_address): Use static tocrel_base_oac
11702 and tocrel_offset_oac.
11703 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
11706 2017-06-29 Maya Rashish <coypu@sdf.org>
11708 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
11710 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
11712 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
11713 objects, take into account only the alignment of 'op0' and 'mode1' if
11716 2017-06-29 Steve Ellcey <sellcey@cavium.com>
11718 * ccmp.c (ccmp_tree_comparison_p): New function.
11719 (ccmp_candidate_p): Update to use above function.
11720 (get_compare_parts): New function.
11721 (expand_ccmp_next): Update to use new functions.
11722 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
11724 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
11725 take mode as argument.
11726 * ccmp.h (expand_ccmp_expr): Add mode as argument.
11727 * expr.c (expand_expr_real_1): Pass mode as argument.
11729 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
11731 * combine.c (combine_instructions): Print insns to dump_file, together
11734 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11736 * asan.c (asan_emit_stack_protection): Update.
11737 (create_cond_insert_point): Update.
11738 * auto-profile.c (afdo_propagate_circuit): Update.
11739 * basic-block.h (struct edge_def): Turn probability to
11740 profile_probability.
11741 (EDGE_FREQUENCY): Update.
11742 * bb-reorder.c (find_traces_1_round): Update.
11743 (better_edge_p): Update.
11744 (sanitize_hot_paths): Update.
11745 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
11746 (make_single_succ_edge): Update.
11747 (check_bb_profile): Update.
11748 (dump_edge_info): Update.
11749 (update_bb_profile_for_threading): Update.
11750 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
11751 probabilitycount to 0.
11752 * cfgbuild.c (compute_outgoing_frequencies): Update.
11753 * cfgcleanup.c (try_forward_edges): Update.
11754 (outgoing_edges_match): Update.
11755 (try_crossjump_to_edge): Update.
11756 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
11757 (expand_gimple_tailcall): Update.
11758 (construct_init_block): Use make_single_succ_edge.
11759 (construct_exit_block): Use make_single_succ_edge.
11760 * cfghooks.c (verify_flow_info): Update.
11761 (redirect_edge_succ_nodup): Update.
11762 (split_edge): Update.
11763 (account_profile_record): Update.
11764 * cfgloopanal.c (single_likely_exit): Update.
11765 * cfgloopmanip.c (scale_loop_profile): Update.
11766 (set_zero_probability): Remove.
11767 (duplicate_loop_to_header_edge): Update.
11768 * cfgloopmanip.h (loop_version): Update prototype.
11769 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
11770 (force_nonfallthru_and_redirect): Update.
11771 (update_br_prob_note): Update.
11772 (rtl_verify_edges): Update.
11773 (purge_dead_edges): Update.
11774 (rtl_lv_add_condition_to_bb): Update.
11775 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
11776 * cgraphunit.c (init_lowered_empty_function): Update.
11777 (cgraph_node::expand_thunk): Update.
11778 * cilk-common.c: Include profile-count.h
11779 * dojump.c (inv): Remove.
11780 (jumpifnot): Update.
11781 (jumpifnot_1): Update.
11782 (do_jump_1): Update.
11784 (do_jump_by_parts_greater_rtx): Update.
11785 (do_compare_rtx_and_jump): Update.
11786 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
11787 do_jump_1. do_compare_rtx_and_jump): Update prototype.
11788 * dwarf2cfi.c: Include profile-count.h
11789 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
11790 (sjlj_emit_dispatch_table): Likewise.
11791 * explow.c: Include profile-count.h
11792 * expmed.c (emit_store_flag_force): Update.
11793 (do_cmp_and_jump): Update.
11794 * expr.c (compare_by_pieces_d::generate): Update.
11795 (compare_by_pieces_d::finish_mode): Update.
11796 (emit_block_move_via_loop): Update.
11797 (store_expr_with_bounds): Update.
11798 (store_constructor): Update.
11799 (expand_expr_real_2): Update.
11800 (expand_expr_real_1): Update.
11801 * expr.h (try_casesi, try_tablejump): Update prototypes.
11802 * gimple-pretty-print.c (dump_probability): Update.
11803 (dump_profile): New.
11804 (dump_gimple_label): Update.
11805 (dump_gimple_bb_header): Update.
11806 * graph.c (draw_cfg_node_succ_edges): Update.
11807 * hsa-gen.c (convert_switch_statements): Update.
11808 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
11809 (find_if_case_1): Update.
11810 (find_if_case_2): Update.
11811 * internal-fn.c (expand_arith_overflow_result_store): Update.
11812 (expand_addsub_overflow): Update.
11813 (expand_neg_overflow): Update.
11814 (expand_mul_overflow): Update.
11815 (expand_vector_ubsan_overflow): Update.
11816 * ipa-cp.c (good_cloning_opportunity_p): Update.
11817 * ipa-split.c (split_function): Use make_single_succ_edge.
11818 * ipa-utils.c (ipa_merge_profiles): Update.
11819 * loop-doloop.c (add_test): Update.
11820 (doloop_modify): Update.
11821 * loop-unroll.c (compare_and_jump_seq): Update.
11822 (unroll_loop_runtime_iterations): Update.
11823 * lra-constraints.c (lra_inheritance): Update.
11824 * lto-streamer-in.c (input_cfg): Update.
11825 * lto-streamer-out.c (output_cfg): Update.
11826 * mcf.c (adjust_cfg_counts): Update.
11827 * modulo-sched.c (sms_schedule): Update.
11828 * omp-expand.c (expand_omp_for_init_counts): Update.
11829 (extract_omp_for_update_vars): Update.
11830 (expand_omp_ordered_sink): Update.
11831 (expand_omp_for_ordered_loops): Update.
11832 (expand_omp_for_generic): Update.
11833 (expand_omp_for_static_nochunk): Update.
11834 (expand_omp_for_static_chunk): Update.
11835 (expand_cilk_for): Update.
11836 (expand_omp_simd): Update.
11837 (expand_omp_taskloop_for_outer): Update.
11838 (expand_omp_taskloop_for_inner): Update.
11839 * omp-simd-clone.c (simd_clone_adjust): Update.
11840 * optabs.c (expand_doubleword_shift): Update.
11841 (expand_abs): Update.
11842 (emit_cmp_and_jump_insn_1): Update.
11843 (expand_compare_and_swap_loop): Update.
11844 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
11845 * predict.c (predictable_edge_p): Update.
11846 (edge_probability_reliable_p): Update.
11847 (set_even_probabilities): Update.
11848 (combine_predictions_for_insn): Update.
11849 (combine_predictions_for_bb): Update.
11850 (propagate_freq): Update.
11851 (estimate_bb_frequencies): Update.
11852 (force_edge_cold): Update.
11853 * profile-count.c (profile_count::dump): Add missing space into dump.
11854 (profile_count::debug): Add newline.
11855 (profile_count::differs_from_p): Explicitly convert to unsigned.
11856 (profile_count::stream_in): Update.
11857 (profile_probability::dump): New member function.
11858 (profile_probability::debug): New member function.
11859 (profile_probability::differs_from_p): New member function.
11860 (profile_probability::differs_lot_from_p): New member function.
11861 (profile_probability::stream_in): New member function.
11862 (profile_probability::stream_out): New member function.
11863 * profile-count.h (profile_count_quality): Rename to ...
11864 (profile_quality): ... this one.
11865 (profile_probability): New.
11866 (profile_count): Update.
11867 * profile.c (compute_branch_probabilities): Update.
11868 * recog.c (peep2_attempt): Update.
11869 * sched-ebb.c (schedule_ebbs): Update.
11870 * sched-rgn.c (find_single_block_region): Update.
11871 (compute_dom_prob_ps): Update.
11872 (schedule_region): Update.
11873 * sel-sched-ir.c (compute_succs_info): Update.
11874 * stmt.c (struct case_node): Update.
11875 (do_jump_if_equal): Update.
11876 (get_outgoing_edge_probs): Update.
11877 (conditional_probability): Update.
11878 (emit_case_dispatch_table): Update.
11879 (expand_case): Update.
11880 (expand_sjlj_dispatch_table): Update.
11881 (emit_case_nodes): Update.
11882 * targhooks.c: Update.
11883 * tracer.c (better_p): Update.
11884 (find_best_successor): Update.
11885 * trans-mem.c (expand_transaction): Update.
11886 * tree-call-cdce.c: Update.
11887 * tree-cfg.c (gimple_split_edge): Upate.
11888 (move_sese_region_to_fn): Upate.
11889 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
11890 * tree-eh.c (lower_resx): Upate.
11891 (cleanup_empty_eh_move_lp): Upate.
11892 * tree-if-conv.c (version_loop_for_if_conversion): Update.
11893 * tree-inline.c (copy_edges_for_bb): Update.
11894 (copy_cfg_body): Update.
11895 * tree-parloops.c (gen_parallel_loop): Update.
11896 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
11897 (gimple_gen_time_profiler): Update.
11898 * tree-ssa-dce.c (remove_dead_stmt): Update.
11899 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11900 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11901 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11902 (unloop_loops): Update.
11903 (try_peel_loop): Update.
11904 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11905 * tree-ssa-loop-split.c (connect_loops): Update.
11906 (split_loop): Update.
11907 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
11908 (hoist_guard): Update.
11909 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11910 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11911 (value_replacement): Update.
11912 * tree-ssa-reassoc.c (branch_fixup): Update.
11913 * tree-ssa-tail-merge.c (replace_block_by): Update.
11914 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11915 (create_edge_and_update_destination_phis): Update.
11916 (compute_path_counts): Update.
11917 (recompute_probabilities): Update.
11918 (update_joiner_offpath_counts): Update.
11919 (freqs_to_counts_path): Update.
11920 (duplicate_thread_path): Update.
11921 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11922 (struct switch_conv_info): Update.
11923 (gen_inbound_check): Update.
11924 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11925 (vect_do_peeling): Update.
11926 (vect_loop_versioning): Update.
11927 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11928 (optimize_mask_stores): Update.
11929 * ubsan.c (ubsan_expand_null_ifn): Update.
11930 * value-prof.c (gimple_divmod_fixed_value): Update.
11931 (gimple_divmod_fixed_value_transform): Update.
11932 (gimple_mod_pow2): Update.
11933 (gimple_mod_pow2_value_transform): Update.
11934 (gimple_mod_subtract): Update.
11935 (gimple_mod_subtract_transform): Update.
11936 (gimple_ic): Update.
11937 (gimple_stringop_fixed_value): Update.
11938 (gimple_stringops_transform): Update.
11939 * value-prof.h: Update.
11941 2017-06-29 Carl Love <cel@us.ibm.com>
11943 * config/rs6000/rs6000-c.c: Add support for built-in functions
11944 vector signed int vec_signed (vector float);
11945 vector signed long long vec_signed (vector double);
11946 vector signed int vec_signed2 (vector double, vector double);
11947 vector signed int vec_signede (vector double);
11948 vector signed int vec_signedo (vector double);
11949 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
11950 instruction generator.
11951 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
11952 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
11953 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
11955 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
11956 vunsignede_v2df): Add define_expands.
11957 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
11958 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
11959 VEC_UNSIGNEDO): Add definitions.
11960 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
11961 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
11962 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
11963 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
11964 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
11965 * config/rs6000/altivec.h (vec_signed, vec_signed2,
11966 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
11967 vec_unsignede, vec_unsignedo): Add builtin defines.
11968 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
11970 * doc/extend.texi: Update the built-in documentation file for the
11971 new built-in functions.
11973 2017-06-29 Richard Biener <rguenther@suse.de>
11975 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
11976 reduction chains to LOOP_VINFO_REDUCTIONS.
11977 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
11978 SLP reductions after processing reduction chains.
11980 2017-06-29 Nathan Sidwell <nathan@acm.org>
11982 * builtins.c (fold_builtin_FUNCTION): Use
11983 lang_hooks.decl_printable_name.
11985 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
11987 PR middle-end/81194
11988 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
11989 with only one label.
11990 * stmt.c (expand_case): Assert NCASES is greater than one.
11992 2017-06-29 Richard Biener <rguenther@suse.de>
11994 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
11996 (group_case_labels): Likewise.
11997 (find_taken_edge): Push sanity checking on val to workers...
11998 (find_taken_edge_cond_expr): ... here
11999 (find_taken_edge_switch_expr): ... and here, handle cases
12000 with just a default label.
12001 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
12002 (group_case_labels): Likewise.
12003 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
12004 group_case_labels does anything cleanup the CFG again.
12006 2017-06-29 Bin Cheng <bin.cheng@arm.com>
12008 PR tree-optimization/81196
12009 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
12010 exit condition comparing two IVs.
12012 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
12014 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
12015 profile to the dummy entry at the end of the list of architectures.
12016 * config/arm/arm-cpu-cdata.h: Regenerated.
12018 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12019 Michael Collison <michael.collison@arm.com>
12022 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
12024 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
12025 (*aarch64_reg_<mode>3_minus_mask): New pattern.
12026 (*aarch64_<optab>_reg_di3_mask2): New pattern.
12027 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
12028 of shift when the shift amount is masked with constant equal to
12029 the size of the mode.
12030 * config/aarch64/predicates.md (subreg_lowpart_operator): New
12033 2017-06-29 Martin Liska <mliska@suse.cz>
12035 * config/i386/i386.opt: Change range from [1,5] to [0,5].
12037 2017-06-29 Yury Gribov <tetra2005@gmail.com>
12040 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
12043 (edge_growth_cache_entry::edge_growth_cache_entry): New
12045 (reset_edge_growth_cache): Update to use constructor.
12047 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12049 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
12050 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
12051 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
12053 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
12055 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
12056 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
12058 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
12060 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
12061 (*-linux-uclibc*): Add t-uclibc tmake_file.
12062 * config/t-musl: New.
12063 * config/t-uclibc: New.
12065 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
12067 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
12069 (gen_comm_data): Emit architectural setting of arch_prof.
12070 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
12072 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
12073 (armv8-m.base, armv8-m.main): Likewise.
12074 * arm-protos.h (arm_build_target): Add profile field.
12075 (arch_option): Likewise.
12076 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
12078 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
12079 arm_active_target.profile.
12081 2017-06-28 Richard Biener <rguenther@suse.de>
12083 PR middle-end/81227
12084 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
12085 TYPE_OVERFLOW_WRAPS.
12086 * match.pd (negate_expr_p): Likewise.
12087 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
12088 fold_build2, not fold_binary.
12090 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
12092 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
12093 Convert memory address to Pmode.
12094 (aarch64_print_operand): Assert MEM operands are always Pmode.
12096 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
12099 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
12100 Remove redundant if.
12101 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
12102 * config/arm/aarch-common-protos.h
12103 (aarch_forward_to_shift_is_not_shifted_re): Remove.
12104 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
12106 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
12109 * multiple_target.c (create_dispatcher_calls): Set the default
12110 clone to be static, not public.
12112 2017-06-28 Richard Biener <rguenther@suse.de>
12114 * tree-vect-loop.c (vectorizable_reduction): Move special
12115 cond reduction IV var creation ...
12116 (vect_create_epilog_for_reduction): ... here. Remove induction_index
12117 parameter. Use STMT_VINFO_VECTYPE.
12118 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
12121 2017-06-28 Martin Liska <mliska@suse.cz>
12124 * ipa-visibility.c (non_local_p): Handle visibility.
12126 2017-06-28 Martin Liska <mliska@suse.cz>
12129 * common.opt: Add IntegerRange to various options.
12130 * opt-functions.awk (integer_range_info): New function.
12131 * optc-gen.awk: Add integer_range_info to cl_options struct.
12132 * opts-common.c (decode_cmdline_option): Handle
12133 CL_ERR_INT_RANGE_ARG.
12134 (cmdline_handle_error): Likewise.
12135 * opts.c (print_filtered_help): Show valid interval in
12136 when --help is provided.
12137 * opts.h (struct cl_option): Add range_min and range_max fields.
12138 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
12140 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
12142 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
12143 (x * C EQ/NE y * C): New transformation.
12145 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
12147 * genmultilib (combination_space): Accept '+' in option names.
12149 2017-06-28 Martin Liska <mliska@suse.cz>
12152 * asan.c (instrument_derefs): Bail out inner references
12153 that are hard register variables.
12155 2017-06-28 Jakub Jelinek <jakub@redhat.com>
12158 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
12159 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
12161 2017-06-28 Richard Biener <rguenther@suse.de>
12163 * tree-vectorizer.h (vect_get_vec_defs): Remove.
12164 (vect_get_slp_defs): Adjust.
12165 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
12167 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
12169 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
12170 get_initial_defs_for_reduction instead of vect_get_vec_defs.
12171 (vectorizable_reduction): Adjust.
12172 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
12174 (vect_get_slp_defs): Likewise.
12175 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
12176 (vectorizable_bswap): Adjust.
12177 (vectorizable_call): Likewise.
12178 (vectorizable_conversion): Likewise.
12179 (vectorizable_assignment): Likewise.
12180 (vectorizable_shift): Likewise.
12181 (vectorizable_operation): Likewise.
12182 (vectorizable_store): Likewise.
12183 (vectorizable_condition): Likewise.
12184 (vectorizable_comparison): Likewise.
12186 2017-06-28 Michael Collison <michael.collison@arm.com>
12189 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
12191 (arm_gen_movmemqi): Removed unused variable 'i'.
12192 Convert 'for' loop into 'while' loop.
12193 (arm_expand_prologue): Remove last unnecessary set of insn.
12194 (thumb_pop): Remove unused variable 'pushed_words'.
12195 (thumb_exit): Remove last unnecessary set of regs_to_pop.
12197 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12199 * config/s390/predicates.md: Use s390_rel_address_ok_p.
12200 * config/s390/s390-protos.h: Add prototype of
12201 s390_rel_address_ok_p.
12202 * config/s390/s390.c (s390_got_symbol): New function.
12203 (s390_rel_address_ok_p): New function.
12204 (legitimize_pic_address): Use s390_rel_address_ok_p.
12205 (s390_load_got): Use s390_got_symbol.
12206 (s390_option_override): Issue error if
12207 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
12208 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
12210 * config/s390/s390.opt: New option mpic-data-is-text-relative.
12212 2017-06-27 Andrew Pinski <apinski@cavium.com>
12214 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
12215 (X * copysign (1.0, X)): New pattern.
12216 (X * copysign (1.0, -X)): New pattern.
12217 (copysign (-1.0, CST)): New pattern.
12219 2017-06-27 Joseph Myers <joseph@codesourcery.com>
12221 * genmultilib (combination_space): Remove variable.
12222 Validate reuse rules against regular expression for any sequence
12223 of multilib options in any order.
12225 2017-06-27 Michael Collison <michael.collison@arm.com>
12227 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
12228 call aarch64_split_simd_combine.
12229 * (aarch64_combine_internal<mode>): Delete pattern.
12230 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
12231 Allow register and subreg operands.
12233 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12235 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
12236 specific need, just fallback on defaults.
12237 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
12239 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12240 Olivier Hainque <hainque@adacore.com>
12242 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
12244 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
12245 targets. Pick a default if no particular attempt applied.
12246 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
12249 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12251 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
12252 (x86_64-wrs-vxworks7): Likewise.
12254 2017-06-27 Marek Polacek <polacek@redhat.com>
12257 * ubsan.c (instrument_null): Check get_base_address's result for null.
12259 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
12261 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
12263 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
12265 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
12266 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
12267 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
12268 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
12269 New function types.
12270 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
12271 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
12272 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
12273 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
12274 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
12275 BUILT_IN_FEUPDATEENV): New builtins.
12276 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
12277 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
12278 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
12279 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
12281 (builtin_structptr_types): Adjust size.
12282 * tree.c (builtin_structptr_types): Add four entries.
12284 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12285 Olivier Hainque <hainque@adacore.com>
12287 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
12288 (TLS_SYM): New local macro, forcing reference to __tls__ on
12289 link command lines for VxWorks 7 RTPs, triggering initialization
12291 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
12292 OS features TLS support, true for RTPs on VxWorks 7.
12293 * config/vxworks.c (vxworks_override_options): Setup emutls
12296 2017-06-27 Jakub Jelinek <jakub@redhat.com>
12298 * predict.c (test_prediction_value_range): Use -1U instead of -1
12299 to avoid narrowing conversion warning.
12300 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
12301 to avoid narrowing conversion warning.
12302 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
12304 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
12306 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12308 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
12309 64bit configurations.
12310 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
12311 (SIZE_TYPE): Likewise.
12312 * config/vxworks.c (vxworks_emutls_var_fields): Use
12313 long_unsigned_type_node instead of unsigned_type_node as the offset
12314 field type, which is "pointer" mode in emutls.c.
12316 2017-06-27 Jakub Jelinek <jakub@redhat.com>
12319 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
12321 PR middle-end/81207
12322 * gimple-fold.c (replace_call_with_call_and_fold): Handle
12323 gimple_vuse copying separately from gimple_vdef copying.
12325 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12327 * value-prof.c (free_hist): Remove call to memset and the enclosing if
12330 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
12331 Olivier Hainque <hainque@adacore.com>
12333 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
12334 for all vxworks7 targets.
12335 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
12336 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
12337 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
12338 variations for VX6/VX7 and 32/64bits later on in ...
12339 (VXWORKS_LIB_SPEC): Leverage new macros.
12340 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
12341 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
12343 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
12345 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
12346 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
12348 2017-06-26 Carl Love <cel@us.ibm.com>
12350 * config/rs6000/rs6000-c.c: Add support for built-in functions
12351 vector bool char vec_reve (vector bool char);
12352 vector signed char vec_reve (vector signed char);
12353 vector unsigned char vec_reve (vector unsigned char);
12354 vector bool int vec_reve (vector bool int);
12355 vector signed int vec_reve (vector signed int);
12356 vector unsigned int vec_reve (vector unsigned int);
12357 vector bool long long vec_reve (vector bool long long);
12358 vector signed long long vec_reve (vector signed long long);
12359 vector unsigned long long vec_reve (vector unsigned long long);
12360 vector bool short vec_reve (vector bool short);
12361 vector signed short vec_reve (vector signed short);
12362 vector double vec_reve (vector double);
12363 vector float vec_reve (vector float);
12364 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
12365 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
12366 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
12367 (altivec_vreve): New pattern.
12368 * config/rs6000/altivec.h (vec_reve): New define.
12369 * doc/extend.texi (vec_rev): Update the built-in documentation file
12370 for the new built-in functions.
12372 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12374 PR tree-optimization/71815
12375 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
12377 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
12379 (slsr_process_phi): Likewise.
12380 (replace_uncond_cands_and_profitable_phis): Don't replace a
12381 multiply candidate with a stride of 1 (copy or cast).
12382 (phi_incr_cost): Call uses_consumed_by_stmt rather than
12384 (lowest_cost_path): Likewise.
12385 (total_savings): Likewise.
12387 2017-06-26 Richard Biener <rguenther@suse.de>
12390 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
12391 Use def_builtin_pure for all gather builtins.
12393 2017-06-26 Richard Biener <rguenther@suse.de>
12395 PR tree-optimization/81203
12396 * tree-tailcall.c (find_tail_calls): Do not move stmts into
12397 non-dominating BBs.
12399 2017-06-26 Marek Polacek <polacek@redhat.com>
12402 * doc/invoke.texi: Document -Wmultistatement-macros.
12404 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
12406 * doc/sourcebuild.texi (ARM-specific attributes): Document new
12407 arm_neon_ok_no_float_abi effective target.
12409 2017-06-26 Richard Biener <rguenther@suse.de>
12411 PR tree-optimization/80928
12412 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
12413 (copy_bbs): Set BB_DUPLICATED flag early.
12414 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
12416 (execute_on_shrinking_pred): Likewise.
12417 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
12418 BB_DUPLICATED blocks.
12419 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
12420 iterate over all PHIs considering removal of *gsi.
12422 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
12424 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
12427 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12429 * config/rs6000/rs6000-string.c: (expand_block_clear,
12430 do_load_for_compare, select_block_compare_mode,
12431 compute_current_alignment, expand_block_compare,
12432 expand_strncmp_align_check, expand_strn_compare,
12433 expand_block_move, rs6000_output_load_multiple)
12434 Move functions related to string/block move/compare
12435 to a separate file.
12436 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
12437 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
12438 for this function which is now used in two files.
12439 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
12440 * config.gcc: Add rs6000-string.o to extra_objs for
12441 targets powerpc*-*-* and rs6000*-*-*.
12443 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
12446 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
12447 32-bit, since indexed is not valid for DImode.
12448 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
12449 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
12450 (define_peephole2 for Altivec d-form load): Add 32-bit support.
12451 (define_peephole2 for Altivec d-form store): Likewise.
12454 * multiple_target.c (create_dispatcher_calls): Only create the
12455 dispatcher call if the function is the default clone of a
12456 versioned function.
12458 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
12460 PR middle-end/80902
12461 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
12462 a call, force the call to not be a tail call.
12464 2017-06-23 Jeff Law <law@redhat.com>
12466 * doc/contrib.texi: Add entry for Steven Pemberton's work on
12469 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
12471 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
12472 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
12473 handling for early expansion of vector shifts (sl,sr,sra,rl).
12474 (builtin_function_type): Add vector shift right instructions
12475 to the unsigned argument list.
12477 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
12479 rtl-optimizatoin/79286
12480 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
12481 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
12482 trap. PIC register plus a const unspec without offset can never trap.
12484 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
12486 * tree.h (builtin_structptr_type): New type.
12487 (builtin_structptr_types): Declare new array.
12488 * tree.c (builtin_structptr_types): New array.
12489 (free_lang_data, build_common_tree_nodes): Use it.
12491 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
12494 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
12497 2017-06-22 Matt Turner <mattst88@gmail.com>
12499 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
12500 Lake models to skylake case. Assume skylake for unknown
12501 models with clflushopt.
12503 2017-06-22 Jeff Law <law@redhat.com>
12505 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
12506 frame sizes that do not satisfy aarch64_uimm12_shift.
12508 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
12510 * profile-count.h (apply_probability,
12511 apply_scale, probability_in): Fix checks for zero.
12513 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12515 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
12516 * doc/cppdiropts.texi (-I @var{dir}): Document it.
12518 2016-06-22 Richard Biener <rguenther@suse.de>
12520 * tree-vect-loop.c (vect_model_reduction_cost): Handle
12521 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
12522 REDUC_MAX_EXPR support.
12523 (vectorizable_reduction): Likewise.
12524 (vect_create_epilog_for_reduction): Likewise.
12526 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
12528 * match.pd (A / (1 << B) -> A >> B): New.
12529 * generic-match-head.c: Include optabs-tree.h.
12530 * gimple-match-head.c: Likewise.
12531 * optabs-tree.h (target_supports_op_p): New.
12532 * optabs-tree.c (target_supports_op_p): New.
12534 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12536 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
12537 $gcc_cv_ld --help output.
12538 (gcc_cv_ld_demangle): Likewise.
12539 (gcc_cv_ld_eh_frame_hdr): Likewise.
12540 (gcc_cv_ld_pie): Likewise.
12541 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
12542 (gcc_cv_ld_buildid): Likewise.
12543 (gcc_cv_ld_sysroot): Likewise.
12544 (ld_bndplt_support): Likewise.
12545 (ld_pushpopstate_support): Likewise.
12546 * configure: Regenerate.
12547 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
12549 2017-06-21 Jakub Jelinek <jakub@redhat.com>
12552 * config/i386/sse.md (round<mode>2): Renumber match_dup and
12553 operands indexes to avoid gap between operands and match_dups.
12555 2017-06-21 Andrew Pinski <apinski@cavium.com>
12557 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
12558 Increment Arith_shift and Arith_shift_reg by 1.
12559 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
12561 * config/aarch64/aarch64.c (thunderx_tunings): Enable
12562 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
12563 (aarch64_strip_extend): Add new argument and test for it.
12564 (aarch64_cheap_mult_shift_p): New function.
12565 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
12566 add a cost if it is true.
12567 Update calls to aarch64_strip_extend.
12568 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
12570 2017-06-21 Andrew Pinski <apinski@cavium.com>
12572 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
12574 (thunderxt88): Likewise.
12575 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
12576 (thunderx_prefetch_tune): New variable.
12577 (thunderx2t99_prefetch_tune): Update for the correct values.
12578 (thunderxt88_tunings): New variable.
12579 (thunderx_tunings): Use thunderx_prefetch_tune instead of
12580 generic_prefetch_tune.
12581 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
12583 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12585 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
12586 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
12587 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
12588 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
12589 (aarch64_atomic_cas<mode>, GPI): Likewise.
12591 2017-06-21 Martin Liska <mliska@suse.cz>
12593 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
12594 statements on cold and hot labels.
12595 * predict.c (tree_estimate_probability_bb): Remove the
12596 prediction from this place.
12598 2017-06-21 Martin Liska <mliska@suse.cz>
12600 PR tree-optimization/79489
12601 * gimplify.c (maybe_add_early_return_predict_stmt): New
12603 (gimplify_return_expr): Call the function.
12604 * predict.c (tree_estimate_probability_bb): Remove handling
12606 * predict.def: Update comment about early return predictor.
12607 * gimple-predict.h (is_gimple_predict): New function.
12608 * predict.def: Change default value of early return to 66.
12609 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
12611 * passes.def: Put pass_strip_predict_hints to the beginning of
12614 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
12616 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
12617 FUNCTION_DECL declarations.
12618 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
12620 (dwaf2out_decl): Likewise.
12621 * godump.c (go_early_global_decl): Skip call to the real debug hook
12622 for FUNCTION_DECL declarations.
12623 * passes.c (rest_of_decl_compilation): Skip call to the
12624 early_global_decl debug hook for FUNCTION_DECL declarations, unless
12625 -fdump-go-spec is passed.
12627 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
12629 * config/i386/i386.c (struct builtin_isa): New field pure_p.
12630 Reorder for compactness.
12631 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
12632 (def_builtin_pure, def_builtin_pure2): New functions.
12633 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
12635 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
12637 * match.pd (nop_convert): New predicate.
12638 ((A +- CST1) +- CST2): Allow some NOP conversions.
12640 2017-06-21 Jakub Jelinek <jakub@redhat.com>
12643 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
12644 with ctors/dtors if GOVD_SHARED is set.
12646 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12648 * config/aarch64/aarch64.md (movti_aarch64):
12649 Emit mov rather than orr.
12650 (movtf_aarch64): Likewise.
12651 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
12652 Emit mov rather than orr.
12654 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12656 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
12657 Swap alternatives, make integer dup more expensive.
12659 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12661 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
12662 Return true for non-tls symbols.
12664 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
12666 * config/aarch64/aarch64-cores.def (cortex-a55): New.
12667 (cortex-a75): Likewise.
12668 (cortex-a75.cortex-a55): Likewise.
12669 * config/aarch64/aarch64-tune.md: Regenerate.
12670 * doc/invoke.texi (-mtune): Document new values for -mtune.
12672 2017-06-21 Tom de Vries <tom@codesourcery.com>
12674 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
12675 stack_size feature.
12676 (Effective-Target Keywords, Other attributes): Suggest using
12677 dg-add-options stack_size feature to get stack limit in stack_size
12678 effective target documentation.
12680 2017-06-21 Julian Brown <julian@codesourcery.com>
12681 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12683 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
12684 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
12685 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
12687 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
12688 attribute type list for neon_multiply.
12689 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
12690 attribute type list for neon_multiply.
12691 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
12692 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
12693 attribute type list for neon_multiply.
12694 * config/arm/types.md (crypto_pmull): Add.
12695 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
12696 attribute type list.
12698 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
12700 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
12703 2017-06-20 Jakub Jelinek <jakub@redhat.com>
12705 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
12706 to make sure not to dereference a NULL cost_classes_ptr pointer.
12708 2017-06-20 Carl Love <cel@us.ibm.com>
12710 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12711 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12712 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12713 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12714 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12715 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
12716 VMULOSW): New enum "unspec" values.
12717 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
12718 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
12719 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
12720 altivec_vmulosw): New patterns.
12721 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12722 VMULOSW): Add definitions.
12724 2017-06-20 Julia Koval <julia.koval@intel.com>
12726 * config/i386/i386.c: Fix rounding expand for new pattern.
12727 * config/i386/subst.md: Fix pattern (parallel -> unspec).
12729 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12731 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
12732 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
12734 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12736 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
12739 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12741 * config/aarch64/aarch64-cores.def: Rearrange to sort by
12742 architecture, then by implementer ID.
12743 * config/aarch64/aarch64-tune.md: Regenerate.
12745 2017-06-20 Richard Biener <rguenther@suse.de>
12747 PR middle-end/81097
12748 * fold-const.c (split_tree): Fold to type before negating.
12750 2017-06-20 David Malcolm <dmalcolm@redhat.com>
12752 * diagnostic-show-locus.c
12753 (selftest::test_fixit_deletion_affecting_newline): New function.
12754 (selftest::diagnostic_show_locus_c_tests): Call it.
12756 2017-06-20 Andreas Schwab <schwab@suse.de>
12759 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
12762 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
12764 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
12765 __ARM_FEATURE_COPROC according to support.
12767 2017-06-20 Jakub Jelinek <jakub@redhat.com>
12769 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
12770 Rewritten to avoid overflow for > 32-bit pointers.
12773 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
12774 by removing enum keyword.
12775 (ubsan_type_descriptor): Likewise. Formatting fix.
12778 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
12779 splitter): Require TARGET_SSE2 in the condition.
12781 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
12784 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
12785 for doing vector set of SFmode on ISA 3.0.
12786 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
12787 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
12789 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
12790 SFmode value into a V4SF variable that was extracted from another
12791 V4SF variable without converting the element to double precision
12792 and back to single precision vector format.
12793 (vsx_insert_extract_v4sf_p9_2): Likewise.
12795 2017-06-19 Jakub Jelinek <jakub@redhat.com>
12797 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
12798 in UWHI to avoid undefined overflow.
12801 * ubsan.h (enum ubsan_encode_value_phase): New.
12802 (ubsan_encode_value): Change second argument to
12803 enum ubsan_encode_value_phase with default value of
12804 UBSAN_ENCODE_VALUE_GENERIC.
12805 * ubsan.c (ubsan_encode_value): Change second argument to
12806 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
12807 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
12808 create_tmp_var_raw instead of create_tmp_var and use a
12810 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
12811 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
12812 ubsan_encode_value callers.
12815 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
12816 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
12817 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
12819 2017-06-19 Richard Biener <rguenther@suse.de>
12821 PR middle-end/81118
12822 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
12823 estimates if we changed anything.
12825 2017-06-19 Richard Biener <rguenther@suse.de>
12827 PR tree-optimization/80887
12828 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
12829 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
12830 simplified lookups, then reset mprts_hook.
12831 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
12833 (try_to_simplify): Likewise.
12835 2017-06-19 Martin Liska <mliska@suse.cz>
12838 * gimplify.c (gimplify_switch_expr):
12839 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
12841 2017-06-19 Martin Liska <mliska@suse.cz>
12843 * doc/install.texi: Document that PGO runs in 4 stages.
12845 2017-06-19 Martin Liska <mliska@suse.cz>
12848 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
12849 to dispatcher function name.
12850 * multiple_target.c (replace_function_decl): New function.
12851 (create_dispatcher_calls): Redirect both edges and references.
12853 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
12855 * profile-count.c (profile_count::dump): Dump quality.
12856 (profile_count::differs_from_p): Update for unsigned val.
12857 * profile-count.h (profile_count_quality): New enum.
12858 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
12860 2017-06-19 Richard Biener <rguenther@suse.de>
12862 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
12863 struct function as arg.
12864 (estimate_numbers_of_iterations): Export overload with loop arg.
12865 (free_numbers_of_iterations_estimates_loop): Use an overload of
12866 free_numbers_of_iterations_estimates instead.
12867 * tree-cfg.c (remove_bb): Adjust.
12868 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
12869 * tree-parloops.c (gen_parallel_loop): Likewise.
12870 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
12872 (tree_unroll_loops_completely): Likewise.
12873 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
12874 Use an overload instead and export.
12875 (estimated_loop_iterations): Adjust.
12876 (max_loop_iterations): Likewise.
12877 (likely_max_loop_iterations): Likewise.
12878 (estimate_numbers_of_iterations): Take struct function as arg
12880 (loop_exits_before_overflow): Adjust.
12881 (free_numbers_of_iterations_estimates_loop): Use an overload.
12882 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
12883 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
12885 2017-06-19 Richard Biener <rguenther@suse.de>
12888 * ipa-prop.c (find_constructor_constant_at_offset): Handle
12889 RANGE_EXPR conservatively.
12891 2017-06-16 Carl Love <cel@us.ibm.com>
12893 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12894 definitions for vec_float, vec_float2, vec_floato,
12895 vec_floate built-ins.
12896 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
12897 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
12899 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
12900 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
12901 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
12902 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
12903 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
12904 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
12905 vec_floato): Add builtin defines.
12906 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
12907 Update the built-in documentation file for the new built-in
12910 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12912 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
12913 (mthumb): Mark as the negative of -marm.
12915 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12917 * doc/invoke.texi (ARM Options, -mcpu): Document supported
12919 (ARM Options, -mtune): Document that this accepts the same
12920 extension options as -mcpu.
12921 (ARM Options, -mfpu): Document addition of -mfpu=auto.
12923 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12925 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
12926 permitted extensions.
12928 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12930 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
12931 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
12932 (armv8-m.main): Add option +nodsp.
12933 * config/arm/arm-cpu-cdata.h: Regenerated.
12935 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12937 * config/arm/t-fuchsia: New file.
12938 * config.gcc (arm*-*-fuchsia*): Use it.
12940 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12942 * config/arm/t-symbian: Rewrite for new option infrastructure.
12944 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12946 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
12947 (MULTILIB_REQUIRED): Likewise.
12949 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12951 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
12952 (MULTILIB_RESUE): Likewise.
12953 (MULTILIB_MATCHES): Likewise.
12954 (MULTLIB_REQUIRED): Likewise.
12956 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12958 * config/arm/t-rtems: Rewrite for new option framework.
12960 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12962 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
12963 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
12964 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
12965 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
12966 * config/arm/t-multilib: ... here.
12967 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
12968 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
12969 armv7-a and armv8*-a when A-profile libraries have not been built.
12970 * config/arm/t-rmprofile: Rewrite.
12972 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12974 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
12975 with a backslash. Remove the backslash after substituting unescaped
12977 * doc/fragments.texi (MULTILIB_REUSE): Document it.
12979 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12981 * config.gcc: (arm*-*-*): When building a-profile libraries, force
12982 the driver to pass through the default setting of -mfloat-abi.
12983 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
12985 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
12986 (all_feat_combs): New rule.
12987 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
12989 * config/arm/t-aprofile: Rewrite.
12991 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12993 * config/arm/arm.h (FPUTYPE_AUTO): Define.
12994 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
12995 fpu is not specified by the user/command-line.
12996 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
12997 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
12998 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
12999 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
13000 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
13001 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
13003 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13005 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
13006 * config/arm/t-arm-elf: Rewritten.
13008 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13010 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
13011 have some floating-point instructions.
13012 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
13013 (TARGET_MAYBE_HARD_FLOAT): New macro.
13014 * config/arm/arm-builtins.c (arm_init_builtins): Use
13015 TARGET_MAYBE_HARD_FLOAT.
13016 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
13018 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13020 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
13021 (configargs.h): Include it.
13022 (arm_print_hint_for_fpu_option): New function.
13023 (arm_parse_fpu_option): New function.
13024 (candidate_extension): New class.
13025 (arm_canon_for_multilib): New function.
13026 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
13027 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
13028 (ARCH_CANONICAL_SPECS): New macro.
13029 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
13031 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13033 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
13034 are set after handling multilib fragments. Set target_cpu_default2
13037 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13039 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
13041 (arm*-*-*): Set target_cpu_default2 to a quoted string.
13042 * config/arm/parsecpu.awk (check_cpu): Validate any extension
13044 (check_arch): Likewise.
13045 * config/arm/arm.c (arm_configure_build_target): Handle
13046 TARGET_CPU_DEFAULT being a string constant. Scan any feature
13047 options in the default.
13049 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13051 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
13052 when an option is an alias of another.
13053 * config/arm/parsecpu.awk (optalias): New parser token.
13054 (gen_comm_data): Mark non-alias options as such. Emit entries
13055 for extension aliases.
13056 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
13057 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
13058 (armv6kz, armv6zk, armv6t2): Likewise.
13059 (armv7): Make vfpv3-d16 an alias.
13060 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
13062 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
13063 Sort in canonical order.
13064 (armv8-a): Sort in canonical order.
13065 (armv8.1-a, armv8.2-a): Likewise.
13066 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
13068 (cortex-a9): Sort in canonical order.
13069 * config/arm/arm.c (selftests.h): Include it.
13070 (arm_test_cpu_arch_data): New function.
13071 (arm_run_self_tests): New function.
13072 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
13073 (targetm): Move declaration to the end of the file.
13074 * arm-cpu-cdata.h: Regenerated.
13076 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13078 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
13079 call to target_mode_check describing the type of option passed.
13080 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
13081 (arm_target_thumb_only): Use arm_parse_arch_option_name or
13082 arm_parse_cpu_option_name to match parameters against list of
13084 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
13085 arm_arch_core_flags data structure.
13086 * config/arm/arm-cpu_cdata.h: Regenerated.
13088 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13090 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
13092 (arm_print_hint_for_cpu_option): Likewise.
13093 (arm_print_hint_for_arch_option): Likewise.
13094 (arm_parse_cpu_option_name): Likewise.
13095 (arm_parse_arch_option_name): Likewise.
13096 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
13097 of entries in the all_fpus list.
13098 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
13099 (arm_parse_cpu_option_name): Declare.
13100 (arm_parse_arch_option_name): Declare.
13101 (arm_parse_option_features): Declare.
13102 (arm_intialize_isa): Declare.
13103 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
13105 (gen_comm_data): ... here. Make definitions non-static.
13106 * config/arm/arm-cpu-data.h: Regenerated.
13107 * config/arm/arm-cpu-cdata.h: Regenerated.
13109 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13111 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
13112 (cpu_arch_extension): New structure.
13113 (cpu_arch_option, arch_option, cpu_option): New structures.
13114 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
13115 architecture types.
13116 (gen_data): Generate new format data tables.
13117 * config/arm/arm.c (cpu_tune): New structure.
13118 (cpu_option, processors): Delete.
13119 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
13120 (arm_print_hint_for_cpu_option): ... this and ...
13121 (arm_print_hint_for_arch_option): ... this.
13122 (arm_parse_arch_cpu_name): Delete. Replace with ...
13123 (arm_parse_cpu_option_name): ... this and ...
13124 (arm_parse_arch_option_name): ... this.
13125 (arm_unrecognized_feature): Change type of target parameter to
13127 (arm_parse_arch_cpu_features): Delete. Replace with ...
13128 (arm_parse_option_features): ... this.
13129 (arm_configure_build_target): Rework to use new configuration data
13131 (arm_print_tune_info): Rework for new configuration data tables.
13132 * config/arm/arm-cpu-data.h: Regenerated.
13133 * config/arm/arm-cpu.h: Regenerated.
13135 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13137 * Makefile.in (OBJS): Move sbitmap.o from here ...
13138 (OBJS-libcommon): ... to here.
13140 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13142 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
13143 (ISA_ALL_CRYPTO): New macro.
13144 (ISA_ALL_SIMD): New macro
13145 (ISA_ALL_FP): New macro.
13146 * config/arm/arm.c (fpu_bitlist): Update initializer.
13147 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
13149 (arm9e): Add fpu. Add option for nofp
13150 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
13151 (arm926ej-s, arm1026ej-s): Likewise.
13152 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
13153 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
13154 neon-fp16, neon-vfpv4, nofp and nosimd.
13155 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
13156 (cortex-a8): Add fpu. Add option for nofp.
13157 (cortex-a9): Add fpu. Add options for nosimd and nofp.
13158 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
13159 (cortex-r4f): Add fpu.
13160 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
13161 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
13163 (cortex-r8): Likewise.
13164 (cortex-m4): Add fpu. Add option for nofp.
13165 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
13166 (cortex-a17.cortex-a7): Likewise.
13167 (cortex-a32): Add fpu. Add options for crypto and nofp.
13168 (cortex-a35, cortex-a53): Likewise.
13169 (cortex-a57): Add fpu. Add option for crypto.
13170 (cortex-a72, cortex-a73): Likewise.
13171 (exynos-m1): Likewise.
13172 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
13173 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
13174 (cortex-m33): Add fpu. Add option for nofp.
13175 * config/arm/arm-cpu-cdata.h: Regenerated
13176 * config/arm/arm-cpu-data.h: Regenerated.
13178 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13180 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
13181 (armv5te, armv5tej): Likewise.
13182 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
13183 (armv7): Add options fp and vfpv3-d16.
13184 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
13185 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
13187 (armv7ve): Likewise.
13188 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
13189 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
13190 (armv8-a): Add nocrypto option.
13191 (armv8.1-a, armv8.2-a): Likewise.
13192 (armv8-m.main): add options fp, fp.dp and nofp.
13194 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13196 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
13198 (armv8-a+crc): Delete.
13199 (armv8.1-a): Add options simd, crypto and nofp.
13200 (armv8.2-a): Add options fp16, simd, crypto and nofp.
13201 (armv8.2-a+fp16): Delete.
13202 (armv8-m.main): Add option dsp.
13203 (armv8-m.main+dsp): Delete.
13204 (cortex-a8): Add fpu. Add nofp option.
13205 (cortex-a9): Add fpu. Add nofp and nosimd options.
13206 * config/arm/parsecpu.awk (gen_data): Generate option tables and
13207 link to main cpu and architecture data structures.
13208 (gen_comm_data): Only put isa attributes from the main architecture
13210 (option): New statement for architecture and CPU entries.
13211 * arm.c (struct cpu_option): New structure.
13212 (struct processors): Add entry for options.
13213 (arm_unrecognized_feature): New function.
13214 (arm_parse_arch_cpu_name): Ignore any characters after the first
13216 (arm_parse_arch_cpu_feature): New function.
13217 (arm_configure_build_target): Separate out any CPU and architecture
13218 features and parse separately. Don't error out if -mfpu=auto is
13219 used with only an architecture string.
13220 (arm_print_asm_arch_directives): New function.
13221 (arm_file_start): Call it.
13222 * config/arm/arm-cpu-cdata.h: Regenerated.
13223 * config/arm/arm-cpu-data.h: Likewise.
13224 * config/arm/arm-tables.opt: Likewise.
13226 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13228 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
13229 assembler when it is not -mfpu=auto.
13231 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13233 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
13234 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
13235 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
13236 (ASM_CPU_SPEC): Rewrite.
13237 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
13238 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
13239 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
13240 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
13241 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
13242 copied string is NUL-terminated. Also strip any characters prefixed
13244 (arm_rewrite_selected_arch): New function.
13245 (arm_rewrite_march): New function.
13247 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13249 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
13250 (x_arm_cpu_string, x_arm_tune_string): Likewise.
13251 (march, mcpu, mtune): Convert to string-based options.
13252 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
13253 (arm_parse_arch_cpu_name): New function.
13254 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
13255 identify selected architecture or CPU.
13256 (arm_option_save): New function.
13257 (TARGET_OPTION_SAVE): Redefine.
13258 (arm_option_restore): Restore string options.
13259 (arm_option_print): Print string options.
13261 2017-06-16 Martin Sebor <msebor@redhat.com>
13263 PR tree-optimization/80933
13264 PR tree-optimization/80934
13265 * builtins.c (fold_builtin_3): Do not handle bcmp here.
13266 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
13267 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
13268 (gimple_fold_builtin): Call them.
13270 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13272 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
13273 as unlikely; update profile.
13275 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13277 * predict.c (force_edge_cold): Handle declaring edges impossible
13280 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13282 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
13284 (try_unroll_loop_completely): Fix reporting.
13286 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13288 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
13290 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
13293 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
13294 if given a non-constant argument for an intrinsic which requires a
13297 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13299 * profile.c (compare_freqs): New function.
13300 (branch_prob): Sort edge list.
13301 (find_spanning_tree): Assume that the list is priority sorted.
13303 2017-06-16 Richard Biener <rguenther@suse.de>
13305 PR tree-optimization/81090
13306 * passes.def (pass_record_bounds): Remove.
13307 * tree-pass.h (make_pass_record_bounds): Likewise.
13308 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
13309 make_pass_record_bounds): Likewise.
13310 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
13311 not free niter estimates at the beginning but at the end.
13312 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
13314 2017-06-16 Richard Biener <rguenther@suse.de>
13316 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
13317 initializer to workaround ICE in host GCC 4.8.
13319 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13321 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
13323 (clone_inlined_nodes): Update.
13325 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13327 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
13328 prefetch settings, and enable prefetching by default at -O3.
13330 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13332 * config/aarch64/aarch64.c (aarch64_override_options_internal):
13333 Set flag_prefetch_loop_arrays according to tuning data.
13335 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13337 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
13338 New tune structure.
13339 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
13340 [Unrelated to main purpose of the patch] Place the pointer field last
13341 to enable type checking errors when tune structure are wrongly merged.
13342 * config/aarch64/aarch64.c (generic_prefetch_tune,)
13343 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
13344 (thunderx2t99_prefetch_tune): New tune constants.
13345 (tune_params *_tunings): Update all tunings (no functional change).
13346 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
13347 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
13348 from tunings structures.
13350 2017-06-16 Jakub Jelinek <jakub@redhat.com>
13353 * ubsan.c (instrument_null): Add T argument, use it instead
13354 of computing it based on IS_LHS.
13355 (instrument_object_size): Likewise.
13356 (pass_ubsan::execute): Adjust instrument_null and
13357 instrument_object_size callers to pass gimple_get_lhs or
13358 gimple_assign_rhs1 result to it. Use instrument_null instead of
13359 calling get_base_address and instrument_mem_ref. Handle
13360 aggregate call arguments for object-size sanitization.
13362 2017-06-16 Yury Gribov <tetra2005@gmail.com>
13364 PR tree-optimization/81089
13365 * tree-vrp.c (is_masked_range_test): Validate operands of
13368 2017-06-15 Martin Sebor <msebor@redhat.com>
13371 * dumpfile.c (dump_register): Avoid calling memset to initialize
13372 a class with a default ctor.
13373 * gcc.c (struct compiler): Remove const qualification.
13374 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
13375 * hash-table.h: Ditto.
13376 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
13378 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
13379 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
13381 * params.h (struct param_info): Make struct members non-const.
13382 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
13383 with default initialization.
13384 * vec.h (vec_copy_construct, vec_default_construct): New helper
13386 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
13387 with vec_copy_construct.
13388 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
13389 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
13390 * doc/invoke.texi (-Wclass-memaccess): Document.
13392 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13394 * emit-rtl.h (is_leaf): Update comment about local
13395 register allocator.
13397 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13400 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
13401 for a variable to have a section before checking if the section has a
13403 Set section to.persistent if persistent attribute is set.
13404 Warn if .persistent attribute is used on an automatic variable.
13406 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
13408 PR rtl-optimization/80474
13409 * reorg.c (update_block): Do not ignore instructions in a delay slot.
13411 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
13413 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
13416 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
13418 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
13419 (casesi): Emit bounds checking as RTL.
13420 (casesi_internal_mips16_<mode>): Remove bounds checking.
13422 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
13424 * config/xtensa/xtensa.c (xtensa_option_override): Append
13425 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
13426 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
13427 xtensa_doloop_hooks): Define unconditionally.
13428 (xtensa_reorg_loops): Only call reorg_loops in the presence of
13430 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
13431 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
13432 for it in xtensa_option_override.
13433 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
13434 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
13436 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
13438 * doc/cppopts.texi: Document '-' special value to -MF.
13440 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
13442 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
13443 (cortex_a53_fconst): Likewise.
13444 (cortex_a53_fpmul): Likewise.
13445 (cortex_a53_f_load_64): Likewise.
13446 (cortex_a53_f_load_many): Likewise.
13447 (cortex_a53_advsimd_alu): Likewise.
13448 (cortex_a53_advsimd_alu_q): Likewise.
13449 (cortex_a53_advsimd_mul): Likewise.
13450 (cortex_a53_advsimd_mul_q): Likewise.
13451 (fpmac bypass): Add new bypass for fpmac-fpmac case.
13452 Add missing fmul, r2f_cvt and fconst cases.
13454 2017-06-14 Richard Biener <rguenther@suse.de>
13456 PR middle-end/81088
13457 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
13459 (fold_binary_loc): When associating do not treat pre-existing
13460 TREE_OVERFLOW on literal constants as a reason to allow
13461 TREE_OVERFLOW on associated literal constants.
13463 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13465 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
13466 (MASK_FEATURES): New macro.
13467 * config/sparc/sparc.c (sparc_option_override): Remove the special
13468 handling of -mfpu and generalize it to all MASK_FEATURES switches.
13470 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13472 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
13473 a division of 0 if non-call exceptions are enabled.
13475 2017-06-14 Andrew Pinski <apinski@cavium.com>
13476 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13479 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
13480 Improve vector initialization code gen for only variable case.
13482 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13484 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
13486 2017-06-14 Richard Biener <rguenther@suse.de>
13488 PR tree-optimization/81083
13489 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
13492 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13494 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
13495 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
13496 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
13497 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
13498 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
13499 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
13501 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13503 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
13504 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
13506 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13508 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
13510 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13512 * config/rs6000/t-rtems: Don't handle SPE.
13514 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13516 * config/rs6000/t-linux: Don't handle SPE.
13518 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13520 * config/rs6000/eabispe.h: Delete file.
13522 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13524 * config/rs6000/t-spe: Delete file.
13526 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13528 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
13529 (rs6000_legitimate_offset_address_p): Return false for anything in
13530 V2SImode or V2SFmode.
13532 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13534 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
13535 except V2SF and V2SI. Rearrange the vector modes, and add comments.
13536 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
13538 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
13539 (rs6000_legitimate_offset_address_p): Ditto.
13540 (rs6000_emit_move): Ditto.
13541 (rs6000_init_builtins): Remove V4HI_type_node.
13543 2017-06-13 Martin Liska <mliska@suse.cz>
13546 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
13547 (gate_asan): Likewise.
13548 * asan.h (asan_no_sanitize_address_p): Remove the function.
13549 (sanitize_flags_p): New function.
13550 * builtins.def: Fix coding style.
13551 * common.opt: Use renamed enum value.
13552 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
13553 * doc/extend.texi: Document no_sanitize attribute.
13554 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
13555 to SANITIZE_UNDEFINED_NONDEFAULT.
13556 * gcc.c (sanitize_spec_function): Use the renamed enum value.
13557 * gimple-fold.c (optimize_atomic_compare_exchange_p):
13558 Use sanitize_flags_p.
13559 * gimplify.c (gimplify_function_tree): Likewise.
13560 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
13561 * opts.c (parse_no_sanitize_attribute): New function.
13562 (common_handle_option): Use renamed enum value.
13563 * opts.h (parse_no_sanitize_attribute): Declare.
13564 * tree.c (sanitize_flags_p): New function.
13565 * tree.h: Declared here.
13566 * tsan.c: Use sanitize_flags_p.
13567 * ubsan.c (ubsan_expand_null_ifn): Likewise.
13568 (instrument_mem_ref): Likewise.
13569 (instrument_bool_enum_load): Likewise.
13570 (do_ubsan_in_current_function): Remove the function.
13571 (pass_ubsan::execute): Use sanitize_flags_p.
13572 * ubsan.h: Remove do_ubsan_in_current_function
13573 * tree-cfg.c (print_no_sanitize_attr_value): New function.
13574 (dump_function_to_file): Use it here.
13576 2017-06-13 Martin Jambor <mjambor@suse.cz>
13578 PR tree-optimization/80803
13579 PR tree-optimization/81063
13580 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
13581 (propagate_subaccesses_across_link): Enqueue subtree whenever
13582 necessary instead of relying on the caller.
13584 2017-06-13 Martin Jambor <mjambor@suse.cz>
13586 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
13587 that have a first_link.
13588 (sort_and_splice_var_accesses): Do not check first_link before
13590 (subtree_mark_written_and_enqueue): Likewise.
13591 (propagate_all_subaccesses): Likewise and do not stop at first
13592 parent with a first_link.
13594 2017-06-13 Martin Jambor <mjambor@suse.cz>
13596 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
13599 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13601 * match.pd: New pattern.
13603 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13605 * tree-vrp.c (is_masked_range_test): New function.
13606 (register_edge_assert_for): Determine ranges for
13609 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13611 PR tree-optimization/67328
13612 * fold-const.c (maskable_range_p): New function.
13613 (build_range_check): Generate bittests if possible.
13615 2017-06-13 Martin Liska <mliska@suse.cz>
13617 * gimple-pretty-print.c (dump_probability): Add new argument.
13618 (dump_edge_probability): Dump both probability and count.
13619 (dump_gimple_label): Likewise.
13620 (dump_gimple_bb_header): Likewise.
13622 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
13625 * config/avr/avr-devices.c: Fix indentation.
13626 * config/avr/gen-avr-mmcu-specs.c: Dito.
13628 2017-06-13 Richard Biener <rguenther@suse.de>
13630 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
13631 instead get vector type from stmt_info.
13632 (vectorizable_reduction): Adjust. Remove dead code.
13634 2017-06-13 Richard Biener <rguenther@suse.de>
13636 PR middle-end/81065
13637 * fold-const.c (extract_muldiv_1): Remove bogus distribution
13638 case of C * (x * C2 + C3).
13639 (fold_addr_of_array_ref_difference): Properly fold index difference.
13641 2017-06-12 David S. Miller <davem@davemloft.net>
13644 * config/sparc/sparc.md (return expander): Emit frame blockage if
13645 function uses alloca.
13647 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
13649 * combine.c (make_field_assignment): Check len rather than the mode
13650 precision when calling force_to_mode.
13652 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
13654 Support multilibs and devices that see flash in RAM address range.
13657 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
13658 (avr_mcu_t) <flash_pm_offset>: New field.
13659 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
13660 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
13661 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
13662 (AVR_TINY_PM_OFFSET): Remove macro.
13663 * config/avr/avr.opt (-mshort-calls): New option.
13664 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
13665 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
13666 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
13667 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
13668 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
13669 instead of avr_arch->have_jmp_call.
13670 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
13671 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
13672 avr_arch->flash_pm_offset to define.
13673 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
13674 new field flash_pm_offset. Add entry for avrxmega3.
13675 (avr_texinfo): Add entry for avrxmega3.
13676 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
13677 attiny212, attiny214,
13678 attiny412, attiny414, attiny416, attiny417,
13679 attiny814, attiny816, attiny817,
13680 attiny1614, attiny1616, attiny1617,
13681 attiny3214, attiny3216, attiny3217.
13682 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
13683 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
13684 (avr_print_operand_address) [AVR_TINY]: Same.
13685 (avr_asm_init_sections) <readonly_data_section>: Only patch
13686 callback if avr_arch->flash_pm_offset = 0.
13687 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
13688 for rodata if avr_arch->flash_pm_offset != 0.
13689 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
13690 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
13691 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
13692 (m_options): Append opt_rcall.
13693 (m_dirnames): Append dir_rcall.
13694 * config/avr/t-multilib: Regenerate.
13696 * configure.ac [target=avr]: Check whether avrxmega3 default
13697 linker description file works as needed.
13698 * configure: Regenerate.
13699 * doc/avr-mmcu.texi: Regenerate.
13700 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
13701 <__AVR_ARCH__>: Document avrxmega3 and 103.
13702 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
13703 <__AVR_SHORT_CALLS__>: Document it.
13704 <__AVR_PM_BASE_ADDRESS__>: Document it.
13705 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
13706 (AVR Variable Attributes) <progmem>: Document this is
13707 not needed for avrxmega3.
13708 (AVR Named Address Spaces) <__flash>: Dito.
13710 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
13712 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
13714 2017-06-12 Doug Rupp <rupp@adacore.com>
13716 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
13717 Append vxworks-stdint.h to the tm_file list.
13718 * config/vxworks-stdint.h: New file.
13720 2017-06-12 Martin Liska <mliska@suse.cz>
13722 PR tree-optimization/81041
13723 * tree-profile.c (gimple_gen_ic_func_profiler):
13724 Create an extra BB in profile-generate
13725 (gimple_gen_time_profiler): Likewise.
13727 2017-06-12 Jakub Jelinek <jakub@redhat.com>
13729 PR tree-optimization/81003
13730 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
13731 (update_range_test): Use it instead of force_gimple_operand_gsi.
13733 2017-06-12 Richard Biener <rguenther@suse.de>
13735 PR tree-optimization/81053
13736 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
13737 with backedge value not defined in loop. Simplify def stmt
13740 2017-06-11 Tom de Vries <tom@codesourcery.com>
13743 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
13745 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
13746 nvptx_cannot_force_const_mem.
13748 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13750 * opts.c (finish_options): Move test for flag_split_stack after
13751 it has been initialized.
13753 2017-06-11 Jason Merrill <jason@redhat.com>
13755 * tree.h (id_equal): New.
13756 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
13757 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
13758 instead of strcmp of IDENTIFIER_POINTER.
13760 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13762 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
13763 (mark_all_inlined_calls_cdtor): Fix formating.
13764 (inline_transform): Rescale profile before inlining.
13766 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13768 * cgraph.h (cgraph_edge::clone): Update prototype.
13769 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
13770 (cgraph_node::create_clone): Update.
13771 (cgraph_node::create_version_clone): Update.
13772 * tree-inline.c (copy_bb): Update.
13773 (expand_call_inline): Update.
13775 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
13777 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
13778 factored out from ...
13779 (rs6000_emit_prologue): ... here.
13781 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
13783 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
13784 factored out from ...
13785 (rs6000_emit_prologue): ... here.
13787 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13789 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
13791 (handle_missing_profiles): Fix computation of tp_first_run.
13792 (counts_to_freqs): Do not touch freqs when count is 0.
13794 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13796 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
13799 2017-06-10 Tom de Vries <tom@codesourcery.com>
13801 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
13802 attributes): Document signal effective target.
13804 2017-06-10 Tom de Vries <tom@codesourcery.com>
13806 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
13807 Document effective target stack_size.
13809 2017-06-09 David Malcolm <dmalcolm@redhat.com>
13811 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
13812 to the edit_context if they can be auto-applied.
13814 2017-06-9 Ian Lance Taylor <iant@golang.org>
13816 * opts.c (finish_options): If -fsplit-stack, disable implicit
13817 -forder-blocks-and-partition.
13818 * doc/invoke.texi (Optimize Options): Document that when using
13819 -fsplit-stack -forder-blocks-and-partition is not implicitly
13822 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13824 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
13825 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
13826 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
13827 * builtins.def (abort, trap, unreachable): Declare cold.
13828 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
13829 * tree-core.h (ECF_COLD): New.
13830 * tree.c (set_call_expr_flags): Handle ECF_COLD.
13831 (build_common_builtin_nodes): Mark unreachable and abort as cold.
13833 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13835 * predict.c (unlikely_executed_stmt_p): Cleanup.
13837 2017-06-09 Richard Biener <rguenther@suse.de>
13839 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
13840 model if the ref is always written to.
13842 2017-06-09 Tamar Christina <tamar.christina@arm.com>
13844 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
13846 2017-06-09 Tamar Christina <tamar.christina@arm.com>
13848 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
13851 2017-06-09 Tom de Vries <tom@codesourcery.com>
13854 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
13855 "target cannot support label values" when encountering LABEL_REF.
13857 2017-06-09 Martin Liska <mliska@suse.cz>
13859 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
13860 (gimple_gen_ic_func_profiler): Emit direct comparison
13861 of __gcov_indirect_call_callee with NULL.
13862 (gimple_gen_time_profiler): Change probability from
13863 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
13865 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13867 * profile.c (edge_gcov_counts): Turn to pointer.
13868 (compute_branch_probabilities, compute_branch_probabilities): Update.
13869 (branch_prob): Do not clear edge_gcov_count.
13870 * profile.h (edge_gcov_counts): Turn to pointer.
13871 (edge_gcov_count): Update.
13873 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13875 * gimple.h (gimple_check_failed): Mark cold.
13877 2017-06-09 Richard Biener <rguenther@suse.de>
13879 PR tree-optimization/66623
13880 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
13881 refactor check_reduction into two parts, properly computing
13882 whether we have to check reduction validity for outer loop
13885 2017-06-09 Richard Biener <rguenther@suse.de>
13887 PR tree-optimization/79483
13888 * graphite-scop-detection.c (order): New global.
13889 (get_order): Compute bb to order mapping that satisfies code
13890 generation constraints.
13891 (cmp_pbbs): New helper.
13892 (build_scops): Start domwalk at entry block, sort generated
13895 2017-06-09 Richard Biener <rguenther@suse.de>
13897 PR middle-end/81007
13898 * ipa-polymorphic-call.c
13899 (ipa_polymorphic_call_context::restrict_to_inner_class):
13900 Skip FIELD_DECLs with error_mark_node type.
13901 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
13904 2017-06-09 Martin Liska <mliska@suse.cz>
13906 * predict.c (struct branch_predictor): New struct.
13907 (test_prediction_value_range): New test.
13908 (predict_c_tests): New function.
13909 * selftest-run-tests.c (selftest::run_tests): Run the function.
13910 * selftest.h: Declare new tests.
13912 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
13915 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
13916 gen_add3_insn did not fail.
13917 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
13918 r0, construct that number in a temporary reg and add that reg to r0.
13919 If asked to put the result in r0 as well, fail.
13921 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
13923 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13924 for early expansion of vec_eqv.
13926 2017-06-08 Jakub Jelinek <jakub@redhat.com>
13928 PR middle-end/81005
13929 * ubsan.c (instrument_null): Avoid pointless code temporary.
13930 (pass_ubsan::execute): Instrument aggregate arguments of calls.
13932 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
13936 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
13939 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
13942 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13944 * predict.c (unlikely_executed_edge_p): Move ahead.
13945 (probably_never_executed_edge_p): Use it.
13947 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
13949 PR middle-end/79988
13950 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
13951 gimple_call_builtin_p call.
13953 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13955 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
13956 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
13957 rtl_check_failed_type2, rtl_check_failed_code1,
13958 rtl_check_failed_code2, rtl_check_failed_code_mode,
13959 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
13960 rtvec_check_failed_bounds, rtl_check_failed_flag,
13961 _fatal_insn_not_found, _fatal_insn): Likewise.
13962 * tree.h (tree_contains_struct_check_failed,
13963 tree_check_failed, tree_not_check_failed,
13964 tree_class_check_failed, tree_range_check_failed,
13965 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
13966 tree_vec_elt_check_failed, phi_node_elt_check_failed,
13967 tree_operand_check_failed, omp_clause_check_failed,
13968 omp_clause_operand_check_failed, omp_clause_range_check_failed):
13971 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13973 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
13974 flag_branch_probabilities.
13975 * ipa-inline.c (edge_badness): Likewise.
13976 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
13977 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
13978 * predict.c (maybe_hot_frequency_p): Likewise.
13979 (probably_never_executed): Likewise.
13980 * sched-ebb.c (schedule_ebbs): Likewise.
13981 * sched-rgn.c (find_single_block_region): Likewise.
13982 * tracer.c (tail_duplicate): Likewise.
13984 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13986 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
13987 longer requires x_flag_profile_use.
13989 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13991 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
13992 instead of flag_reorder_blocks_and_partition.
13993 * dbxout.c (dbxout_function_end): Likewise.
13994 * dwarf2out.c (gen_subprogram_die): Likewise.
13995 * haifa-sched.c (sched_create_recovery_edges): Likewise.
13996 * hw-doloop.c (reorg_loops): Likewise.
13997 * varasm.c (assemble_start_function,
13998 assemble_end_function): Likewise.
13999 (decide_function_section): Do not check for
14000 flag_reorder_blocks_and_partition.
14002 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
14004 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
14006 (chkp_get_hard_register_fake_addr_expr): Ditto.
14007 (chkp_build_addr_expr): Add check for hard reg case.
14008 (chkp_parse_array_and_component_ref): Ditto.
14009 (chkp_find_bounds_1): Ditto.
14010 (chkp_process_stmt): Don't generate bounds store for
14013 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14015 * predict.c (maybe_hot_bb_p): Do not check profile status.
14016 (maybe_hot_edge_p): Likewise.
14017 (probably_never_executed): Check for zero counts even if profile
14019 (unlikely_executed_edge_p): New function.
14020 (unlikely_executed_stmt_p): New function.
14021 (unlikely_executed_bb_p): New function.
14022 (set_even_probabilities): Use unlikely predicates.
14023 (combine_predictions_for_bb): Likewise.
14024 (predict_paths_for_bb): Likewise.
14025 (predict_paths_leading_to_edge): Likewise.
14026 (determine_unlikely_bbs): New function.
14027 (estimate_bb_frequencies): Use it.
14028 (compute_function_frequency): Use zero counts even if profile is
14030 * profile-count.h: Fix typo.
14032 2017-08-08 Julia Koval <julia.koval@intel.com>
14034 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
14035 _mm512_mask_cvtsepi16_storeu_epi8,
14036 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
14037 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
14038 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
14039 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
14040 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
14041 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
14042 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
14043 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
14044 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
14045 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
14046 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
14047 __builtin_ia32_pmovuswb256mem_mask,
14048 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
14049 __builtin_ia32_pmovwb512mem_mask): New builtins.
14051 2017-08-08 Julia Koval <julia.koval@intel.com>
14053 PR target/73350,80862
14054 * config/i386/subst.md (round): Fix round pattern.
14055 * config/i386/i386.c (ix86_erase_embedded_rounding):
14056 Fix erasing rounding for the fixed pattern.
14058 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14060 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
14062 2017-06-08 Martin Liska <mliska@suse.cz>
14064 PR gcov-profile/80911
14065 * gcov.c (block_info::block_info): New constructor.
14067 2017-06-07 Carl Love <cel@us.ibm.com>
14069 * config/rs6000/rs6000-c: The return type of the following
14070 built-in functions was implemented as int not long long. Fix sign
14071 of return value for the unsigned version of vec_mulo and vec_mule.
14072 vector unsigned long long vec_bperm (vector unsigned long long,
14073 vector unsigned char)
14074 vector signed long long vec_mule (vector signed int,
14076 vector unsigned long long vec_mule (vector unsigned int,
14077 vector unsigned int)
14078 vector signed long long vec_mulo (vector signed int,
14080 vector unsigned long long vec_mulo (vector unsigned int,
14081 vector unsigned int)
14082 * doc/extend.texi: Fix the documentation for the built-in
14085 2017-06-07 Carl Love <cel@us.ibm.com>
14088 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
14091 2017-06-07 Carl Love <cel@us.ibm.com>
14093 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
14094 support, Generate doublehv for signed int/float for BE case only.
14096 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
14098 * doc/invoke.texi (mcx16): Rewrite.
14100 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14102 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
14103 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
14104 *mov<mode>_softfloat, and an anonymous splitter): Use
14105 nonimmediate_operand instead of rs6000_nonimmediate_operand.
14107 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14109 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
14111 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
14112 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
14113 (rs6000_debug_reg_global): Adjust.
14114 (rs6000_init_hard_regno_mode_ok): Adjust.
14115 (rs6000_dbx_register_number): Adjust.
14116 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
14117 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
14118 Remove SPE_ACC and SPEFSCR.
14119 (REG_ALLOC_ORDER): Ditto.
14120 (FRAME_POINTER_REGNUM): Change to 111.
14121 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
14122 (REG_CLASS_NAMES): Ditto.
14123 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
14124 (REGISTER_NAMES): Ditto.
14125 (ADDITIONAL_REG_NAMES): Ditto.
14126 (rs6000_reg_names): Ditto.
14127 * config/rs6000/rs6000.md: Renumber some register number
14130 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14132 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
14134 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
14135 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
14137 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
14138 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
14139 Delete the SPE high registers.
14140 (REG_ALLOC_ORDER): Ditto.
14141 (enum reg_class): Remove SPE_HIGH_REGS.
14142 (REG_CLASS_NAMES): Ditto.
14143 (REG_CLASS_CONTENTS): Delete the SPE high registers.
14144 (REGISTER_NAMES): Ditto.
14145 (rs6000_reg_names): Ditto.
14146 * doc/tm.texi.in: Remove SPE as example.
14147 * doc/tm.texi: Regenerate.
14149 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14151 * config/rs6000/8540.md (ppc8540_brinc): Delete.
14152 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
14153 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
14154 * config/rs6000/rs6000.md (type): Remove "brinc".
14156 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14158 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
14159 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
14160 * config/rs6000/linuxspe.h: Delete file.
14161 * config/rs6000/rs6000.md: Don't include spe.md.
14162 * config/rs6000/spe.h: Delete file.
14163 * config/rs6000/spe.md: Delete file.
14164 * config/rs6000/t-rs6000: Remove spe.md.
14166 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14168 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
14169 (reg_or_none500mem_operand): Delete.
14170 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
14171 instead of reg_or_none500mem_operand.
14173 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14175 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
14176 handling of SPE flags.
14177 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
14179 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14181 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
14183 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
14184 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
14185 paired_divv2sf3): Similar.
14186 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
14187 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
14188 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
14190 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
14191 Rename the paired_* instruction patterns.
14192 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
14194 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
14195 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
14196 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
14197 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
14198 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
14199 PAIRED_VECTOR_MODE.
14200 (struct machine_function): Delete field spe_insn_chain_scanned_p.
14201 (spe_func_has_64bit_regs_p): Delete.
14202 (spe_expand_predicate_builtin): Delete.
14203 (spe_expand_evsel_builtin): Delete.
14204 (TARGET_DWARF_REGISTER_SPAN): Do not define.
14205 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
14206 (invalid_e500_subreg): Delete.
14207 (rs6000_legitimize_address): Always force_reg op2 as well, for
14208 paired single memory accesses.
14209 (rs6000_member_type_forces_blk): Delete.
14210 (rs6000_spe_function_arg): Delete.
14211 (rs6000_expand_unop_builtin): Delete SPE handling.
14212 (rs6000_expand_binop_builtin): Ditto.
14213 (spe_expand_stv_builtin): Delete.
14214 (bdesc_2arg_spe): Delete.
14215 (spe_expand_builtin): Delete.
14216 (spe_expand_predicate_builtin): Delete.
14217 (spe_expand_evsel_builtin): Delete.
14218 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
14219 (spe_init_builtins): Delete.
14220 (spe_func_has_64bit_regs_p): Delete.
14221 (savres_routine_name): Delete "info" parameter. Adjust callers.
14222 (rs6000_emit_stack_reset): Ditto.
14223 (rs6000_dwarf_register_span): Delete.
14224 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
14225 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
14226 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
14228 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
14230 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
14231 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
14232 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
14233 mulv2sf3, divv2sf3): Delete expanders.
14235 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14237 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
14239 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14241 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
14242 * config/rs6000/rs6000.c: Ditto.
14244 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14246 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
14247 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
14248 comparison_operator.
14250 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14252 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
14253 * config/rs6000/rs6000.opt: Ditto.
14254 * config/rs6000/t-rtems: Ditto.
14256 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14258 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
14259 TARGET_E500_SINGLE by 0, simplify.
14260 * config/rs6000/rs6000.c: Ditto.
14261 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
14262 (spe_build_register_parallel): Delete.
14263 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
14264 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
14265 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
14266 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
14267 (E500_CONVERT): Delete.
14268 * config/rs6000/spe.md: Remove many patterns and all define_constants.
14270 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14272 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
14273 * config/rs6000/dfp.md: Ditto.
14274 (negdd2, *negdd2_fpr): Merge.
14275 (absdd2, *absdd2_fpr): Merge.
14276 (negtd2, *negtd2_fpr): Merge.
14277 (abstd2, *abstd2_fpr): Merge.
14278 * config/rs6000/e500.h: Delete file.
14279 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
14280 TARGET_FPRS by 1 and simplify.
14281 * config/rs6000/rs6000-c.c: Ditto.
14282 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
14283 TARGET_DF_SPE by 0.
14284 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
14286 * config/rs6000/rs6000.md: Ditto.
14287 (floatdidf2, *floatdidf2_fpr): Merge.
14288 (move_from_CR_gt_bit): Delete.
14289 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
14290 (E500_CR_IOR_COMPARE): Delete.
14291 (All patterns that require !TARGET_FPRS): Delete.
14292 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
14294 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14296 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
14298 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14300 * graphds.c (add_edge): Intitialize edge's attached data.
14301 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
14302 pointer parameter. Call pointed function on each edge during
14303 graph traversing. Skip traversing the edge when the function
14305 (graphds_dfs, graphds_scc): Ditto.
14306 (for_each_edge): New parameter. Pass the new parameter to callback
14308 * graphds.h (skip_edge_callback): New function pointer type.
14309 (graphds_dfs, graphds_scc): New function pointer parameter.
14310 (graphds_edge_callback, for_each_edge): New parameter.
14312 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14314 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
14315 out code checking if runtime alias check is possible to below ...
14316 Call the new function.
14317 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
14318 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
14320 2017-06-07 Marek Polacek <polacek@redhat.com>
14323 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
14324 TYPE_OVERFLOW_WRAPS check.
14326 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14328 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
14329 if versioning is required.
14330 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
14331 peeling with the check for versioning.
14333 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14335 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
14336 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
14337 Set true to new parameter if new ssa variable is defined.
14338 (vect_gen_vector_loop_niters): Refactor. Set range information
14339 for the new vector loop bound variable.
14340 (vect_do_peeling): Ditto.
14342 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14344 * tree-affine.c (ssa.h): Include header file.
14345 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
14346 has wrapping overflow behavior.
14348 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14350 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
14352 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14354 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
14355 (tree_to_aff_combination): ... here.
14357 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14359 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
14360 reg_pressure model function.
14361 (ivopts_global_cost_for_size): Delete.
14362 (determine_set_costs, iv_ca_recount_cost): Call new model function
14363 ivopts_estimate_reg_pressure.
14365 2017-06-07 Tamar Christina <tamar.christina@arm.com>
14367 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
14368 expensive than udiv. Remove floating point cases from mod.
14370 2017-06-07 Tamar Christina <tamar.christina@arm.com>
14372 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
14373 Increase idiv cost.
14375 2017-06-07 Tamar Christina <tamar.christina@arm.com>
14377 * config/aarch64/aarch64.md
14378 (copysignsf3): Fix mask generation.
14380 2017-06-07 Jakub Jelinek <jakub@redhat.com>
14382 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
14384 (class dump_manager): Add register_dumps method.
14385 * dumpfile.c: Include langhooks.h.
14386 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
14387 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
14388 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
14389 (dump_manager::dump_register): Start with 512 entries instead of 32.
14390 (dump_manager::register_dumps): New method.
14391 * toplev.c (general_init): Instead of invoking register_dumps
14392 langhook, invoke register_dumps method on the dump manager.
14393 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
14396 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
14398 * doc/md.texi: Clarify the restrictions on a define_insn condition.
14399 Say that # requires an associated define_split to exist, and that
14400 the define_split must be suitable for use after register allocation.
14402 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14404 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
14405 (compute_outgoing_frequencies): Also initialize zero counts.
14406 (find_many_sub_basic_blocks): Do not produce uninitialized profile
14407 around loops; preserve more of profile when nothing changes.
14409 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
14411 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
14413 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
14414 * config/arm/arm-cpu-cdata.h: Regenerate.
14415 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
14416 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
14417 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
14418 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
14419 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
14421 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
14423 * config/arm/t-rmprofile: Likewise.
14424 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
14426 2017-06-06 David S. Miller <davem@davemloft.net>
14429 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
14430 blockage if function uses alloca.
14432 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14434 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
14435 New "uid" fields to hold pretty-print IDs of group and ref.
14436 Memory references are now identified as <group_id>:<ref_id>
14437 instead of using [random] addresses.
14438 (dump_mem_details): Simplify, no functional change.
14439 (dump_mem_ref): Simplify and make output more concise.
14440 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
14441 (find_or_create_group): Initialize group uid.
14442 (record_ref): Initialize ref uid. Improve debug output.
14443 (prune_group_by_reuse, should_issue_prefetch_p,)
14444 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
14445 (mark_nontemporal_store, determine_loop_nest_reuse):
14446 Improve debug output.
14448 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14450 * dbgcnt.def (prefetch): New debug counter.
14451 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
14452 (schedule_prefetches): Stop issueing prefetches if debug counter
14455 2017-06-06 Tom de Vries <tom@codesourcery.com>
14457 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
14458 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
14460 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14462 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
14463 Use aarch64_reg_or_zero predicate for operand 4.
14464 (aarch64_compare_and_swap<mode> define_insn_and_split):
14465 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
14466 (aarch64_store_exclusive<mode>): Likewise for operand 2.
14468 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
14470 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
14471 (arm_compute_save_core_reg_mask): This.
14472 (thumb1_compute_save_reg_mask): Rename into ...
14473 (thumb1_compute_save_core_reg_mask): This.
14474 (arm_compute_save_reg0_reg12_mask): Adapt comment.
14475 (arm_compute_frame_layout): Likewise.
14477 2017-06-06 Richard Biener <rguenther@suse.de>
14479 PR tree-optimization/80974
14480 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
14481 keep or clear leaders SSA info.
14483 2017-06-06 Tom de Vries <tom@codesourcery.com>
14485 * config/nvptx/nvptx.c (split_mode_p): New function.
14486 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
14488 2017-06-06 Tom de Vries <tom@codesourcery.com>
14490 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
14492 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14495 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
14498 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14500 * shrink-wrap.c (handle_simple_exit): Update profile.
14501 (try_shrink_wrapping): Upate profile.
14503 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14505 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
14506 (tree_guess_outgoing_edge_probabilities): New.
14507 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
14508 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
14510 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14512 * ipa-split.c (split_function): Initialize return bb profile.
14514 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14516 * profile.c (compute_branch_probabilities): Also initialize
14517 EXIT_BLOCK profile.
14519 2017-06-06 Richard Biener <rguenther@suse.de>
14521 PR tree-optimization/80928
14522 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
14523 (vect_analyze_loop_operations): Properly guard analysis for
14525 (vect_transform_loop): Likewise.
14526 (vect_analyze_loop_2): Also reset SLP type on PHIs.
14527 (vect_model_induction_cost): Do not cost for pure SLP.
14528 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
14529 of induction in inner loop vectorization.
14530 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
14531 (vect_get_and_check_slp_defs): Handle vect_induction_def.
14532 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
14534 (vect_analyze_slp_cost_1): Cost induction.
14535 (vect_detect_hybrid_slp_stmts): Handle PHIs.
14536 (vect_get_slp_vect_defs): Likewise.
14537 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
14538 (vect_transform_stmt): Handle SLP reductions.
14539 * tree-vectorizer.h (vectorizable_induction): Adjust.
14541 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
14543 * config/rs6000/rs6000.c (make_resolver_func): Update
14544 init_lowered_empty_function call.
14546 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
14548 * doc/invoke.texi: Document the -fprofile-abs-path option.
14549 * common.opt (fprofile-abs-path): New option.
14550 * gcov-io.h (gcov_write_filename): Declare.
14551 * gcov-io.c (gcov_write_filename): New function.
14552 * coverage.c (coverage_begin_function): Use gcov_write_filename.
14553 * profile.c (output_location): Likewise.
14555 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14557 * shring-wrap.c: Revert accidental commit.
14559 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
14561 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
14563 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14565 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
14567 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
14568 profile in callgraph edge.
14569 * profile-count.h (apply_probability): If THIS is 0, then result is 0
14570 (apply_scale): Likewise.
14571 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
14572 Also scale profile when inlining function with zero profile.
14573 (initialize_cfun): Update exit block profile even when it is zero.
14574 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
14575 when profile is read.
14577 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
14579 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
14580 (CLONE_*): New constants to define the processors we can generate
14581 code for with the target_clone attribute.
14582 (rs6000_clone_map): New array to identify which clone processors
14583 the current program is running on.
14584 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
14585 target_clone attribute.
14586 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
14587 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
14588 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
14589 (cpu_expand_builtin): Add support for target_clone attribute.
14590 (rs6000_valid_attribute_p): Allow "default" attribute.
14591 (get_decl_name): New debug function to simplify printing the
14592 current function name in debugging statements.
14593 (rs6000_clone_priority): New functions to support the target_clone
14594 attribute, and be able to generate code to switch between ISA 2.05
14595 through ISA 3.0 (power6 through power9).
14596 (rs6000_compare_version_priority): Likewise.
14597 (rs6000_get_function_versions_dispatcher): Likewise.
14598 (make_resolver_func): Likewise.
14599 (add_condition_to_bb): Likewise.
14600 (dispatch_function_versions): Likewise.
14601 (rs6000_generate_version_dispatcher_body): Likewise.
14602 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
14603 (fusion_gpr_load_p): Fix a spacing issue.
14604 * doc/extend.texi (Common Function Attributes): Document that the
14605 PowerPC supports the target_clone attribute.
14607 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
14609 * config/arm/arm.h: explain F symbol found in description of ARM
14610 register allocation in its legend.
14612 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14614 * config/mips/frame-header-opt.c: Include profile-count.h.
14615 * config/riscv/riscv.c: Include profile-count.h
14617 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14619 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
14621 (sm_set_flag_if_changed): Add bbs field.
14622 (execute_sm_if_changed_flag_set): Pass BBS.
14623 (execute_sm): Update.
14625 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14627 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
14630 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14632 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
14633 (peephole2): New peephole2 to emit the above.
14634 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
14636 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14638 * config/aarch64/aarch64.c (define_peephole2 above
14639 *sub_<shift>_<mode>): New peephole.
14641 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14643 * config/i386/i386.c (make_resolver_func): Update.
14644 * Makefile.in: Add profile-count.h and profile-count.o
14645 * auto-profile.c (afdo_indirect_call): Update to new API.
14646 (afdo_set_bb_count): Update.
14647 (afdo_propagate_edge): Update.
14648 (afdo_propagate_circuit): Update.
14649 (afdo_calculate_branch_prob): Update.
14650 (afdo_annotate_cfg): Update.
14651 * basic-block.h: Include profile-count.h
14652 (struct edge_def): Turn count to profile_count.
14653 (struct basic_block_def): Likewie.
14654 (REG_BR_PROB_BASE): Move to profile-count.h
14655 (RDIV): Move to profile-count.h
14656 * bb-reorder.c (max_entry_count): Turn to profile_count.
14657 (find_traces): Update.
14658 (rotate_loop):Update.
14659 (connect_traces):Update.
14660 (sanitize_hot_paths):Update.
14661 * bt-load.c (migrate_btr_defs): Update.
14662 * cfg.c (RDIV): Remove.
14663 (init_flow): Use alloc_block.
14664 (alloc_block): Uninitialize count.
14665 (unchecked_make_edge): Uninitialize count.
14666 (check_bb_profile): Update.
14667 (dump_edge_info): Update.
14668 (dump_bb_info): Update.
14669 (update_bb_profile_for_threading): Update.
14670 (scale_bbs_frequencies_int): Update.
14671 (scale_bbs_frequencies_gcov_type): Update.
14672 (scale_bbs_frequencies_profile_count): New.
14673 * cfg.h (update_bb_profile_for_threading): Update.
14674 (scale_bbs_frequencies_profile_count): Declare.
14675 * cfgbuild.c (compute_outgoing_frequencies): Update.
14676 (find_many_sub_basic_blocks): Update.
14677 * cfgcleanup.c (try_forward_edges): Update.
14678 (try_crossjump_to_edge): Update.
14679 * cfgexpand.c (expand_gimple_tailcall): Update.
14680 (construct_exit_block): Update.
14681 * cfghooks.c (verify_flow_info): Update.
14682 (dump_bb_for_graph): Update.
14683 (split_edge): Update.
14684 (make_forwarder_block): Update.
14685 (duplicate_block): Update.
14686 (account_profile_record): Update.
14687 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
14688 (get_estimated_loop_iterations): Update.
14689 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
14690 (single_likely_exit): Update.
14691 * cfgloopmanip.c (scale_loop_profile): Update.
14693 (set_zero_probability): Update.
14694 (lv_adjust_loop_entry_edge): Update.
14695 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
14696 (purge_dead_edges): Update.
14697 (rtl_account_profile_record): Update.
14698 * cgraph.c (cgraph_node::create): Uninitialize count.
14699 (symbol_table::create_edge): Uninitialize count.
14700 (cgraph_update_edges_for_call_stmt_node): Update.
14701 (cgraph_edge::dump_edge_flags): Update.
14702 (cgraph_node::dump): Update.
14703 (cgraph_edge::maybe_hot_p): Update.
14704 * cgraph.h: Include profile-count.h
14705 (create_clone), create_edge, create_indirect_edge): Update.
14706 (cgraph_node): Turn count to profile_count.
14707 (cgraph_edge0: Likewise.
14708 (make_speculative, clone): Update.
14709 (create_edge): Update.
14710 (init_lowered_empty_function): Update.
14711 * cgraphclones.c (cgraph_edge::clone): Update.
14712 (duplicate_thunk_for_node): Update.
14713 (cgraph_node::create_clone): Update.
14714 * cgraphunit.c (cgraph_node::analyze): Update.
14715 (cgraph_node::expand_thunk): Update.
14716 * final.c (dump_basic_block_info): Update.
14717 * gimple-streamer-in.c (input_bb): Update.
14718 * gimple-streamer-out.c (output_bb): Update.
14719 * graphite.c (print_global_statistics): Update.
14720 (print_graphite_scop_statistics): Update.
14721 * hsa-brig.c: Include basic-block.h.
14722 * hsa-dump.c: Include basic-block.h.
14723 * hsa-gen.c (T sum_slice): Update.
14724 (convert_switch_statements):Update.
14725 * hsa-regalloc.c: Include basic-block.h.
14726 * ipa-chkp.c (chkp_produce_thunks): Update.
14727 * ipa-cp.c (struct caller_statistics): Update.
14728 (init_caller_stats): Update.
14729 (gather_caller_stats): Update.
14730 (ipcp_cloning_candidate_p): Update.
14731 (good_cloning_opportunity_p): Update.
14732 (get_info_about_necessary_edges): Update.
14733 (dump_profile_updates): Update.
14734 (update_profiling_info): Update.
14735 (update_specialized_profile): Update.
14736 (perhaps_add_new_callers): Update.
14737 (decide_about_value): Update.
14738 (ipa_cp_c_finalize): Update.
14739 * ipa-devirt.c (struct odr_type_warn_count): Update.
14740 (struct decl_warn_count): Update.
14741 (struct final_warning_record): Update.
14742 (possible_polymorphic_call_targets): Update.
14743 (ipa_devirt): Update.
14744 * ipa-fnsummary.c (redirect_to_unreachable): Update.
14745 * ipa-icf.c (sem_function::merge): Update.
14746 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
14747 * ipa-inline.c (compute_uninlined_call_time): Update.
14748 (compute_inlined_call_time): Update.
14749 (want_inline_small_function_p): Update.
14750 (want_inline_self_recursive_call_p): Update.
14751 (edge_badness): Update.
14752 (lookup_recursive_calls): Update.
14753 (recursive_inlining): Update.
14754 (inline_small_functions): Update.
14755 (dump_overall_stats): Update.
14756 (dump_inline_stats): Update.
14757 * ipa-profile.c (ipa_profile_generate_summary): Update.
14758 (ipa_propagate_frequency): Update.
14759 (ipa_profile): Update.
14760 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14761 * ipa-utils.c (ipa_merge_profiles): Update.
14762 * loop-doloop.c (doloop_modify): Update.
14763 * loop-unroll.c (report_unroll): Update.
14764 (unroll_loop_runtime_iterations): Update.
14765 * lto-cgraph.c (lto_output_edge): Update.
14766 (lto_output_node): Update.
14767 (input_node): Update.
14768 (input_edge): Update.
14769 (merge_profile_summaries): Update.
14770 * lto-streamer-in.c (input_cfg): Update.
14771 * lto-streamer-out.c (output_cfg): Update.
14772 * mcf.c (create_fixup_graph): Update.
14773 (adjust_cfg_counts): Update.
14774 (sum_edge_counts): Update.
14775 * modulo-sched.c (sms_schedule): Update.
14776 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
14777 * predict.c (maybe_hot_count_p): Update.
14778 (probably_never_executed): Update.
14779 (dump_prediction): Update.
14780 (combine_predictions_for_bb): Update.
14781 (propagate_freq): Update.
14782 (handle_missing_profiles): Update.
14783 (counts_to_freqs): Update.
14784 (rebuild_frequencies): Update.
14785 (force_edge_cold): Update.
14786 * predict.h: Include profile-count.h
14787 (maybe_hot_count_p, counts_to_freqs): UPdate.
14788 * print-rtl-function.c: Do not include cfg.h
14789 * print-rtl.c: Include basic-block.h
14790 * profile-count.c: New file.
14791 * profile-count.h: New file.
14792 * profile.c (is_edge_inconsistent): Update.
14793 (correct_negative_edge_counts): Update.
14794 (is_inconsistent): Update.
14795 (set_bb_counts): Update.
14796 (read_profile_edge_counts): Update.
14797 (compute_frequency_overlap): Update.
14798 (compute_branch_probabilities): Update; Initialize and deinitialize
14800 (branch_prob): Update.
14801 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
14802 (edge_gcov_count): New.
14803 (bb_gcov_count): New.
14804 * shrink-wrap.c (try_shrink_wrapping): Update.
14805 * tracer.c (better_p): Update.
14806 * trans-mem.c (expand_transaction): Update.
14807 (ipa_tm_insert_irr_call): Update.
14808 (ipa_tm_insert_gettmclone_call): Update.
14809 * tree-call-cdce.c: Update.
14810 * tree-cfg.c (gimple_duplicate_sese_region): Update.
14811 (gimple_duplicate_sese_tail): Update.
14812 (gimple_account_profile_record): Update.
14813 (execute_fixup_cfg): Update.
14814 * tree-inline.c (copy_bb): Update.
14815 (copy_edges_for_bb): Update.
14816 (initialize_cfun): Update.
14817 (freqs_to_counts): Update.
14818 (copy_cfg_body): Update.
14819 (expand_call_inline): Update.
14820 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
14821 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
14822 (try_unroll_loop_completely): Update.
14823 (try_peel_loop): Update.
14824 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14825 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
14826 * tree-ssa-loop-split.c (connect_loops): Update.
14827 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
14828 * tree-ssa-reassoc.c (branch_fixup): Update.
14829 * tree-ssa-tail-merge.c (replace_block_by): Update.
14830 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
14831 (compute_path_counts): Update.
14832 (update_profile): Update.
14833 (recompute_probabilities): Update.
14834 (update_joiner_offpath_counts): Update.
14835 (estimated_freqs_path): Update.
14836 (freqs_to_counts_path): Update.
14837 (clear_counts_path): Update.
14838 (ssa_fix_duplicate_block_edges): Update.
14839 (duplicate_thread_path): Update.
14840 * tree-switch-conversion.c (case_bit_test_cmp): Update.
14841 (struct switch_conv_info): Update.
14842 * tree-tailcall.c (decrease_profile): Update.
14843 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
14844 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14845 * value-prof.c (check_counter): Update.
14846 (gimple_divmod_fixed_value): Update.
14847 (gimple_mod_pow2): Update.
14848 (gimple_mod_subtract): Update.
14849 (gimple_ic_transform): Update.
14850 (gimple_stringop_fixed_value): Update.
14851 * value-prof.h (gimple_ic): Update.
14853 2017-06-02 Carl Love <cel@us.ibm.com>
14855 * config/rs6000/rs6000-c: Add support for built-in functions
14856 vector double vec_doublee (vector signed int);
14857 vector double vec_doublee (vector unsigned int);
14858 vector double vec_doublee (vector float);
14859 vector double vec_doubleh (vector signed int);
14860 vector double vec_doubleh (vector unsigned int);
14861 vector double vec_doubleh (vector float);
14862 vector double vec_doublel (vector signed int);
14863 vector double vec_doublel (vector unsigned int);
14864 vector double vec_doublel (vector float);
14865 vector double vec_doubleo (vector signed int);
14866 vector double vec_doubleo (vector unsigned int);
14867 vector double vec_doubleo (vector float);.
14868 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
14869 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
14871 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
14872 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
14873 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
14875 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
14876 vec_doublel, vec_doubleh.
14877 * doc/extend.texi: Update the built-in documentation file for the
14878 new built-in functions.
14880 2017-06-02 David Malcolm <dmalcolm@redhat.com>
14883 * ipa-inline-analysis.c (free_growth_caches): Set
14884 edge_removal_hook_holder to NULL after removing it.
14886 2017-06-02 Sudakshina Das <sudi.das@arm.com>
14888 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
14889 comparision with zero.
14891 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
14892 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14893 for early expansion of vec_min and vec_max builtins.
14894 (builtin_function_type): Add min/max unsigned variants to those
14895 identified as having unsigned arguments.
14897 2017-06-02 Olivier Hainque <hainque@adacore.com>
14899 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
14901 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14903 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
14904 Use VALL_F16 iterator rather than VALL.
14906 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14908 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
14909 Emit CBNZ inside loop when doing a strong exchange and comparing
14910 against zero. Generate the CC flags after the loop.
14912 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
14914 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
14915 (dl_section_ref): New.
14916 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
14917 On AIX, append an expression to subtract the size of the
14918 section length to dl_section_ref.
14920 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
14922 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14923 for early expansion of vector absolute builtins.
14925 2017-06-02 Richard Biener <rguenther@suse.de>
14927 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
14928 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
14930 2017-06-02 Richard Biener <rguenther@suse.de>
14932 PR tree-optimization/80948
14933 * tree-tailcall.c (find_tail_calls): Track stmts to move in
14934 stmt order as well.
14936 2017-06-02 Richard Biener <rguenther@suse.de>
14938 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
14940 * tree-vect-stmts.c (process_use): Do not mark backedge defs
14941 for inductions as relevant.
14943 2017-06-02 Richard Biener <rguenther@suse.de>
14945 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
14946 (vectorizable_induction): ... this. Remove dead code.
14948 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
14950 * builtins. (expand_builtin_alloca): Remove second parameter and
14951 infer its value from the first parameter instead.
14952 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
14954 2017-06-02 Jakub Jelinek <jakub@redhat.com>
14956 PR rtl-optimization/80903
14957 * loop-doloop.c (add_test): Unshare sequence.
14959 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
14961 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
14963 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
14965 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
14967 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
14968 xlogue_layout::get_instance, logue_layout::xlogue_layout,
14969 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
14970 (xlogue_layout::get_stub_rtx): Make static.
14971 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
14972 (xlogue_layout::compute_stub_managed_regs): Rename to...
14973 (xlogue_layout::count_stub_managed_regs): ...this.
14974 (xlogue_layout::is_stub_managed_reg): New function.
14975 (xlogue_layout::m_stub_names): Rename to...
14976 (xlogue_layout::s_stub_names): ...this, make static.
14977 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
14978 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
14979 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
14980 xlogue_layout::s_stub_names): Instantiate statics.
14981 (stub_managed_regs): Remove.
14982 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
14983 (disable_call_ms2sysv_xlogues): Rename to...
14984 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
14985 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
14987 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
14988 change after reload_completed.
14989 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
14991 (ix86_expand_prologue): Likewise.
14992 (ix86_expand_epilogue): Likewise.
14993 (ix86_expand_split_stack_prologue): Likewise.
14994 (ix86_compute_frame_layout): Remove frame parameter ...
14995 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
14996 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
14998 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
14999 (ix86_frame): Move from here ...
15000 * config/i386/i386.h (ix86_frame): ... to here.
15001 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
15002 complete ix86_frame data structure instead. Remove some_ld_name.
15004 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
15006 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
15007 symbols that hold a DECL_VALUE_EXPR.
15009 2017-06-01 Martin Jambor <mjambor@suse.cz>
15011 PR tree-optimization/80898
15012 * tree-sra.c (process_subtree_disqualification): Removed.
15013 (disqualify_candidate): Do not acll
15014 process_subtree_disqualification.
15015 (subtree_mark_written_and_enqueue): New function.
15016 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
15017 RHS has been disqualified and re-queue LHS if necessary. Apart
15018 from that, ignore disqualified RHS.
15020 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15022 * config/s390/s390.c (s390_emit_epilogue): Disable early return
15023 address fetch for z10 or later.
15025 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15027 * config/arc/arc.md (tst_movb): Add guard when splitting.
15029 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15031 * config/arc/arc.c (arc_can_eliminate): Test against
15032 arc_frame_pointer_needed.
15034 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15036 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
15037 to prevent store reordering.
15038 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
15039 (type): Add block type.
15040 (stack_tie): Define special instruction to be used in
15043 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15045 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
15046 constraint. It is not valid for the pattern.
15047 (noncommutative_binary_comparison): Likewise.
15049 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15051 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
15054 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15056 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
15057 be used by the reg-alloc.
15059 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15061 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
15062 reg-alloc when having mul64 or mul32x16 instructions.
15063 (mulsidi3): Likewise.
15064 (umulsidi3): Likewise.
15065 (mulsi32x16): New pattern.
15066 (mulsi64): Likewise.
15067 (mulsidi64): Likewise.
15068 (umulsidi64): Likewise.
15069 (MUL32x16_REG): Define.
15070 (mul64_600): Use MUL32x16_REG.
15071 (mac64_600): Likewise.
15072 (umul64_600): Likewise.
15073 (umac64_600): Likewise.
15075 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15077 * config/arc/arc.md (mulsi3_700): Make it commutative.
15079 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
15081 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
15083 (*sign_extendsidi2_insn): Likewise for movstosw.
15085 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
15087 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
15088 the type of the input discriminant value. Convert the
15089 discriminant value of signedness vary.
15091 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
15093 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
15094 Add to -Wall section.
15096 2017-06-01 Richard Biener <rguenther@suse.de>
15098 PR middle-end/66313
15099 * fold-const.c (fold_plusminus_mult_expr): If the factored
15100 factor may be zero use a wrapping type for the inner operation.
15101 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
15102 and handle moved defs.
15103 (process_assignment): Properly guard the unary op case. Return a
15104 tri-state indicating that moving the stmt before the call may allow
15105 to continue. Pass through to_move.
15106 (find_tail_calls): Handle moving unrelated defs before
15109 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
15112 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
15113 splitter result in the canonical way.
15115 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
15117 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
15118 also for 32bit target. Update insn attributes.
15119 (zero-extendsidi2 splitter): Allow all registers for operand 1.
15121 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
15123 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
15124 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
15125 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
15126 (_mm_maskz_min_ss): New intrinsics.
15128 2017-05-31 Martin Liska <mliska@suse.cz>
15130 * tree-vect-loop.c (vect_create_epilog_for_reduction):
15131 Change comment style to one we normally use.
15132 (vectorizable_reduction): Likewise.
15133 (vectorizable_induction): Likewise.
15134 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15135 (vectorizable_call): Likewise.
15136 (vectorizable_simd_clone_call): Likewise.
15137 (vectorizable_conversion): Likewise.
15138 (vectorizable_assignment): Likewise.
15139 (vectorizable_shift): Likewise.
15140 (vectorizable_operation): Likewise.
15141 (vectorizable_store): Likewise.
15142 (vectorizable_load): Likewise.
15143 * tree-vectorizer.h: Likewise.
15145 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
15147 * passes.c (emergency_dump_function): New.
15148 * tree-pass.h (emergency_dump_function): Declare.
15149 * plugin.c (plugins_internal_error_function): Remove.
15150 * plugin.h (plugins_internal_error_function): Remove declaration.
15151 * toplev.c (internal_error_function): New static function. Use it...
15152 (general_init): ...here.
15154 2017-05-31 Graham Markall <graham.markall@embecosm.com>
15156 * config/arc/arc.c (arc_print_operand): Handle constant operands.
15157 (arc_rtx_costs): Add costs for new patterns.
15158 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
15159 * config/arc/predicates.md: Add _1_2_3_operand predicate.
15161 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
15163 * tree-ssa-strlen.c (get_next_strinfo): New function.
15164 (get_stridx_plus_constant): Use it.
15165 (zero_length_string): Likewise.
15166 (adjust_related_strinfos): Likewise.
15167 (adjust_last_stmt): Likewise.
15169 2017-05-31 Richard Biener <rguenther@suse.de>
15172 * config/i386/i386.c (ix86_expand_builtin): Remove assert
15173 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
15175 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
15177 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
15178 loop_vinfo argument and use of dependence distance vectors.
15179 Check instead whether the two references differ only in their
15180 initial value and assume that they have the same alignment if the
15181 difference is a multiple of the vector alignment.
15182 (vect_analyze_data_refs_alignment): Update call accordingly.
15184 2017-05-31 Martin Liska <mliska@suse.cz>
15187 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
15189 2017-05-31 Bin Cheng <bin.cheng@arm.com>
15191 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
15192 (create_intersect_range_checks): Move from ...
15193 * tree-data-ref.c (create_intersect_range_checks_index)
15194 (create_intersect_range_checks): ... to here.
15195 (create_runtime_alias_checks): New function factored from ...
15196 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
15197 here. Call above function.
15198 * tree-data-ref.h (create_runtime_alias_checks): New function.
15200 2017-05-31 Bin Cheng <bin.cheng@arm.com>
15202 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
15203 segment length for dr_b and compute it in wide_int.
15205 2017-05-31 Richard Biener <rguenther@suse.de>
15207 PR tree-optimization/80906
15208 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
15209 and pass through iv_map.
15210 (copy_bb_and_scalar_dependences): Adjust.
15211 (translate_pending_phi_nodes): Likewise.
15212 (copy_loop_close_phi_args): Handle code-generating IVs instead
15215 2017-05-30 David Malcolm <dmalcolm@redhat.com>
15217 * diagnostic-color.c (color_dict): Add "type-diff".
15218 (parse_gcc_colors): Update comment.
15219 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
15220 -fdiagnostics-show-template-tree and -fno-elide-type.
15221 (GCC_COLORS): Add type-diff to example.
15223 (-fdiagnostics-show-template-tree): New.
15224 (-fno-elide-type): New.
15225 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
15226 the pp_format_decoder callback. Call any m_format_postprocessor's
15228 (pretty_printer::pretty_printer): Initialize
15229 m_format_postprocessor.
15230 (pretty_printer::~pretty_printer): Delete any
15231 m_format_postprocessor.
15232 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
15233 (class format_postprocessor): New class.
15234 (struct pretty_printer::format_decoder): Document the new parameters.
15235 (struct pretty_printer::m_format_postprocessor): New field.
15236 * tree-diagnostic.c (default_tree_printer): Update for new
15237 bool and const char ** params.
15238 * tree-diagnostic.h (default_tree_printer): Likewise.
15240 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
15242 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
15243 (lwa_operand): Delete rs6000_gen_cell_microcode test.
15244 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
15245 rs6000_gen_cell_microcode code.
15246 (rs6000_final_prescan_insn): Delete.
15247 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
15248 "warn-cell-microcode" entries.
15249 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
15250 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
15251 throughout. Change cc_reg_not_micro_cr0_operand to
15252 cc_reg_not_cr0_operand throughout.
15253 (*extendhi<mode>2_noload): Delete.
15254 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
15255 (mwarn-cell-microcode): Delete.
15256 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
15257 -mgen-cell-microcode and -mwarn-cell-microcode.
15259 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
15262 * config/i386/constraints.md (Yd): New constraint.
15264 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
15265 and (?Yd, r) alternatives. Update insn attributes.
15266 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
15267 and (?*Yd, r) alternatives. Update insn attributes.
15268 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
15270 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
15272 * gimplify.c (gimplify_modify_expr): Don't create a
15273 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
15276 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
15278 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
15280 2017-05-30 Richard Biener <rguenther@suse.de>
15282 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
15283 and reduc_def fields.
15284 (STMT_VINFO_REDUC_TYPE): New define.
15285 (STMT_VINFO_REDUC_DEF): Likewise.
15286 (vect_force_simple_reduction): Adjust prototype.
15287 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
15288 (vect_is_simple_reduction): Remove check_reduction argument.
15289 (vect_force_simple_reduction): Adjust and set
15290 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
15291 (vectorizable_reduction): Do not re-do reduction analysis
15292 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
15293 * tree-parloops.c (gather_scalar_reductions): Adjust.
15295 2017-05-30 Richard Biener <rguenther@suse.de>
15297 PR middle-end/80901
15298 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
15301 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
15303 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
15304 Introduce unknown_misalignment parameter and remove vf.
15305 (vect_peeling_hash_get_lowest_cost):
15306 Pass unknown_misalignment parameter.
15307 (vect_enhance_data_refs_alignment):
15308 Fix unsupportable data ref treatment.
15310 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15312 * tree-vect-data-refs.c (vect_get_data_access_cost):
15313 Workaround for SLP handling.
15314 (vect_enhance_data_refs_alignment):
15315 Compute costs for doing no peeling at all, compare to the best
15316 peeling costs so far and avoid peeling if cheaper.
15318 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15320 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
15321 Return peeling info and set costs to zero for unlimited cost
15323 (vect_enhance_data_refs_alignment): Also inspect all datarefs
15324 with unknown misalignment. Compute and costs for unknown
15325 misalignment, compare them to the costs for known misalignment
15326 and choose the cheapest for peeling.
15328 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15330 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
15331 (vect_get_peeling_costs_all_drs): Create function.
15332 (vect_peeling_hash_get_lowest_cost):
15333 Use vect_get_peeling_costs_all_drs.
15334 (vect_peeling_supportable): Create function.
15335 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
15337 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15339 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
15340 DR_HAS_NEGATIVE_STEP.
15341 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
15342 (vect_enhance_data_refs_alignment): Use.
15343 (vect_duplicate_ssa_name_ptr_info): Use.
15344 * tree-vectorizer.h (dr_misalignment): Use.
15345 (known_alignment_for_access_p): Use.
15347 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15350 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
15352 (has_section_name): New function.
15354 2017-05-30 Martin Liska <mliska@suse.cz>
15357 * auto-profile.c (get_function_decl_from_block): Fix
15360 2017-05-30 Richard Biener <rguenther@suse.de>
15362 PR middle-end/80876
15363 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
15365 2017-05-30 Martin Liska <mliska@suse.cz>
15367 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
15368 * dumpfile.h (struct dump_file_info): Remove ctors.
15370 2017-05-30 Martin Liska <mliska@suse.cz>
15372 * predict.def: Fix GNU coding style.
15374 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
15376 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
15377 Mark 'to' argument with ATTRIBUTE_UNUSED.
15379 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
15381 * config/xtensa/xtensa.c (xtensa_emit_call): Use
15382 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
15383 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
15386 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
15388 * doc/install.texi (Options specification): Restore entry of
15389 --enable-sjlj-exceptions.
15391 2017-05-27 Michael Eager <eager@eagercon.com>
15394 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
15396 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
15398 * config/microblaze/microblaze.h
15399 (FIXED_REGISTERS): Update in macro.
15400 (CALL_USED_REGISTERS): Update in macro.
15402 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
15404 * doc/install.texi: Add links to macOS binary distributions.
15406 2017-05-27 Jakub Jelinek <jakub@redhat.com>
15410 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15412 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
15414 2017-05-26 Martin Liska <mliska@suse.cz>
15416 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
15418 2017-05-26 Martin Liska <mliska@suse.cz>
15420 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
15421 always leading ';; '.
15422 (dump_bb_info): Likewise.
15423 (brief_dump_cfg): Likewise.
15424 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
15425 * dumpfile.c: Remove usage of TDF_VERBOSE.
15426 * dumpfile.h (enum dump_kind): Likewise.
15427 (dump_gimple_bb_header): Do not use TDF_COMMENT.
15428 * print-tree.c (debug_verbose): Remove.
15429 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
15430 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
15431 * tree-diagnostic.c (default_tree_printer): Replace
15432 TDF_DIAGNOSTIC with TDF_SLIM.
15434 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15436 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
15437 in parameter loop, rather than loop_vinfo.
15438 (create_intersect_range_checks): Ditto.
15439 (vect_create_cond_for_alias_checks): Update call to above functions.
15441 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15443 PR tree-optimization/80815
15444 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
15445 for merging runtime alias checks. Handle negative DR_STEPs.
15447 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15449 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
15451 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
15452 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
15453 out code pruning runtime alias checks.
15454 * tree-data-ref.c (prune_runtime_alias_test_list): New function
15455 factored out from above.
15456 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
15458 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
15460 (prune_runtime_alias_test_list): New decalaration.
15462 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15464 * tree-vect-data-refs.c (compare_tree): Rename and move ...
15465 * tree-data-ref.c (data_ref_compare_tree): ... to here.
15466 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
15467 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
15468 (operator==, comp_dr_with_seg_len_pair): Ditto.
15469 (vect_prune_runtime_alias_test_list): Ditto.
15471 2017-05-26 Martin Liska <mliska@suse.cz>
15474 * params.def: Bound partial-inlining-entry-probability param.
15476 2017-05-26 Marek Polacek <polacek@redhat.com>
15479 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
15482 2017-05-26 Richard Biener <rguenther@suse.de>
15484 PR tree-optimization/80842
15485 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
15488 2017-05-26 Richard Biener <rguenther@suse.de>
15490 PR tree-optimization/80844
15491 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
15493 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
15495 * doc/md.texi (Machine Constraints): Update x86 family
15496 machine constraints section to match 'config/i386/constraints.md'.
15498 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
15500 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
15502 2017-05-25 Nathan Sidwell <nathan@acm.org>
15504 * doc/invoke.texi (--enable-languages): Update documentation.
15506 2017-05-25 Martin Liska <mliska@suse.cz>
15508 * dumpfile.c: Add TDF_FOLDING.
15509 * dumpfile.h (enum dump_kind): Likewise.
15510 * genmatch.c (dt_simplify::gen_1): Use it.
15512 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15514 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
15516 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15518 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
15519 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
15521 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15523 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
15524 * match.pd (X == C): Rewrite it here.
15525 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
15526 with_certain_nonzero_bits2): New predicates.
15527 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
15529 2017-05-24 Nathan Sidwell <nathan@acm.org>
15531 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
15534 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
15537 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
15539 * config/powerpcspe: New port. Files are copied from the rs6000
15540 port, with "rs6000" in filenames replaced by "powerpcspe".
15542 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
15544 PR rtl-optimization/80754
15545 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
15547 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
15549 * config/sparc/sparc.md (length): Return the correct value for -mflat
15550 sibcalls to match output_sibcall.
15552 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
15556 * config/rs6000/rs6000.c (struct machine_function): Add new field
15558 (rs6000_get_separate_components): Init that field, use it.
15559 (rs6000_components_for_bb): Use the field.
15561 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
15563 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
15565 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
15567 PR middle-end/80823
15568 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
15570 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15573 * config/s390/s390.c (s390_check_qrst_address): Check incoming
15574 address against address_operand predicate.
15575 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
15577 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
15579 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
15580 parameters passed indirectly.
15582 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
15584 * config/i386/i386.md (*movdi_internal): Remove SSE4
15585 alternative 18 (?r, *v). Update insn attributes.
15586 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
15587 Update insn attributes.
15588 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
15589 Update insn attributes.
15590 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
15591 alternative 1 (r, v). Remove isa attribute.
15592 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
15593 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
15594 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
15596 2017-05-23 Tom de Vries <tom@codesourcery.com>
15598 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
15601 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15603 * cgraphunit.c (symbol_table::process_new_functions): Update.
15604 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
15605 (inline_generate_summary): Rename to ...
15606 (ipa_fn_summary_generate): ... this one.
15607 (inline_read_summary): Rename to ...
15608 (ipa_fn_summary_read): ... this one.
15609 (inline_write_summary): Rename to ...
15610 (ipa_fn_summary_write): ... this one.
15611 (inline_free_summary): Rename to ...
15612 (ipa_free_fn_summary): ... this one.
15613 (pass_data_local_fn_summary, pass_local_fn_summary,
15614 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
15615 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
15616 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
15617 make_pass_ipa_fn_summary): New.
15618 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
15619 inline_write_summary, inline_free_summary): Remove.
15620 (ipa_free_fn_summary) : New.
15621 * ipa-inline.c (ipa_inline): Update.
15622 (pass_ipa_inline): Do not generate summaries.
15623 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
15625 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
15626 and add pass_ipa_fn_summary.
15627 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
15629 (make_pass_inline_parameters): Remove.
15631 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
15633 * omp-low.c (struct omp_context): Remove "default_kind" member.
15636 * omp-offload.c (execute_oacc_device_lower): Remove the
15637 parallelism dimensions function attributes for unparallelized
15638 OpenACC kernels constructs.
15640 2017-05-23 Martin Liska <mliska@suse.cz>
15642 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
15644 (cgraph_edge::make_speculative): Likewise.
15645 (cgraph_edge::resolve_speculation): Likewise.
15646 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
15647 (cgraph_node::dump): Likewise.
15648 * cgraph.h: Likewise.
15649 * cgraphunit.c (analyze_functions): Likewise.
15650 (symbol_table::compile): Likewise.
15651 * ipa-cp.c (print_all_lattices): Likewise.
15652 (determine_versionability): Likewise.
15653 (initialize_node_lattices): Likewise.
15654 (ipcp_verify_propagated_values): Likewise.
15655 (estimate_local_effects): Likewise.
15656 (update_profiling_info): Likewise.
15657 (create_specialized_node): Likewise.
15658 (perhaps_add_new_callers): Likewise.
15659 (decide_about_value): Likewise.
15660 (decide_whether_version_node): Likewise.
15661 (identify_dead_nodes): Likewise.
15662 (ipcp_store_bits_results): Likewise.
15663 * ipa-devirt.c (dump_targets): Likewise.
15664 (ipa_devirt): Likewise.
15665 * ipa-icf.c (sem_item::dump): Likewise.
15666 (sem_function::equals): Likewise.
15667 (sem_variable::equals): Likewise.
15668 (sem_item_optimizer::read_section): Likewise.
15669 (sem_item_optimizer::execute): Likewise.
15670 (congruence_class::dump): Likewise.
15671 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
15672 (dump_inline_summary): Likewise.
15673 (estimate_node_size_and_time): Likewise.
15674 (inline_analyze_function): Likewise.
15675 * ipa-inline-transform.c (inline_call): Likewise.
15676 * ipa-inline.c (report_inline_failed_reason): Likewise.
15677 (want_early_inline_function_p): Likewise.
15678 (edge_badness): Likewise.
15679 (update_edge_key): Likewise.
15680 (inline_small_functions): Likewise.
15681 * ipa-profile.c (ipa_profile): Likewise.
15682 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
15683 (ipa_make_edge_direct_to_target): Likewise.
15684 (remove_described_reference): Likewise.
15685 (ipa_impossible_devirt_target): Likewise.
15686 (propagate_controlled_uses): Likewise.
15687 (ipa_print_node_params): Likewise.
15688 (ipcp_transform_function): Likewise.
15689 * ipa-pure-const.c (pure_const_read_summary): Likewise.
15690 (propagate_pure_const): Likewise.
15691 * ipa-reference.c (generate_summary): Likewise.
15692 (read_write_all_from_decl): Likewise.
15693 (propagate): Likewise.
15694 (ipa_reference_read_optimization_summary): Likewise.
15695 * ipa-utils.c (ipa_merge_profiles): Likewise.
15696 * ipa.c (walk_polymorphic_call_targets): Likewise.
15697 (symbol_table::remove_unreachable_nodes): Likewise.
15698 (ipa_single_use): Likewise.
15699 * passes.c (execute_todo): Likewise.
15700 * predict.c (drop_profile): Likewise.
15701 * symtab.c (symtab_node::get_dump_name): New function.
15702 (symtab_node::dump_name): Likewise.
15703 (symtab_node::dump_asm_name): Likewise.
15704 (symtab_node::dump_references): Likewise.
15705 (symtab_node::dump_referring): Likewise.
15706 (symtab_node::dump_base): Likewise.
15707 (symtab_node::debug_symtab): Likewise.
15708 * tree-sra.c (convert_callers_for_node): Likewise.
15709 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
15710 * value-prof.c (init_node_map): Likewise.
15712 2017-05-23 Martin Liska <mliska@suse.cz>
15714 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
15715 and symtab_node::debug_symtab to symbol_table::debug.
15716 * cgraphunit.c (analyze_functions): Use the renamed function.
15717 (symbol_table::compile): Likewise.
15718 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
15719 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
15720 * passes.c (execute_todo): Likewise.
15721 * symtab.c (symbol_table::dump): New function.
15722 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
15724 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15726 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
15727 that nonconst implies exec.
15729 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15731 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
15732 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
15733 (inline_edge_summary_vec): Turn into ...
15734 (ipa_call_summaries): ... this one.
15735 (redirect_to_unreachable, edge_set_predicate,
15736 evaluate_properties_for_edge, inline_summary_alloc,
15737 reset_ipa_call_summary, reset_inline_summary,
15738 inline_summary_t::duplicate): Update.
15739 (inline_edge_duplication_hook): Turn to ...
15740 (ipa_call_summary_t::duplicate): ... this one.
15741 (inline_edge_removal_hook): Turn to ...
15742 (ipa_call_summary_t::remove): ... this one.
15743 (dump_inline_edge_summary): Turn to ...
15744 (dump_ipa_call_summary): ... this one.
15745 (estimate_function_body_sizes): Update.
15746 (inline_update_callee_summaries): Update.
15747 (remap_edge_change_prob): Update.
15748 (remap_edge_summaries): Update.
15749 (inline_merge_summary): Update.
15750 (do_estimate_edge_time): Update.
15751 (inline_generate_summary): Update.
15752 (inline_read_section): Update.
15753 (inline_read_summary): Update.
15754 (inline_free_summary): Update.
15755 * ipa-inline.c (can_inline_edge_p): Update.
15756 (compute_inlined_call_time): Update.
15757 (want_inline_small_function_p): Update.
15758 (edge_badness): Update.
15759 (early_inliner): Update.
15760 * ipa-inline.h (inline_edge_summary): Turn to ...
15761 (ipa_call_summary): ... this one.
15762 (ipa_call_summary_t): New class.
15763 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
15764 (ipa_call_summaries): New.
15765 (inline_edge_summary): Remove.
15766 (estimate_edge_growth): Update.
15767 * ipa-profile.c (ipa_propagate_frequency_1): Update.
15768 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15769 * ipa-split.c (execute_split_functions): Update.
15770 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
15772 2017-05-23 Tom de Vries <tom@codesourcery.com>
15774 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
15775 attributes): Document rdrand effective target.
15777 2017-05-23 Tom de Vries <tom@codesourcery.com>
15779 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
15780 attributes): Sort alphabetically.
15782 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
15784 * config/avr/genmultilib.awk: Use gsub instead of gensub.
15786 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
15789 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
15790 V2DF/V2DI splat into two separate patterns, one that handles
15791 registers, and the other that only handles memory. Drop support
15792 for splatting from a GPR on ISA 2.07 and then splitting the
15793 splat into direct move and splat.
15794 (vsx_splat_<mode>_reg): Likewise.
15795 (vsx_splat_<mode>_mem): Likewise.
15797 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
15799 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
15801 2017-05-22 Jakub Jelinek <jakub@redhat.com>
15803 PR middle-end/80809
15804 * omp-low.c (finish_taskreg_remap): New function.
15805 (finish_taskreg_scan): If unit size of ctx->record_type
15806 is non-constant, unshare the size expression and replace
15807 decls in it with possible outer var refs.
15809 PR middle-end/80809
15810 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
15811 GOVD_SHARED rather than GOVD_PRIVATE with it.
15812 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
15813 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
15815 PR middle-end/80853
15816 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
15817 as last argument to build_outer_var_ref for pointer bases of array
15818 section reductions.
15820 2017-05-19 Martin Sebor <msebor@redhat.com>
15822 * print-tree.c (print_node): Print DECL_READ_P flag.
15824 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15826 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
15827 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
15828 * cgraph.c: Likewise.
15829 * cgraphunit.c: Likewise.
15830 * gengtype.c: Likewise.
15831 * ipa-cp.c: Likewise.
15832 * ipa-devirt.c: Likewise.
15833 * ipa-icf.c: Likewise.
15834 * ipa-predicate.c: Likewise.
15835 * ipa-profile.c: Likewise.
15836 * ipa-prop.c: Likewise.
15837 * ipa-split.c: Likewise.
15839 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
15840 edge_predicate_pool, dump_inline_hints,
15841 inline_summary::account_size_time, redirect_to_unreachable,
15842 edge_set_predicate, set_hint_predicate,
15843 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
15844 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
15845 inline_summary_t::remove, remap_hint_predicate_after_duplication,
15846 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
15847 ipa_call_summary_t::remove, initialize_growth_caches,
15848 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
15849 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
15850 mark_modified, unmodified_parm_1, unmodified_parm,
15851 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
15852 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
15853 compute_bb_predicates, will_be_nonconstant_expr_predicate,
15854 will_be_nonconstant_predicate, record_modified_bb_info,
15855 get_minimal_bb, record_modified, param_change_prob,
15856 phi_result_unknown_predicate, predicate_for_phi_result,
15857 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
15858 estimate_function_body_sizes, compute_inline_parameters,
15859 compute_inline_parameters_for_curren, pass_data_inline_parameters,
15860 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
15861 inline_update_callee_summaries, remap_edge_change_prob,
15862 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
15863 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
15864 inline_analyze_function, inline_summary_t::insert,
15865 inline_generate_summary, read_ipa_call_summary, inline_read_section,
15866 inline_read_summary, write_ipa_call_summary, inline_write_summary,
15867 inline_free_summary): Move to ipa-fnsummary.h
15868 (predicate_t): Remove.
15869 * ipa-fnsummary.c: New file.
15870 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
15871 (enum inline_hints_vals, inline_hints, agg_position_info,
15872 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
15873 inline_summaries, ipa_call_summary, ipa_call_summary_t,
15874 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
15875 dump_inline_summary, dump_inline_hints, inline_generate_summary,
15876 inline_read_summary, inline_write_summary, inline_free_summary,
15877 inline_analyze_function, initialize_inline_failed,
15878 inline_merge_summary, inline_update_overall_summary,
15879 compute_inline_parameters): Move to ipa-fnsummary.h
15880 * ipa-fnsummary.h: New file.
15881 * ipa-inline-transform.h: Include ipa-inline.h.
15882 * ipa-inline.c: LIkewise.
15884 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15886 * ipa-inline.c (edge_badness): Use inlined_time instead of
15887 inline_summaries->get.
15889 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15891 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
15893 2017-05-22 Nathan Sidwell <nathan@acm.org>
15895 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
15896 (fdump-lang): Document 'raw' option.
15897 * dumpfile.h (TDI_tu): Delete.
15898 * dumpfile.c (dump_files): Remove translation-unit.
15899 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
15901 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
15903 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
15904 command option from $(AWK) call.
15905 * config/avr/genmultilib.awk: Simplify and rewrite so that it
15906 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
15907 [FORMAT]: Remove handling of variable.
15908 * config/avr/t-multilib: Regenerate.
15910 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15912 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
15914 (dump_inline_summary): Do not print self_time.
15915 (estimate_function_body_sizes): Do not set self_time.
15916 (compute_inline_parameters): Likewise.
15917 (inline_read_section, inline_write_summary): Do not stream self_time.
15918 * ipa-inline.h (inline_summary): Drop self_time.
15920 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15922 * ipa-inline-analysis.c (account_size_time): Rename to ...
15923 (inline_summary::account_size_time): ... this one.
15924 (reset_ipa_call_summary): Turn to ...
15925 (ipa_call_summary::reset): ... this one.
15926 (reset_inline_summary): Turn to ...
15927 (inline_summary::reset): ... this one.
15928 (inline_summary_t::remove): Update.
15929 (inline_summary_t::duplicate): Update.
15930 (ipa_call_summary_t::remove): Update.
15931 (dump_inline_summary): Update.
15932 (estimate_function_body_sizes): Update.
15933 (compute_inline_parameters): Update.
15934 (estimate_node_size_and_time): Update.
15935 (inline_merge_summary): Update.
15936 (inline_update_overall_summary): Update.
15937 (inline_read_section): Update.
15938 (inline_write_summary): Update.
15939 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
15940 add account_size_time and reset member functions.
15941 (ipa_call_summary): Add reset function.
15942 * ipa-predicate.h (predicate::operator &): Constify.
15944 2017-05-22 Richard Biener <rguenther@suse.de>
15946 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
15948 2017-05-19 Jason Merrill <jason@redhat.com>
15950 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
15952 2017-05-19 Marek Polacek <polacek@redhat.com>
15955 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
15956 TYPE_OVERFLOW_WRAPS checks.
15958 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
15960 * tree-core.h (enum omp_clause_default_kind): Add
15961 "OMP_CLAUSE_DEFAULT_PRESENT".
15962 * tree-pretty-print.c (dump_omp_clause): Handle it.
15963 * gimplify.c (enum gimplify_omp_var_data): Add
15964 "GOVD_MAP_FORCE_PRESENT".
15965 (gimplify_adjust_omp_clauses_1): Map it to
15966 "GOMP_MAP_FORCE_PRESENT".
15967 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
15969 * gimplify.c (oacc_default_clause): Clarify.
15971 2017-05-19 Nathan Sidwell <nathan@acm.org>
15973 LANG_HOOK_REGISTER_DUMPS
15974 * toplev.c (general_init): Call register dump lang hook.
15975 * doc/invoke.texi: Document -fdump-lang option family.
15976 * dumpfile.c (dump_files): Remove class dump here.
15977 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
15978 * dumpfile.h (tree_dump_index): Remove TDI_class.
15979 * langhooks-def.h (lhd_register_dumps): Declare.
15980 (LANG_HOOKS_REGISTER_DUMPS): Define.
15981 (LANG_HOOKS_INITIALIZER): Add it.
15982 * langhooks.c (lhd_register_dumps): Define.
15983 * langhooks.h (struct lang_hooks): Add register_dumps.
15985 2017-05-19 Nathan Sidwell <nathan@acm.org>
15987 * context.h (context::set_passes): New.
15988 * context.c (context::context): Do not create pass manager.
15989 * toplev.c (general_init): Create pass manager here.
15991 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
15993 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
15994 use this splitter if two add or or instructions would also work for
15995 the constant we want to generate.
15997 2017-05-19 Richard Biener <rguenther@suse.de>
16000 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
16001 predicate evaluation.
16003 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
16005 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
16007 * ipa-inline.c (want_inline_small_function_p): Do not cast to
16010 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
16012 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
16013 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
16014 (inline_edge_summary_vec): Turn into ...
16015 (ipa_call_summaries): ... this one.
16016 (redirect_to_unreachable, edge_set_predicate,
16017 evaluate_properties_for_edge, inline_summary_alloc,
16018 reset_ipa_call_summary, reset_inline_summary,
16019 inline_summary_t::duplicate): Update.
16020 (inline_edge_duplication_hook): Turn to ...
16021 (ipa_call_summary_t::duplicate): ... this one.
16022 (inline_edge_removal_hook): Turn to ...
16023 (ipa_call_summary_t::remove): ... this one.
16024 (dump_inline_edge_summary): Turn to ...
16025 (dump_ipa_call_summary): ... this one.
16026 (estimate_function_body_sizes): Update.
16027 (inline_update_callee_summaries): Update.
16028 (remap_edge_change_prob): Update.
16029 (remap_edge_summaries): Update.
16030 (inline_merge_summary): Update.
16031 (do_estimate_edge_time): Update.
16032 (inline_generate_summary): Update.
16033 (inline_read_section): Update.
16034 (inline_read_summary): Update.
16035 (inline_free_summary): Update.
16036 * ipa-inline.c (can_inline_edge_p): Update.
16037 (compute_inlined_call_time): Update.
16038 (want_inline_small_function_p): Update.
16039 (edge_badness): Update.
16040 (early_inliner): Update.
16041 * ipa-inline.h (inline_edge_summary): Turn to ...
16042 (ipa_call_summary): ... this one.
16043 (ipa_call_summary_t): New class.
16044 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
16045 (ipa_call_summaries): New.
16046 (inline_edge_summary): Remove.
16047 (estimate_edge_growth): Update.
16048 * ipa-profile.c (ipa_propagate_frequency_1): Update.
16049 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
16050 * ipa-split.c (execute_split_functions): Update.
16051 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
16053 2017-05-19 Richard Biener <rguenther@suse.de>
16055 PR middle-end/80764
16056 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
16058 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
16060 * config/rs6000/rs6000.c (struct machine_function): Add field
16061 fpr_is_wrapped_separately.
16062 (rs6000_get_separate_components): Use 64 components. Handle the
16063 new FPR components.
16064 (rs6000_components_for_bb): Handle the FPR components.
16065 (rs6000_emit_prologue_components): Handle the FPR components.
16066 (rs6000_emit_epilogue_components): Handle the FPR components.
16067 (rs6000_set_handled_components): Handle the FPR components.
16068 (rs6000_emit_prologue): Don't output prologue code for those FPRs
16069 that are already separately shrink-wrapped.
16070 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
16071 that are already separately shrink-wrapped.
16073 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
16076 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
16079 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
16080 (define_peephole2 for Altivec d-form load): Add peepholes to catch
16081 cases where the register allocator uses a move and an offsettable
16082 memory operation to/from a FPR register on ISA 2.06/2.07.
16083 (define_peephole2 for Altivec d-form store): Likewise.
16085 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
16088 * config/i386/mmx.md (*mov<mode>_internal): Enable
16089 alternatives 11, 12, 13 and 14 also for 32bit targets.
16090 Remove alternatives 15, 16, 17 and 18.
16091 * config/i386/sse.md (vec_concatv2di): Change
16092 alternative (!x, *y) to (x, ?!*Yn).
16094 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
16096 * dumpfile.h (enum dump_kind): Remove stray comma.
16098 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
16100 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
16101 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
16102 predicate::num_conditions
16103 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
16104 (CHANGED): turn into predicate::changed.
16105 (agg_position_info): Move to ipa-predicate.h
16106 (add_condition, predicate::add_clause, predicate::operator &=,
16107 predicate::or_with, predicate::evaluate, predicate::probability,
16108 dump_condition, dump_clause, predicate::dump,
16109 predicate::remap_after_duplication, predicate::remap_after_inlining,
16110 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
16111 (evaluate_conditions_for_known_args): Update.
16112 (set_cond_stmt_execution_predicate): Update.
16113 * ipa-inline.h: Include ipa-predicate.h
16114 (condition, inline_param_summary, conditions, agg_position_info,
16115 predicate): Move to ipa-predicate.h
16116 * ipa-predicate.c: New file.
16117 * ipa-predicate.h: New file.
16119 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
16121 * final.c (leaf_function_p): Check we are not in a sequence.
16123 2017-05-18 Martin Liska <mliska@suse.cz>
16125 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
16126 * dumpfile.c (dump_register): Use new enum dump_kind.
16127 (get_dump_file_name): Likewise.
16128 (dump_enable_all): Likewise.
16129 (dump_switch_p_1): Likewise.
16130 (enable_rtl_dump_file): Remove usage of TDF_RTL.
16131 * dumpfile.h (enum dump_kind): New enum type.
16132 (struct dump_file_info): Create constructor and
16133 format fields and comments.
16134 * passes.c (pass_manager::register_one_dump_file):
16136 * statistics.c (statistics_early_init): Likewise.
16137 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
16138 TDF_TREE with TDF_SLIM.
16139 (gather_memory_references_ref): Likewise.
16141 2017-05-18 Martin Liska <mliska@suse.cz>
16143 * vec.h (struct vnull): Use it.
16145 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
16147 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
16148 (true_predicate, false_predicate, true_predicate_p,
16149 false_predicate_p): Remove.
16150 (single_cond_predicate, not_inlined_predicate): Turn to member function
16152 (add_condition): Update.
16153 (add_clause): Turn to...
16154 (predicate::add_clause): ... this one; update; allow passing NULL
16156 (and_predicates): Turn to ...
16157 (predicate::operator &=): ... this one.
16158 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
16159 (or_predicates): Turn to ...
16160 (predicate::or_with): ... this one.
16161 (evaluate_predicate): Turn to ...
16162 (predicate::evaluate): ... this one.
16163 (predicate_probability): Turn to ...
16164 (predicate::probability): ... this one.
16165 (dump_condition): Update.
16166 (dump_predicate): Turn to ...
16167 (predicate::dump): ... this one.
16168 (account_size_time): Update.
16169 (edge_set_predicate): Update.
16170 (set_hint_predicate): UPdate.
16171 (evaluate_conditions_for_known_args): Update.
16172 (evaluate_properties_for_edge): Update.
16173 (remap_predicate_after_duplication): Turn to...
16174 (predicate::remap_after_duplication): ... this one.
16175 (remap_hint_predicate_after_duplication): Update.
16176 (inline_summary_t::duplicate): UPdate.
16177 (dump_inline_edge_summary): Update.
16178 (dump_inline_summary): Update.
16179 (set_cond_stmt_execution_predicate): Update.
16180 (set_switch_stmt_execution_predicate): Update.
16181 (compute_bb_predicates): Update.
16182 (will_be_nonconstant_expr_predicate): Update.
16183 (will_be_nonconstant_predicate): Update.
16184 (phi_result_unknown_predicate): Update.
16185 (predicate_for_phi_result): Update.
16186 (array_index_predicate): Update.
16187 (estimate_function_body_sizes): Update.
16188 (estimate_node_size_and_time): Update.
16189 (estimate_ipcp_clone_size_and_time): Update.
16190 (remap_predicate): Rename to ...
16191 (predicate::remap_after_inlining): ... this one.
16192 (remap_hint_predicate): Update.
16193 (inline_merge_summary): Update.
16194 (inline_update_overall_summary): Update.
16195 (estimate_size_after_inlining): Update.
16196 (read_predicate): Rename to ...
16197 (predicate::stream_in): ... this one.
16198 (read_inline_edge_summary): Update.
16199 (write_predicate): Rename to ...
16200 (predicate::stream_out): ... this one.
16201 (write_inline_edge_summary): Update.
16202 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
16203 (clause_t): Turn to uint32_t
16204 (predicate): Turn to class; implement constructor and operators
16206 (size_time_entry): Update.
16207 (inline_summary): Update.
16208 (inline_edge_summary): Update.
16210 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
16212 * fold-const.c (fold_binary_loc): Move transformation...
16213 * match.pd (C - X CMP X): ... here.
16215 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
16217 * config/sparc/sparc.c (sparc_option_override): Set function
16218 alignment for -mcpu=niagara7 to 64 to match the I$ line.
16219 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
16221 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
16223 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
16225 2017-05-18 Marek Polacek <polacek@redhat.com>
16228 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
16229 (pass_ubsan::execute): Call gimple_assign_single_p instead of
16230 gimple_assign_load_p.
16232 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
16234 PR middle-end/80692
16235 * real.c (do_compare): Give decimal_do_compare preference over
16236 comparing just the signs.
16238 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
16240 * doc/md.texi (Canonicalization of Instructions): Describe the
16241 canonical form of instructions that inherently set a condition
16244 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
16246 PR middle-end/80775
16247 * tree-cfg.c: Move deletion of unreachable case statements to after
16248 the merging of consecutive case labels.
16250 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
16252 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
16253 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
16254 restoring of callee-saved registers.
16256 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
16258 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
16259 * config/visium/visium.c (single_set_and_flags): Likewise.
16260 * config/visium/visium.md (Substitutions): Likewise.
16262 2017-05-17 Martin Liska <mliska@suse.cz>
16264 * cfg.c: Introduce dump_flags_t type and
16265 use it instead of int type.
16267 * cfghooks.c: Likewise.
16268 * cfghooks.h (struct cfg_hooks): Likewise.
16269 * cfgrtl.c: Likewise.
16270 * cfgrtl.h: Likewise.
16271 * cgraph.c (cgraph_node::get_body): Likewise.
16272 * coretypes.h: Likewise.
16273 * domwalk.c: Likewise.
16274 * domwalk.h: Likewise.
16275 * dumpfile.c (struct dump_option_value_info): Likewise.
16276 (dump_enable_all): Likewise.
16277 (dump_switch_p_1): Likewise.
16278 (opt_info_switch_p): Likewise.
16279 * dumpfile.h (enum tree_dump_index): Likewise.
16280 (struct dump_file_info): Likewise.
16281 * genemit.c: Likewise.
16282 * generic-match-head.c: Likewise.
16283 * gengtype.c (open_base_files): Likewise.
16284 * gimple-pretty-print.c: Likewise.
16285 * gimple-pretty-print.h: Likewise.
16286 * graph.c (print_graph_cfg): Likewise.
16287 * graphite-scop-detection.c (dot_all_sese): Likewise.
16288 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
16289 * loop-unroll.c (report_unroll): Likewise.
16290 * passes.c (pass_manager::register_one_dump_file): Likewise.
16291 * print-tree.c: Likewise.
16292 * statistics.c: Likewise.
16293 * tree-cfg.c: Likewise.
16294 * tree-cfg.h: Likewise.
16295 * tree-dfa.c: Likewise.
16296 * tree-dfa.h: Likewise.
16297 * tree-dump.c (dump_function): Likewise.
16298 * tree-dump.h (struct dump_info): Likewise.
16299 * tree-pretty-print.c: Likewise.
16300 * tree-pretty-print.h: Likewise.
16301 * tree-ssa-live.c: Likewise.
16302 * tree-ssa-live.h: Likewise.
16303 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
16304 * tree-vect-loop.c: Likewise.
16305 * tree-vect-slp.c: Likewise.
16307 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
16308 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16310 PR tree-optimization/80457
16311 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
16312 of all arguments to a statement as scalar_to_vec operations.
16313 (vectorizable_call): Adjust call to vect_model_simple_cost for
16315 (vectorizable_conversion): Likewise.
16316 (vectorizable_assignment): Likewise.
16317 (vectorizable_shift): Likewise.
16318 (vectorizable_operation): Likewise.
16319 (vectorizable_comparison): Likewise.
16320 (vect_is_simple_cond): Record the def types for operands.
16321 (vectorizable_condition): Likewise, call vect_model_simple_cost.
16322 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
16323 for statement argument count.
16325 2017-05-16 Carl Love <cel@us.ibm.com>
16327 * config/rs6000/rs6000-c: Add support for built-in functions
16328 vector unsigned long long vec_bperm (vector unsigned long long,
16329 vector unsigned char)
16330 vector signed long long vec_mule (vector signed int,
16332 vector unsigned long long vec_mule (vector unsigned int,
16333 vector unsigned int)
16334 vector signed long long vec_mulo (vector signed int,
16336 vector unsigned long long vec_mulo (vector unsigned int,
16337 vector unsigned int)
16338 vector signed char vec_sldw (vector signed char,
16339 vector signed char,
16341 vector unsigned char vec_sldw (vector unsigned char,
16342 vector unsigned char,
16344 vector signed short vec_sldw (vector signed short,
16345 vector signed short,
16347 vector unsigned short vec_sldw (vector unsigned short,
16348 vector unsigned short,
16350 vector signed int vec_sldw (vector signed int,
16353 vector unsigned int vec_sldw (vector unsigned int,
16354 vector unsigned int,
16356 vector signed long long vec_sldw (vector signed long long,
16357 vector signed long long,
16359 vector unsigned long long vec_sldw (vector unsigned long long,
16360 vector unsigned long long,
16362 * config/rs6000/rs6000-c: Add support for built-in functions
16363 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
16364 * config/rs6000/altivec.h: Add defintion for vec_sldw.
16365 * doc/extend.texi: Update the built-in documentation for the
16366 new built-in functions.
16368 2017-05-16 Marek Polacek <polacek@redhat.com>
16372 * tree.c (save_expr): Don't fold the expression.
16374 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
16376 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
16377 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
16378 and (?*y,m). Update insn attributes.
16380 2017-05-16 Martin Liska <mliska@suse.cz>
16382 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
16383 flags argument of print_gimple_stmt, print_gimple_expr,
16384 print_generic_stmt and print_generic_expr.
16385 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
16386 * coretypes.h: Likewise.
16387 * except.c (dump_eh_tree): Likewise.
16388 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
16389 * gimple-pretty-print.h: Likewise.
16390 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
16391 (backprop::push_to_worklist): Likewise.
16392 (backprop::pop_from_worklist): Likewise.
16393 (backprop::process_use): Likewise.
16394 (backprop::intersect_uses): Likewise.
16395 (note_replacement): Likewise.
16396 * gimple-ssa-store-merging.c
16397 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
16398 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
16399 (pass_store_merging::execute): Likewise.
16400 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
16401 (ssa_base_cand_dump_callback): Likewise.
16402 (dump_incr_vec): Likewise.
16403 (replace_refs): Likewise.
16404 (replace_mult_candidate): Likewise.
16405 (create_add_on_incoming_edge): Likewise.
16406 (create_phi_basis): Likewise.
16407 (insert_initializers): Likewise.
16408 (all_phi_incrs_profitable): Likewise.
16409 (introduce_cast_before_cand): Likewise.
16410 (replace_one_candidate): Likewise.
16411 * gimplify.c (gimplify_expr): Likewise.
16412 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
16413 (set_rename): Likewise.
16414 (rename_uses): Likewise.
16415 (copy_loop_phi_nodes): Likewise.
16416 (add_close_phis_to_merge_points): Likewise.
16417 (copy_loop_close_phi_args): Likewise.
16418 (copy_cond_phi_args): Likewise.
16419 (graphite_copy_stmts_from_block): Likewise.
16420 (translate_pending_phi_nodes): Likewise.
16421 * graphite-poly.c (print_pdr): Likewise.
16422 (dump_gbb_cases): Likewise.
16423 (dump_gbb_conditions): Likewise.
16424 (print_scop_params): Likewise.
16425 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
16426 (build_cross_bb_scalars_use): Likewise.
16427 (gather_bbs::before_dom_children): Likewise.
16428 * hsa-dump.c (dump_hsa_immed): Likewise.
16429 * ipa-cp.c (print_ipcp_constant_value): Likewise.
16430 (get_replacement_map): Likewise.
16431 * ipa-inline-analysis.c (dump_condition): Likewise.
16432 (estimate_function_body_sizes): Likewise.
16433 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
16434 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
16435 * ipa-prop.c (ipa_dump_param): Likewise.
16436 (ipa_print_node_jump_functions_for_edge): Likewise.
16437 (ipa_modify_call_arguments): Likewise.
16438 (ipa_modify_expr): Likewise.
16439 (ipa_dump_param_adjustments): Likewise.
16440 (ipa_dump_agg_replacement_values): Likewise.
16441 (ipcp_modif_dom_walker::before_dom_children): Likewise.
16442 * ipa-pure-const.c (check_stmt): Likewise.
16443 (pass_nothrow::execute): Likewise.
16444 * ipa-split.c (execute_split_functions): Likewise.
16445 * omp-offload.c (dump_oacc_loop_part): Likewise.
16446 (dump_oacc_loop): Likewise.
16447 * trans-mem.c (tm_log_emit): Likewise.
16448 (tm_memopt_accumulate_memops): Likewise.
16449 (dump_tm_memopt_set): Likewise.
16450 (dump_tm_memopt_transform): Likewise.
16451 * tree-cfg.c (gimple_verify_flow_info): Likewise.
16452 (print_loop): Likewise.
16453 * tree-chkp-opt.c (chkp_print_addr): Likewise.
16454 (chkp_gather_checks_info): Likewise.
16455 (chkp_get_check_result): Likewise.
16456 (chkp_remove_check_if_pass): Likewise.
16457 (chkp_use_outer_bounds_if_possible): Likewise.
16458 (chkp_reduce_bounds_lifetime): Likewise.
16459 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
16460 (chkp_mark_completed_bounds): Likewise.
16461 (chkp_register_incomplete_bounds): Likewise.
16462 (chkp_mark_invalid_bounds): Likewise.
16463 (chkp_maybe_copy_and_register_bounds): Likewise.
16464 (chkp_build_returned_bound): Likewise.
16465 (chkp_get_bound_for_parm): Likewise.
16466 (chkp_build_bndldx): Likewise.
16467 (chkp_get_bounds_by_definition): Likewise.
16468 (chkp_generate_extern_var_bounds): Likewise.
16469 (chkp_get_bounds_for_decl_addr): Likewise.
16470 * tree-chrec.c (chrec_apply): Likewise.
16471 * tree-data-ref.c (dump_data_reference): Likewise.
16472 (dump_subscript): Likewise.
16473 (dump_data_dependence_relation): Likewise.
16474 (analyze_overlapping_iterations): Likewise.
16475 * tree-inline.c (expand_call_inline): Likewise.
16476 (tree_function_versioning): Likewise.
16477 * tree-into-ssa.c (dump_defs_stack): Likewise.
16478 (dump_currdefs): Likewise.
16479 (dump_names_replaced_by): Likewise.
16480 (dump_update_ssa): Likewise.
16481 (update_ssa): Likewise.
16482 * tree-object-size.c (pass_object_sizes::execute): Likewise.
16483 * tree-parloops.c (build_new_reduction): Likewise.
16484 (try_create_reduction_list): Likewise.
16485 (ref_conflicts_with_region): Likewise.
16486 (oacc_entry_exit_ok_1): Likewise.
16487 (oacc_entry_exit_single_gang): Likewise.
16488 * tree-pretty-print.h: Likewise.
16489 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
16490 (get_scalar_evolution): Likewise.
16491 (add_to_evolution): Likewise.
16492 (get_loop_exit_condition): Likewise.
16493 (analyze_evolution_in_loop): Likewise.
16494 (analyze_initial_condition): Likewise.
16495 (analyze_scalar_evolution): Likewise.
16496 (instantiate_scev): Likewise.
16497 (number_of_latch_executions): Likewise.
16498 (gather_chrec_stats): Likewise.
16499 (final_value_replacement_loop): Likewise.
16500 (scev_const_prop): Likewise.
16501 * tree-sra.c (dump_access): Likewise.
16502 (disqualify_candidate): Likewise.
16503 (create_access): Likewise.
16504 (reject): Likewise.
16505 (maybe_add_sra_candidate): Likewise.
16506 (create_access_replacement): Likewise.
16507 (analyze_access_subtree): Likewise.
16508 (analyze_all_variable_accesses): Likewise.
16509 (sra_modify_assign): Likewise.
16510 (initialize_constant_pool_replacements): Likewise.
16511 (find_param_candidates): Likewise.
16512 (decide_one_param_reduction): Likewise.
16513 (replace_removed_params_ssa_names): Likewise.
16514 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
16515 * tree-ssa-copy.c (dump_copy_of): Likewise.
16516 (copy_prop_visit_cond_stmt): Likewise.
16517 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
16518 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
16519 (record_equivalences_from_stmt): Likewise.
16520 * tree-ssa-dse.c (compute_trims): Likewise.
16521 (delete_dead_call): Likewise.
16522 (delete_dead_assignment): Likewise.
16523 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
16524 (forward_propagate_into_cond): Likewise.
16525 (pass_forwprop::execute): Likewise.
16526 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16527 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
16529 (move_computations_worker): Likewise.
16530 (execute_sm): Likewise.
16531 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
16532 (remove_exits_and_undefined_stmts): Likewise.
16533 (remove_redundant_iv_tests): Likewise.
16534 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
16535 (adjust_iv_update_pos): Likewise.
16536 * tree-ssa-math-opts.c (bswap_replace): Likewise.
16537 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
16538 (value_replacement): Likewise.
16539 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
16540 * tree-ssa-pre.c (print_pre_expr): Likewise.
16541 (get_representative_for): Likewise.
16542 (create_expression_by_pieces): Likewise.
16543 (insert_into_preds_of_block): Likewise.
16544 (eliminate_insert): Likewise.
16545 (eliminate_dom_walker::before_dom_children): Likewise.
16546 (eliminate): Likewise.
16547 (remove_dead_inserted_code): Likewise.
16548 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
16549 * tree-ssa-reassoc.c (get_rank): Likewise.
16550 (eliminate_duplicate_pair): Likewise.
16551 (eliminate_plus_minus_pair): Likewise.
16552 (eliminate_not_pairs): Likewise.
16553 (undistribute_ops_list): Likewise.
16554 (eliminate_redundant_comparison): Likewise.
16555 (update_range_test): Likewise.
16556 (optimize_range_tests_var_bound): Likewise.
16557 (optimize_vec_cond_expr): Likewise.
16558 (rewrite_expr_tree): Likewise.
16559 (rewrite_expr_tree_parallel): Likewise.
16560 (linearize_expr): Likewise.
16561 (break_up_subtract): Likewise.
16562 (linearize_expr_tree): Likewise.
16563 (attempt_builtin_powi): Likewise.
16564 (attempt_builtin_copysign): Likewise.
16565 (transform_stmt_to_copy): Likewise.
16566 (transform_stmt_to_multiply): Likewise.
16567 (dump_ops_vector): Likewise.
16568 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
16569 (print_scc): Likewise.
16570 (set_ssa_val_to): Likewise.
16571 (visit_reference_op_store): Likewise.
16572 (visit_use): Likewise.
16573 (sccvn_dom_walker::before_dom_children): Likewise.
16574 (run_scc_vn): Likewise.
16575 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
16577 (expr_hash_elt::print): Likewise.
16578 (const_and_copies::pop_to_marker): Likewise.
16579 (const_and_copies::record_const_or_copy_raw): Likewise.
16580 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
16581 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
16582 (dump_predicates): Likewise.
16583 (find_uninit_use): Likewise.
16584 (warn_uninitialized_phi): Likewise.
16585 (pass_late_warn_uninitialized::execute): Likewise.
16586 * tree-ssa.c (verify_vssa): Likewise.
16587 (verify_ssa): Likewise.
16588 (maybe_optimize_var): Likewise.
16589 * tree-vrp.c (dump_value_range): Likewise.
16590 (dump_all_value_ranges): Likewise.
16591 (dump_asserts_for): Likewise.
16592 (register_edge_assert_for_2): Likewise.
16593 (vrp_visit_cond_stmt): Likewise.
16594 (vrp_visit_switch_stmt): Likewise.
16595 (vrp_visit_stmt): Likewise.
16596 (vrp_visit_phi_node): Likewise.
16597 (simplify_cond_using_ranges_1): Likewise.
16598 (fold_predicate_in): Likewise.
16599 (evrp_dom_walker::before_dom_children): Likewise.
16600 (evrp_dom_walker::push_value_range): Likewise.
16601 (evrp_dom_walker::pop_value_range): Likewise.
16602 (execute_early_vrp): Likewise.
16604 2017-05-16 Richard Biener <rguenther@suse.de>
16606 * dwarf2out.c (loc_list_from_tree_1): Do not create
16607 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
16609 2017-05-16 Richard Biener <rguenther@suse.de>
16611 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
16613 (note_variable_value_in_expr): If we resolved the decl ref
16614 do not push to the stack.
16616 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
16618 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
16619 operations in fast-math mode.
16620 (vaddq_f16): Likewise.
16621 (vmul_f16): Likewise.
16622 (vmulq_f16): Likewise.
16623 (vsub_f16): Likewise.
16624 (vsubq_f16): Likewise.
16625 * config/arm/neon.md (add<mode>3): New.
16627 (fma:<VH:mode>3): New. Also remove outdated comment.
16630 2017-05-16 Martin Liska <mliska@suse.cz>
16634 * ipa-devirt.c (warn_types_mismatch): Fix typo.
16635 (odr_types_equivalent_p): Likewise.
16637 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
16639 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
16641 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
16644 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
16645 non-interunit SSE move alternatives with '?'.
16646 (zero-extendsidi peephole2): New peephole to skip intermediate
16647 general register in SSE zero-extend sequence.
16649 2017-05-15 Jeff Law <law@redhat.com>
16651 * reorg.c (relax_delay_slots): Create a new variable to hold
16652 the temporary target rather than clobbering TARGET_LABEL.
16654 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
16655 missing argument to extract_bit_field call.
16656 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
16658 2017-05-15 Martin Liska <mliska@suse.cz>
16661 * gcc.c (process_command): Do not allow empty argument of -o option.
16663 2017-05-15 Renlin Li <renlin.li@arm.com>
16665 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
16666 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
16667 * config/aarch64/constraints.md (Usf): Add long call check.
16668 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
16669 (call_value): Likewise.
16670 (sibcall): Likewise.
16671 (sibcall_value): Likewise.
16673 (call_value_insn): New.
16674 (sibcall_insn): Update rtx pattern.
16675 (sibcall_value_insn): Likewise.
16676 (call_internal): Remove.
16677 (call_value_internal): Likewise.
16678 (sibcall_internal): Likewise.
16679 (sibcall_value_internal): Likewise.
16680 (call_reg): Likewise.
16681 (call_symbol): Likewise.
16682 (call_value_reg): Likewise.
16683 (call_value_symbol): Likewise.
16685 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
16688 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
16690 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
16692 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
16693 compatible with CCGOCmode and with CCZmode.
16695 2017-05-14 Martin Sebor <msebor@redhat.com>
16697 PR middle-end/77671
16698 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
16699 (gimple_fold_builtin_snprintf): Same.
16700 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
16701 (gimple_fold_builtin_snprintf): Same.
16702 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
16703 of character types.
16704 (is_call_safe): New function.
16705 (try_substitute_return_value): Call it.
16706 (try_simplify_call): New function.
16707 (pass_sprintf_length::handle_gimple_call): Call it.
16709 2017-05-14 Martin Sebor <msebor@redhat.com>
16711 PR middle-end/80669
16712 * builtins.c (expand_builtin_stpncpy): Simplify.
16714 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
16716 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
16717 * config/i386/i386.h
16718 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
16719 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
16720 (struct machine_function): Add new members call_ms2sysv,
16721 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
16722 (struct machine_frame_state): New fields sp_realigned and
16723 sp_realigned_offset.
16724 * config/i386/i386.c
16725 (enum xlogue_stub): New enum.
16726 (enum xlogue_stub_sets): New enum.
16727 (class xlogue_layout): New class.
16728 (struct ix86_frame): New fields stack_realign_allocate_offset,
16729 stack_realign_offset and outlined_save_offset. Modify comments to
16730 detail stack layout when using out-of-line stubs.
16731 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
16732 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
16733 -mcall-ms2sysv-xlogues.
16734 (stub_managed_regs): New static variable.
16735 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
16736 registers managed by out-of-line stub.
16737 (disable_call_ms2sysv_xlogues): New function.
16738 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
16739 m->call_ms2sysv when appropriate and compute frame layout for
16741 (sp_valid_at, fp_valid_at): New inline functions.
16742 (choose_basereg): New function.
16743 (choose_baseaddr): Add align parameter, use choose_basereg and modify
16745 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
16746 Use align parameter of choose_baseaddr to generated aligned SSE movs
16748 (pro_epilogue_adjust_stack): Modify to track
16749 machine_frame_state::sp_realigned.
16750 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
16751 (ix86_nsaved_sseregs): Likewise.
16752 (ix86_emit_save_regs): Likewise.
16753 (ix86_emit_save_regs_using_mov): Likewise.
16754 (ix86_emit_save_sse_regs_using_mov): Likewise.
16755 (get_scratch_register_on_entry): Likewise.
16756 (gen_frame_set): New function.
16757 (gen_frame_load): Likewise.
16758 (gen_frame_store): Likewise.
16759 (emit_outlined_ms2sysv_save): Likewise.
16760 (emit_outlined_ms2sysv_restore): Likewise.
16761 (ix86_expand_prologue): Modify stack re-alignment code and call
16762 emit_outlined_ms2sysv_save when appropriate.
16763 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
16764 parameter rtx_insn *insn, which allows the function to be used to only
16765 generate the notes.
16766 (ix86_expand_epilogue): Modify validity checks of frame and stack
16767 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
16768 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
16769 * config/i386/predicates.md
16770 (save_multiple): New predicate.
16771 (restore_multiple): Likewise.
16772 * config/i386/sse.md
16773 (save_multiple<mode>): New pattern.
16774 (save_multiple_realign<mode>): Likewise.
16775 (restore_multiple<mode>): Likewise.
16776 (restore_multiple_and_return<mode>): Likewise.
16777 (restore_multiple_leave_return<mode>): Likewise.
16778 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
16780 2017-05-14 Julia Koval <julia.koval@intel.com>
16782 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
16783 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
16784 (__builtin_ia32_xsetbv): New builtins.
16785 * config/i386/i386.c (ix86_expand_special_args_builtin):
16787 (ix86_expand_builtin): Special expand for new intrinsics.
16788 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
16789 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
16790 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
16792 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16794 * cfganal.c (inverted_post_order_compute): Change argument type
16796 * cfganal.h (inverted_post_order_compute): Adjust prototype.
16797 * df-core.c (rest_of_handle_df_initialize): Adjust.
16798 (rest_of_handle_df_finish): Likewise.
16799 (df_analyze_1): Likewise.
16800 (df_analyze): Likewise.
16801 (loop_inverted_post_order_compute): Change argument to be a vec *.
16802 (df_analyze_loop): Adjust.
16803 (df_get_n_blocks): Likewise.
16804 (df_get_postorder): Likewise.
16805 * df.h (struct df_d): Change field to be a vec.
16806 * lcm.c (compute_laterin): Adjust.
16807 (compute_available): Likewise.
16808 * lra-lives.c (lra_create_live_ranges_1): Likewise.
16809 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
16810 * tree-ssa-pre.c (compute_antic): Likewise.
16812 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16814 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
16815 (depth_first_search::depth_first_search): Change structure init
16816 function to this constructor.
16817 (depth_first_search::add_bb): Rename function to this member.
16818 (depth_first_search::execute): Likewise.
16819 (flow_dfs_compute_reverse_finish): Adjust.
16821 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16823 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
16824 (longest_simple_path): Likewise.
16825 * shrink-wrap.c (spread_components): Likewise.
16826 (disqualify_problematic_components): Likewise.
16827 (emit_common_heads_for_components): Likewise.
16828 (emit_common_tails_for_components): Likewise.
16829 (insert_prologue_epilogue_for_components): Likewise.
16831 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16833 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
16836 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16838 * df-core.c (df_set_blocks): Start using auto_bitmap.
16839 (df_compact_blocks): Likewise.
16840 * df-problems.c (df_rd_confluence_n): Likewise.
16841 * df-scan.c (df_insn_rescan_all): Likewise.
16842 (df_process_deferred_rescans): Likewise.
16843 (df_update_entry_block_defs): Likewise.
16844 (df_update_exit_block_uses): Likewise.
16845 (df_entry_block_bitmap_verify): Likewise.
16846 (df_exit_block_bitmap_verify): Likewise.
16847 (df_scan_verify): Likewise.
16848 * lra-constraints.c (lra_constraints): Likewise.
16849 (undo_optional_reloads): Likewise.
16850 (lra_undo_inheritance): Likewise.
16851 * lra-remat.c (calculate_gen_cands): Likewise.
16852 (do_remat): Likewise.
16853 * lra-spills.c (assign_spill_hard_regs): Likewise.
16854 (spill_pseudos): Likewise.
16855 * tree-ssa-pre.c (bitmap_set_and): Likewise.
16856 (bitmap_set_subtract_values): Likewise.
16858 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16860 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
16861 management with auto_bitmap.
16862 (fix_inter_tick): Likewise.
16863 (fix_recovery_deps): Likewise.
16864 * ira.c (add_store_equivs): Likewise.
16865 (find_moveable_pseudos): Likewise.
16866 (split_live_ranges_for_shrink_wrap): Likewise.
16867 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
16868 (rtx_reuse_manager::seen_def_p): Likewise.
16869 (rtx_reuse_manager::set_seen_def): Likewise.
16870 * print-rtl.h (class rtx_reuse_manager): Likewise.
16872 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16874 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
16876 (migrate_btr_def): Likewise.
16877 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
16878 * df-core.c (loop_post_order_compute): Likewise.
16879 (loop_inverted_post_order_compute): Likewise.
16880 * hsa-common.h: Likewise.
16881 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
16882 * init-regs.c (initialize_uninitialized_regs): Likewise.
16883 * ipa-inline.c (resolve_noninline_speculation): Likewise.
16884 (inline_small_functions): Likewise.
16885 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
16886 * ira.c (combine_and_move_insns): Likewise.
16887 (build_insn_chain): Likewise.
16888 * loop-invariant.c (find_invariants): Likewise.
16889 * lower-subreg.c (propagate_pseudo_copies): Likewise.
16890 * predict.c (tree_predict_by_opcode): Likewise.
16891 (predict_paths_leading_to): Likewise.
16892 (predict_paths_leading_to_edge): Likewise.
16893 (estimate_loops_at_level): Likewise.
16894 (estimate_loops): Likewise.
16895 * shrink-wrap.c (try_shrink_wrapping): Likewise.
16896 (spread_components): Likewise.
16897 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
16898 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
16899 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
16900 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
16901 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
16902 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
16903 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
16904 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
16905 (mark_threaded_blocks): Likewise.
16906 (thread_through_all_blocks): Likewise.
16907 * tree-ssa.c (verify_ssa): Likewise.
16908 (execute_update_addresses_taken): Likewise.
16909 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
16911 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16913 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
16915 (post_order_compute): Likewise.
16916 (inverted_post_order_compute): Likewise.
16917 (pre_and_rev_post_order_compute_fn): Likewise.
16919 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16921 * genrecog.c (int_set::int_set): Explicitly construct our
16922 auto_vec base class.
16923 * vec.h (auto_vec::auto_vec): New constructor.
16925 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16927 * bitmap.h (class auto_bitmap): New constructor taking
16928 bitmap_obstack * argument.
16930 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16932 * bitmap.h (class auto_bitmap): Change type of m_bits to
16933 bitmap_head, and adjust ctor / dtor and member operators.
16935 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
16937 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
16938 when returned register mode doesn't match original mode.
16940 2017-05-12 Jeff Law <law@redhat.com>
16941 Jakub Jelinek <jakub@redhat.com>
16943 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
16944 we look for cc setter after the compare-elim changes.
16945 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
16946 within the vector to match what compare-elim now expects.
16947 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
16948 (xorsi3_flags, one_cmplsi2_flags): Likewise.
16950 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
16951 after the compare-elim changes.
16952 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
16953 the vector to match what compare-elim now expects.
16954 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
16955 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
16956 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
16957 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
16958 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
16960 * config/visium/visium.c (single_set_and_flags): Fix where
16961 we look for cc setter after the compare-elim changes.
16962 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
16963 with the vector to match what compare-elim now expects.
16964 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
16965 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
16966 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
16967 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
16968 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
16969 (neg<mode>2_insn_set_overflow): Likewise.
16971 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
16973 PR middle-end/79794
16974 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
16975 maybe_expand_insn call, set ops[0].target. If still set after call,
16976 set alt_rtl. Add extra arg to recursive calls.
16977 (extract_bit_field): Add alt_rtl argument. Pass to
16979 * expmed.h (extract_bit_field): Fix prototype.
16980 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
16981 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
16982 to extract_bit_field_calls.
16983 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
16984 Pass alt_rtl to extract_bit_field calls.
16985 * calls.c (store_unaligned_arguments_into_psuedos)
16986 load_register_parameters): Pass extra NULL to extract_bit_field calls.
16987 * optabs.c (maybe_legitimize_operand): Clear op->target when call
16989 * optabs.h (struct expand_operand): Add target bitfield.
16991 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
16993 * compare-elim.c (try_eliminate_compare): Canonicalize
16994 operation with embedded compare to
16995 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
16996 (set (reg) (operation)].
16998 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
17000 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
17003 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
17004 cost of adding a carry flag for ADC instruction.
17005 [case MINUS]: Ignore the cost of subtracting a carry flag
17006 for SBB instruction.
17008 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
17010 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
17012 * config/rs6000/bmiintrin.h: New file.
17013 * config/rs6000/bmi2intrin.h: New file.
17014 * config/rs6000/x86intrin.h: New file.
17016 2017-05-12 Jeff Law <law@redhat.com>
17018 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
17021 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
17023 PR middle-end/80707
17024 * tree-cfg.c: Remove cfg edges of unreachable case statements.
17026 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17028 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
17029 early expansion of vector divide builtins.
17030 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
17031 builtins identified as having unsigned arguments.
17033 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17035 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
17036 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
17037 expansion of vector logical operations (and, andc, or, xor,
17040 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17042 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
17043 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
17045 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17047 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
17048 early GIMPLE expansion of vector multiplies.
17050 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17052 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
17053 TARGET_HAVE_MOVT conditional.
17054 (movt splitter): Likewise.
17056 2017-05-12 Richard Biener <rguenther@suse.de>
17058 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
17059 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17060 Fold all stmts not inplace.
17062 2017-05-12 Richard Biener <rguenther@suse.de>
17064 PR tree-optimization/80713
17065 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
17066 inserted_exprs bit for not removed stmts.
17068 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
17070 PR middle-end/69921
17071 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
17072 parallelized" attribute for parallelized OpenACC kernels.
17073 * omp-offload.c (execute_oacc_device_lower): Use it.
17075 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
17076 Set "oacc kernels" attribute.
17077 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
17078 parameter. Adjust all users.
17079 (oacc_fn_attrib_kernels_p): Remove function.
17080 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
17081 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
17082 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
17083 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
17084 assert "oacc kernels" attribute is set.
17086 2017-05-11 Carl Love <cel@us.ibm.com>
17088 * config/rs6000/rs6000-c: Add support for built-in functions
17089 vector unsigned char vec_popcnt (vector signed char)
17090 vector unsigned char vec_popcnt (vector unsigned char)
17091 vector unsigned short vec_popcnt (vector signed short)
17092 vector unsigned short vec_popcnt (vector unsigned short)
17093 vector unsigned int vec_popcnt (vector signed int)
17094 vector unsigned int vec_popcnt (vector unsigned int)
17095 vector unsigned long long vec_popcnt (vector signed long long)
17096 vector unsigned long long vec_popcnt (vector unsigned long long)
17097 vector signed long long vec_slo (vector signed long long,
17098 vector signed char)
17099 vector signed long long vec_slo (vector signed long long,
17100 vector unsigned char)
17101 vector unsigned long long vec_slo (vector unsigned long long,
17102 vector signed char)
17103 vector unsigned long long vec_slo (vector unsigned long long,
17104 vector unsigned char)
17105 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
17106 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
17107 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
17108 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
17109 * doc/extend.texi: Update the built-in documentation file for the
17110 new built-in functions.
17112 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
17114 * attribs.h (sorted_attr_string): Move machine independent
17115 functions for target clone support from the i386 port to common
17116 code. Rename ix86_function_versions to common_function_versions.
17117 Rename make_name to make_unique_name.
17118 (common_function_versions): Likewise.
17119 (make_unique_name): Likewise.
17120 (make_dispatcher_decl): Likewise.
17121 (is_function_default_version): Likewise.
17122 * attribs.c (attr_strcmp): Likewise.
17123 (sorted_attr_string): Likewise.
17124 (common_function_versions): Likewise.
17125 (make_unique_name): Likewise.
17126 (make_dispatcher_decl): Likewise.
17127 (is_function_default_version): Likewise.
17128 * config/i386/i386.c (attr_strcmp): Likewise.
17129 (sorted_attr_string): Likewise.
17130 (ix86_function_versions): Likewise.
17131 (make_name): Likewise.
17132 (make_dispatcher_decl): Likewise.
17133 (is_function_default_version): Likewise.
17134 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
17136 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17139 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
17140 Account for direct move costs for vec_construct of integer
17143 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
17146 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
17147 (UNSPEC_STX_ATOMIC): Ditto.
17148 (loaddi_via_sse): New insn.
17149 (storedi_via_sse): Ditto.
17150 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
17151 Update corresponding peephole2 patterns.
17152 (atomic_storedi_fpu): Ditto.
17154 2017-05-11 Julia Koval <julia.koval@intel.com>
17156 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
17157 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
17159 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
17160 (__builtin_ia32_rsqrt14ss_mask): New builtins.
17161 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
17163 2017-05-11 Nathan Sidwell <nathan@acm.org>
17165 * graphite-poly.c: Include dumpfile.h.
17167 * dumpfle.h (dump_function): Declare here ...
17168 * tree-dump.h (dump_function): ... not here.
17169 * dumpfile.c: #include tree-cfg.h.
17170 (dump_function): Move here from ...
17171 * tree-dump.c (dump_function): ... here.
17172 * gimplify.c: #include splay-tree.h, not tree-dump.h.
17173 * graphite-poly.c: Don't include tree-dump.h.
17174 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
17175 * print-tree.c: Likewise.
17176 * stor-layout.c: Likewise.
17177 * tree-nested.c: Likewise.
17179 * dumpfile.c (dump_start): Use TDF_FLAGS.
17180 (dump_enable_all): Fix TDF_KIND check thinko.
17182 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
17184 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17185 array entries to represent two legal parameterizations of the
17186 overloaded __builtin_cmpb function, as represented by the
17187 P6_OV_BUILTIN_CMPB constant.
17188 (altivec_resolve_overloaded_builtin): Add special case handling
17189 for the __builtin_cmpb function, as represented by the
17190 P6_OV_BUILTIN_CMPB constant.
17191 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
17192 (BU_P6_64BIT_2): New macro.
17193 (BU_P6_OVERLOAD_2): New macro
17194 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
17195 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
17196 (CMPB): Add overload support to represent both 32-bit and 64-bit
17197 compare-bytes function.
17198 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
17199 support for TARGET_CMPB.
17200 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
17201 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17202 documentation of the __builtin_cmpb overloaded built-in function.
17204 2017-05-11 Richard Biener <rguenther@suse.de>
17206 PR tree-optimization/80705
17207 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
17208 bases are not vectorizable.
17210 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17212 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
17213 when counting register pressure.
17215 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17217 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
17218 (iv_ca_more_deps): Renamed to ...
17219 (iv_ca_compare_deps): ... this.
17220 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
17222 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17224 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
17226 (determine_group_iv_costs): ... here.
17227 (find_inv_vars_cb): Record inv var if it's not recorded before.
17229 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17231 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
17232 (get_shiftadd_cost): Ditto.
17234 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17236 * tree-ssa-address.c: Include header file.
17237 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
17239 (add_to_parts): Refactor.
17240 (addr_to_parts): New parameter. Update use of move_hint_to_base.
17241 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
17244 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17246 PR tree-optimization/53090
17247 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
17249 (extract_cond_operands): Detect condition with IV on both sides
17250 and return COMP_IV_EXPR_2.
17251 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
17252 (rewrite_use_compare): Simplify by removing call to function
17253 extract_cond_operands.
17255 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17257 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
17258 (extract_cond_operands): Detect condition comparing against non-
17259 invariant bound and return appropriate enum value.
17260 (find_interesting_uses_cond): Update use of extract_cond_operands.
17261 Handle its return value accordingly.
17262 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
17264 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17266 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
17267 nonlinear iv_use computation in loop invariant sensitive way.
17269 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17271 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
17272 (find_iv_candidates): Call relate_compare_use_with_all_cands.
17274 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17276 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
17277 (dump_cand): Support iv_cand.inv_exprs.
17278 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
17280 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
17283 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17285 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
17287 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
17288 as local function. Include necessary header files.
17289 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
17291 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17293 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
17295 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17297 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
17298 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
17299 RSHIFT_EXPR and BIT_NOT_EXPR.
17301 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17303 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
17304 (adjust_setup_cost): New parameter supporting round up adjustment.
17305 (struct address_cost_data): Delete.
17306 (force_expr_to_var_cost): Don't bound cost with spill_cost.
17307 (split_address_cost, ptr_difference_cost): Delete.
17308 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
17309 (struct ainc_cost_data): New struct.
17310 (get_address_cost_ainc): New function.
17311 (get_address_cost, get_computation_cost): Reimplement.
17312 (determine_group_iv_cost_address): Record inv_expr for all uses of
17314 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
17315 (iv_ca_has_deps): Reimplemented to ...
17316 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
17318 (iv_ca_extend): Call iv_ca_more_deps.
17320 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17322 * tree-ssa-address.c (struct mem_address): Move to header file.
17323 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
17324 * tree-ssa-address.h (struct mem_address): Move from C file.
17325 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
17327 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17329 * tree-affine.h (aff_combination_type): New interface.
17330 (aff_combination_zero_p): Remove static.
17331 (aff_combination_const_p): New interface.
17332 (aff_combination_singleton_var_p): New interfaces.
17334 2017-05-11 Richard Biener <rguenther@suse.de>
17336 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17337 Skip unreachable blocks and destinations.
17338 (eliminate): Move stmt removal and fixup ...
17339 (fini_eliminate): ... here. Skip inserted exprs.
17340 (pass_pre::execute): Move fini_pre after fini_eliminate.
17341 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
17342 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
17343 PRE to get rid of dead code that has invalid SSA form and
17344 split critical edges again.
17346 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17348 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17350 2017-05-11 Richard Biener <rguenther@suse.de>
17352 * passes.c (execute_function_todo): Verify loops if they are
17353 said to be up-to-date.
17354 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
17355 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
17357 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
17360 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
17361 handle calling assemble_external ourself.
17364 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
17365 modes with zero size. Enhance comment.
17367 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17369 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
17370 built-ins for vec_xl and vec_xst with short and char pointer
17373 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
17375 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
17376 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
17377 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
17378 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
17379 (_mm_maskz_min_round_ss): New intrinsics.
17380 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
17381 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
17382 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
17383 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
17384 (__builtin_ia32_minss_mask_round): New builtins.
17385 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17386 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
17387 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
17389 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
17390 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
17392 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
17395 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
17397 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
17398 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
17399 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
17400 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
17401 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
17402 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
17403 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
17404 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17405 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
17406 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
17407 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
17408 (__builtin_ia32_mulss_mask_round): New builtins.
17409 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17410 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
17411 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
17413 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
17414 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
17416 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
17419 2017-05-10 Julia Koval <julia.koval@intel.com>
17421 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
17422 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
17423 (_mm256_setr_m128i): New intrinsics.
17425 2017-05-10 Julia Koval <julia.koval@intel.com>
17427 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
17428 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
17429 (_mm_maskz_rcp14_ss): New intrinsics.
17430 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
17431 (__builtin_ia32_rcp14ss_mask): New builtins.
17432 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
17434 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
17436 PR tree-optimization/51513
17437 * tree-cfg.c (gimple_seq_unreachable_p): New function.
17438 (assert_unreachable_fallthru_edge_p): Use it.
17439 (group_case_labels_stmt): Likewise.
17440 * tree-cfg.h: Prototype it.
17441 * stmt.c: Include cfghooks.h and tree-cfg.h.
17442 (emit_case_dispatch_table) <gap_label>: New local variable.
17443 Use it to fill dispatch table gaps.
17444 Test for default_label before updating probabilities.
17445 (expand_case) <default_label>: Remove unneeded initialization.
17446 Test for unreachable default case statement and remove its edge.
17447 Set default_label accordingly.
17448 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
17450 2017-05-10 Carl Love <cel@us.ibm.com>
17452 * config/rs6000/rs6000-c: Add support for built-in functions
17453 vector signed char vec_neg (vector signed char)
17454 vector signed short int vec_neg (vector short int)
17455 vector signed int vec_neg (vector signed int)
17456 vector signed long long vec_neg (vector signed long long)
17457 vector float vec_neg (vector float)
17458 vector double vec_neg (vector double)
17459 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
17461 * config/rs6000/altivec.h: Add define for vec_neg
17462 * doc/extend.texi: Update the built-in documentation for the
17463 new built-in functions.
17465 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17467 PR tree-optimization/77644
17468 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
17470 2017-05-10 Nathan Sidwell <nathan@acm.org>
17472 * dumpfile.h (TDI_lang_all): New.
17473 (TDF_KIND): New. Renumber others
17474 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
17476 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
17478 (get_dump_file_name): Adjust suffix generation.
17479 (dump_enable_all): Use TDF_KIND.
17480 * doc/invoke.texi (-fdump-lang-all): Document.
17482 * dumpfile.h: Tabify.
17484 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
17487 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
17488 Move member access before delete.
17490 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
17492 * tree-inline.c (expand_call_inline): Split block at stmt
17495 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
17498 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
17499 are now unused after splitting mov{sf,sd}_hardfloat.
17500 (f32_lr2): Likewise.
17501 (f32_lm): Likewise.
17502 (f32_lm2): Likewise.
17503 (f32_li): Likewise.
17504 (f32_li2): Likewise.
17505 (f32_lv): Likewise.
17506 (f32_sr): Likewise.
17507 (f32_sr2): Likewise.
17508 (f32_sm): Likewise.
17509 (f32_sm2): Likewise.
17510 (f32_si): Likewise.
17511 (f32_si2): Likewise.
17512 (f32_sv): Likewise.
17513 (f32_dm): Likewise.
17514 (f32_vsx): Likewise.
17515 (f32_av): Likewise.
17516 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
17517 For movsf, order stores so the VSX stores occur before the GPR
17518 store which encourages the register allocator to use a traditional
17519 FPR instead of a GPR. For movsd, order the stores so that the GPR
17520 store comes before the VSX stores to allow the power6 to work.
17521 This is due to the power6 not having a 32-bit integer store
17522 instruction from a FPR.
17523 (movsf_hardfloat): Likewise.
17524 (movsd_hardfloat): Likewise.
17526 2017-05-09 Martin Sebor <msebor@redhat.com>
17528 PR translation/80280
17529 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
17532 PR translation/80280
17533 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
17534 data member added in r247778.
17535 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
17537 2017-05-09 Nathan Sidwell <nathan@acm.org>
17539 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
17541 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
17544 2017-05-09 Marek Polacek <polacek@redhat.com>
17546 * doc/invoke.texi: Fix typo.
17548 2017-05-09 Richard Biener <rguenther@suse.de>
17550 * tree-vrp.c (vrp_val_is_max): Adjust comment.
17551 (vrp_val_is_min): Likewise.
17552 (set_value_range_to_value): Likewise.
17553 (set_value_range_to_nonnegative): Likewise.
17554 (gimple_assign_nonzero_p): Likewise.
17555 (gimple_stmt_nonzero_p): Likewise.
17556 (vrp_int_const_binop): Likewise. Remove unreachable case.
17557 (adjust_range_with_scev): Adjust comments.
17558 (compare_range_with_value): Likewise.
17559 (extract_range_from_phi_node): Likewise.
17560 (test_for_singularity): Likewise.
17562 2017-05-09 Richard Biener <rguenther@suse.de>
17564 * tree-vrp.c (get_single_symbol): Add assert that we don't
17565 get overflowed constants as invariant part.
17566 (compare_values_warnv): Add comment before the TREE_NO_WARNING
17567 checks. Use wi::cmp instead of recursing for integer constants.
17568 (compare_values): Just ignore whether we assumed undefined
17569 overflow instead of failing the compare.
17570 (extract_range_for_var_from_comparison_expr): Add comment before the
17571 TREE_NO_WARNING sets.
17572 (test_for_singularity): Likewise.
17573 (extract_range_from_comparison): Do not disable optimization
17574 when we assumed undefined overflow.
17575 (extract_range_basic): Remove init of unused var.
17577 2017-05-09 Richard Biener <rguenther@suse.de>
17579 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
17580 (extract_range_from_multiplicative_op_1): Adjust.
17581 (extract_range_from_binary_expr_1): Use int_const_binop.
17583 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
17586 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
17587 rs6000_store_data_bypass_p in seven define_bypass directives and
17588 in several comments.
17589 * config/rs6000/rs6000-protos.h: Add prototype for
17590 rs6000_store_data_bypass_p function.
17591 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
17592 function implements slightly different (rs6000-specific) semantics
17593 than store_data_bypass_p, returning false rather than aborting
17594 with assertion error when arguments do not satisfy the
17595 requirements of store data bypass.
17596 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
17597 rs6000_store_data_bypass_p.
17599 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
17601 * config/xtensa/xtensa-protos.h
17602 (xtensa_initial_elimination_offset): New declaration.
17603 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
17604 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
17605 macro definition, add case for FRAME_POINTER_REGNUM when
17606 FRAME_GROWS_DOWNWARD.
17607 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
17608 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
17609 xtensa_initial_elimination_offset.
17611 2017-05-08 Nathan Sidwell <nathan@acm.org>
17613 * doc/invoke.texi: Alphabetize -fdump options.
17615 2017-05-08 Martin Sebor <msebor@redhat.com>
17617 PR translation/80280
17618 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
17620 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
17622 * target.def (compute_frame_layout): New optional target hook.
17623 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
17624 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
17625 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
17627 * reload1.c (verify_initial_elim_offsets): Likewise.
17628 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
17629 (use_simple_return_p): Call arm_compute_frame_layout if needed.
17630 (arm_get_frame_offsets): Split up into this ...
17631 (arm_compute_frame_layout): ... and this function.
17633 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
17635 * config/aarch64/constraints.md (Usa): New constraint.
17636 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
17638 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
17640 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
17641 with_multilib_list after it has been checked.
17643 2017-05-08 Richard Biener <rguenther@suse.de>
17645 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
17646 (bitmap_set_subtract_values): Likewise.
17648 2017-05-08 Richard Biener <rguenther@suse.de>
17650 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
17651 (gimple_assign_nonzero): ... this and remove strict_overflow_p
17653 (gimple_stmt_nonzero_warnv_p): Rename to ...
17654 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
17656 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
17657 (extract_range_basic): Adjust, do not disable propagation on
17658 strict overflow sensitive simplification.
17659 (vrp_visit_cond_stmt): Likewise.
17661 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
17663 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
17664 body size unconditionally.
17666 2017-05-07 Jeff Law <law@redhat.com>
17669 2017-05-06 Jeff Law <law@redhat.com>
17670 PR tree-optimization/78496
17671 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
17674 PR tree-optimization/78496
17675 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
17676 (simplify_stmt_using_ranges): Call it.
17677 (vrp_dom_walker::before_dom_children): Extract equivalences
17678 from an ASSERT_EXPR with an equality comparison against a
17681 2017-05-06 Jeff Law <law@redhat.com>
17683 PR tree-optimization/78496
17684 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
17687 PR tree-optimization/78496
17688 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
17689 (simplify_stmt_using_ranges): Call it.
17690 (vrp_dom_walker::before_dom_children): Extract equivalences
17691 from an ASSERT_EXPR with an equality comparison against a
17694 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
17696 * lra-constraints.c (lra_copy_reg_equiv): New function.
17697 (split_reg): Use it to copy equivalence information from the
17698 original register to the spill register.
17700 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
17702 PR rtl-optimization/75964
17703 * simplify-rtx.c (simplify_const_relational_operation): Remove
17704 invalid handling of comparisons of integer ABS.
17706 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
17708 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
17709 initialize to zero.
17710 (init_regs): Remove declaration.
17711 (function_arg_advance_32): Initialize error_p as boolean variable.
17713 2017-05-05 Nathan Sidwell <nathan@acm.org>
17715 * store-motion.c (remove_reachable_equiv_notes): Reformat long
17716 lines. Use for (;;).
17718 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17720 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
17721 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
17722 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
17723 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
17724 VF=2 that require versioning.
17726 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17728 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
17731 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17733 * diagnostic.h (diagnostic_override_option_index): Convert from
17734 macro to inline function.
17736 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17738 * diagnostic.c (last_module_changed_p): New function.
17739 (set_last_module): New function.
17740 (diagnostic_report_current_module): Convert macro usage to
17741 the above functions.
17742 * diagnostic.h (diagnostic_context::last_module): Strengthen
17743 from const line_map * to const line_map_ordinary *.
17744 (diagnostic_last_module_changed): Delete macro.
17745 (diagnostic_set_last_module): Delete macro.
17747 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17749 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
17750 with diagnostic_report_diagnostic.
17751 (diagnostic_n_impl_richloc): Likewise.
17752 * diagnostic.h (report_diagnostic): Delete macro.
17753 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
17754 with diagnostic_report_diagnostic.
17755 * substring-locations.c (format_warning_va): Likewise.
17757 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17759 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
17760 save/restor of format_spec. Move option-printing code to...
17761 (print_option_information): ...this new function, and
17762 reimplement by simply printing to the pretty_printer,
17763 rather than appending to the format string.
17765 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17767 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
17768 handling logic into...
17769 (update_effective_level_from_pragmas): ...this new function.
17771 2017-05-04 Andrew Waterman <andrew@sifive.com>
17773 * config/riscv/riscv.opt (mstrict-align): New option.
17774 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
17775 (SLOW_UNALIGNED_ACCESS): Define.
17776 (riscv_slow_unaligned_access): Declare.
17777 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
17779 (riscv_slow_unaligned_access): New variable.
17780 (rocket_tune_info): Set slow_unaligned_access to true.
17781 (optimize_size_tune_info): Set slow_unaligned_access to false.
17782 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
17783 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
17784 (riscv_option_override): Set riscv_slow_unaligned_access.
17785 * doc/invoke.texi: Add -mstrict-align to RISC-V.
17787 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
17789 * config/riscv/riscv.md: Unify indentation.
17791 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
17796 * config/rs6000/rs6000.md (u code attribute): Add FIX and
17798 (extendsi<mode>2): Add support for doing sign extension via
17799 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
17800 don't have ISA 3.0 instructions.
17801 (extendsi<mode>2 splitter): Likewise.
17802 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
17803 generate the normal insns since SImode can now go in vector
17804 registers. Disallow the special UNSPECs needed for previous
17805 machines to hide SImode being used. Add new insns
17806 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
17807 (fix_trunc<mode>si2_stfiwx): Likewise.
17808 (fix_trunc<mode>si2_internal): Likewise.
17809 (fixuns_trunc<mode>si2): Likewise.
17810 (fixuns_trunc<mode>si2_stfiwx): Likewise.
17811 (fctiw<u>z_<mode>_smallint): Likewise.
17812 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
17813 of floating point to 32-bit integer from doing a direct move to
17814 the GPR registers to do a store.
17815 (fctiwz_<mode>): Break long line.
17817 2017-05-05 Bin Cheng <bin.cheng@arm.com>
17819 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
17820 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
17821 (addr_list, addr_offset_valid_p): New.
17822 (split_address_groups): Check offset validity with above function.
17823 (gt-tree-ssa-loop-ivopts.h): Include header file.
17825 2017-05-05 Nathan Sidwell <nathan@acm.org>
17827 * config.gcc (arm*-*-*): Add missing 'fi'.
17829 2017-05-05 Steve Ellcey <sellcey@cavium.com>
17831 * doc/invoke.texi (-fopt-info): Explicitly say order of options
17832 included in -fopt-info does not matter.
17833 * doc/optinfo.texi (-fopt-info): Fix description of default
17834 behavour. Explicitly say order of options included in -fopt-info
17837 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
17839 * config.gcc: Allow combinations of aprofile and rmprofile values for
17840 --with-multilib-list.
17841 * config/arm/t-multilib: New file.
17842 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
17843 variables. Remove setting of ISA and floating-point ABI in
17844 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
17845 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
17846 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
17847 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
17849 * config/arm/t-rmprofile: Likewise except for the matches changes.
17850 * doc/install.texi (--with-multilib-list): Document the combination of
17851 aprofile and rmprofile values and warn about pitfalls in doing that.
17853 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
17855 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
17856 (movdi_aarch64): Likewise.
17858 2017-05-05 Jakub Jelinek <jakub@redhat.com>
17860 PR tree-optimization/80632
17861 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
17863 (build_arrays): Initialize it for virtual phis.
17864 (fix_phi_nodes): Use it for virtual phis.
17866 PR tree-optimization/80558
17867 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
17868 [x, y] op z into [x op, y op z] for op & or | if conditions
17871 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
17872 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17875 * config/arm/arm.md (use_literal_pool): Remove.
17876 (64-bit immediate split): No longer takes cost into consideration
17877 if arm_disable_literal_pool is enabled.
17878 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
17879 used when arm_disable_literal_pool is enabled.
17880 (arm_max_const_double_inline_cost): Remove use of
17881 arm_disable_literal_pool.
17882 (push_minipool_fix): Add assert.
17883 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
17884 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
17885 (no_literal_pool_sf_immediate): New.
17887 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17889 PR tree-optimization/80613
17890 * tree-ssa-dce.c (propagate_necessity): Remove cases for
17891 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
17893 2017-05-05 Richard Biener <rguenther@suse.de>
17895 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
17897 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
17899 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
17900 of this flag from insn conditions due to removal from r247495.
17902 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
17904 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
17906 (arm_early_store_addr_dep_ptr): Likewise.
17907 * config/arm/aarch-common-protos.h
17908 (arm_early_load_addr_dep_ptr): Add prototype.
17909 (arm_early_store_addr_dep_ptr): Likewise.
17910 * config/arm/cortex-a53.md: Add new bypasses.
17912 2017-05-05 Jakub Jelinek <jakub@redhat.com>
17914 * tree.c (next_type_uid): Change type to unsigned.
17915 (type_hash_canon): Decrement back next_type_uid if
17916 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
17917 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
17920 2017-05-04 Martin Sebor <msebor@redhat.com>
17922 * builtins.c: Fix a trivial typo in a comment.
17924 PR middle-end/79234
17925 * builtins.c (check_sizes): Adjust to handle reading past the end.
17926 Avoid printing excessive upper bound of ranges. Use %E to print
17927 tree nodes instead of converting them to %wu.
17928 (expand_builtin_memchr): New function.
17929 (compute_dest_size): Rename...
17930 (compute_objsize): ...to this.
17931 (expand_builtin_memcpy): Adjust.
17932 (expand_builtin_mempcpy): Adjust.
17933 (expand_builtin_strcat): Adjust.
17934 (expand_builtin_strcpy): Adjust.
17935 (check_strncat_sizes): Adjust.
17936 (expand_builtin_strncat): Adjust.
17937 (expand_builtin_strncpy): Adjust and simplify.
17938 (expand_builtin_memset): Adjust.
17939 (expand_builtin_bzero): Adjust.
17940 (expand_builtin_memcmp): Adjust.
17941 (expand_builtin): Handle memcmp.
17942 (maybe_emit_chk_warning): Check strncat just once.
17944 2017-05-04 Martin Sebor <msebor@redhat.com>
17946 PR preprocessor/79214
17947 PR middle-end/79222
17948 PR middle-end/79223
17949 * builtins.c (check_sizes): Add inlining context and issue
17950 warnings even when -Wno-system-headers is set.
17951 (check_strncat_sizes): Same.
17952 (expand_builtin_strncat): Same.
17953 (expand_builtin_memmove): New function.
17954 (expand_builtin_stpncpy): Same.
17955 (expand_builtin): Handle memmove and stpncpy.
17957 2017-05-04 Bin Cheng <bin.cheng@arm.com>
17959 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
17960 which is not used any more.
17962 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17964 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
17966 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17968 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
17969 (cortexa53_tunings): Likewise.
17970 (cortexa57_tunings): Likewise.
17971 (cortexa72_tunings): Likewise.
17972 (cortexa73_tunings): Likewise.
17974 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17976 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
17977 Set loop alignment to 8.
17979 2017-05-04 Martin Sebor <msebor@redhat.com>
17981 PR translation/80280
17982 * builtins.c (expand_builtin_object_size): Add missing quoting to
17983 %D and like directives.
17984 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
17985 (hsa_type_for_tree_type): Same.
17986 (verify_function_arguments): Same.
17987 * symtab.c (symbol_table::change_decl_assembler_name): Same.
17988 * varasm.c (get_section): Same.
17991 2017-05-04 Martin Sebor <msebor@redhat.com>
17993 PR translation/80280
17994 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
17996 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17998 * config/aarch64/aarch64.c (generic_addrcost_table):
17999 Change HI/TI mode setting.
18001 2017-05-04 Martin Jambor <mjambor@suse.cz>
18003 PR tree-optimization/80622
18004 * tree-sra.c (comes_initialized_p): New function.
18005 (build_accesses_from_assign): Only set write lazily when
18006 comes_initialized_p is false.
18007 (analyze_access_subtree): Use comes_initialized_p.
18008 (propagate_subaccesses_across_link): Assert !comes_initialized_p
18009 instead of testing for PARM_DECL.
18011 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18013 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
18014 constraint on operand 0 to allow more general addressing modes.
18015 Adjust output template.
18016 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
18018 * config/aarch64/aarch64-protos.h
18019 (aarch64_address_valid_for_prefetch_p): Declare prototype.
18020 * config/aarch64/constraints.md (Dp): New address constraint.
18021 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
18024 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
18026 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
18027 update use of estimate_ipcp_clone_size_and_time.
18028 (estimate_local_effects): Update use of
18029 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
18030 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
18031 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
18032 Return nonspecialized time.
18034 2017-05-04 Richard Biener <rguenther@suse.de>
18036 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
18037 for the last VUSE which def dominates the PHI. Directly call
18039 (get_continuation_for_phi_1): Remove.
18041 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
18043 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
18044 to explain the use of truncating division. Cap the number of
18045 iterations to the maximum given by nb_iterations_upper_bound,
18048 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18050 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
18051 * configure: Regenerate.
18052 * config.in: Regenerate.
18053 * config/i386/driver-mingw32.c: new file.
18054 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
18055 * config.host: Link driver-mingw32.o on MinGW host.
18056 * doc/install.texi: Document new --enable-mingw-wildcard configure
18059 2017-05-04 Marek Polacek <polacek@redhat.com>
18061 PR tree-optimization/80612
18062 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
18064 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18065 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
18067 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
18068 (movt splitter): Likewise.
18069 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
18070 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
18071 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
18072 block for Thumb-1 with MOVT.
18073 (thumb2_legitimate_address_p): Move code block ...
18074 (can_avoid_literal_pool_for_label_p): ... into this new function.
18075 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
18077 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
18078 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
18079 "M-profile targets with the MOVT instruction".
18081 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18083 * config/arm/arm-builtins.c (arm_init_builtins): Rename
18084 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
18085 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
18087 2017-05-04 Martin Liska <mliska@suse.cz>
18089 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
18090 variable cond_code.
18092 2017-05-04 Richard Biener <rguenther@suse.de>
18094 * tree.c (array_at_struct_end_p): Handle arrays at struct
18095 end with flexarrays more conservatively. Refactor and treat
18096 arrays of arrays or aggregates more strict. Fix
18097 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
18098 * tree.c (array_at_struct_end_p): Adjust prototype.
18099 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
18100 * gimple-fold.c (get_range_strlen): Likewise.
18101 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
18103 2017-05-04 Richard Biener <rguenther@suse.de>
18105 PR tree-optimization/31130
18106 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
18108 (supports_overflow_infinity): Likewise.
18109 (is_negative_overflow_infinity): Likewise.
18110 (is_positive_overflow_infinity): Likewise.
18111 (is_overflow_infinity): Likewise.
18112 (stmt_overflow_infinity): Likewise.
18113 (overflow_infinity_range_p): Likewise.
18114 (usable_range_p): Remove as always returning true.
18115 (make_overflow_infinity): Remove.
18116 (negative_overflow_infinity): Likewise.
18117 (positive_overflow_infinity): Likewise.
18118 (avoid_overflow_infinity): Likewise.
18119 (set_value_range): Adjust accordingly.
18120 (set_value_range_to_nonnegative): Likewise, remove now unused
18121 overflow_infinity arg.
18122 (vrp_operand_equal_p): Adjust.
18123 (update_value_range): Likewise.
18124 (range_int_cst_singleton_p): Likewise.
18125 (operand_less_p): Likewise.
18126 (compare_values_warnv): Likewise.
18127 (extract_range_for_var_from_comparison_expr): Likewise.
18128 (vrp_int_const_binop): Likewise.
18129 (zero_nonzero_bits_from_vr): Likewise.
18130 (extract_range_from_multiplicative_op_1): Likewise.
18131 (extract_range_from_binary_expr_1): Likewise.
18132 (extract_range_from_unary_expr): Likewise.
18133 (extract_range_from_comparison): Likewise.
18134 (extract_range_basic): Likewise.
18135 (adjust_range_with_scev): Likewise.
18136 (compare_ranges): Likewise.
18137 (compare_range_with_value): Likewise.
18138 (dump_value_range): Likewise.
18139 (test_for_singularity): Likewise, remove strict_overflow_p parameter
18141 (simplify_cond_using_ranges): Adjust.
18143 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
18145 * brig-builtins.def: Added a builtin for class_f64.
18146 * builtin-types.def: Added a builtin type needed by class_f64.
18148 2017-05-03 Jason Merrill <jason@redhat.com>
18150 * timevar.def: Add TV_CONSTEXPR.
18152 2017-05-03 David Malcolm <dmalcolm@redhat.com>
18154 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
18156 2017-05-03 Martin Jambor <mjambor@suse.cz>
18158 * ipa-prop.c (ipa_update_after_lto_read): Removed.
18159 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
18160 * ipa-cp.c (ipcp_propagate_stage): Do not call
18161 ipa_update_after_lto_read.
18162 * ipa-inline.c (ipa_inline): Likewise.
18164 2017-05-03 Martin Jambor <mjambor@suse.cz>
18166 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
18167 tag. Added a default constructor and a destructor.
18168 (ipa_edge_args_sum_t): New class;
18169 (ipa_edge_args_sum): Declare.
18170 (ipa_edge_args_vector): Remove declaration.
18171 (IPA_EDGE_REF): Use ipa_edge_args_sum.
18172 (ipa_free_edge_args_substructures): Remove declaration.
18173 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
18174 (ipa_edge_args_info_available_for_edge_p): Likewise.
18175 * ipa-prop.c (ipa_edge_args_vector): Removed.
18176 (edge_removal_hook_holder): Likewise.
18177 (edge_duplication_hook_holder): Likewise.
18178 (ipa_edge_args_sum): New variable.
18179 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
18180 ipa_edge_args_vector.
18181 (ipa_free_edge_args_substructures): Likewise.
18182 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
18183 ipa_edge_args_vector.
18184 (ipa_edge_removal_hook): Turned into method
18185 ipa_edge_args_sum_t::remove.
18186 (ipa_edge_duplication_hook): Turned into method
18187 ipa_edge_args_sum_t::duplicate.
18188 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
18189 registering edge hooks.
18190 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
18191 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
18192 ipa_edge_args_sum instead of ipa_edge_args_vector.
18193 * ipa-profile.c (ipa_profile): Likewise.
18195 2017-05-03 Martin Jambor <mjambor@suse.cz>
18197 * symbol-summary.h (function_summary): New method exists.
18198 (function_summary::symtab_removal): Deallocate through release.
18199 (call_summary): New class.
18200 (gt_ggc_mx): New overload.
18201 (gt_pch_nx): Likewise.
18202 (gt_pch_nx): Likewise.
18204 2017-05-03 Jeff Law <law@redhat.com>
18206 PR tree-optimization/78496
18207 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
18208 from simplify_cond_using_ranges. Split off code to walk
18209 backwards through casts into ...
18210 (simplify_cond_using_ranges_2): New function.
18211 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
18212 (execute_vrp): After identifying jump threads, call
18213 simplify_cond_using_ranges_2.
18215 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
18218 * ipa-inline.h (inline_summary): Add ctor.
18219 (create_ggc): Do not use ggc_cleared_alloc.
18221 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
18222 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18224 * gcc.c (handle_braces): Support escaping in switch matching
18226 * doc/invoke.texi (Spec Files): Document it.
18227 Remove superfluous @code markup in items.
18229 2017-05-03 David Malcolm <dmalcolm@redhat.com>
18231 * diagnostic-show-locus.c (struct column_range): New struct.
18232 (get_affected_columns): New function.
18233 (get_printed_columns): New function.
18234 (struct correction): New struct.
18235 (correction::ensure_capacity): New function.
18236 (correction::ensure_terminated): New function.
18237 (struct line_corrections): New struct.
18238 (line_corrections::~line_corrections): New dtor.
18239 (line_corrections::add_hint): New function.
18240 (layout::print_trailing_fixits): Reimplement in terms of the new
18242 (selftest::test_overlapped_fixit_printing): New function.
18243 (selftest::diagnostic_show_locus_c_tests): Call it.
18245 2017-05-03 Nathan Sidwell <nathan@acm.org>
18247 Canonicalize canonical type hashing
18248 * tree.h (type_hash_canon_hash): Declare.
18249 * tree.c (type_hash_list, attribute_hash_list): Move into
18250 type_hash_canon_hash.
18251 (build_type_attribute_qual_variant): Break out hash code calc into
18252 type_hash_canon_hash.
18253 (type_hash_canon_hash): New. Generic type hash computation.
18254 (build_range_type_1, build_array_type_1, build_function_type,
18255 build_method_type_directly, build_offset_type, build_complex_type,
18256 make_vector_type): Call it.
18258 2017-05-03 Richard Biener <rguenther@suse.de>
18260 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
18261 When all DRs have unknown misaligned do not always peel
18262 when there is a store but apply the same costing model as if
18263 there were only loads.
18265 2017-05-03 Richard Biener <rguenther@suse.de>
18268 PR tree-optimization/80492
18269 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
18270 compare_base_decls returning dont-know properly.
18272 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
18274 * config/arm/iterators.md (CCSI): New mode iterator.
18275 (arch): New mode attribute.
18276 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
18277 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
18278 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
18279 code iterator for success result mode.
18280 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
18281 the corresponding new insn generators.
18283 2017-05-03 Bin Cheng <bin.cheng@arm.com>
18286 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18287 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18289 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
18291 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
18292 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
18293 (DDR_A): Wrap DDR argument in brackets.
18294 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
18295 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
18296 (DDR_REVERSED_P): Likewise.
18298 2017-05-03 Jakub Jelinek <jakub@redhat.com>
18300 PR tree-optimization/79472
18301 * tree-switch-conversion.c (struct switch_conv_info): Add
18302 contiguous_range and default_case_nonstandard fields.
18303 (collect_switch_conv_info): Compute contiguous_range and
18304 default_case_nonstandard fields, don't clear final_bb if
18305 contiguous_range and only the default case doesn't have the required
18307 (check_all_empty_except_final): Set default_case_nonstandard instead
18308 of failing if contiguous_range and the default case doesn't have empty
18310 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
18311 and only the default case doesn't have the required constants. Skip
18313 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
18314 if default_case_nonstandard.
18315 (build_constructors): Build constant 1 just once. Assert that default
18316 values aren't inserted in between cases if contiguous_range. Skip
18318 (build_arrays): Skip virtual phis.
18319 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
18320 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
18321 Handle virtual phis.
18322 (gen_inbound_check): Handle default_case_nonstandard case.
18323 (process_switch): Adjust check_final_bb caller. Call
18324 gather_default_values with the first non-default case instead of
18325 default case if default_case_nonstandard.
18327 2017-05-02 Nathan Sidwell <nathan@acm.org>
18329 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
18330 check. Fix formatting.
18332 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
18334 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
18335 errors when comparing specialized and unspecialized times.
18337 2017-05-02 David Malcolm <dmalcolm@redhat.com>
18339 * diagnostic-show-locus.c
18340 (layout::should_print_annotation_line_p): Make private.
18341 (layout::print_annotation_line): Make private.
18342 (layout::annotation_line_showed_range_p): Make private.
18343 (layout::show_ruler): Make private.
18344 (layout::print_source_line): Make private. Pass in line and
18345 line_width, rather than calling location_get_source_line. Drop
18347 (layout::print_leading_fixits): New method.
18348 (layout::print_any_fixits): Rename to...
18349 (layout::print_trailing_fixits): ...this, and make private.
18350 Don't print newline fixits.
18351 (diagnostic_show_locus): Move logic for printing one row into...
18352 (layout::print_line): ...this new function. Move the
18353 location_get_source_line call and error-handling from
18354 print_source_line to here. Call print_leading_fixits, and rename
18355 print_any_fixits to print_trailing_fixits.
18356 (selftest::test_fixit_insert_containing_newline): Update now that
18357 newlines are partially supported.
18358 (selftest::test_fixit_insert_containing_newline_2): New test.
18359 (selftest::test_fixit_replace_containing_newline): Update comments.
18360 (selftest::diagnostic_show_locus_c_tests): Call the new test.
18361 * edit-context.c (class added_line): New class.
18362 (class edited_line): Describe newline handling in comment.
18363 (edited_line::actually_edited_p): New method.
18364 (edited_line::print_content): Delete redundant decl.
18365 (edited_line::m_predecessors): New field.
18366 (edited_file::print_content): Call edited_line::print_content.
18367 (edited_file::print_diff): Update to support newlines.
18368 (edited_file::print_diff_hunk): Likewise.
18369 (edited_file::print_run_of_changed_lines): New function.
18370 (edited_file::print_diff_line): Convert to...
18371 (print_diff_line): ...this.
18372 (edited_file::get_effective_line_count): New function.
18373 (edited_line::edited_line): Initialize new field m_predecessors.
18374 (edited_line::~edited_line): Clean up m_predecessors.
18375 (edited_line::apply_fixit): Handle newlines.
18376 (edited_line::get_effective_line_count): New function.
18377 (edited_line::print_content): New function.
18378 (edited_line::print_diff_lines): New function.
18379 (selftest::test_applying_fixits_insert_containing_newline): New
18381 (selftest::test_applying_fixits_replace_containing_newline): New
18383 (selftest::insert_line): New function.
18384 (selftest::test_applying_fixits_multiple_lines): Add example of
18386 (selftest::edit_context_c_tests): Call the new tests.
18388 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18390 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
18391 parameter cand. Update dump information.
18392 (get_computation_cost): Update uses.
18394 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18396 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
18397 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
18398 (get_computation_at, rewrite_use_address): Update use of
18399 get_computation_aff.
18401 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18403 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
18404 (get_computation): Delete.
18405 (get_computation_cost): Implement like get_computation_cost_at.
18406 Use get_computation_at.
18407 (get_computation_cost_at): Delete.
18408 (rewrite_use_nonlinear_expr): Use get_computation_at.
18409 (rewrite_use_compare, remove_unused_ivs): Ditto.
18411 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18413 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
18415 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18417 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
18418 (ivopts_global_cost_for_size): Rename parameter and update uses.
18419 (iv_ca_recount_cost): Update uses.
18420 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
18421 candidates seperately in n_invs and n_cands.
18422 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
18424 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18426 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
18427 (find_inv_vars_cb): New.
18428 (find_depends): Renamed to ...
18429 (find_inv_vars): ... this.
18430 (add_candidate_1, force_var_cost): Call find_inv_vars.
18431 (split_address_cost, determine_group_iv_cost_cond): Ditto.
18433 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18435 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
18436 inv_vars. Add inv_exprs.
18437 (struct iv_cand): Rename depends_on to inv_vars.
18438 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
18439 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
18440 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
18441 (dump_cand): Dump inv_vars.
18442 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
18443 (record_invariant, find_depends, add_candidate_1): Ditto.
18444 (set_group_iv_cost, force_var_cost): Ditto.
18445 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
18446 (get_computation_cost_at, get_computation_cost): Ditto.
18447 (determine_group_iv_cost_generic): Ditto.
18448 (determine_group_iv_cost_address): Ditto.
18449 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
18450 (determine_group_iv_costs): Ditto.
18451 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
18452 (iv_ca_set_remove_invariants): Renamed to ...
18453 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
18454 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
18455 (iv_ca_set_add_invariants): Renamed to ...
18456 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
18457 (iv_ca_set_cp): Use iv_ca_set_add_invs.
18458 (iv_ca_has_deps): Support inv_vars and inv_exprs.
18459 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
18460 (create_new_ivs): Remove useless dump.
18462 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18464 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
18466 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
18467 (iv_ca_set_no_cp, create_new_iv): Ditto.
18469 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18471 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18473 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
18475 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
18476 function tree_check2.
18478 2017-05-02 Martin Liska <mliska@suse.cz>
18480 * doc/gcov.texi: Add missing preposition.
18481 * gcov.c (function_info::function_info): Properly fill up
18482 all member variables.
18484 2017-05-02 Tamar Christina <tamar.christina@arm.com>
18486 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
18488 2017-05-02 Tamar Christina <tamar.christina@arm.com>
18490 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
18492 2017-05-02 Martin Liska <mliska@suse.cz>
18495 * lto-streamer-in.c (lto_read_tree_1): Remove
18496 LTO_STREAMER_DEBUG.
18497 * lto-streamer.c (struct tree_hash_entry): Likewise.
18498 (struct tree_entry_hasher): Likewise.
18499 (tree_entry_hasher::hash): Likewise.
18500 (tree_entry_hasher::equal): Likewise.
18501 (lto_streamer_init): Likewise.
18502 (lto_orig_address_map): Likewise.
18503 (lto_orig_address_get): Likewise.
18504 (lto_orig_address_remove): Likewise.
18505 * lto-streamer.h: Likewise.
18506 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
18507 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
18509 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
18511 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
18512 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
18513 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
18514 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
18515 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
18516 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
18517 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
18518 (mm_maskz_sub_ss): New intrinsics.
18519 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18520 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
18521 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
18522 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
18523 (__builtin_ia32_subss_mask_round): New builtins.
18524 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18525 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18526 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
18528 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
18529 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
18531 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
18534 2017-05-02 Martin Jambor <mjambor@suse.cz>
18536 PR tree-optimization/78687
18537 * tree-sra.c (access): New field parent.
18538 (process_subtree_disqualification): New function.
18539 (disqualify_candidate): Call it.
18540 (build_accesses_from_assign): Reset write flag if creating an
18542 (build_access_subtree): Fill in parent field and also prpagate
18543 down grp_write flag.
18544 (create_artificial_child_access): New parameter set_grp_write, set
18545 grp_write to its value.
18546 (propagate_subaccesses_across_link): Also propagate grp_write flag
18548 (propagate_all_subaccesses): Push the closest parent back to work
18549 queue if add_access_to_work_queue returned true.
18551 2017-05-02 Richard Biener <rguenther@suse.de>
18553 * common.opt (fstrict-overflow): Alias negative to fwrapv.
18554 * doc/invoke.texi (fstrict-overflow): Remove all traces of
18555 -fstrict-overflow documentation.
18556 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
18557 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
18558 flag_strict_overflow.
18559 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
18560 * lto-opts.c (lto_write_options): Do not stream it.
18561 * lto-wrapper.c (merge_and_complain): Do not handle it.
18562 * opts.c (default_options_table): Do not set -fstrict-overflow.
18563 (finish_options): Likewise do not clear it when sanitizing.
18564 * simplify-rtx.c (simplify_const_relational_operation): Do not
18565 test flag_strict_overflow.
18567 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
18569 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
18570 using enabled attribute.
18571 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
18572 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
18573 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
18574 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
18575 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
18576 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
18577 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
18578 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
18579 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
18580 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
18582 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
18584 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
18586 2017-05-02 Richard Biener <rguenther@suse.de>
18588 PR tree-optimization/80591
18590 2017-04-10 Richard Biener <rguenther@suse.de>
18592 * tree-ssa-structalias.c (find_func_aliases): Properly handle
18595 2017-05-02 Richard Biener <rguenther@suse.de>
18597 PR tree-optimization/80549
18598 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
18599 (cleanup_tree_cfg_noloop): Create forwarders to known loop
18600 headers if they do not have a preheader.
18602 2017-05-02 Martin Liska <mliska@suse.cz>
18605 * common.opt: Fix typo.
18606 * doc/invoke.texi: Likewise.
18608 2017-05-01 Jan Beulich <jbeulich@suse.com>
18610 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
18611 swapping, add (x,x,m,x,n) alternative.
18613 2017-05-01 Nathan Sidwell <nathan@acm.org>
18615 * calls.c (combine_pending_stack_adjustment_and_call): Remove
18616 unnecessary unadjusted_alignment check.
18618 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
18621 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
18623 * gimplify.c (gimplify_cilk_detach): New function.
18624 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
18625 * tree-core.h: Document EXPR_CILK_SPAWN.
18626 * tree.h (EXPR_CILK_SPAWN): Define.
18628 2017-05-01 David Malcolm <dmalcolm@redhat.com>
18630 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
18631 to use new fixit_hint representation, using the "replace" logic.
18632 (get_line_span_for_fixit_hint): Likewise.
18633 (layout::print_any_fixits): Likewise.
18634 (selftest::test_one_liner_many_fixits): Rename to...
18635 (selftest::test_one_liner_many_fixits_1): ...this, and update
18636 comment and expected output to reflect that the multiple fix-it
18637 hints are now consolidated into one insertion.
18638 (selftest::test_one_liner_many_fixits_2): New test.
18639 (selftest::test_diagnostic_show_locus_one_liner): Update for
18641 (selftest::test_fixit_consolidation): Update for fix-it API
18643 * diagnostic.c (print_parseable_fixits): Likewise.
18644 * edit-context.c (edited_line::m_line_events): Convert from
18645 auto_vec <line_event *> to auto_vec <line_event>.
18646 (class line_event): Convert from abstract base class to a concrete
18647 class, taking over the role of replace_event.
18648 (class insert_event): Delete.
18649 (class replace_event): Rename to class line_event. Convert to
18651 (edit_context::add_fixits): Reimplement.
18652 (edit_context::apply_insert): Delete.
18653 (edit_context::apply_replace): Rename to...
18654 (edit_context::apply_fixit): ...this. Convert to half-open range.
18655 (edited_file::apply_insert): Delete.
18656 (edited_file::apply_replace): Rename to...
18657 (edited_file::apply_fixit): ...this.
18658 (edited_line::~edited_line): Drop deletion of events.
18659 (edited_line::apply_insert): Delete.
18660 (edited_line::apply_replace): Rename to...
18661 (edited_line::apply_fixit): ...this. Convert to half-open range.
18662 Update for change to type of m_line_events.
18663 * edit-context.h (edit_context::apply_insert): Delete.
18664 (edit_context::apply_replace): Rename to...
18665 (edit_context::apply_fixit): ...this.
18667 2017-05-01 Martin Sebor <msebor@redhat.com>
18669 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
18672 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
18675 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
18676 __get_cpuid_max returns 0.
18677 (__get_cpuid_count): Ditto.
18679 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
18681 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
18682 replacement expression is another instance of one of its arguments.
18684 2017-05-01 Jakub Jelinek <jakub@redhat.com>
18687 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
18688 check for stack push/pop autoinc.
18689 * config/i386/i386.c (ix86_agi_dependent): Return false
18690 if the only reason why modified_in_p returned true is that
18691 addr is SP based and set_insn is a push or pop.
18693 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
18695 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
18698 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
18701 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
18702 (account_size_time): Use two predicates - exec_pred and
18704 (evaluate_conditions_for_known_args): Compute both clause and
18706 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
18707 (inline_summary_t::duplicate): Update.
18708 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
18710 (compute_inline_parameters): Likewise.
18711 (estimate_edge_size_and_time): Update caluclation of time.
18712 (estimate_node_size_and_time): Compute both time and nonspecialized
18714 (estimate_ipcp_clone_size_and_time): Update.
18715 (inline_merge_summary): Update.
18716 (do_estimate_edge_time): Update.
18717 (do_estimate_edge_size): Update.
18718 (do_estimate_edge_hints): Update.
18719 (inline_read_section, inline_write_summary): Stream both new predicates.
18720 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
18722 (compute_inlined_call_time): Cleanup.
18723 (big_speedup_p): Update.
18724 (edge_badness): Update.
18725 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
18726 (size_time_entry): Replace predicate by exec_predicate and
18727 nonconst_predicate.
18728 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
18729 (estimate_edge_time): Return also nonspec_time.
18730 (reset_edge_growth_cache): Update.
18732 2017-04-29 Jakub Jelinek <jakub@redhat.com>
18734 PR rtl-optimization/80491
18735 * ifcvt.c (noce_process_if_block): When looking for x setter
18736 with missing else_bb, don't check only the insn right before
18737 cond_earliest, but look for the last insn that x is modified in
18738 within the same bb.
18740 PR rtl-optimization/80491
18741 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
18743 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
18745 PR tree-optimization/80487
18746 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
18748 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18750 PR tree-optimization/79697
18751 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
18752 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
18753 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
18755 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
18756 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
18758 2017-04-28 Martin Sebor <msebor@redhat.com>
18760 PR tree-optimization/80523
18761 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
18762 (init_target_to_host_charmap, target_to_host, target_strtol10): New
18764 (maybe_warn, format_directive, parse_directive): Use new functions.
18765 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
18767 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
18769 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
18771 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
18773 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
18774 target_header_dir): Set correctly.
18775 * configure: Regenerated.
18776 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
18777 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
18778 instead of SYSTEM_HEADER_DIR.
18780 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
18782 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
18783 (estimate_local_effects): Likewise.
18784 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
18785 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
18786 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
18787 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
18788 do_estimate_edge_time, estimate_edge_time): Likewise.
18789 * ipa-inline-analysis.c (estimate_node_size_and_time,
18790 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
18791 (estimate_time_after_inlining): Remove.
18793 2017-04-28 Martin Liska <mliska@suse.cz>
18795 * doc/gcov.texi: Enhance documentation of gcov.
18797 2017-04-28 Martin Liska <mliska@suse.cz>
18799 * doc/gcov.texi: Sort options in alphabetic order.
18800 * doc/gcov-dump.texi: Likewise.
18801 * doc/gcov-tool.texi: Likewise.
18802 * gcov.c (print_usage): Likewise.
18803 * gcov-dump.c (print_usage): Likewise.
18804 * gcov-tool.c (print_merge_usage_message): Likewise.
18805 (print_rewrite_usage_message): Likewise.
18806 (print_overlap_usage_message): Likewise.
18808 2017-04-28 Martin Liska <mliska@suse.cz>
18810 PR gcov-profile/53915
18811 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
18813 2017-04-28 Martin Liska <mliska@suse.cz>
18815 PR gcov-profile/79891
18816 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
18817 is marked by compiler as living on a line.
18818 (get_cycles_count): Remove usage of the union.
18819 (output_intermediate_file): Likewise.
18820 (find_source): Fix GNU coding style.
18821 (accumulate_line_counts): Remove old non-all block mode.
18822 (output_lines): Remove usage of the union.
18823 * profile.c (output_location): Include all BBs, even if
18824 belonging to a same line (and file) as a previous BB.
18826 2017-04-28 Martin Liska <mliska@suse.cz>
18828 * gcov.c (process_args): Handle new argument 'w'.
18829 (read_graph_file): Assign ID to BBs.
18830 (output_branch_count): Display BB # if verbose flag is set.
18831 (output_lines): Likewise for arcs.
18832 (print_usage): Add '--verbose' option help.
18833 * doc/gcov.texi: Document --verbose (-w) option.
18835 2017-04-28 Martin Liska <mliska@suse.cz>
18837 * gcov.c (struct block_location_info): New struct.
18838 (process_file): Fill up the new structure.
18839 (read_graph_file): Replace usage of encoding by the newly added
18841 (add_line_counts): Likewise.
18842 (accumulate_line_counts): Remove usage of the union.
18843 (function_info::function_info): New function.
18844 (function_info::~function_info): Likewise.
18845 (process_file): Call delete instead of release_function.
18846 (release_function): Release the function.
18847 (release_structures): Call delete instead of release_function.
18848 (solve_flow_graph): Replace usage of num_blocks.
18849 (find_exception_blocks): Likewise.
18850 (output_lines): Fix GNU coding style.
18852 2017-04-28 Martin Liska <mliska@suse.cz>
18855 * coverage.c (coverage_remove_note_file): New function.
18856 * coverage.h: Declare the function.
18857 * toplev.c (finalize): Clean if an error has been seen.
18859 2017-04-28 Martin Liska <mliska@suse.cz>
18861 PR gcov-profile/80031
18862 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
18863 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
18864 * gcov.c (read_graph_file): Read just number of blocks.
18865 * profile.c (branch_prob): Do not stream 0 flags per a basic
18868 2017-04-28 Martin Liska <mliska@suse.cz>
18870 * gcov-dump.c (tag_*): Add new argument to declarations.
18871 (dump_gcov_file): Likewise.
18872 (tag_blocks): Add and use new argument depth.
18873 (tag_arcs): Likewise.
18874 (tag_lines): Likewise.
18875 (tag_counters): Likewise.
18876 (tag_summary): Likewise.
18877 (dump_working_sets): Use depth to do a proper indentation.
18879 2017-04-28 Jakub Jelinek <jakub@redhat.com>
18882 * cgraph.h (symtab_node::debug_symtab): No longer inline.
18883 * symtab.c (symtab_node::debug_symtab): Move definition here.
18885 2017-04-28 Richard Biener <rguenther@suse.de>
18887 * lto-streamer.h (LTO_major_version): Bump to 7.
18889 2017-04-28 Richard Biener <rguenther@suse.de>
18891 * tree-vrp.c (assert_info): New struct.
18892 (add_assert_info): New helper.
18893 (register_edge_assert_for_2): Refactor to add asserts to a vector
18895 (register_edge_assert_for_1): Likewise.
18896 (register_edge_assert_for): Likewise.
18897 (finish_register_edge_assert_for): New helper actually registering
18898 asserts where live on edge.
18899 (find_conditional_asserts): Adjust.
18900 (find_switch_asserts): Likewise.
18901 (evrp_dom_walker::try_find_new_range): Generalize.
18902 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
18904 2017-04-27 Marek Polacek <polacek@redhat.com>
18907 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
18908 arg10 and arg11 to itype.
18910 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
18912 * doc/extend.texi (Object Size Checking): Improve grammar.
18914 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
18917 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
18918 that the logic for permitting reciprocal estimates matches that
18921 2017-04-27 Jakub Jelinek <jakub@redhat.com>
18924 * tree.c (type_cache_hasher::equal): Only compare
18925 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
18926 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
18927 non-aggregate element types.
18928 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
18929 about the flag on ARRAY_TYPEs in the comment, formatting fix.
18931 2017-04-27 Richard Biener <rguenther@suse.de>
18933 PR middle-end/80533
18934 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
18935 stripping ARRAY_REFs from MEM_EXPR make sure we're not
18936 keeping a reference to a trailing array.
18938 2017-04-27 Richard Biener <rguenther@suse.de>
18940 PR middle-end/80539
18941 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
18942 being in loop-closed SSA form conservatively.
18943 (chrec_fold_multiply_poly_poly): Likewise.
18945 2017-04-27 Tamar Christina <tamar.christina@arm.com>
18947 PR middle-end/79665
18948 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
18949 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
18951 2017-04-27 Jakub Jelinek <jakub@redhat.com>
18954 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
18955 (aarch64_function_arg_alignment): Return unsigned int again, but still
18956 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
18957 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
18958 Don't emit -Wpsabi note.
18959 (aarch64_function_arg_boundary): Likewise.
18960 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
18963 2017-04-26 Nathan Sidwell <nathan@acm.org>
18965 * tree.h (crc32_unsigned_n): Declare.
18966 (crc32_unsigned, crc32_unsigned): Make inline.
18967 * tree.c (crc32_unsigned_bits): Replace with ...
18968 (crc32_unsigned_n): ... this.
18969 (crc32_unsigned, crc32_byte): Remove.
18970 (crc32_string): Remove unnecessary braces.
18972 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
18974 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
18975 * ipa-inline-analysis.c (MAX_TIME): Remove.
18976 (account_size_time): Use sreal for time.
18977 (dump_inline_summary): Update.
18978 (estimate_function_body_sizes): Update.
18979 (estimate_edge_size_and_time): Update.
18980 (estimate_calls_size_and_time): Update.
18981 (estimate_node_size_and_time): Update.
18982 (inline_merge_summary): Update.
18983 (inline_update_overall_summary): Update.
18984 (estimate_time_after_inlining): Update.
18985 (inline_read_section): Update.
18986 (inline_write_summary): Update.
18987 * ipa-inline.c (compute_uninlined_call_time): Update.
18988 (compute_inlined_call_time): Update.
18989 (recursive_inlining): Update.
18990 (inline_small_functions): Update.
18991 (dump_overall_stats): Update.
18992 * ipa-inline.h: Include sreal.h.
18993 (size_time_entry): Turn time to sreal.
18994 (inline_summary): Turn self_time nad time to sreal.
18996 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
18998 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
19000 (sreal::stream_out, sreal::stream_in): New.
19001 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
19003 2017-04-25 Jakub Jelinek <jakub@redhat.com>
19005 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
19008 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
19011 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
19013 (dimode_scalar_chain::compute_convert_gain): Ditto.
19014 (dimode_scalar_chain::make_vector_copies): Ditto.
19015 (dimode_scalar_chain::convert_reg): Ditto.
19016 (dimode_scalar_chain::convert_insn): Ditto.
19017 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
19018 (VI248_AVX512BW_1): New mode iterator.
19019 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
19020 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
19023 2017-04-25 Martin Sebor <msebor@redhat.com>
19025 PR tree-optimization/80497
19026 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
19027 constants are representable in HOST_WIDE_INT.
19028 (parse_directive): Ditto.
19030 2017-04-25 Martin Sebor <msebor@redhat.com>
19033 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
19034 (new_zero_array): Adjust signature.
19035 (dom_info::dom_init): Used unsigned rather that size_t.
19036 (dom_info::dom_info): Same.
19038 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19039 Jakub Jelinek <jakub@redhat.com>
19042 * config/arm/arm.c: Include gimple.h.
19043 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
19044 returns negative, increment ncrn only if it returned positive.
19045 (arm_needs_doubleword_align): Return int instead of bool,
19046 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
19047 members, but if there is any such non-FIELD_DECL
19048 > PARM_BOUNDARY aligned decl, return -1 instead of false.
19049 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
19050 returns negative, increment nregs only if it returned positive.
19051 (arm_setup_incoming_varargs): Likewise.
19052 (arm_function_arg_boundary): Emit -Wpsabi note if
19053 arm_needs_doubleword_align returns negative, return
19054 DOUBLEWORD_ALIGNMENT only if it returned positive.
19056 2017-04-25 Marek Polacek <polacek@redhat.com>
19059 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
19060 first argument to type.
19062 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
19065 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
19066 type checks to test for compatibility instead of equality.
19068 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19069 Jakub Jelinek <jakub@redhat.com>
19072 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
19074 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
19075 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
19076 the alignment computation, but return their maximum in warn_alignment.
19077 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
19078 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
19080 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
19081 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
19084 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19086 * config/arc/simdext.md (dmpyh): Fix typo.
19088 2017-04-25 Richard Biener <rguenther@suse.de>
19090 PR tree-optimization/80492
19091 * alias.c (compare_base_decls): Handle registers with asm
19092 specification conservatively.
19093 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
19094 compare_base_decls returning dont-know properly.
19096 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19098 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
19099 (legitimate_offset_address_p): New function.
19100 (arc_legitimate_address_p): Use above function.
19102 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19104 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
19106 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19108 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
19109 ACCH registers whenever they are available.
19111 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19113 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
19114 double regs fix when not used.
19116 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19118 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
19120 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
19121 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
19123 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19125 * config/arc/arc.c (arc_output_addsi): Check for h-register class
19126 when emitting short ADD instructions.
19128 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19130 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
19132 (cmpsi_cc_c_insn): Likewise.
19133 (cbranchsi4_scratch): Compute proper instruction length using
19134 compact_hreg_operand.
19135 * config/arc/predicates.md (compact_hreg_operand): New predicate.
19137 2017-04-25 Richard Biener <rguenther@suse.de>
19139 PR middle-end/80509
19140 * passes.c (pass_manager::pass_manager): Initialize
19141 m_name_to_pass_map.
19143 2017-04-25 Richard Biener <rguenther@suse.de>
19145 PR tree-optimization/79201
19146 * tree-ssa-sink.c (statement_sink_location): Handle calls.
19148 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19151 * config/s390/vector.md: Split MEM->GPR vector moves for
19152 non-s_operand addresses.
19154 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19157 * config/s390/predicates.md (reload_const_wide_int_operand): New
19159 * config/s390/s390.md ("movti"): Remove d/P alternative.
19160 ("movti_bigconst"): New pattern definition.
19162 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
19165 * s390-protos.h (s390_expand_cs_hqi): Removed.
19166 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
19167 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
19168 modes as well as CCZ1mode and CCZmode.
19169 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
19170 signature of s390_emit_compare_and_swap.
19171 (s390_expand_cs_hqi): Likewise, make static.
19172 (s390_expand_cs_tdsi): Generate an explicit compare before trying
19173 compare-and-swap, in some cases.
19174 (s390_expand_cs): Wrapper function.
19175 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
19177 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
19178 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
19179 patterns for small and large integers. Forbid symref memory operands.
19180 Move expander to s390.c. Require cc register.
19181 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
19182 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
19183 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
19184 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
19185 symref memory operands. Remove CC mode and call s390_match_ccmode
19187 ("atomic_exchange<mode>"): Allow and implement all integer modes.
19189 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
19191 * config/s390/s390.md (define_peephole2): New peephole to help
19192 combining the load-and-test pattern with volatile memory.
19194 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
19196 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
19197 with CCZmode for TARGET_Z196.
19199 2017-04-25 Jakub Jelinek <jakub@redhat.com>
19201 PR rtl-optimization/80501
19202 * combine.c (make_compound_operation_int): Set subreg_code to SET
19203 even for AND with mask of the sign bit of mode.
19205 PR rtl-optimization/80500
19206 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
19207 sum's initial value.
19209 2017-04-25 Julian Brown <julian@codesourcery.com>
19210 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19212 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
19214 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
19216 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
19218 2017-04-25 Julian Brown <julian@codesourcery.com>
19219 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19221 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
19222 (thunderx2t99_sha): New Reservation.
19224 2017-04-25 Julian Brown <julian@codesourcery.com>
19225 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19227 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
19228 type for 1-element load.
19230 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
19232 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
19234 2017-04-24 Martin Jambor <mjambor@suse.cz>
19236 PR tree-optimization/80293
19237 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
19238 char arrays not totally scalarizable if it is false.
19239 (analyze_all_variable_accesses): Pass correct value in the new
19240 parameter. Add a statistics counter.
19242 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
19244 PR middle-end/79931
19245 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
19247 2017-04-24 Richard Biener <rguenther@suse.de>
19249 PR tree-optimization/80494
19250 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
19251 out for complex types.
19253 2017-04-24 Richard Biener <rguenther@suse.de>
19255 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
19256 * tree-ssa-sccvn.c (print_scc): Print SCC size.
19257 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
19258 (DFS): Adjust and never fail.
19259 (sccvn_dom_walker::fail): Remove.
19260 (sccvn_dom_walker::before_dom_children): Adjust.
19261 (run_scc_vn): Likewise and never fail.
19262 * tree-ssa-pre.c (pass_pre::execute): Adjust.
19263 (pass_fre::execute): Likewise.
19265 2017-04-24 Richard Biener <rguenther@suse.de>
19267 PR tree-optimization/79725
19268 * tree-ssa-sink.c (statement_sink_location): Return whether
19269 failure reason was zero uses. Move that check later.
19270 (sink_code_in_bb): Deal with zero uses by removing the stmt
19273 2017-04-24 Richard Biener <rguenther@suse.de>
19276 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
19277 pointer-based references.
19279 2017-04-24 Richard Biener <rguenther@suse.de>
19282 * pass_manager.h (pass_manager::operator new): Remove.
19283 (pass_manager::operator delete): Likewise.
19284 * passes.c (pass_manager::operator new): Remove.
19285 (pass_manager::operator delete): Likewise.
19286 (pass_manager::pass_manager): Zero individual pass members.
19288 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
19291 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
19292 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
19293 Check "XEXP (src, 1)" operand here.
19294 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
19295 Check "XEXP (src, 1)" operand here.
19296 (dimode_scalar_chain::make_vector_copies): Detect count register
19297 of a shift instruction. Zero extend count register from QImode
19298 to DImode to satisfy vector shift pattern count operand predicate.
19299 Substitute vector shift count operand with a DImode copy.
19300 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
19303 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
19305 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
19306 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
19307 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
19308 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
19309 (UNSPEC_NOREX_MEM): Remove definition.
19311 2017-04-21 Richard Biener <rguenther@suse.de>
19313 PR tree-optimization/79547
19314 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
19315 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
19316 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
19317 without any constraints.
19319 2017-04-21 Richard Biener <rguenther@suse.de>
19321 PR tree-optimization/78847
19322 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
19324 2017-04-21 Richard Biener <rguenther@suse.de>
19326 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
19327 (build_distinct_type_copy): Likewise.
19328 (build_variant_type_copy): Likewise.
19329 * tree.c (build_qualified_type): Pass down mem-stat info.
19330 (build_distinct_type_copy): Likewise.
19331 (build_variant_type_copy): Likewise.
19333 2017-04-21 Richard Biener <rguenther@suse.de>
19335 PR tree-optimization/80237
19336 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
19338 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
19339 for a simplified result.
19341 2016-04-21 Richard Biener <rguenther@suse.de>
19343 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
19344 sth as strict as a simple_iv but a chrec without symbols and an
19345 operand defined in the loop we are peeling (and not some subloop).
19346 (propagate_constants_for_unrolling): Propagate all constants.
19348 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
19351 * config/i386/i386.c (print_reg): Remove assert for disalowed
19352 regno values, call output_operand_lossage instead.
19354 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
19357 * config/i386/constraints.md (Yc): New register constraint.
19358 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
19359 Use Yc constraint for alternative 2 of operand 0. Remove
19360 preferred_for_speed attribute.
19362 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
19364 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
19365 lastprivate clauses in SIMT case.
19367 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
19369 * doc/invoke.texi (-Wextra-semi): Document new warning option.
19371 2017-04-20 Richard Biener <rguenther@suse.de>
19373 PR tree-optimization/57796
19374 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
19375 as N scalar stores.
19376 (vect_model_load_cost): Cost gathers as N scalar loads.
19378 2017-04-20 Richard Biener <rguenther@suse.de>
19380 * ggc-page.c (ggc_allocated_p): Rename to ...
19381 (safe_lookup_page_table_entry): ... this and return the lookup
19383 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
19385 2017-04-20 Richard Biener <rguenther@suse.de>
19387 PR tree-optimization/80453
19388 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
19389 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
19390 from the conditions.
19391 (vn_phi_eq): Pass them down.
19392 (vn_phi_lookup): Record them.
19393 (vn_phi_insert): Likewise.
19395 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
19397 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
19398 uninitialized variable warning to avoid buffer overrun.
19400 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
19403 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
19404 is suppressed for '{ 0 }' in C.
19406 2017-04-20 Jakub Jelinek <jakub@redhat.com>
19408 * BASE-VER: Set to 8.0.0.
19410 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
19412 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
19413 priority .init_array and .fini_array section with SECTION_NOTYPE
19416 2017-04-20 Jakub Jelinek <jakub@redhat.com>
19418 PR middle-end/80423
19419 * tree.h (build_array_type): Add typeless_storage default argument.
19420 * tree.c (type_cache_hasher::equal): Also compare
19421 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
19422 (build_array_type): Add typeless_storage argument, set
19423 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
19425 (build_nonshared_array_type): Adjust build_array_type_1 caller.
19426 (build_array_type): Likewise. Add typeless_storage argument.
19428 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
19429 Jakub Jelinek <jakub@redhat.com>
19431 PR tree-optimization/80426
19432 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
19433 operation on symbolic operands, also compute the overflow for the
19434 invariant part when the operation degenerates into a negation.
19436 2017-04-19 Jakub Jelinek <jakub@redhat.com>
19439 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
19440 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
19443 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
19445 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
19448 * config/avr/avr.c (tree.h): Include it.
19449 (cgraph.h): Include it.
19450 (avr_encode_section_info): Don't warn for uninitialized progmem
19451 variable if it's just an alias.
19453 2017-04-19 Richard Biener <rguenther@suse.de>
19456 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
19457 when needed by AutoPGO.
19459 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
19462 * doc/lto.texi: Remove an extra 'that'.
19464 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
19466 PR rtl-optimization/80429
19467 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
19468 are only used in debug insns.
19470 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
19471 Vladimir Makarov <vmakarov@redhat.com>
19473 * config/sparc/predicates.md (input_operand): Add comment. Return
19474 true for any memory operand when LRA is in progress.
19475 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
19477 2017-04-18 Jeff Law <law@redhat.com>
19480 * mips.md ({return,simple_return}_internal): Do not overwrite
19483 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19485 PR tree-optimization/80443
19486 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
19487 instead of adding 1, subtract -1 and similarly instead of subtracting
19490 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
19492 PR rtl-optimization/80357
19493 * haifa-sched.c (tmp_bitmap): New variable.
19494 (model_recompute): Handle duplicate use records.
19495 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
19496 (free_global_sched_pressure_data): Free it.
19498 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
19501 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
19502 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
19503 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
19504 instead of SYSTEM_HEADER_DIR.
19506 2017-04-18 Jeff Law <law@redhat.com>
19508 PR middle-end/80422
19509 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
19510 predecessors after walking up the insn chain.
19512 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19515 * dwarf2out.c (modified_type_die): Try harder not to emit internal
19516 sizetype type into debug info.
19518 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
19521 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
19522 unneeded test for TARGET_UPPER_REGS_SF.
19523 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
19525 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19528 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
19529 instead of gsi_after_labels.
19531 2017-04-18 Jeff Law <law@redhat.com>
19533 * regcprop.c (maybe_mode_change): Avoid creating copies of the
19537 2017-04-13 Jeff Law <law@redhat.com>
19538 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
19539 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
19541 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
19544 * config/avr/avr.c (intl.h): Include it.
19545 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
19547 2017-04-18 Martin Liska <mliska@suse.cz>
19549 PR gcov-profile/78783
19550 * gcov-tool.c (gcov_output_files): Validate that destination
19551 file is either removed by the tool or by a user.
19553 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
19554 Guy Benyei <guybe@mellanox.com>
19556 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
19557 block, and do not negate it, the stored id is already negative.
19559 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
19561 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
19563 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
19566 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
19567 masks of options that should be turned off if the VSX vector
19568 options are turned off.
19569 (OTHER_P8_VECTOR_MASKS): Likewise.
19570 (OTHER_VSX_VECTOR_MASKS): Likewise.
19571 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
19572 rs6000_disable_incompatible_switches to validate no type switches
19574 (rs6000_incompatible_switch): New function to disallow turning on
19575 other vector options if -mno-vsx, -mno-power8-vector, or
19576 -mno-power9-vector are specified.
19578 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19580 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
19582 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19584 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
19585 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
19586 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
19587 (ARG_POINTER_CFA_OFFSET): Likewise.
19589 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19591 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
19592 conditions to take advantage of various optimizations.
19594 2017-04-13 Jeff Law <law@redhat.com>
19596 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
19597 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
19598 (zero_extendsidi2_dext): Likewise.
19600 2017-04-13 Jakub Jelinek <jakub@redhat.com>
19603 * fold-const.c (fold_ternary_loc): Revert
19604 use op0 instead of fold_convert_loc (loc, type, arg0) part of
19607 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
19609 PR rtl-optimization/80343
19610 * lra-remat.c (update_scratch_ops): Assign original hard reg to
19611 new scratch pseudo.
19613 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
19616 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
19617 to ubsan_encode_value.
19619 2017-04-13 Jeff Law <law@redhat.com>
19621 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
19622 appearing in DEBUG_INSNs.
19624 2017-04-13 Martin Liska <mliska@suse.cz>
19626 PR gcov-profile/80413
19627 * gcov-io.c (gcov_write_string): Copy to buffer just when
19628 allocated size is greater than zero.
19630 2017-04-13 Jakub Jelinek <jakub@redhat.com>
19633 * dwarf2out.c (decls_for_scope): Ignore declarations of
19634 current_function_decl in BLOCK_NONLOCALIZED_VARS.
19636 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
19639 * ipa-visibility.c (non_local_p): Fix typos.
19640 (localize_node): When localizing symbol in same comdat group,
19641 dissolve the group only when we know external symbols are going
19643 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
19645 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19647 PR tree-optimization/79390
19648 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
19649 order does not result in usable sequence, retry with reversed operand
19655 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
19656 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
19657 op0 instead of fold_convert_loc (loc, type, arg0).
19659 2017-04-12 Jeff Law <law@redhat.com>
19661 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
19662 has a delay slot in the generated code.
19664 * config/cris/cris.md (cris_preferred_reload_class): Return
19665 GENNONACR_REGS rather than GENERAL_REGS.
19667 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19670 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
19671 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
19672 signedness of the result type.
19674 2017-04-12 Richard Biener <rguenther@suse.de>
19675 Jeff Law <law@redhat.com>
19677 PR tree-optimization/80359
19678 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
19679 trim stores to TARGET_MEM_REFs.
19681 2017-04-12 Richard Biener <rguenther@suse.de>
19683 PR tree-optimization/79390
19684 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
19685 threading case even more.
19687 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
19690 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
19691 for quad_address_p for TImode, instead of just not indexed_address.
19693 2017-04-12 Richard Biener <rguenther@suse.de>
19694 Bernd Edlinger <bernd.edlinger@hotmail.de>
19696 PR middle-end/79671
19697 * alias.c (component_uses_parent_alias_set_from): Handle
19698 TYPE_TYPELESS_STORAGE.
19699 (get_alias_set): Likewise.
19700 * tree-core.h (tree_type_common): Add typeless_storage flag.
19701 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
19702 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
19703 for types containing members with TYPE_TYPELESS_STORAGE.
19704 (place_field): Likewise.
19705 (layout_type): Likewise for ARRAY_TYPE.
19706 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
19707 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
19708 TYPE_TYPELESS_STORAGE.
19709 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
19711 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19714 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
19715 first argument to type.
19717 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19721 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
19722 CONST0_RTX (mode) rather than const0_rtx where appropriate.
19723 (rs6000_expand_binop_builtin): Likewise.
19724 (rs6000_expand_ternop_builtin): Likewise; also add missing
19725 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
19726 vshasigma built-ins.
19727 * doc/extend.texi: Document that vec_xxpermdi's third argument
19728 must be a constant.
19730 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
19732 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
19733 Use shift_const cost parameter when calculating gain of STV shifts.
19735 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
19737 PR rtl-optimization/70478
19738 * lra-constraints.c (process_alt_operands): Check memory for
19739 disfavoring memory insn operand.
19741 2017-04-11 Jakub Jelinek <jakub@redhat.com>
19743 PR middle-end/80100
19744 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
19745 left shift in unsigned HOST_WIDE_INT type.
19747 PR rtl-optimization/80385
19748 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
19749 (not (neg X)) into (plus X -1) for complex or non-integral modes.
19752 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
19753 if they have any depend clauses.
19755 2017-04-11 Martin Liska <mliska@suse.cz>
19758 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
19759 * ipa-split.c (split_function): Create a local comdat symbol
19760 if caller is in a comdat group.
19762 2017-04-11 Martin Liska <mliska@suse.cz>
19765 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
19768 2017-04-11 Martin Sebor <msebor@redhat.com>
19770 PR middle-end/80364
19771 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
19772 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
19774 (directive::set_width, directive::set_precision, format_character):
19776 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
19779 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
19782 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
19783 conflict, set target->arch_name instead of target->cpu_name.
19785 2017-04-11 Richard Biener <rguenther@suse.de>
19787 PR tree-optimization/80374
19788 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
19789 build_zero_cst, remove fold_convertible_p check again.
19791 2017-04-11 Martin Liska <mliska@suse.cz>
19794 * ubsan.c (instrument_object_size): Do not instrument register
19797 2017-04-11 Jakub Jelinek <jakub@redhat.com>
19800 * config/i386/i386-builtin-types.def
19801 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
19802 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
19803 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
19804 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
19805 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
19806 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
19807 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
19808 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
19809 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
19810 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
19811 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
19812 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
19813 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
19814 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
19815 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
19816 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
19817 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
19818 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
19819 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
19820 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
19821 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
19822 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
19823 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
19824 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
19825 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
19826 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
19827 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
19828 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
19829 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
19830 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
19831 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
19832 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
19833 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
19834 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
19835 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
19836 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
19837 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
19838 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
19839 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
19840 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
19841 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
19842 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
19843 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
19844 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
19845 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
19846 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
19848 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
19849 flag to second_arg_count, handle 4 argument function type _COUNT
19850 aliases, handle second_arg_count on second argument rather than last.
19852 2017-04-10 Jeff Law <law@redhat.com>
19854 PR tree-optimization/80374
19855 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
19856 record anything if we can not convert integer_zero_node to the
19859 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
19862 * config/rs6000/rs6000.c (rs6000_option_override_internal):
19863 Enhance special handling given to the TARGET_P9_MINMAX option in
19864 relation to certain other options.
19866 2017-04-10 Bin Cheng <bin.cheng@arm.com>
19868 PR tree-optimization/80153
19869 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
19870 remove POINTER_PLUS_EXPR's base part directly, rather than through
19873 2017-04-10 Richard Biener <rguenther@suse.de>
19874 Bin Cheng <bin.cheng@arm.com>
19876 PR tree-optimization/80153
19877 * tree-affine.c (aff_combination_to_tree): Get base pointer from
19878 the first element of pointer type aff_tree. Build result expr in
19880 (add_elt_to_tree): Convert to type unconditionally. Remove other
19881 fold_convert calls.
19882 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
19883 (rewrite_use_nonlinear_expr): Check invariant using iv information.
19885 2017-04-10 Richard Biener <rguenther@suse.de>
19887 * tree-ssa-structalias.c (find_func_aliases): Properly handle
19890 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
19892 PR rtl-optimization/70478
19893 * lra-constraints.c (curr_small_class_check): New.
19894 (update_and_check_small_class_inputs): New.
19895 (process_alt_operands): Update curr_small_class_check. Disfavor
19896 alternative insn memory operands. Check available regs for small
19899 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
19902 * config/mips/mips.opt (-mvirt): Update description.
19903 * doc/invoke.texi (-mvirt): Likewise.
19905 2017-04-10 Richard Biener <rguenther@suse.de>
19907 PR middle-end/80362
19908 * fold-const.c (fold_binary_loc): Look at unstripped ops when
19909 looking for NEGATE_EXPR in -A / -B to A / B folding.
19911 2017-04-10 Martin Liska <mliska@suse.cz>
19913 PR gcov-profile/80224
19914 * gcov.c (print_usage): Fix usage string.
19915 (get_gcov_intermediate_filename): Remove.
19916 (output_gcov_file): Use both for normal and intermediate format.
19917 (generate_results): Do not initialize special file for
19918 intermediate format.
19920 2017-04-10 Richard Biener <rguenther@suse.de>
19922 PR tree-optimization/80304
19923 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
19926 2017-04-10 Nathan Sidwell <nathan@acm.org>
19929 * config/rs6000/rs6000.c (rs6000_vector_type): New.
19930 (rs6000_init_builtins): Use it.
19932 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19934 * config/arm/arm.md (<mrc>): Add mode to SET source.
19935 (<mrrc>): Likewise.
19937 2017-04-10 Richard Biener <rguenther@suse.de>
19939 PR middle-end/80344
19940 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
19942 2017-04-10 Jakub Jelinek <jakub@redhat.com>
19945 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
19946 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
19947 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
19948 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
19949 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
19950 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
19951 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
19952 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
19953 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
19954 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
19955 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
19956 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
19957 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
19958 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
19959 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
19960 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
19961 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
19962 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
19963 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
19964 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
19965 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
19966 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
19967 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
19969 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
19971 PR rtl-optimization/70478
19972 * lra-constraints.c: Reverse the last patch.
19974 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
19976 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
19977 Add comment for WCHAR_T.
19979 2017-04-08 Martin Liska <mliska@suse.cz>
19982 2017-04-07 Martin Liska <mliska@suse.cz>
19985 * ipa-split.c (split_function): Add function part to a same comdat
19988 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19991 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
19993 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
19995 * rs6000/rs6000.c (vec_load_pendulum): Rename...
19996 (vec_pairing): ...to this.
19997 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
19998 (rs6000_sched_init): Adjust for name change.
19999 (struct rs6000_sched_context): Likewise.
20000 (rs6000_init_sched_context): Likewise.
20001 (rs6000_set_sched_context): Likewise.
20003 2017-04-07 Jakub Jelinek <jakub@redhat.com>
20009 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
20011 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
20012 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
20013 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
20015 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
20017 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
20019 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
20021 PR rtl-optimization/70703
20022 * ira-color.c (update_conflict_hard_regno_costs): Use
20023 int64_t instead of HOST_WIDE_INT.
20025 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
20027 PR rtl-optimization/70478
20028 * lra-constraints.c (process_alt_operands): Disfavor alternative
20029 insn memory operands.
20031 2017-04-07 Jeff Law <law@redhat.com>
20033 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
20034 CALL and NOTE_INSN_CALL_ARG_LOCATION.
20036 2017-04-07 Martin Liska <mliska@suse.cz>
20039 * config/aarch64/aarch64.c (aarch64_process_target_attr):
20040 Show error message instead of an ICE.
20042 2017-04-07 Martin Liska <mliska@suse.cz>
20045 * ipa-split.c (split_function): Add function part to a same comdat
20048 2017-04-07 Richard Biener <rguenther@suse.de>
20050 PR middle-end/80341
20051 * tree.c (get_unwidened): Also handle ! for_type case for
20053 * convert.c (do_narrow): Split out from ...
20054 (convert_to_integer_1): ... here. Do not pass final truncation
20055 type to get_unwidened for TRUNC_DIV_EXPR.
20057 2017-04-07 Richard Biener <rguenther@suse.de>
20059 * tree-affine.c (wide_int_ext_for_comb): Take type rather
20061 (aff_combination_const): Adjust.
20062 (aff_combination_scale): Likewise.
20063 (aff_combination_add_elt): Likewise.
20064 (aff_combination_add_cst): Likewise.
20065 (aff_combination_convert): Likewise.
20066 (add_elt_to_tree): Likewise. Remove unused argument.
20067 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
20069 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
20071 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
20073 * config/arm/arm.c (arm_default_short_enums): Use
20074 ARM_DEFAULT_SHORT_ENUMS.
20075 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
20077 2017-04-06 Jakub Jelinek <jakub@redhat.com>
20080 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
20081 members with redundant out-of-class redeclaration.
20083 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
20086 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
20087 * config/i386/i386.md (*zero_extendsidi2):
20088 Add (?*x,*x) and (?*v,*v) alternatives.
20090 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
20093 * config/i386/i386.c (ix86_expand_builtin)
20094 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
20095 mode from insn data. Convert operands to insn operand mode.
20096 Copy operands that don't satisfy insn predicate to a register.
20098 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
20100 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
20103 2017-04-06 Richard Biener <rguenther@suse.de>
20105 PR tree-optimization/80334
20106 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
20107 preserve alignment of accesses.
20109 2017-04-06 Richard Biener <rguenther@suse.de>
20111 PR tree-optimization/80262
20112 * tree-sra.c (build_ref_for_offset): Preserve address-space
20114 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
20115 Drop useless address-space information on MEM_REF offsets.
20117 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
20119 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
20121 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
20123 PR rtl-optimization/70703
20124 * ira-color.c (update_conflict_hard_regno_costs): Use
20125 HOST_WIDE_INT instead of long.
20127 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
20130 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
20131 not defined for x86_64 target. Add -mmmx target option when __SSE2__
20133 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
20134 for x86_64 target. Handle -m3dnowa option.
20136 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
20138 PR rtl-optimization/70703
20139 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
20140 (update_conflict_hard_regno_costs): Use long instead of unsigned
20141 arithmetic for cost calculation.
20143 2017-04-05 Jakub Jelinek <jakub@redhat.com>
20144 Bernd Edlinger <bernd.edlinger@hotmail.de>
20147 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
20150 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
20153 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
20154 ptr_mode with Pmode throughout.
20155 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
20156 into probe_stack_range and use DImode.
20158 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
20161 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
20162 call_eh_return is true.
20164 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20166 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
20167 Initialize last_match_fntype_index.
20169 2017-04-05 Jakub Jelinek <jakub@redhat.com>
20172 * tree-nvr.c: Include internal-fn.h.
20173 (pass_return_slot::execute): Ignore internal calls without
20176 2017-04-04 Jakub Jelinek <jakub@redhat.com>
20177 Richard Biener <rguenther@suse.de>
20180 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
20181 captures used multiple times, except for the last use.
20182 * generic-match-head.c: Include gimplify.h.
20184 2017-04-04 Jakub Jelinek <jakub@redhat.com>
20186 PR tree-optimization/79390
20187 * target.h (struct noce_if_info): Declare.
20188 * targhooks.h (default_noce_conversion_profitable_p): Declare.
20189 * target.def (noce_conversion_profitable_p): New target hook.
20190 * ifcvt.h (struct noce_if_info): New type, moved from ...
20191 * ifcvt.c (struct noce_if_info): ... here.
20192 (noce_conversion_profitable_p): Renamed to ...
20193 (default_noce_conversion_profitable_p): ... this. No longer
20195 (noce_try_store_flag_constants, noce_try_addcc,
20196 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
20197 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
20198 instead of noce_conversion_profitable_p.
20199 * config/i386/i386.c: Include ifcvt.h.
20200 (ix86_option_override_internal): Don't override
20201 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
20202 (ix86_noce_conversion_profitable_p): New function.
20203 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
20204 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
20205 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
20206 * doc/tm.texi: Regenerated.
20208 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20210 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
20213 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20216 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
20217 instructions for small multiply cores.
20219 2017-04-04 Jeff Law <law@redhat.com>
20221 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
20223 (mips_expand_vec_perm_const): Initialize elements in orig_perm
20224 that are not set by the loop over the elements.
20226 2017-04-04 Jakub Jelinek <jakub@redhat.com>
20229 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
20230 int mode, convert_modes it to mode as unsigned, otherwise use
20231 lowpart_subreg to mode rather than SImode.
20232 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
20233 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
20234 Use DImode instead of SImode for the shift count operand.
20235 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
20238 2017-04-04 Richard Biener <rguenther@suse.de>
20240 PR middle-end/80281
20241 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
20242 arithmetic done for the negate or the plus. Simplify.
20243 (A - (-B) -> A + B): Likewise.
20244 * fold-const.c (split_tree): Make sure to not negate pointers.
20246 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
20248 PR rtl-optimization/60818
20249 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
20250 a compare of comparisons with the thing compared if this results
20251 in a different machine mode.
20253 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
20255 * alias.c (base_alias_check): Fix typo in comment.
20256 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
20257 * cgraphunit.c (symbol_table::compile): Likewise.
20258 * collect2.c (maybe_run_lto_and_relink): Likewise.
20259 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
20260 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
20261 * config/avr/avr.c (avr_map_op_t): Likewise.
20262 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
20263 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
20264 * config/epiphany/epiphany.md (movcc): Likewise.
20265 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
20266 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
20268 * config/mips/mips.c (mips_save_restore_reg): Likewise.
20269 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
20270 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
20271 * config/sh/sh.c (sh_rtx_costs): Likewise.
20272 * fold-const.c (fold_truth_andor): Likewise.
20273 * genautomata.c (collapse_flag): Likewise.
20274 * gengtype.h (struct type::u::s): Likewise.
20275 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
20276 * input.c (FORMAT_AMOUNT): Likewise.
20277 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
20278 (known_aggs_to_agg_replacement_list): Likewise.
20279 * ipa-inline-analysis.c: Likewise.
20280 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
20281 * ipa-polymorphic-call.c
20282 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
20283 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
20284 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
20286 * modulo-sched.c (apply_reg_moves): Likewise.
20287 * omp-expand.c (build_omp_regions_1): Likewise.
20288 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
20289 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
20290 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
20291 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
20292 * value-prof.c: Likewise.
20293 * var-tracking.c (val_reset): Likewise.
20295 2017-04-03 Richard Biener <rguenther@suse.de>
20297 PR tree-optimization/80275
20298 * fold-const.c (split_address_to_core_and_offset): Handle
20301 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
20303 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
20304 descriptors is at least equal to that of functions.
20306 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
20308 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
20310 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
20313 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
20314 (mov<IMOD4:mode>): New expander.
20315 (*mov<IMOD4:mode>_internal): New insn and split pattern.
20317 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
20319 PR rtl-optimization/79405
20320 * fwprop.c (propagations_left): New variable.
20321 (forward_propagate_into): Decrement it.
20322 (fwprop_init): Initialize it.
20323 (fw_prop): If the variable has reached zero, stop propagating.
20324 (fwprop_addr): Ditto.
20326 2017-03-31 Jakub Jelinek <jakub@redhat.com>
20329 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
20330 a FUNCTION_DECL, pass it as decl instead of origin to
20333 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
20335 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
20338 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
20341 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
20342 TARGET_VSX_SMALL_INTEGER.
20344 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20346 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20347 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
20349 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
20351 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
20352 extraction from odd-numbered MSA register.
20354 2017-03-31 Jakub Jelinek <jakub@redhat.com>
20356 PR middle-end/80173
20357 * expmed.c (store_bit_field_1): Don't attempt to create
20358 a word subreg out of hard registers wider than word if they
20359 have HARD_REGNO_NREGS of 1 for their mode.
20361 PR middle-end/80163
20362 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
20363 conversions to integer types wider than word and pointer.
20366 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
20367 (rtx_equal_for_cselib_p): Pass 0 to it.
20368 * cselib.c (cselib_hasher::equal): Likewise.
20369 (rtx_equal_for_cselib_1): Add depth argument. If depth
20370 is 128, don't look up VALUE locs and punt. Increment
20371 depth in recursive calls when walking VALUE locs.
20373 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
20375 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
20376 (make_gcov_file_name): Use the canonical path name for generating
20378 (read_line): Fix handling of files with ascii null bytes.
20380 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
20382 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
20383 to initialise a vector register instead
20384 of using a const_int.
20386 2017-03-30 Jakub Jelinek <jakub@redhat.com>
20388 PR translation/80189
20389 * gimplify.c (omp_default_clause): Use %qs instead of %s in
20390 diagnostic messages.
20392 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
20395 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
20396 (dfp_diex_<mode>): Update mode of operand 1.
20397 * doc/extend.texi (dxex, dxexq): Document change to return type.
20398 (diex, diexq): Document change to argument type.
20400 2017-03-30 Martin Jambor <mjambor@suse.cz>
20403 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
20404 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
20405 it reflects the signature changes performed at the callee side.
20406 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
20407 to cgraph_build_function_type_skip_args.
20408 (build_function_decl_skip_args): Adjust call to the above function.
20410 2017-03-30 Jakub Jelinek <jakub@redhat.com>
20413 * config/i386/sse.md
20414 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
20415 register as dest whenever it is a MEM not rtx_equal_p to the
20416 corresponding dup operand, and when forcing into reg move the
20417 reg into the memory afterwards.
20418 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
20419 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
20420 for the force_reg mode.
20421 (avx512vl_vextractf128<mode>): Use register as dest either
20422 always when a MEM, or when it is a MEM not rtx_equal_p to the
20423 corresponding dup operand, or even not when it is a CONST_VECTOR
20424 depending on the mode and lo vs. hi.
20425 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
20427 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
20428 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
20429 Likewise. Require that operands[2] is even.
20430 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
20431 Remove extraneous parens. Require that operands[2] is a multiple
20433 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
20434 operands[0] is a MEM if <mask_applied>, the predicates/constraints
20435 disallow memory then.
20437 2017-03-30 Richard Biener <rguenther@suse.de>
20439 PR tree-optimization/77498
20440 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
20441 to non-constants over backedges.
20443 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
20445 PR rtl-optimization/80233
20446 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
20447 as last_combined_insn. Do not test for BARRIER_P separately.
20449 2017-03-29 Andreas Schwab <schwab@suse.de>
20452 * calls.c (prepare_call_address): Convert funexp to Pmode before
20453 copying to temp reg.
20455 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20457 PR tree-optimization/80158
20458 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
20459 Handle possible future case of more than one alternate
20461 (replace_rhs_if_not_dup): Likewise.
20462 (replace_one_candidate): Likewise.
20464 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
20466 PR rtl-optimization/80193
20467 * ira.c (ira): Do not check allocation for LRA.
20469 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
20471 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
20472 (nvptx_output_simt_exit): Declare.
20473 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
20474 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
20475 (init_softstack_frame): Move initialization of crtl->is_leaf to...
20476 (nvptx_declare_function_name): ...here. Emit declaration of local
20477 memory space buffer for omp_simt_enter insn.
20478 (nvptx_output_unisimt_switch): New.
20479 (nvptx_output_softstack_switch): New.
20480 (nvptx_output_simt_enter): New.
20481 (nvptx_output_simt_exit): New.
20482 * config/nvptx/nvptx.h (struct machine_function): New fields
20483 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
20484 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
20485 (UNSPECV_SIMT_EXIT): Ditto.
20486 (omp_simt_enter_insn): New insn.
20487 (omp_simt_enter): New expansion.
20488 (omp_simt_exit): New insn.
20489 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
20491 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
20492 (expand_GOMP_SIMT_ENTER_ALLOC): New.
20493 (expand_GOMP_SIMT_EXIT): New.
20494 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
20495 (GOMP_SIMT_ENTER_ALLOC): Ditto.
20496 (GOMP_SIMT_EXIT): Ditto.
20497 * target-insns.def (omp_simt_enter): New insn.
20498 (omp_simt_exit): Ditto.
20499 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
20501 (lower_rec_simd_input_clauses): Implement SIMT privatization.
20502 (lower_rec_input_clauses): Likewise.
20503 (lower_lastprivate_clauses): Handle SIMT privatization.
20505 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
20506 (ompdevlow_adjust_simt_enter): New.
20507 (find_simtpriv_var_op): New.
20508 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
20509 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
20511 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
20512 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
20513 (copy_decl_for_dup_finish): Ditto.
20515 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
20517 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
20520 * config/i386/i386.c (ix86_option_override_internal): Always
20521 allow -mpreferred-stack-boundary=3 for 64-bit targets.
20523 2017-03-28 Bin Cheng <bin.cheng@arm.com>
20525 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
20527 2017-03-28 Bin Cheng <bin.cheng@arm.com>
20529 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
20530 mark new edge's irreducible flag accordign to it.
20531 (vect_do_peeling): Check loop preheader edge's irreducible flag
20532 and pass it to function slpeel_add_loop_guard.
20534 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
20536 PR tree-optimization/80218
20537 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
20538 Update block frequencies and counts.
20540 2017-03-28 Richard Biener <rguenther@suse.de>
20542 PR tree-optimization/78644
20543 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
20544 of a simplification result we may not use it at all.
20546 2017-03-28 Richard Biener <rguenther@suse.de>
20549 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
20550 without arguments, generate default definition of a SSA name.
20552 2017-03-28 Richard Biener <rguenther@suse.de>
20554 PR middle-end/80222
20555 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
20556 TYPE_REF_CAN_ALIAS_ALL references.
20557 * fold-const.c (fold_indirect_ref_1): Likewise.
20559 2017-03-28 Martin Liska <mliska@suse.cz>
20562 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
20563 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
20565 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
20566 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
20568 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
20569 (EXTRA_SPECS): Define.
20570 (SUBTARGET_EXTRA_SPECS): Likewise.
20571 (SUBTARGET_CPP_SPEC): Likewise.
20572 * config/arc/elf.h (EXTRA_SPECS): Renamed to
20573 SUBTARGET_EXTRA_SPECS.
20574 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
20576 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
20578 * config/arc/simdext.md (vst64_insn): Update pattern.
20579 (vld32wh_insn): Likewise.
20580 (vld32wl_insn): Likewise.
20581 (vld64_insn): Likewise.
20582 (vld32_insn): Likewise.
20584 2017-03-28 Marek Polacek <polacek@redhat.com>
20587 * fold-const.c (fold_comparison): Use protected_set_expr_location
20588 instead of SET_EXPR_LOCATION.
20590 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
20592 * tree.c (add_expr): Avoid name lookup warning.
20594 2017-03-27 Jeff Law <law@redhat.com>
20596 PR tree-optimization/80216
20597 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
20598 function name. Limit recursion depth.
20599 (record_temporary_equivalences): Corresponding changes.
20601 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
20603 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
20606 2017-03-27 Jakub Jelinek <jakub@redhat.com>
20609 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
20611 * cfgcleanup.c (reg_note_cfa_p): New array.
20612 (insns_have_identical_cfa_notes): New function.
20613 (old_insns_match_p): Don't cross-jump in between /f
20614 and non-/f instructions. If both i1 and i2 are frame related,
20615 verify all CFA notes, their order and content.
20617 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
20620 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
20621 HImode and SImode with zero extend to DImode to one insn.
20622 (bswap<mode>2_extenddi): Likewise.
20623 (bswapsi2_extenddi): Likewise.
20624 (bswaphi2_extendsi): Likewise.
20625 (bswaphi2): Combine bswap HImode and SImode into one insn.
20626 Separate memory insns from swapping register.
20627 (bswapsi2): Likewise.
20628 (bswap<mode>2): Likewise.
20629 (bswaphi2_internal): Delete, no longer used.
20630 (bswapsi2_internal): Likewise.
20631 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
20632 store, and gpr<-gpr swap insns.
20633 (bswap<mode>2_store): Likewise.
20634 (bswaphi2_reg): Register only splitter, combine with the splitter.
20635 (bswaphi2 splitter): Likewise.
20636 (bswapsi2_reg): Likewise.
20637 (bswapsi2 splitter): Likewise.
20638 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
20639 the insns into load, store, and register/register insns.
20640 (bswapdi2_ldbrx): Likewise.
20641 (bswapdi2_load): Likewise.
20642 (bswapdi2_store): Likewise.
20643 (bswapdi2_reg): Likewise.
20645 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
20647 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
20648 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
20650 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
20653 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
20655 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
20656 special handling for target option conflicts between dform
20657 options (-mpower9-dform, -mpower9-dform-vector,
20658 -mpower9-dform-scalar) and -mno-direct-move.
20660 2017-03-27 Richard Biener <rguenther@suse.de>
20662 PR tree-optimization/80181
20663 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
20665 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20667 * config/arc/predicates.md (move_double_src_operand): Replace the
20668 call to move_double_src_operand with a call to address_operand.
20670 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20672 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
20673 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
20674 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
20676 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20678 * config/arc/predicates.md (long_immediate_loadstore_operand):
20679 Consider scaled addresses cases.
20681 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20683 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
20684 restored when in interrupt.
20685 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
20686 doesn't have delay slot.
20688 2017-03-27 Richard Biener <rguenther@suse.de>
20691 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
20692 inlined thunk clones.
20694 2017-03-27 Jakub Jelinek <jakub@redhat.com>
20697 * asan.c (instrument_derefs): Copy over last operand from
20698 original COMPONENT_REF to the new COMPONENT_REF with
20699 DECL_BIT_FIELD_REPRESENTATIVE.
20700 * ubsan.c (instrument_object_size): Likewise.
20702 2017-03-27 Richard Biener <rguenther@suse.de>
20704 PR tree-optimization/80170
20705 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
20706 sure DR/SCEV didnt fold in constants we do not see when looking
20707 at the reference base alignment.
20709 2017-03-27 Richard Biener <rguenther@suse.de>
20711 PR middle-end/80171
20712 * gimple-fold.c (fold_ctor_reference): Properly guard against
20713 NULL return value from canonicalize_constructor_val.
20715 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
20718 * config/i386/i386.c (ix86_expand_builtin)
20719 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
20720 flags reg setting and flags reg using instructions.
20721 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
20722 clobbering instructions to zero extend op2.
20724 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
20726 * doc/install.texi (Configuration) <--with-aix-soname>:
20727 Update link to AIX ld.
20729 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
20731 PR rtl-optimization/80160
20732 PR rtl-optimization/80159
20733 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
20734 reg_alternate_class into account.
20736 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
20739 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
20740 to consider in curr_insn_transform.
20742 2017-03-24 Jakub Jelinek <jakub@redhat.com>
20744 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
20745 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
20746 and emit_mode_inner.
20748 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20750 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
20751 argument to the overloaded builtin variants. Use the new flag to
20752 deprecate certain builtin variants.
20753 * config/s390/s390-builtin-types.def: Add new builtin types.
20754 * config/s390/s390-builtins.h: Support new flags field for
20755 overloaded builtins.
20756 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
20757 (s390_macro_to_expand): Enable vector float data type.
20758 (s390_cpu_cpp_builtins_internal): Indicate support of the new
20759 builtins by incrementing the __VEC__ version number.
20760 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
20762 (s390_resolve_overloaded_builtin): Emit error messages depending
20763 on the builtin flags.
20764 * config/s390/s390.c (s390_expand_builtin): Support additional
20765 flags argument. Change error message to match the messages
20766 emitted in s390-c.c.
20767 * config/s390/s390.md: New UNSPEC_* constants.
20768 (op_type): Add new instruction types.
20769 * config/s390/vecintrin.h: Add new builtins and test data class
20771 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
20772 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
20773 (VEC_INEXACT, VEC_NOINEXACT): New constants.
20774 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
20775 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
20776 ("vec_mergel<mode>"): V_HW -> VEC_HW.
20778 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
20779 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
20780 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
20781 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
20783 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
20784 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
20785 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
20786 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
20788 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
20789 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
20790 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
20791 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
20792 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
20793 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
20794 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
20796 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
20797 ("vec_scatter_element<V_HW_4:mode>_DI")
20798 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
20799 ("vec_fpint<mode>", "vflls")
20800 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
20801 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
20802 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
20803 ("*vec_cmphe<mode>_cc"): ... these.
20805 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
20806 mode constant instead of magic value.
20808 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20810 * config/s390/s390.c (s390_expand_vec_compare): Support other
20811 vector floating point modes than just V2DF.
20812 (s390_expand_vcond): Likewise.
20813 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
20814 (s390_cannot_change_mode_class): Prevent mode changes between TF
20815 and V1TF in vector registers.
20816 * config/s390/s390.md (DF, SF): New mode attributes.
20817 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
20818 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
20819 SFmode support for VRs.
20820 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
20822 (VFT, VF_HW): New mode iterators.
20823 (vw, sdx): New mode attributes.
20824 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
20825 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
20826 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
20827 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
20828 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
20829 also the new vector floating point modes. Renaming to ...
20831 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
20832 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
20833 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
20834 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
20835 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
20836 ("vec_unordered<mode>"): ... these.
20838 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
20839 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
20840 ("*vec_extendv2df"): New insn definitions.
20842 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20844 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
20845 ("mulditi3_2", "*muldi3_sign"): New patterns.
20846 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
20847 rename the pattern definition.
20849 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20851 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
20853 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
20855 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20857 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
20858 instruction if possible.
20859 * config/s390/vector.md (vec_halfnumelts): New mode
20861 ("*vec_vllezlf<mode>"): New pattern.
20863 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20865 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
20866 ("popcountv4si2", "popcountv2di2"): Rename to ...
20867 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
20868 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
20870 ("popcount<mode>2_vxe"): New pattern.
20872 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20874 * common/config/s390/s390-common.c (processor_flags_table): Add
20876 * config.gcc: Add arch12.
20877 * config/s390/driver-native.c (s390_host_detect_local_cpu):
20878 Default to arch12 for unknown CPU model numbers.
20879 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
20880 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
20881 PROCESSOR_max sanity check.
20882 * config/s390/s390-opts.h (enum processor_type): Add
20884 * config/s390/s390.c (processor_table): Add arch12.
20885 (s390_expand_builtin): Add check for B_VXE flag.
20886 (s390_issue_rate): Add PROCESSOR_ARCH12.
20887 (s390_get_sched_attrmask): Likewise.
20888 (s390_get_unit_mask): Likewise.
20889 (s390_sched_score): Enable z13 scheduling for arch12.
20890 (s390_sched_reorder): Likewise.
20891 (s390_sched_variable_issue): Likewise.
20892 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
20894 (s390_tune_attr): Use z13 scheduling also for arch12.
20895 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
20896 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
20897 (TARGET_VXE_P): New macros.
20898 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
20899 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
20900 * config/s390/s390.opt: Add arch12 as processor_type.
20902 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20904 * config/s390/s390.md
20905 ("fixuns_truncdddi2", "fixuns_trunctddi2")
20906 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
20907 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
20909 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
20910 Rename expanders to ...
20912 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
20913 ("fixuns_truncdddi2_emu"): ... these.
20915 ("fixuns_trunc<mode>si2_emu"): New expander.
20917 ("*fixuns_truncdfdi2_z13"): Rename to ...
20918 ("*fixuns_truncdfdi2_vx"): ... this.
20920 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20922 * config/s390/2964.md: Remove the single element vector compare
20923 instructions which are no longer used.
20924 * config/s390/s390.c (s390_select_ccmode): Remove handling of
20926 (s390_canonicalize_comparison): Remove handling of DFmode
20928 (s390_expand_vec_compare_scalar): Remove function.
20929 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
20930 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
20932 ("*cmp<mode>_ccs"): Add wfcdb instruction.
20934 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20936 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
20938 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
20939 will anyway by matched by mov<mode>_64dfp.
20941 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20943 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
20944 vlef/vstef. Add missing operand to vleif.
20946 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20948 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
20949 pair for all vector types with 64 bit elements.
20950 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
20951 * config/s390/vector.md (V_HW_64): ... here.
20952 (V_128_NOSINGLE): New mode iterator.
20953 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
20954 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
20955 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
20956 ("*vec_load_pairv2di"): Change to ...
20957 ("*vec_load_pair<mode>"): ... this one.
20959 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20961 * config/s390/constraints.md: Add comments.
20962 (jKK): Reject element sizes > 8 bytes.
20963 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
20965 * config/s390/s390.md: Add the s_operand checks formerly in
20966 s390_split_ok_p to various splitters where they are still
20968 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
20969 for 128 bit vectors. Plus two splitters.
20971 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20973 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
20976 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20979 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
20980 error if the boundary argument is not constant.
20982 2017-03-24 Jakub Jelinek <jakub@redhat.com>
20984 PR rtl-optimization/80112
20985 * loop-doloop.c (doloop_condition_get): Don't check condition
20986 if cmp isn't SET with IF_THEN_ELSE src.
20988 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20990 PR tree-optimization/80158
20991 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
20992 replacing a candidate statement, also replace it for the
20993 candidate's alternate interpretation.
20994 (replace_rhs_if_not_dup): Likewise.
20995 (replace_one_candidate): Likewise.
20997 2017-03-24 Richard Biener <rguenther@suse.de>
20999 PR tree-optimization/80167
21000 * graphite-isl-ast-to-gimple.c
21001 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
21003 (translate_isl_ast_to_gimple::get_rename): Likewise.
21005 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
21007 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
21008 handling of certain combinations of target options, including the
21009 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
21010 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
21012 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21015 * config/arm/arm.md (*load_multiple): Add reload_completed to
21016 matching condition.
21018 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21019 Richard Biener <rguenth@suse.de>
21021 PR tree-optimization/79908
21022 PR tree-optimization/80136
21023 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
21024 been cast away, gimplify_and_add suffices.
21026 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
21028 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
21030 2017-03-23 Richard Biener <rguenther@suse.de>
21032 PR tree-optimization/80032
21033 * gimplify.c (gimple_push_cleanup): Forced unconditional
21034 cleanups still have to go to the conditional_cleanups
21037 2017-03-22 Jakub Jelinek <jakub@redhat.com>
21039 PR tree-optimization/80072
21040 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
21042 (next_operand_entry_id): Change type to unsigned int.
21043 (sort_by_operand_rank): Make sure to return the right return value
21044 even if unsigned fields are bigger than INT_MAX.
21045 (struct oecount): Change cnt and id type to unsigned int.
21046 (oecount_hasher::equal): Formatting fix.
21047 (oecount_cmp): Make sure to return the right return value
21048 even if unsigned fields are bigger than INT_MAX.
21049 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
21052 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
21053 TREE_READONLY on result if writing it more than once.
21056 * doc/invoke.texi (-fsanitize=thread): Document that with
21057 -fnon-call-exceptions atomics are not able to throw
21061 * tsan.c: Include tree-eh.h.
21062 (instrument_builtin_call): Call maybe_clean_eh_stmt or
21063 maybe_clean_or_replace_eh_stmt where needed.
21064 (instrument_memory_accesses): Add cfg_changed argument.
21065 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
21066 if it returned true.
21067 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
21069 PR rtl-optimization/63191
21070 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
21071 wrapper function, moved the whole old content into ...
21072 (ix86_delegitimize_address_1): ... this. New inline function.
21073 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
21074 true as last argument instead of ix86_delegitimize_address.
21076 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
21078 * config/aarch64/aarch64.c (generic_branch_cost): Copy
21079 cortexa57_branch_cost.
21081 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
21083 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
21085 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21088 * doc/md.texi (Constraints): Document wA constraint.
21089 * config/rs6000/constraints.md (wA): New.
21090 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
21091 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
21092 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
21093 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
21095 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
21098 * gimplify.c (is_oacc_declared): New function.
21099 (oacc_default_clause): Use it to set default flags for acc declared
21100 variables inside parallel regions.
21101 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
21102 declared variables.
21103 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
21104 declare attribute to any decl as necessary.
21106 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
21109 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
21110 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
21111 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
21112 (arm_arch_lpae): This.
21113 * config/arm/arm.c (arm_arch7ve): Rename into ...
21114 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
21115 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
21118 2017-03-22 Martin Liska <mliska@suse.cz>
21121 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
21122 error message instead of an ICE.
21124 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21126 * doc/extend.texi (6.11 Additional Floating Types): Revise.
21128 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
21130 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
21132 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
21135 2017-03-21 Martin Sebor <msebor@redhat.com>
21137 * doc/extend.texi: Use "cannot" instead of "can't."
21138 * doc/hostconfig.texi: Same.
21139 * doc/install.texi: Same.
21140 * doc/invoke.texi: Same.
21141 * doc/loop.texi: Same.
21142 * doc/md.texi: Same.
21143 * doc/objc.texi: Same.
21144 * doc/rtl.texi: Same.
21145 * doc/tm.texi: Same.
21146 * doc/tm.texi.in: Same.
21147 * doc/trouble.texi: Same.
21149 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
21152 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
21153 (collect_checksum_attributes): Set it.
21154 (die_checksum_ordered): Use it.
21156 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21158 PR tree-optimization/79908
21159 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
21160 change: For a VA_ARG whose LHS has been cast away, use
21161 force_gimple_operand to construct the side effects.
21163 2017-03-21 David Malcolm <dmalcolm@redhat.com>
21165 PR translation/80001
21166 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
21167 more amenable to translation.
21168 (oacc_loop_auto_partitions): Likewise.
21170 2017-03-21 Marek Polacek <polacek@redhat.com>
21171 Martin Sebor <msebor@redhat.com>
21173 PR tree-optimization/80109
21174 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
21175 on INTEGRAL_TYPE_P.
21177 2017-03-21 Jakub Jelinek <jakub@redhat.com>
21178 Segher Boessenkool <segher@kernel.crashing.org>
21181 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
21182 check reg_used_between_p between insn and one of succ or succ2
21183 depending on if succ is artificial insn not inserted into insn
21186 2017-03-21 Martin Liska <mliska@suse.cz>
21188 PR gcov-profile/80081
21189 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
21190 * doc/gcc.texi: Include gcov-dump stuff.
21191 * doc/gcov-dump.texi: New file.
21193 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
21195 PR rtl-optimization/79150
21196 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
21197 conditional jump, if the jump is the last insn of the loop.
21199 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21200 Richard Biener <rguenth@suse.de>
21202 PR tree-optimization/79908
21203 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
21204 been cast away, use force_gimple_operand to construct the side
21207 2017-03-21 Martin Liska <mliska@suse.cz>
21209 PR libfortran/79956
21210 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
21213 2017-03-21 Brad Spengler <spender@grsecurity.net>
21216 * plugin.c (htab_hash_plugin): New function.
21217 (add_new_plugin): Use it and adjust.
21218 (parse_plugin_arg_opt): Adjust.
21219 (init_one_plugin): Likewise.
21221 2017-03-21 Richard Biener <rguenther@suse.de>
21223 PR tree-optimization/80032
21224 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
21225 if set force the cleanup to happen unconditionally.
21226 (gimplify_target_expr): Push inserted clobbers with force_uncond
21227 to avoid them being removed by control-dependent DCE.
21229 2017-03-21 Richard Biener <rguenther@suse.de>
21231 PR tree-optimization/80122
21232 * tree-inline.c (copy_bb): Do not expans va-arg packs or
21233 va_arg_pack_len when the inlined call stmt requires pack
21235 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
21237 2017-03-21 Jakub Jelinek <jakub@redhat.com>
21240 * tsan.c (instrument_builtin_call): If the memory model argument
21241 is not a constant, assume it is valid.
21244 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
21247 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
21249 PR rtl-optimization/79910
21250 * combine.c (can_combine_p): Do not allow combining an I0 or I1
21251 if its dest is used by an insn before I2 (other than the combined
21252 insns themselves, which are properly handled already).
21254 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
21257 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
21259 * combine.c (record_used_regs): New static function.
21260 (try_combine): Handle situations where there is an additional
21261 instruction between I2 and I3 which needs to have a LOG_LINK
21265 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
21267 * combine.c (try_combine): Delete redundant i1 test. Call
21268 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
21270 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21273 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
21274 alternatives 13/14.
21276 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21278 PR tree-optimization/80054
21279 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
21280 the optimization if a PHI or any of its arguments is not dominated
21281 by the candidate's basis. Use gphi* rather than gimple* as
21283 (replace_profitable_candidates): Clean up a gimple* variable that
21284 should be a gphi* variable.
21286 2017-03-20 Martin Sebor <msebor@redhat.com>
21289 * doc/extend.texi (attribute constructor): Document present limitation.
21291 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
21294 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
21295 __POWER9_VECTOR__ #ifdef control, change template definition to
21296 use Power9-specific built-in function.
21297 (vec_any_eq): Likewise.
21298 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
21299 to control outcomes from this test.
21300 (vector_ae_<mode>p): For VEC_F modes, likewise.
21302 2017-03-20 Ian Lance Taylor <iant@google.com>
21304 * config/i386/i386.c (ix86_function_regparm): Save an extra
21305 register for -fsplit-stack with DECL_STATIC_CHAIN.
21307 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
21310 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
21311 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
21313 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
21315 * config/riscv/riscv.c (riscv_print_operand): Use "fence
21317 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
21320 2017-03-20 Marek Polacek <polacek@redhat.com>
21323 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
21325 2017-03-20 Richard Biener <rguenther@suse.de>
21327 PR tree-optimization/80113
21328 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
21329 allocate extra SSA name for PHI def.
21330 (add_close_phis_to_outer_loops): Likewise.
21331 (add_close_phis_to_merge_points): Likewise.
21332 (copy_loop_close_phi_args): Likewise.
21333 (copy_cond_phi_nodes): Likewise.
21335 2017-03-20 Martin Liska <mliska@suse.cz>
21337 PR middle-end/79753
21338 * tree-chkp.c (chkp_build_returned_bound): Do not build
21339 returned bounds for a LHS that's not a BOUNDED_P type.
21341 2017-03-20 Martin Liska <mliska@suse.cz>
21345 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
21346 COMPLEX_CST and VECTOR_CST.
21348 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21351 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
21352 target operand. A new splitter adds the clobber statement in case
21353 the target operand is dead anyway.
21355 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
21357 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
21358 to age-old versions of binutils and glibc.
21360 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
21362 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
21364 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
21366 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
21368 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
21370 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
21371 requirement for binutils 2.13.
21373 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
21375 * combine.c (try_combine): Delete redundant i1 test. Call
21376 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
21378 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
21380 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
21381 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
21383 <riscv64-*-elf>: Re-arrange section
21384 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
21385 <riscv32-*-linux>: Likewise.
21386 <riscv64-*-elf>: Likewise
21387 <riscv64-*-linux>: Likewise.
21389 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
21392 * aarch64.opt(verbose-cost-dump): Fix typo.
21394 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
21397 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
21398 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
21400 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
21402 * reload.c (find_reloads): When reloading a nonoffsettable address,
21403 use RELOAD_OTHER for it and its address reloads.
21405 PR rtl-optimization/79910
21406 * combine.c (record_used_regs): New static function.
21407 (try_combine): Handle situations where there is an additional
21408 instruction between I2 and I3 which needs to have a LOG_LINK
21411 2017-03-17 Jeff Law <law@redhat.com>
21413 PR tree-optimization/71437
21414 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
21415 conditional in the hash table first.
21416 (vrp_dom_walker::before_dom_children): Extract condition from
21417 ASSERT_EXPR. Record condition, its inverion and any implied
21418 conditions as well.
21420 2017-03-17 Marek Polacek <polacek@redhat.com>
21421 Markus Trippelsdorf <markus@trippelsdorf.de>
21423 PR tree-optimization/80079
21424 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
21427 2017-03-17 Richard Biener <rguenther@suse.de>
21429 PR middle-end/80075
21430 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
21431 Properly verify the LHS before the RHS possibly claims to be
21433 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
21436 2017-03-17 Martin Jambor <mjambor@suse.cz>
21438 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
21439 (List of -O2 options): Likewise.
21440 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
21443 2017-03-17 Tom de Vries <tom@codesourcery.com>
21445 * gcov-dump.c (print_usage): Print bug_report_url.
21447 2017-03-17 Richard Biener <rguenther@suse.de>
21449 PR middle-end/80050
21450 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
21451 (parser::peek): Likewise.
21453 2017-03-17 Richard Biener <rguenther@suse.de>
21455 PR tree-optimization/80048
21456 * sese.c (free_sese_info): Properly release rename_map and
21457 copied_bb_map elements.
21459 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
21461 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
21462 Add linked-list forward and backlinks. Insert on
21463 construction, remove on destruction.
21464 (class pass_store_merging): Add m_stores_head field.
21465 (pass_store_merging::terminate_and_process_all_chains):
21466 Iterate over m_stores_head list.
21467 (pass_store_merging::terminate_all_aliasing_chains):
21469 (pass_store_merging::execute): Check for debug stmts first.
21470 Push new chains onto the m_stores_head stack.
21472 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
21475 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
21476 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
21477 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
21479 2017-03-16 Jeff Law <law@redhat.com>
21481 PR tree-optimization/71437
21482 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
21483 member function. Implementation moved into after_dom_children
21484 member function and into the threader's thread_outgoing_edges
21486 (dom_opt_dom_walker::after_dom_children): Simplify by moving
21487 some code into new thread_outgoing_edges.
21488 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
21489 definition. Simplify marker handling (do it here). Assume we always
21490 have the available expression and the const/copies tables.
21491 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
21493 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
21494 * tree-vrp.c (equiv_stack): No longer file scoped.
21495 (vrp_dom_walker): New class.
21496 (vrp_dom_walker::before_dom_children): New member function.
21497 (vrp_dom_walker::after_dom_children): Likewise.
21498 (identify_jump_threads): Setup domwalker. Use it rather than
21499 walking edges in a random order by hand. Simplify setup/finalization.
21500 (finalize_jump_threads): Remove.
21501 (vrp_finalize): Do not call identify_jump_threads here.
21502 (execute_vrp): Do it here instead and call thread_through_all_blocks
21505 PR tree-optimization/71437
21506 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
21508 (simplify_stmt_for_jump_threading): Add basic_block argument. All
21510 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
21511 (dom_opt_dom_walker::thread_across_edge): Remove
21512 handle_dominating_asserts argument. All callers changed.
21513 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
21514 changes. Remove calls to lhs_of_dominating_assert. Other
21515 uses of handle_dominating_asserts turn into unconditional code
21516 (simplify_control_stmt_condition_1): Likewise.
21517 (simplify_control_stmt_condition): Likewise.
21518 (thread_through_normal_block, thread_across_edge): Likewise.
21519 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
21520 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
21521 object if it is not an SSA_NAME.
21522 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
21523 before calling into the VRP specific simplifiers.
21524 (identify_jump_threads): Remove handle_dominating_asserts
21527 2017-03-16 Jakub Jelinek <jakub@redhat.com>
21530 * tree-diagnostic.c (default_tree_printer): No longer static.
21531 * tree-diagnostic.h (default_tree_printer): New prototype.
21533 2017-03-16 Tamar Christina <tamar.christina@arm.com>
21535 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
21536 Change ins into fmov.
21538 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21540 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
21541 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
21542 Use h_con constraint for operand 1.
21543 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
21544 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
21546 2017-03-15 Jeff Law <law@redhat.com>
21548 PR tree-optimization/71437
21549 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
21550 (record_temporary_equivalences): Use it.
21552 PR tree-optimization/71437
21553 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
21554 tree-ssa-scopedtables.
21555 (lookup_avail_expr, build_and_record_new_cond): Likewise.
21556 (record_conditions, record_cond, vuse_eq): Likewise.
21557 (record_edge_info): Adjust to API tweak of record_conditions.
21558 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
21559 (record_temporary_equivalences, optimize_stmt): Likewise.
21560 (eliminate_redundant_computations): Likewise.
21561 (record_equivalences_from_stmt): Likewise.
21562 * tree-ssa-scopedtables.c: Include options.h and params.h.
21563 (vuse_eq): New function, moved from tree-ssa-dom.c
21564 (build_and_record_new_cond): Likewise.
21565 (record_conditions): Likewise. Accept vector of conditions rather
21566 than edge_equivalence structure for first argument.
21567 for the first argument.
21568 (avail_exprs_stack::lookup_avail_expr): New member function, moved
21569 from tree-ssa-dom.c.
21570 (avail_exprs_stack::record_cond): Likewise.
21571 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
21572 from tree-ssa-dom.c.
21573 (avail_exprs_stack): Add new member functions lookup_avail_expr
21575 (record_conditions): Declare.
21577 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
21580 * lra-constraints.c (process_alt_operands): Increase reject for
21581 reloading an input/output operand.
21583 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
21586 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
21587 insns to convert from signed/unsigned char/short to IEEE 128-bit
21589 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
21591 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
21594 * config/i386/i386.c (ix86_vector_duplicate_value): Create
21595 subreg of inner mode for values already in registers.
21597 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
21599 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
21600 iteration reg is used after the loop.
21602 2017-03-14 Martin Sebor <msebor@redhat.com>
21604 PR tree-optimization/79800
21605 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
21606 precision in negative-positive range.
21607 (format_floating): Call non-const overload with adjusted precision.
21609 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
21612 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
21615 2017-03-14 Martin Sebor <msebor@redhat.com>
21617 PR middle-end/80020
21618 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
21619 * builtins.def (aligned_alloc): Use it.
21622 * Makefile.in (GTFILES): Add calls.c.
21623 * calls.c: Include "gt-calls.h".
21625 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
21627 PR rtl-optimization/79728
21628 * regs.h (struct target_regs): New field
21629 x_contains_allocatable_regs_of_mode.
21630 (contains_allocatable_regs_of_mode): New macro.
21631 * reginfo.c (init_reg_sets_1): Initialize it, and change
21632 contains_reg_of_mode so it includes global regs as well.
21633 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
21634 rather than contains_regs_of_mode.
21636 2017-03-14 Martin Liska <mliska@suse.cz>
21638 * doc/invoke.texi: Document options that can't be combined with
21639 -fcheck-pointer-bounds.
21641 2017-03-14 Martin Liska <mliska@suse.cz>
21643 PR middle-end/79831
21644 * doc/invoke.texi (-Wchkp): Document the option.
21646 2017-03-14 Martin Liska <mliska@suse.cz>
21648 * Makefile.in: Install gcov-dump.
21650 2017-03-14 Martin Liska <mliska@suse.cz>
21652 * multiple_target.c (expand_target_clones): Bail out for
21653 an invalid attribute.
21655 2017-03-14 Richard Biener <rguenther@suse.de>
21657 * alias.c (struct alias_set_entry): Pack properly.
21658 * cfgloop.h (struct loop): Likewise.
21659 * cse.c (struct set): Likewise.
21660 * ipa-utils.c (struct searchc_env): Likewise.
21661 * loop-invariant.c (struct invariant): Likewise.
21662 * lra-remat.c (struct cand): Likewise.
21663 * recog.c (struct change_t): Likewise.
21664 * rtl.h (struct address_info): Likewise.
21665 * symbol-summary.h (function_summary): Likewise.
21666 * tree-loop-distribution.c (struct partition): Likewise.
21667 * tree-object-size.c (struct object_size_info): Likewise.
21668 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
21669 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
21670 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
21671 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
21672 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
21673 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
21674 (struct _stmt_vec_info): Likewise.
21676 2017-03-14 Martin Liska <mliska@suse.cz>
21679 * multiple_target.c (create_dispatcher_calls): Check that
21680 a target can create a function dispatcher.
21682 2017-03-14 Martin Liska <mliska@suse.cz>
21685 * multiple_target.c (expand_target_clones): Drop local.local
21686 flag for default implementation.
21688 2017-03-14 Richard Biener <rguenther@suse.de>
21690 PR tree-optimization/80030
21691 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
21693 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
21695 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
21696 gcc_fallthrough() instead of __attribute__((fallthrough));
21698 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
21700 * doc/gcc.texi: Remove "up" link to (DIR).
21701 * doc/gccint.texi: Ditto.
21703 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
21705 * doc/install.texi (Specific) <avr>: Remove reference to
21708 2017-03-13 Jeff Law <law@redhat.com>
21710 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
21711 attribute rather than comments.
21713 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
21714 match_scratch operand is highest.
21716 2017-03-13 Martin Liska <mliska@suse.cz>
21718 PR middle-end/78339
21719 * ipa-pure-const.c (warn_function_noreturn): If the declarations
21720 is a CHKP clone, use original declaration.
21722 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21724 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
21725 (arc_conditional_register_usage): Use a different allocation order
21726 when optimizing for size.
21727 * common/config/arc/arc-common.c (arc_option_optimization_table):
21728 Section anchors default on when optimizing for size.
21730 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21732 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
21734 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21736 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
21737 * config/arc/arc.md (cpu_facility): Add cd variant.
21738 (*movqi_insn): Add code density variant.
21739 (*movhi_insn): Likewise.
21740 (*movqi_insn): Likewise.
21741 (*addsi3_mixed): Likewise.
21742 (subsi3_insn): Likewise.
21744 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21746 * config/arc/arc.md (movsi_cond_exec): Update constraint.
21748 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21750 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
21751 expressions with MINUS and UNARY ops.
21753 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21756 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
21758 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
21759 between vec_select and vector argument.
21760 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
21761 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
21762 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
21763 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
21764 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
21765 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
21767 2017-03-13 Richard Biener <rguenther@suse.de>
21770 * params.def (vect-max-peeling-for-alignment): Fix typo.
21772 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21774 * doc/install.texi (Specific) <mips-*-*>: Remove description of
21775 issue that only occurred with binutils below 2.18.
21777 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21779 * doc/install.texi (Specific) <cris-axis-elf>: No longer
21780 refer to binutils 2.11/2.12 minimum.
21782 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21784 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
21785 ftp.kernel.org and simplify binutils requirement.
21787 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
21789 * doc/invoke.texi (Warning Options): Fix spelling of link-time
21791 (Optimize Options): Ditto. Also remove redundancy.
21793 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21795 PR translation/79848
21796 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
21798 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
21799 to G_ to avoid double translation.
21801 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21803 PR translation/79923
21804 * auto-profile.c (get_combined_location): Convert leading
21805 character of diagnostics to lower case and remove trailing period.
21806 (read_profile): Likewise for various diagnostics.
21807 * config/arm/arm.c (arm_option_override): Remove trailing period
21808 from various diagnostics.
21809 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
21810 (msp430_expand_delay_cycles): Likewise.
21812 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21815 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
21816 full command-line argument, rather than just "str".
21817 (aarch64_validate_march): Likewise.
21818 (aarch64_validate_mtune): Likewise.
21820 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
21822 PR rtl-optimization/78911
21823 * lra-assigns.c (must_not_spill_p): New function.
21824 (spill_for): Use it.
21826 2017-03-10 Jakub Jelinek <jakub@redhat.com>
21828 PR tree-optimization/79981
21829 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
21830 ATOMIC_COMPARE_EXCHANGE ifn result.
21831 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
21832 IFN_ATOMIC_COMPARE_EXCHANGE.
21834 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21837 * opts.c (parse_sanitizer_options): Add missing question mark to
21838 "did you mean" message.
21840 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21842 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
21844 (VMULEUH_UNS): Likewise.
21845 (VMULOUB_UNS): Likewise.
21846 (VMULOUH_UNS): Likewise.
21847 * config/rs6000/rs6000.c (builtin_function_type): Remove
21848 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
21850 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21853 * read-rtl-function.c (function_reader::read_rtx_operand): Update
21854 x with result of extra_parsing_for_operand_code_0.
21855 (function_reader::extra_parsing_for_operand_code_0): Convert
21856 return type from void to rtx, returning x. When reading
21857 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
21858 larger size containing struct block_symbol.
21860 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
21862 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
21863 -mfloat128-hardware without -m64.
21865 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
21868 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
21869 entries to the case statement that marks unsigned arguments to
21870 overloaded functions.
21872 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21874 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
21875 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
21877 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
21880 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
21881 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
21883 2017-03-10 Martin Liska <mliska@suse.cz>
21887 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
21888 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
21891 2017-03-10 Olivier Hainque <hainque@adacore.com>
21893 * tree-switch-conversion (array_value_type): Start by resetting
21894 candidate type to it's main variant.
21896 2017-03-10 Jakub Jelinek <jakub@redhat.com>
21898 PR rtl-optimization/79909
21899 * combine.c (try_combine): Use simplify_replace_rtx on individual
21900 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
21901 of the whole CALL_INSN_FUNCTION_USAGE.
21903 PR tree-optimization/79972
21904 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
21905 get_range_info on SSA_NAMEs. Formatting fixes.
21907 2017-03-10 Richard Biener <rguenther@suse.de>
21908 Jakub Jelinek <jakub@redhat.com>
21910 PR tree-optimization/77975
21911 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
21912 edge to be constant.
21913 (get_val_for): For constant x return it. Formatting fix.
21914 (loop_niter_by_eval): Avoid pointless looping if the next iteration
21915 would use the same bases as the current one.
21917 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21919 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
21920 instead of vec_select for V1TImode.
21921 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
21923 (VSX_LE_128): Add V1TI to this mode iterator.
21924 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
21925 (*vsx_le_perm_store_<mode>): Likewise.
21926 (pre-reload splitter for VSX stores): Likewise.
21927 (post-reload splitter for VSX stores): Likewise.
21928 (*vsx_xxpermdi2_le_<mode>): Likewise.
21929 (*vsx_lxvd2x2_le_<mode>): Likewise.
21930 (*vsx_stxvd2x2_le_<mode>): Likewise.
21932 2017-03-09 Michael Eager <eager@eagercon.com>
21934 Correct failures with --enable-checking=yes,rtl.
21936 * config/microblaze/microblaze.c (microblaze_expand_shift):
21937 Replace GET_CODE test with CONST_INT_P and INTVAL test with
21938 test for const0_rtx.
21939 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
21940 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
21942 2017-03-09 Richard Biener <rguenther@suse.de>
21944 PR tree-optimization/79977
21945 * graphite-scop-detection.c (scop_detection::merge_sese):
21946 Handle the case of extra exits to blocks dominating the entry.
21948 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
21950 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21953 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
21955 PR rtl-optimization/79949
21956 * lra-constraints.c (process_alt_operands): Check memory when
21957 trying to predict a cycle. Print about the overall increase.
21959 2017-03-09 Richard Biener <rguenther@suse.de>
21961 PR middle-end/79971
21962 * gimple-expr.c (useless_type_conversion_p): Preserve
21963 TYPE_SATURATING for fixed-point types.
21965 2017-03-09 Richard Biener <rguenther@suse.de>
21968 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
21969 alignment of BLKmode params.
21971 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21974 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
21975 (VALL_NO_V2Q): Likewise.
21977 * config/aarch64/aarch64-simd.md
21978 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
21980 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
21981 VALL_NO_V2Q mode iterator.
21982 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
21984 2017-03-09 Martin Liska <mliska@suse.cz>
21986 PR tree-optimization/79631
21987 * tree-chkp-opt.c (chkp_is_constant_addr): Call
21988 tree_int_cst_sign_bit just for INTEGER constants.
21990 2017-03-09 Martin Liska <mliska@suse.cz>
21994 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
21997 2017-03-09 Marek Polacek <polacek@redhat.com>
22000 * tree.c (inchash::add_expr): Handle TREE_VEC.
22002 2017-03-09 Martin Liska <mliska@suse.cz>
22005 (chkp_narrow_size_and_offset): New function.
22006 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
22007 (void chkp_parse_bit_field_ref): New function.
22008 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
22009 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
22011 2017-03-09 Martin Liska <mliska@suse.cz>
22014 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
22015 (chkp_find_bounds_1): Remove gcc_unreachable.
22017 2017-03-09 Jakub Jelinek <jakub@redhat.com>
22020 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
22021 BUILT_IN_SYNC*, determine the access type from the size suffix and
22022 always build a MEM_REF with that type. Handle forgotten
22023 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
22026 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
22027 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
22028 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
22029 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
22030 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
22031 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
22032 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
22033 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
22034 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
22035 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
22036 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
22037 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
22038 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
22039 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
22040 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
22041 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
22042 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
22043 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
22044 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
22045 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
22046 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
22047 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
22048 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
22049 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
22050 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
22051 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
22052 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
22053 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
22054 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
22055 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
22056 definitions outside of __OPTIMIZE__ guarded section.
22059 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
22060 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
22061 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
22062 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
22065 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22067 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
22068 ("vfenez<mode>"): Add missing constraints.
22070 2017-03-08 Martin Sebor <msebor@redhat.com>
22073 * config/nds32/nds32.c (nds32_option_override):
22074 Fix misspelled diagnostic.
22076 2017-03-08 Jakub Jelinek <jakub@redhat.com>
22079 * gimplify.c (gimplify_omp_for): Replace index var in outer
22080 taskloop statement with an artificial variable and add
22081 OMP_CLAUSE_PRIVATE clause for it.
22083 2017-03-08 Richard Biener <rguenther@suse.de>
22085 PR tree-optimization/79955
22086 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
22087 for accesses that are completely outside of the variable.
22089 2017-03-08 Andrew Haley <aph@redhat.com>
22091 PR tree-optimization/79943
22092 * tree-ssa-loop-split.c (compute_new_first_bound): When
22093 calculating the new upper bound, (END-BEG) should be added, not
22096 2017-03-08 Jakub Jelinek <jakub@redhat.com>
22098 * config/avr/avr.md (setmemhi): Make sure match_dup
22099 operand number comes before match_scratch.
22101 2017-03-08 Richard Biener <rguenther@suse.de>
22103 PR tree-optimization/79920
22104 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
22105 with ncopies == 1 to ...
22106 (vect_transform_slp_perm_load): ... here. Properly compute
22107 all element loads by iterating VF times over the group. Do
22108 not handle ncopies (computed in a broken way) in
22109 vect_create_mask_and_perm.
22111 2017-03-08 Jakub Jelinek <jakub@redhat.com>
22114 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
22115 is a uniform vector, use uniform_vector_p return value instead of
22116 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
22118 2017-03-07 Marek Polacek <polacek@redhat.com>
22120 PR middle-end/79809
22121 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
22122 (alloca_call_type): Likewise.
22124 2017-03-07 Martin Liska <mliska@suse.cz>
22126 * gcov.c (process_args): Put comment to correct location.
22128 2017-03-07 Martin Liska <mliska@suse.cz>
22130 PR middle-end/68270
22131 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
22132 Use array_at_struct_end_p instead of DECL_CHAIN (field).
22133 (chkp_narrow_bounds_for_field): Likewise.
22134 (chkp_parse_array_and_component_ref): Pass one more argument to
22137 2017-03-07 Richard Biener <rguenther@suse.de>
22139 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
22142 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
22144 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
22145 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
22147 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22150 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
22151 to end of description.
22152 (PARAM_MAX_STORES_TO_MERGE): Likewise.
22154 2017-03-07 Jakub Jelinek <jakub@redhat.com>
22156 PR rtl-optimization/79901
22157 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
22159 (*avx512f_<code><mode>3<mask_name>): ... this.
22160 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
22161 iterator instead of VI8_AVX2_AVX512BW.
22163 PR rtl-optimization/79901
22164 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
22165 min/max expander, expand it using expand_vec_cond_expr.
22168 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
22171 2017-03-06 Jakub Jelinek <jakub@redhat.com>
22174 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
22175 to void * for PCH reasons.
22176 * dwarf2out.c (output_loc_operands, output_die): Cast
22177 v.val_vec.array to unsigned char *.
22179 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
22182 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
22185 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
22187 PR rtl-optimization/79571
22188 * lra-constraints.c (process_alt_operands): Calculate static
22189 reject and subtract it from overall when only addresses will be
22192 2017-03-06 Julia Koval <julia.koval@intel.com>
22195 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
22196 incoming stack boundary to 128 for 64-bit targets.
22198 2017-03-06 Richard Biener <rguenther@suse.de>
22200 PR tree-optimization/79894
22201 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
22202 to NULL after folding it.
22204 2017-03-06 Richard Biener <rguenther@suse.de>
22206 PR tree-optimization/79824
22207 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
22208 check disabling peeling for gaps.
22210 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
22212 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
22213 attributes): Document gettimeofday.
22215 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
22217 * config/s390/s390.c (s390_option_override_internal): Set
22218 PARAM_MIN_VECT_LOOP_BOUND
22220 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
22222 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
22223 * config/s390/s390.md: Likewise.
22225 2017-03-06 Jakub Jelinek <jakub@redhat.com>
22228 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
22229 (<avx2_avx512>_perm<mode>): Rename to ...
22230 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
22232 (<avx512>_perm<mode>_mask): Rename to ...
22233 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
22235 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
22236 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
22237 instead of VI8F_256_512.
22238 (avx512f_perm<mode>): New define_expand.
22239 (avx512f_perm<mode>_mask): Likewise.
22240 (avx512f_perm<mode>_1<mask_name>): New define_insn.
22241 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
22243 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22245 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
22246 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
22248 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
22250 2017-03-06 Martin Liska <mliska@suse.cz>
22253 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
22254 when having a SSA NAME w/o VAR_DECL assigned to it.
22256 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22258 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
22259 msa_dpsub_<su>_d): Fix MODE for vec_select.
22261 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22263 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
22265 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
22267 2017-03-06 Richard Biener <rguenther@suse.de>
22269 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
22270 * plugin.c (register_plugin_info): Likewise.
22271 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
22273 2017-03-05 Jakub Jelinek <jakub@redhat.com>
22275 * config/i386/sse.md (sse_storehps, sse_storelps,
22276 avx_<castmode><avxsizesuffix>_<castmode>,
22277 avx512f_<castmode><avxsizesuffix>_<castmode>,
22278 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
22279 in condition that at least one operand is not a MEM.
22281 2017-03-03 Jakub Jelinek <jakub@redhat.com>
22283 PR middle-end/79805
22284 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
22285 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
22287 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
22288 gimple_call_nothrow_p flag based on whether original builtin can throw.
22289 If it can, emit following stmts on the fallthrough edge.
22290 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
22291 don't create new bb if inserting just debug stmts on the edge, try to
22292 insert them on the fallthru bb or just reset debug stmts.
22294 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
22297 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
22298 restore recog_data (including the operand rtxes inside it) around
22299 the call to get_insn_template.
22301 2017-03-03 Martin Sebor <msebor@redhat.com>
22303 PR tree-optimization/79699
22304 * context.c (context::~context): Free MPFR caches to avoid
22305 a memory leak on program exit.
22307 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22309 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
22310 Use wide_int::ulow () instead of .elt (0).
22312 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
22314 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
22315 (*pushxf): Limit oF constraint to 32bit targets and add oC
22316 constraint for 64bit targets.
22317 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
22318 (*pushdf): Change rmF constraint to rmC.
22320 2017-03-03 Martin Liska <mliska@suse.cz>
22322 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
22323 Remove unused variable.
22325 2017-03-03 Jakub Jelinek <jakub@redhat.com>
22328 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
22329 is a memory operand, increase num_memory.
22330 (ix86_expand_args_builtin): Likewise.
22332 2017-03-03 Jan Hubicka <jh@suse.cz>
22335 * ipa-devirt.c (maybe_record_node): Properly handle
22336 __cxa_pure_virtual visibility.
22338 2017-03-03 Martin Liska <mliska@suse.cz>
22340 PR tree-optimization/79803
22341 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
22343 (pass_loop_prefetch::execute): Disabled optimization if an
22344 assumption about L1 cache size is not met.
22346 2017-03-03 Martin Liska <mliska@suse.cz>
22348 PR rtl-optimization/79574
22349 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
22350 (hash_scan_set): Likewise.
22351 (dump_hash_table): Likewise.
22352 (hoist_code): Likewise.
22354 2017-03-03 Richard Biener <rguenther@suse.de>
22356 * fixed-value.c (fixed_from_string): Restore use of elt (1)
22357 in place of uhigh ().
22358 (fixed_convert_from_real): Likewise.
22360 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
22363 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
22365 2017-03-03 Richard Biener <rguenther@suse.de>
22367 PR middle-end/79818
22368 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
22369 TYPE_OVERFLOW_UNDEFINED check.
22371 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22373 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
22375 (vector_ae_<mode>_p): Likewise.
22376 (vector_nez_<mode>_p): Likewise.
22377 (vector_ne_v2di_p): Likewise.
22378 (vector_ae_v2di_p): Likewise.
22379 (vector_ne_<mode>_p): Likewise.
22380 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
22382 (vsx_tsqrt<mode>2_fe): Likewise.
22384 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
22387 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
22389 2017-03-02 Jakub Jelinek <jakub@redhat.com>
22391 PR rtl-optimization/79780
22392 * cprop.c (one_cprop_pass): When second and further conditional trap
22393 in a single basic block is turned into an unconditional trap, turn it
22394 into a deleted note to avoid RTL verification failures.
22396 2017-03-02 Richard Biener <rguenther@suse.de>
22398 * fold-const.c (const_binop): Use ulow () instead of elt (0).
22400 2017-03-02 Richard Biener <rguenther@suse.de>
22402 PR tree-optimization/79345
22404 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
22405 param and abort the walk, returning -1 if it is hit.
22406 (walk_aliased_vdefs): Take a limit param and pass it on.
22407 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
22408 defaulting to 0 and return a signed int.
22409 * tree-ssa-uninit.c (struct check_defs_data): New struct.
22410 (check_defs): New helper.
22411 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
22412 about uninitialized memory.
22413 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
22414 bogus uninitialized warning.
22415 (fixed_convert_from_real): Likewise.
22417 2017-03-02 Bin Cheng <bin.cheng@arm.com>
22419 PR tree-optimization/66768
22420 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
22421 iv_use if base object can't be determined.
22423 2017-03-02 Jakub Jelinek <jakub@redhat.com>
22425 PR tree-optimization/79345
22426 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
22427 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
22428 (get_pattern_stats): Initialize it.
22429 * genemit.c (gen_expand): Verify match_scratch numbers come after
22430 match_operand/match_dup numbers.
22431 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
22432 match_scratch numbers.
22433 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
22435 * config/s390/s390.md (trunctdsd2): Likewise.
22437 2017-03-02 Richard Biener <rguenther@suse.de>
22439 * wide-int.h (wide_int_storage::operator=): Implement in terms
22442 2017-03-02 Richard Biener <rguenther@suse.de>
22444 PR tree-optimization/79777
22445 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
22446 the to insert expression to sth existing.
22448 2017-03-01 Martin Sebor <msebor@redhat.com>
22450 PR middle-end/79692
22451 * gimple-ssa-sprintf.c
22452 (directive::known_width_and_precision): New function.
22453 (format_integer): Use it.
22454 (get_mpfr_format_length): Consider the full range of precision
22455 when computing %g output with the # flag. Set the likely byte
22456 count to 3 rather than 1 when precision is indeterminate.
22457 (format_floating): Correct the lower bound of precision.
22459 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22461 * doc/invoke.texi: Document default code model for 64-bit Linux.
22463 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22466 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
22467 udiv rather than div since input pattern is unsigned.
22469 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
22471 * config/i386/i386.c (print_reg): Warn for values of
22472 unsupported size in integer register.
22474 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
22477 * config/rs6000/predicates.md (current_file_function_operand): Do
22478 not allow self calls to be local if the function is replaceable.
22480 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
22483 * config/rs6000/altivec.h (vec_ctz and others): Change the
22484 preprocessor macro that controls conditional compilation from
22485 _ARCH_PWR9 to __POWER9_VECTOR__.
22486 (vec_all_ne): Change parameterization of __altivec_scalar_pred
22487 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
22488 control (instead of _ARCH_PWR9 control) so that template
22489 definition uses power9-specific function.
22490 (vec_any_eq): Likewise.
22491 (vec_all_ne): Change macro definition to use a power9-specific
22492 expansion under #ifdef __POWER9_VECTOR__ control (instead of
22493 _ARCH_PWR9 control).
22494 (vec_any_eq) Likewise.
22495 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
22496 expansion for CMPNEF to remove support for xvcmpnesp instruction.
22497 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
22498 support for xvcmpnedp instruction.
22499 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
22500 macro expansion so that Power9 implementation of vec_all_ne does
22501 not use the AltiVec predicate framework.
22502 (VCMPNEH_P): Likewise.
22503 (VCMPNEW_P): Likewise.
22504 (VCMPNED_P): Likewise.
22505 (VCMPNEFP_P): Likewise.
22506 (VCMPNEDP_P): Likewise.
22507 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
22508 implementation of vec_any_eq to not use AltiVec predicate
22510 (VCMPAEH_P): Likewise.
22511 (VCMPAEW_P): Likewise.
22512 (VCMPAED_P): Likewise.
22513 (VCMPAEFP_P): Likewise.
22514 (VCMPAEDP_P): Likewise.
22515 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
22516 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
22517 not use the AltiVec predicate framework.
22518 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
22519 of vec_any_eq to not use AltiVec predicate framework.
22520 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
22521 support for predefined __POWER9_VECTOR__ macro to indicate that
22522 Power9 instruction selection is enabled.
22523 (altivec_overloaded_builtins): Remove extraneous
22524 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
22525 function argument types RS6000_BTI_bool_V16QI and
22526 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
22527 entry for overloaded function argument types RS6000_BTI_bool_V4SI
22528 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
22529 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
22530 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
22531 Power9 for implementations of vec_cmpne. Change the signature for
22532 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
22533 (representing vec_all_ne) to remove the previously described first
22534 argument of type RS6000_BTI_INTSI, as this was an artifact of
22535 reliance on the AltiVec predicate framework, which is no longer
22536 used in the implementation of these functions. Add
22537 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
22538 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
22539 since, unlike the AltiVec predicate framework implementation, we
22540 do not share function descriptors between vec_alle and vec_anyeq.
22541 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
22542 set of modes that receive special treatment even when
22543 TARGET_P9_VECTOR is true. The special treatment emits code that
22544 does not depend on Power9 instructions.
22545 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
22546 define_expand to not rely on AltiVec predicate framework.
22547 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
22549 (vector_ne_v2di_p): Change this define_expand to not rely on
22550 AltiVec predicate framework.
22551 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
22553 (vector_ne_<mode>_p): Change this define_expand to not rely on
22554 AltiVec predicate framework.
22555 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
22557 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
22558 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
22559 define_insn pattern.
22560 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
22561 define_insn pattern because the xvcmpne<VSs>. instruction is not
22563 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
22564 instruction is not supported.
22566 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22568 * config/nvptx/nvptx.c: Include intl.h.
22570 2017-03-01 Martin Jambor <mjambor@suse.cz>
22573 * ipa-prop.h (ipa_bits): Removed field known.
22574 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
22575 to pointers. Adjusted their comments to warn about their sharing.
22576 (ipcp_transformation_summary): Change bits to a vector of pointers.
22577 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
22578 (ipa_get_ipa_bits_for_value): Declare.
22579 * tree-vrp.h (value_range): Mark as GTY((for_user)).
22580 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
22581 (ipa_bits_hash_table): Likewise.
22582 (ipa_vr_ggc_hash_traits): Likewise.
22583 (ipa_vr_hash_table): Likewise.
22584 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
22585 being pointers and vr_known being removed.
22586 (ipa_set_jf_unknown): Likewise.
22587 (ipa_get_ipa_bits_for_value): New function.
22588 (ipa_set_jfunc_bits): Likewise.
22589 (ipa_get_value_range): New overloaded functions.
22590 (ipa_set_jfunc_vr): Likewise.
22591 (ipa_compute_jump_functions_for_edge): Use the above functions to
22592 construct bits and vr parts of jump functions.
22593 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
22594 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
22596 (ipcp_grow_transformations_if_necessary): Also allocate
22597 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
22599 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
22600 them. Fix too long lines.
22601 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
22602 vr_known being removed.
22603 (ipa_read_jump_function): Use new setter functions to construct bits
22604 and vr parts of jump functions or set them to NULL.
22605 (write_ipcp_transformation_info): Adjust for bits being pointers.
22606 (read_ipcp_transformation_info): Likewise.
22607 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
22609 Include gt-ipa-prop.h.
22610 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
22612 (ipcp_store_bits_results): Likewise.
22613 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
22614 Do not write to existing jump functions but use a temporary instead.
22616 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22619 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
22620 attempt to use its first operand as BIT_FIELD_REF base.
22622 2017-03-01 Richard Biener <rguenther@suse.de>
22624 PR middle-end/79721
22625 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
22626 interpolating formula in wrapping arithmetic.
22627 (chrec_apply): Convert chrec_evaluate return value to wanted type.
22629 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22631 PR tree-optimization/79734
22632 * tree-vect-generic.c (expand_vector_condition): Optimize
22633 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
22634 Handle VEC_COND_EXPR where comparison has different inner width from
22635 type's inner width.
22637 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
22639 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
22640 markup, and similar issues. Remove @opindex entries for things
22641 that aren't options. Add missing -mmpy-option entries.
22643 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22645 PR tree-optimization/79737
22646 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
22647 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
22648 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
22649 instead of byte_size. Formatting fix.
22650 (shift_bytes_in_array_right): Formatting fix.
22652 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
22655 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
22656 condition on optimize for the leaf function test.
22658 2017-02-28 Martin Liska <mliska@suse.cz>
22661 * read-rtl-function.c (function_reader::handle_unknown_directive):
22662 Bail out when one uses -flto.
22664 2017-02-28 Martin Liska <mliska@suse.cz>
22666 * common.opt: Replace space with tabular for options of <number>
22668 * config/i386/i386.opt: Show <number> value for
22669 -mlarge-data-threshold.
22670 * opts.c (print_filtered_help): Do not display number in hexadecimal
22673 2017-02-28 Martin Liska <mliska@suse.cz>
22675 * common.opt: Fix --help=option -Q for options which are of
22678 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
22680 * config/i386/i386.c (print_reg): Error out for values
22681 of 8-bit size in invalid integer register.
22683 2017-02-28 Martin Sebor <msebor@redhat.com>
22685 PR tree-optimization/79691
22686 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
22688 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22691 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
22692 gcc_unreachable with output_operand_lossage.
22694 2017-02-28 Richard Biener <rguenther@suse.de>
22696 PR tree-optimization/79740
22697 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
22699 (visit_nary_op): Insert the nary into the hashtable if we
22700 pattern-matched sth.
22701 * tree-ssa-pre.c (eliminate_insert): Robustify.
22703 2017-02-28 Richard Biener <rguenther@suse.de>
22705 PR middle-end/79731
22706 * fold-const.c (decode_field_reference): Reject out-of-bound
22709 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22711 * config/i386/i386.c: Include intl.h.
22712 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
22713 instead of just cond ? "..." : "...".
22714 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
22715 * coverage.c (read_counts_file): Likewise.
22716 * omp-offload.c: Include intl.h.
22717 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
22718 of just cond ? "..." : "...".
22719 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
22720 of just cond ? "..." : "...".
22722 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
22725 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
22727 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
22728 'tune for' CPU name.
22729 * config/arm/arm-cpu-data.h: Regenerated.
22731 2017-02-28 Richard Biener <rguenther@suse.de>
22733 PR tree-optimization/79732
22734 * tree-inline.c (expand_call_inline): Do not shadow var.
22736 2017-02-28 Richard Biener <rguenther@suse.de>
22738 PR tree-optimization/79723
22739 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
22740 address-space properly.
22742 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
22744 * doc/optinfo.texi (Optimization groups): Fix option used for
22746 * doc/invoke.texi (-fopt-info): Document "omp".
22747 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
22748 (OPTGROUP_ALL): Add OPTGROUP_OMP.
22749 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
22750 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
22751 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
22753 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
22755 * dumpfile.c (optgroup_options): Instead of "openmp", associate
22756 OPTGROUP_OMP with "omp".
22758 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
22761 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
22762 for arithmetic shift of unsigned V2DI.
22764 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
22766 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
22767 arc/linux.h headers.
22768 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
22769 (LINK_SPEC): Likewise.
22770 (ARC_TLS_EXTRA_START_SPEC): Likewise.
22771 (EXTRA_SPECS): Likewise.
22772 (STARTFILE_SPEC): Likewise.
22773 (ENDFILE_SPEC): Likewise.
22774 (LIB_SPEC): Likewise.
22775 (TARGET_SDATA_DEFAULT): Likewise.
22776 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
22777 (MULTILIB_DEFAULTS): Likewise.
22778 (DWARF2_UNWIND_INFO): Likewise.
22779 * config/arc/big.h: New file.
22780 * config/arc/elf.h: Likewise.
22781 * config/arc/linux.h: Likewise.
22782 * config/arc/t-uClibc: Remove.
22784 2017-02-27 Bin Cheng <bin.cheng@arm.com>
22786 PR tree-optimization/77536
22787 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
22788 (tree_transform_and_unroll_loop): Use above function to compute the
22789 estimated niter of unrolled loop and use it when scaling profile.
22790 Also use count info rather than frequency if it's non-zero.
22791 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
22792 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
22793 (vect_transform_loop): Call above function.
22795 2017-02-27 Richard Biener <rguenther@suse.de>
22797 PR tree-optimization/45397
22798 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
22799 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
22800 (visit_nary_op): Add pattern matching for CSEing sign-changed
22801 or truncated operations with wider ones.
22803 2017-02-27 Richard Biener <rguenther@suse.de>
22805 PR tree-optimization/79690
22806 * tree-vect-stmts.c (vectorizable_store): Use vector type
22807 built from the DR with address-space.
22809 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
22811 * doc/invoke.texi (Optimize Options): Refine the description
22812 of asan-use-after-return.
22814 2017-02-25 Alan Modra <amodra@gmail.com>
22816 PR rtl-optimization/79584
22817 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
22818 base, not ad->base_term, the reg within base. Remove assertion
22819 that ad->base == ad->base_term. Replace gen_int_mode using
22820 bogus mode with const0_rtx.
22822 2017-02-25 Jakub Jelinek <jakub@redhat.com>
22824 PR middle-end/79396
22825 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
22826 FMA_EXPR like tcc_binary or tcc_unary.
22828 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
22831 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
22833 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
22834 (output_loc_operands): Handle DW_OP_call_ref and
22835 DW_OP_GNU_variable_value.
22836 (struct variable_value_struct): New type.
22837 (struct variable_value_hasher): Likewise.
22838 (variable_value_hash): New variable.
22839 (string_types): Remove.
22840 (copy_loc_descr): New function.
22841 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
22842 (prepend_loc_descr_to_each): New function.
22843 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
22844 instead of add_loc_descr_to_each if the first argument is single
22845 location list and the second has multiple.
22846 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
22847 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
22848 when looking for variable value which doesn't have other location info.
22849 (loc_list_from_tree): Formatting fix.
22850 (gen_array_type_die): Simplify DW_AT_string_length handling.
22851 (adjust_string_types): Remove.
22852 (gen_subprogram_die): Don't call adjust_string_types nor test/set
22853 string_types. Call resolve_variable_values.
22854 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
22855 (resolve_addr_in_expr): Likewise. Add A argument.
22856 (copy_deref_exprloc): Remove deref argument. Adjust for the
22857 original expression being DW_OP_GNU_variable_value with optionally
22858 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
22859 optionally after it.
22860 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
22861 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
22862 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
22863 (variable_value_hasher::hash, variable_value_hasher::equal): New
22865 (resolve_variable_value_in_expr, resolve_variable_value,
22866 resolve_variable_values, note_variable_value_in_expr,
22867 note_variable_value): New functions.
22868 (dwarf2out_early_finish): Call note_variable_value on all toplevel
22871 2017-02-24 Jakub Jelinek <jakub@redhat.com>
22874 * opts.h (handle_generated_option): Add GENERATED_P argument.
22875 * opts-common.c (handle_option): Adjust function comment.
22876 (handle_generated_option): Add GENERATED_P argument, pass it to
22878 (control_warning_option): Pass false to handle_generated_option
22880 * opts.c (maybe_default_option): Pass true to handle_generated_option
22882 * optc-gen.awk: Likewise.
22884 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22886 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
22887 a REG, look at the REG it is a SUBREG of.
22888 (splitter for cmpeqsi_t): Ditto.
22890 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22892 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
22893 the special USEs with the pattern of the insn, not the insn itself.
22895 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
22898 * doc/invoke.texi: Document -mload-store-pairs.
22900 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22901 Sandra Loosemore <sandra@codesourcery.com>
22903 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
22904 argument isn't a CONST_INT.
22905 (nios2_alternate_compare_const): Assert op is a CONST_INT.
22906 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
22907 (nios2_validate_compare): Bypass alternate compare logic if *op2
22908 is not a CONST_INT.
22909 (ldstwm_operation_p): Return false if first_base is not a REG or
22910 if first_offset is not a CONST_INT.
22912 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22914 * config/cris/cris.md: Use correct operand in a define_peephole2.
22916 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22918 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
22920 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22922 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
22923 this_insn if it is an INSN or JUMP_INSN.
22924 (force_offsettable): Look at base, not at addr.
22925 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
22926 on things that aren't necessarily CONST_INTs.
22928 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
22930 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
22931 -mfpmath=sse is the default also for x86-32 targets with SSE2
22932 instruction set when @option{-ffast-math} is enabled
22934 2017-02-24 Jeff Law <law@redhat.com>
22936 PR rtl-optimizatoin/79286
22937 * ira.c (update_equiv_regs): Drop may_trap_p exception to
22940 2017-02-24 Richard Biener <rguenther@suse.de>
22942 PR tree-optimization/79389
22943 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
22946 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
22948 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
22949 function comment to reflect reality.
22950 (loop_exits_before_overflow): Fix typo in function description.
22952 2017-02-24 Richard Biener <rguenther@suse.de>
22954 PR tree-optimization/79389
22955 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
22956 properly that a threading opportunity exists. Detect conditional
22957 copy/constant propagation opportunities.
22959 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
22961 * config/visium/visium.md (type): Add trap.
22962 (b): New mode attribute.
22963 (*btst): Rename into...
22964 (*btst<mode>): ...this and adjust.
22965 (*cbranchsi4_btst_insn): Rename into...
22966 (*cbranch<mode>4_btst_insn): ...this and adjust.
22967 (trap): New define_insn.
22969 2017-02-23 Jakub Jelinek <jakub@redhat.com>
22971 PR tree-optimization/79389
22972 * ifcvt.c (struct noce_if_info): Add rev_cond field.
22973 (noce_reversed_cond_code): New function.
22974 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
22975 reversed_comparison_code. Formatting fix.
22976 (noce_try_store_flag): Test rev_cond != NULL in addition to
22977 reversed_comparison_code.
22978 (noce_try_store_flag_constants): Likewise.
22979 (noce_try_store_flag_mask): Likewise.
22980 (noce_try_addcc): Use rev_cond if non-NULL instead of
22981 reversed_comparison_code.
22982 (noce_try_cmove_arith): Likewise. Formatting fixes.
22983 (noce_try_minmax, noce_try_abs): Clear rev_cond.
22984 (noce_find_if_block): Initialize rev_cond.
22985 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
22986 instead of false as last argument never attempt to reverse it
22989 2017-02-23 Bin Cheng <bin.cheng@arm.com>
22991 PR tree-optimization/79663
22992 * tree-predcom.c (combine_chains): Process refs in reverse order
22993 only for ZERO length chains, and add explaining comment.
22995 2017-02-23 Jeff Law <law@redhat.com>
22997 PR tree-optimization/79578
22998 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
22999 in call to operand_equal_p.
23001 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
23004 * config/i386/cpuid.h: Fix another undefined behavior.
23006 2017-02-23 Richard Biener <rguenther@suse.de>
23008 PR tree-optimization/79683
23009 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
23010 vector types for data-refs.
23012 2017-02-23 Martin Liska <mliska@suse.cz>
23014 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
23016 2017-02-23 Jakub Jelinek <jakub@redhat.com>
23018 PR middle-end/79665
23019 * internal-fn.c (get_range_pos_neg): Moved to ...
23020 * tree.c (get_range_pos_neg): ... here. No longer static.
23021 * tree.h (get_range_pos_neg): New prototype.
23022 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
23023 are known to be in between 0 and signed maximum inclusive, try to
23024 expand both unsigned and signed divmod and use the cheaper one from
23027 2017-02-22 Jeff Law <law@redhat.com>
23029 PR tree-optimization/79578
23030 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
23031 to compare base operands.
23033 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
23036 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
23037 gpc_reg_operand instead of fpr_reg_operand.
23039 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
23041 * config/mips/mips.c (mips_return_in_memory): Force FP
23042 vector types to be returned in memory for o32 ABI.
23044 2017-02-22 Jakub Jelinek <jakub@redhat.com>
23046 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
23047 instead of DW_TAG_member for static data member declarations and don't
23048 set no_linkage_name for static inline data members.
23049 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
23052 2017-02-22 Martin Liska <mliska@suse.cz>
23054 * doc/invoke.texi: Replace inequality signs with square brackets
23057 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23059 PR tree-optimization/68644
23060 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
23062 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
23065 * lra-constraints.c (simplify_operand_subreg): Handle
23066 WORD_REGISTER_OPERATIONS targets.
23068 2017-02-22 Jakub Jelinek <jakub@redhat.com>
23071 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
23072 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
23073 elimination by swapping fld*.
23075 2017-02-22 Richard Biener <rguenther@suse.de>
23077 PR tree-optimization/79673
23078 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
23079 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
23080 irrelevant address-space qualifiers and avoiding a
23081 ADDR_SPACE_CONVERT_EXPR from fold_convert.
23083 2017-02-22 Richard Biener <rguenther@suse.de>
23085 PR tree-optimization/79666
23086 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
23087 to not symbolically negate if that may introduce undefined
23090 2017-02-22 Martin Liska <mliska@suse.cz>
23093 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
23094 * data-streamer-out.c (streamer_write_gcov_count_stream):
23096 * value-prof.c (stream_out_histogram_value): Make assert more
23097 precise based on type of counter.
23099 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
23102 * config/i386/i386.md (standard_x87sse_constant_load splitter):
23103 Use nonimmediate_operand instead of memory_operand for operand 1.
23104 (float-extend standard_x87sse_constant_load splitter): Ditto.
23106 2017-02-21 Jeff Law <law@redhat.com>
23108 PR tree-optimization/79621
23109 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
23110 blocks with edges to themselves.
23112 2017-02-21 Jakub Jelinek <jakub@redhat.com>
23115 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
23116 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
23117 Use gimple_call_builtin_p.
23120 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
23121 on temporarily removed DEBUG_INSNs.
23123 PR tree-optimization/79649
23124 * tree-loop-distribution.c (classify_partition): Give up on
23125 non-generic address space loads/stores.
23127 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
23129 * doc/loop.texi (Loop manipulation): Remove nonexistent
23130 tree_ssa_loop_version from the documentation.
23131 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
23133 2017-02-21 Jakub Jelinek <jakub@redhat.com>
23136 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
23137 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
23138 * config/rs6000/rs6000.c: Include except.h.
23139 (rs6000_expand_split_stack_prologue): Call
23140 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
23142 2017-02-21 Martin Jambor <mjambor@suse.cz>
23145 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
23146 have been analyzed.
23148 2017-02-21 Martin Jambor <mjambor@suse.cz>
23150 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
23151 for backward compatibility only.
23152 * doc/invoke.texi (Option Summary): Remove all references to
23153 -fipa-cp-alignment.
23155 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
23159 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23161 * lra-constraints.c (curr_insn_transform): Handle
23162 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
23164 2017-02-21 Martin Liska <mliska@suse.cz>
23166 * config/i386/i386.opt: Replace -masm-dialect with -masm.
23168 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
23170 PR translation/79638
23171 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
23173 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
23176 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
23177 (arm_function_ok_for_sibcall): Return false for an indirect call by
23178 descriptor if all the argument registers are used.
23179 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
23180 alignment of the function.
23182 2017-02-21 Jakub Jelinek <jakub@redhat.com>
23184 PR tree-optimization/61441
23185 * simplify-rtx.c (simplify_const_unary_operation): For
23186 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
23187 the sNaN unmodified.
23189 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
23191 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
23192 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
23193 instead of SYSTEM_HEADER_DIR.
23195 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
23196 Martin Liška <mliska@suse.cz>
23198 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
23199 Fix typos and grammar, use active voice, and clarify.
23201 2017-02-20 Marek Polacek <polacek@redhat.com>
23203 PR middle-end/79537
23204 * gimplify.c (gimplify_expr): Handle unused *&&L;.
23207 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
23209 2017-02-20 Jakub Jelinek <jakub@redhat.com>
23212 * config/i386/i386.c (ix86_expand_builtin): Handle
23213 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
23214 ix86_builtins_isa[fcode].isa as a requirement of those
23215 flags and any other flag in the bitmask.
23216 (ix86_init_mmx_sse_builtins): Use 0 instead of
23217 ~OPTION_MASK_ISA_64BIT as mask.
23218 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
23219 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
23220 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
23221 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
23223 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23226 * lra-constraints.c (split_reg): Check requested split mode
23227 is supported by the register.
23229 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23231 * lra-constraints.c (simplify_operand_subreg): Remove early
23234 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23237 * lra-constraints.c (curr_insn_transform): Tighten condition
23238 for converting SUBREG reloads from OP_OUT to OP_INOUT.
23240 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23243 * lra-constraints.c (curr_insn_transform): Handle
23244 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
23246 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
23249 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
23251 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
23253 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
23256 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
23259 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23261 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
23264 2017-02-18 Jakub Jelinek <jakub@redhat.com>
23266 * final.c (last_columnnum, override_columnnum): New variables.
23267 (final_start_function): Set last_columnnum, pass it to begin_prologue
23268 hook and pass 0 to dwarf2out_begin_prologue.
23269 (final_scan_insn): Update override_columnnum. Pass last_columnnum
23270 to source_line debug hook.
23271 (notice_source_line): Compute last_columnnum and for debug_column_info
23272 return true on column changes.
23273 * debug.h (struct gcc_debug_hooks): Add column argument to
23274 source_line and begin_prologue hooks.
23275 (debug_nothing_int_charstar_int_bool): Remove prototype.
23276 (debug_nothing_int_int_charstar,
23277 debug_nothing_int_int_charstar_int_bool): New prototypes.
23278 (dwarf2out_begin_prologue): Add column argument.
23279 * debug.c (do_nothing_debug_hooks): Adjust source_line and
23280 begin_prologue hooks.
23281 (debug_nothing_int_charstar_int_bool): Remove.
23282 (debug_nothing_int_int_charstar,
23283 debug_nothing_int_int_charstar_int_bool): New functions.
23284 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
23285 through to dwarf2out_source_line.
23286 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
23287 (dwarf2out_source_line): Add column argument, emit it if requested.
23288 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
23290 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
23291 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
23292 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
23293 through to dwarf2out_begin_prologue.
23294 (vmsdbgout_source_line): Add column argument, pass it through to
23295 dwarf2out_source_line.
23296 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
23297 dbxout_source_line caller.
23298 (dbxout_source_line): Add column argument.
23300 * common.opt (gno-column-info, gcolumn-info): New options.
23301 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
23302 (check_die): Also test for multiple DW_AT_decl_column attributes.
23303 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
23304 DW_AT_decl_column if requested.
23305 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
23307 (gen_variable_die): Likewise.
23308 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
23309 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
23312 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
23313 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
23314 (ix86_handle_option): Handle OPT_m3dnowa.
23315 * doc/invoke.texi (-m3dnowa): Document.
23316 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
23317 -m3dnowa instead of -m3dnow -march=athlon.
23320 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
23321 instead of gcc_assert for K, r and R code checks. Formatting fixes.
23323 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23326 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
23327 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
23328 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
23329 generator for vsx_xxpermdi_<mode>_be.
23330 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
23331 force big-endian semantics.
23332 (vsx_xxpermdi_<mode>_be): New define_expand with same
23333 implementation as previous version of vsx_xxpermdi_<mode>.
23335 2017-02-17 Jakub Jelinek <jakub@redhat.com>
23337 PR tree-optimization/79327
23338 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
23339 variable, its initialization and use.
23341 2017-02-17 Julia Koval <julia.koval@intel.com>
23343 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
23344 (OPTION_MASK_ISA_PKU_UNSET): New.
23345 (ix86_handle_option): Handle -mrdpid.
23346 * config/i386/cpuid.h (bit_RDPID): New.
23347 * config/i386/driver-i386.c (host_detect_local_cpu):
23348 Detect RDPID feature.
23349 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
23350 * config/i386/i386-c.c (ix86_target_macros_internal):
23352 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
23353 (ix86_valid_target_attribute_inner_p): Add "rdpid".
23354 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
23355 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
23356 * config/i386/i386.md (define_insn "rdpid"): New.
23357 * config/i386/i386.opt Add -mrdpid.
23358 * config/i386/immintrin.h (_rdpid_u32): New.
23360 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
23362 PR rtl-optimization/79541
23363 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
23364 instead of transforming it into USE.
23366 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
23368 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
23369 If HONOR_SNANS (SFmode) force the input to a register.
23370 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
23371 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
23372 an frsp or similar insn.
23374 2017-02-17 Martin Liska <mliska@suse.cz>
23376 PR rtl-optimization/79577
23377 * params.def (selsched-max-sched-times): Increase minimum to 1.
23379 2017-02-17 Martin Liska <mliska@suse.cz>
23381 PR rtl-optimization/79574
23382 * gcse.c (want_to_gcse_p): Prevent integer overflow.
23384 2017-02-17 Martin Liska <mliska@suse.cz>
23386 PR tree-optimization/79529
23387 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
23388 ssa_defined_default_def_p to handle cases which are implicitly
23390 * tree-ssa.c (ssa_defined_default_def_p): New function.
23391 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
23392 which are implicitly defined.
23393 * tree-ssa.h (ssa_defined_default_def_p): Declare.
23395 2017-02-17 Richard Biener <rguenther@suse.de>
23397 PR middle-end/79576
23398 * params.def (max-ssa-name-query-depth): Limit to 10.
23400 2017-02-17 Richard Biener <rguenther@suse.de>
23402 PR tree-optimization/79552
23403 * tree-ssa-structalias.c (visit_loadstore): Properly verify
23406 2017-02-17 Richard Biener <rguenther@suse.de>
23409 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
23411 2017-02-17 Marek Polacek <polacek@redhat.com>
23413 PR middle-end/79536
23414 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
23415 (fold_negate_expr): New wrapper.
23417 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
23419 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
23420 Correct terminology and de-emphasize pre-standard behavior.
23422 2017-02-16 Alan Modra <amodra@gmail.com>
23424 PR rtl-optimization/79286
23425 * ira.c (def_dominates_uses): New function.
23426 (update_equiv_regs): Don't create an equivalence for insns that
23427 may trap where the register def does not dominate the use.
23429 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
23431 PR rtl-optimization/78127
23432 * lra.c (lra): Call lra_eliminate before finish the loop after
23435 2017-02-16 Richard Biener <rguenther@suse.de>
23437 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
23439 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
23440 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
23441 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
23442 (isl_val_int_from_wi): New function.
23443 (extract_affine_gmp): Rename to ...
23444 (extract_affine_wi): ... this, take a widest_int.
23445 (extract_affine_int): Just wrap extract_affine_wi.
23446 (add_param_constraints): Use isl_val_int_from_wi.
23447 (add_loop_constraints): Likewise, and extract_affine_wi.
23449 2017-02-15 Jeff Law <law@redhat.com>
23451 PR middle-end/79521
23452 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
23453 ira_init_register_move_cost_if_necessary.
23455 2017-02-15 Martin Sebor <msebor@redhat.com>
23457 PR middle-end/32003
23458 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
23459 removed in a prior commit.
23461 2017-02-15 Bin Cheng <bin.cheng@arm.com>
23463 PR tree-optimization/79347
23464 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
23465 counters during peeling.
23467 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
23469 * Makefile.in (site.exp): Remove "set ISLVER".
23471 2017-02-15 Jakub Jelinek <jakub@redhat.com>
23474 * real.c (real_from_integer): Call real_convert even for decimal.
23476 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
23479 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
23481 2017-02-14 Andrew Pinski <apinski@cavium.com>
23483 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
23484 cores and change the partno/implementer to be correct.
23485 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
23486 the 'B" as the implementer.
23487 * config/aarch64/aarch64-tune.md: Regenerate.
23489 2017-02-14 Carl Love <cel@us.ibm.com>
23491 * config/rs6000/rs6000.c: Add case statement entry to make the
23492 xvcvuxdsp built-in argument unsigned.
23493 * config/rs6000/vsx.md: Fix the source and return operand types so they
23494 match the instruction definitions from the ISA document. Fix typo
23495 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
23498 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
23501 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
23502 member early_clobber_alts.
23503 * lra-lives.c (reg_early_clobber_p): New.
23504 (process_bb_lives): Use it.
23505 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
23506 (debug_operand_data): Initialize early_clobber_alts.
23507 (setup_operand_alternative): Set up early_clobber_alts.
23508 (collect_non_operand_hard_regs): Ditto. Pass early clobber
23509 alternatives to new_insn_reg.
23510 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
23512 (lra_update_insn_regno_info): Pass the new arg.
23514 2017-02-14 Jakub Jelinek <jakub@redhat.com>
23516 PR middle-end/79505
23517 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
23518 (new_oacc_loop_raw): Don't clear already cleared fields.
23521 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
23522 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
23523 _mm512_prefetch_i64gather_ps): New inline functions and macros.
23525 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
23528 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
23530 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
23533 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
23534 the extra instruction to the right place to store 128-bit constant
23537 2017-02-14 Martin Sebor <msebor@redhat.com>
23539 PR middle-end/79448
23540 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
23541 warning for strings of unknown length.
23543 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
23545 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
23547 2017-02-14 Jeff Law <law@redhat.com>
23550 * ira-costs.c (scan_one_insn): Initialize register move costs
23551 for pseudos seen in USE/CLOBBER insns.
23553 PR tree-optimization/79095
23554 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
23555 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
23556 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
23557 if the operands are known to be not equal, then the resulting range
23559 (intersect_ranges): If the new range is ~[0,0] and the old range is
23560 wide, then prefer ~[0,0].
23561 * tree-vrp.c (overflow_comparison_p_1): New function.
23562 (overflow_comparison_p): New function.
23563 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
23564 if NAME is used in an overflow test.
23565 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
23566 overflow check that can be expressed as an equality test, then adjust
23567 ops to be that equality test.
23569 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23571 * config/s390/s390-builtin-types.def: Remove flags argument.
23572 * config/s390/s390.c (s390_init_builtins): Likewise.
23574 2017-02-14 Martin Liska <mliska@suse.cz>
23576 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
23577 vector. Fix trailing white spaces.
23579 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
23581 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
23584 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23586 PR rtl-optimization/68664
23587 * config/arm/arm.c (arm_sched_can_speculate_insn):
23588 New function. Declare prototype.
23589 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
23591 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23593 PR rtl-optimization/68664
23594 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
23596 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
23598 2017-02-14 Amit Pawar <amit.pawar@amd.com>
23600 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
23601 max skip bytes for function, loop and jump.
23603 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23605 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
23606 ABS_EXPR for gimple dump.
23608 2017-02-14 Jakub Jelinek <jakub@redhat.com>
23611 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
23613 PR tree-optimization/79408
23614 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
23615 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
23616 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
23617 also if rhs1 is INTEGER_CST.
23619 2017-02-14 Richard Biener <rguenther@suse.de>
23621 PR middle-end/79432
23622 * tree-into-ssa.c (insert_phi_nodes): When the function can
23623 have abnormal edges rewrite SSA names with broken use-def
23624 dominance out of SSA and register them for PHI insertion.
23626 2017-02-13 Martin Sebor <msebor@redhat.com>
23628 PR middle-end/79496
23629 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
23630 clearing info.nowrite flag when snprintf size argument is a range.
23632 2017-02-13 Jakub Jelinek <jakub@redhat.com>
23634 * cprop.c (cprop_jump): Add missing space in string literal.
23635 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
23636 (get_constraint_for_component_ref): Likewise.
23637 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
23638 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
23639 * lra-constraints.c (process_alt_operands): Likewise.
23640 * ipa-inline.c (inline_small_functions): Likewise.
23641 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
23642 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23643 * trans-mem.c (diagnose_tm_1_op): Likewise.
23644 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
23645 (grid_parallel_clauses_gridifiable): Likewise.
23647 * config/nvptx/mkoffload.c (process): Add space in between
23650 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
23651 "MOD4_SSE_REGS" and "ALL_REGS".
23653 * spellcheck.c (test_data): Add , in between "foo" and "food".
23655 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23658 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
23659 boundary crossing check and subsequent code generation agree.
23661 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23663 * config/aarch64/aarch64.c (has_memory_op): Delete.
23664 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
23667 2017-02-13 Jakub Jelinek <jakub@redhat.com>
23669 PR rtl-optimization/79388
23670 PR rtl-optimization/79450
23671 * combine.c (distribute_notes): When removing TEM_INSN for which
23672 corresponding dest has last value recorded, invalidate that last
23675 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23677 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
23678 of explicit '@'. Add missing assembly comment marker on branch costs
23681 2017-02-13 Nathan Sidwell <nathan@acm.org>
23683 * gengtype-lex.l (<in_struct>): Add '/'.
23685 2017-02-13 Martin Liska <mliska@suse.cz>
23688 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
23690 2017-02-13 Richard Biener <rguenther@suse.de>
23692 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
23694 * configure: Re-generate.
23695 * config.in: Likewise.
23696 * graphite-dependences.c: Simplify as if
23697 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
23698 * graphite-isl-ast-to-gimple.c: Likewise.
23699 * graphite-optimize-isl.c: Likewise.
23700 * graphite-poly.c: Likewise.
23701 * graphite-sese-to-poly.c: Likewise.
23702 * graphite.h: Likewise.
23703 * toplev.c: Include isl/version.h and use isl_version () for
23704 printing the ISL version.
23705 * doc/install.texi: Update ISL requirement.
23707 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
23709 * doc/standards.texi (Standards): Update reference to
23712 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
23714 * doc/extend.texi (Named Address Spaces): sourceware.org now
23716 * doc/install.texi (Binaries): Ditto.
23719 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
23721 * doc/cpp.texi: Replace "stringify"/"stringification" with C
23722 standard terminology "stringize"/"stringizing" throughout.
23723 * doc/cppinternals.texi: Likewise.
23725 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
23727 * doc/extend.texi: Fix some spelling mistakes and typos.
23728 * doc/invoke.texi: Likewise.
23730 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
23733 * params.def (inline-min-speedup) Change from 10 to 8.
23735 2017-02-11 Jakub Jelinek <jakub@redhat.com>
23737 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
23740 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
23743 * ipa-inline-analysis.c (get_minimal_bb): New function.
23744 (record_modified): Use it.
23745 (remap_edge_change_prob): Handle also ancestor functions.
23747 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
23749 * doc/contrib.texi (Contributors): Remove broken link into
23750 the Mauve CVS repository.
23752 2017-02-11 Jakub Jelinek <jakub@redhat.com>
23754 PR middle-end/79454
23755 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
23756 result computation whenever lhs doesn't have vector mode, not
23757 just when it has BLKmode.
23759 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
23761 * doc/makefile.texi (profiledbootstrap): Refer to the
23762 installation instructions only in textual form.
23764 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23767 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
23769 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
23771 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
23772 (Specific): Update mingw-w64 reference.
23774 (Specific): Remove broken link to Renesas RX processor.
23776 2017-02-10 Richard Biener <rguenther@suse.de>
23778 * toplev.c (process_options): Do not mention obsolete graphite
23779 options when printing sorry message about missing graphite support.
23780 Mention -floop-nest-optimize.
23782 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
23784 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
23785 (vtst_p16): Likewise.
23786 (vtstq_p8): Likewise.
23787 (vtstq_p16): Likewise.
23789 (vtstq_p64): Likewise.
23790 * config/arm/arm_neon.h (vgetq_lane_p64): New.
23791 (vset_lane_p64): New.
23792 (vsetq_lane_p64): New.
23794 2017-02-10 Jakub Jelinek <jakub@redhat.com>
23796 PR tree-optimization/79411
23797 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
23798 stmt operands are SSA_NAMEs used in abnormal phis.
23799 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
23802 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
23805 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
23808 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
23810 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
23812 2017-02-09 Jakub Jelinek <jakub@redhat.com>
23814 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
23818 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
23819 not arbitrary TREE_CONSTANT.
23822 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
23823 "omp declare target link" attribute unless is_global_var.
23824 * omp-offload.c (find_link_var_op): Likewise.
23826 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
23827 Chung-Lin Tang <cltang@codesourcery.com>
23829 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
23831 (gimplify_adjust_omp_clauses): Don't delete TILE.
23832 (gimplify_omp_for): Deal with TILE.
23833 * internal-fn.c (expand_GOACC_TILE): New function.
23834 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
23836 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
23837 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
23839 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
23840 avoid DIV for outermost collapse var.
23841 (expand_oacc_for): Insert tile element loop as needed. Adjust.
23842 Remove out of date comments, fix whitespace.
23843 * omp-general.c (omp_extract_for_data): Deal with tiling.
23844 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
23845 adjust OLF_DIM_BASE value.
23846 (struct omp_for_data): Add tiling field.
23847 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
23848 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
23849 for auto loops. Remove default auto determining, moved to
23850 oacc_loop_fixed_partitions.
23851 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
23852 stmts, add e_mask field.
23853 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
23854 (oacc_thread_numbers): Use oacc_dim_call.
23855 (oacc_xform_tile): New.
23856 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
23857 (finish_oacc_loop): Adjust for ifns vector.
23858 (oacc_loop_discover_walk): Append loop abstraction sites to list,
23859 add case for GOACC_TILE fns.
23860 (oacc_loop_xform_loop): Delete.
23861 (oacc_loop_process): Iterate over call list directly, and add
23862 handling for GOACC_TILE fns.
23863 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
23865 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
23866 vector partitioning to outer loops. Assign 2 partitions to loops
23867 when available. Add TILE handling.
23868 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
23869 (execite_oacc_device_lower): Process GOACC_TILE fns,
23870 ignore unknown specs.
23871 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
23872 * tree.c (omp_clause_num_ops): Adjust TILE ops.
23873 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
23875 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
23877 * configure.ac (ACX_BUGURL): Update.
23878 * configure: Regenerate.
23880 2017-02-09 Richard Biener <rguenther@suse.de>
23882 PR tree-optimization/69823
23883 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
23884 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
23886 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
23888 * config/arc/arc-c.def: Add __NPS400__ definition.
23889 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
23890 (TARGET_NPS400): Define.
23892 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
23894 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
23896 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
23897 pointer, arch_info.
23898 (arc_cpu_types): Fill the arch_info field with a pointer into the
23899 arc_arch_types table.
23900 (arc_selected_cpu): Declare.
23901 * config/arc/arc.c (arc_selected_cpu): Make global.
23902 (arc_selected_arch): Delete.
23903 (arc_base_cpu): Delete.
23904 (arc_override_options): Remove references to deleted variables,
23905 update access to arch information.
23906 (ARC_OPT): Update access to arch information.
23907 (ARC_OPTX): Likewise.
23908 * config/arc/arc.h (arc_base_cpu): Remove declaration.
23909 (TARGET_ARC600): Update access to arch information.
23910 (TARGET_ARC601): Likewise.
23911 (TARGET_ARC700): Likewise.
23912 (TARGET_EM): Likewise.
23913 (TARGET_HS): Likewise.
23914 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
23917 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
23920 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
23921 condition/operands for integer GE/LE/GEU/LEU operations.
23923 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
23925 PR translation/79397
23926 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
23929 2017-02-08 Martin Jambor <mjambor@suse.cz>
23932 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
23933 whether allocation happened.
23934 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
23935 nothing was allocated.
23937 2017-02-08 Jakub Jelinek <jakub@redhat.com>
23939 PR tree-optimization/79408
23940 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
23941 constant, but SSA_NAME with a known integer range, use the minimum
23942 of that range instead of op1 to determine if modulo can be replaced
23943 with its first operand.
23945 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23947 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
23949 2017-02-08 Richard Biener <rguenther@suse.de>
23951 PR tree-optimization/71824
23952 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
23953 Check all loops contained in the merged region.
23955 2017-02-07 Andrew Pinski <apinski@cavium.com>
23957 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
23959 2017-02-07 Andrew Pinski <apinski@cavium.com>
23961 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
23962 (thunderxt88): Likewise.
23963 (thunderxt81): Disable LSE and change v8.1 to v8.
23964 (thunderxt83): Likewise.
23966 2017-02-07 Jakub Jelinek <jakub@redhat.com>
23967 Richard Biener <rguenther@suse.de>
23969 PR middle-end/79399
23970 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
23971 type from int to size_t.
23972 * ira-costs.c (struct_costs_size): Change type from int to size_t.
23974 2017-02-07 Jakub Jelinek <jakub@redhat.com>
23976 PR rtl-optimization/79386
23977 * cprop.c (bypass_conditional_jumps): Initialize
23978 bypass_last_basic_block already before splitting bbs after
23979 unconditional traps...
23980 (bypass_conditional_jumps): ... rather than here.
23983 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
23984 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
23985 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
23986 fix -masm=intel patterns.
23988 2017-02-07 Richard Biener <rguenther@suse.de>
23990 PR tree-optimization/79256
23991 PR middle-end/79278
23992 * builtins.c (get_object_alignment_2): Use min_align_of_type
23993 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
23994 and ADJUST_FIELD_ALIGN.
23996 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
23998 * doc/tm.texi: Regenerate.
23999 * stor-layout.c (layout_decl): Adjust.
24000 (update_alignment_for_field): Likewise.
24001 (place_field): Likewise.
24002 (min_align_of_type): Likewise.
24003 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
24004 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
24005 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
24006 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
24007 * config/frv/frv.c (frv_adjust_field_align): Likewise.
24008 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
24009 * config/i386/i386.c (x86_field_alignment): Likewise.
24010 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
24011 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
24012 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
24013 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
24014 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
24015 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
24019 2017-01-30 Richard Biener <rguenther@suse.de>
24021 PR tree-optimization/79256
24022 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
24023 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
24026 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
24028 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
24029 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
24030 builtins to SImode and emit a zero-extend, if necessary.
24032 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
24034 * docs/invoke.texi (RISC-V Options): Alphabetize.
24036 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
24038 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
24041 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
24043 * config/riscv/riscv.c: New file.
24044 * common/config/riscv/riscv-common.c: Likewise.
24045 * config.gcc: Likewise.
24046 * config/riscv/constraints.md: Likewise.
24047 * config/riscv/elf.h: Likewise.
24048 * config/riscv/generic.md: Likewise.
24049 * config/riscv/linux.h: Likewise.
24050 * config/riscv/multilib-generator: Likewise.
24051 * config/riscv/peephole.md: Likewise.
24052 * config/riscv/pic.md: Likewise.
24053 * config/riscv/predicates.md: Likewise.
24054 * config/riscv/riscv-builtins.c: Likewise.
24055 * config/riscv/riscv-c.c: Likewise.
24056 * config/riscv/riscv-ftypes.def: Likewise.
24057 * config/riscv/riscv-modes.def: Likewise.
24058 * config/riscv/riscv-opts.h: Likewise.
24059 * config/riscv/riscv-protos.h: Likewise.
24060 * config/riscv/riscv.h: Likewise.
24061 * config/riscv/riscv.md: Likewise.
24062 * config/riscv/riscv.opt: Likewise.
24063 * config/riscv/sync.md: Likewise.
24064 * config/riscv/t-elf-multilib: Likewise.
24065 * config/riscv/t-linux: Likewise.
24066 * config/riscv/t-linux-multilib: Likewise.
24067 * config/riscv/t-riscv: Likewise.
24068 * configure.ac: Likewise.
24069 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
24070 Waterman as RISC-V maintainers.
24071 * doc/install.texi: Add RISC-V entries.
24072 * doc/invoke.texi: Add RISC-V options section.
24073 * doc/md.texi: Add RISC-V constraints section.
24074 * configure: Regenerated.
24076 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
24079 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
24080 false values to be constant vectors with all 0 or all 1 bits set.
24081 (vcondu<mode><mode>): Likewise.
24082 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
24084 (fpmask_comparison_operator): Update comment.
24085 (vecint_comparison_operator): New predicate.
24086 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
24087 vector conditionals when the true and false values are constant
24088 vectors with all 0 bits or all 1 bits set.
24090 2017-02-06 Martin Sebor <msebor@redhat.com>
24092 PR tree-optimization/79376
24093 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
24095 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
24097 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
24098 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
24099 to simplify split condition.
24101 2017-02-06 Jakub Jelinek <jakub@redhat.com>
24103 * omp-expand.c (oxpand_omp_atomic_fetch_op,
24104 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
24107 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
24109 PR rtl-optimization/68664
24110 * target.def (can_speculate_insn): New hook.
24111 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
24112 * doc/tm.texi: Regenerate.
24113 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
24114 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
24115 (rs6000_sched_can_speculate_insn): New function.
24117 2017-02-06 Jakub Jelinek <jakub@redhat.com>
24119 PR tree-optimization/79284
24120 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
24121 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
24122 vectorizable_mask_load_store, vectorizable_operation,
24123 vect_is_simple_cond, get_same_sized_vectype): Use it instead
24124 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
24125 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
24126 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
24127 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
24128 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
24129 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
24130 is_gimple_assign (stmt). Replace another such test with
24131 is_gimple_assign (stmt).
24133 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
24136 * config/avr/avr.c (rtl-iter.h): Include it.
24137 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
24138 (avr_legitimate_combined_insn): ...and implementation.
24140 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
24142 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
24143 * config/s390/s390.c (s390_const_operand_ok)
24144 (s390_canonicalize_comparison, s390_extract_part)
24145 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
24146 (s390_contiguous_bitmask_p, s390_rtx_costs)
24147 (legitimize_pic_address): Likewise.
24148 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
24149 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
24150 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
24151 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
24152 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
24154 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
24156 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
24157 REGNO($0) == REGNO($1).
24159 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24161 * config/s390/linux.h(SIZE_TYPE): Add comment.
24163 2017-02-06 Julian Brown <julian@codesourcery.com>
24164 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24165 Virendra Pathak <virendra.pathak@broadcom.com>
24167 * config/aarch64/aarch64-cores.def: Change the scheduler
24169 * config/aarch64/aarch64.md: Include thunderx2t99.md.
24170 * config/aarch64/thunderx2t99.md: New file.
24172 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
24174 * doc/standards.texi (Go Language): Update link to language
24177 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
24179 * tree-eh.c (lower_resx): Sanitize profile.
24180 (cleanup_empty_eh_move_lp): Likewise.
24182 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
24185 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
24187 * cfgloopmanip.h (loop_version): Update prototype.
24188 * modulo-sched.c (sms_schedule): Update call of loop_version.
24189 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
24190 * tree-parloops.c (gen_parallel_loop): Likewise.
24191 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
24192 * tree-ssa-loop-split.c (split_loop): Likewise.
24193 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
24194 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24196 2017-02-05 Martin Liska <mliska@suse.cz>
24199 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
24201 (print_operand_address): Initialize a struct to zero.
24203 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
24205 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
24206 garbage collector only in textual form.
24208 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
24210 * doc/extend.texi (x86 specific memory model extensions for
24211 transactional memory): Simplify a phrase.
24213 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
24216 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
24217 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
24218 (atomic_storedi_1): Likewise.
24220 2017-02-04 Jakub Jelinek <jakub@redhat.com>
24222 PR tree-optimization/79338
24223 * tree-parloops.c (gather_scalar_reductions): Don't call
24224 vect_analyze_loop_form for loop->inner before destroying loop's
24227 2017-02-03 Martin Sebor <msebor@redhat.com>
24229 PR tree-optimization/79327
24230 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
24231 when precision has resulted in leading zeros.
24232 (format_integer): Adjust the likely counter to assume an unknown
24233 argument that may be zero is non-zero.
24235 2017-02-03 Jason Merrill <jason@redhat.com>
24238 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
24239 avoid copying non-taken branch.
24241 2017-02-03 Jakub Jelinek <jakub@redhat.com>
24243 PR tree-optimization/79340
24244 * tree-vect-loop.c (vectorizable_reduction): Release
24245 vec_defs elements after safe_splicing them into other vectors.
24248 PR tree-optimization/79327
24249 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
24250 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
24252 (format_integer): Use wide_int_to_tree instead of build_int_cst
24253 + to_?hwi. If argmin is NULL, just set argmin and argmax to
24254 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
24255 of shortest and longest sequence.
24257 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
24259 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
24260 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
24262 2017-02-03 Walter Lee <walt@tilera.com>
24265 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
24266 after initial stackframe link reg save.
24267 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
24269 2017-02-03 Jakub Jelinek <jakub@redhat.com>
24272 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
24273 wu for stxssp alternative.
24275 2017-02-03 Martin Sebor <msebor@redhat.com>
24277 PR tree-optimization/79352
24278 * gimple-fold.c (get_range_strlen): Add argument.
24279 (get_range_strlen): Change return type to bool.
24280 (get_maxval_strlen): Pass in a dummy argument.
24281 * gimple-fold.h (get_range_strlen): Change return type to bool.
24282 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
24283 * tree.h (array_at_struct_end_p): Add argument.
24284 * tree.c (array_at_struct_end_p): Handle it.
24286 2017-02-03 Martin Liska <mliska@suse.cz>
24289 * multiple_target.c (create_dispatcher_calls): Redirect edge
24290 from a caller of a dispatcher.
24291 (expand_target_clones): Make the clones local.
24292 (ipa_target_clone): Do both target clones and resolvers.
24293 (ipa_dispatcher_calls): Remove the pass.
24294 (pass_dispatcher_calls::gate): Likewise.
24295 (make_pass_dispatcher_calls): Likewise.
24296 * passes.def (pass_target_clone): Put as very first IPA early
24299 2017-02-03 Martin Liska <mliska@suse.cz>
24301 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
24302 in case of a function with ifunc attribute.
24304 2017-02-03 Martin Liska <mliska@suse.cz>
24306 * cgraph.c (cgraph_node::dump): Dump function version info.
24307 * symtab.c (symtab_node::dump_base): Add missing new line.
24309 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
24311 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
24312 (ifcombine_ifandif): Use it.
24314 2017-02-03 Martin Liska <mliska@suse.cz>
24316 * doc/invoke.texi: Document default value for
24317 use-after-scope-direct-emission-threshold.
24319 2017-02-03 Martin Liska <mliska@suse.cz>
24321 PR tree-optimization/79339
24322 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
24323 (format_floating): Likewise.
24325 2017-02-03 Martin Liska <mliska@suse.cz>
24328 * ipa-prop.c (ipa_node_params_t::insert): Remove current
24330 (ipa_node_params_t::remove): Likewise.
24331 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
24332 initialization from removed ipa_node_params_t::insert.
24333 (ipa_node_params::~ipa_node_params): Move from removed
24334 ipa_node_params_t::release.
24335 * symbol-summary.h (symbol_summary::m_released): New member.
24336 Do not release a summary twice. Do not allow to call finalizer
24337 for types of a summary that live in GGC memory.
24339 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24341 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
24344 2017-02-02 Martin Sebor <msebor@redhat.com>
24346 PR middle-end/79275
24347 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
24348 (format_string): Tighten up the range of output for non-constant
24349 strings and correct the expected range for wide non-constant strings.
24351 2017-02-02 Martin Sebor <msebor@redhat.com>
24353 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
24355 PR middle-end/32003
24356 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
24358 (-fdump-tree-@var): Add to index and document how to come up
24359 with pass-specific option and dump file names.
24360 (-fdump-passes): Clarify where to look for output.
24362 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
24364 PR middle-end/77445
24365 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
24366 statistics of the analyzed path; allow threading for speed when
24367 any of BBs along the path are optimized for speed.
24369 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
24371 PR middle-end/78468
24372 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
24373 settings of the virtual registers.
24376 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
24378 * explow.c (get_dynamic_stack_size): Take known alignment of stack
24379 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
24382 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24384 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
24385 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
24387 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24389 * config/s390/s390.md: Add missing comments with the expanded
24391 * config/s390/vector.md: Likewise.
24392 * config/s390/vx-builtins.md: Likewise.
24394 2017-02-02 Jakub Jelinek <jakub@redhat.com>
24397 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
24398 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
24399 conditions on a single line.
24401 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24403 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
24404 __S390_VX__ to __VX__.
24406 2017-02-01 Andrew Pinski <apinski@cavium.com>
24408 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
24409 stmt_info to record_stmt_cost.
24410 (vect_get_known_peeling_cost): Pass stmt_info if known to
24412 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
24413 cpu_vector_cost field into
24414 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
24415 field into vec_int_stmt_cost and vec_fp_stmt_cost.
24416 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
24417 splitting of scalar_stmt_cost and vec_stmt_cost.
24418 (thunderx_vector_cost): Likewise.
24419 (cortexa57_vector_cost): LIkewise.
24420 (exynosm1_vector_cost): Likewise.
24421 (xgene1_vector_cost): Likewise.
24422 (thunderx2t99_vector_cost): Improve after the splitting of the two
24424 (aarch64_builtin_vectorization_cost): Update for the splitting of
24425 scalar_stmt_cost and vec_stmt_cost.
24427 2017-02-01 Torvald Riegel <triegel@redhat.com>
24428 Richard Henderson <rth@redhat.com>
24430 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
24431 conditional on existance of a fast atomic load.
24432 * optabs-query.c (can_atomic_load_p): New function.
24433 * optabs-query.h (can_atomic_load_p): Declare it.
24434 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
24435 no fast atomic load is available for the particular size of access.
24436 (expand_atomic_compare_and_swap): Likewise.
24437 (expand_atomic_load): Likewise.
24438 (expand_atomic_store): Likewise.
24439 (expand_atomic_fetch_op): Likewise.
24440 * testsuite/lib/target-supports.exp
24441 (check_effective_target_sync_int_128): Remove x86 because it provides
24442 no fast atomic load.
24443 (check_effective_target_sync_int_128_runtime): Likewise.
24445 2017-02-01 Richard Biener <rguenther@suse.de>
24447 * graphite.c: Include tree-vectorizer.h for find_loop_location.
24448 (graphite_transform_loops): Provide opt-info for optimized nests.
24449 * tree-parloop.c (parallelize_loops): Provide opt-info for
24450 parallelized loops.
24452 2017-02-01 Richard Biener <rguenther@suse.de>
24454 PR middle-end/79315
24455 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
24456 was not set before.
24458 2017-02-01 Richard Biener <rguenther@suse.de>
24460 PR tree-optimization/71824
24461 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
24462 Verify the loops are valid in the merged SESE region.
24463 (scop_detection::can_represent_loop_1): Check analyzing the
24464 evolution of the number of iterations in the region succeeds.
24466 2017-01-31 Ian Lance Taylor <iant@golang.org>
24468 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
24469 REG_ARGS_SIZE note to 32-bit push insns and call insn.
24471 2017-01-31 David Malcolm <dmalcolm@redhat.com>
24473 PR preprocessor/79210
24474 * input.c (get_substring_ranges_for_loc): Replace line_width
24475 assertion with error-handling.
24477 2017-01-31 Richard Biener <rguenther@suse.de>
24479 PR tree-optimization/77318
24480 * graphite-sese-to-poly.c (extract_affine): Fix assert.
24481 (create_pw_aff_from_tree): Take loop parameter.
24482 (add_condition_to_pbb): Pass loop of the condition to
24483 create_pw_aff_from_tree.
24485 2017-01-31 Jakub Jelinek <jakub@redhat.com>
24487 * config/s390/s390.c (s390_asan_shadow_offset): New function.
24488 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
24490 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
24494 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
24496 (convert_int_to_float128): Likewise.
24497 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
24498 (convert_int_to_float128): Likewise.
24499 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
24500 (UNSPEC_IEEE128_CONVERT): Likewise.
24501 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
24502 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
24503 Use local variables for IBM extended format.
24504 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
24505 (fix_trunc<mode>si2_fprs): Likewise.
24506 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
24507 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
24508 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
24509 to know that we can now have integers of all sizes in vector
24511 (fix<uns>_<mode>di2_hw): Likewise.
24512 (float<uns>_<mode>si2_hw): Likewise.
24513 (fix_<mode>si2_hw): Likewise.
24514 (fixuns_<mode>si2_hw): Likewise.
24515 (float<uns>_<mode>di2_hw): Likewise.
24516 (float_<mode>di2_hw): Likewise.
24517 (float_<mode>si2_hw): Likewise.
24518 (floatuns_<mode>di2_hw): Likewise.
24519 (floatuns_<mode>si2_hw): Likewise.
24520 (xscvqp<su>wz_<mode>): Delete, no longer used.
24521 (xscvqp<su>dz_<mode>): Likewise.
24522 (xscv<su>dqp_<mode>): Likewise.
24523 (ieee128_mfvsrd_64bit): Likewise.
24524 (ieee128_mfvsrd_32bit): Likewise.
24525 (ieee128_mfvsrwz): Likewise.
24526 (ieee128_mtvsrw): Likewise.
24527 (ieee128_mtvsrd_64bit): Likewise.
24528 (ieee128_mtvsrd_32bit): Likewise.
24530 2017-01-31 Martin Liska <mliska@suse.cz>
24533 * ipa-prop.c (ipa_free_all_node_params): Call release method
24534 instead of ~sumbol_summary to not to trigger double times
24537 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
24539 PR tree-optimization/71691
24540 * bitmap.h (class auto_bitmap): New.
24541 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
24542 is_maybe_undefined instead of ssa_undefined_value_p.
24544 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24546 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
24547 __S390_ARCH_LEVEL__ to __ARCH__.
24549 2017-01-31 Jakub Jelinek <jakub@redhat.com>
24551 PR tree-optimization/79267
24552 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
24553 if should_remove_lhs_p is true.
24555 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
24558 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
24559 (add_alignment_attribute): New.
24560 (base_type_die): Add alignment attribute.
24561 (subrange_type_die): Likewise.
24562 (modified_type_die): Likewise.
24563 (gen_array_type_die): Likewise.
24564 (gen_descr_array_type_die: Likewise.
24565 (gen_enumeration_type_die): Likewise.
24566 (gen_subprogram_die): Likewise.
24567 (gen_variable_die): Likewise.
24568 (gen_field_die): Likewise.
24569 (gen_ptr_to_mbr_type_die): Likewise.
24570 (gen_struct_or_union_type_die): Likewise.
24571 (gen_subroutine_type_die): Likewise.
24572 (gen_typedef_die): Likewise.
24573 (base_type_cmp): Compare alignment attribute.
24575 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24578 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
24579 (setb_unsigned) New pattern for setb with CCUNS.
24580 * config/rs6000/rs6000.c (expand_block_compare): Use a different
24581 subfc./subfe sequence to avoid overflow problems. Generate a
24582 shorter sequence with cmpld/setb for power9.
24583 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
24584 for generating subfc. instruction.
24585 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
24586 now uses this instruction.
24588 2017-01-30 Ian Lance Taylor <iant@google.com>
24591 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
24592 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
24594 2017-01-30 Martin Sebor <msebor@redhat.com>
24596 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
24597 Move constant to the right of a relational operator.
24598 (get_mpfr_format_length, format_character, format_string): Ditto.
24599 (should_warn_p, maybe_warn): Same.
24601 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
24603 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
24606 * asan.c (get_translation_unit_decl): Remove function.
24607 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
24609 2017-01-30 Martin Liska <mliska@suse.cz>
24611 PR gcov-profile/79259
24612 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
24613 -fprofile-generate.
24615 2017-01-30 Martin Liska <mliska@suse.cz>
24618 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
24619 Initialize variables with NULL value.
24621 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
24624 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
24626 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
24628 2017-01-30 Richard Biener <rguenther@suse.de>
24630 PR tree-optimization/79276
24631 * tree-vrp.c (process_assert_insertions): Properly adjust common
24632 when removing a duplicate.
24634 * gcc.dg/torture/pr79276.c: New testcase.
24636 2017-01-30 Richard Biener <rguenther@suse.de>
24638 PR tree-optimization/79256
24639 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
24640 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
24642 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
24644 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
24647 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
24648 ("*r<noxa>sbg_<mode>_sll_bitmask")
24649 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
24650 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
24651 Use contiguous_bitmask_nowrap_operand.
24653 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24656 * config/rs6000/altivec.h (vec_xl): Revise #define.
24657 (vec_xst): Likewise.
24659 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
24661 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
24663 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
24665 PR rtl-optimization/79194
24666 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
24667 traps before call to bypass_conditional_jumps.
24669 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
24671 PR tree-optimization/71374
24672 * lra-constraints.c (check_conflict_input_operands): New.
24673 (match_reload): Use it.
24675 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
24678 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
24679 account to calculate conflict_set.
24681 2017-01-27 Bin Cheng <bin.cheng@arm.com>
24683 PR rtl-optimization/78559
24684 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
24685 other_insn in combine.
24687 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24689 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
24690 uint16_type_node for BT_UINT16.
24692 2017-01-27 David Malcolm <dmalcolm@redhat.com>
24694 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
24695 "RTL Tests" to menu.
24696 (GIMPLE Tests): New node.
24697 (RTL Tests): New node.
24699 2017-01-27 Richard Biener <rguenther@suse.de>
24701 PR tree-optimization/79245
24702 * tree-loop-distribution.c (distribute_loop): Apply cost
24703 modeling also to detected patterns.
24705 2017-01-27 Richard Biener <rguenther@suse.de>
24707 PR tree-optimization/71433
24708 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
24709 (compare_assert_loc): New function.
24710 (process_assert_insertions): Sort and optimize assert locations
24711 to remove duplicates and push down identical assertions on
24712 edges to their destination block.
24714 2017-01-27 Richard Biener <rguenther@suse.de>
24716 PR tree-optimization/79244
24717 * tree-vrp.c (remove_range_assertions): Forcefully propagate
24718 out SSA names even if abnormal.
24720 2017-01-27 Jakub Jelinek <jakub@redhat.com>
24722 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
24723 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
24724 instead of MPFR_RNDN.
24726 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
24729 * arm.c (arm_option_override): Don't call build_target_option_node
24730 until after doing all option overrides.
24731 (arm_valid_target_attribute_tree): Likewise.
24733 2017-01-27 Martin Liska <mliska@suse.cz>
24735 * doc/invoke.texi (-fprofile-arcs): Document profiling support
24736 for {cd}tors and C++ {cd}tors.
24738 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
24740 * config/s390/s390.md ("*setmem_long_and")
24741 ("*setmem_long_and_31z"): Use zero_extend instead of and.
24743 2017-01-26 Martin Sebor <msebor@redhat.com>
24745 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
24748 2017-01-26 Martin Sebor <msebor@redhat.com>
24750 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
24751 HAVE_DFmode before using XFmode or DFmode.
24752 (parse_directive): Avoid using the z length modifier to avoid
24753 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
24755 PR middle-end/78703
24756 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
24757 to accept adjustment as an array.
24758 (get_int_range): New function.
24759 (struct directive): Make width and prec arrays.
24760 (directive::set_width, directive::set_precision): Call get_int_range.
24761 (format_integer, format_floating): Handle width and precision ranges.
24762 (format_string, parse_directive): Same.
24764 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24767 * dwarf2out.c (generate_skeleton_bottom_up): For children with
24768 comdat_type_p set, just clone them, but keep the children in the
24772 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
24773 which have direct callers with -fvar-tracking-assignments enabled
24775 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
24776 inside of type units.
24778 2017-01-26 Martin Sebor <msebor@redhat.com>
24780 PR middle-end/78703
24781 * gimple-ssa-sprintf.c (struct result_range): Add likely and
24783 (struct format_result): Replace number_chars, number_chars_min,
24784 and number_chars_max with a single member of struct result_range.
24786 (format_result::operator+=): Adjust.
24787 (struct fmtresult): Remove bounded. Handle likely and unlikely
24789 (fmtresult::adjust_for_width_or_precision): New function.
24790 (fmtresult:type_max_digits): New function.
24791 (bytes_remaining): Handle likely and unlikely counters.
24792 (min_bytes_remaining): Remove.
24793 (format_percent): Simplify.
24794 (format_integer, format_floating): Set likely and unlikely counters.
24795 (get_string_length, format_character, format_string): Same.
24796 (format_plain, should_warn_p): New function.
24797 (maybe_warn): Call should_warn_p. Update diagnostic messages
24798 and handle those for all directives, including plain strings.
24799 (format_directive): Handle likely and unlikely counters.
24800 Remove unnecessary quoting from diagnostics. Add an informational
24802 (add_bytes): Remove.
24803 (pass_sprintf_length::compute_format_length): Simplify.
24804 (try_substitute_return_value): Handle likely and unlikely counters.
24806 2017-01-26 Carl Love <cel@us.ibm.com>
24808 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
24809 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
24811 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
24814 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
24815 endianess for subregs into account.
24816 * lra-constraints.c (lra_constraints): Do risky transformations
24817 always on the first iteration.
24818 * lra-lives.c (check_pseudos_live_through_calls): Add arg
24819 last_call_used_reg_set.
24820 (process_bb_lives): Define and use last_call_used_reg_set.
24821 * lra.c (lra): Always continue after lra_constraints on the first
24824 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
24826 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
24828 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
24830 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24832 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
24833 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
24834 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
24835 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
24836 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
24837 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
24838 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
24839 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
24840 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
24842 2017-01-26 Marek Polacek <polacek@redhat.com>
24845 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
24846 for the third operand.
24848 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24850 PR middle-end/79236
24851 * omp-low.c (struct omp_context): Add simt_stmt field.
24852 (scan_omp_for): Return omp_context *.
24853 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
24854 context to the _simt_ SIMD stmt.
24855 (lower_omp_for): For combined SIMD with sibling _simt_
24856 SIMD, make sure to use the same decls in _looptemp_
24857 clauses as in the sibling.
24859 2017-01-26 David Sherwood <david.sherwood@arm.com>
24861 PR middle-end/79212
24862 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
24865 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24868 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
24869 emit fld b; fld a; if possible.
24871 * brig-builtins.def: Update copyright years.
24872 * config/arm/arm_acle_builtins.def: Update copyright years.
24874 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
24877 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
24878 constraint instead of o for the stxsd instruction.
24880 2017-01-25 Carl Love <cel@us.ibm.com>
24882 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
24883 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
24885 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
24887 * doc/invoke.texi (C++ Dialect Options): Fix typo.
24889 2017-01-25 Richard Biener <rguenther@suse.de>
24891 PR tree-optimization/69264
24892 * target.def (vector_alignment_reachable): Improve documentation.
24893 * doc/tm.texi: Regenerate.
24894 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
24896 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
24897 earlier changes with respect to TYPE_USER_ALIGN.
24898 (vector_alignment_reachable_p): Likewise. Improve dumping.
24900 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24903 * config/arm/arm.md (xordi3): Force constant operand into a register
24906 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24908 * doc/invoke.texi (-fstore-merging): Correct default optimization
24909 levels at which it is enabled.
24910 (-O): Move -fstore-merging from list to...
24913 2017-01-25 Richard Biener <rguenther@suse.de>
24916 * omp-expand.c: Include debug.h.
24917 (expand_omp_taskreg): Make sure to generate early debug before
24918 outlining anything from a function.
24919 (expand_omp_target): Likewise.
24920 (grid_expand_target_grid_body): Likewise.
24922 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
24925 * asan.c (get_translation_unit_decl): New function.
24926 (asan_add_global): Extract modules file name from globals
24927 TRANSLATION_UNIT_DECL name.
24929 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
24932 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
24933 for long calls with APCS frame and VFP.
24935 2017-01-24 David Malcolm <dmalcolm@redhat.com>
24937 * cfg.c (original_copy_tables_initialized_p): New function.
24938 * cfg.h (original_copy_tables_initialized_p): New decl.
24939 * cfgrtl.c (relink_block_chain): Guard the call to
24940 free_original_copy_tables with a call to
24941 original_copy_tables_initialized_p.
24942 * cgraph.h (symtab_node::native_rtl_p): New decl.
24943 * cgraphunit.c (symtab_node::native_rtl_p): New function.
24944 (symtab_node::needed_p): Don't assert for early assembly output
24945 for __RTL functions.
24946 (cgraph_node::finalize_function): Set "force_output" for __RTL
24948 (cgraph_node::analyze): Bail out early for __RTL functions.
24949 (analyze_functions): Update assertion to support __RTL functions.
24950 (cgraph_node::expand): Bail out early for __RTL functions.
24951 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
24953 * function.h (struct function): Update comment for field
24955 * gimple-expr.c: Include "tree-pass.h".
24956 (gimple_has_body_p): Return false for __RTL functions.
24957 * Makefile.in (OBJS): Add run-rtl-passes.o.
24958 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
24960 (gcc::pass_manager::get_clean_slate): New accessor.
24961 * passes.c: Include "insn-addr.h".
24962 (should_skip_pass_p): Add logging. Update logic for running
24963 "expand" to be compatible with both __GIMPLE and __RTL. Guard
24964 property-provider override so it is only done for gimple passes.
24966 (skip_pass): New function.
24967 (execute_one_pass): Call skip_pass when skipping passes.
24968 * read-md.c (md_reader::read_char): Support filtering
24969 the input to a subset of line numbers.
24970 (md_reader::md_reader): Initialize fields
24971 m_first_line and m_last_line.
24972 (md_reader::read_file_fragment): New function.
24973 * read-md.h (md_reader::read_file_fragment): New decl.
24974 (md_reader::m_first_line): New field.
24975 (md_reader::m_last_line): New field.
24976 * read-rtl-function.c (function_reader::create_function): Only
24977 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
24978 curr_properties. Set DECL_INITIAL to a dummy block.
24979 (read_rtl_function_body_from_file_range): New function.
24980 * read-rtl-function.h (read_rtl_function_body_from_file_range):
24982 * run-rtl-passes.c: New file.
24983 * run-rtl-passes.h: New file.
24985 2017-01-24 Jeff Law <law@redhat.com>
24987 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
24990 2017-01-24 Bin Cheng <bin.cheng@arm.com>
24992 PR tree-optimization/79159
24993 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
24994 (record_nonwrapping_iv): Improve boundary using above function if no
24995 value range information.
24997 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
24998 Martin Jambor <mjambor@suse.cz>
25000 * brig-builtins.def: New file.
25001 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
25002 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
25003 (DEF_HSAIL_SAT_BUILTIN): Likewise.
25004 (DEF_HSAIL_INTR_BUILTIN): Likewise.
25005 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
25006 * builtin-types.def (BT_INT8): New.
25007 (BT_INT16): Likewise.
25008 (BT_UINT8): Likewise.
25009 (BT_UINT16): Likewise.
25010 (BT_FN_ULONG): Likewise.
25011 (BT_FN_UINT_INT): Likewise.
25012 (BT_FN_UINT_ULONG): Likewise.
25013 (BT_FN_UINT_LONG): Likewise.
25014 (BT_FN_UINT_PTR): Likewise.
25015 (BT_FN_ULONG_PTR): Likewise.
25016 (BT_FN_INT8_FLOAT): Likewise.
25017 (BT_FN_INT16_FLOAT): Likewise.
25018 (BT_FN_UINT32_FLOAT): Likewise.
25019 (BT_FN_UINT16_FLOAT): Likewise.
25020 (BT_FN_UINT8_FLOAT): Likewise.
25021 (BT_FN_UINT64_FLOAT): Likewise.
25022 (BT_FN_UINT16_UINT32): Likewise.
25023 (BT_FN_UINT32_UINT16): Likewise.
25024 (BT_FN_UINT16_UINT16_UINT16): Likewise.
25025 (BT_FN_INT_PTR_INT): Likewise.
25026 (BT_FN_UINT_PTR_UINT): Likewise.
25027 (BT_FN_LONG_PTR_LONG): Likewise.
25028 (BT_FN_ULONG_PTR_ULONG): Likewise.
25029 (BT_FN_VOID_UINT64_UINT64): Likewise.
25030 (BT_FN_UINT8_UINT8_UINT8): Likewise.
25031 (BT_FN_INT8_INT8_INT8): Likewise.
25032 (BT_FN_INT16_INT16_INT16): Likewise.
25033 (BT_FN_INT_INT_INT): Likewise.
25034 (BT_FN_UINT_FLOAT_UINT): Likewise.
25035 (BT_FN_FLOAT_UINT_UINT): Likewise.
25036 (BT_FN_ULONG_UINT_UINT): Likewise.
25037 (BT_FN_ULONG_UINT_PTR): Likewise.
25038 (BT_FN_ULONG_ULONG_ULONG): Likewise.
25039 (BT_FN_UINT_UINT_UINT): Likewise.
25040 (BT_FN_VOID_UINT_PTR): Likewise.
25041 (BT_FN_UINT_UINT_PTR: Likewise.
25042 (BT_FN_UINT32_UINT64_PTR): Likewise.
25043 (BT_FN_INT_INT_UINT_UINT): Likewise.
25044 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
25045 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
25046 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
25047 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
25048 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
25049 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
25050 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
25051 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
25052 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
25053 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
25054 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
25055 * doc/frontends.texi: List BRIG FE.
25056 * doc/install.texi (Testing): Add BRIG tesring requirements.
25057 * doc/invoke.texi (Overall Options): Mention BRIG.
25058 * doc/standards.texi (Standards): Doucment BRIG HSA version.
25060 2017-01-24 Richard Biener <rguenther@suse.de>
25062 PR translation/79208
25063 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
25065 2017-01-24 Martin Jambor <mjambor@suse.cz>
25068 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
25069 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
25070 and known_contexts.
25072 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
25074 PR middle-end/79123
25075 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
25076 casts from signed to unsigned really don't have a range.
25078 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
25080 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
25081 GMP_RNDx for compatiblity.
25083 2017-01-24 Martin Liska <mliska@suse.cz>
25086 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
25087 that would prevent us to call alloca with -1 as argument.
25089 2017-01-24 Jakub Jelinek <jakub@redhat.com>
25091 * dwarf2out.c (output_compilation_unit_header, output_file_names):
25092 Avoid -Wformat-security warning.
25094 2017-01-23 Andrew Pinski <apinski@cavium.com>
25096 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
25099 2017-01-23 Martin Sebor <msebor@redhat.com>
25101 PR middle-end/78703
25102 * gimple-ssa-sprintf.c (warn_level): New global.
25103 (format_integer): Use it here and throughout the rest of the file.
25104 Use the same switch to compute sign as base.
25105 (maybe_warn): New function.
25106 (format_directive): Factor out warnings into maybe_warn.
25107 Add debugging output. Use warn_level.
25108 (add_bytes): Use warn_level.
25109 (pass_sprintf_length::compute_format_length): Add debugging output.
25110 (try_substitute_return_value): Same.
25111 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
25113 PR middle-end/78703
25114 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
25115 (struct fmtresult, format_integer, format_floating): Adjust.
25116 (fmtresult::fmtresult): Set max correctly in two argument ctor.
25117 (get_string_length, format_string,format_directive): Same.
25118 (pass_sprintf_length::compute_format_length): Same.
25119 (try_substitute_return_value): Simplify slightly.
25121 PR middle-end/78703
25122 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
25123 (fmtresult::operator+=): Outlined.
25124 (struct fmtresult): Add ctors.
25125 (struct conversion_spec): Rename...
25126 (struct directive): ...to this. Add and remove data members.
25127 (directive::set_width, directive::set_precision): New functions.
25128 (format_percent): Use fmtresult ctor.
25129 (get_width_and_precision): Remove.
25130 (format_integer): Make naming changes. Avoid computing width and
25132 (format_floating): Same. Adjust indentation.
25133 (format_character, format_none): New functions.
25134 (format_string): Moved character handling to format_character.
25135 (format_directive): Remove arguments, change return type.
25136 (parse_directive): New function.
25137 (pass_sprintf_length::compute_format_length): Move directive
25138 parsing to parse_directive.
25140 2017-01-23 Jakub Jelinek <jakub@redhat.com>
25142 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
25143 (assign_assembler_name_if_needed): ... this.
25144 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
25145 (assign_assembler_name_if_needed): ... this.
25146 (free_lang_data_in_cgraph): Adjust callers.
25147 * cgraphunit.c (cgraph_node::analyze): Likewise.
25148 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
25151 2017-01-23 Richard Biener <rguenther@suse.de>
25153 PR tree-optimization/79088
25154 PR tree-optimization/79188
25155 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
25156 resetting loop bounds after last path deletion. Reset loop
25157 bounds of the target loop, make code match the comments.
25158 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
25159 Make sure loops need no fixups.
25161 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
25163 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
25164 exponent support with double type for first argument.
25165 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
25166 type returned by __builtin_vec_extract_sig,
25167 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
25168 functions from "vector int" to "vector unsigned int" or from
25169 "vector long long int" to "vector unsigned long long int".
25170 Changed type returned by __builtin_vec_extract_exp,
25171 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
25172 functions from "vector int" to "vector unsigned int" or from
25173 "vector long long int" to "vector unsigned long long int".
25174 Changed return type of __builtin_vec_test_data_class,
25175 __builtin_vec_test_data_class_sp, and
25176 __builtin_vec_test_data_class_dp from "vector int" to
25177 "vector bool int" or from "vector long long int" to "vector bool
25178 long long int" and changed second argument type from "unsigned
25179 int" to "int". Added new overloaded function forms "vector float
25180 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
25181 "vector float __builtin_vec_insert_exp_sp (vector float, vector
25182 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
25183 double, vector unsigned long long int)" and "vector double
25184 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
25185 long int)". Changed return type of
25186 __builtin_scalar_test_data_class and
25187 __builtin_scalar_test_data_class_sp and
25188 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
25189 int" and changed second argument from "unsigned int" to "int".
25190 Changed type returned by __builtin_scalar_test_neg,
25191 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
25192 from "int" to "bool int". Added new overloaded function form
25193 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
25194 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
25195 exponent double-precision with floating point first argument.
25196 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
25197 documentation of scalar_test_data_class, scalar_test_neg,
25198 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
25199 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
25200 vec_test_data_class built-in functions to reflect refinements in
25201 their type signatures.
25203 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
25205 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
25207 (aarch64_elf_asm_destructor): Likewise.
25209 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
25211 PR rtl-optimization/78634
25212 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
25213 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
25214 * ifcvt.c (noce_try_cmove): Add missing cost check.
25216 PR rtl-optimization/71724
25217 * combine.c (if_then_else_cond): Look for situations where it is
25218 beneficial to undo the work of one of the recursive calls.
25220 2017-01-23 Bin Cheng <bin.cheng@arm.com>
25222 PR tree-optimization/70754
25223 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
25224 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
25225 combined stmt before it if not NULL.
25226 (combine_chains): Process refs reversely and compute dominance point
25229 2017-01-23 Martin Liska <mliska@suse.cz>
25231 PR tree-optimization/79196
25232 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
25233 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
25235 (strlen_optimize_stmt): Call the renamed function.
25237 2017-01-23 Michael Matz <matz@suse.de>
25239 PR tree-optimization/78384
25240 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
25242 2017-01-23 Richard Biener <rguenther@suse.de>
25244 PR tree-optimization/79186
25245 * tree-vrp.c (register_new_assert_for): Make sure we've seen
25246 both incoming edges before moving an assert.
25248 2017-01-23 Martin Jambor <mjambor@suse.cz>
25250 * ipa-prop.c (load_from_param_1): Removed.
25251 (load_from_unmodified_param): Bits from load_from_param_1 put back
25253 (load_from_param): Removed.
25254 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
25255 with stmt. Reverted back to use of load_from_unmodified_param.
25257 2017-01-23 Martin Jambor <mjambor@suse.cz>
25260 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
25261 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
25262 field a pointer to garbage collected vector, mark lattices and
25263 ipcp_orig_node with GTY((skip)).
25264 (ipa_get_param_count): Adjust to descriptors being a pointer.
25265 (ipa_get_param): Likewise.
25266 (ipa_get_type): Likewise.
25267 (ipa_get_param_move_cost): Likewise.
25268 (ipa_set_param_used): Likewise.
25269 (ipa_get_controlled_uses): Likewise.
25270 (ipa_set_controlled_uses): Likewise.
25271 (ipa_is_param_used): Likewise.
25272 (ipa_node_params_t): Move into garbage collector. New methods insert
25274 (ipa_node_params_sum): Annotate wth GTY(()).
25275 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
25277 (ipa_load_from_parm_agg): Adjust declaration.
25278 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
25279 * ipa-profile.c (ipa_profile): Likewise.
25280 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
25281 (ipa_populate_param_decls): Make descriptors parameter garbage
25283 (ipa_dump_param): Adjust to descriptors being a pointer.
25284 (ipa_alloc_node_params): Likewise.
25285 (ipa_initialize_node_params): Likewise.
25286 (load_from_param_1): Make descriptors parameter garbage collected.
25287 (load_from_unmodified_param): Likewise.
25288 (load_from_param): Likewise.
25289 (ipa_load_from_parm_agg): Likewise.
25290 (ipa_node_params::~ipa_node_params): Removed.
25291 (ipa_free_all_node_params): Remove call to delete operator.
25292 (ipa_node_params_t::insert): New.
25293 (ipa_node_params_t::remove): Likewise.
25294 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
25295 copy known_csts and known_contexts vectors.
25296 (ipa_read_node_info): Adjust to descriptors being a pointer.
25297 (ipcp_modif_dom_walker): Make m_descriptors field garbage
25299 (ipcp_transform_function): Make descriptors variable garbage
25302 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
25304 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
25305 * config/i386/avx512dqintrin.h: Ditto.
25306 * config/i386/avx512fintrin.h: Ditto.
25307 * config/i386/i386.c: Handle new builtins.
25308 * config/i386/i386-builtin.def: Add new builtins.
25309 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
25310 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
25312 2017-01-23 Jakub Jelinek <jakub@redhat.com>
25313 Martin Liska <mliska@suse.cz>
25315 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
25316 * asan.c (asan_expand_poison_ifn): Support stores and use
25317 appropriate ASAN report function.
25318 * internal-fn.c (expand_ASAN_POISON_USE): New function.
25319 * internal-fn.def (ASAN_POISON_USE): Declare.
25320 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
25321 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
25322 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
25323 ASAN_POISON calls w/o LHS.
25324 * tree-ssa.c (execute_update_addresses_taken): Create clobber
25325 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
25326 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
25327 * gimplify.c (asan_poison_variables): Add attribute
25328 use_after_scope_memory to variables that really needs to live
25330 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
25331 having the attribute.
25333 2017-01-23 Martin Liska <mliska@suse.cz>
25335 * asan.c (create_asan_shadow_var): New function.
25336 (asan_expand_poison_ifn): Likewise.
25337 * asan.h (asan_expand_poison_ifn): New declaration.
25338 * internal-fn.c (expand_ASAN_POISON): Likewise.
25339 * internal-fn.def (ASAN_POISON): New builtin.
25340 * sanopt.c (pass_sanopt::execute): Expand
25341 asan_expand_poison_ifn.
25342 * tree-inline.c (copy_decl_for_dup_finish): Make function
25344 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
25345 * tree-ssa.c (is_asan_mark_p): New function.
25346 (execute_update_addresses_taken): Rewrite local variables
25347 (identified just by use-after-scope as addressable) into SSA.
25349 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
25351 * doc/install.texi (Specific): opensource.apple.com uses https
25352 now. Remove trailing slash.
25354 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
25356 * README.Portability: Remove note on an Irix compatibility issue.
25358 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
25360 * gcov.c (INCLUDE_ALGORITHM): Define.
25361 (INCLUDE_VECTOR): Define.
25362 No longer include <vector> and <algorithm> directly.
25364 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
25366 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
25368 * doc/invoke.texi (Code Gen Options): Ditto.
25370 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
25373 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
25375 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
25377 rtl-optimization/79125
25378 * cprop.c (local_cprop_pass): Handle cases where we make an
25379 unconditional trap.
25381 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
25385 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
25386 read from, for big endian.
25388 2017-01-20 Jiong Wang <jiong.wang@arm.com>
25390 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
25391 register pauth builtins for LP64 only.
25393 2017-01-20 Marek Polacek <polacek@redhat.com>
25396 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
25399 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
25401 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
25403 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
25404 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
25405 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
25407 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25410 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
25411 in big-endian mode when they are not a single duplicated value.
25413 2017-01-20 Richard Biener <rguenther@suse.de>
25415 * BASE-VER: Bump to 7.0.1.
25417 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
25419 * omp-low.c (omplow_simd_context): New struct. Use it...
25420 (lower_rec_simd_input_clauses): ...here and...
25421 (lower_rec_input_clauses): ...here to hold common data. Adjust all
25422 references to idx, lane, max_vf, is_simt.
25424 2017-01-20 Graham Markall <graham.markall@embecosm.com>
25426 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
25429 2017-01-20 Martin Jambor <mjambor@suse.cz>
25431 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
25432 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
25434 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
25435 (GTFILES): Rename hsa.c to hsa-common.c.
25436 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
25437 * hsa-dump.c: Likewise.
25438 * hsa-gen.c: Likewise.
25439 * hsa-regalloc.c: Likewise.
25440 * ipa-hsa.c: Likewise.
25441 * omp-expand.c: Likewise.
25442 * omp-low.c: Likewise.
25443 * toplev.c: Likewise.
25445 2017-01-20 Marek Polacek <polacek@redhat.com>
25448 * doc/invoke.texi: Document -Wduplicated-branches.
25449 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
25450 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
25451 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
25452 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
25453 return 0 only when not OEP_LEXICOGRAPHIC.
25454 (fold_build_cleanup_point_expr): Use the expression
25455 location when building CLEANUP_POINT_EXPR.
25456 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
25457 * tree.c (add_expr): Handle error_mark_node.
25459 2017-01-20 Martin Liska <mliska@suse.cz>
25462 * tree-profile.c (init_ic_make_global_vars): Do not call
25464 (gimple_init_gcov_profiler): Likewise.
25466 2017-01-20 Martin Liska <mliska@suse.cz>
25469 * cgraph.h (maybe_create_reference): Remove argument and
25471 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
25473 * ipa-cp.c (create_specialized_node): Likewise.
25474 * symtab.c (symtab_node::maybe_create_reference): Handle
25475 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
25477 2017-01-20 Martin Liska <mliska@suse.cz>
25479 * read-rtl-function.c (function_reader::create_function): Use
25480 build_decl instread of build_decl_stat.
25482 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
25484 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
25485 * config/i386/avx512dqintrin.h: Ditto.
25486 * config/i386/avx512fintrin.h: Ditto.
25487 * config/i386/i386-builtin-types.def: Add new types.
25488 * config/i386/i386.c: Handle new types.
25489 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
25490 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
25491 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
25492 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
25493 (__builtin_ia32_kshiftridi): New.
25494 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
25496 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
25500 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
25501 define to rs6000_init_stack_protect_guard.
25502 (rs6000_init_stack_protect_guard): New function.
25504 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
25505 Yunqiang Su <yunqiang.su@imgtec.com>
25507 * config.gcc (supported_defaults): Add madd4.
25508 (with_madd4): Add validation.
25509 (all_defaults): Add madd4.
25510 * config/mips/mips.opt (mmadd4): New option.
25511 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
25513 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
25515 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
25516 (ISA_HAS_FUSED_MADD4): Likewise.
25517 * doc/invoke.texi (-mmadd4): Document the new option.
25518 * doc/install.texi (--with-madd4): Document the new option.
25520 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25522 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
25523 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
25524 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
25525 (aarch64_init_pauth_hint_builtins): New.
25526 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
25527 (aarch64_expand_builtin): Expand new builtins.
25529 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25531 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
25532 * combine-stack-adj.c (no_unhandled_cfa): Handle
25533 REG_CFA_TOGGLE_RA_MANGLE.
25534 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
25535 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
25536 info for return address signing.
25537 (aarch64_expand_epilogue): Likewise.
25539 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25541 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
25542 * config/aarch64/aarch64-protos.h
25543 (aarch64_return_address_signing_enabled): New declaration.
25544 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
25546 (aarch64_expand_prologue): Sign return address before it's pushed onto
25548 (aarch64_expand_epilogue): Authenticate return address fetched from
25550 (aarch64_override_options): Sanity check for ILP32 and ISA level.
25551 (aarch64_attributes): New function attributes for "sign-return-address".
25552 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
25553 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
25554 ("*do_return"): Generate combined instructions according to key index.
25555 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
25556 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
25558 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
25559 * config/aarch64/aarch64.opt (msign-return-address=): New.
25560 * doc/extend.texi (AArch64 Function Attributes): Documents
25561 "sign-return-address=".
25562 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
25564 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
25566 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
25567 overall option summary.
25569 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25571 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
25572 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
25573 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
25574 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
25576 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
25578 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
25579 -mpower9-minmax by default for -mcpu=power9.
25580 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
25581 128-bit floating point.
25583 2017-01-20 Alan Modra <amodra@gmail.com>
25585 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
25586 optimizing for size.
25588 2017-01-20 Alan Modra <amodra@gmail.com>
25591 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
25592 for strcmp and strncmp from corresponding builtin decl.
25594 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
25596 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
25597 instead of i386/rtems-64.h.
25598 * config/i386/rtems-64.h: Remove.
25600 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
25604 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
25606 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
25608 2017-01-19 Tamar Christina <tamar.christina@arm.com>
25610 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
25611 Change int to HOST_WIDE_INT.
25612 * config/aarch64/aarch64-protos.h
25613 (aarch64_simd_gen_const_vector_dup): Likewise.
25614 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
25616 2017-01-19 David Malcolm <dmalcolm@redhat.com>
25618 * langhooks-def.h (lhd_type_for_size): New decl.
25619 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
25620 * langhooks.c (lhd_type_for_size): New function, taken from
25623 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
25625 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
25626 define_bypass for CR latency.
25627 (power9-cracked-alu): Update bypass latency and remove power9-branch.
25628 (power9-alu2): Add define_bypass for CR latency.
25630 (power9-mul): Update insn latency.
25631 (power9-mul-compare): Update insn latency, bypass latency and remove
25634 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25636 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
25638 * config/aarch64/aarch64.md
25639 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
25640 aarch64_nopcrelative_literal_loads.
25641 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
25643 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
25645 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
25646 TARGET_LOONGSON_3A.
25647 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
25649 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
25652 * config.gcc (supported_defaults): Add lxc1-sxc1.
25653 (with_lxc1_sxc1): Add validation.
25654 (all_defaults): Add lxc1-sxc1.
25655 * config/mips/mips.opt (mlxc1-sxc1): New option.
25656 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
25658 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
25659 __mips_no_lxc1_sxc1.
25660 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
25661 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
25662 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
25664 2017-01-19 Richard Biener <rguenther@suse.de>
25666 PR tree-optimization/72488
25667 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
25668 sure to restore SSA info.
25669 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
25671 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
25673 PR rtl-optimization/79121
25674 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
25675 of the inner type when shifting an extended value.
25677 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
25680 * symtab.c (symtab_node::equal_address_to): Fix comparing of
25681 interposable aliases.
25683 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
25686 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
25687 Use the evmergelohi instruction.
25688 (mov_si<mode>_e500_subreg4_2_le): Likewise.
25689 (mov_sitf_e500_subreg8_2_be): Likewise.
25690 (mov_sitf_e500_subreg12_2_le): Likewise.
25691 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
25692 (mov_si<mode>_e500_subreg4_2_be): Likewise.
25693 (mov_sitf_e500_subreg8_2_le): Likewise.
25694 (mov_sitf_e500_subreg12_2_be): Likewise.
25696 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25698 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
25699 attribute from vecsimple to vecperm.
25700 (altivec_vbpermq2): Likewise.
25702 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25705 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
25707 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25708 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
25709 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
25710 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
25711 case where N arg is SIZE_MAX.
25712 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
25713 (cmpstrsi): Add pattern.
25715 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25717 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25718 __builtin_vec_revb builtins.
25719 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
25720 built-in functions to support generation of the ISA 3.0 XXBR<x>
25721 vector byte reverse instructions.
25722 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
25723 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
25724 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
25725 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
25726 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
25727 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
25728 (P9V_BUILTIN_VEC_REVB): Likewise.
25729 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
25730 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
25731 (p9_xxbrq_v16qi): Likewise.
25732 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
25733 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
25734 (p9_xxbrh_v8hi): Likewise.
25735 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
25736 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
25737 vec_revb built-in functions.
25739 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
25741 PR rtl-optimization/78952
25742 * config/i386/i386.md (any_extract): New code iterator.
25743 (*insvqi_2): Use any_extract for source operand.
25744 (*insvqi_3): Use any_shiftrt for source operand.
25746 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
25748 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
25750 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
25752 2017-01-18 Matthias Klose <doko@ubuntu.com>
25754 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
25756 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25758 * config/rs6000/altivec.h (vec_bperm): Change #define.
25759 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
25760 (altivec_vbpermq2): New define_insn.
25761 (altivec_vbpermd): Likewise.
25762 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
25763 function interface.
25764 (VBPERMD): Likewise.
25765 (VBPERM): New polymorphic function interface.
25766 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
25767 Add entries for P9V_BUILTIN_VEC_VBPERM.
25768 * doc/extend.texi: Add interfaces for vec_bperm.
25770 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25772 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
25773 first letter of error messages.
25774 (s390_resolve_overloaded_builtin): Likewise.
25775 * config/s390/s390.c (s390_expand_builtin): Likewise.
25776 (s390_invalid_arg_for_unprototyped_fn): Likewise.
25777 (s390_valid_target_attribute_inner_p): Likewise.
25778 * config/s390/s390.md ("tabort"): Likewise.
25780 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
25782 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
25783 (ISA_AVOID_DIV_HILO): New macro.
25784 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
25785 (ISA_HAS_DDIV): Likewise.
25787 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25789 * doc/invoke.texi (fabi-version): Correct number of occurrences.
25791 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25793 * doc/invoke.texi (fabi-version): Spelling fix.
25795 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25798 * doc/invoke.texi (fabi-version): Mention mangling fix for
25801 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25804 * doc/invoke.texi (fabi-version): Document discriminator mangling.
25806 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
25809 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
25810 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
25812 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
25814 (stack_protect_test): Ditto.
25815 * config/rs6000/rs6000.opt (mstack-protector-guard=,
25816 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
25818 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
25819 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
25820 -mstack-protector-guard-offset=.
25821 (RS/6000 and PowerPC Options): Ditto.
25823 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
25825 * config/i386/i386.h (MASK_CLASS_P): New define.
25826 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
25827 there are no registers from different register sets also when
25828 mask registers are used. Update function comment.
25829 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
25830 to (*k/*r) and (*k/*km) alternatives.
25832 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
25834 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
25835 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
25836 (EH_RETURN_HANDLER_RTX): New define.
25837 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
25838 Force frame pointer in EH return functions.
25839 (aarch64_expand_epilogue): Add barrier for eh_return.
25840 (aarch64_final_eh_return_addr): Remove.
25841 (aarch64_eh_return_handler_rtx): New function.
25842 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
25844 (aarch64_eh_return_handler_rtx): New prototype.
25846 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25848 * config/rs6000/altivec.h (vec_rlmi): New #define.
25849 (vec_vrlnm): Likewise.
25850 (vec_rlnm): Likewise.
25851 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
25852 (UNSPEC_VRLNM): Likewise.
25853 (VIlong): New mode iterator.
25854 (altivec_vrl<VI_char>mi): New define_insn.
25855 (altivec_vrl<VI_char>nm): Likewise.
25856 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
25858 (VRLDNM): Likewise.
25859 (RLNM): New polymorphic function entry.
25860 (VRLWMI): New monomorphic function entry.
25861 (VRLDMI): Likewise.
25862 (RLMI): New polymorphic function entry.
25863 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
25864 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
25865 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
25868 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25871 * dwarf2out.c (field_byte_offset): Restore the
25872 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
25873 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
25874 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
25877 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
25880 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
25882 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25885 * dwarf2out.c (add_data_member_location_attribute): For constant
25886 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
25887 instead of DW_AT_data_member_location, DW_AT_bit_offset and
25888 DW_AT_byte_size attributes.
25890 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
25892 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
25893 after forcing to constant memory when the code model is medium.
25895 2017-01-17 Julia Koval <julia.koval@intel.com>
25898 * config/i386/avx512fintrin.h
25899 (_mm512_i32gather_ps): Change __addr type to void const*.
25900 (_mm512_mask_i32gather_ps): Ditto.
25901 (_mm512_i32gather_pd): Ditto.
25902 (_mm512_mask_i32gather_pd): Ditto.
25903 (_mm512_i64gather_ps): Ditto.
25904 (_mm512_mask_i64gather_ps): Ditto.
25905 (_mm512_i64gather_pd): Ditto.
25906 (_mm512_mask_i64gather_pd): Ditto.
25907 (_mm512_i32gather_epi32): Ditto.
25908 (_mm512_mask_i32gather_epi32): Ditto.
25909 (_mm512_i32gather_epi64): Ditto.
25910 (_mm512_mask_i32gather_epi64): Ditto.
25911 (_mm512_i64gather_epi32): Ditto.
25912 (_mm512_mask_i64gather_epi32): Ditto.
25913 (_mm512_i64gather_epi64): Ditto.
25914 (_mm512_mask_i64gather_epi64): Ditto.
25915 (_mm512_i32scatter_ps): Change __addr type to void*.
25916 (_mm512_mask_i32scatter_ps): Ditto.
25917 (_mm512_i32scatter_pd): Ditto.
25918 (_mm512_mask_i32scatter_pd): Ditto.
25919 (_mm512_i64scatter_ps): Ditto.
25920 (_mm512_mask_i64scatter_ps): Ditto.
25921 (_mm512_i64scatter_pd): Ditto.
25922 (_mm512_mask_i64scatter_pd): Ditto.
25923 (_mm512_i32scatter_epi32): Ditto.
25924 (_mm512_mask_i32scatter_epi32): Ditto.
25925 (_mm512_i32scatter_epi64): Ditto.
25926 (_mm512_mask_i32scatter_epi64): Ditto.
25927 (_mm512_i64scatter_epi32): Ditto.
25928 (_mm512_mask_i64scatter_epi32): Ditto.
25929 (_mm512_i64scatter_epi64): Ditto.
25930 (_mm512_mask_i64scatter_epi64): Ditto.
25931 * config/i386/avx512pfintrin.h
25932 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
25933 (_mm512_mask_prefetch_i32gather_ps): Ditto.
25934 (_mm512_mask_prefetch_i64gather_pd): Ditto.
25935 (_mm512_mask_prefetch_i64gather_ps): Ditto.
25936 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
25937 (_mm512_prefetch_i32scatter_ps): Ditto.
25938 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
25939 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
25940 (_mm512_prefetch_i64scatter_pd): Ditto.
25941 (_mm512_prefetch_i64scatter_ps): Ditto.
25942 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
25943 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
25944 * config/i386/avx512vlintrin.h
25945 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
25946 (_mm_mmask_i32gather_ps): Ditto.
25947 (_mm256_mmask_i32gather_pd): Ditto.
25948 (_mm_mmask_i32gather_pd): Ditto.
25949 (_mm256_mmask_i64gather_ps): Ditto.
25950 (_mm_mmask_i64gather_ps): Ditto.
25951 (_mm256_mmask_i64gather_pd): Ditto.
25952 (_mm_mmask_i64gather_pd): Ditto.
25953 (_mm256_mmask_i32gather_epi32): Ditto.
25954 (_mm_mmask_i32gather_epi32): Ditto.
25955 (_mm256_mmask_i32gather_epi64): Ditto.
25956 (_mm_mmask_i32gather_epi64): Ditto.
25957 (_mm256_mmask_i64gather_epi32): Ditto.
25958 (_mm_mmask_i64gather_epi32): Ditto.
25959 (_mm256_mmask_i64gather_epi64): Ditto.
25960 (_mm_mmask_i64gather_epi64): Ditto.
25961 (_mm256_i32scatter_ps): Change __addr type to void*.
25962 (_mm256_mask_i32scatter_ps): Ditto.
25963 (_mm_i32scatter_ps): Ditto.
25964 (_mm_mask_i32scatter_ps): Ditto.
25965 (_mm256_i32scatter_pd): Ditto.
25966 (_mm256_mask_i32scatter_pd): Ditto.
25967 (_mm_i32scatter_pd): Ditto.
25968 (_mm_mask_i32scatter_pd): Ditto.
25969 (_mm256_i64scatter_ps): Ditto.
25970 (_mm256_mask_i64scatter_ps): Ditto.
25971 (_mm_i64scatter_ps): Ditto.
25972 (_mm_mask_i64scatter_ps): Ditto.
25973 (_mm256_i64scatter_pd): Ditto.
25974 (_mm256_mask_i64scatter_pd): Ditto.
25975 (_mm_i64scatter_pd): Ditto.
25976 (_mm_mask_i64scatter_pd): Ditto.
25977 (_mm256_i32scatter_epi32): Ditto.
25978 (_mm256_mask_i32scatter_epi32): Ditto.
25979 (_mm_i32scatter_epi32): Ditto.
25980 (_mm_mask_i32scatter_epi32): Ditto.
25981 (_mm256_i32scatter_epi64): Ditto.
25982 (_mm256_mask_i32scatter_epi64): Ditto.
25983 (_mm_i32scatter_epi64): Ditto.
25984 (_mm_mask_i32scatter_epi64): Ditto.
25985 (_mm256_i64scatter_epi32): Ditto.
25986 (_mm256_mask_i64scatter_epi32): Ditto.
25987 (_mm_i64scatter_epi32): Ditto.
25988 (_mm_mask_i64scatter_epi32): Ditto.
25989 (_mm256_i64scatter_epi64): Ditto.
25990 (_mm256_mask_i64scatter_epi64): Ditto.
25991 (_mm_i64scatter_epi64): Ditto.
25992 (_mm_mask_i64scatter_epi64): Ditto.
25993 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
25994 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
25995 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
25996 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
25997 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
25998 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
25999 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
26000 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
26001 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
26002 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
26003 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
26004 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
26005 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
26006 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
26007 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
26008 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
26009 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
26010 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
26011 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
26012 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
26013 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
26014 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
26015 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
26016 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
26017 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
26018 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
26019 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
26020 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
26021 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
26022 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
26023 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
26024 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
26025 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
26026 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
26027 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
26028 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
26029 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
26030 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
26031 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
26032 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
26033 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
26034 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
26035 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
26036 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
26037 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
26038 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
26039 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
26040 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
26041 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
26042 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
26043 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
26044 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
26045 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
26046 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
26047 definitions accordingly.
26049 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
26050 Kuan-Lin Chen <kuanlinchentw@gmail.com>
26053 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
26056 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
26059 * ira-conflicts.c (ira_build_conflicts): Update total conflict
26060 hard regs for inner regno.
26062 2017-01-17 Martin Liska <mliska@suse.cz>
26065 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
26066 assumption and add comment.
26068 2017-01-17 Nathan Sidwell <nathan@acm.org>
26070 * ipa-visibility.c (localize_node): New function, broken out of ...
26071 (function_and_variable_visibility): ... here. Call it.
26073 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
26075 PR middle-end/77445
26076 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
26077 correctly set frequency of oudgoing edge.
26078 (duplicate_thread_path): Fix profile updating.
26080 2017-01-17 Jakub Jelinek <jakub@redhat.com>
26083 * configure.ac: Add GCC_BASE_VER.
26084 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
26085 version from BASE-VER file.
26086 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
26087 (gcc.o): Depend on $(BASEVER).
26088 * common.opt (dumpfullversion): New option.
26089 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
26090 * doc/invoke.texi: Document -dumpfullversion.
26091 * doc/install.texi: Document --with-gcc-major-version-only.
26092 * configure: Regenerated.
26094 2017-01-17 Richard Biener <rguenther@suse.de>
26096 PR tree-optimization/71433
26097 * tree-vrp.c (register_new_assert_for): Merge same asserts
26098 on all incoming edges.
26099 (process_assert_insertions_for): Handle insertions at the
26102 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
26104 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
26105 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
26107 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
26110 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
26113 2017-01-17 Alan Modra <amodra@gmail.com>
26116 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
26117 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
26118 symbolic stack limit when pic.
26120 2017-01-16 Martin Sebor <msebor@redhat.com>
26122 PR tree-optimization/78608
26123 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
26125 2017-01-16 Jeff Law <law@redhat.com>
26128 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
26129 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
26130 for several include directories that may be relative to sysroot.
26131 * config/i386/x-mingw32 (gplus_includedir): Define.
26132 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
26133 (native_system_includedir): Likewise.
26134 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
26135 override if TARGET_SYSTEM_ROOT is defined.
26136 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
26138 PR tree-optimization/79090
26139 PR tree-optimization/33562
26140 PR tree-optimization/61912
26141 PR tree-optimization/77485
26142 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
26143 and computed trims into the dump file.
26145 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
26147 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
26149 2017-01-16 Jakub Jelinek <jakub@redhat.com>
26152 * gimplify.c (gimplify_init_constructor): If want_value and
26153 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
26157 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
26158 sequence. Formatting fixes.
26159 (doloop_optimize): Formatting fixes.
26162 * gcc.c (debug_level_greater_than_spec_func): New function.
26163 (static_spec_functions): Add debug-level-gt spec function.
26164 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
26166 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
26167 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
26168 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
26169 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
26170 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
26171 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
26173 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
26175 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
26176 QImode fixups to general and mask registers only.
26178 2017-01-16 Carl Love <cel@us.ibm.com>
26180 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
26181 for built-in functions
26182 vector signed char vec_nabs (vector signed char)
26183 vector signed short vec_nabs (vector signed short)
26184 vector signed int vec_nabs (vector signed int)
26185 vector signed long long vec_nabs (vector signed long long)
26186 vector float vec_nabs (vector float)
26187 vector double vec_nabs (vector double)
26188 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
26190 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
26191 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
26192 * doc/extend.texi: Update the documentation file for the new built-in
26195 2017-01-16 Martin Sebor <msebor@redhat.com>
26197 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
26200 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26202 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
26203 UNSPEC_VSX__XXSPLTD to require special splat handling.
26205 2017-01-16 David Malcolm <dmalcolm@redhat.com>
26208 * system.h: Poison strndup.
26210 2017-01-16 Alan Modra <amodra@gmail.com>
26213 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
26216 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
26218 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
26220 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
26222 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
26223 call recog here. Assert that INSN_CODE (insn) is non-negative.
26225 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
26228 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
26230 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
26231 in the currently scheduled RTL fragment.
26233 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
26235 PR rtl-optimization/78751
26236 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
26239 2017-01-14 Jeff Law <law@redhat.com>
26241 PR tree-optimization/79090
26242 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
26243 variable length stores.
26244 (compute_trims): Delete dead assignment to *trim_tail.
26245 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
26248 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
26250 PR rtl-optimization/78626
26251 PR rtl-optimization/78727
26252 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
26253 of a block, and split such blocks after everything else is finished.
26255 2017-01-14 Alan Modra <amodra@gmail.com>
26258 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
26259 target legitimate_combined_insn.
26260 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
26261 (rs6000_legitimate_combined_insn): New function.
26262 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
26264 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
26265 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
26266 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
26268 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
26270 * doc/frontends.texi (G++ and GCC): Remove references to Java.
26272 2017-01-13 Jeff Law <law@redhat.com>
26274 PR tree-optimization/33562
26275 PR tree-optimization/61912
26276 PR tree-optimization/77485
26277 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
26279 (delete_dead_assignment): Likewise.
26280 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
26281 statement to delete_dead_call and delete_dead_assignment.
26283 2017-01-13 David Malcolm <dmalcolm@redhat.com>
26286 * substring-locations.c (format_warning_va): Strengthen case 1 so
26287 that both endpoints of the substring must be within the format
26288 range for just the substring to be printed.
26290 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
26292 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
26293 * config/i386/i386.c (ix86_target_string): Add missing options
26294 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
26295 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
26296 flags_other and ix86_target_other to flags2_other. Display unknown
26298 (ix86_valid_target_attribute_inner_p): Add missing options and
26299 reorder options by implied ISAs, as in ix86_target_string.
26301 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
26303 * hash-table.h (hash_table::too_empty_p): New function.
26304 (hash_table::expand): Use it.
26305 (hash_table::traverse): Likewise.
26306 (hash_table::empty_slot): Use sizeof (value_type) instead of
26307 sizeof (PTR) to convert bytes to elements. Shrink the table
26308 if the current size is excessive for the current number of
26311 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
26313 * ira-costs.c (record_reg_classes): Break from the inner loop
26314 early once alt_fail is known to be true. Update outer loop
26315 handling accordingly.
26317 2017-01-13 Jeff Law <law@redhat.com>
26319 * tree-ssa-dse.c (decrement_count): New function.
26320 (increment_start_addr, maybe_trim_memstar_call): Likewise.
26321 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
26322 when we know the partially dead statement is a mem* function.
26324 PR tree-optimization/61912
26325 PR tree-optimization/77485
26326 * tree-ssa-dse.c: Include expr.h.
26327 (maybe_trim_constructor_store): New function.
26328 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
26330 PR tree-optimization/33562
26331 PR tree-optimization/61912
26332 PR tree-optimization/77485
26333 * doc/invoke.texi: Document new dse-max-object-size param.
26334 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
26335 * tree-ssa-dse.c: Include params.h.
26336 (dse_store_status): New enum.
26337 (initialize_ao_ref_for_dse): New, partially extracted from
26339 (valid_ao_ref_for_dse, normalize_ref): New.
26340 (setup_live_bytes_from_ref, compute_trims): Likewise.
26341 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
26342 (maybe_trim_partially_dead_store): Likewise.
26343 (maybe_trim_complex_store): Likewise.
26344 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
26345 Track what bytes live from the original store. Return tri-state
26346 for dead, partially dead or live.
26347 (dse_dom_walker): Add constructor, destructor and new private members.
26348 (delete_dead_call, delete_dead_assignment): New extracted from
26350 (dse_optimize_stmt): Make a member of dse_dom_walker.
26351 Use initialize_ao_ref_for_dse.
26353 PR tree-optimization/33562
26354 PR tree-optimization/61912
26355 PR tree-optimization/77485
26356 * sbitmap.h (bitmap_count_bits): Prototype.
26357 (bitmap_clear_range, bitmap_set_range): Likewise.
26358 * sbitmap.c (bitmap_clear_range): New function.
26359 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
26361 2017-01-13 Martin Liska <mliska@suse.cz>
26364 * function.c (set_cfun): Add new argument force.
26365 * function.h (set_cfun): Likewise.
26366 * ipa-inline-transform.c (inline_call): Use the function when
26367 strict alising from is dropped for function we inline to.
26369 2017-01-13 Richard Biener <rguenther@suse.de>
26371 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
26372 for dumping GIMPLE INTEGER_CSTs.
26374 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26376 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
26377 to 201112L since C++17.
26379 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
26382 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
26383 if -fsanitize=kernel-address is present.
26385 2017-01-13 Richard Biener <rguenther@suse.de>
26387 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
26388 as _Literal ( type ) number in case usual suffixes do not
26389 preserve all information.
26391 2017-01-13 Richard Biener <rguenther@suse.de>
26393 PR tree-optimization/77283
26394 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
26395 and ssa-iterators.h.
26396 (is_feasible_trace): Implement a cost model based on joiner
26399 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
26402 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
26403 char or short to __float128/_Float128 directly.
26405 2017-01-12 Martin Sebor <msebor@redhat.com>
26407 to -Wformat-overflow.
26408 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
26409 (min_bytes_remaining): Same.
26410 (get_string_length): Same.
26411 (format_string): Same.
26412 (format_directive): Same.
26414 (pass_sprintf_length::handle_gimple_call): Same.
26416 2017-01-12 Jakub Jelinek <jakub@redhat.com>
26418 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
26419 info.nowrite calls with no lhs that can't throw. Return bool
26420 whether gsi_remove has been called or not.
26421 (pass_sprintf_length::handle_gimple_call): Return bool whether
26422 try_substitute_return_value called gsi_remove. Formatting fix.
26423 (pass_sprintf_length::execute): Don't use gsi_remove if
26424 handle_gimple_call returned true.
26427 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
26428 be removed due to side-effects, don't remove following barrier nor
26429 turn the successor edge into fallthru edge.
26431 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26434 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
26435 element-reversing loads and stores as not swappable.
26437 2017-01-12 Nathan Sidwell <nathan@acm.org>
26438 Nicolai Stange <nicstange@gmail.com>
26440 * combine.c (try_combine): Don't ignore result of overlap checking
26441 loop. Combine overlap & asm check into single loop.
26443 2017-01-12 Richard Biener <rguenther@suse.de>
26445 * tree-pretty-print.c (dump_generic_node): Provide -gimple
26446 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
26448 2017-01-12 Richard Biener <rguenther@suse.de>
26450 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
26451 and TS_TARGET_OPTION directly derive from TS_BASE.
26452 * tree-core.h (tree_optimization_option): Derive from tree_base.
26453 (tree_target_option): Likewise.
26455 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
26457 * config/i386/i386.c (memory_address_length): Increase len
26458 only when rip_relative_addr_p returns false.
26460 2017-01-11 Julia Koval <julia.koval@intel.com>
26462 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
26463 (OPTION_MASK_ISA_SGX_SET): New.
26464 (ix86_handle_option): Handle OPT_msgx.
26465 * config.gcc: Added sgxintrin.h.
26466 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
26467 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
26468 * config/i386/i386.c (ix86_target_string): Add -msgx.
26470 (ix86_option_override_internal): Handle new options.
26471 (ix86_valid_target_attribute_inner_p): Add sgx.
26472 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
26473 * config/i386/i386.opt: Add msgx.
26474 * config/i386/sgxintrin.h: New file.
26475 * config/i386/x86intrin.h: Add sgxintrin.h.
26477 2017-01-11 Jakub Jelinek <jakub@redhat.com>
26480 * fold-const.c (maybe_nonzero_address): Return 1 for function
26482 (tree_single_nonzero_warnv_p): Don't handle function local objects
26486 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
26489 2017-01-11 David Malcolm <dmalcolm@redhat.com>
26492 * opts.c: Include "spellcheck.h"
26493 (struct string_fragment): New struct.
26494 (struct edit_distance_traits<const string_fragment &>): New
26496 (get_closest_sanitizer_option): New function.
26497 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
26499 2017-01-11 Jakub Jelinek <jakub@redhat.com>
26501 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
26503 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
26504 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
26505 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
26506 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
26507 for initial die_offset if dwarf_split_debug_info.
26508 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
26509 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
26511 (output_skeleton_debug_sections): Formatting fix. Use
26512 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
26513 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
26515 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
26517 * config/arm/cortex-a53.md: Add bypasses for
26518 cortex_a53_r2f_cvt.
26519 (cortex_a53_r2f): Only use for transfers.
26520 (cortex_a53_f2r): Likewise.
26521 (cortex_a53_r2f_cvt): Add reservation for conversions.
26522 (cortex_a53_f2r_cvt): Likewise.
26524 2017-01-11 Tamar Christina <tamar.christina@arm.com>
26526 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
26527 to all inlined functions, change static to extern.
26529 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
26532 * config/arm/arm.c (legitimize_pic_address): Handle reference to
26534 (arm_assemble_integer): Likewise.
26536 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
26538 * config.gcc: Use new awk script to check CPU, FPU and architecture
26539 parameters for --with-... options.
26540 * config/arm/parsecpu.awk: New file
26541 * config/arm/arm-cpus.in: New file.
26542 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
26544 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
26546 * config/arm/t-arm: Update dependency rules.
26547 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
26548 of processing .def files.
26549 * config/arm/genopt.sh: Deleted.
26550 * config/arm/gentune.sh: Deleted.
26551 * config/arm/arm-cores.def: Deleted.
26552 * config/arm/arm-arches.def: Deleted.
26553 * config/arm/arm-fpus.def: Deleted.
26554 * config/arm/arm-tune.md: Regenerated.
26555 * config/arm/arm-tables.opt: Regenerated.
26556 * config/arm/arm-cpu.h: New generated file.
26557 * config/arm/arm-cpu-data.h: New generated file.
26558 * config/arm/arm-cpu-cdata.h: New generated file.
26560 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
26563 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
26565 (input_varpool_node): Unpack dynamically_initialized bit.
26567 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
26569 PR rtl-optimization/79032
26570 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
26571 the alignment of the adjusted memory reference against that of MODE,
26572 instead of the alignment of the original memory reference.
26574 2017-01-11 Martin Jambor <mjambor@suse.cz>
26576 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
26578 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
26579 decorated functions.
26581 2017-01-11 Richard Biener <rguenther@suse.de>
26583 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
26584 set range/nonnull info for PHI results. Do not set it on
26585 stmts marked for removal.
26587 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
26589 * expr.c (store_field): In the bitfield case, fetch the return value
26590 from the registers before applying a single big-endian adjustment.
26591 Always do a final load for a BLKmode value not larger than a word.
26593 2017-01-10 David Malcolm <dmalcolm@redhat.com>
26596 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
26597 that we correctly handle column numbers greater than
26598 LINE_MAP_MAX_COLUMN_NUMBER.
26600 2017-01-10 Martin Sebor <msebor@redhat.com>
26602 PR middle-end/78245
26603 * gimple-ssa-sprintf.c (get_destination_size): Call
26604 {init,fini}object_sizes.
26605 * tree-object-size.c (addr_object_size): Adjust.
26606 (pass_through_call): Adjust.
26607 (pass_object_sizes::execute): Adjust.
26608 * tree-object-size.h (fini_object_sizes): Declare.
26610 2017-01-10 Martin Sebor <msebor@redhat.com>
26612 PR tree-optimization/78775
26613 * builtins.c (get_size_range): Move...
26614 * calls.c: ...to here.
26615 (alloc_max_size): Accept zero argument.
26616 (operand_signed_p): Remove.
26617 (maybe_warn_alloc_args_overflow): Call get_size_range.
26618 * calls.h (get_size_range): Declare.
26620 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
26622 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
26623 from TI's devices.csv file as of September 2016.
26624 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
26626 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
26628 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
26629 * doc/invoke.texi: Likewise.
26630 * doc/md.texi: Likewise.
26631 * doc/objc.texi: Likewise.
26633 2017-01-10 Joshua Conner <joshconner@google.com>
26635 * config/arm/fuchsia-elf.h: New file.
26636 * config/fuchsia.h: New file.
26637 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
26638 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
26640 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
26642 2016-01-10 Richard Biener <rguenther@suse.de>
26644 PR tree-optimization/79034
26645 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
26646 Propagate out degenerate PHIs in the joiner.
26648 2017-01-10 Martin Liska <mliska@suse.cz>
26650 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
26651 (sort_congruence_classes_by_decl_uid): Likewise.
26652 (sort_congruence_class_groups_by_decl_uid): Likewise.
26653 (sem_item_optimizer::merge_classes): Sort class, groups in these
26654 classes and members in the groups by DECL_UID of declarations.
26655 This would make merge operations stable.
26657 2017-01-10 Martin Liska <mliska@suse.cz>
26659 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
26660 usage of m_classes_vec.
26661 (sem_item_optimizer::~sem_item_optimizer): Likewise.
26662 (sem_item_optimizer::get_group_by_hash): Likewise.
26663 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
26664 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
26665 (sem_item_optimizer::verify_classes): Likewise.
26666 (sem_item_optimizer::process_cong_reduction): Likewise.
26667 (sem_item_optimizer::dump_cong_classes): Likewise.
26668 (sem_item_optimizer::merge_classes): Likewise.
26669 * ipa-icf.h (congruence_class_hash): Rename from
26670 congruence_class_group_hash. Remove declaration of m_classes_vec.
26672 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
26674 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
26675 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
26676 * config.gcc: Add avx512vpopcntdqintrin.h.
26677 * config/i386/avx512vpopcntdqintrin.h: New.
26678 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
26679 * config/i386/i386-builtin-types.def: Add new types.
26680 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
26681 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
26682 __builtin_ia32_vpopcountq_v8di_mask): New.
26683 * config/i386/i386-c.c (ix86_target_macros_internal): Define
26684 __AVX512VPOPCNTDQ__.
26685 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
26686 (PTA_AVX512VPOPCNTDQ): Define.
26687 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
26688 TARGET_AVX512VPOPCNTDQ_P): Define.
26689 * config/i386/i386.opt: Add mavx512vpopcntdq.
26690 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
26691 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
26693 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26695 PR middle-end/77484
26696 * predict.def (PRED_CALL): Set to 67.
26698 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
26700 * expr.c (store_field): In the bitfield case, if the value comes from
26701 a function call and is of an aggregate type returned in registers, do
26702 not modify the field mode; extract the value in all cases if the mode
26703 is BLKmode and the size is not larger than a word.
26705 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
26708 * config/i386/cpuid.h: Fix undefined behavior.
26710 2017-01-04 Jeff Law <law@redhat.com>
26712 PR tree-optimization/79007
26713 PR tree-optimization/67955
26714 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
26715 conservative for pt.null when flag_non_call_exceptions is on.
26717 2017-01-09 Jakub Jelinek <jakub@redhat.com>
26719 PR translation/79019
26720 PR translation/79020
26721 * params.def (PARAM_INLINE_MIN_SPEEDUP,
26722 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
26723 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
26725 * config/avr/avr.opt (maccumulate-args): Likewise.
26726 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
26727 * common.opt (freport-bug): Likewise.
26728 * cif-code.def (CIF_FINAL_ERROR): Likewise.
26729 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
26730 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
26731 translatable string.
26732 * config/i386/i386.c (function_value_32): Likewise.
26733 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
26734 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
26736 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
26737 * common/config/msp430/msp430-common.c (msp430_handle_option):
26739 * symtab.c (symtab_node::verify_base): Likewise.
26740 * opts.c (set_debug_level): Likewise.
26741 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
26742 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
26743 missing whitespace to translatable strings.
26744 * config/avr/avr.md (bswapsi2): Fix typo in comment.
26745 * config/sh/superh.h: Likewise.
26746 * config/i386/xopintrin.h: Likewise.
26747 * config/i386/znver1.md: Likewise.
26748 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
26749 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
26750 * double-int.h (struct double_int): Likewise.
26751 * double-int.c (div_and_round_double): Likewise.
26752 * wide-int.cc: Likewise.
26753 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
26754 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
26755 * cfgcleanup.c (crossjumps_occured): Renamed to ...
26756 (crossjumps_occurred): ... this.
26757 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
26760 PR tree-optimization/78899
26761 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
26762 returning bool return struct loop *, NULL for failure and the new
26764 (versionable_outer_loop_p): Don't version outer loop if it has
26765 dont_vectorized bit set.
26766 (tree_if_conversion): When versioning outer loop, ensure
26767 tree_if_conversion is performed also on the inner loop of the
26768 non-vectorizable outer loop copy.
26769 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
26770 LOOP_VECTORIZED in inner loop of the scalar outer loop and
26771 prevent vectorization of it.
26772 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
26773 the outer loop vectorization of the non-scalar version is attempted
26774 before vectorization of the inner loop in scalar version. If
26775 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
26776 vectorization of its inner loop.
26777 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
26778 has 2 inner loops, rename also on edges from bb whose single pred
26779 is outer_loop->header. Fix typo in function comment.
26781 2017-01-09 Martin Sebor <msebor@redhat.com>
26784 * asan.c (asan_emit_stack_protection): Increase local buffer size
26785 to avoid snprintf truncation warning.
26787 2017-01-09 Andrew Pinski <apinski@cavium.com>
26789 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
26790 to reference thunderx2t99 for the tuning structure
26791 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
26793 (thunderx2t99_extra_costs): This.
26794 * config/aarch64/aarch64-tune.md: Regenerate.
26795 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
26796 (vulcan_addrcost_table): This.
26797 (vulcan_regmove_cost): Rename to ...
26798 (thunderx2t99_regmove_cost): This.
26799 (vulcan_vector_cost): Rename to ...
26800 (thunderx2t99_vector_cost): this.
26801 (vulcan_branch_cost): Rename to ...
26802 (thunderx2t99_branch_cost): This.
26803 (vulcan_tunings): Rename to ...
26804 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
26805 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
26807 2017-01-09 Martin Jambor <mjambor@suse.cz>
26811 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
26812 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
26813 (propagate_vr_accross_jump_function): Use the above function for all
26814 value range computations for pass-through jump functions and type
26815 converasion from explicit value range values.
26816 (ipcp_propagate_stage): Do not attempt to deduce types of formal
26817 parameters from TYPE_ARG_TYPES.
26818 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
26819 (ipa_write_node_info): Stream type of the actual argument.
26820 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
26822 2017-01-09 Martin Liska <mliska@suse.cz>
26825 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
26826 (lookup_compiler): Do not show error message with have_E.
26828 2017-01-09 Jakub Jelinek <jakub@redhat.com>
26830 PR tree-optimization/78938
26831 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
26832 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
26833 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
26834 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
26837 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26839 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
26842 2017-01-09 Richard Biener <rguenther@suse.de>
26844 PR tree-optimization/78997
26845 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
26846 name condition properly.
26848 2017-01-09 Richard Biener <rguenther@suse.de>
26851 * dwarf2out.c (is_cxx): New overload with context.
26852 (is_naming_typedef_decl): Use it.
26854 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26856 * invoke.texi (Option Summary): Correct spacing in option lists
26857 and add line breaks to fix over-long lines.
26859 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26861 PR middle-end/17660
26863 * extend.texi (Common Variable Attributes): Add xref to GCC
26864 Internals manual to explain mode attribute keywords.
26866 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26869 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
26870 and Preprocessor Options.
26871 (Options for Linking): Document -pthread here....
26872 (RS/6000 and PowerPC Options): ...not here.
26873 (Solaris 2 Options): ...or here.
26874 * doc/cppopts.texi: Document -pthread.
26876 2017-01-08 Martin Sebor <msebor@redhat.com>
26878 PR middle-end/77708
26879 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
26880 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
26881 New member functions.
26882 (format_directive): Used them.
26884 (pass_sprintf_length::handle_gimple_call): Same.
26885 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
26886 to avoid truncation for any argument.
26887 (extract_affine_mul): Same.
26888 * tree.c (get_file_function_name): Same.
26890 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26892 PR middle-end/77484
26893 * predict.def (PRED_INDIR_CALL): Set to 86.
26895 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26897 PR preprocessor/54124
26898 * doc/cppopts.texi: Reformat -d subtable to list the full name
26899 of the options. Add cross-reference to the docs for the general
26900 compiler -d options.
26901 * doc/invoke.texi (Developer Options): Add cross-reference to the
26902 preprocessor-specific -d option documentation.
26904 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26906 PR preprocessor/13498
26907 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
26908 redudant material, and reflect new command-line options.
26909 (System Headers): Likewise.
26911 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26913 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
26914 -isystem, and -idirafter. Copy-edit.
26915 * doc/cppopts.texi: Copy-edit. Remove contradiction about
26916 default for -ftrack-macro-expansion. Delete obsolete and
26917 badly-formatted implementation details about -fdebug-cpp output.
26918 * doc/cppwarnopts.texi: Copy-edit.
26920 2017-01-07 David Malcolm <dmalcolm@redhat.com>
26923 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
26924 that the transition from a max line width >= 1<<10 to narrower
26925 lines works correctly.
26927 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
26929 * doc/options.texi (PerFunction): New.
26930 * opt-functions.awk (switch_flags): Map both Optimization and
26931 PerFunction to CL_OPTIMIZATION.
26932 * opth-gen.awk: Test for PerFunction flag along with
26934 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
26935 it only when the latter is present. Skip those that don't in
26936 the hash function generator.
26937 * common.opt (fvar-tracking): Mark as PerFunction instead of
26939 (fvar-tracking-assignments): Likewise.
26940 (fvar-tracking-assignments-toggle): Likewise.
26941 (fvar-tracking-uninit): Likewise.
26943 2017-01-07 Jakub Jelinek <jakub@redhat.com>
26945 PR translation/79018
26946 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
26949 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
26952 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
26953 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
26954 TARGET_LONG_JUMP_TABLE_OFFSETS.
26955 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
26956 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
26957 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
26958 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
26959 * config/m68k/m68k.md (tablejump expander): Likewise.
26960 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
26961 TARGET_LONG_JUMP_TABLE_OFFSETS.
26962 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
26963 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
26965 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
26966 David Holsgrove <david.holsgrove@xilinx.com>
26968 * common/config/microblaze/microblaze-common.c
26969 (TARGET_EXCEPT_UNWIND_INFO): Remove.
26970 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
26972 * config/microblaze/microblaze.c (microblaze_must_save_register)
26973 (microblaze_expand_epilogue, microblaze_return_addr): Handle
26975 (microblaze_eh_return): New function.
26976 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
26977 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
26978 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
26979 * config/microblaze/microblaze.md (eh_return): New pattern.
26981 2017-01-06 Jakub Jelinek <jakub@redhat.com>
26983 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
26984 GCC_DIAGNOSTIC_STRINGIFY): Define.
26986 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
26988 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26990 * config/arm/arm.md (<mcrr>): New.
26992 * config/arm/arm.c (arm_arch5te): New.
26993 (arm_option_override): Set arm_arch5te.
26994 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
26996 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
26997 (arm_mcrr_qualifiers): ... this. New.
26998 (MRRC_QUALIFIERS): Define to...
26999 (arm_mrrc_qualifiers): ... this. New.
27000 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
27002 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
27003 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
27004 (MRRCI, mrrc, MRRC): New.
27005 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
27006 VUNSPEC_MRRC2): New.
27008 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27010 * config/arm/arm.md (<mcr>): New.
27012 * config/arm/arm.c (arm_coproc_builtin_available): Add
27013 support for mcr, mrc, mcr2 and mrc2.
27014 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
27015 (arm_mcr_qualifiers): ... this. New.
27016 (MRC_QUALIFIERS): Define to ...
27017 (arm_mrc_qualifiers): ... this. New.
27018 (MCR_QUALIFIERS): Define to ...
27019 (arm_mcr_qualifiers): ... this. New.
27020 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
27022 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
27023 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
27024 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
27025 VUNSPEC_MRC2): New.
27027 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27029 * config/arm/arm.md (*ldc): New.
27033 * config/arm/arm.c (arm_coproc_builtin_available): Add
27034 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
27035 (arm_coproc_ldc_stc_legitimate_address): New.
27036 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
27037 'qualifier_const_pointer'.
27038 (LDC_QUALIFIERS): Define to...
27039 (arm_ldc_qualifiers): ... this. New.
27040 (STC_QUALIFIERS): Define to...
27041 (arm_stc_qualifiers): ... this. New.
27042 * config/arm/arm-protos.h
27043 (arm_coproc_ldc_stc_legitimate_address): New.
27044 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
27045 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
27046 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
27047 stc2, stcl, stc2l): New.
27048 * config/arm/constraints.md (Uz): New.
27049 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
27050 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
27051 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
27052 VUNSPEC_STC2L): New.
27054 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27056 * config/arm/arm.md (<cdp>): New.
27057 * config/arm/arm.c (neon_const_bounds): Rename this ...
27058 (arm_const_bounds): ... this.
27059 (arm_coproc_builtin_available): New.
27060 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
27061 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
27062 (CDP_QUALIFIERS): Define to...
27063 (arm_cdp_qualifiers): ... this. New.
27065 (arm_expand_builtin_args): Add case for 6 arguments.
27066 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
27067 (arm_const_bounds): ... this.
27068 (arm_coproc_builtin_available): New.
27069 * config/arm/arm_acle.h (__arm_cdp): New.
27071 * config/arm/arm_acle_builtins.def (cdp): New.
27073 * config/arm/iterators.md (CDPI,CDP,cdp): New.
27074 * config/arm/neon.md: Rename all 'neon_const_bounds' to
27075 'arm_const_bounds'.
27076 * config/arm/types.md (coproc): New.
27077 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
27078 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
27079 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
27080 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
27082 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27084 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
27085 (UBINOP_QUALIFIERS): New.
27087 (acle_builtin_data): New. Change comment.
27088 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
27089 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
27090 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
27091 arm_acle_builtins.def.
27092 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
27093 (arm_init_acle_builtins): New.
27094 (CRC32_BUILTIN): Remove.
27095 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
27096 crc32cb, crc32ch and crc32cw.
27097 (arm_init_crc32_builtins): Remove.
27098 (arm_init_builtins): Use arm_init_acle_builtins rather
27099 than arm_init_crc32_builtins.
27100 (arm_expand_acle_builtin): New.
27101 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
27102 * config/arm/arm_acle_builtins.def: New.
27104 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27106 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
27107 (arm_builtin_datum): ... this.
27108 (arm_init_neon_builtin): Rename to ...
27109 (arm_init_builtin): ... this. Add a new parameters PREFIX
27110 and USE_SIG_IN_NAME.
27111 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
27112 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
27113 'arm_builtin_datum'.
27114 (arm_init_vfp_builtins): Likewise.
27115 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
27116 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
27117 (arm_expand_neon_args): Rename to ...
27118 (arm_expand_builtin_args): ... this. Rename builtin_arg
27119 enum values and differentiate between ARG_BUILTIN_MEMORY
27120 and ARG_BUILTIN_NEON_MEMORY.
27121 (arm_expand_neon_builtin_1): Rename to ...
27122 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
27123 values, arm_expand_builtin_args and add bool parameter NEON.
27124 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
27125 (arm_expand_vfp_builtin): Likewise.
27126 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
27128 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27130 PR middle-end/77484
27131 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
27132 * predict.c (tree_estimate_probability_bb): Reverse direction of
27133 polymorphic call predictor.
27135 2017-01-06 David Malcolm <dmalcolm@redhat.com>
27137 * passes.c (execute_one_pass): Split out pass-skipping logic into...
27138 (determine_pass_name_match): ...this new function and...
27139 (should_skip_pass_p): ...this new function.
27141 2017-01-06 Nathan Sidwell <nathan@acm.org>
27143 * ipa-visibility.c (function_and_variable_visibility): Reformat
27144 comments and long lines. Remove extrneous if.
27145 * symtab.c (symtab_node::make_decl_local): Fix code format.
27146 (symtab_node::set_section_for_node): Fix comment typo.
27148 2017-01-06 Martin Liska <mliska@suse.cz>
27151 * lra-constraints.c: Rename invariant to lra_invariant.
27152 * predict.c (set_even_probabilities): Initialize e to NULL.
27154 2017-01-05 Martin Sebor <msebor@redhat.com>
27156 PR tree-optimization/78910
27157 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
27158 (format_integer): Correct off-by-one error in the handling
27159 of precision with negative numbers in signed conversions..
27161 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
27163 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
27165 2017-01-05 Jakub Jelinek <jakub@redhat.com>
27167 PR tree-optimization/71016
27168 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
27169 factor_out_conditional_conversion. Formatting fix.
27170 (factor_out_conditional_conversion): Add cond_stmt argument.
27171 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
27172 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
27175 2017-01-05 David Malcolm <dmalcolm@redhat.com>
27177 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
27178 read-rtl-function.o, and selftest-rtl.o.
27179 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
27180 (selftest::aarch64_test_loading_full_dump): New function.
27181 (selftest::aarch64_run_selftests): New function.
27182 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
27183 selftest::aarch64_run_selftests.
27184 * config/i386/i386.c
27185 (selftest::ix86_test_loading_dump_fragment_1): New function.
27186 (selftest::ix86_test_loading_call_insn): New function.
27187 (selftest::ix86_test_loading_full_dump): New function.
27188 (selftest::ix86_test_loading_unspec): New function.
27189 (selftest::ix86_run_selftests): Call the new functions.
27190 * emit-rtl.c (maybe_set_max_label_num): New function.
27191 * emit-rtl.h (maybe_set_max_label_num): New decl.
27192 * function.c (instantiate_decls): Guard call to
27193 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
27194 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
27196 * gensupport.c (gen_reader::gen_reader): Pass "false"
27197 for new "compact" param of rtx_reader.
27198 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
27199 rather than an empty string for NULL strings.
27200 * read-md.c: Potentially include config.h rather than bconfig.h.
27201 Wrap include of errors.h with #ifdef GENERATOR_FILE.
27202 (have_error): New global, copied from errors.c.
27203 (md_reader::read_name): Rename to...
27204 (md_reader::read_name_1): ...this, adding "out_loc" param,
27205 and converting "missing name or number" to returning false, rather
27207 (md_reader::read_name): Reimplement in terms of read_name_1.
27208 (md_reader::read_name_or_nil): New function.
27209 (md_reader::read_string): Handle "(nil)" by returning NULL.
27210 (md_reader::md_reader): Add new param "compact".
27211 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
27212 (md_reader::read_file): New method.
27213 * read-md.h (md_reader::md_reader): Add new param "compact".
27214 (md_reader::read_file): New method.
27215 (md_reader::is_compact): New accessor.
27216 (md_reader::read_name): Convert return type from void to file_location.
27217 (md_reader::read_name_or_nil): New decl.
27218 (md_reader::read_name_1): New decl.
27219 (md_reader::m_compact): New field.
27220 (noop_reader::noop_reader): Pass "false" for new "compact" param
27222 (rtx_reader::rtx_reader): Add new "compact" param.
27223 (rtx_reader::read_rtx_operand): Make virtual and convert return
27224 type from void to rtx.
27225 (rtx_reader::read_until): New decl.
27226 (rtx_reader::handle_any_trailing_information): New virtual function.
27227 (rtx_reader::postprocess): New virtual function.
27228 (rtx_reader::finalize_string): New virtual function.
27229 (rtx_reader::m_in_call_function_usage): New field.
27230 (rtx_reader::m_reuse_rtx_by_id): New field.
27231 * read-rtl-function.c: New file.
27232 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
27233 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
27234 (selftest::verify_three_block_rtl_cfg): New decl.
27235 * read-rtl-function.h: New file.
27236 * read-rtl.c: Potentially include config.h rather than bconfig.h.
27237 For host, include function.h, memmodel.h, and emit-rtl.h.
27238 (one_time_initialization): New function.
27239 (struct compact_insn_name): New struct.
27240 (compact_insn_names): New array.
27241 (find_code): Handle insn codes in compact dumps.
27242 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
27243 (bind_subst_iter_and_attr): Likewise.
27244 (add_condition_to_string): Likewise.
27245 (add_condition_to_rtx): Likewise.
27246 (apply_attribute_uses): Likewise.
27247 (add_current_iterators): Likewise.
27248 (apply_iterators): Likewise.
27249 (initialize_iterators): Guard usage of apply_subst_iterator with
27250 #ifdef GENERATOR_FILE.
27251 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
27252 (md_reader::read_mapping): Likewise.
27253 (add_define_attr_for_define_subst): Likewise.
27254 (add_define_subst_attr): Likewise.
27255 (read_subst_mapping): Likewise.
27256 (check_code_iterator): Likewise.
27257 (rtx_reader::read_rtx): Likewise. Move one-time initialization
27259 (one_time_initialization): New function.
27260 (rtx_reader::read_until): New method.
27261 (read_flags): New function.
27262 (parse_reg_note_name): New function.
27263 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
27264 Handle reuse_rtx ids.
27265 Wrap iterator lookup within #ifdef GENERATOR_FILE.
27266 Add parsing support for RTL dumps, mirroring the special-cases in
27267 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
27268 values, and calling handle_any_trailing_information.
27269 (rtx_reader::read_rtx_operand): Convert return type from void
27270 to rtx, returning return_rtx. Handle case 'e'. Call
27271 finalize_string on XSTR and XTMPL fields.
27272 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
27273 "(nil)" values were omitted. Call the postprocess vfunc on the
27275 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
27276 class ctor. Initialize m_in_call_function_usage. Call
27277 one_time_initialization.
27278 * rtl-tests.c (selftest::test_uncond_jump): Call
27279 set_new_first_and_last_insn.
27280 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
27281 * selftest-rtl.c: New file.
27282 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
27283 (selftest::get_insn_by_uid): New decl.
27284 * selftest-run-tests.c (selftest::run_tests): Call
27285 read_rtl_function_c_tests.
27286 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
27287 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
27290 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
27292 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
27293 operands in a special way. Assert that pos+len <= mode precision.
27295 2017-01-05 Jakub Jelinek <jakub@redhat.com>
27297 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
27298 3 argument Alias with unlimited for the negative form.
27299 (fno-vect-cost-model): Removed.
27301 2017-01-05 Martin Liska <mliska@suse.cz>
27303 * hsa-gen.c (gen_hsa_divmod): New function.
27304 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
27306 2017-01-05 Martin Liska <mliska@suse.cz>
27309 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
27312 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27314 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
27315 small constant length operands.
27317 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27319 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
27320 between loop iterations.
27322 2017-01-05 Martin Liska <mliska@suse.cz>
27325 * gimplify.c (gimplify_decl_expr): Compare to
27326 asan_poisoned_variables instread of checking flags.
27327 (gimplify_target_expr): Likewise.
27328 (gimplify_expr): Likewise.
27329 (gimplify_function_tree): Conditionally initialize
27330 asan_poisoned_variables.
27332 2017-01-04 Jeff Law <law@redhat.com>
27334 PR tree-optimizatin/78812
27335 * rtl.h (contains_mem_rtx_p): Prototype.
27336 * ifcvt.c (containts_mem_rtx_p): Move from here to...
27337 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
27338 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
27339 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
27340 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
27342 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27344 * input.c (assert_char_at_range): Default-initialize actual_range.
27346 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27348 * df-scan.c (df_ref_create_structure): Make regno unsigned,
27349 to match the caller.
27351 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27353 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
27354 insns after final jump in test to emit dummy move.
27356 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27358 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
27359 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
27361 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27363 * multiple_target.c (create_dispatcher_calls): Init e_next.
27364 * tree-ssa-loop-split.c (split_loop): Init border.
27365 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
27368 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
27373 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
27374 (altivec_register_operand): Do not return true if the operand
27375 contains a SUBREG mixing SImode and SFmode.
27376 (vsx_register_operand): Likewise.
27377 (vsx_reg_sfsubreg_ok): New predicate.
27378 (vfloat_operand): Do not return true if the operand contains a
27379 SUBREG mixing SImode and SFmode.
27380 (vint_operand): Likewise.
27381 (vlogical_operand): Likewise.
27382 (gpc_reg_operand): Likewise.
27383 (int_reg_operand): Likewise.
27384 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
27385 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
27386 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
27388 (rs6000_emit_move_si_sf_subreg): New helper function.
27389 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
27390 fixup SUBREGs involving SImode and SFmode.
27391 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
27392 numbers for the new peephole2 optimization.
27393 (peephole2 for SFmode unions): New peephole2 to optimize cases in
27394 the GLIBC math library that do AND/IOR/XOR operations on single
27395 precision floating point.
27396 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
27397 target macros to say whether we need to avoid SUBREGs mixing
27399 (TARGET_ALLOW_SF_SUBREG): Likewise.
27400 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
27401 (UNSPEC_SI_FROM_SF): Likewise.
27402 (iorxor): Change spacing.
27403 (and_ior_xor): New iterator for AND, IOR, and XOR.
27404 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
27405 (movdi_from_sf_zero_ext): Likewise.
27406 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
27407 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
27408 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
27409 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
27410 (fms<mode>4): Likewise.
27411 (fnma<mode>4): Likewise.
27412 (fnms<mode>4): Likewise.
27413 (nfma<mode>4): Likewise.
27414 (nfms<mode>4): Likewise.
27416 2017-01-04 Marek Polacek <polacek@redhat.com>
27419 * doc/invoke.texi: Document -Wpointer-compare.
27421 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27423 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
27426 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
27427 descriptions for -gdwarf-5 and emit them as uleb128 instead of
27430 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
27433 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
27434 documentation of the powerpc_popcntb_ok attribute.
27435 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27436 code to issue warning messages if a requested CPU configuration is
27437 not supported by the binary (assembler and loader) toolchain.
27438 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
27439 made to define a built-in function that has been disabled.
27440 (paired_init_builtins): Add assertion to prevent ICE if attempt is
27441 made to define a built-in function that has been disabled.
27442 (altivec_init_builtins): Add comment explaining why definition
27443 of the DST built-in functions is not preceded by an assertion
27444 check. Add assertions to prevent ICE if attempts are made to
27445 define an altivec predicate or an abs* built-in function that has
27447 (htm_init_builtins): Add comment explaining why definition of the
27448 htm built-in functions is not preceded by an assertion check.
27450 2017-01-04 Jeff Law <law@redhat.com>
27452 PR tree-optimizatin/67955
27453 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
27454 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
27455 the points-to solution does not include pt_null. Use DECL_PT_UID
27458 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
27460 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
27461 Use gen_int_mode instead of gen_lopwart for const_int operands.
27463 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27465 PR tree-optimization/71563
27466 * match.pd: Simplify X << Y into X if Y is known to be 0 or
27467 out of range value - has low bits known to be zero.
27469 2017-01-04 Alan Modra <amodra@gmail.com>
27471 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
27472 * configure: Regenerate.
27473 * config.in: Regenerate.
27475 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27478 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
27479 a substring of the message, but strcmp with the whole message. Ifdef
27480 ENABLE_NLS, translate the message first using dgettext.
27482 2017-01-03 Jeff Law <law@redhat.com>
27484 PR tree-optimizatin/78856
27485 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
27486 (mark_threaded_blocks): Remove code to truncate thread paths that
27487 cross multiple loop headers. Instead invalidate the cached loop
27488 iteration information and handle case of a thread path walking
27489 into an irreducible region.
27491 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
27494 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
27495 assertions. Add support for doing the signbit if the IEEE 128-bit
27496 floating point value is in a GPR.
27497 * config/rs6000/rs6000.md (Fsignbit): Delete.
27498 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
27499 Update the length attribute if the value is in a GPR.
27500 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
27501 the sign or zero extension instruction, since the value is always 0/1.
27502 (signbit<mode>2_dm2): Delete using <Fsignbit>.
27505 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
27506 extracting SImode to a GPR register so that we can generate a
27507 store, limit the vector to be in a traditional Altivec register
27508 for the vextuwrx instruction.
27510 2017-01-03 Ian Lance Taylor <iant@google.com>
27512 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
27514 2017-01-03 Martin Sebor <msebor@redhat.com>
27516 PR tree-optimization/78696
27517 * gimple-ssa-sprintf.c (format_floating): Correct handling of
27518 precision. Use MPFR for %f for greater fidelity. Correct handling
27520 (pass_sprintf_length::compute_format_length): Set width and precision
27521 specified by asrerisk to void_node for vararg functions.
27522 (try_substitute_return_value): Adjust dump output.
27524 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
27526 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
27528 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
27530 * doc/invoke.texi (SPARC options): Document -mlra as the default.
27531 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
27532 -mlra/-mno-lra was passed to the compiler.
27534 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
27536 PR rtl-optimization/65618
27537 * emit-rtl.c (try_split): Move initialization of "before" and
27538 "after" to just before the call to emit_insn_after_setloc.
27540 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
27542 * doc/md.texi (Standard Names): Remove reference to Java frontend.
27544 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
27546 * dwarf2out.c (gen_enumeration_type_die): When
27547 -gno-strict-dwarf, add a DW_AT_encoding attribute.
27549 2017-01-03 Jakub Jelinek <jakub@redhat.com>
27551 PR tree-optimization/78965
27552 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
27553 Change first argument from const call_info & to call_info &. For %n
27554 set info.nowrite to false.
27556 PR middle-end/78901
27557 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
27558 possibly throwing calls.
27560 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
27561 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
27562 and fns handling, rather than in a separate case SSA_NAME.
27564 2017-01-02 Jeff Law <law@redhat.com>
27566 * config/darwin-driver.c (darwin_driver_init): Const-correctness
27567 fixes for first_period and second_period variables.
27569 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
27572 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
27573 (*insvqi_1): New insn pattern.
27574 (*insvqi_1_mem_rex64): Ditto.
27575 (*insvqi_2): Ditto.
27576 (*insvqi_3): Rename from *insvqi.
27578 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
27580 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
27582 * doc/cfg.texi (Edges): Remove reference to Java.
27583 (Maintaining the CFG): Ditto.
27585 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27587 PR middle-end/77674
27588 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
27589 transparent aliases.
27591 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27593 PR middle-end/77484
27594 * predict.def (PRED_CALL): Update hitrate.
27595 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
27596 * predict.c (tree_estimate_probability_bb): Split CALL predictor
27597 into direct/indirect/polymorphic variants.
27599 2017-01-01 Jakub Jelinek <jakub@redhat.com>
27601 Update copyright years.
27603 * gcc.c (process_command): Update copyright notice dates.
27604 * gcov-dump.c (print_version): Ditto.
27605 * gcov.c (print_version): Ditto.
27606 * gcov-tool.c (print_version): Ditto.
27607 * gengtype.c (create_file): Ditto.
27608 * doc/cpp.texi: Bump @copying's copyright year.
27609 * doc/cppinternals.texi: Ditto.
27610 * doc/gcc.texi: Ditto.
27611 * doc/gccint.texi: Ditto.
27612 * doc/gcov.texi: Ditto.
27613 * doc/install.texi: Ditto.
27614 * doc/invoke.texi: Ditto.
27616 Copyright (C) 2017 Free Software Foundation, Inc.
27618 Copying and distribution of this file, with or without modification,
27619 are permitted in any medium without royalty provided the copyright
27620 notice and this notice are preserved.