1 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
3 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
5 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
7 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
9 * sel-sched-ir.c: Ditto.
11 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
13 * toplev.h (set_random_seed): Adjust return type.
14 * toplev.c (init_local_tick): Move eager initialization of random_seed
15 to get_random_seed. Adjust comment.
16 (init_random_seed): Inline to get_random_seed, delete.
17 (get_random_seed): Initialize random_seed lazily.
18 (set_random_seed): Do not return previous value.
19 (print_switch_value): Do not call get_random_seed.
21 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
23 * cgraph.c (delete_function_version): New, broken out from...
24 (cgraph_node::delete_function_version): ...here. Rename to
25 cgraph_node::delete_function_version_by_decl. Update all uses.
26 (cgraph_node::remove): Call delete_function_version.
28 2017-09-21 Jakub Jelinek <jakub@redhat.com>
31 * tree-inline.c (expand_call_inline): Emit clobber stmts for
32 VAR_DECLs to which addressable non-volatile parameters are mapped
33 and for id->retvar after the return value assignment. Clear
34 id->retval and id->retbnd after inlining.
36 2017-09-21 Richard Biener <rguenther@suse.de>
38 PR tree-optimization/82276
39 PR tree-optimization/82244
40 * tree-vrp.c (build_assert_expr_for): Set
41 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
43 (remove_range_assertions): Revert earlier change.
45 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
48 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
50 2017-09-21 Richard Biener <rguenther@suse.de>
52 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
53 Restore valid IL after code generation errors.
54 * graphite.c (graphite_transform_loops): Diagnose code
55 generation issues as MSG_MISSED_OPTIMIZATION and continue
56 with processing SCOPs.
58 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
59 Alan Hayward <alan.hayward@arm.com>
60 David Sherwood <david.sherwood@arm.com>
62 * calls.c (compute_argument_addresses): Use simplify_gen_binary
63 rather than choosing between plus_constant and gen_rtx_<CODE>.
64 * expr.c (emit_push_insn): Likewise.
65 (expand_expr_real_2): Likewise.
67 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
68 Alan Hayward <alan.hayward@arm.com>
69 David Sherwood <david.sherwood@arm.com>
71 * loop-unroll.c (split_iv): Call copy_rtx on the step.
73 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
74 Alan Hayward <alan.hayward@arm.com>
75 David Sherwood <david.sherwood@arm.com>
77 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
80 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
81 Alan Hayward <alan.hayward@arm.com>
82 David Sherwood <david.sherwood@arm.com>
84 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
85 INTEGER_CST rather than a negative test for ADDR_EXPR.
87 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
88 Alan Hayward <alan.hayward@arm.com>
89 David Sherwood <david.sherwood@arm.com>
91 * tree-vrp.c (extract_range_from_binary_expr_1): Check
92 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
94 2017-09-21 Richard Biener <rguenther@suse.de>
96 PR tree-optimization/71351
97 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
98 graphite_create_new_loop_guard): Remove, fold remaining parts
100 (translate_isl_ast_node_for): ... here and simplify.
102 2017-09-21 Jakub Jelinek <jakub@redhat.com>
105 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
106 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
107 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
108 alternative always use QI mode, for -Os imov (=R,R) alternative
109 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
112 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
113 Jeff Law <law@redhat.com>
115 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
116 (allocate_stack_space): New function, partially extracted from
118 (s390_emit_prologue): Track offset to most recent stack probe.
119 Code to allocate space moved into allocate_stack_space.
120 Dump actions when no stack is allocated.
121 (s390_prologue_plus_offset): New function.
122 (s390_emit_stack_probe): Likewise.
124 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
126 * common.opt (Wa, Wl, Wp, g, gz=): Add
128 (gno-column-info): Remove.
129 (gcolumn-info): Drop RejectNegative.
131 (gno-record-gcc-switches): Remove.
132 (grecord-gcc-switches): Drop RejectNegative.
133 (gno-split-dwarf): Remove.
134 (gsplit-dwarf): Drop RejectNegative.
135 (gno-strict-dwarf): Remove.
136 (gstrict-dwarf): Drop RejectNegative.
137 * config/darwin.opt (gfull, gused): Add RejectNegative.
138 * dwarf2out.c (gen_producer_string): Drop
139 gno-record-gcc-switches handler.
140 * optc-gen.awk: Add g to prefixes with negative forms.
141 * opts-common.c (remapping_prefix_p): New.
142 (find_opt): Check it.
143 (generate_canonical_option): Test g prefix.
144 (option_map): Add -gno- mapping.
145 (add_misspelling_candidates): Check remapping_prefix_p.
147 2017-09-20 Jeff Law <law@redhat.com>
149 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
150 thinko in stack clash protection support.
152 * explow.c (compute_stack_clash_protection_loop_data): Use
153 CONST_INT_P instead of explicit test. Verify object is a
154 CONST_INT_P before looking at INTVAL.
155 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
156 instead of explicit test.
158 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
161 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
162 address instead of to r1 and r11.
164 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
166 * config.gcc: Support "knm".
167 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
168 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
170 * config/i386/i386.c (m_KNM): Define.
171 (processor_target_table): Add "knm".
173 (ix86_option_override_internal): Add "knm".
174 (ix86_issue_rate): Add PROCESSOR_KNM.
175 (ix86_adjust_cost): Ditto.
176 (ia32_multipass_dfa_lookahead): Ditto.
177 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
178 (fold_builtin_cpu): Add M_INTEL_KNM.
179 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
180 (processor_type): Add PROCESSOR_KNM.
181 * config/i386/x86-tune.def: Add m_KNM.
182 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
184 2017-09-20 Richard Biener <rguenther@suse.de>
186 PR tree-optimization/80213
187 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
188 are allowed in empty BBs as well.
189 (canonicalize_loop_closed_ssa): Also look for other complex
191 (scop_detection::get_sese): Include the loop-closed PHI block
193 (scop_detection::merge_sese): Remove code adding extra blocks.
194 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
195 (build_scops): Assert the final returned scop is invalid.
197 2017-09-20 Richard Biener <rguenther@suse.de>
199 PR tree-optimization/82264
200 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
202 (vn_phi_lookup): Likewise.
203 (vn_phi_insert): Likewise.
205 2017-09-20 Jakub Jelinek <jakub@redhat.com>
207 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
208 that fits into uhwi or shwi, add DW_AT_const_value regardless
209 of early_dwarf without going through RTL, using add_AT_unsigned
212 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
213 (DEBUG_LTO_ABBREV_SECTION): Likewise.
214 (DEBUG_LTO_MACINFO_SECTION): Likewise.
215 (DEBUG_MACRO_SECTION): Likewise.
216 (DEBUG_LTO_MACRO_SECTION): Likewise.
217 (DEBUG_STR_DWO_SECTION): Likewise.
218 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
219 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
220 (DEBUG_LTO_DWO_LINE_SECTION): Define.
221 (DEBUG_LTO_LINE_STR_SECTION): Define.
222 (init_sections_and_labels): Initialize debug_line_str_section
223 variable. Initialize debug_loc_section for -gdwarf-5 to
224 DEBUG_LOCLISTS_SECTION. Formatting fixes.
226 2017-09-20 Richard Biener <rguenther@suse.de>
228 * graphite-sese-to-poly.c (extract_affine): Properly handle
229 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
231 2017-09-20 Richard Biener <rguenther@suse.de>
233 PR tree-optimization/81373
234 * graphite-scop-detection.c (build_cross_bb_scalars_def):
235 Force SESE live-out defs to be handled even if they are
238 2017-09-19 Jeff Law <law@redhat.com>
240 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
241 nothing for stack adjustments with REG_STACK_CHECK.
242 * sched-deps.c (parse_add_or_inc): Reject insns with
243 REG_STACK_CHECK from dependency breaking.
244 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
245 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
246 * reg-notes.def (STACK_CHECK): New note.
248 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
249 (ix86_expand_prologue): Dump stack clash info as needed.
250 Call ix86_adjust_stack_and_probe_stack_clash as needed.
252 * function.c (dump_stack_clash_frame_info): New function.
253 * function.h (dump_stack_clash_frame_info): Prototype.
254 (enum stack_clash_probes): New enum.
256 * config/alpha/alpha.c (alpha_expand_prologue): Also check
257 flag_stack_clash_protection.
258 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
259 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
260 (arm_frame_pointer_required): Likewise.
261 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
262 (ia64_expand_prologue): Likewise.
263 * config/mips/mips.c (mips_expand_prologue): Likewise.
264 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
265 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
266 (sparc_flat_expand_prologue): Likewise.
267 * config/spu/spu.c (spu_expand_prologue): Likewise.
269 * explow.c: Include "params.h".
270 (anti_adjust_stack_and_probe_stack_clash): New function.
271 (get_stack_check_protect): Likewise.
272 (compute_stack_clash_protection_loop_data): Likewise.
273 (emit_stack_clash_protection_loop_start): Likewise.
274 (emit_stack_clash_protection_loop_end): Likewise.
275 (allocate_dynamic_stack_space): Use get_stack_check_protect.
276 Use anti_adjust_stack_and_probe_stack_clash.
277 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
278 (emit_stack_clash_protection_loop_start): Likewise.
279 (emit_stack_clash_protection_loop_end): Likewise.
280 * rtl.h (get_stack_check_protect): Prototype.
281 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
282 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
283 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
285 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
287 * doc/tm.texi: Rebuilt.
288 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
289 get_stack_check_protect.
290 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
291 * config/arm/arm.c (arm_expand_prologue): Likewise.
292 (arm_frame_pointer_required): Likewise.
293 * config/i386/i386.c (ix86_expand_prologue): Likewise.
294 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
295 * config/mips/mips.c (mips_expand_prologue): Likewise.
296 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
297 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
298 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
299 (sparc_flat_expand_prologue): Likewise.
301 * common.opt (-fstack-clash-protection): New option.
302 * flag-types.h (enum stack_check_type): Note difference between
303 -fstack-check= and -fstack-clash-protection.
304 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
305 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
306 * toplev.c (process_options): Issue warnings/errors for cases
307 not handled with -fstack-clash-protection.
308 * doc/invoke.texi (-fstack-clash-protection): Document new option.
309 (-fstack-check): Note additional problem with -fstack-check=generic.
310 Note that -fstack-check is primarily for Ada and refer users
311 to -fstack-clash-protection for stack-clash-protection.
312 Document new params for stack clash protection.
314 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
316 * config/i386/i386.c (ix86_split_long_move): Do not handle
317 address used for LEA in a special way.
319 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
321 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
323 2017-09-19 Martin Sebor <msebor@redhat.com>
326 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
327 of incompatible types.
329 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
331 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
332 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
333 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
334 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
336 2017-09-19 Richard Biener <rguenther@suse.de>
338 PR tree-optimization/82244
339 * tree-vrp.c (remove_range_assertions): Do not propagate
340 a constant to abnormals but replace the assert with a copy.
342 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
344 PR rtl-optimization/57878
345 PR rtl-optimization/68988
346 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
347 avoidance test involving non_reload_pseudos. Move frequency test
348 below the general fragmentation avoidance test.
350 2017-09-19 Richard Biener <rguenther@suse.de>
352 PR tree-optimization/69728
353 * graphite-sese-to-poly.c (schedule_error): New global.
354 (add_loop_schedule): Handle empty domain by failing the
356 (build_original_schedule): Handle schedule_error.
358 2017-09-19 Richard Biener <rguenther@suse.de>
360 * graphite-scop-detection.c (scop_detection::can_represent_loop):
361 Do not iterate to sibling loops but only to siblings of inner
364 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
367 * config/m68k/m68k.md (moveq feeding equality comparison): Check
368 that the registers are different.
370 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
372 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
373 to processor_model and "amdfam17h" to arch_names_table.
374 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
376 2017-09-18 Jakub Jelinek <jakub@redhat.com>
379 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
381 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
382 Alan Hayward <alan.hayward@arm.com>
383 David Sherwood <david.sherwood@arm.com>
385 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
387 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
389 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
390 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
391 vect_schedule_slp_instance.
392 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
393 Update call to vect_slp_analyze_node_operations. Simplify return
395 (vect_slp_analyze_bb_1): Update call accordingly.
396 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
397 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
398 (vect_schedule_slp): Update call accordingly.
400 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
401 Alan Hayward <alan.hayward@arm.com>
402 David Sherwood <david.sherwood@arm.com>
404 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
405 with types that aren't in fact scalar.
407 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
409 * tree-vect-slp.c (vect_record_max_nunits): New function,
411 (vect_build_slp_tree_1): ...here.
412 (vect_build_slp_tree_2): Call it for phis too.
414 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
416 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
417 to vect_get_vec_def_for_operand when getting the mask operand.
419 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
420 Alan Hayward <alan.hayward@arm.com>
421 David Sherwood <david.sherwood@arm.com>
423 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
426 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
427 Alan Hayward <alan.hayward@arm.com>
428 David Sherwood <david.sherwood@arm.com>
430 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
431 calculation for vector booleans.
433 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
434 Alan Hayward <alan.hayward@arm.com>
435 David Sherwood <david.sherwood@arm.com>
437 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
439 (vect_transform_stmt): ...here.
440 (vect_analyze_stmt): Use it instead of calling
441 vectorizable_live_operation directly.
443 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
445 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
446 non-SIMT targets in acc vector loops.
448 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
450 * configure.ac: Add arc and check if assembler supports gdwarf2.
451 * configure: Regenerate.
453 2017-09-18 Richard Biener <rguenther@suse.de>
455 PR tree-optimization/82220
456 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
457 epilogue niters from the min_profitable_iters compute.
459 2017-09-18 Jakub Jelinek <jakub@redhat.com>
462 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
463 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
464 (ix86_init_pic_reg): Revert 2017-09-01 changes.
466 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
469 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
470 switching to a new text section.
472 2017-09-18 Richard Biener <rguenther@suse.de>
474 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
476 (build_alias_set): Reject aliases with no access function.
478 2017-09-18 Richard Biener <rguenther@suse.de>
480 PR tree-optimization/79622
481 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
483 (build_cross_bb_scalars_use): Likewise.
485 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
487 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
489 2017-09-18 Alan Modra <amodra@gmail.com>
492 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
493 stack_pointer_rtx for count 0. Update comments. Break up
494 large rtl expression.
496 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
498 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
499 Increase to 20 bytes.
500 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
501 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
502 or avx version of the stub.
504 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
507 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
508 compute the minimum stack alignment. Also update preferred stack
509 boundary for leaf functions.
511 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
513 PR tree-optimization/82228
514 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
517 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
519 * common/config/nds32/nds32-common.c
520 (nds32_option_optimization_table): Refine formatting.
521 (nds32_option_optimization_table): Use -fsched-pressure and
522 -fomit-frame-pointer for specific optimization level.
524 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
526 * config/nds32/nds32.c: Refine formatting and comments.
527 * config/nds32/nds32.h: Likewise.
528 * config/nds32/nds32.md: Likewise.
529 * config/nds32/nds32-cost.c: Likewise.
530 * config/nds32/nds32-isr.c: Likewise.
531 * config/nds32/nds32-md-auxiliary.c: Likewise.
532 * config/nds32/nds32-multiple.md: Likewise.
533 * config/nds32/nds32-predicates.c: Likewise.
535 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
536 Jakub Jelinek <jakub@redhat.com>
538 Add support for -std=c++2a.
539 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
541 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
543 2017-09-15 Steve Ellcey <sellcey@cavium.com>
546 * doc/extend.texi (Common Function Attributes): Add
547 references to ARM, AArch64, and S/390 specific attributes.
548 (Function Specific Option Pragmas): Add AArch64 and S/390
549 to list of back ends that support the target pragma.
551 2017-09-15 Nathan Sidwell <nathan@acm.org>
553 * doc/standards.texi: Fix C++17 description. Update URLs for
556 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
558 * common.opt (Wcast-align=strict): New warning option.
559 * doc/invoke.texi: Document -Wcast-align=strict.
561 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
563 * cgraph.h (cgraph_thunk_info): Add comments.
564 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
565 assert for VIRTUAL_* arguments stricter.
567 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
569 PR tree-optimization/71026
570 * match.pd: Move RDIV patterns from fold-const.c
571 * fold-const.c (distribute_real_division): Removed.
572 (fold_binary_loc): Remove calls to distribute_real_divison.
574 2017-09-15 Jakub Jelinek <jakub@redhat.com>
576 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
577 c++1z and gnu++1z as deprecated. Change other references to
578 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
579 Change -Wc++1z-compat to -Wc++17-compat.
580 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
581 * dwarf2out.c (highest_c_language): Handle C++17.
582 (gen_compile_unit_die): Likewise.
584 2017-09-15 Jakub Jelinek <jakub@redhat.com>
586 PR rtl-optimization/82192
587 * combine.c (make_extraction): Don't look through non-paradoxical
588 SUBREGs or TRUNCATE if pos + len is or might be bigger than
591 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
592 Alan Hayward <alan.hayward@arm.com>
593 David Sherwood <david.sherwood@arm.com>
595 * target.def (function_arg_offset): New hook.
596 * targhooks.h (default_function_arg_offset): Declare.
597 * targhooks.c (default_function_arg_offset): New function.
598 * function.c (locate_and_pad_parm): Use
599 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
600 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
601 (TARGET_FUNCTION_ARG_OFFSET): ...this.
602 * doc/tm.texi: Regenerate.
603 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
604 * config/spu/spu.c (spu_function_arg_offset): New function.
605 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
606 * system.h (FUNCTION_ARG_OFFSET): Poison.
608 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
609 Alan Hayard <alan.hayward@arm.com>
610 David Sherwood <david.sherwood@arm.com>
612 * target.def (truly_noop_truncation): New hook.
613 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
614 than TRULY_NOOP_TRUNCATION.
615 * hooks.h (hook_bool_uint_uint_true): Declare.
616 * hooks.c (hook_bool_uint_uint_true): New function.
617 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
618 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
619 * doc/tm.texi: Regenerate.
620 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
621 rather than TRULY_NOOP_TRUNCATION in comments.
622 (simplify_comparison): Likewise.
623 (record_truncated_value): Likewise.
624 * expmed.c (extract_bit_field_1): Likewise.
625 (extract_split_bit_field): Likewise.
626 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
627 instead of TRULY_NOOP_TRUNCATION.
628 * function.c (assign_parm_setup_block): Likewise.
629 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
630 * rtlhooks.c: Include target.h.
631 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
632 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
633 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
634 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
635 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
636 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
637 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
638 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
639 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
640 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
641 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
642 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
643 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
644 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
645 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
646 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
647 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
648 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
649 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
650 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
651 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
652 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
653 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
654 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
655 * config/mips/mips.c (mips_truly_noop_truncation): New function.
656 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
657 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
658 rather than TRULY_NOOP_TRUNCATION in comments.
659 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
660 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
661 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
662 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
663 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
664 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
665 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
666 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
667 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
668 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
669 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
670 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
671 rather than TRULY_NOOP_TRUNCATION in comments.
672 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
673 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
674 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
675 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
676 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
677 TRULY_NOOP_TRUNCATION condition.
678 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
679 (TRULY_NOOP_TRUNCATION): Delete.
680 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
681 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
682 * config/spu/spu.c (spu_truly_noop_truncation): New function.
683 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
684 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
685 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
686 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
687 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
688 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
689 rather than TRULY_NOOP_TRUNCATION in comments.
690 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
691 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
692 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
693 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
694 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
695 * system.h (TRULY_NOOP_TRUNCATION): Poison.
697 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
700 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
701 (*cmp_ior): Likewise.
702 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
703 (*ior_scc_scc_cmp): Likewise.
704 (*and_scc_scc): Likewise.
705 (*and_scc_scc_cmp): Likewise.
707 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
708 Alan Hayard <alan.hayward@arm.com>
709 David Sherwood <david.sherwood@arm.com>
711 * target.def (can_change_mode_class): New hook.
712 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
713 (hard_regno_nregs): Likewise.
714 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
715 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
716 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
717 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
718 (LOAD_EXTEND_OP): Update accordingly.
719 * doc/tm.texi: Regenerate.
720 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
721 CANNOT_CHANGE_MODE_CLASS.
722 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
723 (REG_CAN_CHANGE_MODE_P): ...this new macro.
724 * combine.c (simplify_set): Update accordingly.
725 * emit-rtl.c (validate_subreg): Likewise.
726 * recog.c (general_operand): Likewise.
727 * regcprop.c (mode_change_ok): Likewise.
728 * reload1.c (choose_reload_regs): Likewise.
729 (inherit_piecemeal_p): Likewise.
730 * rtlanal.c (simplify_subreg_regno): Likewise.
731 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
732 instead of CANNOT_CHANGE_MODE_CLASS.
733 (reload_cse_simplify_operands): Likewise.
734 * reload.c (push_reload): Use targetm.can_change_mode_class
735 instead of CANNOT_CHANGE_MODE_CLASS.
736 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
737 REG_CANNOT_CHANGE_MODE_P.
738 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
739 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
740 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
741 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
742 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
743 (arm_can_change_mode_class): New function.
744 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
745 than CANNOT_CHANGE_MODE_CLASS in comments.
746 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
747 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
748 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
749 (ix86_can_change_mode_class): ...this new function, inverting the
750 sense of the return value.
751 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
752 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
753 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
754 (ia64_can_change_mode_class): New function.
755 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
756 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
757 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
758 (m32c_can_change_mode_class): ...this new function, inverting the
759 sense of the return value.
760 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
761 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
762 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
763 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
764 (mips_can_change_mode_class): ...this new function, inverting the
765 sense of the return value.
766 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
767 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
768 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
769 (msp430_can_change_mode_class): New function.
770 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
771 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
772 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
773 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
774 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
775 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
776 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
777 (pa_cannot_change_mode_class): Replace with...
778 (pa_can_change_mode_class): ...this new function, inverting the
779 sense of the return value.
780 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
781 than CANNOT_CHANGE_MODE_CLASS in comments.
782 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
783 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
784 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
785 (pdp11_cannot_change_mode_class): Replace with...
786 (pdp11_can_change_mode_class): ...this new function, inverting the
787 sense of the return value.
788 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
789 * config/powerpcspe/powerpcspe-protos.h
790 (rs6000_cannot_change_mode_class_ptr): Delete.
791 * config/powerpcspe/powerpcspe.c
792 (rs6000_cannot_change_mode_class_ptr): Delete.
793 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
794 (rs6000_option_override_internal): Assign to
795 targetm.can_change_mode_class instead of
796 rs6000_cannot_change_mode_class_ptr.
797 (rs6000_cannot_change_mode_class): Replace with...
798 (rs6000_can_change_mode_class): ...this new function, inverting the
799 sense of the return value.
800 (rs6000_debug_cannot_change_mode_class): Replace with...
801 (rs6000_debug_can_change_mode_class): ...this new function.
802 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
803 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
804 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
805 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
806 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
808 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
809 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
810 (rs6000_option_override_internal): Assign to
811 targetm.can_change_mode_class instead of
812 rs6000_cannot_change_mode_class_ptr.
813 (rs6000_cannot_change_mode_class): Replace with...
814 (rs6000_can_change_mode_class): ...this new function, inverting the
815 sense of the return value.
816 (rs6000_debug_cannot_change_mode_class): Replace with...
817 (rs6000_debug_can_change_mode_class): ...this new function.
818 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
819 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
820 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
821 (s390_can_change_mode_class): ...this new function, inverting the
822 sense of the return value.
823 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
824 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
825 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
826 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
827 (sh_cannot_change_mode_class): Replace with...
828 (sh_can_change_mode_class): ...this new function, inverting the
829 sense of the return value.
830 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
831 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
832 (sparc_can_change_mode_class): New function.
833 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
834 * config/spu/spu.c (spu_can_change_mode_class): New function.
835 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
836 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
837 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
838 (visium_can_change_mode_class): New function.
839 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
841 2017-09-15 Richard Biener <rguenther@suse.de>
843 PR tree-optimization/82217
844 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
845 but not undefined case.
847 2017-09-15 Jakub Jelinek <jakub@redhat.com>
850 * postreload.c (reload_cse_simplify_operands): Skip
851 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
853 2017-09-15 Richard Biener <rguenther@suse.de>
855 PR tree-optimization/68823
856 * graphite-scop-detection.c (build_alias_set): If we have a
857 possible dependence check whether we can handle them by just
858 looking at the DRs DR_ACCESS_FNs.
859 (build_scops): If build_alias_set fails, fail the SCOP.
861 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
863 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
864 to support float128 built-in functions that require the ISA 3.0
866 (BU_FLOAT128_3_HW): Likewise.
867 (SQRTF128): Add support for the IEEE 128-bit square root and fma
871 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
872 support for built-in functions that need the ISA 3.0 IEEE 128-bit
873 floating point instructions.
874 (rs6000_invalid_builtin): Likewise.
875 (rs6000_builtin_mask_names): Likewise.
876 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
877 (RS6000_BTM_FLOAT128_HW): Likewise.
878 (RS6000_BTM_COMMON): Likewise.
879 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
881 * doc/extend.texi (RS/6000 built-in functions): Document the
882 IEEE 128-bit floating point square root and fused multiply-add
885 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
887 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
888 reg (r2) isn't in the set of registers defined in the prologue.
890 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
891 Alan Hayward <alan.hayward@arm.com>
892 David Sherwood <david.sherwood@arm.com>
894 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
895 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
896 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
897 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
898 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
900 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
901 max_vectorization_factor.
902 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
904 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
905 Alan Hayward <alan.hayward@arm.com>
906 David Sherwood <david.sherwood@arm.com>
908 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
909 (vect_worthwhile_without_simd_p): Declare.
910 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
911 (vectorizable_reduction): Use it.
912 * tree-vect-stmts.c (vectorizable_shift): Likewise.
913 (vectorizable_operation): Likewise.
915 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
916 Alan Hayward <alan.hayward@arm.com>
917 David Sherwood <david.sherwood@arm.com>
919 * tree-vectorizer.h (vect_get_num_copies): New function.
920 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
921 * tree-vect-loop.c (vectorizable_reduction): Likewise.
922 (vectorizable_induction): Likewise.
923 (vectorizable_live_operation): Likewise.
924 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
925 (vectorizable_bswap): Likewise.
926 (vectorizable_call): Likewise.
927 (vectorizable_conversion): Likewise.
928 (vectorizable_assignment): Likewise.
929 (vectorizable_shift): Likewise.
930 (vectorizable_operation): Likewise.
931 (vectorizable_store): Likewise.
932 (vectorizable_load): Likewise.
933 (vectorizable_condition): Likewise.
934 (vectorizable_comparison): Likewise.
935 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
937 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
938 Alan Hayward <alan.hayward@arm.com>
939 David Sherwood <david.sherwood@arm.com>
941 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
944 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
945 Alan Hayward <alan.hayward@arm.com>
946 David Sherwood <david.sherwood@arm.com>
948 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
949 an inline wrapper that provides a location.
950 (gimple_build_vector): Likewise.
951 * gimple-fold.c (gimple_build_vector_from_val): New function.
952 (gimple_build_vector): Likewise.
953 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
954 functions to build the initial value. Always return a gimple value.
955 (get_initial_defs_for_reduction): Likewise. Only compute
957 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
958 vect_init_vector on the results from get_initial_def(s)_for_reduction.
959 (vectorizable_induction): Use gimple_build_vector rather than
962 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
963 Alan Hayward <alan.hayward@arm.com>
964 David Sherwood <david.sherwood@arm.com>
966 * target.h (vec_perm_indices): New typedef.
967 (auto_vec_perm_indices): Likewise.
968 * optabs-query.h: Include target.h
969 (can_vec_perm_p): Take a vec_perm_indices *.
970 * optabs-query.c (can_vec_perm_p): Likewise.
971 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
972 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
973 * tree-vect-generic.c (lower_vec_perm): Likewise.
974 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
975 (vect_grouped_load_supported): Likewise.
976 (vect_shift_permute_load_chain): Likewise.
977 (vect_permute_store_chain): Use auto_vec_perm_indices.
978 (vect_permute_load_chain): Likewise.
979 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
980 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
981 Update uses of can_vec_perm_p.
982 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
983 mode with a number of elements. Take a vec_perm_indices *.
984 (vect_create_epilog_for_reduction): Update accordingly.
985 Use auto_vec_perm_indices.
986 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
987 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
988 (vect_transform_slp_perm_load): Likewise.
989 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
990 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
991 (vect_gen_perm_mask_checked): Likewise.
992 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
993 (vect_gen_perm_mask_checked): Likewise.
994 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
995 (vectorizable_store): Likewise.
996 (vectorizable_load): Likewise.
997 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
998 (vectorizable_bswap): Likewise.
1000 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1001 Alan Hayward <alan.hayward@arm.com>
1002 David Sherwood <david.sherwood@arm.com>
1004 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
1005 * tree.c (build_vector): Likewise.
1006 (build_vector_from_ctor): Update accordingly.
1007 (build_vector_from_val): Likewise.
1008 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
1009 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
1010 * tree-vect-generic.c (add_rshift): Likewise.
1011 (expand_vector_divmod): Likewise.
1012 (optimize_vector_constructor): Likewise.
1013 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
1014 (vect_transform_slp_perm_load): Likewise.
1015 (vect_schedule_slp_instance): Likewise.
1016 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
1017 (vectorizable_call): Likewise.
1018 (vect_gen_perm_mask_any): Likewise. Add elements in order.
1019 * expmed.c (make_tree): Likewise.
1020 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
1021 a vector passed to build_vector.
1022 (fold_convert_const): Likewise.
1023 (exact_inverse): Likewise.
1024 (fold_ternary_loc): Likewise.
1025 (fold_relational_const): Likewise.
1026 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
1027 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
1028 (const_unop): Likewise. Store the reduction accumulator in a
1029 variable rather than an array.
1030 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
1031 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
1032 the new vector, rather than constructing it after the input arrays.
1033 (native_interpret_vector): Use auto_vec<tree> when building
1034 a vector passed to build_vector. Add elements in order.
1035 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
1036 auto_vec<tree> when building a vector passed to build_vector.
1037 (vect_create_epilog_for_reduction): Likewise.
1038 (vectorizable_induction): Likewise.
1039 (get_initial_def_for_reduction): Likewise. Fix indentation of
1041 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
1043 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
1044 passed to build_vector.
1046 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1047 Alan Hayward <alan.hayward@arm.com>
1048 David Sherwood <david.sherwood@arm.com>
1050 * tree-core.h (tree_base::u): Add an "nelts" field.
1051 (tree_vector): Use VECTOR_CST_NELTS as the length.
1052 * tree.c (tree_size): Likewise.
1053 (make_vector): Initialize VECTOR_CST_NELTS.
1054 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
1055 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
1056 TYPE_VECTOR_SUBPARTS.
1057 * expr.c (const_vector_mask_from_tree): Consistently use "units"
1058 as the number of units, setting it from VECTOR_CST_NELTS.
1059 (const_vector_from_tree): Likewise.
1060 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
1061 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
1062 (fold_negate_expr_1): Likewise.
1063 (fold_convert_const): Likewise.
1064 (const_binop): Likewise. Differentiate the number of output and
1066 (const_unop): Likewise.
1067 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
1068 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
1069 in cases that did the opposite.
1071 2017-09-14 Richard Biener <rguenther@suse.de>
1073 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
1076 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
1078 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
1080 2017-09-14 Jakub Jelinek <jakub@redhat.com>
1083 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
1084 if and where to split a bb, except for splitting before debug insn
1085 sequences followed by non-label real insn. Delete debug insns
1086 in between basic blocks.
1088 * combine.c (make_compound_operation_int): Formatting fixes.
1090 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
1091 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1092 * config/netbsd.h (LINK_EH_SPEC): Likewise.
1093 * config/sol2.h (LINK_EH_SPEC): Likewise.
1094 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1095 * config/s390/linux.h (LINK_SPEC): Likewise.
1096 * config/freebsd.h (LINK_EH_SPEC): Likewise.
1097 * config/openbsd.h (LINK_EH_SPEC): Likewise.
1098 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1099 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
1100 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
1101 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1102 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
1103 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
1105 2017-09-13 Jakub Jelinek <jakub@redhat.com>
1107 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
1109 (ENDFILE_LINUX_SPEC): Likewise.
1110 (LINK_EH_SPEC): Likewise.
1111 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
1112 (LINK_OS_LINUX_SPEC32): Likewise.
1113 (LINK_OS_LINUX_SPEC64): Likewise.
1114 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
1115 (LINK_OS_LINUX_SPEC): Likewise.
1117 2017-09-13 Martin Liska <mliska@suse.cz>
1120 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
1121 CASE_HIGH is NULL_TREE.
1123 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
1124 Alan Hayward <alan.hayward@arm.com>
1125 David Sherwood <david.sherwood@arm.com>
1127 * target.def (secondary_memory_needed): New hook.
1128 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
1129 instead of SECONDARY_MEMORY_NEEDED.
1130 (secondary_memory_needed_mode): Likewise.
1131 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
1132 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
1133 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
1134 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
1135 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
1136 * doc/tm.texi: Regenerate.
1137 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
1138 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
1139 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1140 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
1141 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
1142 * config/i386/i386.c (inline_secondary_memory_needed): Put the
1143 mode argument first and change the reg_class arguments to reg_class_t.
1144 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
1145 Make static. Update the call to inline_secondary_memory_needed.
1146 (ix86_register_move_cost): Update the call to
1147 inline_secondary_memory_needed.
1148 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1149 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
1151 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
1152 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
1154 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
1155 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
1156 * config/mips/mips.c (mips_secondary_memory_needed): Make static
1157 and match hook interface. Add comment from mips.h.
1158 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1159 * config/mmix/mmix.md (truncdfsf2): Refer to
1160 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
1162 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
1163 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
1164 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1165 (pa_secondary_memory_needed): New function.
1166 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
1167 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
1168 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1169 (pdp11_secondary_memory_needed): Make static and match hook interface.
1170 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
1171 * config/powerpcspe/powerpcspe-protos.h
1172 (rs6000_secondary_memory_needed_ptr): Delete.
1173 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
1175 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1176 (rs6000_option_override_internal): Assign to
1177 targetm.secondary_memory_needed rather than
1178 rs6000_secondary_memory_needed_ptr.
1179 (rs6000_secondary_memory_needed): Match hook interface.
1180 (rs6000_debug_secondary_memory_needed): Likewise.
1181 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
1182 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
1183 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
1184 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1185 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
1186 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
1188 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
1189 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1190 (rs6000_option_override_internal): Assign to
1191 targetm.secondary_memory_needed rather than
1192 rs6000_secondary_memory_needed_ptr.
1193 (rs6000_secondary_memory_needed): Match hook interface.
1194 (rs6000_debug_secondary_memory_needed): Likewise.
1195 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
1196 * config/s390/s390.c (s390_secondary_memory_needed): New function.
1197 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1198 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
1199 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1200 (sparc_secondary_memory_needed): New function.
1201 * lra-constraints.c (check_and_process_move): Refer to
1202 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
1204 (curr_insn_transform): Likewise.
1205 (process_alt_operands): Use targetm.secondary_memory_needed
1206 instead of TARGET_SECONDARY_MEMORY_NEEDED.
1207 (check_secondary_memory_needed_p): Likewise.
1208 (choose_split_class): Likewise.
1209 * reload.c: Unconditionally include code that was previously
1210 conditional on SECONDARY_MEMORY_NEEDED.
1211 (push_secondary_reload): Use targetm.secondary_memory_needed
1212 instead of TARGET_SECONDARY_MEMORY_NEEDED.
1213 (push_reload): Likewise.
1214 * reload1.c: Unconditionally include code that was previously
1215 conditional on SECONDARY_MEMORY_NEEDED.
1216 (choose_reload_regs): Use targetm.secondary_memory_needed
1217 instead of TARGET_SECONDARY_MEMORY_NEEDED.
1218 (gen_reload): Likewise.
1219 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
1221 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
1222 Alan Hayward <alan.hayward@arm.com>
1223 David Sherwood <david.sherwood@arm.com>
1225 * target.def (secondary_memory_needed_mode): New hook:
1226 * targhooks.c (default_secondary_memory_needed_mode): Declare.
1227 * targhooks.h (default_secondary_memory_needed_mode): New function.
1228 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
1229 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
1230 * doc/tm.texi: Regenerate.
1231 * lra-constraints.c (check_and_process_move): Use
1232 targetm.secondary_memory_needed_mode instead of
1233 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
1234 (curr_insn_transform): Likewise.
1235 * reload.c (get_secondary_mem): Likewise.
1236 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1237 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
1239 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1240 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1241 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
1242 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1243 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
1245 * config/powerpcspe/powerpcspe-protos.h
1246 (rs6000_secondary_memory_needed_mode): Delete.
1247 * config/powerpcspe/powerpcspe.c
1248 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1249 (rs6000_secondary_memory_needed_mode): Make static.
1250 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1251 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
1253 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
1255 (rs6000_secondary_memory_needed_mode): Make static.
1256 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1257 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
1258 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1259 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1260 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
1262 (sparc_secondary_memory_needed_mode): New function.
1263 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
1265 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
1267 * config/aarch64/constraints.md (Umq): New constraint.
1268 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
1270 (mov<mode>): Update condition.
1272 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1274 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
1275 when bitposition is the same.
1277 2017-09-13 Richard Biener <rguenther@suse.de>
1279 * dwarf2out.c (output_die_symbol): Remove.
1280 (output_die): Do not output a DIEs symbol.
1282 2017-09-13 Richard Biener <rguenther@suse.de>
1285 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
1286 default-def to avoid breaking iterator update with the weird
1287 interaction with cgraph_update_edges_for_call_stmt_node.
1289 2017-09-13 Richard Biener <rguenther@suse.de>
1291 * tree-cfg.c (verify_gimple_assign_binary): Add verification
1292 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
1293 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
1294 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
1296 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
1298 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
1299 Disable pc relative literal load irrespective of
1300 TARGET_FIX_ERR_A53_84341 for default.
1302 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
1304 * config/sparc/sparc.c (output_return): Output the source location of
1305 the insn in the delay slot, if any.
1306 (output_sibcall): Likewise.
1308 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
1311 * common.opt (-static-pie): New alias.
1312 (shared): Negate static-pie.
1313 (-no-pie): Update help text.
1315 (static-pie): New option.
1316 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
1317 -static-pie support.
1318 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
1319 (LINK_EH_SPEC): Likewise.
1320 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1321 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
1322 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
1323 * gcc.c (LINK_COMMAND_SPEC): Likewise.
1324 (init_gcc_specs): Likewise.
1325 (init_spec): Likewise.
1326 (display_help): Update help message for -pie.
1327 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
1330 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
1332 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
1333 (movdi_aarch64): Likewise.
1334 (movti_aarch64): Likewise.
1336 2017-09-12 Simon Wright <simon@pushface.org>
1339 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
1340 calculation of the minor version, always output as 0.
1342 2017-09-12 Jakub Jelinek <jakub@redhat.com>
1345 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
1346 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
1347 on it early, rather than manual conversion late. For
1348 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
1349 instead of performing manual conversion.
1351 2017-09-12 Carl Love <cel@us.ibm.com>
1353 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
1354 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
1356 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
1357 VMULOSW): Add definitions.
1358 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1359 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
1360 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
1361 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
1362 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
1364 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
1366 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
1368 (movti_aarch64): Likewise.
1369 (movdf_aarch64): Likewise.
1370 (movtf_aarch64): Likewise.
1371 (load_pairdi): Likewise.
1372 (store_pairdi): Likewise.
1373 (load_pairdf): Likewise.
1374 (store_pairdf): Likewise.
1375 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
1376 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
1377 (ldr_got_small_<mode>): Likewise.
1378 (ldr_got_small_28k_<mode>): Likewise.
1379 (ldr_got_tiny): Likewise.
1380 * config/aarch64/iterators.md (ldst_sz): New.
1381 (ldpstp_sz): Likewise.
1382 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
1384 (thunderx_load): Split load_8 to load_16.
1385 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
1387 (thunderx2t99_storepair_basic): Split store_8 to store_16.
1388 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
1389 (xgene1_store_pair): Split store_8 to store_16.
1390 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
1391 (falkor_st_0_st_sd): Split store_8 to store_16.
1393 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
1395 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
1396 and store1/2/3/4 to store_4/8/12/16.
1397 * config/aarch64/aarch64.md: Update for rename.
1398 * config/arm/arm.md: Likewise.: Likewise.
1399 * config/arm/arm.c: Likewise.
1400 * config/arm/thumb1.md: Likewise.
1401 * config/arm/thumb2.md: Likewise.
1402 * config/arm/vfp.md: Likewise.
1403 * config/arm/arm-generic.md: Likewise.
1404 * config/arm/arm1020e.md: Likewise.
1405 * config/arm/arm1026ejs.md: Likewise.
1406 * config/arm/arm1136jfs.md: Likewise.
1407 * config/arm/arm926ejs.md: Likewise.
1408 * config/arm/cortex-a15.md: Likewise.
1409 * config/arm/cortex-a17.md: Likewise.
1410 * config/arm/cortex-a5.md: Likewise.
1411 * config/arm/cortex-a53.md: Likewise.
1412 * config/arm/cortex-a57.md: Likewise.
1413 * config/arm/cortex-a7.md: Likewise.
1414 * config/arm/cortex-a8.md: Likewise.
1415 * config/arm/cortex-a9.md: Likewise.
1416 * config/arm/cortex-m4.md: Likewise.
1417 * config/arm/cortex-m7.md: Likewise.
1418 * config/arm/cortex-r4.md: Likewise.
1419 * config/arm/exynos-m1.md: Likewise.
1420 * config/arm/fa526.md: Likewise.
1421 * config/arm/fa606te.md: Likewise.
1422 * config/arm/fa626te.md: Likewise.
1423 * config/arm/fa726te.md: Likewise.
1424 * config/arm/fmp626.md: Likewise.
1425 * config/arm/iwmmxt.md: Likewise.
1426 * config/arm/ldmstm.md: Likewise.
1427 * config/arm/marvell-pj4.md: Likewise.
1428 * config/arm/xgene1.md: Likewise.
1429 * config/aarch64/thunderx.md: Likewise.
1430 * config/aarch64/thunderx2t99.md: Likewise.
1431 * config/aarch64/falkor.md: Likewise.
1433 2017-09-12 Martin Liska <mliska@suse.cz>
1435 * attribs.c (private_lookup_attribute): New function.
1436 * attribs.h (private_lookup_attribute): Declared here.
1437 (lookup_attribute): Called from this place.
1439 2017-09-12 Richard Biener <rguenther@suse.de>
1441 PR tree-optimization/82157
1442 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
1443 stmts with side-effects.
1445 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1446 Alan Hayward <alan.hayward@arm.com>
1447 David Sherwood <david.sherwood@arm.com>
1449 * target.def (hard_regno_nregs): New hook.
1450 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
1451 * targhooks.h (default_hard_regno_nregs): Declare.
1452 * targhooks.c (default_hard_regno_nregs): New function.
1453 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
1454 (TARGET_HARD_REGNO_NREGS): ...this hook.
1455 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
1456 (CLASS_MAX_NREGS): Likewise.
1457 * doc/tm.texi: Regenerate.
1458 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
1459 instead of HARD_REGNO_NREGS.
1460 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
1461 HARD_REGNO_NREGS in the comment.
1462 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
1463 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
1464 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
1465 Return an unsigned int.
1466 (TARGET_HARD_REGNO_NREGS): Redefine.
1467 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
1468 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
1469 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
1470 (arc_hard_regno_nregs): New function.
1471 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
1472 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
1473 (arm_hard_regno_nregs): New function.
1474 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
1475 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
1476 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
1477 (TARGET_HARD_REGNO_NREGS): Redefine.
1478 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
1479 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
1480 (HARD_REGNO_NREGS): Delete.
1481 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
1482 (cr16_hard_regno_nregs): New function.
1483 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
1484 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
1485 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
1486 (cris_hard_regno_nregs): New function.
1487 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
1488 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
1489 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
1490 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
1491 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
1492 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
1493 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
1494 (frv_hard_regno_nregs): Make static. Take and return an
1496 (frv_class_max_nregs): Remove outdated copy of documentation.
1497 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
1498 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
1499 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
1500 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
1501 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
1502 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
1503 (TARGET_HARD_REGNO_NREGS): Redefine.
1504 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
1505 (CLASS_MAX_NREGS): Update comment.
1506 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
1507 (ia64_hard_regno_nregs): New function.
1508 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
1509 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
1510 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
1511 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
1512 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
1514 (m32c_hard_regno_nregs): Likewise. Make static.
1515 (TARGET_HARD_REGNO_NREGS): Redefine.
1516 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
1517 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
1518 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
1519 (m68k_hard_regno_nregs): New function.
1520 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
1521 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
1522 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
1523 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
1524 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
1525 Take and return an unsigned int.
1526 (TARGET_HARD_REGNO_NREGS): Redefine.
1527 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
1528 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
1529 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
1530 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
1531 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
1532 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
1533 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
1534 (msp430_hard_regno_nregs): Make static. Take and return an
1536 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
1537 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
1538 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
1539 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
1540 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
1541 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
1542 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
1543 (TARGET_HARD_REGNO_NREGS): Redefine.
1544 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
1545 (PA_HARD_REGNO_NREGS): ...this.
1546 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
1547 (PA_HARD_REGNO_NREGS): ...this.
1548 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
1549 (pa_hard_regno_nregs): New function.
1550 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
1551 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
1552 (pdp11_hard_regno_nregs): New function.
1553 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
1554 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
1555 (rs6000_hard_regno_nregs_hook): New function.
1556 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
1557 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1558 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
1559 Take and return an unsigned int. Move earlier in file.
1560 (TARGET_HARD_REGNO_NREGS): Redefine.
1561 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
1562 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
1563 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
1564 (rl78_hard_regno_nregs): Make static. Take and return an
1566 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
1567 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
1568 (rs6000_hard_regno_nregs_hook): New function.
1569 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
1570 * config/rx/rx.c (rx_hard_regno_nregs): New function.
1571 (TARGET_HARD_REGNO_NREGS): Redefine.
1572 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
1573 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
1574 instead of HARD_REGNO_NREGS.
1575 (s390_hard_regno_nregs): New function.
1576 (s390_hard_regno_mode_ok): Add comment from s390.h.
1577 (TARGET_HARD_REGNO_NREGS): Redefine.
1578 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
1579 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
1580 (sh_hard_regno_nregs): New function.
1581 (sh_pass_in_reg_p): Use it.
1582 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
1583 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
1584 (sparc_hard_regno_nregs): New function.
1585 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
1586 * config/spu/spu.c (spu_hard_regno_nregs): New function.
1587 (spu_function_arg_advance): Use it, supplying a valid register number.
1588 (TARGET_HARD_REGNO_NREGS): Redefine.
1589 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
1590 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
1591 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
1592 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
1593 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
1594 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
1595 (CLASS_MAX_NREGS): Remove copy of old documentation.
1596 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
1597 (visium_hard_regno_nregs): New function.
1598 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
1599 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
1600 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
1601 xtensa_hard_regno_nregs): New function.
1602 * system.h (HARD_REGNO_NREGS): Poison.
1604 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1606 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
1607 hard_regno_nregs instead of HARD_REGNO_NREGS.
1608 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
1609 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
1610 (c6x_expand_epilogue): Likewise.
1611 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
1612 (frv_read_iacc_argument): Likewise.
1613 * config/sh/sh.c: Include regs.h.
1614 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
1615 (regs_used): Likewise.
1616 (output_stack_adjust): Likewise.
1617 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
1618 * expmed.c: Include regs.h.
1619 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
1620 * ree.c: Include regs.h.
1621 (combine_reaching_defs): Use hard_regno_nregs instead of
1623 (add_removable_extension): Likewise.
1625 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1627 * regs.h (hard_regno_nregs): Turn into a function.
1628 (end_hard_regno): Update accordingly.
1629 * caller-save.c (setup_save_areas): Likewise.
1630 (save_call_clobbered_regs): Likewise.
1631 (replace_reg_with_saved_mem): Likewise.
1632 (insert_restore): Likewise.
1633 (insert_save): Likewise.
1634 * combine.c (can_change_dest_mode): Likewise.
1635 (move_deaths): Likewise.
1636 (distribute_notes): Likewise.
1637 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
1638 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
1639 (rs6000_split_multireg_move): Likewise.
1640 (rs6000_register_move_cost): Likewise.
1641 (rs6000_memory_move_cost): Likewise.
1642 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
1643 (rs6000_split_multireg_move): Likewise.
1644 (rs6000_register_move_cost): Likewise.
1645 (rs6000_memory_move_cost): Likewise.
1646 * cselib.c (cselib_reset_table): Likewise.
1647 (cselib_lookup_1): Likewise.
1648 * emit-rtl.c (set_mode_and_regno): Likewise.
1649 * function.c (aggregate_value_p): Likewise.
1650 * ira-color.c (setup_profitable_hard_regs): Likewise.
1651 (check_hard_reg_p): Likewise.
1652 (calculate_saved_nregs): Likewise.
1653 (assign_hard_reg): Likewise.
1654 (improve_allocation): Likewise.
1655 (calculate_spill_cost): Likewise.
1656 * ira-emit.c (modify_move_list): Likewise.
1657 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
1658 (ira_hard_reg_in_set_p): Likewise.
1659 * ira.c (setup_reg_mode_hard_regset): Likewise.
1660 (clarify_prohibited_class_mode_regs): Likewise.
1661 (check_allocation): Likewise.
1662 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1663 (lra_setup_reg_renumber): Likewise.
1664 (setup_try_hard_regno_pseudos): Likewise.
1665 (spill_for): Likewise.
1666 (assign_hard_regno): Likewise.
1667 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1668 * lra-constraints.c (in_class_p): Likewise.
1669 (lra_constraint_offset): Likewise.
1670 (simplify_operand_subreg): Likewise.
1671 (lra_constraints): Likewise.
1672 (split_reg): Likewise.
1673 (split_if_necessary): Likewise.
1674 (invariant_p): Likewise.
1675 (inherit_in_ebb): Likewise.
1676 * lra-lives.c (process_bb_lives): Likewise.
1677 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
1678 (get_hard_regs): Likewise.
1679 (do_remat): Likewise.
1680 * lra-spills.c (assign_spill_hard_regs): Likewise.
1681 * mode-switching.c (create_pre_exit): Likewise.
1682 * postreload.c (reload_combine_recognize_pattern): Likewise.
1683 * recog.c (peep2_find_free_register): Likewise.
1684 * regcprop.c (kill_value_regno): Likewise.
1685 (set_value_regno): Likewise.
1686 (copy_value): Likewise.
1687 (maybe_mode_change): Likewise.
1688 (find_oldest_value_reg): Likewise.
1689 (copyprop_hardreg_forward_1): Likewise.
1690 * regrename.c (check_new_reg_p): Likewise.
1691 (regrename_do_replace): Likewise.
1692 * reload.c (push_reload): Likewise.
1693 (combine_reloads): Likewise.
1694 (find_dummy_reload): Likewise.
1695 (operands_match_p): Likewise.
1696 (find_reloads): Likewise.
1697 (find_equiv_reg): Likewise.
1698 (reload_adjust_reg_for_mode): Likewise.
1699 * reload1.c (count_pseudo): Likewise.
1700 (count_spilled_pseudo): Likewise.
1701 (find_reg): Likewise.
1702 (clear_reload_reg_in_use): Likewise.
1703 (free_for_value_p): Likewise.
1704 (allocate_reload_reg): Likewise.
1705 (choose_reload_regs): Likewise.
1706 (reload_adjust_reg_for_temp): Likewise.
1707 (emit_reload_insns): Likewise.
1708 (delete_output_reload): Likewise.
1709 * rtlanal.c (subreg_get_info): Likewise.
1710 * sched-deps.c (sched_analyze_reg): Likewise.
1711 * sel-sched.c (init_regs_for_mode): Likewise.
1712 (mark_unavailable_hard_regs): Likewise.
1713 (choose_best_reg_1): Likewise.
1714 (verify_target_availability): Likewise.
1715 * valtrack.c (dead_debug_insert_temp): Likewise.
1716 * var-tracking.c (track_loc_p): Likewise.
1717 (emit_note_insn_var_location): Likewise.
1718 * varasm.c (make_decl_rtl): Likewise.
1719 * reginfo.c (choose_hard_reg_mode): Likewise.
1720 (init_reg_modes_target): Refer directly to
1721 this_target_regs->x_hard_regno_nregs.
1723 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1725 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
1726 instead of hard_regno_nregs.
1728 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1730 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
1731 end_hard_regno instead of hard_regno_nregs.
1732 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
1733 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
1734 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
1735 * ira-color.c (improve_allocation): Likewise.
1736 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1737 * lra-lives.c (mark_regno_live): Likewise.
1738 (mark_regno_dead): Likewise.
1739 * lra-remat.c (operand_to_remat): Likewise.
1740 * lra.c (collect_non_operand_hard_regs): Likewise.
1741 * postreload.c (reload_combine_note_store): Likewise.
1742 (move2add_valid_value_p): Likewise.
1743 * reload.c (regno_clobbered_p): Likewise.
1745 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1747 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
1749 * config/v850/v850.c (v850_reorg): Likewise.
1750 * reload.c (refers_to_regno_for_reload_p): Likewise.
1751 (find_equiv_reg): Likewise.
1752 * reload1.c (reload_reg_reaches_end_p): Likewise.
1754 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1756 * caller-save.c (add_used_regs): Use REG_NREGS instead of
1758 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
1759 * config/arm/arm.c (output_move_neon): Likewise.
1760 (arm_attr_length_move_neon): Likewise.
1761 (neon_split_vcombine): Likewise.
1762 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
1763 (c6x_mark_reg_written): Likewise.
1764 (c6x_dwarf_register_span): Likewise.
1765 * config/i386/i386.c (ix86_save_reg): Likewise.
1766 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
1767 (rws_access_reg): Likewise.
1768 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
1769 * mode-switching.c (create_pre_exit): Likewise.
1770 * ree.c (combine_reaching_defs): Likewise.
1771 (add_removable_extension): Likewise.
1772 * regcprop.c (find_oldest_value_reg): Likewise.
1773 (copyprop_hardreg_forward_1): Likewise.
1774 * reload.c (reload_inner_reg_of_subreg): Likewise.
1775 (push_reload): Likewise.
1776 (combine_reloads): Likewise.
1777 (find_dummy_reload): Likewise.
1778 (reload_adjust_reg_for_mode): Likewise.
1779 * reload1.c (find_reload_regs): Likewise.
1780 (forget_old_reloads_1): Likewise.
1781 (reload_reg_free_for_value_p): Likewise.
1782 (reload_adjust_reg_for_temp): Likewise.
1783 (emit_reload_insns): Likewise.
1784 (delete_output_reload): Likewise.
1785 * sel-sched.c (choose_best_reg_1): Likewise.
1786 (choose_best_pseudo_reg): Likewise.
1788 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1789 Alan Hayward <alan.hayward@arm.com>
1790 David Sherwood <david.sherwood@arm.com>
1792 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
1793 * target.def (slow_unaligned_access): New hook.
1794 * targhooks.h (default_slow_unaligned_access): Declare.
1795 * targhooks.c (default_slow_unaligned_access): New function.
1796 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
1797 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
1798 * doc/tm.texi: Regenerate.
1799 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
1800 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
1801 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
1803 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
1804 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
1806 (rs6000_slow_unaligned_access): New function.
1807 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
1808 (expand_block_compare): Likewise.
1809 (expand_strn_compare): Likewise.
1810 (rs6000_rtx_costs): Likewise.
1811 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
1812 (riscv_slow_unaligned_access): Likewise.
1813 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
1814 (riscv_slow_unaligned_access_p): ...this and make static.
1815 (riscv_option_override): Update accordingly.
1816 (riscv_slow_unaligned_access): New function.
1817 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
1818 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
1819 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
1820 (rs6000_slow_unaligned_access): New function.
1821 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
1822 (rs6000_rtx_costs): Likewise.
1823 * config/rs6000/rs6000-string.c (expand_block_compare)
1824 (expand_strn_compare): Use targetm.slow_unaligned_access instead
1825 of SLOW_UNALIGNED_ACCESS.
1826 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
1827 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
1828 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
1829 of SLOW_UNALIGNED_ACCESS.
1830 * expmed.c (simple_mem_bitfield_p): Likewise.
1831 * expr.c (alignment_for_piecewise_move): Likewise.
1832 (emit_group_load_1): Likewise.
1833 (emit_group_store): Likewise.
1834 (copy_blkmode_from_reg): Likewise.
1835 (emit_push_insn): Likewise.
1836 (expand_assignment): Likewise.
1837 (store_field): Likewise.
1838 (expand_expr_real_1): Likewise.
1839 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1840 * lra-constraints.c (simplify_operand_subreg): Likewise.
1841 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
1842 * gimple-ssa-store-merging.c: Likewise in block comment at start
1844 * tree-ssa-strlen.c: Include target.h.
1845 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
1846 of SLOW_UNALIGNED_ACCESS.
1847 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
1849 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1851 PR rtl-optimization/82185
1852 * expmed.c (emit_store_flag_int): Only test tem if it has been
1855 2017-09-12 Richard Biener <rguenther@suse.de>
1858 * match.pd ((FTYPE) N CMP CST): Fix typo.
1860 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
1862 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
1864 (mips_near_type_p): Add 'short_call' attribute as a synonym
1866 * doc/extend.texi (short_call): Document new function attribute.
1868 2017-09-12 Jakub Jelinek <jakub@redhat.com>
1871 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
1872 assertion check that in the condition.
1873 (get_atomic_generic_size): Likewise. Before testing if parameter
1874 has pointer type, if it has array type, call for C++
1875 default_conversion to perform array-to-pointer conversion.
1877 2017-09-12 Richard Biener <rguenther@suse.de>
1879 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
1880 for operations we cannot scalarize.
1882 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
1884 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
1885 vectors heap vectors. Clean up comments.
1886 Make visited_bbs a reference.
1887 (profitable_jump_thread_path): Make GC
1888 vectors heap vectors. Clean up comments.
1890 (convert_and_register_jump_thread_path): Make GC vectors heap
1892 (check_subpath_and_update_thread_path): Same. Clean up comments.
1893 Make visited_bbs a reference.
1894 (handle_phi): Abstract common code to to
1895 register_jump_thread_path_if_profitable.
1896 Rename VAR_BB to DEF_BB.
1898 Make GC vectors heap vectors.
1899 Make visited_bbs a reference.
1900 (handle_assignment): Same.
1901 (register_jump_thread_path_if_profitable): New.
1902 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
1904 Make GC vectors heap vectors. Clean up comments.
1905 Make visited_bbs a reference.
1906 (find_jump_threads_backwards): Make visited_bbs live in the stack.
1907 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
1910 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
1913 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
1914 words of E_DImode object are reachable by xtensa_uimm8x4 access.
1916 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
1918 Revert r251800 and r251799.
1920 2017-09-11 Martin Jambor <mjambor@suse.cz>
1923 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
1924 arguments in advance.
1925 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
1926 use it to find predecessor edges.
1927 (naive_outof_ssa): Collect vector of predecessors.
1929 2017-09-08 Jason Merrill <jason@redhat.com>
1931 PR c++/70029 - ICE with ref-qualifier and -flto
1932 * langhooks.h (struct lang_hooks_for_types): Add
1933 copy_lang_qualifiers.
1934 * attribs.c (build_type_attribute_qual_variant): Use it.
1935 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
1937 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
1938 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
1940 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
1943 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
1944 (*mulsi3_sp64): New instruction.
1945 (mulsi3): New expander.
1947 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
1949 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
1951 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1953 * sancov.c: Include memmodel.h.
1955 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
1958 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
1961 2017-09-07 Carl Love <cel@us.ibm.com>
1963 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
1964 the sldi instruction.
1966 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
1968 * sancov.c: Include tm_p.h.
1970 2017-09-07 Jakub Jelinek <jakub@redhat.com>
1973 * output.h (switch_to_other_text_partition): New declaration.
1974 * varasm.c (switch_to_other_text_partition): New function.
1975 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
1976 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
1977 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
1978 to the other text partition before emitting LCL label and switch back
1979 after emitting the word after it.
1981 2017-09-07 Richard Biener <rguenther@suse.de>
1983 * passes.def (pass_split_crit_edges): Remove instance before PRE.
1984 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
1985 critical edges here, after loop init.
1986 (pass_data_pre): Remove PROP_no_crit_edges flags.
1987 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
1988 for valueization of call args to avoid leaking VN_TOP.
1989 (visit_use): Assert we do not visit default defs.
1990 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
1991 Use error_mark_node to more easily detect leaking VN_TOP.
1992 All default-defs are varying, not VN_TOP. Mark them visited.
1993 (run_scc_vn): Make code match comment.
1995 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
1997 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
1998 OPTION_MASK_FLOAT128_KEYWORD.
1999 (POWERPC_MASKS): Likewise.
2000 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
2001 support for the -mfloat128-type option, and make -mfloat128
2002 default on PowerPC Linux systems. Define or undefine
2003 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
2004 Define __float128 to be __ieee128 if IEEE 128-bit support is
2005 enabled, or undefine it.
2006 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
2007 Delete defining __FLOAT128_TYPE__.
2008 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
2009 -mfloat128-type option and make -mfloat128 default on PowerPC
2011 (TARGET_FLOAT128_TYPE): Likewise.
2012 (-mfloat128-type): Likewise.
2013 * config/rs6000/rs6000.c (rs6000_option_override_internal):
2014 Delete the -mfloat128-type option and make -mfloat128 default on
2015 PowerPC Linux systems. Always use __ieee128 to be the keyword for
2016 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
2017 128-bit floating point is enabled. Change tests from using
2018 -mfloat128-type to -mfloat128.
2019 (rs6000_mangle_type): Use the correct mangling for the __float128
2020 type even if normal long double is restricted to 64-bits.
2021 (floatn_mode): Enable the _Float128 type by default on VSX Linux
2023 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
2024 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
2025 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
2027 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
2028 documentation for -mfloat128.
2030 2017-09-06 Olivier Hainque <hainque@adacore.com>
2032 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
2034 2017-09-06 Wish Wu <wishwu007@gmail.com>
2035 Jakub Jelinek <jakub@redhat.com>
2037 * asan.c (initialize_sanitizer_builtins): Add
2038 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
2039 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
2040 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
2041 BT_FN_VOID_UINT64_PTR variables.
2042 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
2043 (BT_FN_VOID_UINT16_UINT16): Likewise.
2044 (BT_FN_VOID_UINT32_UINT32): Likewise.
2045 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
2046 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
2047 (BT_FN_VOID_UINT64_PTR): Likewise.
2048 * common.opt (flag_sanitize_coverage): New variable.
2049 (fsanitize-coverage=trace-pc): Remove.
2050 (fsanitize-coverage=): Add.
2051 * flag-types.h (enum sanitize_coverage_code): New enum.
2052 * fold-const.c (fold_range_test): Disable non-short-circuit
2053 optimization if flag_sanitize_coverage.
2054 (fold_truth_andor): Likewise.
2055 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
2056 * opts.c (COVERAGE_SANITIZER_OPT): Define.
2057 (coverage_sanitizer_opts): New array.
2058 (get_closest_sanitizer_option): Add OPTS argument, handle also
2059 OPT_fsanitize_coverage_.
2060 (parse_sanitizer_options): Adjusted to also handle
2061 OPT_fsanitize_coverage_.
2062 (common_handle_option): Add OPT_fsanitize_coverage_.
2063 * sancov.c (instrument_comparison, instrument_switch): New function.
2064 (sancov_pass): Add trace-cmp support.
2065 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
2066 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
2067 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
2068 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
2069 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
2070 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
2071 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
2072 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
2073 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
2074 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
2076 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
2078 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
2079 error. Only quit immediately if parsing is complete.
2080 (BEGIN): Initialize fatal_err and parse_done.
2081 (begin fpu, end fpu): Check number of arguments.
2082 (begin arch, end arch): Likewise.
2083 (begin cpu, end cpu): Likewise.
2084 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
2085 (optalias): Likewise.
2087 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
2089 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
2090 * config/arm/arm-isa.h: Delete. Move definitions to ...
2091 * arm-cpus.in: ... here. Use new feature and fgroup values.
2092 * config/arm/arm.c (arm_option_override): Use lower case for feature
2094 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
2095 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
2096 * config/arm/parsecpu.awk (END): Add new command 'isa'.
2098 (print_isa_bits_for): New function.
2099 (gen_isa): New function.
2100 (gen_comm_data): Use print_isa_bits_for.
2101 (define feature): New keyword.
2102 (define fgroup): New keyword.
2103 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
2104 (arm-isa.h): Add rule to generate file.
2105 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
2106 case for feature bit names.
2108 2017-09-06 Richard Biener <rguenther@suse.de>
2110 * tree-ssa-pre.c (NECESSARY): Remove.
2111 (create_expression_by_pieces): Do not touch pass-local flags.
2112 (insert_into_preds_of_block): Likewise.
2113 (do_pre_regular_insertion): Likewise.
2114 (eliminate_insert): Likewise.
2115 (eliminate_dom_walker::before_dom_children): Likewise.
2116 (fini_eliminate): Do not look at inserted_exprs.
2117 (mark_operand_necessary): Remove.
2118 (remove_dead_inserted_code): Replace with simple work-list
2119 algorithm based on inserted_exprs and SSA uses.
2120 (pass_pre::execute): Re-order fini_eliminate and
2121 remove_dead_inserted_code.
2123 2017-09-06 Olivier Hainque <hainque@adacore.com>
2125 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
2126 for VxWorks 7. Adjust surrounding comments.
2128 2017-09-06 Richard Biener <rguenther@suse.de>
2130 * gimple-ssa-strength-reduction.c
2131 (find_candidates_dom_walker::before_dom_children): Also allow
2134 2017-09-06 Richard Biener <rguenther@suse.de>
2136 PR tree-optimization/82108
2137 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
2138 for gap in the non-permutation SLP case.
2140 2017-09-06 Martin Jambor <mjambor@suse.cz>
2142 PR tree-optimization/82078
2143 * tree-sra.c (sort_and_splice_var_accesses): Move call to
2144 add_access_to_work_queue...
2145 (build_accesses_from_assign): ...here.
2146 (propagate_all_subaccesses): Make sure racc is the group
2147 representative, if there is one.
2149 2017-09-06 Jakub Jelinek <jakub@redhat.com>
2152 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
2155 2017-09-06 Richard Biener <rguenther@suse.de>
2157 * gimple-ssa-strength-reduction.c
2158 (find_candidates_dom_walker::before_doom_children): Use a
2159 type and not a mode check.
2161 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
2164 * config/arm/predicates.md (arm_general_adddi_operand): Create new
2166 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
2168 2017-09-05 Jeff Law <law@redhat.com>
2170 PR tree-optimization/64910
2171 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
2172 cases where we have 3 or more operands.
2174 2017-09-05 Jakub Jelinek <jakub@redhat.com>
2177 * omp-low.c (lower_omp_for): Recompute tree invariant if
2178 gimple_omp_for_initial/final is ADDR_EXPR.
2181 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
2182 into gimple val before gimplification fo the COND_EXPR.
2184 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
2186 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
2187 REGION_COPY argument.
2188 (thread_through_all_blocks): Remove unused argument to
2189 duplicate_thread_path.
2191 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2192 Alan Hayward <alan.hayward@arm.com>
2193 David Sherwood <david.sherwood@arm.com>
2195 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
2196 Take a scalar_mode rather than a machine_mode.
2197 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2198 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
2199 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2200 (aarch64_gen_adjusted_ldpstp): Likewise.
2201 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
2203 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2204 Alan Hayward <alan.hayward@arm.com>
2205 David Sherwood <david.sherwood@arm.com>
2207 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
2208 Take a scalar_int_mode instead of a machine_mode.
2209 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
2210 (aarch64_output_scalar_simd_mov_immediate): Likewise.
2211 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
2212 (aarch64_simd_attr_length_rglist): Delete.
2213 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
2214 a scalar_int_mode instead of a machine_mode.
2215 (aarch64_add_offset): Likewise.
2216 (aarch64_internal_mov_immediate): Likewise
2217 (aarch64_add_constant_internal): Likewise.
2218 (aarch64_add_constant): Likewise.
2219 (aarch64_movw_imm): Likewise.
2220 (aarch64_rtx_arith_op_extract_p): Likewise.
2221 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
2222 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
2223 Remove assert that the mode isn't a vector.
2224 (aarch64_output_scalar_simd_mov_immediate): Likewise.
2225 (aarch64_expand_mov_immediate): Update calls after above changes.
2226 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
2227 (aarch64_and_bitmask_imm): Check for scalar integer modes.
2228 (aarch64_move_imm): Likewise.
2229 (aarch64_can_const_movi_rtx_p): Likewise.
2230 (aarch64_strip_extend): Likewise.
2231 (aarch64_extr_rtx_p): Likewise.
2232 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
2233 a CONST_INT when the mode parameter is VOIDmode.
2234 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
2236 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2238 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
2239 * stor-layout.c (bitwise_mode_for_mode): Likewise.
2240 (bitwise_type_for_mode): Update accordingly.
2242 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2244 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
2245 * stor-layout.c (mode_for_size_tree): Likewise.
2246 (mode_for_array): Update accordingly.
2247 (layout_decl): Likewise.
2248 (compute_record_mode): Likewise. Only set the mode once.
2250 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2252 * target.def (get_mask_mode): Change return type to opt_mode.
2254 * doc/tm.texi: Regenerate.
2255 * targhooks.h (default_get_mask_mode): Return an opt_mode.
2256 * targhooks.c (default_get_mask_mode): Likewise.
2257 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
2258 * optabs-query.c (can_vec_mask_load_store_p): Update use of
2259 targetm.get_mask_mode.
2260 * tree.c (build_truth_vector_type): Likewise.
2262 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2264 * machmode.h (mode_for_vector): Return an opt_mode.
2265 * stor-layout.c (mode_for_vector): Likewise.
2266 (mode_for_int_vector): Update accordingly.
2267 (layout_type): Likewise.
2268 * config/i386/i386.c (emit_memmov): Likewise.
2269 (ix86_expand_set_or_movmem): Likewise.
2270 (ix86_expand_vector_init): Likewise.
2271 (ix86_get_mask_mode): Likewise.
2272 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
2274 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
2275 * expmed.c (extract_bit_field_1): Likewise.
2276 * expr.c (expand_expr_real_2): Likewise.
2277 * optabs-query.c (can_vec_perm_p): Likewise.
2278 (can_vec_mask_load_store_p): Likewise.
2279 * optabs.c (expand_vec_perm): Likewise.
2280 * targhooks.c (default_get_mask_mode): Likewise.
2281 * tree-vect-stmts.c (vectorizable_store): Likewise.
2282 (vectorizable_load): Likewise.
2283 (get_vectype_for_scalar_type_and_size): Likewise.
2285 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2287 * machmode.h (mode_for_int_vector): New function.
2288 * stor-layout.c (mode_for_int_vector): Likewise.
2289 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
2290 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
2291 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
2292 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
2293 (s390_expand_vcond): Likewise.
2295 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2297 * machmode.h (opt_machine_mode): New type.
2298 (opt_mode<T>): Allow construction from anything that can be
2300 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
2301 (mode_for_size): Return an opt_machine_mode.
2302 * stor-layout.c (mode_for_size): Likewise.
2303 (mode_for_size_tree): Update call accordingly.
2304 (bitwise_mode_for_mode): Likewise.
2305 (make_fract_type): Likewise.
2306 (make_accum_type): Likewise.
2307 * caller-save.c (replace_reg_with_saved_mem): Update call
2309 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2310 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2311 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2312 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2313 * expmed.c (extract_bit_field_1): Likewise.
2314 * reload.c (get_secondary_mem): Likewise.
2315 * varasm.c (assemble_integer): Likewise.
2316 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
2319 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2321 * machmode.h (decimal_float_mode_for_size): New function.
2322 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
2323 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
2324 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
2325 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
2326 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
2327 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
2329 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2331 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
2332 (get_builtin_sync_mode): Likewise.
2333 (expand_ifn_atomic_compare_exchange): Likewise.
2334 (expand_builtin_atomic_clear): Likewise.
2335 (expand_builtin_atomic_test_and_set): Likewise.
2336 (fold_builtin_atomic_always_lock_free): Likewise.
2337 * calls.c (compute_argument_addresses): Likewise.
2338 (emit_library_call_value_1): Likewise.
2339 (store_one_arg): Likewise.
2340 * combine.c (combine_instructions): Likewise.
2341 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
2342 * config/arm/arm.c (arm_function_value): Likewise.
2343 (aapcs_allocate_return_reg): Likewise.
2344 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
2345 * config/i386/i386.c (construct_container): Likewise.
2346 (ix86_gimplify_va_arg): Likewise.
2347 (ix86_expand_sse_cmp): Likewise.
2348 (emit_memmov): Likewise.
2349 (emit_memset): Likewise.
2350 (expand_small_movmem_or_setmem): Likewise.
2351 (ix86_expand_pextr): Likewise.
2352 (ix86_expand_pinsr): Likewise.
2353 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
2354 * config/microblaze/microblaze.c (microblaze_block_move_straight):
2356 * config/mips/mips.c (mips_function_value_1) Likewise.
2357 (mips_block_move_straight): Likewise.
2358 (mips_expand_ins_as_unaligned_store): Likewise.
2359 * config/powerpcspe/powerpcspe.c
2360 (rs6000_darwin64_record_arg_advance_flush): Likewise.
2361 (rs6000_darwin64_record_arg_flush): Likewise.
2362 * config/rs6000/rs6000.c
2363 (rs6000_darwin64_record_arg_advance_flush): Likewise.
2364 (rs6000_darwin64_record_arg_flush): Likewise.
2365 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
2366 (sparc_function_value_1): Likewise.
2367 * config/spu/spu.c (adjust_operand): Likewise.
2368 (spu_emit_branch_or_set): Likewise.
2369 (arith_immediate_p): Likewise.
2370 * emit-rtl.c (gen_lowpart_common): Likewise.
2371 * expr.c (expand_expr_real_1): Likewise.
2372 * function.c (assign_parm_setup_block): Likewise.
2373 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
2374 * reload1.c (alter_reg): Likewise.
2375 * stor-layout.c (mode_for_vector): Likewise.
2376 (layout_type): Likewise.
2378 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2380 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
2381 (spu_convert_move): Likewise.
2382 * lower-subreg.c (resolve_simple_move): Likewise.
2384 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2387 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
2388 define_insn to a define_expand.
2389 (altivec_vsum2sws_direct): New define_insn.
2390 (altivec_vsumsws): Convert from a define_insn to a define_expand.
2392 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
2394 * config/arm/arm.c (arm_option_params_internal): Improve setting of
2397 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
2402 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
2403 realign stack if stack alignment needed is less than incoming
2406 2017-09-05 Marek Polacek <polacek@redhat.com>
2409 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
2412 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
2414 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
2416 2017-09-05 Richard Biener <rguenther@suse.de>
2418 PR tree-optimization/82084
2419 * fold-const.c (can_native_encode_string_p): Handle wide characters.
2421 2017-09-05 Richard Biener <rguenther@suse.de>
2423 PR tree-optimization/82102
2424 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
2426 2017-09-05 Martin Liska <mliska@suse.cz>
2428 PR tree-optimization/82032
2429 * tree-cfg.c (generate_range_test): New function.
2430 * tree-cfg.h (generate_range_test): Declared here.
2431 * tree-cfgcleanup.c (convert_single_case_switch): New function.
2432 (cleanup_control_expr_graph): Use it.
2433 * tree-switch-conversion.c (try_switch_expansion): Remove
2435 (emit_case_nodes): Use generate_range_test.
2437 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
2440 * config/i386/i386.md (*<btsc><mode>_mask): Add
2441 TARGET_USE_BT to insn constraint.
2442 (*btr<mode>_mask): Ditto.
2444 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
2446 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
2447 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
2449 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2452 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
2453 TARGET_NEON and TARGET_IWMMXT.
2454 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
2455 TARGET_NEON and TARGET_IWMMXT.
2456 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
2458 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
2460 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
2461 (ix86_rewrite_tls_address): Ditto.
2462 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
2463 (ix86_rewrite_tls_address_1): Ditto.
2464 (ix86_rewrite_tls_address): Ditto.
2465 * config/i386/predicates.md (tls_address_pattern): New predicate.
2466 * config/i386/i386.md (TLS address splitter): New splitter.
2468 2017-09-04 Richard Biener <rguenther@suse.de>
2470 PR tree-optimization/82084
2471 * fold-const.h (can_native_encode_string_p): Declare.
2472 * fold-const.c (can_native_encode_string_p): Factor out from ...
2473 (native_encode_string): ... here.
2474 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
2475 vectorizing stores from constants we later cannot handle.
2477 2017-09-04 Marek Polacek <polacek@redhat.com>
2480 * doc/invoke.texi: Update -Wtautological-compare documentation.
2482 2017-09-04 Jeff Law <law@redhat.com>
2484 PR tree-optimization/64910
2485 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
2486 swap the first and last operand if the last is a constant.
2488 2017-09-04 Marek Polacek <polacek@redhat.com>
2491 * convert.c (do_narrow): When sanitizing signed integer overflows,
2492 bail out for signed types.
2493 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
2495 2017-09-04 Richard Biener <rguenther@suse.de>
2497 PR tree-optimization/82060
2498 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
2499 Move devirtualization after stmt folding and before EH/AB/noreturn
2500 cleanup to get the stmt refs canonicalized. Use a bool instead
2501 of gimple_modified_p since that doesn't work for NOPs. Schedule
2502 NOPs generated by folding for removal.
2504 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2505 Alan Hayward <alan.hayward@arm.com>
2506 David Sherwood <david.sherwood@arm.com>
2508 * coretypes.h (pad_direction): New enum.
2509 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
2510 (FUNCTION_ARG_PADDING): Likewise.
2511 * target.def (function_arg_padding): New hook.
2512 * targhooks.h (default_function_arg_padding): Declare.
2513 * targhooks.c (default_function_arg_padding): New function.
2514 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
2515 (TARGET_FUNCTION_ARG_PADDING): ...this.
2516 * doc/tm.texi: Regenerate.
2517 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
2518 instead of direction.
2519 (compute_argument_addresses): Likewise.
2520 (load_register_parameters): Likewise.
2521 (emit_library_call_value_1): Likewise.
2522 (store_one_arg): Use targetm.calls.function_arg_padding instead
2523 of FUNCTION_ARG_PADDING.
2524 (must_pass_in_stack_var_size_or_pad): Likewise.
2525 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
2526 (emit_group_store): Likewise.
2527 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
2528 instead of FUNCTION_ARG_PADDING.
2529 (emit_push_insn): Likewise, and propagate enum change throughout
2531 * function.h (direction): Delete.
2532 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
2534 * function.c (assign_parm_find_stack_rtl): Likewise.
2535 (assign_parm_setup_block_p): Likewise.
2536 (assign_parm_setup_block): Likewise.
2537 (gimplify_parameters): Likewise.
2538 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
2539 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
2541 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
2542 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2543 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
2544 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
2545 (aarch64_function_arg_padding): ...this new function.
2546 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
2547 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2548 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
2549 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2550 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
2551 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2552 (arm_pad_arg_upward): Replace with...
2553 (arm_function_arg_padding): ...this new function.
2554 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
2556 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
2557 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
2558 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2559 (ia64_hpux_function_arg_padding): Replace with...
2560 (ia64_function_arg_padding): ...this new function. Use pad_direction
2561 instead of direction. Check for TARGET_HPUX.
2562 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
2563 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2564 (iq2000_function_arg_padding): New function.
2565 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
2566 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
2567 (mips_function_arg_padding): ...this new function.
2568 (mips_pad_reg_upward): Update accordingly.
2569 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2570 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
2571 targetm.calls.function_arg_padding.
2572 (FUNCTION_ARG_PADDING): Delete.
2573 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2574 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
2575 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2576 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
2577 (nios2_block_reg_padding): Return pad_direction instead of direction.
2578 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
2579 instead of direction.
2580 (nios2_function_arg_padding): Likewise. Make static.
2581 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2582 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
2583 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
2584 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
2585 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2586 (pa_function_arg_padding): Make static. Return pad_direction instead
2588 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
2589 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2590 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
2591 instead of direction. Use targetm.calls.function_arg_padding.
2592 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
2593 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
2594 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
2595 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
2596 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
2598 (function_arg_padding): Rename to...
2599 (rs6000_function_arg_padding): ...this. Make static. Return
2600 pad_direction instead of direction.
2601 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
2602 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
2603 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2604 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
2605 instead of direction. Use targetm.calls.function_arg_padding.
2606 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
2607 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
2608 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
2609 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
2610 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2611 (function_arg_padding): Rename to...
2612 (rs6000_function_arg_padding): ...this. Make static. Return
2613 pad_direction instead of direction.
2614 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
2615 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
2616 * config/s390/s390.c (s390_function_arg_padding): New function.
2617 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2618 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
2619 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
2620 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2621 (function_arg_padding): Rename to...
2622 (sparc_function_arg_padding): ...this. Make static. Return
2623 pad_direction instead of direction.
2624 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
2625 * config/spu/spu.c (spu_function_arg_padding): New function.
2626 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2627 * system.h (FUNCTION_ARG_PADDING): Poison.
2629 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2630 Alan Hayward <alan.hayward@arm.com>
2631 David Sherwood <david.sherwood@arm.com>
2633 * target.def (modes_tieable_p): New hook.
2634 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
2635 (TARGET_MODES_TIEABLE_P): ...this.
2636 * doc/tm.texi.in: Regenerate.
2637 * hooks.h (hook_bool_mode_mode_true): Declare.
2638 * hooks.c (hook_bool_mode_mode_true): New function.
2639 * combine.c (subst): Use targetm.modes_tieable_p instead of
2641 * dse.c (find_shift_sequence): Likewise.
2642 * expmed.c (extract_low_bits): Likewise.
2643 * lower-subreg.c: Include target.h.
2644 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
2646 * rtlanal.c (rtx_cost): Likewise.
2647 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
2648 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
2649 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
2650 (TARGET_MODES_TIEABLE_P): Redefine.
2651 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
2652 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
2653 (TARGET_MODES_TIEABLE_P): Redefine.
2654 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
2655 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
2656 (arc_modes_tieable_p): New function.
2657 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
2658 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
2659 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
2660 (arm_modes_tieable_p): Make static.
2661 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
2662 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
2663 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
2664 (TARGET_MODES_TIEABLE_P): Redefine.
2665 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
2666 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
2667 (TARGET_MODES_TIEABLE_P): Redefine.
2668 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
2669 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
2670 (cr16_modes_tieable_p): New function.
2671 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
2672 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
2673 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
2674 (TRULY_NOOP_TRUNCATION): Update comment.
2675 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
2676 (TRULY_NOOP_TRUNCATION): Update comment.
2677 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
2678 (frv_modes_tieable_p): New function.
2679 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
2680 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
2681 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
2682 (TARGET_MODES_TIEABLE_P): Redefine.
2683 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
2684 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
2685 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
2686 (TARGET_MODES_TIEABLE_P): Redefine.
2687 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
2688 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
2689 (ia64_modes_tieable_p): New function.
2690 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
2691 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
2692 (iq2000_modes_tieable_p): New function.
2693 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
2694 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
2695 (lm32_modes_tieable_p): New function.
2696 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
2697 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
2698 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
2699 (TARGET_MODES_TIEABLE_P): Redefine.
2700 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
2701 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
2702 (m32r_modes_tieable_p): New function.
2703 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
2704 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
2705 (m68k_modes_tieable_p): New function.
2706 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
2707 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
2708 (mcore_modes_tieable_p): New function.
2709 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
2710 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
2712 (TARGET_MODES_TIEABLE_P): Redefine.
2713 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
2714 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
2715 * config/mips/mips.c (mips_modes_tieable_p): Make static.
2716 (TARGET_MODES_TIEABLE_P): Redefine.
2717 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
2718 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
2719 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
2720 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
2721 (mn10300_modes_tieable_p): ...this and make static.
2722 (TARGET_MODES_TIEABLE_P): Redefine.
2723 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
2724 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
2725 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
2726 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
2727 (msp430_modes_tieable_p): Make static.
2728 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
2729 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
2730 (TARGET_MODES_TIEABLE_P): Redefine.
2731 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
2732 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
2733 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
2734 (TARGET_MODES_TIEABLE_P): Redefine.
2735 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
2736 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
2737 * config/pa/pa.c (pa_modes_tieable_p): Make static.
2738 (TARGET_MODES_TIEABLE_P): Redefine.
2739 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
2740 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
2741 (pdp11_modes_tieable_p): New function.
2742 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
2743 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
2744 (rs6000_modes_tieable_p): New function.
2745 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
2746 * config/powerpcspe/powerpcspe.md: Update comment.
2747 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
2748 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
2749 (TARGET_MODES_TIEABLE_P): Redefine.
2750 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
2751 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
2752 (rl78_modes_tieable_p): New function.
2753 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
2754 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
2755 (rs6000_modes_tieable_p): New function.
2756 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
2757 * config/rs6000/rs6000.md: Update comment.
2758 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
2759 * config/rx/rx.c (rx_modes_tieable_p): New function.
2760 (TARGET_MODES_TIEABLE_P): Redefine.
2761 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
2762 * config/s390/s390.c (s390_modes_tieable_p): New function.
2763 (TARGET_MODES_TIEABLE_P): Redefine.
2764 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
2765 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
2766 (sh_modes_tieable_p): New function.
2767 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
2768 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
2769 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
2770 (sparc_modes_tieable_p): Make static.
2771 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
2772 * config/spu/spu.c (spu_modes_tieable_p): New function.
2773 (TARGET_MODES_TIEABLE_P): Redefine.
2774 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
2775 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
2776 (TARGET_MODES_TIEABLE_P): Redefine.
2777 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
2778 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
2779 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
2780 * config/v850/v850.c (v850_modes_tieable_p): New function.
2781 (TARGET_MODES_TIEABLE_P): Redefine.
2782 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
2783 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
2784 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
2785 (visium_modes_tieable_p): New function.
2786 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
2787 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
2788 (xtensa_modes_tieable_p): New function.
2789 * system.h (MODES_TIEABLE_P): Poison.
2791 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2792 Alan Hayward <alan.hayward@arm.com>
2793 David Sherwood <david.sherwood@arm.com>
2795 * target.def (hard_regno_mode_ok): New hook.
2796 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
2797 (TARGET_HARD_REGNO_MODE_OK): ...this.
2798 * doc/tm.texi.in: Regenerate.
2799 * hooks.h (hook_bool_uint_mode_true): Declare.
2800 * hooks.c (hook_bool_uint_mode_true): New function.
2801 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
2803 * genpreds.c (write_insn_preds_c): Add an include of target.h.
2804 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
2805 instead of HARD_REGNO_MODE_OK.
2806 * caller-save.c: Include target.h.
2807 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
2809 * combine.c (can_combine_p): Likewise.
2810 (combinable_i3pat): Likewise.
2811 (can_change_dest_mode): Likewise.
2812 * expr.c (init_expr_target): Likewise.
2813 (convert_move): Likewise.
2814 (convert_modes): Likewise.
2815 * ira.c (setup_prohibited_class_mode_regs): Likewise.
2816 (setup_prohibited_mode_move_regs): Likewise.
2817 * ira.h (target_ira): Likewise.
2818 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2819 * lra-constraints.c (process_alt_operands): Likewise.
2820 (split_reg): Likewise.
2821 * recog.c (peep2_find_free_register): Likewise.
2822 * ree.c (combine_reaching_defs): Likewise.
2823 * regcprop.c (maybe_mode_change): Likewise.
2824 * reginfo.c (init_reg_sets_1): Likewise.
2825 (choose_hard_reg_mode): Likewise.
2826 (simplifiable_subregs): Likewise.
2827 * regrename.c (check_new_reg_p): Likewise.
2828 * reload.c (find_valid_class): Likewise.
2829 (find_valid_class_1): Likewise.
2830 (reload_inner_reg_of_subreg): Likewise.
2831 (push_reload): Likewise.
2832 (combine_reloads): Likewise.
2833 (find_dummy_reload): Likewise.
2834 (find_reloads): Likewise.
2835 * reload1.c (find_reg): Likewise.
2836 (set_reload_reg): Likewise.
2837 (allocate_reload_reg): Likewise.
2838 (choose_reload_regs): Likewise.
2839 (reload_adjust_reg_for_temp): Likewise.
2840 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
2841 (simplify_subreg_regno): Likewise.
2842 * sel-sched.c (init_regs_for_mode): Likewise.
2843 * varasm.c (make_decl_rtl): Likewise.
2844 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
2845 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
2847 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
2848 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
2849 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2850 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
2851 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
2852 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2853 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
2854 (arc_mode_class): Delete.
2855 (HARD_REGNO_MODE_OK): Delete.
2856 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2857 (arc_hard_regno_mode_ok): Rename old array to...
2858 (arc_hard_regno_mode_ok_modes): ...this.
2859 (arc_conditional_register_usage): Update accordingly.
2860 (arc_mode_class): Make static.
2861 (arc_hard_regno_mode_ok): New function.
2862 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
2863 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
2864 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2865 (arm_hard_regno_mode_ok): Make static.
2866 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
2868 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
2869 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
2870 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
2872 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2873 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
2874 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
2875 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
2876 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
2877 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2878 * config/bfin/predicates.md (valid_reg_operand): Use
2879 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
2880 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
2881 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
2882 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2883 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
2884 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
2885 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2886 (cr16_hard_regno_mode_ok): Make static and return a bool.
2887 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
2888 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2889 (cris_hard_regno_mode_ok): New function.
2890 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
2891 (epiphany_mode_class): Delete.
2892 (HARD_REGNO_MODE_OK): Delete.
2893 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
2894 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2895 (hard_regno_mode_ok): Rename to...
2896 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
2897 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
2898 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
2900 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
2901 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
2902 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2903 (frv_hard_regno_mode_ok): Make static and return a bool.
2904 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
2906 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
2907 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
2908 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
2909 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
2911 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2912 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
2913 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
2914 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
2916 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2917 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
2918 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2919 (ia64_hard_regno_mode_ok): New function.
2920 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
2921 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2922 (iq2000_hard_regno_mode_ok): New function.
2923 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
2924 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2925 (lm32_hard_regno_mode_ok): New function.
2926 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
2927 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
2928 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
2929 instead of HARD_REGNO_MODE_OK.
2930 (m32c_hard_regno_ok): Rename to...
2931 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
2932 (m32c_cannot_change_mode_class): Update accordingly.
2933 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2934 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
2935 (m32r_mode_class): Delete.
2936 (HARD_REGNO_MODE_OK): Delete.
2937 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2938 (m32r_hard_regno_mode_ok): Rename to...
2939 (m32r_hard_regno_modes): ...this.
2940 (m32r_mode_class): Make static.
2941 (m32r_hard_regno_mode_ok): New function.
2942 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
2943 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
2944 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2945 (m68k_hard_regno_mode_ok): Make static.
2946 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
2947 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2948 (mcore_hard_regno_mode_ok): New function.
2949 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
2950 (HARD_REGNO_MODE_OK): Delete.
2951 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
2953 (microblaze_hard_regno_mode_ok_p): ...this and make static.
2954 (microblaze_hard_regno_mode_ok): New function.
2955 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2956 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
2957 (mips_hard_regno_mode_ok): Delete.
2958 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
2959 (mips_hard_regno_mode_ok_p): ...this and make static.
2960 (mips_hard_regno_mode_ok_p): Rename to...
2961 (mips_hard_regno_mode_ok_uncached): ...this.
2962 (mips_hard_regno_mode_ok): New function.
2963 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
2964 of HARD_REGNO_MODE_OK.
2965 (mips_option_override): Update after above name changes.
2966 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2967 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
2968 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
2969 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
2970 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
2971 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2972 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
2973 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
2974 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
2975 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2976 (msp430_hard_regno_mode_ok): Make static and return a bool.
2977 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
2978 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
2979 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
2981 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2982 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
2983 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
2984 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
2985 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
2986 (PA_HARD_REGNO_MODE_OK): ...this
2987 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
2988 (PA_HARD_REGNO_MODE_OK): ...this.
2989 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2990 (pa_hard_regno_mode_ok): New function.
2991 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
2992 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2993 (pdp11_hard_regno_mode_ok): New function.
2994 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
2995 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
2997 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
2999 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3000 (rs6000_hard_regno_mode_ok): Rename to...
3001 (rs6000_hard_regno_mode_ok_uncached): ...this.
3002 (rs6000_init_hard_regno_mode_ok): Update accordingly.
3003 (rs6000_hard_regno_mode_ok): New function.
3004 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
3005 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
3006 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
3007 (riscv_hard_regno_mode_ok): ...this and make static.
3008 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3009 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
3010 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
3011 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3012 (rl78_hard_regno_mode_ok): Make static and return bool.
3013 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
3014 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
3016 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
3017 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3018 (rs6000_hard_regno_mode_ok): Rename to...
3019 (rs6000_hard_regno_mode_ok_uncached): ...this.
3020 (rs6000_init_hard_regno_mode_ok): Update accordingly.
3021 (rs6000_hard_regno_mode_ok): New function.
3022 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
3023 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
3024 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3025 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
3026 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
3027 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
3028 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3029 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
3030 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
3031 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3032 (sh_hard_regno_mode_ok): Make static.
3033 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
3034 instead of HARD_REGNO_MODE_OK.
3035 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
3036 (sparc_mode_class): Delete.
3037 (HARD_REGNO_MODE_OK): Delete.
3038 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3039 (hard_regno_mode_classes): Make static.
3040 (sparc_mode_class): Likewise.
3041 (sparc_hard_regno_mode_ok): New function.
3042 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
3043 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
3044 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
3046 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3047 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
3048 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
3049 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
3050 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
3051 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3052 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
3053 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
3054 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3055 (visium_hard_regno_mode_ok): New function.
3056 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
3057 instead of HARD_REGNO_MODE_OK.
3058 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
3059 (HARD_REGNO_MODE_OK): Delete.
3060 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
3061 (xtensa_hard_regno_mode_ok_p): ...this and make static.
3062 (xtensa_option_override): Update accordingly.
3063 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3064 (xtensa_hard_regno_mode_ok): New function.
3065 * system.h (HARD_REGNO_MODE_OK): Poison.
3067 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3068 Alan Hayward <alan.hayward@arm.com>
3069 David Sherwood <david.sherwood@arm.com>
3071 * target.def (hard_regno_call_part_clobbered): New hook.
3072 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
3073 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
3074 * doc/tm.texi: Regenerate.
3075 * hooks.h (hook_bool_uint_mode_false): Declare.
3076 * hooks.c (hook_bool_uint_mode_false): New function.
3077 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3078 * cselib.c (cselib_process_insn): Use
3079 targetm.hard_regno_call_part_clobbered instead of
3080 HARD_REGNO_CALL_PART_CLOBBERED.
3081 * ira-conflicts.c (ira_build_conflicts): Likewise.
3082 * ira-costs.c (ira_tune_allocno_costs): Likewise.
3083 * lra-constraints.c (need_for_call_save_p): Likewise.
3084 * lra-lives.c: Include target.h.
3085 (check_pseudos_live_through_calls): Use
3086 targetm.hard_regno_call_part_clobbered instead of
3087 HARD_REGNO_CALL_PART_CLOBBERED.
3088 * regcprop.c: Include target.h.
3089 (copyprop_hardreg_forward_1): Use
3090 targetm.hard_regno_call_part_clobbered instead of
3091 HARD_REGNO_CALL_PART_CLOBBERED.
3092 * reginfo.c (choose_hard_reg_mode): Likewise.
3093 * regrename.c (check_new_reg_p): Likewise.
3094 * reload.c (find_equiv_reg): Likewise.
3095 * reload1.c (emit_reload_insns): Likewise.
3096 * sched-deps.c (deps_analyze_insn): Likewise.
3097 * sel-sched.c (init_regs_for_mode): Likewise.
3098 (mark_unavailable_hard_regs): Likewise.
3099 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
3100 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3101 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
3103 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3104 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3105 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
3107 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
3109 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3110 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3111 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
3113 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3114 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3115 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
3117 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3118 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
3120 * config/powerpcspe/powerpcspe.c
3121 (rs6000_hard_regno_call_part_clobbered): New function.
3122 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3123 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3124 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
3126 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3127 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3128 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
3130 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3131 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3132 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
3134 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3135 Alan Hayward <alan.hayward@arm.com>
3136 David Sherwood <david.sherwood@arm.com>
3138 * rtl.h (subreg_memory_offset): Declare.
3139 * emit-rtl.c (subreg_memory_offset): New function.
3140 * expmed.c (store_bit_field_1): Use it.
3141 * expr.c (undefined_operand_subword_p): Likewise.
3142 * simplify-rtx.c (simplify_subreg): Likewise.
3144 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
3146 PR rtl-optimization/57448
3149 * optabs.c (expand_atomic_load): Place compiler memory barriers if
3150 using atomic_load pattern.
3151 (expand_atomic_store): Likewise.
3153 2017-09-04 Jakub Jelinek <jakub@redhat.com>
3156 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
3157 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
3158 handling. Use replace_call_with_value with NULL instead of
3159 gsi_replace, unlink_stmt_vdef and release_defs.
3161 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
3164 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
3166 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3169 * rtl.h (emit_library_call_value_1): Declare.
3170 (emit_library_call): Replace declaration with a series of overloads.
3171 Remove the parameter count argument.
3172 (emit_library_call_value): Likewise.
3173 * calls.c (emit_library_call_value_1): Make global. Replace varargs
3174 with an "rtx_mode_t *".
3175 (emit_library_call_value): Delete.
3176 (emit_library_call): Likewise.
3177 * asan.c (asan_emit_stack_protection): Update calls accordingly.
3178 (asan_emit_allocas_unpoison): Likewise.
3179 * builtins.c (expand_builtin_powi): Likewise.
3180 (expand_asan_emit_allocas_unpoison): Likewise.
3181 * cfgexpand.c (expand_main_function): Likewise.
3182 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
3183 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
3184 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
3185 * config/arm/arm.c (arm_trampoline_init): Likewise.
3186 (arm_call_tls_get_addr): Likewise.
3187 (arm_expand_divmod_libfunc): Likewise.
3188 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
3189 (smulsi3_highpart): Likewise.
3190 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
3191 (c6x_expand_compare): Likewise.
3192 (c6x_expand_movmem): Likewise.
3193 * config/frv/frv.c (frv_trampoline_init): Likewise.
3194 * config/i386/i386.c (ix86_trampoline_init): Likewise.
3195 (ix86_expand_divmod_libfunc): Likewise.
3196 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
3197 (ia64_expand_compare): Likewise.
3198 (ia64_profile_hook): Likewise.
3199 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
3200 (nonlocal_goto): Likewise.
3201 (restore_stack_nonlocal): Likewise.
3202 * config/m32r/m32r.c (block_move_call): Likewise.
3203 (m32r_trampoline_init): Likewise.
3204 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
3205 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
3206 (m68k_call_m68k_read_tp): Likewise.
3207 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
3208 (microblaze_expand_divide): Likewise.
3209 * config/mips/mips.h (mips_args): Likewise.
3210 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
3211 (MIPS_ICACHE_SYNC): Likewise.
3212 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
3213 (nios2_trampoline_init): Likewise.
3214 * config/pa/pa.c (hppa_tls_call): Likewise.
3215 (pa_trampoline_init): Likewise.
3216 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
3217 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
3218 (expand_strn_compare): Likewise.
3219 (rs6000_generate_compare): Likewise.
3220 (rs6000_expand_float128_convert): Likewise.
3221 (output_profile_hook): Likewise.
3222 (rs6000_trampoline_init): Likewise.
3223 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
3224 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
3225 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
3226 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
3227 (rs6000_generate_compare): Likewise.
3228 (rs6000_expand_float128_convert): Likewise.
3229 (output_profile_hook): Likewise.
3230 (rs6000_trampoline_init): Likewise.
3231 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
3232 * config/sh/sh.c (sh_trampoline_init): Likewise.
3233 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
3234 (sparc_emit_float_lib_cmp): Likewise.
3235 (sparc32_initialize_trampoline): Likewise.
3236 (sparc64_initialize_trampoline): Likewise.
3237 (sparc_profile_hook): Likewise.
3238 * config/spu/spu.c (ea_load_store): Likewise.
3239 * config/spu/spu.md (floatunssidf2): Likewise.
3240 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
3241 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
3242 * config/visium/visium.c (expand_block_move_4): Likewise.
3243 (expand_block_move_2): Likewise.
3244 (expand_block_move_1): Likewise.
3245 (expand_block_set_4): Likewise.
3246 (expand_block_set_2): Likewise.
3247 (expand_block_set_1): Likewise.
3248 (visium_trampoline_init): Likewise.
3249 (visium_profile_hook): Likewise.
3250 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
3251 (xtensa_setup_frame_addresses): Likewise.
3252 (xtensa_trampoline_init): Likewise.
3253 * except.c (sjlj_emit_function_enter): Likewise.
3254 (sjlj_emit_function_exit): Likewise.
3255 * explow.c (allocate_dynamic_stack_space): Likewise.
3256 (probe_stack_range): Likewise.
3257 * expr.c (convert_mode_scalar): Likewise.
3258 * optabs.c (expand_binop): Likewise.
3259 (expand_twoval_binop_libfunc): Likewise.
3260 (expand_unop): Likewise.
3261 (prepare_cmp_insn): Likewise.
3262 (prepare_float_lib_cmp): Likewise.
3263 (expand_float): Likewise.
3264 (expand_fix): Likewise.
3265 (expand_fixed_convert): Likewise.
3266 (maybe_emit_sync_lock_test_and_set): Likewise.
3267 (expand_atomic_compare_and_swap): Likewise.
3268 (expand_mem_thread_fence): Likewise.
3269 (expand_atomic_fetch_op): Likewise.
3271 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
3273 * doc/generic.texi (OpenACC): Adjust URL.
3274 * doc/invoke.texi (C Dialect Options): Ditto.
3276 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
3278 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
3279 predicate for operand 1. Add (m,<S>) constraint.
3280 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
3281 Prevent memory operand 1 with register operand 2.
3283 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
3285 PR rtl-optimization/82024
3286 * combine.c (try_combine): If the combination result is a PARALLEL,
3287 and we only need to retain the SET in there that would be placed
3288 at I2, check that we can place that at I3 instead, before doing so.
3290 2017-09-01 Jakub Jelinek <jakub@redhat.com>
3293 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
3295 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
3296 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
3299 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
3300 Jeff Law <law@redhat.com>
3302 * varasm.c (bss_initializer_p): Do not put constants into .bss
3303 (categorize_decl_for_section): Handle bss_initializer_p returning
3304 false when DECL_INITIAL is NULL.
3306 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3309 * config/s390/s390.c (s390_can_inline_p): New function.
3311 2017-09-01 Jeff Law <law@redhat.com>
3313 PR tree-optimization/82052
3314 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
3315 Always initialize the returned slot after a hash table miss
3316 when INSERT is true.
3318 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
3320 * config/s390/s390.md (mem_signal_fence): Remove.
3321 * doc/md.texi (mem_signal_fence): Remove.
3322 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
3324 * target-insns.def (mem_signal_fence): Remove.
3326 2017-09-01 Jakub Jelinek <jakub@redhat.com>
3329 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
3332 * asan.c (create_odr_indicator): Strip name encoding from assembler
3333 name before appending it after __odr_asan_.
3335 2017-09-01 Martin Liska <mliska@suse.cz>
3337 PR tree-optimization/82059
3338 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
3339 frequency only when an edge is redirected.
3341 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3343 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
3344 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
3345 (arc_conditional_register_usage): Remove ARC600 lp_count
3347 (arc_file_start): Emit Tag_ARC_CPU_variation.
3348 (arc_can_use_doloop_p): New conditions to use ZOLs.
3349 (hwloop_fail): New function.
3350 (hwloop_optimize): Likewise.
3351 (hwloop_pattern_reg): Likewise.
3352 (arc_doloop_hooks): New struct, to be used with reorg_loops.
3353 (arc_reorg_loops): New function, calls reorg_loops.
3354 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
3355 (arc600_corereg_hazard): Remove ZOL checking, case handled by
3357 (arc_loop_hazard): Remove function, functionality moved into
3359 (arc_hazard): Remove arc_loop_hazard call.
3360 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
3361 into hwloop_optimize.
3362 (arc_label_align): Remove ZOL handling.
3363 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
3364 * config/arc/arc.md (doloop_begin): Remove pattern.
3365 (doloop_begin_i): Likewise.
3366 (doloop_end_i): Likewise.
3367 (doloop_fallback): Likewise.
3368 (doloop_fallback_m): Likewise.
3369 (doloop_end): Reimplement expand.
3370 (arc_lp): New pattern for LP instruction.
3371 (loop_end): New pattern.
3372 (loop_fail): Likewise.
3373 (decrement_and_branch_until_zero): Likewise.
3374 * config/arc/arc.opt (mlpc-width): New option.
3375 * doc/invoke.texi (mlpc-width): Document option.
3377 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3379 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
3380 (arc_ccfsm_advance): Fix checking for delay slots.
3381 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
3383 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3385 * config/arc/arc.md (movqi_insn): Add stores to save constant long
3387 (movhi_insn): Update store instruction constraint which are saving
3388 6-bit short immediates.
3389 (movsi_insn): Consider also short scaled load operations.
3390 (zero_extendhisi2_i): Use Usd constraint instead of T.
3391 (extendhisi2_i): Add q constraint.
3392 (arc_clzsi2): Add type and length attributes.
3393 (arc_ctzsi2): Likewise.
3394 * config/arc/constraints.md (Usc): Update constraint, the
3395 assembler can parse two relocations for a single instruction.
3397 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3399 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
3400 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
3402 2017-08-31 Olivier Hainque <hainque@adacore.com>
3404 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
3405 match as powerpc-wrs-vxworks*.
3407 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
3409 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
3410 register constraint for by-element operand.
3411 (aarch64_mls_elt_merge<mode>): Likewise.
3413 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
3415 * config/arc/arc.c (arc_can_follow_jump): Check for short
3418 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
3420 * config.gcc: Use g.opt for arc.
3421 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
3422 functionality moved to ...
3423 (legitimate_scaled_address_p): New function, ...here.
3424 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
3425 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
3426 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
3428 (arc_override_options): Handle G option.
3429 (arc_output_pic_addr_const): Correct function definition.
3430 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
3431 (arc_decl_anon_ns_mem_p): Delete.
3432 (arc_in_small_data_p): Overhaul this function to take into
3433 consideration the value given via G option.
3434 (arc_rewrite_small_data_1): Renamed and corrected old
3435 arc_rewrite_small_data function.
3436 (arc_rewrite_small_data): New function.
3437 (small_data_pattern): Don't use pic_offset_table_rtx.
3438 * config/arc/arc.h (CC1_SPEC): Recognize G option.
3439 * config/arc/simdext.md (movmisalignv2hi): Use
3440 prepare_move_operands function.
3442 (movmisalign*): Likewise.
3443 * doc/invoke.texi (ARC options): Document -G option.
3445 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
3447 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
3449 * config/arc/arc.c (arc_print_operand): Output scalled address for
3450 sdata whenever is possible.
3451 (arc_in_small_data_p): Allow sdata for 64bit datum when double
3452 load/stores are available.
3453 (compact_sda_memory_operand): Check for the alignment required by
3454 code density instructions.
3455 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
3457 * config/arc/constraints.md (Usd): Update constraint.
3458 (Us0): New constraint.
3459 (Usc): Update constraint.
3461 2017-08-31 Richard Biener <rguenther@suse.de>
3464 * dwarf2out.c (dwarf2out_early_global_decl): Process each
3467 2017-08-31 Tamar Christina <tamar.christina@arm.com>
3469 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
3470 Resize type_signature.
3472 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3473 Alan Hayward <alan.hayward@arm.com>
3474 David Sherwood <david.sherwood@arm.com>
3476 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
3477 subregs whose inner modes can be stored in GPRs.
3478 (aarch64_classify_index): Likewise.
3480 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3481 Alan Hayward <alan.hayward@arm.com>
3482 David Sherwood <david.sherwood@arm.com>
3484 * config/aarch64/iterators.md (V_cmp_result): Rename to...
3485 (V_INT_EQUIV): ...this.
3486 (v_cmp_result): Rename to...
3487 (v_int_equiv): ...this.
3488 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
3489 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
3490 (copysign<mode>3): Likewise.
3491 (aarch64_simd_bsl<mode>_internal): Likewise.
3492 (aarch64_simd_bsl<mode>): Likewise.
3493 (vec_cmp<mode><mode>): Likewise.
3494 (vcond<mode><mode>): Likewise.
3495 (vcond<v_cmp_mixed><mode>): Likewise.
3496 (vcondu<mode><v_cmp_mixed>): Likewise.
3497 (aarch64_cm<optab><mode>): Likewise.
3498 (aarch64_cmtst<mode>): Likewise.
3499 (aarch64_fac<optab><mode>): Likewise.
3500 (vec_perm_const<mode>): Likewise.
3501 (vcond_mask_<mode><v_cmp_result>): Rename to...
3502 (vcond_mask_<mode><v_int_equiv>): ...this.
3503 (vec_cmp<mode><v_cmp_result>): Rename to...
3504 (vec_cmp<mode><v_int_equiv>): ...this.
3506 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3507 Alan Hayward <alan.hayward@arm.com>
3508 David Sherwood <david.sherwood@arm.com>
3510 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
3512 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
3513 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
3514 (UNSPEC_LD4_DREG): New unspecs.
3515 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
3516 (aarch64_ld2<mode>_dreg_be): Replace with...
3517 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
3519 (aarch64_ld3<mode>_dreg_le)
3520 (aarch64_ld3<mode>_dreg_be): Replace with...
3521 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
3523 (aarch64_ld4<mode>_dreg_le)
3524 (aarch64_ld4<mode>_dreg_be): Replace with...
3525 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
3528 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3530 PR tree-optimization/81987
3531 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
3532 insert an initializer in a location not dominated by the stride
3535 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
3537 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
3538 on the entire header of the finally block in the fallthru case.
3540 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
3542 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
3544 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
3546 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
3547 rs6000_emit_move_from_cr and call renamed function.
3548 (rs6000_emit_prologue): Call renamed functions.
3549 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
3550 movesi_from_cr, remove volatile CRs.
3552 2017-08-30 Jon Beniston <jon@beniston.com>
3553 Richard Biener <rguenther@suse.de>
3555 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
3556 of VECTOR_MODE_P check.
3557 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
3558 element vector types.
3560 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3562 * df.h (df_read_modify_subreg_p): Remove in favor of...
3563 * rtl.h (read_modify_subreg_p): ...this new function. Take a
3564 const_rtx instead of an rtx.
3565 * cprop.c (local_cprop_find_used_regs): Update accordingly.
3566 * df-problems.c (df_word_lr_mark_ref): Likewise.
3567 * ira-lives.c (mark_pseudo_reg_live): Likewise.
3568 (mark_pseudo_reg_dead): Likewise.
3569 (mark_ref_dead): Likewise.
3570 * reginfo.c (init_subregs_of_mode): Likewise.
3571 * sched-deps.c (sched_analyze_1): Likewise.
3572 * df-scan.c (df_def_record_1): Likewise.
3573 (df_uses_record): Likewise.
3574 (df_read_modify_subreg_p): Remove in favor of...
3575 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
3576 const_rtx instead of an rtx.
3578 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3579 Alan Hayward <alan.hayward@arm.com>
3580 David Sherwood <david.sherwood@arm.com>
3582 * rtl.h (partial_subreg_p): New function.
3583 * caller-save.c (save_call_clobbered_regs): Use it.
3584 * calls.c (expand_call): Likewise.
3585 * combine.c (combinable_i3pat): Likewise.
3586 (simplify_set): Likewise.
3587 (make_extraction): Likewise.
3588 (make_compound_operation_int): Likewise.
3589 (gen_lowpart_or_truncate): Likewise.
3590 (force_to_mode): Likewise.
3591 (make_field_assignment): Likewise.
3592 (reg_truncated_to_mode): Likewise.
3593 (record_truncated_value): Likewise.
3594 (move_deaths): Likewise.
3595 * cse.c (record_jump_cond): Likewise.
3596 (cse_insn): Likewise.
3597 * cselib.c (cselib_lookup_1): Likewise.
3598 * expmed.c (extract_bit_field_using_extv): Likewise.
3599 * function.c (assign_parm_setup_reg): Likewise.
3600 * ifcvt.c (noce_convert_multiple_sets): Likewise.
3601 * ira-build.c (create_insn_allocnos): Likewise.
3602 * lra-coalesce.c (merge_pseudos): Likewise.
3603 * lra-constraints.c (match_reload): Likewise.
3604 (simplify_operand_subreg): Likewise.
3605 (curr_insn_transform): Likewise.
3606 * lra-lives.c (process_bb_lives): Likewise.
3607 * lra.c (new_insn_reg): Likewise.
3608 (lra_substitute_pseudo): Likewise.
3609 * regcprop.c (mode_change_ok): Likewise.
3610 (maybe_mode_change): Likewise.
3611 (copyprop_hardreg_forward_1): Likewise.
3612 * reload.c (push_reload): Likewise.
3613 (find_reloads): Likewise.
3614 (find_reloads_subreg_address): Likewise.
3615 * reload1.c (alter_reg): Likewise.
3616 (eliminate_regs_1): Likewise.
3617 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3619 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
3621 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
3622 back to if statements, including unpack.
3624 2017-08-30 Martin Liska <mliska@suse.cz>
3627 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
3629 (func_checker::compare_asm_inputs_outputs): ... this function.
3630 (func_checker::compare_gimple_asm): Use the function to compare
3631 also ASM constrains.
3632 * ipa-icf-gimple.h: Rename the function.
3634 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3635 Alan Hayward <alan.hayward@arm.com>
3636 David Sherwood <david.sherwood@arm.com>
3638 * coretypes.h (complex_mode): New type.
3639 * gdbhooks.py (build_pretty_printer): Handle it.
3640 * machmode.h (complex_mode): New class.
3641 (complex_mode::includes_p): New function.
3642 (is_complex_int_mode): Likewise.
3643 (is_complex_float_mode): Likewise.
3644 * genmodes.c (get_mode_class): Handle complex mode classes.
3645 * function.c (expand_function_end): Use is_complex_int_mode.
3647 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3648 Alan Hayward <alan.hayward@arm.com>
3649 David Sherwood <david.sherwood@arm.com>
3651 * coretypes.h (scalar_mode_pod): New typedef.
3652 * gdbhooks.py (build_pretty_printer): Handle it.
3653 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
3654 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
3655 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
3656 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
3659 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3660 Alan Hayward <alan.hayward@arm.com>
3661 David Sherwood <david.sherwood@arm.com>
3663 * machmode.h (mode_for_vector): Take a scalar_mode instead
3665 * stor-layout.c (mode_for_vector): Likewise.
3666 * explow.c (promote_mode): Use as_a <scalar_mode>.
3667 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
3669 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3670 Alan Hayward <alan.hayward@arm.com>
3671 David Sherwood <david.sherwood@arm.com>
3673 * target.def (preferred_simd_mode): Take a scalar_mode
3674 instead of a machine_mode.
3675 * targhooks.h (default_preferred_simd_mode): Likewise.
3676 * targhooks.c (default_preferred_simd_mode): Likewise.
3677 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
3678 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
3679 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
3680 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
3681 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
3682 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
3683 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
3684 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
3686 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
3687 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
3688 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
3689 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
3690 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
3691 * doc/tm.texi: Regenerate.
3692 * optabs-query.c (can_vec_mask_load_store_p): Return false for
3695 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3696 Alan Hayward <alan.hayward@arm.com>
3697 David Sherwood <david.sherwood@arm.com>
3699 * target.def (scalar_mode_supported_p): Take a scalar_mode
3700 instead of a machine_mode.
3701 * targhooks.h (default_scalar_mode_supported_p): Likewise.
3702 * targhooks.c (default_scalar_mode_supported_p): Likewise.
3703 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
3704 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
3705 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
3706 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
3707 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
3708 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
3709 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
3710 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
3711 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
3712 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
3713 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
3714 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
3716 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
3717 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
3718 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
3719 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
3720 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
3722 * doc/tm.texi: Regenerate.
3724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3725 Alan Hayward <alan.hayward@arm.com>
3726 David Sherwood <david.sherwood@arm.com>
3728 * coretypes.h (opt_scalar_mode): New typedef.
3729 * gdbhooks.py (build_pretty_printers): Handle it.
3730 * machmode.h (mode_iterator::get_2xwider): Add overload for
3732 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
3734 * expr.c (convert_mode_scalar): Likewise.
3735 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3736 * optabs.c (expand_float): Likewise.
3737 (expand_fix): Likewise.
3738 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3740 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3741 Alan Hayward <alan.hayward@arm.com>
3742 David Sherwood <david.sherwood@arm.com>
3744 * optabs.c (expand_float): Explicitly check for scalars before
3745 using a branching expansion.
3746 (expand_fix): Likewise.
3748 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3749 Alan Hayward <alan.hayward@arm.com>
3750 David Sherwood <david.sherwood@arm.com>
3752 * expr.c (convert_mode): Split scalar handling out into...
3753 (convert_mode_scalar): ...this new function. Treat the modes
3756 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3757 Alan Hayward <alan.hayward@arm.com>
3758 David Sherwood <david.sherwood@arm.com>
3760 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
3762 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
3764 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3765 Alan Hayward <alan.hayward@arm.com>
3766 David Sherwood <david.sherwood@arm.com>
3768 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
3769 rather than a machine_mode.
3770 (fixed_from_string): Likewise.
3771 (fixed_convert): Likewise.
3772 (fixed_convert_from_int): Likewise.
3773 (fixed_convert_from_real): Likewise.
3774 (real_convert_from_fixed): Likewise.
3775 * fixed-value.c (fixed_from_double_int): Likewise.
3776 (fixed_from_string): Likewise.
3777 (fixed_convert): Likewise.
3778 (fixed_convert_from_int): Likewise.
3779 (fixed_convert_from_real): Likewise.
3780 (real_convert_from_fixed): Likewise.
3781 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
3783 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3784 Alan Hayward <alan.hayward@arm.com>
3785 David Sherwood <david.sherwood@arm.com>
3787 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
3788 of separate mode class checks. Do not allow vector modes here.
3789 (immed_wide_int_const): Use as_a <scalar_mode>.
3790 * explow.c (trunc_int_for_mode): Likewise.
3791 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
3792 (wi::shwi): Likewise.
3793 (wi::min_value): Likewise.
3794 (wi::max_value): Likewise.
3795 * dwarf2out.c (loc_descriptor): Likewise.
3796 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
3799 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3800 Alan Hayward <alan.hayward@arm.com>
3801 David Sherwood <david.sherwood@arm.com>
3803 * tree.h (SCALAR_TYPE_MODE): New macro.
3804 * expr.c (expand_expr_addr_expr_1): Use it.
3805 (expand_expr_real_2): Likewise.
3806 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
3807 (fold_convert_const_fixed_from_int): Likewise.
3808 (fold_convert_const_fixed_from_real): Likewise.
3809 (native_encode_fixed): Likewise
3810 (native_encode_complex): Likewise
3811 (native_encode_vector): Likewise.
3812 (native_interpret_fixed): Likewise.
3813 (native_interpret_real): Likewise.
3814 (native_interpret_complex): Likewise.
3815 (native_interpret_vector): Likewise.
3816 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
3817 (simd_clone_adjust_argument_types): Likewise.
3818 (simd_clone_init_simd_arrays): Likewise.
3819 (simd_clone_adjust): Likewise.
3820 * stor-layout.c (layout_type): Likewise.
3821 * tree.c (build_minus_one_cst): Likewise.
3822 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
3823 * tree-inline.c (estimate_move_cost): Likewise.
3824 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
3825 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
3826 (vectorizable_reduction): Likewise.
3827 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
3828 (vect_recog_mixed_size_cond_pattern): Likewise.
3829 (check_bool_pattern): Likewise.
3830 (adjust_bool_pattern): Likewise.
3831 (search_type_for_mask_1): Likewise.
3832 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
3833 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3834 (vectorizable_load): Likewise.
3835 (vectorizable_store): Likewise.
3836 * ubsan.c (ubsan_encode_value): Likewise.
3837 * varasm.c (output_constant): Likewise.
3839 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3840 Alan Hayward <alan.hayward@arm.com>
3841 David Sherwood <david.sherwood@arm.com>
3843 * coretypes.h (scalar_mode): New class.
3844 * machmode.h (scalar_mode): Likewise.
3845 (scalar_mode::includes_p): New function.
3846 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
3847 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
3848 * genmodes.c (get_mode_class): Handle remaining scalar modes.
3849 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
3850 * expmed.c (store_bit_field_1): Likewise.
3851 (extract_bit_field_1): Likewise.
3852 * expr.c (write_complex_part): Likewise.
3853 (read_complex_part): Likewise.
3854 (emit_move_complex_push): Likewise.
3855 (expand_expr_real_2): Likewise.
3856 * function.c (assign_parm_setup_reg): Likewise.
3857 (assign_parms_unsplit_complex): Likewise.
3858 * optabs.c (expand_binop): Likewise.
3859 * rtlanal.c (subreg_get_info): Likewise.
3860 * simplify-rtx.c (simplify_immed_subreg): Likewise.
3861 * varasm.c (output_constant_pool_2): Likewise.
3863 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3864 Alan Hayward <alan.hayward@arm.com>
3865 David Sherwood <david.sherwood@arm.com>
3867 * expmed.c (extract_high_half): Use scalar_int_mode and remove
3869 (expmed_mult_highpart_optab): Likewise.
3870 (expmed_mult_highpart): Likewise.
3872 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3873 Alan Hayward <alan.hayward@arm.com>
3874 David Sherwood <david.sherwood@arm.com>
3876 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
3877 instead of a machine_mode.
3878 (builtin_memset_read_str): Likewise.
3879 * builtins.c (c_readstr): Likewise.
3880 (builtin_memcpy_read_str): Likewise.
3881 (builtin_strncpy_read_str): Likewise.
3882 (builtin_memset_read_str): Likewise.
3883 (builtin_memset_gen_str): Likewise.
3884 (expand_builtin_signbit): Use scalar_int_mode for local variables.
3885 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
3886 instead of a machine_mode.
3887 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
3889 (make_extraction): Likewise.
3890 (try_widen_shift_mode): Take and return scalar_int_modes instead
3892 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
3893 a scalar_int_mode instead of a machine_mode.
3894 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
3895 (avr_addr_space_pointer_mode): Likewise.
3896 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
3897 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
3898 (msp430_unwind_word_mode): Likewise.
3899 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
3900 (spu_addr_space_pointer_mode): Likewise.
3901 (spu_addr_space_address_mode): Likewise.
3902 (spu_libgcc_cmp_return_mode): Likewise.
3903 (spu_libgcc_shift_count_mode): Likewise.
3904 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
3905 (rl78_addr_space_pointer_mode): Likewise.
3906 (fl78_unwind_word_mode): Likewise.
3907 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
3909 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
3910 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
3911 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
3912 (mips_valid_pointer_mode): Likewise.
3913 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
3914 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
3915 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
3917 (ft32_addr_space_address_mode): Likewise.
3918 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
3919 scalar_int_mode instead of a machine_mode.
3920 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
3922 (m32c_addr_space_address_mode): Likewise.
3923 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
3924 (rs6000_eh_return_filter_mode): Likewise.
3925 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
3926 (rs6000_eh_return_filter_mode): Likewise.
3927 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
3928 (s390_libgcc_shift_count_mode): Likewise.
3929 (s390_unwind_word_mode): Likewise.
3930 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
3932 * target.def (mode_rep_extended): Likewise.
3933 (valid_pointer_mode): Likewise.
3934 (addr_space.valid_pointer_mode): Likewise.
3935 (eh_return_filter_mode): Return a scalar_int_mode rather than
3937 (libgcc_cmp_return_mode): Likewise.
3938 (libgcc_shift_count_mode): Likewise.
3939 (unwind_word_mode): Likewise.
3940 (addr_space.pointer_mode): Likewise.
3941 (addr_space.address_mode): Likewise.
3942 * doc/tm.texi: Regenerate.
3943 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
3945 (do_jump): Use scalar_int_mode for local variables.
3946 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
3947 rather than a machine_mode.
3948 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
3949 (scompare_loc_descriptor_wide): Likewise.
3950 (scompare_loc_descriptor_narrow): Likewise.
3951 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
3953 * except.c (sjlj_emit_dispatch_table): Likewise.
3954 (expand_builtin_eh_copy_values): Likewise.
3955 * explow.c (convert_memory_address_addr_space_1): Likewise.
3956 Take a scalar_int_mode rather than a machine_mode.
3957 (convert_memory_address_addr_space): Take a scalar_int_mode rather
3958 than a machine_mode.
3959 (memory_address_addr_space): Use scalar_int_mode for local variables.
3960 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
3961 rather than a machine_mode.
3962 * expmed.c (mask_rtx): Likewise.
3963 (init_expmed_one_conv): Likewise.
3964 (expand_mult_highpart_adjust): Likewise.
3965 (extract_high_half): Likewise.
3966 (expmed_mult_highpart_optab): Likewise.
3967 (expmed_mult_highpart): Likewise.
3968 (expand_smod_pow2): Likewise.
3969 (expand_sdiv_pow2): Likewise.
3970 (emit_store_flag_int): Likewise.
3971 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
3973 (extract_low_bits): Likewise.
3974 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
3976 * expr.c (pieces_addr::adjust): Likewise.
3977 (can_store_by_pieces): Likewise.
3978 (store_by_pieces): Likewise.
3979 (clear_by_pieces_1): Likewise.
3980 (expand_expr_addr_expr_1): Likewise.
3981 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
3982 (expand_expr_real_1): Likewise.
3983 (try_casesi): Likewise.
3984 * final.c (shorten_branches): Likewise.
3985 * fold-const.c (fold_convert_const_int_from_fixed): Change the
3986 type of "mode" to machine_mode.
3987 * internal-fn.c (expand_arith_overflow_result_store): Take a
3988 scalar_int_mode rather than a machine_mode.
3989 (expand_mul_overflow): Use scalar_int_mode for local variables.
3990 * loop-doloop.c (doloop_modify): Likewise.
3991 (doloop_optimize): Likewise.
3992 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
3993 than a machine_mode.
3994 (expand_doubleword_shift_condmove): Likewise.
3995 (expand_doubleword_shift): Likewise.
3996 (expand_doubleword_clz): Likewise.
3997 (expand_doubleword_popcount): Likewise.
3998 (expand_doubleword_parity): Likewise.
3999 (expand_absneg_bit): Use scalar_int_mode for local variables.
4000 (prepare_float_lib_cmp): Likewise.
4001 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
4002 rather than a machine_mode.
4003 (convert_memory_address_addr_space): Likewise.
4004 (get_mode_bounds): Likewise.
4005 (get_address_mode): Return a scalar_int_mode rather than a
4007 * rtlanal.c (get_address_mode): Likewise.
4008 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
4009 than a machine_mode.
4010 * targhooks.c (default_mode_rep_extended): Likewise.
4011 (default_valid_pointer_mode): Likewise.
4012 (default_addr_space_valid_pointer_mode): Likewise.
4013 (default_eh_return_filter_mode): Return a scalar_int_mode rather
4014 than a machine_mode.
4015 (default_libgcc_cmp_return_mode): Likewise.
4016 (default_libgcc_shift_count_mode): Likewise.
4017 (default_unwind_word_mode): Likewise.
4018 (default_addr_space_pointer_mode): Likewise.
4019 (default_addr_space_address_mode): Likewise.
4020 * targhooks.h (default_eh_return_filter_mode): Likewise.
4021 (default_libgcc_cmp_return_mode): Likewise.
4022 (default_libgcc_shift_count_mode): Likewise.
4023 (default_unwind_word_mode): Likewise.
4024 (default_addr_space_pointer_mode): Likewise.
4025 (default_addr_space_address_mode): Likewise.
4026 (default_mode_rep_extended): Take a scalar_int_mode rather than
4028 (default_valid_pointer_mode): Likewise.
4029 (default_addr_space_valid_pointer_mode): Likewise.
4030 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
4032 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
4033 rather than a machine_mode.
4034 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
4035 for local variables.
4036 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4037 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
4038 than a machine_mode.
4040 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4041 Alan Hayward <alan.hayward@arm.com>
4042 David Sherwood <david.sherwood@arm.com>
4044 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
4045 the mode argument to scalar_int_mode.
4046 (do_jump_by_parts_zero_rtx): Likewise.
4047 (do_jump_by_parts_equality_rtx): Likewise.
4048 (do_jump_by_parts_greater): Take a mode argument.
4049 (do_jump_by_parts_equality): Likewise.
4050 (do_jump_1): Update calls accordingly.
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 * is-a.h (safe_dyn_cast): New function.
4057 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
4058 (jump_table_for_label): Likewise.
4059 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
4060 instead of an rtx_insn *.
4061 (shorten_branches): Use dyn_cast instead of LABEL_P and
4062 JUMP_TABLE_DATA_P. Use jump_table_for_label and
4063 rtx_jump_table_data::get_data_mode.
4064 (final_scan_insn): Likewise.
4066 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4067 Alan Hayward <alan.hayward@arm.com>
4068 David Sherwood <david.sherwood@arm.com>
4070 * combine.c (try_combine): Use is_a <scalar_int_mode> when
4071 trying to combine a full-register integer set with a subreg
4074 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4075 Alan Hayward <alan.hayward@arm.com>
4076 David Sherwood <david.sherwood@arm.com>
4078 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
4079 that is always either address_mode or pointer_mode.
4081 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4082 Alan Hayward <alan.hayward@arm.com>
4083 David Sherwood <david.sherwood@arm.com>
4085 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
4086 when the two are known to be equal.
4088 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4089 Alan Hayward <alan.hayward@arm.com>
4090 David Sherwood <david.sherwood@arm.com>
4092 * simplify-rtx.c (simplify_const_unary_operation): Use
4093 is_a <scalar_int_mode> instead of checking for a nonzero
4094 precision. Forcibly convert op_mode to a scalar_int_mode
4095 in that case. More clearly differentiate the operand and
4096 result modes and use the former when deciding what the value
4097 of a count-bits operation should be. Use is_int_mode instead
4098 of checking for a MODE_INT. Remove redundant check for whether
4099 this mode has a zero precision.
4101 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4102 Alan Hayward <alan.hayward@arm.com>
4103 David Sherwood <david.sherwood@arm.com>
4105 * optabs.c (widen_leading): Change the type of the mode argument
4106 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
4107 (widen_bswap): Likewise.
4108 (expand_parity): Likewise.
4109 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
4110 (expand_ffs): Likewise.
4111 (epand_unop): Check for scalar integer modes before calling the
4114 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4115 Alan Hayward <alan.hayward@arm.com>
4116 David Sherwood <david.sherwood@arm.com>
4118 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
4120 (expand_expr_real_1): Update call accordingly.
4122 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4123 Alan Hayward <alan.hayward@arm.com>
4124 David Sherwood <david.sherwood@arm.com>
4126 * expmed.c (store_bit_field_using_insv): Add op0_mode and
4127 value_mode arguments. Use scalar_int_mode internally.
4128 (store_bit_field_1): Rename the new integer mode from imode
4129 to op0_mode and use it instead of GET_MODE (op0). Update calls
4130 to store_split_bit_field, store_bit_field_using_insv and
4131 store_fixed_bit_field.
4132 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
4133 Use scalar_int_mode internally. Use a bit count rather than a mode
4134 when calculating the largest bit size for get_best_mode.
4135 Update calls to store_split_bit_field and store_fixed_bit_field_1.
4136 (store_fixed_bit_field_1): Add mode and value_mode arguments.
4137 Remove assertion that OP0 has a scalar integer mode.
4138 (store_split_bit_field): Add op0_mode and value_mode arguments.
4139 Update calls to extract_fixed_bit_field.
4140 (extract_bit_field_using_extv): Add an op0_mode argument.
4141 Use scalar_int_mode internally.
4142 (extract_bit_field_1): Rename the new integer mode from imode to
4143 op0_mode and use it instead of GET_MODE (op0). Update calls to
4144 extract_split_bit_field, extract_bit_field_using_extv and
4145 extract_fixed_bit_field.
4146 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
4147 to extract_split_bit_field and extract_fixed_bit_field_1.
4148 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
4149 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
4151 (extract_split_bit_field): Add an op0_mode argument. Update call
4152 to extract_fixed_bit_field.
4154 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4155 Alan Hayward <alan.hayward@arm.com>
4156 David Sherwood <david.sherwood@arm.com>
4158 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
4159 * explow.c (hard_function_value): Likewise.
4160 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
4161 convert_to_mode call outside the loop.
4162 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
4163 for the mode iterator. Require the mode specified by max_pieces
4165 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
4167 (copy_blkmode_to_reg): Likewise.
4168 (set_storage_via_setmem): Likewise.
4169 * optabs.c (prepare_cmp_insn): Likewise.
4170 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4171 * stor-layout.c (finish_bitfield_representative): Likewise.
4173 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4174 Alan Hayward <alan.hayward@arm.com>
4175 David Sherwood <david.sherwood@arm.com>
4177 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
4178 * expr.c (convert_move): Use them.
4179 (convert_modes): Likewise.
4180 (store_expr_with_bounds): Likewise.
4182 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4183 Alan Hayward <alan.hayward@arm.com>
4184 David Sherwood <david.sherwood@arm.com>
4186 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
4187 parameter for the mode of "x". Remove the "known_x", "known_mode"
4188 and "known_ret" arguments. Change the type of the mode argument
4190 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
4191 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
4192 (reg_num_sign_bit_copies_for_combine): Likewise.
4193 * rtlanal.c (nonzero_bits1): Likewise.
4194 (num_sign_bit_copies1): Likewise.
4195 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
4196 (reg_num_sign_bit_copies_general): Likewise.
4197 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
4198 (reg_nonzero_bits_general): Likewise.
4200 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4201 Alan Hayward <alan.hayward@arm.com>
4202 David Sherwood <david.sherwood@arm.com>
4204 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
4205 than in subroutines. Return 1 for non-integer modes.
4206 (cached_num_sign_bit_copies): Change the type of the mode parameter
4208 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
4209 classes. Handle CONST_INT_P first and then check whether X also
4210 has a scalar integer mode. Check the same thing for inner registers
4211 of a SUBREG and for values that are being extended or truncated.
4213 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4214 Alan Hayward <alan.hayward@arm.com>
4215 David Sherwood <david.sherwood@arm.com>
4217 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
4218 in subroutines. Return the mode mask for non-integer modes.
4219 (cached_nonzero_bits): Change the type of the mode parameter
4221 (nonzero_bits1): Likewise. Remove early exit for other mode
4222 classes. Handle CONST_INT_P first and then check whether X
4223 also has a scalar integer mode.
4225 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4226 Alan Hayward <alan.hayward@arm.com>
4227 David Sherwood <david.sherwood@arm.com>
4229 * expr.c (widest_int_mode_for_size): Make the comment match the code.
4230 Return a scalar_int_mode and assert that the size is greater than
4232 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
4233 (op_by_pieces_d::op_by_pieces_d): Likewise.
4234 (op_by_pieces_d::run): Likewise.
4235 (can_store_by_pieces): Likewise.
4237 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4238 Alan Hayward <alan.hayward@arm.com>
4239 David Sherwood <david.sherwood@arm.com>
4241 * combine.c (extract_left_shift): Add a mode argument and update
4243 (make_compound_operation_int): Change the type of the mode parameter
4244 to scalar_int_mode and update the call to extract_left_shift.
4246 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4247 Alan Hayward <alan.hayward@arm.com>
4248 David Sherwood <david.sherwood@arm.com>
4250 * combine.c (simplify_and_const_int): Change the type of the mode
4251 parameter to scalar_int_mode.
4252 (simplify_and_const_int_1): Likewise. Update recursive call.
4254 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4255 Alan Hayward <alan.hayward@arm.com>
4256 David Sherwood <david.sherwood@arm.com>
4258 * combine.c (simplify_compare_const): Check that the mode is a
4259 scalar_int_mode (rather than VOIDmode) before testing its
4261 (simplify_comparison): Move COMPARISON_P handling out of the
4262 loop and restrict the latter part of the loop to scalar_int_modes.
4263 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
4264 and when considering SUBREG_REGs. Use is_int_mode instead of
4265 checking GET_MODE_CLASS against MODE_INT.
4267 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4268 Alan Hayward <alan.hayward@arm.com>
4269 David Sherwood <david.sherwood@arm.com>
4271 * combine.c (try_widen_shift_mode): Move check for equal modes to...
4272 (simplify_shift_const_1): ...here. Use scalar_int_mode for
4273 shift_unit_mode and for modes involved in scalar shifts.
4275 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4276 Alan Hayward <alan.hayward@arm.com>
4277 David Sherwood <david.sherwood@arm.com>
4279 * combine.c (force_int_to_mode): New function, split out from...
4280 (force_to_mode): ...here. Keep xmode up-to-date and use it
4281 instead of GET_MODE (x).
4283 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4284 Alan Hayward <alan.hayward@arm.com>
4285 David Sherwood <david.sherwood@arm.com>
4287 * optabs-query.h (extraction_insn::struct_mode): Change type to
4288 opt_scalar_int_mode and update comment.
4289 (extraction_insn::field_mode): Change type to scalar_int_mode.
4290 (extraction_insn::pos_mode): Likewise.
4291 * combine.c (make_extraction): Update accordingly.
4292 * optabs-query.c (get_traditional_extraction_insn): Likewise.
4293 (get_optab_extraction_insn): Likewise.
4294 * recog.c (simplify_while_replacing): Likewise.
4295 * expmed.c (narrow_bit_field_mem): Change the type of the mode
4296 parameter to opt_scalar_int_mode.
4298 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4299 Alan Hayward <alan.hayward@arm.com>
4300 David Sherwood <david.sherwood@arm.com>
4302 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
4303 to a scalar_int_mode instead of a machine_mode.
4304 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
4305 (get_best_mode): Return a boolean and use a pointer argument to store
4306 the selected mode. Replace the limit mode parameter with a bit limit.
4307 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
4308 for the values returned by bit_field_mode_iterator::next_mode.
4309 (store_bit_field): Update call to get_best_mode.
4310 (store_fixed_bit_field): Likewise.
4311 (extract_fixed_bit_field): Likewise.
4312 * expr.c (optimize_bitfield_assignment_op): Likewise.
4313 * fold-const.c (optimize_bit_field_compare): Likewise.
4314 (fold_truth_andor_1): Likewise.
4315 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
4316 Update for new type of m_mode.
4317 (get_best_mode): As above.
4319 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4320 Alan Hayward <alan.hayward@arm.com>
4321 David Sherwood <david.sherwood@arm.com>
4323 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
4324 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
4325 (store_bit_field): Check is_a <scalar_int_mode> before calling
4326 strict_volatile_bitfield_p.
4327 (extract_bit_field): Likewise.
4329 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4330 Alan Hayward <alan.hayward@arm.com>
4331 David Sherwood <david.sherwood@arm.com>
4333 * target.def (cstore_mode): Return a scalar_int_mode.
4334 * doc/tm.texi: Regenerate.
4335 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
4336 * targhooks.h (default_cstore_mode): Likewise.
4337 * targhooks.c (default_cstore_mode): Likewise, using a forced
4339 * expmed.c (emit_cstore): Expect the target of the cstore to be
4342 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4343 Alan Hayward <alan.hayward@arm.com>
4344 David Sherwood <david.sherwood@arm.com>
4346 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
4348 (niter_desc): Likewise mode.
4349 (iv_analyze): Add a mode parameter.
4351 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
4352 and change its type to scalar_int_mode.
4353 * loop-iv.c: Update commentary at head of file.
4354 (iv_constant): Pass the mode paraeter before the rtx it describes
4355 and change its type to scalar_int_mode. Remove VOIDmode handling.
4356 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
4357 (iv_extend): Likewise.
4358 (shorten_into_mode): Likewise.
4359 (iv_add): Use scalar_int_mode.
4360 (iv_mult): Likewise.
4361 (iv_shift): Likewise.
4362 (canonicalize_iv_subregs): Likewise.
4363 (get_biv_step_1): Pass the outer_mode parameter before the rtx
4364 it describes and change its mode to scalar_int_mode. Also change
4365 the type of the returned inner_mode to scalar_int_mode.
4366 (get_biv_step): Likewise, turning outer_mode from a pointer
4367 into a direct parameter. Update call to get_biv_step_1.
4368 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
4369 iv_constant and get_biv_step.
4370 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
4371 and change its type to scalar_int_mode. Don't initialise iv->mode
4372 to VOIDmode and remove later checks for its still being VOIDmode.
4373 Update calls to iv_analyze_op and iv_analyze_expr. Check
4374 is_a <scalar_int_mode> when changing the mode under consideration.
4375 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
4376 Update call to iv_analyze_expr.
4377 (iv_analyze_op): Add a mode parameter. Reject subregs whose
4378 inner register is not also a scalar_int_mode. Update call to
4380 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
4381 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
4382 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
4383 separate mode class checks. Update calls to iv_analyze. Remove
4384 fix-up of VOIDmodes after iv_analyze_biv.
4385 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
4386 don't have a scalar_int_mode. Update call to biv_p.
4388 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4389 Alan Hayward <alan.hayward@arm.com>
4390 David Sherwood <david.sherwood@arm.com>
4392 * cfgexpand.c (convert_debug_memory_address): Use
4393 as_a <scalar_int_mode>.
4394 * combine.c (expand_compound_operation): Likewise.
4395 (make_extraction): Likewise.
4396 (change_zero_ext): Likewise.
4397 (simplify_comparison): Likewise.
4398 * cse.c (cse_insn): Likewise.
4399 * dwarf2out.c (minmax_loc_descriptor): Likewise.
4400 (mem_loc_descriptor): Likewise.
4401 (loc_descriptor): Likewise.
4402 * expmed.c (init_expmed_one_mode): Likewise.
4403 (synth_mult): Likewise.
4404 (emit_store_flag_1): Likewise.
4405 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
4406 of a comparison with size.
4407 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
4408 (reduce_to_bit_field_precision): Likewise.
4409 * function.c (expand_function_end): Likewise.
4410 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
4411 * loop-doloop.c (doloop_modify): Likewise.
4412 * optabs.c (expand_binop): Likewise.
4413 (expand_unop): Likewise.
4414 (expand_copysign_absneg): Likewise.
4415 (prepare_cmp_insn): Likewise.
4416 (maybe_legitimize_operand): Likewise.
4417 * recog.c (const_scalar_int_operand): Likewise.
4418 * rtlanal.c (get_address_mode): Likewise.
4419 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4420 (simplify_cond_clz_ctz): Likewise.
4421 * tree-nested.c (get_nl_goto_field): Likewise.
4422 * tree.c (build_vector_type_for_mode): Likewise.
4423 * var-tracking.c (use_narrower_mode): Likewise.
4425 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4426 Alan Hayward <alan.hayward@arm.com>
4427 David Sherwood <david.sherwood@arm.com>
4429 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
4430 * builtins.c (expand_builtin_signbit): Use it.
4431 * cfgexpand.c (expand_debug_expr): Likewise.
4432 * dojump.c (do_jump): Likewise.
4433 (do_compare_and_jump): Likewise.
4434 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
4435 * expmed.c (make_tree): Likewise.
4436 * expr.c (expand_expr_real_2): Likewise.
4437 (expand_expr_real_1): Likewise.
4438 (try_casesi): Likewise.
4439 * fold-const-call.c (fold_const_call_ss): Likewise.
4440 * fold-const.c (unextend): Likewise.
4441 (extract_muldiv_1): Likewise.
4442 (fold_single_bit_test): Likewise.
4443 (native_encode_int): Likewise.
4444 (native_encode_string): Likewise.
4445 (native_interpret_int): Likewise.
4446 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
4447 * internal-fn.c (expand_addsub_overflow): Likewise.
4448 (expand_neg_overflow): Likewise.
4449 (expand_mul_overflow): Likewise.
4450 (expand_arith_overflow): Likewise.
4451 * match.pd: Likewise.
4452 * stor-layout.c (layout_type): Likewise.
4453 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
4454 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
4455 * tree-ssanames.c (get_range_info): Likewise.
4456 * tree-switch-conversion.c (array_value_type) Likewise.
4457 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
4458 (vect_recog_divmod_pattern): Likewise.
4459 (vect_recog_mixed_size_cond_pattern): Likewise.
4460 * tree-vrp.c (extract_range_basic): Likewise.
4461 (simplify_float_conversion_using_ranges): Likewise.
4462 * tree.c (int_fits_type_p): Likewise.
4463 * ubsan.c (instrument_bool_enum_load): Likewise.
4464 * varasm.c (mergeable_string_section): Likewise.
4465 (narrowing_initializer_constant_valid_p): Likewise.
4466 (output_constant): Likewise.
4468 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4469 Alan Hayward <alan.hayward@arm.com>
4470 David Sherwood <david.sherwood@arm.com>
4472 * machmode.h (NARROWEST_INT_MODE): New macro.
4473 * expr.c (alignment_for_piecewise_move): Use it instead of
4474 GET_CLASS_NARROWEST_MODE (MODE_INT).
4475 (push_block): Likewise.
4476 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
4478 * tree-vrp.c (simplify_float_conversion_using_ranges): 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 * postreload.c (move2add_valid_value_p): Change the type of the
4485 mode parameter to scalar_int_mode.
4486 (move2add_use_add2_insn): Add a mode parameter and use it instead
4488 (move2add_use_add3_insn): Likewise.
4489 (reload_cse_move2add): Update accordingly.
4491 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4492 Alan Hayward <alan.hayward@arm.com>
4493 David Sherwood <david.sherwood@arm.com>
4495 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
4497 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
4498 * optabs.c (expand_unop): Likewise.
4500 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4501 Alan Hayward <alan.hayward@arm.com>
4502 David Sherwood <david.sherwood@arm.com>
4504 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
4505 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
4506 (popcount_loc_descriptor): Likewise.
4507 (bswap_loc_descriptor): Likewise.
4508 (rotate_loc_descriptor): Likewise.
4509 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
4510 calling the functions above.
4512 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4513 Alan Hayward <alan.hayward@arm.com>
4514 David Sherwood <david.sherwood@arm.com>
4516 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
4518 (try_combine): Likewise.
4519 (simplify_if_then_else): Likewise.
4520 * cse.c (cse_insn): Likewise.
4521 * dwarf2out.c (mem_loc_descriptor): Likewise.
4522 * emit-rtl.c (gen_lowpart_common): Likewise.
4523 * simplify-rtx.c (simplify_truncation): Likewise.
4524 (simplify_binary_operation_1): Likewise.
4525 (simplify_const_relational_operation): Likewise.
4526 (simplify_ternary_operation): Likewise.
4527 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
4529 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4530 Alan Hayward <alan.hayward@arm.com>
4531 David Sherwood <david.sherwood@arm.com>
4533 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
4534 * reload.c (push_reload): Likewise.
4535 (find_reloads): Likewise.
4537 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4538 Alan Hayward <alan.hayward@arm.com>
4539 David Sherwood <david.sherwood@arm.com>
4541 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
4542 (make_compound_operation_int): Likewise.
4543 (change_zero_ext): Likewise.
4544 * expr.c (convert_move): Likewise.
4545 (convert_modes): Likewise.
4546 * fwprop.c (forward_propagate_subreg): Likewise.
4547 * loop-iv.c (get_biv_step_1): Likewise.
4548 * optabs.c (widen_operand): Likewise.
4549 * postreload.c (move2add_valid_value_p): Likewise.
4550 * recog.c (simplify_while_replacing): Likewise.
4551 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4552 (simplify_binary_operation_1): Likewise. Remove redundant
4553 mode equality check.
4555 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4556 Alan Hayward <alan.hayward@arm.com>
4557 David Sherwood <david.sherwood@arm.com>
4559 * combine.c (combine_simplify_rtx): Add checks for
4560 is_a <scalar_int_mode>.
4561 (simplify_if_then_else): Likewise.
4562 (make_field_assignment): Likewise.
4563 (simplify_comparison): Likewise.
4564 * ifcvt.c (noce_try_bitop): Likewise.
4565 * loop-invariant.c (canonicalize_address_mult): Likewise.
4566 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4568 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4569 Alan Hayward <alan.hayward@arm.com>
4570 David Sherwood <david.sherwood@arm.com>
4572 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
4573 is_a <scalar_int_mode> instead of != BLKmode.
4575 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4576 Alan Hayward <alan.hayward@arm.com>
4577 David Sherwood <david.sherwood@arm.com>
4579 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
4580 instead of != VOIDmode.
4581 * combine.c (if_then_else_cond): Likewise.
4582 (change_zero_ext): Likewise.
4583 * dwarf2out.c (mem_loc_descriptor): Likewise.
4584 (loc_descriptor): Likewise.
4585 * rtlanal.c (canonicalize_condition): Likewise.
4586 * simplify-rtx.c (simplify_relational_operation_1): 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 * simplify-rtx.c (simplify_binary_operation_1): Use
4593 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
4595 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4596 Alan Hayward <alan.hayward@arm.com>
4597 David Sherwood <david.sherwood@arm.com>
4599 * wide-int.h (int_traits<unsigned char>) New class.
4600 (int_traits<unsigned short>) Likewise.
4601 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
4602 Use GET_MODE_UNIT_PRECISION and remove redundant test for
4604 * combine.c (set_nonzero_bits_and_sign_copies): Use
4605 is_a <scalar_int_mode>.
4606 (find_split_point): Likewise.
4607 (combine_simplify_rtx): Likewise.
4608 (simplify_logical): Likewise.
4609 (expand_compound_operation): Likewise.
4610 (expand_field_assignment): Likewise.
4611 (make_compound_operation): Likewise.
4612 (extended_count): Likewise.
4613 (change_zero_ext): Likewise.
4614 (simplify_comparison): Likewise.
4615 * dwarf2out.c (scompare_loc_descriptor): Likewise.
4616 (ucompare_loc_descriptor): Likewise.
4617 (minmax_loc_descriptor): Likewise.
4618 (mem_loc_descriptor): Likewise.
4619 (loc_descriptor): Likewise.
4620 * expmed.c (init_expmed_one_mode): Likewise.
4621 * lra-constraints.c (lra_constraint_offset): Likewise.
4622 * optabs.c (prepare_libcall_arg): Likewise.
4623 * postreload.c (move2add_note_store): Likewise.
4624 * reload.c (operands_match_p): Likewise.
4625 * rtl.h (load_extend_op): Likewise.
4626 * rtlhooks.c (gen_lowpart_general): Likewise.
4627 * simplify-rtx.c (simplify_truncation): Likewise.
4628 (simplify_unary_operation_1): Likewise.
4629 (simplify_binary_operation_1): Likewise.
4630 (simplify_const_binary_operation): Likewise.
4631 (simplify_const_relational_operation): Likewise.
4632 (simplify_subreg): Likewise.
4633 * stor-layout.c (bitwise_mode_for_mode): Likewise.
4634 * var-tracking.c (adjust_mems): Likewise.
4635 (prepare_call_arguments): Likewise.
4637 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4638 Alan Hayward <alan.hayward@arm.com>
4639 David Sherwood <david.sherwood@arm.com>
4641 * machmode.h (is_int_mode): New fuction.
4642 * combine.c (find_split_point): Use it.
4643 (combine_simplify_rtx): Likewise.
4644 (simplify_if_then_else): Likewise.
4645 (simplify_set): Likewise.
4646 (simplify_shift_const_1): Likewise.
4647 (simplify_comparison): Likewise.
4648 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
4649 * cse.c (notreg_cost): Likewise.
4650 (cse_insn): Likewise.
4651 * cselib.c (cselib_lookup_1): Likewise.
4652 * dojump.c (do_jump_1): Likewise.
4653 (do_compare_rtx_and_jump): Likewise.
4654 * dse.c (get_call_args): Likewise.
4655 * dwarf2out.c (rtl_for_decl_init): Likewise.
4656 (native_encode_initializer): Likewise.
4657 * expmed.c (emit_store_flag_1): Likewise.
4658 (emit_store_flag): Likewise.
4659 * expr.c (convert_modes): Likewise.
4660 (store_field): Likewise.
4661 (expand_expr_real_1): Likewise.
4662 * fold-const.c (fold_read_from_constant_string): Likewise.
4663 * gimple-ssa-sprintf.c (get_format_string): Likewise.
4664 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
4665 * optabs.c (expand_binop): Likewise.
4666 (expand_unop): Likewise.
4667 (expand_abs_nojump): Likewise.
4668 (expand_one_cmpl_abs_nojump): Likewise.
4669 * simplify-rtx.c (mode_signbit_p): Likewise.
4670 (val_signbit_p): Likewise.
4671 (val_signbit_known_set_p): Likewise.
4672 (val_signbit_known_clear_p): Likewise.
4673 (simplify_relational_operation_1): Likewise.
4674 * tree.c (vector_type_mode): Likewise.
4676 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4677 Alan Hayward <alan.hayward@arm.com>
4678 David Sherwood <david.sherwood@arm.com>
4680 * machmode.h (smallest_mode_for_size): Fix formatting.
4681 (smallest_int_mode_for_size): New function.
4682 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
4683 instead of smallest_mode_for_size.
4684 * combine.c (make_extraction): Likewise.
4685 * config/arc/arc.c (arc_expand_movmem): Likewise.
4686 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
4687 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
4688 * config/s390/s390.c (s390_expand_insv): Likewise.
4689 * config/sparc/sparc.c (assign_int_registers): Likewise.
4690 * config/spu/spu.c (spu_function_value): Likewise.
4691 (spu_function_arg): Likewise.
4692 * coverage.c (get_gcov_type): Likewise.
4693 (get_gcov_unsigned_t): Likewise.
4694 * dse.c (find_shift_sequence): Likewise.
4695 * expmed.c (store_bit_field_1): Likewise.
4696 * expr.c (convert_move): Likewise.
4697 (store_field): Likewise.
4698 * internal-fn.c (expand_arith_overflow): Likewise.
4699 * optabs-query.c (get_best_extraction_insn): Likewise.
4700 * optabs.c (expand_twoval_binop_libfunc): Likewise.
4701 * stor-layout.c (layout_type): Likewise.
4702 (initialize_sizetypes): Likewise.
4703 * targhooks.c (default_get_mask_mode): Likewise.
4704 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
4706 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4707 Alan Hayward <alan.hayward@arm.com>
4708 David Sherwood <david.sherwood@arm.com>
4710 * machmode.h (opt_mode::else_blk): New function.
4711 (int_mode_for_mode): Declare.
4712 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
4713 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
4715 * cfgexpand.c (expand_debug_expr): Likewise.
4716 * combine.c (gen_lowpart_or_truncate): Likewise.
4717 (gen_lowpart_for_combine): Likewise.
4718 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
4719 * config/avr/avr.c (avr_to_int_mode): Likewise.
4720 (avr_out_plus_1): Likewise.
4721 (avr_out_plus): Likewise.
4722 (avr_out_round): Likewise.
4723 * config/i386/i386.c (ix86_split_to_parts): Likewise.
4724 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
4725 (s390_expand_vcond): Likewise.
4726 * config/spu/spu.c (spu_split_immediate): Likewise.
4727 (spu_expand_mov): Likewise.
4728 * dse.c (get_stored_val): Likewise.
4729 * expmed.c (store_bit_field_1): Likewise.
4730 (convert_extracted_bit_field): Use int_mode_for_mode instead of
4732 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
4733 (extract_low_bits): Likewise.
4734 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
4735 handling rather than repeating the check.
4736 (emit_group_store): Likewise.
4737 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
4738 * optabs.c (expand_absneg_bit): Likewise.
4739 (expand_copysign_absneg): Likewise.
4740 (expand_copysign_bit): Likewise.
4741 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
4742 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
4743 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
4744 * var-tracking.c (prepare_call_arguments): Likewise.
4745 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
4746 int_mode_for_mode instead of mode_for_size.
4747 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
4749 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4750 Alan Hayward <alan.hayward@arm.com>
4751 David Sherwood <david.sherwood@arm.com>
4753 * machmode.h (int_mode_for_size): New function.
4754 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
4755 instead of mode_for_size.
4756 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
4758 * combine.c (expand_field_assignment): Use int_mode_for_size
4759 instead of mode_for_size.
4760 (make_extraction): Likewise.
4761 (simplify_shift_const_1): Likewise.
4762 (simplify_comparison): Likewise.
4763 * dojump.c (do_jump): Likewise.
4764 * dwarf2out.c (mem_loc_descriptor): Likewise.
4765 * emit-rtl.c (init_derived_machine_modes): Likewise.
4766 * expmed.c (flip_storage_order): Likewise.
4767 (convert_extracted_bit_field): Likewise.
4768 * expr.c (copy_blkmode_from_reg): Likewise.
4769 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
4770 * internal-fn.c (expand_mul_overflow): Likewise.
4771 * lower-subreg.c (simple_move): Likewise.
4772 * optabs-libfuncs.c (init_optabs): Likewise.
4773 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4774 * tree.c (vector_type_mode): Likewise.
4775 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
4776 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
4777 * tree-vect-generic.c (expand_vector_parallel): Likewise.
4778 * tree-vect-stmts.c (vectorizable_load): Likewise.
4779 (vectorizable_store): Likewise.
4781 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4782 Alan Hayward <alan.hayward@arm.com>
4783 David Sherwood <david.sherwood@arm.com>
4785 * coretypes.h (pod_mode): New type.
4786 (scalar_int_mode_pod): New typedef.
4787 * machmode.h (pod_mode): New class.
4788 (int_n_data_t::m): Change type to scalar_int_mode_pod.
4789 * genmodes.c (emit_mode_int_n): Update accordingly.
4790 * lower-subreg.h (target_lower_subreg): Change type to
4791 scalar_int_mode_pod.
4792 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
4793 scalar_int_mode_pod.
4795 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4796 Alan Hayward <alan.hayward@arm.com>
4797 David Sherwood <david.sherwood@arm.com>
4799 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
4800 machine_mode to scalar_int_mode.
4801 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
4802 (rs6000_option_override_internal): Remove cast to int.
4803 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
4804 machine_mode to scalar_int_mode.
4805 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
4806 (rs6000_option_override_internal): Remove cast to int.
4807 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
4808 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
4810 * config/s390/s390.c (s390_expand_builtin): Likewise.
4811 * coretypes.h (scalar_int_mode): New type.
4812 (opt_scalar_int_mode): New typedef.
4813 * machmode.h (scalar_int_mode): New class.
4814 (scalar_int_mode::includes_p): New function.
4815 (byte_mode): Change type to scalar_int_mode.
4816 (word_mode): Likewise.
4817 (ptr_mode): Likewise.
4818 * emit-rtl.c (byte_mode): Likewise.
4819 (word_mode): Likewise.
4820 (ptr_mode): Likewise.
4821 (init_derived_machine_modes): Update accordingly.
4822 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
4823 and MODE_PARTIAL_INT.
4824 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
4825 opt_scalar_int_mode.
4827 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4828 Alan Hayward <alan.hayward@arm.com>
4829 David Sherwood <david.sherwood@arm.com>
4831 * target.def (libgcc_floating_mode_supported_p): Take a
4833 * doc/tm.texi: Regenerate.
4834 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
4836 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
4837 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
4840 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4841 Alan Hayward <alan.hayward@arm.com>
4842 David Sherwood <david.sherwood@arm.com>
4844 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
4845 * doc/tm.texi: Regenerate.
4846 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
4847 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
4848 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
4849 * targhooks.h (default_floatn_mode): Likewise.
4850 * targhooks.c (default_floatn_mode): Likewise.
4851 * tree.c (build_common_tree_nodes): Update accordingly.
4853 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4854 Alan Hayward <alan.hayward@arm.com>
4855 David Sherwood <david.sherwood@arm.com>
4857 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
4858 (mode_iterator::iterate_p): Likewise.
4859 (mode_iterator::get_wider): Likewise.
4860 * expr.c (init_expr_target): Use opt_scalar_float_mode.
4862 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4863 Alan Hayward <alan.hayward@arm.com>
4864 David Sherwood <david.sherwood@arm.com>
4866 * coretypes.h (opt_scalar_float_mode): New typedef.
4867 * machmode.h (float_mode_for_size): New function.
4868 * emit-rtl.c (double_mode): Delete.
4869 (init_emit_once): Use float_mode_for_size.
4870 * stor-layout.c (layout_type): Likewise.
4871 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
4873 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4874 Alan Hayward <alan.hayward@arm.com>
4875 David Sherwood <david.sherwood@arm.com>
4877 * output.h (assemble_real): Take a scalar_float_mode.
4878 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
4879 * config/arm/arm.md (consttable_4): Likewise.
4880 (consttable_8): Likewise.
4881 (consttable_16): Likewise.
4882 * config/mips/mips.md (consttable_float): Likewise.
4883 * config/s390/s390.c (s390_output_pool_entry): Likewise.
4884 * varasm.c (assemble_real): Take a scalar_float_mode.
4885 (output_constant_pool_2): Update accordingly.
4886 (output_constant): Likewise.
4888 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4889 Alan Hayward <alan.hayward@arm.com>
4890 David Sherwood <david.sherwood@arm.com>
4892 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
4893 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
4894 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
4895 (native_encode_real): Likewise.
4896 (native_interpret_real): Likewise.
4897 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
4898 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4900 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4901 Alan Hayward <alan.hayward@arm.com>
4902 David Sherwood <david.sherwood@arm.com>
4904 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
4905 <scalar_float_mode>. Simplify.
4906 (gen_extend_conv_libfunc): Likewise.
4908 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4909 Alan Hayward <alan.hayward@arm.com>
4910 David Sherwood <david.sherwood@arm.com>
4912 * coretypes.h (scalar_float_mode): New type.
4913 * machmode.h (mode_traits::from_int): Use machine_mode if
4914 USE_ENUM_MODES is defined.
4915 (is_a): New function.
4917 (dyn_cast): Likewise.
4918 (scalar_float_mode): New class.
4919 (scalar_float_mode::includes_p): New function.
4920 (is_float_mode): Likewise.
4921 * gdbhooks.py (MachineModePrinter): New class.
4922 (build_pretty_printer): Use it for scalar_float_mode.
4923 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
4924 (format_helper::format_helper): Turn into a template.
4925 * genmodes.c (get_mode_class): New function.
4926 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
4927 or machine_mode if none.
4928 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
4929 as_a <scalar_float_mode>.
4930 * dwarf2out.c (mem_loc_descriptor): Likewise.
4931 (insert_float): Likewise.
4932 (add_const_value_attribute): Likewise.
4933 * simplify-rtx.c (simplify_immed_subreg): Likewise.
4934 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
4935 (expand_unop): Update accordingly.
4936 (expand_abs_nojump): Likewise.
4937 (expand_copysign_absneg): Take a scalar_float_mode.
4938 (expand_copysign_bit): Likewise.
4939 (expand_copysign): Update accordingly.
4941 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4942 Alan Hayward <alan.hayward@arm.com>
4943 David Sherwood <david.sherwood@arm.com>
4945 * coretypes.h (opt_mode): New class.
4946 * machmode.h (opt_mode): Likewise.
4947 (opt_mode::else_void): New function.
4948 (opt_mode::require): Likewise.
4949 (opt_mode::exists): Likewise.
4950 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
4951 (GET_MODE_2XWIDER_MODE): Likewise.
4952 (mode_iterator::get_wider): Update accordingly.
4953 (mode_iterator::get_2xwider): Likewise.
4954 (mode_iterator::get_known_wider): Likewise, turning into a template.
4955 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
4956 forcing a wider mode to exist.
4957 * config/cr16/cr16.h (LONG_REG_P): Likewise.
4958 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4959 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
4960 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
4961 * lower-subreg.c (init_lower_subreg): Likewise.
4962 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
4963 on the final iteration.
4964 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
4965 a wider mode exists before asking for a move pattern.
4966 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
4967 forcing a wider mode to exist.
4968 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
4969 returning false if no such mode exists.
4970 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
4971 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
4972 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
4973 Avoid checking for a MODE_INT if we already know the mode is not a
4975 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
4976 forcing a wider mode to exist.
4977 (expmed_mult_highpart_optab): Likewise.
4978 (expmed_mult_highpart): Likewise.
4979 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
4981 * lto-streamer-in.c (lto_input_mode_table): Likewise.
4982 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
4983 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
4984 * internal-fn.c (expand_mul_overflow): Update use of
4985 GET_MODE_2XWIDER_MODE.
4986 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4987 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
4988 GET_MODE_WIDER_MODE.
4989 (convert_plusminus_to_widen): Likewise.
4990 * tree-switch-conversion.c (array_value_type): Likewise.
4991 * var-tracking.c (emit_note_insn_var_location): Likewise.
4992 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4993 Return false inside rather than outside the loop if no wider mode
4995 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
4996 and GET_MODE_2XWIDER_MODE
4997 (can_compare_p): Use else_void.
4998 * gdbhooks.py (OptMachineModePrinter): New class.
4999 (build_pretty_printer): Use it for opt_mode.
5001 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5002 Alan Hayward <alan.hayward@arm.com>
5003 David Sherwood <david.sherwood@arm.com>
5005 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
5006 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
5008 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5009 Alan Hayward <alan.hayward@arm.com>
5010 David Sherwood <david.sherwood@arm.com>
5012 * machmode.h (mode_traits): New structure.
5013 (get_narrowest_mode): New function.
5014 (mode_iterator::start): Likewise.
5015 (mode_iterator::iterate_p): Likewise.
5016 (mode_iterator::get_wider): Likewise.
5017 (mode_iterator::get_known_wider): Likewise.
5018 (mode_iterator::get_2xwider): Likewise.
5019 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
5020 (FOR_EACH_MODE): Likewise.
5021 (FOR_EACH_MODE_FROM): Likewise.
5022 (FOR_EACH_MODE_UNTIL): Likewise.
5023 (FOR_EACH_WIDER_MODE): Likewise.
5024 (FOR_EACH_2XWIDER_MODE): Likewise.
5025 * builtins.c (expand_builtin_strlen): Use new mode iterators.
5026 * combine.c (simplify_comparison): Likewise
5027 * config/i386/i386.c (type_natural_mode): Likewise.
5028 * cse.c (cse_insn): Likewise.
5029 * dse.c (find_shift_sequence): Likewise.
5030 * emit-rtl.c (init_derived_machine_modes): Likewise.
5031 (init_emit_once): Likewise.
5032 * explow.c (hard_function_value): Likewise.
5033 * expmed.c (extract_fixed_bit_field_1): Likewise.
5034 (extract_bit_field_1): Likewise.
5035 (expand_divmod): Likewise.
5036 (emit_store_flag_1): Likewise.
5037 * expr.c (init_expr_target): Likewise.
5038 (convert_move): Likewise.
5039 (alignment_for_piecewise_move): Likewise.
5040 (widest_int_mode_for_size): Likewise.
5041 (emit_block_move_via_movmem): Likewise.
5042 (copy_blkmode_to_reg): Likewise.
5043 (set_storage_via_setmem): Likewise.
5044 (compress_float_constant): Likewise.
5045 * omp-low.c (omp_clause_aligned_alignment): Likewise.
5046 * optabs-query.c (get_best_extraction_insn): Likewise.
5047 * optabs.c (expand_binop): Likewise.
5048 (expand_twoval_unop): Likewise.
5049 (expand_twoval_binop): Likewise.
5050 (widen_leading): Likewise.
5051 (widen_bswap): Likewise.
5052 (expand_parity): Likewise.
5053 (expand_unop): Likewise.
5054 (prepare_cmp_insn): Likewise.
5055 (prepare_float_lib_cmp): Likewise.
5056 (expand_float): Likewise.
5057 (expand_fix): Likewise.
5058 (expand_sfix_optab): Likewise.
5059 * postreload.c (move2add_use_add2_insn): Likewise.
5060 * reg-stack.c (reg_to_stack): Likewise.
5061 * reginfo.c (choose_hard_reg_mode): Likewise.
5062 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
5063 * stor-layout.c (mode_for_size): Likewise.
5064 (smallest_mode_for_size): Likewise.
5065 (mode_for_vector): Likewise.
5066 (finish_bitfield_representative): Likewise.
5067 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
5068 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
5069 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5070 * var-tracking.c (prepare_call_arguments): Likewise.
5072 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5073 Alan Hayward <alan.hayward@arm.com>
5074 David Sherwood <david.sherwood@arm.com>
5076 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
5077 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
5078 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
5079 * machmode.h (mode_size): Move earlier in file.
5080 (mode_precision): Likewise.
5081 (mode_inner): Likewise.
5082 (mode_nunits): Likewise.
5083 (mode_unit_size): Likewise.
5084 (unit_unit_precision): Likewise.
5085 (mode_wider): Likewise.
5086 (mode_2xwider): Likewise.
5087 (machine_mode): New class.
5088 (mode_to_bytes): New function.
5089 (mode_to_bits): Likewise.
5090 (mode_to_precision): Likewise.
5091 (mode_to_inner): Likewise.
5092 (mode_to_unit_size): Likewise.
5093 (mode_to_unit_precision): Likewise.
5094 (mode_to_nunits): Likewise.
5095 (GET_MODE_SIZE): Use mode_to_bytes.
5096 (GET_MODE_BITSIZE): Use mode_to_bits.
5097 (GET_MODE_PRECISION): Use mode_to_precision.
5098 (GET_MODE_INNER): Use mode_to_inner.
5099 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
5100 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
5101 (GET_MODE_NUNITS): Use mode_to_nunits.
5102 * system.h (ALWAYS_INLINE): New macro.
5103 * config/powerpcspe/powerpcspe-c.c
5104 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
5105 int for arg1_mode and arg2_mode.
5107 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5108 Alan Hayward <alan.hayward@arm.com>
5109 David Sherwood <david.sherwood@arm.com>
5111 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
5112 Prefix mode names with E_ in case statements.
5113 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5114 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
5115 (aarch64_split_simd_move): Likewise.
5116 (aarch64_gen_storewb_pair): Likewise.
5117 (aarch64_gen_loadwb_pair): Likewise.
5118 (aarch64_gen_store_pair): Likewise.
5119 (aarch64_gen_load_pair): Likewise.
5120 (aarch64_get_condition_code_1): Likewise.
5121 (aarch64_constant_pool_reload_icode): Likewise.
5122 (get_rsqrte_type): Likewise.
5123 (get_rsqrts_type): Likewise.
5124 (get_recpe_type): Likewise.
5125 (get_recps_type): Likewise.
5126 (aarch64_gimplify_va_arg_expr): Likewise.
5127 (aarch64_simd_container_mode): Likewise.
5128 (aarch64_emit_load_exclusive): Likewise.
5129 (aarch64_emit_store_exclusive): Likewise.
5130 (aarch64_expand_compare_and_swap): Likewise.
5131 (aarch64_gen_atomic_cas): Likewise.
5132 (aarch64_emit_bic): Likewise.
5133 (aarch64_emit_atomic_swap): Likewise.
5134 (aarch64_emit_atomic_load_op): Likewise.
5135 (aarch64_evpc_trn): Likewise.
5136 (aarch64_evpc_uzp): Likewise.
5137 (aarch64_evpc_zip): Likewise.
5138 (aarch64_evpc_ext): Likewise.
5139 (aarch64_evpc_rev): Likewise.
5140 (aarch64_evpc_dup): Likewise.
5141 (aarch64_gen_ccmp_first): Likewise.
5142 (aarch64_gen_ccmp_next): Likewise.
5143 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
5144 (alpha_emit_xfloating_libcall): Likewise.
5145 (emit_insxl): Likewise.
5146 (alpha_arg_type): Likewise.
5147 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
5148 (arc_preferred_simd_mode): Likewise.
5149 (arc_secondary_reload): Likewise.
5150 (get_arc_condition_code): Likewise.
5151 (arc_print_operand): Likewise.
5152 (arc_legitimate_constant_p): Likewise.
5153 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5154 * config/arc/arc.md (casesi_load): Likewise.
5155 (casesi_compact_jump): Likewise.
5156 * config/arc/predicates.md (proper_comparison_operator): Likewise.
5157 (cc_use_register): Likewise.
5158 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5159 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
5160 (arm_init_iwmmxt_builtins): Likewise.
5161 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
5162 (neon_expand_vector_init): Likewise.
5163 (arm_attr_length_move_neon): Likewise.
5164 (maybe_get_arm_condition_code): Likewise.
5165 (arm_emit_vector_const): Likewise.
5166 (arm_preferred_simd_mode): Likewise.
5167 (arm_output_iwmmxt_tinsr): Likewise.
5168 (thumb1_output_casesi): Likewise.
5169 (thumb2_output_casesi): Likewise.
5170 (arm_emit_load_exclusive): Likewise.
5171 (arm_emit_store_exclusive): Likewise.
5172 (arm_expand_compare_and_swap): Likewise.
5173 (arm_evpc_neon_vuzp): Likewise.
5174 (arm_evpc_neon_vzip): Likewise.
5175 (arm_evpc_neon_vrev): Likewise.
5176 (arm_evpc_neon_vtrn): Likewise.
5177 (arm_evpc_neon_vext): Likewise.
5178 (arm_validize_comparison): Likewise.
5179 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
5180 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
5181 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
5182 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
5183 (c6x_preferred_simd_mode): Likewise.
5184 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
5185 (epiphany_rtx_costs): Likewise.
5186 * config/epiphany/predicates.md (proper_comparison_operator):
5188 * config/frv/frv.c (condexec_memory_operand): Likewise.
5189 (frv_emit_move): Likewise.
5190 (output_move_single): Likewise.
5191 (output_condmove_single): Likewise.
5192 (frv_hard_regno_mode_ok): Likewise.
5193 (frv_matching_accg_mode): Likewise.
5194 * config/h8300/h8300.c (split_adds_subs): Likewise.
5195 (h8300_rtx_costs): Likewise.
5196 (h8300_print_operand): Likewise.
5197 (compute_mov_length): Likewise.
5198 (output_logical_op): Likewise.
5199 (compute_logical_op_length): Likewise.
5200 (compute_logical_op_cc): Likewise.
5201 (h8300_shift_needs_scratch_p): Likewise.
5202 (output_a_shift): Likewise.
5203 (compute_a_shift_length): Likewise.
5204 (compute_a_shift_cc): Likewise.
5205 (expand_a_rotate): Likewise.
5206 (output_a_rotate): Likewise.
5207 * config/i386/i386.c (classify_argument): Likewise.
5208 (function_arg_advance_32): Likewise.
5209 (function_arg_32): Likewise.
5210 (function_arg_64): Likewise.
5211 (function_value_64): Likewise.
5212 (ix86_gimplify_va_arg): Likewise.
5213 (ix86_legitimate_constant_p): Likewise.
5214 (put_condition_code): Likewise.
5215 (split_double_mode): Likewise.
5216 (ix86_avx256_split_vector_move_misalign): Likewise.
5217 (ix86_expand_vector_logical_operator): Likewise.
5218 (ix86_split_idivmod): Likewise.
5219 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
5220 (ix86_build_const_vector): Likewise.
5221 (ix86_build_signbit_mask): Likewise.
5222 (ix86_match_ccmode): Likewise.
5223 (ix86_cc_modes_compatible): Likewise.
5224 (ix86_expand_branch): Likewise.
5225 (ix86_expand_sse_cmp): Likewise.
5226 (ix86_expand_sse_movcc): Likewise.
5227 (ix86_expand_int_sse_cmp): Likewise.
5228 (ix86_expand_vec_perm_vpermi2): Likewise.
5229 (ix86_expand_vec_perm): Likewise.
5230 (ix86_expand_sse_unpack): Likewise.
5231 (ix86_expand_int_addcc): Likewise.
5232 (ix86_split_to_parts): Likewise.
5233 (ix86_vectorize_builtin_gather): Likewise.
5234 (ix86_vectorize_builtin_scatter): Likewise.
5235 (avx_vpermilp_parallel): Likewise.
5236 (inline_memory_move_cost): Likewise.
5237 (ix86_tieable_integer_mode_p): Likewise.
5238 (x86_maybe_negate_const_int): Likewise.
5239 (ix86_expand_vector_init_duplicate): Likewise.
5240 (ix86_expand_vector_init_one_nonzero): Likewise.
5241 (ix86_expand_vector_init_one_var): Likewise.
5242 (ix86_expand_vector_init_concat): Likewise.
5243 (ix86_expand_vector_init_interleave): Likewise.
5244 (ix86_expand_vector_init_general): Likewise.
5245 (ix86_expand_vector_set): Likewise.
5246 (ix86_expand_vector_extract): Likewise.
5247 (emit_reduc_half): Likewise.
5248 (ix86_emit_i387_round): Likewise.
5249 (ix86_mangle_type): Likewise.
5250 (ix86_expand_round_sse4): Likewise.
5251 (expand_vec_perm_blend): Likewise.
5252 (canonicalize_vector_int_perm): Likewise.
5253 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
5254 (expand_vec_perm_1): Likewise.
5255 (expand_vec_perm_interleave3): Likewise.
5256 (expand_vec_perm_even_odd_pack): Likewise.
5257 (expand_vec_perm_even_odd_1): Likewise.
5258 (expand_vec_perm_broadcast_1): Likewise.
5259 (ix86_vectorize_vec_perm_const_ok): Likewise.
5260 (ix86_expand_vecop_qihi): Likewise.
5261 (ix86_expand_mul_widen_hilo): Likewise.
5262 (ix86_expand_sse2_abs): Likewise.
5263 (ix86_expand_pextr): Likewise.
5264 (ix86_expand_pinsr): Likewise.
5265 (ix86_preferred_simd_mode): Likewise.
5266 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
5267 * config/i386/sse.md (*andnot<mode>3): Likewise.
5268 (<mask_codefor><code><mode>3<mask_name>): Likewise.
5269 (*<code><mode>3): Likewise.
5270 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
5271 (ia64_expand_atomic_op): Likewise.
5272 (ia64_arg_type): Likewise.
5273 (ia64_mode_to_int): Likewise.
5274 (ia64_scalar_mode_supported_p): Likewise.
5275 (ia64_vector_mode_supported_p): Likewise.
5276 (expand_vec_perm_broadcast): Likewise.
5277 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
5278 (iq2000_function_arg_advance): Likewise.
5279 (iq2000_function_arg): Likewise.
5280 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
5281 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
5282 (m68k_libcall_value): Likewise.
5283 (m68k_function_value): Likewise.
5284 (sched_attr_op_type): Likewise.
5285 * config/mcore/mcore.c (mcore_output_move): Likewise.
5286 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
5288 (microblaze_function_arg): Likewise.
5289 * config/mips/mips.c (mips16_build_call_stub): Likewise.
5290 (mips_print_operand): Likewise.
5291 (mips_mode_ok_for_mov_fmt_p): Likewise.
5292 (mips_vector_mode_supported_p): Likewise.
5293 (mips_preferred_simd_mode): Likewise.
5294 (mips_expand_vpc_loongson_even_odd): Likewise.
5295 (mips_expand_vec_unpack): Likewise.
5296 (mips_expand_vi_broadcast): Likewise.
5297 (mips_expand_vector_init): Likewise.
5298 (mips_expand_vec_reduc): Likewise.
5299 (mips_expand_msa_cmp): Likewise.
5300 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
5301 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
5302 (cc_flags_for_mode): Likewise.
5303 * config/msp430/msp430.c (msp430_print_operand): Likewise.
5304 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
5305 (nds32_output_casesi_pc_relative): Likewise.
5306 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5307 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
5308 (nvptx_gen_unpack): Likewise.
5309 (nvptx_gen_pack): Likewise.
5310 (nvptx_gen_shuffle): Likewise.
5311 (nvptx_gen_wcast): Likewise.
5312 (nvptx_preferred_simd_mode): Likewise.
5313 * config/pa/pa.c (pa_secondary_reload): Likewise.
5314 * config/pa/predicates.md (base14_operand): Likewise.
5315 * config/powerpcspe/powerpcspe-c.c
5316 (altivec_resolve_overloaded_builtin): Likewise.
5317 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
5319 (rs6000_preferred_simd_mode): Likewise.
5320 (output_vec_const_move): Likewise.
5321 (rs6000_expand_vector_extract): Likewise.
5322 (rs6000_split_vec_extract_var): Likewise.
5323 (reg_offset_addressing_ok_p): Likewise.
5324 (rs6000_legitimate_offset_address_p): Likewise.
5325 (rs6000_legitimize_address): Likewise.
5326 (rs6000_emit_set_const): Likewise.
5327 (rs6000_const_vec): Likewise.
5328 (rs6000_emit_move): Likewise.
5329 (spe_build_register_parallel): Likewise.
5330 (rs6000_darwin64_record_arg_recurse): Likewise.
5331 (swap_selector_for_mode): Likewise.
5332 (spe_init_builtins): Likewise.
5333 (paired_init_builtins): Likewise.
5334 (altivec_init_builtins): Likewise.
5335 (do_load_for_compare): Likewise.
5336 (rs6000_generate_compare): Likewise.
5337 (rs6000_expand_float128_convert): Likewise.
5338 (emit_load_locked): Likewise.
5339 (emit_store_conditional): Likewise.
5340 (rs6000_output_function_epilogue): Likewise.
5341 (rs6000_handle_altivec_attribute): Likewise.
5342 (rs6000_function_value): Likewise.
5343 (emit_fusion_gpr_load): Likewise.
5344 (emit_fusion_p9_load): Likewise.
5345 (emit_fusion_p9_store): Likewise.
5346 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
5347 (fusion_gpr_mem_load): Likewise.
5348 (fusion_addis_mem_combo_load): Likewise.
5349 (fusion_addis_mem_combo_store): Likewise.
5350 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
5351 (fusion_gpr_mem_load): Likewise.
5352 (fusion_addis_mem_combo_load): Likewise.
5353 (fusion_addis_mem_combo_store): Likewise.
5354 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5356 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
5357 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
5358 (rs6000_preferred_simd_mode): Likewise.
5359 (output_vec_const_move): Likewise.
5360 (rs6000_expand_vector_extract): Likewise.
5361 (rs6000_split_vec_extract_var): Likewise.
5362 (reg_offset_addressing_ok_p): Likewise.
5363 (rs6000_legitimate_offset_address_p): Likewise.
5364 (rs6000_legitimize_address): Likewise.
5365 (rs6000_emit_set_const): Likewise.
5366 (rs6000_const_vec): Likewise.
5367 (rs6000_emit_move): Likewise.
5368 (rs6000_darwin64_record_arg_recurse): Likewise.
5369 (swap_selector_for_mode): Likewise.
5370 (paired_init_builtins): Likewise.
5371 (altivec_init_builtins): Likewise.
5372 (rs6000_expand_float128_convert): Likewise.
5373 (emit_load_locked): Likewise.
5374 (emit_store_conditional): Likewise.
5375 (rs6000_output_function_epilogue): Likewise.
5376 (rs6000_handle_altivec_attribute): Likewise.
5377 (rs6000_function_value): Likewise.
5378 (emit_fusion_gpr_load): Likewise.
5379 (emit_fusion_p9_load): Likewise.
5380 (emit_fusion_p9_store): Likewise.
5381 * config/rx/rx.c (rx_gen_move_template): Likewise.
5382 (flags_from_mode): Likewise.
5383 * config/s390/predicates.md (s390_alc_comparison): Likewise.
5384 (s390_slb_comparison): Likewise.
5385 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
5386 (s390_vector_mode_supported_p): Likewise.
5387 (s390_cc_modes_compatible): Likewise.
5388 (s390_match_ccmode_set): Likewise.
5389 (s390_canonicalize_comparison): Likewise.
5390 (s390_emit_compare_and_swap): Likewise.
5391 (s390_branch_condition_mask): Likewise.
5392 (s390_rtx_costs): Likewise.
5393 (s390_secondary_reload): Likewise.
5394 (__SECONDARY_RELOAD_CASE): Likewise.
5395 (s390_expand_cs): Likewise.
5396 (s390_preferred_simd_mode): Likewise.
5397 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
5398 * config/sh/sh.c (sh_print_operand): Likewise.
5399 (dump_table): Likewise.
5400 (sh_secondary_reload): Likewise.
5401 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5402 * config/sh/sh.md (casesi_worker_1): Likewise.
5403 (casesi_worker_2): Likewise.
5404 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
5405 (fcc_comparison_operator): Likewise.
5406 * config/sparc/sparc.c (sparc_expand_move): Likewise.
5407 (emit_soft_tfmode_cvt): Likewise.
5408 (sparc_preferred_simd_mode): Likewise.
5409 (output_cbranch): Likewise.
5410 (sparc_print_operand): Likewise.
5411 (sparc_expand_vec_perm_bmask): Likewise.
5412 (vector_init_bshuffle): Likewise.
5413 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
5414 (spu_vector_mode_supported_p): Likewise.
5415 (spu_expand_insv): Likewise.
5416 (spu_emit_branch_or_set): Likewise.
5417 (spu_handle_vector_attribute): Likewise.
5418 (spu_builtin_splats): Likewise.
5419 (spu_builtin_extract): Likewise.
5420 (spu_builtin_promote): Likewise.
5421 (spu_expand_sign_extend): Likewise.
5422 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
5423 (tilegx_simd_int): Likewise.
5424 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
5425 (tilepro_simd_int): Likewise.
5426 * config/v850/v850.c (const_double_split): Likewise.
5427 (v850_print_operand): Likewise.
5428 (ep_memory_offset): Likewise.
5429 * config/vax/vax.c (vax_rtx_costs): Likewise.
5430 (vax_output_int_move): Likewise.
5431 (vax_output_int_add): Likewise.
5432 (vax_output_int_subtract): Likewise.
5433 * config/visium/predicates.md (visium_branch_operator): Likewise.
5434 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
5435 (visium_print_operand_address): Likewise.
5436 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5437 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
5438 (xtensa_expand_conditional_branch): Likewise.
5439 (xtensa_copy_incoming_a7): Likewise.
5440 (xtensa_output_literal): Likewise.
5441 * dfp.c (decimal_real_maxval): Likewise.
5442 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
5444 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5445 Alan Hayward <alan.hayward@arm.com>
5446 David Sherwood <david.sherwood@arm.com>
5448 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
5449 (mode_nunits_inline): Likewise.
5450 (mode_inner_inline): Likewise.
5451 (mode_unit_size_inline): Likewise.
5452 (mode_unit_precision_inline): Likewise.
5453 (emit_insn_modes_h): Likewise. Also emit a #define of the
5455 (emit_mode_wider): Add an E_ prefix to mode names.
5456 (emit_mode_complex): Likewise.
5457 (emit_mode_inner): Likewise.
5458 (emit_mode_adjustments): Likewise.
5459 (emit_mode_int_n): Likewise.
5460 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
5461 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
5462 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
5463 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
5464 (CRC32_BUILTIN, ENTRY): Likewise.
5465 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
5466 (aarch64_pop_regs): Likewise.
5467 (aarch64_process_components): Likewise.
5468 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
5469 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
5470 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
5471 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
5472 * config/arm/arm.c (arm_init_libfuncs): Likewise.
5473 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
5475 * config/i386/i386-builtin.def (pcmpestr): Likewise.
5476 (pcmpistr): Likewise.
5477 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
5478 * config/mmix/mmix.c (mmix_output_condition): Likewise.
5479 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
5481 * config/rl78/rl78.c (mduc_regs): Likewise.
5482 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
5483 (htm_expand_builtin): Likewise.
5484 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
5485 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
5486 * config/xtensa/xtensa.c (print_operand): Likewise.
5487 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
5488 (NUM_MODE_VECTOR_INT): Likewise.
5489 * genoutput.c (null_operand): Likewise.
5490 (output_operand_data): Likewise.
5491 * genrecog.c (print_parameter_value): Likewise.
5492 * lra.c (debug_operand_data): Likewise.
5494 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5496 * dwarf2out.c (scompare_loc_descriptor_wide)
5497 (scompare_loc_descriptor_narrow): New functions, split out from...
5498 (scompare_loc_descriptor): ...here.
5499 * expmed.c (emit_store_flag_int): New function, split out from...
5500 (emit_store_flag): ...here.
5502 2017-08-30 Richard Biener <rguenther@suse.de>
5504 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
5505 (dwarf2out_early_finish): Move setting of AT_pubnames from
5506 early debug output to early finish.
5508 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5510 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
5511 and -mdata-region to the assembler.
5513 2017-08-30 Richard Biener <rguenther@suse.de>
5515 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
5517 (gen_subprogram_die): Add DW_AT_object_pointer only early.
5518 (dwarf2out_early_global_decl): Only generate a DIE for the
5519 abstract origin if it doesn't already exist or is a declaration DIE.
5520 (resolve_addr): Do not add the linkage name twice when
5521 generating a stub DIE for the DW_TAG_GNU_call_site target.
5523 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5525 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5526 Use machine_mode rather than int for arg1_mode.
5528 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
5531 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
5532 that the second argument of the built-in functions to unpack
5533 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
5534 switch statement instead a lot of if statements.
5535 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
5536 Allow 64-bit values to be in Altivec registers as well as
5537 traditional floating point registers.
5538 (pack<mode>, FMOVE128_VSX iterator): Likewise.
5540 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
5542 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
5543 MAX_REGS_PER_ADDRESS == 1.
5545 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
5547 * config/i386/i386.opt (flag_fentry): Do not init to -1.
5548 * config/i386/i386.c (ix86_option_override_internal): Simplify
5549 setting of opts->x_flag_entry.
5551 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5552 Jakub Jelinek <jakub@redhat.com>
5553 Richard Biener <rguenther@suse.de>
5555 PR tree-optimization/81503
5556 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
5557 folded constant fits in the target type; reorder tests for clarity.
5559 2017-08-29 Martin Liska <mliska@suse.cz>
5561 * passes.def: Include pass_lower_switch.
5562 * stmt.c (dump_case_nodes): Remove and move to
5563 tree-switch-conversion.
5564 (case_values_threshold): Likewise.
5565 (expand_switch_as_decision_tree_p): Likewise.
5566 (emit_case_decision_tree): Likewise.
5567 (expand_case): Likewise.
5568 (balance_case_nodes): Likewise.
5569 (node_has_low_bound): Likewise.
5570 (node_has_high_bound): Likewise.
5571 (node_is_bounded): Likewise.
5572 (emit_case_nodes): Likewise.
5573 (struct simple_case_node): New struct.
5574 (add_case_node): Remove.
5575 (emit_case_dispatch_table): Use vector instead of case_list.
5576 (reset_out_edges_aux): Remove.
5577 (compute_cases_per_edge): Likewise.
5578 (expand_case): Build list of simple_case_node.
5579 (expand_sjlj_dispatch_table): Use it.
5580 * tree-switch-conversion.c (struct case_node): Moved from
5581 stmt.c and adjusted.
5582 (emit_case_nodes): Likewise.
5583 (node_has_low_bound): Likewise.
5584 (node_has_high_bound): Likewise.
5585 (node_is_bounded): Likewise.
5586 (case_values_threshold): Likewise.
5587 (reset_out_edges_aux): Likewise.
5588 (compute_cases_per_edge): Likewise.
5589 (add_case_node): Likewise.
5590 (dump_case_nodes): Likewise.
5591 (balance_case_nodes): Likewise.
5592 (expand_switch_as_decision_tree_p): Likewise.
5593 (emit_jump): Likewise.
5594 (emit_case_decision_tree): Likewise.
5595 (try_switch_expansion): Likewise.
5596 (do_jump_if_equal): Likewise.
5597 (emit_cmp_and_jump_insns): Likewise.
5598 (fix_phi_operands_for_edge): New function.
5599 (record_phi_operand_mapping): Likewise.
5600 (class pass_lower_switch): New pass.
5601 (pass_lower_switch::execute): New function.
5602 (make_pass_lower_switch): Likewise.
5603 (conditional_probability):
5604 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
5605 * tree-pass.h: Add make_pass_lower_switch.
5607 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5610 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
5613 2017-08-29 Richard Biener <rguenther@suse.de>
5615 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
5616 we do not add a DW_AT_inline attribute twice.
5617 (gen_subprogram_die): Remove code setting DW_AT_inline on
5618 DECL_ABSTRACT_P nodes.
5620 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
5622 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
5623 calls to internal functions.
5624 (gimplify_modify_expr): Likewise.
5625 * tree-call-cdce.c (use_internal_fn): Likewise.
5626 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
5627 (convert_to_divmod): Set the nothrow flag.
5628 * tree-if-conv.c (predicate_mem_writes): Likewise.
5629 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
5630 (vectorizable_call): Likewise.
5631 (vectorizable_store): Likewise.
5632 (vectorizable_load): Likewise.
5633 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
5634 (vect_recog_mask_conversion_pattern): Likewise.
5636 2017-08-29 Martin Liska <mliska@suse.cz>
5639 * gcc.c (driver_handle_option): Add new argument.
5640 * opts-common.c (handle_option): Pass
5641 target_option_override_hook.
5642 * opts-global.c (lang_handle_option): Add new option.
5643 (set_default_handlers): Add new argument.
5644 (decode_options): Likewise.
5645 * opts.c (target_handle_option): Likewise.
5646 (common_handle_option): Call target_option_override_hook.
5647 * opts.h (struct cl_option_handler_func): Add hook for
5648 target option override.
5649 (struct cl_option_handlers): Likewise.
5650 (set_default_handlers): Add new argument.
5651 (decode_options): Likewise.
5652 (common_handle_option): Likewise.
5653 (target_handle_option): Likewise.
5654 * toplev.c (toplev::main): Pass targetm.target_option.override
5657 2017-08-29 Richard Biener <rguenther@suse.de>
5658 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5660 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
5661 life to the active subtree.
5663 2017-08-28 Jeff Law <law@redhat.com>
5665 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
5666 derive_equivalences.
5667 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
5669 (edge_info::derive_equivalences): New private member function
5671 * tree-ssa-dom.c (class edge_info): Changed from a struct
5672 to a class. Add ctor/dtor, methods and data members.
5673 (edge_info::edge_info): Renamed from allocate_edge_info.
5674 Initialize additional members.
5675 (edge_info::~edge_info): New.
5676 (free_dom_edge_info): Delete the edge info.
5677 (record_edge_info): Use new class & associated member functions.
5678 Tighten forms for testing for edge equivalences.
5679 (record_temporary_equivalences): Iterate over the simple
5680 equivalences rather than assuming there's only one per edge.
5681 (cprop_into_successor_phis): Iterate over the simple
5682 equivalences rather than assuming there's only one per edge.
5683 (optimize_stmt): Use operand_equal_p rather than pointer
5684 equality for mini-DSE code.
5686 2017-08-28 Nathan Sidwell <nathan@acm.org>
5688 * gcc.c (execute): Fold SIGPIPE handling into switch
5689 statement. Adjust internal error message.
5691 2017-08-28 Richard Biener <rguenther@suse.de>
5694 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
5695 Do nothing for removed DIEs.
5697 2017-08-28 Richard Biener <rguenther@suse.de>
5699 PR tree-optimization/81977
5700 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
5703 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
5706 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
5707 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
5708 using targetm.gen_mem_thread_fence.
5710 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
5713 * config/i386/i386.md (*<btsc><mode>): Change operand 2
5714 predicate to register_operand. Reorder operands.
5715 (*btr<mode>): Ditto.
5716 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
5717 (*btr<mode>_mask): Ditto.
5719 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
5721 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
5722 * config/rs6000/xmmintrin.h: New file.
5723 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
5725 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5728 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
5729 parameter and_insn and return it.
5730 (recombine_lvx_pattern): Insert a copy to ensure availability of
5731 the base register of the copied masking operation at the point of
5732 the instruction replacement.
5733 (recombine_stvx_pattern): Likewise.
5735 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
5737 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
5738 undocumented switches.
5739 (-mpower9-dform-vector): Likewise.
5740 (-mpower9-dform): Likewise.
5741 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
5742 comments to delete references to -mpower9-dform* switches.
5743 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
5744 Delete reference to -mpower9-dform* switches, test for
5745 -mpower9-vector instead.
5746 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
5747 (OTHER_P9_VECTOR_MASKS): Likewise.
5748 (POWERPC_MASKS): Likewise.
5749 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
5750 tests against -mpower9-dform* to -mpower9-vector. Delete code
5751 that checked for -mpower9-dform* consistancy with other options.
5752 Add test for -mpower9-misc to enable other power9 switches.
5753 (rs6000_init_hard_regno_mode_ok): Likewise.
5754 (rs6000_option_override_internal): Likewise.
5755 (rs6000_emit_prologue): Likewise.
5756 (rs6000_emit_epilogue): Likewise.
5757 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
5758 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
5759 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
5761 (emit_fusion_p9_store): Likewise.
5762 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
5763 resetting these macros if the assembler does not support ISA 3.0
5765 (TARGET_P9_DFORM_VECTOR): Likewise.
5766 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
5767 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
5769 2017-08-25 Alan Modra <amodra@gmail.com>
5771 PR rtl-optimization/81747
5772 * cse.c (cse_extended_basic_block): Don't attempt to record
5773 equivalences for degenerate conditional branches that branch
5774 to their fall-through.
5776 2017-08-24 Martin Sebor <msebor@redhat.com>
5779 * gimple-fold.c (size_must_be_zero_p): New function.
5780 (gimple_fold_builtin_memory_op): Call it.
5782 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
5784 * config/rs6000/mm_malloc.h: New file.
5786 2017-08-24 Bin Cheng <bin.cheng@arm.com>
5788 PR tree-optimization/81913
5789 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
5790 analysis when either IVs in condition can wrap.
5792 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
5794 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
5795 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
5797 2017-08-24 Richard Biener <rguenther@suse.de>
5800 * targhooks.c (default_target_can_inline_p): Properly
5801 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
5802 is present and always compare.
5803 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
5804 infer -mfpmath=sse from TARGET_SSE_P.
5805 (ix86_can_inline_p): Properly use target_option_default_node when
5806 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
5808 2017-08-24 Richard Biener <rguenther@suse.de>
5811 * dwarf2out.c (output_die): Handle flag_generate_offload like
5813 (output_comp_unit): Likewise.
5814 (gen_array_type_die): Likewise.
5815 (dwarf2out_early_finish): Likewise.
5816 (note_variable_value_in_expr): Likewise.
5817 (dwarf2out_finish): Likewise. Adjust assert.
5818 * cgraphunit.c (symbol_table::compile): Move setting of
5819 flag_generate_offload earlier ...
5820 (symbol_table::finalize_compilation_unit): ... here, before
5821 early debug finalization.
5823 2017-08-24 Richard Biener <rguenther@suse.de>
5825 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
5826 and ipa-fnsummary.h.
5827 (ix86_can_inline_p): When ix86_fpmath flags do not match
5828 check whether the callee uses FP math at all.
5830 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
5833 * tree-ssanames.c (get_nonzero_bits): Use element_precision
5834 instead of TYPE_PRECISION.
5836 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
5837 Alan Hayward <alan.hayward@arm.com>
5838 David Sherwood <david.sherwood@arm.com>
5840 * combine.c (make_extraction): Use subreg_offset_from_lsb.
5842 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
5844 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
5846 (ix86_frame::stack_realign_allocate): New field.
5847 (struct machine_frame_state): Modify comments.
5848 (machine_frame_state::sp_realigned_fp_end): New field.
5849 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
5851 (sp_valid_at): Add assertion to assure no attempt to access invalid
5852 offset of a realigned stack.
5853 (fp_valid_at): Likewise.
5854 (choose_baseaddr): Modify comments.
5855 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
5856 ix86_expand_prologue.
5857 (ix86_expand_prologue): Modify stack realignment and allocation.
5858 (ix86_expand_epilogue): Modify comments.
5859 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
5860 avx2_runtime, avx512f, and avx512f_runtime.
5862 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
5864 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
5865 (mstackrealign): Do not init to -1.
5866 * config/i386/i386.c (ix86_option_override_internal):
5867 Check opts_set, not opts when setting default value of
5868 opts->x_ix86_force_align_arg_pointer.
5870 2017-08-23 Richard Biener <rguenther@suse.de>
5872 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
5873 lang_hooks.decl_printable_name.
5874 * print-rtl-function.c (print_rtx_function): Likewise.
5875 * tree-pretty-print.c (dump_function_header): Likewise.
5877 2017-08-23 Richard Biener <rguenther@suse.de>
5880 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
5881 -g0 at compile-time.
5883 2017-08-23 Tamar Christina <tamar.christina@arm.com>
5886 * doc/sourcebuild.texi (Other hardware attributes):
5889 2017-08-23 Tamar Christina <tamar.christina@arm.com>
5892 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
5893 Fix single-use check.
5895 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
5897 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
5899 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
5901 * doc/install.texi: Modify to add more details on running selected
5904 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
5906 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
5907 is combined with -mabi=ms.
5908 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
5911 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5913 PR tree-optimization/81488
5914 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
5915 and cached_basis fields.
5916 (MAX_SPREAD): New constant.
5917 (alloc_cand_and_find_basis): Initialize new fields.
5918 (clear_visited): New function.
5919 (create_phi_basis_1): Rename from create_phi_basis, set visited
5920 and cached_basis fields.
5921 (create_phi_basis): New wrapper function.
5922 (phi_add_costs_1): Rename from phi_add_costs, add spread
5923 parameter, set visited field, short-circuit when limits reached.
5924 (phi_add_costs): New wrapper function.
5925 (record_phi_increments_1): Rename from record_phi_increments, set
5927 (record_phi_increments): New wrapper function.
5928 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
5929 (phi_incr_cost): New wrapper function.
5930 (all_phi_incrs_profitable_1): Rename from
5931 all_phi_incrs_profitable, set visited field.
5932 (all_phi_incrs_profitable): New wrapper function.
5934 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
5935 Alan Hayward <alan.hayward@arm.com>
5936 David Sherwood <david.sherwood@arm.com>
5938 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
5939 that takes the outer and inner modes.
5940 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
5941 comparison as the canonical test for a paradoxical subreg.
5942 * combine.c (simplify_set): Use paradoxical_subreg_p.
5943 (make_extraction): Likewise.
5944 (force_to_mode): Likewise.
5945 (rtx_equal_for_field_assignment_p): Likewise.
5946 (gen_lowpart_for_combine): Likewise.
5947 (simplify_comparison): Likewise.
5948 * cse.c (equiv_constant): Likewise.
5949 * expmed.c (store_bit_field_1): Likewise.
5950 * final.c (alter_subreg): Likewise.
5951 * fwprop.c (propagate_rtx): Likewise.
5952 (forward_propagate_subreg): Likewise.
5953 * ira-conflicts.c (ira_build_conflicts): Likewise.
5954 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
5955 * lra-constraints.c (curr_insn_transform): Likewise.
5956 (split_reg): Likewise.
5957 * lra-eliminations.c (move_plus_up): Likewise.
5958 (lra_eliminate_regs_1): Likewise.
5959 * recog.c (general_operand): Likewise.
5960 * ree.c (combine_reaching_defs): Likewise.
5961 * reload.c (push_reload): Likewise.
5962 (find_reloads): Likewise.
5963 * reload1.c (elimination_effects): Likewise.
5964 (compute_reload_subreg_offset): Likewise.
5965 (choose_reload_regs): Likewise.
5966 * rtlanal.c (subreg_lsb_1): Likewise.
5967 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5968 (simplify_subreg): Likewise.
5969 * var-tracking.c (track_loc_p): Likewise.
5970 * emit-rtl.c (byte_lowpart_offset): Likewise.
5971 (paradoxical_subreg_p): Delete out-of-line definition.
5973 2017-08-22 Jeff Law <law@redhat.com>
5975 PR tree-optimization/81741
5976 PR tree-optimization/71947
5977 * tree-ssa-dom.c: Include tree-inline.h.
5978 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
5979 equivalences if one is more expensive to compute than the other.
5980 * tree-ssa-scopedtables.h (class const_or_copies): Make
5981 record_const_or_copy_raw method private.
5982 (class avail_exprs_stack): New method simplify_binary_operation.
5983 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
5984 avail_exprs_stack::simplify_binary_operation as needed.
5985 (avail_exprs_stack::simplify_binary_operation): New function.
5987 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
5989 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
5990 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
5991 (DOT_SYMBOLS): Likewise.
5992 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
5993 (RELOCATABLE_NEEDS_FIXUP): Likewise.
5994 (RS6000_ABI_NAME): Likewise.
5995 (TARGET_CMODEL): Likewise.
5996 (TOC_SECTION_ASM_OP): Likewise.
5997 (SET_CMODEL): New macro.
5998 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
6000 2017-08-22 Richard Biener <rguenther@suse.de>
6002 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
6003 to allow for free-lang-data replacements similar to verify_type_variant.
6005 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
6006 Alan Hayward <alan.hayward@arm.com>
6007 David Sherwood <david.sherwood@arm.com>
6009 * config/aarch64/aarch64.md (casesi): Use DImode rather than
6010 VOIDmode for the LABEL_REF.
6012 2017-08-22 Richard Biener <rguenther@suse.de>
6014 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
6016 2017-08-22 Richard Biener <rguenther@suse.de>
6018 * common.opt (feliminate-dwarf2-dups): Ignore.
6019 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
6020 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
6021 same_die_p_wrap, compute_section_prefix,
6022 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
6023 (comdat_symbol_id, comdat_symbol_number): Likewise.
6024 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
6026 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
6027 (output_die): Mark unreachable path unreachable.
6028 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
6029 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
6030 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
6031 (dwarf2out_early_finish): Likewise.
6033 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
6035 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
6037 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
6040 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
6041 not VAR_P. Filter attribute warnings with OPT_Wattributes.
6042 (avr_attribute_table) <io, io_low, address>: Initialize
6043 .decl_required with true.
6045 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
6047 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
6048 undocumented debugging options.
6049 (-mvsx-scalar-double): Likewise.
6050 (-mallow-df-permute): Likewise.
6051 (-mvectorize-builtins): Likewise.
6052 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
6053 (rs6000_builtin_vectorized_function): Likewise.
6054 (rs6000_builtin_md_vectorized_function): Likewise.
6055 (rs6000_opt_vars): Likewise.
6057 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
6060 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
6061 (*btrq_imm): Rename from *btrq.
6062 (*btcq_imm): Rename from *btcq.
6063 (btsc): New code attribute.
6064 (*<btsc><mode>): New insn pattern.
6065 (*btr<mode>): Ditto.
6066 (*<btsc><mode>_mask): New insn_and_split pattern.
6067 (*btr<mode>_mask): Ditto.
6069 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6070 Alan Hayward <alan.hayward@arm.com>
6071 David Sherwood <david.sherwood@arm.com>
6073 * function.c (pad_below): Simplify padding calculation.
6075 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6076 Alan Hayward <alan.hayward@arm.com>
6077 David Sherwood <david.sherwood@arm.com>
6079 * target.def (function_prologue): Remove frame size argument.
6080 (function_epilogue): Likewise.
6081 * doc/tm.texi: Regenerate.
6082 * final.c (final_start_function): Update call to function_prologue.
6083 (final_end_function): Update call to function_epilogue.
6084 (default_function_pro_epilogue): Remove frame size argument.
6085 * output.h (default_function_pro_epilogue): Likewise.
6086 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
6087 (arm_output_function_prologue): Likewise.
6088 * config/frv/frv.c (frv_function_prologue): Likewise.
6089 (frv_function_epilogue): Likewise.
6090 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
6091 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
6092 (ia64_output_function_epilogue): Likewise.
6093 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
6094 (m32r_output_function_epilogue): Likewise.
6095 * config/microblaze/microblaze.c (microblaze_function_prologue)
6096 (microblaze_function_epilogue): Likewise.
6097 * config/mips/mips.c (mips_output_function_prologue): Likewise.
6098 (mips_output_function_epilogue): Likewise.
6099 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
6100 (mmix_target_asm_function_epilogue): Likewise.
6101 * config/msp430/msp430.c (msp430_start_function): Likewise.
6102 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
6103 (nds32_asm_function_epilogue): Likewise.
6104 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
6105 * config/pa/pa.c (pa_output_function_prologue): Likewise.
6106 (pa_output_function_epilogue): Likewise.
6107 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
6108 (rs6000_output_function_epilogue): Likewise.
6109 * config/rl78/rl78.c (rl78_start_function): Likewise.
6110 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
6111 (rs6000_output_function_epilogue): Likewise.
6112 * config/rx/rx.c (rx_output_function_prologue): Likewise.
6113 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
6114 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
6115 (sparc_asm_function_epilogue): Likewise.
6117 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6119 * tree.h (type_has_mode_precision_p): New function.
6120 * convert.c (convert_to_integer_1): Use it.
6121 * expr.c (expand_expr_real_2): Likewise.
6122 (expand_expr_real_1): Likewise.
6123 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
6124 * match.pd: Likewise.
6125 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
6126 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
6127 * tree-tailcall.c (process_assignment): Likewise.
6128 * tree-vect-loop.c (vectorizable_reduction): Likewise.
6129 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
6130 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
6131 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6132 (vectorizable_assignment): Likewise.
6133 (vectorizable_shift): Likewise.
6134 (vectorizable_operation): Likewise.
6135 * tree-vrp.c (register_edge_assert_for_2): Likewise.
6137 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
6139 * match.pd: Add pow (C, x) simplification.
6141 2017-08-21 Richard Biener <rguenther@suse.de>
6143 PR tree-optimization/81900
6144 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
6145 for blocks with abnormal predecessors.
6146 (compute_antic): Do not set visited flag prematurely.
6148 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
6151 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
6153 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6155 * stor-layout.h (vector_type_mode): Move to...
6156 * tree.h (vector_type_mode): ...here.
6157 * stor-layout.c (vector_type_mode): Move to...
6158 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
6160 2017-08-21 Richard Biener <rguenther@suse.de>
6162 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
6163 register_external_die hooks.
6164 (debug_false_tree_charstarstar_uhwistar): Declare.
6165 (debug_nothing_tree_charstar_uhwi): Likewise.
6166 * debug.c (do_nothing_debug_hooks): Adjust.
6167 (debug_false_tree_charstarstar_uhwistar): New do nothing.
6168 (debug_nothing_tree_charstar_uhwi): Likewise.
6169 * dbxout.c (dbx_debug_hooks): Adjust.
6170 (xcoff_debug_hooks): Likewise.
6171 * sdbout.c (sdb_debug_hooks): Likewise.
6172 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
6173 * dwarf2out.c (macinfo_label_base): New global.
6174 (dwarf2out_register_external_die): New function for the
6175 register_external_die hook.
6176 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
6177 (dwarf2_debug_hooks): Use them.
6178 (dwarf2_lineno_debug_hooks): Adjust.
6179 (struct die_struct): Add with_offset flag.
6180 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
6181 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
6182 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
6183 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
6184 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
6185 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
6186 defining section names for the early LTO debug variants.
6187 (reset_indirect_string): New helper.
6188 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
6189 (print_dw_val): Add support for offsetted symbol references.
6190 (get_ultimate_context): Split out from is_cxx.
6191 (is_cxx): Use get_ultimate_context.
6192 (is_fortran): Add decl overload.
6193 (compute_comp_unit_symbol): Split out worker from
6194 compute_section_prefix.
6195 (compute_section_prefix): Call compute_comp_unit_symbol and
6196 set comdat_type_p here.
6197 (output_die): Skip DIE symbol output for the LTO added one.
6198 Handle DIE symbol references with offset.
6199 (output_comp_unit): Guard section name mangling properly.
6200 For LTO debug sections emit a symbol at the section beginning
6201 which we use to refer to its DIEs.
6202 (add_abstract_origin_attribute): For DIEs registered via
6203 dwarf2out_register_external_die directly refer to the early
6204 DIE rather than indirectly through the shadow one we created.
6205 Remove obsolete call to dwarf2out_abstract_function for
6206 non-function/block origins.
6207 (gen_array_type_die): When generating early LTO debug do
6208 not emit DW_AT_string_length.
6209 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
6210 late when in LTO. As suggested place a gcc_unreachable for
6211 the DECL_ABSTRACT_P case.
6212 (gen_subprogram_die): Avoid another specification DIE
6213 for early built declarations/definitions for the late LTO case.
6214 (gen_variable_die): Add type references for late duplicated VLA dies
6216 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
6217 we have the abstract instance already.
6218 (process_scope_var): Adjust decl DIE contexts in LTO which
6219 first puts them in limbo.
6220 (gen_decl_die): Do not generate type DIEs late apart from
6221 types for VLAs or for decls we do not yet have a DIE. Do not
6222 call dwarf2out_abstract_function late.
6223 (dwarf2out_early_global_decl): Make sure to create DIEs
6224 for abstract instances of a decl first.
6225 (dwarf2out_late_global_decl): Adjust comment.
6226 (output_macinfo_op): With multiple macro sections use
6227 macinfo_label_base to distinguish labels.
6228 (output_macinfo): Likewise. Update macinfo_label_base.
6229 Pass in the line info label.
6230 (note_variable_value_in_expr): When generating LTO resolve
6231 all variable values here by generating DIEs as needed.
6232 (init_sections_and_labels): Add early LTO debug flag parameter
6233 and generate different sections and names if set. Add generation
6234 counter for the labels so we can have multiple of them.
6235 (reset_dies): Helper to allow DIEs to be output multiple times.
6236 (dwarf2out_finish): When outputting DIEs to the fat part of an
6237 LTO object first reset DIEs.
6238 (dwarf2out_early_finish): Output early DIEs when generating LTO.
6239 (modified_type_die): Check for decl_ultimate_origin being self
6241 (gen_type_die_with_usage): Likewise.
6242 (gen_typedef_die): Allow decl_ultimate_origin being self.
6243 (set_decl_abstract_flags): Remove.
6244 (set_block_abstract_flags): Likewise.
6245 (dwarf2out_abstract_function): Treat the early generated DIEs
6246 as the abstract copy and only add DW_AT_inline and
6247 DW_AT_artificial here and call set_decl_origin_self.
6248 If the DIE has an abstract origin don't do anything.
6249 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
6250 if we have none yet (Go fails to build one, PR78628).
6251 (variably_modified_type_p): Prevent endless recursion for Ada
6252 cyclic pointer types.
6253 * lto-streamer-in.c: Include debug.h.
6254 (dref_queue): New global.
6255 (lto_read_tree_1): Stream in DIE references.
6256 (lto_input_tree): Register DIE references.
6257 (input_function): Stream DECL_DEBUG_ARGS.
6258 * lto-streamer-out.c: Include debug.h.
6259 (lto_write_tree_1): Output DIE references.
6260 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
6261 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
6262 (output_function): Stream DECL_DEBUG_ARGS.
6263 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
6264 Stream DECL_ABSTRACT_ORIGIN.
6265 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
6266 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
6267 DECL_CONTEXT for file-scope decls.
6268 * lto-streamer.h (struct dref_entry): Declare.
6269 (dref_queue): Likewise.
6270 * cfgexpand.c (pass_expand::execute): Do not call the
6271 outlining_inline_function hook here.
6272 * lto-wrapper.c (debug_obj): New global.
6273 (tool_cleanup): Unlink it if required.
6274 (debug_objcopy): New function.
6275 (run_gcc): Handle early debug sections in the IL files by
6276 extracting them to separate files, partially linkin them and
6277 feeding the result back as result to the linker.
6278 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
6279 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
6280 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
6281 sections into a separate segment.
6282 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
6284 (darwin_asm_dwarf_section): Likewise.
6285 (darwin_asm_output_dwarf_offset): Likewise.
6286 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
6288 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6289 Alan Hayward <alan.hayward@arm.com>
6290 David Sherwood <david.sherwood@arm.com>
6292 * read-md.h (md_reader::record_potential_iterator_use): Replace
6293 pointer argument with an rtx and an index.
6294 * read-rtl.c (iterator_group::apply_iterator): Likewise.
6295 (apply_mode_iterator): Likewise.
6296 (apply_code_iterator): Likewise.
6297 (apply_int_iterator): Likewise.
6298 (apply_subst_iterator): Likewise.
6299 (record_iterator_use): Likewise.
6300 (record_attribute_use): Likewise.
6301 (md_reader::record_potential_iterator_use): Likewise. Update calls
6302 to record_iterator_use and apply_iterator.
6303 (iterator_use): Replace ptr with x and index.
6304 (attribute_use): Likewise.
6305 (apply_attribute_uses): Update calls to apply_iterator.
6306 (apply_iterators): Likewise. Update initialization of iterator_use.
6307 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
6308 and record_potential_iterator_use.
6309 (rtx_reader::read_rtx_operand): Likewise.
6311 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6312 Alan Hayward <alan.hayward@arm.com>
6313 David Sherwood <david.sherwood@arm.com>
6315 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
6318 2017-08-21 Richard Biener <rguenther@suse.de>
6321 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
6322 at struct end conservatively when comparing common bases.
6324 2017-08-21 Richard Biener <rguenther@suse.de>
6326 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
6327 (mem_ref_in_stmt): Remove.
6328 (determine_max_movement): Use ref index to get at the reference.
6329 (invariantness_dom_walker::before_dom_children): Deal with
6330 lim data already initialized.
6331 (gather_mem_refs_stmt): Initialize lim data and record ref index.
6333 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
6335 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
6336 (TARGET_ISA_ROUND): Ditto.
6337 (TARGET_ROUND): Ditto.
6338 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
6339 * config/i386/i386.md: Ditto.
6340 * config/i386/sse.md: Ditto.
6341 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
6342 with OPTION_MASK_ISA_SSE4_1.
6344 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
6347 * doc/extend.texi (x86 Built-in Functions): Correct the name of
6348 __builtin_ia32_lzcnt_u16.
6350 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
6353 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
6354 (rs6000_set_current_function): Rewrite function to use it.
6356 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
6359 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
6360 and TYPE_WARN_IF_NOT_ALIGN.
6361 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
6362 (handle_warn_if_not_align): New.
6363 (place_union_field): Call handle_warn_if_not_align.
6364 (place_field): Call handle_warn_if_not_align.
6365 Copy TYPE_WARN_IF_NOT_ALIGN.
6366 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
6367 (layout_type): Likewise.
6368 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
6370 (tree_decl_common): Add warn_if_not_align.
6371 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
6372 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
6373 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
6374 (DECL_WARN_IF_NOT_ALIGN): Likewise.
6375 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
6376 * doc/extend.texi: Document warn_if_not_aligned attribute.
6377 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
6379 2017-08-17 Martin Liska <mliska@suse.cz>
6382 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
6383 (get_data_dependence): Use it as pointer type.
6384 (distribute_loop): Likewise.
6386 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6388 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
6389 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
6390 (p8_vmrgow_<mode>_direct): New define_insn.
6391 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
6392 handle endianness for vmrgew and vmrgow permute patterns.
6394 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
6396 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
6397 * config/rs6000/rs6000-cpus.def: Remove comment.
6398 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
6399 (POWERPC_MASKS): Likewise.
6400 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
6401 use of TARGET_VSX_TIMODE.
6402 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
6403 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
6404 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
6405 (rs6000_option_override_internal): Remove dead code.
6406 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
6407 (rs6000_legitimize_reload_address): Likewise.
6408 (rs6000_legitimate_address_p): Likewise.
6409 (rs6000_opt_masks): Delete "vsx-timode".
6410 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
6411 from function comment.
6412 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
6413 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
6414 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
6416 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
6417 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
6418 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
6421 (define_peephole2): Likewise.
6423 2017-08-17 Martin Sebor <msebor@redhat.com>
6426 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
6427 past the end of an array.
6428 (test_pp_format): Add test cases.
6430 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
6432 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
6433 missing ECF_NOTHROW flags.
6435 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
6438 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
6439 operands residing in integer registers.
6440 (*vsx_le_perm_load_<mode>): Likewise.
6441 (*vsx_le_perm_store_<mode>): Likewise.
6442 (define_peephole2): Add peepholes to optimize the above.
6444 2017-08-17 Marek Polacek <polacek@redhat.com>
6447 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
6448 to mimic what shorten_compare did. Change the return type to bool.
6449 (fold_cond_expr_with_comparison): Update call to
6450 operand_equal_for_comparison_p.
6451 (fold_ternary_loc): Likewise.
6453 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
6455 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
6457 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
6459 2017-08-17 Richard Biener <rguenther@suse.de>
6461 * tree-ssa-structalias.c (solve_graph): When propagating
6462 to successors update the graphs succ edges and avoid duplicate work.
6464 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
6467 * config/i386/i386.c (ix86_option_override_internal): Save target
6468 specific options after ix86_stack_protector_guard_reg was changed.
6470 2017-08-17 Richard Biener <rguenther@suse.de>
6472 PR tree-optimization/81827
6473 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
6474 (new_var_info): Initialize it conservatively.
6475 (get_call_vi): Mark register vars.
6476 (new_scalar_tmp_constraint_exp): Likewise.
6477 (handle_rhs_call): Likewise.
6478 (handle_const_call): Likewise.
6479 (create_function_info_for): Likewise.
6480 (solve_constraints): Sort varinfos to separate register from
6481 non-register vars to pack points-to solution bitmaps during
6484 2017-08-17 Marek Polacek <polacek@redhat.com>
6486 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
6488 2017-08-17 Richard Biener <rguenther@suse.de>
6490 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
6491 to true when overflow is undefined and we saturated the result.
6493 2017-08-17 Alan Modra <amodra@gmail.com>
6496 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
6497 Don't use store multiple if only one reg needs saving.
6498 (interesting_frame_related_regno): New function.
6499 (rs6000_frame_related): Don't emit frame info for regs that
6501 (rs6000_emit_epilogue): Likewise.
6503 2017-08-16 Nathan Sidwell <nathan@acm.org>
6505 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
6506 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
6507 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
6508 * tree.c (free_lang_data_in_type): Use else-if chain. Always
6509 clear TYPE_LANG_1. Remove obsolete member-function stripping.
6510 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
6511 (verify_type): Adjust for TYPE_BINFO move.
6512 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
6513 process TYPE_BINFO directly.
6514 (hash_tree): Likewise.
6515 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
6517 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
6520 2017-08-16 David Malcolm <dmalcolm@redhat.com>
6522 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
6524 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
6527 * config/i386/i386.md (*anddi_1_btr): Change predicates of
6528 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
6529 Add ix86_binary_operator_ok to insn constraint.
6530 (*iordi_1_bts): Ditto.
6531 (*xordi_1_btc): Ditto.
6532 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
6533 Update corresponding peephole2 pattern.
6537 2017-08-16 Bin Cheng <bin.cheng@arm.com>
6539 PR tree-optimization/81832
6540 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
6541 copy loop header which has IFN_LOOP_DIST_ALIAS call.
6543 2017-08-16 Marek Polacek <polacek@redhat.com>
6546 * fold-const.c (fold_indirect_ref_1): Restore original behavior
6547 regarding size_zero_node.
6549 2017-08-16 Martin Liska <mliska@suse.cz>
6552 * config.gcc: Respect previously set extra_objs in case
6555 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
6557 PR tree-optimization/81835
6558 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
6559 the phi SSA_NAME. Check that the condition in a COND_EXPR does
6560 not depend on the phi.
6562 2017-08-16 Alan Modra <amodra@gmail.com>
6564 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
6567 2017-08-16 Alan Modra <amodra@gmail.com>
6569 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
6570 (save_reg_p): ..into this. Update all callers.
6571 (first_reg_to_save): Simplify.
6573 2017-08-16 Alan Modra <amodra@gmail.com>
6575 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
6578 2017-08-15 Joseph Myers <joseph@codesourcery.com>
6582 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
6583 constant count if that count is less than 32.
6585 2017-08-15 Nathan Sidwell <nathan@acm.org>
6587 * gcc.c (execute): Emit friendlier message if inferior is killed
6588 by an external cause.
6590 2017-08-15 Richard Biener <rguenther@suse.de>
6592 PR tree-optimization/81790
6593 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
6594 CONSTRUCTORs from simplifying and VN.
6596 2017-08-14 Martin Sebor <msebor@redhat.com>
6598 * builtin-attrs.def: Add comments.
6600 2017-08-14 Martin Sebor <msebor@redhat.com>
6603 * doc/extend.texi (attribute nonstring): Document new attribute.
6605 2017-08-14 Martin Sebor <msebor@redhat.com>
6608 * tree-diagnostic.c (default_tree_printer): Handle %G.
6609 * gimple-pretty-print.h (percent_G_format): Declare new function.
6610 * gimple-pretty-print.c (percent_G_format): Define.
6611 * tree-pretty-print.c (percent_K_format): Add argument.
6613 2017-08-14 Martin Sebor <msebor@redhat.com>
6615 PR translation/79998
6616 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
6617 Remove a stray space.
6619 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
6622 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
6623 (*iordi_1_bts): Ditto.
6624 (*xordi_1_btc): Ditto.
6626 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6629 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
6630 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6632 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
6633 quoted strings, and make more translator-friendly.
6634 (darwin_rs6000_override_options): Likewise.
6635 (rs6000_option_override_internal): Likewise.
6636 (rs6000_return_in_memory): Fix overlong line.
6637 (init_cmulative_args): Use quoted strings, and make more
6638 translator-friendly.
6639 (rs6000_pass_by_reference): Fix overlong line.
6640 (def_builtin): Use quoted strings.
6641 (altivec_expand_predicate_builtin): Use quoted strings, and make
6642 more translator-friendly.
6643 (htm_expand_builtin): Use quoted strings.
6644 (cpu_expand_builtin): Use quoted strings, and make more
6645 translator-friendly.
6646 (altivec_expand_builtin): Likewise.
6647 (paired_expand_predicate_builtin): Likewise.
6648 (rs6000_invalid_builtin): Likewise.
6649 (builtin_function_type): Use quoted strings.
6650 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
6651 more translator-friendly.
6652 (rs6000_trampoline_init): Likewise.
6653 (rs6000_handle_altivec_attribute): Likewise.
6654 (rs6000_inner_target_options): Use quoted strings.
6655 (rs6000_disable_incompatible_switches): Likewise.
6656 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
6657 strings, and make more translator-friendly.
6658 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
6660 2017-08-14 Bin Cheng <bin.cheng@arm.com>
6662 PR tree-optimization/81799
6663 * tree-loop-distribution.c (version_loop_by_alias_check): Force
6664 cond_expr to simple gimple operand.
6666 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
6669 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
6671 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
6675 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
6676 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
6677 TARGET_GASISR_PROLOGUES.
6678 * config/avr/avr.c (avr_option_override): Same.
6679 (avr_pass_pre_proep::execute): Same.
6681 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
6684 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
6685 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
6687 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
6689 * config/i386/i386.md (*load_tp_<mode>): Redefine as
6690 define_insn_and_split. Split to a memory load from 0 in
6691 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
6692 using PTR mode iterator.
6693 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
6694 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
6695 (*add_tp_<mode>): Redefine as define_insn_and_split.
6696 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
6697 address space. Merge with *add_tp_x32 using PTR mode iterator.
6698 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
6699 Split to an add with a memory load from 0 in
6700 DEFAULT_TLS_SEG_REG address space.
6702 2017-08-12 Andrew Pinski <apinski@cavium.com>
6704 * config/aarch64/aarch64-option-extensions.def (rdma):
6705 Fix feature string to what Linux prints out in /proc/cpuinfo.
6707 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
6710 * dwarf2out.c (modified_type_die): For C typedef types that have
6711 an ultimate origin, process the ultimate origin instead of the
6713 (gen_typedef_die): Assert that input DECLs have no ultimate
6715 (gen_type_die_with_usage): For typedef variants that have an
6716 ultimate origin, just call gen_decl_die on the original DECL.
6717 (process_scope_var): Avoid creating DIEs for local typedefs and
6718 concrete static variables.
6720 2017-08-12 Alan Modra <amodra@gmail.com>
6724 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
6725 match gnu-user.h startfile.
6726 (ENDFILE_LINUX_SPEC): Similarly.
6728 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
6731 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
6733 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
6735 2017-08-11 Tamar Christina <tamar.christina@arm.com>
6736 * config/aarch64/aarch64.md (mov<mode>): Change.
6737 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
6738 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
6739 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
6741 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
6743 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
6744 for storage order barriers.
6746 2017-08-11 Martin Liska <mliska@suse.cz>
6749 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
6750 variables of void type.
6752 2017-08-11 Martin Liska <mliska@suse.cz>
6754 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
6755 TARGET_SUPPORTS_ALIASES.
6756 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
6757 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
6758 (optimize_weakref): Likewise.
6759 * symtab.c (symtab_node::noninterposable_alias): Likewise.
6760 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
6761 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
6763 2017-08-11 Martin Liska <mliska@suse.cz>
6766 * config/i386/i386.c (make_resolver_func): Do complete
6767 refactoring of the function.
6769 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
6772 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
6773 * config/i386/i386.c (ix86_stack_protect_guard): Use
6774 ix86_stack_protect_guard_symbol_str to generate varible declaration.
6775 * doc/invoke.texi (x86 Options): Document
6776 -mstack-protector-guard-symbol= option.
6778 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
6780 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
6781 * config/i386/i386.c (ix86_split_stack_guard): New function.
6782 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
6783 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
6784 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
6785 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
6786 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
6787 (split_stack_space_check): Call ix86_split_stack_guard.
6789 2017-08-10 Martin Sebor <msebor@redhat.com>
6791 * print-tree.c (print_node): Print location using the established
6793 Replace spaces with colons.
6794 (debug_raw, debug): Ditto.
6796 2017-08-10 Martin Sebor <msebor@redhat.com>
6799 * pretty-print.c (pp_format): Correct the handling of %s precision.
6801 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
6804 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
6806 (ix86_finalize_stack_frame_flags): This. Also clear
6807 frame_pointer_needed if -fno-omit-frame-pointer is used without
6809 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
6810 with ix86_finalize_stack_frame_flags.
6811 (ix86_expand_epilogue): Likewise.
6812 (ix86_expand_split_stack_prologue): Likewise.
6813 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
6815 2017-08-10 Martin Liska <mliska@suse.cz>
6818 * c-attribs.c (handle_target_attribute):
6819 Report warning for an empty string argument of target attribute.
6821 2017-08-09 Jakub Jelinek <jakub@redhat.com>
6824 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
6826 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
6827 or DECL_NONLOCAL labels.
6828 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
6829 or DECL_NONLOCAL labels here.
6831 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
6833 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
6834 to indicate when early gimple folding has been disabled.
6835 (rs6000_gimple_fold_builtin): Add debug content.
6836 (rs6000_invalid_builtin): Fix whitespace.
6837 (rs6000_expand_builtin): Fix whitespace.
6838 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
6840 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
6843 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
6844 SAVE_MULTIPLE if not all the registers that saves, should be saved.
6846 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
6848 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
6849 (qdf24xx): Likewise.
6850 * config/aarch64/aarch64.md: Include falkor.md.
6851 * config/aarch64/falkor.md: New.
6853 2017-08-09 Marek Polacek <polacek@redhat.com>
6856 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
6857 * diagnostic.c (emit_diagnostic): Add a comment.
6858 (emit_diagnostic_valist): New function.
6860 2017-08-09 Marek Polacek <polacek@redhat.com>
6863 * input.c (make_location): New overload.
6864 * input.h (make_location): Declare.
6866 2017-08-08 Alan Modra <amodra@gmail.com>
6867 H.J. Lu <hongjiu.lu@intel.com>
6870 * gcc.c (NO_PIE_SPEC): Delete.
6871 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
6873 (LINK_PIE_SPEC): ..to here.
6874 (LINK_COMMAND_SPEC): Support -no-pie.
6875 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
6876 chain of crtbegin*.o selection, update for PIE_SPEC changes and
6878 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
6879 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
6880 (ENDFILE_CRTEND_SPEC): Similarly.
6882 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
6885 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
6886 (mstack-protector-guard-offset=): Ditto.
6887 * config/i386/i386.c (ix86_option_override): Handle
6888 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
6890 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
6891 ix86_stack_protect_guard_offset variables.
6892 (TARGET_STACK_PROTECT_GUARD): Always define.
6893 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
6894 and -mstack-protector-guard-offset= options.
6896 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6898 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
6899 boundary case for the last candidate.
6901 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6903 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
6904 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
6906 2017-08-08 Tamar Christina <tamar.christina@arm.com>
6909 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
6910 * config/aarch64/aarch64-builtins.c
6911 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
6912 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
6913 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
6915 2017-08-08 Tamar Christina <tamar.christina@arm.com>
6916 Andrew Pinski <pinskia@gmail.com>
6919 * internal-fn.def (XORSIGN): New.
6920 * optabs.def (xorsign_optab): New.
6921 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
6922 (convert_expand_mult_copysign): New.
6923 (pass_optimize_widening_mul::execute): Call
6924 convert_expand_mult_copysign.
6926 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6928 PR tree-optimization/81354
6929 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
6930 Insert on edges rather than explicitly creating landing pads.
6931 (analyze_candidates_and_replace): Commit edge inserts.
6933 2017-08-08 Richard Biener <rguenther@suse.de>
6936 * tree-ssa-loop-niter.c: Include tree-dfa.h.
6937 (expand_simple_operations): Also look through ADDR_EXPRs with
6938 MEM_REF bases treating them as POINTER_PLUS_EXPR.
6940 2017-08-08 Richard Biener <rguenther@suse.de>
6942 PR tree-optimization/81723
6943 * tree-vect-slp.c (struct bst_traits): New hash traits.
6944 (bst_fail): New global.
6945 (vect_build_slp_tree_2): New worker, split out from ...
6946 (vect_build_slp_tree): ... this now wrapping it with using
6947 bst_fail set to cache SLP tree build fails. Properly handle
6949 (vect_analyze_slp_instance): Allocate and free bst_fail.
6951 2017-08-08 Martin Liska <mliska@suse.cz>
6954 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
6955 LABEL_DECLs that can be from a different function.
6957 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6959 PR tree-optimization/81744
6960 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
6961 loop's number of iterations.
6963 2017-08-08 Martin Liska <mliska@suse.cz>
6965 * asan.c: Include header files.
6966 * attribs.c (build_decl_attribute_variant): New function moved
6968 (build_type_attribute_qual_variant): Likewise.
6969 (cmp_attrib_identifiers): Likewise.
6970 (simple_cst_list_equal): Likewise.
6971 (omp_declare_simd_clauses_equal): Likewise.
6972 (attribute_value_equal): Likewise.
6973 (comp_type_attributes): Likewise.
6974 (build_type_attribute_variant): Likewise.
6975 (lookup_ident_attribute): Likewise.
6976 (remove_attribute): Likewise.
6977 (merge_attributes): Likewise.
6978 (merge_type_attributes): Likewise.
6979 (merge_decl_attributes): Likewise.
6980 (merge_dllimport_decl_attributes): Likewise.
6981 (handle_dll_attribute): Likewise.
6982 (attribute_list_equal): Likewise.
6983 (attribute_list_contained): Likewise.
6984 * attribs.h (lookup_attribute): New function moved from tree.[ch].
6985 (lookup_attribute_by_prefix): Likewise.
6986 * bb-reorder.c: Include header files.
6987 * builtins.c: Likewise.
6988 * calls.c: Likewise.
6989 * cfgexpand.c: Likewise.
6990 * cgraph.c: Likewise.
6991 * cgraphunit.c: Likewise.
6992 * convert.c: Likewise.
6993 * dwarf2out.c: Likewise.
6994 * final.c: Likewise.
6995 * fold-const.c: Likewise.
6996 * function.c: Likewise.
6997 * gimple-expr.c: Likewise.
6998 * gimple-fold.c: Likewise.
6999 * gimple-pretty-print.c: Likewise.
7000 * gimple.c: Likewise.
7001 * gimplify.c: Likewise.
7002 * hsa-common.c: Likewise.
7003 * hsa-gen.c: Likewise.
7004 * internal-fn.c: Likewise.
7005 * ipa-chkp.c: Likewise.
7006 * ipa-cp.c: Likewise.
7007 * ipa-devirt.c: Likewise.
7008 * ipa-fnsummary.c: Likewise.
7009 * ipa-inline.c: Likewise.
7010 * ipa-visibility.c: Likewise.
7012 * lto-cgraph.c: Likewise.
7013 * omp-expand.c: Likewise.
7014 * omp-general.c: Likewise.
7015 * omp-low.c: Likewise.
7016 * omp-offload.c: Likewise.
7017 * omp-simd-clone.c: Likewise.
7018 * opts-global.c: Likewise.
7019 * passes.c: Likewise.
7020 * predict.c: Likewise.
7021 * sancov.c: Likewise.
7022 * sanopt.c: Likewise.
7023 * symtab.c: Likewise.
7024 * toplev.c: Likewise.
7025 * trans-mem.c: Likewise.
7026 * tree-chkp.c: Likewise.
7027 * tree-eh.c: Likewise.
7028 * tree-into-ssa.c: Likewise.
7029 * tree-object-size.c: Likewise.
7030 * tree-parloops.c: Likewise.
7031 * tree-profile.c: Likewise.
7032 * tree-ssa-ccp.c: Likewise.
7033 * tree-ssa-live.c: Likewise.
7034 * tree-ssa-loop.c: Likewise.
7035 * tree-ssa-sccvn.c: Likewise.
7036 * tree-ssa-structalias.c: Likewise.
7037 * tree-ssa.c: Likewise.
7038 * tree-streamer-in.c: Likewise.
7039 * tree-vectorizer.c: Likewise.
7040 * tree-vrp.c: Likewise.
7042 * ubsan.c: Likewise.
7043 * varasm.c: Likewise.
7044 * varpool.c: Likewise.
7045 * tree.c: Remove functions moved to attribs.[ch].
7047 * config/aarch64/aarch64.c: Add attrs.h header file.
7048 * config/alpha/alpha.c: Likewise.
7049 * config/arc/arc.c: Likewise.
7050 * config/arm/arm.c: Likewise.
7051 * config/avr/avr.c: Likewise.
7052 * config/bfin/bfin.c: Likewise.
7053 * config/c6x/c6x.c: Likewise.
7054 * config/cr16/cr16.c: Likewise.
7055 * config/cris/cris.c: Likewise.
7056 * config/darwin.c: Likewise.
7057 * config/epiphany/epiphany.c: Likewise.
7058 * config/fr30/fr30.c: Likewise.
7059 * config/frv/frv.c: Likewise.
7060 * config/ft32/ft32.c: Likewise.
7061 * config/h8300/h8300.c: Likewise.
7062 * config/i386/winnt.c: Likewise.
7063 * config/ia64/ia64.c: Likewise.
7064 * config/iq2000/iq2000.c: Likewise.
7065 * config/lm32/lm32.c: Likewise.
7066 * config/m32c/m32c.c: Likewise.
7067 * config/m32r/m32r.c: Likewise.
7068 * config/m68k/m68k.c: Likewise.
7069 * config/mcore/mcore.c: Likewise.
7070 * config/microblaze/microblaze.c: Likewise.
7071 * config/mips/mips.c: Likewise.
7072 * config/mmix/mmix.c: Likewise.
7073 * config/mn10300/mn10300.c: Likewise.
7074 * config/moxie/moxie.c: Likewise.
7075 * config/msp430/msp430.c: Likewise.
7076 * config/nds32/nds32-isr.c: Likewise.
7077 * config/nds32/nds32.c: Likewise.
7078 * config/nios2/nios2.c: Likewise.
7079 * config/nvptx/nvptx.c: Likewise.
7080 * config/pa/pa.c: Likewise.
7081 * config/pdp11/pdp11.c: Likewise.
7082 * config/powerpcspe/powerpcspe.c: Likewise.
7083 * config/riscv/riscv.c: Likewise.
7084 * config/rl78/rl78.c: Likewise.
7085 * config/rx/rx.c: Likewise.
7086 * config/s390/s390.c: Likewise.
7087 * config/sh/sh.c: Likewise.
7088 * config/sol2.c: Likewise.
7089 * config/sparc/sparc.c: Likewise.
7090 * config/spu/spu.c: Likewise.
7091 * config/stormy16/stormy16.c: Likewise.
7092 * config/tilegx/tilegx.c: Likewise.
7093 * config/tilepro/tilepro.c: Likewise.
7094 * config/v850/v850.c: Likewise.
7095 * config/vax/vax.c: Likewise.
7096 * config/visium/visium.c: Likewise.
7097 * config/xtensa/xtensa.c: Likewise.
7099 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
7102 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
7103 constraints since the -mupper-regs-* switches have been
7105 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
7106 into a vector from a double word element that was extracted from
7107 another vector, and eliminate extra XXPERMDI instructions.
7108 (vsx_concat_<mode>_2): Likewise.
7109 (vsx_concat_<mode>_3): Likewise.
7110 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
7111 concat to allow optimizing inserts from previous extracts.
7113 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
7115 * config/i386/i386.c (ix86_stack_protect_guard): Generate
7116 memory reference to a SSP offset in TLS address space.
7117 (ix86_print_operand) <case '@'>: Remove.
7118 (ix86_print_operand_punct_valid_p): Remove '@' code.
7119 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
7121 (stack_tls_protect_set_<mode>): Remove.
7122 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
7123 (stack_tls_protect_test_<mode>): Remove.
7124 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
7126 2017-08-07 Olivier Hainque <hainque@adacore.com>
7129 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
7131 2017-08-07 Douglas Rupp <rupp@adacore.com>
7133 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
7134 variable was referenced as multidir in command.
7136 2017-08-07 Jakub Jelinek <jakub@redhat.com>
7139 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
7142 2017-08-07 Martin Liska <mliska@suse.cz>
7144 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
7145 * config/rl78/rl78.c: Add include of attribs.h.
7146 * config/sh/sh.c: Likewise.
7147 * config/v850/v850.c: Likewise.
7149 2017-08-07 Tom de Vries <tom@codesourcery.com>
7152 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
7154 2017-08-07 Martin Liska <mliska@suse.cz>
7156 * config/mips/mips.c: Include attribs.h.
7158 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
7161 * doc/invoke.texi: Document change in behvaior for -Ofast for
7164 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
7166 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
7168 (aarch64_pop_regs): Likewise.
7169 (aarch64_gen_load_pair): Likewise.
7170 (aarch64_save_callee_saves): Likewise.
7171 (aarch64_restore_callee_saves): Likewise.
7173 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
7175 * config/i386/i386.c: Revert the last change.
7177 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
7180 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
7182 (ix86_finalize_stack_frame_flags): This. Also clear
7183 frame_pointer_needed if -fno-omit-frame-pointer is used without
7185 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
7186 with ix86_finalize_stack_frame_flags.
7187 (ix86_expand_epilogue): Likewise.
7188 (ix86_expand_split_stack_prologue): Likewise.
7190 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
7193 * config/i386/i386.c (get_builtin_code_for_version): Set priority
7194 to P_AES for Westmere.
7196 2017-08-07 Jonathan Yong <10walls@gmail.com>
7198 * config/i386/mingw.opt (fset-stack-executable): Removed.
7199 * config/i386/cygming.opt (fset-stack-executable): Moved
7201 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
7203 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
7205 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
7207 2017-08-07 Marek Polacek <polacek@redhat.com>
7210 * fold-const.c (fold_indirect_ref_1): Check type_domain.
7212 2017-08-07 Martin Liska <mliska@suse.cz>
7214 * attribs.h (canonicalize_attr_name): New function.
7215 (cmp_attribs): Move from c-format.c and adjusted.
7216 (is_attribute_p): Moved from tree.h.
7217 * tree-inline.c: Add new includes.
7218 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
7219 (private_is_attribute_p): Remove.
7220 (private_lookup_attribute): Likewise.
7221 (private_lookup_attribute_by_prefix): Simplify.
7222 (remove_attribute): Use is_attribute_p.
7223 * tree.h: Remove removed declarations.
7225 2017-08-07 Jakub Jelinek <jakub@redhat.com>
7228 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
7229 instead of computing it in the function. Formatting fix.
7230 (expand_case): Don't rely on default_edge being the first edge,
7231 clear it if removing it, pass default_edge to
7232 emit_case_dispatch_table.
7233 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
7236 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
7238 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
7239 insn in the function, emit NOP after the insn.
7241 2017-08-06 Tom de Vries <tom@codesourcery.com>
7243 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
7246 2017-08-06 Tom de Vries <tom@codesourcery.com>
7248 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
7251 2017-08-04 Yury Gribov <tetra2005@gmail.com>
7253 PR tree-optimization/57371
7254 * match.pd: New pattern.
7256 2017-08-04 Marek Polacek <polacek@redhat.com>
7259 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
7260 perform the computation in offset_int.
7262 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7264 PR tree-optimization/81136
7265 * tree-vectorizer.h: Include tree-hash-traits.h.
7266 (vec_base_alignments): New typedef.
7267 (vec_info): Add a base_alignments field.
7268 (vect_record_base_alignments): Declare.
7269 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
7271 (DR_IS_CONDITIONAL_IN_STMT): New macro.
7272 (create_data_ref): Add an is_conditional_in_stmt argument.
7273 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
7274 the is_conditional_in_stmt field.
7275 (data_ref_loc): Add an is_conditional_in_stmt field.
7276 (get_references_in_stmt): Set the is_conditional_in_stmt field.
7277 (find_data_references_in_stmt): Update call to create_data_ref.
7278 (graphite_find_data_references_in_stmt): Likewise.
7279 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
7280 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
7281 (vect_record_base_alignment): New function.
7282 (vect_record_base_alignments): Likewise.
7283 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
7284 for nested statements even if we fail to compute a misalignment.
7285 Use pooled base alignments for unconditional references.
7286 (vect_find_same_alignment_drs): Compare base addresses instead
7288 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
7289 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
7291 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7293 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
7294 Add an explicit name for the enum. Use auto_vec for slp_instances
7296 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
7298 (_bb_vec_info): Add a constructor and destructor.
7299 (vinfo_for_stmt): Return NULL for uids of -1 as well.
7300 (destroy_loop_vec_info): Delete.
7301 (vect_destroy_datarefs): Likewise.
7302 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
7303 (vec_info::vec_info): New function.
7304 (vec_info::~vec_info): Likewise.
7305 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
7306 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
7307 destroy_loop_vec_info.
7308 * tree-vect-loop.c (new_loop_vec_info): Replace with...
7309 (_loop_vec_info::_loop_vec_info): ...this.
7310 (destroy_loop_vec_info): Replace with...
7311 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
7312 the stmt_vec_infos. Leave handling of vec_info information to its
7313 destructor. Remove explicit vector releases.
7314 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
7315 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
7316 * tree-vect-slp.c (new_bb_vec_info): Replace with...
7317 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
7318 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
7319 (destroy_bb_vec_info): Replace with...
7320 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
7321 information to its destructor.
7322 (vect_slp_analyze_bb_1): Use new and delete instead of
7323 new_bb_vec_info and destroy_bb_vec_info.
7324 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
7327 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7329 * tree-data-ref.h (subscript): Add access_fn field.
7330 (data_dependence_relation): Add could_be_independent_p.
7331 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
7332 (same_access_functions): Move to tree-data-ref.c.
7333 * tree-data-ref.c (ref_contains_union_access_p): New function.
7334 (access_fn_component_p): Likewise.
7335 (access_fn_components_comparable_p): Likewise.
7336 (dr_analyze_indices): Add a reference to access_fn_component_p.
7337 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
7339 (constant_access_functions): Likewise.
7340 (add_other_self_distances): Likewise.
7341 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
7342 (initialize_data_dependence_relation): Use XCNEW and remove
7343 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
7344 of access functions that have the same type. Allow the
7345 subsequence to end with different bases in some circumstances.
7346 Record the chosen access functions in SUB_ACCESS_FN.
7347 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
7348 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
7349 (subscript_dependence_tester_1): Likewise dra and drb.
7350 (build_classic_dist_vector): Update calls accordingly.
7351 (subscript_dependence_tester): Likewise.
7352 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
7353 DDR_COULD_BE_INDEPENDENT_P.
7354 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
7355 comp_alias_ddrs instead of may_alias_ddrs.
7356 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
7358 (vect_analyze_data_ref_dependence): Use it if
7359 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
7360 distance vectors if that fails.
7361 (dependence_distance_ge_vf): New function.
7362 (vect_prune_runtime_alias_test_list): Use it. Don't clear
7363 LOOP_VINFO_MAY_ALIAS_DDRS.
7365 2017-08-04 Richard Biener <rguenther@suse.de>
7368 * fold-const.c (fold_binary_loc): Properly restrict
7369 minus_var0 && minus_var1 case when associating undefined overflow
7372 2017-08-04 Tom de Vries <tom@codesourcery.com>
7374 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
7376 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7378 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7379 Don't start diagnostic messages with a capital letter.
7380 * config/rs6000/rs6000.c (rs6000_option_override_internal):
7382 (rs6000_invalid_builtin): Likewise.
7383 (rs6000_trampoline_init): Likewise.
7385 2017-08-03 Jakub Jelinek <jakub@redhat.com>
7388 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
7389 after setting changeable df flags.
7391 2017-08-03 Richard Biener <rguenther@suse.de>
7393 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
7394 up if the use is in USE - X.
7396 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
7398 * toplev.c (dumpfile.h): New include.
7399 (internal_error_reentered): New static function. Use it...
7400 (internal_error_function): ...here to handle reentered internal_error.
7402 2017-08-03 Richard Biener <rguenther@suse.de>
7405 * fold-const.c (split_tree): Add minus_var and minus_con
7406 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
7407 here but always use minus_*.
7408 (associate_trees): Assert we never associate with MINUS_EXPR
7409 and NULL first operand. Do not recurse for PLUS_EXPR operands
7410 when associating as MINUS_EXPR either.
7411 (fold_binary_loc): Track minus_var and minus_con.
7413 2017-08-03 Tom de Vries <tom@codesourcery.com>
7416 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
7417 ACCEL_COMPILER, apply finish_options on
7418 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
7420 2017-08-03 Tom de Vries <tom@codesourcery.com>
7423 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
7424 function_entry_patch_area_size > 0.
7426 2017-08-03 Jakub Jelinek <jakub@redhat.com>
7429 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
7430 instead of 10??LU, perform unit multiplication in wide_int,
7431 don't change alloc_object_size_limit if the limit is larger
7434 PR tree-optimization/81655
7435 PR tree-optimization/81588
7436 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
7437 the case when ranges[i].low and high are 1 for unsigned type with
7441 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
7442 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
7444 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7446 * tree-vrp.h: Add include guard.
7448 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
7451 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
7452 (ud2): New insn pattern.
7453 * config/i386/i386.c (ix86_expand_epilogue):
7454 For naked functions, generate ud2 instead of trap insn.
7456 2017-08-02 Marek Polacek <polacek@redhat.com>
7459 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
7461 2017-08-02 Tom de Vries <tom@codesourcery.com>
7462 Cesar Philippidis <cesar@codesourcery.com>
7464 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
7465 Add missing edge probabilities.
7467 2017-08-02 Tamar Christina <tamar.christina@arm.com>
7469 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
7472 2017-08-02 Jakub Jelinek <jakub@redhat.com>
7475 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
7476 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
7477 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
7478 prologue_seq sequences - if any.
7480 2017-08-02 Richard Biener <rguenther@suse.de>
7482 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
7483 via vectors if supported, integer extracts via punning if supported
7484 or otherwise vector extracts.
7486 2017-08-02 Richard Biener <rguenther@suse.de>
7488 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
7490 (bitmap_insert_into_set): ... this.
7492 2017-08-02 Richard Biener <rguenther@suse.de>
7494 PR tree-optimization/81633
7496 2015-08-17 Alan Hayward <alan.hayward@arm.com>
7498 PR tree-optimization/71752
7499 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
7501 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
7503 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
7504 (machine_function::call_ms2sysv_pad_out): Remove field.
7505 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
7506 (ix86_compute_frame_layout): Likewise.
7508 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
7511 * config/i386/i386.c (ix86_set_func_type): Disallow naked
7512 attribute with interrupt attribute.
7514 2017-08-01 Andrew Pinski <apinski@cavium.com>
7516 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
7517 BIT_INSERT_EXPR's operand 1
7518 to see if the types precision matches.
7520 2017-08-01 Martin Liska <mliska@suse.cz>
7523 * builtins.c (expand_builtin_memcpy_args): Remove.
7524 (expand_builtin_memcpy): Call newly added function
7525 expand_builtin_memory_copy_args.
7526 (expand_builtin_memcpy_with_bounds): Likewise.
7527 (expand_builtin_mempcpy): Remove last argument.
7528 (expand_builtin_mempcpy_with_bounds): Likewise.
7529 (expand_builtin_memory_copy_args): New function created from
7530 expand_builtin_mempcpy_args with small modifications.
7531 (expand_builtin_mempcpy_args): Remove.
7532 (expand_builtin_stpcpy): Remove unused argument.
7533 (expand_builtin): Likewise.
7534 (expand_builtin_with_bounds): Likewise.
7536 2017-08-01 Martin Liska <mliska@suse.cz>
7539 Make mempcpy more optimal (PR middle-end/70140).
7541 2017-08-01 Jakub Jelinek <jakub@redhat.com>
7544 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
7545 __builtin_vec_cmpne verify both arguments are compatible vectors
7546 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
7547 verify arg1_type is a pointer or array type. For __builtin_vec_st,
7548 move computation of aligned to after checking the argument types.
7552 * config/rs6000/vsx.md (vextract_fp_from_shorth,
7553 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
7556 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
7557 Doug Rupp <rupp@adacore.com>
7558 Olivier Hainque <hainque@adacore.com>
7560 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
7561 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
7563 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
7564 for TARGET_OS_CPP_BUILTIN.
7565 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
7566 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
7568 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
7569 passing required abi options to the assembler for EABI configurations.
7570 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
7571 of .text.hot and .text.unlikely sections for kernel modules when
7572 using ARM style exceptions.
7573 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
7574 options. Add EXTRA_CC1_SPEC.
7575 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
7577 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
7579 (ARM_TARGET2_DWARF_FORMAT): Define.
7580 * config/arm/t-vxworks: Adjust multilib control to removal of the
7581 Diab command line options.
7583 2017-08-01 Martin Liska <mliska@suse.cz>
7585 PR gcov-profile/81561
7586 * gcov.c (unblock): Make unblocking safe as we need to preserve
7587 index correspondence of blocks and block_lists.
7589 2017-08-01 Richard Biener <rguenther@suse.de>
7591 PR tree-optimization/81181
7592 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7593 (compute_antic): ... end of iteration here.
7595 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
7597 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
7598 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
7599 (ftree-slp-vectorize): Likewise.
7600 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
7601 can no longer be set independent of flag_tree_loop_vectorize.
7602 * omp-general.c (emp_max_vf): Likewise.
7603 * opts.c (enable_fdo_optimizations): Remove references to
7604 flag_tree_vectorize, these are now implicit.
7605 (common_handle_option): Remove handling for OPT_ftree_vectorize,
7606 and leave it for the options machinery.
7608 2017-08-01 Martin Liska <mliska@suse.cz>
7611 * builtins.c (expand_builtin_memcpy_args): Remove.
7612 (expand_builtin_memcpy): Call newly added function
7613 expand_builtin_memory_copy_args.
7614 (expand_builtin_memcpy_with_bounds): Likewise.
7615 (expand_builtin_mempcpy): Remove last argument.
7616 (expand_builtin_mempcpy_with_bounds): Likewise.
7617 (expand_builtin_memory_copy_args): New function created from
7618 expand_builtin_mempcpy_args with small modifications.
7619 (expand_builtin_mempcpy_args): Remove.
7620 (expand_builtin_stpcpy): Remove unused argument.
7621 (expand_builtin): Likewise.
7622 (expand_builtin_with_bounds): Likewise.
7624 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
7627 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
7628 print "ds:" only for immediates in generic address space.
7630 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
7633 * config/i386/i386.c (ix86_funciton_naked): New prototype.
7634 (ix86_function_ok_for_sibcall): Return false for naked functions.
7636 2017-08-01 Richard Biener <rguenther@suse.de>
7638 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
7639 (compute_antic): Seed worklist with exit block predecessors.
7640 * cfganal.c (dfs_find_deadend): For a cycle return the source
7641 of the edge closing it.
7643 2017-08-01 Tamar Christina <tamar.christina@arm.com>
7645 * config/aarch64/aarch64.c
7646 (aarch64_can_const_movi_rtx_p): Move 0 check.
7648 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7650 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
7651 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
7653 * match.pd: Ditto in address comparison pattern.
7655 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7657 PR tree-optimization/81627
7658 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
7659 closed ssa form for store-store chain.
7661 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7663 PR tree-optimization/81620
7664 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
7665 for store-store chain.
7667 2017-08-01 Jakub Jelinek <jakub@redhat.com>
7669 PR tree-optimization/81588
7670 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
7671 ranges[i].in_p, invert comparison code ccode. For >/>=,
7672 swap rhs1 and rhs2 and comparison code unconditionally,
7673 for </<= don't do that. Don't swap rhs1/rhs2 again if
7674 ranges[i].in_p, instead invert comparison code ccode if
7675 opcode or oe->rank is BIT_IOR_EXPR.
7678 * optabs.def (vec_extract_optab, vec_init_optab): Change from
7679 a direct optab to conversion optab.
7680 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
7681 with GET_MODE_INNER as last argument instead of optab_handler.
7682 * expmed.c (extract_bit_field_1): Likewise. Use vector from
7683 vector extraction if possible and optab is available.
7684 * expr.c (store_constructor): Use convert_optab_handler instead
7685 of optab_handler. Use vector initialization from smaller
7686 vectors if possible and optab is available.
7687 * tree-vect-stmts.c (vectorizable_load): Likewise.
7688 * doc/md.texi (vec_extract, vec_init): Document that the optabs
7690 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
7691 of vec_init from half-sized vectors with the same element mode.
7692 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
7693 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
7694 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
7695 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
7696 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
7697 after mode in gen_vec_extract* calls.
7698 (vec_extract<mode>): Renamed to ...
7699 (vec_extract<mode><ssescalarmodelower>): ... this.
7700 (vec_extract<mode><ssehalfvecmodelower>): New expander.
7701 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
7702 element mode after mode in gen_vec_init* calls.
7703 (VEC_INIT_HALF_MODE): New mode iterator.
7704 (vec_init<mode>): Renamed to ...
7705 (vec_init<mode><ssescalarmodelower>): ... this.
7706 (vec_init<mode><ssehalfvecmodelower>): New expander.
7707 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
7708 (vec_extractv2sfsf): ... this.
7709 (vec_initv2sf): Renamed to ...
7710 (vec_initv2sfsf): ... this.
7711 (vec_extractv2si): Renamed to ...
7712 (vec_extractv2sisi): ... this.
7713 (vec_initv2si): Renamed to ...
7714 (vec_initv2sisi): ... this.
7715 (vec_extractv4hi): Renamed to ...
7716 (vec_extractv4hihi): ... this.
7717 (vec_initv4hi): Renamed to ...
7718 (vec_initv4hihi): ... this.
7719 (vec_extractv8qi): Renamed to ...
7720 (vec_extractv8qiqi): ... this.
7721 (vec_initv8qi): Renamed to ...
7722 (vec_initv8qiqi): ... this.
7723 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
7724 (vec_init<mode>): Renamed to ...
7725 (vec_init<mode><VEC_base_l>): ... this.
7726 (vec_extract<mode>): Renamed to ...
7727 (vec_extract<mode><VEC_base_l>): ... this.
7728 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
7729 (vec_initv2sfsf): ... this.
7730 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
7731 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
7732 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
7733 element mode after mode in gen_vec_init* calls.
7734 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
7735 (vec_init<mode><Vel>): ... this.
7736 (vec_extract<mode>): Renamed to ...
7737 (vec_extract<mode><Vel>): ... this.
7738 * config/aarch64/iterators.md (Vel): New mode attribute.
7739 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
7740 Add element mode after mode in gen_vec_extract* calls.
7741 * config/s390/vector.md (non_vec_l): New mode attribute.
7742 (vec_extract<mode>): Renamed to ...
7743 (vec_extract<mode><non_vec_l>): ... this.
7744 (vec_init<mode>): Renamed to ...
7745 (vec_init<mode><non_vec_l>): ... this.
7746 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
7747 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
7749 * config/arm/iterators.md (V_elem_l): New mode attribute.
7750 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
7751 (vec_extract<mode><V_elem_l>): ... this.
7752 (vec_extractv2di): Renamed to ...
7753 (vec_extractv2didi): ... this.
7754 (vec_init<mode>): Renamed to ...
7755 (vec_init<mode><V_elem_l>): ... this.
7756 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
7757 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
7758 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
7759 Add element mode after gen_vec_extract* calls.
7760 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
7761 (vec_init<mode><unitmode>): ... this.
7762 (vec_extract<mode>): Renamed to ...
7763 (vec_extract<mode><unitmode>): ... this.
7764 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
7765 (vec_init<mode><unitmode>): ... this.
7766 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
7767 (vec_initv2sfsf): ... this.
7768 (vec_extractv2sf): Renamed to ...
7769 (vec_extractv2sfsf): ... this.
7770 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
7771 Add element mode after gen_vec_extract* calls.
7772 * config/mips/mips.md (unitmode): New mode iterator.
7773 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
7774 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
7775 * config/spu/spu.md (inner_l): New mode attribute.
7776 (vec_init<mode>): Renamed to ...
7777 (vec_init<mode><inner_l>): ... this.
7778 (vec_extract<mode>): Renamed to ...
7779 (vec_extract<mode><inner_l>): ... this.
7780 * config/sparc/sparc.md (veltmode): New mode iterator.
7781 (vec_init<VMALL:mode>): Renamed to ...
7782 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
7783 * config/ia64/vect.md (vec_initv2si): Renamed to ...
7784 (vec_initv2sisi): ... this.
7785 (vec_initv2sf): Renamed to ...
7786 (vec_initv2sfsf): ... this.
7787 (vec_extractv2sf): Renamed to ...
7788 (vec_extractv2sfsf): ... this.
7789 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
7790 (vec_init<mode>): Renamed to ...
7791 (vec_init<mode><VEC_base_l>): ... this.
7792 (vec_extract<mode>): Renamed to ...
7793 (vec_extract<mode><VEC_base_l>): ... this.
7794 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
7795 (vec_initv2sfsf): ... this.
7796 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
7797 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
7798 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
7799 gen_vec_init* calls.
7801 2017-08-01 Richard Biener <rguenther@suse.de>
7803 PR tree-optimization/81297
7804 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
7805 TREE_OVERFLOW from INTEGER_CSTs.
7807 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
7809 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
7811 2017-07-31 Carl Love <cel@us.ibm.com>
7813 * config/rs6000/rs6000-c: Add support for built-in functions
7814 vector signed char vec_xl_be (signed long long, signed char *);
7815 vector unsigned char vec_xl_be (signed long long, unsigned char *);
7816 vector signed int vec_xl_be (signed long long, signed int *);
7817 vector unsigned int vec_xl_be (signed long long, unsigned int *);
7818 vector signed long long vec_xl_be (signed long long, signed long long *);
7819 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
7820 vector signed short vec_xl_be (signed long long, signed short *);
7821 vector unsigned short vec_xl_be (signed long long, unsigned short *);
7822 vector double vec_xl_be (signed long long, double *);
7823 vector float vec_xl_be (signed long long, float *);
7824 * config/rs6000/altivec.h (vec_xl_be): Add #define.
7825 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
7826 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
7828 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
7829 (altivec_expand_builtin): Add switch statement to call
7830 altivec_expand_xl_be for each builtin.
7831 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
7832 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
7833 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
7834 __builtin_vsx_le_be_v16qi.
7835 * doc/extend.texi: Update the built-in documentation file for the
7836 new built-in functions.
7838 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
7841 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
7843 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
7845 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7847 * config.gcc: Add z14.
7848 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
7849 CPU model numbers for z13s and z14.
7850 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
7852 * config/s390/s390-opts.h (enum processor_type): Rename
7853 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
7854 * config/s390/s390.c (processor_table): Add field for CPU name to
7855 be passed to Binutils.
7856 (s390_asm_output_machine_for_arch): Use the new field in
7857 processor_table for Binutils.
7858 (s390_expand_builtin): Replace arch12 with z14.
7859 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
7860 (s390_get_sched_attrmask): Likewise.
7861 (s390_get_unit_mask): Likewise.
7862 * config/s390/s390.opt: Add z14 to processor_type enum.
7864 2017-07-31 Martin Jambor <mjambor@suse.cz>
7867 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
7868 regardless of optimization level.
7870 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
7871 Martin Liska <mliska@suse.cz>
7873 * predict.def: Remove old comment and adjust probability.
7874 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
7877 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
7880 * config/i386/i386.c (ix86_function_naked): New function.
7881 (ix86_can_use_return_insn_p): Return false for naked functions.
7882 (ix86_expand_prologue): Skip prologue for naked functions.
7883 (ix86_expand_epilogue): Skip epilogue for naked functions
7884 and emit trap instruction.
7885 (ix86_warn_func_return): New function.
7886 (ix86_attribute_table): Add "naked" attribute specification.
7887 (TARGET_WARN_FUNC_RETURN): Define.
7888 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
7890 2017-07-31 Martin Liska <mliska@suse.cz>
7892 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
7893 (dump_gimple_bb_header): Always dump BB info.
7894 (pp_cfg_jump): Do not append info about BB when dumping a jump.
7896 2017-07-31 Martin Liska <mliska@suse.cz>
7899 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
7900 also with current_function_decl non-null equality.
7902 2017-07-31 Jakub Jelinek <jakub@redhat.com>
7905 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
7906 change type to the element type, instead add eltype variable and
7907 use it where we are interested in the element type.
7909 PR tree-optimization/81603
7910 * ipa-polymorphic-call.c
7911 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
7912 offset arithmetic in offset_int, bail out if the resulting bit offset
7913 doesn't fit into shwi.
7915 2017-07-31 Martin Liska <mliska@suse.cz>
7917 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
7918 (gimplify_save_expr): Fix comment.
7920 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
7923 * config/i386/i386.c (ix86_function_arg): Update arguments for
7925 (ix86_compute_frame_layout): Set the initial stack offset to
7926 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
7927 INCOMING_FRAME_SP_OFFSET.
7928 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
7929 stack before exception handler returns.
7930 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
7931 the 'ERROR_CODE' for exception handler.
7933 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
7935 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
7936 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
7937 (ASM_OUTPUT_REG_POP): Ditto.
7938 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
7939 instead of asm_fprintf to output pure string.
7941 2017-07-29 Jakub Jelinek <jakub@redhat.com>
7943 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
7944 to imported_module_or_decl hook.
7945 (debug_nothing_tree_tree_tree_bool): Remove.
7946 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
7947 * debug.c (do_nothing_debug_hooks): Use
7948 debug_nothing_tree_tree_tree_bool_bool instead of
7949 debug_nothing_tree_tree_tree_bool.
7950 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
7951 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
7952 * sdbout.c (sdb_debug_hooks): Likewise.
7953 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
7954 (gen_namespace_die): Add DW_AT_export_symbols attribute if
7956 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
7957 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
7958 attribute, don't add anything.
7960 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7962 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
7963 (fold_build2_stat_loc): Likewise.
7964 (fold_build3_stat_loc): Likewise.
7965 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
7966 (fold_build1_loc): Remove macro.
7967 (fold_build2_loc): Likewise.
7968 (fold_build3_loc): Likewise.
7970 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7972 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
7973 (gimple_build_debug_bind_source_stat): Likewise.
7974 * gimple.h (gimple_build_debug_bind): Remove macro.
7975 (gimple_build_debug_bind_source): Likewise.
7977 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7979 * bitmap.c (bitmap_alloc): Adjust.
7980 (bitmap_gc_alloc): Likewise.
7981 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
7983 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7985 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
7986 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
7987 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
7988 (bitmap_gc_alloc_stat): Likewise.
7989 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
7991 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7993 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
7994 * rtl.h (shallow_copy_rtx): Remove macro.
7996 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7998 * emit-rtl.c (gen_raw_REG): Adjust.
7999 * gengenrtl.c (gendef): Likewise.
8000 * rtl.c (rtx_alloc_stat): Remove _stat from name.
8001 * rtl.h (rtx_alloc): Remove macro.
8003 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8005 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
8006 (build_tree_list_stat): Likewise.
8007 * tree.h (build_tree_list): Remove macro.
8008 (build_tree_list_vec): Likewise.
8010 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8012 * tree.c (make_vector_stat): Remove _stat from name.
8013 (build_vector_stat): Likewise.
8014 * tree.h (make_vector_stat): Remove macro.
8015 (build_vector_stat): Likewise.
8017 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8019 * tree.h (build_var_debug_value): Remove prototype.
8021 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8023 * tree.c (tree_cons_stat): Remove _stat from name.
8024 * tree.h (tree_cons): Remove macro.
8026 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8028 * tree.c (build_vl_exp_stat): Remove _stat from name.
8029 * tree.h (build_vl_exp): Remove macro.
8031 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8033 * tree.c (build_decl_stat): Remove _stat from name.
8034 * tree.h (build_decl): Remove macro.
8036 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8038 * gimple.c (gimple_build_with_ops_stat): Adjust.
8039 (gimple_alloc_stat): Remove _stat from name.
8040 * gimple.h (gimple_alloc): Remove macro.
8042 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8044 * tree.c (make_tree_vec_stat): Remove _stat from name.
8045 (grow_tree_vec_stat): Likewise.
8046 * tree.h (make_tree_vec_stat): Adjust prototype.
8047 (grow_tree_vec_stat): Likewise.
8048 (make_tree_vec): Remove macro.
8049 (grow_tree_vec): Likewise.
8051 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8053 * fold-const.c (fold_build1_stat_loc): Adjust.
8054 (fold_build2_stat_loc): Likewise.
8055 (fold_build3_stat_loc): Likewise.
8056 * tree.c (build0_stat): Remove _stat from name.
8057 (build1_stat): Likewise.
8058 (build2_stat): Likewise.
8059 (build3_stat): Likewise.
8060 (build4_stat): Likewise.
8061 (build5_stat): Likewise.
8062 * tree.h (build1_loc): Remove macro, and rename _stat function
8064 (build2_loc): Likewise.
8065 (build3_loc): Likewise.
8066 (build4_loc): Likewise.
8067 (build5_loc): Likewise.
8069 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8071 * tree.c (make_int_cst_stat): Remove _stat from name.
8072 * tree.h (make_int_cst_stat): Adjust prototype.
8073 (make_int_cst): Remove macro.
8075 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8077 * tree.c (make_tre_binfo_stat): Remove _stat from name.
8078 * tree.h (make_tree_binfo_stat): Adjust prototype.
8079 (make_tree_binfo): Remove.
8081 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8083 * tree.c (copy_node_stat): Rename to copy_node.
8084 (build_distinct_type_copy): Adjust.
8085 * tree.h (copy_node_stat): Adjust prototype.
8086 (copy_node): Remove macro.
8088 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8090 * tree.c (make_node_stat): rename to make_node.
8091 (build_tree_list_stat): Adjust.
8092 (build0_stat): Likewise.
8093 (build2_stat): Likewise.
8094 (build3_stat): Likewise.
8095 (build4_stat): Likewise.
8096 (build5_stat): Likewise.
8097 (build_decl_stat): Likewise.
8098 * tree.h (make_node_stat): Adjust prototype.
8099 (make_node): remove macro.
8101 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
8103 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
8104 (PPC_FEATURE2_SCV): Likewise.
8105 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
8107 2017-07-28 Tamar Christina <tamar.christina@arm.com>
8109 * config/aarch64/aarch64.c
8110 (aarch64_internal_mov_immediate): Add new special pattern.
8111 * config/aarch64/aarch64.md (*movdi_aarch64):
8112 Add reg/32bit const mov case.
8114 2017-07-28 Tamar Christina <tamar.christina@arm.com>
8115 Richard Sandiford <richard.sandiford@linaro.org>
8117 * config/aarch64/aarch64.md (mov<mode>): Generalize.
8118 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
8119 Add integer and movi cases.
8120 (movi-split-hf-df-sf split, fp16): New.
8121 (enabled): Added TARGET_FP_F16INST.
8122 * config/aarch64/iterators.md (GPF_HF): New.
8123 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
8125 2017-07-28 Tamar Christina <tamar.christina@arm.com>
8127 * config/aarch64/aarch64.c
8128 (aarch64_simd_container_mode): Add prototype.
8129 (aarch64_expand_mov_immediate): Add HI support.
8130 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
8131 (aarch64_can_const_movi_rtx_p): New.
8132 (aarch64_preferred_reload_class):
8133 Remove restrictions of using FP registers for certain SIMD operations.
8134 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
8135 (aarch64_valid_floating_const): Add integer move validation.
8136 (aarch64_simd_imm_scalar_p): Remove.
8137 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
8138 (aarch64_legitimate_constant_p): Expand list of supported cases.
8139 * config/aarch64/aarch64-protos.h
8140 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
8141 (aarch64_reinterpret_float_as_int): New.
8142 (aarch64_simd_imm_scalar_p): Remove.
8143 * config/aarch64/constraints.md (Uvi): New.
8144 (Dd): Split into Ds and new Dd.
8145 * config/aarch64/aarch64.md (*movsi_aarch64):
8147 (*movdi_aarch64): Add SIMD mov case.
8149 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8151 * tree-predcom.c: (struct chain): Handle store-store chain in which
8152 stores for elimination only store loop invariant values.
8153 (execute_pred_commoning_chain): Ditto.
8154 (prepare_initializers_chain_store_elim): Ditto.
8155 (prepare_finalizers): Ditto.
8156 (is_inv_store_elimination_chain): New function.
8157 (initialize_root_vars_store_elim_1): New function.
8159 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8161 * tree-predcom.c: Revise general description of the pass.
8162 (enum chain_type): New enum type for store elimination.
8163 (struct chain): New field supporting store elimination.
8164 (struct component): Ditto.
8165 (dump_chain): Dump store-stores chain.
8166 (release_chain): Release resources.
8167 (split_data_refs_to_components): Compute and create component
8168 contains only stores for elimination.
8169 (get_chain_last_ref_at): New function.
8170 (make_invariant_chain): Initialization.
8171 (make_rooted_chain): Specify chain type in parameter and record it.
8172 (add_looparound_copies): Skip for store-stores chain.
8173 (determine_roots_comp): Compute type of chain and pass it to
8175 (initialize_root_vars_store_elim_2): New function.
8176 (finalize_eliminated_stores): New function.
8177 (remove_stmt): Handle store for elimination.
8178 (execute_pred_commoning_chain): Execute predictive commoning on
8180 (determine_unroll_factor): Skip unroll for store-stores chain.
8181 (prepare_initializers_chain_store_elim): New function.
8182 (prepare_initializers_chain): Hanlde store-store chain.
8183 (prepare_finalizers_chain, prepare_finalizers): New function.
8184 (tree_predictive_commoning_loop): Return integer value indicating
8185 if loop is unrolled or lcssa form is corrupted.
8186 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
8188 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8190 * tree-predcom.c (initialize_root): Delete.
8191 (execute_pred_commoning_chain): Initialize root vars and replace
8192 reference of non-combined chain directly, rather than call above
8195 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8197 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
8198 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
8200 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8202 * tree-predcom.c (struct chain): New field init_seq.
8203 (release_chain): Release init_seq.
8204 (prepare_initializers_chain): Record intialization stmts in above
8206 (insert_init_seqs): New function.
8207 (tree_predictive_commoning_loop): Call insert_init_seqs.
8209 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8211 * tree-predcom.c (determine_roots_comp): Skip trivial components.
8213 2017-07-28 Richard Biener <rguenther@suse.de>
8215 * match.pd: Remove superfluous :c.
8216 * genmatch.c (simplify::id): Add member.
8217 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
8219 (current_id): New global.
8220 (dt_node::parent): Move from ...
8221 (dt_operand::parent): ... here. Add for_id member.
8222 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
8223 (decision_tree::find_node): Relax order requirement when
8224 merging DT_TRUE nodes to ones inbetween the current simplify
8225 and the one we try to merge with. Add diagnostic whenever
8226 we need to enforce pattern order by not merging.
8227 (decision_tree::insert): Set current_id.
8228 (decision_tree::print_node): Dump parent node and for_id.
8229 (parser::last_id): Add member.
8230 (parser::push_simplify): Assign unique id.
8231 (parser::parser): Initialize last_id.
8233 2017-07-28 Martin Liska <mliska@suse.cz>
8236 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
8237 gimple_build_debug_bind.
8239 2017-07-28 Richard Biener <rguenther@suse.de>
8241 PR tree-optimization/81502
8242 * match.pd: Add pattern combining BIT_INSERT_EXPR with
8244 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
8246 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
8247 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
8248 for BIT_FIELD_REF args.
8249 * fold-const.c (make_bit_field_ref): Likewise.
8250 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8252 2017-07-28 Jakub Jelinek <jakub@redhat.com>
8255 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
8256 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
8257 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
8258 Or it into SANITIZER_UNDEFINED.
8259 * ubsan.c: Include gimple-fold.h and varasm.h.
8260 (ubsan_expand_ptr_ifn): New function.
8261 (instrument_pointer_overflow): New function.
8262 (maybe_instrument_pointer_overflow): New function.
8263 (instrument_object_size): Formatting fix.
8264 (pass_ubsan::execute): Call instrument_pointer_overflow
8265 and maybe_instrument_pointer_overflow.
8266 * internal-fn.c (expand_UBSAN_PTR): New function.
8267 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
8268 * sanitizer.def (__ubsan_handle_pointer_overflow,
8269 __ubsan_handle_pointer_overflow_abort): New builtins.
8270 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
8271 * internal-fn.def (UBSAN_PTR): New internal function.
8272 * opts.c (sanitizer_opts): Add pointer-overflow.
8273 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
8274 * fold-const.c (build_range_check): Compute pointer range check in
8275 integral type if pointer arithmetics would be needed. Formatting
8278 2017-07-28 Martin Liska <mliska@suse.cz>
8281 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
8282 parameters that are of a variable-length.
8284 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
8286 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
8288 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
8289 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
8290 (CRT_CALL_STATIC_FUNCTION): Likewise.
8291 (ASM_DEFAULT_SPEC): New define.
8292 (ASM_SPEC32): Likewise.
8293 (ASM_SPEC64): Likewise.
8294 (ASM_SPEC_COMMON): Likewise.
8295 (ASM_SPEC): Likewise.
8296 (INVALID_64BIT): Likewise.
8297 (LINK_OS_DEFAULT_SPEC): Likewise.
8298 (LINK_OS_SPEC32): Likewise.
8299 (LINK_OS_SPEC64): Likewise.
8300 (POWERPC_LINUX): Likewise.
8301 (PTRDIFF_TYPE): Likewise.
8302 (RESTORE_FP_PREFIX): Likewise.
8303 (RESTORE_FP_SUFFIX): Likewise.
8304 (SAVE_FP_PREFIX): Likewise.
8305 (SAVE_FP_SUFFIX): Likewise.
8306 (SIZE_TYPE): Likewise.
8307 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
8308 (TARGET_64BIT): Likewise.
8309 (TARGET_64BIT): Likewise.
8310 (TARGET_AIX): Likewise.
8311 (WCHAR_TYPE_SIZE): Likewise.
8312 (WCHAR_TYPE): Undefine.
8313 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
8314 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
8315 (CPP_OS_RTEMS_SPEC): Delete.
8316 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
8317 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
8319 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
8321 2017-07-28 Jakub Jelinek <jakub@redhat.com>
8323 PR tree-optimization/81578
8324 * tree-parloops.c (build_new_reduction): Bail out if
8325 reduction_code isn't one of the standard OpenMP reductions.
8326 Move the details printing after that decision.
8328 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8330 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
8331 related to reload_in_progress.
8332 (splat_input_operand): Likewise.
8333 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
8335 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
8337 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
8338 (TARGET_INSTANTIATE_DECLS): Likewise.
8339 (legitimate_indexed_address_p): Delete reload_in_progress code.
8340 (rs6000_debug_legitimate_address_p): Likewise.
8341 (rs6000_eliminate_indexed_memrefs): Likewise.
8342 (rs6000_emit_le_vsx_store): Likewise.
8343 (rs6000_emit_move_si_sf_subreg): Likewise.
8344 (rs6000_emit_move): Likewise.
8345 (register_to_reg_type): Likewise.
8346 (rs6000_pre_atomic_barrier): Likewise.
8347 (rs6000_machopic_legitimize_pic_address): Likewise.
8348 (rs6000_allocate_stack_temp): Likewise.
8349 (rs6000_address_for_fpconvert): Likewise.
8350 (rs6000_address_for_altivec): Likewise.
8351 (rs6000_secondary_memory_needed_rtx): Delete function.
8352 (rs6000_check_sdmode): Likewise.
8353 (rs6000_alloc_sdmode_stack_slot): Likewise.
8354 (rs6000_instantiate_decls): Likewise.
8355 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
8356 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
8357 Delete reload_in_progress.
8358 (*vec_reload_and_plus_<mptrsize>): Likewise.
8359 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
8360 (vsx_div_v2di): Likewise.
8361 (vsx_udiv_v2di): Likewise.
8363 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8365 * config/rs6000/rs6000.opt (mlra): Replace with stub.
8366 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
8367 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
8368 (rs6000_debug_reg_global): Delete print of LRA status.
8369 (rs6000_option_override_internal): Delete dead LRA related code.
8370 (rs6000_lra_p): Delete function.
8371 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
8373 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
8375 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
8376 * config/riscv/rtems.h: New file.
8378 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8379 Sudakshina Das <sudi.das@arm.com>
8381 * config/aarch64/aarch64.md
8382 (define_split for and<mode>3nr_compare): Move
8383 non aarch64_logical_operand to a register.
8384 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
8385 register immediate operand to a register.
8386 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
8388 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8391 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
8393 2017-07-27 Richard Biener <rguenther@suse.de>
8395 PR tree-optimization/81573
8396 PR tree-optimization/81494
8397 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
8398 multi defuse cycle case.
8400 2017-07-27 Richard Biener <rguenther@suse.de>
8402 PR tree-optimization/81571
8403 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
8406 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
8408 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
8409 earlier and only if MASK_FPU is set. Adjust formatting.
8411 2017-07-27 Martin Liska <mliska@suse.cz>
8413 * opt-functions.awk: Add validation of value of Init.
8414 * optc-gen.awk: Pass new argument.
8416 2017-07-27 Martin Liska <mliska@suse.cz>
8418 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
8419 Fix wrong condition.
8421 2017-07-27 Martin Liska <mliska@suse.cz>
8423 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
8424 BBs and edges seen by autoFDO.
8426 2017-07-27 Richard Biener <rguenther@suse.de>
8428 PR tree-optimization/81502
8429 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
8430 with incompatible but same sized type.
8431 (execute_update_addresses_taken): Likewise.
8433 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
8435 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
8436 flag_tree_loop_vectorize rather than flag_tree_vectorize.
8438 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8441 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
8442 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
8443 Change s_operand to memory_operand.
8445 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
8447 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
8448 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
8449 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
8450 Emit instructions rather than returning an expression. Handle TFmode
8451 and KFmode by casting to TImode.
8452 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
8453 (rs6000_emit_le_vsx_store): Likewise.
8454 * config/rs6000/vsx.md (VSX_TI): New iterator.
8455 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
8456 (*vsx_le_undo_permute_<mode>): Likewise.
8457 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
8458 emit the split sequence.
8459 (*vsx_le_perm_store_<mode>): Likewise.
8461 2017-07-27 Jakub Jelinek <jakub@redhat.com>
8463 PR tree-optimization/81555
8464 PR tree-optimization/81556
8465 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
8466 if true, force CHANGED for the recursive invocation.
8467 (reassociate_bb): Remember original length of ops array, pass
8468 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
8470 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
8471 attributes for noipa attribute. For naked attribute use
8472 lookup_attribute first before lookup_attribute_spec.
8473 * final.c (rest_of_handle_final): Disable IPA RA for functions with
8475 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
8476 for functions with noipa attribute.
8477 (cgraph_externally_visible_p): Return true for functions with noipa
8479 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
8480 for functions with noipa attribute.
8481 * doc/extend.texi: Document noipa function attribute.
8482 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
8483 also for functions with noipa attribute.
8484 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
8486 2017-07-26 Andrew Pinski <apinski@cavium.com>
8488 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
8489 vec_unalign_load_cost and vec_unalign_store_cost.
8491 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8493 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
8494 -mvsx-small-integer option.
8495 (ISA_3_0_MASKS_IEEE): Likewise.
8496 (OTHER_VSX_VECTOR_MASKS): Likewise.
8497 (POWERPC_MASKS): Likewise.
8498 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
8499 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
8500 code, only testing for DImode being allowed in non-VSX floating
8502 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
8503 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
8505 (rs6000_option_override_internal): Delete -mvsx-small-integer.
8506 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
8507 TARGET_P8_VECTOR test.
8508 (rs6000_secondary_reload_simple_move): Likewise.
8509 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
8510 since TARGET_P9_VECTOR was already tested.
8511 (rs6000_opt_masks): Remove -mvsx-small-integer.
8512 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
8513 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8515 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
8516 test for TARGET_VEXTRACTUB was used, and that uses
8518 (p9 extract splitter): Likewise.
8519 (vsx_extract_<mode>_di_p9): Likewise.
8520 (vsx_extract_<mode>_store_p9): Likewise.
8521 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
8522 for TARGET_P9_VECTOR was used. Delete code that is now dead with
8523 the elimination of TARGET_VSX_SMALL_INTEGER.
8524 (vsx_extract_<mode>_p8): Likewise.
8525 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
8526 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
8527 (vsx_set_<mode>_p9): Likewise.
8528 (vsx_set_v4sf_p9): Likewise.
8529 (vsx_set_v4sf_p9_zero): Likewise.
8530 (vsx_insert_extract_v4sf_p9): Likewise.
8531 (vsx_insert_extract_v4sf_p9_2): Likewise.
8532 * config/rs6000/rs6000.md (sign extend splitter): Change
8533 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
8534 (floatsi<mode>2_lfiwax_mem): Likewise.
8535 (floatunssi<mode>2_lfiwzx_mem): Likewise.
8536 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
8537 since a test for TARGET_P9_VECTOR was used.
8538 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8539 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
8540 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8541 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
8542 TARGET_P8_VECTOR test.
8543 (fix_trunc<mode>si2_stfiwx): Likewise.
8544 (fix_trunc<mode>si2_internal): Likewise.
8545 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
8546 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8548 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
8549 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
8550 TARGET_P8_VECTOR test.
8551 (fixuns_trunc<mode>si2_stfiwx): Likewise.
8552 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
8553 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8555 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
8556 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
8557 since a test for TARGET_P9_VECTOR was used.
8558 (splitter for loading small constants): Likewise.
8560 2017-07-26 Andrew Pinski <apinski@cavium.com>
8562 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
8565 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
8568 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
8569 (fp_valid_at): Likewise.
8571 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
8573 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
8574 (qdf24xx_addrcost_table): Likewise.
8575 (cortexa57_tunings): Update to use generic_branch_cost.
8576 (cortexa72_tunings): Likewise.
8577 (cortexa73_tunings): Likewise.
8578 (qdf24xx_tunings): Likewise.
8580 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
8582 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
8583 (thunderx2t99_branch_cost): Likewise.
8584 (cortexa35_tunings): Update to use generic_branch_cost.
8585 (cortexa53_tunings): Likewise.
8586 (cortexa57_tunings): Likewise.
8587 (cortexa72_tunings): Likewise.
8588 (cortexa73_tunings): Likewise.
8589 (thunderx2t99_tunings): Likewise.
8591 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
8593 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
8594 (sparc_option_override): Honour MASK_FSMULD.
8595 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
8596 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
8597 * config/sparc/sparc.opt (mfsmuld): New option.
8598 * doc/invoke.texi (mfsmuld): Document option.
8600 2017-07-26 Marek Polacek <polacek@redhat.com>
8603 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
8605 2017-07-26 Richard Biener <rguenther@suse.de>
8607 * gimple-match-head.c (do_valueize): Return OP if valueize
8609 (get_def): New helper to get at the def stmt of a SSA name
8611 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
8612 do_valueize to get at the def stmt.
8613 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
8615 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
8618 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
8620 2017-07-26 Martin Liska <mliska@suse.cz>
8623 * function.c (expand_function_start): Make expansion of
8624 nonlocal_goto_save_area after parm_birth_insn.
8626 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
8628 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
8629 from all CPU target flags enable members.
8631 2017-07-26 Richard Biener <rguenther@suse.de>
8633 * genmatch.c (dt_simplify::gen): Make iterator vars const.
8634 (decision_tree::gen): Make 'type' const.
8635 (write_predicate): Likewise.
8637 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
8639 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
8640 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
8641 (rs6000_option_override_internal): Likewise.
8642 (rs6000_expand_vector_set): Likewise.
8643 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
8644 (TARGET_UPPER_REGS_SF): Likewise.
8645 (TARGET_UPPER_REGS_DI): Likewise.
8646 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
8647 (TARGET_DIRECT_MOVE_64BIT): Likewise.
8648 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
8649 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8650 (Splitters for DI constants in Altivec registers): Likewise.
8651 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
8652 (vsx_set_v4sf_p9): Likewise.
8653 (vsx_set_v4sf_p9_zero): Likewise.
8654 (vsx_insert_extract_v4sf_p9): Likewise.
8655 (vsx_insert_extract_v4sf_p9_2): Likewise.
8657 2017-07-25 Carl Love <cel@us.ibm.com>
8659 * doc/extend.texi: Update the built-in documentation file for the
8660 existing built-in functions
8661 vector signed char vec_cnttz (vector signed char);
8662 vector unsigned char vec_cnttz (vector unsigned char);
8663 vector signed short vec_cnttz (vector signed short);
8664 vector unsigned short vec_cnttz (vector unsigned short);
8665 vector signed int vec_cnttz (vector signed int);
8666 vector unsigned int vec_cnttz (vector unsigned int);
8667 vector signed long long vec_cnttz (vector signed long long);
8668 vector unsigned long long vec_cnttz (vector unsigned long long);
8670 2017-07-25 Andrew Pinski <apinski@cavium.com>
8672 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
8673 accesses where the use is for the first operand of a BIT_INSERT.
8675 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
8678 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
8679 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
8681 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
8683 * config/i386/gstabs.h: Delete.
8684 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
8686 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
8688 * config/i386/i386.c (ix86_decompose_address): Do not check for
8689 register RTX when looking at index_reg or base_reg.
8690 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
8692 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
8694 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
8695 to update EH info here.
8697 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8699 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
8701 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8703 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
8705 2017-07-25 Torsten Duwe <duwe@suse.de>
8707 * common.opt: Introduce -fpatchable-function-entry
8708 command line option, and its variables function_entry_patch_area_size
8709 and function_entry_patch_area_start.
8710 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
8711 including a two-value parser.
8712 * target.def (print_patchable_function_entry): New target hook.
8713 * targhooks.h (default_print_patchable_function_entry): New function.
8714 * targhooks.c (default_print_patchable_function_entry): Likewise.
8715 * toplev.c (process_options): Switch off IPA-RA if
8716 patchable function entries are being generated.
8717 * varasm.c (assemble_start_function): Look at the
8718 patchable-function-entry command line switch and current
8719 function attributes and maybe generate NOP instructions by
8720 calling the print_patchable_function_entry hook.
8721 * doc/extend.texi: Document patchable_function_entry attribute.
8722 * doc/invoke.texi: Document -fpatchable_function_entry
8723 command line option.
8724 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
8726 * doc/tm.texi: Re-generate.
8728 2017-07-25 Jakub Jelinek <jakub@redhat.com>
8731 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
8732 TARGET_AVX512DQ rather than TARGET_AVX512BW.
8734 2017-07-25 Tamar Christina <tamar.christina@arm.com>
8736 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
8738 2017-07-25 Richard Biener <rguenther@suse.de>
8740 PR tree-optimization/81455
8741 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
8742 not walk in cycles when looking for guards.
8744 2017-07-25 Richard Biener <rguenther@suse.de>
8746 PR tree-optimization/81529
8747 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
8748 when optimizing backedge uses.
8750 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
8752 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
8754 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
8755 to dl_section_ref. On AIX, append an expression to subtract
8756 the size of the section length to dl_section_ref.
8758 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
8760 * configure.ac: If any of the config.* scripts fail, exit 1.
8761 * configure: Regenerate.
8763 2017-07-25 Richard Biener <rguenther@suse.de>
8766 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
8767 of immediate uses, be more verbose on errors.
8769 2017-07-25 Richard Biener <rguenther@suse.de>
8771 PR tree-optimization/81510
8772 * tree-vect-loop.c (vect_is_simple_reduction): When the
8773 reduction stmt is not inside the loop bail out.
8775 2017-07-25 Richard Biener <rguenther@suse.de>
8777 PR tree-optimization/81303
8778 * tree-vect-loop-manip.c (vect_loop_versioning): Build
8779 profitability check against LOOP_VINFO_NITERSM1.
8781 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8783 * domwalk.c (cmp_bb_postorder): Simplify.
8784 (sort_bbs_postorder): New function. Use it...
8785 (dom_walker::walk): ...here to optimize common cases.
8787 2017-07-25 Martin Liska <mliska@suse.cz>
8790 * ipa-visibility.c (function_and_variable_visibility): Make the
8791 redirection just on target that supports aliasing.
8792 Fix GNU coding style.
8794 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
8797 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
8799 * config/arm/rtems.h: Likewise.
8800 * config/bfin/rtems.h: Likewise.
8801 * config/i386/rtemself.h: Likewise.
8802 * config/lm32/rtems.h: Likewise.
8803 * config/m32c/rtems.h: Likewise.
8804 * config/m68k/rtemself.h: Likewise.
8805 * config/microblaze/rtems.h: Likewise.
8806 * config/mips/rtems.h: Likewise.
8807 * config/moxie/rtems.h: Likewise.
8808 * config/nios2/rtems.h: Likewise.
8809 * config/powerpcspe/rtems.h: Likewise.
8810 * config/rs6000/rtems.h: Likewise.
8811 * config/rtems.h: Likewise.
8812 * config/sh/rtems.h: Likewise.
8813 * config/sh/rtemself.h: Likewise.
8814 * config/sparc/rtemself.h: Likewise.
8816 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
8819 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
8820 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
8821 * tree-ssa-structalias.c (alias_get_name): Same.
8823 2017-07-25 Bin Cheng <bin.cheng@arm.com>
8826 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
8827 instructions if no du chain is found.
8829 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
8831 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
8833 2017-07-25 Richard Biener <rguenther@suse.de>
8836 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
8839 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
8841 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
8843 (ISA_2_7_MASKS_SERVER): Likewise.
8844 (ISA_3_0_MASKS_IEEE): Likewise.
8845 (OTHER_P8_VECTOR_MASKS): Likewise.
8846 (OTHER_VSX_VECTOR_MASKS): Likewise.
8847 (POWERPC_MASKS): Likewise.
8848 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
8849 duplicate list of options.
8850 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
8851 explicit -mupper-regs options.
8852 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
8853 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
8854 alias for -mupper-regs-df.
8855 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
8856 (rs6000_init_hard_regno_mode_ok): Likewise.
8857 (rs6000_option_override_internal): Likewise.
8858 (rs6000_opt_masks): Likewise.
8859 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
8860 options in terms of whether -mvsx or -mpower8-vector was used.
8861 (TARGET_UPPER_REGS_DI): Likewise.
8862 (TARGET_UPPER_REGS_SF): Likewise.
8863 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
8864 -mupper-regs-* options.
8866 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
8868 * passes.c (emergency_dump_function): Print some empty lines and a
8869 header before the RTL dump.
8871 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
8873 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
8875 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
8878 * config/aarch64/aarch64.c (aarch64_classify_symbol):
8879 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
8881 2017-07-24 Carl Love <cel@us.ibm.com>
8883 * config/rs6000/rs6000-c.c: Add support for built-in functions
8884 vector float vec_extract_fp32_from_shorth (vector unsigned short);
8885 vector float vec_extract_fp32_from_shortl (vector unsigned short);
8886 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
8887 vec_extract_fp_from_shortl): Add defines for the two builtins.
8888 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
8889 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
8891 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
8892 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
8893 * doc/extend.texi: Update the built-in documentation file for the
8894 new built-in function.
8896 2017-07-24 Jakub Jelinek <jakub@redhat.com>
8899 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
8901 * doc/generic.texi: Likewise.
8902 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
8903 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
8905 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
8907 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
8908 (aarch64_mls_elt_merge<mode>): Likewise.
8910 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
8912 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
8913 having __cxa_atexit.
8915 2017-07-23 Michael Collison <michael.collison@arm.com>
8917 * config/arm/arm.c (arm_option_override): Deprecate
8918 use of -mstructure-size-boundary.
8919 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
8920 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
8922 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8925 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
8926 Reduce cost estimate for direct moves.
8928 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
8931 * config/i386/i386.c (ix86_option_override_internal): Disable
8932 BMI, BMI2 and TBM instructions for -m16.
8934 2017-07-21 Carl Love <cel@us.ibm.com>
8936 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8937 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8938 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8939 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8940 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8941 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8942 VMULOSW): New enum "unspec" values.
8943 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8944 altivec_vmulosw): New patterns.
8945 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8946 VMULOSW): Add definitions.
8948 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
8950 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
8951 (qdf24xx): Likewise.
8952 * config/aarch64/aarch64-options-extensions.def (rdma); New.
8953 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
8954 (AARCH64_FL_V8_1): Renumber.
8955 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
8956 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
8957 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
8958 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
8959 rdma to feature modifiers list.
8961 2017-07-21 Yury Gribov <tetra2005@gmail.com>
8964 * ipa-visibility (function_and_variable_visibility): Convert
8965 recursive PLT call to direct call if appropriate.
8967 2017-07-21 Andrew Pinski <apinski@cavium.com>
8969 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
8970 operand 1 to see if the types precision matches.
8971 * fold-const.c (operand_equal_p): Likewise.
8973 2017-07-21 Richard Biener <rguenther@suse.de>
8975 PR tree-optimization/81303
8976 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
8977 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
8978 (vect_peeling_hash_get_lowest_cost): Adjust.
8979 (vect_enhance_data_refs_alignment): Likewise. Use
8980 vect_get_peeling_costs_all_drs to compute the penalty for no
8981 peeling to match up costs.
8983 2017-07-21 Richard Biener <rguenther@suse.de>
8985 PR tree-optimization/81500
8986 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
8987 we didn't identify a reduction path.
8989 2017-07-21 Tom de Vries <tom@codesourcery.com>
8990 Cesar Philippidis <cesar@codesourcery.com>
8992 PR gcov-profile/81442
8993 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
8996 2017-07-21 Tom de Vries <tom@codesourcery.com>
8999 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
9001 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
9002 nvptx_override_options_after_change.
9004 2017-07-21 Ulrich Drepper <drepper@redhat.com>
9006 * dwarf2out.c (output_file_names): Avoid double testing for
9009 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
9011 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
9013 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
9015 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
9017 (try_crossjump_to_edge): Do not punt on partitioned functions.
9019 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
9021 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
9022 Put all BBs reachable only via paths crossing cold region to cold
9024 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
9026 2016-07-21 Richard Biener <rguenther@suse.de>
9028 PR tree-optimization/81303
9029 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
9030 into account prologue and epilogue iterations when raising
9031 min_profitable_iters to sth at least covering one vector iteration.
9033 2017-07-21 Tamar Christina <tamar.christina@arm.com>
9035 * config/arm/arm.c (arm_test_cpu_arch_dat):
9038 2017-07-20 Nathan Sidwell <nathan@acm.org>
9040 Remove TYPE_METHODS.
9041 * tree.h (TYPE_METHODS): Delete.
9042 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
9043 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
9044 (dbxout_type_methods): Scan TYPE_FIELDS.
9045 (dbxout_type): Don't check TYPE_METHODS here.
9046 * function.c (use_register_for_decl): Always ignore register for
9047 class types when not optimizing.
9048 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
9049 * tree.c (free_lang_data_in_type): Stitch out member functions and
9050 templates from TYPE_FIELDS.
9051 (build_distinct_type_copy, verify_type_variant,
9052 verify_type): Member fns are on TYPE_FIELDS.
9053 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
9054 * tree-pretty-print.c (dump_generic_node): Likewise.
9056 2017-07-20 Jakub Jelinek <jakub@redhat.com>
9059 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
9060 V2TImode and V4TImode.
9061 (ix86_expand_vector_extract): Likewise.
9062 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
9063 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
9064 (ssescalarmode): Handle V4TImode and V2TImode.
9065 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
9066 (*vec_extractv2ti, *vec_extractv4ti): New insns.
9067 (VEXTRACTI128_MODE): New mode iterator.
9068 (splitter for *vec_extractv?ti first element): New.
9069 (VEC_INIT_MODE): New mode iterator.
9070 (vec_init<mode>): Consolidate 3 expanders into one using
9071 VEC_INIT_MODE mode iterator.
9073 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
9075 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
9076 non_spilled_static_chain_regno_p.
9078 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
9080 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
9082 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
9084 * bb-reorder.c (connect_traces): Allow copying of blocks within
9087 2017-07-20 Richard Biener <rguenther@suse.de>
9089 * gimple.h (gimple_phi_result): Add gphi * overload.
9090 (gimple_phi_result_ptr): Likewise.
9091 (gimple_phi_arg): Likewise. Adjust index assert to only
9092 allow actual argument accesses rather than all slots available
9094 (gimple_phi_arg_def): Add gphi * overload.
9095 * tree-phinodes.c (make_phi_node): Initialize only actual
9097 (resize_phi_node): Clear memory not covered by old node,
9098 do not initialize excess argument slots.
9099 (reserve_phi_args_for_new_edge): Initialize new argument slot
9102 2017-07-20 Bin Cheng <bin.cheng@arm.com>
9104 PR tree-optimization/81388
9106 2016-07-21 Bin Cheng <bin.cheng@arm.com>
9108 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
9109 by removing computation of may_be_zero.
9111 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
9112 Tom de Vries <tom@codesourcery.com>
9115 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
9116 when gimple level profile disagrees with what RTL expander did.
9118 2017-07-20 Richard Biener <rguenther@suse.de>
9120 PR tree-optimization/61171
9121 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
9122 (vect_analyze_stmt): Add slp instance parameter.
9123 (vectorizable_reduction): Likewise.
9124 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
9125 (vect_is_simple_reduction): Deal with chains not detected
9126 as SLP reduction chain, specifically not properly associated
9127 chains containing a mix of plus/minus.
9128 (get_reduction_op): Remove.
9129 (get_initial_defs_for_reduction): Simplify, pass in whether
9130 this is a reduction chain, pass in the SLP node for the PHIs.
9131 (vect_create_epilog_for_reduction): Get the SLP instance as
9133 (vectorizable_reduction): Get the SLP instance as arg.
9134 During analysis remember the SLP node with the PHIs in the
9135 instance. Simplify getting at the vectorized reduction PHIs.
9136 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
9137 through SLP instance.
9138 (vect_slp_analyze_operations): Likewise.
9139 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
9140 (vect_transform_stmt): Likewise.
9142 2017-07-20 Tom de Vries <tom@codesourcery.com>
9144 PR tree-optimization/81489
9145 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
9146 read of phi arg location to before loop that modifies phi.
9148 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9150 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
9153 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
9156 * except.c (execute): Fix ordering issue.
9158 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
9160 PR rtl-optimization/81423
9161 * combine.c (make_compound_operation_int): Don't try to optimize
9162 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
9164 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
9166 PR rtl-optimization/81423
9167 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
9168 with a constant that is -1 in the truncated to mode.
9170 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
9172 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
9173 (determine_unlikely_bbs): ... here.
9174 * predict.h (propagate_unlikely_bbs_forward): Declare.
9175 * cfgexpand.c (pass_expand::execute): Use it.
9176 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
9178 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
9179 propagate_unlikely_bbs_forward.
9181 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
9184 * except.c (maybe_add_nop_after_section_switch): New function.
9187 2017-07-19 Tom de Vries <tom@codesourcery.com>
9189 * gimple.h (gimple_phi_set_arg): Make assert more strict.
9191 2017-07-19 Tom de Vries <tom@codesourcery.com>
9193 * gimple.h (gimple_phi_arg): Make assert more strict.
9195 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
9197 * config.gcc (powerpc*-*-*): Add mmintrin.h.
9198 * config/rs6000/mmintrin.h: New file.
9199 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
9201 2017-07-19 Jakub Jelinek <jakub@redhat.com>
9203 PR tree-optimization/81346
9204 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
9206 2017-07-19 Tom de Vries <tom@codesourcery.com>
9208 * config/nvptx/nvptx.md (VECIM): Add V2DI.
9210 2017-07-19 Tom de Vries <tom@codesourcery.com>
9212 * config/nvptx/nvptx-modes.def: Add V2DImode.
9213 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
9214 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
9215 (nvptx_output_mov_insn): Handle lack of mov.b128.
9216 (nvptx_print_operand): Handle 'H' and 'L' codes.
9217 (nvptx_vector_mode_supported): Allow V2DImode.
9218 (nvptx_preferred_simd_mode): New function.
9219 (nvptx_data_alignment): New function.
9220 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
9221 nvptx_preferred_simd_mode.
9222 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
9224 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
9226 2017-07-19 Tom de Vries <tom@codesourcery.com>
9228 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
9229 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
9230 (nvptx_vector_mode_supported): New function. Allow V2SImode.
9231 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
9232 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
9233 (mov<VECIM>_insn): New define_insn.
9234 (define_expand "mov<VECIM>): New define_expand.
9236 2017-07-19 Tom de Vries <tom@codesourcery.com>
9238 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
9240 2017-07-19 Jakub Jelinek <jakub@redhat.com>
9242 PR tree-optimization/81346
9243 * fold-const.h (fold_div_compare, range_check_type): Declare.
9244 * fold-const.c (range_check_type): New function.
9245 (build_range_check): Use range_check_type.
9246 (fold_div_compare): No longer static, rewritten into
9247 a match.pd helper function.
9248 (fold_comparison): Don't call fold_div_compare here.
9249 * match.pd (X / C1 op C2): New optimization using fold_div_compare
9252 2017-07-19 Nathan Sidwell <nathan@acm.org>
9254 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
9255 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
9256 * tree.c (find_decls_types_r, verify_type): Use
9257 TYPE_{MIN,MAX}_VALUE_RAW.
9258 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
9259 (hash_tree): Likewise.
9260 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
9262 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
9265 2017-07-18 Tom de Vries <tom@codesourcery.com>
9268 * omp-expand.c (expand_omp_for_static_chunk): Handle
9269 equal-argument loop exit phi.
9271 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
9274 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
9275 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
9276 operand 2 predicate.
9277 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
9278 operand 2 predicate.
9279 (ror,rol -> rorx splitters): Use const_int_operand as
9280 operand 2 predicate.
9282 2017-06-18 Richard Biener <rguenther@suse.de>
9284 PR tree-optimization/81410
9285 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
9286 the gap in the ! slp_perm SLP case after each group.
9288 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
9291 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
9294 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
9297 * predict.c (set_even_probabilities): Cleanup; do not affect
9298 probabilities that are already known.
9299 (combine_predictions_for_bb): Call even when count is set.
9301 2017-07-18 Nathan Sidwell <nathan@acm.org>
9303 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
9306 2017-07-18 Bin Cheng <bin.cheng@arm.com>
9309 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
9310 optimization for loop niter analysis.
9312 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
9315 * config/avr/avr.c (avr_optimize_casesi): Don't use
9316 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
9318 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
9320 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
9321 body_cost_vec from _vect_peel_extended_info.
9322 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
9323 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
9326 2017-07-18 Bin Cheng <bin.cheng@arm.com>
9328 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
9330 2017-07-18 Richard Biener <rguenther@suse.de>
9332 PR tree-optimization/80620
9333 PR tree-optimization/81403
9334 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
9335 info when re-using a VN table entry.
9337 2017-07-18 Richard Biener <rguenther@suse.de>
9339 PR tree-optimization/81418
9340 * tree-vect-loop.c (vectorizable_reduction): Properly compute
9341 vectype_in. Verify that with lane-reducing reduction operations
9342 we have a single def-use cycle.
9344 2017-07-17 Carl Love <cel@us.ibm.com>
9346 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
9348 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9349 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9350 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9351 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9352 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9353 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
9354 VMULOSW): New enum "unspec" values.
9355 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
9356 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
9357 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
9358 altivec_vmulosw): New patterns.
9359 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9360 VMULOSW): Add definitions.
9361 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
9363 * config/alpha/alpha.c: Include predict.h.
9365 2017-07-17 Yury Gribov <tetra2005@gmail.com>
9367 * tree-vrp.c (compare_assert_loc): Fix comparison function
9368 to return predictable results.
9370 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9372 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
9375 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
9376 * doc/invoke.texi (mexpand-adddi): Update text.
9378 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9380 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
9381 that also clobbers the CC register. The old expand code is moved
9383 (*arc_clzsi2): ... here.
9384 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
9385 the CC register. The old expand code is moved to ...
9386 (arc_ctzsi2): ... here.
9388 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9390 * config/arc/arc.opt (mindexed-loads): Use initial value
9391 TARGET_INDEXED_LOADS_DEFAULT.
9392 (mauto-modify-reg): Use initial value
9393 TARGET_AUTO_MODIFY_REG_DEFAULT.
9394 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
9395 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
9396 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
9397 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
9399 2017-07-17 Martin Liska <mliska@suse.cz>
9402 * opts.c (finish_options): Do not allow -fgnu-tm
9403 w/ -fsanitize={kernel-,}address. Say sorry.
9405 2017-07-17 Bin Cheng <bin.cheng@arm.com>
9408 * tree-loop-distribution.c (classify_partition): Only assert on
9409 numer of iterations.
9410 (merge_dep_scc_partitions): Delete prameter. Update function call.
9411 (distribute_loop): Remove code handling loop with unknown niters.
9412 (pass_loop_distribution::execute): Skip loop with unknown niters.
9414 2017-07-17 Bin Cheng <bin.cheng@arm.com>
9417 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
9418 function sort_partitions_by_post_order.
9420 2017-07-17 Bin Cheng <bin.cheng@arm.com>
9422 PR tree-optimization/81374
9423 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
9424 the max index of basic blocks, rather than number of basic blocks.
9426 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9428 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
9430 (arc_legitimate_pic_operand_p): Likewise.
9431 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
9433 (arc_needs_pcl_p): Likewise.
9434 (arc_legitimate_pc_offset_p): Likewise.
9435 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
9436 function is also used in constrains.md.
9437 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
9438 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
9439 PLUS. Only return true/false in known cases, otherwise assert.
9440 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
9441 is already called in arc_legitimate_constant_p.
9442 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
9444 (LEGITIMATE_PIC_OPERAND_P): Use
9445 arc_raw_symbolic_reference_mentioned_p function.
9446 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
9451 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9452 Andrew Burgess <andrew.burgess@embecosm.com>
9454 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
9455 (arc_return_address_register): New function.
9456 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
9457 (arc_handle_fndecl_attribute): Add naked attribute.
9458 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
9459 (TARGET_WARN_FUNC_RETURN): Likewise.
9460 (arc_allocate_stack_slots_for_args): New function.
9461 (arc_warn_func_return): Likewise.
9462 (machine_function): Change type fn_type.
9463 (arc_compute_function_type): Consider new naked function type,
9464 change function return type.
9465 (arc_must_save_register): Adapt to handle new
9466 arc_compute_function_type's return type.
9467 (arc_expand_prologue): Likewise.
9468 (arc_expand_epilogue): Likewise.
9469 (arc_return_address_regs): Delete.
9470 (arc_return_address_register): New function.
9471 (arc_epilogue_uses): Use above function.
9472 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
9473 (arc_function_type): Change encoding, add naked type.
9474 (ARC_INTERRUPT_P): Change to handle the new encoding.
9475 (ARC_FAST_INTERRUPT_P): Likewise.
9476 (ARC_NORMAL_P): Define.
9477 (ARC_NAKED_P): Likewise.
9478 (arc_compute_function_type): Delete prototype.
9479 * config/arc/arc.md (in_ret_delay_slot): Use
9480 arc_return_address_register function.
9481 (simple_return): Likewise.
9482 (p_return_i): Likewise.
9484 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9486 PR tree-optimization/81428
9487 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
9488 can't be built for those types.
9490 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
9492 Remove stuff dead since r239246.
9494 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
9495 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
9496 (avr_inform_devices): Remove dead stuff.
9498 2017-07-17 Tamar Christina <tamar.christina@arm.com>
9500 * config/arm/arm_neon.h: Fix softp typo.
9502 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9504 PR tree-optimization/81365
9505 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
9506 aggregate moves onto bb predecessor edges, make sure there are no
9507 loads that could alias the lhs in between the start of bb and the
9510 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
9513 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
9514 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
9515 [LSHIFTRT, outer_code = TRUNCATE]: Same.
9517 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9519 PR tree-optimization/81396
9520 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
9521 (init_symbolic_number): Initialize it to 1.
9522 (perform_symbolic_merge): Add n_ops from both operands into the new
9524 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
9525 without base_addr as useless if they need more than one operation.
9526 (bswap_replace): Handle !bswap case for NULL base_addr.
9528 2017-07-17 Tom de Vries <tom@codesourcery.com>
9531 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
9534 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
9536 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
9537 conditional builtin define __FIX_LEON3FT_B2BST.
9539 2017-07-17 Daniel Cederman <cederman@gaisler.com>
9541 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
9542 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
9545 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
9547 PR rtl-optimization/81424
9548 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
9549 to remove potential trapping from operands if -fnon-call-exceptions.
9551 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9553 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
9554 profile_proability for scalling.
9555 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
9557 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9559 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
9561 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9563 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
9564 fixpoint arithmetics.
9566 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9568 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
9569 fixpoint arithmetics.
9571 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9573 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
9574 fixpoint arithmetics.
9576 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9578 * profile-count.h (profile_probability::from_reg_br_prob_note,
9579 profile_probability::to_reg_br_prob_note): New functions.
9580 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
9581 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
9582 * predict.c (probability_reliable_p): Update.
9583 (edge_probability_reliable_p): Update.
9584 (br_prob_note_reliable_p): Update.
9585 (invert_br_probabilities): Update.
9586 (add_reg_br_prob_note): New function.
9587 (combine_predictions_for_insn): Update.
9588 * asan.c (asan_clear_shadow): Update.
9589 * cfgbuild.c (compute_outgoing_frequencies): Update.
9590 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
9591 (update_br_prob_note): Update.
9592 (rtl_verify_edges): Update.
9593 (purge_dead_edges): Update.
9594 (fixup_reorder_chain): Update.
9595 * emit-rtl.c (try_split): Update.
9596 * ifcvt.c (cond_exec_process_insns): Update.
9597 (cond_exec_process_if_block): Update.
9598 (dead_or_predicable): Update.
9599 * internal-fn.c (expand_addsub_overflow): Update.
9600 (expand_neg_overflow): Update.
9601 (expand_mul_overflow): Update.
9602 * loop-doloop.c (doloop_modify): Update.
9603 * loop-unroll.c (compare_and_jump_seq): Update.
9604 * optabs.c (emit_cmp_and_jump_insn_1): Update.
9605 * predict.h: Update.
9606 * reorg.c (mostly_true_jump): Update.
9608 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
9609 * config/alpha/alpha.c (emit_unlikely_jump): Update.
9610 * config/arc/arc.c: (emit_unlikely_jump): Update.
9611 * config/arm/arm.c: (emit_unlikely_jump): Update.
9612 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
9613 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
9614 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
9615 (ix86_print_operand): Update.
9616 (ix86_split_fp_branch): Update.
9617 (predict_jump): Update.
9618 * config/ia64/ia64.c (ia64_print_operand): Update.
9619 * config/mmix/mmix.c (mmix_print_operand): Update.
9620 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
9621 (rs6000_expand_split_stack_prologue): Update.
9622 * config/rs6000/rs6000.c: Update.
9623 * config/s390/s390.c (s390_expand_vec_strlen): Update.
9624 (s390_expand_vec_movstr): Update.
9625 (s390_expand_cs_tdsi): Update.
9626 (s390_expand_split_stack_prologue): Update.
9627 * config/sh/sh.c (sh_print_operand): Update.
9628 (expand_cbranchsi4): Update.
9629 (expand_cbranchdi4): Update.
9630 * config/sparc/sparc.c (output_v9branch): Update.
9631 * config/spu/spu.c (get_branch_target): Update.
9632 (ea_load_store_inline): Update.
9633 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
9634 * config/tilepro/tilepro.c: Update.
9636 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
9638 * gimplify.c (mostly_copy_tree_r): Revert latest change.
9639 (gimplify_save_expr): Likewise.
9641 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9643 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
9645 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9647 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
9649 * timevar.def (TV_IPA_FNSUMMARY): Define.
9651 2017-07-16 Daniel Cederman <cederman@gaisler.com>
9653 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
9654 to back store errata sensitive sequence from being generated.
9655 (sqrtdf2_fix): Likewise.
9657 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9659 * tree-ssa-threadupdate.c (compute_path_counts,
9660 update_joiner_offpath_counts): Use profile_probability.
9662 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
9665 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9667 * config/arm/arm-c.c (arm_cpu_builtins): Define
9668 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
9670 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
9672 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9673 array entries to represent __ieee128 versions of the
9674 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
9675 scalar_extract_sig, and scalar_insert_exp built-in functions.
9676 (altivec_resolve_overloaded_builtin): Add special case handling
9677 for the __builtin_scalar_insert_exp function, as represented by
9678 the P9V_BUILTIN_VEC_VSIEDP constant.
9679 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
9680 exponent support for __ieee128 argument.
9681 (VSESQP): Add scalar extract signature support for __ieee128
9683 (VSTDCNQP): Add scalar test negative support for __ieee128
9685 (VSIEQP): Add scalar insert exponent support for __int128 argument
9686 with __ieee128 result.
9687 (VSIEQPF): Add scalar insert exponent support for __ieee128
9688 argument with __ieee128 result.
9689 (VSTDCQP): Add scalar test data class support for __ieee128
9691 (VSTDCNQP): Add overload support for scalar test negative with
9693 (VSTDCQP): Add overload support for scalar test data class
9695 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
9697 (UNSPEC_VSX_SIEXPQP): New constant.
9698 (xsxexpqp): New insn for VSX scalar extract exponent quad
9700 (xsxsigqp): New insn for VSX scalar extract significand quad
9702 (xsiexpqpf): New insn for VSX scalar insert exponent quad
9703 precision with floating point argument.
9704 (xststdcqp): New expand for VSX scalar test data class quad
9706 (xststdcnegqp): New expand for VSX scalar test negative quad
9708 (xststdcqp): New insn to match expansions for VSX scalar test data
9709 class quad precision and VSX scalar test negative quad precision.
9710 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
9711 special case operand checking to enforce that second operand of
9712 VSX scalar test data class with quad precision argument is a 7-bit
9714 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
9715 prototypes and descriptions of __ieee128 versions of
9716 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
9717 scalar_test_data_class, and scalar_test_neg built-in functions.
9719 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9721 PR tree-optimization/81162
9722 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
9723 replace a negate with an add.
9725 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
9727 * doc/invoke.texi (arm/-mcpu): Document +crypto.
9729 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9731 * config/arm/arm-c.c (arm_cpu_builtins): Define
9732 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
9734 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9736 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
9737 (armv8-r): Set ARM Cortex-R52 as default CPU.
9738 * config/arm/arm-tables.opt: Regenerate.
9739 * config/arm/arm-tune.md: Regenerate.
9740 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
9742 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
9743 extension for -mcpu=cortex-r52.
9745 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9747 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
9748 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
9749 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
9750 (fp-armv8): Define it as FP_ARMv8 only.
9751 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
9752 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
9754 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
9755 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
9756 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
9757 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
9758 than TARGET_FPU_ARMV8.
9759 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
9760 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
9761 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
9763 * config/arm/neon.md (neon_vrint): Likewise.
9764 (neon_vcvt): Likewise.
9765 (neon_<fmaxmin_op><mode>): Likewise.
9766 (<fmaxmin><mode>3): Likewise.
9767 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
9768 * config/arm/predicates.md (arm_cond_move_operator): Check against
9769 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
9771 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
9773 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
9776 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9778 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
9779 loop in comment with memset.
9781 2017-07-14 Martin Liska <mliska@suse.cz>
9783 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
9784 * dwarf2out.c (is_java): Remove the function.
9785 (output_pubname): Remove usage of the function.
9786 (lower_bound_default): Remove usage of DW_LANG_Java.
9787 (gen_compile_unit_die): Likewise.
9788 * gcc.c: Remove compiler defaults for .java and .zip files.
9789 * gimple-expr.c (remove_suffix): Change as there's no longer
9790 extension than 4-letter one.
9791 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
9792 (gimplify_save_expr): Likewise.
9793 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
9794 as it's possible even for other languages than Java.
9795 * langhooks.h (struct lang_hooks): Remove Java from a comment.
9796 * lto-opts.c (lto_write_options): Remove reference to Java.
9797 * opts.c (strip_off_ending): Update file extension handling.
9798 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
9799 * tree-eh.c (lower_resx): Likewise.
9800 * tree.c (free_lang_data_in_type): Remove dead code.
9801 (find_decls_types_r): Likewise.
9802 (build_common_builtin_nodes): Remove Java from a comment.
9803 (verify_type): Remove dead code.
9804 * varasm.c (assemble_external): Remove Java from a comment.
9806 2017-07-14 Martin Liska <mliska@suse.cz>
9808 * opts.c (finish_options): Add quotes.
9809 (common_handle_option): Likewise.
9811 2017-07-14 Martin Liska <mliska@suse.cz>
9813 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
9814 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
9816 * dwarf2out.c (lower_bound_default): Do not handle
9818 (gen_compile_unit_die): Likewise.
9819 * gcc.c: Remove default extension binding for GNU Pascal.
9820 * stmt.c: Remove Pascal language from a comment.
9821 * xcoffout.c: Likewise.
9823 2017-07-13 David Malcolm <dmalcolm@redhat.com>
9826 * diagnostic-show-locus.c (fixit_cmp): New function.
9827 (layout::layout): Sort m_fixit_hints.
9828 (column_range::column_range): Assert that the values are valid.
9829 (struct char_span): New struct.
9830 (correction::overwrite): New method.
9831 (struct source_line): New struct.
9832 (line_corrections::add_hint): Add assertions. Reimplement memcpy
9833 calls in terms of classes source_line and char_span, and
9834 correction::overwrite.
9835 (selftest::test_overlapped_fixit_printing_2): New function.
9836 (selftest::diagnostic_show_locus_c_tests): Call it.
9838 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
9840 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
9841 early if there is no lhs.
9843 2017-07-13 Martin Liska <mliska@suse.cz>
9845 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
9846 (gen_reference_type_die): Likewise.
9847 * stor-layout.c: Remove Pascal-related comment.
9849 2017-07-13 Martin Liska <mliska@suse.cz>
9851 * opts.c (finish_options): Add quotes to error messages.
9852 (parse_sanitizer_options): Likewise.
9854 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
9856 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
9858 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
9860 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
9862 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
9864 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
9866 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
9868 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
9871 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
9872 provides the hardware capability bits, define the macro
9873 __BUILTIN_CPU_SUPPORTS__.
9874 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
9875 if GLIBC does not provide the hardware capability bits. Add a
9876 gcc_unreachable call if the built-in cpu function is neither
9877 __builtin_cpu_is nor __builtin_cpu_supports.
9878 (rs6000_get_function_versions_dispatcher): Change the warning
9879 that an old GLIBC is used which does not export the capability
9880 bits to be an error.
9881 * doc/extend.texi (target_clones attribute): Document the
9882 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
9883 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
9884 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
9885 the macros defined by GCC if the newer GLIBC is available.
9887 2017-07-12 Jeff Law <law@redhat.com>
9889 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
9890 remaining includes slightly.
9891 * config/riscv/riscv-builtins.c: Include profile-count.h.
9893 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
9896 * config/avr/avr.c (avr_set_current_function): In diagnostic
9897 messages: Quote keywords and (parts of) identifiers.
9898 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
9901 2017-07-12 Carl Love <cel@us.ibm.com>
9903 * config/rs6000/rs6000-c.c: Add support for built-in functions
9904 vector bool char vec_revb (vector bool char);
9905 vector bool short vec_revb (vector short char);
9906 vector bool int vec_revb (vector bool int);
9907 vector bool long long vec_revb (vector bool long long);
9908 * doc/extend.texi: Update the built-in documentation file for the
9909 new built-in functions.
9911 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9913 * config/s390/s390.md: Remove movcc splitter.
9915 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9917 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
9918 load/store on condition.
9920 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
9923 * config/avr/avr.c (avr_encode_section_info)
9924 [progmem && !TREE_READONLY]: Error if progmem object needs
9927 2017-07-11 Michael Collison <michael.collison@arm.com>
9929 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
9932 2017-07-11 Carl Love <cel@us.ibm.com>
9934 * config/rs6000/rs6000-c.c: Add support for builtins
9935 vector unsigned int vec_parity_lsbb (vector signed int);
9936 vector unsigned int vec_parity_lsbb (vector unsigned int);
9937 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
9938 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
9939 vector unsigned long long vec_parity_lsbb (vector signed long long);
9940 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
9941 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
9942 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
9943 * doc/extend.texi: Update the built-in documentation file for the
9944 new built-in functions.
9946 2017-07-11 David Malcolm <dmalcolm@redhat.com>
9948 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
9949 (layout::m_primary_loc): New field.
9950 (layout::layout): Initialize new field. Move location filtering
9951 logic from here to...
9952 (layout::maybe_add_location_range): ...this new method. Add
9953 support for filtering to just the lines already specified by other
9955 (layout::will_show_line_p): New method.
9956 (gcc_rich_location::add_location_if_nearby): New method.
9957 (selftest::test_add_location_if_nearby): New test function.
9958 (selftest::diagnostic_show_locus_c_tests): Call it.
9959 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
9962 2017-07-11 Tom de Vries <tom@codesourcery.com>
9964 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
9965 (bb_first_real_insn): New function.
9966 (nvptx_single): Add extra initialization of broadcasted condition
9969 2017-07-11 Nathan Sidwell <nathan@acm.org>
9971 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
9973 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
9975 * doc/extend.texi (AVR Function Attributes): Remove weblink to
9976 Binutils doc as TEXI will mess them up.
9977 * doc/invoke.texi (AVR Options): Same here.
9979 2017-07-11 Daniel Cederman <cederman@gaisler.com>
9981 * config/sparc/sparc.opt (mfix-ut700): New option.
9982 (mfix-gr712rc): Likewise.
9983 (sparc_fix_b2bst): New variable.
9984 * doc/invoke.texi (SPARC options): Document them.
9985 (ARM options): Fix warnings.
9986 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
9987 instructions to prevent sequences that can trigger the store-store
9988 errata for certain LEON3FT processors.
9989 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
9990 (sparc_option_override): Set sparc_fix_b2bst appropriately.
9991 * config/sparc/sparc.md (fix_b2bst): New attribute.
9992 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
9994 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
9997 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
9999 (*rsqrtsf2_sse): Ditto.
10001 (div<mode>3): Macroize insn from divdf3 and divsf3
10002 using MODEF mode iterator.
10004 2017-07-10 Martin Sebor <msebor@redhat.com>
10006 PR tree-optimization/80397
10007 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
10008 instead of testing for equality to INTEGER_TYPE.
10010 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
10012 * config.gcc: Remove uclibc from arc target spec.
10014 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
10016 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
10018 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10021 * lto-wrapper.c (remove_option): New function.
10022 (merge_and_complain): Merge PIC/PIE options more realistically.
10024 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
10026 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
10030 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
10031 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
10032 * config.in: Regenerate.
10033 * configure: Regenerate.
10034 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
10035 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
10036 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
10037 (TARGET_GASISR_PROLOGUES): ...target mask.
10038 * common/config/avr/avr-common.c
10039 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
10040 Set -mgas-isr-prologues.
10041 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
10042 INSERT_PASS_BEFORE for it.
10043 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
10044 * config/avr/avr.c (avr_option_override)
10045 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
10046 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
10047 (avr_attribute_table) <no_gccisr>: Add new function attribute.
10048 (avr_set_current_function) <is_no_gccisr>: Init machine field.
10049 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
10051 (make_avr_pass_pre_proep): New function.
10052 (emit_push_sfr) <treg>: Add argument to function and use it
10053 instead of TMP_REG.
10054 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
10055 and set machine->gasisr.yes.
10056 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
10057 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
10058 __gcc_isr.n_pushed to .L__stack_usage.
10059 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
10060 (avr_asm_final_postscan_insn): ...this new static function.
10061 * config/avr/avr.h (machine_function)
10062 <is_no_gccisr, use_L__stack_usage>: New fields.
10063 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
10064 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
10065 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
10066 (gasisr, *gasisr): New expander and insn.
10067 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
10068 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
10069 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
10071 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
10073 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
10076 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
10078 Move jump-tables out of .text again.
10081 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
10082 (ASM_OUTPUT_ADDR_VEC): New function.
10083 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
10084 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
10085 INSN_ADDRESSes as asm comment.
10086 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
10087 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
10088 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
10089 * config/avr/avr.md (*tablejump): Adjust comment.
10090 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
10091 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
10093 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
10094 (avr_output_addr_vec): New proto.
10095 (avr_log_t) <insn_addresses>: New field.
10097 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
10100 * config/i386/i386.c (ix86_function_arg_advance): Set
10101 outgoing_args_on_stack to true if there are outgoing arguments
10103 (ix86_function_arg): Likewise.
10104 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
10105 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
10106 * config/i386/i386.h (machine_function): Add
10107 outgoing_args_on_stack.
10109 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
10111 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
10112 supporting pthreds.
10113 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
10115 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
10117 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
10118 (REAL_H): Remove $(MACHMODE_H).
10119 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
10121 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
10122 $(MACHMODE_H) and double-int.h.
10123 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
10125 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
10128 2017-07-07 Andrew Pinski <apinski@cavium.com>
10130 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
10131 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
10133 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10135 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
10136 Add warning if GCC was not configured to link against a GLIBC that
10137 exports the hardware capability bits.
10138 (make_resolver_func): Make resolver function private and not a
10139 COMDAT function. Create the name with clone_function_name instead
10140 of make_unique_name.
10143 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
10144 correct operand in doing the split.
10146 2017-07-07 Carl Love <cel@us.ibm.com>
10148 * config/rs6000/rs6000-c: Add support for built-in function
10149 vector unsigned short vec_pack_to_short_fp32 (vector float,
10151 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
10152 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
10153 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
10154 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
10155 (convert_4f32_8i16): Add define_expand.
10156 * doc/extend.texi: Update the built-in documentation file for the
10157 new built-in function.
10159 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10161 * config/sparc/m8.md: New file.
10162 * config/sparc/sparc.md: Include m8.md.
10164 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10166 * config/sparc/sparc.opt: New option -mvis4b.
10167 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
10168 (sparc_option_override): Handle VIS4B.
10169 (enum sparc_builtins): Define
10170 SPARC_BUILTIN_DICTUNPACK{8,16,32},
10171 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
10172 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
10173 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
10174 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
10175 (check_constant_argument): New function.
10176 (sparc_vis_init_builtins): Define builtins
10177 __builtin_vis_dictunpack{8,16,32},
10178 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
10179 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
10180 __builtin_vis_fpcmpde{8,16,32}shl and
10181 __builtin_vis_fpcmpur{8,16,32}shl.
10182 (sparc_expand_builtin): Check that the constant operands to
10183 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
10184 constant and in range.
10185 * config/sparc/sparc-c.c (sparc_target_macros): Handle
10187 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
10188 (SPARC_IMM5_P): Likewise.
10189 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
10190 (enabled): Handle vis4b.
10191 (UNSPEC_DICTUNPACK): New unspec.
10192 (UNSPEC_FPCMPSHL): Likewise.
10193 (UNSPEC_FPUCMPSHL): Likewise.
10194 (UNSPEC_FPCMPDESHL): Likewise.
10195 (UNSPEC_FPCMPURSHL): Likewise.
10196 (cpu_feature): New CPU feature `vis4b'.
10197 (dictunpack{8,16,32}): New insns.
10198 (FPCSMODE): New mode iterator.
10199 (fpcscond): New code iterator.
10200 (fpcsucond): Likewise.
10201 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
10202 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
10203 (fpcmpde{8,16,32}{si,di}shl): Likewise.
10204 (fpcmpur{8,16,32}{si,di}shl): Likewise.
10205 * config/sparc/constraints.md: Define constraints `q' for unsigned
10206 2-bit integer constants and `t' for unsigned 5-bit integer
10208 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
10210 (imm5_operand_dictunpack16): Likewise.
10211 (imm5_operand_dictunpack32): Likewise.
10212 (imm2_operand): Likewise.
10213 * doc/invoke.texi (SPARC Options): Document -mvis4b.
10214 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
10215 ditunpack* and fpcmp*shl builtins.
10217 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10219 * config.gcc: Handle m8 in --with-{cpu,tune} options.
10220 * config.in: Add HAVE_AS_SPARC6 define.
10221 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
10223 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
10225 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
10226 (CPP_CPU_SPEC): Handle m8.
10227 (ASM_CPU_SPEC): Likewise.
10228 * config/sparc/sparc-opts.h (enum processor_type): Add
10230 * config/sparc/sparc.c (m8_costs): New struct.
10231 (sparc_option_override): Handle TARGET_CPU_m8.
10232 (sparc32_initialize_trampoline): Likewise.
10233 (sparc64_initialize_trampoline): Likewise.
10234 (sparc_issue_rate): Likewise.
10235 (sparc_register_move_cost): Likewise.
10236 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
10237 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
10238 (ASM_CPU64_DEFAULT_SPEC): Likewise.
10239 (CPP_CPU_SPEC): Handle M8.
10240 (ASM_CPU_SPEC): Likewise.
10241 (AS_M8_FLAG): Define.
10242 * config/sparc/sparc.md: Add m8 to the cpu attribute.
10243 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
10244 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
10246 * configure: Regenerate.
10247 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
10250 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10252 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
10254 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
10255 ("*movdi_insn_sp32"): Do not set v3pipe.
10256 ("*movsi_insn"): Likewise.
10257 ("*movdi_insn_sp64"): Likewise.
10258 ("*movsf_insn"): Likewise.
10259 ("*movdf_insn_sp32"): Likewise.
10260 ("*movdf_insn_sp64"): Likewise.
10261 ("*zero_extendsidi2_insn_sp64"): Likewise.
10262 ("*sign_extendsidi2_insn"): Likewise.
10263 ("*mov<VM32:mode>_insn"): Likewise.
10264 ("*mov<VM64:mode>_insn_sp64"): Likewise.
10265 ("*mov<VM64:mode>_insn_sp32"): Likewise.
10266 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
10267 ("<vlop:code><VL:mode>3"): Likewise.
10268 ("*not_<vlop:code><VL:mode>3"): Likewise.
10269 ("*nand<VL:mode>_vis"): Likewise.
10270 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
10271 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
10272 ("one_cmpl<VL:mode>2"): Likewise.
10273 ("faligndata<VM64:mode>_vis"): Likewise.
10274 ("alignaddrsi_vis"): Likewise.
10275 ("alignaddrdi_vis"): Likweise.
10276 ("alignaddrlsi_vis"): Likewise.
10277 ("alignaddrldi_vis"): Likewise.
10278 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
10279 ("bmaskdi_vis"): Likewise.
10280 ("bmasksi_vis"): Likewise.
10281 ("bshuffle<VM64:mode>_vis"): Likewise.
10282 ("cmask8<P:mode>_vis"): Likewise.
10283 ("cmask16<P:mode>_vis"): Likewise.
10284 ("cmask32<P:mode>_vis"): Likewise.
10285 ("pdistn<P:mode>_vis"): Likewise.
10286 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
10288 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10290 * config/sparc/sparc.md ("subtype"): New insn attribute.
10291 ("*wrgsr_sp64"): Set insn subtype.
10292 ("*rdgsr_sp64"): Likewise.
10293 ("alignaddrsi_vis"): Likewise.
10294 ("alignaddrdi_vis"): Likewise.
10295 ("alignaddrlsi_vis"): Likewise.
10296 ("alignaddrldi_vis"): Likewise.
10297 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
10298 ("fexpand_vis"): Likewise.
10299 ("fpmerge_vis"): Likewise.
10300 ("faligndata<VM64:mode>_vis"): Likewise.
10301 ("bshuffle<VM64:mode>_vis"): Likewise.
10302 ("cmask8<P:mode>_vis"): Likewise.
10303 ("cmask16<P:mode>_vis"): Likewise.
10304 ("cmask32<P:mode>_vis"): Likewise.
10305 ("fchksm16_vis"): Likewise.
10306 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
10307 ("fmean16_vis"): Likewise.
10308 ("fp<plusminus_insn>64_vis"): Likewise.
10309 ("<plusminus_insn>v8qi3"): Likewise.
10310 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
10311 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
10312 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
10313 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
10314 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
10315 ("*movqi_insn"): Likewise.
10316 ("*movhi_insn"): Likewise.
10317 ("*movsi_insn"): Likewise.
10318 ("movsi_pic_gotdata_op"): Likewise.
10319 ("*movdi_insn_sp32"): Likewise.
10320 ("*movdi_insn_sp64"): Likewise.
10321 ("movdi_pic_gotdata_op"): Likewise.
10322 ("*movsf_insn"): Likewise.
10323 ("*movdf_insn_sp32"): Likewise.
10324 ("*movdf_insn_sp64"): Likewise.
10325 ("*zero_extendhisi2_insn"): Likewise.
10326 ("*zero_extendqihi2_insn"): Likewise.
10327 ("*zero_extendqisi2_insn"): Likewise.
10328 ("*zero_extendqidi2_insn"): Likewise.
10329 ("*zero_extendhidi2_insn"): Likewise.
10330 ("*zero_extendsidi2_insn_sp64"): Likewise.
10331 ("ldfsr"): Likewise.
10332 ("prefetch_64"): Likewise.
10333 ("prefetch_32"): Likewise.
10334 ("tie_ld32"): Likewise.
10335 ("tie_ld64"): Likewise.
10336 ("*tldo_ldub_sp32"): Likewise.
10337 ("*tldo_ldub1_sp32"): Likewise.
10338 ("*tldo_ldub2_sp32"): Likewise.
10339 ("*tldo_ldub_sp64"): Likewise.
10340 ("*tldo_ldub1_sp64"): Likewise.
10341 ("*tldo_ldub2_sp64"): Likewise.
10342 ("*tldo_ldub3_sp64"): Likewise.
10343 ("*tldo_lduh_sp32"): Likewise.
10344 ("*tldo_lduh1_sp32"): Likewise.
10345 ("*tldo_lduh_sp64"): Likewise.
10346 ("*tldo_lduh1_sp64"): Likewise.
10347 ("*tldo_lduh2_sp64"): Likewise.
10348 ("*tldo_lduw_sp32"): Likewise.
10349 ("*tldo_lduw_sp64"): Likewise.
10350 ("*tldo_lduw1_sp64"): Likewise.
10351 ("*tldo_ldx_sp64"): Likewise.
10352 ("*mov<VM32:mode>_insn"): Likewise.
10353 ("*mov<VM64:mode>_insn_sp64"): Likewise.
10354 ("*mov<VM64:mode>_insn_sp32"): Likewise.
10356 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10358 * config/sparc/sparc.md ("type"): New insn type viscmp.
10359 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
10361 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
10362 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
10363 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
10364 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
10366 ("n7_vis_logical_11cycle"): Likewise.
10367 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
10368 * config/sparc/niagara2.md ("niag3_vis": Likewise.
10369 * config/sparc/niagara.md ("niag_vis"): Likewise.
10370 * config/sparc/ultra3.md ("us3_fga"): Likewise.
10371 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
10373 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10375 * config/sparc/sparc.md: New instruction type `bmask'.
10376 (bmaskdi_vis): Use the `bmask' type.
10377 (bmasksi_vis): Likewise.
10378 * config/sparc/ultra3.md (us3_array): Likewise.
10379 * config/sparc/niagara7.md (n7_array): Likewise.
10380 * config/sparc/niagara4.md (n4_array): Likewise.
10381 * config/sparc/niagara2.md (niag2_vis): Likewise.
10382 (niag3_vis): Likewise.
10383 * config/sparc/niagara.md (niag_vis): Likewise.
10385 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10387 * ipa-comdats.c: Remove optimize check from gate.
10388 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
10389 for functions not optimized.
10390 (ipa_fn_summary_read): Skip optimize check.
10391 (ipa_fn_summary_write): Likewise.
10392 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
10394 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
10396 (can_inline_edge_p): Check flag_pcc_struct_return for match.
10397 (check_callers): Give up on caller which is not optimized.
10398 (inline_small_functions): Likewise.
10399 (ipa_inline): Do not give up when not optimizing.
10400 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
10401 away unoptimizes cdtors.
10402 (whole_program_function_and_variable_visibility): Do
10403 ipa_discover_readonly_nonaddressable_vars in LTO mode.
10404 * ipa.c (process_references): Do not check optimize.
10405 (symbol_table::remove_unreachable_nodes): Update optimize check.
10406 (set_writeonly_bit): Update optimize check.
10407 (pass_ipa_cdtor_merge::gate): Do not check optimize.
10408 (pass_ipa_single_use::gate): Remove.
10410 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10412 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
10413 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
10414 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
10415 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
10416 permute_load, permute_store, adjust_extract, adjust_splat,
10417 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
10418 replace_swap_with_copy, dump_swap_insn_table,
10419 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
10420 recombine_lvx_pattern, recombine_stvx_pattern,
10421 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
10422 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
10423 to file rs6000-p8swap.c.
10424 * config/rs6000/rs6000-p8swap.c: New file.
10425 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
10426 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
10427 and rs6000*-*-* targets.
10429 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10431 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
10433 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10435 * lto-wrapper.c (merge_and_complain): Do not merge
10436 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
10437 fsigned_zeros, ftrapping_math, fwrapv.
10438 (append_compiler_options): Do not track these options.
10439 (append_linker_options): Likewie
10441 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10443 * cgraphunit.c (cgraph_node::finalize_function): When
10444 !flag_toplevel_reorde set no_reorder flag.
10445 (varpool_node::finalize_decl): Likewise.
10446 (symbol_table::compile): Drop no toplevel reorder path.
10448 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10450 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
10451 edges; zero probability is not better than uninitialized.
10453 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
10455 * asan.h (asan_sanitize_allocas_p): Declare.
10456 * asan.c (asan_sanitize_allocas_p): New function.
10457 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
10458 (handle_builtin_alloca): Likewise.
10459 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
10460 if !asan_sanitize_allocas_p.
10461 * params.def (asan-instrument-allocas): Add new option.
10462 * params.h (ASAN_PROTECT_ALLOCAS): Define.
10463 * opts.c (common_handle_option): Disable allocas sanitization for
10466 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
10468 * asan.c: Include gimple-fold.h.
10469 (get_last_alloca_addr): New function.
10470 (handle_builtin_stackrestore): Likewise.
10471 (handle_builtin_alloca): Likewise.
10472 (asan_emit_allocas_unpoison): Likewise.
10473 (get_mem_refs_of_builtin_call): Add new parameter, remove const
10474 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
10475 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
10476 (instrument_builtin_call): Pass gimple iterator to
10477 get_mem_refs_of_builtin_call.
10478 (last_alloca_addr): New global.
10479 * asan.h (asan_emit_allocas_unpoison): Declare.
10480 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
10481 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
10482 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
10483 if function calls alloca.
10484 * gimple-fold.c (replace_call_with_value): Remove static keyword.
10485 * gimple-fold.h (replace_call_with_value): Declare.
10486 * internal-fn.c: Include asan.h.
10487 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
10488 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
10490 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10492 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
10493 (C_SELFTEST_FLAGS): New.
10494 (CPP_SELFTEST_FLAGS): New.
10495 (SELFTEST_DEPS): New, from deps of s-selftest.
10496 (C_SELFTEST_DEPS): New, from deps of s-selftest.
10497 (CPP_SELFTEST_DEPS): New.
10498 (selftest): Add dependency on s-selftest-c++.
10499 (s-selftest): Rename to...
10500 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
10501 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
10502 than SELFTEST_FLAGS.
10503 (selftest-gdb): Rename to...
10504 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
10506 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
10507 (selftest-valgrind): Rename to...
10508 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
10510 (selftest-valgrind): Reintroduce as an alias for
10511 selftest-c-valgrind.
10512 (s-selftest-c++): New.
10513 (selftest-c++-gdb): New.
10514 (selftest-c++-valgrind): New.
10516 2017-07-06 Olivier Hainque <hainque@adacore.com>
10518 * gcc.c (process_command): When deciding if undefined variables
10519 should be ignored when processing specs, accept "gcc -v" as well.
10521 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10523 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
10524 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
10526 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
10528 * config/arm/arm-cpus.in (armv8-r): Add new entry.
10529 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
10530 * config/arm/arm-tables.opt: Regenerate.
10531 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
10533 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
10535 2017-07-06 Carl Love <cel@us.ibm.com>
10537 * ChangeLog: Clean up from mid air collision
10539 2017-07-06 Carl Love <cel@us.ibm.com>
10541 * config/rs6000/rs6000-c.c: Add support for built-in functions
10542 vector signed int vec_subc (vector signed int, vector signed int);
10543 vector signed __int128 vec_subc (vector signed __int128,
10544 vector signed __int128);
10545 vector unsigned __int128 vec_subc (vector unsigned __int128,
10546 vector unsigned __int128);
10547 vector signed int vec_sube (vector signed int, vector signed int,
10548 vector signed int);
10549 vector unsigned int vec_sube (vector unsigned int,
10550 vector unsigned int,
10551 vector unsigned int);
10552 vector signed __int128 vec_sube (vector signed __int128,
10553 vector signed __int128,
10554 vector signed__int128);
10555 vector unsigned __int128 vec_sube (vector unsigned __int128,
10556 vector unsigned __int128,
10557 vector unsigned __int128);
10558 vector signed int vec_subec (vector signed int, vector signed int,
10559 vector signed int);
10560 vector unsigned int vec_subec (vector unsigned int,
10561 vector unsigned int,
10562 vector unsigned int);
10563 vector signed __int128 vec_subec (vector signed __int128,
10564 vector signed __int128,
10565 vector signed__int128);
10566 vector unsigned __int128 vec_subec (vector unsigned __int128,
10567 vector unsigned __int128,
10568 vector unsigned __int128);
10569 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
10570 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
10571 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
10572 BU_ALTIVEC_OVERLOAD_X definitions.
10573 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
10574 * doc/extend.texi: Update the built-in documentation file for the new
10575 built-in functions.
10577 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10580 * diagnostic-show-locus.c (layout::layout): Use start and finish
10581 spelling location for the start and finish of each range.
10582 * genmatch.c (linemap_client_expand_location_to_spelling_point):
10583 Add unused aspect param.
10584 * input.c (expand_location_1): Add "aspect" param, and use it
10585 to access the correct part of the location.
10586 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
10588 (expand_location_to_spelling_point): Likewise.
10589 (linemap_client_expand_location_to_spelling_point): Add "aspect"
10590 param, and pass it to expand_location_1.
10592 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
10594 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
10595 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
10596 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
10597 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
10598 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
10599 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
10600 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
10601 _mm_maskz_getmant_ss): New intrinsics.
10602 (__builtin_ia32_getexpss128_mask): Changed to ...
10603 __builtin_ia32_getexpss128_round ... this.
10604 (__builtin_ia32_getexpsd128_mask): Changed to ...
10605 __builtin_ia32_getexpsd128_round ... this.
10606 * config/i386/i386-builtin-types.def
10607 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
10608 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
10609 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
10610 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
10611 __builtin_ia32_getmantss_mask_round): New builtins.
10612 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
10613 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
10614 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
10615 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
10616 * config/i386/sse.md
10617 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
10618 avx512f_sgetexp<mode><mask_scalar_name>
10619 <round_saeonly_scalar_name> ... this.
10620 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
10621 %0, %1, %2<round_saeonly_op3>}): Changed to ...
10622 vgetexp<ssescalarmodesuffix>
10623 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10624 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
10625 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
10626 avx512f_vgetmant<mode><mask_scalar_name>
10627 <round_saeonly_scalar_name> ... this.
10628 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
10629 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
10630 vgetmant<ssescalarmodesuffix>
10631 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
10632 %0<mask_scalar_operand4>, %1, %2
10633 <round_saeonly_scalar_mask_op4>, %3} ... this.
10634 * config/i386/subst.md (mask_scalar_operand4,
10635 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
10636 round_saeonly_scalar_nimm_predicate): New subst attributes.
10638 2017-07-06 Julia Koval <julia.koval@intel.com>
10640 * config/i386/i386.c (ix86_erase_embedded_rounding):
10641 Remove code for old rounding pattern.
10643 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
10645 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
10647 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
10649 * doc/sourcebuild.texi (Test Directives, Variants of
10650 dg-require-support): Add documentation for dg-require-stack-check.
10652 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
10654 * config/i386/subst.md (mask_scalar, round_scalar,
10655 round_saeonly_scalar): New meta-templates.
10656 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
10657 round_scalar_mask_operand3, round_scalar_mask_op3,
10658 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
10659 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
10660 round_saeonly_scalar_constraint,
10661 round_saeonly_scalar_prefix): New subst attribute.
10662 * config/i386/sse.md
10663 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
10664 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
10665 <round_scalar_name> ... this.
10666 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
10667 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
10668 <round_scalar_name> ... this.
10669 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
10670 <sse>_vm<code><mode>3<mask_scalar_name>
10671 <round_saeonly_scalar_name> ... this.
10672 (v<plusminus_mnemonic><ssescalarmodesuffix>
10673 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
10674 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
10675 v<plusminus_mnemonic><ssescalarmodesuffix>
10676 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10677 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
10678 (v<multdiv_mnemonic><ssescalarmodesuffix>
10679 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
10680 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
10681 v<multdiv_mnemonic><ssescalarmodesuffix>
10682 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10683 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
10684 (v<maxmin_float><ssescalarmodesuffix>
10685 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
10686 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
10687 v<maxmin_float><ssescalarmodesuffix>
10688 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10689 %0<mask_scalar_operand3>, %1, %<iptr>2
10690 <round_saeonly_scalar_mask_op3>} ... this.
10692 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
10694 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
10695 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
10697 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
10698 Alan Hayward <alan.hayward@arm.com>
10699 David Sherwood <david.sherwood@arm.com>
10701 * combine.c (simplify_if_then_else): Remove "enum" before
10703 * compare-elim.c (can_eliminate_compare): Likewise.
10704 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
10706 (aarch64_lookup_simd_builtin_type): Likewise.
10707 (aarch64_simd_builtin_type): Likewise.
10708 (aarch64_init_simd_builtin_types): Likewise.
10709 (aarch64_simd_expand_args): Likewise.
10710 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
10712 (aarch64_reverse_mask): Likewise.
10713 (aarch64_simd_emit_reg_reg_move): Likewise.
10714 (aarch64_gen_adjusted_ldpstp): Likewise.
10715 (aarch64_ccmp_mode_to_code): Likewise.
10716 (aarch64_operands_ok_for_ldpstp): Likewise.
10717 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10718 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
10720 (aarch64_min_divisions_for_recip_mul): Likewise.
10721 (aarch64_reassociation_width): Likewise.
10722 (aarch64_get_condition_code_1): Likewise.
10723 (aarch64_simd_emit_reg_reg_move): Likewise.
10724 (aarch64_simd_attr_length_rglist): Likewise.
10725 (aarch64_reverse_mask): Likewise.
10726 (aarch64_operands_ok_for_ldpstp): Likewise.
10727 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10728 (aarch64_gen_adjusted_ldpstp): Likewise.
10729 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
10731 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
10732 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
10733 (arm_lookup_simd_builtin_type): Likewise.
10734 (arm_simd_builtin_type): Likewise.
10735 (arm_init_simd_builtin_types): Likewise.
10736 (arm_expand_builtin_args): Likewise.
10737 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
10738 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
10739 (ft32_setup_incoming_varargs): Likewise.
10740 (ft32_function_arg): Likewise.
10741 (ft32_function_arg_advance): Likewise.
10742 (ft32_pass_by_reference): Likewise.
10743 (ft32_arg_partial_bytes): Likewise.
10744 (ft32_valid_pointer_mode): Likewise.
10745 (ft32_addr_space_pointer_mode): Likewise.
10746 (ft32_addr_space_legitimate_address_p): Likewise.
10747 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
10749 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
10750 (ix86_emit_outlined_ms2sysv_restore): Likewise.
10751 (iamcu_alignment): Likewise.
10752 (canonicalize_vector_int_perm): Likewise.
10753 (ix86_noce_conversion_profitable_p): Likewise.
10754 (ix86_mpx_bound_mode): Likewise.
10755 (ix86_operands_ok_for_move_multiple): Likewise.
10756 * config/microblaze/microblaze-protos.h
10757 (microblaze_expand_conditional_branch_reg): Likewise.
10758 * config/microblaze/microblaze.c
10759 (microblaze_expand_conditional_branch_reg): Likewise.
10760 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
10762 (rs6000_reassociation_width): Likewise.
10763 (rs6000_invalid_binary_op): Likewise.
10764 (fusion_p9_p): Likewise.
10765 (emit_fusion_p9_load): Likewise.
10766 (emit_fusion_p9_store): Likewise.
10767 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
10769 (riscv_hard_regno_mode_ok_p): Likewise.
10770 (riscv_address_insns): Likewise.
10771 (riscv_split_symbol): Likewise.
10772 (riscv_legitimize_move): Likewise.
10773 (riscv_function_value): Likewise.
10774 (riscv_hard_regno_nregs): Likewise.
10775 (riscv_expand_builtin): Likewise.
10776 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
10777 (riscv_build_integer): Likewise.
10778 (riscv_split_integer): Likewise.
10779 (riscv_legitimate_constant_p): Likewise.
10780 (riscv_cannot_force_const_mem): Likewise.
10781 (riscv_regno_mode_ok_for_base_p): Likewise.
10782 (riscv_valid_base_register_p): Likewise.
10783 (riscv_valid_offset_p): Likewise.
10784 (riscv_valid_lo_sum_p): Likewise.
10785 (riscv_classify_address): Likewise.
10786 (riscv_legitimate_address_p): Likewise.
10787 (riscv_address_insns): Likewise.
10788 (riscv_load_store_insns): Likewise.
10789 (riscv_force_binary): Likewise.
10790 (riscv_split_symbol): Likewise.
10791 (riscv_force_address): Likewise.
10792 (riscv_legitimize_address): Likewise.
10793 (riscv_move_integer): Likewise.
10794 (riscv_legitimize_const_move): Likewise.
10795 (riscv_legitimize_move): Likewise.
10796 (riscv_address_cost): Likewise.
10797 (riscv_subword): Likewise.
10798 (riscv_output_move): Likewise.
10799 (riscv_canonicalize_int_order_test): Likewise.
10800 (riscv_emit_int_order_test): Likewise.
10801 (riscv_function_arg_boundary): Likewise.
10802 (riscv_pass_mode_in_fpr_p): Likewise.
10803 (riscv_pass_fpr_single): Likewise.
10804 (riscv_pass_fpr_pair): Likewise.
10805 (riscv_get_arg_info): Likewise.
10806 (riscv_function_arg): Likewise.
10807 (riscv_function_arg_advance): Likewise.
10808 (riscv_arg_partial_bytes): Likewise.
10809 (riscv_function_value): Likewise.
10810 (riscv_pass_by_reference): Likewise.
10811 (riscv_setup_incoming_varargs): Likewise.
10812 (riscv_print_operand): Likewise.
10813 (riscv_elf_select_rtx_section): Likewise.
10814 (riscv_save_restore_reg): Likewise.
10815 (riscv_for_each_saved_reg): Likewise.
10816 (riscv_register_move_cost): Likewise.
10817 (riscv_hard_regno_mode_ok_p): Likewise.
10818 (riscv_hard_regno_nregs): Likewise.
10819 (riscv_class_max_nregs): Likewise.
10820 (riscv_memory_move_cost): Likewise.
10821 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
10822 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
10823 (rl78_addr_space_address_mode): Likewise.
10824 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10826 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
10827 (rs6000_reassociation_width): Likewise.
10828 (rs6000_invalid_binary_op): Likewise.
10829 (fusion_p9_p): Likewise.
10830 (emit_fusion_p9_load): Likewise.
10831 (emit_fusion_p9_store): Likewise.
10832 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
10833 (ok_for_simple_move_operands): Likewise.
10834 (ok_for_simple_move_strict_operands): Likewise.
10835 (ok_for_simple_arith_logic_operands): Likewise.
10836 (visium_legitimize_reload_address): Likewise.
10837 (visium_select_cc_mode): Likewise.
10838 (output_cbranch): Likewise.
10839 (visium_split_double_move): Likewise.
10840 (visium_expand_copysign): Likewise.
10841 (visium_expand_int_cstore): Likewise.
10842 (visium_expand_fp_cstore): Likewise.
10843 * config/visium/visium.c (visium_pass_by_reference): Likewise.
10844 (visium_function_arg): Likewise.
10845 (visium_function_arg_advance): Likewise.
10846 (visium_libcall_value): Likewise.
10847 (visium_setup_incoming_varargs): Likewise.
10848 (visium_legitimate_constant_p): Likewise.
10849 (visium_legitimate_address_p): Likewise.
10850 (visium_legitimize_address): Likewise.
10851 (visium_secondary_reload): Likewise.
10852 (visium_register_move_cost): Likewise.
10853 (visium_memory_move_cost): Likewise.
10854 (prepare_move_operands): Likewise.
10855 (ok_for_simple_move_operands): Likewise.
10856 (ok_for_simple_move_strict_operands): Likewise.
10857 (ok_for_simple_arith_logic_operands): Likewise.
10858 (visium_function_value_1): Likewise.
10859 (rtx_ok_for_offset_p): Likewise.
10860 (visium_legitimize_reload_address): Likewise.
10861 (visium_split_double_move): Likewise.
10862 (visium_expand_copysign): Likewise.
10863 (visium_expand_int_cstore): Likewise.
10864 (visium_expand_fp_cstore): Likewise.
10865 (visium_split_cstore): Likewise.
10866 (visium_select_cc_mode): Likewise.
10867 (visium_split_cbranch): Likewise.
10868 (output_cbranch): Likewise.
10869 (visium_print_operand_address): Likewise.
10870 * expmed.c (flip_storage_order): Likewise.
10871 * expmed.h (emit_cstore): Likewise.
10872 (flip_storage_order): Likewise.
10873 * genrecog.c (validate_pattern): Likewise.
10874 * hsa-gen.c (gen_hsa_addr): Likewise.
10875 * internal-fn.c (expand_arith_overflow): Likewise.
10876 * ira-color.c (allocno_copy_cost_saving): Likewise.
10877 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10878 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
10879 (process_invariant_for_inheritance): Likewise.
10880 * lra-eliminations.c (move_plus_up): Likewise.
10881 * omp-low.c (lower_oacc_reductions): Likewise.
10882 * simplify-rtx.c (simplify_subreg): Likewise.
10883 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
10884 (TARGET_CHKP_BOUND_MODE): Likewise..
10885 * targhooks.c (default_chkp_bound_mode): Likewise.
10886 (default_setup_incoming_vararg_bounds): Likewise.
10887 * targhooks.h (default_chkp_bound_mode): Likewise.
10888 (default_setup_incoming_vararg_bounds): Likewise.
10889 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
10890 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
10891 (have_whole_vector_shift): Likewise.
10892 * tree-vect-stmts.c (vectorizable_load): Likewise.
10893 * doc/tm.texi: Regenerate.
10895 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
10897 Graceful degrade if Binutils PR21472 is not available.
10900 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
10901 .rodata in flash test fails.
10902 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
10903 * confgure: Regenerate.
10904 * config.in: Regenerate.
10905 * config/avr/avr.c (avr_asm_named_section)
10906 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
10907 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
10908 (avr_asm_init_sections): Same.
10910 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10912 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
10913 (fma<VH:mode>4_intrinsic): Likewise.
10914 (*fmsub<VCVTF:mode>4): Likewise.
10915 (*fmsub<VH:mode>4_intrinsic): Likewise.
10917 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
10920 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
10921 Don't depend on "optimize > 0".
10922 (out_movhi_r_mr, out_movqi_mr_r): Same.
10923 (out_movhi_mr_r, out_movqi_r_mr): Same.
10924 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
10925 io_address_operand on "optimize > 0".
10927 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10929 * tree-loop-distribution.c: Add general explanantion on the pass.
10930 (generate_loops_for_partition): Mark distributed loop.
10931 (pg_add_dependence_edges): New parameter. Handle alias data
10932 dependence specially and record it in the parameter if asked.
10933 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
10934 (init_partition_graph_vertices, add_partition_graph_edge): New.
10935 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
10936 (free_partition_graph_vdata, build_partition_graph): New.
10937 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
10938 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
10939 (data_ref_segment_size, latch_dominated_by_data_ref): New.
10940 (compute_alias_check_pairs, version_loop_by_alias_check): New.
10941 (version_for_distribution_p, finalize_partitions): New.
10942 (distribute_loop): Handle alias data dependence specially. Factor
10943 out loop fusion code as functions and call these functions.
10945 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10947 * tree-loop-distribution.c (classify_partition): New parameter and
10948 better handle reduction statement.
10949 (rdg_build_partitions): Revise comment.
10950 (distribute_loop): Compute statements in all partitions and pass it
10951 to classify_partition.
10953 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10955 * tree-loop-distribution.c (enum partition_type): New.
10956 (struct partition): New field type.
10957 (partition_merge_into): Add parameter. Update partition type.
10958 (data_dep_in_cycle_p, update_type_for_merge): New functions.
10959 (build_rdg_partition_for_vertex): Compute partition type.
10960 (rdg_build_partitions): Dump partition type.
10961 (distribute_loop): Update calls to partition_merge_into.
10963 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10965 * tree-loop-distribution.c (struct ddr_hasher): New.
10966 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
10968 (classify_partition): Call get_data_dependence.
10969 (pg_add_dependence_edges): Ditto.
10970 (distribute_loop): Release data dependence hash table.
10972 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10974 * tree-loop-distribution.c (ref_base_address): Delete.
10975 (similar_memory_accesses): Rename ...
10976 (share_memory_accesses): ... to this. Check if partitions access
10977 the same memory reference.
10978 (distribute_loop): Call share_memory_accesses.
10980 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10982 * tree-loop-distribution.c (struct partition): New field recording
10983 its data reference.
10984 (partition_alloc, partition_free): Init and release data refs.
10985 (partition_merge_into): Merge data refs.
10986 (build_rdg_partition_for_vertex): Collect data refs for partition.
10987 (pg_add_dependence_edges): Change parameters from vector to bitmap.
10989 (distribute_loop): Remve data refs from vertice data of partition
10992 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10994 * tree-loop-distribution.c (params.h): Include header file.
10995 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
10996 (datarefs_vec): New global var.
10997 (create_rdg_vertices): Use datarefs_vec directly.
10998 (free_rdg): Don't free data references.
10999 (build_rdg): Update use. Don't free data references.
11000 (distribute_loop): Compute global variable for data references.
11001 Bail out if there are too many data references.
11003 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11005 * tree-loop-distribution.c (loop_nest): New global var.
11006 (build_rdg): Use loop directly, rather than loop nest.
11007 (pg_add_dependence_edges): Remove loop nest parameter. Use global
11009 (distribute_loop): Compute global variable loop nest. Update use.
11011 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11013 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
11014 (partition_merge_into): New parameter. Dump reason for fusion.
11015 (distribute_loop): Update use of partition_merge_into.
11017 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11019 * tree-loop-distribution.c (bb_top_order_index): New.
11020 (bb_top_order_index_size, bb_top_order_cmp): New.
11021 (stmts_from_loop): Use topological order.
11022 (pass_loop_distribution::execute): Compute and release topological
11023 order for basic blocks.
11025 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11027 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
11030 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11032 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
11033 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
11034 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
11035 * internal-fn.def (LOOP_DIST_ALIAS): New.
11036 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
11037 (fold_loop_internal_call): ... this.
11038 (vect_loop_dist_alias_call): New function.
11039 (set_uid_loop_bbs): Call fold_loop_internal_call.
11040 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
11043 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
11046 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
11047 Require dead FLAGS_REG at the beginning of a peephole.
11049 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
11052 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
11053 arguments in the call to __builtin_ia32_sbb_u32.
11054 (_subborrow_u64): Swap _X and _Y arguments in the call to
11055 __builtin_ia32_sbb_u64.
11057 2017-07-04 Jakub Jelinek <jakub@redhat.com>
11060 * tree-vrp.c (compare_assert_loc): Turn into a function template
11061 with stable template parameter. Only test if a->e is NULL,
11062 !a->e == !b->e has been verified already. Use e == NULL or
11063 e != NULL instead of e or ! e tests. If stable is true, don't use
11064 iterative_hash_expr, on the other side allow a or b or both NULL
11065 and sort the NULLs last.
11066 (process_assert_insertions): Sort using compare_assert_loc<false>
11067 instead of compare_assert_loc, later sort using
11068 compare_assert_loc<true> before calling process_assert_insertions_for
11069 in a loop. Use break instead of continue once seen NULL pointer.
11071 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
11073 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
11074 Cortex-R7 and Cortex-R8 processors.
11076 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
11078 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
11079 uninitialized while src is not.
11081 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
11083 * common/config/arm/arm-common.c: Adjust include path for
11085 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
11086 (arm-cpu.h): Create in build directory. Adjust dependency rules.
11087 (arm-cpu-data.h): Likewise.
11088 (arm-cpu-cdata.h): Likewise.
11089 * config/arm/arm-cpu.h: Delete.
11090 * config/arm/arm-cpu-cdata.h: Delete.
11091 * config/arm/arm-cpu-data.h: Delete.
11093 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
11095 * config/arm/arm-cpus.in (cortex-a55): New.
11096 (cortex-a75): Likewise.
11097 (cortex-a75.cortex-a55): Likewise.
11098 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
11100 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
11101 * config/arm/arm-cpu-cdata.h: Regenerate.
11102 * config/arm/arm-cpu-data.h: Regenerate.
11103 * config/arm/arm-cpu.h: Regenerate.
11104 * config/arm/arm-tables.opt: Regenerate.
11105 * config/arm/arm-tune.md: Regenerate.
11107 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
11109 * haifa-sched.c (sched_create_recovery_edges): Update profile.
11111 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
11113 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
11116 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
11118 PR tree-optimization/81292
11119 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
11120 full_string_p, also call adjust_related_strinfos if the adjustment
11121 is simple, otherwise invalidate related strinfos.
11123 2017-07-04 Martin Liska <mliska@suse.cz>
11126 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
11127 newly created variable as DECL_IGNORED_P.
11129 2017-07-04 Martin Liska <mliska@suse.cz>
11132 * ipa-inline.c (inline_small_functions):
11133 Use xstrdup_for_dump.
11135 2017-07-04 Tom de Vries <tom@codesourcery.com>
11137 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
11139 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
11142 * config/darwin.c (darwin_function_switched_text_sections):
11145 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
11147 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
11149 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
11151 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
11153 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11155 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
11156 min_profitable_iters, and th as inclusive lower bounds.
11157 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
11158 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
11159 for min_profitable_iters and min_profitable_estimate.
11160 (vect_transform_loop): Treat th as an inclusive lower bound.
11161 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
11163 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
11166 * config/darwin.c (darwin_function_switched_text_sections):
11167 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
11168 in two pieces, and suppress the use of buf.
11170 2017-07-03 Nathan Sidwell <nathan@acm.org>
11172 * hash-table.h (hash_table_mod1): Fix indentation.
11174 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11176 PR middle-end/81290
11177 * predict.c (force_edge_cold): Be more careful about propagation
11179 * profile-count.h (profile_probability::guessed,
11180 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
11182 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
11184 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
11186 * doc/invoke.texi (rcpc architecture extension): Document it.
11188 2017-07-03 Richard Biener <rguenther@suse.de>
11190 PR tree-optimization/60510
11191 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
11192 the scalar reduction PHI and use it.
11193 (vectorizable_reduction): Properly guard the single_defuse_cycle
11194 path for non-SLP reduction chains where we cannot use it.
11195 Rework reduc_def/index and vector type deduction. Rework
11196 vector operand gathering during reduction op code-gen.
11197 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
11198 chains dissolve the chain and leave it to non-SLP reduction
11201 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11203 * tree-data-ref.h (dr_alignment): Declare.
11204 * tree-data-ref.c (dr_alignment): New function.
11205 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
11206 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
11208 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
11210 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11212 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
11213 and base_misalignment fields.
11214 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
11215 * tree-data-ref.c: Include builtins.h.
11216 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
11217 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
11218 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
11219 * tree-vect-data-refs.c: Include tree-cfg.h.
11220 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
11221 fields instead of calculating an alignment here.
11222 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
11223 innermost_loop_behavior fields.
11225 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11227 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
11229 (DR_STEP_ALIGNMENT): New macro.
11230 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
11231 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
11232 (create_data_ref): Print it.
11233 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
11234 to tell whether the step preserves vector (mis)alignment.
11235 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
11236 Move the check for an integer step and generalise to all INTEGER_CST.
11237 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
11238 Print the outer step alignment.
11240 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11242 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
11243 with offset_alignment.
11244 (DR_ALIGNED_TO): Delete.
11245 (DR_OFFSET_ALIGNMENT): New macro.
11246 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
11247 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
11248 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
11249 (create_data_ref): Likewise.
11250 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
11251 (vect_analyze_data_refs): Likewise.
11252 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
11253 creating dummy innermost behavior.
11255 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11257 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
11258 with a "innermost_loop_behavior *" and refeence tree.
11259 * tree-data-ref.c (dr_analyze_innermost): Likewise.
11260 (create_data_ref): Update call accordingly.
11261 * tree-predcom.c (find_looparound_phi): Likewise.
11263 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11265 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
11266 fields with dr_wrt_vec_loop.
11267 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
11268 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
11269 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
11270 (vect_dr_behavior): New function.
11271 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
11272 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
11273 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
11274 track whether the step preserves the misalignment.
11275 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
11276 Use vect_dr_behavior.
11277 (vect_setup_realignment): Update call accordingly.
11278 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
11279 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
11280 call to vect_create_addr_base_for_vector_ref.
11281 (vect_create_cond_for_align_checks): Likewise.
11282 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
11283 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
11284 (vect_recog_mask_conversion_pattern): Likewise.
11285 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
11286 (new_stmt_vec_info): Remove redundant zeroing.
11288 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
11290 * common/config/arm/arm-common.c (arm_be8_option): New function.
11291 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
11292 (ISA_ARMv6): Add isa_bit_be8.
11293 * config/arm/arm.h (arm_be8_option): Add prototype.
11294 (BE8_SPEC_FUNCTION): New define.
11295 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
11296 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
11297 (mlittle-endian): Similarly.
11298 (mbe8, mbe32): New options.
11299 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
11300 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
11302 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11304 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
11306 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11308 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
11309 (cleanup_tree_cfg_bb): Use it.
11310 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
11312 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
11314 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11317 * loop-doloop.c (add_test): Update profile.
11319 2017-07-03 Martin Liska <mliska@suse.cz>
11322 * sanopt.c (rewrite_usage_of_param): New function.
11323 (sanitize_rewrite_addressable_params): Likewise.
11324 (pass_sanopt::execute): Call rewrite_usage_of_param.
11326 2017-07-03 Richard Biener <rguenther@suse.de>
11328 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
11329 back to using VIEW_CONVERT_EXPR.
11331 2017-07-03 Martin Liska <mliska@suse.cz>
11334 * doc/extend.texi: Document when a resolver function is
11335 generated for target_clones.
11337 2017-07-03 Martin Liska <mliska@suse.cz>
11339 * asan.c (asan_emit_stack_protection): Unpoison just red zones
11340 and shadow memory of auto variables which are subject of
11341 use-after-scope sanitization.
11342 (asan_expand_mark_ifn): Add do set only when is_poison.
11344 2016-07-03 Richard Biener <rguenther@suse.de>
11346 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
11348 (vect_force_simple_reduction): Record reduction def -> phi mapping.
11349 (vectorizable_reduction): Perform reduction PHI creation when
11350 visiting a reduction PHI and adjust and simplify code generation
11351 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
11352 (vect_transform_loop): Visit reduction PHIs.
11353 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
11354 defs into the SLP tree.
11355 (vect_build_slp_tree): Reduction defs terminate the recursion.
11356 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
11358 (vect_get_vec_defs_for_stmt_copy): Export.
11359 (vect_get_vec_defs): Likewise.
11360 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
11362 (vect_get_vec_defs_for_stmt_copy): Declare.
11363 (vect_get_vec_defs): Likewise.
11365 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11367 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
11368 parameter with a "loop" parameter and use it instead of the
11369 loop containing DR_STMT. Don't check simple_iv when doing
11370 BB analysis. Describe the two analysis modes in the comment.
11372 2017-07-03 Tom de Vries <tom@codesourcery.com>
11374 PR tree-optimization/69468
11375 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
11376 (find_same_succ_bb): Handle ignore_edge_flags.
11378 2017-07-03 Tom de Vries <tom@codesourcery.com>
11380 PR tree-optimization/81192
11381 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
11383 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
11385 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
11387 2017-07-03 Tom de Vries <tom@codesourcery.com>
11389 PR tree-optimization/81192
11390 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
11391 BB_SAME_SUCC (bb) == NULL.
11393 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11395 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
11398 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11400 * dumpfile.c: Include profile-count.h
11401 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
11403 (insert_cond_bb): Update profile.
11404 * tree-cfg.h (insert_cond_bb): Update prototype.
11405 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
11406 * tree-dump.c: Do not include tree-cfg.
11408 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11410 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
11412 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11414 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
11417 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11419 * tree-complex.c (expand_complex_div_wide): update profile.
11421 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11422 Alan Hayward <alan.hayward@arm.com>
11423 David Sherwood <david.sherwood@arm.com>
11425 * Makefile.in (MACHMODE_H): Remove insn-modes.h
11426 (CORETYPES_H): New define.
11427 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
11428 (insn-modes-inline.h, s-modes-inline-h): New rules.
11429 (generated_files): Add insn-modes-inline.h.
11430 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
11431 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
11432 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
11433 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
11434 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
11435 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
11436 (build/gencodes.o, build/genconditions.o): Likewise.
11437 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
11438 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
11439 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
11440 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
11441 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
11442 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
11443 * coretypes.h: Include everything up to real.h for generators.
11444 Include insn-modes.h first. Include wide-int-print.h after
11445 wide-int.h. Include insn-modes-inline.h and then machmode.h.
11446 * machmode.h: Don't include insn-modes.h here.
11447 * function-tests.c: Remove includes of signop.h, machmode.h,
11448 double-int.h and wide-int.h.
11450 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
11452 * optc-save-gen.awk: Likewise.
11453 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
11454 * godump.c: Remove include of wide-int-print.h.
11455 * pretty-print.h: Likewise.
11456 * wide-int-print.cc: Likewise.
11457 * wide-int.cc: Likewise.
11458 * hash-map-tests.c: Remove include of signop.h.
11459 * hash-set-tests.c: Likewise.
11460 * rtl-tests.c: Likewise.
11461 * mkconfig.sh: Remove include of machmode.h.
11462 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
11464 (emit_insn_modes_inline_h): ...this new function. Emit the code
11465 into an insn-modes-inline.h header file, adding appropriate
11466 include guards and end comments.
11467 (emit_insn_modes_c_header): Remove include of machmode.h.
11468 (emit_min_insn_modes_c_header): Include coretypes.h rather than
11470 (main): Handle -i flag and call emit_insn_modes_inline_h when
11473 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11475 * tree-ssa-strlen.c (strinfo): Rename the length field to
11476 nonzero_chars. Add a full_string_p field.
11477 (compare_nonzero_chars, zero_length_string_p): New functions.
11478 (get_addr_stridx): Add an offset_out parameter.
11479 Use compare_nonzero_chars.
11480 (get_stridx): Update accordingly. Use compare_nonzero_chars.
11481 (new_strinfo): Update after above changes to strinfo.
11482 (set_endptr_and_length): Set full_string_p.
11483 (get_string_length): Update after above changes to strinfo.
11484 (unshare_strinfo): Update call to new_strinfo.
11485 (maybe_invalidate): Likewise.
11486 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
11487 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
11488 as a uhwi instead of an shwi. Update after above changes to
11489 strinfo and new_strinfo.
11490 (zero_length_string): Assert that chainsi contains full strings.
11491 Use zero_length_string_p. Update call to new_strinfo.
11492 (adjust_related_strinfos): Update after above changes to strinfo.
11493 Copy full_string_p from origsi.
11494 (adjust_last_stmt): Use zero_length_string_p.
11495 (handle_builtin_strlen): Update after above changes to strinfo and
11496 new_strinfo. Install the lhs as the string length if the previous
11497 entry didn't describe a full string.
11498 (handle_builtin_strchr): Update after above changes to strinfo
11500 (handle_builtin_strcpy): Likewise.
11501 (handle_builtin_strcat): Likewise.
11502 (handle_builtin_malloc): Likewise.
11503 (handle_pointer_plus): Likewise.
11504 (handle_builtin_memcpy): Likewise. Track nonzero characters
11505 that aren't necessarily followed by a nul terminator.
11506 (handle_char_store): Likewise.
11508 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11510 PR tree-optimization/80769
11511 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
11512 for malloc and calloc. Document the new invariant that all related
11513 strinfos have delayed lengths or none do.
11514 (verify_related_strinfos): Move earlier in file.
11515 (set_endptr_and_length): New function, split out from...
11516 (get_string_length): ...here. Also set the lengths of related
11518 (zero_length_string): Assert that chainsi has known (rather than
11520 (adjust_related_strinfos): Likewise.
11522 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11524 PR tree-optimization/81136
11525 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
11526 assert that two references with the same misalignment have the same
11527 compile-time misalignment if those compile-time misalignments
11530 2017-07-01 Andi Kleen <ak@linux.intel.com>
11532 * print-tree.c (print_node): Print all attributes.
11534 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11536 * cfg.c (scale_bbs_frequencies): New function.
11537 * cfg.h (scale_bbs_frequencies): Declare it.
11538 * cfgloopanal.c (single_likely_exit): Cleanup.
11539 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
11541 (scale_loop_profile): Likewise.
11542 (loop_version): Likewise.
11543 (create_empty_loop_on_edge): Update.
11544 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
11545 scale_loop_frequencies, scale_loop_profile, loopify,
11546 loop_version): Update prototypes.
11547 * modulo-sched.c (sms_schedule): Update.
11548 * predict.c (unlikely_executed_edge_p): Also check probability.
11549 (probably_never_executed_edge_p): Fix typo.
11550 * tree-if-conv.c (version_loop_for_if_conversion): Update.
11551 * tree-parloops.c (gen_parallel_loop): Update.
11552 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
11553 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11554 * tree-ssa-loop-split.c (split_loop): Update.
11555 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
11556 * tree-vect-loop-manip.c (vect_do_peeling): Update.
11557 (vect_loop_versioning): Update.
11558 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11560 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11562 * trans-mem.c (split_bb_make_tm_edge): Update profile.
11564 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11566 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
11567 to keep profile consistent.
11569 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11571 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
11572 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
11573 * profile-count.h (max_safe_multiplier): Make unsigned.
11574 (profile_count::guessed_zero): New.
11576 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11578 * bb-reorder.c (fix_up_crossing_landing_pad,
11579 fix_crossing_conditional_branches): Use make_single_succ_edge
11580 to keep profile consistent.
11582 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11584 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
11587 2017-07-01 Jakub Jelinek <jakub@redhat.com>
11590 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
11591 the right scopes, make sure cond_jump isn't preserved between multiple
11592 iterations. Search for fallthru edge whenever there are 3+ edges and
11593 use find_fallthru_edge for it.
11595 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11597 Patch by Alexander Monakov <amonakov@ispras.ru>
11598 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
11599 probabilities consistently.
11601 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11603 * pa.c (pa_expand_compare_and_swap_loop): Update call of
11604 emit_cmp_and_jump_insns.
11606 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11609 * tree-inline.c (expand_call_inline): Combine profile statuses.
11611 2017-06-30 Andrew Pinski <apinski@cavium.com>
11613 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
11614 fold_stmt returned true.
11616 2017-06-30 Nathan Sidwell <nathan@acm.org>
11618 * ggc.h (empty_string): Delete.
11619 * cfgexpand.c (expand_asm_stmt): Use plain "".
11620 * optabs.c (expand_asm_memory_barrier): Likewise.
11621 * stringpool.c (empty_string): Delete.
11622 (digit_vector, digit_string): Delete.
11623 (ggc_alloc_string): Use plain "", don't optimize single digit
11624 strings. Use ggc_alloc_atomic.
11626 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
11628 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
11629 comparison set and one other set, use the cost of the non-comparison
11632 2017-06-30 Nathan Sidwell <nathan@acm.org>
11634 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
11637 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
11639 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
11640 loops. Remove now unneeded calls to gimple_switch_set_label() that
11641 just set removed labels to NULL_TREE.
11643 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
11645 * tree-ssanames.c (set_range_info_raw): Abstract from ...
11646 (set_range_info): ...here. Only call set_range_info_raw if domain
11648 (set_nonzero_bits): Call set_range_info_raw.
11649 * tree-ssanames.h (set_range_info_raw): New.
11651 2017-06-30 Jakub Jelinek <jakub@redhat.com>
11654 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
11655 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
11656 of nonimmediate_operand and <store_mask_constraint> instead of m
11657 for the input operand. For V8FI iterator, always split if input
11658 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
11659 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
11660 <store_mask_predicate> instead of register_operand and
11661 <store_mask_constraint> instead of v for the input operand. Make
11662 sure both operands aren't MEMs for if not <mask_applied>.
11664 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
11666 * lto-wrapper.c (copy_file) Close both file descriptors before
11669 2017-06-30 Martin Liska <mliska@suse.cz>
11672 * multiple_target.c (create_dispatcher_calls): Make ifunc
11673 also for function that don't have calls or are not referenced.
11675 2017-06-30 Richard Biener <rguenther@suse.de>
11677 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
11678 analyze the first scalar stmt. Move vector type computation
11679 for the BB case here from ...
11680 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
11681 live operation processing in the SLP case properly.
11683 2017-06-30 Richard Biener <rguenther@suse.de>
11685 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
11687 2017-06-30 Martin Liska <mliska@suse.cz>
11690 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
11691 before BUILT_IN_UNWIND_RESUME when ASAN is used.
11693 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
11695 * doc/invoke.texi (AArch64): Add missing options and remove redundant
11698 2017-06-30 Richard Biener <rguenther@suse.de>
11700 PR tree-optimization/81249
11701 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
11702 condition reduction result to original scalar type.
11704 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11706 * profile-count.h (enum profile_quality): Fix typos and whitespace
11709 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11711 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
11712 type for branch probabilities.
11714 2017-06-29 Julian Brown <julian@codesourcery.com>
11715 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11717 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
11718 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
11719 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
11720 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
11722 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11724 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
11725 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
11726 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
11727 CC usage from generic code to here.
11728 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
11729 CC usage into the target macros.
11731 2017-06-29 Maya Rashish <coypu@sdf.org>
11733 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
11736 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11738 * arm/arm-builtins.c: Include profile-count.h
11739 * except.c (sjlj_emit_function_enter): Use
11740 profile_probability::unlikely.
11742 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11744 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
11745 and tocrel_offset be pointer args rather than implicitly using
11747 (legitimate_constant_pool_address_p, rs6000_emit_move,
11748 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
11749 tocrel_offset and use in toc_relative_expr_p call.
11750 (print_operand, print_operand_address): Use static tocrel_base_oac
11751 and tocrel_offset_oac.
11752 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
11755 2017-06-29 Maya Rashish <coypu@sdf.org>
11757 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
11759 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
11761 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
11762 objects, take into account only the alignment of 'op0' and 'mode1' if
11765 2017-06-29 Steve Ellcey <sellcey@cavium.com>
11767 * ccmp.c (ccmp_tree_comparison_p): New function.
11768 (ccmp_candidate_p): Update to use above function.
11769 (get_compare_parts): New function.
11770 (expand_ccmp_next): Update to use new functions.
11771 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
11773 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
11774 take mode as argument.
11775 * ccmp.h (expand_ccmp_expr): Add mode as argument.
11776 * expr.c (expand_expr_real_1): Pass mode as argument.
11778 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
11780 * combine.c (combine_instructions): Print insns to dump_file, together
11783 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11785 * asan.c (asan_emit_stack_protection): Update.
11786 (create_cond_insert_point): Update.
11787 * auto-profile.c (afdo_propagate_circuit): Update.
11788 * basic-block.h (struct edge_def): Turn probability to
11789 profile_probability.
11790 (EDGE_FREQUENCY): Update.
11791 * bb-reorder.c (find_traces_1_round): Update.
11792 (better_edge_p): Update.
11793 (sanitize_hot_paths): Update.
11794 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
11795 (make_single_succ_edge): Update.
11796 (check_bb_profile): Update.
11797 (dump_edge_info): Update.
11798 (update_bb_profile_for_threading): Update.
11799 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
11800 probabilitycount to 0.
11801 * cfgbuild.c (compute_outgoing_frequencies): Update.
11802 * cfgcleanup.c (try_forward_edges): Update.
11803 (outgoing_edges_match): Update.
11804 (try_crossjump_to_edge): Update.
11805 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
11806 (expand_gimple_tailcall): Update.
11807 (construct_init_block): Use make_single_succ_edge.
11808 (construct_exit_block): Use make_single_succ_edge.
11809 * cfghooks.c (verify_flow_info): Update.
11810 (redirect_edge_succ_nodup): Update.
11811 (split_edge): Update.
11812 (account_profile_record): Update.
11813 * cfgloopanal.c (single_likely_exit): Update.
11814 * cfgloopmanip.c (scale_loop_profile): Update.
11815 (set_zero_probability): Remove.
11816 (duplicate_loop_to_header_edge): Update.
11817 * cfgloopmanip.h (loop_version): Update prototype.
11818 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
11819 (force_nonfallthru_and_redirect): Update.
11820 (update_br_prob_note): Update.
11821 (rtl_verify_edges): Update.
11822 (purge_dead_edges): Update.
11823 (rtl_lv_add_condition_to_bb): Update.
11824 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
11825 * cgraphunit.c (init_lowered_empty_function): Update.
11826 (cgraph_node::expand_thunk): Update.
11827 * cilk-common.c: Include profile-count.h
11828 * dojump.c (inv): Remove.
11829 (jumpifnot): Update.
11830 (jumpifnot_1): Update.
11831 (do_jump_1): Update.
11833 (do_jump_by_parts_greater_rtx): Update.
11834 (do_compare_rtx_and_jump): Update.
11835 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
11836 do_jump_1. do_compare_rtx_and_jump): Update prototype.
11837 * dwarf2cfi.c: Include profile-count.h
11838 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
11839 (sjlj_emit_dispatch_table): Likewise.
11840 * explow.c: Include profile-count.h
11841 * expmed.c (emit_store_flag_force): Update.
11842 (do_cmp_and_jump): Update.
11843 * expr.c (compare_by_pieces_d::generate): Update.
11844 (compare_by_pieces_d::finish_mode): Update.
11845 (emit_block_move_via_loop): Update.
11846 (store_expr_with_bounds): Update.
11847 (store_constructor): Update.
11848 (expand_expr_real_2): Update.
11849 (expand_expr_real_1): Update.
11850 * expr.h (try_casesi, try_tablejump): Update prototypes.
11851 * gimple-pretty-print.c (dump_probability): Update.
11852 (dump_profile): New.
11853 (dump_gimple_label): Update.
11854 (dump_gimple_bb_header): Update.
11855 * graph.c (draw_cfg_node_succ_edges): Update.
11856 * hsa-gen.c (convert_switch_statements): Update.
11857 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
11858 (find_if_case_1): Update.
11859 (find_if_case_2): Update.
11860 * internal-fn.c (expand_arith_overflow_result_store): Update.
11861 (expand_addsub_overflow): Update.
11862 (expand_neg_overflow): Update.
11863 (expand_mul_overflow): Update.
11864 (expand_vector_ubsan_overflow): Update.
11865 * ipa-cp.c (good_cloning_opportunity_p): Update.
11866 * ipa-split.c (split_function): Use make_single_succ_edge.
11867 * ipa-utils.c (ipa_merge_profiles): Update.
11868 * loop-doloop.c (add_test): Update.
11869 (doloop_modify): Update.
11870 * loop-unroll.c (compare_and_jump_seq): Update.
11871 (unroll_loop_runtime_iterations): Update.
11872 * lra-constraints.c (lra_inheritance): Update.
11873 * lto-streamer-in.c (input_cfg): Update.
11874 * lto-streamer-out.c (output_cfg): Update.
11875 * mcf.c (adjust_cfg_counts): Update.
11876 * modulo-sched.c (sms_schedule): Update.
11877 * omp-expand.c (expand_omp_for_init_counts): Update.
11878 (extract_omp_for_update_vars): Update.
11879 (expand_omp_ordered_sink): Update.
11880 (expand_omp_for_ordered_loops): Update.
11881 (expand_omp_for_generic): Update.
11882 (expand_omp_for_static_nochunk): Update.
11883 (expand_omp_for_static_chunk): Update.
11884 (expand_cilk_for): Update.
11885 (expand_omp_simd): Update.
11886 (expand_omp_taskloop_for_outer): Update.
11887 (expand_omp_taskloop_for_inner): Update.
11888 * omp-simd-clone.c (simd_clone_adjust): Update.
11889 * optabs.c (expand_doubleword_shift): Update.
11890 (expand_abs): Update.
11891 (emit_cmp_and_jump_insn_1): Update.
11892 (expand_compare_and_swap_loop): Update.
11893 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
11894 * predict.c (predictable_edge_p): Update.
11895 (edge_probability_reliable_p): Update.
11896 (set_even_probabilities): Update.
11897 (combine_predictions_for_insn): Update.
11898 (combine_predictions_for_bb): Update.
11899 (propagate_freq): Update.
11900 (estimate_bb_frequencies): Update.
11901 (force_edge_cold): Update.
11902 * profile-count.c (profile_count::dump): Add missing space into dump.
11903 (profile_count::debug): Add newline.
11904 (profile_count::differs_from_p): Explicitly convert to unsigned.
11905 (profile_count::stream_in): Update.
11906 (profile_probability::dump): New member function.
11907 (profile_probability::debug): New member function.
11908 (profile_probability::differs_from_p): New member function.
11909 (profile_probability::differs_lot_from_p): New member function.
11910 (profile_probability::stream_in): New member function.
11911 (profile_probability::stream_out): New member function.
11912 * profile-count.h (profile_count_quality): Rename to ...
11913 (profile_quality): ... this one.
11914 (profile_probability): New.
11915 (profile_count): Update.
11916 * profile.c (compute_branch_probabilities): Update.
11917 * recog.c (peep2_attempt): Update.
11918 * sched-ebb.c (schedule_ebbs): Update.
11919 * sched-rgn.c (find_single_block_region): Update.
11920 (compute_dom_prob_ps): Update.
11921 (schedule_region): Update.
11922 * sel-sched-ir.c (compute_succs_info): Update.
11923 * stmt.c (struct case_node): Update.
11924 (do_jump_if_equal): Update.
11925 (get_outgoing_edge_probs): Update.
11926 (conditional_probability): Update.
11927 (emit_case_dispatch_table): Update.
11928 (expand_case): Update.
11929 (expand_sjlj_dispatch_table): Update.
11930 (emit_case_nodes): Update.
11931 * targhooks.c: Update.
11932 * tracer.c (better_p): Update.
11933 (find_best_successor): Update.
11934 * trans-mem.c (expand_transaction): Update.
11935 * tree-call-cdce.c: Update.
11936 * tree-cfg.c (gimple_split_edge): Upate.
11937 (move_sese_region_to_fn): Upate.
11938 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
11939 * tree-eh.c (lower_resx): Upate.
11940 (cleanup_empty_eh_move_lp): Upate.
11941 * tree-if-conv.c (version_loop_for_if_conversion): Update.
11942 * tree-inline.c (copy_edges_for_bb): Update.
11943 (copy_cfg_body): Update.
11944 * tree-parloops.c (gen_parallel_loop): Update.
11945 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
11946 (gimple_gen_time_profiler): Update.
11947 * tree-ssa-dce.c (remove_dead_stmt): Update.
11948 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11949 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11950 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11951 (unloop_loops): Update.
11952 (try_peel_loop): Update.
11953 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11954 * tree-ssa-loop-split.c (connect_loops): Update.
11955 (split_loop): Update.
11956 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
11957 (hoist_guard): Update.
11958 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11959 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11960 (value_replacement): Update.
11961 * tree-ssa-reassoc.c (branch_fixup): Update.
11962 * tree-ssa-tail-merge.c (replace_block_by): Update.
11963 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11964 (create_edge_and_update_destination_phis): Update.
11965 (compute_path_counts): Update.
11966 (recompute_probabilities): Update.
11967 (update_joiner_offpath_counts): Update.
11968 (freqs_to_counts_path): Update.
11969 (duplicate_thread_path): Update.
11970 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11971 (struct switch_conv_info): Update.
11972 (gen_inbound_check): Update.
11973 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11974 (vect_do_peeling): Update.
11975 (vect_loop_versioning): Update.
11976 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11977 (optimize_mask_stores): Update.
11978 * ubsan.c (ubsan_expand_null_ifn): Update.
11979 * value-prof.c (gimple_divmod_fixed_value): Update.
11980 (gimple_divmod_fixed_value_transform): Update.
11981 (gimple_mod_pow2): Update.
11982 (gimple_mod_pow2_value_transform): Update.
11983 (gimple_mod_subtract): Update.
11984 (gimple_mod_subtract_transform): Update.
11985 (gimple_ic): Update.
11986 (gimple_stringop_fixed_value): Update.
11987 (gimple_stringops_transform): Update.
11988 * value-prof.h: Update.
11990 2017-06-29 Carl Love <cel@us.ibm.com>
11992 * config/rs6000/rs6000-c.c: Add support for built-in functions
11993 vector signed int vec_signed (vector float);
11994 vector signed long long vec_signed (vector double);
11995 vector signed int vec_signed2 (vector double, vector double);
11996 vector signed int vec_signede (vector double);
11997 vector signed int vec_signedo (vector double);
11998 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
11999 instruction generator.
12000 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
12001 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
12002 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
12004 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
12005 vunsignede_v2df): Add define_expands.
12006 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
12007 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
12008 VEC_UNSIGNEDO): Add definitions.
12009 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
12010 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
12011 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
12012 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
12013 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
12014 * config/rs6000/altivec.h (vec_signed, vec_signed2,
12015 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
12016 vec_unsignede, vec_unsignedo): Add builtin defines.
12017 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
12019 * doc/extend.texi: Update the built-in documentation file for the
12020 new built-in functions.
12022 2017-06-29 Richard Biener <rguenther@suse.de>
12024 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
12025 reduction chains to LOOP_VINFO_REDUCTIONS.
12026 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
12027 SLP reductions after processing reduction chains.
12029 2017-06-29 Nathan Sidwell <nathan@acm.org>
12031 * builtins.c (fold_builtin_FUNCTION): Use
12032 lang_hooks.decl_printable_name.
12034 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
12036 PR middle-end/81194
12037 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
12038 with only one label.
12039 * stmt.c (expand_case): Assert NCASES is greater than one.
12041 2017-06-29 Richard Biener <rguenther@suse.de>
12043 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
12045 (group_case_labels): Likewise.
12046 (find_taken_edge): Push sanity checking on val to workers...
12047 (find_taken_edge_cond_expr): ... here
12048 (find_taken_edge_switch_expr): ... and here, handle cases
12049 with just a default label.
12050 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
12051 (group_case_labels): Likewise.
12052 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
12053 group_case_labels does anything cleanup the CFG again.
12055 2017-06-29 Bin Cheng <bin.cheng@arm.com>
12057 PR tree-optimization/81196
12058 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
12059 exit condition comparing two IVs.
12061 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
12063 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
12064 profile to the dummy entry at the end of the list of architectures.
12065 * config/arm/arm-cpu-cdata.h: Regenerated.
12067 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12068 Michael Collison <michael.collison@arm.com>
12071 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
12073 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
12074 (*aarch64_reg_<mode>3_minus_mask): New pattern.
12075 (*aarch64_<optab>_reg_di3_mask2): New pattern.
12076 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
12077 of shift when the shift amount is masked with constant equal to
12078 the size of the mode.
12079 * config/aarch64/predicates.md (subreg_lowpart_operator): New
12082 2017-06-29 Martin Liska <mliska@suse.cz>
12084 * config/i386/i386.opt: Change range from [1,5] to [0,5].
12086 2017-06-29 Yury Gribov <tetra2005@gmail.com>
12089 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
12092 (edge_growth_cache_entry::edge_growth_cache_entry): New
12094 (reset_edge_growth_cache): Update to use constructor.
12096 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12098 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
12099 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
12100 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
12102 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
12104 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
12105 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
12107 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
12109 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
12110 (*-linux-uclibc*): Add t-uclibc tmake_file.
12111 * config/t-musl: New.
12112 * config/t-uclibc: New.
12114 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
12116 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
12118 (gen_comm_data): Emit architectural setting of arch_prof.
12119 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
12121 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
12122 (armv8-m.base, armv8-m.main): Likewise.
12123 * arm-protos.h (arm_build_target): Add profile field.
12124 (arch_option): Likewise.
12125 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
12127 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
12128 arm_active_target.profile.
12130 2017-06-28 Richard Biener <rguenther@suse.de>
12132 PR middle-end/81227
12133 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
12134 TYPE_OVERFLOW_WRAPS.
12135 * match.pd (negate_expr_p): Likewise.
12136 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
12137 fold_build2, not fold_binary.
12139 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
12141 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
12142 Convert memory address to Pmode.
12143 (aarch64_print_operand): Assert MEM operands are always Pmode.
12145 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
12148 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
12149 Remove redundant if.
12150 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
12151 * config/arm/aarch-common-protos.h
12152 (aarch_forward_to_shift_is_not_shifted_re): Remove.
12153 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
12155 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
12158 * multiple_target.c (create_dispatcher_calls): Set the default
12159 clone to be static, not public.
12161 2017-06-28 Richard Biener <rguenther@suse.de>
12163 * tree-vect-loop.c (vectorizable_reduction): Move special
12164 cond reduction IV var creation ...
12165 (vect_create_epilog_for_reduction): ... here. Remove induction_index
12166 parameter. Use STMT_VINFO_VECTYPE.
12167 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
12170 2017-06-28 Martin Liska <mliska@suse.cz>
12173 * ipa-visibility.c (non_local_p): Handle visibility.
12175 2017-06-28 Martin Liska <mliska@suse.cz>
12178 * common.opt: Add IntegerRange to various options.
12179 * opt-functions.awk (integer_range_info): New function.
12180 * optc-gen.awk: Add integer_range_info to cl_options struct.
12181 * opts-common.c (decode_cmdline_option): Handle
12182 CL_ERR_INT_RANGE_ARG.
12183 (cmdline_handle_error): Likewise.
12184 * opts.c (print_filtered_help): Show valid interval in
12185 when --help is provided.
12186 * opts.h (struct cl_option): Add range_min and range_max fields.
12187 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
12189 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
12191 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
12192 (x * C EQ/NE y * C): New transformation.
12194 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
12196 * genmultilib (combination_space): Accept '+' in option names.
12198 2017-06-28 Martin Liska <mliska@suse.cz>
12201 * asan.c (instrument_derefs): Bail out inner references
12202 that are hard register variables.
12204 2017-06-28 Jakub Jelinek <jakub@redhat.com>
12207 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
12208 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
12210 2017-06-28 Richard Biener <rguenther@suse.de>
12212 * tree-vectorizer.h (vect_get_vec_defs): Remove.
12213 (vect_get_slp_defs): Adjust.
12214 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
12216 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
12218 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
12219 get_initial_defs_for_reduction instead of vect_get_vec_defs.
12220 (vectorizable_reduction): Adjust.
12221 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
12223 (vect_get_slp_defs): Likewise.
12224 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
12225 (vectorizable_bswap): Adjust.
12226 (vectorizable_call): Likewise.
12227 (vectorizable_conversion): Likewise.
12228 (vectorizable_assignment): Likewise.
12229 (vectorizable_shift): Likewise.
12230 (vectorizable_operation): Likewise.
12231 (vectorizable_store): Likewise.
12232 (vectorizable_condition): Likewise.
12233 (vectorizable_comparison): Likewise.
12235 2017-06-28 Michael Collison <michael.collison@arm.com>
12238 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
12240 (arm_gen_movmemqi): Removed unused variable 'i'.
12241 Convert 'for' loop into 'while' loop.
12242 (arm_expand_prologue): Remove last unnecessary set of insn.
12243 (thumb_pop): Remove unused variable 'pushed_words'.
12244 (thumb_exit): Remove last unnecessary set of regs_to_pop.
12246 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12248 * config/s390/predicates.md: Use s390_rel_address_ok_p.
12249 * config/s390/s390-protos.h: Add prototype of
12250 s390_rel_address_ok_p.
12251 * config/s390/s390.c (s390_got_symbol): New function.
12252 (s390_rel_address_ok_p): New function.
12253 (legitimize_pic_address): Use s390_rel_address_ok_p.
12254 (s390_load_got): Use s390_got_symbol.
12255 (s390_option_override): Issue error if
12256 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
12257 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
12259 * config/s390/s390.opt: New option mpic-data-is-text-relative.
12261 2017-06-27 Andrew Pinski <apinski@cavium.com>
12263 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
12264 (X * copysign (1.0, X)): New pattern.
12265 (X * copysign (1.0, -X)): New pattern.
12266 (copysign (-1.0, CST)): New pattern.
12268 2017-06-27 Joseph Myers <joseph@codesourcery.com>
12270 * genmultilib (combination_space): Remove variable.
12271 Validate reuse rules against regular expression for any sequence
12272 of multilib options in any order.
12274 2017-06-27 Michael Collison <michael.collison@arm.com>
12276 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
12277 call aarch64_split_simd_combine.
12278 * (aarch64_combine_internal<mode>): Delete pattern.
12279 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
12280 Allow register and subreg operands.
12282 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12284 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
12285 specific need, just fallback on defaults.
12286 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
12288 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12289 Olivier Hainque <hainque@adacore.com>
12291 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
12293 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
12294 targets. Pick a default if no particular attempt applied.
12295 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
12298 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12300 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
12301 (x86_64-wrs-vxworks7): Likewise.
12303 2017-06-27 Marek Polacek <polacek@redhat.com>
12306 * ubsan.c (instrument_null): Check get_base_address's result for null.
12308 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
12310 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
12312 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
12314 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
12315 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
12316 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
12317 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
12318 New function types.
12319 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
12320 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
12321 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
12322 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
12323 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
12324 BUILT_IN_FEUPDATEENV): New builtins.
12325 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
12326 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
12327 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
12328 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
12330 (builtin_structptr_types): Adjust size.
12331 * tree.c (builtin_structptr_types): Add four entries.
12333 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12334 Olivier Hainque <hainque@adacore.com>
12336 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
12337 (TLS_SYM): New local macro, forcing reference to __tls__ on
12338 link command lines for VxWorks 7 RTPs, triggering initialization
12340 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
12341 OS features TLS support, true for RTPs on VxWorks 7.
12342 * config/vxworks.c (vxworks_override_options): Setup emutls
12345 2017-06-27 Jakub Jelinek <jakub@redhat.com>
12347 * predict.c (test_prediction_value_range): Use -1U instead of -1
12348 to avoid narrowing conversion warning.
12349 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
12350 to avoid narrowing conversion warning.
12351 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
12353 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
12355 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12357 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
12358 64bit configurations.
12359 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
12360 (SIZE_TYPE): Likewise.
12361 * config/vxworks.c (vxworks_emutls_var_fields): Use
12362 long_unsigned_type_node instead of unsigned_type_node as the offset
12363 field type, which is "pointer" mode in emutls.c.
12365 2017-06-27 Jakub Jelinek <jakub@redhat.com>
12368 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
12370 PR middle-end/81207
12371 * gimple-fold.c (replace_call_with_call_and_fold): Handle
12372 gimple_vuse copying separately from gimple_vdef copying.
12374 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12376 * value-prof.c (free_hist): Remove call to memset and the enclosing if
12379 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
12380 Olivier Hainque <hainque@adacore.com>
12382 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
12383 for all vxworks7 targets.
12384 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
12385 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
12386 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
12387 variations for VX6/VX7 and 32/64bits later on in ...
12388 (VXWORKS_LIB_SPEC): Leverage new macros.
12389 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
12390 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
12392 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
12394 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
12395 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
12397 2017-06-26 Carl Love <cel@us.ibm.com>
12399 * config/rs6000/rs6000-c.c: Add support for built-in functions
12400 vector bool char vec_reve (vector bool char);
12401 vector signed char vec_reve (vector signed char);
12402 vector unsigned char vec_reve (vector unsigned char);
12403 vector bool int vec_reve (vector bool int);
12404 vector signed int vec_reve (vector signed int);
12405 vector unsigned int vec_reve (vector unsigned int);
12406 vector bool long long vec_reve (vector bool long long);
12407 vector signed long long vec_reve (vector signed long long);
12408 vector unsigned long long vec_reve (vector unsigned long long);
12409 vector bool short vec_reve (vector bool short);
12410 vector signed short vec_reve (vector signed short);
12411 vector double vec_reve (vector double);
12412 vector float vec_reve (vector float);
12413 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
12414 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
12415 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
12416 (altivec_vreve): New pattern.
12417 * config/rs6000/altivec.h (vec_reve): New define.
12418 * doc/extend.texi (vec_rev): Update the built-in documentation file
12419 for the new built-in functions.
12421 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12423 PR tree-optimization/71815
12424 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
12426 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
12428 (slsr_process_phi): Likewise.
12429 (replace_uncond_cands_and_profitable_phis): Don't replace a
12430 multiply candidate with a stride of 1 (copy or cast).
12431 (phi_incr_cost): Call uses_consumed_by_stmt rather than
12433 (lowest_cost_path): Likewise.
12434 (total_savings): Likewise.
12436 2017-06-26 Richard Biener <rguenther@suse.de>
12439 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
12440 Use def_builtin_pure for all gather builtins.
12442 2017-06-26 Richard Biener <rguenther@suse.de>
12444 PR tree-optimization/81203
12445 * tree-tailcall.c (find_tail_calls): Do not move stmts into
12446 non-dominating BBs.
12448 2017-06-26 Marek Polacek <polacek@redhat.com>
12451 * doc/invoke.texi: Document -Wmultistatement-macros.
12453 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
12455 * doc/sourcebuild.texi (ARM-specific attributes): Document new
12456 arm_neon_ok_no_float_abi effective target.
12458 2017-06-26 Richard Biener <rguenther@suse.de>
12460 PR tree-optimization/80928
12461 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
12462 (copy_bbs): Set BB_DUPLICATED flag early.
12463 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
12465 (execute_on_shrinking_pred): Likewise.
12466 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
12467 BB_DUPLICATED blocks.
12468 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
12469 iterate over all PHIs considering removal of *gsi.
12471 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
12473 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
12476 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12478 * config/rs6000/rs6000-string.c: (expand_block_clear,
12479 do_load_for_compare, select_block_compare_mode,
12480 compute_current_alignment, expand_block_compare,
12481 expand_strncmp_align_check, expand_strn_compare,
12482 expand_block_move, rs6000_output_load_multiple)
12483 Move functions related to string/block move/compare
12484 to a separate file.
12485 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
12486 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
12487 for this function which is now used in two files.
12488 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
12489 * config.gcc: Add rs6000-string.o to extra_objs for
12490 targets powerpc*-*-* and rs6000*-*-*.
12492 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
12495 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
12496 32-bit, since indexed is not valid for DImode.
12497 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
12498 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
12499 (define_peephole2 for Altivec d-form load): Add 32-bit support.
12500 (define_peephole2 for Altivec d-form store): Likewise.
12503 * multiple_target.c (create_dispatcher_calls): Only create the
12504 dispatcher call if the function is the default clone of a
12505 versioned function.
12507 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
12509 PR middle-end/80902
12510 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
12511 a call, force the call to not be a tail call.
12513 2017-06-23 Jeff Law <law@redhat.com>
12515 * doc/contrib.texi: Add entry for Steven Pemberton's work on
12518 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
12520 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
12521 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
12522 handling for early expansion of vector shifts (sl,sr,sra,rl).
12523 (builtin_function_type): Add vector shift right instructions
12524 to the unsigned argument list.
12526 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
12528 rtl-optimizatoin/79286
12529 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
12530 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
12531 trap. PIC register plus a const unspec without offset can never trap.
12533 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
12535 * tree.h (builtin_structptr_type): New type.
12536 (builtin_structptr_types): Declare new array.
12537 * tree.c (builtin_structptr_types): New array.
12538 (free_lang_data, build_common_tree_nodes): Use it.
12540 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
12543 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
12546 2017-06-22 Matt Turner <mattst88@gmail.com>
12548 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
12549 Lake models to skylake case. Assume skylake for unknown
12550 models with clflushopt.
12552 2017-06-22 Jeff Law <law@redhat.com>
12554 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
12555 frame sizes that do not satisfy aarch64_uimm12_shift.
12557 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
12559 * profile-count.h (apply_probability,
12560 apply_scale, probability_in): Fix checks for zero.
12562 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12564 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
12565 * doc/cppdiropts.texi (-I @var{dir}): Document it.
12567 2016-06-22 Richard Biener <rguenther@suse.de>
12569 * tree-vect-loop.c (vect_model_reduction_cost): Handle
12570 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
12571 REDUC_MAX_EXPR support.
12572 (vectorizable_reduction): Likewise.
12573 (vect_create_epilog_for_reduction): Likewise.
12575 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
12577 * match.pd (A / (1 << B) -> A >> B): New.
12578 * generic-match-head.c: Include optabs-tree.h.
12579 * gimple-match-head.c: Likewise.
12580 * optabs-tree.h (target_supports_op_p): New.
12581 * optabs-tree.c (target_supports_op_p): New.
12583 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12585 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
12586 $gcc_cv_ld --help output.
12587 (gcc_cv_ld_demangle): Likewise.
12588 (gcc_cv_ld_eh_frame_hdr): Likewise.
12589 (gcc_cv_ld_pie): Likewise.
12590 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
12591 (gcc_cv_ld_buildid): Likewise.
12592 (gcc_cv_ld_sysroot): Likewise.
12593 (ld_bndplt_support): Likewise.
12594 (ld_pushpopstate_support): Likewise.
12595 * configure: Regenerate.
12596 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
12598 2017-06-21 Jakub Jelinek <jakub@redhat.com>
12601 * config/i386/sse.md (round<mode>2): Renumber match_dup and
12602 operands indexes to avoid gap between operands and match_dups.
12604 2017-06-21 Andrew Pinski <apinski@cavium.com>
12606 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
12607 Increment Arith_shift and Arith_shift_reg by 1.
12608 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
12610 * config/aarch64/aarch64.c (thunderx_tunings): Enable
12611 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
12612 (aarch64_strip_extend): Add new argument and test for it.
12613 (aarch64_cheap_mult_shift_p): New function.
12614 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
12615 add a cost if it is true.
12616 Update calls to aarch64_strip_extend.
12617 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
12619 2017-06-21 Andrew Pinski <apinski@cavium.com>
12621 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
12623 (thunderxt88): Likewise.
12624 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
12625 (thunderx_prefetch_tune): New variable.
12626 (thunderx2t99_prefetch_tune): Update for the correct values.
12627 (thunderxt88_tunings): New variable.
12628 (thunderx_tunings): Use thunderx_prefetch_tune instead of
12629 generic_prefetch_tune.
12630 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
12632 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12634 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
12635 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
12636 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
12637 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
12638 (aarch64_atomic_cas<mode>, GPI): Likewise.
12640 2017-06-21 Martin Liska <mliska@suse.cz>
12642 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
12643 statements on cold and hot labels.
12644 * predict.c (tree_estimate_probability_bb): Remove the
12645 prediction from this place.
12647 2017-06-21 Martin Liska <mliska@suse.cz>
12649 PR tree-optimization/79489
12650 * gimplify.c (maybe_add_early_return_predict_stmt): New
12652 (gimplify_return_expr): Call the function.
12653 * predict.c (tree_estimate_probability_bb): Remove handling
12655 * predict.def: Update comment about early return predictor.
12656 * gimple-predict.h (is_gimple_predict): New function.
12657 * predict.def: Change default value of early return to 66.
12658 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
12660 * passes.def: Put pass_strip_predict_hints to the beginning of
12663 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
12665 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
12666 FUNCTION_DECL declarations.
12667 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
12669 (dwaf2out_decl): Likewise.
12670 * godump.c (go_early_global_decl): Skip call to the real debug hook
12671 for FUNCTION_DECL declarations.
12672 * passes.c (rest_of_decl_compilation): Skip call to the
12673 early_global_decl debug hook for FUNCTION_DECL declarations, unless
12674 -fdump-go-spec is passed.
12676 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
12678 * config/i386/i386.c (struct builtin_isa): New field pure_p.
12679 Reorder for compactness.
12680 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
12681 (def_builtin_pure, def_builtin_pure2): New functions.
12682 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
12684 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
12686 * match.pd (nop_convert): New predicate.
12687 ((A +- CST1) +- CST2): Allow some NOP conversions.
12689 2017-06-21 Jakub Jelinek <jakub@redhat.com>
12692 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
12693 with ctors/dtors if GOVD_SHARED is set.
12695 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12697 * config/aarch64/aarch64.md (movti_aarch64):
12698 Emit mov rather than orr.
12699 (movtf_aarch64): Likewise.
12700 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
12701 Emit mov rather than orr.
12703 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12705 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
12706 Swap alternatives, make integer dup more expensive.
12708 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12710 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
12711 Return true for non-tls symbols.
12713 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
12715 * config/aarch64/aarch64-cores.def (cortex-a55): New.
12716 (cortex-a75): Likewise.
12717 (cortex-a75.cortex-a55): Likewise.
12718 * config/aarch64/aarch64-tune.md: Regenerate.
12719 * doc/invoke.texi (-mtune): Document new values for -mtune.
12721 2017-06-21 Tom de Vries <tom@codesourcery.com>
12723 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
12724 stack_size feature.
12725 (Effective-Target Keywords, Other attributes): Suggest using
12726 dg-add-options stack_size feature to get stack limit in stack_size
12727 effective target documentation.
12729 2017-06-21 Julian Brown <julian@codesourcery.com>
12730 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12732 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
12733 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
12734 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
12736 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
12737 attribute type list for neon_multiply.
12738 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
12739 attribute type list for neon_multiply.
12740 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
12741 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
12742 attribute type list for neon_multiply.
12743 * config/arm/types.md (crypto_pmull): Add.
12744 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
12745 attribute type list.
12747 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
12749 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
12752 2017-06-20 Jakub Jelinek <jakub@redhat.com>
12754 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
12755 to make sure not to dereference a NULL cost_classes_ptr pointer.
12757 2017-06-20 Carl Love <cel@us.ibm.com>
12759 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12760 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12761 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12762 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12763 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12764 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
12765 VMULOSW): New enum "unspec" values.
12766 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
12767 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
12768 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
12769 altivec_vmulosw): New patterns.
12770 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12771 VMULOSW): Add definitions.
12773 2017-06-20 Julia Koval <julia.koval@intel.com>
12775 * config/i386/i386.c: Fix rounding expand for new pattern.
12776 * config/i386/subst.md: Fix pattern (parallel -> unspec).
12778 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12780 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
12781 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
12783 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12785 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
12788 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12790 * config/aarch64/aarch64-cores.def: Rearrange to sort by
12791 architecture, then by implementer ID.
12792 * config/aarch64/aarch64-tune.md: Regenerate.
12794 2017-06-20 Richard Biener <rguenther@suse.de>
12796 PR middle-end/81097
12797 * fold-const.c (split_tree): Fold to type before negating.
12799 2017-06-20 David Malcolm <dmalcolm@redhat.com>
12801 * diagnostic-show-locus.c
12802 (selftest::test_fixit_deletion_affecting_newline): New function.
12803 (selftest::diagnostic_show_locus_c_tests): Call it.
12805 2017-06-20 Andreas Schwab <schwab@suse.de>
12808 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
12811 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
12813 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
12814 __ARM_FEATURE_COPROC according to support.
12816 2017-06-20 Jakub Jelinek <jakub@redhat.com>
12818 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
12819 Rewritten to avoid overflow for > 32-bit pointers.
12822 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
12823 by removing enum keyword.
12824 (ubsan_type_descriptor): Likewise. Formatting fix.
12827 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
12828 splitter): Require TARGET_SSE2 in the condition.
12830 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
12833 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
12834 for doing vector set of SFmode on ISA 3.0.
12835 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
12836 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
12838 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
12839 SFmode value into a V4SF variable that was extracted from another
12840 V4SF variable without converting the element to double precision
12841 and back to single precision vector format.
12842 (vsx_insert_extract_v4sf_p9_2): Likewise.
12844 2017-06-19 Jakub Jelinek <jakub@redhat.com>
12846 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
12847 in UWHI to avoid undefined overflow.
12850 * ubsan.h (enum ubsan_encode_value_phase): New.
12851 (ubsan_encode_value): Change second argument to
12852 enum ubsan_encode_value_phase with default value of
12853 UBSAN_ENCODE_VALUE_GENERIC.
12854 * ubsan.c (ubsan_encode_value): Change second argument to
12855 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
12856 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
12857 create_tmp_var_raw instead of create_tmp_var and use a
12859 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
12860 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
12861 ubsan_encode_value callers.
12864 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
12865 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
12866 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
12868 2017-06-19 Richard Biener <rguenther@suse.de>
12870 PR middle-end/81118
12871 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
12872 estimates if we changed anything.
12874 2017-06-19 Richard Biener <rguenther@suse.de>
12876 PR tree-optimization/80887
12877 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
12878 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
12879 simplified lookups, then reset mprts_hook.
12880 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
12882 (try_to_simplify): Likewise.
12884 2017-06-19 Martin Liska <mliska@suse.cz>
12887 * gimplify.c (gimplify_switch_expr):
12888 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
12890 2017-06-19 Martin Liska <mliska@suse.cz>
12892 * doc/install.texi: Document that PGO runs in 4 stages.
12894 2017-06-19 Martin Liska <mliska@suse.cz>
12897 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
12898 to dispatcher function name.
12899 * multiple_target.c (replace_function_decl): New function.
12900 (create_dispatcher_calls): Redirect both edges and references.
12902 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
12904 * profile-count.c (profile_count::dump): Dump quality.
12905 (profile_count::differs_from_p): Update for unsigned val.
12906 * profile-count.h (profile_count_quality): New enum.
12907 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
12909 2017-06-19 Richard Biener <rguenther@suse.de>
12911 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
12912 struct function as arg.
12913 (estimate_numbers_of_iterations): Export overload with loop arg.
12914 (free_numbers_of_iterations_estimates_loop): Use an overload of
12915 free_numbers_of_iterations_estimates instead.
12916 * tree-cfg.c (remove_bb): Adjust.
12917 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
12918 * tree-parloops.c (gen_parallel_loop): Likewise.
12919 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
12921 (tree_unroll_loops_completely): Likewise.
12922 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
12923 Use an overload instead and export.
12924 (estimated_loop_iterations): Adjust.
12925 (max_loop_iterations): Likewise.
12926 (likely_max_loop_iterations): Likewise.
12927 (estimate_numbers_of_iterations): Take struct function as arg
12929 (loop_exits_before_overflow): Adjust.
12930 (free_numbers_of_iterations_estimates_loop): Use an overload.
12931 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
12932 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
12934 2017-06-19 Richard Biener <rguenther@suse.de>
12937 * ipa-prop.c (find_constructor_constant_at_offset): Handle
12938 RANGE_EXPR conservatively.
12940 2017-06-16 Carl Love <cel@us.ibm.com>
12942 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12943 definitions for vec_float, vec_float2, vec_floato,
12944 vec_floate built-ins.
12945 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
12946 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
12948 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
12949 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
12950 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
12951 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
12952 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
12953 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
12954 vec_floato): Add builtin defines.
12955 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
12956 Update the built-in documentation file for the new built-in
12959 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12961 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
12962 (mthumb): Mark as the negative of -marm.
12964 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12966 * doc/invoke.texi (ARM Options, -mcpu): Document supported
12968 (ARM Options, -mtune): Document that this accepts the same
12969 extension options as -mcpu.
12970 (ARM Options, -mfpu): Document addition of -mfpu=auto.
12972 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12974 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
12975 permitted extensions.
12977 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12979 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
12980 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
12981 (armv8-m.main): Add option +nodsp.
12982 * config/arm/arm-cpu-cdata.h: Regenerated.
12984 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12986 * config/arm/t-fuchsia: New file.
12987 * config.gcc (arm*-*-fuchsia*): Use it.
12989 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12991 * config/arm/t-symbian: Rewrite for new option infrastructure.
12993 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12995 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
12996 (MULTILIB_REQUIRED): Likewise.
12998 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13000 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
13001 (MULTILIB_RESUE): Likewise.
13002 (MULTILIB_MATCHES): Likewise.
13003 (MULTLIB_REQUIRED): Likewise.
13005 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13007 * config/arm/t-rtems: Rewrite for new option framework.
13009 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13011 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
13012 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
13013 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
13014 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
13015 * config/arm/t-multilib: ... here.
13016 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
13017 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
13018 armv7-a and armv8*-a when A-profile libraries have not been built.
13019 * config/arm/t-rmprofile: Rewrite.
13021 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13023 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
13024 with a backslash. Remove the backslash after substituting unescaped
13026 * doc/fragments.texi (MULTILIB_REUSE): Document it.
13028 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13030 * config.gcc: (arm*-*-*): When building a-profile libraries, force
13031 the driver to pass through the default setting of -mfloat-abi.
13032 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
13034 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
13035 (all_feat_combs): New rule.
13036 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
13038 * config/arm/t-aprofile: Rewrite.
13040 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13042 * config/arm/arm.h (FPUTYPE_AUTO): Define.
13043 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
13044 fpu is not specified by the user/command-line.
13045 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
13046 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
13047 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
13048 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
13049 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
13050 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
13052 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13054 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
13055 * config/arm/t-arm-elf: Rewritten.
13057 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13059 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
13060 have some floating-point instructions.
13061 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
13062 (TARGET_MAYBE_HARD_FLOAT): New macro.
13063 * config/arm/arm-builtins.c (arm_init_builtins): Use
13064 TARGET_MAYBE_HARD_FLOAT.
13065 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
13067 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13069 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
13070 (configargs.h): Include it.
13071 (arm_print_hint_for_fpu_option): New function.
13072 (arm_parse_fpu_option): New function.
13073 (candidate_extension): New class.
13074 (arm_canon_for_multilib): New function.
13075 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
13076 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
13077 (ARCH_CANONICAL_SPECS): New macro.
13078 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
13080 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13082 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
13083 are set after handling multilib fragments. Set target_cpu_default2
13086 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13088 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
13090 (arm*-*-*): Set target_cpu_default2 to a quoted string.
13091 * config/arm/parsecpu.awk (check_cpu): Validate any extension
13093 (check_arch): Likewise.
13094 * config/arm/arm.c (arm_configure_build_target): Handle
13095 TARGET_CPU_DEFAULT being a string constant. Scan any feature
13096 options in the default.
13098 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13100 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
13101 when an option is an alias of another.
13102 * config/arm/parsecpu.awk (optalias): New parser token.
13103 (gen_comm_data): Mark non-alias options as such. Emit entries
13104 for extension aliases.
13105 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
13106 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
13107 (armv6kz, armv6zk, armv6t2): Likewise.
13108 (armv7): Make vfpv3-d16 an alias.
13109 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
13111 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
13112 Sort in canonical order.
13113 (armv8-a): Sort in canonical order.
13114 (armv8.1-a, armv8.2-a): Likewise.
13115 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
13117 (cortex-a9): Sort in canonical order.
13118 * config/arm/arm.c (selftests.h): Include it.
13119 (arm_test_cpu_arch_data): New function.
13120 (arm_run_self_tests): New function.
13121 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
13122 (targetm): Move declaration to the end of the file.
13123 * arm-cpu-cdata.h: Regenerated.
13125 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13127 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
13128 call to target_mode_check describing the type of option passed.
13129 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
13130 (arm_target_thumb_only): Use arm_parse_arch_option_name or
13131 arm_parse_cpu_option_name to match parameters against list of
13133 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
13134 arm_arch_core_flags data structure.
13135 * config/arm/arm-cpu_cdata.h: Regenerated.
13137 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13139 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
13141 (arm_print_hint_for_cpu_option): Likewise.
13142 (arm_print_hint_for_arch_option): Likewise.
13143 (arm_parse_cpu_option_name): Likewise.
13144 (arm_parse_arch_option_name): Likewise.
13145 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
13146 of entries in the all_fpus list.
13147 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
13148 (arm_parse_cpu_option_name): Declare.
13149 (arm_parse_arch_option_name): Declare.
13150 (arm_parse_option_features): Declare.
13151 (arm_intialize_isa): Declare.
13152 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
13154 (gen_comm_data): ... here. Make definitions non-static.
13155 * config/arm/arm-cpu-data.h: Regenerated.
13156 * config/arm/arm-cpu-cdata.h: Regenerated.
13158 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13160 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
13161 (cpu_arch_extension): New structure.
13162 (cpu_arch_option, arch_option, cpu_option): New structures.
13163 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
13164 architecture types.
13165 (gen_data): Generate new format data tables.
13166 * config/arm/arm.c (cpu_tune): New structure.
13167 (cpu_option, processors): Delete.
13168 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
13169 (arm_print_hint_for_cpu_option): ... this and ...
13170 (arm_print_hint_for_arch_option): ... this.
13171 (arm_parse_arch_cpu_name): Delete. Replace with ...
13172 (arm_parse_cpu_option_name): ... this and ...
13173 (arm_parse_arch_option_name): ... this.
13174 (arm_unrecognized_feature): Change type of target parameter to
13176 (arm_parse_arch_cpu_features): Delete. Replace with ...
13177 (arm_parse_option_features): ... this.
13178 (arm_configure_build_target): Rework to use new configuration data
13180 (arm_print_tune_info): Rework for new configuration data tables.
13181 * config/arm/arm-cpu-data.h: Regenerated.
13182 * config/arm/arm-cpu.h: Regenerated.
13184 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13186 * Makefile.in (OBJS): Move sbitmap.o from here ...
13187 (OBJS-libcommon): ... to here.
13189 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13191 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
13192 (ISA_ALL_CRYPTO): New macro.
13193 (ISA_ALL_SIMD): New macro
13194 (ISA_ALL_FP): New macro.
13195 * config/arm/arm.c (fpu_bitlist): Update initializer.
13196 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
13198 (arm9e): Add fpu. Add option for nofp
13199 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
13200 (arm926ej-s, arm1026ej-s): Likewise.
13201 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
13202 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
13203 neon-fp16, neon-vfpv4, nofp and nosimd.
13204 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
13205 (cortex-a8): Add fpu. Add option for nofp.
13206 (cortex-a9): Add fpu. Add options for nosimd and nofp.
13207 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
13208 (cortex-r4f): Add fpu.
13209 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
13210 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
13212 (cortex-r8): Likewise.
13213 (cortex-m4): Add fpu. Add option for nofp.
13214 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
13215 (cortex-a17.cortex-a7): Likewise.
13216 (cortex-a32): Add fpu. Add options for crypto and nofp.
13217 (cortex-a35, cortex-a53): Likewise.
13218 (cortex-a57): Add fpu. Add option for crypto.
13219 (cortex-a72, cortex-a73): Likewise.
13220 (exynos-m1): Likewise.
13221 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
13222 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
13223 (cortex-m33): Add fpu. Add option for nofp.
13224 * config/arm/arm-cpu-cdata.h: Regenerated
13225 * config/arm/arm-cpu-data.h: Regenerated.
13227 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13229 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
13230 (armv5te, armv5tej): Likewise.
13231 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
13232 (armv7): Add options fp and vfpv3-d16.
13233 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
13234 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
13236 (armv7ve): Likewise.
13237 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
13238 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
13239 (armv8-a): Add nocrypto option.
13240 (armv8.1-a, armv8.2-a): Likewise.
13241 (armv8-m.main): add options fp, fp.dp and nofp.
13243 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13245 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
13247 (armv8-a+crc): Delete.
13248 (armv8.1-a): Add options simd, crypto and nofp.
13249 (armv8.2-a): Add options fp16, simd, crypto and nofp.
13250 (armv8.2-a+fp16): Delete.
13251 (armv8-m.main): Add option dsp.
13252 (armv8-m.main+dsp): Delete.
13253 (cortex-a8): Add fpu. Add nofp option.
13254 (cortex-a9): Add fpu. Add nofp and nosimd options.
13255 * config/arm/parsecpu.awk (gen_data): Generate option tables and
13256 link to main cpu and architecture data structures.
13257 (gen_comm_data): Only put isa attributes from the main architecture
13259 (option): New statement for architecture and CPU entries.
13260 * arm.c (struct cpu_option): New structure.
13261 (struct processors): Add entry for options.
13262 (arm_unrecognized_feature): New function.
13263 (arm_parse_arch_cpu_name): Ignore any characters after the first
13265 (arm_parse_arch_cpu_feature): New function.
13266 (arm_configure_build_target): Separate out any CPU and architecture
13267 features and parse separately. Don't error out if -mfpu=auto is
13268 used with only an architecture string.
13269 (arm_print_asm_arch_directives): New function.
13270 (arm_file_start): Call it.
13271 * config/arm/arm-cpu-cdata.h: Regenerated.
13272 * config/arm/arm-cpu-data.h: Likewise.
13273 * config/arm/arm-tables.opt: Likewise.
13275 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13277 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
13278 assembler when it is not -mfpu=auto.
13280 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13282 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
13283 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
13284 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
13285 (ASM_CPU_SPEC): Rewrite.
13286 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
13287 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
13288 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
13289 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
13290 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
13291 copied string is NUL-terminated. Also strip any characters prefixed
13293 (arm_rewrite_selected_arch): New function.
13294 (arm_rewrite_march): New function.
13296 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13298 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
13299 (x_arm_cpu_string, x_arm_tune_string): Likewise.
13300 (march, mcpu, mtune): Convert to string-based options.
13301 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
13302 (arm_parse_arch_cpu_name): New function.
13303 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
13304 identify selected architecture or CPU.
13305 (arm_option_save): New function.
13306 (TARGET_OPTION_SAVE): Redefine.
13307 (arm_option_restore): Restore string options.
13308 (arm_option_print): Print string options.
13310 2017-06-16 Martin Sebor <msebor@redhat.com>
13312 PR tree-optimization/80933
13313 PR tree-optimization/80934
13314 * builtins.c (fold_builtin_3): Do not handle bcmp here.
13315 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
13316 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
13317 (gimple_fold_builtin): Call them.
13319 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13321 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
13322 as unlikely; update profile.
13324 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13326 * predict.c (force_edge_cold): Handle declaring edges impossible
13329 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13331 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
13333 (try_unroll_loop_completely): Fix reporting.
13335 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13337 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
13339 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
13342 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
13343 if given a non-constant argument for an intrinsic which requires a
13346 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13348 * profile.c (compare_freqs): New function.
13349 (branch_prob): Sort edge list.
13350 (find_spanning_tree): Assume that the list is priority sorted.
13352 2017-06-16 Richard Biener <rguenther@suse.de>
13354 PR tree-optimization/81090
13355 * passes.def (pass_record_bounds): Remove.
13356 * tree-pass.h (make_pass_record_bounds): Likewise.
13357 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
13358 make_pass_record_bounds): Likewise.
13359 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
13360 not free niter estimates at the beginning but at the end.
13361 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
13363 2017-06-16 Richard Biener <rguenther@suse.de>
13365 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
13366 initializer to workaround ICE in host GCC 4.8.
13368 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13370 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
13372 (clone_inlined_nodes): Update.
13374 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13376 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
13377 prefetch settings, and enable prefetching by default at -O3.
13379 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13381 * config/aarch64/aarch64.c (aarch64_override_options_internal):
13382 Set flag_prefetch_loop_arrays according to tuning data.
13384 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13386 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
13387 New tune structure.
13388 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
13389 [Unrelated to main purpose of the patch] Place the pointer field last
13390 to enable type checking errors when tune structure are wrongly merged.
13391 * config/aarch64/aarch64.c (generic_prefetch_tune,)
13392 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
13393 (thunderx2t99_prefetch_tune): New tune constants.
13394 (tune_params *_tunings): Update all tunings (no functional change).
13395 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
13396 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
13397 from tunings structures.
13399 2017-06-16 Jakub Jelinek <jakub@redhat.com>
13402 * ubsan.c (instrument_null): Add T argument, use it instead
13403 of computing it based on IS_LHS.
13404 (instrument_object_size): Likewise.
13405 (pass_ubsan::execute): Adjust instrument_null and
13406 instrument_object_size callers to pass gimple_get_lhs or
13407 gimple_assign_rhs1 result to it. Use instrument_null instead of
13408 calling get_base_address and instrument_mem_ref. Handle
13409 aggregate call arguments for object-size sanitization.
13411 2017-06-16 Yury Gribov <tetra2005@gmail.com>
13413 PR tree-optimization/81089
13414 * tree-vrp.c (is_masked_range_test): Validate operands of
13417 2017-06-15 Martin Sebor <msebor@redhat.com>
13420 * dumpfile.c (dump_register): Avoid calling memset to initialize
13421 a class with a default ctor.
13422 * gcc.c (struct compiler): Remove const qualification.
13423 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
13424 * hash-table.h: Ditto.
13425 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
13427 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
13428 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
13430 * params.h (struct param_info): Make struct members non-const.
13431 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
13432 with default initialization.
13433 * vec.h (vec_copy_construct, vec_default_construct): New helper
13435 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
13436 with vec_copy_construct.
13437 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
13438 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
13439 * doc/invoke.texi (-Wclass-memaccess): Document.
13441 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13443 * emit-rtl.h (is_leaf): Update comment about local
13444 register allocator.
13446 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13449 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
13450 for a variable to have a section before checking if the section has a
13452 Set section to.persistent if persistent attribute is set.
13453 Warn if .persistent attribute is used on an automatic variable.
13455 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
13457 PR rtl-optimization/80474
13458 * reorg.c (update_block): Do not ignore instructions in a delay slot.
13460 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
13462 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
13465 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
13467 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
13468 (casesi): Emit bounds checking as RTL.
13469 (casesi_internal_mips16_<mode>): Remove bounds checking.
13471 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
13473 * config/xtensa/xtensa.c (xtensa_option_override): Append
13474 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
13475 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
13476 xtensa_doloop_hooks): Define unconditionally.
13477 (xtensa_reorg_loops): Only call reorg_loops in the presence of
13479 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
13480 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
13481 for it in xtensa_option_override.
13482 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
13483 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
13485 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
13487 * doc/cppopts.texi: Document '-' special value to -MF.
13489 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
13491 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
13492 (cortex_a53_fconst): Likewise.
13493 (cortex_a53_fpmul): Likewise.
13494 (cortex_a53_f_load_64): Likewise.
13495 (cortex_a53_f_load_many): Likewise.
13496 (cortex_a53_advsimd_alu): Likewise.
13497 (cortex_a53_advsimd_alu_q): Likewise.
13498 (cortex_a53_advsimd_mul): Likewise.
13499 (cortex_a53_advsimd_mul_q): Likewise.
13500 (fpmac bypass): Add new bypass for fpmac-fpmac case.
13501 Add missing fmul, r2f_cvt and fconst cases.
13503 2017-06-14 Richard Biener <rguenther@suse.de>
13505 PR middle-end/81088
13506 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
13508 (fold_binary_loc): When associating do not treat pre-existing
13509 TREE_OVERFLOW on literal constants as a reason to allow
13510 TREE_OVERFLOW on associated literal constants.
13512 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13514 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
13515 (MASK_FEATURES): New macro.
13516 * config/sparc/sparc.c (sparc_option_override): Remove the special
13517 handling of -mfpu and generalize it to all MASK_FEATURES switches.
13519 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13521 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
13522 a division of 0 if non-call exceptions are enabled.
13524 2017-06-14 Andrew Pinski <apinski@cavium.com>
13525 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13528 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
13529 Improve vector initialization code gen for only variable case.
13531 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13533 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
13535 2017-06-14 Richard Biener <rguenther@suse.de>
13537 PR tree-optimization/81083
13538 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
13541 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13543 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
13544 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
13545 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
13546 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
13547 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
13548 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
13550 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13552 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
13553 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
13555 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13557 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
13559 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13561 * config/rs6000/t-rtems: Don't handle SPE.
13563 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13565 * config/rs6000/t-linux: Don't handle SPE.
13567 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13569 * config/rs6000/eabispe.h: Delete file.
13571 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13573 * config/rs6000/t-spe: Delete file.
13575 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13577 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
13578 (rs6000_legitimate_offset_address_p): Return false for anything in
13579 V2SImode or V2SFmode.
13581 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13583 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
13584 except V2SF and V2SI. Rearrange the vector modes, and add comments.
13585 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
13587 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
13588 (rs6000_legitimate_offset_address_p): Ditto.
13589 (rs6000_emit_move): Ditto.
13590 (rs6000_init_builtins): Remove V4HI_type_node.
13592 2017-06-13 Martin Liska <mliska@suse.cz>
13595 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
13596 (gate_asan): Likewise.
13597 * asan.h (asan_no_sanitize_address_p): Remove the function.
13598 (sanitize_flags_p): New function.
13599 * builtins.def: Fix coding style.
13600 * common.opt: Use renamed enum value.
13601 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
13602 * doc/extend.texi: Document no_sanitize attribute.
13603 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
13604 to SANITIZE_UNDEFINED_NONDEFAULT.
13605 * gcc.c (sanitize_spec_function): Use the renamed enum value.
13606 * gimple-fold.c (optimize_atomic_compare_exchange_p):
13607 Use sanitize_flags_p.
13608 * gimplify.c (gimplify_function_tree): Likewise.
13609 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
13610 * opts.c (parse_no_sanitize_attribute): New function.
13611 (common_handle_option): Use renamed enum value.
13612 * opts.h (parse_no_sanitize_attribute): Declare.
13613 * tree.c (sanitize_flags_p): New function.
13614 * tree.h: Declared here.
13615 * tsan.c: Use sanitize_flags_p.
13616 * ubsan.c (ubsan_expand_null_ifn): Likewise.
13617 (instrument_mem_ref): Likewise.
13618 (instrument_bool_enum_load): Likewise.
13619 (do_ubsan_in_current_function): Remove the function.
13620 (pass_ubsan::execute): Use sanitize_flags_p.
13621 * ubsan.h: Remove do_ubsan_in_current_function
13622 * tree-cfg.c (print_no_sanitize_attr_value): New function.
13623 (dump_function_to_file): Use it here.
13625 2017-06-13 Martin Jambor <mjambor@suse.cz>
13627 PR tree-optimization/80803
13628 PR tree-optimization/81063
13629 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
13630 (propagate_subaccesses_across_link): Enqueue subtree whenever
13631 necessary instead of relying on the caller.
13633 2017-06-13 Martin Jambor <mjambor@suse.cz>
13635 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
13636 that have a first_link.
13637 (sort_and_splice_var_accesses): Do not check first_link before
13639 (subtree_mark_written_and_enqueue): Likewise.
13640 (propagate_all_subaccesses): Likewise and do not stop at first
13641 parent with a first_link.
13643 2017-06-13 Martin Jambor <mjambor@suse.cz>
13645 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
13648 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13650 * match.pd: New pattern.
13652 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13654 * tree-vrp.c (is_masked_range_test): New function.
13655 (register_edge_assert_for): Determine ranges for
13658 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13660 PR tree-optimization/67328
13661 * fold-const.c (maskable_range_p): New function.
13662 (build_range_check): Generate bittests if possible.
13664 2017-06-13 Martin Liska <mliska@suse.cz>
13666 * gimple-pretty-print.c (dump_probability): Add new argument.
13667 (dump_edge_probability): Dump both probability and count.
13668 (dump_gimple_label): Likewise.
13669 (dump_gimple_bb_header): Likewise.
13671 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
13674 * config/avr/avr-devices.c: Fix indentation.
13675 * config/avr/gen-avr-mmcu-specs.c: Dito.
13677 2017-06-13 Richard Biener <rguenther@suse.de>
13679 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
13680 instead get vector type from stmt_info.
13681 (vectorizable_reduction): Adjust. Remove dead code.
13683 2017-06-13 Richard Biener <rguenther@suse.de>
13685 PR middle-end/81065
13686 * fold-const.c (extract_muldiv_1): Remove bogus distribution
13687 case of C * (x * C2 + C3).
13688 (fold_addr_of_array_ref_difference): Properly fold index difference.
13690 2017-06-12 David S. Miller <davem@davemloft.net>
13693 * config/sparc/sparc.md (return expander): Emit frame blockage if
13694 function uses alloca.
13696 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
13698 * combine.c (make_field_assignment): Check len rather than the mode
13699 precision when calling force_to_mode.
13701 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
13703 Support multilibs and devices that see flash in RAM address range.
13706 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
13707 (avr_mcu_t) <flash_pm_offset>: New field.
13708 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
13709 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
13710 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
13711 (AVR_TINY_PM_OFFSET): Remove macro.
13712 * config/avr/avr.opt (-mshort-calls): New option.
13713 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
13714 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
13715 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
13716 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
13717 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
13718 instead of avr_arch->have_jmp_call.
13719 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
13720 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
13721 avr_arch->flash_pm_offset to define.
13722 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
13723 new field flash_pm_offset. Add entry for avrxmega3.
13724 (avr_texinfo): Add entry for avrxmega3.
13725 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
13726 attiny212, attiny214,
13727 attiny412, attiny414, attiny416, attiny417,
13728 attiny814, attiny816, attiny817,
13729 attiny1614, attiny1616, attiny1617,
13730 attiny3214, attiny3216, attiny3217.
13731 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
13732 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
13733 (avr_print_operand_address) [AVR_TINY]: Same.
13734 (avr_asm_init_sections) <readonly_data_section>: Only patch
13735 callback if avr_arch->flash_pm_offset = 0.
13736 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
13737 for rodata if avr_arch->flash_pm_offset != 0.
13738 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
13739 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
13740 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
13741 (m_options): Append opt_rcall.
13742 (m_dirnames): Append dir_rcall.
13743 * config/avr/t-multilib: Regenerate.
13745 * configure.ac [target=avr]: Check whether avrxmega3 default
13746 linker description file works as needed.
13747 * configure: Regenerate.
13748 * doc/avr-mmcu.texi: Regenerate.
13749 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
13750 <__AVR_ARCH__>: Document avrxmega3 and 103.
13751 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
13752 <__AVR_SHORT_CALLS__>: Document it.
13753 <__AVR_PM_BASE_ADDRESS__>: Document it.
13754 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
13755 (AVR Variable Attributes) <progmem>: Document this is
13756 not needed for avrxmega3.
13757 (AVR Named Address Spaces) <__flash>: Dito.
13759 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
13761 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
13763 2017-06-12 Doug Rupp <rupp@adacore.com>
13765 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
13766 Append vxworks-stdint.h to the tm_file list.
13767 * config/vxworks-stdint.h: New file.
13769 2017-06-12 Martin Liska <mliska@suse.cz>
13771 PR tree-optimization/81041
13772 * tree-profile.c (gimple_gen_ic_func_profiler):
13773 Create an extra BB in profile-generate
13774 (gimple_gen_time_profiler): Likewise.
13776 2017-06-12 Jakub Jelinek <jakub@redhat.com>
13778 PR tree-optimization/81003
13779 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
13780 (update_range_test): Use it instead of force_gimple_operand_gsi.
13782 2017-06-12 Richard Biener <rguenther@suse.de>
13784 PR tree-optimization/81053
13785 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
13786 with backedge value not defined in loop. Simplify def stmt
13789 2017-06-11 Tom de Vries <tom@codesourcery.com>
13792 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
13794 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
13795 nvptx_cannot_force_const_mem.
13797 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13799 * opts.c (finish_options): Move test for flag_split_stack after
13800 it has been initialized.
13802 2017-06-11 Jason Merrill <jason@redhat.com>
13804 * tree.h (id_equal): New.
13805 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
13806 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
13807 instead of strcmp of IDENTIFIER_POINTER.
13809 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13811 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
13812 (mark_all_inlined_calls_cdtor): Fix formating.
13813 (inline_transform): Rescale profile before inlining.
13815 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13817 * cgraph.h (cgraph_edge::clone): Update prototype.
13818 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
13819 (cgraph_node::create_clone): Update.
13820 (cgraph_node::create_version_clone): Update.
13821 * tree-inline.c (copy_bb): Update.
13822 (expand_call_inline): Update.
13824 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
13826 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
13827 factored out from ...
13828 (rs6000_emit_prologue): ... here.
13830 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
13832 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
13833 factored out from ...
13834 (rs6000_emit_prologue): ... here.
13836 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13838 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
13840 (handle_missing_profiles): Fix computation of tp_first_run.
13841 (counts_to_freqs): Do not touch freqs when count is 0.
13843 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13845 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
13848 2017-06-10 Tom de Vries <tom@codesourcery.com>
13850 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
13851 attributes): Document signal effective target.
13853 2017-06-10 Tom de Vries <tom@codesourcery.com>
13855 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
13856 Document effective target stack_size.
13858 2017-06-09 David Malcolm <dmalcolm@redhat.com>
13860 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
13861 to the edit_context if they can be auto-applied.
13863 2017-06-9 Ian Lance Taylor <iant@golang.org>
13865 * opts.c (finish_options): If -fsplit-stack, disable implicit
13866 -forder-blocks-and-partition.
13867 * doc/invoke.texi (Optimize Options): Document that when using
13868 -fsplit-stack -forder-blocks-and-partition is not implicitly
13871 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13873 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
13874 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
13875 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
13876 * builtins.def (abort, trap, unreachable): Declare cold.
13877 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
13878 * tree-core.h (ECF_COLD): New.
13879 * tree.c (set_call_expr_flags): Handle ECF_COLD.
13880 (build_common_builtin_nodes): Mark unreachable and abort as cold.
13882 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13884 * predict.c (unlikely_executed_stmt_p): Cleanup.
13886 2017-06-09 Richard Biener <rguenther@suse.de>
13888 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
13889 model if the ref is always written to.
13891 2017-06-09 Tamar Christina <tamar.christina@arm.com>
13893 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
13895 2017-06-09 Tamar Christina <tamar.christina@arm.com>
13897 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
13900 2017-06-09 Tom de Vries <tom@codesourcery.com>
13903 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
13904 "target cannot support label values" when encountering LABEL_REF.
13906 2017-06-09 Martin Liska <mliska@suse.cz>
13908 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
13909 (gimple_gen_ic_func_profiler): Emit direct comparison
13910 of __gcov_indirect_call_callee with NULL.
13911 (gimple_gen_time_profiler): Change probability from
13912 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
13914 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13916 * profile.c (edge_gcov_counts): Turn to pointer.
13917 (compute_branch_probabilities, compute_branch_probabilities): Update.
13918 (branch_prob): Do not clear edge_gcov_count.
13919 * profile.h (edge_gcov_counts): Turn to pointer.
13920 (edge_gcov_count): Update.
13922 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13924 * gimple.h (gimple_check_failed): Mark cold.
13926 2017-06-09 Richard Biener <rguenther@suse.de>
13928 PR tree-optimization/66623
13929 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
13930 refactor check_reduction into two parts, properly computing
13931 whether we have to check reduction validity for outer loop
13934 2017-06-09 Richard Biener <rguenther@suse.de>
13936 PR tree-optimization/79483
13937 * graphite-scop-detection.c (order): New global.
13938 (get_order): Compute bb to order mapping that satisfies code
13939 generation constraints.
13940 (cmp_pbbs): New helper.
13941 (build_scops): Start domwalk at entry block, sort generated
13944 2017-06-09 Richard Biener <rguenther@suse.de>
13946 PR middle-end/81007
13947 * ipa-polymorphic-call.c
13948 (ipa_polymorphic_call_context::restrict_to_inner_class):
13949 Skip FIELD_DECLs with error_mark_node type.
13950 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
13953 2017-06-09 Martin Liska <mliska@suse.cz>
13955 * predict.c (struct branch_predictor): New struct.
13956 (test_prediction_value_range): New test.
13957 (predict_c_tests): New function.
13958 * selftest-run-tests.c (selftest::run_tests): Run the function.
13959 * selftest.h: Declare new tests.
13961 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
13964 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
13965 gen_add3_insn did not fail.
13966 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
13967 r0, construct that number in a temporary reg and add that reg to r0.
13968 If asked to put the result in r0 as well, fail.
13970 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
13972 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13973 for early expansion of vec_eqv.
13975 2017-06-08 Jakub Jelinek <jakub@redhat.com>
13977 PR middle-end/81005
13978 * ubsan.c (instrument_null): Avoid pointless code temporary.
13979 (pass_ubsan::execute): Instrument aggregate arguments of calls.
13981 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
13985 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
13988 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
13991 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13993 * predict.c (unlikely_executed_edge_p): Move ahead.
13994 (probably_never_executed_edge_p): Use it.
13996 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
13998 PR middle-end/79988
13999 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
14000 gimple_call_builtin_p call.
14002 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14004 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
14005 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
14006 rtl_check_failed_type2, rtl_check_failed_code1,
14007 rtl_check_failed_code2, rtl_check_failed_code_mode,
14008 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
14009 rtvec_check_failed_bounds, rtl_check_failed_flag,
14010 _fatal_insn_not_found, _fatal_insn): Likewise.
14011 * tree.h (tree_contains_struct_check_failed,
14012 tree_check_failed, tree_not_check_failed,
14013 tree_class_check_failed, tree_range_check_failed,
14014 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
14015 tree_vec_elt_check_failed, phi_node_elt_check_failed,
14016 tree_operand_check_failed, omp_clause_check_failed,
14017 omp_clause_operand_check_failed, omp_clause_range_check_failed):
14020 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14022 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
14023 flag_branch_probabilities.
14024 * ipa-inline.c (edge_badness): Likewise.
14025 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
14026 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
14027 * predict.c (maybe_hot_frequency_p): Likewise.
14028 (probably_never_executed): Likewise.
14029 * sched-ebb.c (schedule_ebbs): Likewise.
14030 * sched-rgn.c (find_single_block_region): Likewise.
14031 * tracer.c (tail_duplicate): Likewise.
14033 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14035 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
14036 longer requires x_flag_profile_use.
14038 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14040 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
14041 instead of flag_reorder_blocks_and_partition.
14042 * dbxout.c (dbxout_function_end): Likewise.
14043 * dwarf2out.c (gen_subprogram_die): Likewise.
14044 * haifa-sched.c (sched_create_recovery_edges): Likewise.
14045 * hw-doloop.c (reorg_loops): Likewise.
14046 * varasm.c (assemble_start_function,
14047 assemble_end_function): Likewise.
14048 (decide_function_section): Do not check for
14049 flag_reorder_blocks_and_partition.
14051 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
14053 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
14055 (chkp_get_hard_register_fake_addr_expr): Ditto.
14056 (chkp_build_addr_expr): Add check for hard reg case.
14057 (chkp_parse_array_and_component_ref): Ditto.
14058 (chkp_find_bounds_1): Ditto.
14059 (chkp_process_stmt): Don't generate bounds store for
14062 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14064 * predict.c (maybe_hot_bb_p): Do not check profile status.
14065 (maybe_hot_edge_p): Likewise.
14066 (probably_never_executed): Check for zero counts even if profile
14068 (unlikely_executed_edge_p): New function.
14069 (unlikely_executed_stmt_p): New function.
14070 (unlikely_executed_bb_p): New function.
14071 (set_even_probabilities): Use unlikely predicates.
14072 (combine_predictions_for_bb): Likewise.
14073 (predict_paths_for_bb): Likewise.
14074 (predict_paths_leading_to_edge): Likewise.
14075 (determine_unlikely_bbs): New function.
14076 (estimate_bb_frequencies): Use it.
14077 (compute_function_frequency): Use zero counts even if profile is
14079 * profile-count.h: Fix typo.
14081 2017-08-08 Julia Koval <julia.koval@intel.com>
14083 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
14084 _mm512_mask_cvtsepi16_storeu_epi8,
14085 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
14086 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
14087 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
14088 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
14089 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
14090 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
14091 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
14092 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
14093 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
14094 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
14095 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
14096 __builtin_ia32_pmovuswb256mem_mask,
14097 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
14098 __builtin_ia32_pmovwb512mem_mask): New builtins.
14100 2017-08-08 Julia Koval <julia.koval@intel.com>
14102 PR target/73350,80862
14103 * config/i386/subst.md (round): Fix round pattern.
14104 * config/i386/i386.c (ix86_erase_embedded_rounding):
14105 Fix erasing rounding for the fixed pattern.
14107 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14109 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
14111 2017-06-08 Martin Liska <mliska@suse.cz>
14113 PR gcov-profile/80911
14114 * gcov.c (block_info::block_info): New constructor.
14116 2017-06-07 Carl Love <cel@us.ibm.com>
14118 * config/rs6000/rs6000-c: The return type of the following
14119 built-in functions was implemented as int not long long. Fix sign
14120 of return value for the unsigned version of vec_mulo and vec_mule.
14121 vector unsigned long long vec_bperm (vector unsigned long long,
14122 vector unsigned char)
14123 vector signed long long vec_mule (vector signed int,
14125 vector unsigned long long vec_mule (vector unsigned int,
14126 vector unsigned int)
14127 vector signed long long vec_mulo (vector signed int,
14129 vector unsigned long long vec_mulo (vector unsigned int,
14130 vector unsigned int)
14131 * doc/extend.texi: Fix the documentation for the built-in
14134 2017-06-07 Carl Love <cel@us.ibm.com>
14137 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
14140 2017-06-07 Carl Love <cel@us.ibm.com>
14142 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
14143 support, Generate doublehv for signed int/float for BE case only.
14145 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
14147 * doc/invoke.texi (mcx16): Rewrite.
14149 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14151 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
14152 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
14153 *mov<mode>_softfloat, and an anonymous splitter): Use
14154 nonimmediate_operand instead of rs6000_nonimmediate_operand.
14156 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14158 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
14160 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
14161 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
14162 (rs6000_debug_reg_global): Adjust.
14163 (rs6000_init_hard_regno_mode_ok): Adjust.
14164 (rs6000_dbx_register_number): Adjust.
14165 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
14166 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
14167 Remove SPE_ACC and SPEFSCR.
14168 (REG_ALLOC_ORDER): Ditto.
14169 (FRAME_POINTER_REGNUM): Change to 111.
14170 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
14171 (REG_CLASS_NAMES): Ditto.
14172 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
14173 (REGISTER_NAMES): Ditto.
14174 (ADDITIONAL_REG_NAMES): Ditto.
14175 (rs6000_reg_names): Ditto.
14176 * config/rs6000/rs6000.md: Renumber some register number
14179 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14181 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
14183 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
14184 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
14186 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
14187 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
14188 Delete the SPE high registers.
14189 (REG_ALLOC_ORDER): Ditto.
14190 (enum reg_class): Remove SPE_HIGH_REGS.
14191 (REG_CLASS_NAMES): Ditto.
14192 (REG_CLASS_CONTENTS): Delete the SPE high registers.
14193 (REGISTER_NAMES): Ditto.
14194 (rs6000_reg_names): Ditto.
14195 * doc/tm.texi.in: Remove SPE as example.
14196 * doc/tm.texi: Regenerate.
14198 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14200 * config/rs6000/8540.md (ppc8540_brinc): Delete.
14201 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
14202 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
14203 * config/rs6000/rs6000.md (type): Remove "brinc".
14205 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14207 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
14208 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
14209 * config/rs6000/linuxspe.h: Delete file.
14210 * config/rs6000/rs6000.md: Don't include spe.md.
14211 * config/rs6000/spe.h: Delete file.
14212 * config/rs6000/spe.md: Delete file.
14213 * config/rs6000/t-rs6000: Remove spe.md.
14215 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14217 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
14218 (reg_or_none500mem_operand): Delete.
14219 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
14220 instead of reg_or_none500mem_operand.
14222 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14224 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
14225 handling of SPE flags.
14226 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
14228 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14230 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
14232 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
14233 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
14234 paired_divv2sf3): Similar.
14235 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
14236 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
14237 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
14239 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
14240 Rename the paired_* instruction patterns.
14241 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
14243 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
14244 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
14245 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
14246 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
14247 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
14248 PAIRED_VECTOR_MODE.
14249 (struct machine_function): Delete field spe_insn_chain_scanned_p.
14250 (spe_func_has_64bit_regs_p): Delete.
14251 (spe_expand_predicate_builtin): Delete.
14252 (spe_expand_evsel_builtin): Delete.
14253 (TARGET_DWARF_REGISTER_SPAN): Do not define.
14254 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
14255 (invalid_e500_subreg): Delete.
14256 (rs6000_legitimize_address): Always force_reg op2 as well, for
14257 paired single memory accesses.
14258 (rs6000_member_type_forces_blk): Delete.
14259 (rs6000_spe_function_arg): Delete.
14260 (rs6000_expand_unop_builtin): Delete SPE handling.
14261 (rs6000_expand_binop_builtin): Ditto.
14262 (spe_expand_stv_builtin): Delete.
14263 (bdesc_2arg_spe): Delete.
14264 (spe_expand_builtin): Delete.
14265 (spe_expand_predicate_builtin): Delete.
14266 (spe_expand_evsel_builtin): Delete.
14267 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
14268 (spe_init_builtins): Delete.
14269 (spe_func_has_64bit_regs_p): Delete.
14270 (savres_routine_name): Delete "info" parameter. Adjust callers.
14271 (rs6000_emit_stack_reset): Ditto.
14272 (rs6000_dwarf_register_span): Delete.
14273 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
14274 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
14275 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
14277 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
14279 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
14280 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
14281 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
14282 mulv2sf3, divv2sf3): Delete expanders.
14284 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14286 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
14288 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14290 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
14291 * config/rs6000/rs6000.c: Ditto.
14293 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14295 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
14296 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
14297 comparison_operator.
14299 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14301 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
14302 * config/rs6000/rs6000.opt: Ditto.
14303 * config/rs6000/t-rtems: Ditto.
14305 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14307 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
14308 TARGET_E500_SINGLE by 0, simplify.
14309 * config/rs6000/rs6000.c: Ditto.
14310 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
14311 (spe_build_register_parallel): Delete.
14312 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
14313 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
14314 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
14315 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
14316 (E500_CONVERT): Delete.
14317 * config/rs6000/spe.md: Remove many patterns and all define_constants.
14319 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14321 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
14322 * config/rs6000/dfp.md: Ditto.
14323 (negdd2, *negdd2_fpr): Merge.
14324 (absdd2, *absdd2_fpr): Merge.
14325 (negtd2, *negtd2_fpr): Merge.
14326 (abstd2, *abstd2_fpr): Merge.
14327 * config/rs6000/e500.h: Delete file.
14328 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
14329 TARGET_FPRS by 1 and simplify.
14330 * config/rs6000/rs6000-c.c: Ditto.
14331 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
14332 TARGET_DF_SPE by 0.
14333 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
14335 * config/rs6000/rs6000.md: Ditto.
14336 (floatdidf2, *floatdidf2_fpr): Merge.
14337 (move_from_CR_gt_bit): Delete.
14338 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
14339 (E500_CR_IOR_COMPARE): Delete.
14340 (All patterns that require !TARGET_FPRS): Delete.
14341 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
14343 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14345 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
14347 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14349 * graphds.c (add_edge): Intitialize edge's attached data.
14350 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
14351 pointer parameter. Call pointed function on each edge during
14352 graph traversing. Skip traversing the edge when the function
14354 (graphds_dfs, graphds_scc): Ditto.
14355 (for_each_edge): New parameter. Pass the new parameter to callback
14357 * graphds.h (skip_edge_callback): New function pointer type.
14358 (graphds_dfs, graphds_scc): New function pointer parameter.
14359 (graphds_edge_callback, for_each_edge): New parameter.
14361 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14363 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
14364 out code checking if runtime alias check is possible to below ...
14365 Call the new function.
14366 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
14367 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
14369 2017-06-07 Marek Polacek <polacek@redhat.com>
14372 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
14373 TYPE_OVERFLOW_WRAPS check.
14375 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14377 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
14378 if versioning is required.
14379 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
14380 peeling with the check for versioning.
14382 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14384 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
14385 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
14386 Set true to new parameter if new ssa variable is defined.
14387 (vect_gen_vector_loop_niters): Refactor. Set range information
14388 for the new vector loop bound variable.
14389 (vect_do_peeling): Ditto.
14391 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14393 * tree-affine.c (ssa.h): Include header file.
14394 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
14395 has wrapping overflow behavior.
14397 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14399 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
14401 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14403 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
14404 (tree_to_aff_combination): ... here.
14406 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14408 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
14409 reg_pressure model function.
14410 (ivopts_global_cost_for_size): Delete.
14411 (determine_set_costs, iv_ca_recount_cost): Call new model function
14412 ivopts_estimate_reg_pressure.
14414 2017-06-07 Tamar Christina <tamar.christina@arm.com>
14416 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
14417 expensive than udiv. Remove floating point cases from mod.
14419 2017-06-07 Tamar Christina <tamar.christina@arm.com>
14421 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
14422 Increase idiv cost.
14424 2017-06-07 Tamar Christina <tamar.christina@arm.com>
14426 * config/aarch64/aarch64.md
14427 (copysignsf3): Fix mask generation.
14429 2017-06-07 Jakub Jelinek <jakub@redhat.com>
14431 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
14433 (class dump_manager): Add register_dumps method.
14434 * dumpfile.c: Include langhooks.h.
14435 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
14436 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
14437 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
14438 (dump_manager::dump_register): Start with 512 entries instead of 32.
14439 (dump_manager::register_dumps): New method.
14440 * toplev.c (general_init): Instead of invoking register_dumps
14441 langhook, invoke register_dumps method on the dump manager.
14442 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
14445 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
14447 * doc/md.texi: Clarify the restrictions on a define_insn condition.
14448 Say that # requires an associated define_split to exist, and that
14449 the define_split must be suitable for use after register allocation.
14451 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14453 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
14454 (compute_outgoing_frequencies): Also initialize zero counts.
14455 (find_many_sub_basic_blocks): Do not produce uninitialized profile
14456 around loops; preserve more of profile when nothing changes.
14458 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
14460 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
14462 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
14463 * config/arm/arm-cpu-cdata.h: Regenerate.
14464 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
14465 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
14466 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
14467 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
14468 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
14470 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
14472 * config/arm/t-rmprofile: Likewise.
14473 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
14475 2017-06-06 David S. Miller <davem@davemloft.net>
14478 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
14479 blockage if function uses alloca.
14481 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14483 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
14484 New "uid" fields to hold pretty-print IDs of group and ref.
14485 Memory references are now identified as <group_id>:<ref_id>
14486 instead of using [random] addresses.
14487 (dump_mem_details): Simplify, no functional change.
14488 (dump_mem_ref): Simplify and make output more concise.
14489 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
14490 (find_or_create_group): Initialize group uid.
14491 (record_ref): Initialize ref uid. Improve debug output.
14492 (prune_group_by_reuse, should_issue_prefetch_p,)
14493 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
14494 (mark_nontemporal_store, determine_loop_nest_reuse):
14495 Improve debug output.
14497 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14499 * dbgcnt.def (prefetch): New debug counter.
14500 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
14501 (schedule_prefetches): Stop issueing prefetches if debug counter
14504 2017-06-06 Tom de Vries <tom@codesourcery.com>
14506 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
14507 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
14509 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14511 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
14512 Use aarch64_reg_or_zero predicate for operand 4.
14513 (aarch64_compare_and_swap<mode> define_insn_and_split):
14514 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
14515 (aarch64_store_exclusive<mode>): Likewise for operand 2.
14517 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
14519 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
14520 (arm_compute_save_core_reg_mask): This.
14521 (thumb1_compute_save_reg_mask): Rename into ...
14522 (thumb1_compute_save_core_reg_mask): This.
14523 (arm_compute_save_reg0_reg12_mask): Adapt comment.
14524 (arm_compute_frame_layout): Likewise.
14526 2017-06-06 Richard Biener <rguenther@suse.de>
14528 PR tree-optimization/80974
14529 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
14530 keep or clear leaders SSA info.
14532 2017-06-06 Tom de Vries <tom@codesourcery.com>
14534 * config/nvptx/nvptx.c (split_mode_p): New function.
14535 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
14537 2017-06-06 Tom de Vries <tom@codesourcery.com>
14539 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
14541 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14544 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
14547 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14549 * shrink-wrap.c (handle_simple_exit): Update profile.
14550 (try_shrink_wrapping): Upate profile.
14552 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14554 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
14555 (tree_guess_outgoing_edge_probabilities): New.
14556 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
14557 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
14559 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14561 * ipa-split.c (split_function): Initialize return bb profile.
14563 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14565 * profile.c (compute_branch_probabilities): Also initialize
14566 EXIT_BLOCK profile.
14568 2017-06-06 Richard Biener <rguenther@suse.de>
14570 PR tree-optimization/80928
14571 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
14572 (vect_analyze_loop_operations): Properly guard analysis for
14574 (vect_transform_loop): Likewise.
14575 (vect_analyze_loop_2): Also reset SLP type on PHIs.
14576 (vect_model_induction_cost): Do not cost for pure SLP.
14577 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
14578 of induction in inner loop vectorization.
14579 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
14580 (vect_get_and_check_slp_defs): Handle vect_induction_def.
14581 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
14583 (vect_analyze_slp_cost_1): Cost induction.
14584 (vect_detect_hybrid_slp_stmts): Handle PHIs.
14585 (vect_get_slp_vect_defs): Likewise.
14586 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
14587 (vect_transform_stmt): Handle SLP reductions.
14588 * tree-vectorizer.h (vectorizable_induction): Adjust.
14590 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
14592 * config/rs6000/rs6000.c (make_resolver_func): Update
14593 init_lowered_empty_function call.
14595 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
14597 * doc/invoke.texi: Document the -fprofile-abs-path option.
14598 * common.opt (fprofile-abs-path): New option.
14599 * gcov-io.h (gcov_write_filename): Declare.
14600 * gcov-io.c (gcov_write_filename): New function.
14601 * coverage.c (coverage_begin_function): Use gcov_write_filename.
14602 * profile.c (output_location): Likewise.
14604 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14606 * shring-wrap.c: Revert accidental commit.
14608 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
14610 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
14612 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14614 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
14616 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
14617 profile in callgraph edge.
14618 * profile-count.h (apply_probability): If THIS is 0, then result is 0
14619 (apply_scale): Likewise.
14620 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
14621 Also scale profile when inlining function with zero profile.
14622 (initialize_cfun): Update exit block profile even when it is zero.
14623 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
14624 when profile is read.
14626 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
14628 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
14629 (CLONE_*): New constants to define the processors we can generate
14630 code for with the target_clone attribute.
14631 (rs6000_clone_map): New array to identify which clone processors
14632 the current program is running on.
14633 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
14634 target_clone attribute.
14635 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
14636 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
14637 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
14638 (cpu_expand_builtin): Add support for target_clone attribute.
14639 (rs6000_valid_attribute_p): Allow "default" attribute.
14640 (get_decl_name): New debug function to simplify printing the
14641 current function name in debugging statements.
14642 (rs6000_clone_priority): New functions to support the target_clone
14643 attribute, and be able to generate code to switch between ISA 2.05
14644 through ISA 3.0 (power6 through power9).
14645 (rs6000_compare_version_priority): Likewise.
14646 (rs6000_get_function_versions_dispatcher): Likewise.
14647 (make_resolver_func): Likewise.
14648 (add_condition_to_bb): Likewise.
14649 (dispatch_function_versions): Likewise.
14650 (rs6000_generate_version_dispatcher_body): Likewise.
14651 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
14652 (fusion_gpr_load_p): Fix a spacing issue.
14653 * doc/extend.texi (Common Function Attributes): Document that the
14654 PowerPC supports the target_clone attribute.
14656 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
14658 * config/arm/arm.h: explain F symbol found in description of ARM
14659 register allocation in its legend.
14661 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14663 * config/mips/frame-header-opt.c: Include profile-count.h.
14664 * config/riscv/riscv.c: Include profile-count.h
14666 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14668 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
14670 (sm_set_flag_if_changed): Add bbs field.
14671 (execute_sm_if_changed_flag_set): Pass BBS.
14672 (execute_sm): Update.
14674 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14676 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
14679 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14681 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
14682 (peephole2): New peephole2 to emit the above.
14683 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
14685 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14687 * config/aarch64/aarch64.c (define_peephole2 above
14688 *sub_<shift>_<mode>): New peephole.
14690 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14692 * config/i386/i386.c (make_resolver_func): Update.
14693 * Makefile.in: Add profile-count.h and profile-count.o
14694 * auto-profile.c (afdo_indirect_call): Update to new API.
14695 (afdo_set_bb_count): Update.
14696 (afdo_propagate_edge): Update.
14697 (afdo_propagate_circuit): Update.
14698 (afdo_calculate_branch_prob): Update.
14699 (afdo_annotate_cfg): Update.
14700 * basic-block.h: Include profile-count.h
14701 (struct edge_def): Turn count to profile_count.
14702 (struct basic_block_def): Likewie.
14703 (REG_BR_PROB_BASE): Move to profile-count.h
14704 (RDIV): Move to profile-count.h
14705 * bb-reorder.c (max_entry_count): Turn to profile_count.
14706 (find_traces): Update.
14707 (rotate_loop):Update.
14708 (connect_traces):Update.
14709 (sanitize_hot_paths):Update.
14710 * bt-load.c (migrate_btr_defs): Update.
14711 * cfg.c (RDIV): Remove.
14712 (init_flow): Use alloc_block.
14713 (alloc_block): Uninitialize count.
14714 (unchecked_make_edge): Uninitialize count.
14715 (check_bb_profile): Update.
14716 (dump_edge_info): Update.
14717 (dump_bb_info): Update.
14718 (update_bb_profile_for_threading): Update.
14719 (scale_bbs_frequencies_int): Update.
14720 (scale_bbs_frequencies_gcov_type): Update.
14721 (scale_bbs_frequencies_profile_count): New.
14722 * cfg.h (update_bb_profile_for_threading): Update.
14723 (scale_bbs_frequencies_profile_count): Declare.
14724 * cfgbuild.c (compute_outgoing_frequencies): Update.
14725 (find_many_sub_basic_blocks): Update.
14726 * cfgcleanup.c (try_forward_edges): Update.
14727 (try_crossjump_to_edge): Update.
14728 * cfgexpand.c (expand_gimple_tailcall): Update.
14729 (construct_exit_block): Update.
14730 * cfghooks.c (verify_flow_info): Update.
14731 (dump_bb_for_graph): Update.
14732 (split_edge): Update.
14733 (make_forwarder_block): Update.
14734 (duplicate_block): Update.
14735 (account_profile_record): Update.
14736 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
14737 (get_estimated_loop_iterations): Update.
14738 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
14739 (single_likely_exit): Update.
14740 * cfgloopmanip.c (scale_loop_profile): Update.
14742 (set_zero_probability): Update.
14743 (lv_adjust_loop_entry_edge): Update.
14744 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
14745 (purge_dead_edges): Update.
14746 (rtl_account_profile_record): Update.
14747 * cgraph.c (cgraph_node::create): Uninitialize count.
14748 (symbol_table::create_edge): Uninitialize count.
14749 (cgraph_update_edges_for_call_stmt_node): Update.
14750 (cgraph_edge::dump_edge_flags): Update.
14751 (cgraph_node::dump): Update.
14752 (cgraph_edge::maybe_hot_p): Update.
14753 * cgraph.h: Include profile-count.h
14754 (create_clone), create_edge, create_indirect_edge): Update.
14755 (cgraph_node): Turn count to profile_count.
14756 (cgraph_edge0: Likewise.
14757 (make_speculative, clone): Update.
14758 (create_edge): Update.
14759 (init_lowered_empty_function): Update.
14760 * cgraphclones.c (cgraph_edge::clone): Update.
14761 (duplicate_thunk_for_node): Update.
14762 (cgraph_node::create_clone): Update.
14763 * cgraphunit.c (cgraph_node::analyze): Update.
14764 (cgraph_node::expand_thunk): Update.
14765 * final.c (dump_basic_block_info): Update.
14766 * gimple-streamer-in.c (input_bb): Update.
14767 * gimple-streamer-out.c (output_bb): Update.
14768 * graphite.c (print_global_statistics): Update.
14769 (print_graphite_scop_statistics): Update.
14770 * hsa-brig.c: Include basic-block.h.
14771 * hsa-dump.c: Include basic-block.h.
14772 * hsa-gen.c (T sum_slice): Update.
14773 (convert_switch_statements):Update.
14774 * hsa-regalloc.c: Include basic-block.h.
14775 * ipa-chkp.c (chkp_produce_thunks): Update.
14776 * ipa-cp.c (struct caller_statistics): Update.
14777 (init_caller_stats): Update.
14778 (gather_caller_stats): Update.
14779 (ipcp_cloning_candidate_p): Update.
14780 (good_cloning_opportunity_p): Update.
14781 (get_info_about_necessary_edges): Update.
14782 (dump_profile_updates): Update.
14783 (update_profiling_info): Update.
14784 (update_specialized_profile): Update.
14785 (perhaps_add_new_callers): Update.
14786 (decide_about_value): Update.
14787 (ipa_cp_c_finalize): Update.
14788 * ipa-devirt.c (struct odr_type_warn_count): Update.
14789 (struct decl_warn_count): Update.
14790 (struct final_warning_record): Update.
14791 (possible_polymorphic_call_targets): Update.
14792 (ipa_devirt): Update.
14793 * ipa-fnsummary.c (redirect_to_unreachable): Update.
14794 * ipa-icf.c (sem_function::merge): Update.
14795 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
14796 * ipa-inline.c (compute_uninlined_call_time): Update.
14797 (compute_inlined_call_time): Update.
14798 (want_inline_small_function_p): Update.
14799 (want_inline_self_recursive_call_p): Update.
14800 (edge_badness): Update.
14801 (lookup_recursive_calls): Update.
14802 (recursive_inlining): Update.
14803 (inline_small_functions): Update.
14804 (dump_overall_stats): Update.
14805 (dump_inline_stats): Update.
14806 * ipa-profile.c (ipa_profile_generate_summary): Update.
14807 (ipa_propagate_frequency): Update.
14808 (ipa_profile): Update.
14809 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14810 * ipa-utils.c (ipa_merge_profiles): Update.
14811 * loop-doloop.c (doloop_modify): Update.
14812 * loop-unroll.c (report_unroll): Update.
14813 (unroll_loop_runtime_iterations): Update.
14814 * lto-cgraph.c (lto_output_edge): Update.
14815 (lto_output_node): Update.
14816 (input_node): Update.
14817 (input_edge): Update.
14818 (merge_profile_summaries): Update.
14819 * lto-streamer-in.c (input_cfg): Update.
14820 * lto-streamer-out.c (output_cfg): Update.
14821 * mcf.c (create_fixup_graph): Update.
14822 (adjust_cfg_counts): Update.
14823 (sum_edge_counts): Update.
14824 * modulo-sched.c (sms_schedule): Update.
14825 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
14826 * predict.c (maybe_hot_count_p): Update.
14827 (probably_never_executed): Update.
14828 (dump_prediction): Update.
14829 (combine_predictions_for_bb): Update.
14830 (propagate_freq): Update.
14831 (handle_missing_profiles): Update.
14832 (counts_to_freqs): Update.
14833 (rebuild_frequencies): Update.
14834 (force_edge_cold): Update.
14835 * predict.h: Include profile-count.h
14836 (maybe_hot_count_p, counts_to_freqs): UPdate.
14837 * print-rtl-function.c: Do not include cfg.h
14838 * print-rtl.c: Include basic-block.h
14839 * profile-count.c: New file.
14840 * profile-count.h: New file.
14841 * profile.c (is_edge_inconsistent): Update.
14842 (correct_negative_edge_counts): Update.
14843 (is_inconsistent): Update.
14844 (set_bb_counts): Update.
14845 (read_profile_edge_counts): Update.
14846 (compute_frequency_overlap): Update.
14847 (compute_branch_probabilities): Update; Initialize and deinitialize
14849 (branch_prob): Update.
14850 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
14851 (edge_gcov_count): New.
14852 (bb_gcov_count): New.
14853 * shrink-wrap.c (try_shrink_wrapping): Update.
14854 * tracer.c (better_p): Update.
14855 * trans-mem.c (expand_transaction): Update.
14856 (ipa_tm_insert_irr_call): Update.
14857 (ipa_tm_insert_gettmclone_call): Update.
14858 * tree-call-cdce.c: Update.
14859 * tree-cfg.c (gimple_duplicate_sese_region): Update.
14860 (gimple_duplicate_sese_tail): Update.
14861 (gimple_account_profile_record): Update.
14862 (execute_fixup_cfg): Update.
14863 * tree-inline.c (copy_bb): Update.
14864 (copy_edges_for_bb): Update.
14865 (initialize_cfun): Update.
14866 (freqs_to_counts): Update.
14867 (copy_cfg_body): Update.
14868 (expand_call_inline): Update.
14869 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
14870 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
14871 (try_unroll_loop_completely): Update.
14872 (try_peel_loop): Update.
14873 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14874 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
14875 * tree-ssa-loop-split.c (connect_loops): Update.
14876 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
14877 * tree-ssa-reassoc.c (branch_fixup): Update.
14878 * tree-ssa-tail-merge.c (replace_block_by): Update.
14879 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
14880 (compute_path_counts): Update.
14881 (update_profile): Update.
14882 (recompute_probabilities): Update.
14883 (update_joiner_offpath_counts): Update.
14884 (estimated_freqs_path): Update.
14885 (freqs_to_counts_path): Update.
14886 (clear_counts_path): Update.
14887 (ssa_fix_duplicate_block_edges): Update.
14888 (duplicate_thread_path): Update.
14889 * tree-switch-conversion.c (case_bit_test_cmp): Update.
14890 (struct switch_conv_info): Update.
14891 * tree-tailcall.c (decrease_profile): Update.
14892 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
14893 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14894 * value-prof.c (check_counter): Update.
14895 (gimple_divmod_fixed_value): Update.
14896 (gimple_mod_pow2): Update.
14897 (gimple_mod_subtract): Update.
14898 (gimple_ic_transform): Update.
14899 (gimple_stringop_fixed_value): Update.
14900 * value-prof.h (gimple_ic): Update.
14902 2017-06-02 Carl Love <cel@us.ibm.com>
14904 * config/rs6000/rs6000-c: Add support for built-in functions
14905 vector double vec_doublee (vector signed int);
14906 vector double vec_doublee (vector unsigned int);
14907 vector double vec_doublee (vector float);
14908 vector double vec_doubleh (vector signed int);
14909 vector double vec_doubleh (vector unsigned int);
14910 vector double vec_doubleh (vector float);
14911 vector double vec_doublel (vector signed int);
14912 vector double vec_doublel (vector unsigned int);
14913 vector double vec_doublel (vector float);
14914 vector double vec_doubleo (vector signed int);
14915 vector double vec_doubleo (vector unsigned int);
14916 vector double vec_doubleo (vector float);.
14917 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
14918 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
14920 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
14921 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
14922 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
14924 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
14925 vec_doublel, vec_doubleh.
14926 * doc/extend.texi: Update the built-in documentation file for the
14927 new built-in functions.
14929 2017-06-02 David Malcolm <dmalcolm@redhat.com>
14932 * ipa-inline-analysis.c (free_growth_caches): Set
14933 edge_removal_hook_holder to NULL after removing it.
14935 2017-06-02 Sudakshina Das <sudi.das@arm.com>
14937 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
14938 comparision with zero.
14940 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
14941 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14942 for early expansion of vec_min and vec_max builtins.
14943 (builtin_function_type): Add min/max unsigned variants to those
14944 identified as having unsigned arguments.
14946 2017-06-02 Olivier Hainque <hainque@adacore.com>
14948 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
14950 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14952 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
14953 Use VALL_F16 iterator rather than VALL.
14955 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14957 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
14958 Emit CBNZ inside loop when doing a strong exchange and comparing
14959 against zero. Generate the CC flags after the loop.
14961 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
14963 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
14964 (dl_section_ref): New.
14965 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
14966 On AIX, append an expression to subtract the size of the
14967 section length to dl_section_ref.
14969 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
14971 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14972 for early expansion of vector absolute builtins.
14974 2017-06-02 Richard Biener <rguenther@suse.de>
14976 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
14977 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
14979 2017-06-02 Richard Biener <rguenther@suse.de>
14981 PR tree-optimization/80948
14982 * tree-tailcall.c (find_tail_calls): Track stmts to move in
14983 stmt order as well.
14985 2017-06-02 Richard Biener <rguenther@suse.de>
14987 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
14989 * tree-vect-stmts.c (process_use): Do not mark backedge defs
14990 for inductions as relevant.
14992 2017-06-02 Richard Biener <rguenther@suse.de>
14994 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
14995 (vectorizable_induction): ... this. Remove dead code.
14997 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
14999 * builtins. (expand_builtin_alloca): Remove second parameter and
15000 infer its value from the first parameter instead.
15001 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
15003 2017-06-02 Jakub Jelinek <jakub@redhat.com>
15005 PR rtl-optimization/80903
15006 * loop-doloop.c (add_test): Unshare sequence.
15008 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
15010 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
15012 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
15014 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
15016 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
15017 xlogue_layout::get_instance, logue_layout::xlogue_layout,
15018 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
15019 (xlogue_layout::get_stub_rtx): Make static.
15020 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
15021 (xlogue_layout::compute_stub_managed_regs): Rename to...
15022 (xlogue_layout::count_stub_managed_regs): ...this.
15023 (xlogue_layout::is_stub_managed_reg): New function.
15024 (xlogue_layout::m_stub_names): Rename to...
15025 (xlogue_layout::s_stub_names): ...this, make static.
15026 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
15027 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
15028 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
15029 xlogue_layout::s_stub_names): Instantiate statics.
15030 (stub_managed_regs): Remove.
15031 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
15032 (disable_call_ms2sysv_xlogues): Rename to...
15033 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
15034 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
15036 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
15037 change after reload_completed.
15038 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
15040 (ix86_expand_prologue): Likewise.
15041 (ix86_expand_epilogue): Likewise.
15042 (ix86_expand_split_stack_prologue): Likewise.
15043 (ix86_compute_frame_layout): Remove frame parameter ...
15044 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
15045 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
15047 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
15048 (ix86_frame): Move from here ...
15049 * config/i386/i386.h (ix86_frame): ... to here.
15050 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
15051 complete ix86_frame data structure instead. Remove some_ld_name.
15053 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
15055 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
15056 symbols that hold a DECL_VALUE_EXPR.
15058 2017-06-01 Martin Jambor <mjambor@suse.cz>
15060 PR tree-optimization/80898
15061 * tree-sra.c (process_subtree_disqualification): Removed.
15062 (disqualify_candidate): Do not acll
15063 process_subtree_disqualification.
15064 (subtree_mark_written_and_enqueue): New function.
15065 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
15066 RHS has been disqualified and re-queue LHS if necessary. Apart
15067 from that, ignore disqualified RHS.
15069 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15071 * config/s390/s390.c (s390_emit_epilogue): Disable early return
15072 address fetch for z10 or later.
15074 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15076 * config/arc/arc.md (tst_movb): Add guard when splitting.
15078 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15080 * config/arc/arc.c (arc_can_eliminate): Test against
15081 arc_frame_pointer_needed.
15083 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15085 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
15086 to prevent store reordering.
15087 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
15088 (type): Add block type.
15089 (stack_tie): Define special instruction to be used in
15092 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15094 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
15095 constraint. It is not valid for the pattern.
15096 (noncommutative_binary_comparison): Likewise.
15098 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15100 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
15103 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15105 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
15106 be used by the reg-alloc.
15108 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15110 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
15111 reg-alloc when having mul64 or mul32x16 instructions.
15112 (mulsidi3): Likewise.
15113 (umulsidi3): Likewise.
15114 (mulsi32x16): New pattern.
15115 (mulsi64): Likewise.
15116 (mulsidi64): Likewise.
15117 (umulsidi64): Likewise.
15118 (MUL32x16_REG): Define.
15119 (mul64_600): Use MUL32x16_REG.
15120 (mac64_600): Likewise.
15121 (umul64_600): Likewise.
15122 (umac64_600): Likewise.
15124 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15126 * config/arc/arc.md (mulsi3_700): Make it commutative.
15128 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
15130 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
15132 (*sign_extendsidi2_insn): Likewise for movstosw.
15134 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
15136 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
15137 the type of the input discriminant value. Convert the
15138 discriminant value of signedness vary.
15140 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
15142 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
15143 Add to -Wall section.
15145 2017-06-01 Richard Biener <rguenther@suse.de>
15147 PR middle-end/66313
15148 * fold-const.c (fold_plusminus_mult_expr): If the factored
15149 factor may be zero use a wrapping type for the inner operation.
15150 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
15151 and handle moved defs.
15152 (process_assignment): Properly guard the unary op case. Return a
15153 tri-state indicating that moving the stmt before the call may allow
15154 to continue. Pass through to_move.
15155 (find_tail_calls): Handle moving unrelated defs before
15158 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
15161 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
15162 splitter result in the canonical way.
15164 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
15166 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
15167 also for 32bit target. Update insn attributes.
15168 (zero-extendsidi2 splitter): Allow all registers for operand 1.
15170 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
15172 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
15173 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
15174 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
15175 (_mm_maskz_min_ss): New intrinsics.
15177 2017-05-31 Martin Liska <mliska@suse.cz>
15179 * tree-vect-loop.c (vect_create_epilog_for_reduction):
15180 Change comment style to one we normally use.
15181 (vectorizable_reduction): Likewise.
15182 (vectorizable_induction): Likewise.
15183 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15184 (vectorizable_call): Likewise.
15185 (vectorizable_simd_clone_call): Likewise.
15186 (vectorizable_conversion): Likewise.
15187 (vectorizable_assignment): Likewise.
15188 (vectorizable_shift): Likewise.
15189 (vectorizable_operation): Likewise.
15190 (vectorizable_store): Likewise.
15191 (vectorizable_load): Likewise.
15192 * tree-vectorizer.h: Likewise.
15194 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
15196 * passes.c (emergency_dump_function): New.
15197 * tree-pass.h (emergency_dump_function): Declare.
15198 * plugin.c (plugins_internal_error_function): Remove.
15199 * plugin.h (plugins_internal_error_function): Remove declaration.
15200 * toplev.c (internal_error_function): New static function. Use it...
15201 (general_init): ...here.
15203 2017-05-31 Graham Markall <graham.markall@embecosm.com>
15205 * config/arc/arc.c (arc_print_operand): Handle constant operands.
15206 (arc_rtx_costs): Add costs for new patterns.
15207 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
15208 * config/arc/predicates.md: Add _1_2_3_operand predicate.
15210 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
15212 * tree-ssa-strlen.c (get_next_strinfo): New function.
15213 (get_stridx_plus_constant): Use it.
15214 (zero_length_string): Likewise.
15215 (adjust_related_strinfos): Likewise.
15216 (adjust_last_stmt): Likewise.
15218 2017-05-31 Richard Biener <rguenther@suse.de>
15221 * config/i386/i386.c (ix86_expand_builtin): Remove assert
15222 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
15224 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
15226 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
15227 loop_vinfo argument and use of dependence distance vectors.
15228 Check instead whether the two references differ only in their
15229 initial value and assume that they have the same alignment if the
15230 difference is a multiple of the vector alignment.
15231 (vect_analyze_data_refs_alignment): Update call accordingly.
15233 2017-05-31 Martin Liska <mliska@suse.cz>
15236 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
15238 2017-05-31 Bin Cheng <bin.cheng@arm.com>
15240 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
15241 (create_intersect_range_checks): Move from ...
15242 * tree-data-ref.c (create_intersect_range_checks_index)
15243 (create_intersect_range_checks): ... to here.
15244 (create_runtime_alias_checks): New function factored from ...
15245 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
15246 here. Call above function.
15247 * tree-data-ref.h (create_runtime_alias_checks): New function.
15249 2017-05-31 Bin Cheng <bin.cheng@arm.com>
15251 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
15252 segment length for dr_b and compute it in wide_int.
15254 2017-05-31 Richard Biener <rguenther@suse.de>
15256 PR tree-optimization/80906
15257 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
15258 and pass through iv_map.
15259 (copy_bb_and_scalar_dependences): Adjust.
15260 (translate_pending_phi_nodes): Likewise.
15261 (copy_loop_close_phi_args): Handle code-generating IVs instead
15264 2017-05-30 David Malcolm <dmalcolm@redhat.com>
15266 * diagnostic-color.c (color_dict): Add "type-diff".
15267 (parse_gcc_colors): Update comment.
15268 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
15269 -fdiagnostics-show-template-tree and -fno-elide-type.
15270 (GCC_COLORS): Add type-diff to example.
15272 (-fdiagnostics-show-template-tree): New.
15273 (-fno-elide-type): New.
15274 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
15275 the pp_format_decoder callback. Call any m_format_postprocessor's
15277 (pretty_printer::pretty_printer): Initialize
15278 m_format_postprocessor.
15279 (pretty_printer::~pretty_printer): Delete any
15280 m_format_postprocessor.
15281 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
15282 (class format_postprocessor): New class.
15283 (struct pretty_printer::format_decoder): Document the new parameters.
15284 (struct pretty_printer::m_format_postprocessor): New field.
15285 * tree-diagnostic.c (default_tree_printer): Update for new
15286 bool and const char ** params.
15287 * tree-diagnostic.h (default_tree_printer): Likewise.
15289 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
15291 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
15292 (lwa_operand): Delete rs6000_gen_cell_microcode test.
15293 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
15294 rs6000_gen_cell_microcode code.
15295 (rs6000_final_prescan_insn): Delete.
15296 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
15297 "warn-cell-microcode" entries.
15298 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
15299 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
15300 throughout. Change cc_reg_not_micro_cr0_operand to
15301 cc_reg_not_cr0_operand throughout.
15302 (*extendhi<mode>2_noload): Delete.
15303 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
15304 (mwarn-cell-microcode): Delete.
15305 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
15306 -mgen-cell-microcode and -mwarn-cell-microcode.
15308 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
15311 * config/i386/constraints.md (Yd): New constraint.
15313 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
15314 and (?Yd, r) alternatives. Update insn attributes.
15315 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
15316 and (?*Yd, r) alternatives. Update insn attributes.
15317 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
15319 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
15321 * gimplify.c (gimplify_modify_expr): Don't create a
15322 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
15325 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
15327 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
15329 2017-05-30 Richard Biener <rguenther@suse.de>
15331 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
15332 and reduc_def fields.
15333 (STMT_VINFO_REDUC_TYPE): New define.
15334 (STMT_VINFO_REDUC_DEF): Likewise.
15335 (vect_force_simple_reduction): Adjust prototype.
15336 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
15337 (vect_is_simple_reduction): Remove check_reduction argument.
15338 (vect_force_simple_reduction): Adjust and set
15339 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
15340 (vectorizable_reduction): Do not re-do reduction analysis
15341 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
15342 * tree-parloops.c (gather_scalar_reductions): Adjust.
15344 2017-05-30 Richard Biener <rguenther@suse.de>
15346 PR middle-end/80901
15347 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
15350 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
15352 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
15353 Introduce unknown_misalignment parameter and remove vf.
15354 (vect_peeling_hash_get_lowest_cost):
15355 Pass unknown_misalignment parameter.
15356 (vect_enhance_data_refs_alignment):
15357 Fix unsupportable data ref treatment.
15359 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15361 * tree-vect-data-refs.c (vect_get_data_access_cost):
15362 Workaround for SLP handling.
15363 (vect_enhance_data_refs_alignment):
15364 Compute costs for doing no peeling at all, compare to the best
15365 peeling costs so far and avoid peeling if cheaper.
15367 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15369 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
15370 Return peeling info and set costs to zero for unlimited cost
15372 (vect_enhance_data_refs_alignment): Also inspect all datarefs
15373 with unknown misalignment. Compute and costs for unknown
15374 misalignment, compare them to the costs for known misalignment
15375 and choose the cheapest for peeling.
15377 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15379 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
15380 (vect_get_peeling_costs_all_drs): Create function.
15381 (vect_peeling_hash_get_lowest_cost):
15382 Use vect_get_peeling_costs_all_drs.
15383 (vect_peeling_supportable): Create function.
15384 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
15386 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15388 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
15389 DR_HAS_NEGATIVE_STEP.
15390 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
15391 (vect_enhance_data_refs_alignment): Use.
15392 (vect_duplicate_ssa_name_ptr_info): Use.
15393 * tree-vectorizer.h (dr_misalignment): Use.
15394 (known_alignment_for_access_p): Use.
15396 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15399 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
15401 (has_section_name): New function.
15403 2017-05-30 Martin Liska <mliska@suse.cz>
15406 * auto-profile.c (get_function_decl_from_block): Fix
15409 2017-05-30 Richard Biener <rguenther@suse.de>
15411 PR middle-end/80876
15412 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
15414 2017-05-30 Martin Liska <mliska@suse.cz>
15416 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
15417 * dumpfile.h (struct dump_file_info): Remove ctors.
15419 2017-05-30 Martin Liska <mliska@suse.cz>
15421 * predict.def: Fix GNU coding style.
15423 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
15425 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
15426 Mark 'to' argument with ATTRIBUTE_UNUSED.
15428 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
15430 * config/xtensa/xtensa.c (xtensa_emit_call): Use
15431 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
15432 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
15435 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
15437 * doc/install.texi (Options specification): Restore entry of
15438 --enable-sjlj-exceptions.
15440 2017-05-27 Michael Eager <eager@eagercon.com>
15443 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
15445 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
15447 * config/microblaze/microblaze.h
15448 (FIXED_REGISTERS): Update in macro.
15449 (CALL_USED_REGISTERS): Update in macro.
15451 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
15453 * doc/install.texi: Add links to macOS binary distributions.
15455 2017-05-27 Jakub Jelinek <jakub@redhat.com>
15459 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15461 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
15463 2017-05-26 Martin Liska <mliska@suse.cz>
15465 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
15467 2017-05-26 Martin Liska <mliska@suse.cz>
15469 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
15470 always leading ';; '.
15471 (dump_bb_info): Likewise.
15472 (brief_dump_cfg): Likewise.
15473 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
15474 * dumpfile.c: Remove usage of TDF_VERBOSE.
15475 * dumpfile.h (enum dump_kind): Likewise.
15476 (dump_gimple_bb_header): Do not use TDF_COMMENT.
15477 * print-tree.c (debug_verbose): Remove.
15478 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
15479 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
15480 * tree-diagnostic.c (default_tree_printer): Replace
15481 TDF_DIAGNOSTIC with TDF_SLIM.
15483 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15485 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
15486 in parameter loop, rather than loop_vinfo.
15487 (create_intersect_range_checks): Ditto.
15488 (vect_create_cond_for_alias_checks): Update call to above functions.
15490 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15492 PR tree-optimization/80815
15493 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
15494 for merging runtime alias checks. Handle negative DR_STEPs.
15496 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15498 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
15500 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
15501 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
15502 out code pruning runtime alias checks.
15503 * tree-data-ref.c (prune_runtime_alias_test_list): New function
15504 factored out from above.
15505 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
15507 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
15509 (prune_runtime_alias_test_list): New decalaration.
15511 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15513 * tree-vect-data-refs.c (compare_tree): Rename and move ...
15514 * tree-data-ref.c (data_ref_compare_tree): ... to here.
15515 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
15516 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
15517 (operator==, comp_dr_with_seg_len_pair): Ditto.
15518 (vect_prune_runtime_alias_test_list): Ditto.
15520 2017-05-26 Martin Liska <mliska@suse.cz>
15523 * params.def: Bound partial-inlining-entry-probability param.
15525 2017-05-26 Marek Polacek <polacek@redhat.com>
15528 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
15531 2017-05-26 Richard Biener <rguenther@suse.de>
15533 PR tree-optimization/80842
15534 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
15537 2017-05-26 Richard Biener <rguenther@suse.de>
15539 PR tree-optimization/80844
15540 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
15542 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
15544 * doc/md.texi (Machine Constraints): Update x86 family
15545 machine constraints section to match 'config/i386/constraints.md'.
15547 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
15549 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
15551 2017-05-25 Nathan Sidwell <nathan@acm.org>
15553 * doc/invoke.texi (--enable-languages): Update documentation.
15555 2017-05-25 Martin Liska <mliska@suse.cz>
15557 * dumpfile.c: Add TDF_FOLDING.
15558 * dumpfile.h (enum dump_kind): Likewise.
15559 * genmatch.c (dt_simplify::gen_1): Use it.
15561 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15563 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
15565 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15567 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
15568 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
15570 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15572 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
15573 * match.pd (X == C): Rewrite it here.
15574 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
15575 with_certain_nonzero_bits2): New predicates.
15576 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
15578 2017-05-24 Nathan Sidwell <nathan@acm.org>
15580 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
15583 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
15586 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
15588 * config/powerpcspe: New port. Files are copied from the rs6000
15589 port, with "rs6000" in filenames replaced by "powerpcspe".
15591 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
15593 PR rtl-optimization/80754
15594 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
15596 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
15598 * config/sparc/sparc.md (length): Return the correct value for -mflat
15599 sibcalls to match output_sibcall.
15601 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
15605 * config/rs6000/rs6000.c (struct machine_function): Add new field
15607 (rs6000_get_separate_components): Init that field, use it.
15608 (rs6000_components_for_bb): Use the field.
15610 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
15612 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
15614 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
15616 PR middle-end/80823
15617 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
15619 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15622 * config/s390/s390.c (s390_check_qrst_address): Check incoming
15623 address against address_operand predicate.
15624 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
15626 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
15628 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
15629 parameters passed indirectly.
15631 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
15633 * config/i386/i386.md (*movdi_internal): Remove SSE4
15634 alternative 18 (?r, *v). Update insn attributes.
15635 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
15636 Update insn attributes.
15637 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
15638 Update insn attributes.
15639 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
15640 alternative 1 (r, v). Remove isa attribute.
15641 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
15642 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
15643 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
15645 2017-05-23 Tom de Vries <tom@codesourcery.com>
15647 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
15650 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15652 * cgraphunit.c (symbol_table::process_new_functions): Update.
15653 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
15654 (inline_generate_summary): Rename to ...
15655 (ipa_fn_summary_generate): ... this one.
15656 (inline_read_summary): Rename to ...
15657 (ipa_fn_summary_read): ... this one.
15658 (inline_write_summary): Rename to ...
15659 (ipa_fn_summary_write): ... this one.
15660 (inline_free_summary): Rename to ...
15661 (ipa_free_fn_summary): ... this one.
15662 (pass_data_local_fn_summary, pass_local_fn_summary,
15663 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
15664 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
15665 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
15666 make_pass_ipa_fn_summary): New.
15667 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
15668 inline_write_summary, inline_free_summary): Remove.
15669 (ipa_free_fn_summary) : New.
15670 * ipa-inline.c (ipa_inline): Update.
15671 (pass_ipa_inline): Do not generate summaries.
15672 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
15674 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
15675 and add pass_ipa_fn_summary.
15676 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
15678 (make_pass_inline_parameters): Remove.
15680 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
15682 * omp-low.c (struct omp_context): Remove "default_kind" member.
15685 * omp-offload.c (execute_oacc_device_lower): Remove the
15686 parallelism dimensions function attributes for unparallelized
15687 OpenACC kernels constructs.
15689 2017-05-23 Martin Liska <mliska@suse.cz>
15691 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
15693 (cgraph_edge::make_speculative): Likewise.
15694 (cgraph_edge::resolve_speculation): Likewise.
15695 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
15696 (cgraph_node::dump): Likewise.
15697 * cgraph.h: Likewise.
15698 * cgraphunit.c (analyze_functions): Likewise.
15699 (symbol_table::compile): Likewise.
15700 * ipa-cp.c (print_all_lattices): Likewise.
15701 (determine_versionability): Likewise.
15702 (initialize_node_lattices): Likewise.
15703 (ipcp_verify_propagated_values): Likewise.
15704 (estimate_local_effects): Likewise.
15705 (update_profiling_info): Likewise.
15706 (create_specialized_node): Likewise.
15707 (perhaps_add_new_callers): Likewise.
15708 (decide_about_value): Likewise.
15709 (decide_whether_version_node): Likewise.
15710 (identify_dead_nodes): Likewise.
15711 (ipcp_store_bits_results): Likewise.
15712 * ipa-devirt.c (dump_targets): Likewise.
15713 (ipa_devirt): Likewise.
15714 * ipa-icf.c (sem_item::dump): Likewise.
15715 (sem_function::equals): Likewise.
15716 (sem_variable::equals): Likewise.
15717 (sem_item_optimizer::read_section): Likewise.
15718 (sem_item_optimizer::execute): Likewise.
15719 (congruence_class::dump): Likewise.
15720 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
15721 (dump_inline_summary): Likewise.
15722 (estimate_node_size_and_time): Likewise.
15723 (inline_analyze_function): Likewise.
15724 * ipa-inline-transform.c (inline_call): Likewise.
15725 * ipa-inline.c (report_inline_failed_reason): Likewise.
15726 (want_early_inline_function_p): Likewise.
15727 (edge_badness): Likewise.
15728 (update_edge_key): Likewise.
15729 (inline_small_functions): Likewise.
15730 * ipa-profile.c (ipa_profile): Likewise.
15731 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
15732 (ipa_make_edge_direct_to_target): Likewise.
15733 (remove_described_reference): Likewise.
15734 (ipa_impossible_devirt_target): Likewise.
15735 (propagate_controlled_uses): Likewise.
15736 (ipa_print_node_params): Likewise.
15737 (ipcp_transform_function): Likewise.
15738 * ipa-pure-const.c (pure_const_read_summary): Likewise.
15739 (propagate_pure_const): Likewise.
15740 * ipa-reference.c (generate_summary): Likewise.
15741 (read_write_all_from_decl): Likewise.
15742 (propagate): Likewise.
15743 (ipa_reference_read_optimization_summary): Likewise.
15744 * ipa-utils.c (ipa_merge_profiles): Likewise.
15745 * ipa.c (walk_polymorphic_call_targets): Likewise.
15746 (symbol_table::remove_unreachable_nodes): Likewise.
15747 (ipa_single_use): Likewise.
15748 * passes.c (execute_todo): Likewise.
15749 * predict.c (drop_profile): Likewise.
15750 * symtab.c (symtab_node::get_dump_name): New function.
15751 (symtab_node::dump_name): Likewise.
15752 (symtab_node::dump_asm_name): Likewise.
15753 (symtab_node::dump_references): Likewise.
15754 (symtab_node::dump_referring): Likewise.
15755 (symtab_node::dump_base): Likewise.
15756 (symtab_node::debug_symtab): Likewise.
15757 * tree-sra.c (convert_callers_for_node): Likewise.
15758 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
15759 * value-prof.c (init_node_map): Likewise.
15761 2017-05-23 Martin Liska <mliska@suse.cz>
15763 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
15764 and symtab_node::debug_symtab to symbol_table::debug.
15765 * cgraphunit.c (analyze_functions): Use the renamed function.
15766 (symbol_table::compile): Likewise.
15767 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
15768 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
15769 * passes.c (execute_todo): Likewise.
15770 * symtab.c (symbol_table::dump): New function.
15771 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
15773 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15775 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
15776 that nonconst implies exec.
15778 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15780 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
15781 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
15782 (inline_edge_summary_vec): Turn into ...
15783 (ipa_call_summaries): ... this one.
15784 (redirect_to_unreachable, edge_set_predicate,
15785 evaluate_properties_for_edge, inline_summary_alloc,
15786 reset_ipa_call_summary, reset_inline_summary,
15787 inline_summary_t::duplicate): Update.
15788 (inline_edge_duplication_hook): Turn to ...
15789 (ipa_call_summary_t::duplicate): ... this one.
15790 (inline_edge_removal_hook): Turn to ...
15791 (ipa_call_summary_t::remove): ... this one.
15792 (dump_inline_edge_summary): Turn to ...
15793 (dump_ipa_call_summary): ... this one.
15794 (estimate_function_body_sizes): Update.
15795 (inline_update_callee_summaries): Update.
15796 (remap_edge_change_prob): Update.
15797 (remap_edge_summaries): Update.
15798 (inline_merge_summary): Update.
15799 (do_estimate_edge_time): Update.
15800 (inline_generate_summary): Update.
15801 (inline_read_section): Update.
15802 (inline_read_summary): Update.
15803 (inline_free_summary): Update.
15804 * ipa-inline.c (can_inline_edge_p): Update.
15805 (compute_inlined_call_time): Update.
15806 (want_inline_small_function_p): Update.
15807 (edge_badness): Update.
15808 (early_inliner): Update.
15809 * ipa-inline.h (inline_edge_summary): Turn to ...
15810 (ipa_call_summary): ... this one.
15811 (ipa_call_summary_t): New class.
15812 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
15813 (ipa_call_summaries): New.
15814 (inline_edge_summary): Remove.
15815 (estimate_edge_growth): Update.
15816 * ipa-profile.c (ipa_propagate_frequency_1): Update.
15817 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15818 * ipa-split.c (execute_split_functions): Update.
15819 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
15821 2017-05-23 Tom de Vries <tom@codesourcery.com>
15823 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
15824 attributes): Document rdrand effective target.
15826 2017-05-23 Tom de Vries <tom@codesourcery.com>
15828 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
15829 attributes): Sort alphabetically.
15831 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
15833 * config/avr/genmultilib.awk: Use gsub instead of gensub.
15835 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
15838 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
15839 V2DF/V2DI splat into two separate patterns, one that handles
15840 registers, and the other that only handles memory. Drop support
15841 for splatting from a GPR on ISA 2.07 and then splitting the
15842 splat into direct move and splat.
15843 (vsx_splat_<mode>_reg): Likewise.
15844 (vsx_splat_<mode>_mem): Likewise.
15846 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
15848 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
15850 2017-05-22 Jakub Jelinek <jakub@redhat.com>
15852 PR middle-end/80809
15853 * omp-low.c (finish_taskreg_remap): New function.
15854 (finish_taskreg_scan): If unit size of ctx->record_type
15855 is non-constant, unshare the size expression and replace
15856 decls in it with possible outer var refs.
15858 PR middle-end/80809
15859 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
15860 GOVD_SHARED rather than GOVD_PRIVATE with it.
15861 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
15862 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
15864 PR middle-end/80853
15865 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
15866 as last argument to build_outer_var_ref for pointer bases of array
15867 section reductions.
15869 2017-05-19 Martin Sebor <msebor@redhat.com>
15871 * print-tree.c (print_node): Print DECL_READ_P flag.
15873 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15875 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
15876 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
15877 * cgraph.c: Likewise.
15878 * cgraphunit.c: Likewise.
15879 * gengtype.c: Likewise.
15880 * ipa-cp.c: Likewise.
15881 * ipa-devirt.c: Likewise.
15882 * ipa-icf.c: Likewise.
15883 * ipa-predicate.c: Likewise.
15884 * ipa-profile.c: Likewise.
15885 * ipa-prop.c: Likewise.
15886 * ipa-split.c: Likewise.
15888 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
15889 edge_predicate_pool, dump_inline_hints,
15890 inline_summary::account_size_time, redirect_to_unreachable,
15891 edge_set_predicate, set_hint_predicate,
15892 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
15893 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
15894 inline_summary_t::remove, remap_hint_predicate_after_duplication,
15895 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
15896 ipa_call_summary_t::remove, initialize_growth_caches,
15897 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
15898 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
15899 mark_modified, unmodified_parm_1, unmodified_parm,
15900 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
15901 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
15902 compute_bb_predicates, will_be_nonconstant_expr_predicate,
15903 will_be_nonconstant_predicate, record_modified_bb_info,
15904 get_minimal_bb, record_modified, param_change_prob,
15905 phi_result_unknown_predicate, predicate_for_phi_result,
15906 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
15907 estimate_function_body_sizes, compute_inline_parameters,
15908 compute_inline_parameters_for_curren, pass_data_inline_parameters,
15909 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
15910 inline_update_callee_summaries, remap_edge_change_prob,
15911 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
15912 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
15913 inline_analyze_function, inline_summary_t::insert,
15914 inline_generate_summary, read_ipa_call_summary, inline_read_section,
15915 inline_read_summary, write_ipa_call_summary, inline_write_summary,
15916 inline_free_summary): Move to ipa-fnsummary.h
15917 (predicate_t): Remove.
15918 * ipa-fnsummary.c: New file.
15919 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
15920 (enum inline_hints_vals, inline_hints, agg_position_info,
15921 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
15922 inline_summaries, ipa_call_summary, ipa_call_summary_t,
15923 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
15924 dump_inline_summary, dump_inline_hints, inline_generate_summary,
15925 inline_read_summary, inline_write_summary, inline_free_summary,
15926 inline_analyze_function, initialize_inline_failed,
15927 inline_merge_summary, inline_update_overall_summary,
15928 compute_inline_parameters): Move to ipa-fnsummary.h
15929 * ipa-fnsummary.h: New file.
15930 * ipa-inline-transform.h: Include ipa-inline.h.
15931 * ipa-inline.c: LIkewise.
15933 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15935 * ipa-inline.c (edge_badness): Use inlined_time instead of
15936 inline_summaries->get.
15938 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15940 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
15942 2017-05-22 Nathan Sidwell <nathan@acm.org>
15944 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
15945 (fdump-lang): Document 'raw' option.
15946 * dumpfile.h (TDI_tu): Delete.
15947 * dumpfile.c (dump_files): Remove translation-unit.
15948 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
15950 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
15952 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
15953 command option from $(AWK) call.
15954 * config/avr/genmultilib.awk: Simplify and rewrite so that it
15955 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
15956 [FORMAT]: Remove handling of variable.
15957 * config/avr/t-multilib: Regenerate.
15959 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15961 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
15963 (dump_inline_summary): Do not print self_time.
15964 (estimate_function_body_sizes): Do not set self_time.
15965 (compute_inline_parameters): Likewise.
15966 (inline_read_section, inline_write_summary): Do not stream self_time.
15967 * ipa-inline.h (inline_summary): Drop self_time.
15969 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15971 * ipa-inline-analysis.c (account_size_time): Rename to ...
15972 (inline_summary::account_size_time): ... this one.
15973 (reset_ipa_call_summary): Turn to ...
15974 (ipa_call_summary::reset): ... this one.
15975 (reset_inline_summary): Turn to ...
15976 (inline_summary::reset): ... this one.
15977 (inline_summary_t::remove): Update.
15978 (inline_summary_t::duplicate): Update.
15979 (ipa_call_summary_t::remove): Update.
15980 (dump_inline_summary): Update.
15981 (estimate_function_body_sizes): Update.
15982 (compute_inline_parameters): Update.
15983 (estimate_node_size_and_time): Update.
15984 (inline_merge_summary): Update.
15985 (inline_update_overall_summary): Update.
15986 (inline_read_section): Update.
15987 (inline_write_summary): Update.
15988 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
15989 add account_size_time and reset member functions.
15990 (ipa_call_summary): Add reset function.
15991 * ipa-predicate.h (predicate::operator &): Constify.
15993 2017-05-22 Richard Biener <rguenther@suse.de>
15995 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
15997 2017-05-19 Jason Merrill <jason@redhat.com>
15999 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
16001 2017-05-19 Marek Polacek <polacek@redhat.com>
16004 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
16005 TYPE_OVERFLOW_WRAPS checks.
16007 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
16009 * tree-core.h (enum omp_clause_default_kind): Add
16010 "OMP_CLAUSE_DEFAULT_PRESENT".
16011 * tree-pretty-print.c (dump_omp_clause): Handle it.
16012 * gimplify.c (enum gimplify_omp_var_data): Add
16013 "GOVD_MAP_FORCE_PRESENT".
16014 (gimplify_adjust_omp_clauses_1): Map it to
16015 "GOMP_MAP_FORCE_PRESENT".
16016 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
16018 * gimplify.c (oacc_default_clause): Clarify.
16020 2017-05-19 Nathan Sidwell <nathan@acm.org>
16022 LANG_HOOK_REGISTER_DUMPS
16023 * toplev.c (general_init): Call register dump lang hook.
16024 * doc/invoke.texi: Document -fdump-lang option family.
16025 * dumpfile.c (dump_files): Remove class dump here.
16026 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
16027 * dumpfile.h (tree_dump_index): Remove TDI_class.
16028 * langhooks-def.h (lhd_register_dumps): Declare.
16029 (LANG_HOOKS_REGISTER_DUMPS): Define.
16030 (LANG_HOOKS_INITIALIZER): Add it.
16031 * langhooks.c (lhd_register_dumps): Define.
16032 * langhooks.h (struct lang_hooks): Add register_dumps.
16034 2017-05-19 Nathan Sidwell <nathan@acm.org>
16036 * context.h (context::set_passes): New.
16037 * context.c (context::context): Do not create pass manager.
16038 * toplev.c (general_init): Create pass manager here.
16040 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
16042 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
16043 use this splitter if two add or or instructions would also work for
16044 the constant we want to generate.
16046 2017-05-19 Richard Biener <rguenther@suse.de>
16049 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
16050 predicate evaluation.
16052 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
16054 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
16056 * ipa-inline.c (want_inline_small_function_p): Do not cast to
16059 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
16061 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
16062 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
16063 (inline_edge_summary_vec): Turn into ...
16064 (ipa_call_summaries): ... this one.
16065 (redirect_to_unreachable, edge_set_predicate,
16066 evaluate_properties_for_edge, inline_summary_alloc,
16067 reset_ipa_call_summary, reset_inline_summary,
16068 inline_summary_t::duplicate): Update.
16069 (inline_edge_duplication_hook): Turn to ...
16070 (ipa_call_summary_t::duplicate): ... this one.
16071 (inline_edge_removal_hook): Turn to ...
16072 (ipa_call_summary_t::remove): ... this one.
16073 (dump_inline_edge_summary): Turn to ...
16074 (dump_ipa_call_summary): ... this one.
16075 (estimate_function_body_sizes): Update.
16076 (inline_update_callee_summaries): Update.
16077 (remap_edge_change_prob): Update.
16078 (remap_edge_summaries): Update.
16079 (inline_merge_summary): Update.
16080 (do_estimate_edge_time): Update.
16081 (inline_generate_summary): Update.
16082 (inline_read_section): Update.
16083 (inline_read_summary): Update.
16084 (inline_free_summary): Update.
16085 * ipa-inline.c (can_inline_edge_p): Update.
16086 (compute_inlined_call_time): Update.
16087 (want_inline_small_function_p): Update.
16088 (edge_badness): Update.
16089 (early_inliner): Update.
16090 * ipa-inline.h (inline_edge_summary): Turn to ...
16091 (ipa_call_summary): ... this one.
16092 (ipa_call_summary_t): New class.
16093 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
16094 (ipa_call_summaries): New.
16095 (inline_edge_summary): Remove.
16096 (estimate_edge_growth): Update.
16097 * ipa-profile.c (ipa_propagate_frequency_1): Update.
16098 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
16099 * ipa-split.c (execute_split_functions): Update.
16100 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
16102 2017-05-19 Richard Biener <rguenther@suse.de>
16104 PR middle-end/80764
16105 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
16107 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
16109 * config/rs6000/rs6000.c (struct machine_function): Add field
16110 fpr_is_wrapped_separately.
16111 (rs6000_get_separate_components): Use 64 components. Handle the
16112 new FPR components.
16113 (rs6000_components_for_bb): Handle the FPR components.
16114 (rs6000_emit_prologue_components): Handle the FPR components.
16115 (rs6000_emit_epilogue_components): Handle the FPR components.
16116 (rs6000_set_handled_components): Handle the FPR components.
16117 (rs6000_emit_prologue): Don't output prologue code for those FPRs
16118 that are already separately shrink-wrapped.
16119 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
16120 that are already separately shrink-wrapped.
16122 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
16125 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
16128 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
16129 (define_peephole2 for Altivec d-form load): Add peepholes to catch
16130 cases where the register allocator uses a move and an offsettable
16131 memory operation to/from a FPR register on ISA 2.06/2.07.
16132 (define_peephole2 for Altivec d-form store): Likewise.
16134 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
16137 * config/i386/mmx.md (*mov<mode>_internal): Enable
16138 alternatives 11, 12, 13 and 14 also for 32bit targets.
16139 Remove alternatives 15, 16, 17 and 18.
16140 * config/i386/sse.md (vec_concatv2di): Change
16141 alternative (!x, *y) to (x, ?!*Yn).
16143 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
16145 * dumpfile.h (enum dump_kind): Remove stray comma.
16147 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
16149 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
16150 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
16151 predicate::num_conditions
16152 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
16153 (CHANGED): turn into predicate::changed.
16154 (agg_position_info): Move to ipa-predicate.h
16155 (add_condition, predicate::add_clause, predicate::operator &=,
16156 predicate::or_with, predicate::evaluate, predicate::probability,
16157 dump_condition, dump_clause, predicate::dump,
16158 predicate::remap_after_duplication, predicate::remap_after_inlining,
16159 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
16160 (evaluate_conditions_for_known_args): Update.
16161 (set_cond_stmt_execution_predicate): Update.
16162 * ipa-inline.h: Include ipa-predicate.h
16163 (condition, inline_param_summary, conditions, agg_position_info,
16164 predicate): Move to ipa-predicate.h
16165 * ipa-predicate.c: New file.
16166 * ipa-predicate.h: New file.
16168 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
16170 * final.c (leaf_function_p): Check we are not in a sequence.
16172 2017-05-18 Martin Liska <mliska@suse.cz>
16174 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
16175 * dumpfile.c (dump_register): Use new enum dump_kind.
16176 (get_dump_file_name): Likewise.
16177 (dump_enable_all): Likewise.
16178 (dump_switch_p_1): Likewise.
16179 (enable_rtl_dump_file): Remove usage of TDF_RTL.
16180 * dumpfile.h (enum dump_kind): New enum type.
16181 (struct dump_file_info): Create constructor and
16182 format fields and comments.
16183 * passes.c (pass_manager::register_one_dump_file):
16185 * statistics.c (statistics_early_init): Likewise.
16186 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
16187 TDF_TREE with TDF_SLIM.
16188 (gather_memory_references_ref): Likewise.
16190 2017-05-18 Martin Liska <mliska@suse.cz>
16192 * vec.h (struct vnull): Use it.
16194 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
16196 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
16197 (true_predicate, false_predicate, true_predicate_p,
16198 false_predicate_p): Remove.
16199 (single_cond_predicate, not_inlined_predicate): Turn to member function
16201 (add_condition): Update.
16202 (add_clause): Turn to...
16203 (predicate::add_clause): ... this one; update; allow passing NULL
16205 (and_predicates): Turn to ...
16206 (predicate::operator &=): ... this one.
16207 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
16208 (or_predicates): Turn to ...
16209 (predicate::or_with): ... this one.
16210 (evaluate_predicate): Turn to ...
16211 (predicate::evaluate): ... this one.
16212 (predicate_probability): Turn to ...
16213 (predicate::probability): ... this one.
16214 (dump_condition): Update.
16215 (dump_predicate): Turn to ...
16216 (predicate::dump): ... this one.
16217 (account_size_time): Update.
16218 (edge_set_predicate): Update.
16219 (set_hint_predicate): UPdate.
16220 (evaluate_conditions_for_known_args): Update.
16221 (evaluate_properties_for_edge): Update.
16222 (remap_predicate_after_duplication): Turn to...
16223 (predicate::remap_after_duplication): ... this one.
16224 (remap_hint_predicate_after_duplication): Update.
16225 (inline_summary_t::duplicate): UPdate.
16226 (dump_inline_edge_summary): Update.
16227 (dump_inline_summary): Update.
16228 (set_cond_stmt_execution_predicate): Update.
16229 (set_switch_stmt_execution_predicate): Update.
16230 (compute_bb_predicates): Update.
16231 (will_be_nonconstant_expr_predicate): Update.
16232 (will_be_nonconstant_predicate): Update.
16233 (phi_result_unknown_predicate): Update.
16234 (predicate_for_phi_result): Update.
16235 (array_index_predicate): Update.
16236 (estimate_function_body_sizes): Update.
16237 (estimate_node_size_and_time): Update.
16238 (estimate_ipcp_clone_size_and_time): Update.
16239 (remap_predicate): Rename to ...
16240 (predicate::remap_after_inlining): ... this one.
16241 (remap_hint_predicate): Update.
16242 (inline_merge_summary): Update.
16243 (inline_update_overall_summary): Update.
16244 (estimate_size_after_inlining): Update.
16245 (read_predicate): Rename to ...
16246 (predicate::stream_in): ... this one.
16247 (read_inline_edge_summary): Update.
16248 (write_predicate): Rename to ...
16249 (predicate::stream_out): ... this one.
16250 (write_inline_edge_summary): Update.
16251 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
16252 (clause_t): Turn to uint32_t
16253 (predicate): Turn to class; implement constructor and operators
16255 (size_time_entry): Update.
16256 (inline_summary): Update.
16257 (inline_edge_summary): Update.
16259 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
16261 * fold-const.c (fold_binary_loc): Move transformation...
16262 * match.pd (C - X CMP X): ... here.
16264 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
16266 * config/sparc/sparc.c (sparc_option_override): Set function
16267 alignment for -mcpu=niagara7 to 64 to match the I$ line.
16268 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
16270 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
16272 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
16274 2017-05-18 Marek Polacek <polacek@redhat.com>
16277 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
16278 (pass_ubsan::execute): Call gimple_assign_single_p instead of
16279 gimple_assign_load_p.
16281 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
16283 PR middle-end/80692
16284 * real.c (do_compare): Give decimal_do_compare preference over
16285 comparing just the signs.
16287 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
16289 * doc/md.texi (Canonicalization of Instructions): Describe the
16290 canonical form of instructions that inherently set a condition
16293 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
16295 PR middle-end/80775
16296 * tree-cfg.c: Move deletion of unreachable case statements to after
16297 the merging of consecutive case labels.
16299 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
16301 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
16302 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
16303 restoring of callee-saved registers.
16305 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
16307 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
16308 * config/visium/visium.c (single_set_and_flags): Likewise.
16309 * config/visium/visium.md (Substitutions): Likewise.
16311 2017-05-17 Martin Liska <mliska@suse.cz>
16313 * cfg.c: Introduce dump_flags_t type and
16314 use it instead of int type.
16316 * cfghooks.c: Likewise.
16317 * cfghooks.h (struct cfg_hooks): Likewise.
16318 * cfgrtl.c: Likewise.
16319 * cfgrtl.h: Likewise.
16320 * cgraph.c (cgraph_node::get_body): Likewise.
16321 * coretypes.h: Likewise.
16322 * domwalk.c: Likewise.
16323 * domwalk.h: Likewise.
16324 * dumpfile.c (struct dump_option_value_info): Likewise.
16325 (dump_enable_all): Likewise.
16326 (dump_switch_p_1): Likewise.
16327 (opt_info_switch_p): Likewise.
16328 * dumpfile.h (enum tree_dump_index): Likewise.
16329 (struct dump_file_info): Likewise.
16330 * genemit.c: Likewise.
16331 * generic-match-head.c: Likewise.
16332 * gengtype.c (open_base_files): Likewise.
16333 * gimple-pretty-print.c: Likewise.
16334 * gimple-pretty-print.h: Likewise.
16335 * graph.c (print_graph_cfg): Likewise.
16336 * graphite-scop-detection.c (dot_all_sese): Likewise.
16337 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
16338 * loop-unroll.c (report_unroll): Likewise.
16339 * passes.c (pass_manager::register_one_dump_file): Likewise.
16340 * print-tree.c: Likewise.
16341 * statistics.c: Likewise.
16342 * tree-cfg.c: Likewise.
16343 * tree-cfg.h: Likewise.
16344 * tree-dfa.c: Likewise.
16345 * tree-dfa.h: Likewise.
16346 * tree-dump.c (dump_function): Likewise.
16347 * tree-dump.h (struct dump_info): Likewise.
16348 * tree-pretty-print.c: Likewise.
16349 * tree-pretty-print.h: Likewise.
16350 * tree-ssa-live.c: Likewise.
16351 * tree-ssa-live.h: Likewise.
16352 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
16353 * tree-vect-loop.c: Likewise.
16354 * tree-vect-slp.c: Likewise.
16356 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
16357 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16359 PR tree-optimization/80457
16360 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
16361 of all arguments to a statement as scalar_to_vec operations.
16362 (vectorizable_call): Adjust call to vect_model_simple_cost for
16364 (vectorizable_conversion): Likewise.
16365 (vectorizable_assignment): Likewise.
16366 (vectorizable_shift): Likewise.
16367 (vectorizable_operation): Likewise.
16368 (vectorizable_comparison): Likewise.
16369 (vect_is_simple_cond): Record the def types for operands.
16370 (vectorizable_condition): Likewise, call vect_model_simple_cost.
16371 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
16372 for statement argument count.
16374 2017-05-16 Carl Love <cel@us.ibm.com>
16376 * config/rs6000/rs6000-c: Add support for built-in functions
16377 vector unsigned long long vec_bperm (vector unsigned long long,
16378 vector unsigned char)
16379 vector signed long long vec_mule (vector signed int,
16381 vector unsigned long long vec_mule (vector unsigned int,
16382 vector unsigned int)
16383 vector signed long long vec_mulo (vector signed int,
16385 vector unsigned long long vec_mulo (vector unsigned int,
16386 vector unsigned int)
16387 vector signed char vec_sldw (vector signed char,
16388 vector signed char,
16390 vector unsigned char vec_sldw (vector unsigned char,
16391 vector unsigned char,
16393 vector signed short vec_sldw (vector signed short,
16394 vector signed short,
16396 vector unsigned short vec_sldw (vector unsigned short,
16397 vector unsigned short,
16399 vector signed int vec_sldw (vector signed int,
16402 vector unsigned int vec_sldw (vector unsigned int,
16403 vector unsigned int,
16405 vector signed long long vec_sldw (vector signed long long,
16406 vector signed long long,
16408 vector unsigned long long vec_sldw (vector unsigned long long,
16409 vector unsigned long long,
16411 * config/rs6000/rs6000-c: Add support for built-in functions
16412 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
16413 * config/rs6000/altivec.h: Add defintion for vec_sldw.
16414 * doc/extend.texi: Update the built-in documentation for the
16415 new built-in functions.
16417 2017-05-16 Marek Polacek <polacek@redhat.com>
16421 * tree.c (save_expr): Don't fold the expression.
16423 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
16425 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
16426 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
16427 and (?*y,m). Update insn attributes.
16429 2017-05-16 Martin Liska <mliska@suse.cz>
16431 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
16432 flags argument of print_gimple_stmt, print_gimple_expr,
16433 print_generic_stmt and print_generic_expr.
16434 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
16435 * coretypes.h: Likewise.
16436 * except.c (dump_eh_tree): Likewise.
16437 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
16438 * gimple-pretty-print.h: Likewise.
16439 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
16440 (backprop::push_to_worklist): Likewise.
16441 (backprop::pop_from_worklist): Likewise.
16442 (backprop::process_use): Likewise.
16443 (backprop::intersect_uses): Likewise.
16444 (note_replacement): Likewise.
16445 * gimple-ssa-store-merging.c
16446 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
16447 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
16448 (pass_store_merging::execute): Likewise.
16449 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
16450 (ssa_base_cand_dump_callback): Likewise.
16451 (dump_incr_vec): Likewise.
16452 (replace_refs): Likewise.
16453 (replace_mult_candidate): Likewise.
16454 (create_add_on_incoming_edge): Likewise.
16455 (create_phi_basis): Likewise.
16456 (insert_initializers): Likewise.
16457 (all_phi_incrs_profitable): Likewise.
16458 (introduce_cast_before_cand): Likewise.
16459 (replace_one_candidate): Likewise.
16460 * gimplify.c (gimplify_expr): Likewise.
16461 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
16462 (set_rename): Likewise.
16463 (rename_uses): Likewise.
16464 (copy_loop_phi_nodes): Likewise.
16465 (add_close_phis_to_merge_points): Likewise.
16466 (copy_loop_close_phi_args): Likewise.
16467 (copy_cond_phi_args): Likewise.
16468 (graphite_copy_stmts_from_block): Likewise.
16469 (translate_pending_phi_nodes): Likewise.
16470 * graphite-poly.c (print_pdr): Likewise.
16471 (dump_gbb_cases): Likewise.
16472 (dump_gbb_conditions): Likewise.
16473 (print_scop_params): Likewise.
16474 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
16475 (build_cross_bb_scalars_use): Likewise.
16476 (gather_bbs::before_dom_children): Likewise.
16477 * hsa-dump.c (dump_hsa_immed): Likewise.
16478 * ipa-cp.c (print_ipcp_constant_value): Likewise.
16479 (get_replacement_map): Likewise.
16480 * ipa-inline-analysis.c (dump_condition): Likewise.
16481 (estimate_function_body_sizes): Likewise.
16482 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
16483 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
16484 * ipa-prop.c (ipa_dump_param): Likewise.
16485 (ipa_print_node_jump_functions_for_edge): Likewise.
16486 (ipa_modify_call_arguments): Likewise.
16487 (ipa_modify_expr): Likewise.
16488 (ipa_dump_param_adjustments): Likewise.
16489 (ipa_dump_agg_replacement_values): Likewise.
16490 (ipcp_modif_dom_walker::before_dom_children): Likewise.
16491 * ipa-pure-const.c (check_stmt): Likewise.
16492 (pass_nothrow::execute): Likewise.
16493 * ipa-split.c (execute_split_functions): Likewise.
16494 * omp-offload.c (dump_oacc_loop_part): Likewise.
16495 (dump_oacc_loop): Likewise.
16496 * trans-mem.c (tm_log_emit): Likewise.
16497 (tm_memopt_accumulate_memops): Likewise.
16498 (dump_tm_memopt_set): Likewise.
16499 (dump_tm_memopt_transform): Likewise.
16500 * tree-cfg.c (gimple_verify_flow_info): Likewise.
16501 (print_loop): Likewise.
16502 * tree-chkp-opt.c (chkp_print_addr): Likewise.
16503 (chkp_gather_checks_info): Likewise.
16504 (chkp_get_check_result): Likewise.
16505 (chkp_remove_check_if_pass): Likewise.
16506 (chkp_use_outer_bounds_if_possible): Likewise.
16507 (chkp_reduce_bounds_lifetime): Likewise.
16508 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
16509 (chkp_mark_completed_bounds): Likewise.
16510 (chkp_register_incomplete_bounds): Likewise.
16511 (chkp_mark_invalid_bounds): Likewise.
16512 (chkp_maybe_copy_and_register_bounds): Likewise.
16513 (chkp_build_returned_bound): Likewise.
16514 (chkp_get_bound_for_parm): Likewise.
16515 (chkp_build_bndldx): Likewise.
16516 (chkp_get_bounds_by_definition): Likewise.
16517 (chkp_generate_extern_var_bounds): Likewise.
16518 (chkp_get_bounds_for_decl_addr): Likewise.
16519 * tree-chrec.c (chrec_apply): Likewise.
16520 * tree-data-ref.c (dump_data_reference): Likewise.
16521 (dump_subscript): Likewise.
16522 (dump_data_dependence_relation): Likewise.
16523 (analyze_overlapping_iterations): Likewise.
16524 * tree-inline.c (expand_call_inline): Likewise.
16525 (tree_function_versioning): Likewise.
16526 * tree-into-ssa.c (dump_defs_stack): Likewise.
16527 (dump_currdefs): Likewise.
16528 (dump_names_replaced_by): Likewise.
16529 (dump_update_ssa): Likewise.
16530 (update_ssa): Likewise.
16531 * tree-object-size.c (pass_object_sizes::execute): Likewise.
16532 * tree-parloops.c (build_new_reduction): Likewise.
16533 (try_create_reduction_list): Likewise.
16534 (ref_conflicts_with_region): Likewise.
16535 (oacc_entry_exit_ok_1): Likewise.
16536 (oacc_entry_exit_single_gang): Likewise.
16537 * tree-pretty-print.h: Likewise.
16538 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
16539 (get_scalar_evolution): Likewise.
16540 (add_to_evolution): Likewise.
16541 (get_loop_exit_condition): Likewise.
16542 (analyze_evolution_in_loop): Likewise.
16543 (analyze_initial_condition): Likewise.
16544 (analyze_scalar_evolution): Likewise.
16545 (instantiate_scev): Likewise.
16546 (number_of_latch_executions): Likewise.
16547 (gather_chrec_stats): Likewise.
16548 (final_value_replacement_loop): Likewise.
16549 (scev_const_prop): Likewise.
16550 * tree-sra.c (dump_access): Likewise.
16551 (disqualify_candidate): Likewise.
16552 (create_access): Likewise.
16553 (reject): Likewise.
16554 (maybe_add_sra_candidate): Likewise.
16555 (create_access_replacement): Likewise.
16556 (analyze_access_subtree): Likewise.
16557 (analyze_all_variable_accesses): Likewise.
16558 (sra_modify_assign): Likewise.
16559 (initialize_constant_pool_replacements): Likewise.
16560 (find_param_candidates): Likewise.
16561 (decide_one_param_reduction): Likewise.
16562 (replace_removed_params_ssa_names): Likewise.
16563 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
16564 * tree-ssa-copy.c (dump_copy_of): Likewise.
16565 (copy_prop_visit_cond_stmt): Likewise.
16566 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
16567 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
16568 (record_equivalences_from_stmt): Likewise.
16569 * tree-ssa-dse.c (compute_trims): Likewise.
16570 (delete_dead_call): Likewise.
16571 (delete_dead_assignment): Likewise.
16572 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
16573 (forward_propagate_into_cond): Likewise.
16574 (pass_forwprop::execute): Likewise.
16575 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16576 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
16578 (move_computations_worker): Likewise.
16579 (execute_sm): Likewise.
16580 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
16581 (remove_exits_and_undefined_stmts): Likewise.
16582 (remove_redundant_iv_tests): Likewise.
16583 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
16584 (adjust_iv_update_pos): Likewise.
16585 * tree-ssa-math-opts.c (bswap_replace): Likewise.
16586 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
16587 (value_replacement): Likewise.
16588 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
16589 * tree-ssa-pre.c (print_pre_expr): Likewise.
16590 (get_representative_for): Likewise.
16591 (create_expression_by_pieces): Likewise.
16592 (insert_into_preds_of_block): Likewise.
16593 (eliminate_insert): Likewise.
16594 (eliminate_dom_walker::before_dom_children): Likewise.
16595 (eliminate): Likewise.
16596 (remove_dead_inserted_code): Likewise.
16597 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
16598 * tree-ssa-reassoc.c (get_rank): Likewise.
16599 (eliminate_duplicate_pair): Likewise.
16600 (eliminate_plus_minus_pair): Likewise.
16601 (eliminate_not_pairs): Likewise.
16602 (undistribute_ops_list): Likewise.
16603 (eliminate_redundant_comparison): Likewise.
16604 (update_range_test): Likewise.
16605 (optimize_range_tests_var_bound): Likewise.
16606 (optimize_vec_cond_expr): Likewise.
16607 (rewrite_expr_tree): Likewise.
16608 (rewrite_expr_tree_parallel): Likewise.
16609 (linearize_expr): Likewise.
16610 (break_up_subtract): Likewise.
16611 (linearize_expr_tree): Likewise.
16612 (attempt_builtin_powi): Likewise.
16613 (attempt_builtin_copysign): Likewise.
16614 (transform_stmt_to_copy): Likewise.
16615 (transform_stmt_to_multiply): Likewise.
16616 (dump_ops_vector): Likewise.
16617 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
16618 (print_scc): Likewise.
16619 (set_ssa_val_to): Likewise.
16620 (visit_reference_op_store): Likewise.
16621 (visit_use): Likewise.
16622 (sccvn_dom_walker::before_dom_children): Likewise.
16623 (run_scc_vn): Likewise.
16624 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
16626 (expr_hash_elt::print): Likewise.
16627 (const_and_copies::pop_to_marker): Likewise.
16628 (const_and_copies::record_const_or_copy_raw): Likewise.
16629 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
16630 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
16631 (dump_predicates): Likewise.
16632 (find_uninit_use): Likewise.
16633 (warn_uninitialized_phi): Likewise.
16634 (pass_late_warn_uninitialized::execute): Likewise.
16635 * tree-ssa.c (verify_vssa): Likewise.
16636 (verify_ssa): Likewise.
16637 (maybe_optimize_var): Likewise.
16638 * tree-vrp.c (dump_value_range): Likewise.
16639 (dump_all_value_ranges): Likewise.
16640 (dump_asserts_for): Likewise.
16641 (register_edge_assert_for_2): Likewise.
16642 (vrp_visit_cond_stmt): Likewise.
16643 (vrp_visit_switch_stmt): Likewise.
16644 (vrp_visit_stmt): Likewise.
16645 (vrp_visit_phi_node): Likewise.
16646 (simplify_cond_using_ranges_1): Likewise.
16647 (fold_predicate_in): Likewise.
16648 (evrp_dom_walker::before_dom_children): Likewise.
16649 (evrp_dom_walker::push_value_range): Likewise.
16650 (evrp_dom_walker::pop_value_range): Likewise.
16651 (execute_early_vrp): Likewise.
16653 2017-05-16 Richard Biener <rguenther@suse.de>
16655 * dwarf2out.c (loc_list_from_tree_1): Do not create
16656 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
16658 2017-05-16 Richard Biener <rguenther@suse.de>
16660 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
16662 (note_variable_value_in_expr): If we resolved the decl ref
16663 do not push to the stack.
16665 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
16667 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
16668 operations in fast-math mode.
16669 (vaddq_f16): Likewise.
16670 (vmul_f16): Likewise.
16671 (vmulq_f16): Likewise.
16672 (vsub_f16): Likewise.
16673 (vsubq_f16): Likewise.
16674 * config/arm/neon.md (add<mode>3): New.
16676 (fma:<VH:mode>3): New. Also remove outdated comment.
16679 2017-05-16 Martin Liska <mliska@suse.cz>
16683 * ipa-devirt.c (warn_types_mismatch): Fix typo.
16684 (odr_types_equivalent_p): Likewise.
16686 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
16688 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
16690 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
16693 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
16694 non-interunit SSE move alternatives with '?'.
16695 (zero-extendsidi peephole2): New peephole to skip intermediate
16696 general register in SSE zero-extend sequence.
16698 2017-05-15 Jeff Law <law@redhat.com>
16700 * reorg.c (relax_delay_slots): Create a new variable to hold
16701 the temporary target rather than clobbering TARGET_LABEL.
16703 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
16704 missing argument to extract_bit_field call.
16705 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
16707 2017-05-15 Martin Liska <mliska@suse.cz>
16710 * gcc.c (process_command): Do not allow empty argument of -o option.
16712 2017-05-15 Renlin Li <renlin.li@arm.com>
16714 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
16715 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
16716 * config/aarch64/constraints.md (Usf): Add long call check.
16717 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
16718 (call_value): Likewise.
16719 (sibcall): Likewise.
16720 (sibcall_value): Likewise.
16722 (call_value_insn): New.
16723 (sibcall_insn): Update rtx pattern.
16724 (sibcall_value_insn): Likewise.
16725 (call_internal): Remove.
16726 (call_value_internal): Likewise.
16727 (sibcall_internal): Likewise.
16728 (sibcall_value_internal): Likewise.
16729 (call_reg): Likewise.
16730 (call_symbol): Likewise.
16731 (call_value_reg): Likewise.
16732 (call_value_symbol): Likewise.
16734 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
16737 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
16739 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
16741 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
16742 compatible with CCGOCmode and with CCZmode.
16744 2017-05-14 Martin Sebor <msebor@redhat.com>
16746 PR middle-end/77671
16747 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
16748 (gimple_fold_builtin_snprintf): Same.
16749 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
16750 (gimple_fold_builtin_snprintf): Same.
16751 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
16752 of character types.
16753 (is_call_safe): New function.
16754 (try_substitute_return_value): Call it.
16755 (try_simplify_call): New function.
16756 (pass_sprintf_length::handle_gimple_call): Call it.
16758 2017-05-14 Martin Sebor <msebor@redhat.com>
16760 PR middle-end/80669
16761 * builtins.c (expand_builtin_stpncpy): Simplify.
16763 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
16765 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
16766 * config/i386/i386.h
16767 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
16768 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
16769 (struct machine_function): Add new members call_ms2sysv,
16770 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
16771 (struct machine_frame_state): New fields sp_realigned and
16772 sp_realigned_offset.
16773 * config/i386/i386.c
16774 (enum xlogue_stub): New enum.
16775 (enum xlogue_stub_sets): New enum.
16776 (class xlogue_layout): New class.
16777 (struct ix86_frame): New fields stack_realign_allocate_offset,
16778 stack_realign_offset and outlined_save_offset. Modify comments to
16779 detail stack layout when using out-of-line stubs.
16780 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
16781 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
16782 -mcall-ms2sysv-xlogues.
16783 (stub_managed_regs): New static variable.
16784 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
16785 registers managed by out-of-line stub.
16786 (disable_call_ms2sysv_xlogues): New function.
16787 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
16788 m->call_ms2sysv when appropriate and compute frame layout for
16790 (sp_valid_at, fp_valid_at): New inline functions.
16791 (choose_basereg): New function.
16792 (choose_baseaddr): Add align parameter, use choose_basereg and modify
16794 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
16795 Use align parameter of choose_baseaddr to generated aligned SSE movs
16797 (pro_epilogue_adjust_stack): Modify to track
16798 machine_frame_state::sp_realigned.
16799 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
16800 (ix86_nsaved_sseregs): Likewise.
16801 (ix86_emit_save_regs): Likewise.
16802 (ix86_emit_save_regs_using_mov): Likewise.
16803 (ix86_emit_save_sse_regs_using_mov): Likewise.
16804 (get_scratch_register_on_entry): Likewise.
16805 (gen_frame_set): New function.
16806 (gen_frame_load): Likewise.
16807 (gen_frame_store): Likewise.
16808 (emit_outlined_ms2sysv_save): Likewise.
16809 (emit_outlined_ms2sysv_restore): Likewise.
16810 (ix86_expand_prologue): Modify stack re-alignment code and call
16811 emit_outlined_ms2sysv_save when appropriate.
16812 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
16813 parameter rtx_insn *insn, which allows the function to be used to only
16814 generate the notes.
16815 (ix86_expand_epilogue): Modify validity checks of frame and stack
16816 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
16817 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
16818 * config/i386/predicates.md
16819 (save_multiple): New predicate.
16820 (restore_multiple): Likewise.
16821 * config/i386/sse.md
16822 (save_multiple<mode>): New pattern.
16823 (save_multiple_realign<mode>): Likewise.
16824 (restore_multiple<mode>): Likewise.
16825 (restore_multiple_and_return<mode>): Likewise.
16826 (restore_multiple_leave_return<mode>): Likewise.
16827 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
16829 2017-05-14 Julia Koval <julia.koval@intel.com>
16831 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
16832 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
16833 (__builtin_ia32_xsetbv): New builtins.
16834 * config/i386/i386.c (ix86_expand_special_args_builtin):
16836 (ix86_expand_builtin): Special expand for new intrinsics.
16837 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
16838 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
16839 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
16841 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16843 * cfganal.c (inverted_post_order_compute): Change argument type
16845 * cfganal.h (inverted_post_order_compute): Adjust prototype.
16846 * df-core.c (rest_of_handle_df_initialize): Adjust.
16847 (rest_of_handle_df_finish): Likewise.
16848 (df_analyze_1): Likewise.
16849 (df_analyze): Likewise.
16850 (loop_inverted_post_order_compute): Change argument to be a vec *.
16851 (df_analyze_loop): Adjust.
16852 (df_get_n_blocks): Likewise.
16853 (df_get_postorder): Likewise.
16854 * df.h (struct df_d): Change field to be a vec.
16855 * lcm.c (compute_laterin): Adjust.
16856 (compute_available): Likewise.
16857 * lra-lives.c (lra_create_live_ranges_1): Likewise.
16858 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
16859 * tree-ssa-pre.c (compute_antic): Likewise.
16861 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16863 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
16864 (depth_first_search::depth_first_search): Change structure init
16865 function to this constructor.
16866 (depth_first_search::add_bb): Rename function to this member.
16867 (depth_first_search::execute): Likewise.
16868 (flow_dfs_compute_reverse_finish): Adjust.
16870 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16872 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
16873 (longest_simple_path): Likewise.
16874 * shrink-wrap.c (spread_components): Likewise.
16875 (disqualify_problematic_components): Likewise.
16876 (emit_common_heads_for_components): Likewise.
16877 (emit_common_tails_for_components): Likewise.
16878 (insert_prologue_epilogue_for_components): Likewise.
16880 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16882 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
16885 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16887 * df-core.c (df_set_blocks): Start using auto_bitmap.
16888 (df_compact_blocks): Likewise.
16889 * df-problems.c (df_rd_confluence_n): Likewise.
16890 * df-scan.c (df_insn_rescan_all): Likewise.
16891 (df_process_deferred_rescans): Likewise.
16892 (df_update_entry_block_defs): Likewise.
16893 (df_update_exit_block_uses): Likewise.
16894 (df_entry_block_bitmap_verify): Likewise.
16895 (df_exit_block_bitmap_verify): Likewise.
16896 (df_scan_verify): Likewise.
16897 * lra-constraints.c (lra_constraints): Likewise.
16898 (undo_optional_reloads): Likewise.
16899 (lra_undo_inheritance): Likewise.
16900 * lra-remat.c (calculate_gen_cands): Likewise.
16901 (do_remat): Likewise.
16902 * lra-spills.c (assign_spill_hard_regs): Likewise.
16903 (spill_pseudos): Likewise.
16904 * tree-ssa-pre.c (bitmap_set_and): Likewise.
16905 (bitmap_set_subtract_values): Likewise.
16907 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16909 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
16910 management with auto_bitmap.
16911 (fix_inter_tick): Likewise.
16912 (fix_recovery_deps): Likewise.
16913 * ira.c (add_store_equivs): Likewise.
16914 (find_moveable_pseudos): Likewise.
16915 (split_live_ranges_for_shrink_wrap): Likewise.
16916 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
16917 (rtx_reuse_manager::seen_def_p): Likewise.
16918 (rtx_reuse_manager::set_seen_def): Likewise.
16919 * print-rtl.h (class rtx_reuse_manager): Likewise.
16921 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16923 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
16925 (migrate_btr_def): Likewise.
16926 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
16927 * df-core.c (loop_post_order_compute): Likewise.
16928 (loop_inverted_post_order_compute): Likewise.
16929 * hsa-common.h: Likewise.
16930 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
16931 * init-regs.c (initialize_uninitialized_regs): Likewise.
16932 * ipa-inline.c (resolve_noninline_speculation): Likewise.
16933 (inline_small_functions): Likewise.
16934 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
16935 * ira.c (combine_and_move_insns): Likewise.
16936 (build_insn_chain): Likewise.
16937 * loop-invariant.c (find_invariants): Likewise.
16938 * lower-subreg.c (propagate_pseudo_copies): Likewise.
16939 * predict.c (tree_predict_by_opcode): Likewise.
16940 (predict_paths_leading_to): Likewise.
16941 (predict_paths_leading_to_edge): Likewise.
16942 (estimate_loops_at_level): Likewise.
16943 (estimate_loops): Likewise.
16944 * shrink-wrap.c (try_shrink_wrapping): Likewise.
16945 (spread_components): Likewise.
16946 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
16947 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
16948 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
16949 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
16950 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
16951 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
16952 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
16953 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
16954 (mark_threaded_blocks): Likewise.
16955 (thread_through_all_blocks): Likewise.
16956 * tree-ssa.c (verify_ssa): Likewise.
16957 (execute_update_addresses_taken): Likewise.
16958 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
16960 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16962 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
16964 (post_order_compute): Likewise.
16965 (inverted_post_order_compute): Likewise.
16966 (pre_and_rev_post_order_compute_fn): Likewise.
16968 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16970 * genrecog.c (int_set::int_set): Explicitly construct our
16971 auto_vec base class.
16972 * vec.h (auto_vec::auto_vec): New constructor.
16974 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16976 * bitmap.h (class auto_bitmap): New constructor taking
16977 bitmap_obstack * argument.
16979 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16981 * bitmap.h (class auto_bitmap): Change type of m_bits to
16982 bitmap_head, and adjust ctor / dtor and member operators.
16984 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
16986 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
16987 when returned register mode doesn't match original mode.
16989 2017-05-12 Jeff Law <law@redhat.com>
16990 Jakub Jelinek <jakub@redhat.com>
16992 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
16993 we look for cc setter after the compare-elim changes.
16994 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
16995 within the vector to match what compare-elim now expects.
16996 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
16997 (xorsi3_flags, one_cmplsi2_flags): Likewise.
16999 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
17000 after the compare-elim changes.
17001 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
17002 the vector to match what compare-elim now expects.
17003 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
17004 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
17005 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
17006 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
17007 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
17009 * config/visium/visium.c (single_set_and_flags): Fix where
17010 we look for cc setter after the compare-elim changes.
17011 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
17012 with the vector to match what compare-elim now expects.
17013 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
17014 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
17015 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
17016 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
17017 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
17018 (neg<mode>2_insn_set_overflow): Likewise.
17020 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
17022 PR middle-end/79794
17023 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
17024 maybe_expand_insn call, set ops[0].target. If still set after call,
17025 set alt_rtl. Add extra arg to recursive calls.
17026 (extract_bit_field): Add alt_rtl argument. Pass to
17028 * expmed.h (extract_bit_field): Fix prototype.
17029 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
17030 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
17031 to extract_bit_field_calls.
17032 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
17033 Pass alt_rtl to extract_bit_field calls.
17034 * calls.c (store_unaligned_arguments_into_psuedos)
17035 load_register_parameters): Pass extra NULL to extract_bit_field calls.
17036 * optabs.c (maybe_legitimize_operand): Clear op->target when call
17038 * optabs.h (struct expand_operand): Add target bitfield.
17040 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
17042 * compare-elim.c (try_eliminate_compare): Canonicalize
17043 operation with embedded compare to
17044 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
17045 (set (reg) (operation)].
17047 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
17049 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
17052 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
17053 cost of adding a carry flag for ADC instruction.
17054 [case MINUS]: Ignore the cost of subtracting a carry flag
17055 for SBB instruction.
17057 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
17059 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
17061 * config/rs6000/bmiintrin.h: New file.
17062 * config/rs6000/bmi2intrin.h: New file.
17063 * config/rs6000/x86intrin.h: New file.
17065 2017-05-12 Jeff Law <law@redhat.com>
17067 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
17070 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
17072 PR middle-end/80707
17073 * tree-cfg.c: Remove cfg edges of unreachable case statements.
17075 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17077 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
17078 early expansion of vector divide builtins.
17079 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
17080 builtins identified as having unsigned arguments.
17082 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17084 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
17085 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
17086 expansion of vector logical operations (and, andc, or, xor,
17089 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17091 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
17092 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
17094 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17096 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
17097 early GIMPLE expansion of vector multiplies.
17099 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17101 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
17102 TARGET_HAVE_MOVT conditional.
17103 (movt splitter): Likewise.
17105 2017-05-12 Richard Biener <rguenther@suse.de>
17107 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
17108 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17109 Fold all stmts not inplace.
17111 2017-05-12 Richard Biener <rguenther@suse.de>
17113 PR tree-optimization/80713
17114 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
17115 inserted_exprs bit for not removed stmts.
17117 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
17119 PR middle-end/69921
17120 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
17121 parallelized" attribute for parallelized OpenACC kernels.
17122 * omp-offload.c (execute_oacc_device_lower): Use it.
17124 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
17125 Set "oacc kernels" attribute.
17126 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
17127 parameter. Adjust all users.
17128 (oacc_fn_attrib_kernels_p): Remove function.
17129 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
17130 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
17131 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
17132 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
17133 assert "oacc kernels" attribute is set.
17135 2017-05-11 Carl Love <cel@us.ibm.com>
17137 * config/rs6000/rs6000-c: Add support for built-in functions
17138 vector unsigned char vec_popcnt (vector signed char)
17139 vector unsigned char vec_popcnt (vector unsigned char)
17140 vector unsigned short vec_popcnt (vector signed short)
17141 vector unsigned short vec_popcnt (vector unsigned short)
17142 vector unsigned int vec_popcnt (vector signed int)
17143 vector unsigned int vec_popcnt (vector unsigned int)
17144 vector unsigned long long vec_popcnt (vector signed long long)
17145 vector unsigned long long vec_popcnt (vector unsigned long long)
17146 vector signed long long vec_slo (vector signed long long,
17147 vector signed char)
17148 vector signed long long vec_slo (vector signed long long,
17149 vector unsigned char)
17150 vector unsigned long long vec_slo (vector unsigned long long,
17151 vector signed char)
17152 vector unsigned long long vec_slo (vector unsigned long long,
17153 vector unsigned char)
17154 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
17155 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
17156 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
17157 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
17158 * doc/extend.texi: Update the built-in documentation file for the
17159 new built-in functions.
17161 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
17163 * attribs.h (sorted_attr_string): Move machine independent
17164 functions for target clone support from the i386 port to common
17165 code. Rename ix86_function_versions to common_function_versions.
17166 Rename make_name to make_unique_name.
17167 (common_function_versions): Likewise.
17168 (make_unique_name): Likewise.
17169 (make_dispatcher_decl): Likewise.
17170 (is_function_default_version): Likewise.
17171 * attribs.c (attr_strcmp): Likewise.
17172 (sorted_attr_string): Likewise.
17173 (common_function_versions): Likewise.
17174 (make_unique_name): Likewise.
17175 (make_dispatcher_decl): Likewise.
17176 (is_function_default_version): Likewise.
17177 * config/i386/i386.c (attr_strcmp): Likewise.
17178 (sorted_attr_string): Likewise.
17179 (ix86_function_versions): Likewise.
17180 (make_name): Likewise.
17181 (make_dispatcher_decl): Likewise.
17182 (is_function_default_version): Likewise.
17183 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
17185 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17188 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
17189 Account for direct move costs for vec_construct of integer
17192 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
17195 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
17196 (UNSPEC_STX_ATOMIC): Ditto.
17197 (loaddi_via_sse): New insn.
17198 (storedi_via_sse): Ditto.
17199 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
17200 Update corresponding peephole2 patterns.
17201 (atomic_storedi_fpu): Ditto.
17203 2017-05-11 Julia Koval <julia.koval@intel.com>
17205 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
17206 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
17208 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
17209 (__builtin_ia32_rsqrt14ss_mask): New builtins.
17210 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
17212 2017-05-11 Nathan Sidwell <nathan@acm.org>
17214 * graphite-poly.c: Include dumpfile.h.
17216 * dumpfle.h (dump_function): Declare here ...
17217 * tree-dump.h (dump_function): ... not here.
17218 * dumpfile.c: #include tree-cfg.h.
17219 (dump_function): Move here from ...
17220 * tree-dump.c (dump_function): ... here.
17221 * gimplify.c: #include splay-tree.h, not tree-dump.h.
17222 * graphite-poly.c: Don't include tree-dump.h.
17223 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
17224 * print-tree.c: Likewise.
17225 * stor-layout.c: Likewise.
17226 * tree-nested.c: Likewise.
17228 * dumpfile.c (dump_start): Use TDF_FLAGS.
17229 (dump_enable_all): Fix TDF_KIND check thinko.
17231 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
17233 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17234 array entries to represent two legal parameterizations of the
17235 overloaded __builtin_cmpb function, as represented by the
17236 P6_OV_BUILTIN_CMPB constant.
17237 (altivec_resolve_overloaded_builtin): Add special case handling
17238 for the __builtin_cmpb function, as represented by the
17239 P6_OV_BUILTIN_CMPB constant.
17240 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
17241 (BU_P6_64BIT_2): New macro.
17242 (BU_P6_OVERLOAD_2): New macro
17243 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
17244 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
17245 (CMPB): Add overload support to represent both 32-bit and 64-bit
17246 compare-bytes function.
17247 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
17248 support for TARGET_CMPB.
17249 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
17250 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17251 documentation of the __builtin_cmpb overloaded built-in function.
17253 2017-05-11 Richard Biener <rguenther@suse.de>
17255 PR tree-optimization/80705
17256 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
17257 bases are not vectorizable.
17259 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17261 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
17262 when counting register pressure.
17264 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17266 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
17267 (iv_ca_more_deps): Renamed to ...
17268 (iv_ca_compare_deps): ... this.
17269 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
17271 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17273 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
17275 (determine_group_iv_costs): ... here.
17276 (find_inv_vars_cb): Record inv var if it's not recorded before.
17278 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17280 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
17281 (get_shiftadd_cost): Ditto.
17283 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17285 * tree-ssa-address.c: Include header file.
17286 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
17288 (add_to_parts): Refactor.
17289 (addr_to_parts): New parameter. Update use of move_hint_to_base.
17290 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
17293 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17295 PR tree-optimization/53090
17296 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
17298 (extract_cond_operands): Detect condition with IV on both sides
17299 and return COMP_IV_EXPR_2.
17300 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
17301 (rewrite_use_compare): Simplify by removing call to function
17302 extract_cond_operands.
17304 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17306 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
17307 (extract_cond_operands): Detect condition comparing against non-
17308 invariant bound and return appropriate enum value.
17309 (find_interesting_uses_cond): Update use of extract_cond_operands.
17310 Handle its return value accordingly.
17311 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
17313 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17315 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
17316 nonlinear iv_use computation in loop invariant sensitive way.
17318 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17320 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
17321 (find_iv_candidates): Call relate_compare_use_with_all_cands.
17323 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17325 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
17326 (dump_cand): Support iv_cand.inv_exprs.
17327 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
17329 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
17332 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17334 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
17336 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
17337 as local function. Include necessary header files.
17338 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
17340 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17342 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
17344 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17346 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
17347 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
17348 RSHIFT_EXPR and BIT_NOT_EXPR.
17350 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17352 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
17353 (adjust_setup_cost): New parameter supporting round up adjustment.
17354 (struct address_cost_data): Delete.
17355 (force_expr_to_var_cost): Don't bound cost with spill_cost.
17356 (split_address_cost, ptr_difference_cost): Delete.
17357 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
17358 (struct ainc_cost_data): New struct.
17359 (get_address_cost_ainc): New function.
17360 (get_address_cost, get_computation_cost): Reimplement.
17361 (determine_group_iv_cost_address): Record inv_expr for all uses of
17363 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
17364 (iv_ca_has_deps): Reimplemented to ...
17365 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
17367 (iv_ca_extend): Call iv_ca_more_deps.
17369 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17371 * tree-ssa-address.c (struct mem_address): Move to header file.
17372 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
17373 * tree-ssa-address.h (struct mem_address): Move from C file.
17374 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
17376 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17378 * tree-affine.h (aff_combination_type): New interface.
17379 (aff_combination_zero_p): Remove static.
17380 (aff_combination_const_p): New interface.
17381 (aff_combination_singleton_var_p): New interfaces.
17383 2017-05-11 Richard Biener <rguenther@suse.de>
17385 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17386 Skip unreachable blocks and destinations.
17387 (eliminate): Move stmt removal and fixup ...
17388 (fini_eliminate): ... here. Skip inserted exprs.
17389 (pass_pre::execute): Move fini_pre after fini_eliminate.
17390 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
17391 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
17392 PRE to get rid of dead code that has invalid SSA form and
17393 split critical edges again.
17395 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17397 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17399 2017-05-11 Richard Biener <rguenther@suse.de>
17401 * passes.c (execute_function_todo): Verify loops if they are
17402 said to be up-to-date.
17403 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
17404 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
17406 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
17409 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
17410 handle calling assemble_external ourself.
17413 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
17414 modes with zero size. Enhance comment.
17416 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17418 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
17419 built-ins for vec_xl and vec_xst with short and char pointer
17422 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
17424 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
17425 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
17426 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
17427 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
17428 (_mm_maskz_min_round_ss): New intrinsics.
17429 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
17430 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
17431 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
17432 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
17433 (__builtin_ia32_minss_mask_round): New builtins.
17434 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17435 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
17436 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
17438 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
17439 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
17441 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
17444 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
17446 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
17447 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
17448 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
17449 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
17450 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
17451 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
17452 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
17453 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17454 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
17455 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
17456 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
17457 (__builtin_ia32_mulss_mask_round): New builtins.
17458 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17459 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
17460 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
17462 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
17463 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
17465 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
17468 2017-05-10 Julia Koval <julia.koval@intel.com>
17470 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
17471 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
17472 (_mm256_setr_m128i): New intrinsics.
17474 2017-05-10 Julia Koval <julia.koval@intel.com>
17476 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
17477 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
17478 (_mm_maskz_rcp14_ss): New intrinsics.
17479 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
17480 (__builtin_ia32_rcp14ss_mask): New builtins.
17481 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
17483 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
17485 PR tree-optimization/51513
17486 * tree-cfg.c (gimple_seq_unreachable_p): New function.
17487 (assert_unreachable_fallthru_edge_p): Use it.
17488 (group_case_labels_stmt): Likewise.
17489 * tree-cfg.h: Prototype it.
17490 * stmt.c: Include cfghooks.h and tree-cfg.h.
17491 (emit_case_dispatch_table) <gap_label>: New local variable.
17492 Use it to fill dispatch table gaps.
17493 Test for default_label before updating probabilities.
17494 (expand_case) <default_label>: Remove unneeded initialization.
17495 Test for unreachable default case statement and remove its edge.
17496 Set default_label accordingly.
17497 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
17499 2017-05-10 Carl Love <cel@us.ibm.com>
17501 * config/rs6000/rs6000-c: Add support for built-in functions
17502 vector signed char vec_neg (vector signed char)
17503 vector signed short int vec_neg (vector short int)
17504 vector signed int vec_neg (vector signed int)
17505 vector signed long long vec_neg (vector signed long long)
17506 vector float vec_neg (vector float)
17507 vector double vec_neg (vector double)
17508 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
17510 * config/rs6000/altivec.h: Add define for vec_neg
17511 * doc/extend.texi: Update the built-in documentation for the
17512 new built-in functions.
17514 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17516 PR tree-optimization/77644
17517 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
17519 2017-05-10 Nathan Sidwell <nathan@acm.org>
17521 * dumpfile.h (TDI_lang_all): New.
17522 (TDF_KIND): New. Renumber others
17523 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
17525 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
17527 (get_dump_file_name): Adjust suffix generation.
17528 (dump_enable_all): Use TDF_KIND.
17529 * doc/invoke.texi (-fdump-lang-all): Document.
17531 * dumpfile.h: Tabify.
17533 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
17536 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
17537 Move member access before delete.
17539 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
17541 * tree-inline.c (expand_call_inline): Split block at stmt
17544 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
17547 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
17548 are now unused after splitting mov{sf,sd}_hardfloat.
17549 (f32_lr2): Likewise.
17550 (f32_lm): Likewise.
17551 (f32_lm2): Likewise.
17552 (f32_li): Likewise.
17553 (f32_li2): Likewise.
17554 (f32_lv): Likewise.
17555 (f32_sr): Likewise.
17556 (f32_sr2): Likewise.
17557 (f32_sm): Likewise.
17558 (f32_sm2): Likewise.
17559 (f32_si): Likewise.
17560 (f32_si2): Likewise.
17561 (f32_sv): Likewise.
17562 (f32_dm): Likewise.
17563 (f32_vsx): Likewise.
17564 (f32_av): Likewise.
17565 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
17566 For movsf, order stores so the VSX stores occur before the GPR
17567 store which encourages the register allocator to use a traditional
17568 FPR instead of a GPR. For movsd, order the stores so that the GPR
17569 store comes before the VSX stores to allow the power6 to work.
17570 This is due to the power6 not having a 32-bit integer store
17571 instruction from a FPR.
17572 (movsf_hardfloat): Likewise.
17573 (movsd_hardfloat): Likewise.
17575 2017-05-09 Martin Sebor <msebor@redhat.com>
17577 PR translation/80280
17578 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
17581 PR translation/80280
17582 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
17583 data member added in r247778.
17584 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
17586 2017-05-09 Nathan Sidwell <nathan@acm.org>
17588 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
17590 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
17593 2017-05-09 Marek Polacek <polacek@redhat.com>
17595 * doc/invoke.texi: Fix typo.
17597 2017-05-09 Richard Biener <rguenther@suse.de>
17599 * tree-vrp.c (vrp_val_is_max): Adjust comment.
17600 (vrp_val_is_min): Likewise.
17601 (set_value_range_to_value): Likewise.
17602 (set_value_range_to_nonnegative): Likewise.
17603 (gimple_assign_nonzero_p): Likewise.
17604 (gimple_stmt_nonzero_p): Likewise.
17605 (vrp_int_const_binop): Likewise. Remove unreachable case.
17606 (adjust_range_with_scev): Adjust comments.
17607 (compare_range_with_value): Likewise.
17608 (extract_range_from_phi_node): Likewise.
17609 (test_for_singularity): Likewise.
17611 2017-05-09 Richard Biener <rguenther@suse.de>
17613 * tree-vrp.c (get_single_symbol): Add assert that we don't
17614 get overflowed constants as invariant part.
17615 (compare_values_warnv): Add comment before the TREE_NO_WARNING
17616 checks. Use wi::cmp instead of recursing for integer constants.
17617 (compare_values): Just ignore whether we assumed undefined
17618 overflow instead of failing the compare.
17619 (extract_range_for_var_from_comparison_expr): Add comment before the
17620 TREE_NO_WARNING sets.
17621 (test_for_singularity): Likewise.
17622 (extract_range_from_comparison): Do not disable optimization
17623 when we assumed undefined overflow.
17624 (extract_range_basic): Remove init of unused var.
17626 2017-05-09 Richard Biener <rguenther@suse.de>
17628 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
17629 (extract_range_from_multiplicative_op_1): Adjust.
17630 (extract_range_from_binary_expr_1): Use int_const_binop.
17632 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
17635 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
17636 rs6000_store_data_bypass_p in seven define_bypass directives and
17637 in several comments.
17638 * config/rs6000/rs6000-protos.h: Add prototype for
17639 rs6000_store_data_bypass_p function.
17640 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
17641 function implements slightly different (rs6000-specific) semantics
17642 than store_data_bypass_p, returning false rather than aborting
17643 with assertion error when arguments do not satisfy the
17644 requirements of store data bypass.
17645 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
17646 rs6000_store_data_bypass_p.
17648 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
17650 * config/xtensa/xtensa-protos.h
17651 (xtensa_initial_elimination_offset): New declaration.
17652 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
17653 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
17654 macro definition, add case for FRAME_POINTER_REGNUM when
17655 FRAME_GROWS_DOWNWARD.
17656 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
17657 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
17658 xtensa_initial_elimination_offset.
17660 2017-05-08 Nathan Sidwell <nathan@acm.org>
17662 * doc/invoke.texi: Alphabetize -fdump options.
17664 2017-05-08 Martin Sebor <msebor@redhat.com>
17666 PR translation/80280
17667 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
17669 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
17671 * target.def (compute_frame_layout): New optional target hook.
17672 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
17673 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
17674 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
17676 * reload1.c (verify_initial_elim_offsets): Likewise.
17677 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
17678 (use_simple_return_p): Call arm_compute_frame_layout if needed.
17679 (arm_get_frame_offsets): Split up into this ...
17680 (arm_compute_frame_layout): ... and this function.
17682 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
17684 * config/aarch64/constraints.md (Usa): New constraint.
17685 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
17687 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
17689 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
17690 with_multilib_list after it has been checked.
17692 2017-05-08 Richard Biener <rguenther@suse.de>
17694 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
17695 (bitmap_set_subtract_values): Likewise.
17697 2017-05-08 Richard Biener <rguenther@suse.de>
17699 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
17700 (gimple_assign_nonzero): ... this and remove strict_overflow_p
17702 (gimple_stmt_nonzero_warnv_p): Rename to ...
17703 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
17705 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
17706 (extract_range_basic): Adjust, do not disable propagation on
17707 strict overflow sensitive simplification.
17708 (vrp_visit_cond_stmt): Likewise.
17710 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
17712 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
17713 body size unconditionally.
17715 2017-05-07 Jeff Law <law@redhat.com>
17718 2017-05-06 Jeff Law <law@redhat.com>
17719 PR tree-optimization/78496
17720 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
17723 PR tree-optimization/78496
17724 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
17725 (simplify_stmt_using_ranges): Call it.
17726 (vrp_dom_walker::before_dom_children): Extract equivalences
17727 from an ASSERT_EXPR with an equality comparison against a
17730 2017-05-06 Jeff Law <law@redhat.com>
17732 PR tree-optimization/78496
17733 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
17736 PR tree-optimization/78496
17737 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
17738 (simplify_stmt_using_ranges): Call it.
17739 (vrp_dom_walker::before_dom_children): Extract equivalences
17740 from an ASSERT_EXPR with an equality comparison against a
17743 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
17745 * lra-constraints.c (lra_copy_reg_equiv): New function.
17746 (split_reg): Use it to copy equivalence information from the
17747 original register to the spill register.
17749 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
17751 PR rtl-optimization/75964
17752 * simplify-rtx.c (simplify_const_relational_operation): Remove
17753 invalid handling of comparisons of integer ABS.
17755 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
17757 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
17758 initialize to zero.
17759 (init_regs): Remove declaration.
17760 (function_arg_advance_32): Initialize error_p as boolean variable.
17762 2017-05-05 Nathan Sidwell <nathan@acm.org>
17764 * store-motion.c (remove_reachable_equiv_notes): Reformat long
17765 lines. Use for (;;).
17767 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17769 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
17770 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
17771 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
17772 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
17773 VF=2 that require versioning.
17775 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17777 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
17780 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17782 * diagnostic.h (diagnostic_override_option_index): Convert from
17783 macro to inline function.
17785 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17787 * diagnostic.c (last_module_changed_p): New function.
17788 (set_last_module): New function.
17789 (diagnostic_report_current_module): Convert macro usage to
17790 the above functions.
17791 * diagnostic.h (diagnostic_context::last_module): Strengthen
17792 from const line_map * to const line_map_ordinary *.
17793 (diagnostic_last_module_changed): Delete macro.
17794 (diagnostic_set_last_module): Delete macro.
17796 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17798 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
17799 with diagnostic_report_diagnostic.
17800 (diagnostic_n_impl_richloc): Likewise.
17801 * diagnostic.h (report_diagnostic): Delete macro.
17802 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
17803 with diagnostic_report_diagnostic.
17804 * substring-locations.c (format_warning_va): Likewise.
17806 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17808 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
17809 save/restor of format_spec. Move option-printing code to...
17810 (print_option_information): ...this new function, and
17811 reimplement by simply printing to the pretty_printer,
17812 rather than appending to the format string.
17814 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17816 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
17817 handling logic into...
17818 (update_effective_level_from_pragmas): ...this new function.
17820 2017-05-04 Andrew Waterman <andrew@sifive.com>
17822 * config/riscv/riscv.opt (mstrict-align): New option.
17823 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
17824 (SLOW_UNALIGNED_ACCESS): Define.
17825 (riscv_slow_unaligned_access): Declare.
17826 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
17828 (riscv_slow_unaligned_access): New variable.
17829 (rocket_tune_info): Set slow_unaligned_access to true.
17830 (optimize_size_tune_info): Set slow_unaligned_access to false.
17831 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
17832 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
17833 (riscv_option_override): Set riscv_slow_unaligned_access.
17834 * doc/invoke.texi: Add -mstrict-align to RISC-V.
17836 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
17838 * config/riscv/riscv.md: Unify indentation.
17840 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
17845 * config/rs6000/rs6000.md (u code attribute): Add FIX and
17847 (extendsi<mode>2): Add support for doing sign extension via
17848 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
17849 don't have ISA 3.0 instructions.
17850 (extendsi<mode>2 splitter): Likewise.
17851 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
17852 generate the normal insns since SImode can now go in vector
17853 registers. Disallow the special UNSPECs needed for previous
17854 machines to hide SImode being used. Add new insns
17855 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
17856 (fix_trunc<mode>si2_stfiwx): Likewise.
17857 (fix_trunc<mode>si2_internal): Likewise.
17858 (fixuns_trunc<mode>si2): Likewise.
17859 (fixuns_trunc<mode>si2_stfiwx): Likewise.
17860 (fctiw<u>z_<mode>_smallint): Likewise.
17861 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
17862 of floating point to 32-bit integer from doing a direct move to
17863 the GPR registers to do a store.
17864 (fctiwz_<mode>): Break long line.
17866 2017-05-05 Bin Cheng <bin.cheng@arm.com>
17868 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
17869 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
17870 (addr_list, addr_offset_valid_p): New.
17871 (split_address_groups): Check offset validity with above function.
17872 (gt-tree-ssa-loop-ivopts.h): Include header file.
17874 2017-05-05 Nathan Sidwell <nathan@acm.org>
17876 * config.gcc (arm*-*-*): Add missing 'fi'.
17878 2017-05-05 Steve Ellcey <sellcey@cavium.com>
17880 * doc/invoke.texi (-fopt-info): Explicitly say order of options
17881 included in -fopt-info does not matter.
17882 * doc/optinfo.texi (-fopt-info): Fix description of default
17883 behavour. Explicitly say order of options included in -fopt-info
17886 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
17888 * config.gcc: Allow combinations of aprofile and rmprofile values for
17889 --with-multilib-list.
17890 * config/arm/t-multilib: New file.
17891 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
17892 variables. Remove setting of ISA and floating-point ABI in
17893 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
17894 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
17895 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
17896 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
17898 * config/arm/t-rmprofile: Likewise except for the matches changes.
17899 * doc/install.texi (--with-multilib-list): Document the combination of
17900 aprofile and rmprofile values and warn about pitfalls in doing that.
17902 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
17904 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
17905 (movdi_aarch64): Likewise.
17907 2017-05-05 Jakub Jelinek <jakub@redhat.com>
17909 PR tree-optimization/80632
17910 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
17912 (build_arrays): Initialize it for virtual phis.
17913 (fix_phi_nodes): Use it for virtual phis.
17915 PR tree-optimization/80558
17916 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
17917 [x, y] op z into [x op, y op z] for op & or | if conditions
17920 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
17921 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17924 * config/arm/arm.md (use_literal_pool): Remove.
17925 (64-bit immediate split): No longer takes cost into consideration
17926 if arm_disable_literal_pool is enabled.
17927 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
17928 used when arm_disable_literal_pool is enabled.
17929 (arm_max_const_double_inline_cost): Remove use of
17930 arm_disable_literal_pool.
17931 (push_minipool_fix): Add assert.
17932 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
17933 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
17934 (no_literal_pool_sf_immediate): New.
17936 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17938 PR tree-optimization/80613
17939 * tree-ssa-dce.c (propagate_necessity): Remove cases for
17940 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
17942 2017-05-05 Richard Biener <rguenther@suse.de>
17944 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
17946 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
17948 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
17949 of this flag from insn conditions due to removal from r247495.
17951 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
17953 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
17955 (arm_early_store_addr_dep_ptr): Likewise.
17956 * config/arm/aarch-common-protos.h
17957 (arm_early_load_addr_dep_ptr): Add prototype.
17958 (arm_early_store_addr_dep_ptr): Likewise.
17959 * config/arm/cortex-a53.md: Add new bypasses.
17961 2017-05-05 Jakub Jelinek <jakub@redhat.com>
17963 * tree.c (next_type_uid): Change type to unsigned.
17964 (type_hash_canon): Decrement back next_type_uid if
17965 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
17966 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
17969 2017-05-04 Martin Sebor <msebor@redhat.com>
17971 * builtins.c: Fix a trivial typo in a comment.
17973 PR middle-end/79234
17974 * builtins.c (check_sizes): Adjust to handle reading past the end.
17975 Avoid printing excessive upper bound of ranges. Use %E to print
17976 tree nodes instead of converting them to %wu.
17977 (expand_builtin_memchr): New function.
17978 (compute_dest_size): Rename...
17979 (compute_objsize): ...to this.
17980 (expand_builtin_memcpy): Adjust.
17981 (expand_builtin_mempcpy): Adjust.
17982 (expand_builtin_strcat): Adjust.
17983 (expand_builtin_strcpy): Adjust.
17984 (check_strncat_sizes): Adjust.
17985 (expand_builtin_strncat): Adjust.
17986 (expand_builtin_strncpy): Adjust and simplify.
17987 (expand_builtin_memset): Adjust.
17988 (expand_builtin_bzero): Adjust.
17989 (expand_builtin_memcmp): Adjust.
17990 (expand_builtin): Handle memcmp.
17991 (maybe_emit_chk_warning): Check strncat just once.
17993 2017-05-04 Martin Sebor <msebor@redhat.com>
17995 PR preprocessor/79214
17996 PR middle-end/79222
17997 PR middle-end/79223
17998 * builtins.c (check_sizes): Add inlining context and issue
17999 warnings even when -Wno-system-headers is set.
18000 (check_strncat_sizes): Same.
18001 (expand_builtin_strncat): Same.
18002 (expand_builtin_memmove): New function.
18003 (expand_builtin_stpncpy): Same.
18004 (expand_builtin): Handle memmove and stpncpy.
18006 2017-05-04 Bin Cheng <bin.cheng@arm.com>
18008 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
18009 which is not used any more.
18011 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
18013 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
18015 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
18017 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
18018 (cortexa53_tunings): Likewise.
18019 (cortexa57_tunings): Likewise.
18020 (cortexa72_tunings): Likewise.
18021 (cortexa73_tunings): Likewise.
18023 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
18025 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
18026 Set loop alignment to 8.
18028 2017-05-04 Martin Sebor <msebor@redhat.com>
18030 PR translation/80280
18031 * builtins.c (expand_builtin_object_size): Add missing quoting to
18032 %D and like directives.
18033 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
18034 (hsa_type_for_tree_type): Same.
18035 (verify_function_arguments): Same.
18036 * symtab.c (symbol_table::change_decl_assembler_name): Same.
18037 * varasm.c (get_section): Same.
18040 2017-05-04 Martin Sebor <msebor@redhat.com>
18042 PR translation/80280
18043 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
18045 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
18047 * config/aarch64/aarch64.c (generic_addrcost_table):
18048 Change HI/TI mode setting.
18050 2017-05-04 Martin Jambor <mjambor@suse.cz>
18052 PR tree-optimization/80622
18053 * tree-sra.c (comes_initialized_p): New function.
18054 (build_accesses_from_assign): Only set write lazily when
18055 comes_initialized_p is false.
18056 (analyze_access_subtree): Use comes_initialized_p.
18057 (propagate_subaccesses_across_link): Assert !comes_initialized_p
18058 instead of testing for PARM_DECL.
18060 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18062 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
18063 constraint on operand 0 to allow more general addressing modes.
18064 Adjust output template.
18065 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
18067 * config/aarch64/aarch64-protos.h
18068 (aarch64_address_valid_for_prefetch_p): Declare prototype.
18069 * config/aarch64/constraints.md (Dp): New address constraint.
18070 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
18073 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
18075 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
18076 update use of estimate_ipcp_clone_size_and_time.
18077 (estimate_local_effects): Update use of
18078 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
18079 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
18080 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
18081 Return nonspecialized time.
18083 2017-05-04 Richard Biener <rguenther@suse.de>
18085 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
18086 for the last VUSE which def dominates the PHI. Directly call
18088 (get_continuation_for_phi_1): Remove.
18090 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
18092 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
18093 to explain the use of truncating division. Cap the number of
18094 iterations to the maximum given by nb_iterations_upper_bound,
18097 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18099 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
18100 * configure: Regenerate.
18101 * config.in: Regenerate.
18102 * config/i386/driver-mingw32.c: new file.
18103 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
18104 * config.host: Link driver-mingw32.o on MinGW host.
18105 * doc/install.texi: Document new --enable-mingw-wildcard configure
18108 2017-05-04 Marek Polacek <polacek@redhat.com>
18110 PR tree-optimization/80612
18111 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
18113 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18114 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
18116 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
18117 (movt splitter): Likewise.
18118 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
18119 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
18120 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
18121 block for Thumb-1 with MOVT.
18122 (thumb2_legitimate_address_p): Move code block ...
18123 (can_avoid_literal_pool_for_label_p): ... into this new function.
18124 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
18126 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
18127 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
18128 "M-profile targets with the MOVT instruction".
18130 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18132 * config/arm/arm-builtins.c (arm_init_builtins): Rename
18133 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
18134 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
18136 2017-05-04 Martin Liska <mliska@suse.cz>
18138 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
18139 variable cond_code.
18141 2017-05-04 Richard Biener <rguenther@suse.de>
18143 * tree.c (array_at_struct_end_p): Handle arrays at struct
18144 end with flexarrays more conservatively. Refactor and treat
18145 arrays of arrays or aggregates more strict. Fix
18146 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
18147 * tree.c (array_at_struct_end_p): Adjust prototype.
18148 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
18149 * gimple-fold.c (get_range_strlen): Likewise.
18150 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
18152 2017-05-04 Richard Biener <rguenther@suse.de>
18154 PR tree-optimization/31130
18155 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
18157 (supports_overflow_infinity): Likewise.
18158 (is_negative_overflow_infinity): Likewise.
18159 (is_positive_overflow_infinity): Likewise.
18160 (is_overflow_infinity): Likewise.
18161 (stmt_overflow_infinity): Likewise.
18162 (overflow_infinity_range_p): Likewise.
18163 (usable_range_p): Remove as always returning true.
18164 (make_overflow_infinity): Remove.
18165 (negative_overflow_infinity): Likewise.
18166 (positive_overflow_infinity): Likewise.
18167 (avoid_overflow_infinity): Likewise.
18168 (set_value_range): Adjust accordingly.
18169 (set_value_range_to_nonnegative): Likewise, remove now unused
18170 overflow_infinity arg.
18171 (vrp_operand_equal_p): Adjust.
18172 (update_value_range): Likewise.
18173 (range_int_cst_singleton_p): Likewise.
18174 (operand_less_p): Likewise.
18175 (compare_values_warnv): Likewise.
18176 (extract_range_for_var_from_comparison_expr): Likewise.
18177 (vrp_int_const_binop): Likewise.
18178 (zero_nonzero_bits_from_vr): Likewise.
18179 (extract_range_from_multiplicative_op_1): Likewise.
18180 (extract_range_from_binary_expr_1): Likewise.
18181 (extract_range_from_unary_expr): Likewise.
18182 (extract_range_from_comparison): Likewise.
18183 (extract_range_basic): Likewise.
18184 (adjust_range_with_scev): Likewise.
18185 (compare_ranges): Likewise.
18186 (compare_range_with_value): Likewise.
18187 (dump_value_range): Likewise.
18188 (test_for_singularity): Likewise, remove strict_overflow_p parameter
18190 (simplify_cond_using_ranges): Adjust.
18192 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
18194 * brig-builtins.def: Added a builtin for class_f64.
18195 * builtin-types.def: Added a builtin type needed by class_f64.
18197 2017-05-03 Jason Merrill <jason@redhat.com>
18199 * timevar.def: Add TV_CONSTEXPR.
18201 2017-05-03 David Malcolm <dmalcolm@redhat.com>
18203 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
18205 2017-05-03 Martin Jambor <mjambor@suse.cz>
18207 * ipa-prop.c (ipa_update_after_lto_read): Removed.
18208 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
18209 * ipa-cp.c (ipcp_propagate_stage): Do not call
18210 ipa_update_after_lto_read.
18211 * ipa-inline.c (ipa_inline): Likewise.
18213 2017-05-03 Martin Jambor <mjambor@suse.cz>
18215 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
18216 tag. Added a default constructor and a destructor.
18217 (ipa_edge_args_sum_t): New class;
18218 (ipa_edge_args_sum): Declare.
18219 (ipa_edge_args_vector): Remove declaration.
18220 (IPA_EDGE_REF): Use ipa_edge_args_sum.
18221 (ipa_free_edge_args_substructures): Remove declaration.
18222 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
18223 (ipa_edge_args_info_available_for_edge_p): Likewise.
18224 * ipa-prop.c (ipa_edge_args_vector): Removed.
18225 (edge_removal_hook_holder): Likewise.
18226 (edge_duplication_hook_holder): Likewise.
18227 (ipa_edge_args_sum): New variable.
18228 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
18229 ipa_edge_args_vector.
18230 (ipa_free_edge_args_substructures): Likewise.
18231 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
18232 ipa_edge_args_vector.
18233 (ipa_edge_removal_hook): Turned into method
18234 ipa_edge_args_sum_t::remove.
18235 (ipa_edge_duplication_hook): Turned into method
18236 ipa_edge_args_sum_t::duplicate.
18237 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
18238 registering edge hooks.
18239 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
18240 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
18241 ipa_edge_args_sum instead of ipa_edge_args_vector.
18242 * ipa-profile.c (ipa_profile): Likewise.
18244 2017-05-03 Martin Jambor <mjambor@suse.cz>
18246 * symbol-summary.h (function_summary): New method exists.
18247 (function_summary::symtab_removal): Deallocate through release.
18248 (call_summary): New class.
18249 (gt_ggc_mx): New overload.
18250 (gt_pch_nx): Likewise.
18251 (gt_pch_nx): Likewise.
18253 2017-05-03 Jeff Law <law@redhat.com>
18255 PR tree-optimization/78496
18256 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
18257 from simplify_cond_using_ranges. Split off code to walk
18258 backwards through casts into ...
18259 (simplify_cond_using_ranges_2): New function.
18260 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
18261 (execute_vrp): After identifying jump threads, call
18262 simplify_cond_using_ranges_2.
18264 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
18267 * ipa-inline.h (inline_summary): Add ctor.
18268 (create_ggc): Do not use ggc_cleared_alloc.
18270 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
18271 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18273 * gcc.c (handle_braces): Support escaping in switch matching
18275 * doc/invoke.texi (Spec Files): Document it.
18276 Remove superfluous @code markup in items.
18278 2017-05-03 David Malcolm <dmalcolm@redhat.com>
18280 * diagnostic-show-locus.c (struct column_range): New struct.
18281 (get_affected_columns): New function.
18282 (get_printed_columns): New function.
18283 (struct correction): New struct.
18284 (correction::ensure_capacity): New function.
18285 (correction::ensure_terminated): New function.
18286 (struct line_corrections): New struct.
18287 (line_corrections::~line_corrections): New dtor.
18288 (line_corrections::add_hint): New function.
18289 (layout::print_trailing_fixits): Reimplement in terms of the new
18291 (selftest::test_overlapped_fixit_printing): New function.
18292 (selftest::diagnostic_show_locus_c_tests): Call it.
18294 2017-05-03 Nathan Sidwell <nathan@acm.org>
18296 Canonicalize canonical type hashing
18297 * tree.h (type_hash_canon_hash): Declare.
18298 * tree.c (type_hash_list, attribute_hash_list): Move into
18299 type_hash_canon_hash.
18300 (build_type_attribute_qual_variant): Break out hash code calc into
18301 type_hash_canon_hash.
18302 (type_hash_canon_hash): New. Generic type hash computation.
18303 (build_range_type_1, build_array_type_1, build_function_type,
18304 build_method_type_directly, build_offset_type, build_complex_type,
18305 make_vector_type): Call it.
18307 2017-05-03 Richard Biener <rguenther@suse.de>
18309 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
18310 When all DRs have unknown misaligned do not always peel
18311 when there is a store but apply the same costing model as if
18312 there were only loads.
18314 2017-05-03 Richard Biener <rguenther@suse.de>
18317 PR tree-optimization/80492
18318 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
18319 compare_base_decls returning dont-know properly.
18321 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
18323 * config/arm/iterators.md (CCSI): New mode iterator.
18324 (arch): New mode attribute.
18325 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
18326 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
18327 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
18328 code iterator for success result mode.
18329 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
18330 the corresponding new insn generators.
18332 2017-05-03 Bin Cheng <bin.cheng@arm.com>
18335 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18336 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18338 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
18340 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
18341 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
18342 (DDR_A): Wrap DDR argument in brackets.
18343 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
18344 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
18345 (DDR_REVERSED_P): Likewise.
18347 2017-05-03 Jakub Jelinek <jakub@redhat.com>
18349 PR tree-optimization/79472
18350 * tree-switch-conversion.c (struct switch_conv_info): Add
18351 contiguous_range and default_case_nonstandard fields.
18352 (collect_switch_conv_info): Compute contiguous_range and
18353 default_case_nonstandard fields, don't clear final_bb if
18354 contiguous_range and only the default case doesn't have the required
18356 (check_all_empty_except_final): Set default_case_nonstandard instead
18357 of failing if contiguous_range and the default case doesn't have empty
18359 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
18360 and only the default case doesn't have the required constants. Skip
18362 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
18363 if default_case_nonstandard.
18364 (build_constructors): Build constant 1 just once. Assert that default
18365 values aren't inserted in between cases if contiguous_range. Skip
18367 (build_arrays): Skip virtual phis.
18368 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
18369 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
18370 Handle virtual phis.
18371 (gen_inbound_check): Handle default_case_nonstandard case.
18372 (process_switch): Adjust check_final_bb caller. Call
18373 gather_default_values with the first non-default case instead of
18374 default case if default_case_nonstandard.
18376 2017-05-02 Nathan Sidwell <nathan@acm.org>
18378 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
18379 check. Fix formatting.
18381 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
18383 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
18384 errors when comparing specialized and unspecialized times.
18386 2017-05-02 David Malcolm <dmalcolm@redhat.com>
18388 * diagnostic-show-locus.c
18389 (layout::should_print_annotation_line_p): Make private.
18390 (layout::print_annotation_line): Make private.
18391 (layout::annotation_line_showed_range_p): Make private.
18392 (layout::show_ruler): Make private.
18393 (layout::print_source_line): Make private. Pass in line and
18394 line_width, rather than calling location_get_source_line. Drop
18396 (layout::print_leading_fixits): New method.
18397 (layout::print_any_fixits): Rename to...
18398 (layout::print_trailing_fixits): ...this, and make private.
18399 Don't print newline fixits.
18400 (diagnostic_show_locus): Move logic for printing one row into...
18401 (layout::print_line): ...this new function. Move the
18402 location_get_source_line call and error-handling from
18403 print_source_line to here. Call print_leading_fixits, and rename
18404 print_any_fixits to print_trailing_fixits.
18405 (selftest::test_fixit_insert_containing_newline): Update now that
18406 newlines are partially supported.
18407 (selftest::test_fixit_insert_containing_newline_2): New test.
18408 (selftest::test_fixit_replace_containing_newline): Update comments.
18409 (selftest::diagnostic_show_locus_c_tests): Call the new test.
18410 * edit-context.c (class added_line): New class.
18411 (class edited_line): Describe newline handling in comment.
18412 (edited_line::actually_edited_p): New method.
18413 (edited_line::print_content): Delete redundant decl.
18414 (edited_line::m_predecessors): New field.
18415 (edited_file::print_content): Call edited_line::print_content.
18416 (edited_file::print_diff): Update to support newlines.
18417 (edited_file::print_diff_hunk): Likewise.
18418 (edited_file::print_run_of_changed_lines): New function.
18419 (edited_file::print_diff_line): Convert to...
18420 (print_diff_line): ...this.
18421 (edited_file::get_effective_line_count): New function.
18422 (edited_line::edited_line): Initialize new field m_predecessors.
18423 (edited_line::~edited_line): Clean up m_predecessors.
18424 (edited_line::apply_fixit): Handle newlines.
18425 (edited_line::get_effective_line_count): New function.
18426 (edited_line::print_content): New function.
18427 (edited_line::print_diff_lines): New function.
18428 (selftest::test_applying_fixits_insert_containing_newline): New
18430 (selftest::test_applying_fixits_replace_containing_newline): New
18432 (selftest::insert_line): New function.
18433 (selftest::test_applying_fixits_multiple_lines): Add example of
18435 (selftest::edit_context_c_tests): Call the new tests.
18437 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18439 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
18440 parameter cand. Update dump information.
18441 (get_computation_cost): Update uses.
18443 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18445 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
18446 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
18447 (get_computation_at, rewrite_use_address): Update use of
18448 get_computation_aff.
18450 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18452 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
18453 (get_computation): Delete.
18454 (get_computation_cost): Implement like get_computation_cost_at.
18455 Use get_computation_at.
18456 (get_computation_cost_at): Delete.
18457 (rewrite_use_nonlinear_expr): Use get_computation_at.
18458 (rewrite_use_compare, remove_unused_ivs): Ditto.
18460 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18462 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
18464 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18466 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
18467 (ivopts_global_cost_for_size): Rename parameter and update uses.
18468 (iv_ca_recount_cost): Update uses.
18469 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
18470 candidates seperately in n_invs and n_cands.
18471 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
18473 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18475 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
18476 (find_inv_vars_cb): New.
18477 (find_depends): Renamed to ...
18478 (find_inv_vars): ... this.
18479 (add_candidate_1, force_var_cost): Call find_inv_vars.
18480 (split_address_cost, determine_group_iv_cost_cond): Ditto.
18482 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18484 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
18485 inv_vars. Add inv_exprs.
18486 (struct iv_cand): Rename depends_on to inv_vars.
18487 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
18488 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
18489 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
18490 (dump_cand): Dump inv_vars.
18491 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
18492 (record_invariant, find_depends, add_candidate_1): Ditto.
18493 (set_group_iv_cost, force_var_cost): Ditto.
18494 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
18495 (get_computation_cost_at, get_computation_cost): Ditto.
18496 (determine_group_iv_cost_generic): Ditto.
18497 (determine_group_iv_cost_address): Ditto.
18498 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
18499 (determine_group_iv_costs): Ditto.
18500 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
18501 (iv_ca_set_remove_invariants): Renamed to ...
18502 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
18503 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
18504 (iv_ca_set_add_invariants): Renamed to ...
18505 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
18506 (iv_ca_set_cp): Use iv_ca_set_add_invs.
18507 (iv_ca_has_deps): Support inv_vars and inv_exprs.
18508 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
18509 (create_new_ivs): Remove useless dump.
18511 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18513 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
18515 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
18516 (iv_ca_set_no_cp, create_new_iv): Ditto.
18518 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18520 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18522 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
18524 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
18525 function tree_check2.
18527 2017-05-02 Martin Liska <mliska@suse.cz>
18529 * doc/gcov.texi: Add missing preposition.
18530 * gcov.c (function_info::function_info): Properly fill up
18531 all member variables.
18533 2017-05-02 Tamar Christina <tamar.christina@arm.com>
18535 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
18537 2017-05-02 Tamar Christina <tamar.christina@arm.com>
18539 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
18541 2017-05-02 Martin Liska <mliska@suse.cz>
18544 * lto-streamer-in.c (lto_read_tree_1): Remove
18545 LTO_STREAMER_DEBUG.
18546 * lto-streamer.c (struct tree_hash_entry): Likewise.
18547 (struct tree_entry_hasher): Likewise.
18548 (tree_entry_hasher::hash): Likewise.
18549 (tree_entry_hasher::equal): Likewise.
18550 (lto_streamer_init): Likewise.
18551 (lto_orig_address_map): Likewise.
18552 (lto_orig_address_get): Likewise.
18553 (lto_orig_address_remove): Likewise.
18554 * lto-streamer.h: Likewise.
18555 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
18556 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
18558 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
18560 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
18561 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
18562 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
18563 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
18564 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
18565 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
18566 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
18567 (mm_maskz_sub_ss): New intrinsics.
18568 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18569 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
18570 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
18571 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
18572 (__builtin_ia32_subss_mask_round): New builtins.
18573 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18574 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18575 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
18577 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
18578 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
18580 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
18583 2017-05-02 Martin Jambor <mjambor@suse.cz>
18585 PR tree-optimization/78687
18586 * tree-sra.c (access): New field parent.
18587 (process_subtree_disqualification): New function.
18588 (disqualify_candidate): Call it.
18589 (build_accesses_from_assign): Reset write flag if creating an
18591 (build_access_subtree): Fill in parent field and also prpagate
18592 down grp_write flag.
18593 (create_artificial_child_access): New parameter set_grp_write, set
18594 grp_write to its value.
18595 (propagate_subaccesses_across_link): Also propagate grp_write flag
18597 (propagate_all_subaccesses): Push the closest parent back to work
18598 queue if add_access_to_work_queue returned true.
18600 2017-05-02 Richard Biener <rguenther@suse.de>
18602 * common.opt (fstrict-overflow): Alias negative to fwrapv.
18603 * doc/invoke.texi (fstrict-overflow): Remove all traces of
18604 -fstrict-overflow documentation.
18605 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
18606 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
18607 flag_strict_overflow.
18608 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
18609 * lto-opts.c (lto_write_options): Do not stream it.
18610 * lto-wrapper.c (merge_and_complain): Do not handle it.
18611 * opts.c (default_options_table): Do not set -fstrict-overflow.
18612 (finish_options): Likewise do not clear it when sanitizing.
18613 * simplify-rtx.c (simplify_const_relational_operation): Do not
18614 test flag_strict_overflow.
18616 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
18618 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
18619 using enabled attribute.
18620 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
18621 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
18622 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
18623 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
18624 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
18625 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
18626 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
18627 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
18628 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
18629 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
18631 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
18633 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
18635 2017-05-02 Richard Biener <rguenther@suse.de>
18637 PR tree-optimization/80591
18639 2017-04-10 Richard Biener <rguenther@suse.de>
18641 * tree-ssa-structalias.c (find_func_aliases): Properly handle
18644 2017-05-02 Richard Biener <rguenther@suse.de>
18646 PR tree-optimization/80549
18647 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
18648 (cleanup_tree_cfg_noloop): Create forwarders to known loop
18649 headers if they do not have a preheader.
18651 2017-05-02 Martin Liska <mliska@suse.cz>
18654 * common.opt: Fix typo.
18655 * doc/invoke.texi: Likewise.
18657 2017-05-01 Jan Beulich <jbeulich@suse.com>
18659 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
18660 swapping, add (x,x,m,x,n) alternative.
18662 2017-05-01 Nathan Sidwell <nathan@acm.org>
18664 * calls.c (combine_pending_stack_adjustment_and_call): Remove
18665 unnecessary unadjusted_alignment check.
18667 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
18670 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
18672 * gimplify.c (gimplify_cilk_detach): New function.
18673 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
18674 * tree-core.h: Document EXPR_CILK_SPAWN.
18675 * tree.h (EXPR_CILK_SPAWN): Define.
18677 2017-05-01 David Malcolm <dmalcolm@redhat.com>
18679 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
18680 to use new fixit_hint representation, using the "replace" logic.
18681 (get_line_span_for_fixit_hint): Likewise.
18682 (layout::print_any_fixits): Likewise.
18683 (selftest::test_one_liner_many_fixits): Rename to...
18684 (selftest::test_one_liner_many_fixits_1): ...this, and update
18685 comment and expected output to reflect that the multiple fix-it
18686 hints are now consolidated into one insertion.
18687 (selftest::test_one_liner_many_fixits_2): New test.
18688 (selftest::test_diagnostic_show_locus_one_liner): Update for
18690 (selftest::test_fixit_consolidation): Update for fix-it API
18692 * diagnostic.c (print_parseable_fixits): Likewise.
18693 * edit-context.c (edited_line::m_line_events): Convert from
18694 auto_vec <line_event *> to auto_vec <line_event>.
18695 (class line_event): Convert from abstract base class to a concrete
18696 class, taking over the role of replace_event.
18697 (class insert_event): Delete.
18698 (class replace_event): Rename to class line_event. Convert to
18700 (edit_context::add_fixits): Reimplement.
18701 (edit_context::apply_insert): Delete.
18702 (edit_context::apply_replace): Rename to...
18703 (edit_context::apply_fixit): ...this. Convert to half-open range.
18704 (edited_file::apply_insert): Delete.
18705 (edited_file::apply_replace): Rename to...
18706 (edited_file::apply_fixit): ...this.
18707 (edited_line::~edited_line): Drop deletion of events.
18708 (edited_line::apply_insert): Delete.
18709 (edited_line::apply_replace): Rename to...
18710 (edited_line::apply_fixit): ...this. Convert to half-open range.
18711 Update for change to type of m_line_events.
18712 * edit-context.h (edit_context::apply_insert): Delete.
18713 (edit_context::apply_replace): Rename to...
18714 (edit_context::apply_fixit): ...this.
18716 2017-05-01 Martin Sebor <msebor@redhat.com>
18718 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
18721 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
18724 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
18725 __get_cpuid_max returns 0.
18726 (__get_cpuid_count): Ditto.
18728 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
18730 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
18731 replacement expression is another instance of one of its arguments.
18733 2017-05-01 Jakub Jelinek <jakub@redhat.com>
18736 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
18737 check for stack push/pop autoinc.
18738 * config/i386/i386.c (ix86_agi_dependent): Return false
18739 if the only reason why modified_in_p returned true is that
18740 addr is SP based and set_insn is a push or pop.
18742 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
18744 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
18747 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
18750 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
18751 (account_size_time): Use two predicates - exec_pred and
18753 (evaluate_conditions_for_known_args): Compute both clause and
18755 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
18756 (inline_summary_t::duplicate): Update.
18757 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
18759 (compute_inline_parameters): Likewise.
18760 (estimate_edge_size_and_time): Update caluclation of time.
18761 (estimate_node_size_and_time): Compute both time and nonspecialized
18763 (estimate_ipcp_clone_size_and_time): Update.
18764 (inline_merge_summary): Update.
18765 (do_estimate_edge_time): Update.
18766 (do_estimate_edge_size): Update.
18767 (do_estimate_edge_hints): Update.
18768 (inline_read_section, inline_write_summary): Stream both new predicates.
18769 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
18771 (compute_inlined_call_time): Cleanup.
18772 (big_speedup_p): Update.
18773 (edge_badness): Update.
18774 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
18775 (size_time_entry): Replace predicate by exec_predicate and
18776 nonconst_predicate.
18777 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
18778 (estimate_edge_time): Return also nonspec_time.
18779 (reset_edge_growth_cache): Update.
18781 2017-04-29 Jakub Jelinek <jakub@redhat.com>
18783 PR rtl-optimization/80491
18784 * ifcvt.c (noce_process_if_block): When looking for x setter
18785 with missing else_bb, don't check only the insn right before
18786 cond_earliest, but look for the last insn that x is modified in
18787 within the same bb.
18789 PR rtl-optimization/80491
18790 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
18792 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
18794 PR tree-optimization/80487
18795 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
18797 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18799 PR tree-optimization/79697
18800 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
18801 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
18802 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
18804 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
18805 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
18807 2017-04-28 Martin Sebor <msebor@redhat.com>
18809 PR tree-optimization/80523
18810 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
18811 (init_target_to_host_charmap, target_to_host, target_strtol10): New
18813 (maybe_warn, format_directive, parse_directive): Use new functions.
18814 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
18816 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
18818 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
18820 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
18822 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
18823 target_header_dir): Set correctly.
18824 * configure: Regenerated.
18825 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
18826 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
18827 instead of SYSTEM_HEADER_DIR.
18829 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
18831 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
18832 (estimate_local_effects): Likewise.
18833 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
18834 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
18835 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
18836 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
18837 do_estimate_edge_time, estimate_edge_time): Likewise.
18838 * ipa-inline-analysis.c (estimate_node_size_and_time,
18839 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
18840 (estimate_time_after_inlining): Remove.
18842 2017-04-28 Martin Liska <mliska@suse.cz>
18844 * doc/gcov.texi: Enhance documentation of gcov.
18846 2017-04-28 Martin Liska <mliska@suse.cz>
18848 * doc/gcov.texi: Sort options in alphabetic order.
18849 * doc/gcov-dump.texi: Likewise.
18850 * doc/gcov-tool.texi: Likewise.
18851 * gcov.c (print_usage): Likewise.
18852 * gcov-dump.c (print_usage): Likewise.
18853 * gcov-tool.c (print_merge_usage_message): Likewise.
18854 (print_rewrite_usage_message): Likewise.
18855 (print_overlap_usage_message): Likewise.
18857 2017-04-28 Martin Liska <mliska@suse.cz>
18859 PR gcov-profile/53915
18860 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
18862 2017-04-28 Martin Liska <mliska@suse.cz>
18864 PR gcov-profile/79891
18865 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
18866 is marked by compiler as living on a line.
18867 (get_cycles_count): Remove usage of the union.
18868 (output_intermediate_file): Likewise.
18869 (find_source): Fix GNU coding style.
18870 (accumulate_line_counts): Remove old non-all block mode.
18871 (output_lines): Remove usage of the union.
18872 * profile.c (output_location): Include all BBs, even if
18873 belonging to a same line (and file) as a previous BB.
18875 2017-04-28 Martin Liska <mliska@suse.cz>
18877 * gcov.c (process_args): Handle new argument 'w'.
18878 (read_graph_file): Assign ID to BBs.
18879 (output_branch_count): Display BB # if verbose flag is set.
18880 (output_lines): Likewise for arcs.
18881 (print_usage): Add '--verbose' option help.
18882 * doc/gcov.texi: Document --verbose (-w) option.
18884 2017-04-28 Martin Liska <mliska@suse.cz>
18886 * gcov.c (struct block_location_info): New struct.
18887 (process_file): Fill up the new structure.
18888 (read_graph_file): Replace usage of encoding by the newly added
18890 (add_line_counts): Likewise.
18891 (accumulate_line_counts): Remove usage of the union.
18892 (function_info::function_info): New function.
18893 (function_info::~function_info): Likewise.
18894 (process_file): Call delete instead of release_function.
18895 (release_function): Release the function.
18896 (release_structures): Call delete instead of release_function.
18897 (solve_flow_graph): Replace usage of num_blocks.
18898 (find_exception_blocks): Likewise.
18899 (output_lines): Fix GNU coding style.
18901 2017-04-28 Martin Liska <mliska@suse.cz>
18904 * coverage.c (coverage_remove_note_file): New function.
18905 * coverage.h: Declare the function.
18906 * toplev.c (finalize): Clean if an error has been seen.
18908 2017-04-28 Martin Liska <mliska@suse.cz>
18910 PR gcov-profile/80031
18911 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
18912 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
18913 * gcov.c (read_graph_file): Read just number of blocks.
18914 * profile.c (branch_prob): Do not stream 0 flags per a basic
18917 2017-04-28 Martin Liska <mliska@suse.cz>
18919 * gcov-dump.c (tag_*): Add new argument to declarations.
18920 (dump_gcov_file): Likewise.
18921 (tag_blocks): Add and use new argument depth.
18922 (tag_arcs): Likewise.
18923 (tag_lines): Likewise.
18924 (tag_counters): Likewise.
18925 (tag_summary): Likewise.
18926 (dump_working_sets): Use depth to do a proper indentation.
18928 2017-04-28 Jakub Jelinek <jakub@redhat.com>
18931 * cgraph.h (symtab_node::debug_symtab): No longer inline.
18932 * symtab.c (symtab_node::debug_symtab): Move definition here.
18934 2017-04-28 Richard Biener <rguenther@suse.de>
18936 * lto-streamer.h (LTO_major_version): Bump to 7.
18938 2017-04-28 Richard Biener <rguenther@suse.de>
18940 * tree-vrp.c (assert_info): New struct.
18941 (add_assert_info): New helper.
18942 (register_edge_assert_for_2): Refactor to add asserts to a vector
18944 (register_edge_assert_for_1): Likewise.
18945 (register_edge_assert_for): Likewise.
18946 (finish_register_edge_assert_for): New helper actually registering
18947 asserts where live on edge.
18948 (find_conditional_asserts): Adjust.
18949 (find_switch_asserts): Likewise.
18950 (evrp_dom_walker::try_find_new_range): Generalize.
18951 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
18953 2017-04-27 Marek Polacek <polacek@redhat.com>
18956 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
18957 arg10 and arg11 to itype.
18959 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
18961 * doc/extend.texi (Object Size Checking): Improve grammar.
18963 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
18966 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
18967 that the logic for permitting reciprocal estimates matches that
18970 2017-04-27 Jakub Jelinek <jakub@redhat.com>
18973 * tree.c (type_cache_hasher::equal): Only compare
18974 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
18975 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
18976 non-aggregate element types.
18977 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
18978 about the flag on ARRAY_TYPEs in the comment, formatting fix.
18980 2017-04-27 Richard Biener <rguenther@suse.de>
18982 PR middle-end/80533
18983 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
18984 stripping ARRAY_REFs from MEM_EXPR make sure we're not
18985 keeping a reference to a trailing array.
18987 2017-04-27 Richard Biener <rguenther@suse.de>
18989 PR middle-end/80539
18990 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
18991 being in loop-closed SSA form conservatively.
18992 (chrec_fold_multiply_poly_poly): Likewise.
18994 2017-04-27 Tamar Christina <tamar.christina@arm.com>
18996 PR middle-end/79665
18997 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
18998 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
19000 2017-04-27 Jakub Jelinek <jakub@redhat.com>
19003 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
19004 (aarch64_function_arg_alignment): Return unsigned int again, but still
19005 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
19006 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
19007 Don't emit -Wpsabi note.
19008 (aarch64_function_arg_boundary): Likewise.
19009 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
19012 2017-04-26 Nathan Sidwell <nathan@acm.org>
19014 * tree.h (crc32_unsigned_n): Declare.
19015 (crc32_unsigned, crc32_unsigned): Make inline.
19016 * tree.c (crc32_unsigned_bits): Replace with ...
19017 (crc32_unsigned_n): ... this.
19018 (crc32_unsigned, crc32_byte): Remove.
19019 (crc32_string): Remove unnecessary braces.
19021 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
19023 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
19024 * ipa-inline-analysis.c (MAX_TIME): Remove.
19025 (account_size_time): Use sreal for time.
19026 (dump_inline_summary): Update.
19027 (estimate_function_body_sizes): Update.
19028 (estimate_edge_size_and_time): Update.
19029 (estimate_calls_size_and_time): Update.
19030 (estimate_node_size_and_time): Update.
19031 (inline_merge_summary): Update.
19032 (inline_update_overall_summary): Update.
19033 (estimate_time_after_inlining): Update.
19034 (inline_read_section): Update.
19035 (inline_write_summary): Update.
19036 * ipa-inline.c (compute_uninlined_call_time): Update.
19037 (compute_inlined_call_time): Update.
19038 (recursive_inlining): Update.
19039 (inline_small_functions): Update.
19040 (dump_overall_stats): Update.
19041 * ipa-inline.h: Include sreal.h.
19042 (size_time_entry): Turn time to sreal.
19043 (inline_summary): Turn self_time nad time to sreal.
19045 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
19047 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
19049 (sreal::stream_out, sreal::stream_in): New.
19050 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
19052 2017-04-25 Jakub Jelinek <jakub@redhat.com>
19054 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
19057 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
19060 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
19062 (dimode_scalar_chain::compute_convert_gain): Ditto.
19063 (dimode_scalar_chain::make_vector_copies): Ditto.
19064 (dimode_scalar_chain::convert_reg): Ditto.
19065 (dimode_scalar_chain::convert_insn): Ditto.
19066 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
19067 (VI248_AVX512BW_1): New mode iterator.
19068 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
19069 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
19072 2017-04-25 Martin Sebor <msebor@redhat.com>
19074 PR tree-optimization/80497
19075 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
19076 constants are representable in HOST_WIDE_INT.
19077 (parse_directive): Ditto.
19079 2017-04-25 Martin Sebor <msebor@redhat.com>
19082 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
19083 (new_zero_array): Adjust signature.
19084 (dom_info::dom_init): Used unsigned rather that size_t.
19085 (dom_info::dom_info): Same.
19087 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19088 Jakub Jelinek <jakub@redhat.com>
19091 * config/arm/arm.c: Include gimple.h.
19092 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
19093 returns negative, increment ncrn only if it returned positive.
19094 (arm_needs_doubleword_align): Return int instead of bool,
19095 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
19096 members, but if there is any such non-FIELD_DECL
19097 > PARM_BOUNDARY aligned decl, return -1 instead of false.
19098 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
19099 returns negative, increment nregs only if it returned positive.
19100 (arm_setup_incoming_varargs): Likewise.
19101 (arm_function_arg_boundary): Emit -Wpsabi note if
19102 arm_needs_doubleword_align returns negative, return
19103 DOUBLEWORD_ALIGNMENT only if it returned positive.
19105 2017-04-25 Marek Polacek <polacek@redhat.com>
19108 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
19109 first argument to type.
19111 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
19114 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
19115 type checks to test for compatibility instead of equality.
19117 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19118 Jakub Jelinek <jakub@redhat.com>
19121 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
19123 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
19124 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
19125 the alignment computation, but return their maximum in warn_alignment.
19126 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
19127 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
19129 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
19130 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
19133 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19135 * config/arc/simdext.md (dmpyh): Fix typo.
19137 2017-04-25 Richard Biener <rguenther@suse.de>
19139 PR tree-optimization/80492
19140 * alias.c (compare_base_decls): Handle registers with asm
19141 specification conservatively.
19142 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
19143 compare_base_decls returning dont-know properly.
19145 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19147 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
19148 (legitimate_offset_address_p): New function.
19149 (arc_legitimate_address_p): Use above function.
19151 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19153 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
19155 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19157 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
19158 ACCH registers whenever they are available.
19160 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19162 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
19163 double regs fix when not used.
19165 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19167 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
19169 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
19170 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
19172 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19174 * config/arc/arc.c (arc_output_addsi): Check for h-register class
19175 when emitting short ADD instructions.
19177 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19179 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
19181 (cmpsi_cc_c_insn): Likewise.
19182 (cbranchsi4_scratch): Compute proper instruction length using
19183 compact_hreg_operand.
19184 * config/arc/predicates.md (compact_hreg_operand): New predicate.
19186 2017-04-25 Richard Biener <rguenther@suse.de>
19188 PR middle-end/80509
19189 * passes.c (pass_manager::pass_manager): Initialize
19190 m_name_to_pass_map.
19192 2017-04-25 Richard Biener <rguenther@suse.de>
19194 PR tree-optimization/79201
19195 * tree-ssa-sink.c (statement_sink_location): Handle calls.
19197 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19200 * config/s390/vector.md: Split MEM->GPR vector moves for
19201 non-s_operand addresses.
19203 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19206 * config/s390/predicates.md (reload_const_wide_int_operand): New
19208 * config/s390/s390.md ("movti"): Remove d/P alternative.
19209 ("movti_bigconst"): New pattern definition.
19211 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
19214 * s390-protos.h (s390_expand_cs_hqi): Removed.
19215 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
19216 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
19217 modes as well as CCZ1mode and CCZmode.
19218 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
19219 signature of s390_emit_compare_and_swap.
19220 (s390_expand_cs_hqi): Likewise, make static.
19221 (s390_expand_cs_tdsi): Generate an explicit compare before trying
19222 compare-and-swap, in some cases.
19223 (s390_expand_cs): Wrapper function.
19224 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
19226 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
19227 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
19228 patterns for small and large integers. Forbid symref memory operands.
19229 Move expander to s390.c. Require cc register.
19230 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
19231 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
19232 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
19233 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
19234 symref memory operands. Remove CC mode and call s390_match_ccmode
19236 ("atomic_exchange<mode>"): Allow and implement all integer modes.
19238 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
19240 * config/s390/s390.md (define_peephole2): New peephole to help
19241 combining the load-and-test pattern with volatile memory.
19243 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
19245 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
19246 with CCZmode for TARGET_Z196.
19248 2017-04-25 Jakub Jelinek <jakub@redhat.com>
19250 PR rtl-optimization/80501
19251 * combine.c (make_compound_operation_int): Set subreg_code to SET
19252 even for AND with mask of the sign bit of mode.
19254 PR rtl-optimization/80500
19255 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
19256 sum's initial value.
19258 2017-04-25 Julian Brown <julian@codesourcery.com>
19259 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19261 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
19263 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
19265 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
19267 2017-04-25 Julian Brown <julian@codesourcery.com>
19268 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19270 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
19271 (thunderx2t99_sha): New Reservation.
19273 2017-04-25 Julian Brown <julian@codesourcery.com>
19274 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19276 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
19277 type for 1-element load.
19279 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
19281 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
19283 2017-04-24 Martin Jambor <mjambor@suse.cz>
19285 PR tree-optimization/80293
19286 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
19287 char arrays not totally scalarizable if it is false.
19288 (analyze_all_variable_accesses): Pass correct value in the new
19289 parameter. Add a statistics counter.
19291 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
19293 PR middle-end/79931
19294 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
19296 2017-04-24 Richard Biener <rguenther@suse.de>
19298 PR tree-optimization/80494
19299 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
19300 out for complex types.
19302 2017-04-24 Richard Biener <rguenther@suse.de>
19304 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
19305 * tree-ssa-sccvn.c (print_scc): Print SCC size.
19306 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
19307 (DFS): Adjust and never fail.
19308 (sccvn_dom_walker::fail): Remove.
19309 (sccvn_dom_walker::before_dom_children): Adjust.
19310 (run_scc_vn): Likewise and never fail.
19311 * tree-ssa-pre.c (pass_pre::execute): Adjust.
19312 (pass_fre::execute): Likewise.
19314 2017-04-24 Richard Biener <rguenther@suse.de>
19316 PR tree-optimization/79725
19317 * tree-ssa-sink.c (statement_sink_location): Return whether
19318 failure reason was zero uses. Move that check later.
19319 (sink_code_in_bb): Deal with zero uses by removing the stmt
19322 2017-04-24 Richard Biener <rguenther@suse.de>
19325 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
19326 pointer-based references.
19328 2017-04-24 Richard Biener <rguenther@suse.de>
19331 * pass_manager.h (pass_manager::operator new): Remove.
19332 (pass_manager::operator delete): Likewise.
19333 * passes.c (pass_manager::operator new): Remove.
19334 (pass_manager::operator delete): Likewise.
19335 (pass_manager::pass_manager): Zero individual pass members.
19337 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
19340 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
19341 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
19342 Check "XEXP (src, 1)" operand here.
19343 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
19344 Check "XEXP (src, 1)" operand here.
19345 (dimode_scalar_chain::make_vector_copies): Detect count register
19346 of a shift instruction. Zero extend count register from QImode
19347 to DImode to satisfy vector shift pattern count operand predicate.
19348 Substitute vector shift count operand with a DImode copy.
19349 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
19352 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
19354 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
19355 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
19356 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
19357 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
19358 (UNSPEC_NOREX_MEM): Remove definition.
19360 2017-04-21 Richard Biener <rguenther@suse.de>
19362 PR tree-optimization/79547
19363 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
19364 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
19365 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
19366 without any constraints.
19368 2017-04-21 Richard Biener <rguenther@suse.de>
19370 PR tree-optimization/78847
19371 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
19373 2017-04-21 Richard Biener <rguenther@suse.de>
19375 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
19376 (build_distinct_type_copy): Likewise.
19377 (build_variant_type_copy): Likewise.
19378 * tree.c (build_qualified_type): Pass down mem-stat info.
19379 (build_distinct_type_copy): Likewise.
19380 (build_variant_type_copy): Likewise.
19382 2017-04-21 Richard Biener <rguenther@suse.de>
19384 PR tree-optimization/80237
19385 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
19387 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
19388 for a simplified result.
19390 2016-04-21 Richard Biener <rguenther@suse.de>
19392 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
19393 sth as strict as a simple_iv but a chrec without symbols and an
19394 operand defined in the loop we are peeling (and not some subloop).
19395 (propagate_constants_for_unrolling): Propagate all constants.
19397 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
19400 * config/i386/i386.c (print_reg): Remove assert for disalowed
19401 regno values, call output_operand_lossage instead.
19403 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
19406 * config/i386/constraints.md (Yc): New register constraint.
19407 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
19408 Use Yc constraint for alternative 2 of operand 0. Remove
19409 preferred_for_speed attribute.
19411 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
19413 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
19414 lastprivate clauses in SIMT case.
19416 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
19418 * doc/invoke.texi (-Wextra-semi): Document new warning option.
19420 2017-04-20 Richard Biener <rguenther@suse.de>
19422 PR tree-optimization/57796
19423 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
19424 as N scalar stores.
19425 (vect_model_load_cost): Cost gathers as N scalar loads.
19427 2017-04-20 Richard Biener <rguenther@suse.de>
19429 * ggc-page.c (ggc_allocated_p): Rename to ...
19430 (safe_lookup_page_table_entry): ... this and return the lookup
19432 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
19434 2017-04-20 Richard Biener <rguenther@suse.de>
19436 PR tree-optimization/80453
19437 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
19438 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
19439 from the conditions.
19440 (vn_phi_eq): Pass them down.
19441 (vn_phi_lookup): Record them.
19442 (vn_phi_insert): Likewise.
19444 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
19446 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
19447 uninitialized variable warning to avoid buffer overrun.
19449 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
19452 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
19453 is suppressed for '{ 0 }' in C.
19455 2017-04-20 Jakub Jelinek <jakub@redhat.com>
19457 * BASE-VER: Set to 8.0.0.
19459 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
19461 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
19462 priority .init_array and .fini_array section with SECTION_NOTYPE
19465 2017-04-20 Jakub Jelinek <jakub@redhat.com>
19467 PR middle-end/80423
19468 * tree.h (build_array_type): Add typeless_storage default argument.
19469 * tree.c (type_cache_hasher::equal): Also compare
19470 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
19471 (build_array_type): Add typeless_storage argument, set
19472 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
19474 (build_nonshared_array_type): Adjust build_array_type_1 caller.
19475 (build_array_type): Likewise. Add typeless_storage argument.
19477 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
19478 Jakub Jelinek <jakub@redhat.com>
19480 PR tree-optimization/80426
19481 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
19482 operation on symbolic operands, also compute the overflow for the
19483 invariant part when the operation degenerates into a negation.
19485 2017-04-19 Jakub Jelinek <jakub@redhat.com>
19488 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
19489 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
19492 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
19494 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
19497 * config/avr/avr.c (tree.h): Include it.
19498 (cgraph.h): Include it.
19499 (avr_encode_section_info): Don't warn for uninitialized progmem
19500 variable if it's just an alias.
19502 2017-04-19 Richard Biener <rguenther@suse.de>
19505 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
19506 when needed by AutoPGO.
19508 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
19511 * doc/lto.texi: Remove an extra 'that'.
19513 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
19515 PR rtl-optimization/80429
19516 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
19517 are only used in debug insns.
19519 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
19520 Vladimir Makarov <vmakarov@redhat.com>
19522 * config/sparc/predicates.md (input_operand): Add comment. Return
19523 true for any memory operand when LRA is in progress.
19524 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
19526 2017-04-18 Jeff Law <law@redhat.com>
19529 * mips.md ({return,simple_return}_internal): Do not overwrite
19532 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19534 PR tree-optimization/80443
19535 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
19536 instead of adding 1, subtract -1 and similarly instead of subtracting
19539 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
19541 PR rtl-optimization/80357
19542 * haifa-sched.c (tmp_bitmap): New variable.
19543 (model_recompute): Handle duplicate use records.
19544 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
19545 (free_global_sched_pressure_data): Free it.
19547 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
19550 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
19551 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
19552 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
19553 instead of SYSTEM_HEADER_DIR.
19555 2017-04-18 Jeff Law <law@redhat.com>
19557 PR middle-end/80422
19558 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
19559 predecessors after walking up the insn chain.
19561 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19564 * dwarf2out.c (modified_type_die): Try harder not to emit internal
19565 sizetype type into debug info.
19567 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
19570 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
19571 unneeded test for TARGET_UPPER_REGS_SF.
19572 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
19574 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19577 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
19578 instead of gsi_after_labels.
19580 2017-04-18 Jeff Law <law@redhat.com>
19582 * regcprop.c (maybe_mode_change): Avoid creating copies of the
19586 2017-04-13 Jeff Law <law@redhat.com>
19587 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
19588 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
19590 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
19593 * config/avr/avr.c (intl.h): Include it.
19594 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
19596 2017-04-18 Martin Liska <mliska@suse.cz>
19598 PR gcov-profile/78783
19599 * gcov-tool.c (gcov_output_files): Validate that destination
19600 file is either removed by the tool or by a user.
19602 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
19603 Guy Benyei <guybe@mellanox.com>
19605 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
19606 block, and do not negate it, the stored id is already negative.
19608 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
19610 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
19612 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
19615 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
19616 masks of options that should be turned off if the VSX vector
19617 options are turned off.
19618 (OTHER_P8_VECTOR_MASKS): Likewise.
19619 (OTHER_VSX_VECTOR_MASKS): Likewise.
19620 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
19621 rs6000_disable_incompatible_switches to validate no type switches
19623 (rs6000_incompatible_switch): New function to disallow turning on
19624 other vector options if -mno-vsx, -mno-power8-vector, or
19625 -mno-power9-vector are specified.
19627 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19629 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
19631 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19633 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
19634 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
19635 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
19636 (ARG_POINTER_CFA_OFFSET): Likewise.
19638 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19640 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
19641 conditions to take advantage of various optimizations.
19643 2017-04-13 Jeff Law <law@redhat.com>
19645 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
19646 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
19647 (zero_extendsidi2_dext): Likewise.
19649 2017-04-13 Jakub Jelinek <jakub@redhat.com>
19652 * fold-const.c (fold_ternary_loc): Revert
19653 use op0 instead of fold_convert_loc (loc, type, arg0) part of
19656 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
19658 PR rtl-optimization/80343
19659 * lra-remat.c (update_scratch_ops): Assign original hard reg to
19660 new scratch pseudo.
19662 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
19665 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
19666 to ubsan_encode_value.
19668 2017-04-13 Jeff Law <law@redhat.com>
19670 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
19671 appearing in DEBUG_INSNs.
19673 2017-04-13 Martin Liska <mliska@suse.cz>
19675 PR gcov-profile/80413
19676 * gcov-io.c (gcov_write_string): Copy to buffer just when
19677 allocated size is greater than zero.
19679 2017-04-13 Jakub Jelinek <jakub@redhat.com>
19682 * dwarf2out.c (decls_for_scope): Ignore declarations of
19683 current_function_decl in BLOCK_NONLOCALIZED_VARS.
19685 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
19688 * ipa-visibility.c (non_local_p): Fix typos.
19689 (localize_node): When localizing symbol in same comdat group,
19690 dissolve the group only when we know external symbols are going
19692 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
19694 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19696 PR tree-optimization/79390
19697 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
19698 order does not result in usable sequence, retry with reversed operand
19704 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
19705 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
19706 op0 instead of fold_convert_loc (loc, type, arg0).
19708 2017-04-12 Jeff Law <law@redhat.com>
19710 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
19711 has a delay slot in the generated code.
19713 * config/cris/cris.md (cris_preferred_reload_class): Return
19714 GENNONACR_REGS rather than GENERAL_REGS.
19716 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19719 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
19720 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
19721 signedness of the result type.
19723 2017-04-12 Richard Biener <rguenther@suse.de>
19724 Jeff Law <law@redhat.com>
19726 PR tree-optimization/80359
19727 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
19728 trim stores to TARGET_MEM_REFs.
19730 2017-04-12 Richard Biener <rguenther@suse.de>
19732 PR tree-optimization/79390
19733 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
19734 threading case even more.
19736 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
19739 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
19740 for quad_address_p for TImode, instead of just not indexed_address.
19742 2017-04-12 Richard Biener <rguenther@suse.de>
19743 Bernd Edlinger <bernd.edlinger@hotmail.de>
19745 PR middle-end/79671
19746 * alias.c (component_uses_parent_alias_set_from): Handle
19747 TYPE_TYPELESS_STORAGE.
19748 (get_alias_set): Likewise.
19749 * tree-core.h (tree_type_common): Add typeless_storage flag.
19750 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
19751 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
19752 for types containing members with TYPE_TYPELESS_STORAGE.
19753 (place_field): Likewise.
19754 (layout_type): Likewise for ARRAY_TYPE.
19755 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
19756 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
19757 TYPE_TYPELESS_STORAGE.
19758 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
19760 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19763 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
19764 first argument to type.
19766 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19770 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
19771 CONST0_RTX (mode) rather than const0_rtx where appropriate.
19772 (rs6000_expand_binop_builtin): Likewise.
19773 (rs6000_expand_ternop_builtin): Likewise; also add missing
19774 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
19775 vshasigma built-ins.
19776 * doc/extend.texi: Document that vec_xxpermdi's third argument
19777 must be a constant.
19779 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
19781 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
19782 Use shift_const cost parameter when calculating gain of STV shifts.
19784 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
19786 PR rtl-optimization/70478
19787 * lra-constraints.c (process_alt_operands): Check memory for
19788 disfavoring memory insn operand.
19790 2017-04-11 Jakub Jelinek <jakub@redhat.com>
19792 PR middle-end/80100
19793 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
19794 left shift in unsigned HOST_WIDE_INT type.
19796 PR rtl-optimization/80385
19797 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
19798 (not (neg X)) into (plus X -1) for complex or non-integral modes.
19801 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
19802 if they have any depend clauses.
19804 2017-04-11 Martin Liska <mliska@suse.cz>
19807 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
19808 * ipa-split.c (split_function): Create a local comdat symbol
19809 if caller is in a comdat group.
19811 2017-04-11 Martin Liska <mliska@suse.cz>
19814 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
19817 2017-04-11 Martin Sebor <msebor@redhat.com>
19819 PR middle-end/80364
19820 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
19821 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
19823 (directive::set_width, directive::set_precision, format_character):
19825 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
19828 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
19831 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
19832 conflict, set target->arch_name instead of target->cpu_name.
19834 2017-04-11 Richard Biener <rguenther@suse.de>
19836 PR tree-optimization/80374
19837 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
19838 build_zero_cst, remove fold_convertible_p check again.
19840 2017-04-11 Martin Liska <mliska@suse.cz>
19843 * ubsan.c (instrument_object_size): Do not instrument register
19846 2017-04-11 Jakub Jelinek <jakub@redhat.com>
19849 * config/i386/i386-builtin-types.def
19850 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
19851 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
19852 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
19853 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
19854 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
19855 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
19856 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
19857 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
19858 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
19859 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
19860 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
19861 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
19862 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
19863 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
19864 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
19865 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
19866 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
19867 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
19868 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
19869 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
19870 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
19871 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
19872 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
19873 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
19874 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
19875 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
19876 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
19877 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
19878 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
19879 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
19880 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
19881 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
19882 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
19883 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
19884 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
19885 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
19886 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
19887 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
19888 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
19889 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
19890 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
19891 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
19892 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
19893 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
19894 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
19895 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
19897 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
19898 flag to second_arg_count, handle 4 argument function type _COUNT
19899 aliases, handle second_arg_count on second argument rather than last.
19901 2017-04-10 Jeff Law <law@redhat.com>
19903 PR tree-optimization/80374
19904 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
19905 record anything if we can not convert integer_zero_node to the
19908 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
19911 * config/rs6000/rs6000.c (rs6000_option_override_internal):
19912 Enhance special handling given to the TARGET_P9_MINMAX option in
19913 relation to certain other options.
19915 2017-04-10 Bin Cheng <bin.cheng@arm.com>
19917 PR tree-optimization/80153
19918 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
19919 remove POINTER_PLUS_EXPR's base part directly, rather than through
19922 2017-04-10 Richard Biener <rguenther@suse.de>
19923 Bin Cheng <bin.cheng@arm.com>
19925 PR tree-optimization/80153
19926 * tree-affine.c (aff_combination_to_tree): Get base pointer from
19927 the first element of pointer type aff_tree. Build result expr in
19929 (add_elt_to_tree): Convert to type unconditionally. Remove other
19930 fold_convert calls.
19931 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
19932 (rewrite_use_nonlinear_expr): Check invariant using iv information.
19934 2017-04-10 Richard Biener <rguenther@suse.de>
19936 * tree-ssa-structalias.c (find_func_aliases): Properly handle
19939 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
19941 PR rtl-optimization/70478
19942 * lra-constraints.c (curr_small_class_check): New.
19943 (update_and_check_small_class_inputs): New.
19944 (process_alt_operands): Update curr_small_class_check. Disfavor
19945 alternative insn memory operands. Check available regs for small
19948 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
19951 * config/mips/mips.opt (-mvirt): Update description.
19952 * doc/invoke.texi (-mvirt): Likewise.
19954 2017-04-10 Richard Biener <rguenther@suse.de>
19956 PR middle-end/80362
19957 * fold-const.c (fold_binary_loc): Look at unstripped ops when
19958 looking for NEGATE_EXPR in -A / -B to A / B folding.
19960 2017-04-10 Martin Liska <mliska@suse.cz>
19962 PR gcov-profile/80224
19963 * gcov.c (print_usage): Fix usage string.
19964 (get_gcov_intermediate_filename): Remove.
19965 (output_gcov_file): Use both for normal and intermediate format.
19966 (generate_results): Do not initialize special file for
19967 intermediate format.
19969 2017-04-10 Richard Biener <rguenther@suse.de>
19971 PR tree-optimization/80304
19972 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
19975 2017-04-10 Nathan Sidwell <nathan@acm.org>
19978 * config/rs6000/rs6000.c (rs6000_vector_type): New.
19979 (rs6000_init_builtins): Use it.
19981 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19983 * config/arm/arm.md (<mrc>): Add mode to SET source.
19984 (<mrrc>): Likewise.
19986 2017-04-10 Richard Biener <rguenther@suse.de>
19988 PR middle-end/80344
19989 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
19991 2017-04-10 Jakub Jelinek <jakub@redhat.com>
19994 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
19995 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
19996 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
19997 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
19998 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
19999 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
20000 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
20001 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
20002 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
20003 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
20004 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
20005 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
20006 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
20007 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
20008 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
20009 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
20010 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
20011 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
20012 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
20013 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
20014 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
20015 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
20016 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
20018 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
20020 PR rtl-optimization/70478
20021 * lra-constraints.c: Reverse the last patch.
20023 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
20025 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
20026 Add comment for WCHAR_T.
20028 2017-04-08 Martin Liska <mliska@suse.cz>
20031 2017-04-07 Martin Liska <mliska@suse.cz>
20034 * ipa-split.c (split_function): Add function part to a same comdat
20037 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20040 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
20042 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
20044 * rs6000/rs6000.c (vec_load_pendulum): Rename...
20045 (vec_pairing): ...to this.
20046 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
20047 (rs6000_sched_init): Adjust for name change.
20048 (struct rs6000_sched_context): Likewise.
20049 (rs6000_init_sched_context): Likewise.
20050 (rs6000_set_sched_context): Likewise.
20052 2017-04-07 Jakub Jelinek <jakub@redhat.com>
20058 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
20060 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
20061 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
20062 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
20064 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
20066 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
20068 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
20070 PR rtl-optimization/70703
20071 * ira-color.c (update_conflict_hard_regno_costs): Use
20072 int64_t instead of HOST_WIDE_INT.
20074 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
20076 PR rtl-optimization/70478
20077 * lra-constraints.c (process_alt_operands): Disfavor alternative
20078 insn memory operands.
20080 2017-04-07 Jeff Law <law@redhat.com>
20082 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
20083 CALL and NOTE_INSN_CALL_ARG_LOCATION.
20085 2017-04-07 Martin Liska <mliska@suse.cz>
20088 * config/aarch64/aarch64.c (aarch64_process_target_attr):
20089 Show error message instead of an ICE.
20091 2017-04-07 Martin Liska <mliska@suse.cz>
20094 * ipa-split.c (split_function): Add function part to a same comdat
20097 2017-04-07 Richard Biener <rguenther@suse.de>
20099 PR middle-end/80341
20100 * tree.c (get_unwidened): Also handle ! for_type case for
20102 * convert.c (do_narrow): Split out from ...
20103 (convert_to_integer_1): ... here. Do not pass final truncation
20104 type to get_unwidened for TRUNC_DIV_EXPR.
20106 2017-04-07 Richard Biener <rguenther@suse.de>
20108 * tree-affine.c (wide_int_ext_for_comb): Take type rather
20110 (aff_combination_const): Adjust.
20111 (aff_combination_scale): Likewise.
20112 (aff_combination_add_elt): Likewise.
20113 (aff_combination_add_cst): Likewise.
20114 (aff_combination_convert): Likewise.
20115 (add_elt_to_tree): Likewise. Remove unused argument.
20116 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
20118 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
20120 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
20122 * config/arm/arm.c (arm_default_short_enums): Use
20123 ARM_DEFAULT_SHORT_ENUMS.
20124 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
20126 2017-04-06 Jakub Jelinek <jakub@redhat.com>
20129 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
20130 members with redundant out-of-class redeclaration.
20132 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
20135 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
20136 * config/i386/i386.md (*zero_extendsidi2):
20137 Add (?*x,*x) and (?*v,*v) alternatives.
20139 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
20142 * config/i386/i386.c (ix86_expand_builtin)
20143 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
20144 mode from insn data. Convert operands to insn operand mode.
20145 Copy operands that don't satisfy insn predicate to a register.
20147 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
20149 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
20152 2017-04-06 Richard Biener <rguenther@suse.de>
20154 PR tree-optimization/80334
20155 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
20156 preserve alignment of accesses.
20158 2017-04-06 Richard Biener <rguenther@suse.de>
20160 PR tree-optimization/80262
20161 * tree-sra.c (build_ref_for_offset): Preserve address-space
20163 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
20164 Drop useless address-space information on MEM_REF offsets.
20166 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
20168 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
20170 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
20172 PR rtl-optimization/70703
20173 * ira-color.c (update_conflict_hard_regno_costs): Use
20174 HOST_WIDE_INT instead of long.
20176 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
20179 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
20180 not defined for x86_64 target. Add -mmmx target option when __SSE2__
20182 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
20183 for x86_64 target. Handle -m3dnowa option.
20185 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
20187 PR rtl-optimization/70703
20188 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
20189 (update_conflict_hard_regno_costs): Use long instead of unsigned
20190 arithmetic for cost calculation.
20192 2017-04-05 Jakub Jelinek <jakub@redhat.com>
20193 Bernd Edlinger <bernd.edlinger@hotmail.de>
20196 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
20199 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
20202 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
20203 ptr_mode with Pmode throughout.
20204 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
20205 into probe_stack_range and use DImode.
20207 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
20210 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
20211 call_eh_return is true.
20213 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20215 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
20216 Initialize last_match_fntype_index.
20218 2017-04-05 Jakub Jelinek <jakub@redhat.com>
20221 * tree-nvr.c: Include internal-fn.h.
20222 (pass_return_slot::execute): Ignore internal calls without
20225 2017-04-04 Jakub Jelinek <jakub@redhat.com>
20226 Richard Biener <rguenther@suse.de>
20229 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
20230 captures used multiple times, except for the last use.
20231 * generic-match-head.c: Include gimplify.h.
20233 2017-04-04 Jakub Jelinek <jakub@redhat.com>
20235 PR tree-optimization/79390
20236 * target.h (struct noce_if_info): Declare.
20237 * targhooks.h (default_noce_conversion_profitable_p): Declare.
20238 * target.def (noce_conversion_profitable_p): New target hook.
20239 * ifcvt.h (struct noce_if_info): New type, moved from ...
20240 * ifcvt.c (struct noce_if_info): ... here.
20241 (noce_conversion_profitable_p): Renamed to ...
20242 (default_noce_conversion_profitable_p): ... this. No longer
20244 (noce_try_store_flag_constants, noce_try_addcc,
20245 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
20246 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
20247 instead of noce_conversion_profitable_p.
20248 * config/i386/i386.c: Include ifcvt.h.
20249 (ix86_option_override_internal): Don't override
20250 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
20251 (ix86_noce_conversion_profitable_p): New function.
20252 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
20253 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
20254 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
20255 * doc/tm.texi: Regenerated.
20257 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20259 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
20262 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20265 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
20266 instructions for small multiply cores.
20268 2017-04-04 Jeff Law <law@redhat.com>
20270 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
20272 (mips_expand_vec_perm_const): Initialize elements in orig_perm
20273 that are not set by the loop over the elements.
20275 2017-04-04 Jakub Jelinek <jakub@redhat.com>
20278 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
20279 int mode, convert_modes it to mode as unsigned, otherwise use
20280 lowpart_subreg to mode rather than SImode.
20281 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
20282 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
20283 Use DImode instead of SImode for the shift count operand.
20284 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
20287 2017-04-04 Richard Biener <rguenther@suse.de>
20289 PR middle-end/80281
20290 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
20291 arithmetic done for the negate or the plus. Simplify.
20292 (A - (-B) -> A + B): Likewise.
20293 * fold-const.c (split_tree): Make sure to not negate pointers.
20295 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
20297 PR rtl-optimization/60818
20298 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
20299 a compare of comparisons with the thing compared if this results
20300 in a different machine mode.
20302 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
20304 * alias.c (base_alias_check): Fix typo in comment.
20305 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
20306 * cgraphunit.c (symbol_table::compile): Likewise.
20307 * collect2.c (maybe_run_lto_and_relink): Likewise.
20308 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
20309 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
20310 * config/avr/avr.c (avr_map_op_t): Likewise.
20311 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
20312 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
20313 * config/epiphany/epiphany.md (movcc): Likewise.
20314 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
20315 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
20317 * config/mips/mips.c (mips_save_restore_reg): Likewise.
20318 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
20319 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
20320 * config/sh/sh.c (sh_rtx_costs): Likewise.
20321 * fold-const.c (fold_truth_andor): Likewise.
20322 * genautomata.c (collapse_flag): Likewise.
20323 * gengtype.h (struct type::u::s): Likewise.
20324 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
20325 * input.c (FORMAT_AMOUNT): Likewise.
20326 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
20327 (known_aggs_to_agg_replacement_list): Likewise.
20328 * ipa-inline-analysis.c: Likewise.
20329 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
20330 * ipa-polymorphic-call.c
20331 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
20332 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
20333 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
20335 * modulo-sched.c (apply_reg_moves): Likewise.
20336 * omp-expand.c (build_omp_regions_1): Likewise.
20337 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
20338 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
20339 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
20340 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
20341 * value-prof.c: Likewise.
20342 * var-tracking.c (val_reset): Likewise.
20344 2017-04-03 Richard Biener <rguenther@suse.de>
20346 PR tree-optimization/80275
20347 * fold-const.c (split_address_to_core_and_offset): Handle
20350 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
20352 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
20353 descriptors is at least equal to that of functions.
20355 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
20357 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
20359 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
20362 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
20363 (mov<IMOD4:mode>): New expander.
20364 (*mov<IMOD4:mode>_internal): New insn and split pattern.
20366 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
20368 PR rtl-optimization/79405
20369 * fwprop.c (propagations_left): New variable.
20370 (forward_propagate_into): Decrement it.
20371 (fwprop_init): Initialize it.
20372 (fw_prop): If the variable has reached zero, stop propagating.
20373 (fwprop_addr): Ditto.
20375 2017-03-31 Jakub Jelinek <jakub@redhat.com>
20378 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
20379 a FUNCTION_DECL, pass it as decl instead of origin to
20382 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
20384 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
20387 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
20390 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
20391 TARGET_VSX_SMALL_INTEGER.
20393 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20395 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20396 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
20398 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
20400 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
20401 extraction from odd-numbered MSA register.
20403 2017-03-31 Jakub Jelinek <jakub@redhat.com>
20405 PR middle-end/80173
20406 * expmed.c (store_bit_field_1): Don't attempt to create
20407 a word subreg out of hard registers wider than word if they
20408 have HARD_REGNO_NREGS of 1 for their mode.
20410 PR middle-end/80163
20411 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
20412 conversions to integer types wider than word and pointer.
20415 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
20416 (rtx_equal_for_cselib_p): Pass 0 to it.
20417 * cselib.c (cselib_hasher::equal): Likewise.
20418 (rtx_equal_for_cselib_1): Add depth argument. If depth
20419 is 128, don't look up VALUE locs and punt. Increment
20420 depth in recursive calls when walking VALUE locs.
20422 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
20424 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
20425 (make_gcov_file_name): Use the canonical path name for generating
20427 (read_line): Fix handling of files with ascii null bytes.
20429 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
20431 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
20432 to initialise a vector register instead
20433 of using a const_int.
20435 2017-03-30 Jakub Jelinek <jakub@redhat.com>
20437 PR translation/80189
20438 * gimplify.c (omp_default_clause): Use %qs instead of %s in
20439 diagnostic messages.
20441 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
20444 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
20445 (dfp_diex_<mode>): Update mode of operand 1.
20446 * doc/extend.texi (dxex, dxexq): Document change to return type.
20447 (diex, diexq): Document change to argument type.
20449 2017-03-30 Martin Jambor <mjambor@suse.cz>
20452 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
20453 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
20454 it reflects the signature changes performed at the callee side.
20455 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
20456 to cgraph_build_function_type_skip_args.
20457 (build_function_decl_skip_args): Adjust call to the above function.
20459 2017-03-30 Jakub Jelinek <jakub@redhat.com>
20462 * config/i386/sse.md
20463 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
20464 register as dest whenever it is a MEM not rtx_equal_p to the
20465 corresponding dup operand, and when forcing into reg move the
20466 reg into the memory afterwards.
20467 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
20468 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
20469 for the force_reg mode.
20470 (avx512vl_vextractf128<mode>): Use register as dest either
20471 always when a MEM, or when it is a MEM not rtx_equal_p to the
20472 corresponding dup operand, or even not when it is a CONST_VECTOR
20473 depending on the mode and lo vs. hi.
20474 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
20476 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
20477 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
20478 Likewise. Require that operands[2] is even.
20479 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
20480 Remove extraneous parens. Require that operands[2] is a multiple
20482 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
20483 operands[0] is a MEM if <mask_applied>, the predicates/constraints
20484 disallow memory then.
20486 2017-03-30 Richard Biener <rguenther@suse.de>
20488 PR tree-optimization/77498
20489 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
20490 to non-constants over backedges.
20492 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
20494 PR rtl-optimization/80233
20495 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
20496 as last_combined_insn. Do not test for BARRIER_P separately.
20498 2017-03-29 Andreas Schwab <schwab@suse.de>
20501 * calls.c (prepare_call_address): Convert funexp to Pmode before
20502 copying to temp reg.
20504 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20506 PR tree-optimization/80158
20507 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
20508 Handle possible future case of more than one alternate
20510 (replace_rhs_if_not_dup): Likewise.
20511 (replace_one_candidate): Likewise.
20513 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
20515 PR rtl-optimization/80193
20516 * ira.c (ira): Do not check allocation for LRA.
20518 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
20520 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
20521 (nvptx_output_simt_exit): Declare.
20522 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
20523 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
20524 (init_softstack_frame): Move initialization of crtl->is_leaf to...
20525 (nvptx_declare_function_name): ...here. Emit declaration of local
20526 memory space buffer for omp_simt_enter insn.
20527 (nvptx_output_unisimt_switch): New.
20528 (nvptx_output_softstack_switch): New.
20529 (nvptx_output_simt_enter): New.
20530 (nvptx_output_simt_exit): New.
20531 * config/nvptx/nvptx.h (struct machine_function): New fields
20532 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
20533 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
20534 (UNSPECV_SIMT_EXIT): Ditto.
20535 (omp_simt_enter_insn): New insn.
20536 (omp_simt_enter): New expansion.
20537 (omp_simt_exit): New insn.
20538 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
20540 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
20541 (expand_GOMP_SIMT_ENTER_ALLOC): New.
20542 (expand_GOMP_SIMT_EXIT): New.
20543 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
20544 (GOMP_SIMT_ENTER_ALLOC): Ditto.
20545 (GOMP_SIMT_EXIT): Ditto.
20546 * target-insns.def (omp_simt_enter): New insn.
20547 (omp_simt_exit): Ditto.
20548 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
20550 (lower_rec_simd_input_clauses): Implement SIMT privatization.
20551 (lower_rec_input_clauses): Likewise.
20552 (lower_lastprivate_clauses): Handle SIMT privatization.
20554 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
20555 (ompdevlow_adjust_simt_enter): New.
20556 (find_simtpriv_var_op): New.
20557 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
20558 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
20560 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
20561 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
20562 (copy_decl_for_dup_finish): Ditto.
20564 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
20566 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
20569 * config/i386/i386.c (ix86_option_override_internal): Always
20570 allow -mpreferred-stack-boundary=3 for 64-bit targets.
20572 2017-03-28 Bin Cheng <bin.cheng@arm.com>
20574 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
20576 2017-03-28 Bin Cheng <bin.cheng@arm.com>
20578 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
20579 mark new edge's irreducible flag accordign to it.
20580 (vect_do_peeling): Check loop preheader edge's irreducible flag
20581 and pass it to function slpeel_add_loop_guard.
20583 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
20585 PR tree-optimization/80218
20586 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
20587 Update block frequencies and counts.
20589 2017-03-28 Richard Biener <rguenther@suse.de>
20591 PR tree-optimization/78644
20592 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
20593 of a simplification result we may not use it at all.
20595 2017-03-28 Richard Biener <rguenther@suse.de>
20598 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
20599 without arguments, generate default definition of a SSA name.
20601 2017-03-28 Richard Biener <rguenther@suse.de>
20603 PR middle-end/80222
20604 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
20605 TYPE_REF_CAN_ALIAS_ALL references.
20606 * fold-const.c (fold_indirect_ref_1): Likewise.
20608 2017-03-28 Martin Liska <mliska@suse.cz>
20611 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
20612 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
20614 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
20615 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
20617 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
20618 (EXTRA_SPECS): Define.
20619 (SUBTARGET_EXTRA_SPECS): Likewise.
20620 (SUBTARGET_CPP_SPEC): Likewise.
20621 * config/arc/elf.h (EXTRA_SPECS): Renamed to
20622 SUBTARGET_EXTRA_SPECS.
20623 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
20625 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
20627 * config/arc/simdext.md (vst64_insn): Update pattern.
20628 (vld32wh_insn): Likewise.
20629 (vld32wl_insn): Likewise.
20630 (vld64_insn): Likewise.
20631 (vld32_insn): Likewise.
20633 2017-03-28 Marek Polacek <polacek@redhat.com>
20636 * fold-const.c (fold_comparison): Use protected_set_expr_location
20637 instead of SET_EXPR_LOCATION.
20639 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
20641 * tree.c (add_expr): Avoid name lookup warning.
20643 2017-03-27 Jeff Law <law@redhat.com>
20645 PR tree-optimization/80216
20646 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
20647 function name. Limit recursion depth.
20648 (record_temporary_equivalences): Corresponding changes.
20650 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
20652 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
20655 2017-03-27 Jakub Jelinek <jakub@redhat.com>
20658 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
20660 * cfgcleanup.c (reg_note_cfa_p): New array.
20661 (insns_have_identical_cfa_notes): New function.
20662 (old_insns_match_p): Don't cross-jump in between /f
20663 and non-/f instructions. If both i1 and i2 are frame related,
20664 verify all CFA notes, their order and content.
20666 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
20669 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
20670 HImode and SImode with zero extend to DImode to one insn.
20671 (bswap<mode>2_extenddi): Likewise.
20672 (bswapsi2_extenddi): Likewise.
20673 (bswaphi2_extendsi): Likewise.
20674 (bswaphi2): Combine bswap HImode and SImode into one insn.
20675 Separate memory insns from swapping register.
20676 (bswapsi2): Likewise.
20677 (bswap<mode>2): Likewise.
20678 (bswaphi2_internal): Delete, no longer used.
20679 (bswapsi2_internal): Likewise.
20680 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
20681 store, and gpr<-gpr swap insns.
20682 (bswap<mode>2_store): Likewise.
20683 (bswaphi2_reg): Register only splitter, combine with the splitter.
20684 (bswaphi2 splitter): Likewise.
20685 (bswapsi2_reg): Likewise.
20686 (bswapsi2 splitter): Likewise.
20687 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
20688 the insns into load, store, and register/register insns.
20689 (bswapdi2_ldbrx): Likewise.
20690 (bswapdi2_load): Likewise.
20691 (bswapdi2_store): Likewise.
20692 (bswapdi2_reg): Likewise.
20694 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
20696 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
20697 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
20699 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
20702 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
20704 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
20705 special handling for target option conflicts between dform
20706 options (-mpower9-dform, -mpower9-dform-vector,
20707 -mpower9-dform-scalar) and -mno-direct-move.
20709 2017-03-27 Richard Biener <rguenther@suse.de>
20711 PR tree-optimization/80181
20712 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
20714 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20716 * config/arc/predicates.md (move_double_src_operand): Replace the
20717 call to move_double_src_operand with a call to address_operand.
20719 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20721 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
20722 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
20723 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
20725 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20727 * config/arc/predicates.md (long_immediate_loadstore_operand):
20728 Consider scaled addresses cases.
20730 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20732 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
20733 restored when in interrupt.
20734 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
20735 doesn't have delay slot.
20737 2017-03-27 Richard Biener <rguenther@suse.de>
20740 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
20741 inlined thunk clones.
20743 2017-03-27 Jakub Jelinek <jakub@redhat.com>
20746 * asan.c (instrument_derefs): Copy over last operand from
20747 original COMPONENT_REF to the new COMPONENT_REF with
20748 DECL_BIT_FIELD_REPRESENTATIVE.
20749 * ubsan.c (instrument_object_size): Likewise.
20751 2017-03-27 Richard Biener <rguenther@suse.de>
20753 PR tree-optimization/80170
20754 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
20755 sure DR/SCEV didnt fold in constants we do not see when looking
20756 at the reference base alignment.
20758 2017-03-27 Richard Biener <rguenther@suse.de>
20760 PR middle-end/80171
20761 * gimple-fold.c (fold_ctor_reference): Properly guard against
20762 NULL return value from canonicalize_constructor_val.
20764 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
20767 * config/i386/i386.c (ix86_expand_builtin)
20768 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
20769 flags reg setting and flags reg using instructions.
20770 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
20771 clobbering instructions to zero extend op2.
20773 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
20775 * doc/install.texi (Configuration) <--with-aix-soname>:
20776 Update link to AIX ld.
20778 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
20780 PR rtl-optimization/80160
20781 PR rtl-optimization/80159
20782 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
20783 reg_alternate_class into account.
20785 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
20788 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
20789 to consider in curr_insn_transform.
20791 2017-03-24 Jakub Jelinek <jakub@redhat.com>
20793 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
20794 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
20795 and emit_mode_inner.
20797 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20799 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
20800 argument to the overloaded builtin variants. Use the new flag to
20801 deprecate certain builtin variants.
20802 * config/s390/s390-builtin-types.def: Add new builtin types.
20803 * config/s390/s390-builtins.h: Support new flags field for
20804 overloaded builtins.
20805 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
20806 (s390_macro_to_expand): Enable vector float data type.
20807 (s390_cpu_cpp_builtins_internal): Indicate support of the new
20808 builtins by incrementing the __VEC__ version number.
20809 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
20811 (s390_resolve_overloaded_builtin): Emit error messages depending
20812 on the builtin flags.
20813 * config/s390/s390.c (s390_expand_builtin): Support additional
20814 flags argument. Change error message to match the messages
20815 emitted in s390-c.c.
20816 * config/s390/s390.md: New UNSPEC_* constants.
20817 (op_type): Add new instruction types.
20818 * config/s390/vecintrin.h: Add new builtins and test data class
20820 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
20821 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
20822 (VEC_INEXACT, VEC_NOINEXACT): New constants.
20823 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
20824 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
20825 ("vec_mergel<mode>"): V_HW -> VEC_HW.
20827 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
20828 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
20829 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
20830 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
20832 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
20833 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
20834 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
20835 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
20837 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
20838 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
20839 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
20840 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
20841 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
20842 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
20843 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
20845 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
20846 ("vec_scatter_element<V_HW_4:mode>_DI")
20847 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
20848 ("vec_fpint<mode>", "vflls")
20849 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
20850 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
20851 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
20852 ("*vec_cmphe<mode>_cc"): ... these.
20854 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
20855 mode constant instead of magic value.
20857 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20859 * config/s390/s390.c (s390_expand_vec_compare): Support other
20860 vector floating point modes than just V2DF.
20861 (s390_expand_vcond): Likewise.
20862 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
20863 (s390_cannot_change_mode_class): Prevent mode changes between TF
20864 and V1TF in vector registers.
20865 * config/s390/s390.md (DF, SF): New mode attributes.
20866 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
20867 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
20868 SFmode support for VRs.
20869 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
20871 (VFT, VF_HW): New mode iterators.
20872 (vw, sdx): New mode attributes.
20873 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
20874 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
20875 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
20876 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
20877 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
20878 also the new vector floating point modes. Renaming to ...
20880 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
20881 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
20882 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
20883 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
20884 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
20885 ("vec_unordered<mode>"): ... these.
20887 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
20888 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
20889 ("*vec_extendv2df"): New insn definitions.
20891 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20893 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
20894 ("mulditi3_2", "*muldi3_sign"): New patterns.
20895 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
20896 rename the pattern definition.
20898 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20900 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
20902 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
20904 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20906 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
20907 instruction if possible.
20908 * config/s390/vector.md (vec_halfnumelts): New mode
20910 ("*vec_vllezlf<mode>"): New pattern.
20912 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20914 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
20915 ("popcountv4si2", "popcountv2di2"): Rename to ...
20916 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
20917 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
20919 ("popcount<mode>2_vxe"): New pattern.
20921 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20923 * common/config/s390/s390-common.c (processor_flags_table): Add
20925 * config.gcc: Add arch12.
20926 * config/s390/driver-native.c (s390_host_detect_local_cpu):
20927 Default to arch12 for unknown CPU model numbers.
20928 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
20929 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
20930 PROCESSOR_max sanity check.
20931 * config/s390/s390-opts.h (enum processor_type): Add
20933 * config/s390/s390.c (processor_table): Add arch12.
20934 (s390_expand_builtin): Add check for B_VXE flag.
20935 (s390_issue_rate): Add PROCESSOR_ARCH12.
20936 (s390_get_sched_attrmask): Likewise.
20937 (s390_get_unit_mask): Likewise.
20938 (s390_sched_score): Enable z13 scheduling for arch12.
20939 (s390_sched_reorder): Likewise.
20940 (s390_sched_variable_issue): Likewise.
20941 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
20943 (s390_tune_attr): Use z13 scheduling also for arch12.
20944 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
20945 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
20946 (TARGET_VXE_P): New macros.
20947 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
20948 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
20949 * config/s390/s390.opt: Add arch12 as processor_type.
20951 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20953 * config/s390/s390.md
20954 ("fixuns_truncdddi2", "fixuns_trunctddi2")
20955 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
20956 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
20958 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
20959 Rename expanders to ...
20961 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
20962 ("fixuns_truncdddi2_emu"): ... these.
20964 ("fixuns_trunc<mode>si2_emu"): New expander.
20966 ("*fixuns_truncdfdi2_z13"): Rename to ...
20967 ("*fixuns_truncdfdi2_vx"): ... this.
20969 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20971 * config/s390/2964.md: Remove the single element vector compare
20972 instructions which are no longer used.
20973 * config/s390/s390.c (s390_select_ccmode): Remove handling of
20975 (s390_canonicalize_comparison): Remove handling of DFmode
20977 (s390_expand_vec_compare_scalar): Remove function.
20978 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
20979 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
20981 ("*cmp<mode>_ccs"): Add wfcdb instruction.
20983 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20985 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
20987 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
20988 will anyway by matched by mov<mode>_64dfp.
20990 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20992 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
20993 vlef/vstef. Add missing operand to vleif.
20995 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20997 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
20998 pair for all vector types with 64 bit elements.
20999 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
21000 * config/s390/vector.md (V_HW_64): ... here.
21001 (V_128_NOSINGLE): New mode iterator.
21002 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
21003 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
21004 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
21005 ("*vec_load_pairv2di"): Change to ...
21006 ("*vec_load_pair<mode>"): ... this one.
21008 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21010 * config/s390/constraints.md: Add comments.
21011 (jKK): Reject element sizes > 8 bytes.
21012 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
21014 * config/s390/s390.md: Add the s_operand checks formerly in
21015 s390_split_ok_p to various splitters where they are still
21017 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
21018 for 128 bit vectors. Plus two splitters.
21020 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21022 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
21025 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21028 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
21029 error if the boundary argument is not constant.
21031 2017-03-24 Jakub Jelinek <jakub@redhat.com>
21033 PR rtl-optimization/80112
21034 * loop-doloop.c (doloop_condition_get): Don't check condition
21035 if cmp isn't SET with IF_THEN_ELSE src.
21037 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21039 PR tree-optimization/80158
21040 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
21041 replacing a candidate statement, also replace it for the
21042 candidate's alternate interpretation.
21043 (replace_rhs_if_not_dup): Likewise.
21044 (replace_one_candidate): Likewise.
21046 2017-03-24 Richard Biener <rguenther@suse.de>
21048 PR tree-optimization/80167
21049 * graphite-isl-ast-to-gimple.c
21050 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
21052 (translate_isl_ast_to_gimple::get_rename): Likewise.
21054 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
21056 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
21057 handling of certain combinations of target options, including the
21058 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
21059 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
21061 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21064 * config/arm/arm.md (*load_multiple): Add reload_completed to
21065 matching condition.
21067 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21068 Richard Biener <rguenth@suse.de>
21070 PR tree-optimization/79908
21071 PR tree-optimization/80136
21072 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
21073 been cast away, gimplify_and_add suffices.
21075 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
21077 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
21079 2017-03-23 Richard Biener <rguenther@suse.de>
21081 PR tree-optimization/80032
21082 * gimplify.c (gimple_push_cleanup): Forced unconditional
21083 cleanups still have to go to the conditional_cleanups
21086 2017-03-22 Jakub Jelinek <jakub@redhat.com>
21088 PR tree-optimization/80072
21089 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
21091 (next_operand_entry_id): Change type to unsigned int.
21092 (sort_by_operand_rank): Make sure to return the right return value
21093 even if unsigned fields are bigger than INT_MAX.
21094 (struct oecount): Change cnt and id type to unsigned int.
21095 (oecount_hasher::equal): Formatting fix.
21096 (oecount_cmp): Make sure to return the right return value
21097 even if unsigned fields are bigger than INT_MAX.
21098 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
21101 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
21102 TREE_READONLY on result if writing it more than once.
21105 * doc/invoke.texi (-fsanitize=thread): Document that with
21106 -fnon-call-exceptions atomics are not able to throw
21110 * tsan.c: Include tree-eh.h.
21111 (instrument_builtin_call): Call maybe_clean_eh_stmt or
21112 maybe_clean_or_replace_eh_stmt where needed.
21113 (instrument_memory_accesses): Add cfg_changed argument.
21114 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
21115 if it returned true.
21116 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
21118 PR rtl-optimization/63191
21119 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
21120 wrapper function, moved the whole old content into ...
21121 (ix86_delegitimize_address_1): ... this. New inline function.
21122 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
21123 true as last argument instead of ix86_delegitimize_address.
21125 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
21127 * config/aarch64/aarch64.c (generic_branch_cost): Copy
21128 cortexa57_branch_cost.
21130 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
21132 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
21134 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21137 * doc/md.texi (Constraints): Document wA constraint.
21138 * config/rs6000/constraints.md (wA): New.
21139 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
21140 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
21141 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
21142 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
21144 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
21147 * gimplify.c (is_oacc_declared): New function.
21148 (oacc_default_clause): Use it to set default flags for acc declared
21149 variables inside parallel regions.
21150 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
21151 declared variables.
21152 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
21153 declare attribute to any decl as necessary.
21155 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
21158 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
21159 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
21160 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
21161 (arm_arch_lpae): This.
21162 * config/arm/arm.c (arm_arch7ve): Rename into ...
21163 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
21164 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
21167 2017-03-22 Martin Liska <mliska@suse.cz>
21170 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
21171 error message instead of an ICE.
21173 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21175 * doc/extend.texi (6.11 Additional Floating Types): Revise.
21177 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
21179 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
21181 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
21184 2017-03-21 Martin Sebor <msebor@redhat.com>
21186 * doc/extend.texi: Use "cannot" instead of "can't."
21187 * doc/hostconfig.texi: Same.
21188 * doc/install.texi: Same.
21189 * doc/invoke.texi: Same.
21190 * doc/loop.texi: Same.
21191 * doc/md.texi: Same.
21192 * doc/objc.texi: Same.
21193 * doc/rtl.texi: Same.
21194 * doc/tm.texi: Same.
21195 * doc/tm.texi.in: Same.
21196 * doc/trouble.texi: Same.
21198 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
21201 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
21202 (collect_checksum_attributes): Set it.
21203 (die_checksum_ordered): Use it.
21205 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21207 PR tree-optimization/79908
21208 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
21209 change: For a VA_ARG whose LHS has been cast away, use
21210 force_gimple_operand to construct the side effects.
21212 2017-03-21 David Malcolm <dmalcolm@redhat.com>
21214 PR translation/80001
21215 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
21216 more amenable to translation.
21217 (oacc_loop_auto_partitions): Likewise.
21219 2017-03-21 Marek Polacek <polacek@redhat.com>
21220 Martin Sebor <msebor@redhat.com>
21222 PR tree-optimization/80109
21223 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
21224 on INTEGRAL_TYPE_P.
21226 2017-03-21 Jakub Jelinek <jakub@redhat.com>
21227 Segher Boessenkool <segher@kernel.crashing.org>
21230 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
21231 check reg_used_between_p between insn and one of succ or succ2
21232 depending on if succ is artificial insn not inserted into insn
21235 2017-03-21 Martin Liska <mliska@suse.cz>
21237 PR gcov-profile/80081
21238 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
21239 * doc/gcc.texi: Include gcov-dump stuff.
21240 * doc/gcov-dump.texi: New file.
21242 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
21244 PR rtl-optimization/79150
21245 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
21246 conditional jump, if the jump is the last insn of the loop.
21248 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21249 Richard Biener <rguenth@suse.de>
21251 PR tree-optimization/79908
21252 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
21253 been cast away, use force_gimple_operand to construct the side
21256 2017-03-21 Martin Liska <mliska@suse.cz>
21258 PR libfortran/79956
21259 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
21262 2017-03-21 Brad Spengler <spender@grsecurity.net>
21265 * plugin.c (htab_hash_plugin): New function.
21266 (add_new_plugin): Use it and adjust.
21267 (parse_plugin_arg_opt): Adjust.
21268 (init_one_plugin): Likewise.
21270 2017-03-21 Richard Biener <rguenther@suse.de>
21272 PR tree-optimization/80032
21273 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
21274 if set force the cleanup to happen unconditionally.
21275 (gimplify_target_expr): Push inserted clobbers with force_uncond
21276 to avoid them being removed by control-dependent DCE.
21278 2017-03-21 Richard Biener <rguenther@suse.de>
21280 PR tree-optimization/80122
21281 * tree-inline.c (copy_bb): Do not expans va-arg packs or
21282 va_arg_pack_len when the inlined call stmt requires pack
21284 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
21286 2017-03-21 Jakub Jelinek <jakub@redhat.com>
21289 * tsan.c (instrument_builtin_call): If the memory model argument
21290 is not a constant, assume it is valid.
21293 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
21296 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
21298 PR rtl-optimization/79910
21299 * combine.c (can_combine_p): Do not allow combining an I0 or I1
21300 if its dest is used by an insn before I2 (other than the combined
21301 insns themselves, which are properly handled already).
21303 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
21306 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
21308 * combine.c (record_used_regs): New static function.
21309 (try_combine): Handle situations where there is an additional
21310 instruction between I2 and I3 which needs to have a LOG_LINK
21314 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
21316 * combine.c (try_combine): Delete redundant i1 test. Call
21317 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
21319 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21322 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
21323 alternatives 13/14.
21325 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21327 PR tree-optimization/80054
21328 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
21329 the optimization if a PHI or any of its arguments is not dominated
21330 by the candidate's basis. Use gphi* rather than gimple* as
21332 (replace_profitable_candidates): Clean up a gimple* variable that
21333 should be a gphi* variable.
21335 2017-03-20 Martin Sebor <msebor@redhat.com>
21338 * doc/extend.texi (attribute constructor): Document present limitation.
21340 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
21343 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
21344 __POWER9_VECTOR__ #ifdef control, change template definition to
21345 use Power9-specific built-in function.
21346 (vec_any_eq): Likewise.
21347 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
21348 to control outcomes from this test.
21349 (vector_ae_<mode>p): For VEC_F modes, likewise.
21351 2017-03-20 Ian Lance Taylor <iant@google.com>
21353 * config/i386/i386.c (ix86_function_regparm): Save an extra
21354 register for -fsplit-stack with DECL_STATIC_CHAIN.
21356 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
21359 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
21360 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
21362 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
21364 * config/riscv/riscv.c (riscv_print_operand): Use "fence
21366 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
21369 2017-03-20 Marek Polacek <polacek@redhat.com>
21372 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
21374 2017-03-20 Richard Biener <rguenther@suse.de>
21376 PR tree-optimization/80113
21377 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
21378 allocate extra SSA name for PHI def.
21379 (add_close_phis_to_outer_loops): Likewise.
21380 (add_close_phis_to_merge_points): Likewise.
21381 (copy_loop_close_phi_args): Likewise.
21382 (copy_cond_phi_nodes): Likewise.
21384 2017-03-20 Martin Liska <mliska@suse.cz>
21386 PR middle-end/79753
21387 * tree-chkp.c (chkp_build_returned_bound): Do not build
21388 returned bounds for a LHS that's not a BOUNDED_P type.
21390 2017-03-20 Martin Liska <mliska@suse.cz>
21394 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
21395 COMPLEX_CST and VECTOR_CST.
21397 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21400 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
21401 target operand. A new splitter adds the clobber statement in case
21402 the target operand is dead anyway.
21404 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
21406 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
21407 to age-old versions of binutils and glibc.
21409 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
21411 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
21413 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
21415 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
21417 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
21419 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
21420 requirement for binutils 2.13.
21422 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
21424 * combine.c (try_combine): Delete redundant i1 test. Call
21425 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
21427 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
21429 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
21430 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
21432 <riscv64-*-elf>: Re-arrange section
21433 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
21434 <riscv32-*-linux>: Likewise.
21435 <riscv64-*-elf>: Likewise
21436 <riscv64-*-linux>: Likewise.
21438 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
21441 * aarch64.opt(verbose-cost-dump): Fix typo.
21443 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
21446 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
21447 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
21449 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
21451 * reload.c (find_reloads): When reloading a nonoffsettable address,
21452 use RELOAD_OTHER for it and its address reloads.
21454 PR rtl-optimization/79910
21455 * combine.c (record_used_regs): New static function.
21456 (try_combine): Handle situations where there is an additional
21457 instruction between I2 and I3 which needs to have a LOG_LINK
21460 2017-03-17 Jeff Law <law@redhat.com>
21462 PR tree-optimization/71437
21463 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
21464 conditional in the hash table first.
21465 (vrp_dom_walker::before_dom_children): Extract condition from
21466 ASSERT_EXPR. Record condition, its inverion and any implied
21467 conditions as well.
21469 2017-03-17 Marek Polacek <polacek@redhat.com>
21470 Markus Trippelsdorf <markus@trippelsdorf.de>
21472 PR tree-optimization/80079
21473 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
21476 2017-03-17 Richard Biener <rguenther@suse.de>
21478 PR middle-end/80075
21479 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
21480 Properly verify the LHS before the RHS possibly claims to be
21482 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
21485 2017-03-17 Martin Jambor <mjambor@suse.cz>
21487 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
21488 (List of -O2 options): Likewise.
21489 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
21492 2017-03-17 Tom de Vries <tom@codesourcery.com>
21494 * gcov-dump.c (print_usage): Print bug_report_url.
21496 2017-03-17 Richard Biener <rguenther@suse.de>
21498 PR middle-end/80050
21499 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
21500 (parser::peek): Likewise.
21502 2017-03-17 Richard Biener <rguenther@suse.de>
21504 PR tree-optimization/80048
21505 * sese.c (free_sese_info): Properly release rename_map and
21506 copied_bb_map elements.
21508 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
21510 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
21511 Add linked-list forward and backlinks. Insert on
21512 construction, remove on destruction.
21513 (class pass_store_merging): Add m_stores_head field.
21514 (pass_store_merging::terminate_and_process_all_chains):
21515 Iterate over m_stores_head list.
21516 (pass_store_merging::terminate_all_aliasing_chains):
21518 (pass_store_merging::execute): Check for debug stmts first.
21519 Push new chains onto the m_stores_head stack.
21521 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
21524 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
21525 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
21526 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
21528 2017-03-16 Jeff Law <law@redhat.com>
21530 PR tree-optimization/71437
21531 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
21532 member function. Implementation moved into after_dom_children
21533 member function and into the threader's thread_outgoing_edges
21535 (dom_opt_dom_walker::after_dom_children): Simplify by moving
21536 some code into new thread_outgoing_edges.
21537 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
21538 definition. Simplify marker handling (do it here). Assume we always
21539 have the available expression and the const/copies tables.
21540 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
21542 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
21543 * tree-vrp.c (equiv_stack): No longer file scoped.
21544 (vrp_dom_walker): New class.
21545 (vrp_dom_walker::before_dom_children): New member function.
21546 (vrp_dom_walker::after_dom_children): Likewise.
21547 (identify_jump_threads): Setup domwalker. Use it rather than
21548 walking edges in a random order by hand. Simplify setup/finalization.
21549 (finalize_jump_threads): Remove.
21550 (vrp_finalize): Do not call identify_jump_threads here.
21551 (execute_vrp): Do it here instead and call thread_through_all_blocks
21554 PR tree-optimization/71437
21555 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
21557 (simplify_stmt_for_jump_threading): Add basic_block argument. All
21559 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
21560 (dom_opt_dom_walker::thread_across_edge): Remove
21561 handle_dominating_asserts argument. All callers changed.
21562 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
21563 changes. Remove calls to lhs_of_dominating_assert. Other
21564 uses of handle_dominating_asserts turn into unconditional code
21565 (simplify_control_stmt_condition_1): Likewise.
21566 (simplify_control_stmt_condition): Likewise.
21567 (thread_through_normal_block, thread_across_edge): Likewise.
21568 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
21569 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
21570 object if it is not an SSA_NAME.
21571 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
21572 before calling into the VRP specific simplifiers.
21573 (identify_jump_threads): Remove handle_dominating_asserts
21576 2017-03-16 Jakub Jelinek <jakub@redhat.com>
21579 * tree-diagnostic.c (default_tree_printer): No longer static.
21580 * tree-diagnostic.h (default_tree_printer): New prototype.
21582 2017-03-16 Tamar Christina <tamar.christina@arm.com>
21584 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
21585 Change ins into fmov.
21587 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21589 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
21590 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
21591 Use h_con constraint for operand 1.
21592 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
21593 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
21595 2017-03-15 Jeff Law <law@redhat.com>
21597 PR tree-optimization/71437
21598 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
21599 (record_temporary_equivalences): Use it.
21601 PR tree-optimization/71437
21602 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
21603 tree-ssa-scopedtables.
21604 (lookup_avail_expr, build_and_record_new_cond): Likewise.
21605 (record_conditions, record_cond, vuse_eq): Likewise.
21606 (record_edge_info): Adjust to API tweak of record_conditions.
21607 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
21608 (record_temporary_equivalences, optimize_stmt): Likewise.
21609 (eliminate_redundant_computations): Likewise.
21610 (record_equivalences_from_stmt): Likewise.
21611 * tree-ssa-scopedtables.c: Include options.h and params.h.
21612 (vuse_eq): New function, moved from tree-ssa-dom.c
21613 (build_and_record_new_cond): Likewise.
21614 (record_conditions): Likewise. Accept vector of conditions rather
21615 than edge_equivalence structure for first argument.
21616 for the first argument.
21617 (avail_exprs_stack::lookup_avail_expr): New member function, moved
21618 from tree-ssa-dom.c.
21619 (avail_exprs_stack::record_cond): Likewise.
21620 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
21621 from tree-ssa-dom.c.
21622 (avail_exprs_stack): Add new member functions lookup_avail_expr
21624 (record_conditions): Declare.
21626 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
21629 * lra-constraints.c (process_alt_operands): Increase reject for
21630 reloading an input/output operand.
21632 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
21635 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
21636 insns to convert from signed/unsigned char/short to IEEE 128-bit
21638 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
21640 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
21643 * config/i386/i386.c (ix86_vector_duplicate_value): Create
21644 subreg of inner mode for values already in registers.
21646 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
21648 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
21649 iteration reg is used after the loop.
21651 2017-03-14 Martin Sebor <msebor@redhat.com>
21653 PR tree-optimization/79800
21654 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
21655 precision in negative-positive range.
21656 (format_floating): Call non-const overload with adjusted precision.
21658 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
21661 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
21664 2017-03-14 Martin Sebor <msebor@redhat.com>
21666 PR middle-end/80020
21667 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
21668 * builtins.def (aligned_alloc): Use it.
21671 * Makefile.in (GTFILES): Add calls.c.
21672 * calls.c: Include "gt-calls.h".
21674 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
21676 PR rtl-optimization/79728
21677 * regs.h (struct target_regs): New field
21678 x_contains_allocatable_regs_of_mode.
21679 (contains_allocatable_regs_of_mode): New macro.
21680 * reginfo.c (init_reg_sets_1): Initialize it, and change
21681 contains_reg_of_mode so it includes global regs as well.
21682 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
21683 rather than contains_regs_of_mode.
21685 2017-03-14 Martin Liska <mliska@suse.cz>
21687 * doc/invoke.texi: Document options that can't be combined with
21688 -fcheck-pointer-bounds.
21690 2017-03-14 Martin Liska <mliska@suse.cz>
21692 PR middle-end/79831
21693 * doc/invoke.texi (-Wchkp): Document the option.
21695 2017-03-14 Martin Liska <mliska@suse.cz>
21697 * Makefile.in: Install gcov-dump.
21699 2017-03-14 Martin Liska <mliska@suse.cz>
21701 * multiple_target.c (expand_target_clones): Bail out for
21702 an invalid attribute.
21704 2017-03-14 Richard Biener <rguenther@suse.de>
21706 * alias.c (struct alias_set_entry): Pack properly.
21707 * cfgloop.h (struct loop): Likewise.
21708 * cse.c (struct set): Likewise.
21709 * ipa-utils.c (struct searchc_env): Likewise.
21710 * loop-invariant.c (struct invariant): Likewise.
21711 * lra-remat.c (struct cand): Likewise.
21712 * recog.c (struct change_t): Likewise.
21713 * rtl.h (struct address_info): Likewise.
21714 * symbol-summary.h (function_summary): Likewise.
21715 * tree-loop-distribution.c (struct partition): Likewise.
21716 * tree-object-size.c (struct object_size_info): Likewise.
21717 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
21718 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
21719 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
21720 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
21721 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
21722 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
21723 (struct _stmt_vec_info): Likewise.
21725 2017-03-14 Martin Liska <mliska@suse.cz>
21728 * multiple_target.c (create_dispatcher_calls): Check that
21729 a target can create a function dispatcher.
21731 2017-03-14 Martin Liska <mliska@suse.cz>
21734 * multiple_target.c (expand_target_clones): Drop local.local
21735 flag for default implementation.
21737 2017-03-14 Richard Biener <rguenther@suse.de>
21739 PR tree-optimization/80030
21740 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
21742 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
21744 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
21745 gcc_fallthrough() instead of __attribute__((fallthrough));
21747 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
21749 * doc/gcc.texi: Remove "up" link to (DIR).
21750 * doc/gccint.texi: Ditto.
21752 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
21754 * doc/install.texi (Specific) <avr>: Remove reference to
21757 2017-03-13 Jeff Law <law@redhat.com>
21759 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
21760 attribute rather than comments.
21762 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
21763 match_scratch operand is highest.
21765 2017-03-13 Martin Liska <mliska@suse.cz>
21767 PR middle-end/78339
21768 * ipa-pure-const.c (warn_function_noreturn): If the declarations
21769 is a CHKP clone, use original declaration.
21771 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21773 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
21774 (arc_conditional_register_usage): Use a different allocation order
21775 when optimizing for size.
21776 * common/config/arc/arc-common.c (arc_option_optimization_table):
21777 Section anchors default on when optimizing for size.
21779 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21781 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
21783 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21785 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
21786 * config/arc/arc.md (cpu_facility): Add cd variant.
21787 (*movqi_insn): Add code density variant.
21788 (*movhi_insn): Likewise.
21789 (*movqi_insn): Likewise.
21790 (*addsi3_mixed): Likewise.
21791 (subsi3_insn): Likewise.
21793 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21795 * config/arc/arc.md (movsi_cond_exec): Update constraint.
21797 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21799 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
21800 expressions with MINUS and UNARY ops.
21802 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21805 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
21807 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
21808 between vec_select and vector argument.
21809 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
21810 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
21811 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
21812 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
21813 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
21814 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
21816 2017-03-13 Richard Biener <rguenther@suse.de>
21819 * params.def (vect-max-peeling-for-alignment): Fix typo.
21821 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21823 * doc/install.texi (Specific) <mips-*-*>: Remove description of
21824 issue that only occurred with binutils below 2.18.
21826 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21828 * doc/install.texi (Specific) <cris-axis-elf>: No longer
21829 refer to binutils 2.11/2.12 minimum.
21831 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21833 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
21834 ftp.kernel.org and simplify binutils requirement.
21836 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
21838 * doc/invoke.texi (Warning Options): Fix spelling of link-time
21840 (Optimize Options): Ditto. Also remove redundancy.
21842 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21844 PR translation/79848
21845 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
21847 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
21848 to G_ to avoid double translation.
21850 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21852 PR translation/79923
21853 * auto-profile.c (get_combined_location): Convert leading
21854 character of diagnostics to lower case and remove trailing period.
21855 (read_profile): Likewise for various diagnostics.
21856 * config/arm/arm.c (arm_option_override): Remove trailing period
21857 from various diagnostics.
21858 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
21859 (msp430_expand_delay_cycles): Likewise.
21861 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21864 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
21865 full command-line argument, rather than just "str".
21866 (aarch64_validate_march): Likewise.
21867 (aarch64_validate_mtune): Likewise.
21869 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
21871 PR rtl-optimization/78911
21872 * lra-assigns.c (must_not_spill_p): New function.
21873 (spill_for): Use it.
21875 2017-03-10 Jakub Jelinek <jakub@redhat.com>
21877 PR tree-optimization/79981
21878 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
21879 ATOMIC_COMPARE_EXCHANGE ifn result.
21880 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
21881 IFN_ATOMIC_COMPARE_EXCHANGE.
21883 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21886 * opts.c (parse_sanitizer_options): Add missing question mark to
21887 "did you mean" message.
21889 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21891 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
21893 (VMULEUH_UNS): Likewise.
21894 (VMULOUB_UNS): Likewise.
21895 (VMULOUH_UNS): Likewise.
21896 * config/rs6000/rs6000.c (builtin_function_type): Remove
21897 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
21899 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21902 * read-rtl-function.c (function_reader::read_rtx_operand): Update
21903 x with result of extra_parsing_for_operand_code_0.
21904 (function_reader::extra_parsing_for_operand_code_0): Convert
21905 return type from void to rtx, returning x. When reading
21906 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
21907 larger size containing struct block_symbol.
21909 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
21911 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
21912 -mfloat128-hardware without -m64.
21914 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
21917 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
21918 entries to the case statement that marks unsigned arguments to
21919 overloaded functions.
21921 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21923 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
21924 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
21926 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
21929 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
21930 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
21932 2017-03-10 Martin Liska <mliska@suse.cz>
21936 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
21937 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
21940 2017-03-10 Olivier Hainque <hainque@adacore.com>
21942 * tree-switch-conversion (array_value_type): Start by resetting
21943 candidate type to it's main variant.
21945 2017-03-10 Jakub Jelinek <jakub@redhat.com>
21947 PR rtl-optimization/79909
21948 * combine.c (try_combine): Use simplify_replace_rtx on individual
21949 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
21950 of the whole CALL_INSN_FUNCTION_USAGE.
21952 PR tree-optimization/79972
21953 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
21954 get_range_info on SSA_NAMEs. Formatting fixes.
21956 2017-03-10 Richard Biener <rguenther@suse.de>
21957 Jakub Jelinek <jakub@redhat.com>
21959 PR tree-optimization/77975
21960 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
21961 edge to be constant.
21962 (get_val_for): For constant x return it. Formatting fix.
21963 (loop_niter_by_eval): Avoid pointless looping if the next iteration
21964 would use the same bases as the current one.
21966 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21968 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
21969 instead of vec_select for V1TImode.
21970 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
21972 (VSX_LE_128): Add V1TI to this mode iterator.
21973 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
21974 (*vsx_le_perm_store_<mode>): Likewise.
21975 (pre-reload splitter for VSX stores): Likewise.
21976 (post-reload splitter for VSX stores): Likewise.
21977 (*vsx_xxpermdi2_le_<mode>): Likewise.
21978 (*vsx_lxvd2x2_le_<mode>): Likewise.
21979 (*vsx_stxvd2x2_le_<mode>): Likewise.
21981 2017-03-09 Michael Eager <eager@eagercon.com>
21983 Correct failures with --enable-checking=yes,rtl.
21985 * config/microblaze/microblaze.c (microblaze_expand_shift):
21986 Replace GET_CODE test with CONST_INT_P and INTVAL test with
21987 test for const0_rtx.
21988 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
21989 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
21991 2017-03-09 Richard Biener <rguenther@suse.de>
21993 PR tree-optimization/79977
21994 * graphite-scop-detection.c (scop_detection::merge_sese):
21995 Handle the case of extra exits to blocks dominating the entry.
21997 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
21999 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
22002 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
22004 PR rtl-optimization/79949
22005 * lra-constraints.c (process_alt_operands): Check memory when
22006 trying to predict a cycle. Print about the overall increase.
22008 2017-03-09 Richard Biener <rguenther@suse.de>
22010 PR middle-end/79971
22011 * gimple-expr.c (useless_type_conversion_p): Preserve
22012 TYPE_SATURATING for fixed-point types.
22014 2017-03-09 Richard Biener <rguenther@suse.de>
22017 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
22018 alignment of BLKmode params.
22020 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22023 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
22024 (VALL_NO_V2Q): Likewise.
22026 * config/aarch64/aarch64-simd.md
22027 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
22029 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
22030 VALL_NO_V2Q mode iterator.
22031 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
22033 2017-03-09 Martin Liska <mliska@suse.cz>
22035 PR tree-optimization/79631
22036 * tree-chkp-opt.c (chkp_is_constant_addr): Call
22037 tree_int_cst_sign_bit just for INTEGER constants.
22039 2017-03-09 Martin Liska <mliska@suse.cz>
22043 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
22046 2017-03-09 Marek Polacek <polacek@redhat.com>
22049 * tree.c (inchash::add_expr): Handle TREE_VEC.
22051 2017-03-09 Martin Liska <mliska@suse.cz>
22054 (chkp_narrow_size_and_offset): New function.
22055 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
22056 (void chkp_parse_bit_field_ref): New function.
22057 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
22058 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
22060 2017-03-09 Martin Liska <mliska@suse.cz>
22063 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
22064 (chkp_find_bounds_1): Remove gcc_unreachable.
22066 2017-03-09 Jakub Jelinek <jakub@redhat.com>
22069 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
22070 BUILT_IN_SYNC*, determine the access type from the size suffix and
22071 always build a MEM_REF with that type. Handle forgotten
22072 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
22075 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
22076 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
22077 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
22078 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
22079 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
22080 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
22081 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
22082 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
22083 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
22084 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
22085 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
22086 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
22087 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
22088 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
22089 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
22090 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
22091 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
22092 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
22093 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
22094 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
22095 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
22096 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
22097 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
22098 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
22099 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
22100 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
22101 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
22102 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
22103 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
22104 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
22105 definitions outside of __OPTIMIZE__ guarded section.
22108 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
22109 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
22110 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
22111 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
22114 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22116 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
22117 ("vfenez<mode>"): Add missing constraints.
22119 2017-03-08 Martin Sebor <msebor@redhat.com>
22122 * config/nds32/nds32.c (nds32_option_override):
22123 Fix misspelled diagnostic.
22125 2017-03-08 Jakub Jelinek <jakub@redhat.com>
22128 * gimplify.c (gimplify_omp_for): Replace index var in outer
22129 taskloop statement with an artificial variable and add
22130 OMP_CLAUSE_PRIVATE clause for it.
22132 2017-03-08 Richard Biener <rguenther@suse.de>
22134 PR tree-optimization/79955
22135 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
22136 for accesses that are completely outside of the variable.
22138 2017-03-08 Andrew Haley <aph@redhat.com>
22140 PR tree-optimization/79943
22141 * tree-ssa-loop-split.c (compute_new_first_bound): When
22142 calculating the new upper bound, (END-BEG) should be added, not
22145 2017-03-08 Jakub Jelinek <jakub@redhat.com>
22147 * config/avr/avr.md (setmemhi): Make sure match_dup
22148 operand number comes before match_scratch.
22150 2017-03-08 Richard Biener <rguenther@suse.de>
22152 PR tree-optimization/79920
22153 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
22154 with ncopies == 1 to ...
22155 (vect_transform_slp_perm_load): ... here. Properly compute
22156 all element loads by iterating VF times over the group. Do
22157 not handle ncopies (computed in a broken way) in
22158 vect_create_mask_and_perm.
22160 2017-03-08 Jakub Jelinek <jakub@redhat.com>
22163 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
22164 is a uniform vector, use uniform_vector_p return value instead of
22165 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
22167 2017-03-07 Marek Polacek <polacek@redhat.com>
22169 PR middle-end/79809
22170 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
22171 (alloca_call_type): Likewise.
22173 2017-03-07 Martin Liska <mliska@suse.cz>
22175 * gcov.c (process_args): Put comment to correct location.
22177 2017-03-07 Martin Liska <mliska@suse.cz>
22179 PR middle-end/68270
22180 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
22181 Use array_at_struct_end_p instead of DECL_CHAIN (field).
22182 (chkp_narrow_bounds_for_field): Likewise.
22183 (chkp_parse_array_and_component_ref): Pass one more argument to
22186 2017-03-07 Richard Biener <rguenther@suse.de>
22188 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
22191 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
22193 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
22194 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
22196 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22199 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
22200 to end of description.
22201 (PARAM_MAX_STORES_TO_MERGE): Likewise.
22203 2017-03-07 Jakub Jelinek <jakub@redhat.com>
22205 PR rtl-optimization/79901
22206 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
22208 (*avx512f_<code><mode>3<mask_name>): ... this.
22209 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
22210 iterator instead of VI8_AVX2_AVX512BW.
22212 PR rtl-optimization/79901
22213 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
22214 min/max expander, expand it using expand_vec_cond_expr.
22217 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
22220 2017-03-06 Jakub Jelinek <jakub@redhat.com>
22223 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
22224 to void * for PCH reasons.
22225 * dwarf2out.c (output_loc_operands, output_die): Cast
22226 v.val_vec.array to unsigned char *.
22228 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
22231 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
22234 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
22236 PR rtl-optimization/79571
22237 * lra-constraints.c (process_alt_operands): Calculate static
22238 reject and subtract it from overall when only addresses will be
22241 2017-03-06 Julia Koval <julia.koval@intel.com>
22244 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
22245 incoming stack boundary to 128 for 64-bit targets.
22247 2017-03-06 Richard Biener <rguenther@suse.de>
22249 PR tree-optimization/79894
22250 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
22251 to NULL after folding it.
22253 2017-03-06 Richard Biener <rguenther@suse.de>
22255 PR tree-optimization/79824
22256 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
22257 check disabling peeling for gaps.
22259 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
22261 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
22262 attributes): Document gettimeofday.
22264 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
22266 * config/s390/s390.c (s390_option_override_internal): Set
22267 PARAM_MIN_VECT_LOOP_BOUND
22269 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
22271 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
22272 * config/s390/s390.md: Likewise.
22274 2017-03-06 Jakub Jelinek <jakub@redhat.com>
22277 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
22278 (<avx2_avx512>_perm<mode>): Rename to ...
22279 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
22281 (<avx512>_perm<mode>_mask): Rename to ...
22282 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
22284 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
22285 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
22286 instead of VI8F_256_512.
22287 (avx512f_perm<mode>): New define_expand.
22288 (avx512f_perm<mode>_mask): Likewise.
22289 (avx512f_perm<mode>_1<mask_name>): New define_insn.
22290 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
22292 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22294 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
22295 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
22297 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
22299 2017-03-06 Martin Liska <mliska@suse.cz>
22302 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
22303 when having a SSA NAME w/o VAR_DECL assigned to it.
22305 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22307 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
22308 msa_dpsub_<su>_d): Fix MODE for vec_select.
22310 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22312 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
22314 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
22316 2017-03-06 Richard Biener <rguenther@suse.de>
22318 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
22319 * plugin.c (register_plugin_info): Likewise.
22320 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
22322 2017-03-05 Jakub Jelinek <jakub@redhat.com>
22324 * config/i386/sse.md (sse_storehps, sse_storelps,
22325 avx_<castmode><avxsizesuffix>_<castmode>,
22326 avx512f_<castmode><avxsizesuffix>_<castmode>,
22327 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
22328 in condition that at least one operand is not a MEM.
22330 2017-03-03 Jakub Jelinek <jakub@redhat.com>
22332 PR middle-end/79805
22333 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
22334 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
22336 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
22337 gimple_call_nothrow_p flag based on whether original builtin can throw.
22338 If it can, emit following stmts on the fallthrough edge.
22339 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
22340 don't create new bb if inserting just debug stmts on the edge, try to
22341 insert them on the fallthru bb or just reset debug stmts.
22343 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
22346 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
22347 restore recog_data (including the operand rtxes inside it) around
22348 the call to get_insn_template.
22350 2017-03-03 Martin Sebor <msebor@redhat.com>
22352 PR tree-optimization/79699
22353 * context.c (context::~context): Free MPFR caches to avoid
22354 a memory leak on program exit.
22356 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22358 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
22359 Use wide_int::ulow () instead of .elt (0).
22361 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
22363 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
22364 (*pushxf): Limit oF constraint to 32bit targets and add oC
22365 constraint for 64bit targets.
22366 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
22367 (*pushdf): Change rmF constraint to rmC.
22369 2017-03-03 Martin Liska <mliska@suse.cz>
22371 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
22372 Remove unused variable.
22374 2017-03-03 Jakub Jelinek <jakub@redhat.com>
22377 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
22378 is a memory operand, increase num_memory.
22379 (ix86_expand_args_builtin): Likewise.
22381 2017-03-03 Jan Hubicka <jh@suse.cz>
22384 * ipa-devirt.c (maybe_record_node): Properly handle
22385 __cxa_pure_virtual visibility.
22387 2017-03-03 Martin Liska <mliska@suse.cz>
22389 PR tree-optimization/79803
22390 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
22392 (pass_loop_prefetch::execute): Disabled optimization if an
22393 assumption about L1 cache size is not met.
22395 2017-03-03 Martin Liska <mliska@suse.cz>
22397 PR rtl-optimization/79574
22398 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
22399 (hash_scan_set): Likewise.
22400 (dump_hash_table): Likewise.
22401 (hoist_code): Likewise.
22403 2017-03-03 Richard Biener <rguenther@suse.de>
22405 * fixed-value.c (fixed_from_string): Restore use of elt (1)
22406 in place of uhigh ().
22407 (fixed_convert_from_real): Likewise.
22409 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
22412 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
22414 2017-03-03 Richard Biener <rguenther@suse.de>
22416 PR middle-end/79818
22417 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
22418 TYPE_OVERFLOW_UNDEFINED check.
22420 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22422 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
22424 (vector_ae_<mode>_p): Likewise.
22425 (vector_nez_<mode>_p): Likewise.
22426 (vector_ne_v2di_p): Likewise.
22427 (vector_ae_v2di_p): Likewise.
22428 (vector_ne_<mode>_p): Likewise.
22429 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
22431 (vsx_tsqrt<mode>2_fe): Likewise.
22433 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
22436 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
22438 2017-03-02 Jakub Jelinek <jakub@redhat.com>
22440 PR rtl-optimization/79780
22441 * cprop.c (one_cprop_pass): When second and further conditional trap
22442 in a single basic block is turned into an unconditional trap, turn it
22443 into a deleted note to avoid RTL verification failures.
22445 2017-03-02 Richard Biener <rguenther@suse.de>
22447 * fold-const.c (const_binop): Use ulow () instead of elt (0).
22449 2017-03-02 Richard Biener <rguenther@suse.de>
22451 PR tree-optimization/79345
22453 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
22454 param and abort the walk, returning -1 if it is hit.
22455 (walk_aliased_vdefs): Take a limit param and pass it on.
22456 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
22457 defaulting to 0 and return a signed int.
22458 * tree-ssa-uninit.c (struct check_defs_data): New struct.
22459 (check_defs): New helper.
22460 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
22461 about uninitialized memory.
22462 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
22463 bogus uninitialized warning.
22464 (fixed_convert_from_real): Likewise.
22466 2017-03-02 Bin Cheng <bin.cheng@arm.com>
22468 PR tree-optimization/66768
22469 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
22470 iv_use if base object can't be determined.
22472 2017-03-02 Jakub Jelinek <jakub@redhat.com>
22474 PR tree-optimization/79345
22475 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
22476 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
22477 (get_pattern_stats): Initialize it.
22478 * genemit.c (gen_expand): Verify match_scratch numbers come after
22479 match_operand/match_dup numbers.
22480 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
22481 match_scratch numbers.
22482 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
22484 * config/s390/s390.md (trunctdsd2): Likewise.
22486 2017-03-02 Richard Biener <rguenther@suse.de>
22488 * wide-int.h (wide_int_storage::operator=): Implement in terms
22491 2017-03-02 Richard Biener <rguenther@suse.de>
22493 PR tree-optimization/79777
22494 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
22495 the to insert expression to sth existing.
22497 2017-03-01 Martin Sebor <msebor@redhat.com>
22499 PR middle-end/79692
22500 * gimple-ssa-sprintf.c
22501 (directive::known_width_and_precision): New function.
22502 (format_integer): Use it.
22503 (get_mpfr_format_length): Consider the full range of precision
22504 when computing %g output with the # flag. Set the likely byte
22505 count to 3 rather than 1 when precision is indeterminate.
22506 (format_floating): Correct the lower bound of precision.
22508 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22510 * doc/invoke.texi: Document default code model for 64-bit Linux.
22512 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22515 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
22516 udiv rather than div since input pattern is unsigned.
22518 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
22520 * config/i386/i386.c (print_reg): Warn for values of
22521 unsupported size in integer register.
22523 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
22526 * config/rs6000/predicates.md (current_file_function_operand): Do
22527 not allow self calls to be local if the function is replaceable.
22529 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
22532 * config/rs6000/altivec.h (vec_ctz and others): Change the
22533 preprocessor macro that controls conditional compilation from
22534 _ARCH_PWR9 to __POWER9_VECTOR__.
22535 (vec_all_ne): Change parameterization of __altivec_scalar_pred
22536 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
22537 control (instead of _ARCH_PWR9 control) so that template
22538 definition uses power9-specific function.
22539 (vec_any_eq): Likewise.
22540 (vec_all_ne): Change macro definition to use a power9-specific
22541 expansion under #ifdef __POWER9_VECTOR__ control (instead of
22542 _ARCH_PWR9 control).
22543 (vec_any_eq) Likewise.
22544 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
22545 expansion for CMPNEF to remove support for xvcmpnesp instruction.
22546 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
22547 support for xvcmpnedp instruction.
22548 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
22549 macro expansion so that Power9 implementation of vec_all_ne does
22550 not use the AltiVec predicate framework.
22551 (VCMPNEH_P): Likewise.
22552 (VCMPNEW_P): Likewise.
22553 (VCMPNED_P): Likewise.
22554 (VCMPNEFP_P): Likewise.
22555 (VCMPNEDP_P): Likewise.
22556 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
22557 implementation of vec_any_eq to not use AltiVec predicate
22559 (VCMPAEH_P): Likewise.
22560 (VCMPAEW_P): Likewise.
22561 (VCMPAED_P): Likewise.
22562 (VCMPAEFP_P): Likewise.
22563 (VCMPAEDP_P): Likewise.
22564 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
22565 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
22566 not use the AltiVec predicate framework.
22567 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
22568 of vec_any_eq to not use AltiVec predicate framework.
22569 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
22570 support for predefined __POWER9_VECTOR__ macro to indicate that
22571 Power9 instruction selection is enabled.
22572 (altivec_overloaded_builtins): Remove extraneous
22573 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
22574 function argument types RS6000_BTI_bool_V16QI and
22575 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
22576 entry for overloaded function argument types RS6000_BTI_bool_V4SI
22577 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
22578 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
22579 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
22580 Power9 for implementations of vec_cmpne. Change the signature for
22581 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
22582 (representing vec_all_ne) to remove the previously described first
22583 argument of type RS6000_BTI_INTSI, as this was an artifact of
22584 reliance on the AltiVec predicate framework, which is no longer
22585 used in the implementation of these functions. Add
22586 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
22587 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
22588 since, unlike the AltiVec predicate framework implementation, we
22589 do not share function descriptors between vec_alle and vec_anyeq.
22590 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
22591 set of modes that receive special treatment even when
22592 TARGET_P9_VECTOR is true. The special treatment emits code that
22593 does not depend on Power9 instructions.
22594 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
22595 define_expand to not rely on AltiVec predicate framework.
22596 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
22598 (vector_ne_v2di_p): Change this define_expand to not rely on
22599 AltiVec predicate framework.
22600 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
22602 (vector_ne_<mode>_p): Change this define_expand to not rely on
22603 AltiVec predicate framework.
22604 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
22606 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
22607 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
22608 define_insn pattern.
22609 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
22610 define_insn pattern because the xvcmpne<VSs>. instruction is not
22612 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
22613 instruction is not supported.
22615 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22617 * config/nvptx/nvptx.c: Include intl.h.
22619 2017-03-01 Martin Jambor <mjambor@suse.cz>
22622 * ipa-prop.h (ipa_bits): Removed field known.
22623 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
22624 to pointers. Adjusted their comments to warn about their sharing.
22625 (ipcp_transformation_summary): Change bits to a vector of pointers.
22626 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
22627 (ipa_get_ipa_bits_for_value): Declare.
22628 * tree-vrp.h (value_range): Mark as GTY((for_user)).
22629 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
22630 (ipa_bits_hash_table): Likewise.
22631 (ipa_vr_ggc_hash_traits): Likewise.
22632 (ipa_vr_hash_table): Likewise.
22633 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
22634 being pointers and vr_known being removed.
22635 (ipa_set_jf_unknown): Likewise.
22636 (ipa_get_ipa_bits_for_value): New function.
22637 (ipa_set_jfunc_bits): Likewise.
22638 (ipa_get_value_range): New overloaded functions.
22639 (ipa_set_jfunc_vr): Likewise.
22640 (ipa_compute_jump_functions_for_edge): Use the above functions to
22641 construct bits and vr parts of jump functions.
22642 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
22643 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
22645 (ipcp_grow_transformations_if_necessary): Also allocate
22646 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
22648 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
22649 them. Fix too long lines.
22650 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
22651 vr_known being removed.
22652 (ipa_read_jump_function): Use new setter functions to construct bits
22653 and vr parts of jump functions or set them to NULL.
22654 (write_ipcp_transformation_info): Adjust for bits being pointers.
22655 (read_ipcp_transformation_info): Likewise.
22656 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
22658 Include gt-ipa-prop.h.
22659 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
22661 (ipcp_store_bits_results): Likewise.
22662 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
22663 Do not write to existing jump functions but use a temporary instead.
22665 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22668 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
22669 attempt to use its first operand as BIT_FIELD_REF base.
22671 2017-03-01 Richard Biener <rguenther@suse.de>
22673 PR middle-end/79721
22674 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
22675 interpolating formula in wrapping arithmetic.
22676 (chrec_apply): Convert chrec_evaluate return value to wanted type.
22678 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22680 PR tree-optimization/79734
22681 * tree-vect-generic.c (expand_vector_condition): Optimize
22682 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
22683 Handle VEC_COND_EXPR where comparison has different inner width from
22684 type's inner width.
22686 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
22688 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
22689 markup, and similar issues. Remove @opindex entries for things
22690 that aren't options. Add missing -mmpy-option entries.
22692 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22694 PR tree-optimization/79737
22695 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
22696 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
22697 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
22698 instead of byte_size. Formatting fix.
22699 (shift_bytes_in_array_right): Formatting fix.
22701 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
22704 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
22705 condition on optimize for the leaf function test.
22707 2017-02-28 Martin Liska <mliska@suse.cz>
22710 * read-rtl-function.c (function_reader::handle_unknown_directive):
22711 Bail out when one uses -flto.
22713 2017-02-28 Martin Liska <mliska@suse.cz>
22715 * common.opt: Replace space with tabular for options of <number>
22717 * config/i386/i386.opt: Show <number> value for
22718 -mlarge-data-threshold.
22719 * opts.c (print_filtered_help): Do not display number in hexadecimal
22722 2017-02-28 Martin Liska <mliska@suse.cz>
22724 * common.opt: Fix --help=option -Q for options which are of
22727 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
22729 * config/i386/i386.c (print_reg): Error out for values
22730 of 8-bit size in invalid integer register.
22732 2017-02-28 Martin Sebor <msebor@redhat.com>
22734 PR tree-optimization/79691
22735 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
22737 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22740 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
22741 gcc_unreachable with output_operand_lossage.
22743 2017-02-28 Richard Biener <rguenther@suse.de>
22745 PR tree-optimization/79740
22746 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
22748 (visit_nary_op): Insert the nary into the hashtable if we
22749 pattern-matched sth.
22750 * tree-ssa-pre.c (eliminate_insert): Robustify.
22752 2017-02-28 Richard Biener <rguenther@suse.de>
22754 PR middle-end/79731
22755 * fold-const.c (decode_field_reference): Reject out-of-bound
22758 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22760 * config/i386/i386.c: Include intl.h.
22761 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
22762 instead of just cond ? "..." : "...".
22763 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
22764 * coverage.c (read_counts_file): Likewise.
22765 * omp-offload.c: Include intl.h.
22766 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
22767 of just cond ? "..." : "...".
22768 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
22769 of just cond ? "..." : "...".
22771 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
22774 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
22776 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
22777 'tune for' CPU name.
22778 * config/arm/arm-cpu-data.h: Regenerated.
22780 2017-02-28 Richard Biener <rguenther@suse.de>
22782 PR tree-optimization/79732
22783 * tree-inline.c (expand_call_inline): Do not shadow var.
22785 2017-02-28 Richard Biener <rguenther@suse.de>
22787 PR tree-optimization/79723
22788 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
22789 address-space properly.
22791 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
22793 * doc/optinfo.texi (Optimization groups): Fix option used for
22795 * doc/invoke.texi (-fopt-info): Document "omp".
22796 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
22797 (OPTGROUP_ALL): Add OPTGROUP_OMP.
22798 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
22799 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
22800 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
22802 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
22804 * dumpfile.c (optgroup_options): Instead of "openmp", associate
22805 OPTGROUP_OMP with "omp".
22807 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
22810 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
22811 for arithmetic shift of unsigned V2DI.
22813 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
22815 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
22816 arc/linux.h headers.
22817 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
22818 (LINK_SPEC): Likewise.
22819 (ARC_TLS_EXTRA_START_SPEC): Likewise.
22820 (EXTRA_SPECS): Likewise.
22821 (STARTFILE_SPEC): Likewise.
22822 (ENDFILE_SPEC): Likewise.
22823 (LIB_SPEC): Likewise.
22824 (TARGET_SDATA_DEFAULT): Likewise.
22825 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
22826 (MULTILIB_DEFAULTS): Likewise.
22827 (DWARF2_UNWIND_INFO): Likewise.
22828 * config/arc/big.h: New file.
22829 * config/arc/elf.h: Likewise.
22830 * config/arc/linux.h: Likewise.
22831 * config/arc/t-uClibc: Remove.
22833 2017-02-27 Bin Cheng <bin.cheng@arm.com>
22835 PR tree-optimization/77536
22836 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
22837 (tree_transform_and_unroll_loop): Use above function to compute the
22838 estimated niter of unrolled loop and use it when scaling profile.
22839 Also use count info rather than frequency if it's non-zero.
22840 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
22841 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
22842 (vect_transform_loop): Call above function.
22844 2017-02-27 Richard Biener <rguenther@suse.de>
22846 PR tree-optimization/45397
22847 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
22848 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
22849 (visit_nary_op): Add pattern matching for CSEing sign-changed
22850 or truncated operations with wider ones.
22852 2017-02-27 Richard Biener <rguenther@suse.de>
22854 PR tree-optimization/79690
22855 * tree-vect-stmts.c (vectorizable_store): Use vector type
22856 built from the DR with address-space.
22858 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
22860 * doc/invoke.texi (Optimize Options): Refine the description
22861 of asan-use-after-return.
22863 2017-02-25 Alan Modra <amodra@gmail.com>
22865 PR rtl-optimization/79584
22866 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
22867 base, not ad->base_term, the reg within base. Remove assertion
22868 that ad->base == ad->base_term. Replace gen_int_mode using
22869 bogus mode with const0_rtx.
22871 2017-02-25 Jakub Jelinek <jakub@redhat.com>
22873 PR middle-end/79396
22874 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
22875 FMA_EXPR like tcc_binary or tcc_unary.
22877 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
22880 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
22882 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
22883 (output_loc_operands): Handle DW_OP_call_ref and
22884 DW_OP_GNU_variable_value.
22885 (struct variable_value_struct): New type.
22886 (struct variable_value_hasher): Likewise.
22887 (variable_value_hash): New variable.
22888 (string_types): Remove.
22889 (copy_loc_descr): New function.
22890 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
22891 (prepend_loc_descr_to_each): New function.
22892 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
22893 instead of add_loc_descr_to_each if the first argument is single
22894 location list and the second has multiple.
22895 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
22896 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
22897 when looking for variable value which doesn't have other location info.
22898 (loc_list_from_tree): Formatting fix.
22899 (gen_array_type_die): Simplify DW_AT_string_length handling.
22900 (adjust_string_types): Remove.
22901 (gen_subprogram_die): Don't call adjust_string_types nor test/set
22902 string_types. Call resolve_variable_values.
22903 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
22904 (resolve_addr_in_expr): Likewise. Add A argument.
22905 (copy_deref_exprloc): Remove deref argument. Adjust for the
22906 original expression being DW_OP_GNU_variable_value with optionally
22907 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
22908 optionally after it.
22909 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
22910 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
22911 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
22912 (variable_value_hasher::hash, variable_value_hasher::equal): New
22914 (resolve_variable_value_in_expr, resolve_variable_value,
22915 resolve_variable_values, note_variable_value_in_expr,
22916 note_variable_value): New functions.
22917 (dwarf2out_early_finish): Call note_variable_value on all toplevel
22920 2017-02-24 Jakub Jelinek <jakub@redhat.com>
22923 * opts.h (handle_generated_option): Add GENERATED_P argument.
22924 * opts-common.c (handle_option): Adjust function comment.
22925 (handle_generated_option): Add GENERATED_P argument, pass it to
22927 (control_warning_option): Pass false to handle_generated_option
22929 * opts.c (maybe_default_option): Pass true to handle_generated_option
22931 * optc-gen.awk: Likewise.
22933 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22935 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
22936 a REG, look at the REG it is a SUBREG of.
22937 (splitter for cmpeqsi_t): Ditto.
22939 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22941 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
22942 the special USEs with the pattern of the insn, not the insn itself.
22944 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
22947 * doc/invoke.texi: Document -mload-store-pairs.
22949 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22950 Sandra Loosemore <sandra@codesourcery.com>
22952 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
22953 argument isn't a CONST_INT.
22954 (nios2_alternate_compare_const): Assert op is a CONST_INT.
22955 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
22956 (nios2_validate_compare): Bypass alternate compare logic if *op2
22957 is not a CONST_INT.
22958 (ldstwm_operation_p): Return false if first_base is not a REG or
22959 if first_offset is not a CONST_INT.
22961 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22963 * config/cris/cris.md: Use correct operand in a define_peephole2.
22965 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22967 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
22969 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22971 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
22972 this_insn if it is an INSN or JUMP_INSN.
22973 (force_offsettable): Look at base, not at addr.
22974 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
22975 on things that aren't necessarily CONST_INTs.
22977 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
22979 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
22980 -mfpmath=sse is the default also for x86-32 targets with SSE2
22981 instruction set when @option{-ffast-math} is enabled
22983 2017-02-24 Jeff Law <law@redhat.com>
22985 PR rtl-optimizatoin/79286
22986 * ira.c (update_equiv_regs): Drop may_trap_p exception to
22989 2017-02-24 Richard Biener <rguenther@suse.de>
22991 PR tree-optimization/79389
22992 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
22995 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
22997 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
22998 function comment to reflect reality.
22999 (loop_exits_before_overflow): Fix typo in function description.
23001 2017-02-24 Richard Biener <rguenther@suse.de>
23003 PR tree-optimization/79389
23004 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
23005 properly that a threading opportunity exists. Detect conditional
23006 copy/constant propagation opportunities.
23008 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
23010 * config/visium/visium.md (type): Add trap.
23011 (b): New mode attribute.
23012 (*btst): Rename into...
23013 (*btst<mode>): ...this and adjust.
23014 (*cbranchsi4_btst_insn): Rename into...
23015 (*cbranch<mode>4_btst_insn): ...this and adjust.
23016 (trap): New define_insn.
23018 2017-02-23 Jakub Jelinek <jakub@redhat.com>
23020 PR tree-optimization/79389
23021 * ifcvt.c (struct noce_if_info): Add rev_cond field.
23022 (noce_reversed_cond_code): New function.
23023 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
23024 reversed_comparison_code. Formatting fix.
23025 (noce_try_store_flag): Test rev_cond != NULL in addition to
23026 reversed_comparison_code.
23027 (noce_try_store_flag_constants): Likewise.
23028 (noce_try_store_flag_mask): Likewise.
23029 (noce_try_addcc): Use rev_cond if non-NULL instead of
23030 reversed_comparison_code.
23031 (noce_try_cmove_arith): Likewise. Formatting fixes.
23032 (noce_try_minmax, noce_try_abs): Clear rev_cond.
23033 (noce_find_if_block): Initialize rev_cond.
23034 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
23035 instead of false as last argument never attempt to reverse it
23038 2017-02-23 Bin Cheng <bin.cheng@arm.com>
23040 PR tree-optimization/79663
23041 * tree-predcom.c (combine_chains): Process refs in reverse order
23042 only for ZERO length chains, and add explaining comment.
23044 2017-02-23 Jeff Law <law@redhat.com>
23046 PR tree-optimization/79578
23047 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
23048 in call to operand_equal_p.
23050 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
23053 * config/i386/cpuid.h: Fix another undefined behavior.
23055 2017-02-23 Richard Biener <rguenther@suse.de>
23057 PR tree-optimization/79683
23058 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
23059 vector types for data-refs.
23061 2017-02-23 Martin Liska <mliska@suse.cz>
23063 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
23065 2017-02-23 Jakub Jelinek <jakub@redhat.com>
23067 PR middle-end/79665
23068 * internal-fn.c (get_range_pos_neg): Moved to ...
23069 * tree.c (get_range_pos_neg): ... here. No longer static.
23070 * tree.h (get_range_pos_neg): New prototype.
23071 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
23072 are known to be in between 0 and signed maximum inclusive, try to
23073 expand both unsigned and signed divmod and use the cheaper one from
23076 2017-02-22 Jeff Law <law@redhat.com>
23078 PR tree-optimization/79578
23079 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
23080 to compare base operands.
23082 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
23085 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
23086 gpc_reg_operand instead of fpr_reg_operand.
23088 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
23090 * config/mips/mips.c (mips_return_in_memory): Force FP
23091 vector types to be returned in memory for o32 ABI.
23093 2017-02-22 Jakub Jelinek <jakub@redhat.com>
23095 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
23096 instead of DW_TAG_member for static data member declarations and don't
23097 set no_linkage_name for static inline data members.
23098 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
23101 2017-02-22 Martin Liska <mliska@suse.cz>
23103 * doc/invoke.texi: Replace inequality signs with square brackets
23106 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23108 PR tree-optimization/68644
23109 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
23111 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
23114 * lra-constraints.c (simplify_operand_subreg): Handle
23115 WORD_REGISTER_OPERATIONS targets.
23117 2017-02-22 Jakub Jelinek <jakub@redhat.com>
23120 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
23121 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
23122 elimination by swapping fld*.
23124 2017-02-22 Richard Biener <rguenther@suse.de>
23126 PR tree-optimization/79673
23127 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
23128 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
23129 irrelevant address-space qualifiers and avoiding a
23130 ADDR_SPACE_CONVERT_EXPR from fold_convert.
23132 2017-02-22 Richard Biener <rguenther@suse.de>
23134 PR tree-optimization/79666
23135 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
23136 to not symbolically negate if that may introduce undefined
23139 2017-02-22 Martin Liska <mliska@suse.cz>
23142 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
23143 * data-streamer-out.c (streamer_write_gcov_count_stream):
23145 * value-prof.c (stream_out_histogram_value): Make assert more
23146 precise based on type of counter.
23148 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
23151 * config/i386/i386.md (standard_x87sse_constant_load splitter):
23152 Use nonimmediate_operand instead of memory_operand for operand 1.
23153 (float-extend standard_x87sse_constant_load splitter): Ditto.
23155 2017-02-21 Jeff Law <law@redhat.com>
23157 PR tree-optimization/79621
23158 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
23159 blocks with edges to themselves.
23161 2017-02-21 Jakub Jelinek <jakub@redhat.com>
23164 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
23165 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
23166 Use gimple_call_builtin_p.
23169 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
23170 on temporarily removed DEBUG_INSNs.
23172 PR tree-optimization/79649
23173 * tree-loop-distribution.c (classify_partition): Give up on
23174 non-generic address space loads/stores.
23176 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
23178 * doc/loop.texi (Loop manipulation): Remove nonexistent
23179 tree_ssa_loop_version from the documentation.
23180 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
23182 2017-02-21 Jakub Jelinek <jakub@redhat.com>
23185 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
23186 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
23187 * config/rs6000/rs6000.c: Include except.h.
23188 (rs6000_expand_split_stack_prologue): Call
23189 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
23191 2017-02-21 Martin Jambor <mjambor@suse.cz>
23194 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
23195 have been analyzed.
23197 2017-02-21 Martin Jambor <mjambor@suse.cz>
23199 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
23200 for backward compatibility only.
23201 * doc/invoke.texi (Option Summary): Remove all references to
23202 -fipa-cp-alignment.
23204 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
23208 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23210 * lra-constraints.c (curr_insn_transform): Handle
23211 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
23213 2017-02-21 Martin Liska <mliska@suse.cz>
23215 * config/i386/i386.opt: Replace -masm-dialect with -masm.
23217 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
23219 PR translation/79638
23220 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
23222 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
23225 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
23226 (arm_function_ok_for_sibcall): Return false for an indirect call by
23227 descriptor if all the argument registers are used.
23228 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
23229 alignment of the function.
23231 2017-02-21 Jakub Jelinek <jakub@redhat.com>
23233 PR tree-optimization/61441
23234 * simplify-rtx.c (simplify_const_unary_operation): For
23235 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
23236 the sNaN unmodified.
23238 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
23240 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
23241 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
23242 instead of SYSTEM_HEADER_DIR.
23244 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
23245 Martin Liška <mliska@suse.cz>
23247 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
23248 Fix typos and grammar, use active voice, and clarify.
23250 2017-02-20 Marek Polacek <polacek@redhat.com>
23252 PR middle-end/79537
23253 * gimplify.c (gimplify_expr): Handle unused *&&L;.
23256 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
23258 2017-02-20 Jakub Jelinek <jakub@redhat.com>
23261 * config/i386/i386.c (ix86_expand_builtin): Handle
23262 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
23263 ix86_builtins_isa[fcode].isa as a requirement of those
23264 flags and any other flag in the bitmask.
23265 (ix86_init_mmx_sse_builtins): Use 0 instead of
23266 ~OPTION_MASK_ISA_64BIT as mask.
23267 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
23268 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
23269 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
23270 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
23272 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23275 * lra-constraints.c (split_reg): Check requested split mode
23276 is supported by the register.
23278 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23280 * lra-constraints.c (simplify_operand_subreg): Remove early
23283 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23286 * lra-constraints.c (curr_insn_transform): Tighten condition
23287 for converting SUBREG reloads from OP_OUT to OP_INOUT.
23289 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23292 * lra-constraints.c (curr_insn_transform): Handle
23293 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
23295 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
23298 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
23300 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
23302 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
23305 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
23308 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23310 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
23313 2017-02-18 Jakub Jelinek <jakub@redhat.com>
23315 * final.c (last_columnnum, override_columnnum): New variables.
23316 (final_start_function): Set last_columnnum, pass it to begin_prologue
23317 hook and pass 0 to dwarf2out_begin_prologue.
23318 (final_scan_insn): Update override_columnnum. Pass last_columnnum
23319 to source_line debug hook.
23320 (notice_source_line): Compute last_columnnum and for debug_column_info
23321 return true on column changes.
23322 * debug.h (struct gcc_debug_hooks): Add column argument to
23323 source_line and begin_prologue hooks.
23324 (debug_nothing_int_charstar_int_bool): Remove prototype.
23325 (debug_nothing_int_int_charstar,
23326 debug_nothing_int_int_charstar_int_bool): New prototypes.
23327 (dwarf2out_begin_prologue): Add column argument.
23328 * debug.c (do_nothing_debug_hooks): Adjust source_line and
23329 begin_prologue hooks.
23330 (debug_nothing_int_charstar_int_bool): Remove.
23331 (debug_nothing_int_int_charstar,
23332 debug_nothing_int_int_charstar_int_bool): New functions.
23333 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
23334 through to dwarf2out_source_line.
23335 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
23336 (dwarf2out_source_line): Add column argument, emit it if requested.
23337 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
23339 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
23340 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
23341 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
23342 through to dwarf2out_begin_prologue.
23343 (vmsdbgout_source_line): Add column argument, pass it through to
23344 dwarf2out_source_line.
23345 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
23346 dbxout_source_line caller.
23347 (dbxout_source_line): Add column argument.
23349 * common.opt (gno-column-info, gcolumn-info): New options.
23350 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
23351 (check_die): Also test for multiple DW_AT_decl_column attributes.
23352 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
23353 DW_AT_decl_column if requested.
23354 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
23356 (gen_variable_die): Likewise.
23357 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
23358 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
23361 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
23362 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
23363 (ix86_handle_option): Handle OPT_m3dnowa.
23364 * doc/invoke.texi (-m3dnowa): Document.
23365 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
23366 -m3dnowa instead of -m3dnow -march=athlon.
23369 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
23370 instead of gcc_assert for K, r and R code checks. Formatting fixes.
23372 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23375 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
23376 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
23377 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
23378 generator for vsx_xxpermdi_<mode>_be.
23379 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
23380 force big-endian semantics.
23381 (vsx_xxpermdi_<mode>_be): New define_expand with same
23382 implementation as previous version of vsx_xxpermdi_<mode>.
23384 2017-02-17 Jakub Jelinek <jakub@redhat.com>
23386 PR tree-optimization/79327
23387 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
23388 variable, its initialization and use.
23390 2017-02-17 Julia Koval <julia.koval@intel.com>
23392 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
23393 (OPTION_MASK_ISA_PKU_UNSET): New.
23394 (ix86_handle_option): Handle -mrdpid.
23395 * config/i386/cpuid.h (bit_RDPID): New.
23396 * config/i386/driver-i386.c (host_detect_local_cpu):
23397 Detect RDPID feature.
23398 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
23399 * config/i386/i386-c.c (ix86_target_macros_internal):
23401 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
23402 (ix86_valid_target_attribute_inner_p): Add "rdpid".
23403 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
23404 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
23405 * config/i386/i386.md (define_insn "rdpid"): New.
23406 * config/i386/i386.opt Add -mrdpid.
23407 * config/i386/immintrin.h (_rdpid_u32): New.
23409 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
23411 PR rtl-optimization/79541
23412 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
23413 instead of transforming it into USE.
23415 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
23417 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
23418 If HONOR_SNANS (SFmode) force the input to a register.
23419 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
23420 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
23421 an frsp or similar insn.
23423 2017-02-17 Martin Liska <mliska@suse.cz>
23425 PR rtl-optimization/79577
23426 * params.def (selsched-max-sched-times): Increase minimum to 1.
23428 2017-02-17 Martin Liska <mliska@suse.cz>
23430 PR rtl-optimization/79574
23431 * gcse.c (want_to_gcse_p): Prevent integer overflow.
23433 2017-02-17 Martin Liska <mliska@suse.cz>
23435 PR tree-optimization/79529
23436 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
23437 ssa_defined_default_def_p to handle cases which are implicitly
23439 * tree-ssa.c (ssa_defined_default_def_p): New function.
23440 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
23441 which are implicitly defined.
23442 * tree-ssa.h (ssa_defined_default_def_p): Declare.
23444 2017-02-17 Richard Biener <rguenther@suse.de>
23446 PR middle-end/79576
23447 * params.def (max-ssa-name-query-depth): Limit to 10.
23449 2017-02-17 Richard Biener <rguenther@suse.de>
23451 PR tree-optimization/79552
23452 * tree-ssa-structalias.c (visit_loadstore): Properly verify
23455 2017-02-17 Richard Biener <rguenther@suse.de>
23458 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
23460 2017-02-17 Marek Polacek <polacek@redhat.com>
23462 PR middle-end/79536
23463 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
23464 (fold_negate_expr): New wrapper.
23466 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
23468 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
23469 Correct terminology and de-emphasize pre-standard behavior.
23471 2017-02-16 Alan Modra <amodra@gmail.com>
23473 PR rtl-optimization/79286
23474 * ira.c (def_dominates_uses): New function.
23475 (update_equiv_regs): Don't create an equivalence for insns that
23476 may trap where the register def does not dominate the use.
23478 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
23480 PR rtl-optimization/78127
23481 * lra.c (lra): Call lra_eliminate before finish the loop after
23484 2017-02-16 Richard Biener <rguenther@suse.de>
23486 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
23488 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
23489 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
23490 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
23491 (isl_val_int_from_wi): New function.
23492 (extract_affine_gmp): Rename to ...
23493 (extract_affine_wi): ... this, take a widest_int.
23494 (extract_affine_int): Just wrap extract_affine_wi.
23495 (add_param_constraints): Use isl_val_int_from_wi.
23496 (add_loop_constraints): Likewise, and extract_affine_wi.
23498 2017-02-15 Jeff Law <law@redhat.com>
23500 PR middle-end/79521
23501 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
23502 ira_init_register_move_cost_if_necessary.
23504 2017-02-15 Martin Sebor <msebor@redhat.com>
23506 PR middle-end/32003
23507 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
23508 removed in a prior commit.
23510 2017-02-15 Bin Cheng <bin.cheng@arm.com>
23512 PR tree-optimization/79347
23513 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
23514 counters during peeling.
23516 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
23518 * Makefile.in (site.exp): Remove "set ISLVER".
23520 2017-02-15 Jakub Jelinek <jakub@redhat.com>
23523 * real.c (real_from_integer): Call real_convert even for decimal.
23525 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
23528 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
23530 2017-02-14 Andrew Pinski <apinski@cavium.com>
23532 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
23533 cores and change the partno/implementer to be correct.
23534 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
23535 the 'B" as the implementer.
23536 * config/aarch64/aarch64-tune.md: Regenerate.
23538 2017-02-14 Carl Love <cel@us.ibm.com>
23540 * config/rs6000/rs6000.c: Add case statement entry to make the
23541 xvcvuxdsp built-in argument unsigned.
23542 * config/rs6000/vsx.md: Fix the source and return operand types so they
23543 match the instruction definitions from the ISA document. Fix typo
23544 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
23547 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
23550 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
23551 member early_clobber_alts.
23552 * lra-lives.c (reg_early_clobber_p): New.
23553 (process_bb_lives): Use it.
23554 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
23555 (debug_operand_data): Initialize early_clobber_alts.
23556 (setup_operand_alternative): Set up early_clobber_alts.
23557 (collect_non_operand_hard_regs): Ditto. Pass early clobber
23558 alternatives to new_insn_reg.
23559 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
23561 (lra_update_insn_regno_info): Pass the new arg.
23563 2017-02-14 Jakub Jelinek <jakub@redhat.com>
23565 PR middle-end/79505
23566 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
23567 (new_oacc_loop_raw): Don't clear already cleared fields.
23570 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
23571 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
23572 _mm512_prefetch_i64gather_ps): New inline functions and macros.
23574 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
23577 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
23579 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
23582 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
23583 the extra instruction to the right place to store 128-bit constant
23586 2017-02-14 Martin Sebor <msebor@redhat.com>
23588 PR middle-end/79448
23589 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
23590 warning for strings of unknown length.
23592 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
23594 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
23596 2017-02-14 Jeff Law <law@redhat.com>
23599 * ira-costs.c (scan_one_insn): Initialize register move costs
23600 for pseudos seen in USE/CLOBBER insns.
23602 PR tree-optimization/79095
23603 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
23604 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
23605 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
23606 if the operands are known to be not equal, then the resulting range
23608 (intersect_ranges): If the new range is ~[0,0] and the old range is
23609 wide, then prefer ~[0,0].
23610 * tree-vrp.c (overflow_comparison_p_1): New function.
23611 (overflow_comparison_p): New function.
23612 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
23613 if NAME is used in an overflow test.
23614 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
23615 overflow check that can be expressed as an equality test, then adjust
23616 ops to be that equality test.
23618 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23620 * config/s390/s390-builtin-types.def: Remove flags argument.
23621 * config/s390/s390.c (s390_init_builtins): Likewise.
23623 2017-02-14 Martin Liska <mliska@suse.cz>
23625 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
23626 vector. Fix trailing white spaces.
23628 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
23630 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
23633 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23635 PR rtl-optimization/68664
23636 * config/arm/arm.c (arm_sched_can_speculate_insn):
23637 New function. Declare prototype.
23638 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
23640 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23642 PR rtl-optimization/68664
23643 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
23645 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
23647 2017-02-14 Amit Pawar <amit.pawar@amd.com>
23649 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
23650 max skip bytes for function, loop and jump.
23652 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23654 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
23655 ABS_EXPR for gimple dump.
23657 2017-02-14 Jakub Jelinek <jakub@redhat.com>
23660 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
23662 PR tree-optimization/79408
23663 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
23664 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
23665 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
23666 also if rhs1 is INTEGER_CST.
23668 2017-02-14 Richard Biener <rguenther@suse.de>
23670 PR middle-end/79432
23671 * tree-into-ssa.c (insert_phi_nodes): When the function can
23672 have abnormal edges rewrite SSA names with broken use-def
23673 dominance out of SSA and register them for PHI insertion.
23675 2017-02-13 Martin Sebor <msebor@redhat.com>
23677 PR middle-end/79496
23678 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
23679 clearing info.nowrite flag when snprintf size argument is a range.
23681 2017-02-13 Jakub Jelinek <jakub@redhat.com>
23683 * cprop.c (cprop_jump): Add missing space in string literal.
23684 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
23685 (get_constraint_for_component_ref): Likewise.
23686 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
23687 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
23688 * lra-constraints.c (process_alt_operands): Likewise.
23689 * ipa-inline.c (inline_small_functions): Likewise.
23690 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
23691 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23692 * trans-mem.c (diagnose_tm_1_op): Likewise.
23693 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
23694 (grid_parallel_clauses_gridifiable): Likewise.
23696 * config/nvptx/mkoffload.c (process): Add space in between
23699 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
23700 "MOD4_SSE_REGS" and "ALL_REGS".
23702 * spellcheck.c (test_data): Add , in between "foo" and "food".
23704 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23707 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
23708 boundary crossing check and subsequent code generation agree.
23710 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23712 * config/aarch64/aarch64.c (has_memory_op): Delete.
23713 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
23716 2017-02-13 Jakub Jelinek <jakub@redhat.com>
23718 PR rtl-optimization/79388
23719 PR rtl-optimization/79450
23720 * combine.c (distribute_notes): When removing TEM_INSN for which
23721 corresponding dest has last value recorded, invalidate that last
23724 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23726 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
23727 of explicit '@'. Add missing assembly comment marker on branch costs
23730 2017-02-13 Nathan Sidwell <nathan@acm.org>
23732 * gengtype-lex.l (<in_struct>): Add '/'.
23734 2017-02-13 Martin Liska <mliska@suse.cz>
23737 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
23739 2017-02-13 Richard Biener <rguenther@suse.de>
23741 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
23743 * configure: Re-generate.
23744 * config.in: Likewise.
23745 * graphite-dependences.c: Simplify as if
23746 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
23747 * graphite-isl-ast-to-gimple.c: Likewise.
23748 * graphite-optimize-isl.c: Likewise.
23749 * graphite-poly.c: Likewise.
23750 * graphite-sese-to-poly.c: Likewise.
23751 * graphite.h: Likewise.
23752 * toplev.c: Include isl/version.h and use isl_version () for
23753 printing the ISL version.
23754 * doc/install.texi: Update ISL requirement.
23756 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
23758 * doc/standards.texi (Standards): Update reference to
23761 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
23763 * doc/extend.texi (Named Address Spaces): sourceware.org now
23765 * doc/install.texi (Binaries): Ditto.
23768 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
23770 * doc/cpp.texi: Replace "stringify"/"stringification" with C
23771 standard terminology "stringize"/"stringizing" throughout.
23772 * doc/cppinternals.texi: Likewise.
23774 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
23776 * doc/extend.texi: Fix some spelling mistakes and typos.
23777 * doc/invoke.texi: Likewise.
23779 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
23782 * params.def (inline-min-speedup) Change from 10 to 8.
23784 2017-02-11 Jakub Jelinek <jakub@redhat.com>
23786 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
23789 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
23792 * ipa-inline-analysis.c (get_minimal_bb): New function.
23793 (record_modified): Use it.
23794 (remap_edge_change_prob): Handle also ancestor functions.
23796 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
23798 * doc/contrib.texi (Contributors): Remove broken link into
23799 the Mauve CVS repository.
23801 2017-02-11 Jakub Jelinek <jakub@redhat.com>
23803 PR middle-end/79454
23804 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
23805 result computation whenever lhs doesn't have vector mode, not
23806 just when it has BLKmode.
23808 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
23810 * doc/makefile.texi (profiledbootstrap): Refer to the
23811 installation instructions only in textual form.
23813 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23816 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
23818 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
23820 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
23821 (Specific): Update mingw-w64 reference.
23823 (Specific): Remove broken link to Renesas RX processor.
23825 2017-02-10 Richard Biener <rguenther@suse.de>
23827 * toplev.c (process_options): Do not mention obsolete graphite
23828 options when printing sorry message about missing graphite support.
23829 Mention -floop-nest-optimize.
23831 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
23833 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
23834 (vtst_p16): Likewise.
23835 (vtstq_p8): Likewise.
23836 (vtstq_p16): Likewise.
23838 (vtstq_p64): Likewise.
23839 * config/arm/arm_neon.h (vgetq_lane_p64): New.
23840 (vset_lane_p64): New.
23841 (vsetq_lane_p64): New.
23843 2017-02-10 Jakub Jelinek <jakub@redhat.com>
23845 PR tree-optimization/79411
23846 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
23847 stmt operands are SSA_NAMEs used in abnormal phis.
23848 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
23851 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
23854 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
23857 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
23859 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
23861 2017-02-09 Jakub Jelinek <jakub@redhat.com>
23863 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
23867 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
23868 not arbitrary TREE_CONSTANT.
23871 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
23872 "omp declare target link" attribute unless is_global_var.
23873 * omp-offload.c (find_link_var_op): Likewise.
23875 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
23876 Chung-Lin Tang <cltang@codesourcery.com>
23878 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
23880 (gimplify_adjust_omp_clauses): Don't delete TILE.
23881 (gimplify_omp_for): Deal with TILE.
23882 * internal-fn.c (expand_GOACC_TILE): New function.
23883 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
23885 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
23886 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
23888 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
23889 avoid DIV for outermost collapse var.
23890 (expand_oacc_for): Insert tile element loop as needed. Adjust.
23891 Remove out of date comments, fix whitespace.
23892 * omp-general.c (omp_extract_for_data): Deal with tiling.
23893 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
23894 adjust OLF_DIM_BASE value.
23895 (struct omp_for_data): Add tiling field.
23896 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
23897 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
23898 for auto loops. Remove default auto determining, moved to
23899 oacc_loop_fixed_partitions.
23900 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
23901 stmts, add e_mask field.
23902 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
23903 (oacc_thread_numbers): Use oacc_dim_call.
23904 (oacc_xform_tile): New.
23905 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
23906 (finish_oacc_loop): Adjust for ifns vector.
23907 (oacc_loop_discover_walk): Append loop abstraction sites to list,
23908 add case for GOACC_TILE fns.
23909 (oacc_loop_xform_loop): Delete.
23910 (oacc_loop_process): Iterate over call list directly, and add
23911 handling for GOACC_TILE fns.
23912 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
23914 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
23915 vector partitioning to outer loops. Assign 2 partitions to loops
23916 when available. Add TILE handling.
23917 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
23918 (execite_oacc_device_lower): Process GOACC_TILE fns,
23919 ignore unknown specs.
23920 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
23921 * tree.c (omp_clause_num_ops): Adjust TILE ops.
23922 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
23924 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
23926 * configure.ac (ACX_BUGURL): Update.
23927 * configure: Regenerate.
23929 2017-02-09 Richard Biener <rguenther@suse.de>
23931 PR tree-optimization/69823
23932 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
23933 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
23935 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
23937 * config/arc/arc-c.def: Add __NPS400__ definition.
23938 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
23939 (TARGET_NPS400): Define.
23941 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
23943 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
23945 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
23946 pointer, arch_info.
23947 (arc_cpu_types): Fill the arch_info field with a pointer into the
23948 arc_arch_types table.
23949 (arc_selected_cpu): Declare.
23950 * config/arc/arc.c (arc_selected_cpu): Make global.
23951 (arc_selected_arch): Delete.
23952 (arc_base_cpu): Delete.
23953 (arc_override_options): Remove references to deleted variables,
23954 update access to arch information.
23955 (ARC_OPT): Update access to arch information.
23956 (ARC_OPTX): Likewise.
23957 * config/arc/arc.h (arc_base_cpu): Remove declaration.
23958 (TARGET_ARC600): Update access to arch information.
23959 (TARGET_ARC601): Likewise.
23960 (TARGET_ARC700): Likewise.
23961 (TARGET_EM): Likewise.
23962 (TARGET_HS): Likewise.
23963 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
23966 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
23969 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
23970 condition/operands for integer GE/LE/GEU/LEU operations.
23972 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
23974 PR translation/79397
23975 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
23978 2017-02-08 Martin Jambor <mjambor@suse.cz>
23981 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
23982 whether allocation happened.
23983 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
23984 nothing was allocated.
23986 2017-02-08 Jakub Jelinek <jakub@redhat.com>
23988 PR tree-optimization/79408
23989 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
23990 constant, but SSA_NAME with a known integer range, use the minimum
23991 of that range instead of op1 to determine if modulo can be replaced
23992 with its first operand.
23994 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23996 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
23998 2017-02-08 Richard Biener <rguenther@suse.de>
24000 PR tree-optimization/71824
24001 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
24002 Check all loops contained in the merged region.
24004 2017-02-07 Andrew Pinski <apinski@cavium.com>
24006 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
24008 2017-02-07 Andrew Pinski <apinski@cavium.com>
24010 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
24011 (thunderxt88): Likewise.
24012 (thunderxt81): Disable LSE and change v8.1 to v8.
24013 (thunderxt83): Likewise.
24015 2017-02-07 Jakub Jelinek <jakub@redhat.com>
24016 Richard Biener <rguenther@suse.de>
24018 PR middle-end/79399
24019 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
24020 type from int to size_t.
24021 * ira-costs.c (struct_costs_size): Change type from int to size_t.
24023 2017-02-07 Jakub Jelinek <jakub@redhat.com>
24025 PR rtl-optimization/79386
24026 * cprop.c (bypass_conditional_jumps): Initialize
24027 bypass_last_basic_block already before splitting bbs after
24028 unconditional traps...
24029 (bypass_conditional_jumps): ... rather than here.
24032 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
24033 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
24034 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
24035 fix -masm=intel patterns.
24037 2017-02-07 Richard Biener <rguenther@suse.de>
24039 PR tree-optimization/79256
24040 PR middle-end/79278
24041 * builtins.c (get_object_alignment_2): Use min_align_of_type
24042 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
24043 and ADJUST_FIELD_ALIGN.
24045 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
24047 * doc/tm.texi: Regenerate.
24048 * stor-layout.c (layout_decl): Adjust.
24049 (update_alignment_for_field): Likewise.
24050 (place_field): Likewise.
24051 (min_align_of_type): Likewise.
24052 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
24053 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
24054 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
24055 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
24056 * config/frv/frv.c (frv_adjust_field_align): Likewise.
24057 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
24058 * config/i386/i386.c (x86_field_alignment): Likewise.
24059 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
24060 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
24061 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
24062 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
24063 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
24064 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
24068 2017-01-30 Richard Biener <rguenther@suse.de>
24070 PR tree-optimization/79256
24071 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
24072 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
24075 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
24077 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
24078 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
24079 builtins to SImode and emit a zero-extend, if necessary.
24081 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
24083 * docs/invoke.texi (RISC-V Options): Alphabetize.
24085 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
24087 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
24090 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
24092 * config/riscv/riscv.c: New file.
24093 * common/config/riscv/riscv-common.c: Likewise.
24094 * config.gcc: Likewise.
24095 * config/riscv/constraints.md: Likewise.
24096 * config/riscv/elf.h: Likewise.
24097 * config/riscv/generic.md: Likewise.
24098 * config/riscv/linux.h: Likewise.
24099 * config/riscv/multilib-generator: Likewise.
24100 * config/riscv/peephole.md: Likewise.
24101 * config/riscv/pic.md: Likewise.
24102 * config/riscv/predicates.md: Likewise.
24103 * config/riscv/riscv-builtins.c: Likewise.
24104 * config/riscv/riscv-c.c: Likewise.
24105 * config/riscv/riscv-ftypes.def: Likewise.
24106 * config/riscv/riscv-modes.def: Likewise.
24107 * config/riscv/riscv-opts.h: Likewise.
24108 * config/riscv/riscv-protos.h: Likewise.
24109 * config/riscv/riscv.h: Likewise.
24110 * config/riscv/riscv.md: Likewise.
24111 * config/riscv/riscv.opt: Likewise.
24112 * config/riscv/sync.md: Likewise.
24113 * config/riscv/t-elf-multilib: Likewise.
24114 * config/riscv/t-linux: Likewise.
24115 * config/riscv/t-linux-multilib: Likewise.
24116 * config/riscv/t-riscv: Likewise.
24117 * configure.ac: Likewise.
24118 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
24119 Waterman as RISC-V maintainers.
24120 * doc/install.texi: Add RISC-V entries.
24121 * doc/invoke.texi: Add RISC-V options section.
24122 * doc/md.texi: Add RISC-V constraints section.
24123 * configure: Regenerated.
24125 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
24128 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
24129 false values to be constant vectors with all 0 or all 1 bits set.
24130 (vcondu<mode><mode>): Likewise.
24131 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
24133 (fpmask_comparison_operator): Update comment.
24134 (vecint_comparison_operator): New predicate.
24135 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
24136 vector conditionals when the true and false values are constant
24137 vectors with all 0 bits or all 1 bits set.
24139 2017-02-06 Martin Sebor <msebor@redhat.com>
24141 PR tree-optimization/79376
24142 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
24144 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
24146 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
24147 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
24148 to simplify split condition.
24150 2017-02-06 Jakub Jelinek <jakub@redhat.com>
24152 * omp-expand.c (oxpand_omp_atomic_fetch_op,
24153 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
24156 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
24158 PR rtl-optimization/68664
24159 * target.def (can_speculate_insn): New hook.
24160 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
24161 * doc/tm.texi: Regenerate.
24162 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
24163 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
24164 (rs6000_sched_can_speculate_insn): New function.
24166 2017-02-06 Jakub Jelinek <jakub@redhat.com>
24168 PR tree-optimization/79284
24169 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
24170 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
24171 vectorizable_mask_load_store, vectorizable_operation,
24172 vect_is_simple_cond, get_same_sized_vectype): Use it instead
24173 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
24174 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
24175 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
24176 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
24177 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
24178 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
24179 is_gimple_assign (stmt). Replace another such test with
24180 is_gimple_assign (stmt).
24182 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
24185 * config/avr/avr.c (rtl-iter.h): Include it.
24186 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
24187 (avr_legitimate_combined_insn): ...and implementation.
24189 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
24191 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
24192 * config/s390/s390.c (s390_const_operand_ok)
24193 (s390_canonicalize_comparison, s390_extract_part)
24194 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
24195 (s390_contiguous_bitmask_p, s390_rtx_costs)
24196 (legitimize_pic_address): Likewise.
24197 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
24198 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
24199 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
24200 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
24201 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
24203 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
24205 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
24206 REGNO($0) == REGNO($1).
24208 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24210 * config/s390/linux.h(SIZE_TYPE): Add comment.
24212 2017-02-06 Julian Brown <julian@codesourcery.com>
24213 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24214 Virendra Pathak <virendra.pathak@broadcom.com>
24216 * config/aarch64/aarch64-cores.def: Change the scheduler
24218 * config/aarch64/aarch64.md: Include thunderx2t99.md.
24219 * config/aarch64/thunderx2t99.md: New file.
24221 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
24223 * doc/standards.texi (Go Language): Update link to language
24226 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
24228 * tree-eh.c (lower_resx): Sanitize profile.
24229 (cleanup_empty_eh_move_lp): Likewise.
24231 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
24234 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
24236 * cfgloopmanip.h (loop_version): Update prototype.
24237 * modulo-sched.c (sms_schedule): Update call of loop_version.
24238 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
24239 * tree-parloops.c (gen_parallel_loop): Likewise.
24240 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
24241 * tree-ssa-loop-split.c (split_loop): Likewise.
24242 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
24243 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24245 2017-02-05 Martin Liska <mliska@suse.cz>
24248 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
24250 (print_operand_address): Initialize a struct to zero.
24252 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
24254 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
24255 garbage collector only in textual form.
24257 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
24259 * doc/extend.texi (x86 specific memory model extensions for
24260 transactional memory): Simplify a phrase.
24262 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
24265 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
24266 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
24267 (atomic_storedi_1): Likewise.
24269 2017-02-04 Jakub Jelinek <jakub@redhat.com>
24271 PR tree-optimization/79338
24272 * tree-parloops.c (gather_scalar_reductions): Don't call
24273 vect_analyze_loop_form for loop->inner before destroying loop's
24276 2017-02-03 Martin Sebor <msebor@redhat.com>
24278 PR tree-optimization/79327
24279 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
24280 when precision has resulted in leading zeros.
24281 (format_integer): Adjust the likely counter to assume an unknown
24282 argument that may be zero is non-zero.
24284 2017-02-03 Jason Merrill <jason@redhat.com>
24287 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
24288 avoid copying non-taken branch.
24290 2017-02-03 Jakub Jelinek <jakub@redhat.com>
24292 PR tree-optimization/79340
24293 * tree-vect-loop.c (vectorizable_reduction): Release
24294 vec_defs elements after safe_splicing them into other vectors.
24297 PR tree-optimization/79327
24298 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
24299 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
24301 (format_integer): Use wide_int_to_tree instead of build_int_cst
24302 + to_?hwi. If argmin is NULL, just set argmin and argmax to
24303 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
24304 of shortest and longest sequence.
24306 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
24308 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
24309 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
24311 2017-02-03 Walter Lee <walt@tilera.com>
24314 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
24315 after initial stackframe link reg save.
24316 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
24318 2017-02-03 Jakub Jelinek <jakub@redhat.com>
24321 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
24322 wu for stxssp alternative.
24324 2017-02-03 Martin Sebor <msebor@redhat.com>
24326 PR tree-optimization/79352
24327 * gimple-fold.c (get_range_strlen): Add argument.
24328 (get_range_strlen): Change return type to bool.
24329 (get_maxval_strlen): Pass in a dummy argument.
24330 * gimple-fold.h (get_range_strlen): Change return type to bool.
24331 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
24332 * tree.h (array_at_struct_end_p): Add argument.
24333 * tree.c (array_at_struct_end_p): Handle it.
24335 2017-02-03 Martin Liska <mliska@suse.cz>
24338 * multiple_target.c (create_dispatcher_calls): Redirect edge
24339 from a caller of a dispatcher.
24340 (expand_target_clones): Make the clones local.
24341 (ipa_target_clone): Do both target clones and resolvers.
24342 (ipa_dispatcher_calls): Remove the pass.
24343 (pass_dispatcher_calls::gate): Likewise.
24344 (make_pass_dispatcher_calls): Likewise.
24345 * passes.def (pass_target_clone): Put as very first IPA early
24348 2017-02-03 Martin Liska <mliska@suse.cz>
24350 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
24351 in case of a function with ifunc attribute.
24353 2017-02-03 Martin Liska <mliska@suse.cz>
24355 * cgraph.c (cgraph_node::dump): Dump function version info.
24356 * symtab.c (symtab_node::dump_base): Add missing new line.
24358 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
24360 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
24361 (ifcombine_ifandif): Use it.
24363 2017-02-03 Martin Liska <mliska@suse.cz>
24365 * doc/invoke.texi: Document default value for
24366 use-after-scope-direct-emission-threshold.
24368 2017-02-03 Martin Liska <mliska@suse.cz>
24370 PR tree-optimization/79339
24371 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
24372 (format_floating): Likewise.
24374 2017-02-03 Martin Liska <mliska@suse.cz>
24377 * ipa-prop.c (ipa_node_params_t::insert): Remove current
24379 (ipa_node_params_t::remove): Likewise.
24380 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
24381 initialization from removed ipa_node_params_t::insert.
24382 (ipa_node_params::~ipa_node_params): Move from removed
24383 ipa_node_params_t::release.
24384 * symbol-summary.h (symbol_summary::m_released): New member.
24385 Do not release a summary twice. Do not allow to call finalizer
24386 for types of a summary that live in GGC memory.
24388 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24390 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
24393 2017-02-02 Martin Sebor <msebor@redhat.com>
24395 PR middle-end/79275
24396 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
24397 (format_string): Tighten up the range of output for non-constant
24398 strings and correct the expected range for wide non-constant strings.
24400 2017-02-02 Martin Sebor <msebor@redhat.com>
24402 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
24404 PR middle-end/32003
24405 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
24407 (-fdump-tree-@var): Add to index and document how to come up
24408 with pass-specific option and dump file names.
24409 (-fdump-passes): Clarify where to look for output.
24411 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
24413 PR middle-end/77445
24414 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
24415 statistics of the analyzed path; allow threading for speed when
24416 any of BBs along the path are optimized for speed.
24418 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
24420 PR middle-end/78468
24421 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
24422 settings of the virtual registers.
24425 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
24427 * explow.c (get_dynamic_stack_size): Take known alignment of stack
24428 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
24431 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24433 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
24434 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
24436 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24438 * config/s390/s390.md: Add missing comments with the expanded
24440 * config/s390/vector.md: Likewise.
24441 * config/s390/vx-builtins.md: Likewise.
24443 2017-02-02 Jakub Jelinek <jakub@redhat.com>
24446 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
24447 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
24448 conditions on a single line.
24450 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24452 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
24453 __S390_VX__ to __VX__.
24455 2017-02-01 Andrew Pinski <apinski@cavium.com>
24457 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
24458 stmt_info to record_stmt_cost.
24459 (vect_get_known_peeling_cost): Pass stmt_info if known to
24461 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
24462 cpu_vector_cost field into
24463 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
24464 field into vec_int_stmt_cost and vec_fp_stmt_cost.
24465 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
24466 splitting of scalar_stmt_cost and vec_stmt_cost.
24467 (thunderx_vector_cost): Likewise.
24468 (cortexa57_vector_cost): LIkewise.
24469 (exynosm1_vector_cost): Likewise.
24470 (xgene1_vector_cost): Likewise.
24471 (thunderx2t99_vector_cost): Improve after the splitting of the two
24473 (aarch64_builtin_vectorization_cost): Update for the splitting of
24474 scalar_stmt_cost and vec_stmt_cost.
24476 2017-02-01 Torvald Riegel <triegel@redhat.com>
24477 Richard Henderson <rth@redhat.com>
24479 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
24480 conditional on existance of a fast atomic load.
24481 * optabs-query.c (can_atomic_load_p): New function.
24482 * optabs-query.h (can_atomic_load_p): Declare it.
24483 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
24484 no fast atomic load is available for the particular size of access.
24485 (expand_atomic_compare_and_swap): Likewise.
24486 (expand_atomic_load): Likewise.
24487 (expand_atomic_store): Likewise.
24488 (expand_atomic_fetch_op): Likewise.
24489 * testsuite/lib/target-supports.exp
24490 (check_effective_target_sync_int_128): Remove x86 because it provides
24491 no fast atomic load.
24492 (check_effective_target_sync_int_128_runtime): Likewise.
24494 2017-02-01 Richard Biener <rguenther@suse.de>
24496 * graphite.c: Include tree-vectorizer.h for find_loop_location.
24497 (graphite_transform_loops): Provide opt-info for optimized nests.
24498 * tree-parloop.c (parallelize_loops): Provide opt-info for
24499 parallelized loops.
24501 2017-02-01 Richard Biener <rguenther@suse.de>
24503 PR middle-end/79315
24504 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
24505 was not set before.
24507 2017-02-01 Richard Biener <rguenther@suse.de>
24509 PR tree-optimization/71824
24510 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
24511 Verify the loops are valid in the merged SESE region.
24512 (scop_detection::can_represent_loop_1): Check analyzing the
24513 evolution of the number of iterations in the region succeeds.
24515 2017-01-31 Ian Lance Taylor <iant@golang.org>
24517 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
24518 REG_ARGS_SIZE note to 32-bit push insns and call insn.
24520 2017-01-31 David Malcolm <dmalcolm@redhat.com>
24522 PR preprocessor/79210
24523 * input.c (get_substring_ranges_for_loc): Replace line_width
24524 assertion with error-handling.
24526 2017-01-31 Richard Biener <rguenther@suse.de>
24528 PR tree-optimization/77318
24529 * graphite-sese-to-poly.c (extract_affine): Fix assert.
24530 (create_pw_aff_from_tree): Take loop parameter.
24531 (add_condition_to_pbb): Pass loop of the condition to
24532 create_pw_aff_from_tree.
24534 2017-01-31 Jakub Jelinek <jakub@redhat.com>
24536 * config/s390/s390.c (s390_asan_shadow_offset): New function.
24537 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
24539 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
24543 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
24545 (convert_int_to_float128): Likewise.
24546 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
24547 (convert_int_to_float128): Likewise.
24548 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
24549 (UNSPEC_IEEE128_CONVERT): Likewise.
24550 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
24551 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
24552 Use local variables for IBM extended format.
24553 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
24554 (fix_trunc<mode>si2_fprs): Likewise.
24555 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
24556 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
24557 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
24558 to know that we can now have integers of all sizes in vector
24560 (fix<uns>_<mode>di2_hw): Likewise.
24561 (float<uns>_<mode>si2_hw): Likewise.
24562 (fix_<mode>si2_hw): Likewise.
24563 (fixuns_<mode>si2_hw): Likewise.
24564 (float<uns>_<mode>di2_hw): Likewise.
24565 (float_<mode>di2_hw): Likewise.
24566 (float_<mode>si2_hw): Likewise.
24567 (floatuns_<mode>di2_hw): Likewise.
24568 (floatuns_<mode>si2_hw): Likewise.
24569 (xscvqp<su>wz_<mode>): Delete, no longer used.
24570 (xscvqp<su>dz_<mode>): Likewise.
24571 (xscv<su>dqp_<mode>): Likewise.
24572 (ieee128_mfvsrd_64bit): Likewise.
24573 (ieee128_mfvsrd_32bit): Likewise.
24574 (ieee128_mfvsrwz): Likewise.
24575 (ieee128_mtvsrw): Likewise.
24576 (ieee128_mtvsrd_64bit): Likewise.
24577 (ieee128_mtvsrd_32bit): Likewise.
24579 2017-01-31 Martin Liska <mliska@suse.cz>
24582 * ipa-prop.c (ipa_free_all_node_params): Call release method
24583 instead of ~sumbol_summary to not to trigger double times
24586 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
24588 PR tree-optimization/71691
24589 * bitmap.h (class auto_bitmap): New.
24590 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
24591 is_maybe_undefined instead of ssa_undefined_value_p.
24593 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24595 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
24596 __S390_ARCH_LEVEL__ to __ARCH__.
24598 2017-01-31 Jakub Jelinek <jakub@redhat.com>
24600 PR tree-optimization/79267
24601 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
24602 if should_remove_lhs_p is true.
24604 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
24607 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
24608 (add_alignment_attribute): New.
24609 (base_type_die): Add alignment attribute.
24610 (subrange_type_die): Likewise.
24611 (modified_type_die): Likewise.
24612 (gen_array_type_die): Likewise.
24613 (gen_descr_array_type_die: Likewise.
24614 (gen_enumeration_type_die): Likewise.
24615 (gen_subprogram_die): Likewise.
24616 (gen_variable_die): Likewise.
24617 (gen_field_die): Likewise.
24618 (gen_ptr_to_mbr_type_die): Likewise.
24619 (gen_struct_or_union_type_die): Likewise.
24620 (gen_subroutine_type_die): Likewise.
24621 (gen_typedef_die): Likewise.
24622 (base_type_cmp): Compare alignment attribute.
24624 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24627 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
24628 (setb_unsigned) New pattern for setb with CCUNS.
24629 * config/rs6000/rs6000.c (expand_block_compare): Use a different
24630 subfc./subfe sequence to avoid overflow problems. Generate a
24631 shorter sequence with cmpld/setb for power9.
24632 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
24633 for generating subfc. instruction.
24634 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
24635 now uses this instruction.
24637 2017-01-30 Ian Lance Taylor <iant@google.com>
24640 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
24641 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
24643 2017-01-30 Martin Sebor <msebor@redhat.com>
24645 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
24646 Move constant to the right of a relational operator.
24647 (get_mpfr_format_length, format_character, format_string): Ditto.
24648 (should_warn_p, maybe_warn): Same.
24650 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
24652 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
24655 * asan.c (get_translation_unit_decl): Remove function.
24656 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
24658 2017-01-30 Martin Liska <mliska@suse.cz>
24660 PR gcov-profile/79259
24661 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
24662 -fprofile-generate.
24664 2017-01-30 Martin Liska <mliska@suse.cz>
24667 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
24668 Initialize variables with NULL value.
24670 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
24673 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
24675 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
24677 2017-01-30 Richard Biener <rguenther@suse.de>
24679 PR tree-optimization/79276
24680 * tree-vrp.c (process_assert_insertions): Properly adjust common
24681 when removing a duplicate.
24683 * gcc.dg/torture/pr79276.c: New testcase.
24685 2017-01-30 Richard Biener <rguenther@suse.de>
24687 PR tree-optimization/79256
24688 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
24689 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
24691 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
24693 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
24696 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
24697 ("*r<noxa>sbg_<mode>_sll_bitmask")
24698 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
24699 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
24700 Use contiguous_bitmask_nowrap_operand.
24702 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24705 * config/rs6000/altivec.h (vec_xl): Revise #define.
24706 (vec_xst): Likewise.
24708 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
24710 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
24712 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
24714 PR rtl-optimization/79194
24715 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
24716 traps before call to bypass_conditional_jumps.
24718 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
24720 PR tree-optimization/71374
24721 * lra-constraints.c (check_conflict_input_operands): New.
24722 (match_reload): Use it.
24724 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
24727 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
24728 account to calculate conflict_set.
24730 2017-01-27 Bin Cheng <bin.cheng@arm.com>
24732 PR rtl-optimization/78559
24733 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
24734 other_insn in combine.
24736 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24738 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
24739 uint16_type_node for BT_UINT16.
24741 2017-01-27 David Malcolm <dmalcolm@redhat.com>
24743 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
24744 "RTL Tests" to menu.
24745 (GIMPLE Tests): New node.
24746 (RTL Tests): New node.
24748 2017-01-27 Richard Biener <rguenther@suse.de>
24750 PR tree-optimization/79245
24751 * tree-loop-distribution.c (distribute_loop): Apply cost
24752 modeling also to detected patterns.
24754 2017-01-27 Richard Biener <rguenther@suse.de>
24756 PR tree-optimization/71433
24757 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
24758 (compare_assert_loc): New function.
24759 (process_assert_insertions): Sort and optimize assert locations
24760 to remove duplicates and push down identical assertions on
24761 edges to their destination block.
24763 2017-01-27 Richard Biener <rguenther@suse.de>
24765 PR tree-optimization/79244
24766 * tree-vrp.c (remove_range_assertions): Forcefully propagate
24767 out SSA names even if abnormal.
24769 2017-01-27 Jakub Jelinek <jakub@redhat.com>
24771 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
24772 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
24773 instead of MPFR_RNDN.
24775 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
24778 * arm.c (arm_option_override): Don't call build_target_option_node
24779 until after doing all option overrides.
24780 (arm_valid_target_attribute_tree): Likewise.
24782 2017-01-27 Martin Liska <mliska@suse.cz>
24784 * doc/invoke.texi (-fprofile-arcs): Document profiling support
24785 for {cd}tors and C++ {cd}tors.
24787 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
24789 * config/s390/s390.md ("*setmem_long_and")
24790 ("*setmem_long_and_31z"): Use zero_extend instead of and.
24792 2017-01-26 Martin Sebor <msebor@redhat.com>
24794 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
24797 2017-01-26 Martin Sebor <msebor@redhat.com>
24799 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
24800 HAVE_DFmode before using XFmode or DFmode.
24801 (parse_directive): Avoid using the z length modifier to avoid
24802 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
24804 PR middle-end/78703
24805 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
24806 to accept adjustment as an array.
24807 (get_int_range): New function.
24808 (struct directive): Make width and prec arrays.
24809 (directive::set_width, directive::set_precision): Call get_int_range.
24810 (format_integer, format_floating): Handle width and precision ranges.
24811 (format_string, parse_directive): Same.
24813 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24816 * dwarf2out.c (generate_skeleton_bottom_up): For children with
24817 comdat_type_p set, just clone them, but keep the children in the
24821 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
24822 which have direct callers with -fvar-tracking-assignments enabled
24824 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
24825 inside of type units.
24827 2017-01-26 Martin Sebor <msebor@redhat.com>
24829 PR middle-end/78703
24830 * gimple-ssa-sprintf.c (struct result_range): Add likely and
24832 (struct format_result): Replace number_chars, number_chars_min,
24833 and number_chars_max with a single member of struct result_range.
24835 (format_result::operator+=): Adjust.
24836 (struct fmtresult): Remove bounded. Handle likely and unlikely
24838 (fmtresult::adjust_for_width_or_precision): New function.
24839 (fmtresult:type_max_digits): New function.
24840 (bytes_remaining): Handle likely and unlikely counters.
24841 (min_bytes_remaining): Remove.
24842 (format_percent): Simplify.
24843 (format_integer, format_floating): Set likely and unlikely counters.
24844 (get_string_length, format_character, format_string): Same.
24845 (format_plain, should_warn_p): New function.
24846 (maybe_warn): Call should_warn_p. Update diagnostic messages
24847 and handle those for all directives, including plain strings.
24848 (format_directive): Handle likely and unlikely counters.
24849 Remove unnecessary quoting from diagnostics. Add an informational
24851 (add_bytes): Remove.
24852 (pass_sprintf_length::compute_format_length): Simplify.
24853 (try_substitute_return_value): Handle likely and unlikely counters.
24855 2017-01-26 Carl Love <cel@us.ibm.com>
24857 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
24858 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
24860 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
24863 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
24864 endianess for subregs into account.
24865 * lra-constraints.c (lra_constraints): Do risky transformations
24866 always on the first iteration.
24867 * lra-lives.c (check_pseudos_live_through_calls): Add arg
24868 last_call_used_reg_set.
24869 (process_bb_lives): Define and use last_call_used_reg_set.
24870 * lra.c (lra): Always continue after lra_constraints on the first
24873 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
24875 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
24877 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
24879 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24881 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
24882 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
24883 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
24884 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
24885 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
24886 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
24887 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
24888 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
24889 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
24891 2017-01-26 Marek Polacek <polacek@redhat.com>
24894 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
24895 for the third operand.
24897 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24899 PR middle-end/79236
24900 * omp-low.c (struct omp_context): Add simt_stmt field.
24901 (scan_omp_for): Return omp_context *.
24902 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
24903 context to the _simt_ SIMD stmt.
24904 (lower_omp_for): For combined SIMD with sibling _simt_
24905 SIMD, make sure to use the same decls in _looptemp_
24906 clauses as in the sibling.
24908 2017-01-26 David Sherwood <david.sherwood@arm.com>
24910 PR middle-end/79212
24911 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
24914 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24917 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
24918 emit fld b; fld a; if possible.
24920 * brig-builtins.def: Update copyright years.
24921 * config/arm/arm_acle_builtins.def: Update copyright years.
24923 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
24926 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
24927 constraint instead of o for the stxsd instruction.
24929 2017-01-25 Carl Love <cel@us.ibm.com>
24931 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
24932 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
24934 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
24936 * doc/invoke.texi (C++ Dialect Options): Fix typo.
24938 2017-01-25 Richard Biener <rguenther@suse.de>
24940 PR tree-optimization/69264
24941 * target.def (vector_alignment_reachable): Improve documentation.
24942 * doc/tm.texi: Regenerate.
24943 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
24945 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
24946 earlier changes with respect to TYPE_USER_ALIGN.
24947 (vector_alignment_reachable_p): Likewise. Improve dumping.
24949 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24952 * config/arm/arm.md (xordi3): Force constant operand into a register
24955 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24957 * doc/invoke.texi (-fstore-merging): Correct default optimization
24958 levels at which it is enabled.
24959 (-O): Move -fstore-merging from list to...
24962 2017-01-25 Richard Biener <rguenther@suse.de>
24965 * omp-expand.c: Include debug.h.
24966 (expand_omp_taskreg): Make sure to generate early debug before
24967 outlining anything from a function.
24968 (expand_omp_target): Likewise.
24969 (grid_expand_target_grid_body): Likewise.
24971 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
24974 * asan.c (get_translation_unit_decl): New function.
24975 (asan_add_global): Extract modules file name from globals
24976 TRANSLATION_UNIT_DECL name.
24978 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
24981 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
24982 for long calls with APCS frame and VFP.
24984 2017-01-24 David Malcolm <dmalcolm@redhat.com>
24986 * cfg.c (original_copy_tables_initialized_p): New function.
24987 * cfg.h (original_copy_tables_initialized_p): New decl.
24988 * cfgrtl.c (relink_block_chain): Guard the call to
24989 free_original_copy_tables with a call to
24990 original_copy_tables_initialized_p.
24991 * cgraph.h (symtab_node::native_rtl_p): New decl.
24992 * cgraphunit.c (symtab_node::native_rtl_p): New function.
24993 (symtab_node::needed_p): Don't assert for early assembly output
24994 for __RTL functions.
24995 (cgraph_node::finalize_function): Set "force_output" for __RTL
24997 (cgraph_node::analyze): Bail out early for __RTL functions.
24998 (analyze_functions): Update assertion to support __RTL functions.
24999 (cgraph_node::expand): Bail out early for __RTL functions.
25000 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
25002 * function.h (struct function): Update comment for field
25004 * gimple-expr.c: Include "tree-pass.h".
25005 (gimple_has_body_p): Return false for __RTL functions.
25006 * Makefile.in (OBJS): Add run-rtl-passes.o.
25007 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
25009 (gcc::pass_manager::get_clean_slate): New accessor.
25010 * passes.c: Include "insn-addr.h".
25011 (should_skip_pass_p): Add logging. Update logic for running
25012 "expand" to be compatible with both __GIMPLE and __RTL. Guard
25013 property-provider override so it is only done for gimple passes.
25015 (skip_pass): New function.
25016 (execute_one_pass): Call skip_pass when skipping passes.
25017 * read-md.c (md_reader::read_char): Support filtering
25018 the input to a subset of line numbers.
25019 (md_reader::md_reader): Initialize fields
25020 m_first_line and m_last_line.
25021 (md_reader::read_file_fragment): New function.
25022 * read-md.h (md_reader::read_file_fragment): New decl.
25023 (md_reader::m_first_line): New field.
25024 (md_reader::m_last_line): New field.
25025 * read-rtl-function.c (function_reader::create_function): Only
25026 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
25027 curr_properties. Set DECL_INITIAL to a dummy block.
25028 (read_rtl_function_body_from_file_range): New function.
25029 * read-rtl-function.h (read_rtl_function_body_from_file_range):
25031 * run-rtl-passes.c: New file.
25032 * run-rtl-passes.h: New file.
25034 2017-01-24 Jeff Law <law@redhat.com>
25036 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
25039 2017-01-24 Bin Cheng <bin.cheng@arm.com>
25041 PR tree-optimization/79159
25042 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
25043 (record_nonwrapping_iv): Improve boundary using above function if no
25044 value range information.
25046 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
25047 Martin Jambor <mjambor@suse.cz>
25049 * brig-builtins.def: New file.
25050 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
25051 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
25052 (DEF_HSAIL_SAT_BUILTIN): Likewise.
25053 (DEF_HSAIL_INTR_BUILTIN): Likewise.
25054 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
25055 * builtin-types.def (BT_INT8): New.
25056 (BT_INT16): Likewise.
25057 (BT_UINT8): Likewise.
25058 (BT_UINT16): Likewise.
25059 (BT_FN_ULONG): Likewise.
25060 (BT_FN_UINT_INT): Likewise.
25061 (BT_FN_UINT_ULONG): Likewise.
25062 (BT_FN_UINT_LONG): Likewise.
25063 (BT_FN_UINT_PTR): Likewise.
25064 (BT_FN_ULONG_PTR): Likewise.
25065 (BT_FN_INT8_FLOAT): Likewise.
25066 (BT_FN_INT16_FLOAT): Likewise.
25067 (BT_FN_UINT32_FLOAT): Likewise.
25068 (BT_FN_UINT16_FLOAT): Likewise.
25069 (BT_FN_UINT8_FLOAT): Likewise.
25070 (BT_FN_UINT64_FLOAT): Likewise.
25071 (BT_FN_UINT16_UINT32): Likewise.
25072 (BT_FN_UINT32_UINT16): Likewise.
25073 (BT_FN_UINT16_UINT16_UINT16): Likewise.
25074 (BT_FN_INT_PTR_INT): Likewise.
25075 (BT_FN_UINT_PTR_UINT): Likewise.
25076 (BT_FN_LONG_PTR_LONG): Likewise.
25077 (BT_FN_ULONG_PTR_ULONG): Likewise.
25078 (BT_FN_VOID_UINT64_UINT64): Likewise.
25079 (BT_FN_UINT8_UINT8_UINT8): Likewise.
25080 (BT_FN_INT8_INT8_INT8): Likewise.
25081 (BT_FN_INT16_INT16_INT16): Likewise.
25082 (BT_FN_INT_INT_INT): Likewise.
25083 (BT_FN_UINT_FLOAT_UINT): Likewise.
25084 (BT_FN_FLOAT_UINT_UINT): Likewise.
25085 (BT_FN_ULONG_UINT_UINT): Likewise.
25086 (BT_FN_ULONG_UINT_PTR): Likewise.
25087 (BT_FN_ULONG_ULONG_ULONG): Likewise.
25088 (BT_FN_UINT_UINT_UINT): Likewise.
25089 (BT_FN_VOID_UINT_PTR): Likewise.
25090 (BT_FN_UINT_UINT_PTR: Likewise.
25091 (BT_FN_UINT32_UINT64_PTR): Likewise.
25092 (BT_FN_INT_INT_UINT_UINT): Likewise.
25093 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
25094 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
25095 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
25096 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
25097 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
25098 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
25099 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
25100 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
25101 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
25102 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
25103 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
25104 * doc/frontends.texi: List BRIG FE.
25105 * doc/install.texi (Testing): Add BRIG tesring requirements.
25106 * doc/invoke.texi (Overall Options): Mention BRIG.
25107 * doc/standards.texi (Standards): Doucment BRIG HSA version.
25109 2017-01-24 Richard Biener <rguenther@suse.de>
25111 PR translation/79208
25112 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
25114 2017-01-24 Martin Jambor <mjambor@suse.cz>
25117 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
25118 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
25119 and known_contexts.
25121 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
25123 PR middle-end/79123
25124 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
25125 casts from signed to unsigned really don't have a range.
25127 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
25129 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
25130 GMP_RNDx for compatiblity.
25132 2017-01-24 Martin Liska <mliska@suse.cz>
25135 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
25136 that would prevent us to call alloca with -1 as argument.
25138 2017-01-24 Jakub Jelinek <jakub@redhat.com>
25140 * dwarf2out.c (output_compilation_unit_header, output_file_names):
25141 Avoid -Wformat-security warning.
25143 2017-01-23 Andrew Pinski <apinski@cavium.com>
25145 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
25148 2017-01-23 Martin Sebor <msebor@redhat.com>
25150 PR middle-end/78703
25151 * gimple-ssa-sprintf.c (warn_level): New global.
25152 (format_integer): Use it here and throughout the rest of the file.
25153 Use the same switch to compute sign as base.
25154 (maybe_warn): New function.
25155 (format_directive): Factor out warnings into maybe_warn.
25156 Add debugging output. Use warn_level.
25157 (add_bytes): Use warn_level.
25158 (pass_sprintf_length::compute_format_length): Add debugging output.
25159 (try_substitute_return_value): Same.
25160 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
25162 PR middle-end/78703
25163 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
25164 (struct fmtresult, format_integer, format_floating): Adjust.
25165 (fmtresult::fmtresult): Set max correctly in two argument ctor.
25166 (get_string_length, format_string,format_directive): Same.
25167 (pass_sprintf_length::compute_format_length): Same.
25168 (try_substitute_return_value): Simplify slightly.
25170 PR middle-end/78703
25171 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
25172 (fmtresult::operator+=): Outlined.
25173 (struct fmtresult): Add ctors.
25174 (struct conversion_spec): Rename...
25175 (struct directive): ...to this. Add and remove data members.
25176 (directive::set_width, directive::set_precision): New functions.
25177 (format_percent): Use fmtresult ctor.
25178 (get_width_and_precision): Remove.
25179 (format_integer): Make naming changes. Avoid computing width and
25181 (format_floating): Same. Adjust indentation.
25182 (format_character, format_none): New functions.
25183 (format_string): Moved character handling to format_character.
25184 (format_directive): Remove arguments, change return type.
25185 (parse_directive): New function.
25186 (pass_sprintf_length::compute_format_length): Move directive
25187 parsing to parse_directive.
25189 2017-01-23 Jakub Jelinek <jakub@redhat.com>
25191 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
25192 (assign_assembler_name_if_needed): ... this.
25193 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
25194 (assign_assembler_name_if_needed): ... this.
25195 (free_lang_data_in_cgraph): Adjust callers.
25196 * cgraphunit.c (cgraph_node::analyze): Likewise.
25197 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
25200 2017-01-23 Richard Biener <rguenther@suse.de>
25202 PR tree-optimization/79088
25203 PR tree-optimization/79188
25204 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
25205 resetting loop bounds after last path deletion. Reset loop
25206 bounds of the target loop, make code match the comments.
25207 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
25208 Make sure loops need no fixups.
25210 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
25212 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
25213 exponent support with double type for first argument.
25214 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
25215 type returned by __builtin_vec_extract_sig,
25216 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
25217 functions from "vector int" to "vector unsigned int" or from
25218 "vector long long int" to "vector unsigned long long int".
25219 Changed type returned by __builtin_vec_extract_exp,
25220 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
25221 functions from "vector int" to "vector unsigned int" or from
25222 "vector long long int" to "vector unsigned long long int".
25223 Changed return type of __builtin_vec_test_data_class,
25224 __builtin_vec_test_data_class_sp, and
25225 __builtin_vec_test_data_class_dp from "vector int" to
25226 "vector bool int" or from "vector long long int" to "vector bool
25227 long long int" and changed second argument type from "unsigned
25228 int" to "int". Added new overloaded function forms "vector float
25229 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
25230 "vector float __builtin_vec_insert_exp_sp (vector float, vector
25231 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
25232 double, vector unsigned long long int)" and "vector double
25233 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
25234 long int)". Changed return type of
25235 __builtin_scalar_test_data_class and
25236 __builtin_scalar_test_data_class_sp and
25237 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
25238 int" and changed second argument from "unsigned int" to "int".
25239 Changed type returned by __builtin_scalar_test_neg,
25240 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
25241 from "int" to "bool int". Added new overloaded function form
25242 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
25243 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
25244 exponent double-precision with floating point first argument.
25245 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
25246 documentation of scalar_test_data_class, scalar_test_neg,
25247 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
25248 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
25249 vec_test_data_class built-in functions to reflect refinements in
25250 their type signatures.
25252 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
25254 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
25256 (aarch64_elf_asm_destructor): Likewise.
25258 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
25260 PR rtl-optimization/78634
25261 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
25262 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
25263 * ifcvt.c (noce_try_cmove): Add missing cost check.
25265 PR rtl-optimization/71724
25266 * combine.c (if_then_else_cond): Look for situations where it is
25267 beneficial to undo the work of one of the recursive calls.
25269 2017-01-23 Bin Cheng <bin.cheng@arm.com>
25271 PR tree-optimization/70754
25272 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
25273 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
25274 combined stmt before it if not NULL.
25275 (combine_chains): Process refs reversely and compute dominance point
25278 2017-01-23 Martin Liska <mliska@suse.cz>
25280 PR tree-optimization/79196
25281 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
25282 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
25284 (strlen_optimize_stmt): Call the renamed function.
25286 2017-01-23 Michael Matz <matz@suse.de>
25288 PR tree-optimization/78384
25289 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
25291 2017-01-23 Richard Biener <rguenther@suse.de>
25293 PR tree-optimization/79186
25294 * tree-vrp.c (register_new_assert_for): Make sure we've seen
25295 both incoming edges before moving an assert.
25297 2017-01-23 Martin Jambor <mjambor@suse.cz>
25299 * ipa-prop.c (load_from_param_1): Removed.
25300 (load_from_unmodified_param): Bits from load_from_param_1 put back
25302 (load_from_param): Removed.
25303 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
25304 with stmt. Reverted back to use of load_from_unmodified_param.
25306 2017-01-23 Martin Jambor <mjambor@suse.cz>
25309 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
25310 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
25311 field a pointer to garbage collected vector, mark lattices and
25312 ipcp_orig_node with GTY((skip)).
25313 (ipa_get_param_count): Adjust to descriptors being a pointer.
25314 (ipa_get_param): Likewise.
25315 (ipa_get_type): Likewise.
25316 (ipa_get_param_move_cost): Likewise.
25317 (ipa_set_param_used): Likewise.
25318 (ipa_get_controlled_uses): Likewise.
25319 (ipa_set_controlled_uses): Likewise.
25320 (ipa_is_param_used): Likewise.
25321 (ipa_node_params_t): Move into garbage collector. New methods insert
25323 (ipa_node_params_sum): Annotate wth GTY(()).
25324 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
25326 (ipa_load_from_parm_agg): Adjust declaration.
25327 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
25328 * ipa-profile.c (ipa_profile): Likewise.
25329 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
25330 (ipa_populate_param_decls): Make descriptors parameter garbage
25332 (ipa_dump_param): Adjust to descriptors being a pointer.
25333 (ipa_alloc_node_params): Likewise.
25334 (ipa_initialize_node_params): Likewise.
25335 (load_from_param_1): Make descriptors parameter garbage collected.
25336 (load_from_unmodified_param): Likewise.
25337 (load_from_param): Likewise.
25338 (ipa_load_from_parm_agg): Likewise.
25339 (ipa_node_params::~ipa_node_params): Removed.
25340 (ipa_free_all_node_params): Remove call to delete operator.
25341 (ipa_node_params_t::insert): New.
25342 (ipa_node_params_t::remove): Likewise.
25343 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
25344 copy known_csts and known_contexts vectors.
25345 (ipa_read_node_info): Adjust to descriptors being a pointer.
25346 (ipcp_modif_dom_walker): Make m_descriptors field garbage
25348 (ipcp_transform_function): Make descriptors variable garbage
25351 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
25353 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
25354 * config/i386/avx512dqintrin.h: Ditto.
25355 * config/i386/avx512fintrin.h: Ditto.
25356 * config/i386/i386.c: Handle new builtins.
25357 * config/i386/i386-builtin.def: Add new builtins.
25358 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
25359 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
25361 2017-01-23 Jakub Jelinek <jakub@redhat.com>
25362 Martin Liska <mliska@suse.cz>
25364 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
25365 * asan.c (asan_expand_poison_ifn): Support stores and use
25366 appropriate ASAN report function.
25367 * internal-fn.c (expand_ASAN_POISON_USE): New function.
25368 * internal-fn.def (ASAN_POISON_USE): Declare.
25369 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
25370 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
25371 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
25372 ASAN_POISON calls w/o LHS.
25373 * tree-ssa.c (execute_update_addresses_taken): Create clobber
25374 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
25375 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
25376 * gimplify.c (asan_poison_variables): Add attribute
25377 use_after_scope_memory to variables that really needs to live
25379 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
25380 having the attribute.
25382 2017-01-23 Martin Liska <mliska@suse.cz>
25384 * asan.c (create_asan_shadow_var): New function.
25385 (asan_expand_poison_ifn): Likewise.
25386 * asan.h (asan_expand_poison_ifn): New declaration.
25387 * internal-fn.c (expand_ASAN_POISON): Likewise.
25388 * internal-fn.def (ASAN_POISON): New builtin.
25389 * sanopt.c (pass_sanopt::execute): Expand
25390 asan_expand_poison_ifn.
25391 * tree-inline.c (copy_decl_for_dup_finish): Make function
25393 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
25394 * tree-ssa.c (is_asan_mark_p): New function.
25395 (execute_update_addresses_taken): Rewrite local variables
25396 (identified just by use-after-scope as addressable) into SSA.
25398 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
25400 * doc/install.texi (Specific): opensource.apple.com uses https
25401 now. Remove trailing slash.
25403 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
25405 * README.Portability: Remove note on an Irix compatibility issue.
25407 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
25409 * gcov.c (INCLUDE_ALGORITHM): Define.
25410 (INCLUDE_VECTOR): Define.
25411 No longer include <vector> and <algorithm> directly.
25413 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
25415 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
25417 * doc/invoke.texi (Code Gen Options): Ditto.
25419 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
25422 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
25424 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
25426 rtl-optimization/79125
25427 * cprop.c (local_cprop_pass): Handle cases where we make an
25428 unconditional trap.
25430 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
25434 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
25435 read from, for big endian.
25437 2017-01-20 Jiong Wang <jiong.wang@arm.com>
25439 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
25440 register pauth builtins for LP64 only.
25442 2017-01-20 Marek Polacek <polacek@redhat.com>
25445 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
25448 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
25450 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
25452 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
25453 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
25454 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
25456 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25459 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
25460 in big-endian mode when they are not a single duplicated value.
25462 2017-01-20 Richard Biener <rguenther@suse.de>
25464 * BASE-VER: Bump to 7.0.1.
25466 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
25468 * omp-low.c (omplow_simd_context): New struct. Use it...
25469 (lower_rec_simd_input_clauses): ...here and...
25470 (lower_rec_input_clauses): ...here to hold common data. Adjust all
25471 references to idx, lane, max_vf, is_simt.
25473 2017-01-20 Graham Markall <graham.markall@embecosm.com>
25475 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
25478 2017-01-20 Martin Jambor <mjambor@suse.cz>
25480 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
25481 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
25483 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
25484 (GTFILES): Rename hsa.c to hsa-common.c.
25485 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
25486 * hsa-dump.c: Likewise.
25487 * hsa-gen.c: Likewise.
25488 * hsa-regalloc.c: Likewise.
25489 * ipa-hsa.c: Likewise.
25490 * omp-expand.c: Likewise.
25491 * omp-low.c: Likewise.
25492 * toplev.c: Likewise.
25494 2017-01-20 Marek Polacek <polacek@redhat.com>
25497 * doc/invoke.texi: Document -Wduplicated-branches.
25498 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
25499 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
25500 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
25501 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
25502 return 0 only when not OEP_LEXICOGRAPHIC.
25503 (fold_build_cleanup_point_expr): Use the expression
25504 location when building CLEANUP_POINT_EXPR.
25505 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
25506 * tree.c (add_expr): Handle error_mark_node.
25508 2017-01-20 Martin Liska <mliska@suse.cz>
25511 * tree-profile.c (init_ic_make_global_vars): Do not call
25513 (gimple_init_gcov_profiler): Likewise.
25515 2017-01-20 Martin Liska <mliska@suse.cz>
25518 * cgraph.h (maybe_create_reference): Remove argument and
25520 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
25522 * ipa-cp.c (create_specialized_node): Likewise.
25523 * symtab.c (symtab_node::maybe_create_reference): Handle
25524 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
25526 2017-01-20 Martin Liska <mliska@suse.cz>
25528 * read-rtl-function.c (function_reader::create_function): Use
25529 build_decl instread of build_decl_stat.
25531 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
25533 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
25534 * config/i386/avx512dqintrin.h: Ditto.
25535 * config/i386/avx512fintrin.h: Ditto.
25536 * config/i386/i386-builtin-types.def: Add new types.
25537 * config/i386/i386.c: Handle new types.
25538 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
25539 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
25540 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
25541 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
25542 (__builtin_ia32_kshiftridi): New.
25543 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
25545 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
25549 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
25550 define to rs6000_init_stack_protect_guard.
25551 (rs6000_init_stack_protect_guard): New function.
25553 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
25554 Yunqiang Su <yunqiang.su@imgtec.com>
25556 * config.gcc (supported_defaults): Add madd4.
25557 (with_madd4): Add validation.
25558 (all_defaults): Add madd4.
25559 * config/mips/mips.opt (mmadd4): New option.
25560 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
25562 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
25564 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
25565 (ISA_HAS_FUSED_MADD4): Likewise.
25566 * doc/invoke.texi (-mmadd4): Document the new option.
25567 * doc/install.texi (--with-madd4): Document the new option.
25569 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25571 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
25572 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
25573 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
25574 (aarch64_init_pauth_hint_builtins): New.
25575 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
25576 (aarch64_expand_builtin): Expand new builtins.
25578 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25580 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
25581 * combine-stack-adj.c (no_unhandled_cfa): Handle
25582 REG_CFA_TOGGLE_RA_MANGLE.
25583 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
25584 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
25585 info for return address signing.
25586 (aarch64_expand_epilogue): Likewise.
25588 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25590 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
25591 * config/aarch64/aarch64-protos.h
25592 (aarch64_return_address_signing_enabled): New declaration.
25593 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
25595 (aarch64_expand_prologue): Sign return address before it's pushed onto
25597 (aarch64_expand_epilogue): Authenticate return address fetched from
25599 (aarch64_override_options): Sanity check for ILP32 and ISA level.
25600 (aarch64_attributes): New function attributes for "sign-return-address".
25601 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
25602 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
25603 ("*do_return"): Generate combined instructions according to key index.
25604 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
25605 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
25607 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
25608 * config/aarch64/aarch64.opt (msign-return-address=): New.
25609 * doc/extend.texi (AArch64 Function Attributes): Documents
25610 "sign-return-address=".
25611 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
25613 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
25615 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
25616 overall option summary.
25618 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25620 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
25621 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
25622 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
25623 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
25625 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
25627 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
25628 -mpower9-minmax by default for -mcpu=power9.
25629 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
25630 128-bit floating point.
25632 2017-01-20 Alan Modra <amodra@gmail.com>
25634 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
25635 optimizing for size.
25637 2017-01-20 Alan Modra <amodra@gmail.com>
25640 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
25641 for strcmp and strncmp from corresponding builtin decl.
25643 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
25645 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
25646 instead of i386/rtems-64.h.
25647 * config/i386/rtems-64.h: Remove.
25649 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
25653 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
25655 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
25657 2017-01-19 Tamar Christina <tamar.christina@arm.com>
25659 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
25660 Change int to HOST_WIDE_INT.
25661 * config/aarch64/aarch64-protos.h
25662 (aarch64_simd_gen_const_vector_dup): Likewise.
25663 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
25665 2017-01-19 David Malcolm <dmalcolm@redhat.com>
25667 * langhooks-def.h (lhd_type_for_size): New decl.
25668 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
25669 * langhooks.c (lhd_type_for_size): New function, taken from
25672 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
25674 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
25675 define_bypass for CR latency.
25676 (power9-cracked-alu): Update bypass latency and remove power9-branch.
25677 (power9-alu2): Add define_bypass for CR latency.
25679 (power9-mul): Update insn latency.
25680 (power9-mul-compare): Update insn latency, bypass latency and remove
25683 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25685 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
25687 * config/aarch64/aarch64.md
25688 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
25689 aarch64_nopcrelative_literal_loads.
25690 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
25692 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
25694 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
25695 TARGET_LOONGSON_3A.
25696 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
25698 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
25701 * config.gcc (supported_defaults): Add lxc1-sxc1.
25702 (with_lxc1_sxc1): Add validation.
25703 (all_defaults): Add lxc1-sxc1.
25704 * config/mips/mips.opt (mlxc1-sxc1): New option.
25705 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
25707 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
25708 __mips_no_lxc1_sxc1.
25709 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
25710 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
25711 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
25713 2017-01-19 Richard Biener <rguenther@suse.de>
25715 PR tree-optimization/72488
25716 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
25717 sure to restore SSA info.
25718 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
25720 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
25722 PR rtl-optimization/79121
25723 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
25724 of the inner type when shifting an extended value.
25726 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
25729 * symtab.c (symtab_node::equal_address_to): Fix comparing of
25730 interposable aliases.
25732 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
25735 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
25736 Use the evmergelohi instruction.
25737 (mov_si<mode>_e500_subreg4_2_le): Likewise.
25738 (mov_sitf_e500_subreg8_2_be): Likewise.
25739 (mov_sitf_e500_subreg12_2_le): Likewise.
25740 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
25741 (mov_si<mode>_e500_subreg4_2_be): Likewise.
25742 (mov_sitf_e500_subreg8_2_le): Likewise.
25743 (mov_sitf_e500_subreg12_2_be): Likewise.
25745 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25747 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
25748 attribute from vecsimple to vecperm.
25749 (altivec_vbpermq2): Likewise.
25751 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25754 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
25756 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25757 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
25758 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
25759 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
25760 case where N arg is SIZE_MAX.
25761 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
25762 (cmpstrsi): Add pattern.
25764 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25766 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25767 __builtin_vec_revb builtins.
25768 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
25769 built-in functions to support generation of the ISA 3.0 XXBR<x>
25770 vector byte reverse instructions.
25771 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
25772 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
25773 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
25774 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
25775 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
25776 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
25777 (P9V_BUILTIN_VEC_REVB): Likewise.
25778 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
25779 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
25780 (p9_xxbrq_v16qi): Likewise.
25781 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
25782 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
25783 (p9_xxbrh_v8hi): Likewise.
25784 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
25785 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
25786 vec_revb built-in functions.
25788 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
25790 PR rtl-optimization/78952
25791 * config/i386/i386.md (any_extract): New code iterator.
25792 (*insvqi_2): Use any_extract for source operand.
25793 (*insvqi_3): Use any_shiftrt for source operand.
25795 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
25797 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
25799 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
25801 2017-01-18 Matthias Klose <doko@ubuntu.com>
25803 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
25805 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25807 * config/rs6000/altivec.h (vec_bperm): Change #define.
25808 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
25809 (altivec_vbpermq2): New define_insn.
25810 (altivec_vbpermd): Likewise.
25811 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
25812 function interface.
25813 (VBPERMD): Likewise.
25814 (VBPERM): New polymorphic function interface.
25815 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
25816 Add entries for P9V_BUILTIN_VEC_VBPERM.
25817 * doc/extend.texi: Add interfaces for vec_bperm.
25819 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25821 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
25822 first letter of error messages.
25823 (s390_resolve_overloaded_builtin): Likewise.
25824 * config/s390/s390.c (s390_expand_builtin): Likewise.
25825 (s390_invalid_arg_for_unprototyped_fn): Likewise.
25826 (s390_valid_target_attribute_inner_p): Likewise.
25827 * config/s390/s390.md ("tabort"): Likewise.
25829 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
25831 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
25832 (ISA_AVOID_DIV_HILO): New macro.
25833 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
25834 (ISA_HAS_DDIV): Likewise.
25836 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25838 * doc/invoke.texi (fabi-version): Correct number of occurrences.
25840 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25842 * doc/invoke.texi (fabi-version): Spelling fix.
25844 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25847 * doc/invoke.texi (fabi-version): Mention mangling fix for
25850 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25853 * doc/invoke.texi (fabi-version): Document discriminator mangling.
25855 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
25858 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
25859 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
25861 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
25863 (stack_protect_test): Ditto.
25864 * config/rs6000/rs6000.opt (mstack-protector-guard=,
25865 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
25867 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
25868 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
25869 -mstack-protector-guard-offset=.
25870 (RS/6000 and PowerPC Options): Ditto.
25872 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
25874 * config/i386/i386.h (MASK_CLASS_P): New define.
25875 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
25876 there are no registers from different register sets also when
25877 mask registers are used. Update function comment.
25878 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
25879 to (*k/*r) and (*k/*km) alternatives.
25881 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
25883 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
25884 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
25885 (EH_RETURN_HANDLER_RTX): New define.
25886 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
25887 Force frame pointer in EH return functions.
25888 (aarch64_expand_epilogue): Add barrier for eh_return.
25889 (aarch64_final_eh_return_addr): Remove.
25890 (aarch64_eh_return_handler_rtx): New function.
25891 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
25893 (aarch64_eh_return_handler_rtx): New prototype.
25895 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25897 * config/rs6000/altivec.h (vec_rlmi): New #define.
25898 (vec_vrlnm): Likewise.
25899 (vec_rlnm): Likewise.
25900 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
25901 (UNSPEC_VRLNM): Likewise.
25902 (VIlong): New mode iterator.
25903 (altivec_vrl<VI_char>mi): New define_insn.
25904 (altivec_vrl<VI_char>nm): Likewise.
25905 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
25907 (VRLDNM): Likewise.
25908 (RLNM): New polymorphic function entry.
25909 (VRLWMI): New monomorphic function entry.
25910 (VRLDMI): Likewise.
25911 (RLMI): New polymorphic function entry.
25912 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
25913 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
25914 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
25917 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25920 * dwarf2out.c (field_byte_offset): Restore the
25921 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
25922 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
25923 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
25926 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
25929 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
25931 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25934 * dwarf2out.c (add_data_member_location_attribute): For constant
25935 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
25936 instead of DW_AT_data_member_location, DW_AT_bit_offset and
25937 DW_AT_byte_size attributes.
25939 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
25941 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
25942 after forcing to constant memory when the code model is medium.
25944 2017-01-17 Julia Koval <julia.koval@intel.com>
25947 * config/i386/avx512fintrin.h
25948 (_mm512_i32gather_ps): Change __addr type to void const*.
25949 (_mm512_mask_i32gather_ps): Ditto.
25950 (_mm512_i32gather_pd): Ditto.
25951 (_mm512_mask_i32gather_pd): Ditto.
25952 (_mm512_i64gather_ps): Ditto.
25953 (_mm512_mask_i64gather_ps): Ditto.
25954 (_mm512_i64gather_pd): Ditto.
25955 (_mm512_mask_i64gather_pd): Ditto.
25956 (_mm512_i32gather_epi32): Ditto.
25957 (_mm512_mask_i32gather_epi32): Ditto.
25958 (_mm512_i32gather_epi64): Ditto.
25959 (_mm512_mask_i32gather_epi64): Ditto.
25960 (_mm512_i64gather_epi32): Ditto.
25961 (_mm512_mask_i64gather_epi32): Ditto.
25962 (_mm512_i64gather_epi64): Ditto.
25963 (_mm512_mask_i64gather_epi64): Ditto.
25964 (_mm512_i32scatter_ps): Change __addr type to void*.
25965 (_mm512_mask_i32scatter_ps): Ditto.
25966 (_mm512_i32scatter_pd): Ditto.
25967 (_mm512_mask_i32scatter_pd): Ditto.
25968 (_mm512_i64scatter_ps): Ditto.
25969 (_mm512_mask_i64scatter_ps): Ditto.
25970 (_mm512_i64scatter_pd): Ditto.
25971 (_mm512_mask_i64scatter_pd): Ditto.
25972 (_mm512_i32scatter_epi32): Ditto.
25973 (_mm512_mask_i32scatter_epi32): Ditto.
25974 (_mm512_i32scatter_epi64): Ditto.
25975 (_mm512_mask_i32scatter_epi64): Ditto.
25976 (_mm512_i64scatter_epi32): Ditto.
25977 (_mm512_mask_i64scatter_epi32): Ditto.
25978 (_mm512_i64scatter_epi64): Ditto.
25979 (_mm512_mask_i64scatter_epi64): Ditto.
25980 * config/i386/avx512pfintrin.h
25981 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
25982 (_mm512_mask_prefetch_i32gather_ps): Ditto.
25983 (_mm512_mask_prefetch_i64gather_pd): Ditto.
25984 (_mm512_mask_prefetch_i64gather_ps): Ditto.
25985 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
25986 (_mm512_prefetch_i32scatter_ps): Ditto.
25987 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
25988 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
25989 (_mm512_prefetch_i64scatter_pd): Ditto.
25990 (_mm512_prefetch_i64scatter_ps): Ditto.
25991 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
25992 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
25993 * config/i386/avx512vlintrin.h
25994 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
25995 (_mm_mmask_i32gather_ps): Ditto.
25996 (_mm256_mmask_i32gather_pd): Ditto.
25997 (_mm_mmask_i32gather_pd): Ditto.
25998 (_mm256_mmask_i64gather_ps): Ditto.
25999 (_mm_mmask_i64gather_ps): Ditto.
26000 (_mm256_mmask_i64gather_pd): Ditto.
26001 (_mm_mmask_i64gather_pd): Ditto.
26002 (_mm256_mmask_i32gather_epi32): Ditto.
26003 (_mm_mmask_i32gather_epi32): Ditto.
26004 (_mm256_mmask_i32gather_epi64): Ditto.
26005 (_mm_mmask_i32gather_epi64): Ditto.
26006 (_mm256_mmask_i64gather_epi32): Ditto.
26007 (_mm_mmask_i64gather_epi32): Ditto.
26008 (_mm256_mmask_i64gather_epi64): Ditto.
26009 (_mm_mmask_i64gather_epi64): Ditto.
26010 (_mm256_i32scatter_ps): Change __addr type to void*.
26011 (_mm256_mask_i32scatter_ps): Ditto.
26012 (_mm_i32scatter_ps): Ditto.
26013 (_mm_mask_i32scatter_ps): Ditto.
26014 (_mm256_i32scatter_pd): Ditto.
26015 (_mm256_mask_i32scatter_pd): Ditto.
26016 (_mm_i32scatter_pd): Ditto.
26017 (_mm_mask_i32scatter_pd): Ditto.
26018 (_mm256_i64scatter_ps): Ditto.
26019 (_mm256_mask_i64scatter_ps): Ditto.
26020 (_mm_i64scatter_ps): Ditto.
26021 (_mm_mask_i64scatter_ps): Ditto.
26022 (_mm256_i64scatter_pd): Ditto.
26023 (_mm256_mask_i64scatter_pd): Ditto.
26024 (_mm_i64scatter_pd): Ditto.
26025 (_mm_mask_i64scatter_pd): Ditto.
26026 (_mm256_i32scatter_epi32): Ditto.
26027 (_mm256_mask_i32scatter_epi32): Ditto.
26028 (_mm_i32scatter_epi32): Ditto.
26029 (_mm_mask_i32scatter_epi32): Ditto.
26030 (_mm256_i32scatter_epi64): Ditto.
26031 (_mm256_mask_i32scatter_epi64): Ditto.
26032 (_mm_i32scatter_epi64): Ditto.
26033 (_mm_mask_i32scatter_epi64): Ditto.
26034 (_mm256_i64scatter_epi32): Ditto.
26035 (_mm256_mask_i64scatter_epi32): Ditto.
26036 (_mm_i64scatter_epi32): Ditto.
26037 (_mm_mask_i64scatter_epi32): Ditto.
26038 (_mm256_i64scatter_epi64): Ditto.
26039 (_mm256_mask_i64scatter_epi64): Ditto.
26040 (_mm_i64scatter_epi64): Ditto.
26041 (_mm_mask_i64scatter_epi64): Ditto.
26042 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
26043 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
26044 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
26045 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
26046 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
26047 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
26048 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
26049 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
26050 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
26051 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
26052 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
26053 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
26054 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
26055 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
26056 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
26057 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
26058 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
26059 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
26060 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
26061 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
26062 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
26063 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
26064 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
26065 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
26066 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
26067 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
26068 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
26069 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
26070 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
26071 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
26072 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
26073 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
26074 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
26075 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
26076 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
26077 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
26078 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
26079 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
26080 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
26081 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
26082 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
26083 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
26084 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
26085 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
26086 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
26087 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
26088 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
26089 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
26090 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
26091 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
26092 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
26093 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
26094 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
26095 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
26096 definitions accordingly.
26098 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
26099 Kuan-Lin Chen <kuanlinchentw@gmail.com>
26102 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
26105 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
26108 * ira-conflicts.c (ira_build_conflicts): Update total conflict
26109 hard regs for inner regno.
26111 2017-01-17 Martin Liska <mliska@suse.cz>
26114 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
26115 assumption and add comment.
26117 2017-01-17 Nathan Sidwell <nathan@acm.org>
26119 * ipa-visibility.c (localize_node): New function, broken out of ...
26120 (function_and_variable_visibility): ... here. Call it.
26122 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
26124 PR middle-end/77445
26125 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
26126 correctly set frequency of oudgoing edge.
26127 (duplicate_thread_path): Fix profile updating.
26129 2017-01-17 Jakub Jelinek <jakub@redhat.com>
26132 * configure.ac: Add GCC_BASE_VER.
26133 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
26134 version from BASE-VER file.
26135 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
26136 (gcc.o): Depend on $(BASEVER).
26137 * common.opt (dumpfullversion): New option.
26138 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
26139 * doc/invoke.texi: Document -dumpfullversion.
26140 * doc/install.texi: Document --with-gcc-major-version-only.
26141 * configure: Regenerated.
26143 2017-01-17 Richard Biener <rguenther@suse.de>
26145 PR tree-optimization/71433
26146 * tree-vrp.c (register_new_assert_for): Merge same asserts
26147 on all incoming edges.
26148 (process_assert_insertions_for): Handle insertions at the
26151 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
26153 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
26154 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
26156 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
26159 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
26162 2017-01-17 Alan Modra <amodra@gmail.com>
26165 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
26166 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
26167 symbolic stack limit when pic.
26169 2017-01-16 Martin Sebor <msebor@redhat.com>
26171 PR tree-optimization/78608
26172 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
26174 2017-01-16 Jeff Law <law@redhat.com>
26177 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
26178 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
26179 for several include directories that may be relative to sysroot.
26180 * config/i386/x-mingw32 (gplus_includedir): Define.
26181 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
26182 (native_system_includedir): Likewise.
26183 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
26184 override if TARGET_SYSTEM_ROOT is defined.
26185 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
26187 PR tree-optimization/79090
26188 PR tree-optimization/33562
26189 PR tree-optimization/61912
26190 PR tree-optimization/77485
26191 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
26192 and computed trims into the dump file.
26194 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
26196 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
26198 2017-01-16 Jakub Jelinek <jakub@redhat.com>
26201 * gimplify.c (gimplify_init_constructor): If want_value and
26202 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
26206 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
26207 sequence. Formatting fixes.
26208 (doloop_optimize): Formatting fixes.
26211 * gcc.c (debug_level_greater_than_spec_func): New function.
26212 (static_spec_functions): Add debug-level-gt spec function.
26213 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
26215 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
26216 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
26217 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
26218 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
26219 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
26220 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
26222 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
26224 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
26225 QImode fixups to general and mask registers only.
26227 2017-01-16 Carl Love <cel@us.ibm.com>
26229 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
26230 for built-in functions
26231 vector signed char vec_nabs (vector signed char)
26232 vector signed short vec_nabs (vector signed short)
26233 vector signed int vec_nabs (vector signed int)
26234 vector signed long long vec_nabs (vector signed long long)
26235 vector float vec_nabs (vector float)
26236 vector double vec_nabs (vector double)
26237 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
26239 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
26240 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
26241 * doc/extend.texi: Update the documentation file for the new built-in
26244 2017-01-16 Martin Sebor <msebor@redhat.com>
26246 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
26249 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26251 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
26252 UNSPEC_VSX__XXSPLTD to require special splat handling.
26254 2017-01-16 David Malcolm <dmalcolm@redhat.com>
26257 * system.h: Poison strndup.
26259 2017-01-16 Alan Modra <amodra@gmail.com>
26262 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
26265 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
26267 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
26269 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
26271 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
26272 call recog here. Assert that INSN_CODE (insn) is non-negative.
26274 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
26277 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
26279 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
26280 in the currently scheduled RTL fragment.
26282 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
26284 PR rtl-optimization/78751
26285 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
26288 2017-01-14 Jeff Law <law@redhat.com>
26290 PR tree-optimization/79090
26291 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
26292 variable length stores.
26293 (compute_trims): Delete dead assignment to *trim_tail.
26294 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
26297 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
26299 PR rtl-optimization/78626
26300 PR rtl-optimization/78727
26301 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
26302 of a block, and split such blocks after everything else is finished.
26304 2017-01-14 Alan Modra <amodra@gmail.com>
26307 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
26308 target legitimate_combined_insn.
26309 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
26310 (rs6000_legitimate_combined_insn): New function.
26311 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
26313 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
26314 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
26315 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
26317 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
26319 * doc/frontends.texi (G++ and GCC): Remove references to Java.
26321 2017-01-13 Jeff Law <law@redhat.com>
26323 PR tree-optimization/33562
26324 PR tree-optimization/61912
26325 PR tree-optimization/77485
26326 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
26328 (delete_dead_assignment): Likewise.
26329 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
26330 statement to delete_dead_call and delete_dead_assignment.
26332 2017-01-13 David Malcolm <dmalcolm@redhat.com>
26335 * substring-locations.c (format_warning_va): Strengthen case 1 so
26336 that both endpoints of the substring must be within the format
26337 range for just the substring to be printed.
26339 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
26341 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
26342 * config/i386/i386.c (ix86_target_string): Add missing options
26343 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
26344 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
26345 flags_other and ix86_target_other to flags2_other. Display unknown
26347 (ix86_valid_target_attribute_inner_p): Add missing options and
26348 reorder options by implied ISAs, as in ix86_target_string.
26350 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
26352 * hash-table.h (hash_table::too_empty_p): New function.
26353 (hash_table::expand): Use it.
26354 (hash_table::traverse): Likewise.
26355 (hash_table::empty_slot): Use sizeof (value_type) instead of
26356 sizeof (PTR) to convert bytes to elements. Shrink the table
26357 if the current size is excessive for the current number of
26360 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
26362 * ira-costs.c (record_reg_classes): Break from the inner loop
26363 early once alt_fail is known to be true. Update outer loop
26364 handling accordingly.
26366 2017-01-13 Jeff Law <law@redhat.com>
26368 * tree-ssa-dse.c (decrement_count): New function.
26369 (increment_start_addr, maybe_trim_memstar_call): Likewise.
26370 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
26371 when we know the partially dead statement is a mem* function.
26373 PR tree-optimization/61912
26374 PR tree-optimization/77485
26375 * tree-ssa-dse.c: Include expr.h.
26376 (maybe_trim_constructor_store): New function.
26377 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
26379 PR tree-optimization/33562
26380 PR tree-optimization/61912
26381 PR tree-optimization/77485
26382 * doc/invoke.texi: Document new dse-max-object-size param.
26383 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
26384 * tree-ssa-dse.c: Include params.h.
26385 (dse_store_status): New enum.
26386 (initialize_ao_ref_for_dse): New, partially extracted from
26388 (valid_ao_ref_for_dse, normalize_ref): New.
26389 (setup_live_bytes_from_ref, compute_trims): Likewise.
26390 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
26391 (maybe_trim_partially_dead_store): Likewise.
26392 (maybe_trim_complex_store): Likewise.
26393 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
26394 Track what bytes live from the original store. Return tri-state
26395 for dead, partially dead or live.
26396 (dse_dom_walker): Add constructor, destructor and new private members.
26397 (delete_dead_call, delete_dead_assignment): New extracted from
26399 (dse_optimize_stmt): Make a member of dse_dom_walker.
26400 Use initialize_ao_ref_for_dse.
26402 PR tree-optimization/33562
26403 PR tree-optimization/61912
26404 PR tree-optimization/77485
26405 * sbitmap.h (bitmap_count_bits): Prototype.
26406 (bitmap_clear_range, bitmap_set_range): Likewise.
26407 * sbitmap.c (bitmap_clear_range): New function.
26408 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
26410 2017-01-13 Martin Liska <mliska@suse.cz>
26413 * function.c (set_cfun): Add new argument force.
26414 * function.h (set_cfun): Likewise.
26415 * ipa-inline-transform.c (inline_call): Use the function when
26416 strict alising from is dropped for function we inline to.
26418 2017-01-13 Richard Biener <rguenther@suse.de>
26420 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
26421 for dumping GIMPLE INTEGER_CSTs.
26423 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26425 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
26426 to 201112L since C++17.
26428 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
26431 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
26432 if -fsanitize=kernel-address is present.
26434 2017-01-13 Richard Biener <rguenther@suse.de>
26436 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
26437 as _Literal ( type ) number in case usual suffixes do not
26438 preserve all information.
26440 2017-01-13 Richard Biener <rguenther@suse.de>
26442 PR tree-optimization/77283
26443 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
26444 and ssa-iterators.h.
26445 (is_feasible_trace): Implement a cost model based on joiner
26448 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
26451 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
26452 char or short to __float128/_Float128 directly.
26454 2017-01-12 Martin Sebor <msebor@redhat.com>
26456 to -Wformat-overflow.
26457 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
26458 (min_bytes_remaining): Same.
26459 (get_string_length): Same.
26460 (format_string): Same.
26461 (format_directive): Same.
26463 (pass_sprintf_length::handle_gimple_call): Same.
26465 2017-01-12 Jakub Jelinek <jakub@redhat.com>
26467 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
26468 info.nowrite calls with no lhs that can't throw. Return bool
26469 whether gsi_remove has been called or not.
26470 (pass_sprintf_length::handle_gimple_call): Return bool whether
26471 try_substitute_return_value called gsi_remove. Formatting fix.
26472 (pass_sprintf_length::execute): Don't use gsi_remove if
26473 handle_gimple_call returned true.
26476 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
26477 be removed due to side-effects, don't remove following barrier nor
26478 turn the successor edge into fallthru edge.
26480 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26483 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
26484 element-reversing loads and stores as not swappable.
26486 2017-01-12 Nathan Sidwell <nathan@acm.org>
26487 Nicolai Stange <nicstange@gmail.com>
26489 * combine.c (try_combine): Don't ignore result of overlap checking
26490 loop. Combine overlap & asm check into single loop.
26492 2017-01-12 Richard Biener <rguenther@suse.de>
26494 * tree-pretty-print.c (dump_generic_node): Provide -gimple
26495 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
26497 2017-01-12 Richard Biener <rguenther@suse.de>
26499 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
26500 and TS_TARGET_OPTION directly derive from TS_BASE.
26501 * tree-core.h (tree_optimization_option): Derive from tree_base.
26502 (tree_target_option): Likewise.
26504 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
26506 * config/i386/i386.c (memory_address_length): Increase len
26507 only when rip_relative_addr_p returns false.
26509 2017-01-11 Julia Koval <julia.koval@intel.com>
26511 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
26512 (OPTION_MASK_ISA_SGX_SET): New.
26513 (ix86_handle_option): Handle OPT_msgx.
26514 * config.gcc: Added sgxintrin.h.
26515 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
26516 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
26517 * config/i386/i386.c (ix86_target_string): Add -msgx.
26519 (ix86_option_override_internal): Handle new options.
26520 (ix86_valid_target_attribute_inner_p): Add sgx.
26521 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
26522 * config/i386/i386.opt: Add msgx.
26523 * config/i386/sgxintrin.h: New file.
26524 * config/i386/x86intrin.h: Add sgxintrin.h.
26526 2017-01-11 Jakub Jelinek <jakub@redhat.com>
26529 * fold-const.c (maybe_nonzero_address): Return 1 for function
26531 (tree_single_nonzero_warnv_p): Don't handle function local objects
26535 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
26538 2017-01-11 David Malcolm <dmalcolm@redhat.com>
26541 * opts.c: Include "spellcheck.h"
26542 (struct string_fragment): New struct.
26543 (struct edit_distance_traits<const string_fragment &>): New
26545 (get_closest_sanitizer_option): New function.
26546 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
26548 2017-01-11 Jakub Jelinek <jakub@redhat.com>
26550 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
26552 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
26553 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
26554 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
26555 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
26556 for initial die_offset if dwarf_split_debug_info.
26557 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
26558 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
26560 (output_skeleton_debug_sections): Formatting fix. Use
26561 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
26562 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
26564 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
26566 * config/arm/cortex-a53.md: Add bypasses for
26567 cortex_a53_r2f_cvt.
26568 (cortex_a53_r2f): Only use for transfers.
26569 (cortex_a53_f2r): Likewise.
26570 (cortex_a53_r2f_cvt): Add reservation for conversions.
26571 (cortex_a53_f2r_cvt): Likewise.
26573 2017-01-11 Tamar Christina <tamar.christina@arm.com>
26575 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
26576 to all inlined functions, change static to extern.
26578 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
26581 * config/arm/arm.c (legitimize_pic_address): Handle reference to
26583 (arm_assemble_integer): Likewise.
26585 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
26587 * config.gcc: Use new awk script to check CPU, FPU and architecture
26588 parameters for --with-... options.
26589 * config/arm/parsecpu.awk: New file
26590 * config/arm/arm-cpus.in: New file.
26591 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
26593 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
26595 * config/arm/t-arm: Update dependency rules.
26596 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
26597 of processing .def files.
26598 * config/arm/genopt.sh: Deleted.
26599 * config/arm/gentune.sh: Deleted.
26600 * config/arm/arm-cores.def: Deleted.
26601 * config/arm/arm-arches.def: Deleted.
26602 * config/arm/arm-fpus.def: Deleted.
26603 * config/arm/arm-tune.md: Regenerated.
26604 * config/arm/arm-tables.opt: Regenerated.
26605 * config/arm/arm-cpu.h: New generated file.
26606 * config/arm/arm-cpu-data.h: New generated file.
26607 * config/arm/arm-cpu-cdata.h: New generated file.
26609 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
26612 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
26614 (input_varpool_node): Unpack dynamically_initialized bit.
26616 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
26618 PR rtl-optimization/79032
26619 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
26620 the alignment of the adjusted memory reference against that of MODE,
26621 instead of the alignment of the original memory reference.
26623 2017-01-11 Martin Jambor <mjambor@suse.cz>
26625 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
26627 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
26628 decorated functions.
26630 2017-01-11 Richard Biener <rguenther@suse.de>
26632 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
26633 set range/nonnull info for PHI results. Do not set it on
26634 stmts marked for removal.
26636 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
26638 * expr.c (store_field): In the bitfield case, fetch the return value
26639 from the registers before applying a single big-endian adjustment.
26640 Always do a final load for a BLKmode value not larger than a word.
26642 2017-01-10 David Malcolm <dmalcolm@redhat.com>
26645 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
26646 that we correctly handle column numbers greater than
26647 LINE_MAP_MAX_COLUMN_NUMBER.
26649 2017-01-10 Martin Sebor <msebor@redhat.com>
26651 PR middle-end/78245
26652 * gimple-ssa-sprintf.c (get_destination_size): Call
26653 {init,fini}object_sizes.
26654 * tree-object-size.c (addr_object_size): Adjust.
26655 (pass_through_call): Adjust.
26656 (pass_object_sizes::execute): Adjust.
26657 * tree-object-size.h (fini_object_sizes): Declare.
26659 2017-01-10 Martin Sebor <msebor@redhat.com>
26661 PR tree-optimization/78775
26662 * builtins.c (get_size_range): Move...
26663 * calls.c: ...to here.
26664 (alloc_max_size): Accept zero argument.
26665 (operand_signed_p): Remove.
26666 (maybe_warn_alloc_args_overflow): Call get_size_range.
26667 * calls.h (get_size_range): Declare.
26669 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
26671 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
26672 from TI's devices.csv file as of September 2016.
26673 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
26675 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
26677 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
26678 * doc/invoke.texi: Likewise.
26679 * doc/md.texi: Likewise.
26680 * doc/objc.texi: Likewise.
26682 2017-01-10 Joshua Conner <joshconner@google.com>
26684 * config/arm/fuchsia-elf.h: New file.
26685 * config/fuchsia.h: New file.
26686 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
26687 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
26689 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
26691 2016-01-10 Richard Biener <rguenther@suse.de>
26693 PR tree-optimization/79034
26694 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
26695 Propagate out degenerate PHIs in the joiner.
26697 2017-01-10 Martin Liska <mliska@suse.cz>
26699 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
26700 (sort_congruence_classes_by_decl_uid): Likewise.
26701 (sort_congruence_class_groups_by_decl_uid): Likewise.
26702 (sem_item_optimizer::merge_classes): Sort class, groups in these
26703 classes and members in the groups by DECL_UID of declarations.
26704 This would make merge operations stable.
26706 2017-01-10 Martin Liska <mliska@suse.cz>
26708 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
26709 usage of m_classes_vec.
26710 (sem_item_optimizer::~sem_item_optimizer): Likewise.
26711 (sem_item_optimizer::get_group_by_hash): Likewise.
26712 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
26713 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
26714 (sem_item_optimizer::verify_classes): Likewise.
26715 (sem_item_optimizer::process_cong_reduction): Likewise.
26716 (sem_item_optimizer::dump_cong_classes): Likewise.
26717 (sem_item_optimizer::merge_classes): Likewise.
26718 * ipa-icf.h (congruence_class_hash): Rename from
26719 congruence_class_group_hash. Remove declaration of m_classes_vec.
26721 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
26723 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
26724 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
26725 * config.gcc: Add avx512vpopcntdqintrin.h.
26726 * config/i386/avx512vpopcntdqintrin.h: New.
26727 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
26728 * config/i386/i386-builtin-types.def: Add new types.
26729 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
26730 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
26731 __builtin_ia32_vpopcountq_v8di_mask): New.
26732 * config/i386/i386-c.c (ix86_target_macros_internal): Define
26733 __AVX512VPOPCNTDQ__.
26734 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
26735 (PTA_AVX512VPOPCNTDQ): Define.
26736 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
26737 TARGET_AVX512VPOPCNTDQ_P): Define.
26738 * config/i386/i386.opt: Add mavx512vpopcntdq.
26739 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
26740 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
26742 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26744 PR middle-end/77484
26745 * predict.def (PRED_CALL): Set to 67.
26747 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
26749 * expr.c (store_field): In the bitfield case, if the value comes from
26750 a function call and is of an aggregate type returned in registers, do
26751 not modify the field mode; extract the value in all cases if the mode
26752 is BLKmode and the size is not larger than a word.
26754 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
26757 * config/i386/cpuid.h: Fix undefined behavior.
26759 2017-01-04 Jeff Law <law@redhat.com>
26761 PR tree-optimization/79007
26762 PR tree-optimization/67955
26763 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
26764 conservative for pt.null when flag_non_call_exceptions is on.
26766 2017-01-09 Jakub Jelinek <jakub@redhat.com>
26768 PR translation/79019
26769 PR translation/79020
26770 * params.def (PARAM_INLINE_MIN_SPEEDUP,
26771 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
26772 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
26774 * config/avr/avr.opt (maccumulate-args): Likewise.
26775 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
26776 * common.opt (freport-bug): Likewise.
26777 * cif-code.def (CIF_FINAL_ERROR): Likewise.
26778 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
26779 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
26780 translatable string.
26781 * config/i386/i386.c (function_value_32): Likewise.
26782 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
26783 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
26785 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
26786 * common/config/msp430/msp430-common.c (msp430_handle_option):
26788 * symtab.c (symtab_node::verify_base): Likewise.
26789 * opts.c (set_debug_level): Likewise.
26790 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
26791 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
26792 missing whitespace to translatable strings.
26793 * config/avr/avr.md (bswapsi2): Fix typo in comment.
26794 * config/sh/superh.h: Likewise.
26795 * config/i386/xopintrin.h: Likewise.
26796 * config/i386/znver1.md: Likewise.
26797 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
26798 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
26799 * double-int.h (struct double_int): Likewise.
26800 * double-int.c (div_and_round_double): Likewise.
26801 * wide-int.cc: Likewise.
26802 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
26803 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
26804 * cfgcleanup.c (crossjumps_occured): Renamed to ...
26805 (crossjumps_occurred): ... this.
26806 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
26809 PR tree-optimization/78899
26810 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
26811 returning bool return struct loop *, NULL for failure and the new
26813 (versionable_outer_loop_p): Don't version outer loop if it has
26814 dont_vectorized bit set.
26815 (tree_if_conversion): When versioning outer loop, ensure
26816 tree_if_conversion is performed also on the inner loop of the
26817 non-vectorizable outer loop copy.
26818 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
26819 LOOP_VECTORIZED in inner loop of the scalar outer loop and
26820 prevent vectorization of it.
26821 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
26822 the outer loop vectorization of the non-scalar version is attempted
26823 before vectorization of the inner loop in scalar version. If
26824 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
26825 vectorization of its inner loop.
26826 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
26827 has 2 inner loops, rename also on edges from bb whose single pred
26828 is outer_loop->header. Fix typo in function comment.
26830 2017-01-09 Martin Sebor <msebor@redhat.com>
26833 * asan.c (asan_emit_stack_protection): Increase local buffer size
26834 to avoid snprintf truncation warning.
26836 2017-01-09 Andrew Pinski <apinski@cavium.com>
26838 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
26839 to reference thunderx2t99 for the tuning structure
26840 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
26842 (thunderx2t99_extra_costs): This.
26843 * config/aarch64/aarch64-tune.md: Regenerate.
26844 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
26845 (vulcan_addrcost_table): This.
26846 (vulcan_regmove_cost): Rename to ...
26847 (thunderx2t99_regmove_cost): This.
26848 (vulcan_vector_cost): Rename to ...
26849 (thunderx2t99_vector_cost): this.
26850 (vulcan_branch_cost): Rename to ...
26851 (thunderx2t99_branch_cost): This.
26852 (vulcan_tunings): Rename to ...
26853 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
26854 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
26856 2017-01-09 Martin Jambor <mjambor@suse.cz>
26860 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
26861 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
26862 (propagate_vr_accross_jump_function): Use the above function for all
26863 value range computations for pass-through jump functions and type
26864 converasion from explicit value range values.
26865 (ipcp_propagate_stage): Do not attempt to deduce types of formal
26866 parameters from TYPE_ARG_TYPES.
26867 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
26868 (ipa_write_node_info): Stream type of the actual argument.
26869 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
26871 2017-01-09 Martin Liska <mliska@suse.cz>
26874 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
26875 (lookup_compiler): Do not show error message with have_E.
26877 2017-01-09 Jakub Jelinek <jakub@redhat.com>
26879 PR tree-optimization/78938
26880 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
26881 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
26882 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
26883 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
26886 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26888 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
26891 2017-01-09 Richard Biener <rguenther@suse.de>
26893 PR tree-optimization/78997
26894 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
26895 name condition properly.
26897 2017-01-09 Richard Biener <rguenther@suse.de>
26900 * dwarf2out.c (is_cxx): New overload with context.
26901 (is_naming_typedef_decl): Use it.
26903 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26905 * invoke.texi (Option Summary): Correct spacing in option lists
26906 and add line breaks to fix over-long lines.
26908 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26910 PR middle-end/17660
26912 * extend.texi (Common Variable Attributes): Add xref to GCC
26913 Internals manual to explain mode attribute keywords.
26915 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26918 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
26919 and Preprocessor Options.
26920 (Options for Linking): Document -pthread here....
26921 (RS/6000 and PowerPC Options): ...not here.
26922 (Solaris 2 Options): ...or here.
26923 * doc/cppopts.texi: Document -pthread.
26925 2017-01-08 Martin Sebor <msebor@redhat.com>
26927 PR middle-end/77708
26928 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
26929 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
26930 New member functions.
26931 (format_directive): Used them.
26933 (pass_sprintf_length::handle_gimple_call): Same.
26934 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
26935 to avoid truncation for any argument.
26936 (extract_affine_mul): Same.
26937 * tree.c (get_file_function_name): Same.
26939 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26941 PR middle-end/77484
26942 * predict.def (PRED_INDIR_CALL): Set to 86.
26944 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26946 PR preprocessor/54124
26947 * doc/cppopts.texi: Reformat -d subtable to list the full name
26948 of the options. Add cross-reference to the docs for the general
26949 compiler -d options.
26950 * doc/invoke.texi (Developer Options): Add cross-reference to the
26951 preprocessor-specific -d option documentation.
26953 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26955 PR preprocessor/13498
26956 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
26957 redudant material, and reflect new command-line options.
26958 (System Headers): Likewise.
26960 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26962 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
26963 -isystem, and -idirafter. Copy-edit.
26964 * doc/cppopts.texi: Copy-edit. Remove contradiction about
26965 default for -ftrack-macro-expansion. Delete obsolete and
26966 badly-formatted implementation details about -fdebug-cpp output.
26967 * doc/cppwarnopts.texi: Copy-edit.
26969 2017-01-07 David Malcolm <dmalcolm@redhat.com>
26972 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
26973 that the transition from a max line width >= 1<<10 to narrower
26974 lines works correctly.
26976 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
26978 * doc/options.texi (PerFunction): New.
26979 * opt-functions.awk (switch_flags): Map both Optimization and
26980 PerFunction to CL_OPTIMIZATION.
26981 * opth-gen.awk: Test for PerFunction flag along with
26983 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
26984 it only when the latter is present. Skip those that don't in
26985 the hash function generator.
26986 * common.opt (fvar-tracking): Mark as PerFunction instead of
26988 (fvar-tracking-assignments): Likewise.
26989 (fvar-tracking-assignments-toggle): Likewise.
26990 (fvar-tracking-uninit): Likewise.
26992 2017-01-07 Jakub Jelinek <jakub@redhat.com>
26994 PR translation/79018
26995 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
26998 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
27001 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
27002 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
27003 TARGET_LONG_JUMP_TABLE_OFFSETS.
27004 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
27005 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
27006 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
27007 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
27008 * config/m68k/m68k.md (tablejump expander): Likewise.
27009 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
27010 TARGET_LONG_JUMP_TABLE_OFFSETS.
27011 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
27012 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
27014 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
27015 David Holsgrove <david.holsgrove@xilinx.com>
27017 * common/config/microblaze/microblaze-common.c
27018 (TARGET_EXCEPT_UNWIND_INFO): Remove.
27019 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
27021 * config/microblaze/microblaze.c (microblaze_must_save_register)
27022 (microblaze_expand_epilogue, microblaze_return_addr): Handle
27024 (microblaze_eh_return): New function.
27025 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
27026 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
27027 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
27028 * config/microblaze/microblaze.md (eh_return): New pattern.
27030 2017-01-06 Jakub Jelinek <jakub@redhat.com>
27032 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
27033 GCC_DIAGNOSTIC_STRINGIFY): Define.
27035 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
27037 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27039 * config/arm/arm.md (<mcrr>): New.
27041 * config/arm/arm.c (arm_arch5te): New.
27042 (arm_option_override): Set arm_arch5te.
27043 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
27045 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
27046 (arm_mcrr_qualifiers): ... this. New.
27047 (MRRC_QUALIFIERS): Define to...
27048 (arm_mrrc_qualifiers): ... this. New.
27049 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
27051 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
27052 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
27053 (MRRCI, mrrc, MRRC): New.
27054 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
27055 VUNSPEC_MRRC2): New.
27057 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27059 * config/arm/arm.md (<mcr>): New.
27061 * config/arm/arm.c (arm_coproc_builtin_available): Add
27062 support for mcr, mrc, mcr2 and mrc2.
27063 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
27064 (arm_mcr_qualifiers): ... this. New.
27065 (MRC_QUALIFIERS): Define to ...
27066 (arm_mrc_qualifiers): ... this. New.
27067 (MCR_QUALIFIERS): Define to ...
27068 (arm_mcr_qualifiers): ... this. New.
27069 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
27071 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
27072 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
27073 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
27074 VUNSPEC_MRC2): New.
27076 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27078 * config/arm/arm.md (*ldc): New.
27082 * config/arm/arm.c (arm_coproc_builtin_available): Add
27083 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
27084 (arm_coproc_ldc_stc_legitimate_address): New.
27085 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
27086 'qualifier_const_pointer'.
27087 (LDC_QUALIFIERS): Define to...
27088 (arm_ldc_qualifiers): ... this. New.
27089 (STC_QUALIFIERS): Define to...
27090 (arm_stc_qualifiers): ... this. New.
27091 * config/arm/arm-protos.h
27092 (arm_coproc_ldc_stc_legitimate_address): New.
27093 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
27094 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
27095 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
27096 stc2, stcl, stc2l): New.
27097 * config/arm/constraints.md (Uz): New.
27098 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
27099 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
27100 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
27101 VUNSPEC_STC2L): New.
27103 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27105 * config/arm/arm.md (<cdp>): New.
27106 * config/arm/arm.c (neon_const_bounds): Rename this ...
27107 (arm_const_bounds): ... this.
27108 (arm_coproc_builtin_available): New.
27109 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
27110 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
27111 (CDP_QUALIFIERS): Define to...
27112 (arm_cdp_qualifiers): ... this. New.
27114 (arm_expand_builtin_args): Add case for 6 arguments.
27115 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
27116 (arm_const_bounds): ... this.
27117 (arm_coproc_builtin_available): New.
27118 * config/arm/arm_acle.h (__arm_cdp): New.
27120 * config/arm/arm_acle_builtins.def (cdp): New.
27122 * config/arm/iterators.md (CDPI,CDP,cdp): New.
27123 * config/arm/neon.md: Rename all 'neon_const_bounds' to
27124 'arm_const_bounds'.
27125 * config/arm/types.md (coproc): New.
27126 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
27127 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
27128 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
27129 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
27131 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27133 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
27134 (UBINOP_QUALIFIERS): New.
27136 (acle_builtin_data): New. Change comment.
27137 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
27138 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
27139 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
27140 arm_acle_builtins.def.
27141 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
27142 (arm_init_acle_builtins): New.
27143 (CRC32_BUILTIN): Remove.
27144 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
27145 crc32cb, crc32ch and crc32cw.
27146 (arm_init_crc32_builtins): Remove.
27147 (arm_init_builtins): Use arm_init_acle_builtins rather
27148 than arm_init_crc32_builtins.
27149 (arm_expand_acle_builtin): New.
27150 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
27151 * config/arm/arm_acle_builtins.def: New.
27153 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27155 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
27156 (arm_builtin_datum): ... this.
27157 (arm_init_neon_builtin): Rename to ...
27158 (arm_init_builtin): ... this. Add a new parameters PREFIX
27159 and USE_SIG_IN_NAME.
27160 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
27161 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
27162 'arm_builtin_datum'.
27163 (arm_init_vfp_builtins): Likewise.
27164 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
27165 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
27166 (arm_expand_neon_args): Rename to ...
27167 (arm_expand_builtin_args): ... this. Rename builtin_arg
27168 enum values and differentiate between ARG_BUILTIN_MEMORY
27169 and ARG_BUILTIN_NEON_MEMORY.
27170 (arm_expand_neon_builtin_1): Rename to ...
27171 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
27172 values, arm_expand_builtin_args and add bool parameter NEON.
27173 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
27174 (arm_expand_vfp_builtin): Likewise.
27175 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
27177 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27179 PR middle-end/77484
27180 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
27181 * predict.c (tree_estimate_probability_bb): Reverse direction of
27182 polymorphic call predictor.
27184 2017-01-06 David Malcolm <dmalcolm@redhat.com>
27186 * passes.c (execute_one_pass): Split out pass-skipping logic into...
27187 (determine_pass_name_match): ...this new function and...
27188 (should_skip_pass_p): ...this new function.
27190 2017-01-06 Nathan Sidwell <nathan@acm.org>
27192 * ipa-visibility.c (function_and_variable_visibility): Reformat
27193 comments and long lines. Remove extrneous if.
27194 * symtab.c (symtab_node::make_decl_local): Fix code format.
27195 (symtab_node::set_section_for_node): Fix comment typo.
27197 2017-01-06 Martin Liska <mliska@suse.cz>
27200 * lra-constraints.c: Rename invariant to lra_invariant.
27201 * predict.c (set_even_probabilities): Initialize e to NULL.
27203 2017-01-05 Martin Sebor <msebor@redhat.com>
27205 PR tree-optimization/78910
27206 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
27207 (format_integer): Correct off-by-one error in the handling
27208 of precision with negative numbers in signed conversions..
27210 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
27212 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
27214 2017-01-05 Jakub Jelinek <jakub@redhat.com>
27216 PR tree-optimization/71016
27217 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
27218 factor_out_conditional_conversion. Formatting fix.
27219 (factor_out_conditional_conversion): Add cond_stmt argument.
27220 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
27221 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
27224 2017-01-05 David Malcolm <dmalcolm@redhat.com>
27226 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
27227 read-rtl-function.o, and selftest-rtl.o.
27228 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
27229 (selftest::aarch64_test_loading_full_dump): New function.
27230 (selftest::aarch64_run_selftests): New function.
27231 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
27232 selftest::aarch64_run_selftests.
27233 * config/i386/i386.c
27234 (selftest::ix86_test_loading_dump_fragment_1): New function.
27235 (selftest::ix86_test_loading_call_insn): New function.
27236 (selftest::ix86_test_loading_full_dump): New function.
27237 (selftest::ix86_test_loading_unspec): New function.
27238 (selftest::ix86_run_selftests): Call the new functions.
27239 * emit-rtl.c (maybe_set_max_label_num): New function.
27240 * emit-rtl.h (maybe_set_max_label_num): New decl.
27241 * function.c (instantiate_decls): Guard call to
27242 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
27243 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
27245 * gensupport.c (gen_reader::gen_reader): Pass "false"
27246 for new "compact" param of rtx_reader.
27247 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
27248 rather than an empty string for NULL strings.
27249 * read-md.c: Potentially include config.h rather than bconfig.h.
27250 Wrap include of errors.h with #ifdef GENERATOR_FILE.
27251 (have_error): New global, copied from errors.c.
27252 (md_reader::read_name): Rename to...
27253 (md_reader::read_name_1): ...this, adding "out_loc" param,
27254 and converting "missing name or number" to returning false, rather
27256 (md_reader::read_name): Reimplement in terms of read_name_1.
27257 (md_reader::read_name_or_nil): New function.
27258 (md_reader::read_string): Handle "(nil)" by returning NULL.
27259 (md_reader::md_reader): Add new param "compact".
27260 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
27261 (md_reader::read_file): New method.
27262 * read-md.h (md_reader::md_reader): Add new param "compact".
27263 (md_reader::read_file): New method.
27264 (md_reader::is_compact): New accessor.
27265 (md_reader::read_name): Convert return type from void to file_location.
27266 (md_reader::read_name_or_nil): New decl.
27267 (md_reader::read_name_1): New decl.
27268 (md_reader::m_compact): New field.
27269 (noop_reader::noop_reader): Pass "false" for new "compact" param
27271 (rtx_reader::rtx_reader): Add new "compact" param.
27272 (rtx_reader::read_rtx_operand): Make virtual and convert return
27273 type from void to rtx.
27274 (rtx_reader::read_until): New decl.
27275 (rtx_reader::handle_any_trailing_information): New virtual function.
27276 (rtx_reader::postprocess): New virtual function.
27277 (rtx_reader::finalize_string): New virtual function.
27278 (rtx_reader::m_in_call_function_usage): New field.
27279 (rtx_reader::m_reuse_rtx_by_id): New field.
27280 * read-rtl-function.c: New file.
27281 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
27282 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
27283 (selftest::verify_three_block_rtl_cfg): New decl.
27284 * read-rtl-function.h: New file.
27285 * read-rtl.c: Potentially include config.h rather than bconfig.h.
27286 For host, include function.h, memmodel.h, and emit-rtl.h.
27287 (one_time_initialization): New function.
27288 (struct compact_insn_name): New struct.
27289 (compact_insn_names): New array.
27290 (find_code): Handle insn codes in compact dumps.
27291 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
27292 (bind_subst_iter_and_attr): Likewise.
27293 (add_condition_to_string): Likewise.
27294 (add_condition_to_rtx): Likewise.
27295 (apply_attribute_uses): Likewise.
27296 (add_current_iterators): Likewise.
27297 (apply_iterators): Likewise.
27298 (initialize_iterators): Guard usage of apply_subst_iterator with
27299 #ifdef GENERATOR_FILE.
27300 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
27301 (md_reader::read_mapping): Likewise.
27302 (add_define_attr_for_define_subst): Likewise.
27303 (add_define_subst_attr): Likewise.
27304 (read_subst_mapping): Likewise.
27305 (check_code_iterator): Likewise.
27306 (rtx_reader::read_rtx): Likewise. Move one-time initialization
27308 (one_time_initialization): New function.
27309 (rtx_reader::read_until): New method.
27310 (read_flags): New function.
27311 (parse_reg_note_name): New function.
27312 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
27313 Handle reuse_rtx ids.
27314 Wrap iterator lookup within #ifdef GENERATOR_FILE.
27315 Add parsing support for RTL dumps, mirroring the special-cases in
27316 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
27317 values, and calling handle_any_trailing_information.
27318 (rtx_reader::read_rtx_operand): Convert return type from void
27319 to rtx, returning return_rtx. Handle case 'e'. Call
27320 finalize_string on XSTR and XTMPL fields.
27321 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
27322 "(nil)" values were omitted. Call the postprocess vfunc on the
27324 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
27325 class ctor. Initialize m_in_call_function_usage. Call
27326 one_time_initialization.
27327 * rtl-tests.c (selftest::test_uncond_jump): Call
27328 set_new_first_and_last_insn.
27329 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
27330 * selftest-rtl.c: New file.
27331 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
27332 (selftest::get_insn_by_uid): New decl.
27333 * selftest-run-tests.c (selftest::run_tests): Call
27334 read_rtl_function_c_tests.
27335 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
27336 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
27339 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
27341 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
27342 operands in a special way. Assert that pos+len <= mode precision.
27344 2017-01-05 Jakub Jelinek <jakub@redhat.com>
27346 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
27347 3 argument Alias with unlimited for the negative form.
27348 (fno-vect-cost-model): Removed.
27350 2017-01-05 Martin Liska <mliska@suse.cz>
27352 * hsa-gen.c (gen_hsa_divmod): New function.
27353 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
27355 2017-01-05 Martin Liska <mliska@suse.cz>
27358 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
27361 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27363 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
27364 small constant length operands.
27366 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27368 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
27369 between loop iterations.
27371 2017-01-05 Martin Liska <mliska@suse.cz>
27374 * gimplify.c (gimplify_decl_expr): Compare to
27375 asan_poisoned_variables instread of checking flags.
27376 (gimplify_target_expr): Likewise.
27377 (gimplify_expr): Likewise.
27378 (gimplify_function_tree): Conditionally initialize
27379 asan_poisoned_variables.
27381 2017-01-04 Jeff Law <law@redhat.com>
27383 PR tree-optimizatin/78812
27384 * rtl.h (contains_mem_rtx_p): Prototype.
27385 * ifcvt.c (containts_mem_rtx_p): Move from here to...
27386 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
27387 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
27388 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
27389 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
27391 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27393 * input.c (assert_char_at_range): Default-initialize actual_range.
27395 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27397 * df-scan.c (df_ref_create_structure): Make regno unsigned,
27398 to match the caller.
27400 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27402 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
27403 insns after final jump in test to emit dummy move.
27405 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27407 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
27408 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
27410 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27412 * multiple_target.c (create_dispatcher_calls): Init e_next.
27413 * tree-ssa-loop-split.c (split_loop): Init border.
27414 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
27417 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
27422 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
27423 (altivec_register_operand): Do not return true if the operand
27424 contains a SUBREG mixing SImode and SFmode.
27425 (vsx_register_operand): Likewise.
27426 (vsx_reg_sfsubreg_ok): New predicate.
27427 (vfloat_operand): Do not return true if the operand contains a
27428 SUBREG mixing SImode and SFmode.
27429 (vint_operand): Likewise.
27430 (vlogical_operand): Likewise.
27431 (gpc_reg_operand): Likewise.
27432 (int_reg_operand): Likewise.
27433 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
27434 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
27435 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
27437 (rs6000_emit_move_si_sf_subreg): New helper function.
27438 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
27439 fixup SUBREGs involving SImode and SFmode.
27440 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
27441 numbers for the new peephole2 optimization.
27442 (peephole2 for SFmode unions): New peephole2 to optimize cases in
27443 the GLIBC math library that do AND/IOR/XOR operations on single
27444 precision floating point.
27445 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
27446 target macros to say whether we need to avoid SUBREGs mixing
27448 (TARGET_ALLOW_SF_SUBREG): Likewise.
27449 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
27450 (UNSPEC_SI_FROM_SF): Likewise.
27451 (iorxor): Change spacing.
27452 (and_ior_xor): New iterator for AND, IOR, and XOR.
27453 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
27454 (movdi_from_sf_zero_ext): Likewise.
27455 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
27456 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
27457 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
27458 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
27459 (fms<mode>4): Likewise.
27460 (fnma<mode>4): Likewise.
27461 (fnms<mode>4): Likewise.
27462 (nfma<mode>4): Likewise.
27463 (nfms<mode>4): Likewise.
27465 2017-01-04 Marek Polacek <polacek@redhat.com>
27468 * doc/invoke.texi: Document -Wpointer-compare.
27470 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27472 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
27475 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
27476 descriptions for -gdwarf-5 and emit them as uleb128 instead of
27479 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
27482 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
27483 documentation of the powerpc_popcntb_ok attribute.
27484 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27485 code to issue warning messages if a requested CPU configuration is
27486 not supported by the binary (assembler and loader) toolchain.
27487 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
27488 made to define a built-in function that has been disabled.
27489 (paired_init_builtins): Add assertion to prevent ICE if attempt is
27490 made to define a built-in function that has been disabled.
27491 (altivec_init_builtins): Add comment explaining why definition
27492 of the DST built-in functions is not preceded by an assertion
27493 check. Add assertions to prevent ICE if attempts are made to
27494 define an altivec predicate or an abs* built-in function that has
27496 (htm_init_builtins): Add comment explaining why definition of the
27497 htm built-in functions is not preceded by an assertion check.
27499 2017-01-04 Jeff Law <law@redhat.com>
27501 PR tree-optimizatin/67955
27502 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
27503 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
27504 the points-to solution does not include pt_null. Use DECL_PT_UID
27507 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
27509 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
27510 Use gen_int_mode instead of gen_lopwart for const_int operands.
27512 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27514 PR tree-optimization/71563
27515 * match.pd: Simplify X << Y into X if Y is known to be 0 or
27516 out of range value - has low bits known to be zero.
27518 2017-01-04 Alan Modra <amodra@gmail.com>
27520 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
27521 * configure: Regenerate.
27522 * config.in: Regenerate.
27524 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27527 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
27528 a substring of the message, but strcmp with the whole message. Ifdef
27529 ENABLE_NLS, translate the message first using dgettext.
27531 2017-01-03 Jeff Law <law@redhat.com>
27533 PR tree-optimizatin/78856
27534 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
27535 (mark_threaded_blocks): Remove code to truncate thread paths that
27536 cross multiple loop headers. Instead invalidate the cached loop
27537 iteration information and handle case of a thread path walking
27538 into an irreducible region.
27540 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
27543 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
27544 assertions. Add support for doing the signbit if the IEEE 128-bit
27545 floating point value is in a GPR.
27546 * config/rs6000/rs6000.md (Fsignbit): Delete.
27547 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
27548 Update the length attribute if the value is in a GPR.
27549 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
27550 the sign or zero extension instruction, since the value is always 0/1.
27551 (signbit<mode>2_dm2): Delete using <Fsignbit>.
27554 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
27555 extracting SImode to a GPR register so that we can generate a
27556 store, limit the vector to be in a traditional Altivec register
27557 for the vextuwrx instruction.
27559 2017-01-03 Ian Lance Taylor <iant@google.com>
27561 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
27563 2017-01-03 Martin Sebor <msebor@redhat.com>
27565 PR tree-optimization/78696
27566 * gimple-ssa-sprintf.c (format_floating): Correct handling of
27567 precision. Use MPFR for %f for greater fidelity. Correct handling
27569 (pass_sprintf_length::compute_format_length): Set width and precision
27570 specified by asrerisk to void_node for vararg functions.
27571 (try_substitute_return_value): Adjust dump output.
27573 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
27575 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
27577 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
27579 * doc/invoke.texi (SPARC options): Document -mlra as the default.
27580 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
27581 -mlra/-mno-lra was passed to the compiler.
27583 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
27585 PR rtl-optimization/65618
27586 * emit-rtl.c (try_split): Move initialization of "before" and
27587 "after" to just before the call to emit_insn_after_setloc.
27589 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
27591 * doc/md.texi (Standard Names): Remove reference to Java frontend.
27593 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
27595 * dwarf2out.c (gen_enumeration_type_die): When
27596 -gno-strict-dwarf, add a DW_AT_encoding attribute.
27598 2017-01-03 Jakub Jelinek <jakub@redhat.com>
27600 PR tree-optimization/78965
27601 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
27602 Change first argument from const call_info & to call_info &. For %n
27603 set info.nowrite to false.
27605 PR middle-end/78901
27606 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
27607 possibly throwing calls.
27609 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
27610 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
27611 and fns handling, rather than in a separate case SSA_NAME.
27613 2017-01-02 Jeff Law <law@redhat.com>
27615 * config/darwin-driver.c (darwin_driver_init): Const-correctness
27616 fixes for first_period and second_period variables.
27618 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
27621 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
27622 (*insvqi_1): New insn pattern.
27623 (*insvqi_1_mem_rex64): Ditto.
27624 (*insvqi_2): Ditto.
27625 (*insvqi_3): Rename from *insvqi.
27627 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
27629 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
27631 * doc/cfg.texi (Edges): Remove reference to Java.
27632 (Maintaining the CFG): Ditto.
27634 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27636 PR middle-end/77674
27637 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
27638 transparent aliases.
27640 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27642 PR middle-end/77484
27643 * predict.def (PRED_CALL): Update hitrate.
27644 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
27645 * predict.c (tree_estimate_probability_bb): Split CALL predictor
27646 into direct/indirect/polymorphic variants.
27648 2017-01-01 Jakub Jelinek <jakub@redhat.com>
27650 Update copyright years.
27652 * gcc.c (process_command): Update copyright notice dates.
27653 * gcov-dump.c (print_version): Ditto.
27654 * gcov.c (print_version): Ditto.
27655 * gcov-tool.c (print_version): Ditto.
27656 * gengtype.c (create_file): Ditto.
27657 * doc/cpp.texi: Bump @copying's copyright year.
27658 * doc/cppinternals.texi: Ditto.
27659 * doc/gcc.texi: Ditto.
27660 * doc/gccint.texi: Ditto.
27661 * doc/gcov.texi: Ditto.
27662 * doc/install.texi: Ditto.
27663 * doc/invoke.texi: Ditto.
27665 Copyright (C) 2017 Free Software Foundation, Inc.
27667 Copying and distribution of this file, with or without modification,
27668 are permitted in any medium without royalty provided the copyright
27669 notice and this notice are preserved.