1 2017-09-21 Jakub Jelinek <jakub@redhat.com>
4 * tree-inline.c (expand_call_inline): Emit clobber stmts for
5 VAR_DECLs to which addressable non-volatile parameters are mapped
6 and for id->retvar after the return value assignment. Clear
7 id->retval and id->retbnd after inlining.
9 2017-09-21 Richard Biener <rguenther@suse.de>
11 PR tree-optimization/82276
12 PR tree-optimization/82244
13 * tree-vrp.c (build_assert_expr_for): Set
14 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
16 (remove_range_assertions): Revert earlier change.
18 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
21 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
23 2017-09-21 Richard Biener <rguenther@suse.de>
25 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
26 Restore valid IL after code generation errors.
27 * graphite.c (graphite_transform_loops): Diagnose code
28 generation issues as MSG_MISSED_OPTIMIZATION and continue
29 with processing SCOPs.
31 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
32 Alan Hayward <alan.hayward@arm.com>
33 David Sherwood <david.sherwood@arm.com>
35 * calls.c (compute_argument_addresses): Use simplify_gen_binary
36 rather than choosing between plus_constant and gen_rtx_<CODE>.
37 * expr.c (emit_push_insn): Likewise.
38 (expand_expr_real_2): Likewise.
40 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
41 Alan Hayward <alan.hayward@arm.com>
42 David Sherwood <david.sherwood@arm.com>
44 * loop-unroll.c (split_iv): Call copy_rtx on the step.
46 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
47 Alan Hayward <alan.hayward@arm.com>
48 David Sherwood <david.sherwood@arm.com>
50 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
53 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
54 Alan Hayward <alan.hayward@arm.com>
55 David Sherwood <david.sherwood@arm.com>
57 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
58 INTEGER_CST rather than a negative test for ADDR_EXPR.
60 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
61 Alan Hayward <alan.hayward@arm.com>
62 David Sherwood <david.sherwood@arm.com>
64 * tree-vrp.c (extract_range_from_binary_expr_1): Check
65 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
67 2017-09-21 Richard Biener <rguenther@suse.de>
69 PR tree-optimization/71351
70 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
71 graphite_create_new_loop_guard): Remove, fold remaining parts
73 (translate_isl_ast_node_for): ... here and simplify.
75 2017-09-21 Jakub Jelinek <jakub@redhat.com>
78 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
79 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
80 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
81 alternative always use QI mode, for -Os imov (=R,R) alternative
82 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
85 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
86 Jeff Law <law@redhat.com>
88 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
89 (allocate_stack_space): New function, partially extracted from
91 (s390_emit_prologue): Track offset to most recent stack probe.
92 Code to allocate space moved into allocate_stack_space.
93 Dump actions when no stack is allocated.
94 (s390_prologue_plus_offset): New function.
95 (s390_emit_stack_probe): Likewise.
97 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
99 * common.opt (Wa, Wl, Wp, g, gz=): Add
101 (gno-column-info): Remove.
102 (gcolumn-info): Drop RejectNegative.
104 (gno-record-gcc-switches): Remove.
105 (grecord-gcc-switches): Drop RejectNegative.
106 (gno-split-dwarf): Remove.
107 (gsplit-dwarf): Drop RejectNegative.
108 (gno-strict-dwarf): Remove.
109 (gstrict-dwarf): Drop RejectNegative.
110 * config/darwin.opt (gfull, gused): Add RejectNegative.
111 * dwarf2out.c (gen_producer_string): Drop
112 gno-record-gcc-switches handler.
113 * optc-gen.awk: Add g to prefixes with negative forms.
114 * opts-common.c (remapping_prefix_p): New.
115 (find_opt): Check it.
116 (generate_canonical_option): Test g prefix.
117 (option_map): Add -gno- mapping.
118 (add_misspelling_candidates): Check remapping_prefix_p.
120 2017-09-20 Jeff Law <law@redhat.com>
122 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
123 thinko in stack clash protection support.
125 * explow.c (compute_stack_clash_protection_loop_data): Use
126 CONST_INT_P instead of explicit test. Verify object is a
127 CONST_INT_P before looking at INTVAL.
128 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
129 instead of explicit test.
131 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
134 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
135 address instead of to r1 and r11.
137 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
139 * config.gcc: Support "knm".
140 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
141 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
143 * config/i386/i386.c (m_KNM): Define.
144 (processor_target_table): Add "knm".
146 (ix86_option_override_internal): Add "knm".
147 (ix86_issue_rate): Add PROCESSOR_KNM.
148 (ix86_adjust_cost): Ditto.
149 (ia32_multipass_dfa_lookahead): Ditto.
150 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
151 (fold_builtin_cpu): Add M_INTEL_KNM.
152 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
153 (processor_type): Add PROCESSOR_KNM.
154 * config/i386/x86-tune.def: Add m_KNM.
155 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
157 2017-09-20 Richard Biener <rguenther@suse.de>
159 PR tree-optimization/80213
160 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
161 are allowed in empty BBs as well.
162 (canonicalize_loop_closed_ssa): Also look for other complex
164 (scop_detection::get_sese): Include the loop-closed PHI block
166 (scop_detection::merge_sese): Remove code adding extra blocks.
167 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
168 (build_scops): Assert the final returned scop is invalid.
170 2017-09-20 Richard Biener <rguenther@suse.de>
172 PR tree-optimization/82264
173 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
175 (vn_phi_lookup): Likewise.
176 (vn_phi_insert): Likewise.
178 2017-09-20 Jakub Jelinek <jakub@redhat.com>
180 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
181 that fits into uhwi or shwi, add DW_AT_const_value regardless
182 of early_dwarf without going through RTL, using add_AT_unsigned
185 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
186 (DEBUG_LTO_ABBREV_SECTION): Likewise.
187 (DEBUG_LTO_MACINFO_SECTION): Likewise.
188 (DEBUG_MACRO_SECTION): Likewise.
189 (DEBUG_LTO_MACRO_SECTION): Likewise.
190 (DEBUG_STR_DWO_SECTION): Likewise.
191 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
192 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
193 (DEBUG_LTO_DWO_LINE_SECTION): Define.
194 (DEBUG_LTO_LINE_STR_SECTION): Define.
195 (init_sections_and_labels): Initialize debug_line_str_section
196 variable. Initialize debug_loc_section for -gdwarf-5 to
197 DEBUG_LOCLISTS_SECTION. Formatting fixes.
199 2017-09-20 Richard Biener <rguenther@suse.de>
201 * graphite-sese-to-poly.c (extract_affine): Properly handle
202 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
204 2017-09-20 Richard Biener <rguenther@suse.de>
206 PR tree-optimization/81373
207 * graphite-scop-detection.c (build_cross_bb_scalars_def):
208 Force SESE live-out defs to be handled even if they are
211 2017-09-19 Jeff Law <law@redhat.com>
213 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
214 nothing for stack adjustments with REG_STACK_CHECK.
215 * sched-deps.c (parse_add_or_inc): Reject insns with
216 REG_STACK_CHECK from dependency breaking.
217 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
218 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
219 * reg-notes.def (STACK_CHECK): New note.
221 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
222 (ix86_expand_prologue): Dump stack clash info as needed.
223 Call ix86_adjust_stack_and_probe_stack_clash as needed.
225 * function.c (dump_stack_clash_frame_info): New function.
226 * function.h (dump_stack_clash_frame_info): Prototype.
227 (enum stack_clash_probes): New enum.
229 * config/alpha/alpha.c (alpha_expand_prologue): Also check
230 flag_stack_clash_protection.
231 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
232 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
233 (arm_frame_pointer_required): Likewise.
234 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
235 (ia64_expand_prologue): Likewise.
236 * config/mips/mips.c (mips_expand_prologue): Likewise.
237 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
238 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
239 (sparc_flat_expand_prologue): Likewise.
240 * config/spu/spu.c (spu_expand_prologue): Likewise.
242 * explow.c: Include "params.h".
243 (anti_adjust_stack_and_probe_stack_clash): New function.
244 (get_stack_check_protect): Likewise.
245 (compute_stack_clash_protection_loop_data): Likewise.
246 (emit_stack_clash_protection_loop_start): Likewise.
247 (emit_stack_clash_protection_loop_end): Likewise.
248 (allocate_dynamic_stack_space): Use get_stack_check_protect.
249 Use anti_adjust_stack_and_probe_stack_clash.
250 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
251 (emit_stack_clash_protection_loop_start): Likewise.
252 (emit_stack_clash_protection_loop_end): Likewise.
253 * rtl.h (get_stack_check_protect): Prototype.
254 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
255 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
256 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
258 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
260 * doc/tm.texi: Rebuilt.
261 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
262 get_stack_check_protect.
263 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
264 * config/arm/arm.c (arm_expand_prologue): Likewise.
265 (arm_frame_pointer_required): Likewise.
266 * config/i386/i386.c (ix86_expand_prologue): Likewise.
267 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
268 * config/mips/mips.c (mips_expand_prologue): Likewise.
269 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
270 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
271 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
272 (sparc_flat_expand_prologue): Likewise.
274 * common.opt (-fstack-clash-protection): New option.
275 * flag-types.h (enum stack_check_type): Note difference between
276 -fstack-check= and -fstack-clash-protection.
277 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
278 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
279 * toplev.c (process_options): Issue warnings/errors for cases
280 not handled with -fstack-clash-protection.
281 * doc/invoke.texi (-fstack-clash-protection): Document new option.
282 (-fstack-check): Note additional problem with -fstack-check=generic.
283 Note that -fstack-check is primarily for Ada and refer users
284 to -fstack-clash-protection for stack-clash-protection.
285 Document new params for stack clash protection.
287 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
289 * config/i386/i386.c (ix86_split_long_move): Do not handle
290 address used for LEA in a special way.
292 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
294 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
296 2017-09-19 Martin Sebor <msebor@redhat.com>
299 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
300 of incompatible types.
302 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
304 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
305 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
306 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
307 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
309 2017-09-19 Richard Biener <rguenther@suse.de>
311 PR tree-optimization/82244
312 * tree-vrp.c (remove_range_assertions): Do not propagate
313 a constant to abnormals but replace the assert with a copy.
315 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
317 PR rtl-optimization/57878
318 PR rtl-optimization/68988
319 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
320 avoidance test involving non_reload_pseudos. Move frequency test
321 below the general fragmentation avoidance test.
323 2017-09-19 Richard Biener <rguenther@suse.de>
325 PR tree-optimization/69728
326 * graphite-sese-to-poly.c (schedule_error): New global.
327 (add_loop_schedule): Handle empty domain by failing the
329 (build_original_schedule): Handle schedule_error.
331 2017-09-19 Richard Biener <rguenther@suse.de>
333 * graphite-scop-detection.c (scop_detection::can_represent_loop):
334 Do not iterate to sibling loops but only to siblings of inner
337 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
340 * config/m68k/m68k.md (moveq feeding equality comparison): Check
341 that the registers are different.
343 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
345 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
346 to processor_model and "amdfam17h" to arch_names_table.
347 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
349 2017-09-18 Jakub Jelinek <jakub@redhat.com>
352 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
354 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
355 Alan Hayward <alan.hayward@arm.com>
356 David Sherwood <david.sherwood@arm.com>
358 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
360 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
362 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
363 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
364 vect_schedule_slp_instance.
365 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
366 Update call to vect_slp_analyze_node_operations. Simplify return
368 (vect_slp_analyze_bb_1): Update call accordingly.
369 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
370 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
371 (vect_schedule_slp): Update call accordingly.
373 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
374 Alan Hayward <alan.hayward@arm.com>
375 David Sherwood <david.sherwood@arm.com>
377 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
378 with types that aren't in fact scalar.
380 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
382 * tree-vect-slp.c (vect_record_max_nunits): New function,
384 (vect_build_slp_tree_1): ...here.
385 (vect_build_slp_tree_2): Call it for phis too.
387 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
389 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
390 to vect_get_vec_def_for_operand when getting the mask operand.
392 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
393 Alan Hayward <alan.hayward@arm.com>
394 David Sherwood <david.sherwood@arm.com>
396 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
399 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
400 Alan Hayward <alan.hayward@arm.com>
401 David Sherwood <david.sherwood@arm.com>
403 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
404 calculation for vector booleans.
406 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
407 Alan Hayward <alan.hayward@arm.com>
408 David Sherwood <david.sherwood@arm.com>
410 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
412 (vect_transform_stmt): ...here.
413 (vect_analyze_stmt): Use it instead of calling
414 vectorizable_live_operation directly.
416 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
418 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
419 non-SIMT targets in acc vector loops.
421 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
423 * configure.ac: Add arc and check if assembler supports gdwarf2.
424 * configure: Regenerate.
426 2017-09-18 Richard Biener <rguenther@suse.de>
428 PR tree-optimization/82220
429 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
430 epilogue niters from the min_profitable_iters compute.
432 2017-09-18 Jakub Jelinek <jakub@redhat.com>
435 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
436 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
437 (ix86_init_pic_reg): Revert 2017-09-01 changes.
439 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
442 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
443 switching to a new text section.
445 2017-09-18 Richard Biener <rguenther@suse.de>
447 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
449 (build_alias_set): Reject aliases with no access function.
451 2017-09-18 Richard Biener <rguenther@suse.de>
453 PR tree-optimization/79622
454 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
456 (build_cross_bb_scalars_use): Likewise.
458 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
460 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
462 2017-09-18 Alan Modra <amodra@gmail.com>
465 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
466 stack_pointer_rtx for count 0. Update comments. Break up
467 large rtl expression.
469 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
471 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
472 Increase to 20 bytes.
473 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
474 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
475 or avx version of the stub.
477 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
480 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
481 compute the minimum stack alignment. Also update preferred stack
482 boundary for leaf functions.
484 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
486 PR tree-optimization/82228
487 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
490 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
492 * common/config/nds32/nds32-common.c
493 (nds32_option_optimization_table): Refine formatting.
494 (nds32_option_optimization_table): Use -fsched-pressure and
495 -fomit-frame-pointer for specific optimization level.
497 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
499 * config/nds32/nds32.c: Refine formatting and comments.
500 * config/nds32/nds32.h: Likewise.
501 * config/nds32/nds32.md: Likewise.
502 * config/nds32/nds32-cost.c: Likewise.
503 * config/nds32/nds32-isr.c: Likewise.
504 * config/nds32/nds32-md-auxiliary.c: Likewise.
505 * config/nds32/nds32-multiple.md: Likewise.
506 * config/nds32/nds32-predicates.c: Likewise.
508 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
509 Jakub Jelinek <jakub@redhat.com>
511 Add support for -std=c++2a.
512 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
514 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
516 2017-09-15 Steve Ellcey <sellcey@cavium.com>
519 * doc/extend.texi (Common Function Attributes): Add
520 references to ARM, AArch64, and S/390 specific attributes.
521 (Function Specific Option Pragmas): Add AArch64 and S/390
522 to list of back ends that support the target pragma.
524 2017-09-15 Nathan Sidwell <nathan@acm.org>
526 * doc/standards.texi: Fix C++17 description. Update URLs for
529 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
531 * common.opt (Wcast-align=strict): New warning option.
532 * doc/invoke.texi: Document -Wcast-align=strict.
534 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
536 * cgraph.h (cgraph_thunk_info): Add comments.
537 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
538 assert for VIRTUAL_* arguments stricter.
540 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
542 PR tree-optimization/71026
543 * match.pd: Move RDIV patterns from fold-const.c
544 * fold-const.c (distribute_real_division): Removed.
545 (fold_binary_loc): Remove calls to distribute_real_divison.
547 2017-09-15 Jakub Jelinek <jakub@redhat.com>
549 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
550 c++1z and gnu++1z as deprecated. Change other references to
551 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
552 Change -Wc++1z-compat to -Wc++17-compat.
553 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
554 * dwarf2out.c (highest_c_language): Handle C++17.
555 (gen_compile_unit_die): Likewise.
557 2017-09-15 Jakub Jelinek <jakub@redhat.com>
559 PR rtl-optimization/82192
560 * combine.c (make_extraction): Don't look through non-paradoxical
561 SUBREGs or TRUNCATE if pos + len is or might be bigger than
564 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
565 Alan Hayward <alan.hayward@arm.com>
566 David Sherwood <david.sherwood@arm.com>
568 * target.def (function_arg_offset): New hook.
569 * targhooks.h (default_function_arg_offset): Declare.
570 * targhooks.c (default_function_arg_offset): New function.
571 * function.c (locate_and_pad_parm): Use
572 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
573 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
574 (TARGET_FUNCTION_ARG_OFFSET): ...this.
575 * doc/tm.texi: Regenerate.
576 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
577 * config/spu/spu.c (spu_function_arg_offset): New function.
578 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
579 * system.h (FUNCTION_ARG_OFFSET): Poison.
581 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
582 Alan Hayard <alan.hayward@arm.com>
583 David Sherwood <david.sherwood@arm.com>
585 * target.def (truly_noop_truncation): New hook.
586 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
587 than TRULY_NOOP_TRUNCATION.
588 * hooks.h (hook_bool_uint_uint_true): Declare.
589 * hooks.c (hook_bool_uint_uint_true): New function.
590 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
591 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
592 * doc/tm.texi: Regenerate.
593 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
594 rather than TRULY_NOOP_TRUNCATION in comments.
595 (simplify_comparison): Likewise.
596 (record_truncated_value): Likewise.
597 * expmed.c (extract_bit_field_1): Likewise.
598 (extract_split_bit_field): Likewise.
599 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
600 instead of TRULY_NOOP_TRUNCATION.
601 * function.c (assign_parm_setup_block): Likewise.
602 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
603 * rtlhooks.c: Include target.h.
604 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
605 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
606 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
607 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
608 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
609 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
610 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
611 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
612 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
613 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
614 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
615 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
616 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
617 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
618 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
619 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
620 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
621 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
622 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
623 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
624 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
625 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
626 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
627 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
628 * config/mips/mips.c (mips_truly_noop_truncation): New function.
629 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
630 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
631 rather than TRULY_NOOP_TRUNCATION in comments.
632 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
633 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
634 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
635 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
636 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
637 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
638 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
639 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
640 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
641 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
642 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
643 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
644 rather than TRULY_NOOP_TRUNCATION in comments.
645 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
646 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
647 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
648 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
649 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
650 TRULY_NOOP_TRUNCATION condition.
651 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
652 (TRULY_NOOP_TRUNCATION): Delete.
653 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
654 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
655 * config/spu/spu.c (spu_truly_noop_truncation): New function.
656 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
657 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
658 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
659 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
660 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
661 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
662 rather than TRULY_NOOP_TRUNCATION in comments.
663 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
664 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
665 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
666 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
667 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
668 * system.h (TRULY_NOOP_TRUNCATION): Poison.
670 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
673 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
674 (*cmp_ior): Likewise.
675 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
676 (*ior_scc_scc_cmp): Likewise.
677 (*and_scc_scc): Likewise.
678 (*and_scc_scc_cmp): Likewise.
680 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
681 Alan Hayard <alan.hayward@arm.com>
682 David Sherwood <david.sherwood@arm.com>
684 * target.def (can_change_mode_class): New hook.
685 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
686 (hard_regno_nregs): Likewise.
687 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
688 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
689 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
690 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
691 (LOAD_EXTEND_OP): Update accordingly.
692 * doc/tm.texi: Regenerate.
693 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
694 CANNOT_CHANGE_MODE_CLASS.
695 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
696 (REG_CAN_CHANGE_MODE_P): ...this new macro.
697 * combine.c (simplify_set): Update accordingly.
698 * emit-rtl.c (validate_subreg): Likewise.
699 * recog.c (general_operand): Likewise.
700 * regcprop.c (mode_change_ok): Likewise.
701 * reload1.c (choose_reload_regs): Likewise.
702 (inherit_piecemeal_p): Likewise.
703 * rtlanal.c (simplify_subreg_regno): Likewise.
704 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
705 instead of CANNOT_CHANGE_MODE_CLASS.
706 (reload_cse_simplify_operands): Likewise.
707 * reload.c (push_reload): Use targetm.can_change_mode_class
708 instead of CANNOT_CHANGE_MODE_CLASS.
709 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
710 REG_CANNOT_CHANGE_MODE_P.
711 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
712 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
713 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
714 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
715 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
716 (arm_can_change_mode_class): New function.
717 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
718 than CANNOT_CHANGE_MODE_CLASS in comments.
719 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
720 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
721 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
722 (ix86_can_change_mode_class): ...this new function, inverting the
723 sense of the return value.
724 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
725 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
726 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
727 (ia64_can_change_mode_class): New function.
728 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
729 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
730 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
731 (m32c_can_change_mode_class): ...this new function, inverting the
732 sense of the return value.
733 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
734 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
735 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
736 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
737 (mips_can_change_mode_class): ...this new function, inverting the
738 sense of the return value.
739 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
740 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
741 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
742 (msp430_can_change_mode_class): New function.
743 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
744 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
745 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
746 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
747 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
748 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
749 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
750 (pa_cannot_change_mode_class): Replace with...
751 (pa_can_change_mode_class): ...this new function, inverting the
752 sense of the return value.
753 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
754 than CANNOT_CHANGE_MODE_CLASS in comments.
755 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
756 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
757 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
758 (pdp11_cannot_change_mode_class): Replace with...
759 (pdp11_can_change_mode_class): ...this new function, inverting the
760 sense of the return value.
761 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
762 * config/powerpcspe/powerpcspe-protos.h
763 (rs6000_cannot_change_mode_class_ptr): Delete.
764 * config/powerpcspe/powerpcspe.c
765 (rs6000_cannot_change_mode_class_ptr): Delete.
766 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
767 (rs6000_option_override_internal): Assign to
768 targetm.can_change_mode_class instead of
769 rs6000_cannot_change_mode_class_ptr.
770 (rs6000_cannot_change_mode_class): Replace with...
771 (rs6000_can_change_mode_class): ...this new function, inverting the
772 sense of the return value.
773 (rs6000_debug_cannot_change_mode_class): Replace with...
774 (rs6000_debug_can_change_mode_class): ...this new function.
775 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
776 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
777 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
778 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
779 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
781 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
782 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
783 (rs6000_option_override_internal): Assign to
784 targetm.can_change_mode_class instead of
785 rs6000_cannot_change_mode_class_ptr.
786 (rs6000_cannot_change_mode_class): Replace with...
787 (rs6000_can_change_mode_class): ...this new function, inverting the
788 sense of the return value.
789 (rs6000_debug_cannot_change_mode_class): Replace with...
790 (rs6000_debug_can_change_mode_class): ...this new function.
791 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
792 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
793 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
794 (s390_can_change_mode_class): ...this new function, inverting the
795 sense of the return value.
796 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
797 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
798 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
799 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
800 (sh_cannot_change_mode_class): Replace with...
801 (sh_can_change_mode_class): ...this new function, inverting the
802 sense of the return value.
803 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
804 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
805 (sparc_can_change_mode_class): New function.
806 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
807 * config/spu/spu.c (spu_can_change_mode_class): New function.
808 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
809 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
810 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
811 (visium_can_change_mode_class): New function.
812 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
814 2017-09-15 Richard Biener <rguenther@suse.de>
816 PR tree-optimization/82217
817 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
818 but not undefined case.
820 2017-09-15 Jakub Jelinek <jakub@redhat.com>
823 * postreload.c (reload_cse_simplify_operands): Skip
824 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
826 2017-09-15 Richard Biener <rguenther@suse.de>
828 PR tree-optimization/68823
829 * graphite-scop-detection.c (build_alias_set): If we have a
830 possible dependence check whether we can handle them by just
831 looking at the DRs DR_ACCESS_FNs.
832 (build_scops): If build_alias_set fails, fail the SCOP.
834 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
836 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
837 to support float128 built-in functions that require the ISA 3.0
839 (BU_FLOAT128_3_HW): Likewise.
840 (SQRTF128): Add support for the IEEE 128-bit square root and fma
844 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
845 support for built-in functions that need the ISA 3.0 IEEE 128-bit
846 floating point instructions.
847 (rs6000_invalid_builtin): Likewise.
848 (rs6000_builtin_mask_names): Likewise.
849 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
850 (RS6000_BTM_FLOAT128_HW): Likewise.
851 (RS6000_BTM_COMMON): Likewise.
852 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
854 * doc/extend.texi (RS/6000 built-in functions): Document the
855 IEEE 128-bit floating point square root and fused multiply-add
858 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
860 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
861 reg (r2) isn't in the set of registers defined in the prologue.
863 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
864 Alan Hayward <alan.hayward@arm.com>
865 David Sherwood <david.sherwood@arm.com>
867 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
868 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
869 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
870 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
871 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
873 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
874 max_vectorization_factor.
875 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
877 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
878 Alan Hayward <alan.hayward@arm.com>
879 David Sherwood <david.sherwood@arm.com>
881 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
882 (vect_worthwhile_without_simd_p): Declare.
883 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
884 (vectorizable_reduction): Use it.
885 * tree-vect-stmts.c (vectorizable_shift): Likewise.
886 (vectorizable_operation): Likewise.
888 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
889 Alan Hayward <alan.hayward@arm.com>
890 David Sherwood <david.sherwood@arm.com>
892 * tree-vectorizer.h (vect_get_num_copies): New function.
893 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
894 * tree-vect-loop.c (vectorizable_reduction): Likewise.
895 (vectorizable_induction): Likewise.
896 (vectorizable_live_operation): Likewise.
897 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
898 (vectorizable_bswap): Likewise.
899 (vectorizable_call): Likewise.
900 (vectorizable_conversion): Likewise.
901 (vectorizable_assignment): Likewise.
902 (vectorizable_shift): Likewise.
903 (vectorizable_operation): Likewise.
904 (vectorizable_store): Likewise.
905 (vectorizable_load): Likewise.
906 (vectorizable_condition): Likewise.
907 (vectorizable_comparison): Likewise.
908 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
910 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
911 Alan Hayward <alan.hayward@arm.com>
912 David Sherwood <david.sherwood@arm.com>
914 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
917 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
918 Alan Hayward <alan.hayward@arm.com>
919 David Sherwood <david.sherwood@arm.com>
921 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
922 an inline wrapper that provides a location.
923 (gimple_build_vector): Likewise.
924 * gimple-fold.c (gimple_build_vector_from_val): New function.
925 (gimple_build_vector): Likewise.
926 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
927 functions to build the initial value. Always return a gimple value.
928 (get_initial_defs_for_reduction): Likewise. Only compute
930 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
931 vect_init_vector on the results from get_initial_def(s)_for_reduction.
932 (vectorizable_induction): Use gimple_build_vector rather than
935 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
936 Alan Hayward <alan.hayward@arm.com>
937 David Sherwood <david.sherwood@arm.com>
939 * target.h (vec_perm_indices): New typedef.
940 (auto_vec_perm_indices): Likewise.
941 * optabs-query.h: Include target.h
942 (can_vec_perm_p): Take a vec_perm_indices *.
943 * optabs-query.c (can_vec_perm_p): Likewise.
944 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
945 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
946 * tree-vect-generic.c (lower_vec_perm): Likewise.
947 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
948 (vect_grouped_load_supported): Likewise.
949 (vect_shift_permute_load_chain): Likewise.
950 (vect_permute_store_chain): Use auto_vec_perm_indices.
951 (vect_permute_load_chain): Likewise.
952 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
953 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
954 Update uses of can_vec_perm_p.
955 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
956 mode with a number of elements. Take a vec_perm_indices *.
957 (vect_create_epilog_for_reduction): Update accordingly.
958 Use auto_vec_perm_indices.
959 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
960 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
961 (vect_transform_slp_perm_load): Likewise.
962 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
963 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
964 (vect_gen_perm_mask_checked): Likewise.
965 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
966 (vect_gen_perm_mask_checked): Likewise.
967 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
968 (vectorizable_store): Likewise.
969 (vectorizable_load): Likewise.
970 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
971 (vectorizable_bswap): Likewise.
973 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
974 Alan Hayward <alan.hayward@arm.com>
975 David Sherwood <david.sherwood@arm.com>
977 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
978 * tree.c (build_vector): Likewise.
979 (build_vector_from_ctor): Update accordingly.
980 (build_vector_from_val): Likewise.
981 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
982 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
983 * tree-vect-generic.c (add_rshift): Likewise.
984 (expand_vector_divmod): Likewise.
985 (optimize_vector_constructor): Likewise.
986 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
987 (vect_transform_slp_perm_load): Likewise.
988 (vect_schedule_slp_instance): Likewise.
989 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
990 (vectorizable_call): Likewise.
991 (vect_gen_perm_mask_any): Likewise. Add elements in order.
992 * expmed.c (make_tree): Likewise.
993 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
994 a vector passed to build_vector.
995 (fold_convert_const): Likewise.
996 (exact_inverse): Likewise.
997 (fold_ternary_loc): Likewise.
998 (fold_relational_const): Likewise.
999 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
1000 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
1001 (const_unop): Likewise. Store the reduction accumulator in a
1002 variable rather than an array.
1003 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
1004 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
1005 the new vector, rather than constructing it after the input arrays.
1006 (native_interpret_vector): Use auto_vec<tree> when building
1007 a vector passed to build_vector. Add elements in order.
1008 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
1009 auto_vec<tree> when building a vector passed to build_vector.
1010 (vect_create_epilog_for_reduction): Likewise.
1011 (vectorizable_induction): Likewise.
1012 (get_initial_def_for_reduction): Likewise. Fix indentation of
1014 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
1016 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
1017 passed to build_vector.
1019 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1020 Alan Hayward <alan.hayward@arm.com>
1021 David Sherwood <david.sherwood@arm.com>
1023 * tree-core.h (tree_base::u): Add an "nelts" field.
1024 (tree_vector): Use VECTOR_CST_NELTS as the length.
1025 * tree.c (tree_size): Likewise.
1026 (make_vector): Initialize VECTOR_CST_NELTS.
1027 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
1028 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
1029 TYPE_VECTOR_SUBPARTS.
1030 * expr.c (const_vector_mask_from_tree): Consistently use "units"
1031 as the number of units, setting it from VECTOR_CST_NELTS.
1032 (const_vector_from_tree): Likewise.
1033 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
1034 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
1035 (fold_negate_expr_1): Likewise.
1036 (fold_convert_const): Likewise.
1037 (const_binop): Likewise. Differentiate the number of output and
1039 (const_unop): Likewise.
1040 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
1041 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
1042 in cases that did the opposite.
1044 2017-09-14 Richard Biener <rguenther@suse.de>
1046 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
1049 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
1051 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
1053 2017-09-14 Jakub Jelinek <jakub@redhat.com>
1056 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
1057 if and where to split a bb, except for splitting before debug insn
1058 sequences followed by non-label real insn. Delete debug insns
1059 in between basic blocks.
1061 * combine.c (make_compound_operation_int): Formatting fixes.
1063 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
1064 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1065 * config/netbsd.h (LINK_EH_SPEC): Likewise.
1066 * config/sol2.h (LINK_EH_SPEC): Likewise.
1067 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1068 * config/s390/linux.h (LINK_SPEC): Likewise.
1069 * config/freebsd.h (LINK_EH_SPEC): Likewise.
1070 * config/openbsd.h (LINK_EH_SPEC): Likewise.
1071 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1072 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
1073 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
1074 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1075 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
1076 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
1078 2017-09-13 Jakub Jelinek <jakub@redhat.com>
1080 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
1082 (ENDFILE_LINUX_SPEC): Likewise.
1083 (LINK_EH_SPEC): Likewise.
1084 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
1085 (LINK_OS_LINUX_SPEC32): Likewise.
1086 (LINK_OS_LINUX_SPEC64): Likewise.
1087 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
1088 (LINK_OS_LINUX_SPEC): Likewise.
1090 2017-09-13 Martin Liska <mliska@suse.cz>
1093 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
1094 CASE_HIGH is NULL_TREE.
1096 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
1097 Alan Hayward <alan.hayward@arm.com>
1098 David Sherwood <david.sherwood@arm.com>
1100 * target.def (secondary_memory_needed): New hook.
1101 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
1102 instead of SECONDARY_MEMORY_NEEDED.
1103 (secondary_memory_needed_mode): Likewise.
1104 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
1105 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
1106 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
1107 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
1108 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
1109 * doc/tm.texi: Regenerate.
1110 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
1111 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
1112 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1113 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
1114 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
1115 * config/i386/i386.c (inline_secondary_memory_needed): Put the
1116 mode argument first and change the reg_class arguments to reg_class_t.
1117 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
1118 Make static. Update the call to inline_secondary_memory_needed.
1119 (ix86_register_move_cost): Update the call to
1120 inline_secondary_memory_needed.
1121 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1122 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
1124 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
1125 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
1127 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
1128 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
1129 * config/mips/mips.c (mips_secondary_memory_needed): Make static
1130 and match hook interface. Add comment from mips.h.
1131 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1132 * config/mmix/mmix.md (truncdfsf2): Refer to
1133 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
1135 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
1136 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
1137 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1138 (pa_secondary_memory_needed): New function.
1139 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
1140 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
1141 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1142 (pdp11_secondary_memory_needed): Make static and match hook interface.
1143 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
1144 * config/powerpcspe/powerpcspe-protos.h
1145 (rs6000_secondary_memory_needed_ptr): Delete.
1146 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
1148 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1149 (rs6000_option_override_internal): Assign to
1150 targetm.secondary_memory_needed rather than
1151 rs6000_secondary_memory_needed_ptr.
1152 (rs6000_secondary_memory_needed): Match hook interface.
1153 (rs6000_debug_secondary_memory_needed): Likewise.
1154 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
1155 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
1156 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
1157 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1158 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
1159 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
1161 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
1162 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1163 (rs6000_option_override_internal): Assign to
1164 targetm.secondary_memory_needed rather than
1165 rs6000_secondary_memory_needed_ptr.
1166 (rs6000_secondary_memory_needed): Match hook interface.
1167 (rs6000_debug_secondary_memory_needed): Likewise.
1168 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
1169 * config/s390/s390.c (s390_secondary_memory_needed): New function.
1170 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1171 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
1172 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1173 (sparc_secondary_memory_needed): New function.
1174 * lra-constraints.c (check_and_process_move): Refer to
1175 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
1177 (curr_insn_transform): Likewise.
1178 (process_alt_operands): Use targetm.secondary_memory_needed
1179 instead of TARGET_SECONDARY_MEMORY_NEEDED.
1180 (check_secondary_memory_needed_p): Likewise.
1181 (choose_split_class): Likewise.
1182 * reload.c: Unconditionally include code that was previously
1183 conditional on SECONDARY_MEMORY_NEEDED.
1184 (push_secondary_reload): Use targetm.secondary_memory_needed
1185 instead of TARGET_SECONDARY_MEMORY_NEEDED.
1186 (push_reload): Likewise.
1187 * reload1.c: Unconditionally include code that was previously
1188 conditional on SECONDARY_MEMORY_NEEDED.
1189 (choose_reload_regs): Use targetm.secondary_memory_needed
1190 instead of TARGET_SECONDARY_MEMORY_NEEDED.
1191 (gen_reload): Likewise.
1192 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
1194 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
1195 Alan Hayward <alan.hayward@arm.com>
1196 David Sherwood <david.sherwood@arm.com>
1198 * target.def (secondary_memory_needed_mode): New hook:
1199 * targhooks.c (default_secondary_memory_needed_mode): Declare.
1200 * targhooks.h (default_secondary_memory_needed_mode): New function.
1201 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
1202 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
1203 * doc/tm.texi: Regenerate.
1204 * lra-constraints.c (check_and_process_move): Use
1205 targetm.secondary_memory_needed_mode instead of
1206 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
1207 (curr_insn_transform): Likewise.
1208 * reload.c (get_secondary_mem): Likewise.
1209 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1210 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
1212 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1213 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1214 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
1215 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1216 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
1218 * config/powerpcspe/powerpcspe-protos.h
1219 (rs6000_secondary_memory_needed_mode): Delete.
1220 * config/powerpcspe/powerpcspe.c
1221 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1222 (rs6000_secondary_memory_needed_mode): Make static.
1223 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1224 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
1226 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
1228 (rs6000_secondary_memory_needed_mode): Make static.
1229 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1230 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
1231 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1232 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1233 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
1235 (sparc_secondary_memory_needed_mode): New function.
1236 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
1238 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
1240 * config/aarch64/constraints.md (Umq): New constraint.
1241 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
1243 (mov<mode>): Update condition.
1245 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1247 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
1248 when bitposition is the same.
1250 2017-09-13 Richard Biener <rguenther@suse.de>
1252 * dwarf2out.c (output_die_symbol): Remove.
1253 (output_die): Do not output a DIEs symbol.
1255 2017-09-13 Richard Biener <rguenther@suse.de>
1258 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
1259 default-def to avoid breaking iterator update with the weird
1260 interaction with cgraph_update_edges_for_call_stmt_node.
1262 2017-09-13 Richard Biener <rguenther@suse.de>
1264 * tree-cfg.c (verify_gimple_assign_binary): Add verification
1265 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
1266 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
1267 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
1269 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
1271 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
1272 Disable pc relative literal load irrespective of
1273 TARGET_FIX_ERR_A53_84341 for default.
1275 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
1277 * config/sparc/sparc.c (output_return): Output the source location of
1278 the insn in the delay slot, if any.
1279 (output_sibcall): Likewise.
1281 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
1284 * common.opt (-static-pie): New alias.
1285 (shared): Negate static-pie.
1286 (-no-pie): Update help text.
1288 (static-pie): New option.
1289 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
1290 -static-pie support.
1291 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
1292 (LINK_EH_SPEC): Likewise.
1293 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1294 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
1295 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
1296 * gcc.c (LINK_COMMAND_SPEC): Likewise.
1297 (init_gcc_specs): Likewise.
1298 (init_spec): Likewise.
1299 (display_help): Update help message for -pie.
1300 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
1303 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
1305 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
1306 (movdi_aarch64): Likewise.
1307 (movti_aarch64): Likewise.
1309 2017-09-12 Simon Wright <simon@pushface.org>
1312 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
1313 calculation of the minor version, always output as 0.
1315 2017-09-12 Jakub Jelinek <jakub@redhat.com>
1318 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
1319 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
1320 on it early, rather than manual conversion late. For
1321 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
1322 instead of performing manual conversion.
1324 2017-09-12 Carl Love <cel@us.ibm.com>
1326 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
1327 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
1329 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
1330 VMULOSW): Add definitions.
1331 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1332 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
1333 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
1334 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
1335 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
1337 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
1339 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
1341 (movti_aarch64): Likewise.
1342 (movdf_aarch64): Likewise.
1343 (movtf_aarch64): Likewise.
1344 (load_pairdi): Likewise.
1345 (store_pairdi): Likewise.
1346 (load_pairdf): Likewise.
1347 (store_pairdf): Likewise.
1348 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
1349 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
1350 (ldr_got_small_<mode>): Likewise.
1351 (ldr_got_small_28k_<mode>): Likewise.
1352 (ldr_got_tiny): Likewise.
1353 * config/aarch64/iterators.md (ldst_sz): New.
1354 (ldpstp_sz): Likewise.
1355 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
1357 (thunderx_load): Split load_8 to load_16.
1358 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
1360 (thunderx2t99_storepair_basic): Split store_8 to store_16.
1361 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
1362 (xgene1_store_pair): Split store_8 to store_16.
1363 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
1364 (falkor_st_0_st_sd): Split store_8 to store_16.
1366 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
1368 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
1369 and store1/2/3/4 to store_4/8/12/16.
1370 * config/aarch64/aarch64.md: Update for rename.
1371 * config/arm/arm.md: Likewise.: Likewise.
1372 * config/arm/arm.c: Likewise.
1373 * config/arm/thumb1.md: Likewise.
1374 * config/arm/thumb2.md: Likewise.
1375 * config/arm/vfp.md: Likewise.
1376 * config/arm/arm-generic.md: Likewise.
1377 * config/arm/arm1020e.md: Likewise.
1378 * config/arm/arm1026ejs.md: Likewise.
1379 * config/arm/arm1136jfs.md: Likewise.
1380 * config/arm/arm926ejs.md: Likewise.
1381 * config/arm/cortex-a15.md: Likewise.
1382 * config/arm/cortex-a17.md: Likewise.
1383 * config/arm/cortex-a5.md: Likewise.
1384 * config/arm/cortex-a53.md: Likewise.
1385 * config/arm/cortex-a57.md: Likewise.
1386 * config/arm/cortex-a7.md: Likewise.
1387 * config/arm/cortex-a8.md: Likewise.
1388 * config/arm/cortex-a9.md: Likewise.
1389 * config/arm/cortex-m4.md: Likewise.
1390 * config/arm/cortex-m7.md: Likewise.
1391 * config/arm/cortex-r4.md: Likewise.
1392 * config/arm/exynos-m1.md: Likewise.
1393 * config/arm/fa526.md: Likewise.
1394 * config/arm/fa606te.md: Likewise.
1395 * config/arm/fa626te.md: Likewise.
1396 * config/arm/fa726te.md: Likewise.
1397 * config/arm/fmp626.md: Likewise.
1398 * config/arm/iwmmxt.md: Likewise.
1399 * config/arm/ldmstm.md: Likewise.
1400 * config/arm/marvell-pj4.md: Likewise.
1401 * config/arm/xgene1.md: Likewise.
1402 * config/aarch64/thunderx.md: Likewise.
1403 * config/aarch64/thunderx2t99.md: Likewise.
1404 * config/aarch64/falkor.md: Likewise.
1406 2017-09-12 Martin Liska <mliska@suse.cz>
1408 * attribs.c (private_lookup_attribute): New function.
1409 * attribs.h (private_lookup_attribute): Declared here.
1410 (lookup_attribute): Called from this place.
1412 2017-09-12 Richard Biener <rguenther@suse.de>
1414 PR tree-optimization/82157
1415 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
1416 stmts with side-effects.
1418 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1419 Alan Hayward <alan.hayward@arm.com>
1420 David Sherwood <david.sherwood@arm.com>
1422 * target.def (hard_regno_nregs): New hook.
1423 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
1424 * targhooks.h (default_hard_regno_nregs): Declare.
1425 * targhooks.c (default_hard_regno_nregs): New function.
1426 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
1427 (TARGET_HARD_REGNO_NREGS): ...this hook.
1428 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
1429 (CLASS_MAX_NREGS): Likewise.
1430 * doc/tm.texi: Regenerate.
1431 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
1432 instead of HARD_REGNO_NREGS.
1433 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
1434 HARD_REGNO_NREGS in the comment.
1435 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
1436 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
1437 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
1438 Return an unsigned int.
1439 (TARGET_HARD_REGNO_NREGS): Redefine.
1440 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
1441 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
1442 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
1443 (arc_hard_regno_nregs): New function.
1444 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
1445 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
1446 (arm_hard_regno_nregs): New function.
1447 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
1448 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
1449 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
1450 (TARGET_HARD_REGNO_NREGS): Redefine.
1451 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
1452 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
1453 (HARD_REGNO_NREGS): Delete.
1454 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
1455 (cr16_hard_regno_nregs): New function.
1456 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
1457 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
1458 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
1459 (cris_hard_regno_nregs): New function.
1460 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
1461 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
1462 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
1463 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
1464 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
1465 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
1466 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
1467 (frv_hard_regno_nregs): Make static. Take and return an
1469 (frv_class_max_nregs): Remove outdated copy of documentation.
1470 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
1471 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
1472 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
1473 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
1474 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
1475 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
1476 (TARGET_HARD_REGNO_NREGS): Redefine.
1477 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
1478 (CLASS_MAX_NREGS): Update comment.
1479 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
1480 (ia64_hard_regno_nregs): New function.
1481 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
1482 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
1483 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
1484 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
1485 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
1487 (m32c_hard_regno_nregs): Likewise. Make static.
1488 (TARGET_HARD_REGNO_NREGS): Redefine.
1489 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
1490 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
1491 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
1492 (m68k_hard_regno_nregs): New function.
1493 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
1494 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
1495 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
1496 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
1497 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
1498 Take and return an unsigned int.
1499 (TARGET_HARD_REGNO_NREGS): Redefine.
1500 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
1501 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
1502 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
1503 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
1504 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
1505 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
1506 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
1507 (msp430_hard_regno_nregs): Make static. Take and return an
1509 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
1510 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
1511 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
1512 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
1513 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
1514 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
1515 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
1516 (TARGET_HARD_REGNO_NREGS): Redefine.
1517 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
1518 (PA_HARD_REGNO_NREGS): ...this.
1519 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
1520 (PA_HARD_REGNO_NREGS): ...this.
1521 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
1522 (pa_hard_regno_nregs): New function.
1523 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
1524 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
1525 (pdp11_hard_regno_nregs): New function.
1526 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
1527 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
1528 (rs6000_hard_regno_nregs_hook): New function.
1529 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
1530 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1531 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
1532 Take and return an unsigned int. Move earlier in file.
1533 (TARGET_HARD_REGNO_NREGS): Redefine.
1534 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
1535 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
1536 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
1537 (rl78_hard_regno_nregs): Make static. Take and return an
1539 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
1540 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
1541 (rs6000_hard_regno_nregs_hook): New function.
1542 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
1543 * config/rx/rx.c (rx_hard_regno_nregs): New function.
1544 (TARGET_HARD_REGNO_NREGS): Redefine.
1545 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
1546 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
1547 instead of HARD_REGNO_NREGS.
1548 (s390_hard_regno_nregs): New function.
1549 (s390_hard_regno_mode_ok): Add comment from s390.h.
1550 (TARGET_HARD_REGNO_NREGS): Redefine.
1551 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
1552 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
1553 (sh_hard_regno_nregs): New function.
1554 (sh_pass_in_reg_p): Use it.
1555 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
1556 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
1557 (sparc_hard_regno_nregs): New function.
1558 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
1559 * config/spu/spu.c (spu_hard_regno_nregs): New function.
1560 (spu_function_arg_advance): Use it, supplying a valid register number.
1561 (TARGET_HARD_REGNO_NREGS): Redefine.
1562 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
1563 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
1564 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
1565 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
1566 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
1567 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
1568 (CLASS_MAX_NREGS): Remove copy of old documentation.
1569 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
1570 (visium_hard_regno_nregs): New function.
1571 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
1572 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
1573 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
1574 xtensa_hard_regno_nregs): New function.
1575 * system.h (HARD_REGNO_NREGS): Poison.
1577 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1579 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
1580 hard_regno_nregs instead of HARD_REGNO_NREGS.
1581 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
1582 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
1583 (c6x_expand_epilogue): Likewise.
1584 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
1585 (frv_read_iacc_argument): Likewise.
1586 * config/sh/sh.c: Include regs.h.
1587 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
1588 (regs_used): Likewise.
1589 (output_stack_adjust): Likewise.
1590 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
1591 * expmed.c: Include regs.h.
1592 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
1593 * ree.c: Include regs.h.
1594 (combine_reaching_defs): Use hard_regno_nregs instead of
1596 (add_removable_extension): Likewise.
1598 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1600 * regs.h (hard_regno_nregs): Turn into a function.
1601 (end_hard_regno): Update accordingly.
1602 * caller-save.c (setup_save_areas): Likewise.
1603 (save_call_clobbered_regs): Likewise.
1604 (replace_reg_with_saved_mem): Likewise.
1605 (insert_restore): Likewise.
1606 (insert_save): Likewise.
1607 * combine.c (can_change_dest_mode): Likewise.
1608 (move_deaths): Likewise.
1609 (distribute_notes): Likewise.
1610 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
1611 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
1612 (rs6000_split_multireg_move): Likewise.
1613 (rs6000_register_move_cost): Likewise.
1614 (rs6000_memory_move_cost): Likewise.
1615 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
1616 (rs6000_split_multireg_move): Likewise.
1617 (rs6000_register_move_cost): Likewise.
1618 (rs6000_memory_move_cost): Likewise.
1619 * cselib.c (cselib_reset_table): Likewise.
1620 (cselib_lookup_1): Likewise.
1621 * emit-rtl.c (set_mode_and_regno): Likewise.
1622 * function.c (aggregate_value_p): Likewise.
1623 * ira-color.c (setup_profitable_hard_regs): Likewise.
1624 (check_hard_reg_p): Likewise.
1625 (calculate_saved_nregs): Likewise.
1626 (assign_hard_reg): Likewise.
1627 (improve_allocation): Likewise.
1628 (calculate_spill_cost): Likewise.
1629 * ira-emit.c (modify_move_list): Likewise.
1630 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
1631 (ira_hard_reg_in_set_p): Likewise.
1632 * ira.c (setup_reg_mode_hard_regset): Likewise.
1633 (clarify_prohibited_class_mode_regs): Likewise.
1634 (check_allocation): Likewise.
1635 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1636 (lra_setup_reg_renumber): Likewise.
1637 (setup_try_hard_regno_pseudos): Likewise.
1638 (spill_for): Likewise.
1639 (assign_hard_regno): Likewise.
1640 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1641 * lra-constraints.c (in_class_p): Likewise.
1642 (lra_constraint_offset): Likewise.
1643 (simplify_operand_subreg): Likewise.
1644 (lra_constraints): Likewise.
1645 (split_reg): Likewise.
1646 (split_if_necessary): Likewise.
1647 (invariant_p): Likewise.
1648 (inherit_in_ebb): Likewise.
1649 * lra-lives.c (process_bb_lives): Likewise.
1650 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
1651 (get_hard_regs): Likewise.
1652 (do_remat): Likewise.
1653 * lra-spills.c (assign_spill_hard_regs): Likewise.
1654 * mode-switching.c (create_pre_exit): Likewise.
1655 * postreload.c (reload_combine_recognize_pattern): Likewise.
1656 * recog.c (peep2_find_free_register): Likewise.
1657 * regcprop.c (kill_value_regno): Likewise.
1658 (set_value_regno): Likewise.
1659 (copy_value): Likewise.
1660 (maybe_mode_change): Likewise.
1661 (find_oldest_value_reg): Likewise.
1662 (copyprop_hardreg_forward_1): Likewise.
1663 * regrename.c (check_new_reg_p): Likewise.
1664 (regrename_do_replace): Likewise.
1665 * reload.c (push_reload): Likewise.
1666 (combine_reloads): Likewise.
1667 (find_dummy_reload): Likewise.
1668 (operands_match_p): Likewise.
1669 (find_reloads): Likewise.
1670 (find_equiv_reg): Likewise.
1671 (reload_adjust_reg_for_mode): Likewise.
1672 * reload1.c (count_pseudo): Likewise.
1673 (count_spilled_pseudo): Likewise.
1674 (find_reg): Likewise.
1675 (clear_reload_reg_in_use): Likewise.
1676 (free_for_value_p): Likewise.
1677 (allocate_reload_reg): Likewise.
1678 (choose_reload_regs): Likewise.
1679 (reload_adjust_reg_for_temp): Likewise.
1680 (emit_reload_insns): Likewise.
1681 (delete_output_reload): Likewise.
1682 * rtlanal.c (subreg_get_info): Likewise.
1683 * sched-deps.c (sched_analyze_reg): Likewise.
1684 * sel-sched.c (init_regs_for_mode): Likewise.
1685 (mark_unavailable_hard_regs): Likewise.
1686 (choose_best_reg_1): Likewise.
1687 (verify_target_availability): Likewise.
1688 * valtrack.c (dead_debug_insert_temp): Likewise.
1689 * var-tracking.c (track_loc_p): Likewise.
1690 (emit_note_insn_var_location): Likewise.
1691 * varasm.c (make_decl_rtl): Likewise.
1692 * reginfo.c (choose_hard_reg_mode): Likewise.
1693 (init_reg_modes_target): Refer directly to
1694 this_target_regs->x_hard_regno_nregs.
1696 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1698 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
1699 instead of hard_regno_nregs.
1701 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1703 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
1704 end_hard_regno instead of hard_regno_nregs.
1705 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
1706 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
1707 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
1708 * ira-color.c (improve_allocation): Likewise.
1709 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1710 * lra-lives.c (mark_regno_live): Likewise.
1711 (mark_regno_dead): Likewise.
1712 * lra-remat.c (operand_to_remat): Likewise.
1713 * lra.c (collect_non_operand_hard_regs): Likewise.
1714 * postreload.c (reload_combine_note_store): Likewise.
1715 (move2add_valid_value_p): Likewise.
1716 * reload.c (regno_clobbered_p): Likewise.
1718 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1720 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
1722 * config/v850/v850.c (v850_reorg): Likewise.
1723 * reload.c (refers_to_regno_for_reload_p): Likewise.
1724 (find_equiv_reg): Likewise.
1725 * reload1.c (reload_reg_reaches_end_p): Likewise.
1727 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1729 * caller-save.c (add_used_regs): Use REG_NREGS instead of
1731 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
1732 * config/arm/arm.c (output_move_neon): Likewise.
1733 (arm_attr_length_move_neon): Likewise.
1734 (neon_split_vcombine): Likewise.
1735 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
1736 (c6x_mark_reg_written): Likewise.
1737 (c6x_dwarf_register_span): Likewise.
1738 * config/i386/i386.c (ix86_save_reg): Likewise.
1739 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
1740 (rws_access_reg): Likewise.
1741 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
1742 * mode-switching.c (create_pre_exit): Likewise.
1743 * ree.c (combine_reaching_defs): Likewise.
1744 (add_removable_extension): Likewise.
1745 * regcprop.c (find_oldest_value_reg): Likewise.
1746 (copyprop_hardreg_forward_1): Likewise.
1747 * reload.c (reload_inner_reg_of_subreg): Likewise.
1748 (push_reload): Likewise.
1749 (combine_reloads): Likewise.
1750 (find_dummy_reload): Likewise.
1751 (reload_adjust_reg_for_mode): Likewise.
1752 * reload1.c (find_reload_regs): Likewise.
1753 (forget_old_reloads_1): Likewise.
1754 (reload_reg_free_for_value_p): Likewise.
1755 (reload_adjust_reg_for_temp): Likewise.
1756 (emit_reload_insns): Likewise.
1757 (delete_output_reload): Likewise.
1758 * sel-sched.c (choose_best_reg_1): Likewise.
1759 (choose_best_pseudo_reg): Likewise.
1761 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1762 Alan Hayward <alan.hayward@arm.com>
1763 David Sherwood <david.sherwood@arm.com>
1765 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
1766 * target.def (slow_unaligned_access): New hook.
1767 * targhooks.h (default_slow_unaligned_access): Declare.
1768 * targhooks.c (default_slow_unaligned_access): New function.
1769 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
1770 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
1771 * doc/tm.texi: Regenerate.
1772 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
1773 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
1774 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
1776 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
1777 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
1779 (rs6000_slow_unaligned_access): New function.
1780 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
1781 (expand_block_compare): Likewise.
1782 (expand_strn_compare): Likewise.
1783 (rs6000_rtx_costs): Likewise.
1784 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
1785 (riscv_slow_unaligned_access): Likewise.
1786 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
1787 (riscv_slow_unaligned_access_p): ...this and make static.
1788 (riscv_option_override): Update accordingly.
1789 (riscv_slow_unaligned_access): New function.
1790 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
1791 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
1792 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
1793 (rs6000_slow_unaligned_access): New function.
1794 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
1795 (rs6000_rtx_costs): Likewise.
1796 * config/rs6000/rs6000-string.c (expand_block_compare)
1797 (expand_strn_compare): Use targetm.slow_unaligned_access instead
1798 of SLOW_UNALIGNED_ACCESS.
1799 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
1800 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
1801 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
1802 of SLOW_UNALIGNED_ACCESS.
1803 * expmed.c (simple_mem_bitfield_p): Likewise.
1804 * expr.c (alignment_for_piecewise_move): Likewise.
1805 (emit_group_load_1): Likewise.
1806 (emit_group_store): Likewise.
1807 (copy_blkmode_from_reg): Likewise.
1808 (emit_push_insn): Likewise.
1809 (expand_assignment): Likewise.
1810 (store_field): Likewise.
1811 (expand_expr_real_1): Likewise.
1812 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1813 * lra-constraints.c (simplify_operand_subreg): Likewise.
1814 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
1815 * gimple-ssa-store-merging.c: Likewise in block comment at start
1817 * tree-ssa-strlen.c: Include target.h.
1818 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
1819 of SLOW_UNALIGNED_ACCESS.
1820 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
1822 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1824 PR rtl-optimization/82185
1825 * expmed.c (emit_store_flag_int): Only test tem if it has been
1828 2017-09-12 Richard Biener <rguenther@suse.de>
1831 * match.pd ((FTYPE) N CMP CST): Fix typo.
1833 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
1835 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
1837 (mips_near_type_p): Add 'short_call' attribute as a synonym
1839 * doc/extend.texi (short_call): Document new function attribute.
1841 2017-09-12 Jakub Jelinek <jakub@redhat.com>
1844 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
1845 assertion check that in the condition.
1846 (get_atomic_generic_size): Likewise. Before testing if parameter
1847 has pointer type, if it has array type, call for C++
1848 default_conversion to perform array-to-pointer conversion.
1850 2017-09-12 Richard Biener <rguenther@suse.de>
1852 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
1853 for operations we cannot scalarize.
1855 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
1857 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
1858 vectors heap vectors. Clean up comments.
1859 Make visited_bbs a reference.
1860 (profitable_jump_thread_path): Make GC
1861 vectors heap vectors. Clean up comments.
1863 (convert_and_register_jump_thread_path): Make GC vectors heap
1865 (check_subpath_and_update_thread_path): Same. Clean up comments.
1866 Make visited_bbs a reference.
1867 (handle_phi): Abstract common code to to
1868 register_jump_thread_path_if_profitable.
1869 Rename VAR_BB to DEF_BB.
1871 Make GC vectors heap vectors.
1872 Make visited_bbs a reference.
1873 (handle_assignment): Same.
1874 (register_jump_thread_path_if_profitable): New.
1875 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
1877 Make GC vectors heap vectors. Clean up comments.
1878 Make visited_bbs a reference.
1879 (find_jump_threads_backwards): Make visited_bbs live in the stack.
1880 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
1883 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
1886 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
1887 words of E_DImode object are reachable by xtensa_uimm8x4 access.
1889 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
1891 Revert r251800 and r251799.
1893 2017-09-11 Martin Jambor <mjambor@suse.cz>
1896 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
1897 arguments in advance.
1898 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
1899 use it to find predecessor edges.
1900 (naive_outof_ssa): Collect vector of predecessors.
1902 2017-09-08 Jason Merrill <jason@redhat.com>
1904 PR c++/70029 - ICE with ref-qualifier and -flto
1905 * langhooks.h (struct lang_hooks_for_types): Add
1906 copy_lang_qualifiers.
1907 * attribs.c (build_type_attribute_qual_variant): Use it.
1908 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
1910 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
1911 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
1913 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
1916 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
1917 (*mulsi3_sp64): New instruction.
1918 (mulsi3): New expander.
1920 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
1922 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
1924 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1926 * sancov.c: Include memmodel.h.
1928 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
1931 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
1934 2017-09-07 Carl Love <cel@us.ibm.com>
1936 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
1937 the sldi instruction.
1939 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
1941 * sancov.c: Include tm_p.h.
1943 2017-09-07 Jakub Jelinek <jakub@redhat.com>
1946 * output.h (switch_to_other_text_partition): New declaration.
1947 * varasm.c (switch_to_other_text_partition): New function.
1948 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
1949 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
1950 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
1951 to the other text partition before emitting LCL label and switch back
1952 after emitting the word after it.
1954 2017-09-07 Richard Biener <rguenther@suse.de>
1956 * passes.def (pass_split_crit_edges): Remove instance before PRE.
1957 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
1958 critical edges here, after loop init.
1959 (pass_data_pre): Remove PROP_no_crit_edges flags.
1960 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
1961 for valueization of call args to avoid leaking VN_TOP.
1962 (visit_use): Assert we do not visit default defs.
1963 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
1964 Use error_mark_node to more easily detect leaking VN_TOP.
1965 All default-defs are varying, not VN_TOP. Mark them visited.
1966 (run_scc_vn): Make code match comment.
1968 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
1970 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
1971 OPTION_MASK_FLOAT128_KEYWORD.
1972 (POWERPC_MASKS): Likewise.
1973 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
1974 support for the -mfloat128-type option, and make -mfloat128
1975 default on PowerPC Linux systems. Define or undefine
1976 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
1977 Define __float128 to be __ieee128 if IEEE 128-bit support is
1978 enabled, or undefine it.
1979 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
1980 Delete defining __FLOAT128_TYPE__.
1981 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
1982 -mfloat128-type option and make -mfloat128 default on PowerPC
1984 (TARGET_FLOAT128_TYPE): Likewise.
1985 (-mfloat128-type): Likewise.
1986 * config/rs6000/rs6000.c (rs6000_option_override_internal):
1987 Delete the -mfloat128-type option and make -mfloat128 default on
1988 PowerPC Linux systems. Always use __ieee128 to be the keyword for
1989 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
1990 128-bit floating point is enabled. Change tests from using
1991 -mfloat128-type to -mfloat128.
1992 (rs6000_mangle_type): Use the correct mangling for the __float128
1993 type even if normal long double is restricted to 64-bits.
1994 (floatn_mode): Enable the _Float128 type by default on VSX Linux
1996 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
1997 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
1998 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
2000 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
2001 documentation for -mfloat128.
2003 2017-09-06 Olivier Hainque <hainque@adacore.com>
2005 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
2007 2017-09-06 Wish Wu <wishwu007@gmail.com>
2008 Jakub Jelinek <jakub@redhat.com>
2010 * asan.c (initialize_sanitizer_builtins): Add
2011 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
2012 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
2013 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
2014 BT_FN_VOID_UINT64_PTR variables.
2015 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
2016 (BT_FN_VOID_UINT16_UINT16): Likewise.
2017 (BT_FN_VOID_UINT32_UINT32): Likewise.
2018 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
2019 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
2020 (BT_FN_VOID_UINT64_PTR): Likewise.
2021 * common.opt (flag_sanitize_coverage): New variable.
2022 (fsanitize-coverage=trace-pc): Remove.
2023 (fsanitize-coverage=): Add.
2024 * flag-types.h (enum sanitize_coverage_code): New enum.
2025 * fold-const.c (fold_range_test): Disable non-short-circuit
2026 optimization if flag_sanitize_coverage.
2027 (fold_truth_andor): Likewise.
2028 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
2029 * opts.c (COVERAGE_SANITIZER_OPT): Define.
2030 (coverage_sanitizer_opts): New array.
2031 (get_closest_sanitizer_option): Add OPTS argument, handle also
2032 OPT_fsanitize_coverage_.
2033 (parse_sanitizer_options): Adjusted to also handle
2034 OPT_fsanitize_coverage_.
2035 (common_handle_option): Add OPT_fsanitize_coverage_.
2036 * sancov.c (instrument_comparison, instrument_switch): New function.
2037 (sancov_pass): Add trace-cmp support.
2038 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
2039 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
2040 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
2041 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
2042 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
2043 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
2044 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
2045 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
2046 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
2047 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
2049 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
2051 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
2052 error. Only quit immediately if parsing is complete.
2053 (BEGIN): Initialize fatal_err and parse_done.
2054 (begin fpu, end fpu): Check number of arguments.
2055 (begin arch, end arch): Likewise.
2056 (begin cpu, end cpu): Likewise.
2057 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
2058 (optalias): Likewise.
2060 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
2062 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
2063 * config/arm/arm-isa.h: Delete. Move definitions to ...
2064 * arm-cpus.in: ... here. Use new feature and fgroup values.
2065 * config/arm/arm.c (arm_option_override): Use lower case for feature
2067 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
2068 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
2069 * config/arm/parsecpu.awk (END): Add new command 'isa'.
2071 (print_isa_bits_for): New function.
2072 (gen_isa): New function.
2073 (gen_comm_data): Use print_isa_bits_for.
2074 (define feature): New keyword.
2075 (define fgroup): New keyword.
2076 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
2077 (arm-isa.h): Add rule to generate file.
2078 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
2079 case for feature bit names.
2081 2017-09-06 Richard Biener <rguenther@suse.de>
2083 * tree-ssa-pre.c (NECESSARY): Remove.
2084 (create_expression_by_pieces): Do not touch pass-local flags.
2085 (insert_into_preds_of_block): Likewise.
2086 (do_pre_regular_insertion): Likewise.
2087 (eliminate_insert): Likewise.
2088 (eliminate_dom_walker::before_dom_children): Likewise.
2089 (fini_eliminate): Do not look at inserted_exprs.
2090 (mark_operand_necessary): Remove.
2091 (remove_dead_inserted_code): Replace with simple work-list
2092 algorithm based on inserted_exprs and SSA uses.
2093 (pass_pre::execute): Re-order fini_eliminate and
2094 remove_dead_inserted_code.
2096 2017-09-06 Olivier Hainque <hainque@adacore.com>
2098 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
2099 for VxWorks 7. Adjust surrounding comments.
2101 2017-09-06 Richard Biener <rguenther@suse.de>
2103 * gimple-ssa-strength-reduction.c
2104 (find_candidates_dom_walker::before_dom_children): Also allow
2107 2017-09-06 Richard Biener <rguenther@suse.de>
2109 PR tree-optimization/82108
2110 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
2111 for gap in the non-permutation SLP case.
2113 2017-09-06 Martin Jambor <mjambor@suse.cz>
2115 PR tree-optimization/82078
2116 * tree-sra.c (sort_and_splice_var_accesses): Move call to
2117 add_access_to_work_queue...
2118 (build_accesses_from_assign): ...here.
2119 (propagate_all_subaccesses): Make sure racc is the group
2120 representative, if there is one.
2122 2017-09-06 Jakub Jelinek <jakub@redhat.com>
2125 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
2128 2017-09-06 Richard Biener <rguenther@suse.de>
2130 * gimple-ssa-strength-reduction.c
2131 (find_candidates_dom_walker::before_doom_children): Use a
2132 type and not a mode check.
2134 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
2137 * config/arm/predicates.md (arm_general_adddi_operand): Create new
2139 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
2141 2017-09-05 Jeff Law <law@redhat.com>
2143 PR tree-optimization/64910
2144 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
2145 cases where we have 3 or more operands.
2147 2017-09-05 Jakub Jelinek <jakub@redhat.com>
2150 * omp-low.c (lower_omp_for): Recompute tree invariant if
2151 gimple_omp_for_initial/final is ADDR_EXPR.
2154 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
2155 into gimple val before gimplification fo the COND_EXPR.
2157 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
2159 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
2160 REGION_COPY argument.
2161 (thread_through_all_blocks): Remove unused argument to
2162 duplicate_thread_path.
2164 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2165 Alan Hayward <alan.hayward@arm.com>
2166 David Sherwood <david.sherwood@arm.com>
2168 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
2169 Take a scalar_mode rather than a machine_mode.
2170 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2171 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
2172 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2173 (aarch64_gen_adjusted_ldpstp): Likewise.
2174 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
2176 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2177 Alan Hayward <alan.hayward@arm.com>
2178 David Sherwood <david.sherwood@arm.com>
2180 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
2181 Take a scalar_int_mode instead of a machine_mode.
2182 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
2183 (aarch64_output_scalar_simd_mov_immediate): Likewise.
2184 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
2185 (aarch64_simd_attr_length_rglist): Delete.
2186 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
2187 a scalar_int_mode instead of a machine_mode.
2188 (aarch64_add_offset): Likewise.
2189 (aarch64_internal_mov_immediate): Likewise
2190 (aarch64_add_constant_internal): Likewise.
2191 (aarch64_add_constant): Likewise.
2192 (aarch64_movw_imm): Likewise.
2193 (aarch64_rtx_arith_op_extract_p): Likewise.
2194 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
2195 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
2196 Remove assert that the mode isn't a vector.
2197 (aarch64_output_scalar_simd_mov_immediate): Likewise.
2198 (aarch64_expand_mov_immediate): Update calls after above changes.
2199 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
2200 (aarch64_and_bitmask_imm): Check for scalar integer modes.
2201 (aarch64_move_imm): Likewise.
2202 (aarch64_can_const_movi_rtx_p): Likewise.
2203 (aarch64_strip_extend): Likewise.
2204 (aarch64_extr_rtx_p): Likewise.
2205 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
2206 a CONST_INT when the mode parameter is VOIDmode.
2207 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
2209 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2211 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
2212 * stor-layout.c (bitwise_mode_for_mode): Likewise.
2213 (bitwise_type_for_mode): Update accordingly.
2215 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2217 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
2218 * stor-layout.c (mode_for_size_tree): Likewise.
2219 (mode_for_array): Update accordingly.
2220 (layout_decl): Likewise.
2221 (compute_record_mode): Likewise. Only set the mode once.
2223 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2225 * target.def (get_mask_mode): Change return type to opt_mode.
2227 * doc/tm.texi: Regenerate.
2228 * targhooks.h (default_get_mask_mode): Return an opt_mode.
2229 * targhooks.c (default_get_mask_mode): Likewise.
2230 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
2231 * optabs-query.c (can_vec_mask_load_store_p): Update use of
2232 targetm.get_mask_mode.
2233 * tree.c (build_truth_vector_type): Likewise.
2235 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2237 * machmode.h (mode_for_vector): Return an opt_mode.
2238 * stor-layout.c (mode_for_vector): Likewise.
2239 (mode_for_int_vector): Update accordingly.
2240 (layout_type): Likewise.
2241 * config/i386/i386.c (emit_memmov): Likewise.
2242 (ix86_expand_set_or_movmem): Likewise.
2243 (ix86_expand_vector_init): Likewise.
2244 (ix86_get_mask_mode): Likewise.
2245 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
2247 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
2248 * expmed.c (extract_bit_field_1): Likewise.
2249 * expr.c (expand_expr_real_2): Likewise.
2250 * optabs-query.c (can_vec_perm_p): Likewise.
2251 (can_vec_mask_load_store_p): Likewise.
2252 * optabs.c (expand_vec_perm): Likewise.
2253 * targhooks.c (default_get_mask_mode): Likewise.
2254 * tree-vect-stmts.c (vectorizable_store): Likewise.
2255 (vectorizable_load): Likewise.
2256 (get_vectype_for_scalar_type_and_size): Likewise.
2258 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2260 * machmode.h (mode_for_int_vector): New function.
2261 * stor-layout.c (mode_for_int_vector): Likewise.
2262 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
2263 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
2264 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
2265 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
2266 (s390_expand_vcond): Likewise.
2268 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2270 * machmode.h (opt_machine_mode): New type.
2271 (opt_mode<T>): Allow construction from anything that can be
2273 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
2274 (mode_for_size): Return an opt_machine_mode.
2275 * stor-layout.c (mode_for_size): Likewise.
2276 (mode_for_size_tree): Update call accordingly.
2277 (bitwise_mode_for_mode): Likewise.
2278 (make_fract_type): Likewise.
2279 (make_accum_type): Likewise.
2280 * caller-save.c (replace_reg_with_saved_mem): Update call
2282 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2283 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2284 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2285 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2286 * expmed.c (extract_bit_field_1): Likewise.
2287 * reload.c (get_secondary_mem): Likewise.
2288 * varasm.c (assemble_integer): Likewise.
2289 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
2292 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2294 * machmode.h (decimal_float_mode_for_size): New function.
2295 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
2296 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
2297 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
2298 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
2299 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
2300 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
2302 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2304 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
2305 (get_builtin_sync_mode): Likewise.
2306 (expand_ifn_atomic_compare_exchange): Likewise.
2307 (expand_builtin_atomic_clear): Likewise.
2308 (expand_builtin_atomic_test_and_set): Likewise.
2309 (fold_builtin_atomic_always_lock_free): Likewise.
2310 * calls.c (compute_argument_addresses): Likewise.
2311 (emit_library_call_value_1): Likewise.
2312 (store_one_arg): Likewise.
2313 * combine.c (combine_instructions): Likewise.
2314 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
2315 * config/arm/arm.c (arm_function_value): Likewise.
2316 (aapcs_allocate_return_reg): Likewise.
2317 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
2318 * config/i386/i386.c (construct_container): Likewise.
2319 (ix86_gimplify_va_arg): Likewise.
2320 (ix86_expand_sse_cmp): Likewise.
2321 (emit_memmov): Likewise.
2322 (emit_memset): Likewise.
2323 (expand_small_movmem_or_setmem): Likewise.
2324 (ix86_expand_pextr): Likewise.
2325 (ix86_expand_pinsr): Likewise.
2326 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
2327 * config/microblaze/microblaze.c (microblaze_block_move_straight):
2329 * config/mips/mips.c (mips_function_value_1) Likewise.
2330 (mips_block_move_straight): Likewise.
2331 (mips_expand_ins_as_unaligned_store): Likewise.
2332 * config/powerpcspe/powerpcspe.c
2333 (rs6000_darwin64_record_arg_advance_flush): Likewise.
2334 (rs6000_darwin64_record_arg_flush): Likewise.
2335 * config/rs6000/rs6000.c
2336 (rs6000_darwin64_record_arg_advance_flush): Likewise.
2337 (rs6000_darwin64_record_arg_flush): Likewise.
2338 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
2339 (sparc_function_value_1): Likewise.
2340 * config/spu/spu.c (adjust_operand): Likewise.
2341 (spu_emit_branch_or_set): Likewise.
2342 (arith_immediate_p): Likewise.
2343 * emit-rtl.c (gen_lowpart_common): Likewise.
2344 * expr.c (expand_expr_real_1): Likewise.
2345 * function.c (assign_parm_setup_block): Likewise.
2346 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
2347 * reload1.c (alter_reg): Likewise.
2348 * stor-layout.c (mode_for_vector): Likewise.
2349 (layout_type): Likewise.
2351 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2353 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
2354 (spu_convert_move): Likewise.
2355 * lower-subreg.c (resolve_simple_move): Likewise.
2357 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2360 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
2361 define_insn to a define_expand.
2362 (altivec_vsum2sws_direct): New define_insn.
2363 (altivec_vsumsws): Convert from a define_insn to a define_expand.
2365 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
2367 * config/arm/arm.c (arm_option_params_internal): Improve setting of
2370 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
2375 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
2376 realign stack if stack alignment needed is less than incoming
2379 2017-09-05 Marek Polacek <polacek@redhat.com>
2382 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
2385 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
2387 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
2389 2017-09-05 Richard Biener <rguenther@suse.de>
2391 PR tree-optimization/82084
2392 * fold-const.c (can_native_encode_string_p): Handle wide characters.
2394 2017-09-05 Richard Biener <rguenther@suse.de>
2396 PR tree-optimization/82102
2397 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
2399 2017-09-05 Martin Liska <mliska@suse.cz>
2401 PR tree-optimization/82032
2402 * tree-cfg.c (generate_range_test): New function.
2403 * tree-cfg.h (generate_range_test): Declared here.
2404 * tree-cfgcleanup.c (convert_single_case_switch): New function.
2405 (cleanup_control_expr_graph): Use it.
2406 * tree-switch-conversion.c (try_switch_expansion): Remove
2408 (emit_case_nodes): Use generate_range_test.
2410 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
2413 * config/i386/i386.md (*<btsc><mode>_mask): Add
2414 TARGET_USE_BT to insn constraint.
2415 (*btr<mode>_mask): Ditto.
2417 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
2419 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
2420 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
2422 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2425 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
2426 TARGET_NEON and TARGET_IWMMXT.
2427 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
2428 TARGET_NEON and TARGET_IWMMXT.
2429 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
2431 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
2433 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
2434 (ix86_rewrite_tls_address): Ditto.
2435 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
2436 (ix86_rewrite_tls_address_1): Ditto.
2437 (ix86_rewrite_tls_address): Ditto.
2438 * config/i386/predicates.md (tls_address_pattern): New predicate.
2439 * config/i386/i386.md (TLS address splitter): New splitter.
2441 2017-09-04 Richard Biener <rguenther@suse.de>
2443 PR tree-optimization/82084
2444 * fold-const.h (can_native_encode_string_p): Declare.
2445 * fold-const.c (can_native_encode_string_p): Factor out from ...
2446 (native_encode_string): ... here.
2447 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
2448 vectorizing stores from constants we later cannot handle.
2450 2017-09-04 Marek Polacek <polacek@redhat.com>
2453 * doc/invoke.texi: Update -Wtautological-compare documentation.
2455 2017-09-04 Jeff Law <law@redhat.com>
2457 PR tree-optimization/64910
2458 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
2459 swap the first and last operand if the last is a constant.
2461 2017-09-04 Marek Polacek <polacek@redhat.com>
2464 * convert.c (do_narrow): When sanitizing signed integer overflows,
2465 bail out for signed types.
2466 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
2468 2017-09-04 Richard Biener <rguenther@suse.de>
2470 PR tree-optimization/82060
2471 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
2472 Move devirtualization after stmt folding and before EH/AB/noreturn
2473 cleanup to get the stmt refs canonicalized. Use a bool instead
2474 of gimple_modified_p since that doesn't work for NOPs. Schedule
2475 NOPs generated by folding for removal.
2477 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2478 Alan Hayward <alan.hayward@arm.com>
2479 David Sherwood <david.sherwood@arm.com>
2481 * coretypes.h (pad_direction): New enum.
2482 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
2483 (FUNCTION_ARG_PADDING): Likewise.
2484 * target.def (function_arg_padding): New hook.
2485 * targhooks.h (default_function_arg_padding): Declare.
2486 * targhooks.c (default_function_arg_padding): New function.
2487 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
2488 (TARGET_FUNCTION_ARG_PADDING): ...this.
2489 * doc/tm.texi: Regenerate.
2490 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
2491 instead of direction.
2492 (compute_argument_addresses): Likewise.
2493 (load_register_parameters): Likewise.
2494 (emit_library_call_value_1): Likewise.
2495 (store_one_arg): Use targetm.calls.function_arg_padding instead
2496 of FUNCTION_ARG_PADDING.
2497 (must_pass_in_stack_var_size_or_pad): Likewise.
2498 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
2499 (emit_group_store): Likewise.
2500 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
2501 instead of FUNCTION_ARG_PADDING.
2502 (emit_push_insn): Likewise, and propagate enum change throughout
2504 * function.h (direction): Delete.
2505 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
2507 * function.c (assign_parm_find_stack_rtl): Likewise.
2508 (assign_parm_setup_block_p): Likewise.
2509 (assign_parm_setup_block): Likewise.
2510 (gimplify_parameters): Likewise.
2511 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
2512 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
2514 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
2515 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2516 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
2517 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
2518 (aarch64_function_arg_padding): ...this new function.
2519 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
2520 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2521 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
2522 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2523 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
2524 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2525 (arm_pad_arg_upward): Replace with...
2526 (arm_function_arg_padding): ...this new function.
2527 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
2529 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
2530 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
2531 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2532 (ia64_hpux_function_arg_padding): Replace with...
2533 (ia64_function_arg_padding): ...this new function. Use pad_direction
2534 instead of direction. Check for TARGET_HPUX.
2535 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
2536 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2537 (iq2000_function_arg_padding): New function.
2538 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
2539 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
2540 (mips_function_arg_padding): ...this new function.
2541 (mips_pad_reg_upward): Update accordingly.
2542 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2543 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
2544 targetm.calls.function_arg_padding.
2545 (FUNCTION_ARG_PADDING): Delete.
2546 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2547 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
2548 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2549 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
2550 (nios2_block_reg_padding): Return pad_direction instead of direction.
2551 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
2552 instead of direction.
2553 (nios2_function_arg_padding): Likewise. Make static.
2554 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2555 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
2556 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
2557 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
2558 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2559 (pa_function_arg_padding): Make static. Return pad_direction instead
2561 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
2562 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2563 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
2564 instead of direction. Use targetm.calls.function_arg_padding.
2565 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
2566 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
2567 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
2568 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
2569 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
2571 (function_arg_padding): Rename to...
2572 (rs6000_function_arg_padding): ...this. Make static. Return
2573 pad_direction instead of direction.
2574 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
2575 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
2576 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2577 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
2578 instead of direction. Use targetm.calls.function_arg_padding.
2579 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
2580 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
2581 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
2582 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
2583 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2584 (function_arg_padding): Rename to...
2585 (rs6000_function_arg_padding): ...this. Make static. Return
2586 pad_direction instead of direction.
2587 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
2588 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
2589 * config/s390/s390.c (s390_function_arg_padding): New function.
2590 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2591 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
2592 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
2593 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2594 (function_arg_padding): Rename to...
2595 (sparc_function_arg_padding): ...this. Make static. Return
2596 pad_direction instead of direction.
2597 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
2598 * config/spu/spu.c (spu_function_arg_padding): New function.
2599 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2600 * system.h (FUNCTION_ARG_PADDING): Poison.
2602 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2603 Alan Hayward <alan.hayward@arm.com>
2604 David Sherwood <david.sherwood@arm.com>
2606 * target.def (modes_tieable_p): New hook.
2607 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
2608 (TARGET_MODES_TIEABLE_P): ...this.
2609 * doc/tm.texi.in: Regenerate.
2610 * hooks.h (hook_bool_mode_mode_true): Declare.
2611 * hooks.c (hook_bool_mode_mode_true): New function.
2612 * combine.c (subst): Use targetm.modes_tieable_p instead of
2614 * dse.c (find_shift_sequence): Likewise.
2615 * expmed.c (extract_low_bits): Likewise.
2616 * lower-subreg.c: Include target.h.
2617 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
2619 * rtlanal.c (rtx_cost): Likewise.
2620 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
2621 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
2622 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
2623 (TARGET_MODES_TIEABLE_P): Redefine.
2624 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
2625 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
2626 (TARGET_MODES_TIEABLE_P): Redefine.
2627 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
2628 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
2629 (arc_modes_tieable_p): New function.
2630 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
2631 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
2632 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
2633 (arm_modes_tieable_p): Make static.
2634 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
2635 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
2636 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
2637 (TARGET_MODES_TIEABLE_P): Redefine.
2638 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
2639 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
2640 (TARGET_MODES_TIEABLE_P): Redefine.
2641 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
2642 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
2643 (cr16_modes_tieable_p): New function.
2644 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
2645 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
2646 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
2647 (TRULY_NOOP_TRUNCATION): Update comment.
2648 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
2649 (TRULY_NOOP_TRUNCATION): Update comment.
2650 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
2651 (frv_modes_tieable_p): New function.
2652 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
2653 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
2654 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
2655 (TARGET_MODES_TIEABLE_P): Redefine.
2656 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
2657 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
2658 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
2659 (TARGET_MODES_TIEABLE_P): Redefine.
2660 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
2661 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
2662 (ia64_modes_tieable_p): New function.
2663 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
2664 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
2665 (iq2000_modes_tieable_p): New function.
2666 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
2667 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
2668 (lm32_modes_tieable_p): New function.
2669 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
2670 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
2671 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
2672 (TARGET_MODES_TIEABLE_P): Redefine.
2673 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
2674 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
2675 (m32r_modes_tieable_p): New function.
2676 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
2677 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
2678 (m68k_modes_tieable_p): New function.
2679 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
2680 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
2681 (mcore_modes_tieable_p): New function.
2682 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
2683 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
2685 (TARGET_MODES_TIEABLE_P): Redefine.
2686 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
2687 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
2688 * config/mips/mips.c (mips_modes_tieable_p): Make static.
2689 (TARGET_MODES_TIEABLE_P): Redefine.
2690 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
2691 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
2692 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
2693 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
2694 (mn10300_modes_tieable_p): ...this and make static.
2695 (TARGET_MODES_TIEABLE_P): Redefine.
2696 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
2697 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
2698 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
2699 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
2700 (msp430_modes_tieable_p): Make static.
2701 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
2702 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
2703 (TARGET_MODES_TIEABLE_P): Redefine.
2704 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
2705 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
2706 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
2707 (TARGET_MODES_TIEABLE_P): Redefine.
2708 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
2709 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
2710 * config/pa/pa.c (pa_modes_tieable_p): Make static.
2711 (TARGET_MODES_TIEABLE_P): Redefine.
2712 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
2713 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
2714 (pdp11_modes_tieable_p): New function.
2715 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
2716 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
2717 (rs6000_modes_tieable_p): New function.
2718 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
2719 * config/powerpcspe/powerpcspe.md: Update comment.
2720 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
2721 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
2722 (TARGET_MODES_TIEABLE_P): Redefine.
2723 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
2724 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
2725 (rl78_modes_tieable_p): New function.
2726 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
2727 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
2728 (rs6000_modes_tieable_p): New function.
2729 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
2730 * config/rs6000/rs6000.md: Update comment.
2731 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
2732 * config/rx/rx.c (rx_modes_tieable_p): New function.
2733 (TARGET_MODES_TIEABLE_P): Redefine.
2734 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
2735 * config/s390/s390.c (s390_modes_tieable_p): New function.
2736 (TARGET_MODES_TIEABLE_P): Redefine.
2737 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
2738 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
2739 (sh_modes_tieable_p): New function.
2740 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
2741 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
2742 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
2743 (sparc_modes_tieable_p): Make static.
2744 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
2745 * config/spu/spu.c (spu_modes_tieable_p): New function.
2746 (TARGET_MODES_TIEABLE_P): Redefine.
2747 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
2748 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
2749 (TARGET_MODES_TIEABLE_P): Redefine.
2750 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
2751 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
2752 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
2753 * config/v850/v850.c (v850_modes_tieable_p): New function.
2754 (TARGET_MODES_TIEABLE_P): Redefine.
2755 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
2756 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
2757 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
2758 (visium_modes_tieable_p): New function.
2759 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
2760 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
2761 (xtensa_modes_tieable_p): New function.
2762 * system.h (MODES_TIEABLE_P): Poison.
2764 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2765 Alan Hayward <alan.hayward@arm.com>
2766 David Sherwood <david.sherwood@arm.com>
2768 * target.def (hard_regno_mode_ok): New hook.
2769 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
2770 (TARGET_HARD_REGNO_MODE_OK): ...this.
2771 * doc/tm.texi.in: Regenerate.
2772 * hooks.h (hook_bool_uint_mode_true): Declare.
2773 * hooks.c (hook_bool_uint_mode_true): New function.
2774 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
2776 * genpreds.c (write_insn_preds_c): Add an include of target.h.
2777 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
2778 instead of HARD_REGNO_MODE_OK.
2779 * caller-save.c: Include target.h.
2780 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
2782 * combine.c (can_combine_p): Likewise.
2783 (combinable_i3pat): Likewise.
2784 (can_change_dest_mode): Likewise.
2785 * expr.c (init_expr_target): Likewise.
2786 (convert_move): Likewise.
2787 (convert_modes): Likewise.
2788 * ira.c (setup_prohibited_class_mode_regs): Likewise.
2789 (setup_prohibited_mode_move_regs): Likewise.
2790 * ira.h (target_ira): Likewise.
2791 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2792 * lra-constraints.c (process_alt_operands): Likewise.
2793 (split_reg): Likewise.
2794 * recog.c (peep2_find_free_register): Likewise.
2795 * ree.c (combine_reaching_defs): Likewise.
2796 * regcprop.c (maybe_mode_change): Likewise.
2797 * reginfo.c (init_reg_sets_1): Likewise.
2798 (choose_hard_reg_mode): Likewise.
2799 (simplifiable_subregs): Likewise.
2800 * regrename.c (check_new_reg_p): Likewise.
2801 * reload.c (find_valid_class): Likewise.
2802 (find_valid_class_1): Likewise.
2803 (reload_inner_reg_of_subreg): Likewise.
2804 (push_reload): Likewise.
2805 (combine_reloads): Likewise.
2806 (find_dummy_reload): Likewise.
2807 (find_reloads): Likewise.
2808 * reload1.c (find_reg): Likewise.
2809 (set_reload_reg): Likewise.
2810 (allocate_reload_reg): Likewise.
2811 (choose_reload_regs): Likewise.
2812 (reload_adjust_reg_for_temp): Likewise.
2813 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
2814 (simplify_subreg_regno): Likewise.
2815 * sel-sched.c (init_regs_for_mode): Likewise.
2816 * varasm.c (make_decl_rtl): Likewise.
2817 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
2818 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
2820 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
2821 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
2822 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2823 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
2824 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
2825 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2826 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
2827 (arc_mode_class): Delete.
2828 (HARD_REGNO_MODE_OK): Delete.
2829 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2830 (arc_hard_regno_mode_ok): Rename old array to...
2831 (arc_hard_regno_mode_ok_modes): ...this.
2832 (arc_conditional_register_usage): Update accordingly.
2833 (arc_mode_class): Make static.
2834 (arc_hard_regno_mode_ok): New function.
2835 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
2836 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
2837 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2838 (arm_hard_regno_mode_ok): Make static.
2839 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
2841 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
2842 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
2843 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
2845 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2846 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
2847 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
2848 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
2849 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
2850 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2851 * config/bfin/predicates.md (valid_reg_operand): Use
2852 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
2853 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
2854 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
2855 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2856 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
2857 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
2858 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2859 (cr16_hard_regno_mode_ok): Make static and return a bool.
2860 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
2861 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2862 (cris_hard_regno_mode_ok): New function.
2863 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
2864 (epiphany_mode_class): Delete.
2865 (HARD_REGNO_MODE_OK): Delete.
2866 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
2867 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2868 (hard_regno_mode_ok): Rename to...
2869 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
2870 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
2871 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
2873 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
2874 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
2875 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2876 (frv_hard_regno_mode_ok): Make static and return a bool.
2877 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
2879 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
2880 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
2881 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
2882 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
2884 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2885 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
2886 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
2887 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
2889 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2890 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
2891 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2892 (ia64_hard_regno_mode_ok): New function.
2893 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
2894 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2895 (iq2000_hard_regno_mode_ok): New function.
2896 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
2897 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2898 (lm32_hard_regno_mode_ok): New function.
2899 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
2900 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
2901 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
2902 instead of HARD_REGNO_MODE_OK.
2903 (m32c_hard_regno_ok): Rename to...
2904 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
2905 (m32c_cannot_change_mode_class): Update accordingly.
2906 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2907 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
2908 (m32r_mode_class): Delete.
2909 (HARD_REGNO_MODE_OK): Delete.
2910 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2911 (m32r_hard_regno_mode_ok): Rename to...
2912 (m32r_hard_regno_modes): ...this.
2913 (m32r_mode_class): Make static.
2914 (m32r_hard_regno_mode_ok): New function.
2915 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
2916 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
2917 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2918 (m68k_hard_regno_mode_ok): Make static.
2919 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
2920 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2921 (mcore_hard_regno_mode_ok): New function.
2922 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
2923 (HARD_REGNO_MODE_OK): Delete.
2924 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
2926 (microblaze_hard_regno_mode_ok_p): ...this and make static.
2927 (microblaze_hard_regno_mode_ok): New function.
2928 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2929 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
2930 (mips_hard_regno_mode_ok): Delete.
2931 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
2932 (mips_hard_regno_mode_ok_p): ...this and make static.
2933 (mips_hard_regno_mode_ok_p): Rename to...
2934 (mips_hard_regno_mode_ok_uncached): ...this.
2935 (mips_hard_regno_mode_ok): New function.
2936 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
2937 of HARD_REGNO_MODE_OK.
2938 (mips_option_override): Update after above name changes.
2939 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2940 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
2941 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
2942 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
2943 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
2944 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2945 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
2946 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
2947 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
2948 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2949 (msp430_hard_regno_mode_ok): Make static and return a bool.
2950 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
2951 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
2952 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
2954 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2955 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
2956 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
2957 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
2958 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
2959 (PA_HARD_REGNO_MODE_OK): ...this
2960 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
2961 (PA_HARD_REGNO_MODE_OK): ...this.
2962 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2963 (pa_hard_regno_mode_ok): New function.
2964 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
2965 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2966 (pdp11_hard_regno_mode_ok): New function.
2967 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
2968 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
2970 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
2972 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2973 (rs6000_hard_regno_mode_ok): Rename to...
2974 (rs6000_hard_regno_mode_ok_uncached): ...this.
2975 (rs6000_init_hard_regno_mode_ok): Update accordingly.
2976 (rs6000_hard_regno_mode_ok): New function.
2977 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
2978 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
2979 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
2980 (riscv_hard_regno_mode_ok): ...this and make static.
2981 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2982 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
2983 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
2984 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2985 (rl78_hard_regno_mode_ok): Make static and return bool.
2986 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
2987 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
2989 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
2990 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2991 (rs6000_hard_regno_mode_ok): Rename to...
2992 (rs6000_hard_regno_mode_ok_uncached): ...this.
2993 (rs6000_init_hard_regno_mode_ok): Update accordingly.
2994 (rs6000_hard_regno_mode_ok): New function.
2995 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
2996 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
2997 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2998 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
2999 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
3000 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
3001 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3002 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
3003 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
3004 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3005 (sh_hard_regno_mode_ok): Make static.
3006 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
3007 instead of HARD_REGNO_MODE_OK.
3008 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
3009 (sparc_mode_class): Delete.
3010 (HARD_REGNO_MODE_OK): Delete.
3011 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3012 (hard_regno_mode_classes): Make static.
3013 (sparc_mode_class): Likewise.
3014 (sparc_hard_regno_mode_ok): New function.
3015 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
3016 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
3017 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
3019 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3020 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
3021 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
3022 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
3023 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
3024 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3025 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
3026 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
3027 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3028 (visium_hard_regno_mode_ok): New function.
3029 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
3030 instead of HARD_REGNO_MODE_OK.
3031 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
3032 (HARD_REGNO_MODE_OK): Delete.
3033 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
3034 (xtensa_hard_regno_mode_ok_p): ...this and make static.
3035 (xtensa_option_override): Update accordingly.
3036 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3037 (xtensa_hard_regno_mode_ok): New function.
3038 * system.h (HARD_REGNO_MODE_OK): Poison.
3040 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3041 Alan Hayward <alan.hayward@arm.com>
3042 David Sherwood <david.sherwood@arm.com>
3044 * target.def (hard_regno_call_part_clobbered): New hook.
3045 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
3046 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
3047 * doc/tm.texi: Regenerate.
3048 * hooks.h (hook_bool_uint_mode_false): Declare.
3049 * hooks.c (hook_bool_uint_mode_false): New function.
3050 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3051 * cselib.c (cselib_process_insn): Use
3052 targetm.hard_regno_call_part_clobbered instead of
3053 HARD_REGNO_CALL_PART_CLOBBERED.
3054 * ira-conflicts.c (ira_build_conflicts): Likewise.
3055 * ira-costs.c (ira_tune_allocno_costs): Likewise.
3056 * lra-constraints.c (need_for_call_save_p): Likewise.
3057 * lra-lives.c: Include target.h.
3058 (check_pseudos_live_through_calls): Use
3059 targetm.hard_regno_call_part_clobbered instead of
3060 HARD_REGNO_CALL_PART_CLOBBERED.
3061 * regcprop.c: Include target.h.
3062 (copyprop_hardreg_forward_1): Use
3063 targetm.hard_regno_call_part_clobbered instead of
3064 HARD_REGNO_CALL_PART_CLOBBERED.
3065 * reginfo.c (choose_hard_reg_mode): Likewise.
3066 * regrename.c (check_new_reg_p): Likewise.
3067 * reload.c (find_equiv_reg): Likewise.
3068 * reload1.c (emit_reload_insns): Likewise.
3069 * sched-deps.c (deps_analyze_insn): Likewise.
3070 * sel-sched.c (init_regs_for_mode): Likewise.
3071 (mark_unavailable_hard_regs): Likewise.
3072 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
3073 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3074 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
3076 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3077 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3078 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
3080 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
3082 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3083 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3084 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
3086 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3087 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3088 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
3090 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3091 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
3093 * config/powerpcspe/powerpcspe.c
3094 (rs6000_hard_regno_call_part_clobbered): New function.
3095 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3096 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3097 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
3099 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3100 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3101 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
3103 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3104 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3105 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
3107 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3108 Alan Hayward <alan.hayward@arm.com>
3109 David Sherwood <david.sherwood@arm.com>
3111 * rtl.h (subreg_memory_offset): Declare.
3112 * emit-rtl.c (subreg_memory_offset): New function.
3113 * expmed.c (store_bit_field_1): Use it.
3114 * expr.c (undefined_operand_subword_p): Likewise.
3115 * simplify-rtx.c (simplify_subreg): Likewise.
3117 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
3119 PR rtl-optimization/57448
3122 * optabs.c (expand_atomic_load): Place compiler memory barriers if
3123 using atomic_load pattern.
3124 (expand_atomic_store): Likewise.
3126 2017-09-04 Jakub Jelinek <jakub@redhat.com>
3129 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
3130 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
3131 handling. Use replace_call_with_value with NULL instead of
3132 gsi_replace, unlink_stmt_vdef and release_defs.
3134 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
3137 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
3139 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3142 * rtl.h (emit_library_call_value_1): Declare.
3143 (emit_library_call): Replace declaration with a series of overloads.
3144 Remove the parameter count argument.
3145 (emit_library_call_value): Likewise.
3146 * calls.c (emit_library_call_value_1): Make global. Replace varargs
3147 with an "rtx_mode_t *".
3148 (emit_library_call_value): Delete.
3149 (emit_library_call): Likewise.
3150 * asan.c (asan_emit_stack_protection): Update calls accordingly.
3151 (asan_emit_allocas_unpoison): Likewise.
3152 * builtins.c (expand_builtin_powi): Likewise.
3153 (expand_asan_emit_allocas_unpoison): Likewise.
3154 * cfgexpand.c (expand_main_function): Likewise.
3155 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
3156 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
3157 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
3158 * config/arm/arm.c (arm_trampoline_init): Likewise.
3159 (arm_call_tls_get_addr): Likewise.
3160 (arm_expand_divmod_libfunc): Likewise.
3161 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
3162 (smulsi3_highpart): Likewise.
3163 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
3164 (c6x_expand_compare): Likewise.
3165 (c6x_expand_movmem): Likewise.
3166 * config/frv/frv.c (frv_trampoline_init): Likewise.
3167 * config/i386/i386.c (ix86_trampoline_init): Likewise.
3168 (ix86_expand_divmod_libfunc): Likewise.
3169 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
3170 (ia64_expand_compare): Likewise.
3171 (ia64_profile_hook): Likewise.
3172 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
3173 (nonlocal_goto): Likewise.
3174 (restore_stack_nonlocal): Likewise.
3175 * config/m32r/m32r.c (block_move_call): Likewise.
3176 (m32r_trampoline_init): Likewise.
3177 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
3178 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
3179 (m68k_call_m68k_read_tp): Likewise.
3180 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
3181 (microblaze_expand_divide): Likewise.
3182 * config/mips/mips.h (mips_args): Likewise.
3183 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
3184 (MIPS_ICACHE_SYNC): Likewise.
3185 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
3186 (nios2_trampoline_init): Likewise.
3187 * config/pa/pa.c (hppa_tls_call): Likewise.
3188 (pa_trampoline_init): Likewise.
3189 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
3190 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
3191 (expand_strn_compare): Likewise.
3192 (rs6000_generate_compare): Likewise.
3193 (rs6000_expand_float128_convert): Likewise.
3194 (output_profile_hook): Likewise.
3195 (rs6000_trampoline_init): Likewise.
3196 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
3197 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
3198 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
3199 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
3200 (rs6000_generate_compare): Likewise.
3201 (rs6000_expand_float128_convert): Likewise.
3202 (output_profile_hook): Likewise.
3203 (rs6000_trampoline_init): Likewise.
3204 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
3205 * config/sh/sh.c (sh_trampoline_init): Likewise.
3206 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
3207 (sparc_emit_float_lib_cmp): Likewise.
3208 (sparc32_initialize_trampoline): Likewise.
3209 (sparc64_initialize_trampoline): Likewise.
3210 (sparc_profile_hook): Likewise.
3211 * config/spu/spu.c (ea_load_store): Likewise.
3212 * config/spu/spu.md (floatunssidf2): Likewise.
3213 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
3214 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
3215 * config/visium/visium.c (expand_block_move_4): Likewise.
3216 (expand_block_move_2): Likewise.
3217 (expand_block_move_1): Likewise.
3218 (expand_block_set_4): Likewise.
3219 (expand_block_set_2): Likewise.
3220 (expand_block_set_1): Likewise.
3221 (visium_trampoline_init): Likewise.
3222 (visium_profile_hook): Likewise.
3223 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
3224 (xtensa_setup_frame_addresses): Likewise.
3225 (xtensa_trampoline_init): Likewise.
3226 * except.c (sjlj_emit_function_enter): Likewise.
3227 (sjlj_emit_function_exit): Likewise.
3228 * explow.c (allocate_dynamic_stack_space): Likewise.
3229 (probe_stack_range): Likewise.
3230 * expr.c (convert_mode_scalar): Likewise.
3231 * optabs.c (expand_binop): Likewise.
3232 (expand_twoval_binop_libfunc): Likewise.
3233 (expand_unop): Likewise.
3234 (prepare_cmp_insn): Likewise.
3235 (prepare_float_lib_cmp): Likewise.
3236 (expand_float): Likewise.
3237 (expand_fix): Likewise.
3238 (expand_fixed_convert): Likewise.
3239 (maybe_emit_sync_lock_test_and_set): Likewise.
3240 (expand_atomic_compare_and_swap): Likewise.
3241 (expand_mem_thread_fence): Likewise.
3242 (expand_atomic_fetch_op): Likewise.
3244 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
3246 * doc/generic.texi (OpenACC): Adjust URL.
3247 * doc/invoke.texi (C Dialect Options): Ditto.
3249 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
3251 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
3252 predicate for operand 1. Add (m,<S>) constraint.
3253 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
3254 Prevent memory operand 1 with register operand 2.
3256 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
3258 PR rtl-optimization/82024
3259 * combine.c (try_combine): If the combination result is a PARALLEL,
3260 and we only need to retain the SET in there that would be placed
3261 at I2, check that we can place that at I3 instead, before doing so.
3263 2017-09-01 Jakub Jelinek <jakub@redhat.com>
3266 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
3268 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
3269 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
3272 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
3273 Jeff Law <law@redhat.com>
3275 * varasm.c (bss_initializer_p): Do not put constants into .bss
3276 (categorize_decl_for_section): Handle bss_initializer_p returning
3277 false when DECL_INITIAL is NULL.
3279 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3282 * config/s390/s390.c (s390_can_inline_p): New function.
3284 2017-09-01 Jeff Law <law@redhat.com>
3286 PR tree-optimization/82052
3287 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
3288 Always initialize the returned slot after a hash table miss
3289 when INSERT is true.
3291 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
3293 * config/s390/s390.md (mem_signal_fence): Remove.
3294 * doc/md.texi (mem_signal_fence): Remove.
3295 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
3297 * target-insns.def (mem_signal_fence): Remove.
3299 2017-09-01 Jakub Jelinek <jakub@redhat.com>
3302 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
3305 * asan.c (create_odr_indicator): Strip name encoding from assembler
3306 name before appending it after __odr_asan_.
3308 2017-09-01 Martin Liska <mliska@suse.cz>
3310 PR tree-optimization/82059
3311 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
3312 frequency only when an edge is redirected.
3314 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3316 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
3317 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
3318 (arc_conditional_register_usage): Remove ARC600 lp_count
3320 (arc_file_start): Emit Tag_ARC_CPU_variation.
3321 (arc_can_use_doloop_p): New conditions to use ZOLs.
3322 (hwloop_fail): New function.
3323 (hwloop_optimize): Likewise.
3324 (hwloop_pattern_reg): Likewise.
3325 (arc_doloop_hooks): New struct, to be used with reorg_loops.
3326 (arc_reorg_loops): New function, calls reorg_loops.
3327 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
3328 (arc600_corereg_hazard): Remove ZOL checking, case handled by
3330 (arc_loop_hazard): Remove function, functionality moved into
3332 (arc_hazard): Remove arc_loop_hazard call.
3333 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
3334 into hwloop_optimize.
3335 (arc_label_align): Remove ZOL handling.
3336 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
3337 * config/arc/arc.md (doloop_begin): Remove pattern.
3338 (doloop_begin_i): Likewise.
3339 (doloop_end_i): Likewise.
3340 (doloop_fallback): Likewise.
3341 (doloop_fallback_m): Likewise.
3342 (doloop_end): Reimplement expand.
3343 (arc_lp): New pattern for LP instruction.
3344 (loop_end): New pattern.
3345 (loop_fail): Likewise.
3346 (decrement_and_branch_until_zero): Likewise.
3347 * config/arc/arc.opt (mlpc-width): New option.
3348 * doc/invoke.texi (mlpc-width): Document option.
3350 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3352 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
3353 (arc_ccfsm_advance): Fix checking for delay slots.
3354 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
3356 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3358 * config/arc/arc.md (movqi_insn): Add stores to save constant long
3360 (movhi_insn): Update store instruction constraint which are saving
3361 6-bit short immediates.
3362 (movsi_insn): Consider also short scaled load operations.
3363 (zero_extendhisi2_i): Use Usd constraint instead of T.
3364 (extendhisi2_i): Add q constraint.
3365 (arc_clzsi2): Add type and length attributes.
3366 (arc_ctzsi2): Likewise.
3367 * config/arc/constraints.md (Usc): Update constraint, the
3368 assembler can parse two relocations for a single instruction.
3370 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3372 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
3373 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
3375 2017-08-31 Olivier Hainque <hainque@adacore.com>
3377 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
3378 match as powerpc-wrs-vxworks*.
3380 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
3382 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
3383 register constraint for by-element operand.
3384 (aarch64_mls_elt_merge<mode>): Likewise.
3386 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
3388 * config/arc/arc.c (arc_can_follow_jump): Check for short
3391 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
3393 * config.gcc: Use g.opt for arc.
3394 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
3395 functionality moved to ...
3396 (legitimate_scaled_address_p): New function, ...here.
3397 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
3398 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
3399 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
3401 (arc_override_options): Handle G option.
3402 (arc_output_pic_addr_const): Correct function definition.
3403 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
3404 (arc_decl_anon_ns_mem_p): Delete.
3405 (arc_in_small_data_p): Overhaul this function to take into
3406 consideration the value given via G option.
3407 (arc_rewrite_small_data_1): Renamed and corrected old
3408 arc_rewrite_small_data function.
3409 (arc_rewrite_small_data): New function.
3410 (small_data_pattern): Don't use pic_offset_table_rtx.
3411 * config/arc/arc.h (CC1_SPEC): Recognize G option.
3412 * config/arc/simdext.md (movmisalignv2hi): Use
3413 prepare_move_operands function.
3415 (movmisalign*): Likewise.
3416 * doc/invoke.texi (ARC options): Document -G option.
3418 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
3420 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
3422 * config/arc/arc.c (arc_print_operand): Output scalled address for
3423 sdata whenever is possible.
3424 (arc_in_small_data_p): Allow sdata for 64bit datum when double
3425 load/stores are available.
3426 (compact_sda_memory_operand): Check for the alignment required by
3427 code density instructions.
3428 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
3430 * config/arc/constraints.md (Usd): Update constraint.
3431 (Us0): New constraint.
3432 (Usc): Update constraint.
3434 2017-08-31 Richard Biener <rguenther@suse.de>
3437 * dwarf2out.c (dwarf2out_early_global_decl): Process each
3440 2017-08-31 Tamar Christina <tamar.christina@arm.com>
3442 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
3443 Resize type_signature.
3445 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3446 Alan Hayward <alan.hayward@arm.com>
3447 David Sherwood <david.sherwood@arm.com>
3449 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
3450 subregs whose inner modes can be stored in GPRs.
3451 (aarch64_classify_index): Likewise.
3453 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3454 Alan Hayward <alan.hayward@arm.com>
3455 David Sherwood <david.sherwood@arm.com>
3457 * config/aarch64/iterators.md (V_cmp_result): Rename to...
3458 (V_INT_EQUIV): ...this.
3459 (v_cmp_result): Rename to...
3460 (v_int_equiv): ...this.
3461 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
3462 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
3463 (copysign<mode>3): Likewise.
3464 (aarch64_simd_bsl<mode>_internal): Likewise.
3465 (aarch64_simd_bsl<mode>): Likewise.
3466 (vec_cmp<mode><mode>): Likewise.
3467 (vcond<mode><mode>): Likewise.
3468 (vcond<v_cmp_mixed><mode>): Likewise.
3469 (vcondu<mode><v_cmp_mixed>): Likewise.
3470 (aarch64_cm<optab><mode>): Likewise.
3471 (aarch64_cmtst<mode>): Likewise.
3472 (aarch64_fac<optab><mode>): Likewise.
3473 (vec_perm_const<mode>): Likewise.
3474 (vcond_mask_<mode><v_cmp_result>): Rename to...
3475 (vcond_mask_<mode><v_int_equiv>): ...this.
3476 (vec_cmp<mode><v_cmp_result>): Rename to...
3477 (vec_cmp<mode><v_int_equiv>): ...this.
3479 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3480 Alan Hayward <alan.hayward@arm.com>
3481 David Sherwood <david.sherwood@arm.com>
3483 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
3485 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
3486 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
3487 (UNSPEC_LD4_DREG): New unspecs.
3488 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
3489 (aarch64_ld2<mode>_dreg_be): Replace with...
3490 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
3492 (aarch64_ld3<mode>_dreg_le)
3493 (aarch64_ld3<mode>_dreg_be): Replace with...
3494 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
3496 (aarch64_ld4<mode>_dreg_le)
3497 (aarch64_ld4<mode>_dreg_be): Replace with...
3498 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
3501 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3503 PR tree-optimization/81987
3504 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
3505 insert an initializer in a location not dominated by the stride
3508 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
3510 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
3511 on the entire header of the finally block in the fallthru case.
3513 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
3515 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
3517 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
3519 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
3520 rs6000_emit_move_from_cr and call renamed function.
3521 (rs6000_emit_prologue): Call renamed functions.
3522 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
3523 movesi_from_cr, remove volatile CRs.
3525 2017-08-30 Jon Beniston <jon@beniston.com>
3526 Richard Biener <rguenther@suse.de>
3528 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
3529 of VECTOR_MODE_P check.
3530 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
3531 element vector types.
3533 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3535 * df.h (df_read_modify_subreg_p): Remove in favor of...
3536 * rtl.h (read_modify_subreg_p): ...this new function. Take a
3537 const_rtx instead of an rtx.
3538 * cprop.c (local_cprop_find_used_regs): Update accordingly.
3539 * df-problems.c (df_word_lr_mark_ref): Likewise.
3540 * ira-lives.c (mark_pseudo_reg_live): Likewise.
3541 (mark_pseudo_reg_dead): Likewise.
3542 (mark_ref_dead): Likewise.
3543 * reginfo.c (init_subregs_of_mode): Likewise.
3544 * sched-deps.c (sched_analyze_1): Likewise.
3545 * df-scan.c (df_def_record_1): Likewise.
3546 (df_uses_record): Likewise.
3547 (df_read_modify_subreg_p): Remove in favor of...
3548 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
3549 const_rtx instead of an rtx.
3551 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3552 Alan Hayward <alan.hayward@arm.com>
3553 David Sherwood <david.sherwood@arm.com>
3555 * rtl.h (partial_subreg_p): New function.
3556 * caller-save.c (save_call_clobbered_regs): Use it.
3557 * calls.c (expand_call): Likewise.
3558 * combine.c (combinable_i3pat): Likewise.
3559 (simplify_set): Likewise.
3560 (make_extraction): Likewise.
3561 (make_compound_operation_int): Likewise.
3562 (gen_lowpart_or_truncate): Likewise.
3563 (force_to_mode): Likewise.
3564 (make_field_assignment): Likewise.
3565 (reg_truncated_to_mode): Likewise.
3566 (record_truncated_value): Likewise.
3567 (move_deaths): Likewise.
3568 * cse.c (record_jump_cond): Likewise.
3569 (cse_insn): Likewise.
3570 * cselib.c (cselib_lookup_1): Likewise.
3571 * expmed.c (extract_bit_field_using_extv): Likewise.
3572 * function.c (assign_parm_setup_reg): Likewise.
3573 * ifcvt.c (noce_convert_multiple_sets): Likewise.
3574 * ira-build.c (create_insn_allocnos): Likewise.
3575 * lra-coalesce.c (merge_pseudos): Likewise.
3576 * lra-constraints.c (match_reload): Likewise.
3577 (simplify_operand_subreg): Likewise.
3578 (curr_insn_transform): Likewise.
3579 * lra-lives.c (process_bb_lives): Likewise.
3580 * lra.c (new_insn_reg): Likewise.
3581 (lra_substitute_pseudo): Likewise.
3582 * regcprop.c (mode_change_ok): Likewise.
3583 (maybe_mode_change): Likewise.
3584 (copyprop_hardreg_forward_1): Likewise.
3585 * reload.c (push_reload): Likewise.
3586 (find_reloads): Likewise.
3587 (find_reloads_subreg_address): Likewise.
3588 * reload1.c (alter_reg): Likewise.
3589 (eliminate_regs_1): Likewise.
3590 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3592 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
3594 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
3595 back to if statements, including unpack.
3597 2017-08-30 Martin Liska <mliska@suse.cz>
3600 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
3602 (func_checker::compare_asm_inputs_outputs): ... this function.
3603 (func_checker::compare_gimple_asm): Use the function to compare
3604 also ASM constrains.
3605 * ipa-icf-gimple.h: Rename the function.
3607 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3608 Alan Hayward <alan.hayward@arm.com>
3609 David Sherwood <david.sherwood@arm.com>
3611 * coretypes.h (complex_mode): New type.
3612 * gdbhooks.py (build_pretty_printer): Handle it.
3613 * machmode.h (complex_mode): New class.
3614 (complex_mode::includes_p): New function.
3615 (is_complex_int_mode): Likewise.
3616 (is_complex_float_mode): Likewise.
3617 * genmodes.c (get_mode_class): Handle complex mode classes.
3618 * function.c (expand_function_end): Use is_complex_int_mode.
3620 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3621 Alan Hayward <alan.hayward@arm.com>
3622 David Sherwood <david.sherwood@arm.com>
3624 * coretypes.h (scalar_mode_pod): New typedef.
3625 * gdbhooks.py (build_pretty_printer): Handle it.
3626 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
3627 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
3628 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
3629 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
3632 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3633 Alan Hayward <alan.hayward@arm.com>
3634 David Sherwood <david.sherwood@arm.com>
3636 * machmode.h (mode_for_vector): Take a scalar_mode instead
3638 * stor-layout.c (mode_for_vector): Likewise.
3639 * explow.c (promote_mode): Use as_a <scalar_mode>.
3640 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
3642 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3643 Alan Hayward <alan.hayward@arm.com>
3644 David Sherwood <david.sherwood@arm.com>
3646 * target.def (preferred_simd_mode): Take a scalar_mode
3647 instead of a machine_mode.
3648 * targhooks.h (default_preferred_simd_mode): Likewise.
3649 * targhooks.c (default_preferred_simd_mode): Likewise.
3650 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
3651 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
3652 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
3653 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
3654 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
3655 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
3656 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
3657 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
3659 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
3660 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
3661 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
3662 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
3663 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
3664 * doc/tm.texi: Regenerate.
3665 * optabs-query.c (can_vec_mask_load_store_p): Return false for
3668 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3669 Alan Hayward <alan.hayward@arm.com>
3670 David Sherwood <david.sherwood@arm.com>
3672 * target.def (scalar_mode_supported_p): Take a scalar_mode
3673 instead of a machine_mode.
3674 * targhooks.h (default_scalar_mode_supported_p): Likewise.
3675 * targhooks.c (default_scalar_mode_supported_p): Likewise.
3676 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
3677 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
3678 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
3679 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
3680 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
3681 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
3682 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
3683 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
3684 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
3685 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
3686 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
3687 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
3689 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
3690 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
3691 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
3692 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
3693 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
3695 * doc/tm.texi: Regenerate.
3697 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3698 Alan Hayward <alan.hayward@arm.com>
3699 David Sherwood <david.sherwood@arm.com>
3701 * coretypes.h (opt_scalar_mode): New typedef.
3702 * gdbhooks.py (build_pretty_printers): Handle it.
3703 * machmode.h (mode_iterator::get_2xwider): Add overload for
3705 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
3707 * expr.c (convert_mode_scalar): Likewise.
3708 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3709 * optabs.c (expand_float): Likewise.
3710 (expand_fix): Likewise.
3711 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3713 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3714 Alan Hayward <alan.hayward@arm.com>
3715 David Sherwood <david.sherwood@arm.com>
3717 * optabs.c (expand_float): Explicitly check for scalars before
3718 using a branching expansion.
3719 (expand_fix): Likewise.
3721 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3722 Alan Hayward <alan.hayward@arm.com>
3723 David Sherwood <david.sherwood@arm.com>
3725 * expr.c (convert_mode): Split scalar handling out into...
3726 (convert_mode_scalar): ...this new function. Treat the modes
3729 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3730 Alan Hayward <alan.hayward@arm.com>
3731 David Sherwood <david.sherwood@arm.com>
3733 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
3735 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
3737 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3738 Alan Hayward <alan.hayward@arm.com>
3739 David Sherwood <david.sherwood@arm.com>
3741 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
3742 rather than a machine_mode.
3743 (fixed_from_string): Likewise.
3744 (fixed_convert): Likewise.
3745 (fixed_convert_from_int): Likewise.
3746 (fixed_convert_from_real): Likewise.
3747 (real_convert_from_fixed): Likewise.
3748 * fixed-value.c (fixed_from_double_int): Likewise.
3749 (fixed_from_string): Likewise.
3750 (fixed_convert): Likewise.
3751 (fixed_convert_from_int): Likewise.
3752 (fixed_convert_from_real): Likewise.
3753 (real_convert_from_fixed): Likewise.
3754 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
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 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
3761 of separate mode class checks. Do not allow vector modes here.
3762 (immed_wide_int_const): Use as_a <scalar_mode>.
3763 * explow.c (trunc_int_for_mode): Likewise.
3764 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
3765 (wi::shwi): Likewise.
3766 (wi::min_value): Likewise.
3767 (wi::max_value): Likewise.
3768 * dwarf2out.c (loc_descriptor): Likewise.
3769 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
3772 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3773 Alan Hayward <alan.hayward@arm.com>
3774 David Sherwood <david.sherwood@arm.com>
3776 * tree.h (SCALAR_TYPE_MODE): New macro.
3777 * expr.c (expand_expr_addr_expr_1): Use it.
3778 (expand_expr_real_2): Likewise.
3779 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
3780 (fold_convert_const_fixed_from_int): Likewise.
3781 (fold_convert_const_fixed_from_real): Likewise.
3782 (native_encode_fixed): Likewise
3783 (native_encode_complex): Likewise
3784 (native_encode_vector): Likewise.
3785 (native_interpret_fixed): Likewise.
3786 (native_interpret_real): Likewise.
3787 (native_interpret_complex): Likewise.
3788 (native_interpret_vector): Likewise.
3789 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
3790 (simd_clone_adjust_argument_types): Likewise.
3791 (simd_clone_init_simd_arrays): Likewise.
3792 (simd_clone_adjust): Likewise.
3793 * stor-layout.c (layout_type): Likewise.
3794 * tree.c (build_minus_one_cst): Likewise.
3795 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
3796 * tree-inline.c (estimate_move_cost): Likewise.
3797 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
3798 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
3799 (vectorizable_reduction): Likewise.
3800 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
3801 (vect_recog_mixed_size_cond_pattern): Likewise.
3802 (check_bool_pattern): Likewise.
3803 (adjust_bool_pattern): Likewise.
3804 (search_type_for_mask_1): Likewise.
3805 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
3806 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3807 (vectorizable_load): Likewise.
3808 (vectorizable_store): Likewise.
3809 * ubsan.c (ubsan_encode_value): Likewise.
3810 * varasm.c (output_constant): Likewise.
3812 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3813 Alan Hayward <alan.hayward@arm.com>
3814 David Sherwood <david.sherwood@arm.com>
3816 * coretypes.h (scalar_mode): New class.
3817 * machmode.h (scalar_mode): Likewise.
3818 (scalar_mode::includes_p): New function.
3819 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
3820 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
3821 * genmodes.c (get_mode_class): Handle remaining scalar modes.
3822 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
3823 * expmed.c (store_bit_field_1): Likewise.
3824 (extract_bit_field_1): Likewise.
3825 * expr.c (write_complex_part): Likewise.
3826 (read_complex_part): Likewise.
3827 (emit_move_complex_push): Likewise.
3828 (expand_expr_real_2): Likewise.
3829 * function.c (assign_parm_setup_reg): Likewise.
3830 (assign_parms_unsplit_complex): Likewise.
3831 * optabs.c (expand_binop): Likewise.
3832 * rtlanal.c (subreg_get_info): Likewise.
3833 * simplify-rtx.c (simplify_immed_subreg): Likewise.
3834 * varasm.c (output_constant_pool_2): Likewise.
3836 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3837 Alan Hayward <alan.hayward@arm.com>
3838 David Sherwood <david.sherwood@arm.com>
3840 * expmed.c (extract_high_half): Use scalar_int_mode and remove
3842 (expmed_mult_highpart_optab): Likewise.
3843 (expmed_mult_highpart): Likewise.
3845 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3846 Alan Hayward <alan.hayward@arm.com>
3847 David Sherwood <david.sherwood@arm.com>
3849 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
3850 instead of a machine_mode.
3851 (builtin_memset_read_str): Likewise.
3852 * builtins.c (c_readstr): Likewise.
3853 (builtin_memcpy_read_str): Likewise.
3854 (builtin_strncpy_read_str): Likewise.
3855 (builtin_memset_read_str): Likewise.
3856 (builtin_memset_gen_str): Likewise.
3857 (expand_builtin_signbit): Use scalar_int_mode for local variables.
3858 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
3859 instead of a machine_mode.
3860 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
3862 (make_extraction): Likewise.
3863 (try_widen_shift_mode): Take and return scalar_int_modes instead
3865 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
3866 a scalar_int_mode instead of a machine_mode.
3867 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
3868 (avr_addr_space_pointer_mode): Likewise.
3869 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
3870 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
3871 (msp430_unwind_word_mode): Likewise.
3872 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
3873 (spu_addr_space_pointer_mode): Likewise.
3874 (spu_addr_space_address_mode): Likewise.
3875 (spu_libgcc_cmp_return_mode): Likewise.
3876 (spu_libgcc_shift_count_mode): Likewise.
3877 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
3878 (rl78_addr_space_pointer_mode): Likewise.
3879 (fl78_unwind_word_mode): Likewise.
3880 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
3882 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
3883 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
3884 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
3885 (mips_valid_pointer_mode): Likewise.
3886 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
3887 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
3888 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
3890 (ft32_addr_space_address_mode): Likewise.
3891 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
3892 scalar_int_mode instead of a machine_mode.
3893 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
3895 (m32c_addr_space_address_mode): Likewise.
3896 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
3897 (rs6000_eh_return_filter_mode): Likewise.
3898 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
3899 (rs6000_eh_return_filter_mode): Likewise.
3900 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
3901 (s390_libgcc_shift_count_mode): Likewise.
3902 (s390_unwind_word_mode): Likewise.
3903 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
3905 * target.def (mode_rep_extended): Likewise.
3906 (valid_pointer_mode): Likewise.
3907 (addr_space.valid_pointer_mode): Likewise.
3908 (eh_return_filter_mode): Return a scalar_int_mode rather than
3910 (libgcc_cmp_return_mode): Likewise.
3911 (libgcc_shift_count_mode): Likewise.
3912 (unwind_word_mode): Likewise.
3913 (addr_space.pointer_mode): Likewise.
3914 (addr_space.address_mode): Likewise.
3915 * doc/tm.texi: Regenerate.
3916 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
3918 (do_jump): Use scalar_int_mode for local variables.
3919 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
3920 rather than a machine_mode.
3921 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
3922 (scompare_loc_descriptor_wide): Likewise.
3923 (scompare_loc_descriptor_narrow): Likewise.
3924 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
3926 * except.c (sjlj_emit_dispatch_table): Likewise.
3927 (expand_builtin_eh_copy_values): Likewise.
3928 * explow.c (convert_memory_address_addr_space_1): Likewise.
3929 Take a scalar_int_mode rather than a machine_mode.
3930 (convert_memory_address_addr_space): Take a scalar_int_mode rather
3931 than a machine_mode.
3932 (memory_address_addr_space): Use scalar_int_mode for local variables.
3933 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
3934 rather than a machine_mode.
3935 * expmed.c (mask_rtx): Likewise.
3936 (init_expmed_one_conv): Likewise.
3937 (expand_mult_highpart_adjust): Likewise.
3938 (extract_high_half): Likewise.
3939 (expmed_mult_highpart_optab): Likewise.
3940 (expmed_mult_highpart): Likewise.
3941 (expand_smod_pow2): Likewise.
3942 (expand_sdiv_pow2): Likewise.
3943 (emit_store_flag_int): Likewise.
3944 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
3946 (extract_low_bits): Likewise.
3947 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
3949 * expr.c (pieces_addr::adjust): Likewise.
3950 (can_store_by_pieces): Likewise.
3951 (store_by_pieces): Likewise.
3952 (clear_by_pieces_1): Likewise.
3953 (expand_expr_addr_expr_1): Likewise.
3954 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
3955 (expand_expr_real_1): Likewise.
3956 (try_casesi): Likewise.
3957 * final.c (shorten_branches): Likewise.
3958 * fold-const.c (fold_convert_const_int_from_fixed): Change the
3959 type of "mode" to machine_mode.
3960 * internal-fn.c (expand_arith_overflow_result_store): Take a
3961 scalar_int_mode rather than a machine_mode.
3962 (expand_mul_overflow): Use scalar_int_mode for local variables.
3963 * loop-doloop.c (doloop_modify): Likewise.
3964 (doloop_optimize): Likewise.
3965 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
3966 than a machine_mode.
3967 (expand_doubleword_shift_condmove): Likewise.
3968 (expand_doubleword_shift): Likewise.
3969 (expand_doubleword_clz): Likewise.
3970 (expand_doubleword_popcount): Likewise.
3971 (expand_doubleword_parity): Likewise.
3972 (expand_absneg_bit): Use scalar_int_mode for local variables.
3973 (prepare_float_lib_cmp): Likewise.
3974 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
3975 rather than a machine_mode.
3976 (convert_memory_address_addr_space): Likewise.
3977 (get_mode_bounds): Likewise.
3978 (get_address_mode): Return a scalar_int_mode rather than a
3980 * rtlanal.c (get_address_mode): Likewise.
3981 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
3982 than a machine_mode.
3983 * targhooks.c (default_mode_rep_extended): Likewise.
3984 (default_valid_pointer_mode): Likewise.
3985 (default_addr_space_valid_pointer_mode): Likewise.
3986 (default_eh_return_filter_mode): Return a scalar_int_mode rather
3987 than a machine_mode.
3988 (default_libgcc_cmp_return_mode): Likewise.
3989 (default_libgcc_shift_count_mode): Likewise.
3990 (default_unwind_word_mode): Likewise.
3991 (default_addr_space_pointer_mode): Likewise.
3992 (default_addr_space_address_mode): Likewise.
3993 * targhooks.h (default_eh_return_filter_mode): Likewise.
3994 (default_libgcc_cmp_return_mode): Likewise.
3995 (default_libgcc_shift_count_mode): Likewise.
3996 (default_unwind_word_mode): Likewise.
3997 (default_addr_space_pointer_mode): Likewise.
3998 (default_addr_space_address_mode): Likewise.
3999 (default_mode_rep_extended): Take a scalar_int_mode rather than
4001 (default_valid_pointer_mode): Likewise.
4002 (default_addr_space_valid_pointer_mode): Likewise.
4003 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
4005 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
4006 rather than a machine_mode.
4007 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
4008 for local variables.
4009 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4010 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
4011 than a machine_mode.
4013 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4014 Alan Hayward <alan.hayward@arm.com>
4015 David Sherwood <david.sherwood@arm.com>
4017 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
4018 the mode argument to scalar_int_mode.
4019 (do_jump_by_parts_zero_rtx): Likewise.
4020 (do_jump_by_parts_equality_rtx): Likewise.
4021 (do_jump_by_parts_greater): Take a mode argument.
4022 (do_jump_by_parts_equality): Likewise.
4023 (do_jump_1): Update calls accordingly.
4025 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4026 Alan Hayward <alan.hayward@arm.com>
4027 David Sherwood <david.sherwood@arm.com>
4029 * is-a.h (safe_dyn_cast): New function.
4030 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
4031 (jump_table_for_label): Likewise.
4032 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
4033 instead of an rtx_insn *.
4034 (shorten_branches): Use dyn_cast instead of LABEL_P and
4035 JUMP_TABLE_DATA_P. Use jump_table_for_label and
4036 rtx_jump_table_data::get_data_mode.
4037 (final_scan_insn): Likewise.
4039 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4040 Alan Hayward <alan.hayward@arm.com>
4041 David Sherwood <david.sherwood@arm.com>
4043 * combine.c (try_combine): Use is_a <scalar_int_mode> when
4044 trying to combine a full-register integer set with a subreg
4047 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4048 Alan Hayward <alan.hayward@arm.com>
4049 David Sherwood <david.sherwood@arm.com>
4051 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
4052 that is always either address_mode or pointer_mode.
4054 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4055 Alan Hayward <alan.hayward@arm.com>
4056 David Sherwood <david.sherwood@arm.com>
4058 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
4059 when the two are known to be equal.
4061 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4062 Alan Hayward <alan.hayward@arm.com>
4063 David Sherwood <david.sherwood@arm.com>
4065 * simplify-rtx.c (simplify_const_unary_operation): Use
4066 is_a <scalar_int_mode> instead of checking for a nonzero
4067 precision. Forcibly convert op_mode to a scalar_int_mode
4068 in that case. More clearly differentiate the operand and
4069 result modes and use the former when deciding what the value
4070 of a count-bits operation should be. Use is_int_mode instead
4071 of checking for a MODE_INT. Remove redundant check for whether
4072 this mode has a zero precision.
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 * optabs.c (widen_leading): Change the type of the mode argument
4079 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
4080 (widen_bswap): Likewise.
4081 (expand_parity): Likewise.
4082 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
4083 (expand_ffs): Likewise.
4084 (epand_unop): Check for scalar integer modes before calling the
4087 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4088 Alan Hayward <alan.hayward@arm.com>
4089 David Sherwood <david.sherwood@arm.com>
4091 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
4093 (expand_expr_real_1): Update call accordingly.
4095 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4096 Alan Hayward <alan.hayward@arm.com>
4097 David Sherwood <david.sherwood@arm.com>
4099 * expmed.c (store_bit_field_using_insv): Add op0_mode and
4100 value_mode arguments. Use scalar_int_mode internally.
4101 (store_bit_field_1): Rename the new integer mode from imode
4102 to op0_mode and use it instead of GET_MODE (op0). Update calls
4103 to store_split_bit_field, store_bit_field_using_insv and
4104 store_fixed_bit_field.
4105 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
4106 Use scalar_int_mode internally. Use a bit count rather than a mode
4107 when calculating the largest bit size for get_best_mode.
4108 Update calls to store_split_bit_field and store_fixed_bit_field_1.
4109 (store_fixed_bit_field_1): Add mode and value_mode arguments.
4110 Remove assertion that OP0 has a scalar integer mode.
4111 (store_split_bit_field): Add op0_mode and value_mode arguments.
4112 Update calls to extract_fixed_bit_field.
4113 (extract_bit_field_using_extv): Add an op0_mode argument.
4114 Use scalar_int_mode internally.
4115 (extract_bit_field_1): Rename the new integer mode from imode to
4116 op0_mode and use it instead of GET_MODE (op0). Update calls to
4117 extract_split_bit_field, extract_bit_field_using_extv and
4118 extract_fixed_bit_field.
4119 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
4120 to extract_split_bit_field and extract_fixed_bit_field_1.
4121 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
4122 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
4124 (extract_split_bit_field): Add an op0_mode argument. Update call
4125 to extract_fixed_bit_field.
4127 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4128 Alan Hayward <alan.hayward@arm.com>
4129 David Sherwood <david.sherwood@arm.com>
4131 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
4132 * explow.c (hard_function_value): Likewise.
4133 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
4134 convert_to_mode call outside the loop.
4135 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
4136 for the mode iterator. Require the mode specified by max_pieces
4138 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
4140 (copy_blkmode_to_reg): Likewise.
4141 (set_storage_via_setmem): Likewise.
4142 * optabs.c (prepare_cmp_insn): Likewise.
4143 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4144 * stor-layout.c (finish_bitfield_representative): Likewise.
4146 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4147 Alan Hayward <alan.hayward@arm.com>
4148 David Sherwood <david.sherwood@arm.com>
4150 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
4151 * expr.c (convert_move): Use them.
4152 (convert_modes): Likewise.
4153 (store_expr_with_bounds): Likewise.
4155 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4156 Alan Hayward <alan.hayward@arm.com>
4157 David Sherwood <david.sherwood@arm.com>
4159 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
4160 parameter for the mode of "x". Remove the "known_x", "known_mode"
4161 and "known_ret" arguments. Change the type of the mode argument
4163 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
4164 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
4165 (reg_num_sign_bit_copies_for_combine): Likewise.
4166 * rtlanal.c (nonzero_bits1): Likewise.
4167 (num_sign_bit_copies1): Likewise.
4168 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
4169 (reg_num_sign_bit_copies_general): Likewise.
4170 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
4171 (reg_nonzero_bits_general): 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 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
4178 than in subroutines. Return 1 for non-integer modes.
4179 (cached_num_sign_bit_copies): Change the type of the mode parameter
4181 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
4182 classes. Handle CONST_INT_P first and then check whether X also
4183 has a scalar integer mode. Check the same thing for inner registers
4184 of a SUBREG and for values that are being extended or truncated.
4186 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4187 Alan Hayward <alan.hayward@arm.com>
4188 David Sherwood <david.sherwood@arm.com>
4190 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
4191 in subroutines. Return the mode mask for non-integer modes.
4192 (cached_nonzero_bits): Change the type of the mode parameter
4194 (nonzero_bits1): Likewise. Remove early exit for other mode
4195 classes. Handle CONST_INT_P first and then check whether X
4196 also has a scalar integer mode.
4198 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4199 Alan Hayward <alan.hayward@arm.com>
4200 David Sherwood <david.sherwood@arm.com>
4202 * expr.c (widest_int_mode_for_size): Make the comment match the code.
4203 Return a scalar_int_mode and assert that the size is greater than
4205 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
4206 (op_by_pieces_d::op_by_pieces_d): Likewise.
4207 (op_by_pieces_d::run): Likewise.
4208 (can_store_by_pieces): Likewise.
4210 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4211 Alan Hayward <alan.hayward@arm.com>
4212 David Sherwood <david.sherwood@arm.com>
4214 * combine.c (extract_left_shift): Add a mode argument and update
4216 (make_compound_operation_int): Change the type of the mode parameter
4217 to scalar_int_mode and update the call to extract_left_shift.
4219 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4220 Alan Hayward <alan.hayward@arm.com>
4221 David Sherwood <david.sherwood@arm.com>
4223 * combine.c (simplify_and_const_int): Change the type of the mode
4224 parameter to scalar_int_mode.
4225 (simplify_and_const_int_1): Likewise. Update recursive call.
4227 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4228 Alan Hayward <alan.hayward@arm.com>
4229 David Sherwood <david.sherwood@arm.com>
4231 * combine.c (simplify_compare_const): Check that the mode is a
4232 scalar_int_mode (rather than VOIDmode) before testing its
4234 (simplify_comparison): Move COMPARISON_P handling out of the
4235 loop and restrict the latter part of the loop to scalar_int_modes.
4236 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
4237 and when considering SUBREG_REGs. Use is_int_mode instead of
4238 checking GET_MODE_CLASS against MODE_INT.
4240 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4241 Alan Hayward <alan.hayward@arm.com>
4242 David Sherwood <david.sherwood@arm.com>
4244 * combine.c (try_widen_shift_mode): Move check for equal modes to...
4245 (simplify_shift_const_1): ...here. Use scalar_int_mode for
4246 shift_unit_mode and for modes involved in scalar shifts.
4248 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4249 Alan Hayward <alan.hayward@arm.com>
4250 David Sherwood <david.sherwood@arm.com>
4252 * combine.c (force_int_to_mode): New function, split out from...
4253 (force_to_mode): ...here. Keep xmode up-to-date and use it
4254 instead of GET_MODE (x).
4256 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4257 Alan Hayward <alan.hayward@arm.com>
4258 David Sherwood <david.sherwood@arm.com>
4260 * optabs-query.h (extraction_insn::struct_mode): Change type to
4261 opt_scalar_int_mode and update comment.
4262 (extraction_insn::field_mode): Change type to scalar_int_mode.
4263 (extraction_insn::pos_mode): Likewise.
4264 * combine.c (make_extraction): Update accordingly.
4265 * optabs-query.c (get_traditional_extraction_insn): Likewise.
4266 (get_optab_extraction_insn): Likewise.
4267 * recog.c (simplify_while_replacing): Likewise.
4268 * expmed.c (narrow_bit_field_mem): Change the type of the mode
4269 parameter to opt_scalar_int_mode.
4271 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4272 Alan Hayward <alan.hayward@arm.com>
4273 David Sherwood <david.sherwood@arm.com>
4275 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
4276 to a scalar_int_mode instead of a machine_mode.
4277 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
4278 (get_best_mode): Return a boolean and use a pointer argument to store
4279 the selected mode. Replace the limit mode parameter with a bit limit.
4280 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
4281 for the values returned by bit_field_mode_iterator::next_mode.
4282 (store_bit_field): Update call to get_best_mode.
4283 (store_fixed_bit_field): Likewise.
4284 (extract_fixed_bit_field): Likewise.
4285 * expr.c (optimize_bitfield_assignment_op): Likewise.
4286 * fold-const.c (optimize_bit_field_compare): Likewise.
4287 (fold_truth_andor_1): Likewise.
4288 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
4289 Update for new type of m_mode.
4290 (get_best_mode): As above.
4292 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4293 Alan Hayward <alan.hayward@arm.com>
4294 David Sherwood <david.sherwood@arm.com>
4296 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
4297 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
4298 (store_bit_field): Check is_a <scalar_int_mode> before calling
4299 strict_volatile_bitfield_p.
4300 (extract_bit_field): Likewise.
4302 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4303 Alan Hayward <alan.hayward@arm.com>
4304 David Sherwood <david.sherwood@arm.com>
4306 * target.def (cstore_mode): Return a scalar_int_mode.
4307 * doc/tm.texi: Regenerate.
4308 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
4309 * targhooks.h (default_cstore_mode): Likewise.
4310 * targhooks.c (default_cstore_mode): Likewise, using a forced
4312 * expmed.c (emit_cstore): Expect the target of the cstore to be
4315 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4316 Alan Hayward <alan.hayward@arm.com>
4317 David Sherwood <david.sherwood@arm.com>
4319 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
4321 (niter_desc): Likewise mode.
4322 (iv_analyze): Add a mode parameter.
4324 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
4325 and change its type to scalar_int_mode.
4326 * loop-iv.c: Update commentary at head of file.
4327 (iv_constant): Pass the mode paraeter before the rtx it describes
4328 and change its type to scalar_int_mode. Remove VOIDmode handling.
4329 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
4330 (iv_extend): Likewise.
4331 (shorten_into_mode): Likewise.
4332 (iv_add): Use scalar_int_mode.
4333 (iv_mult): Likewise.
4334 (iv_shift): Likewise.
4335 (canonicalize_iv_subregs): Likewise.
4336 (get_biv_step_1): Pass the outer_mode parameter before the rtx
4337 it describes and change its mode to scalar_int_mode. Also change
4338 the type of the returned inner_mode to scalar_int_mode.
4339 (get_biv_step): Likewise, turning outer_mode from a pointer
4340 into a direct parameter. Update call to get_biv_step_1.
4341 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
4342 iv_constant and get_biv_step.
4343 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
4344 and change its type to scalar_int_mode. Don't initialise iv->mode
4345 to VOIDmode and remove later checks for its still being VOIDmode.
4346 Update calls to iv_analyze_op and iv_analyze_expr. Check
4347 is_a <scalar_int_mode> when changing the mode under consideration.
4348 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
4349 Update call to iv_analyze_expr.
4350 (iv_analyze_op): Add a mode parameter. Reject subregs whose
4351 inner register is not also a scalar_int_mode. Update call to
4353 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
4354 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
4355 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
4356 separate mode class checks. Update calls to iv_analyze. Remove
4357 fix-up of VOIDmodes after iv_analyze_biv.
4358 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
4359 don't have a scalar_int_mode. Update call to biv_p.
4361 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4362 Alan Hayward <alan.hayward@arm.com>
4363 David Sherwood <david.sherwood@arm.com>
4365 * cfgexpand.c (convert_debug_memory_address): Use
4366 as_a <scalar_int_mode>.
4367 * combine.c (expand_compound_operation): Likewise.
4368 (make_extraction): Likewise.
4369 (change_zero_ext): Likewise.
4370 (simplify_comparison): Likewise.
4371 * cse.c (cse_insn): Likewise.
4372 * dwarf2out.c (minmax_loc_descriptor): Likewise.
4373 (mem_loc_descriptor): Likewise.
4374 (loc_descriptor): Likewise.
4375 * expmed.c (init_expmed_one_mode): Likewise.
4376 (synth_mult): Likewise.
4377 (emit_store_flag_1): Likewise.
4378 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
4379 of a comparison with size.
4380 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
4381 (reduce_to_bit_field_precision): Likewise.
4382 * function.c (expand_function_end): Likewise.
4383 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
4384 * loop-doloop.c (doloop_modify): Likewise.
4385 * optabs.c (expand_binop): Likewise.
4386 (expand_unop): Likewise.
4387 (expand_copysign_absneg): Likewise.
4388 (prepare_cmp_insn): Likewise.
4389 (maybe_legitimize_operand): Likewise.
4390 * recog.c (const_scalar_int_operand): Likewise.
4391 * rtlanal.c (get_address_mode): Likewise.
4392 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4393 (simplify_cond_clz_ctz): Likewise.
4394 * tree-nested.c (get_nl_goto_field): Likewise.
4395 * tree.c (build_vector_type_for_mode): Likewise.
4396 * var-tracking.c (use_narrower_mode): Likewise.
4398 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4399 Alan Hayward <alan.hayward@arm.com>
4400 David Sherwood <david.sherwood@arm.com>
4402 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
4403 * builtins.c (expand_builtin_signbit): Use it.
4404 * cfgexpand.c (expand_debug_expr): Likewise.
4405 * dojump.c (do_jump): Likewise.
4406 (do_compare_and_jump): Likewise.
4407 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
4408 * expmed.c (make_tree): Likewise.
4409 * expr.c (expand_expr_real_2): Likewise.
4410 (expand_expr_real_1): Likewise.
4411 (try_casesi): Likewise.
4412 * fold-const-call.c (fold_const_call_ss): Likewise.
4413 * fold-const.c (unextend): Likewise.
4414 (extract_muldiv_1): Likewise.
4415 (fold_single_bit_test): Likewise.
4416 (native_encode_int): Likewise.
4417 (native_encode_string): Likewise.
4418 (native_interpret_int): Likewise.
4419 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
4420 * internal-fn.c (expand_addsub_overflow): Likewise.
4421 (expand_neg_overflow): Likewise.
4422 (expand_mul_overflow): Likewise.
4423 (expand_arith_overflow): Likewise.
4424 * match.pd: Likewise.
4425 * stor-layout.c (layout_type): Likewise.
4426 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
4427 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
4428 * tree-ssanames.c (get_range_info): Likewise.
4429 * tree-switch-conversion.c (array_value_type) Likewise.
4430 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
4431 (vect_recog_divmod_pattern): Likewise.
4432 (vect_recog_mixed_size_cond_pattern): Likewise.
4433 * tree-vrp.c (extract_range_basic): Likewise.
4434 (simplify_float_conversion_using_ranges): Likewise.
4435 * tree.c (int_fits_type_p): Likewise.
4436 * ubsan.c (instrument_bool_enum_load): Likewise.
4437 * varasm.c (mergeable_string_section): Likewise.
4438 (narrowing_initializer_constant_valid_p): Likewise.
4439 (output_constant): Likewise.
4441 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4442 Alan Hayward <alan.hayward@arm.com>
4443 David Sherwood <david.sherwood@arm.com>
4445 * machmode.h (NARROWEST_INT_MODE): New macro.
4446 * expr.c (alignment_for_piecewise_move): Use it instead of
4447 GET_CLASS_NARROWEST_MODE (MODE_INT).
4448 (push_block): Likewise.
4449 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
4451 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4453 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4454 Alan Hayward <alan.hayward@arm.com>
4455 David Sherwood <david.sherwood@arm.com>
4457 * postreload.c (move2add_valid_value_p): Change the type of the
4458 mode parameter to scalar_int_mode.
4459 (move2add_use_add2_insn): Add a mode parameter and use it instead
4461 (move2add_use_add3_insn): Likewise.
4462 (reload_cse_move2add): Update accordingly.
4464 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4465 Alan Hayward <alan.hayward@arm.com>
4466 David Sherwood <david.sherwood@arm.com>
4468 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
4470 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
4471 * optabs.c (expand_unop): Likewise.
4473 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4474 Alan Hayward <alan.hayward@arm.com>
4475 David Sherwood <david.sherwood@arm.com>
4477 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
4478 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
4479 (popcount_loc_descriptor): Likewise.
4480 (bswap_loc_descriptor): Likewise.
4481 (rotate_loc_descriptor): Likewise.
4482 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
4483 calling the functions above.
4485 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4486 Alan Hayward <alan.hayward@arm.com>
4487 David Sherwood <david.sherwood@arm.com>
4489 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
4491 (try_combine): Likewise.
4492 (simplify_if_then_else): Likewise.
4493 * cse.c (cse_insn): Likewise.
4494 * dwarf2out.c (mem_loc_descriptor): Likewise.
4495 * emit-rtl.c (gen_lowpart_common): Likewise.
4496 * simplify-rtx.c (simplify_truncation): Likewise.
4497 (simplify_binary_operation_1): Likewise.
4498 (simplify_const_relational_operation): Likewise.
4499 (simplify_ternary_operation): Likewise.
4500 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
4502 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4503 Alan Hayward <alan.hayward@arm.com>
4504 David Sherwood <david.sherwood@arm.com>
4506 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
4507 * reload.c (push_reload): Likewise.
4508 (find_reloads): Likewise.
4510 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4511 Alan Hayward <alan.hayward@arm.com>
4512 David Sherwood <david.sherwood@arm.com>
4514 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
4515 (make_compound_operation_int): Likewise.
4516 (change_zero_ext): Likewise.
4517 * expr.c (convert_move): Likewise.
4518 (convert_modes): Likewise.
4519 * fwprop.c (forward_propagate_subreg): Likewise.
4520 * loop-iv.c (get_biv_step_1): Likewise.
4521 * optabs.c (widen_operand): Likewise.
4522 * postreload.c (move2add_valid_value_p): Likewise.
4523 * recog.c (simplify_while_replacing): Likewise.
4524 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4525 (simplify_binary_operation_1): Likewise. Remove redundant
4526 mode equality check.
4528 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4529 Alan Hayward <alan.hayward@arm.com>
4530 David Sherwood <david.sherwood@arm.com>
4532 * combine.c (combine_simplify_rtx): Add checks for
4533 is_a <scalar_int_mode>.
4534 (simplify_if_then_else): Likewise.
4535 (make_field_assignment): Likewise.
4536 (simplify_comparison): Likewise.
4537 * ifcvt.c (noce_try_bitop): Likewise.
4538 * loop-invariant.c (canonicalize_address_mult): Likewise.
4539 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4541 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4542 Alan Hayward <alan.hayward@arm.com>
4543 David Sherwood <david.sherwood@arm.com>
4545 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
4546 is_a <scalar_int_mode> instead of != BLKmode.
4548 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4549 Alan Hayward <alan.hayward@arm.com>
4550 David Sherwood <david.sherwood@arm.com>
4552 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
4553 instead of != VOIDmode.
4554 * combine.c (if_then_else_cond): Likewise.
4555 (change_zero_ext): Likewise.
4556 * dwarf2out.c (mem_loc_descriptor): Likewise.
4557 (loc_descriptor): Likewise.
4558 * rtlanal.c (canonicalize_condition): Likewise.
4559 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
4561 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4562 Alan Hayward <alan.hayward@arm.com>
4563 David Sherwood <david.sherwood@arm.com>
4565 * simplify-rtx.c (simplify_binary_operation_1): Use
4566 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
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 * wide-int.h (int_traits<unsigned char>) New class.
4573 (int_traits<unsigned short>) Likewise.
4574 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
4575 Use GET_MODE_UNIT_PRECISION and remove redundant test for
4577 * combine.c (set_nonzero_bits_and_sign_copies): Use
4578 is_a <scalar_int_mode>.
4579 (find_split_point): Likewise.
4580 (combine_simplify_rtx): Likewise.
4581 (simplify_logical): Likewise.
4582 (expand_compound_operation): Likewise.
4583 (expand_field_assignment): Likewise.
4584 (make_compound_operation): Likewise.
4585 (extended_count): Likewise.
4586 (change_zero_ext): Likewise.
4587 (simplify_comparison): Likewise.
4588 * dwarf2out.c (scompare_loc_descriptor): Likewise.
4589 (ucompare_loc_descriptor): Likewise.
4590 (minmax_loc_descriptor): Likewise.
4591 (mem_loc_descriptor): Likewise.
4592 (loc_descriptor): Likewise.
4593 * expmed.c (init_expmed_one_mode): Likewise.
4594 * lra-constraints.c (lra_constraint_offset): Likewise.
4595 * optabs.c (prepare_libcall_arg): Likewise.
4596 * postreload.c (move2add_note_store): Likewise.
4597 * reload.c (operands_match_p): Likewise.
4598 * rtl.h (load_extend_op): Likewise.
4599 * rtlhooks.c (gen_lowpart_general): Likewise.
4600 * simplify-rtx.c (simplify_truncation): Likewise.
4601 (simplify_unary_operation_1): Likewise.
4602 (simplify_binary_operation_1): Likewise.
4603 (simplify_const_binary_operation): Likewise.
4604 (simplify_const_relational_operation): Likewise.
4605 (simplify_subreg): Likewise.
4606 * stor-layout.c (bitwise_mode_for_mode): Likewise.
4607 * var-tracking.c (adjust_mems): Likewise.
4608 (prepare_call_arguments): Likewise.
4610 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4611 Alan Hayward <alan.hayward@arm.com>
4612 David Sherwood <david.sherwood@arm.com>
4614 * machmode.h (is_int_mode): New fuction.
4615 * combine.c (find_split_point): Use it.
4616 (combine_simplify_rtx): Likewise.
4617 (simplify_if_then_else): Likewise.
4618 (simplify_set): Likewise.
4619 (simplify_shift_const_1): Likewise.
4620 (simplify_comparison): Likewise.
4621 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
4622 * cse.c (notreg_cost): Likewise.
4623 (cse_insn): Likewise.
4624 * cselib.c (cselib_lookup_1): Likewise.
4625 * dojump.c (do_jump_1): Likewise.
4626 (do_compare_rtx_and_jump): Likewise.
4627 * dse.c (get_call_args): Likewise.
4628 * dwarf2out.c (rtl_for_decl_init): Likewise.
4629 (native_encode_initializer): Likewise.
4630 * expmed.c (emit_store_flag_1): Likewise.
4631 (emit_store_flag): Likewise.
4632 * expr.c (convert_modes): Likewise.
4633 (store_field): Likewise.
4634 (expand_expr_real_1): Likewise.
4635 * fold-const.c (fold_read_from_constant_string): Likewise.
4636 * gimple-ssa-sprintf.c (get_format_string): Likewise.
4637 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
4638 * optabs.c (expand_binop): Likewise.
4639 (expand_unop): Likewise.
4640 (expand_abs_nojump): Likewise.
4641 (expand_one_cmpl_abs_nojump): Likewise.
4642 * simplify-rtx.c (mode_signbit_p): Likewise.
4643 (val_signbit_p): Likewise.
4644 (val_signbit_known_set_p): Likewise.
4645 (val_signbit_known_clear_p): Likewise.
4646 (simplify_relational_operation_1): Likewise.
4647 * tree.c (vector_type_mode): Likewise.
4649 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4650 Alan Hayward <alan.hayward@arm.com>
4651 David Sherwood <david.sherwood@arm.com>
4653 * machmode.h (smallest_mode_for_size): Fix formatting.
4654 (smallest_int_mode_for_size): New function.
4655 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
4656 instead of smallest_mode_for_size.
4657 * combine.c (make_extraction): Likewise.
4658 * config/arc/arc.c (arc_expand_movmem): Likewise.
4659 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
4660 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
4661 * config/s390/s390.c (s390_expand_insv): Likewise.
4662 * config/sparc/sparc.c (assign_int_registers): Likewise.
4663 * config/spu/spu.c (spu_function_value): Likewise.
4664 (spu_function_arg): Likewise.
4665 * coverage.c (get_gcov_type): Likewise.
4666 (get_gcov_unsigned_t): Likewise.
4667 * dse.c (find_shift_sequence): Likewise.
4668 * expmed.c (store_bit_field_1): Likewise.
4669 * expr.c (convert_move): Likewise.
4670 (store_field): Likewise.
4671 * internal-fn.c (expand_arith_overflow): Likewise.
4672 * optabs-query.c (get_best_extraction_insn): Likewise.
4673 * optabs.c (expand_twoval_binop_libfunc): Likewise.
4674 * stor-layout.c (layout_type): Likewise.
4675 (initialize_sizetypes): Likewise.
4676 * targhooks.c (default_get_mask_mode): Likewise.
4677 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
4679 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4680 Alan Hayward <alan.hayward@arm.com>
4681 David Sherwood <david.sherwood@arm.com>
4683 * machmode.h (opt_mode::else_blk): New function.
4684 (int_mode_for_mode): Declare.
4685 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
4686 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
4688 * cfgexpand.c (expand_debug_expr): Likewise.
4689 * combine.c (gen_lowpart_or_truncate): Likewise.
4690 (gen_lowpart_for_combine): Likewise.
4691 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
4692 * config/avr/avr.c (avr_to_int_mode): Likewise.
4693 (avr_out_plus_1): Likewise.
4694 (avr_out_plus): Likewise.
4695 (avr_out_round): Likewise.
4696 * config/i386/i386.c (ix86_split_to_parts): Likewise.
4697 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
4698 (s390_expand_vcond): Likewise.
4699 * config/spu/spu.c (spu_split_immediate): Likewise.
4700 (spu_expand_mov): Likewise.
4701 * dse.c (get_stored_val): Likewise.
4702 * expmed.c (store_bit_field_1): Likewise.
4703 (convert_extracted_bit_field): Use int_mode_for_mode instead of
4705 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
4706 (extract_low_bits): Likewise.
4707 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
4708 handling rather than repeating the check.
4709 (emit_group_store): Likewise.
4710 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
4711 * optabs.c (expand_absneg_bit): Likewise.
4712 (expand_copysign_absneg): Likewise.
4713 (expand_copysign_bit): Likewise.
4714 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
4715 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
4716 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
4717 * var-tracking.c (prepare_call_arguments): Likewise.
4718 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
4719 int_mode_for_mode instead of mode_for_size.
4720 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
4722 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4723 Alan Hayward <alan.hayward@arm.com>
4724 David Sherwood <david.sherwood@arm.com>
4726 * machmode.h (int_mode_for_size): New function.
4727 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
4728 instead of mode_for_size.
4729 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
4731 * combine.c (expand_field_assignment): Use int_mode_for_size
4732 instead of mode_for_size.
4733 (make_extraction): Likewise.
4734 (simplify_shift_const_1): Likewise.
4735 (simplify_comparison): Likewise.
4736 * dojump.c (do_jump): Likewise.
4737 * dwarf2out.c (mem_loc_descriptor): Likewise.
4738 * emit-rtl.c (init_derived_machine_modes): Likewise.
4739 * expmed.c (flip_storage_order): Likewise.
4740 (convert_extracted_bit_field): Likewise.
4741 * expr.c (copy_blkmode_from_reg): Likewise.
4742 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
4743 * internal-fn.c (expand_mul_overflow): Likewise.
4744 * lower-subreg.c (simple_move): Likewise.
4745 * optabs-libfuncs.c (init_optabs): Likewise.
4746 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4747 * tree.c (vector_type_mode): Likewise.
4748 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
4749 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
4750 * tree-vect-generic.c (expand_vector_parallel): Likewise.
4751 * tree-vect-stmts.c (vectorizable_load): Likewise.
4752 (vectorizable_store): Likewise.
4754 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4755 Alan Hayward <alan.hayward@arm.com>
4756 David Sherwood <david.sherwood@arm.com>
4758 * coretypes.h (pod_mode): New type.
4759 (scalar_int_mode_pod): New typedef.
4760 * machmode.h (pod_mode): New class.
4761 (int_n_data_t::m): Change type to scalar_int_mode_pod.
4762 * genmodes.c (emit_mode_int_n): Update accordingly.
4763 * lower-subreg.h (target_lower_subreg): Change type to
4764 scalar_int_mode_pod.
4765 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
4766 scalar_int_mode_pod.
4768 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4769 Alan Hayward <alan.hayward@arm.com>
4770 David Sherwood <david.sherwood@arm.com>
4772 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
4773 machine_mode to scalar_int_mode.
4774 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
4775 (rs6000_option_override_internal): Remove cast to int.
4776 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
4777 machine_mode to scalar_int_mode.
4778 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
4779 (rs6000_option_override_internal): Remove cast to int.
4780 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
4781 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
4783 * config/s390/s390.c (s390_expand_builtin): Likewise.
4784 * coretypes.h (scalar_int_mode): New type.
4785 (opt_scalar_int_mode): New typedef.
4786 * machmode.h (scalar_int_mode): New class.
4787 (scalar_int_mode::includes_p): New function.
4788 (byte_mode): Change type to scalar_int_mode.
4789 (word_mode): Likewise.
4790 (ptr_mode): Likewise.
4791 * emit-rtl.c (byte_mode): Likewise.
4792 (word_mode): Likewise.
4793 (ptr_mode): Likewise.
4794 (init_derived_machine_modes): Update accordingly.
4795 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
4796 and MODE_PARTIAL_INT.
4797 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
4798 opt_scalar_int_mode.
4800 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4801 Alan Hayward <alan.hayward@arm.com>
4802 David Sherwood <david.sherwood@arm.com>
4804 * target.def (libgcc_floating_mode_supported_p): Take a
4806 * doc/tm.texi: Regenerate.
4807 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
4809 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
4810 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
4813 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4814 Alan Hayward <alan.hayward@arm.com>
4815 David Sherwood <david.sherwood@arm.com>
4817 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
4818 * doc/tm.texi: Regenerate.
4819 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
4820 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
4821 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
4822 * targhooks.h (default_floatn_mode): Likewise.
4823 * targhooks.c (default_floatn_mode): Likewise.
4824 * tree.c (build_common_tree_nodes): Update accordingly.
4826 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4827 Alan Hayward <alan.hayward@arm.com>
4828 David Sherwood <david.sherwood@arm.com>
4830 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
4831 (mode_iterator::iterate_p): Likewise.
4832 (mode_iterator::get_wider): Likewise.
4833 * expr.c (init_expr_target): Use opt_scalar_float_mode.
4835 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4836 Alan Hayward <alan.hayward@arm.com>
4837 David Sherwood <david.sherwood@arm.com>
4839 * coretypes.h (opt_scalar_float_mode): New typedef.
4840 * machmode.h (float_mode_for_size): New function.
4841 * emit-rtl.c (double_mode): Delete.
4842 (init_emit_once): Use float_mode_for_size.
4843 * stor-layout.c (layout_type): Likewise.
4844 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
4846 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4847 Alan Hayward <alan.hayward@arm.com>
4848 David Sherwood <david.sherwood@arm.com>
4850 * output.h (assemble_real): Take a scalar_float_mode.
4851 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
4852 * config/arm/arm.md (consttable_4): Likewise.
4853 (consttable_8): Likewise.
4854 (consttable_16): Likewise.
4855 * config/mips/mips.md (consttable_float): Likewise.
4856 * config/s390/s390.c (s390_output_pool_entry): Likewise.
4857 * varasm.c (assemble_real): Take a scalar_float_mode.
4858 (output_constant_pool_2): Update accordingly.
4859 (output_constant): Likewise.
4861 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4862 Alan Hayward <alan.hayward@arm.com>
4863 David Sherwood <david.sherwood@arm.com>
4865 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
4866 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
4867 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
4868 (native_encode_real): Likewise.
4869 (native_interpret_real): Likewise.
4870 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
4871 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
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 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
4878 <scalar_float_mode>. Simplify.
4879 (gen_extend_conv_libfunc): Likewise.
4881 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4882 Alan Hayward <alan.hayward@arm.com>
4883 David Sherwood <david.sherwood@arm.com>
4885 * coretypes.h (scalar_float_mode): New type.
4886 * machmode.h (mode_traits::from_int): Use machine_mode if
4887 USE_ENUM_MODES is defined.
4888 (is_a): New function.
4890 (dyn_cast): Likewise.
4891 (scalar_float_mode): New class.
4892 (scalar_float_mode::includes_p): New function.
4893 (is_float_mode): Likewise.
4894 * gdbhooks.py (MachineModePrinter): New class.
4895 (build_pretty_printer): Use it for scalar_float_mode.
4896 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
4897 (format_helper::format_helper): Turn into a template.
4898 * genmodes.c (get_mode_class): New function.
4899 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
4900 or machine_mode if none.
4901 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
4902 as_a <scalar_float_mode>.
4903 * dwarf2out.c (mem_loc_descriptor): Likewise.
4904 (insert_float): Likewise.
4905 (add_const_value_attribute): Likewise.
4906 * simplify-rtx.c (simplify_immed_subreg): Likewise.
4907 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
4908 (expand_unop): Update accordingly.
4909 (expand_abs_nojump): Likewise.
4910 (expand_copysign_absneg): Take a scalar_float_mode.
4911 (expand_copysign_bit): Likewise.
4912 (expand_copysign): Update accordingly.
4914 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4915 Alan Hayward <alan.hayward@arm.com>
4916 David Sherwood <david.sherwood@arm.com>
4918 * coretypes.h (opt_mode): New class.
4919 * machmode.h (opt_mode): Likewise.
4920 (opt_mode::else_void): New function.
4921 (opt_mode::require): Likewise.
4922 (opt_mode::exists): Likewise.
4923 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
4924 (GET_MODE_2XWIDER_MODE): Likewise.
4925 (mode_iterator::get_wider): Update accordingly.
4926 (mode_iterator::get_2xwider): Likewise.
4927 (mode_iterator::get_known_wider): Likewise, turning into a template.
4928 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
4929 forcing a wider mode to exist.
4930 * config/cr16/cr16.h (LONG_REG_P): Likewise.
4931 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4932 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
4933 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
4934 * lower-subreg.c (init_lower_subreg): Likewise.
4935 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
4936 on the final iteration.
4937 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
4938 a wider mode exists before asking for a move pattern.
4939 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
4940 forcing a wider mode to exist.
4941 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
4942 returning false if no such mode exists.
4943 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
4944 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
4945 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
4946 Avoid checking for a MODE_INT if we already know the mode is not a
4948 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
4949 forcing a wider mode to exist.
4950 (expmed_mult_highpart_optab): Likewise.
4951 (expmed_mult_highpart): Likewise.
4952 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
4954 * lto-streamer-in.c (lto_input_mode_table): Likewise.
4955 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
4956 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
4957 * internal-fn.c (expand_mul_overflow): Update use of
4958 GET_MODE_2XWIDER_MODE.
4959 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4960 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
4961 GET_MODE_WIDER_MODE.
4962 (convert_plusminus_to_widen): Likewise.
4963 * tree-switch-conversion.c (array_value_type): Likewise.
4964 * var-tracking.c (emit_note_insn_var_location): Likewise.
4965 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4966 Return false inside rather than outside the loop if no wider mode
4968 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
4969 and GET_MODE_2XWIDER_MODE
4970 (can_compare_p): Use else_void.
4971 * gdbhooks.py (OptMachineModePrinter): New class.
4972 (build_pretty_printer): Use it for opt_mode.
4974 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4975 Alan Hayward <alan.hayward@arm.com>
4976 David Sherwood <david.sherwood@arm.com>
4978 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
4979 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
4981 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4982 Alan Hayward <alan.hayward@arm.com>
4983 David Sherwood <david.sherwood@arm.com>
4985 * machmode.h (mode_traits): New structure.
4986 (get_narrowest_mode): New function.
4987 (mode_iterator::start): Likewise.
4988 (mode_iterator::iterate_p): Likewise.
4989 (mode_iterator::get_wider): Likewise.
4990 (mode_iterator::get_known_wider): Likewise.
4991 (mode_iterator::get_2xwider): Likewise.
4992 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
4993 (FOR_EACH_MODE): Likewise.
4994 (FOR_EACH_MODE_FROM): Likewise.
4995 (FOR_EACH_MODE_UNTIL): Likewise.
4996 (FOR_EACH_WIDER_MODE): Likewise.
4997 (FOR_EACH_2XWIDER_MODE): Likewise.
4998 * builtins.c (expand_builtin_strlen): Use new mode iterators.
4999 * combine.c (simplify_comparison): Likewise
5000 * config/i386/i386.c (type_natural_mode): Likewise.
5001 * cse.c (cse_insn): Likewise.
5002 * dse.c (find_shift_sequence): Likewise.
5003 * emit-rtl.c (init_derived_machine_modes): Likewise.
5004 (init_emit_once): Likewise.
5005 * explow.c (hard_function_value): Likewise.
5006 * expmed.c (extract_fixed_bit_field_1): Likewise.
5007 (extract_bit_field_1): Likewise.
5008 (expand_divmod): Likewise.
5009 (emit_store_flag_1): Likewise.
5010 * expr.c (init_expr_target): Likewise.
5011 (convert_move): Likewise.
5012 (alignment_for_piecewise_move): Likewise.
5013 (widest_int_mode_for_size): Likewise.
5014 (emit_block_move_via_movmem): Likewise.
5015 (copy_blkmode_to_reg): Likewise.
5016 (set_storage_via_setmem): Likewise.
5017 (compress_float_constant): Likewise.
5018 * omp-low.c (omp_clause_aligned_alignment): Likewise.
5019 * optabs-query.c (get_best_extraction_insn): Likewise.
5020 * optabs.c (expand_binop): Likewise.
5021 (expand_twoval_unop): Likewise.
5022 (expand_twoval_binop): Likewise.
5023 (widen_leading): Likewise.
5024 (widen_bswap): Likewise.
5025 (expand_parity): Likewise.
5026 (expand_unop): Likewise.
5027 (prepare_cmp_insn): Likewise.
5028 (prepare_float_lib_cmp): Likewise.
5029 (expand_float): Likewise.
5030 (expand_fix): Likewise.
5031 (expand_sfix_optab): Likewise.
5032 * postreload.c (move2add_use_add2_insn): Likewise.
5033 * reg-stack.c (reg_to_stack): Likewise.
5034 * reginfo.c (choose_hard_reg_mode): Likewise.
5035 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
5036 * stor-layout.c (mode_for_size): Likewise.
5037 (smallest_mode_for_size): Likewise.
5038 (mode_for_vector): Likewise.
5039 (finish_bitfield_representative): Likewise.
5040 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
5041 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
5042 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5043 * var-tracking.c (prepare_call_arguments): Likewise.
5045 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5046 Alan Hayward <alan.hayward@arm.com>
5047 David Sherwood <david.sherwood@arm.com>
5049 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
5050 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
5051 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
5052 * machmode.h (mode_size): Move earlier in file.
5053 (mode_precision): Likewise.
5054 (mode_inner): Likewise.
5055 (mode_nunits): Likewise.
5056 (mode_unit_size): Likewise.
5057 (unit_unit_precision): Likewise.
5058 (mode_wider): Likewise.
5059 (mode_2xwider): Likewise.
5060 (machine_mode): New class.
5061 (mode_to_bytes): New function.
5062 (mode_to_bits): Likewise.
5063 (mode_to_precision): Likewise.
5064 (mode_to_inner): Likewise.
5065 (mode_to_unit_size): Likewise.
5066 (mode_to_unit_precision): Likewise.
5067 (mode_to_nunits): Likewise.
5068 (GET_MODE_SIZE): Use mode_to_bytes.
5069 (GET_MODE_BITSIZE): Use mode_to_bits.
5070 (GET_MODE_PRECISION): Use mode_to_precision.
5071 (GET_MODE_INNER): Use mode_to_inner.
5072 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
5073 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
5074 (GET_MODE_NUNITS): Use mode_to_nunits.
5075 * system.h (ALWAYS_INLINE): New macro.
5076 * config/powerpcspe/powerpcspe-c.c
5077 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
5078 int for arg1_mode and arg2_mode.
5080 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5081 Alan Hayward <alan.hayward@arm.com>
5082 David Sherwood <david.sherwood@arm.com>
5084 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
5085 Prefix mode names with E_ in case statements.
5086 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5087 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
5088 (aarch64_split_simd_move): Likewise.
5089 (aarch64_gen_storewb_pair): Likewise.
5090 (aarch64_gen_loadwb_pair): Likewise.
5091 (aarch64_gen_store_pair): Likewise.
5092 (aarch64_gen_load_pair): Likewise.
5093 (aarch64_get_condition_code_1): Likewise.
5094 (aarch64_constant_pool_reload_icode): Likewise.
5095 (get_rsqrte_type): Likewise.
5096 (get_rsqrts_type): Likewise.
5097 (get_recpe_type): Likewise.
5098 (get_recps_type): Likewise.
5099 (aarch64_gimplify_va_arg_expr): Likewise.
5100 (aarch64_simd_container_mode): Likewise.
5101 (aarch64_emit_load_exclusive): Likewise.
5102 (aarch64_emit_store_exclusive): Likewise.
5103 (aarch64_expand_compare_and_swap): Likewise.
5104 (aarch64_gen_atomic_cas): Likewise.
5105 (aarch64_emit_bic): Likewise.
5106 (aarch64_emit_atomic_swap): Likewise.
5107 (aarch64_emit_atomic_load_op): Likewise.
5108 (aarch64_evpc_trn): Likewise.
5109 (aarch64_evpc_uzp): Likewise.
5110 (aarch64_evpc_zip): Likewise.
5111 (aarch64_evpc_ext): Likewise.
5112 (aarch64_evpc_rev): Likewise.
5113 (aarch64_evpc_dup): Likewise.
5114 (aarch64_gen_ccmp_first): Likewise.
5115 (aarch64_gen_ccmp_next): Likewise.
5116 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
5117 (alpha_emit_xfloating_libcall): Likewise.
5118 (emit_insxl): Likewise.
5119 (alpha_arg_type): Likewise.
5120 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
5121 (arc_preferred_simd_mode): Likewise.
5122 (arc_secondary_reload): Likewise.
5123 (get_arc_condition_code): Likewise.
5124 (arc_print_operand): Likewise.
5125 (arc_legitimate_constant_p): Likewise.
5126 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5127 * config/arc/arc.md (casesi_load): Likewise.
5128 (casesi_compact_jump): Likewise.
5129 * config/arc/predicates.md (proper_comparison_operator): Likewise.
5130 (cc_use_register): Likewise.
5131 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5132 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
5133 (arm_init_iwmmxt_builtins): Likewise.
5134 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
5135 (neon_expand_vector_init): Likewise.
5136 (arm_attr_length_move_neon): Likewise.
5137 (maybe_get_arm_condition_code): Likewise.
5138 (arm_emit_vector_const): Likewise.
5139 (arm_preferred_simd_mode): Likewise.
5140 (arm_output_iwmmxt_tinsr): Likewise.
5141 (thumb1_output_casesi): Likewise.
5142 (thumb2_output_casesi): Likewise.
5143 (arm_emit_load_exclusive): Likewise.
5144 (arm_emit_store_exclusive): Likewise.
5145 (arm_expand_compare_and_swap): Likewise.
5146 (arm_evpc_neon_vuzp): Likewise.
5147 (arm_evpc_neon_vzip): Likewise.
5148 (arm_evpc_neon_vrev): Likewise.
5149 (arm_evpc_neon_vtrn): Likewise.
5150 (arm_evpc_neon_vext): Likewise.
5151 (arm_validize_comparison): Likewise.
5152 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
5153 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
5154 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
5155 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
5156 (c6x_preferred_simd_mode): Likewise.
5157 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
5158 (epiphany_rtx_costs): Likewise.
5159 * config/epiphany/predicates.md (proper_comparison_operator):
5161 * config/frv/frv.c (condexec_memory_operand): Likewise.
5162 (frv_emit_move): Likewise.
5163 (output_move_single): Likewise.
5164 (output_condmove_single): Likewise.
5165 (frv_hard_regno_mode_ok): Likewise.
5166 (frv_matching_accg_mode): Likewise.
5167 * config/h8300/h8300.c (split_adds_subs): Likewise.
5168 (h8300_rtx_costs): Likewise.
5169 (h8300_print_operand): Likewise.
5170 (compute_mov_length): Likewise.
5171 (output_logical_op): Likewise.
5172 (compute_logical_op_length): Likewise.
5173 (compute_logical_op_cc): Likewise.
5174 (h8300_shift_needs_scratch_p): Likewise.
5175 (output_a_shift): Likewise.
5176 (compute_a_shift_length): Likewise.
5177 (compute_a_shift_cc): Likewise.
5178 (expand_a_rotate): Likewise.
5179 (output_a_rotate): Likewise.
5180 * config/i386/i386.c (classify_argument): Likewise.
5181 (function_arg_advance_32): Likewise.
5182 (function_arg_32): Likewise.
5183 (function_arg_64): Likewise.
5184 (function_value_64): Likewise.
5185 (ix86_gimplify_va_arg): Likewise.
5186 (ix86_legitimate_constant_p): Likewise.
5187 (put_condition_code): Likewise.
5188 (split_double_mode): Likewise.
5189 (ix86_avx256_split_vector_move_misalign): Likewise.
5190 (ix86_expand_vector_logical_operator): Likewise.
5191 (ix86_split_idivmod): Likewise.
5192 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
5193 (ix86_build_const_vector): Likewise.
5194 (ix86_build_signbit_mask): Likewise.
5195 (ix86_match_ccmode): Likewise.
5196 (ix86_cc_modes_compatible): Likewise.
5197 (ix86_expand_branch): Likewise.
5198 (ix86_expand_sse_cmp): Likewise.
5199 (ix86_expand_sse_movcc): Likewise.
5200 (ix86_expand_int_sse_cmp): Likewise.
5201 (ix86_expand_vec_perm_vpermi2): Likewise.
5202 (ix86_expand_vec_perm): Likewise.
5203 (ix86_expand_sse_unpack): Likewise.
5204 (ix86_expand_int_addcc): Likewise.
5205 (ix86_split_to_parts): Likewise.
5206 (ix86_vectorize_builtin_gather): Likewise.
5207 (ix86_vectorize_builtin_scatter): Likewise.
5208 (avx_vpermilp_parallel): Likewise.
5209 (inline_memory_move_cost): Likewise.
5210 (ix86_tieable_integer_mode_p): Likewise.
5211 (x86_maybe_negate_const_int): Likewise.
5212 (ix86_expand_vector_init_duplicate): Likewise.
5213 (ix86_expand_vector_init_one_nonzero): Likewise.
5214 (ix86_expand_vector_init_one_var): Likewise.
5215 (ix86_expand_vector_init_concat): Likewise.
5216 (ix86_expand_vector_init_interleave): Likewise.
5217 (ix86_expand_vector_init_general): Likewise.
5218 (ix86_expand_vector_set): Likewise.
5219 (ix86_expand_vector_extract): Likewise.
5220 (emit_reduc_half): Likewise.
5221 (ix86_emit_i387_round): Likewise.
5222 (ix86_mangle_type): Likewise.
5223 (ix86_expand_round_sse4): Likewise.
5224 (expand_vec_perm_blend): Likewise.
5225 (canonicalize_vector_int_perm): Likewise.
5226 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
5227 (expand_vec_perm_1): Likewise.
5228 (expand_vec_perm_interleave3): Likewise.
5229 (expand_vec_perm_even_odd_pack): Likewise.
5230 (expand_vec_perm_even_odd_1): Likewise.
5231 (expand_vec_perm_broadcast_1): Likewise.
5232 (ix86_vectorize_vec_perm_const_ok): Likewise.
5233 (ix86_expand_vecop_qihi): Likewise.
5234 (ix86_expand_mul_widen_hilo): Likewise.
5235 (ix86_expand_sse2_abs): Likewise.
5236 (ix86_expand_pextr): Likewise.
5237 (ix86_expand_pinsr): Likewise.
5238 (ix86_preferred_simd_mode): Likewise.
5239 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
5240 * config/i386/sse.md (*andnot<mode>3): Likewise.
5241 (<mask_codefor><code><mode>3<mask_name>): Likewise.
5242 (*<code><mode>3): Likewise.
5243 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
5244 (ia64_expand_atomic_op): Likewise.
5245 (ia64_arg_type): Likewise.
5246 (ia64_mode_to_int): Likewise.
5247 (ia64_scalar_mode_supported_p): Likewise.
5248 (ia64_vector_mode_supported_p): Likewise.
5249 (expand_vec_perm_broadcast): Likewise.
5250 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
5251 (iq2000_function_arg_advance): Likewise.
5252 (iq2000_function_arg): Likewise.
5253 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
5254 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
5255 (m68k_libcall_value): Likewise.
5256 (m68k_function_value): Likewise.
5257 (sched_attr_op_type): Likewise.
5258 * config/mcore/mcore.c (mcore_output_move): Likewise.
5259 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
5261 (microblaze_function_arg): Likewise.
5262 * config/mips/mips.c (mips16_build_call_stub): Likewise.
5263 (mips_print_operand): Likewise.
5264 (mips_mode_ok_for_mov_fmt_p): Likewise.
5265 (mips_vector_mode_supported_p): Likewise.
5266 (mips_preferred_simd_mode): Likewise.
5267 (mips_expand_vpc_loongson_even_odd): Likewise.
5268 (mips_expand_vec_unpack): Likewise.
5269 (mips_expand_vi_broadcast): Likewise.
5270 (mips_expand_vector_init): Likewise.
5271 (mips_expand_vec_reduc): Likewise.
5272 (mips_expand_msa_cmp): Likewise.
5273 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
5274 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
5275 (cc_flags_for_mode): Likewise.
5276 * config/msp430/msp430.c (msp430_print_operand): Likewise.
5277 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
5278 (nds32_output_casesi_pc_relative): Likewise.
5279 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5280 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
5281 (nvptx_gen_unpack): Likewise.
5282 (nvptx_gen_pack): Likewise.
5283 (nvptx_gen_shuffle): Likewise.
5284 (nvptx_gen_wcast): Likewise.
5285 (nvptx_preferred_simd_mode): Likewise.
5286 * config/pa/pa.c (pa_secondary_reload): Likewise.
5287 * config/pa/predicates.md (base14_operand): Likewise.
5288 * config/powerpcspe/powerpcspe-c.c
5289 (altivec_resolve_overloaded_builtin): Likewise.
5290 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
5292 (rs6000_preferred_simd_mode): Likewise.
5293 (output_vec_const_move): Likewise.
5294 (rs6000_expand_vector_extract): Likewise.
5295 (rs6000_split_vec_extract_var): Likewise.
5296 (reg_offset_addressing_ok_p): Likewise.
5297 (rs6000_legitimate_offset_address_p): Likewise.
5298 (rs6000_legitimize_address): Likewise.
5299 (rs6000_emit_set_const): Likewise.
5300 (rs6000_const_vec): Likewise.
5301 (rs6000_emit_move): Likewise.
5302 (spe_build_register_parallel): Likewise.
5303 (rs6000_darwin64_record_arg_recurse): Likewise.
5304 (swap_selector_for_mode): Likewise.
5305 (spe_init_builtins): Likewise.
5306 (paired_init_builtins): Likewise.
5307 (altivec_init_builtins): Likewise.
5308 (do_load_for_compare): Likewise.
5309 (rs6000_generate_compare): Likewise.
5310 (rs6000_expand_float128_convert): Likewise.
5311 (emit_load_locked): Likewise.
5312 (emit_store_conditional): Likewise.
5313 (rs6000_output_function_epilogue): Likewise.
5314 (rs6000_handle_altivec_attribute): Likewise.
5315 (rs6000_function_value): Likewise.
5316 (emit_fusion_gpr_load): Likewise.
5317 (emit_fusion_p9_load): Likewise.
5318 (emit_fusion_p9_store): Likewise.
5319 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
5320 (fusion_gpr_mem_load): Likewise.
5321 (fusion_addis_mem_combo_load): Likewise.
5322 (fusion_addis_mem_combo_store): Likewise.
5323 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
5324 (fusion_gpr_mem_load): Likewise.
5325 (fusion_addis_mem_combo_load): Likewise.
5326 (fusion_addis_mem_combo_store): Likewise.
5327 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5329 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
5330 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
5331 (rs6000_preferred_simd_mode): Likewise.
5332 (output_vec_const_move): Likewise.
5333 (rs6000_expand_vector_extract): Likewise.
5334 (rs6000_split_vec_extract_var): Likewise.
5335 (reg_offset_addressing_ok_p): Likewise.
5336 (rs6000_legitimate_offset_address_p): Likewise.
5337 (rs6000_legitimize_address): Likewise.
5338 (rs6000_emit_set_const): Likewise.
5339 (rs6000_const_vec): Likewise.
5340 (rs6000_emit_move): Likewise.
5341 (rs6000_darwin64_record_arg_recurse): Likewise.
5342 (swap_selector_for_mode): Likewise.
5343 (paired_init_builtins): Likewise.
5344 (altivec_init_builtins): Likewise.
5345 (rs6000_expand_float128_convert): Likewise.
5346 (emit_load_locked): Likewise.
5347 (emit_store_conditional): Likewise.
5348 (rs6000_output_function_epilogue): Likewise.
5349 (rs6000_handle_altivec_attribute): Likewise.
5350 (rs6000_function_value): Likewise.
5351 (emit_fusion_gpr_load): Likewise.
5352 (emit_fusion_p9_load): Likewise.
5353 (emit_fusion_p9_store): Likewise.
5354 * config/rx/rx.c (rx_gen_move_template): Likewise.
5355 (flags_from_mode): Likewise.
5356 * config/s390/predicates.md (s390_alc_comparison): Likewise.
5357 (s390_slb_comparison): Likewise.
5358 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
5359 (s390_vector_mode_supported_p): Likewise.
5360 (s390_cc_modes_compatible): Likewise.
5361 (s390_match_ccmode_set): Likewise.
5362 (s390_canonicalize_comparison): Likewise.
5363 (s390_emit_compare_and_swap): Likewise.
5364 (s390_branch_condition_mask): Likewise.
5365 (s390_rtx_costs): Likewise.
5366 (s390_secondary_reload): Likewise.
5367 (__SECONDARY_RELOAD_CASE): Likewise.
5368 (s390_expand_cs): Likewise.
5369 (s390_preferred_simd_mode): Likewise.
5370 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
5371 * config/sh/sh.c (sh_print_operand): Likewise.
5372 (dump_table): Likewise.
5373 (sh_secondary_reload): Likewise.
5374 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5375 * config/sh/sh.md (casesi_worker_1): Likewise.
5376 (casesi_worker_2): Likewise.
5377 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
5378 (fcc_comparison_operator): Likewise.
5379 * config/sparc/sparc.c (sparc_expand_move): Likewise.
5380 (emit_soft_tfmode_cvt): Likewise.
5381 (sparc_preferred_simd_mode): Likewise.
5382 (output_cbranch): Likewise.
5383 (sparc_print_operand): Likewise.
5384 (sparc_expand_vec_perm_bmask): Likewise.
5385 (vector_init_bshuffle): Likewise.
5386 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
5387 (spu_vector_mode_supported_p): Likewise.
5388 (spu_expand_insv): Likewise.
5389 (spu_emit_branch_or_set): Likewise.
5390 (spu_handle_vector_attribute): Likewise.
5391 (spu_builtin_splats): Likewise.
5392 (spu_builtin_extract): Likewise.
5393 (spu_builtin_promote): Likewise.
5394 (spu_expand_sign_extend): Likewise.
5395 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
5396 (tilegx_simd_int): Likewise.
5397 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
5398 (tilepro_simd_int): Likewise.
5399 * config/v850/v850.c (const_double_split): Likewise.
5400 (v850_print_operand): Likewise.
5401 (ep_memory_offset): Likewise.
5402 * config/vax/vax.c (vax_rtx_costs): Likewise.
5403 (vax_output_int_move): Likewise.
5404 (vax_output_int_add): Likewise.
5405 (vax_output_int_subtract): Likewise.
5406 * config/visium/predicates.md (visium_branch_operator): Likewise.
5407 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
5408 (visium_print_operand_address): Likewise.
5409 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5410 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
5411 (xtensa_expand_conditional_branch): Likewise.
5412 (xtensa_copy_incoming_a7): Likewise.
5413 (xtensa_output_literal): Likewise.
5414 * dfp.c (decimal_real_maxval): Likewise.
5415 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
5417 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5418 Alan Hayward <alan.hayward@arm.com>
5419 David Sherwood <david.sherwood@arm.com>
5421 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
5422 (mode_nunits_inline): Likewise.
5423 (mode_inner_inline): Likewise.
5424 (mode_unit_size_inline): Likewise.
5425 (mode_unit_precision_inline): Likewise.
5426 (emit_insn_modes_h): Likewise. Also emit a #define of the
5428 (emit_mode_wider): Add an E_ prefix to mode names.
5429 (emit_mode_complex): Likewise.
5430 (emit_mode_inner): Likewise.
5431 (emit_mode_adjustments): Likewise.
5432 (emit_mode_int_n): Likewise.
5433 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
5434 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
5435 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
5436 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
5437 (CRC32_BUILTIN, ENTRY): Likewise.
5438 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
5439 (aarch64_pop_regs): Likewise.
5440 (aarch64_process_components): Likewise.
5441 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
5442 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
5443 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
5444 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
5445 * config/arm/arm.c (arm_init_libfuncs): Likewise.
5446 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
5448 * config/i386/i386-builtin.def (pcmpestr): Likewise.
5449 (pcmpistr): Likewise.
5450 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
5451 * config/mmix/mmix.c (mmix_output_condition): Likewise.
5452 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
5454 * config/rl78/rl78.c (mduc_regs): Likewise.
5455 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
5456 (htm_expand_builtin): Likewise.
5457 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
5458 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
5459 * config/xtensa/xtensa.c (print_operand): Likewise.
5460 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
5461 (NUM_MODE_VECTOR_INT): Likewise.
5462 * genoutput.c (null_operand): Likewise.
5463 (output_operand_data): Likewise.
5464 * genrecog.c (print_parameter_value): Likewise.
5465 * lra.c (debug_operand_data): Likewise.
5467 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5469 * dwarf2out.c (scompare_loc_descriptor_wide)
5470 (scompare_loc_descriptor_narrow): New functions, split out from...
5471 (scompare_loc_descriptor): ...here.
5472 * expmed.c (emit_store_flag_int): New function, split out from...
5473 (emit_store_flag): ...here.
5475 2017-08-30 Richard Biener <rguenther@suse.de>
5477 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
5478 (dwarf2out_early_finish): Move setting of AT_pubnames from
5479 early debug output to early finish.
5481 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5483 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
5484 and -mdata-region to the assembler.
5486 2017-08-30 Richard Biener <rguenther@suse.de>
5488 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
5490 (gen_subprogram_die): Add DW_AT_object_pointer only early.
5491 (dwarf2out_early_global_decl): Only generate a DIE for the
5492 abstract origin if it doesn't already exist or is a declaration DIE.
5493 (resolve_addr): Do not add the linkage name twice when
5494 generating a stub DIE for the DW_TAG_GNU_call_site target.
5496 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5498 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5499 Use machine_mode rather than int for arg1_mode.
5501 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
5504 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
5505 that the second argument of the built-in functions to unpack
5506 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
5507 switch statement instead a lot of if statements.
5508 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
5509 Allow 64-bit values to be in Altivec registers as well as
5510 traditional floating point registers.
5511 (pack<mode>, FMOVE128_VSX iterator): Likewise.
5513 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
5515 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
5516 MAX_REGS_PER_ADDRESS == 1.
5518 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
5520 * config/i386/i386.opt (flag_fentry): Do not init to -1.
5521 * config/i386/i386.c (ix86_option_override_internal): Simplify
5522 setting of opts->x_flag_entry.
5524 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5525 Jakub Jelinek <jakub@redhat.com>
5526 Richard Biener <rguenther@suse.de>
5528 PR tree-optimization/81503
5529 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
5530 folded constant fits in the target type; reorder tests for clarity.
5532 2017-08-29 Martin Liska <mliska@suse.cz>
5534 * passes.def: Include pass_lower_switch.
5535 * stmt.c (dump_case_nodes): Remove and move to
5536 tree-switch-conversion.
5537 (case_values_threshold): Likewise.
5538 (expand_switch_as_decision_tree_p): Likewise.
5539 (emit_case_decision_tree): Likewise.
5540 (expand_case): Likewise.
5541 (balance_case_nodes): Likewise.
5542 (node_has_low_bound): Likewise.
5543 (node_has_high_bound): Likewise.
5544 (node_is_bounded): Likewise.
5545 (emit_case_nodes): Likewise.
5546 (struct simple_case_node): New struct.
5547 (add_case_node): Remove.
5548 (emit_case_dispatch_table): Use vector instead of case_list.
5549 (reset_out_edges_aux): Remove.
5550 (compute_cases_per_edge): Likewise.
5551 (expand_case): Build list of simple_case_node.
5552 (expand_sjlj_dispatch_table): Use it.
5553 * tree-switch-conversion.c (struct case_node): Moved from
5554 stmt.c and adjusted.
5555 (emit_case_nodes): Likewise.
5556 (node_has_low_bound): Likewise.
5557 (node_has_high_bound): Likewise.
5558 (node_is_bounded): Likewise.
5559 (case_values_threshold): Likewise.
5560 (reset_out_edges_aux): Likewise.
5561 (compute_cases_per_edge): Likewise.
5562 (add_case_node): Likewise.
5563 (dump_case_nodes): Likewise.
5564 (balance_case_nodes): Likewise.
5565 (expand_switch_as_decision_tree_p): Likewise.
5566 (emit_jump): Likewise.
5567 (emit_case_decision_tree): Likewise.
5568 (try_switch_expansion): Likewise.
5569 (do_jump_if_equal): Likewise.
5570 (emit_cmp_and_jump_insns): Likewise.
5571 (fix_phi_operands_for_edge): New function.
5572 (record_phi_operand_mapping): Likewise.
5573 (class pass_lower_switch): New pass.
5574 (pass_lower_switch::execute): New function.
5575 (make_pass_lower_switch): Likewise.
5576 (conditional_probability):
5577 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
5578 * tree-pass.h: Add make_pass_lower_switch.
5580 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5583 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
5586 2017-08-29 Richard Biener <rguenther@suse.de>
5588 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
5589 we do not add a DW_AT_inline attribute twice.
5590 (gen_subprogram_die): Remove code setting DW_AT_inline on
5591 DECL_ABSTRACT_P nodes.
5593 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
5595 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
5596 calls to internal functions.
5597 (gimplify_modify_expr): Likewise.
5598 * tree-call-cdce.c (use_internal_fn): Likewise.
5599 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
5600 (convert_to_divmod): Set the nothrow flag.
5601 * tree-if-conv.c (predicate_mem_writes): Likewise.
5602 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
5603 (vectorizable_call): Likewise.
5604 (vectorizable_store): Likewise.
5605 (vectorizable_load): Likewise.
5606 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
5607 (vect_recog_mask_conversion_pattern): Likewise.
5609 2017-08-29 Martin Liska <mliska@suse.cz>
5612 * gcc.c (driver_handle_option): Add new argument.
5613 * opts-common.c (handle_option): Pass
5614 target_option_override_hook.
5615 * opts-global.c (lang_handle_option): Add new option.
5616 (set_default_handlers): Add new argument.
5617 (decode_options): Likewise.
5618 * opts.c (target_handle_option): Likewise.
5619 (common_handle_option): Call target_option_override_hook.
5620 * opts.h (struct cl_option_handler_func): Add hook for
5621 target option override.
5622 (struct cl_option_handlers): Likewise.
5623 (set_default_handlers): Add new argument.
5624 (decode_options): Likewise.
5625 (common_handle_option): Likewise.
5626 (target_handle_option): Likewise.
5627 * toplev.c (toplev::main): Pass targetm.target_option.override
5630 2017-08-29 Richard Biener <rguenther@suse.de>
5631 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5633 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
5634 life to the active subtree.
5636 2017-08-28 Jeff Law <law@redhat.com>
5638 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
5639 derive_equivalences.
5640 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
5642 (edge_info::derive_equivalences): New private member function
5644 * tree-ssa-dom.c (class edge_info): Changed from a struct
5645 to a class. Add ctor/dtor, methods and data members.
5646 (edge_info::edge_info): Renamed from allocate_edge_info.
5647 Initialize additional members.
5648 (edge_info::~edge_info): New.
5649 (free_dom_edge_info): Delete the edge info.
5650 (record_edge_info): Use new class & associated member functions.
5651 Tighten forms for testing for edge equivalences.
5652 (record_temporary_equivalences): Iterate over the simple
5653 equivalences rather than assuming there's only one per edge.
5654 (cprop_into_successor_phis): Iterate over the simple
5655 equivalences rather than assuming there's only one per edge.
5656 (optimize_stmt): Use operand_equal_p rather than pointer
5657 equality for mini-DSE code.
5659 2017-08-28 Nathan Sidwell <nathan@acm.org>
5661 * gcc.c (execute): Fold SIGPIPE handling into switch
5662 statement. Adjust internal error message.
5664 2017-08-28 Richard Biener <rguenther@suse.de>
5667 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
5668 Do nothing for removed DIEs.
5670 2017-08-28 Richard Biener <rguenther@suse.de>
5672 PR tree-optimization/81977
5673 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
5676 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
5679 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
5680 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
5681 using targetm.gen_mem_thread_fence.
5683 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
5686 * config/i386/i386.md (*<btsc><mode>): Change operand 2
5687 predicate to register_operand. Reorder operands.
5688 (*btr<mode>): Ditto.
5689 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
5690 (*btr<mode>_mask): Ditto.
5692 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
5694 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
5695 * config/rs6000/xmmintrin.h: New file.
5696 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
5698 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5701 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
5702 parameter and_insn and return it.
5703 (recombine_lvx_pattern): Insert a copy to ensure availability of
5704 the base register of the copied masking operation at the point of
5705 the instruction replacement.
5706 (recombine_stvx_pattern): Likewise.
5708 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
5710 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
5711 undocumented switches.
5712 (-mpower9-dform-vector): Likewise.
5713 (-mpower9-dform): Likewise.
5714 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
5715 comments to delete references to -mpower9-dform* switches.
5716 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
5717 Delete reference to -mpower9-dform* switches, test for
5718 -mpower9-vector instead.
5719 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
5720 (OTHER_P9_VECTOR_MASKS): Likewise.
5721 (POWERPC_MASKS): Likewise.
5722 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
5723 tests against -mpower9-dform* to -mpower9-vector. Delete code
5724 that checked for -mpower9-dform* consistancy with other options.
5725 Add test for -mpower9-misc to enable other power9 switches.
5726 (rs6000_init_hard_regno_mode_ok): Likewise.
5727 (rs6000_option_override_internal): Likewise.
5728 (rs6000_emit_prologue): Likewise.
5729 (rs6000_emit_epilogue): Likewise.
5730 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
5731 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
5732 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
5734 (emit_fusion_p9_store): Likewise.
5735 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
5736 resetting these macros if the assembler does not support ISA 3.0
5738 (TARGET_P9_DFORM_VECTOR): Likewise.
5739 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
5740 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
5742 2017-08-25 Alan Modra <amodra@gmail.com>
5744 PR rtl-optimization/81747
5745 * cse.c (cse_extended_basic_block): Don't attempt to record
5746 equivalences for degenerate conditional branches that branch
5747 to their fall-through.
5749 2017-08-24 Martin Sebor <msebor@redhat.com>
5752 * gimple-fold.c (size_must_be_zero_p): New function.
5753 (gimple_fold_builtin_memory_op): Call it.
5755 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
5757 * config/rs6000/mm_malloc.h: New file.
5759 2017-08-24 Bin Cheng <bin.cheng@arm.com>
5761 PR tree-optimization/81913
5762 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
5763 analysis when either IVs in condition can wrap.
5765 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
5767 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
5768 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
5770 2017-08-24 Richard Biener <rguenther@suse.de>
5773 * targhooks.c (default_target_can_inline_p): Properly
5774 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
5775 is present and always compare.
5776 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
5777 infer -mfpmath=sse from TARGET_SSE_P.
5778 (ix86_can_inline_p): Properly use target_option_default_node when
5779 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
5781 2017-08-24 Richard Biener <rguenther@suse.de>
5784 * dwarf2out.c (output_die): Handle flag_generate_offload like
5786 (output_comp_unit): Likewise.
5787 (gen_array_type_die): Likewise.
5788 (dwarf2out_early_finish): Likewise.
5789 (note_variable_value_in_expr): Likewise.
5790 (dwarf2out_finish): Likewise. Adjust assert.
5791 * cgraphunit.c (symbol_table::compile): Move setting of
5792 flag_generate_offload earlier ...
5793 (symbol_table::finalize_compilation_unit): ... here, before
5794 early debug finalization.
5796 2017-08-24 Richard Biener <rguenther@suse.de>
5798 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
5799 and ipa-fnsummary.h.
5800 (ix86_can_inline_p): When ix86_fpmath flags do not match
5801 check whether the callee uses FP math at all.
5803 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
5806 * tree-ssanames.c (get_nonzero_bits): Use element_precision
5807 instead of TYPE_PRECISION.
5809 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
5810 Alan Hayward <alan.hayward@arm.com>
5811 David Sherwood <david.sherwood@arm.com>
5813 * combine.c (make_extraction): Use subreg_offset_from_lsb.
5815 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
5817 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
5819 (ix86_frame::stack_realign_allocate): New field.
5820 (struct machine_frame_state): Modify comments.
5821 (machine_frame_state::sp_realigned_fp_end): New field.
5822 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
5824 (sp_valid_at): Add assertion to assure no attempt to access invalid
5825 offset of a realigned stack.
5826 (fp_valid_at): Likewise.
5827 (choose_baseaddr): Modify comments.
5828 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
5829 ix86_expand_prologue.
5830 (ix86_expand_prologue): Modify stack realignment and allocation.
5831 (ix86_expand_epilogue): Modify comments.
5832 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
5833 avx2_runtime, avx512f, and avx512f_runtime.
5835 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
5837 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
5838 (mstackrealign): Do not init to -1.
5839 * config/i386/i386.c (ix86_option_override_internal):
5840 Check opts_set, not opts when setting default value of
5841 opts->x_ix86_force_align_arg_pointer.
5843 2017-08-23 Richard Biener <rguenther@suse.de>
5845 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
5846 lang_hooks.decl_printable_name.
5847 * print-rtl-function.c (print_rtx_function): Likewise.
5848 * tree-pretty-print.c (dump_function_header): Likewise.
5850 2017-08-23 Richard Biener <rguenther@suse.de>
5853 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
5854 -g0 at compile-time.
5856 2017-08-23 Tamar Christina <tamar.christina@arm.com>
5859 * doc/sourcebuild.texi (Other hardware attributes):
5862 2017-08-23 Tamar Christina <tamar.christina@arm.com>
5865 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
5866 Fix single-use check.
5868 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
5870 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
5872 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
5874 * doc/install.texi: Modify to add more details on running selected
5877 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
5879 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
5880 is combined with -mabi=ms.
5881 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
5884 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5886 PR tree-optimization/81488
5887 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
5888 and cached_basis fields.
5889 (MAX_SPREAD): New constant.
5890 (alloc_cand_and_find_basis): Initialize new fields.
5891 (clear_visited): New function.
5892 (create_phi_basis_1): Rename from create_phi_basis, set visited
5893 and cached_basis fields.
5894 (create_phi_basis): New wrapper function.
5895 (phi_add_costs_1): Rename from phi_add_costs, add spread
5896 parameter, set visited field, short-circuit when limits reached.
5897 (phi_add_costs): New wrapper function.
5898 (record_phi_increments_1): Rename from record_phi_increments, set
5900 (record_phi_increments): New wrapper function.
5901 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
5902 (phi_incr_cost): New wrapper function.
5903 (all_phi_incrs_profitable_1): Rename from
5904 all_phi_incrs_profitable, set visited field.
5905 (all_phi_incrs_profitable): New wrapper function.
5907 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
5908 Alan Hayward <alan.hayward@arm.com>
5909 David Sherwood <david.sherwood@arm.com>
5911 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
5912 that takes the outer and inner modes.
5913 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
5914 comparison as the canonical test for a paradoxical subreg.
5915 * combine.c (simplify_set): Use paradoxical_subreg_p.
5916 (make_extraction): Likewise.
5917 (force_to_mode): Likewise.
5918 (rtx_equal_for_field_assignment_p): Likewise.
5919 (gen_lowpart_for_combine): Likewise.
5920 (simplify_comparison): Likewise.
5921 * cse.c (equiv_constant): Likewise.
5922 * expmed.c (store_bit_field_1): Likewise.
5923 * final.c (alter_subreg): Likewise.
5924 * fwprop.c (propagate_rtx): Likewise.
5925 (forward_propagate_subreg): Likewise.
5926 * ira-conflicts.c (ira_build_conflicts): Likewise.
5927 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
5928 * lra-constraints.c (curr_insn_transform): Likewise.
5929 (split_reg): Likewise.
5930 * lra-eliminations.c (move_plus_up): Likewise.
5931 (lra_eliminate_regs_1): Likewise.
5932 * recog.c (general_operand): Likewise.
5933 * ree.c (combine_reaching_defs): Likewise.
5934 * reload.c (push_reload): Likewise.
5935 (find_reloads): Likewise.
5936 * reload1.c (elimination_effects): Likewise.
5937 (compute_reload_subreg_offset): Likewise.
5938 (choose_reload_regs): Likewise.
5939 * rtlanal.c (subreg_lsb_1): Likewise.
5940 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5941 (simplify_subreg): Likewise.
5942 * var-tracking.c (track_loc_p): Likewise.
5943 * emit-rtl.c (byte_lowpart_offset): Likewise.
5944 (paradoxical_subreg_p): Delete out-of-line definition.
5946 2017-08-22 Jeff Law <law@redhat.com>
5948 PR tree-optimization/81741
5949 PR tree-optimization/71947
5950 * tree-ssa-dom.c: Include tree-inline.h.
5951 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
5952 equivalences if one is more expensive to compute than the other.
5953 * tree-ssa-scopedtables.h (class const_or_copies): Make
5954 record_const_or_copy_raw method private.
5955 (class avail_exprs_stack): New method simplify_binary_operation.
5956 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
5957 avail_exprs_stack::simplify_binary_operation as needed.
5958 (avail_exprs_stack::simplify_binary_operation): New function.
5960 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
5962 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
5963 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
5964 (DOT_SYMBOLS): Likewise.
5965 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
5966 (RELOCATABLE_NEEDS_FIXUP): Likewise.
5967 (RS6000_ABI_NAME): Likewise.
5968 (TARGET_CMODEL): Likewise.
5969 (TOC_SECTION_ASM_OP): Likewise.
5970 (SET_CMODEL): New macro.
5971 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
5973 2017-08-22 Richard Biener <rguenther@suse.de>
5975 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
5976 to allow for free-lang-data replacements similar to verify_type_variant.
5978 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
5979 Alan Hayward <alan.hayward@arm.com>
5980 David Sherwood <david.sherwood@arm.com>
5982 * config/aarch64/aarch64.md (casesi): Use DImode rather than
5983 VOIDmode for the LABEL_REF.
5985 2017-08-22 Richard Biener <rguenther@suse.de>
5987 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
5989 2017-08-22 Richard Biener <rguenther@suse.de>
5991 * common.opt (feliminate-dwarf2-dups): Ignore.
5992 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
5993 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
5994 same_die_p_wrap, compute_section_prefix,
5995 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
5996 (comdat_symbol_id, comdat_symbol_number): Likewise.
5997 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
5999 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
6000 (output_die): Mark unreachable path unreachable.
6001 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
6002 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
6003 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
6004 (dwarf2out_early_finish): Likewise.
6006 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
6008 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
6010 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
6013 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
6014 not VAR_P. Filter attribute warnings with OPT_Wattributes.
6015 (avr_attribute_table) <io, io_low, address>: Initialize
6016 .decl_required with true.
6018 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
6020 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
6021 undocumented debugging options.
6022 (-mvsx-scalar-double): Likewise.
6023 (-mallow-df-permute): Likewise.
6024 (-mvectorize-builtins): Likewise.
6025 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
6026 (rs6000_builtin_vectorized_function): Likewise.
6027 (rs6000_builtin_md_vectorized_function): Likewise.
6028 (rs6000_opt_vars): Likewise.
6030 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
6033 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
6034 (*btrq_imm): Rename from *btrq.
6035 (*btcq_imm): Rename from *btcq.
6036 (btsc): New code attribute.
6037 (*<btsc><mode>): New insn pattern.
6038 (*btr<mode>): Ditto.
6039 (*<btsc><mode>_mask): New insn_and_split pattern.
6040 (*btr<mode>_mask): Ditto.
6042 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6043 Alan Hayward <alan.hayward@arm.com>
6044 David Sherwood <david.sherwood@arm.com>
6046 * function.c (pad_below): Simplify padding calculation.
6048 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6049 Alan Hayward <alan.hayward@arm.com>
6050 David Sherwood <david.sherwood@arm.com>
6052 * target.def (function_prologue): Remove frame size argument.
6053 (function_epilogue): Likewise.
6054 * doc/tm.texi: Regenerate.
6055 * final.c (final_start_function): Update call to function_prologue.
6056 (final_end_function): Update call to function_epilogue.
6057 (default_function_pro_epilogue): Remove frame size argument.
6058 * output.h (default_function_pro_epilogue): Likewise.
6059 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
6060 (arm_output_function_prologue): Likewise.
6061 * config/frv/frv.c (frv_function_prologue): Likewise.
6062 (frv_function_epilogue): Likewise.
6063 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
6064 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
6065 (ia64_output_function_epilogue): Likewise.
6066 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
6067 (m32r_output_function_epilogue): Likewise.
6068 * config/microblaze/microblaze.c (microblaze_function_prologue)
6069 (microblaze_function_epilogue): Likewise.
6070 * config/mips/mips.c (mips_output_function_prologue): Likewise.
6071 (mips_output_function_epilogue): Likewise.
6072 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
6073 (mmix_target_asm_function_epilogue): Likewise.
6074 * config/msp430/msp430.c (msp430_start_function): Likewise.
6075 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
6076 (nds32_asm_function_epilogue): Likewise.
6077 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
6078 * config/pa/pa.c (pa_output_function_prologue): Likewise.
6079 (pa_output_function_epilogue): Likewise.
6080 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
6081 (rs6000_output_function_epilogue): Likewise.
6082 * config/rl78/rl78.c (rl78_start_function): Likewise.
6083 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
6084 (rs6000_output_function_epilogue): Likewise.
6085 * config/rx/rx.c (rx_output_function_prologue): Likewise.
6086 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
6087 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
6088 (sparc_asm_function_epilogue): Likewise.
6090 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6092 * tree.h (type_has_mode_precision_p): New function.
6093 * convert.c (convert_to_integer_1): Use it.
6094 * expr.c (expand_expr_real_2): Likewise.
6095 (expand_expr_real_1): Likewise.
6096 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
6097 * match.pd: Likewise.
6098 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
6099 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
6100 * tree-tailcall.c (process_assignment): Likewise.
6101 * tree-vect-loop.c (vectorizable_reduction): Likewise.
6102 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
6103 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
6104 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6105 (vectorizable_assignment): Likewise.
6106 (vectorizable_shift): Likewise.
6107 (vectorizable_operation): Likewise.
6108 * tree-vrp.c (register_edge_assert_for_2): Likewise.
6110 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
6112 * match.pd: Add pow (C, x) simplification.
6114 2017-08-21 Richard Biener <rguenther@suse.de>
6116 PR tree-optimization/81900
6117 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
6118 for blocks with abnormal predecessors.
6119 (compute_antic): Do not set visited flag prematurely.
6121 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
6124 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
6126 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6128 * stor-layout.h (vector_type_mode): Move to...
6129 * tree.h (vector_type_mode): ...here.
6130 * stor-layout.c (vector_type_mode): Move to...
6131 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
6133 2017-08-21 Richard Biener <rguenther@suse.de>
6135 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
6136 register_external_die hooks.
6137 (debug_false_tree_charstarstar_uhwistar): Declare.
6138 (debug_nothing_tree_charstar_uhwi): Likewise.
6139 * debug.c (do_nothing_debug_hooks): Adjust.
6140 (debug_false_tree_charstarstar_uhwistar): New do nothing.
6141 (debug_nothing_tree_charstar_uhwi): Likewise.
6142 * dbxout.c (dbx_debug_hooks): Adjust.
6143 (xcoff_debug_hooks): Likewise.
6144 * sdbout.c (sdb_debug_hooks): Likewise.
6145 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
6146 * dwarf2out.c (macinfo_label_base): New global.
6147 (dwarf2out_register_external_die): New function for the
6148 register_external_die hook.
6149 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
6150 (dwarf2_debug_hooks): Use them.
6151 (dwarf2_lineno_debug_hooks): Adjust.
6152 (struct die_struct): Add with_offset flag.
6153 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
6154 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
6155 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
6156 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
6157 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
6158 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
6159 defining section names for the early LTO debug variants.
6160 (reset_indirect_string): New helper.
6161 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
6162 (print_dw_val): Add support for offsetted symbol references.
6163 (get_ultimate_context): Split out from is_cxx.
6164 (is_cxx): Use get_ultimate_context.
6165 (is_fortran): Add decl overload.
6166 (compute_comp_unit_symbol): Split out worker from
6167 compute_section_prefix.
6168 (compute_section_prefix): Call compute_comp_unit_symbol and
6169 set comdat_type_p here.
6170 (output_die): Skip DIE symbol output for the LTO added one.
6171 Handle DIE symbol references with offset.
6172 (output_comp_unit): Guard section name mangling properly.
6173 For LTO debug sections emit a symbol at the section beginning
6174 which we use to refer to its DIEs.
6175 (add_abstract_origin_attribute): For DIEs registered via
6176 dwarf2out_register_external_die directly refer to the early
6177 DIE rather than indirectly through the shadow one we created.
6178 Remove obsolete call to dwarf2out_abstract_function for
6179 non-function/block origins.
6180 (gen_array_type_die): When generating early LTO debug do
6181 not emit DW_AT_string_length.
6182 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
6183 late when in LTO. As suggested place a gcc_unreachable for
6184 the DECL_ABSTRACT_P case.
6185 (gen_subprogram_die): Avoid another specification DIE
6186 for early built declarations/definitions for the late LTO case.
6187 (gen_variable_die): Add type references for late duplicated VLA dies
6189 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
6190 we have the abstract instance already.
6191 (process_scope_var): Adjust decl DIE contexts in LTO which
6192 first puts them in limbo.
6193 (gen_decl_die): Do not generate type DIEs late apart from
6194 types for VLAs or for decls we do not yet have a DIE. Do not
6195 call dwarf2out_abstract_function late.
6196 (dwarf2out_early_global_decl): Make sure to create DIEs
6197 for abstract instances of a decl first.
6198 (dwarf2out_late_global_decl): Adjust comment.
6199 (output_macinfo_op): With multiple macro sections use
6200 macinfo_label_base to distinguish labels.
6201 (output_macinfo): Likewise. Update macinfo_label_base.
6202 Pass in the line info label.
6203 (note_variable_value_in_expr): When generating LTO resolve
6204 all variable values here by generating DIEs as needed.
6205 (init_sections_and_labels): Add early LTO debug flag parameter
6206 and generate different sections and names if set. Add generation
6207 counter for the labels so we can have multiple of them.
6208 (reset_dies): Helper to allow DIEs to be output multiple times.
6209 (dwarf2out_finish): When outputting DIEs to the fat part of an
6210 LTO object first reset DIEs.
6211 (dwarf2out_early_finish): Output early DIEs when generating LTO.
6212 (modified_type_die): Check for decl_ultimate_origin being self
6214 (gen_type_die_with_usage): Likewise.
6215 (gen_typedef_die): Allow decl_ultimate_origin being self.
6216 (set_decl_abstract_flags): Remove.
6217 (set_block_abstract_flags): Likewise.
6218 (dwarf2out_abstract_function): Treat the early generated DIEs
6219 as the abstract copy and only add DW_AT_inline and
6220 DW_AT_artificial here and call set_decl_origin_self.
6221 If the DIE has an abstract origin don't do anything.
6222 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
6223 if we have none yet (Go fails to build one, PR78628).
6224 (variably_modified_type_p): Prevent endless recursion for Ada
6225 cyclic pointer types.
6226 * lto-streamer-in.c: Include debug.h.
6227 (dref_queue): New global.
6228 (lto_read_tree_1): Stream in DIE references.
6229 (lto_input_tree): Register DIE references.
6230 (input_function): Stream DECL_DEBUG_ARGS.
6231 * lto-streamer-out.c: Include debug.h.
6232 (lto_write_tree_1): Output DIE references.
6233 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
6234 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
6235 (output_function): Stream DECL_DEBUG_ARGS.
6236 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
6237 Stream DECL_ABSTRACT_ORIGIN.
6238 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
6239 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
6240 DECL_CONTEXT for file-scope decls.
6241 * lto-streamer.h (struct dref_entry): Declare.
6242 (dref_queue): Likewise.
6243 * cfgexpand.c (pass_expand::execute): Do not call the
6244 outlining_inline_function hook here.
6245 * lto-wrapper.c (debug_obj): New global.
6246 (tool_cleanup): Unlink it if required.
6247 (debug_objcopy): New function.
6248 (run_gcc): Handle early debug sections in the IL files by
6249 extracting them to separate files, partially linkin them and
6250 feeding the result back as result to the linker.
6251 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
6252 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
6253 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
6254 sections into a separate segment.
6255 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
6257 (darwin_asm_dwarf_section): Likewise.
6258 (darwin_asm_output_dwarf_offset): Likewise.
6259 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
6261 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6262 Alan Hayward <alan.hayward@arm.com>
6263 David Sherwood <david.sherwood@arm.com>
6265 * read-md.h (md_reader::record_potential_iterator_use): Replace
6266 pointer argument with an rtx and an index.
6267 * read-rtl.c (iterator_group::apply_iterator): Likewise.
6268 (apply_mode_iterator): Likewise.
6269 (apply_code_iterator): Likewise.
6270 (apply_int_iterator): Likewise.
6271 (apply_subst_iterator): Likewise.
6272 (record_iterator_use): Likewise.
6273 (record_attribute_use): Likewise.
6274 (md_reader::record_potential_iterator_use): Likewise. Update calls
6275 to record_iterator_use and apply_iterator.
6276 (iterator_use): Replace ptr with x and index.
6277 (attribute_use): Likewise.
6278 (apply_attribute_uses): Update calls to apply_iterator.
6279 (apply_iterators): Likewise. Update initialization of iterator_use.
6280 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
6281 and record_potential_iterator_use.
6282 (rtx_reader::read_rtx_operand): Likewise.
6284 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6285 Alan Hayward <alan.hayward@arm.com>
6286 David Sherwood <david.sherwood@arm.com>
6288 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
6291 2017-08-21 Richard Biener <rguenther@suse.de>
6294 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
6295 at struct end conservatively when comparing common bases.
6297 2017-08-21 Richard Biener <rguenther@suse.de>
6299 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
6300 (mem_ref_in_stmt): Remove.
6301 (determine_max_movement): Use ref index to get at the reference.
6302 (invariantness_dom_walker::before_dom_children): Deal with
6303 lim data already initialized.
6304 (gather_mem_refs_stmt): Initialize lim data and record ref index.
6306 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
6308 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
6309 (TARGET_ISA_ROUND): Ditto.
6310 (TARGET_ROUND): Ditto.
6311 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
6312 * config/i386/i386.md: Ditto.
6313 * config/i386/sse.md: Ditto.
6314 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
6315 with OPTION_MASK_ISA_SSE4_1.
6317 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
6320 * doc/extend.texi (x86 Built-in Functions): Correct the name of
6321 __builtin_ia32_lzcnt_u16.
6323 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
6326 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
6327 (rs6000_set_current_function): Rewrite function to use it.
6329 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
6332 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
6333 and TYPE_WARN_IF_NOT_ALIGN.
6334 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
6335 (handle_warn_if_not_align): New.
6336 (place_union_field): Call handle_warn_if_not_align.
6337 (place_field): Call handle_warn_if_not_align.
6338 Copy TYPE_WARN_IF_NOT_ALIGN.
6339 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
6340 (layout_type): Likewise.
6341 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
6343 (tree_decl_common): Add warn_if_not_align.
6344 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
6345 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
6346 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
6347 (DECL_WARN_IF_NOT_ALIGN): Likewise.
6348 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
6349 * doc/extend.texi: Document warn_if_not_aligned attribute.
6350 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
6352 2017-08-17 Martin Liska <mliska@suse.cz>
6355 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
6356 (get_data_dependence): Use it as pointer type.
6357 (distribute_loop): Likewise.
6359 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6361 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
6362 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
6363 (p8_vmrgow_<mode>_direct): New define_insn.
6364 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
6365 handle endianness for vmrgew and vmrgow permute patterns.
6367 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
6369 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
6370 * config/rs6000/rs6000-cpus.def: Remove comment.
6371 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
6372 (POWERPC_MASKS): Likewise.
6373 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
6374 use of TARGET_VSX_TIMODE.
6375 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
6376 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
6377 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
6378 (rs6000_option_override_internal): Remove dead code.
6379 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
6380 (rs6000_legitimize_reload_address): Likewise.
6381 (rs6000_legitimate_address_p): Likewise.
6382 (rs6000_opt_masks): Delete "vsx-timode".
6383 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
6384 from function comment.
6385 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
6386 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
6387 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
6389 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
6390 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
6391 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
6394 (define_peephole2): Likewise.
6396 2017-08-17 Martin Sebor <msebor@redhat.com>
6399 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
6400 past the end of an array.
6401 (test_pp_format): Add test cases.
6403 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
6405 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
6406 missing ECF_NOTHROW flags.
6408 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
6411 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
6412 operands residing in integer registers.
6413 (*vsx_le_perm_load_<mode>): Likewise.
6414 (*vsx_le_perm_store_<mode>): Likewise.
6415 (define_peephole2): Add peepholes to optimize the above.
6417 2017-08-17 Marek Polacek <polacek@redhat.com>
6420 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
6421 to mimic what shorten_compare did. Change the return type to bool.
6422 (fold_cond_expr_with_comparison): Update call to
6423 operand_equal_for_comparison_p.
6424 (fold_ternary_loc): Likewise.
6426 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
6428 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
6430 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
6432 2017-08-17 Richard Biener <rguenther@suse.de>
6434 * tree-ssa-structalias.c (solve_graph): When propagating
6435 to successors update the graphs succ edges and avoid duplicate work.
6437 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
6440 * config/i386/i386.c (ix86_option_override_internal): Save target
6441 specific options after ix86_stack_protector_guard_reg was changed.
6443 2017-08-17 Richard Biener <rguenther@suse.de>
6445 PR tree-optimization/81827
6446 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
6447 (new_var_info): Initialize it conservatively.
6448 (get_call_vi): Mark register vars.
6449 (new_scalar_tmp_constraint_exp): Likewise.
6450 (handle_rhs_call): Likewise.
6451 (handle_const_call): Likewise.
6452 (create_function_info_for): Likewise.
6453 (solve_constraints): Sort varinfos to separate register from
6454 non-register vars to pack points-to solution bitmaps during
6457 2017-08-17 Marek Polacek <polacek@redhat.com>
6459 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
6461 2017-08-17 Richard Biener <rguenther@suse.de>
6463 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
6464 to true when overflow is undefined and we saturated the result.
6466 2017-08-17 Alan Modra <amodra@gmail.com>
6469 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
6470 Don't use store multiple if only one reg needs saving.
6471 (interesting_frame_related_regno): New function.
6472 (rs6000_frame_related): Don't emit frame info for regs that
6474 (rs6000_emit_epilogue): Likewise.
6476 2017-08-16 Nathan Sidwell <nathan@acm.org>
6478 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
6479 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
6480 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
6481 * tree.c (free_lang_data_in_type): Use else-if chain. Always
6482 clear TYPE_LANG_1. Remove obsolete member-function stripping.
6483 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
6484 (verify_type): Adjust for TYPE_BINFO move.
6485 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
6486 process TYPE_BINFO directly.
6487 (hash_tree): Likewise.
6488 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
6490 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
6493 2017-08-16 David Malcolm <dmalcolm@redhat.com>
6495 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
6497 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
6500 * config/i386/i386.md (*anddi_1_btr): Change predicates of
6501 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
6502 Add ix86_binary_operator_ok to insn constraint.
6503 (*iordi_1_bts): Ditto.
6504 (*xordi_1_btc): Ditto.
6505 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
6506 Update corresponding peephole2 pattern.
6510 2017-08-16 Bin Cheng <bin.cheng@arm.com>
6512 PR tree-optimization/81832
6513 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
6514 copy loop header which has IFN_LOOP_DIST_ALIAS call.
6516 2017-08-16 Marek Polacek <polacek@redhat.com>
6519 * fold-const.c (fold_indirect_ref_1): Restore original behavior
6520 regarding size_zero_node.
6522 2017-08-16 Martin Liska <mliska@suse.cz>
6525 * config.gcc: Respect previously set extra_objs in case
6528 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
6530 PR tree-optimization/81835
6531 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
6532 the phi SSA_NAME. Check that the condition in a COND_EXPR does
6533 not depend on the phi.
6535 2017-08-16 Alan Modra <amodra@gmail.com>
6537 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
6540 2017-08-16 Alan Modra <amodra@gmail.com>
6542 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
6543 (save_reg_p): ..into this. Update all callers.
6544 (first_reg_to_save): Simplify.
6546 2017-08-16 Alan Modra <amodra@gmail.com>
6548 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
6551 2017-08-15 Joseph Myers <joseph@codesourcery.com>
6555 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
6556 constant count if that count is less than 32.
6558 2017-08-15 Nathan Sidwell <nathan@acm.org>
6560 * gcc.c (execute): Emit friendlier message if inferior is killed
6561 by an external cause.
6563 2017-08-15 Richard Biener <rguenther@suse.de>
6565 PR tree-optimization/81790
6566 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
6567 CONSTRUCTORs from simplifying and VN.
6569 2017-08-14 Martin Sebor <msebor@redhat.com>
6571 * builtin-attrs.def: Add comments.
6573 2017-08-14 Martin Sebor <msebor@redhat.com>
6576 * doc/extend.texi (attribute nonstring): Document new attribute.
6578 2017-08-14 Martin Sebor <msebor@redhat.com>
6581 * tree-diagnostic.c (default_tree_printer): Handle %G.
6582 * gimple-pretty-print.h (percent_G_format): Declare new function.
6583 * gimple-pretty-print.c (percent_G_format): Define.
6584 * tree-pretty-print.c (percent_K_format): Add argument.
6586 2017-08-14 Martin Sebor <msebor@redhat.com>
6588 PR translation/79998
6589 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
6590 Remove a stray space.
6592 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
6595 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
6596 (*iordi_1_bts): Ditto.
6597 (*xordi_1_btc): Ditto.
6599 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6602 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
6603 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6605 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
6606 quoted strings, and make more translator-friendly.
6607 (darwin_rs6000_override_options): Likewise.
6608 (rs6000_option_override_internal): Likewise.
6609 (rs6000_return_in_memory): Fix overlong line.
6610 (init_cmulative_args): Use quoted strings, and make more
6611 translator-friendly.
6612 (rs6000_pass_by_reference): Fix overlong line.
6613 (def_builtin): Use quoted strings.
6614 (altivec_expand_predicate_builtin): Use quoted strings, and make
6615 more translator-friendly.
6616 (htm_expand_builtin): Use quoted strings.
6617 (cpu_expand_builtin): Use quoted strings, and make more
6618 translator-friendly.
6619 (altivec_expand_builtin): Likewise.
6620 (paired_expand_predicate_builtin): Likewise.
6621 (rs6000_invalid_builtin): Likewise.
6622 (builtin_function_type): Use quoted strings.
6623 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
6624 more translator-friendly.
6625 (rs6000_trampoline_init): Likewise.
6626 (rs6000_handle_altivec_attribute): Likewise.
6627 (rs6000_inner_target_options): Use quoted strings.
6628 (rs6000_disable_incompatible_switches): Likewise.
6629 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
6630 strings, and make more translator-friendly.
6631 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
6633 2017-08-14 Bin Cheng <bin.cheng@arm.com>
6635 PR tree-optimization/81799
6636 * tree-loop-distribution.c (version_loop_by_alias_check): Force
6637 cond_expr to simple gimple operand.
6639 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
6642 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
6644 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
6648 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
6649 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
6650 TARGET_GASISR_PROLOGUES.
6651 * config/avr/avr.c (avr_option_override): Same.
6652 (avr_pass_pre_proep::execute): Same.
6654 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
6657 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
6658 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
6660 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
6662 * config/i386/i386.md (*load_tp_<mode>): Redefine as
6663 define_insn_and_split. Split to a memory load from 0 in
6664 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
6665 using PTR mode iterator.
6666 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
6667 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
6668 (*add_tp_<mode>): Redefine as define_insn_and_split.
6669 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
6670 address space. Merge with *add_tp_x32 using PTR mode iterator.
6671 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
6672 Split to an add with a memory load from 0 in
6673 DEFAULT_TLS_SEG_REG address space.
6675 2017-08-12 Andrew Pinski <apinski@cavium.com>
6677 * config/aarch64/aarch64-option-extensions.def (rdma):
6678 Fix feature string to what Linux prints out in /proc/cpuinfo.
6680 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
6683 * dwarf2out.c (modified_type_die): For C typedef types that have
6684 an ultimate origin, process the ultimate origin instead of the
6686 (gen_typedef_die): Assert that input DECLs have no ultimate
6688 (gen_type_die_with_usage): For typedef variants that have an
6689 ultimate origin, just call gen_decl_die on the original DECL.
6690 (process_scope_var): Avoid creating DIEs for local typedefs and
6691 concrete static variables.
6693 2017-08-12 Alan Modra <amodra@gmail.com>
6697 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
6698 match gnu-user.h startfile.
6699 (ENDFILE_LINUX_SPEC): Similarly.
6701 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
6704 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
6706 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
6708 2017-08-11 Tamar Christina <tamar.christina@arm.com>
6709 * config/aarch64/aarch64.md (mov<mode>): Change.
6710 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
6711 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
6712 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
6714 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
6716 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
6717 for storage order barriers.
6719 2017-08-11 Martin Liska <mliska@suse.cz>
6722 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
6723 variables of void type.
6725 2017-08-11 Martin Liska <mliska@suse.cz>
6727 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
6728 TARGET_SUPPORTS_ALIASES.
6729 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
6730 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
6731 (optimize_weakref): Likewise.
6732 * symtab.c (symtab_node::noninterposable_alias): Likewise.
6733 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
6734 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
6736 2017-08-11 Martin Liska <mliska@suse.cz>
6739 * config/i386/i386.c (make_resolver_func): Do complete
6740 refactoring of the function.
6742 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
6745 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
6746 * config/i386/i386.c (ix86_stack_protect_guard): Use
6747 ix86_stack_protect_guard_symbol_str to generate varible declaration.
6748 * doc/invoke.texi (x86 Options): Document
6749 -mstack-protector-guard-symbol= option.
6751 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
6753 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
6754 * config/i386/i386.c (ix86_split_stack_guard): New function.
6755 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
6756 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
6757 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
6758 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
6759 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
6760 (split_stack_space_check): Call ix86_split_stack_guard.
6762 2017-08-10 Martin Sebor <msebor@redhat.com>
6764 * print-tree.c (print_node): Print location using the established
6766 Replace spaces with colons.
6767 (debug_raw, debug): Ditto.
6769 2017-08-10 Martin Sebor <msebor@redhat.com>
6772 * pretty-print.c (pp_format): Correct the handling of %s precision.
6774 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
6777 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
6779 (ix86_finalize_stack_frame_flags): This. Also clear
6780 frame_pointer_needed if -fno-omit-frame-pointer is used without
6782 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
6783 with ix86_finalize_stack_frame_flags.
6784 (ix86_expand_epilogue): Likewise.
6785 (ix86_expand_split_stack_prologue): Likewise.
6786 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
6788 2017-08-10 Martin Liska <mliska@suse.cz>
6791 * c-attribs.c (handle_target_attribute):
6792 Report warning for an empty string argument of target attribute.
6794 2017-08-09 Jakub Jelinek <jakub@redhat.com>
6797 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
6799 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
6800 or DECL_NONLOCAL labels.
6801 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
6802 or DECL_NONLOCAL labels here.
6804 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
6806 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
6807 to indicate when early gimple folding has been disabled.
6808 (rs6000_gimple_fold_builtin): Add debug content.
6809 (rs6000_invalid_builtin): Fix whitespace.
6810 (rs6000_expand_builtin): Fix whitespace.
6811 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
6813 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
6816 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
6817 SAVE_MULTIPLE if not all the registers that saves, should be saved.
6819 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
6821 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
6822 (qdf24xx): Likewise.
6823 * config/aarch64/aarch64.md: Include falkor.md.
6824 * config/aarch64/falkor.md: New.
6826 2017-08-09 Marek Polacek <polacek@redhat.com>
6829 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
6830 * diagnostic.c (emit_diagnostic): Add a comment.
6831 (emit_diagnostic_valist): New function.
6833 2017-08-09 Marek Polacek <polacek@redhat.com>
6836 * input.c (make_location): New overload.
6837 * input.h (make_location): Declare.
6839 2017-08-08 Alan Modra <amodra@gmail.com>
6840 H.J. Lu <hongjiu.lu@intel.com>
6843 * gcc.c (NO_PIE_SPEC): Delete.
6844 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
6846 (LINK_PIE_SPEC): ..to here.
6847 (LINK_COMMAND_SPEC): Support -no-pie.
6848 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
6849 chain of crtbegin*.o selection, update for PIE_SPEC changes and
6851 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
6852 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
6853 (ENDFILE_CRTEND_SPEC): Similarly.
6855 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
6858 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
6859 (mstack-protector-guard-offset=): Ditto.
6860 * config/i386/i386.c (ix86_option_override): Handle
6861 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
6863 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
6864 ix86_stack_protect_guard_offset variables.
6865 (TARGET_STACK_PROTECT_GUARD): Always define.
6866 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
6867 and -mstack-protector-guard-offset= options.
6869 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6871 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
6872 boundary case for the last candidate.
6874 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6876 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
6877 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
6879 2017-08-08 Tamar Christina <tamar.christina@arm.com>
6882 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
6883 * config/aarch64/aarch64-builtins.c
6884 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
6885 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
6886 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
6888 2017-08-08 Tamar Christina <tamar.christina@arm.com>
6889 Andrew Pinski <pinskia@gmail.com>
6892 * internal-fn.def (XORSIGN): New.
6893 * optabs.def (xorsign_optab): New.
6894 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
6895 (convert_expand_mult_copysign): New.
6896 (pass_optimize_widening_mul::execute): Call
6897 convert_expand_mult_copysign.
6899 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6901 PR tree-optimization/81354
6902 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
6903 Insert on edges rather than explicitly creating landing pads.
6904 (analyze_candidates_and_replace): Commit edge inserts.
6906 2017-08-08 Richard Biener <rguenther@suse.de>
6909 * tree-ssa-loop-niter.c: Include tree-dfa.h.
6910 (expand_simple_operations): Also look through ADDR_EXPRs with
6911 MEM_REF bases treating them as POINTER_PLUS_EXPR.
6913 2017-08-08 Richard Biener <rguenther@suse.de>
6915 PR tree-optimization/81723
6916 * tree-vect-slp.c (struct bst_traits): New hash traits.
6917 (bst_fail): New global.
6918 (vect_build_slp_tree_2): New worker, split out from ...
6919 (vect_build_slp_tree): ... this now wrapping it with using
6920 bst_fail set to cache SLP tree build fails. Properly handle
6922 (vect_analyze_slp_instance): Allocate and free bst_fail.
6924 2017-08-08 Martin Liska <mliska@suse.cz>
6927 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
6928 LABEL_DECLs that can be from a different function.
6930 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6932 PR tree-optimization/81744
6933 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
6934 loop's number of iterations.
6936 2017-08-08 Martin Liska <mliska@suse.cz>
6938 * asan.c: Include header files.
6939 * attribs.c (build_decl_attribute_variant): New function moved
6941 (build_type_attribute_qual_variant): Likewise.
6942 (cmp_attrib_identifiers): Likewise.
6943 (simple_cst_list_equal): Likewise.
6944 (omp_declare_simd_clauses_equal): Likewise.
6945 (attribute_value_equal): Likewise.
6946 (comp_type_attributes): Likewise.
6947 (build_type_attribute_variant): Likewise.
6948 (lookup_ident_attribute): Likewise.
6949 (remove_attribute): Likewise.
6950 (merge_attributes): Likewise.
6951 (merge_type_attributes): Likewise.
6952 (merge_decl_attributes): Likewise.
6953 (merge_dllimport_decl_attributes): Likewise.
6954 (handle_dll_attribute): Likewise.
6955 (attribute_list_equal): Likewise.
6956 (attribute_list_contained): Likewise.
6957 * attribs.h (lookup_attribute): New function moved from tree.[ch].
6958 (lookup_attribute_by_prefix): Likewise.
6959 * bb-reorder.c: Include header files.
6960 * builtins.c: Likewise.
6961 * calls.c: Likewise.
6962 * cfgexpand.c: Likewise.
6963 * cgraph.c: Likewise.
6964 * cgraphunit.c: Likewise.
6965 * convert.c: Likewise.
6966 * dwarf2out.c: Likewise.
6967 * final.c: Likewise.
6968 * fold-const.c: Likewise.
6969 * function.c: Likewise.
6970 * gimple-expr.c: Likewise.
6971 * gimple-fold.c: Likewise.
6972 * gimple-pretty-print.c: Likewise.
6973 * gimple.c: Likewise.
6974 * gimplify.c: Likewise.
6975 * hsa-common.c: Likewise.
6976 * hsa-gen.c: Likewise.
6977 * internal-fn.c: Likewise.
6978 * ipa-chkp.c: Likewise.
6979 * ipa-cp.c: Likewise.
6980 * ipa-devirt.c: Likewise.
6981 * ipa-fnsummary.c: Likewise.
6982 * ipa-inline.c: Likewise.
6983 * ipa-visibility.c: Likewise.
6985 * lto-cgraph.c: Likewise.
6986 * omp-expand.c: Likewise.
6987 * omp-general.c: Likewise.
6988 * omp-low.c: Likewise.
6989 * omp-offload.c: Likewise.
6990 * omp-simd-clone.c: Likewise.
6991 * opts-global.c: Likewise.
6992 * passes.c: Likewise.
6993 * predict.c: Likewise.
6994 * sancov.c: Likewise.
6995 * sanopt.c: Likewise.
6996 * symtab.c: Likewise.
6997 * toplev.c: Likewise.
6998 * trans-mem.c: Likewise.
6999 * tree-chkp.c: Likewise.
7000 * tree-eh.c: Likewise.
7001 * tree-into-ssa.c: Likewise.
7002 * tree-object-size.c: Likewise.
7003 * tree-parloops.c: Likewise.
7004 * tree-profile.c: Likewise.
7005 * tree-ssa-ccp.c: Likewise.
7006 * tree-ssa-live.c: Likewise.
7007 * tree-ssa-loop.c: Likewise.
7008 * tree-ssa-sccvn.c: Likewise.
7009 * tree-ssa-structalias.c: Likewise.
7010 * tree-ssa.c: Likewise.
7011 * tree-streamer-in.c: Likewise.
7012 * tree-vectorizer.c: Likewise.
7013 * tree-vrp.c: Likewise.
7015 * ubsan.c: Likewise.
7016 * varasm.c: Likewise.
7017 * varpool.c: Likewise.
7018 * tree.c: Remove functions moved to attribs.[ch].
7020 * config/aarch64/aarch64.c: Add attrs.h header file.
7021 * config/alpha/alpha.c: Likewise.
7022 * config/arc/arc.c: Likewise.
7023 * config/arm/arm.c: Likewise.
7024 * config/avr/avr.c: Likewise.
7025 * config/bfin/bfin.c: Likewise.
7026 * config/c6x/c6x.c: Likewise.
7027 * config/cr16/cr16.c: Likewise.
7028 * config/cris/cris.c: Likewise.
7029 * config/darwin.c: Likewise.
7030 * config/epiphany/epiphany.c: Likewise.
7031 * config/fr30/fr30.c: Likewise.
7032 * config/frv/frv.c: Likewise.
7033 * config/ft32/ft32.c: Likewise.
7034 * config/h8300/h8300.c: Likewise.
7035 * config/i386/winnt.c: Likewise.
7036 * config/ia64/ia64.c: Likewise.
7037 * config/iq2000/iq2000.c: Likewise.
7038 * config/lm32/lm32.c: Likewise.
7039 * config/m32c/m32c.c: Likewise.
7040 * config/m32r/m32r.c: Likewise.
7041 * config/m68k/m68k.c: Likewise.
7042 * config/mcore/mcore.c: Likewise.
7043 * config/microblaze/microblaze.c: Likewise.
7044 * config/mips/mips.c: Likewise.
7045 * config/mmix/mmix.c: Likewise.
7046 * config/mn10300/mn10300.c: Likewise.
7047 * config/moxie/moxie.c: Likewise.
7048 * config/msp430/msp430.c: Likewise.
7049 * config/nds32/nds32-isr.c: Likewise.
7050 * config/nds32/nds32.c: Likewise.
7051 * config/nios2/nios2.c: Likewise.
7052 * config/nvptx/nvptx.c: Likewise.
7053 * config/pa/pa.c: Likewise.
7054 * config/pdp11/pdp11.c: Likewise.
7055 * config/powerpcspe/powerpcspe.c: Likewise.
7056 * config/riscv/riscv.c: Likewise.
7057 * config/rl78/rl78.c: Likewise.
7058 * config/rx/rx.c: Likewise.
7059 * config/s390/s390.c: Likewise.
7060 * config/sh/sh.c: Likewise.
7061 * config/sol2.c: Likewise.
7062 * config/sparc/sparc.c: Likewise.
7063 * config/spu/spu.c: Likewise.
7064 * config/stormy16/stormy16.c: Likewise.
7065 * config/tilegx/tilegx.c: Likewise.
7066 * config/tilepro/tilepro.c: Likewise.
7067 * config/v850/v850.c: Likewise.
7068 * config/vax/vax.c: Likewise.
7069 * config/visium/visium.c: Likewise.
7070 * config/xtensa/xtensa.c: Likewise.
7072 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
7075 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
7076 constraints since the -mupper-regs-* switches have been
7078 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
7079 into a vector from a double word element that was extracted from
7080 another vector, and eliminate extra XXPERMDI instructions.
7081 (vsx_concat_<mode>_2): Likewise.
7082 (vsx_concat_<mode>_3): Likewise.
7083 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
7084 concat to allow optimizing inserts from previous extracts.
7086 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
7088 * config/i386/i386.c (ix86_stack_protect_guard): Generate
7089 memory reference to a SSP offset in TLS address space.
7090 (ix86_print_operand) <case '@'>: Remove.
7091 (ix86_print_operand_punct_valid_p): Remove '@' code.
7092 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
7094 (stack_tls_protect_set_<mode>): Remove.
7095 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
7096 (stack_tls_protect_test_<mode>): Remove.
7097 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
7099 2017-08-07 Olivier Hainque <hainque@adacore.com>
7102 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
7104 2017-08-07 Douglas Rupp <rupp@adacore.com>
7106 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
7107 variable was referenced as multidir in command.
7109 2017-08-07 Jakub Jelinek <jakub@redhat.com>
7112 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
7115 2017-08-07 Martin Liska <mliska@suse.cz>
7117 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
7118 * config/rl78/rl78.c: Add include of attribs.h.
7119 * config/sh/sh.c: Likewise.
7120 * config/v850/v850.c: Likewise.
7122 2017-08-07 Tom de Vries <tom@codesourcery.com>
7125 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
7127 2017-08-07 Martin Liska <mliska@suse.cz>
7129 * config/mips/mips.c: Include attribs.h.
7131 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
7134 * doc/invoke.texi: Document change in behvaior for -Ofast for
7137 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
7139 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
7141 (aarch64_pop_regs): Likewise.
7142 (aarch64_gen_load_pair): Likewise.
7143 (aarch64_save_callee_saves): Likewise.
7144 (aarch64_restore_callee_saves): Likewise.
7146 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
7148 * config/i386/i386.c: Revert the last change.
7150 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
7153 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
7155 (ix86_finalize_stack_frame_flags): This. Also clear
7156 frame_pointer_needed if -fno-omit-frame-pointer is used without
7158 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
7159 with ix86_finalize_stack_frame_flags.
7160 (ix86_expand_epilogue): Likewise.
7161 (ix86_expand_split_stack_prologue): Likewise.
7163 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
7166 * config/i386/i386.c (get_builtin_code_for_version): Set priority
7167 to P_AES for Westmere.
7169 2017-08-07 Jonathan Yong <10walls@gmail.com>
7171 * config/i386/mingw.opt (fset-stack-executable): Removed.
7172 * config/i386/cygming.opt (fset-stack-executable): Moved
7174 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
7176 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
7178 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
7180 2017-08-07 Marek Polacek <polacek@redhat.com>
7183 * fold-const.c (fold_indirect_ref_1): Check type_domain.
7185 2017-08-07 Martin Liska <mliska@suse.cz>
7187 * attribs.h (canonicalize_attr_name): New function.
7188 (cmp_attribs): Move from c-format.c and adjusted.
7189 (is_attribute_p): Moved from tree.h.
7190 * tree-inline.c: Add new includes.
7191 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
7192 (private_is_attribute_p): Remove.
7193 (private_lookup_attribute): Likewise.
7194 (private_lookup_attribute_by_prefix): Simplify.
7195 (remove_attribute): Use is_attribute_p.
7196 * tree.h: Remove removed declarations.
7198 2017-08-07 Jakub Jelinek <jakub@redhat.com>
7201 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
7202 instead of computing it in the function. Formatting fix.
7203 (expand_case): Don't rely on default_edge being the first edge,
7204 clear it if removing it, pass default_edge to
7205 emit_case_dispatch_table.
7206 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
7209 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
7211 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
7212 insn in the function, emit NOP after the insn.
7214 2017-08-06 Tom de Vries <tom@codesourcery.com>
7216 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
7219 2017-08-06 Tom de Vries <tom@codesourcery.com>
7221 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
7224 2017-08-04 Yury Gribov <tetra2005@gmail.com>
7226 PR tree-optimization/57371
7227 * match.pd: New pattern.
7229 2017-08-04 Marek Polacek <polacek@redhat.com>
7232 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
7233 perform the computation in offset_int.
7235 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7237 PR tree-optimization/81136
7238 * tree-vectorizer.h: Include tree-hash-traits.h.
7239 (vec_base_alignments): New typedef.
7240 (vec_info): Add a base_alignments field.
7241 (vect_record_base_alignments): Declare.
7242 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
7244 (DR_IS_CONDITIONAL_IN_STMT): New macro.
7245 (create_data_ref): Add an is_conditional_in_stmt argument.
7246 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
7247 the is_conditional_in_stmt field.
7248 (data_ref_loc): Add an is_conditional_in_stmt field.
7249 (get_references_in_stmt): Set the is_conditional_in_stmt field.
7250 (find_data_references_in_stmt): Update call to create_data_ref.
7251 (graphite_find_data_references_in_stmt): Likewise.
7252 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
7253 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
7254 (vect_record_base_alignment): New function.
7255 (vect_record_base_alignments): Likewise.
7256 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
7257 for nested statements even if we fail to compute a misalignment.
7258 Use pooled base alignments for unconditional references.
7259 (vect_find_same_alignment_drs): Compare base addresses instead
7261 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
7262 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
7264 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7266 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
7267 Add an explicit name for the enum. Use auto_vec for slp_instances
7269 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
7271 (_bb_vec_info): Add a constructor and destructor.
7272 (vinfo_for_stmt): Return NULL for uids of -1 as well.
7273 (destroy_loop_vec_info): Delete.
7274 (vect_destroy_datarefs): Likewise.
7275 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
7276 (vec_info::vec_info): New function.
7277 (vec_info::~vec_info): Likewise.
7278 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
7279 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
7280 destroy_loop_vec_info.
7281 * tree-vect-loop.c (new_loop_vec_info): Replace with...
7282 (_loop_vec_info::_loop_vec_info): ...this.
7283 (destroy_loop_vec_info): Replace with...
7284 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
7285 the stmt_vec_infos. Leave handling of vec_info information to its
7286 destructor. Remove explicit vector releases.
7287 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
7288 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
7289 * tree-vect-slp.c (new_bb_vec_info): Replace with...
7290 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
7291 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
7292 (destroy_bb_vec_info): Replace with...
7293 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
7294 information to its destructor.
7295 (vect_slp_analyze_bb_1): Use new and delete instead of
7296 new_bb_vec_info and destroy_bb_vec_info.
7297 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
7300 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7302 * tree-data-ref.h (subscript): Add access_fn field.
7303 (data_dependence_relation): Add could_be_independent_p.
7304 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
7305 (same_access_functions): Move to tree-data-ref.c.
7306 * tree-data-ref.c (ref_contains_union_access_p): New function.
7307 (access_fn_component_p): Likewise.
7308 (access_fn_components_comparable_p): Likewise.
7309 (dr_analyze_indices): Add a reference to access_fn_component_p.
7310 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
7312 (constant_access_functions): Likewise.
7313 (add_other_self_distances): Likewise.
7314 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
7315 (initialize_data_dependence_relation): Use XCNEW and remove
7316 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
7317 of access functions that have the same type. Allow the
7318 subsequence to end with different bases in some circumstances.
7319 Record the chosen access functions in SUB_ACCESS_FN.
7320 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
7321 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
7322 (subscript_dependence_tester_1): Likewise dra and drb.
7323 (build_classic_dist_vector): Update calls accordingly.
7324 (subscript_dependence_tester): Likewise.
7325 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
7326 DDR_COULD_BE_INDEPENDENT_P.
7327 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
7328 comp_alias_ddrs instead of may_alias_ddrs.
7329 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
7331 (vect_analyze_data_ref_dependence): Use it if
7332 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
7333 distance vectors if that fails.
7334 (dependence_distance_ge_vf): New function.
7335 (vect_prune_runtime_alias_test_list): Use it. Don't clear
7336 LOOP_VINFO_MAY_ALIAS_DDRS.
7338 2017-08-04 Richard Biener <rguenther@suse.de>
7341 * fold-const.c (fold_binary_loc): Properly restrict
7342 minus_var0 && minus_var1 case when associating undefined overflow
7345 2017-08-04 Tom de Vries <tom@codesourcery.com>
7347 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
7349 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7351 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7352 Don't start diagnostic messages with a capital letter.
7353 * config/rs6000/rs6000.c (rs6000_option_override_internal):
7355 (rs6000_invalid_builtin): Likewise.
7356 (rs6000_trampoline_init): Likewise.
7358 2017-08-03 Jakub Jelinek <jakub@redhat.com>
7361 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
7362 after setting changeable df flags.
7364 2017-08-03 Richard Biener <rguenther@suse.de>
7366 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
7367 up if the use is in USE - X.
7369 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
7371 * toplev.c (dumpfile.h): New include.
7372 (internal_error_reentered): New static function. Use it...
7373 (internal_error_function): ...here to handle reentered internal_error.
7375 2017-08-03 Richard Biener <rguenther@suse.de>
7378 * fold-const.c (split_tree): Add minus_var and minus_con
7379 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
7380 here but always use minus_*.
7381 (associate_trees): Assert we never associate with MINUS_EXPR
7382 and NULL first operand. Do not recurse for PLUS_EXPR operands
7383 when associating as MINUS_EXPR either.
7384 (fold_binary_loc): Track minus_var and minus_con.
7386 2017-08-03 Tom de Vries <tom@codesourcery.com>
7389 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
7390 ACCEL_COMPILER, apply finish_options on
7391 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
7393 2017-08-03 Tom de Vries <tom@codesourcery.com>
7396 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
7397 function_entry_patch_area_size > 0.
7399 2017-08-03 Jakub Jelinek <jakub@redhat.com>
7402 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
7403 instead of 10??LU, perform unit multiplication in wide_int,
7404 don't change alloc_object_size_limit if the limit is larger
7407 PR tree-optimization/81655
7408 PR tree-optimization/81588
7409 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
7410 the case when ranges[i].low and high are 1 for unsigned type with
7414 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
7415 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
7417 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7419 * tree-vrp.h: Add include guard.
7421 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
7424 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
7425 (ud2): New insn pattern.
7426 * config/i386/i386.c (ix86_expand_epilogue):
7427 For naked functions, generate ud2 instead of trap insn.
7429 2017-08-02 Marek Polacek <polacek@redhat.com>
7432 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
7434 2017-08-02 Tom de Vries <tom@codesourcery.com>
7435 Cesar Philippidis <cesar@codesourcery.com>
7437 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
7438 Add missing edge probabilities.
7440 2017-08-02 Tamar Christina <tamar.christina@arm.com>
7442 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
7445 2017-08-02 Jakub Jelinek <jakub@redhat.com>
7448 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
7449 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
7450 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
7451 prologue_seq sequences - if any.
7453 2017-08-02 Richard Biener <rguenther@suse.de>
7455 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
7456 via vectors if supported, integer extracts via punning if supported
7457 or otherwise vector extracts.
7459 2017-08-02 Richard Biener <rguenther@suse.de>
7461 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
7463 (bitmap_insert_into_set): ... this.
7465 2017-08-02 Richard Biener <rguenther@suse.de>
7467 PR tree-optimization/81633
7469 2015-08-17 Alan Hayward <alan.hayward@arm.com>
7471 PR tree-optimization/71752
7472 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
7474 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
7476 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
7477 (machine_function::call_ms2sysv_pad_out): Remove field.
7478 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
7479 (ix86_compute_frame_layout): Likewise.
7481 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
7484 * config/i386/i386.c (ix86_set_func_type): Disallow naked
7485 attribute with interrupt attribute.
7487 2017-08-01 Andrew Pinski <apinski@cavium.com>
7489 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
7490 BIT_INSERT_EXPR's operand 1
7491 to see if the types precision matches.
7493 2017-08-01 Martin Liska <mliska@suse.cz>
7496 * builtins.c (expand_builtin_memcpy_args): Remove.
7497 (expand_builtin_memcpy): Call newly added function
7498 expand_builtin_memory_copy_args.
7499 (expand_builtin_memcpy_with_bounds): Likewise.
7500 (expand_builtin_mempcpy): Remove last argument.
7501 (expand_builtin_mempcpy_with_bounds): Likewise.
7502 (expand_builtin_memory_copy_args): New function created from
7503 expand_builtin_mempcpy_args with small modifications.
7504 (expand_builtin_mempcpy_args): Remove.
7505 (expand_builtin_stpcpy): Remove unused argument.
7506 (expand_builtin): Likewise.
7507 (expand_builtin_with_bounds): Likewise.
7509 2017-08-01 Martin Liska <mliska@suse.cz>
7512 Make mempcpy more optimal (PR middle-end/70140).
7514 2017-08-01 Jakub Jelinek <jakub@redhat.com>
7517 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
7518 __builtin_vec_cmpne verify both arguments are compatible vectors
7519 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
7520 verify arg1_type is a pointer or array type. For __builtin_vec_st,
7521 move computation of aligned to after checking the argument types.
7525 * config/rs6000/vsx.md (vextract_fp_from_shorth,
7526 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
7529 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
7530 Doug Rupp <rupp@adacore.com>
7531 Olivier Hainque <hainque@adacore.com>
7533 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
7534 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
7536 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
7537 for TARGET_OS_CPP_BUILTIN.
7538 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
7539 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
7541 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
7542 passing required abi options to the assembler for EABI configurations.
7543 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
7544 of .text.hot and .text.unlikely sections for kernel modules when
7545 using ARM style exceptions.
7546 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
7547 options. Add EXTRA_CC1_SPEC.
7548 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
7550 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
7552 (ARM_TARGET2_DWARF_FORMAT): Define.
7553 * config/arm/t-vxworks: Adjust multilib control to removal of the
7554 Diab command line options.
7556 2017-08-01 Martin Liska <mliska@suse.cz>
7558 PR gcov-profile/81561
7559 * gcov.c (unblock): Make unblocking safe as we need to preserve
7560 index correspondence of blocks and block_lists.
7562 2017-08-01 Richard Biener <rguenther@suse.de>
7564 PR tree-optimization/81181
7565 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7566 (compute_antic): ... end of iteration here.
7568 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
7570 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
7571 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
7572 (ftree-slp-vectorize): Likewise.
7573 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
7574 can no longer be set independent of flag_tree_loop_vectorize.
7575 * omp-general.c (emp_max_vf): Likewise.
7576 * opts.c (enable_fdo_optimizations): Remove references to
7577 flag_tree_vectorize, these are now implicit.
7578 (common_handle_option): Remove handling for OPT_ftree_vectorize,
7579 and leave it for the options machinery.
7581 2017-08-01 Martin Liska <mliska@suse.cz>
7584 * builtins.c (expand_builtin_memcpy_args): Remove.
7585 (expand_builtin_memcpy): Call newly added function
7586 expand_builtin_memory_copy_args.
7587 (expand_builtin_memcpy_with_bounds): Likewise.
7588 (expand_builtin_mempcpy): Remove last argument.
7589 (expand_builtin_mempcpy_with_bounds): Likewise.
7590 (expand_builtin_memory_copy_args): New function created from
7591 expand_builtin_mempcpy_args with small modifications.
7592 (expand_builtin_mempcpy_args): Remove.
7593 (expand_builtin_stpcpy): Remove unused argument.
7594 (expand_builtin): Likewise.
7595 (expand_builtin_with_bounds): Likewise.
7597 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
7600 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
7601 print "ds:" only for immediates in generic address space.
7603 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
7606 * config/i386/i386.c (ix86_funciton_naked): New prototype.
7607 (ix86_function_ok_for_sibcall): Return false for naked functions.
7609 2017-08-01 Richard Biener <rguenther@suse.de>
7611 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
7612 (compute_antic): Seed worklist with exit block predecessors.
7613 * cfganal.c (dfs_find_deadend): For a cycle return the source
7614 of the edge closing it.
7616 2017-08-01 Tamar Christina <tamar.christina@arm.com>
7618 * config/aarch64/aarch64.c
7619 (aarch64_can_const_movi_rtx_p): Move 0 check.
7621 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7623 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
7624 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
7626 * match.pd: Ditto in address comparison pattern.
7628 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7630 PR tree-optimization/81627
7631 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
7632 closed ssa form for store-store chain.
7634 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7636 PR tree-optimization/81620
7637 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
7638 for store-store chain.
7640 2017-08-01 Jakub Jelinek <jakub@redhat.com>
7642 PR tree-optimization/81588
7643 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
7644 ranges[i].in_p, invert comparison code ccode. For >/>=,
7645 swap rhs1 and rhs2 and comparison code unconditionally,
7646 for </<= don't do that. Don't swap rhs1/rhs2 again if
7647 ranges[i].in_p, instead invert comparison code ccode if
7648 opcode or oe->rank is BIT_IOR_EXPR.
7651 * optabs.def (vec_extract_optab, vec_init_optab): Change from
7652 a direct optab to conversion optab.
7653 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
7654 with GET_MODE_INNER as last argument instead of optab_handler.
7655 * expmed.c (extract_bit_field_1): Likewise. Use vector from
7656 vector extraction if possible and optab is available.
7657 * expr.c (store_constructor): Use convert_optab_handler instead
7658 of optab_handler. Use vector initialization from smaller
7659 vectors if possible and optab is available.
7660 * tree-vect-stmts.c (vectorizable_load): Likewise.
7661 * doc/md.texi (vec_extract, vec_init): Document that the optabs
7663 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
7664 of vec_init from half-sized vectors with the same element mode.
7665 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
7666 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
7667 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
7668 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
7669 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
7670 after mode in gen_vec_extract* calls.
7671 (vec_extract<mode>): Renamed to ...
7672 (vec_extract<mode><ssescalarmodelower>): ... this.
7673 (vec_extract<mode><ssehalfvecmodelower>): New expander.
7674 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
7675 element mode after mode in gen_vec_init* calls.
7676 (VEC_INIT_HALF_MODE): New mode iterator.
7677 (vec_init<mode>): Renamed to ...
7678 (vec_init<mode><ssescalarmodelower>): ... this.
7679 (vec_init<mode><ssehalfvecmodelower>): New expander.
7680 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
7681 (vec_extractv2sfsf): ... this.
7682 (vec_initv2sf): Renamed to ...
7683 (vec_initv2sfsf): ... this.
7684 (vec_extractv2si): Renamed to ...
7685 (vec_extractv2sisi): ... this.
7686 (vec_initv2si): Renamed to ...
7687 (vec_initv2sisi): ... this.
7688 (vec_extractv4hi): Renamed to ...
7689 (vec_extractv4hihi): ... this.
7690 (vec_initv4hi): Renamed to ...
7691 (vec_initv4hihi): ... this.
7692 (vec_extractv8qi): Renamed to ...
7693 (vec_extractv8qiqi): ... this.
7694 (vec_initv8qi): Renamed to ...
7695 (vec_initv8qiqi): ... this.
7696 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
7697 (vec_init<mode>): Renamed to ...
7698 (vec_init<mode><VEC_base_l>): ... this.
7699 (vec_extract<mode>): Renamed to ...
7700 (vec_extract<mode><VEC_base_l>): ... this.
7701 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
7702 (vec_initv2sfsf): ... this.
7703 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
7704 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
7705 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
7706 element mode after mode in gen_vec_init* calls.
7707 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
7708 (vec_init<mode><Vel>): ... this.
7709 (vec_extract<mode>): Renamed to ...
7710 (vec_extract<mode><Vel>): ... this.
7711 * config/aarch64/iterators.md (Vel): New mode attribute.
7712 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
7713 Add element mode after mode in gen_vec_extract* calls.
7714 * config/s390/vector.md (non_vec_l): New mode attribute.
7715 (vec_extract<mode>): Renamed to ...
7716 (vec_extract<mode><non_vec_l>): ... this.
7717 (vec_init<mode>): Renamed to ...
7718 (vec_init<mode><non_vec_l>): ... this.
7719 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
7720 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
7722 * config/arm/iterators.md (V_elem_l): New mode attribute.
7723 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
7724 (vec_extract<mode><V_elem_l>): ... this.
7725 (vec_extractv2di): Renamed to ...
7726 (vec_extractv2didi): ... this.
7727 (vec_init<mode>): Renamed to ...
7728 (vec_init<mode><V_elem_l>): ... this.
7729 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
7730 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
7731 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
7732 Add element mode after gen_vec_extract* calls.
7733 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
7734 (vec_init<mode><unitmode>): ... this.
7735 (vec_extract<mode>): Renamed to ...
7736 (vec_extract<mode><unitmode>): ... this.
7737 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
7738 (vec_init<mode><unitmode>): ... this.
7739 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
7740 (vec_initv2sfsf): ... this.
7741 (vec_extractv2sf): Renamed to ...
7742 (vec_extractv2sfsf): ... this.
7743 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
7744 Add element mode after gen_vec_extract* calls.
7745 * config/mips/mips.md (unitmode): New mode iterator.
7746 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
7747 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
7748 * config/spu/spu.md (inner_l): New mode attribute.
7749 (vec_init<mode>): Renamed to ...
7750 (vec_init<mode><inner_l>): ... this.
7751 (vec_extract<mode>): Renamed to ...
7752 (vec_extract<mode><inner_l>): ... this.
7753 * config/sparc/sparc.md (veltmode): New mode iterator.
7754 (vec_init<VMALL:mode>): Renamed to ...
7755 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
7756 * config/ia64/vect.md (vec_initv2si): Renamed to ...
7757 (vec_initv2sisi): ... this.
7758 (vec_initv2sf): Renamed to ...
7759 (vec_initv2sfsf): ... this.
7760 (vec_extractv2sf): Renamed to ...
7761 (vec_extractv2sfsf): ... this.
7762 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
7763 (vec_init<mode>): Renamed to ...
7764 (vec_init<mode><VEC_base_l>): ... this.
7765 (vec_extract<mode>): Renamed to ...
7766 (vec_extract<mode><VEC_base_l>): ... this.
7767 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
7768 (vec_initv2sfsf): ... this.
7769 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
7770 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
7771 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
7772 gen_vec_init* calls.
7774 2017-08-01 Richard Biener <rguenther@suse.de>
7776 PR tree-optimization/81297
7777 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
7778 TREE_OVERFLOW from INTEGER_CSTs.
7780 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
7782 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
7784 2017-07-31 Carl Love <cel@us.ibm.com>
7786 * config/rs6000/rs6000-c: Add support for built-in functions
7787 vector signed char vec_xl_be (signed long long, signed char *);
7788 vector unsigned char vec_xl_be (signed long long, unsigned char *);
7789 vector signed int vec_xl_be (signed long long, signed int *);
7790 vector unsigned int vec_xl_be (signed long long, unsigned int *);
7791 vector signed long long vec_xl_be (signed long long, signed long long *);
7792 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
7793 vector signed short vec_xl_be (signed long long, signed short *);
7794 vector unsigned short vec_xl_be (signed long long, unsigned short *);
7795 vector double vec_xl_be (signed long long, double *);
7796 vector float vec_xl_be (signed long long, float *);
7797 * config/rs6000/altivec.h (vec_xl_be): Add #define.
7798 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
7799 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
7801 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
7802 (altivec_expand_builtin): Add switch statement to call
7803 altivec_expand_xl_be for each builtin.
7804 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
7805 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
7806 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
7807 __builtin_vsx_le_be_v16qi.
7808 * doc/extend.texi: Update the built-in documentation file for the
7809 new built-in functions.
7811 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
7814 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
7816 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
7818 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7820 * config.gcc: Add z14.
7821 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
7822 CPU model numbers for z13s and z14.
7823 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
7825 * config/s390/s390-opts.h (enum processor_type): Rename
7826 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
7827 * config/s390/s390.c (processor_table): Add field for CPU name to
7828 be passed to Binutils.
7829 (s390_asm_output_machine_for_arch): Use the new field in
7830 processor_table for Binutils.
7831 (s390_expand_builtin): Replace arch12 with z14.
7832 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
7833 (s390_get_sched_attrmask): Likewise.
7834 (s390_get_unit_mask): Likewise.
7835 * config/s390/s390.opt: Add z14 to processor_type enum.
7837 2017-07-31 Martin Jambor <mjambor@suse.cz>
7840 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
7841 regardless of optimization level.
7843 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
7844 Martin Liska <mliska@suse.cz>
7846 * predict.def: Remove old comment and adjust probability.
7847 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
7850 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
7853 * config/i386/i386.c (ix86_function_naked): New function.
7854 (ix86_can_use_return_insn_p): Return false for naked functions.
7855 (ix86_expand_prologue): Skip prologue for naked functions.
7856 (ix86_expand_epilogue): Skip epilogue for naked functions
7857 and emit trap instruction.
7858 (ix86_warn_func_return): New function.
7859 (ix86_attribute_table): Add "naked" attribute specification.
7860 (TARGET_WARN_FUNC_RETURN): Define.
7861 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
7863 2017-07-31 Martin Liska <mliska@suse.cz>
7865 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
7866 (dump_gimple_bb_header): Always dump BB info.
7867 (pp_cfg_jump): Do not append info about BB when dumping a jump.
7869 2017-07-31 Martin Liska <mliska@suse.cz>
7872 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
7873 also with current_function_decl non-null equality.
7875 2017-07-31 Jakub Jelinek <jakub@redhat.com>
7878 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
7879 change type to the element type, instead add eltype variable and
7880 use it where we are interested in the element type.
7882 PR tree-optimization/81603
7883 * ipa-polymorphic-call.c
7884 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
7885 offset arithmetic in offset_int, bail out if the resulting bit offset
7886 doesn't fit into shwi.
7888 2017-07-31 Martin Liska <mliska@suse.cz>
7890 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
7891 (gimplify_save_expr): Fix comment.
7893 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
7896 * config/i386/i386.c (ix86_function_arg): Update arguments for
7898 (ix86_compute_frame_layout): Set the initial stack offset to
7899 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
7900 INCOMING_FRAME_SP_OFFSET.
7901 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
7902 stack before exception handler returns.
7903 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
7904 the 'ERROR_CODE' for exception handler.
7906 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
7908 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
7909 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
7910 (ASM_OUTPUT_REG_POP): Ditto.
7911 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
7912 instead of asm_fprintf to output pure string.
7914 2017-07-29 Jakub Jelinek <jakub@redhat.com>
7916 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
7917 to imported_module_or_decl hook.
7918 (debug_nothing_tree_tree_tree_bool): Remove.
7919 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
7920 * debug.c (do_nothing_debug_hooks): Use
7921 debug_nothing_tree_tree_tree_bool_bool instead of
7922 debug_nothing_tree_tree_tree_bool.
7923 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
7924 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
7925 * sdbout.c (sdb_debug_hooks): Likewise.
7926 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
7927 (gen_namespace_die): Add DW_AT_export_symbols attribute if
7929 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
7930 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
7931 attribute, don't add anything.
7933 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7935 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
7936 (fold_build2_stat_loc): Likewise.
7937 (fold_build3_stat_loc): Likewise.
7938 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
7939 (fold_build1_loc): Remove macro.
7940 (fold_build2_loc): Likewise.
7941 (fold_build3_loc): Likewise.
7943 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7945 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
7946 (gimple_build_debug_bind_source_stat): Likewise.
7947 * gimple.h (gimple_build_debug_bind): Remove macro.
7948 (gimple_build_debug_bind_source): Likewise.
7950 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7952 * bitmap.c (bitmap_alloc): Adjust.
7953 (bitmap_gc_alloc): Likewise.
7954 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
7956 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7958 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
7959 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
7960 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
7961 (bitmap_gc_alloc_stat): Likewise.
7962 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
7964 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7966 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
7967 * rtl.h (shallow_copy_rtx): Remove macro.
7969 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7971 * emit-rtl.c (gen_raw_REG): Adjust.
7972 * gengenrtl.c (gendef): Likewise.
7973 * rtl.c (rtx_alloc_stat): Remove _stat from name.
7974 * rtl.h (rtx_alloc): Remove macro.
7976 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7978 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
7979 (build_tree_list_stat): Likewise.
7980 * tree.h (build_tree_list): Remove macro.
7981 (build_tree_list_vec): Likewise.
7983 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7985 * tree.c (make_vector_stat): Remove _stat from name.
7986 (build_vector_stat): Likewise.
7987 * tree.h (make_vector_stat): Remove macro.
7988 (build_vector_stat): Likewise.
7990 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7992 * tree.h (build_var_debug_value): Remove prototype.
7994 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7996 * tree.c (tree_cons_stat): Remove _stat from name.
7997 * tree.h (tree_cons): Remove macro.
7999 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8001 * tree.c (build_vl_exp_stat): Remove _stat from name.
8002 * tree.h (build_vl_exp): Remove macro.
8004 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8006 * tree.c (build_decl_stat): Remove _stat from name.
8007 * tree.h (build_decl): Remove macro.
8009 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8011 * gimple.c (gimple_build_with_ops_stat): Adjust.
8012 (gimple_alloc_stat): Remove _stat from name.
8013 * gimple.h (gimple_alloc): Remove macro.
8015 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8017 * tree.c (make_tree_vec_stat): Remove _stat from name.
8018 (grow_tree_vec_stat): Likewise.
8019 * tree.h (make_tree_vec_stat): Adjust prototype.
8020 (grow_tree_vec_stat): Likewise.
8021 (make_tree_vec): Remove macro.
8022 (grow_tree_vec): Likewise.
8024 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8026 * fold-const.c (fold_build1_stat_loc): Adjust.
8027 (fold_build2_stat_loc): Likewise.
8028 (fold_build3_stat_loc): Likewise.
8029 * tree.c (build0_stat): Remove _stat from name.
8030 (build1_stat): Likewise.
8031 (build2_stat): Likewise.
8032 (build3_stat): Likewise.
8033 (build4_stat): Likewise.
8034 (build5_stat): Likewise.
8035 * tree.h (build1_loc): Remove macro, and rename _stat function
8037 (build2_loc): Likewise.
8038 (build3_loc): Likewise.
8039 (build4_loc): Likewise.
8040 (build5_loc): Likewise.
8042 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8044 * tree.c (make_int_cst_stat): Remove _stat from name.
8045 * tree.h (make_int_cst_stat): Adjust prototype.
8046 (make_int_cst): Remove macro.
8048 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8050 * tree.c (make_tre_binfo_stat): Remove _stat from name.
8051 * tree.h (make_tree_binfo_stat): Adjust prototype.
8052 (make_tree_binfo): Remove.
8054 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8056 * tree.c (copy_node_stat): Rename to copy_node.
8057 (build_distinct_type_copy): Adjust.
8058 * tree.h (copy_node_stat): Adjust prototype.
8059 (copy_node): Remove macro.
8061 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8063 * tree.c (make_node_stat): rename to make_node.
8064 (build_tree_list_stat): Adjust.
8065 (build0_stat): Likewise.
8066 (build2_stat): Likewise.
8067 (build3_stat): Likewise.
8068 (build4_stat): Likewise.
8069 (build5_stat): Likewise.
8070 (build_decl_stat): Likewise.
8071 * tree.h (make_node_stat): Adjust prototype.
8072 (make_node): remove macro.
8074 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
8076 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
8077 (PPC_FEATURE2_SCV): Likewise.
8078 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
8080 2017-07-28 Tamar Christina <tamar.christina@arm.com>
8082 * config/aarch64/aarch64.c
8083 (aarch64_internal_mov_immediate): Add new special pattern.
8084 * config/aarch64/aarch64.md (*movdi_aarch64):
8085 Add reg/32bit const mov case.
8087 2017-07-28 Tamar Christina <tamar.christina@arm.com>
8088 Richard Sandiford <richard.sandiford@linaro.org>
8090 * config/aarch64/aarch64.md (mov<mode>): Generalize.
8091 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
8092 Add integer and movi cases.
8093 (movi-split-hf-df-sf split, fp16): New.
8094 (enabled): Added TARGET_FP_F16INST.
8095 * config/aarch64/iterators.md (GPF_HF): New.
8096 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
8098 2017-07-28 Tamar Christina <tamar.christina@arm.com>
8100 * config/aarch64/aarch64.c
8101 (aarch64_simd_container_mode): Add prototype.
8102 (aarch64_expand_mov_immediate): Add HI support.
8103 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
8104 (aarch64_can_const_movi_rtx_p): New.
8105 (aarch64_preferred_reload_class):
8106 Remove restrictions of using FP registers for certain SIMD operations.
8107 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
8108 (aarch64_valid_floating_const): Add integer move validation.
8109 (aarch64_simd_imm_scalar_p): Remove.
8110 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
8111 (aarch64_legitimate_constant_p): Expand list of supported cases.
8112 * config/aarch64/aarch64-protos.h
8113 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
8114 (aarch64_reinterpret_float_as_int): New.
8115 (aarch64_simd_imm_scalar_p): Remove.
8116 * config/aarch64/constraints.md (Uvi): New.
8117 (Dd): Split into Ds and new Dd.
8118 * config/aarch64/aarch64.md (*movsi_aarch64):
8120 (*movdi_aarch64): Add SIMD mov case.
8122 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8124 * tree-predcom.c: (struct chain): Handle store-store chain in which
8125 stores for elimination only store loop invariant values.
8126 (execute_pred_commoning_chain): Ditto.
8127 (prepare_initializers_chain_store_elim): Ditto.
8128 (prepare_finalizers): Ditto.
8129 (is_inv_store_elimination_chain): New function.
8130 (initialize_root_vars_store_elim_1): New function.
8132 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8134 * tree-predcom.c: Revise general description of the pass.
8135 (enum chain_type): New enum type for store elimination.
8136 (struct chain): New field supporting store elimination.
8137 (struct component): Ditto.
8138 (dump_chain): Dump store-stores chain.
8139 (release_chain): Release resources.
8140 (split_data_refs_to_components): Compute and create component
8141 contains only stores for elimination.
8142 (get_chain_last_ref_at): New function.
8143 (make_invariant_chain): Initialization.
8144 (make_rooted_chain): Specify chain type in parameter and record it.
8145 (add_looparound_copies): Skip for store-stores chain.
8146 (determine_roots_comp): Compute type of chain and pass it to
8148 (initialize_root_vars_store_elim_2): New function.
8149 (finalize_eliminated_stores): New function.
8150 (remove_stmt): Handle store for elimination.
8151 (execute_pred_commoning_chain): Execute predictive commoning on
8153 (determine_unroll_factor): Skip unroll for store-stores chain.
8154 (prepare_initializers_chain_store_elim): New function.
8155 (prepare_initializers_chain): Hanlde store-store chain.
8156 (prepare_finalizers_chain, prepare_finalizers): New function.
8157 (tree_predictive_commoning_loop): Return integer value indicating
8158 if loop is unrolled or lcssa form is corrupted.
8159 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
8161 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8163 * tree-predcom.c (initialize_root): Delete.
8164 (execute_pred_commoning_chain): Initialize root vars and replace
8165 reference of non-combined chain directly, rather than call above
8168 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8170 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
8171 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
8173 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8175 * tree-predcom.c (struct chain): New field init_seq.
8176 (release_chain): Release init_seq.
8177 (prepare_initializers_chain): Record intialization stmts in above
8179 (insert_init_seqs): New function.
8180 (tree_predictive_commoning_loop): Call insert_init_seqs.
8182 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8184 * tree-predcom.c (determine_roots_comp): Skip trivial components.
8186 2017-07-28 Richard Biener <rguenther@suse.de>
8188 * match.pd: Remove superfluous :c.
8189 * genmatch.c (simplify::id): Add member.
8190 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
8192 (current_id): New global.
8193 (dt_node::parent): Move from ...
8194 (dt_operand::parent): ... here. Add for_id member.
8195 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
8196 (decision_tree::find_node): Relax order requirement when
8197 merging DT_TRUE nodes to ones inbetween the current simplify
8198 and the one we try to merge with. Add diagnostic whenever
8199 we need to enforce pattern order by not merging.
8200 (decision_tree::insert): Set current_id.
8201 (decision_tree::print_node): Dump parent node and for_id.
8202 (parser::last_id): Add member.
8203 (parser::push_simplify): Assign unique id.
8204 (parser::parser): Initialize last_id.
8206 2017-07-28 Martin Liska <mliska@suse.cz>
8209 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
8210 gimple_build_debug_bind.
8212 2017-07-28 Richard Biener <rguenther@suse.de>
8214 PR tree-optimization/81502
8215 * match.pd: Add pattern combining BIT_INSERT_EXPR with
8217 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
8219 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
8220 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
8221 for BIT_FIELD_REF args.
8222 * fold-const.c (make_bit_field_ref): Likewise.
8223 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8225 2017-07-28 Jakub Jelinek <jakub@redhat.com>
8228 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
8229 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
8230 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
8231 Or it into SANITIZER_UNDEFINED.
8232 * ubsan.c: Include gimple-fold.h and varasm.h.
8233 (ubsan_expand_ptr_ifn): New function.
8234 (instrument_pointer_overflow): New function.
8235 (maybe_instrument_pointer_overflow): New function.
8236 (instrument_object_size): Formatting fix.
8237 (pass_ubsan::execute): Call instrument_pointer_overflow
8238 and maybe_instrument_pointer_overflow.
8239 * internal-fn.c (expand_UBSAN_PTR): New function.
8240 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
8241 * sanitizer.def (__ubsan_handle_pointer_overflow,
8242 __ubsan_handle_pointer_overflow_abort): New builtins.
8243 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
8244 * internal-fn.def (UBSAN_PTR): New internal function.
8245 * opts.c (sanitizer_opts): Add pointer-overflow.
8246 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
8247 * fold-const.c (build_range_check): Compute pointer range check in
8248 integral type if pointer arithmetics would be needed. Formatting
8251 2017-07-28 Martin Liska <mliska@suse.cz>
8254 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
8255 parameters that are of a variable-length.
8257 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
8259 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
8261 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
8262 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
8263 (CRT_CALL_STATIC_FUNCTION): Likewise.
8264 (ASM_DEFAULT_SPEC): New define.
8265 (ASM_SPEC32): Likewise.
8266 (ASM_SPEC64): Likewise.
8267 (ASM_SPEC_COMMON): Likewise.
8268 (ASM_SPEC): Likewise.
8269 (INVALID_64BIT): Likewise.
8270 (LINK_OS_DEFAULT_SPEC): Likewise.
8271 (LINK_OS_SPEC32): Likewise.
8272 (LINK_OS_SPEC64): Likewise.
8273 (POWERPC_LINUX): Likewise.
8274 (PTRDIFF_TYPE): Likewise.
8275 (RESTORE_FP_PREFIX): Likewise.
8276 (RESTORE_FP_SUFFIX): Likewise.
8277 (SAVE_FP_PREFIX): Likewise.
8278 (SAVE_FP_SUFFIX): Likewise.
8279 (SIZE_TYPE): Likewise.
8280 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
8281 (TARGET_64BIT): Likewise.
8282 (TARGET_64BIT): Likewise.
8283 (TARGET_AIX): Likewise.
8284 (WCHAR_TYPE_SIZE): Likewise.
8285 (WCHAR_TYPE): Undefine.
8286 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
8287 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
8288 (CPP_OS_RTEMS_SPEC): Delete.
8289 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
8290 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
8292 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
8294 2017-07-28 Jakub Jelinek <jakub@redhat.com>
8296 PR tree-optimization/81578
8297 * tree-parloops.c (build_new_reduction): Bail out if
8298 reduction_code isn't one of the standard OpenMP reductions.
8299 Move the details printing after that decision.
8301 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8303 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
8304 related to reload_in_progress.
8305 (splat_input_operand): Likewise.
8306 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
8308 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
8310 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
8311 (TARGET_INSTANTIATE_DECLS): Likewise.
8312 (legitimate_indexed_address_p): Delete reload_in_progress code.
8313 (rs6000_debug_legitimate_address_p): Likewise.
8314 (rs6000_eliminate_indexed_memrefs): Likewise.
8315 (rs6000_emit_le_vsx_store): Likewise.
8316 (rs6000_emit_move_si_sf_subreg): Likewise.
8317 (rs6000_emit_move): Likewise.
8318 (register_to_reg_type): Likewise.
8319 (rs6000_pre_atomic_barrier): Likewise.
8320 (rs6000_machopic_legitimize_pic_address): Likewise.
8321 (rs6000_allocate_stack_temp): Likewise.
8322 (rs6000_address_for_fpconvert): Likewise.
8323 (rs6000_address_for_altivec): Likewise.
8324 (rs6000_secondary_memory_needed_rtx): Delete function.
8325 (rs6000_check_sdmode): Likewise.
8326 (rs6000_alloc_sdmode_stack_slot): Likewise.
8327 (rs6000_instantiate_decls): Likewise.
8328 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
8329 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
8330 Delete reload_in_progress.
8331 (*vec_reload_and_plus_<mptrsize>): Likewise.
8332 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
8333 (vsx_div_v2di): Likewise.
8334 (vsx_udiv_v2di): Likewise.
8336 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8338 * config/rs6000/rs6000.opt (mlra): Replace with stub.
8339 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
8340 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
8341 (rs6000_debug_reg_global): Delete print of LRA status.
8342 (rs6000_option_override_internal): Delete dead LRA related code.
8343 (rs6000_lra_p): Delete function.
8344 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
8346 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
8348 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
8349 * config/riscv/rtems.h: New file.
8351 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8352 Sudakshina Das <sudi.das@arm.com>
8354 * config/aarch64/aarch64.md
8355 (define_split for and<mode>3nr_compare): Move
8356 non aarch64_logical_operand to a register.
8357 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
8358 register immediate operand to a register.
8359 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
8361 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8364 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
8366 2017-07-27 Richard Biener <rguenther@suse.de>
8368 PR tree-optimization/81573
8369 PR tree-optimization/81494
8370 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
8371 multi defuse cycle case.
8373 2017-07-27 Richard Biener <rguenther@suse.de>
8375 PR tree-optimization/81571
8376 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
8379 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
8381 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
8382 earlier and only if MASK_FPU is set. Adjust formatting.
8384 2017-07-27 Martin Liska <mliska@suse.cz>
8386 * opt-functions.awk: Add validation of value of Init.
8387 * optc-gen.awk: Pass new argument.
8389 2017-07-27 Martin Liska <mliska@suse.cz>
8391 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
8392 Fix wrong condition.
8394 2017-07-27 Martin Liska <mliska@suse.cz>
8396 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
8397 BBs and edges seen by autoFDO.
8399 2017-07-27 Richard Biener <rguenther@suse.de>
8401 PR tree-optimization/81502
8402 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
8403 with incompatible but same sized type.
8404 (execute_update_addresses_taken): Likewise.
8406 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
8408 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
8409 flag_tree_loop_vectorize rather than flag_tree_vectorize.
8411 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8414 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
8415 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
8416 Change s_operand to memory_operand.
8418 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
8420 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
8421 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
8422 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
8423 Emit instructions rather than returning an expression. Handle TFmode
8424 and KFmode by casting to TImode.
8425 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
8426 (rs6000_emit_le_vsx_store): Likewise.
8427 * config/rs6000/vsx.md (VSX_TI): New iterator.
8428 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
8429 (*vsx_le_undo_permute_<mode>): Likewise.
8430 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
8431 emit the split sequence.
8432 (*vsx_le_perm_store_<mode>): Likewise.
8434 2017-07-27 Jakub Jelinek <jakub@redhat.com>
8436 PR tree-optimization/81555
8437 PR tree-optimization/81556
8438 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
8439 if true, force CHANGED for the recursive invocation.
8440 (reassociate_bb): Remember original length of ops array, pass
8441 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
8443 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
8444 attributes for noipa attribute. For naked attribute use
8445 lookup_attribute first before lookup_attribute_spec.
8446 * final.c (rest_of_handle_final): Disable IPA RA for functions with
8448 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
8449 for functions with noipa attribute.
8450 (cgraph_externally_visible_p): Return true for functions with noipa
8452 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
8453 for functions with noipa attribute.
8454 * doc/extend.texi: Document noipa function attribute.
8455 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
8456 also for functions with noipa attribute.
8457 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
8459 2017-07-26 Andrew Pinski <apinski@cavium.com>
8461 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
8462 vec_unalign_load_cost and vec_unalign_store_cost.
8464 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8466 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
8467 -mvsx-small-integer option.
8468 (ISA_3_0_MASKS_IEEE): Likewise.
8469 (OTHER_VSX_VECTOR_MASKS): Likewise.
8470 (POWERPC_MASKS): Likewise.
8471 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
8472 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
8473 code, only testing for DImode being allowed in non-VSX floating
8475 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
8476 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
8478 (rs6000_option_override_internal): Delete -mvsx-small-integer.
8479 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
8480 TARGET_P8_VECTOR test.
8481 (rs6000_secondary_reload_simple_move): Likewise.
8482 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
8483 since TARGET_P9_VECTOR was already tested.
8484 (rs6000_opt_masks): Remove -mvsx-small-integer.
8485 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
8486 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8488 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
8489 test for TARGET_VEXTRACTUB was used, and that uses
8491 (p9 extract splitter): Likewise.
8492 (vsx_extract_<mode>_di_p9): Likewise.
8493 (vsx_extract_<mode>_store_p9): Likewise.
8494 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
8495 for TARGET_P9_VECTOR was used. Delete code that is now dead with
8496 the elimination of TARGET_VSX_SMALL_INTEGER.
8497 (vsx_extract_<mode>_p8): Likewise.
8498 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
8499 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
8500 (vsx_set_<mode>_p9): Likewise.
8501 (vsx_set_v4sf_p9): Likewise.
8502 (vsx_set_v4sf_p9_zero): Likewise.
8503 (vsx_insert_extract_v4sf_p9): Likewise.
8504 (vsx_insert_extract_v4sf_p9_2): Likewise.
8505 * config/rs6000/rs6000.md (sign extend splitter): Change
8506 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
8507 (floatsi<mode>2_lfiwax_mem): Likewise.
8508 (floatunssi<mode>2_lfiwzx_mem): Likewise.
8509 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
8510 since a test for TARGET_P9_VECTOR was used.
8511 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8512 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
8513 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8514 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
8515 TARGET_P8_VECTOR test.
8516 (fix_trunc<mode>si2_stfiwx): Likewise.
8517 (fix_trunc<mode>si2_internal): Likewise.
8518 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
8519 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8521 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
8522 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
8523 TARGET_P8_VECTOR test.
8524 (fixuns_trunc<mode>si2_stfiwx): Likewise.
8525 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
8526 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8528 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
8529 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
8530 since a test for TARGET_P9_VECTOR was used.
8531 (splitter for loading small constants): Likewise.
8533 2017-07-26 Andrew Pinski <apinski@cavium.com>
8535 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
8538 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
8541 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
8542 (fp_valid_at): Likewise.
8544 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
8546 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
8547 (qdf24xx_addrcost_table): Likewise.
8548 (cortexa57_tunings): Update to use generic_branch_cost.
8549 (cortexa72_tunings): Likewise.
8550 (cortexa73_tunings): Likewise.
8551 (qdf24xx_tunings): Likewise.
8553 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
8555 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
8556 (thunderx2t99_branch_cost): Likewise.
8557 (cortexa35_tunings): Update to use generic_branch_cost.
8558 (cortexa53_tunings): Likewise.
8559 (cortexa57_tunings): Likewise.
8560 (cortexa72_tunings): Likewise.
8561 (cortexa73_tunings): Likewise.
8562 (thunderx2t99_tunings): Likewise.
8564 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
8566 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
8567 (sparc_option_override): Honour MASK_FSMULD.
8568 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
8569 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
8570 * config/sparc/sparc.opt (mfsmuld): New option.
8571 * doc/invoke.texi (mfsmuld): Document option.
8573 2017-07-26 Marek Polacek <polacek@redhat.com>
8576 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
8578 2017-07-26 Richard Biener <rguenther@suse.de>
8580 * gimple-match-head.c (do_valueize): Return OP if valueize
8582 (get_def): New helper to get at the def stmt of a SSA name
8584 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
8585 do_valueize to get at the def stmt.
8586 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
8588 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
8591 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
8593 2017-07-26 Martin Liska <mliska@suse.cz>
8596 * function.c (expand_function_start): Make expansion of
8597 nonlocal_goto_save_area after parm_birth_insn.
8599 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
8601 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
8602 from all CPU target flags enable members.
8604 2017-07-26 Richard Biener <rguenther@suse.de>
8606 * genmatch.c (dt_simplify::gen): Make iterator vars const.
8607 (decision_tree::gen): Make 'type' const.
8608 (write_predicate): Likewise.
8610 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
8612 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
8613 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
8614 (rs6000_option_override_internal): Likewise.
8615 (rs6000_expand_vector_set): Likewise.
8616 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
8617 (TARGET_UPPER_REGS_SF): Likewise.
8618 (TARGET_UPPER_REGS_DI): Likewise.
8619 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
8620 (TARGET_DIRECT_MOVE_64BIT): Likewise.
8621 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
8622 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8623 (Splitters for DI constants in Altivec registers): Likewise.
8624 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
8625 (vsx_set_v4sf_p9): Likewise.
8626 (vsx_set_v4sf_p9_zero): Likewise.
8627 (vsx_insert_extract_v4sf_p9): Likewise.
8628 (vsx_insert_extract_v4sf_p9_2): Likewise.
8630 2017-07-25 Carl Love <cel@us.ibm.com>
8632 * doc/extend.texi: Update the built-in documentation file for the
8633 existing built-in functions
8634 vector signed char vec_cnttz (vector signed char);
8635 vector unsigned char vec_cnttz (vector unsigned char);
8636 vector signed short vec_cnttz (vector signed short);
8637 vector unsigned short vec_cnttz (vector unsigned short);
8638 vector signed int vec_cnttz (vector signed int);
8639 vector unsigned int vec_cnttz (vector unsigned int);
8640 vector signed long long vec_cnttz (vector signed long long);
8641 vector unsigned long long vec_cnttz (vector unsigned long long);
8643 2017-07-25 Andrew Pinski <apinski@cavium.com>
8645 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
8646 accesses where the use is for the first operand of a BIT_INSERT.
8648 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
8651 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
8652 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
8654 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
8656 * config/i386/gstabs.h: Delete.
8657 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
8659 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
8661 * config/i386/i386.c (ix86_decompose_address): Do not check for
8662 register RTX when looking at index_reg or base_reg.
8663 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
8665 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
8667 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
8668 to update EH info here.
8670 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8672 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
8674 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8676 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
8678 2017-07-25 Torsten Duwe <duwe@suse.de>
8680 * common.opt: Introduce -fpatchable-function-entry
8681 command line option, and its variables function_entry_patch_area_size
8682 and function_entry_patch_area_start.
8683 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
8684 including a two-value parser.
8685 * target.def (print_patchable_function_entry): New target hook.
8686 * targhooks.h (default_print_patchable_function_entry): New function.
8687 * targhooks.c (default_print_patchable_function_entry): Likewise.
8688 * toplev.c (process_options): Switch off IPA-RA if
8689 patchable function entries are being generated.
8690 * varasm.c (assemble_start_function): Look at the
8691 patchable-function-entry command line switch and current
8692 function attributes and maybe generate NOP instructions by
8693 calling the print_patchable_function_entry hook.
8694 * doc/extend.texi: Document patchable_function_entry attribute.
8695 * doc/invoke.texi: Document -fpatchable_function_entry
8696 command line option.
8697 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
8699 * doc/tm.texi: Re-generate.
8701 2017-07-25 Jakub Jelinek <jakub@redhat.com>
8704 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
8705 TARGET_AVX512DQ rather than TARGET_AVX512BW.
8707 2017-07-25 Tamar Christina <tamar.christina@arm.com>
8709 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
8711 2017-07-25 Richard Biener <rguenther@suse.de>
8713 PR tree-optimization/81455
8714 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
8715 not walk in cycles when looking for guards.
8717 2017-07-25 Richard Biener <rguenther@suse.de>
8719 PR tree-optimization/81529
8720 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
8721 when optimizing backedge uses.
8723 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
8725 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
8727 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
8728 to dl_section_ref. On AIX, append an expression to subtract
8729 the size of the section length to dl_section_ref.
8731 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
8733 * configure.ac: If any of the config.* scripts fail, exit 1.
8734 * configure: Regenerate.
8736 2017-07-25 Richard Biener <rguenther@suse.de>
8739 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
8740 of immediate uses, be more verbose on errors.
8742 2017-07-25 Richard Biener <rguenther@suse.de>
8744 PR tree-optimization/81510
8745 * tree-vect-loop.c (vect_is_simple_reduction): When the
8746 reduction stmt is not inside the loop bail out.
8748 2017-07-25 Richard Biener <rguenther@suse.de>
8750 PR tree-optimization/81303
8751 * tree-vect-loop-manip.c (vect_loop_versioning): Build
8752 profitability check against LOOP_VINFO_NITERSM1.
8754 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8756 * domwalk.c (cmp_bb_postorder): Simplify.
8757 (sort_bbs_postorder): New function. Use it...
8758 (dom_walker::walk): ...here to optimize common cases.
8760 2017-07-25 Martin Liska <mliska@suse.cz>
8763 * ipa-visibility.c (function_and_variable_visibility): Make the
8764 redirection just on target that supports aliasing.
8765 Fix GNU coding style.
8767 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
8770 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
8772 * config/arm/rtems.h: Likewise.
8773 * config/bfin/rtems.h: Likewise.
8774 * config/i386/rtemself.h: Likewise.
8775 * config/lm32/rtems.h: Likewise.
8776 * config/m32c/rtems.h: Likewise.
8777 * config/m68k/rtemself.h: Likewise.
8778 * config/microblaze/rtems.h: Likewise.
8779 * config/mips/rtems.h: Likewise.
8780 * config/moxie/rtems.h: Likewise.
8781 * config/nios2/rtems.h: Likewise.
8782 * config/powerpcspe/rtems.h: Likewise.
8783 * config/rs6000/rtems.h: Likewise.
8784 * config/rtems.h: Likewise.
8785 * config/sh/rtems.h: Likewise.
8786 * config/sh/rtemself.h: Likewise.
8787 * config/sparc/rtemself.h: Likewise.
8789 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
8792 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
8793 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
8794 * tree-ssa-structalias.c (alias_get_name): Same.
8796 2017-07-25 Bin Cheng <bin.cheng@arm.com>
8799 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
8800 instructions if no du chain is found.
8802 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
8804 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
8806 2017-07-25 Richard Biener <rguenther@suse.de>
8809 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
8812 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
8814 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
8816 (ISA_2_7_MASKS_SERVER): Likewise.
8817 (ISA_3_0_MASKS_IEEE): Likewise.
8818 (OTHER_P8_VECTOR_MASKS): Likewise.
8819 (OTHER_VSX_VECTOR_MASKS): Likewise.
8820 (POWERPC_MASKS): Likewise.
8821 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
8822 duplicate list of options.
8823 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
8824 explicit -mupper-regs options.
8825 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
8826 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
8827 alias for -mupper-regs-df.
8828 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
8829 (rs6000_init_hard_regno_mode_ok): Likewise.
8830 (rs6000_option_override_internal): Likewise.
8831 (rs6000_opt_masks): Likewise.
8832 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
8833 options in terms of whether -mvsx or -mpower8-vector was used.
8834 (TARGET_UPPER_REGS_DI): Likewise.
8835 (TARGET_UPPER_REGS_SF): Likewise.
8836 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
8837 -mupper-regs-* options.
8839 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
8841 * passes.c (emergency_dump_function): Print some empty lines and a
8842 header before the RTL dump.
8844 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
8846 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
8848 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
8851 * config/aarch64/aarch64.c (aarch64_classify_symbol):
8852 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
8854 2017-07-24 Carl Love <cel@us.ibm.com>
8856 * config/rs6000/rs6000-c.c: Add support for built-in functions
8857 vector float vec_extract_fp32_from_shorth (vector unsigned short);
8858 vector float vec_extract_fp32_from_shortl (vector unsigned short);
8859 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
8860 vec_extract_fp_from_shortl): Add defines for the two builtins.
8861 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
8862 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
8864 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
8865 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
8866 * doc/extend.texi: Update the built-in documentation file for the
8867 new built-in function.
8869 2017-07-24 Jakub Jelinek <jakub@redhat.com>
8872 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
8874 * doc/generic.texi: Likewise.
8875 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
8876 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
8878 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
8880 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
8881 (aarch64_mls_elt_merge<mode>): Likewise.
8883 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
8885 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
8886 having __cxa_atexit.
8888 2017-07-23 Michael Collison <michael.collison@arm.com>
8890 * config/arm/arm.c (arm_option_override): Deprecate
8891 use of -mstructure-size-boundary.
8892 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
8893 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
8895 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8898 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
8899 Reduce cost estimate for direct moves.
8901 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
8904 * config/i386/i386.c (ix86_option_override_internal): Disable
8905 BMI, BMI2 and TBM instructions for -m16.
8907 2017-07-21 Carl Love <cel@us.ibm.com>
8909 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8910 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8911 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8912 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8913 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8914 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8915 VMULOSW): New enum "unspec" values.
8916 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8917 altivec_vmulosw): New patterns.
8918 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8919 VMULOSW): Add definitions.
8921 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
8923 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
8924 (qdf24xx): Likewise.
8925 * config/aarch64/aarch64-options-extensions.def (rdma); New.
8926 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
8927 (AARCH64_FL_V8_1): Renumber.
8928 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
8929 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
8930 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
8931 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
8932 rdma to feature modifiers list.
8934 2017-07-21 Yury Gribov <tetra2005@gmail.com>
8937 * ipa-visibility (function_and_variable_visibility): Convert
8938 recursive PLT call to direct call if appropriate.
8940 2017-07-21 Andrew Pinski <apinski@cavium.com>
8942 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
8943 operand 1 to see if the types precision matches.
8944 * fold-const.c (operand_equal_p): Likewise.
8946 2017-07-21 Richard Biener <rguenther@suse.de>
8948 PR tree-optimization/81303
8949 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
8950 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
8951 (vect_peeling_hash_get_lowest_cost): Adjust.
8952 (vect_enhance_data_refs_alignment): Likewise. Use
8953 vect_get_peeling_costs_all_drs to compute the penalty for no
8954 peeling to match up costs.
8956 2017-07-21 Richard Biener <rguenther@suse.de>
8958 PR tree-optimization/81500
8959 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
8960 we didn't identify a reduction path.
8962 2017-07-21 Tom de Vries <tom@codesourcery.com>
8963 Cesar Philippidis <cesar@codesourcery.com>
8965 PR gcov-profile/81442
8966 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
8969 2017-07-21 Tom de Vries <tom@codesourcery.com>
8972 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
8974 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
8975 nvptx_override_options_after_change.
8977 2017-07-21 Ulrich Drepper <drepper@redhat.com>
8979 * dwarf2out.c (output_file_names): Avoid double testing for
8982 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
8984 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
8986 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
8988 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
8990 (try_crossjump_to_edge): Do not punt on partitioned functions.
8992 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
8994 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
8995 Put all BBs reachable only via paths crossing cold region to cold
8997 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
8999 2016-07-21 Richard Biener <rguenther@suse.de>
9001 PR tree-optimization/81303
9002 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
9003 into account prologue and epilogue iterations when raising
9004 min_profitable_iters to sth at least covering one vector iteration.
9006 2017-07-21 Tamar Christina <tamar.christina@arm.com>
9008 * config/arm/arm.c (arm_test_cpu_arch_dat):
9011 2017-07-20 Nathan Sidwell <nathan@acm.org>
9013 Remove TYPE_METHODS.
9014 * tree.h (TYPE_METHODS): Delete.
9015 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
9016 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
9017 (dbxout_type_methods): Scan TYPE_FIELDS.
9018 (dbxout_type): Don't check TYPE_METHODS here.
9019 * function.c (use_register_for_decl): Always ignore register for
9020 class types when not optimizing.
9021 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
9022 * tree.c (free_lang_data_in_type): Stitch out member functions and
9023 templates from TYPE_FIELDS.
9024 (build_distinct_type_copy, verify_type_variant,
9025 verify_type): Member fns are on TYPE_FIELDS.
9026 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
9027 * tree-pretty-print.c (dump_generic_node): Likewise.
9029 2017-07-20 Jakub Jelinek <jakub@redhat.com>
9032 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
9033 V2TImode and V4TImode.
9034 (ix86_expand_vector_extract): Likewise.
9035 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
9036 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
9037 (ssescalarmode): Handle V4TImode and V2TImode.
9038 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
9039 (*vec_extractv2ti, *vec_extractv4ti): New insns.
9040 (VEXTRACTI128_MODE): New mode iterator.
9041 (splitter for *vec_extractv?ti first element): New.
9042 (VEC_INIT_MODE): New mode iterator.
9043 (vec_init<mode>): Consolidate 3 expanders into one using
9044 VEC_INIT_MODE mode iterator.
9046 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
9048 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
9049 non_spilled_static_chain_regno_p.
9051 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
9053 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
9055 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
9057 * bb-reorder.c (connect_traces): Allow copying of blocks within
9060 2017-07-20 Richard Biener <rguenther@suse.de>
9062 * gimple.h (gimple_phi_result): Add gphi * overload.
9063 (gimple_phi_result_ptr): Likewise.
9064 (gimple_phi_arg): Likewise. Adjust index assert to only
9065 allow actual argument accesses rather than all slots available
9067 (gimple_phi_arg_def): Add gphi * overload.
9068 * tree-phinodes.c (make_phi_node): Initialize only actual
9070 (resize_phi_node): Clear memory not covered by old node,
9071 do not initialize excess argument slots.
9072 (reserve_phi_args_for_new_edge): Initialize new argument slot
9075 2017-07-20 Bin Cheng <bin.cheng@arm.com>
9077 PR tree-optimization/81388
9079 2016-07-21 Bin Cheng <bin.cheng@arm.com>
9081 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
9082 by removing computation of may_be_zero.
9084 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
9085 Tom de Vries <tom@codesourcery.com>
9088 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
9089 when gimple level profile disagrees with what RTL expander did.
9091 2017-07-20 Richard Biener <rguenther@suse.de>
9093 PR tree-optimization/61171
9094 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
9095 (vect_analyze_stmt): Add slp instance parameter.
9096 (vectorizable_reduction): Likewise.
9097 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
9098 (vect_is_simple_reduction): Deal with chains not detected
9099 as SLP reduction chain, specifically not properly associated
9100 chains containing a mix of plus/minus.
9101 (get_reduction_op): Remove.
9102 (get_initial_defs_for_reduction): Simplify, pass in whether
9103 this is a reduction chain, pass in the SLP node for the PHIs.
9104 (vect_create_epilog_for_reduction): Get the SLP instance as
9106 (vectorizable_reduction): Get the SLP instance as arg.
9107 During analysis remember the SLP node with the PHIs in the
9108 instance. Simplify getting at the vectorized reduction PHIs.
9109 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
9110 through SLP instance.
9111 (vect_slp_analyze_operations): Likewise.
9112 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
9113 (vect_transform_stmt): Likewise.
9115 2017-07-20 Tom de Vries <tom@codesourcery.com>
9117 PR tree-optimization/81489
9118 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
9119 read of phi arg location to before loop that modifies phi.
9121 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9123 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
9126 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
9129 * except.c (execute): Fix ordering issue.
9131 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
9133 PR rtl-optimization/81423
9134 * combine.c (make_compound_operation_int): Don't try to optimize
9135 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
9137 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
9139 PR rtl-optimization/81423
9140 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
9141 with a constant that is -1 in the truncated to mode.
9143 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
9145 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
9146 (determine_unlikely_bbs): ... here.
9147 * predict.h (propagate_unlikely_bbs_forward): Declare.
9148 * cfgexpand.c (pass_expand::execute): Use it.
9149 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
9151 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
9152 propagate_unlikely_bbs_forward.
9154 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
9157 * except.c (maybe_add_nop_after_section_switch): New function.
9160 2017-07-19 Tom de Vries <tom@codesourcery.com>
9162 * gimple.h (gimple_phi_set_arg): Make assert more strict.
9164 2017-07-19 Tom de Vries <tom@codesourcery.com>
9166 * gimple.h (gimple_phi_arg): Make assert more strict.
9168 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
9170 * config.gcc (powerpc*-*-*): Add mmintrin.h.
9171 * config/rs6000/mmintrin.h: New file.
9172 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
9174 2017-07-19 Jakub Jelinek <jakub@redhat.com>
9176 PR tree-optimization/81346
9177 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
9179 2017-07-19 Tom de Vries <tom@codesourcery.com>
9181 * config/nvptx/nvptx.md (VECIM): Add V2DI.
9183 2017-07-19 Tom de Vries <tom@codesourcery.com>
9185 * config/nvptx/nvptx-modes.def: Add V2DImode.
9186 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
9187 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
9188 (nvptx_output_mov_insn): Handle lack of mov.b128.
9189 (nvptx_print_operand): Handle 'H' and 'L' codes.
9190 (nvptx_vector_mode_supported): Allow V2DImode.
9191 (nvptx_preferred_simd_mode): New function.
9192 (nvptx_data_alignment): New function.
9193 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
9194 nvptx_preferred_simd_mode.
9195 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
9197 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
9199 2017-07-19 Tom de Vries <tom@codesourcery.com>
9201 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
9202 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
9203 (nvptx_vector_mode_supported): New function. Allow V2SImode.
9204 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
9205 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
9206 (mov<VECIM>_insn): New define_insn.
9207 (define_expand "mov<VECIM>): New define_expand.
9209 2017-07-19 Tom de Vries <tom@codesourcery.com>
9211 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
9213 2017-07-19 Jakub Jelinek <jakub@redhat.com>
9215 PR tree-optimization/81346
9216 * fold-const.h (fold_div_compare, range_check_type): Declare.
9217 * fold-const.c (range_check_type): New function.
9218 (build_range_check): Use range_check_type.
9219 (fold_div_compare): No longer static, rewritten into
9220 a match.pd helper function.
9221 (fold_comparison): Don't call fold_div_compare here.
9222 * match.pd (X / C1 op C2): New optimization using fold_div_compare
9225 2017-07-19 Nathan Sidwell <nathan@acm.org>
9227 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
9228 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
9229 * tree.c (find_decls_types_r, verify_type): Use
9230 TYPE_{MIN,MAX}_VALUE_RAW.
9231 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
9232 (hash_tree): Likewise.
9233 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
9235 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
9238 2017-07-18 Tom de Vries <tom@codesourcery.com>
9241 * omp-expand.c (expand_omp_for_static_chunk): Handle
9242 equal-argument loop exit phi.
9244 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
9247 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
9248 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
9249 operand 2 predicate.
9250 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
9251 operand 2 predicate.
9252 (ror,rol -> rorx splitters): Use const_int_operand as
9253 operand 2 predicate.
9255 2017-06-18 Richard Biener <rguenther@suse.de>
9257 PR tree-optimization/81410
9258 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
9259 the gap in the ! slp_perm SLP case after each group.
9261 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
9264 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
9267 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
9270 * predict.c (set_even_probabilities): Cleanup; do not affect
9271 probabilities that are already known.
9272 (combine_predictions_for_bb): Call even when count is set.
9274 2017-07-18 Nathan Sidwell <nathan@acm.org>
9276 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
9279 2017-07-18 Bin Cheng <bin.cheng@arm.com>
9282 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
9283 optimization for loop niter analysis.
9285 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
9288 * config/avr/avr.c (avr_optimize_casesi): Don't use
9289 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
9291 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
9293 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
9294 body_cost_vec from _vect_peel_extended_info.
9295 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
9296 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
9299 2017-07-18 Bin Cheng <bin.cheng@arm.com>
9301 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
9303 2017-07-18 Richard Biener <rguenther@suse.de>
9305 PR tree-optimization/80620
9306 PR tree-optimization/81403
9307 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
9308 info when re-using a VN table entry.
9310 2017-07-18 Richard Biener <rguenther@suse.de>
9312 PR tree-optimization/81418
9313 * tree-vect-loop.c (vectorizable_reduction): Properly compute
9314 vectype_in. Verify that with lane-reducing reduction operations
9315 we have a single def-use cycle.
9317 2017-07-17 Carl Love <cel@us.ibm.com>
9319 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
9321 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9322 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9323 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9324 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9325 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9326 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
9327 VMULOSW): New enum "unspec" values.
9328 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
9329 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
9330 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
9331 altivec_vmulosw): New patterns.
9332 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9333 VMULOSW): Add definitions.
9334 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
9336 * config/alpha/alpha.c: Include predict.h.
9338 2017-07-17 Yury Gribov <tetra2005@gmail.com>
9340 * tree-vrp.c (compare_assert_loc): Fix comparison function
9341 to return predictable results.
9343 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9345 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
9348 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
9349 * doc/invoke.texi (mexpand-adddi): Update text.
9351 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9353 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
9354 that also clobbers the CC register. The old expand code is moved
9356 (*arc_clzsi2): ... here.
9357 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
9358 the CC register. The old expand code is moved to ...
9359 (arc_ctzsi2): ... here.
9361 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9363 * config/arc/arc.opt (mindexed-loads): Use initial value
9364 TARGET_INDEXED_LOADS_DEFAULT.
9365 (mauto-modify-reg): Use initial value
9366 TARGET_AUTO_MODIFY_REG_DEFAULT.
9367 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
9368 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
9369 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
9370 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
9372 2017-07-17 Martin Liska <mliska@suse.cz>
9375 * opts.c (finish_options): Do not allow -fgnu-tm
9376 w/ -fsanitize={kernel-,}address. Say sorry.
9378 2017-07-17 Bin Cheng <bin.cheng@arm.com>
9381 * tree-loop-distribution.c (classify_partition): Only assert on
9382 numer of iterations.
9383 (merge_dep_scc_partitions): Delete prameter. Update function call.
9384 (distribute_loop): Remove code handling loop with unknown niters.
9385 (pass_loop_distribution::execute): Skip loop with unknown niters.
9387 2017-07-17 Bin Cheng <bin.cheng@arm.com>
9390 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
9391 function sort_partitions_by_post_order.
9393 2017-07-17 Bin Cheng <bin.cheng@arm.com>
9395 PR tree-optimization/81374
9396 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
9397 the max index of basic blocks, rather than number of basic blocks.
9399 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9401 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
9403 (arc_legitimate_pic_operand_p): Likewise.
9404 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
9406 (arc_needs_pcl_p): Likewise.
9407 (arc_legitimate_pc_offset_p): Likewise.
9408 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
9409 function is also used in constrains.md.
9410 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
9411 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
9412 PLUS. Only return true/false in known cases, otherwise assert.
9413 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
9414 is already called in arc_legitimate_constant_p.
9415 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
9417 (LEGITIMATE_PIC_OPERAND_P): Use
9418 arc_raw_symbolic_reference_mentioned_p function.
9419 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
9424 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9425 Andrew Burgess <andrew.burgess@embecosm.com>
9427 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
9428 (arc_return_address_register): New function.
9429 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
9430 (arc_handle_fndecl_attribute): Add naked attribute.
9431 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
9432 (TARGET_WARN_FUNC_RETURN): Likewise.
9433 (arc_allocate_stack_slots_for_args): New function.
9434 (arc_warn_func_return): Likewise.
9435 (machine_function): Change type fn_type.
9436 (arc_compute_function_type): Consider new naked function type,
9437 change function return type.
9438 (arc_must_save_register): Adapt to handle new
9439 arc_compute_function_type's return type.
9440 (arc_expand_prologue): Likewise.
9441 (arc_expand_epilogue): Likewise.
9442 (arc_return_address_regs): Delete.
9443 (arc_return_address_register): New function.
9444 (arc_epilogue_uses): Use above function.
9445 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
9446 (arc_function_type): Change encoding, add naked type.
9447 (ARC_INTERRUPT_P): Change to handle the new encoding.
9448 (ARC_FAST_INTERRUPT_P): Likewise.
9449 (ARC_NORMAL_P): Define.
9450 (ARC_NAKED_P): Likewise.
9451 (arc_compute_function_type): Delete prototype.
9452 * config/arc/arc.md (in_ret_delay_slot): Use
9453 arc_return_address_register function.
9454 (simple_return): Likewise.
9455 (p_return_i): Likewise.
9457 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9459 PR tree-optimization/81428
9460 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
9461 can't be built for those types.
9463 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
9465 Remove stuff dead since r239246.
9467 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
9468 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
9469 (avr_inform_devices): Remove dead stuff.
9471 2017-07-17 Tamar Christina <tamar.christina@arm.com>
9473 * config/arm/arm_neon.h: Fix softp typo.
9475 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9477 PR tree-optimization/81365
9478 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
9479 aggregate moves onto bb predecessor edges, make sure there are no
9480 loads that could alias the lhs in between the start of bb and the
9483 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
9486 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
9487 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
9488 [LSHIFTRT, outer_code = TRUNCATE]: Same.
9490 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9492 PR tree-optimization/81396
9493 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
9494 (init_symbolic_number): Initialize it to 1.
9495 (perform_symbolic_merge): Add n_ops from both operands into the new
9497 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
9498 without base_addr as useless if they need more than one operation.
9499 (bswap_replace): Handle !bswap case for NULL base_addr.
9501 2017-07-17 Tom de Vries <tom@codesourcery.com>
9504 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
9507 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
9509 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
9510 conditional builtin define __FIX_LEON3FT_B2BST.
9512 2017-07-17 Daniel Cederman <cederman@gaisler.com>
9514 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
9515 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
9518 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
9520 PR rtl-optimization/81424
9521 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
9522 to remove potential trapping from operands if -fnon-call-exceptions.
9524 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9526 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
9527 profile_proability for scalling.
9528 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
9530 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9532 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
9534 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9536 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
9537 fixpoint arithmetics.
9539 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9541 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
9542 fixpoint arithmetics.
9544 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9546 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
9547 fixpoint arithmetics.
9549 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9551 * profile-count.h (profile_probability::from_reg_br_prob_note,
9552 profile_probability::to_reg_br_prob_note): New functions.
9553 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
9554 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
9555 * predict.c (probability_reliable_p): Update.
9556 (edge_probability_reliable_p): Update.
9557 (br_prob_note_reliable_p): Update.
9558 (invert_br_probabilities): Update.
9559 (add_reg_br_prob_note): New function.
9560 (combine_predictions_for_insn): Update.
9561 * asan.c (asan_clear_shadow): Update.
9562 * cfgbuild.c (compute_outgoing_frequencies): Update.
9563 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
9564 (update_br_prob_note): Update.
9565 (rtl_verify_edges): Update.
9566 (purge_dead_edges): Update.
9567 (fixup_reorder_chain): Update.
9568 * emit-rtl.c (try_split): Update.
9569 * ifcvt.c (cond_exec_process_insns): Update.
9570 (cond_exec_process_if_block): Update.
9571 (dead_or_predicable): Update.
9572 * internal-fn.c (expand_addsub_overflow): Update.
9573 (expand_neg_overflow): Update.
9574 (expand_mul_overflow): Update.
9575 * loop-doloop.c (doloop_modify): Update.
9576 * loop-unroll.c (compare_and_jump_seq): Update.
9577 * optabs.c (emit_cmp_and_jump_insn_1): Update.
9578 * predict.h: Update.
9579 * reorg.c (mostly_true_jump): Update.
9581 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
9582 * config/alpha/alpha.c (emit_unlikely_jump): Update.
9583 * config/arc/arc.c: (emit_unlikely_jump): Update.
9584 * config/arm/arm.c: (emit_unlikely_jump): Update.
9585 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
9586 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
9587 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
9588 (ix86_print_operand): Update.
9589 (ix86_split_fp_branch): Update.
9590 (predict_jump): Update.
9591 * config/ia64/ia64.c (ia64_print_operand): Update.
9592 * config/mmix/mmix.c (mmix_print_operand): Update.
9593 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
9594 (rs6000_expand_split_stack_prologue): Update.
9595 * config/rs6000/rs6000.c: Update.
9596 * config/s390/s390.c (s390_expand_vec_strlen): Update.
9597 (s390_expand_vec_movstr): Update.
9598 (s390_expand_cs_tdsi): Update.
9599 (s390_expand_split_stack_prologue): Update.
9600 * config/sh/sh.c (sh_print_operand): Update.
9601 (expand_cbranchsi4): Update.
9602 (expand_cbranchdi4): Update.
9603 * config/sparc/sparc.c (output_v9branch): Update.
9604 * config/spu/spu.c (get_branch_target): Update.
9605 (ea_load_store_inline): Update.
9606 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
9607 * config/tilepro/tilepro.c: Update.
9609 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
9611 * gimplify.c (mostly_copy_tree_r): Revert latest change.
9612 (gimplify_save_expr): Likewise.
9614 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9616 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
9618 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9620 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
9622 * timevar.def (TV_IPA_FNSUMMARY): Define.
9624 2017-07-16 Daniel Cederman <cederman@gaisler.com>
9626 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
9627 to back store errata sensitive sequence from being generated.
9628 (sqrtdf2_fix): Likewise.
9630 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9632 * tree-ssa-threadupdate.c (compute_path_counts,
9633 update_joiner_offpath_counts): Use profile_probability.
9635 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
9638 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9640 * config/arm/arm-c.c (arm_cpu_builtins): Define
9641 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
9643 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
9645 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9646 array entries to represent __ieee128 versions of the
9647 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
9648 scalar_extract_sig, and scalar_insert_exp built-in functions.
9649 (altivec_resolve_overloaded_builtin): Add special case handling
9650 for the __builtin_scalar_insert_exp function, as represented by
9651 the P9V_BUILTIN_VEC_VSIEDP constant.
9652 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
9653 exponent support for __ieee128 argument.
9654 (VSESQP): Add scalar extract signature support for __ieee128
9656 (VSTDCNQP): Add scalar test negative support for __ieee128
9658 (VSIEQP): Add scalar insert exponent support for __int128 argument
9659 with __ieee128 result.
9660 (VSIEQPF): Add scalar insert exponent support for __ieee128
9661 argument with __ieee128 result.
9662 (VSTDCQP): Add scalar test data class support for __ieee128
9664 (VSTDCNQP): Add overload support for scalar test negative with
9666 (VSTDCQP): Add overload support for scalar test data class
9668 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
9670 (UNSPEC_VSX_SIEXPQP): New constant.
9671 (xsxexpqp): New insn for VSX scalar extract exponent quad
9673 (xsxsigqp): New insn for VSX scalar extract significand quad
9675 (xsiexpqpf): New insn for VSX scalar insert exponent quad
9676 precision with floating point argument.
9677 (xststdcqp): New expand for VSX scalar test data class quad
9679 (xststdcnegqp): New expand for VSX scalar test negative quad
9681 (xststdcqp): New insn to match expansions for VSX scalar test data
9682 class quad precision and VSX scalar test negative quad precision.
9683 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
9684 special case operand checking to enforce that second operand of
9685 VSX scalar test data class with quad precision argument is a 7-bit
9687 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
9688 prototypes and descriptions of __ieee128 versions of
9689 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
9690 scalar_test_data_class, and scalar_test_neg built-in functions.
9692 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9694 PR tree-optimization/81162
9695 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
9696 replace a negate with an add.
9698 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
9700 * doc/invoke.texi (arm/-mcpu): Document +crypto.
9702 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9704 * config/arm/arm-c.c (arm_cpu_builtins): Define
9705 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
9707 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9709 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
9710 (armv8-r): Set ARM Cortex-R52 as default CPU.
9711 * config/arm/arm-tables.opt: Regenerate.
9712 * config/arm/arm-tune.md: Regenerate.
9713 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
9715 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
9716 extension for -mcpu=cortex-r52.
9718 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9720 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
9721 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
9722 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
9723 (fp-armv8): Define it as FP_ARMv8 only.
9724 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
9725 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
9727 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
9728 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
9729 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
9730 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
9731 than TARGET_FPU_ARMV8.
9732 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
9733 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
9734 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
9736 * config/arm/neon.md (neon_vrint): Likewise.
9737 (neon_vcvt): Likewise.
9738 (neon_<fmaxmin_op><mode>): Likewise.
9739 (<fmaxmin><mode>3): Likewise.
9740 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
9741 * config/arm/predicates.md (arm_cond_move_operator): Check against
9742 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
9744 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
9746 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
9749 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9751 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
9752 loop in comment with memset.
9754 2017-07-14 Martin Liska <mliska@suse.cz>
9756 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
9757 * dwarf2out.c (is_java): Remove the function.
9758 (output_pubname): Remove usage of the function.
9759 (lower_bound_default): Remove usage of DW_LANG_Java.
9760 (gen_compile_unit_die): Likewise.
9761 * gcc.c: Remove compiler defaults for .java and .zip files.
9762 * gimple-expr.c (remove_suffix): Change as there's no longer
9763 extension than 4-letter one.
9764 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
9765 (gimplify_save_expr): Likewise.
9766 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
9767 as it's possible even for other languages than Java.
9768 * langhooks.h (struct lang_hooks): Remove Java from a comment.
9769 * lto-opts.c (lto_write_options): Remove reference to Java.
9770 * opts.c (strip_off_ending): Update file extension handling.
9771 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
9772 * tree-eh.c (lower_resx): Likewise.
9773 * tree.c (free_lang_data_in_type): Remove dead code.
9774 (find_decls_types_r): Likewise.
9775 (build_common_builtin_nodes): Remove Java from a comment.
9776 (verify_type): Remove dead code.
9777 * varasm.c (assemble_external): Remove Java from a comment.
9779 2017-07-14 Martin Liska <mliska@suse.cz>
9781 * opts.c (finish_options): Add quotes.
9782 (common_handle_option): Likewise.
9784 2017-07-14 Martin Liska <mliska@suse.cz>
9786 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
9787 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
9789 * dwarf2out.c (lower_bound_default): Do not handle
9791 (gen_compile_unit_die): Likewise.
9792 * gcc.c: Remove default extension binding for GNU Pascal.
9793 * stmt.c: Remove Pascal language from a comment.
9794 * xcoffout.c: Likewise.
9796 2017-07-13 David Malcolm <dmalcolm@redhat.com>
9799 * diagnostic-show-locus.c (fixit_cmp): New function.
9800 (layout::layout): Sort m_fixit_hints.
9801 (column_range::column_range): Assert that the values are valid.
9802 (struct char_span): New struct.
9803 (correction::overwrite): New method.
9804 (struct source_line): New struct.
9805 (line_corrections::add_hint): Add assertions. Reimplement memcpy
9806 calls in terms of classes source_line and char_span, and
9807 correction::overwrite.
9808 (selftest::test_overlapped_fixit_printing_2): New function.
9809 (selftest::diagnostic_show_locus_c_tests): Call it.
9811 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
9813 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
9814 early if there is no lhs.
9816 2017-07-13 Martin Liska <mliska@suse.cz>
9818 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
9819 (gen_reference_type_die): Likewise.
9820 * stor-layout.c: Remove Pascal-related comment.
9822 2017-07-13 Martin Liska <mliska@suse.cz>
9824 * opts.c (finish_options): Add quotes to error messages.
9825 (parse_sanitizer_options): Likewise.
9827 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
9829 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
9831 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
9833 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
9835 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
9837 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
9839 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
9841 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
9844 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
9845 provides the hardware capability bits, define the macro
9846 __BUILTIN_CPU_SUPPORTS__.
9847 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
9848 if GLIBC does not provide the hardware capability bits. Add a
9849 gcc_unreachable call if the built-in cpu function is neither
9850 __builtin_cpu_is nor __builtin_cpu_supports.
9851 (rs6000_get_function_versions_dispatcher): Change the warning
9852 that an old GLIBC is used which does not export the capability
9853 bits to be an error.
9854 * doc/extend.texi (target_clones attribute): Document the
9855 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
9856 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
9857 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
9858 the macros defined by GCC if the newer GLIBC is available.
9860 2017-07-12 Jeff Law <law@redhat.com>
9862 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
9863 remaining includes slightly.
9864 * config/riscv/riscv-builtins.c: Include profile-count.h.
9866 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
9869 * config/avr/avr.c (avr_set_current_function): In diagnostic
9870 messages: Quote keywords and (parts of) identifiers.
9871 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
9874 2017-07-12 Carl Love <cel@us.ibm.com>
9876 * config/rs6000/rs6000-c.c: Add support for built-in functions
9877 vector bool char vec_revb (vector bool char);
9878 vector bool short vec_revb (vector short char);
9879 vector bool int vec_revb (vector bool int);
9880 vector bool long long vec_revb (vector bool long long);
9881 * doc/extend.texi: Update the built-in documentation file for the
9882 new built-in functions.
9884 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9886 * config/s390/s390.md: Remove movcc splitter.
9888 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9890 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
9891 load/store on condition.
9893 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
9896 * config/avr/avr.c (avr_encode_section_info)
9897 [progmem && !TREE_READONLY]: Error if progmem object needs
9900 2017-07-11 Michael Collison <michael.collison@arm.com>
9902 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
9905 2017-07-11 Carl Love <cel@us.ibm.com>
9907 * config/rs6000/rs6000-c.c: Add support for builtins
9908 vector unsigned int vec_parity_lsbb (vector signed int);
9909 vector unsigned int vec_parity_lsbb (vector unsigned int);
9910 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
9911 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
9912 vector unsigned long long vec_parity_lsbb (vector signed long long);
9913 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
9914 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
9915 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
9916 * doc/extend.texi: Update the built-in documentation file for the
9917 new built-in functions.
9919 2017-07-11 David Malcolm <dmalcolm@redhat.com>
9921 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
9922 (layout::m_primary_loc): New field.
9923 (layout::layout): Initialize new field. Move location filtering
9924 logic from here to...
9925 (layout::maybe_add_location_range): ...this new method. Add
9926 support for filtering to just the lines already specified by other
9928 (layout::will_show_line_p): New method.
9929 (gcc_rich_location::add_location_if_nearby): New method.
9930 (selftest::test_add_location_if_nearby): New test function.
9931 (selftest::diagnostic_show_locus_c_tests): Call it.
9932 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
9935 2017-07-11 Tom de Vries <tom@codesourcery.com>
9937 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
9938 (bb_first_real_insn): New function.
9939 (nvptx_single): Add extra initialization of broadcasted condition
9942 2017-07-11 Nathan Sidwell <nathan@acm.org>
9944 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
9946 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
9948 * doc/extend.texi (AVR Function Attributes): Remove weblink to
9949 Binutils doc as TEXI will mess them up.
9950 * doc/invoke.texi (AVR Options): Same here.
9952 2017-07-11 Daniel Cederman <cederman@gaisler.com>
9954 * config/sparc/sparc.opt (mfix-ut700): New option.
9955 (mfix-gr712rc): Likewise.
9956 (sparc_fix_b2bst): New variable.
9957 * doc/invoke.texi (SPARC options): Document them.
9958 (ARM options): Fix warnings.
9959 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
9960 instructions to prevent sequences that can trigger the store-store
9961 errata for certain LEON3FT processors.
9962 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
9963 (sparc_option_override): Set sparc_fix_b2bst appropriately.
9964 * config/sparc/sparc.md (fix_b2bst): New attribute.
9965 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
9967 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
9970 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
9972 (*rsqrtsf2_sse): Ditto.
9974 (div<mode>3): Macroize insn from divdf3 and divsf3
9975 using MODEF mode iterator.
9977 2017-07-10 Martin Sebor <msebor@redhat.com>
9979 PR tree-optimization/80397
9980 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
9981 instead of testing for equality to INTEGER_TYPE.
9983 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
9985 * config.gcc: Remove uclibc from arc target spec.
9987 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
9989 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
9991 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9994 * lto-wrapper.c (remove_option): New function.
9995 (merge_and_complain): Merge PIC/PIE options more realistically.
9997 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
9999 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
10003 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
10004 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
10005 * config.in: Regenerate.
10006 * configure: Regenerate.
10007 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
10008 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
10009 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
10010 (TARGET_GASISR_PROLOGUES): ...target mask.
10011 * common/config/avr/avr-common.c
10012 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
10013 Set -mgas-isr-prologues.
10014 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
10015 INSERT_PASS_BEFORE for it.
10016 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
10017 * config/avr/avr.c (avr_option_override)
10018 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
10019 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
10020 (avr_attribute_table) <no_gccisr>: Add new function attribute.
10021 (avr_set_current_function) <is_no_gccisr>: Init machine field.
10022 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
10024 (make_avr_pass_pre_proep): New function.
10025 (emit_push_sfr) <treg>: Add argument to function and use it
10026 instead of TMP_REG.
10027 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
10028 and set machine->gasisr.yes.
10029 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
10030 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
10031 __gcc_isr.n_pushed to .L__stack_usage.
10032 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
10033 (avr_asm_final_postscan_insn): ...this new static function.
10034 * config/avr/avr.h (machine_function)
10035 <is_no_gccisr, use_L__stack_usage>: New fields.
10036 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
10037 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
10038 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
10039 (gasisr, *gasisr): New expander and insn.
10040 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
10041 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
10042 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
10044 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
10046 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
10049 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
10051 Move jump-tables out of .text again.
10054 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
10055 (ASM_OUTPUT_ADDR_VEC): New function.
10056 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
10057 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
10058 INSN_ADDRESSes as asm comment.
10059 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
10060 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
10061 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
10062 * config/avr/avr.md (*tablejump): Adjust comment.
10063 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
10064 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
10066 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
10067 (avr_output_addr_vec): New proto.
10068 (avr_log_t) <insn_addresses>: New field.
10070 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
10073 * config/i386/i386.c (ix86_function_arg_advance): Set
10074 outgoing_args_on_stack to true if there are outgoing arguments
10076 (ix86_function_arg): Likewise.
10077 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
10078 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
10079 * config/i386/i386.h (machine_function): Add
10080 outgoing_args_on_stack.
10082 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
10084 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
10085 supporting pthreds.
10086 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
10088 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
10090 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
10091 (REAL_H): Remove $(MACHMODE_H).
10092 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
10094 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
10095 $(MACHMODE_H) and double-int.h.
10096 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
10098 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
10101 2017-07-07 Andrew Pinski <apinski@cavium.com>
10103 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
10104 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
10106 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10108 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
10109 Add warning if GCC was not configured to link against a GLIBC that
10110 exports the hardware capability bits.
10111 (make_resolver_func): Make resolver function private and not a
10112 COMDAT function. Create the name with clone_function_name instead
10113 of make_unique_name.
10116 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
10117 correct operand in doing the split.
10119 2017-07-07 Carl Love <cel@us.ibm.com>
10121 * config/rs6000/rs6000-c: Add support for built-in function
10122 vector unsigned short vec_pack_to_short_fp32 (vector float,
10124 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
10125 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
10126 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
10127 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
10128 (convert_4f32_8i16): Add define_expand.
10129 * doc/extend.texi: Update the built-in documentation file for the
10130 new built-in function.
10132 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10134 * config/sparc/m8.md: New file.
10135 * config/sparc/sparc.md: Include m8.md.
10137 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10139 * config/sparc/sparc.opt: New option -mvis4b.
10140 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
10141 (sparc_option_override): Handle VIS4B.
10142 (enum sparc_builtins): Define
10143 SPARC_BUILTIN_DICTUNPACK{8,16,32},
10144 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
10145 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
10146 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
10147 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
10148 (check_constant_argument): New function.
10149 (sparc_vis_init_builtins): Define builtins
10150 __builtin_vis_dictunpack{8,16,32},
10151 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
10152 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
10153 __builtin_vis_fpcmpde{8,16,32}shl and
10154 __builtin_vis_fpcmpur{8,16,32}shl.
10155 (sparc_expand_builtin): Check that the constant operands to
10156 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
10157 constant and in range.
10158 * config/sparc/sparc-c.c (sparc_target_macros): Handle
10160 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
10161 (SPARC_IMM5_P): Likewise.
10162 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
10163 (enabled): Handle vis4b.
10164 (UNSPEC_DICTUNPACK): New unspec.
10165 (UNSPEC_FPCMPSHL): Likewise.
10166 (UNSPEC_FPUCMPSHL): Likewise.
10167 (UNSPEC_FPCMPDESHL): Likewise.
10168 (UNSPEC_FPCMPURSHL): Likewise.
10169 (cpu_feature): New CPU feature `vis4b'.
10170 (dictunpack{8,16,32}): New insns.
10171 (FPCSMODE): New mode iterator.
10172 (fpcscond): New code iterator.
10173 (fpcsucond): Likewise.
10174 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
10175 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
10176 (fpcmpde{8,16,32}{si,di}shl): Likewise.
10177 (fpcmpur{8,16,32}{si,di}shl): Likewise.
10178 * config/sparc/constraints.md: Define constraints `q' for unsigned
10179 2-bit integer constants and `t' for unsigned 5-bit integer
10181 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
10183 (imm5_operand_dictunpack16): Likewise.
10184 (imm5_operand_dictunpack32): Likewise.
10185 (imm2_operand): Likewise.
10186 * doc/invoke.texi (SPARC Options): Document -mvis4b.
10187 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
10188 ditunpack* and fpcmp*shl builtins.
10190 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10192 * config.gcc: Handle m8 in --with-{cpu,tune} options.
10193 * config.in: Add HAVE_AS_SPARC6 define.
10194 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
10196 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
10198 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
10199 (CPP_CPU_SPEC): Handle m8.
10200 (ASM_CPU_SPEC): Likewise.
10201 * config/sparc/sparc-opts.h (enum processor_type): Add
10203 * config/sparc/sparc.c (m8_costs): New struct.
10204 (sparc_option_override): Handle TARGET_CPU_m8.
10205 (sparc32_initialize_trampoline): Likewise.
10206 (sparc64_initialize_trampoline): Likewise.
10207 (sparc_issue_rate): Likewise.
10208 (sparc_register_move_cost): Likewise.
10209 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
10210 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
10211 (ASM_CPU64_DEFAULT_SPEC): Likewise.
10212 (CPP_CPU_SPEC): Handle M8.
10213 (ASM_CPU_SPEC): Likewise.
10214 (AS_M8_FLAG): Define.
10215 * config/sparc/sparc.md: Add m8 to the cpu attribute.
10216 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
10217 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
10219 * configure: Regenerate.
10220 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
10223 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10225 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
10227 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
10228 ("*movdi_insn_sp32"): Do not set v3pipe.
10229 ("*movsi_insn"): Likewise.
10230 ("*movdi_insn_sp64"): Likewise.
10231 ("*movsf_insn"): Likewise.
10232 ("*movdf_insn_sp32"): Likewise.
10233 ("*movdf_insn_sp64"): Likewise.
10234 ("*zero_extendsidi2_insn_sp64"): Likewise.
10235 ("*sign_extendsidi2_insn"): Likewise.
10236 ("*mov<VM32:mode>_insn"): Likewise.
10237 ("*mov<VM64:mode>_insn_sp64"): Likewise.
10238 ("*mov<VM64:mode>_insn_sp32"): Likewise.
10239 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
10240 ("<vlop:code><VL:mode>3"): Likewise.
10241 ("*not_<vlop:code><VL:mode>3"): Likewise.
10242 ("*nand<VL:mode>_vis"): Likewise.
10243 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
10244 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
10245 ("one_cmpl<VL:mode>2"): Likewise.
10246 ("faligndata<VM64:mode>_vis"): Likewise.
10247 ("alignaddrsi_vis"): Likewise.
10248 ("alignaddrdi_vis"): Likweise.
10249 ("alignaddrlsi_vis"): Likewise.
10250 ("alignaddrldi_vis"): Likewise.
10251 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
10252 ("bmaskdi_vis"): Likewise.
10253 ("bmasksi_vis"): Likewise.
10254 ("bshuffle<VM64:mode>_vis"): Likewise.
10255 ("cmask8<P:mode>_vis"): Likewise.
10256 ("cmask16<P:mode>_vis"): Likewise.
10257 ("cmask32<P:mode>_vis"): Likewise.
10258 ("pdistn<P:mode>_vis"): Likewise.
10259 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
10261 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10263 * config/sparc/sparc.md ("subtype"): New insn attribute.
10264 ("*wrgsr_sp64"): Set insn subtype.
10265 ("*rdgsr_sp64"): Likewise.
10266 ("alignaddrsi_vis"): Likewise.
10267 ("alignaddrdi_vis"): Likewise.
10268 ("alignaddrlsi_vis"): Likewise.
10269 ("alignaddrldi_vis"): Likewise.
10270 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
10271 ("fexpand_vis"): Likewise.
10272 ("fpmerge_vis"): Likewise.
10273 ("faligndata<VM64:mode>_vis"): Likewise.
10274 ("bshuffle<VM64:mode>_vis"): Likewise.
10275 ("cmask8<P:mode>_vis"): Likewise.
10276 ("cmask16<P:mode>_vis"): Likewise.
10277 ("cmask32<P:mode>_vis"): Likewise.
10278 ("fchksm16_vis"): Likewise.
10279 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
10280 ("fmean16_vis"): Likewise.
10281 ("fp<plusminus_insn>64_vis"): Likewise.
10282 ("<plusminus_insn>v8qi3"): Likewise.
10283 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
10284 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
10285 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
10286 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
10287 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
10288 ("*movqi_insn"): Likewise.
10289 ("*movhi_insn"): Likewise.
10290 ("*movsi_insn"): Likewise.
10291 ("movsi_pic_gotdata_op"): Likewise.
10292 ("*movdi_insn_sp32"): Likewise.
10293 ("*movdi_insn_sp64"): Likewise.
10294 ("movdi_pic_gotdata_op"): Likewise.
10295 ("*movsf_insn"): Likewise.
10296 ("*movdf_insn_sp32"): Likewise.
10297 ("*movdf_insn_sp64"): Likewise.
10298 ("*zero_extendhisi2_insn"): Likewise.
10299 ("*zero_extendqihi2_insn"): Likewise.
10300 ("*zero_extendqisi2_insn"): Likewise.
10301 ("*zero_extendqidi2_insn"): Likewise.
10302 ("*zero_extendhidi2_insn"): Likewise.
10303 ("*zero_extendsidi2_insn_sp64"): Likewise.
10304 ("ldfsr"): Likewise.
10305 ("prefetch_64"): Likewise.
10306 ("prefetch_32"): Likewise.
10307 ("tie_ld32"): Likewise.
10308 ("tie_ld64"): Likewise.
10309 ("*tldo_ldub_sp32"): Likewise.
10310 ("*tldo_ldub1_sp32"): Likewise.
10311 ("*tldo_ldub2_sp32"): Likewise.
10312 ("*tldo_ldub_sp64"): Likewise.
10313 ("*tldo_ldub1_sp64"): Likewise.
10314 ("*tldo_ldub2_sp64"): Likewise.
10315 ("*tldo_ldub3_sp64"): Likewise.
10316 ("*tldo_lduh_sp32"): Likewise.
10317 ("*tldo_lduh1_sp32"): Likewise.
10318 ("*tldo_lduh_sp64"): Likewise.
10319 ("*tldo_lduh1_sp64"): Likewise.
10320 ("*tldo_lduh2_sp64"): Likewise.
10321 ("*tldo_lduw_sp32"): Likewise.
10322 ("*tldo_lduw_sp64"): Likewise.
10323 ("*tldo_lduw1_sp64"): Likewise.
10324 ("*tldo_ldx_sp64"): Likewise.
10325 ("*mov<VM32:mode>_insn"): Likewise.
10326 ("*mov<VM64:mode>_insn_sp64"): Likewise.
10327 ("*mov<VM64:mode>_insn_sp32"): Likewise.
10329 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10331 * config/sparc/sparc.md ("type"): New insn type viscmp.
10332 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
10334 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
10335 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
10336 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
10337 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
10339 ("n7_vis_logical_11cycle"): Likewise.
10340 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
10341 * config/sparc/niagara2.md ("niag3_vis": Likewise.
10342 * config/sparc/niagara.md ("niag_vis"): Likewise.
10343 * config/sparc/ultra3.md ("us3_fga"): Likewise.
10344 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
10346 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10348 * config/sparc/sparc.md: New instruction type `bmask'.
10349 (bmaskdi_vis): Use the `bmask' type.
10350 (bmasksi_vis): Likewise.
10351 * config/sparc/ultra3.md (us3_array): Likewise.
10352 * config/sparc/niagara7.md (n7_array): Likewise.
10353 * config/sparc/niagara4.md (n4_array): Likewise.
10354 * config/sparc/niagara2.md (niag2_vis): Likewise.
10355 (niag3_vis): Likewise.
10356 * config/sparc/niagara.md (niag_vis): Likewise.
10358 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10360 * ipa-comdats.c: Remove optimize check from gate.
10361 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
10362 for functions not optimized.
10363 (ipa_fn_summary_read): Skip optimize check.
10364 (ipa_fn_summary_write): Likewise.
10365 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
10367 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
10369 (can_inline_edge_p): Check flag_pcc_struct_return for match.
10370 (check_callers): Give up on caller which is not optimized.
10371 (inline_small_functions): Likewise.
10372 (ipa_inline): Do not give up when not optimizing.
10373 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
10374 away unoptimizes cdtors.
10375 (whole_program_function_and_variable_visibility): Do
10376 ipa_discover_readonly_nonaddressable_vars in LTO mode.
10377 * ipa.c (process_references): Do not check optimize.
10378 (symbol_table::remove_unreachable_nodes): Update optimize check.
10379 (set_writeonly_bit): Update optimize check.
10380 (pass_ipa_cdtor_merge::gate): Do not check optimize.
10381 (pass_ipa_single_use::gate): Remove.
10383 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10385 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
10386 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
10387 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
10388 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
10389 permute_load, permute_store, adjust_extract, adjust_splat,
10390 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
10391 replace_swap_with_copy, dump_swap_insn_table,
10392 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
10393 recombine_lvx_pattern, recombine_stvx_pattern,
10394 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
10395 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
10396 to file rs6000-p8swap.c.
10397 * config/rs6000/rs6000-p8swap.c: New file.
10398 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
10399 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
10400 and rs6000*-*-* targets.
10402 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10404 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
10406 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10408 * lto-wrapper.c (merge_and_complain): Do not merge
10409 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
10410 fsigned_zeros, ftrapping_math, fwrapv.
10411 (append_compiler_options): Do not track these options.
10412 (append_linker_options): Likewie
10414 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10416 * cgraphunit.c (cgraph_node::finalize_function): When
10417 !flag_toplevel_reorde set no_reorder flag.
10418 (varpool_node::finalize_decl): Likewise.
10419 (symbol_table::compile): Drop no toplevel reorder path.
10421 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10423 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
10424 edges; zero probability is not better than uninitialized.
10426 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
10428 * asan.h (asan_sanitize_allocas_p): Declare.
10429 * asan.c (asan_sanitize_allocas_p): New function.
10430 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
10431 (handle_builtin_alloca): Likewise.
10432 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
10433 if !asan_sanitize_allocas_p.
10434 * params.def (asan-instrument-allocas): Add new option.
10435 * params.h (ASAN_PROTECT_ALLOCAS): Define.
10436 * opts.c (common_handle_option): Disable allocas sanitization for
10439 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
10441 * asan.c: Include gimple-fold.h.
10442 (get_last_alloca_addr): New function.
10443 (handle_builtin_stackrestore): Likewise.
10444 (handle_builtin_alloca): Likewise.
10445 (asan_emit_allocas_unpoison): Likewise.
10446 (get_mem_refs_of_builtin_call): Add new parameter, remove const
10447 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
10448 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
10449 (instrument_builtin_call): Pass gimple iterator to
10450 get_mem_refs_of_builtin_call.
10451 (last_alloca_addr): New global.
10452 * asan.h (asan_emit_allocas_unpoison): Declare.
10453 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
10454 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
10455 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
10456 if function calls alloca.
10457 * gimple-fold.c (replace_call_with_value): Remove static keyword.
10458 * gimple-fold.h (replace_call_with_value): Declare.
10459 * internal-fn.c: Include asan.h.
10460 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
10461 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
10463 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10465 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
10466 (C_SELFTEST_FLAGS): New.
10467 (CPP_SELFTEST_FLAGS): New.
10468 (SELFTEST_DEPS): New, from deps of s-selftest.
10469 (C_SELFTEST_DEPS): New, from deps of s-selftest.
10470 (CPP_SELFTEST_DEPS): New.
10471 (selftest): Add dependency on s-selftest-c++.
10472 (s-selftest): Rename to...
10473 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
10474 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
10475 than SELFTEST_FLAGS.
10476 (selftest-gdb): Rename to...
10477 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
10479 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
10480 (selftest-valgrind): Rename to...
10481 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
10483 (selftest-valgrind): Reintroduce as an alias for
10484 selftest-c-valgrind.
10485 (s-selftest-c++): New.
10486 (selftest-c++-gdb): New.
10487 (selftest-c++-valgrind): New.
10489 2017-07-06 Olivier Hainque <hainque@adacore.com>
10491 * gcc.c (process_command): When deciding if undefined variables
10492 should be ignored when processing specs, accept "gcc -v" as well.
10494 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10496 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
10497 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
10499 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
10501 * config/arm/arm-cpus.in (armv8-r): Add new entry.
10502 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
10503 * config/arm/arm-tables.opt: Regenerate.
10504 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
10506 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
10508 2017-07-06 Carl Love <cel@us.ibm.com>
10510 * ChangeLog: Clean up from mid air collision
10512 2017-07-06 Carl Love <cel@us.ibm.com>
10514 * config/rs6000/rs6000-c.c: Add support for built-in functions
10515 vector signed int vec_subc (vector signed int, vector signed int);
10516 vector signed __int128 vec_subc (vector signed __int128,
10517 vector signed __int128);
10518 vector unsigned __int128 vec_subc (vector unsigned __int128,
10519 vector unsigned __int128);
10520 vector signed int vec_sube (vector signed int, vector signed int,
10521 vector signed int);
10522 vector unsigned int vec_sube (vector unsigned int,
10523 vector unsigned int,
10524 vector unsigned int);
10525 vector signed __int128 vec_sube (vector signed __int128,
10526 vector signed __int128,
10527 vector signed__int128);
10528 vector unsigned __int128 vec_sube (vector unsigned __int128,
10529 vector unsigned __int128,
10530 vector unsigned __int128);
10531 vector signed int vec_subec (vector signed int, vector signed int,
10532 vector signed int);
10533 vector unsigned int vec_subec (vector unsigned int,
10534 vector unsigned int,
10535 vector unsigned int);
10536 vector signed __int128 vec_subec (vector signed __int128,
10537 vector signed __int128,
10538 vector signed__int128);
10539 vector unsigned __int128 vec_subec (vector unsigned __int128,
10540 vector unsigned __int128,
10541 vector unsigned __int128);
10542 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
10543 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
10544 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
10545 BU_ALTIVEC_OVERLOAD_X definitions.
10546 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
10547 * doc/extend.texi: Update the built-in documentation file for the new
10548 built-in functions.
10550 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10553 * diagnostic-show-locus.c (layout::layout): Use start and finish
10554 spelling location for the start and finish of each range.
10555 * genmatch.c (linemap_client_expand_location_to_spelling_point):
10556 Add unused aspect param.
10557 * input.c (expand_location_1): Add "aspect" param, and use it
10558 to access the correct part of the location.
10559 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
10561 (expand_location_to_spelling_point): Likewise.
10562 (linemap_client_expand_location_to_spelling_point): Add "aspect"
10563 param, and pass it to expand_location_1.
10565 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
10567 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
10568 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
10569 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
10570 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
10571 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
10572 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
10573 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
10574 _mm_maskz_getmant_ss): New intrinsics.
10575 (__builtin_ia32_getexpss128_mask): Changed to ...
10576 __builtin_ia32_getexpss128_round ... this.
10577 (__builtin_ia32_getexpsd128_mask): Changed to ...
10578 __builtin_ia32_getexpsd128_round ... this.
10579 * config/i386/i386-builtin-types.def
10580 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
10581 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
10582 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
10583 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
10584 __builtin_ia32_getmantss_mask_round): New builtins.
10585 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
10586 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
10587 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
10588 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
10589 * config/i386/sse.md
10590 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
10591 avx512f_sgetexp<mode><mask_scalar_name>
10592 <round_saeonly_scalar_name> ... this.
10593 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
10594 %0, %1, %2<round_saeonly_op3>}): Changed to ...
10595 vgetexp<ssescalarmodesuffix>
10596 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10597 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
10598 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
10599 avx512f_vgetmant<mode><mask_scalar_name>
10600 <round_saeonly_scalar_name> ... this.
10601 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
10602 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
10603 vgetmant<ssescalarmodesuffix>
10604 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
10605 %0<mask_scalar_operand4>, %1, %2
10606 <round_saeonly_scalar_mask_op4>, %3} ... this.
10607 * config/i386/subst.md (mask_scalar_operand4,
10608 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
10609 round_saeonly_scalar_nimm_predicate): New subst attributes.
10611 2017-07-06 Julia Koval <julia.koval@intel.com>
10613 * config/i386/i386.c (ix86_erase_embedded_rounding):
10614 Remove code for old rounding pattern.
10616 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
10618 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
10620 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
10622 * doc/sourcebuild.texi (Test Directives, Variants of
10623 dg-require-support): Add documentation for dg-require-stack-check.
10625 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
10627 * config/i386/subst.md (mask_scalar, round_scalar,
10628 round_saeonly_scalar): New meta-templates.
10629 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
10630 round_scalar_mask_operand3, round_scalar_mask_op3,
10631 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
10632 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
10633 round_saeonly_scalar_constraint,
10634 round_saeonly_scalar_prefix): New subst attribute.
10635 * config/i386/sse.md
10636 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
10637 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
10638 <round_scalar_name> ... this.
10639 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
10640 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
10641 <round_scalar_name> ... this.
10642 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
10643 <sse>_vm<code><mode>3<mask_scalar_name>
10644 <round_saeonly_scalar_name> ... this.
10645 (v<plusminus_mnemonic><ssescalarmodesuffix>
10646 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
10647 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
10648 v<plusminus_mnemonic><ssescalarmodesuffix>
10649 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10650 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
10651 (v<multdiv_mnemonic><ssescalarmodesuffix>
10652 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
10653 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
10654 v<multdiv_mnemonic><ssescalarmodesuffix>
10655 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10656 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
10657 (v<maxmin_float><ssescalarmodesuffix>
10658 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
10659 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
10660 v<maxmin_float><ssescalarmodesuffix>
10661 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10662 %0<mask_scalar_operand3>, %1, %<iptr>2
10663 <round_saeonly_scalar_mask_op3>} ... this.
10665 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
10667 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
10668 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
10670 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
10671 Alan Hayward <alan.hayward@arm.com>
10672 David Sherwood <david.sherwood@arm.com>
10674 * combine.c (simplify_if_then_else): Remove "enum" before
10676 * compare-elim.c (can_eliminate_compare): Likewise.
10677 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
10679 (aarch64_lookup_simd_builtin_type): Likewise.
10680 (aarch64_simd_builtin_type): Likewise.
10681 (aarch64_init_simd_builtin_types): Likewise.
10682 (aarch64_simd_expand_args): Likewise.
10683 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
10685 (aarch64_reverse_mask): Likewise.
10686 (aarch64_simd_emit_reg_reg_move): Likewise.
10687 (aarch64_gen_adjusted_ldpstp): Likewise.
10688 (aarch64_ccmp_mode_to_code): Likewise.
10689 (aarch64_operands_ok_for_ldpstp): Likewise.
10690 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10691 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
10693 (aarch64_min_divisions_for_recip_mul): Likewise.
10694 (aarch64_reassociation_width): Likewise.
10695 (aarch64_get_condition_code_1): Likewise.
10696 (aarch64_simd_emit_reg_reg_move): Likewise.
10697 (aarch64_simd_attr_length_rglist): Likewise.
10698 (aarch64_reverse_mask): Likewise.
10699 (aarch64_operands_ok_for_ldpstp): Likewise.
10700 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10701 (aarch64_gen_adjusted_ldpstp): Likewise.
10702 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
10704 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
10705 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
10706 (arm_lookup_simd_builtin_type): Likewise.
10707 (arm_simd_builtin_type): Likewise.
10708 (arm_init_simd_builtin_types): Likewise.
10709 (arm_expand_builtin_args): Likewise.
10710 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
10711 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
10712 (ft32_setup_incoming_varargs): Likewise.
10713 (ft32_function_arg): Likewise.
10714 (ft32_function_arg_advance): Likewise.
10715 (ft32_pass_by_reference): Likewise.
10716 (ft32_arg_partial_bytes): Likewise.
10717 (ft32_valid_pointer_mode): Likewise.
10718 (ft32_addr_space_pointer_mode): Likewise.
10719 (ft32_addr_space_legitimate_address_p): Likewise.
10720 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
10722 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
10723 (ix86_emit_outlined_ms2sysv_restore): Likewise.
10724 (iamcu_alignment): Likewise.
10725 (canonicalize_vector_int_perm): Likewise.
10726 (ix86_noce_conversion_profitable_p): Likewise.
10727 (ix86_mpx_bound_mode): Likewise.
10728 (ix86_operands_ok_for_move_multiple): Likewise.
10729 * config/microblaze/microblaze-protos.h
10730 (microblaze_expand_conditional_branch_reg): Likewise.
10731 * config/microblaze/microblaze.c
10732 (microblaze_expand_conditional_branch_reg): Likewise.
10733 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
10735 (rs6000_reassociation_width): Likewise.
10736 (rs6000_invalid_binary_op): Likewise.
10737 (fusion_p9_p): Likewise.
10738 (emit_fusion_p9_load): Likewise.
10739 (emit_fusion_p9_store): Likewise.
10740 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
10742 (riscv_hard_regno_mode_ok_p): Likewise.
10743 (riscv_address_insns): Likewise.
10744 (riscv_split_symbol): Likewise.
10745 (riscv_legitimize_move): Likewise.
10746 (riscv_function_value): Likewise.
10747 (riscv_hard_regno_nregs): Likewise.
10748 (riscv_expand_builtin): Likewise.
10749 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
10750 (riscv_build_integer): Likewise.
10751 (riscv_split_integer): Likewise.
10752 (riscv_legitimate_constant_p): Likewise.
10753 (riscv_cannot_force_const_mem): Likewise.
10754 (riscv_regno_mode_ok_for_base_p): Likewise.
10755 (riscv_valid_base_register_p): Likewise.
10756 (riscv_valid_offset_p): Likewise.
10757 (riscv_valid_lo_sum_p): Likewise.
10758 (riscv_classify_address): Likewise.
10759 (riscv_legitimate_address_p): Likewise.
10760 (riscv_address_insns): Likewise.
10761 (riscv_load_store_insns): Likewise.
10762 (riscv_force_binary): Likewise.
10763 (riscv_split_symbol): Likewise.
10764 (riscv_force_address): Likewise.
10765 (riscv_legitimize_address): Likewise.
10766 (riscv_move_integer): Likewise.
10767 (riscv_legitimize_const_move): Likewise.
10768 (riscv_legitimize_move): Likewise.
10769 (riscv_address_cost): Likewise.
10770 (riscv_subword): Likewise.
10771 (riscv_output_move): Likewise.
10772 (riscv_canonicalize_int_order_test): Likewise.
10773 (riscv_emit_int_order_test): Likewise.
10774 (riscv_function_arg_boundary): Likewise.
10775 (riscv_pass_mode_in_fpr_p): Likewise.
10776 (riscv_pass_fpr_single): Likewise.
10777 (riscv_pass_fpr_pair): Likewise.
10778 (riscv_get_arg_info): Likewise.
10779 (riscv_function_arg): Likewise.
10780 (riscv_function_arg_advance): Likewise.
10781 (riscv_arg_partial_bytes): Likewise.
10782 (riscv_function_value): Likewise.
10783 (riscv_pass_by_reference): Likewise.
10784 (riscv_setup_incoming_varargs): Likewise.
10785 (riscv_print_operand): Likewise.
10786 (riscv_elf_select_rtx_section): Likewise.
10787 (riscv_save_restore_reg): Likewise.
10788 (riscv_for_each_saved_reg): Likewise.
10789 (riscv_register_move_cost): Likewise.
10790 (riscv_hard_regno_mode_ok_p): Likewise.
10791 (riscv_hard_regno_nregs): Likewise.
10792 (riscv_class_max_nregs): Likewise.
10793 (riscv_memory_move_cost): Likewise.
10794 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
10795 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
10796 (rl78_addr_space_address_mode): Likewise.
10797 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10799 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
10800 (rs6000_reassociation_width): Likewise.
10801 (rs6000_invalid_binary_op): Likewise.
10802 (fusion_p9_p): Likewise.
10803 (emit_fusion_p9_load): Likewise.
10804 (emit_fusion_p9_store): Likewise.
10805 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
10806 (ok_for_simple_move_operands): Likewise.
10807 (ok_for_simple_move_strict_operands): Likewise.
10808 (ok_for_simple_arith_logic_operands): Likewise.
10809 (visium_legitimize_reload_address): Likewise.
10810 (visium_select_cc_mode): Likewise.
10811 (output_cbranch): Likewise.
10812 (visium_split_double_move): Likewise.
10813 (visium_expand_copysign): Likewise.
10814 (visium_expand_int_cstore): Likewise.
10815 (visium_expand_fp_cstore): Likewise.
10816 * config/visium/visium.c (visium_pass_by_reference): Likewise.
10817 (visium_function_arg): Likewise.
10818 (visium_function_arg_advance): Likewise.
10819 (visium_libcall_value): Likewise.
10820 (visium_setup_incoming_varargs): Likewise.
10821 (visium_legitimate_constant_p): Likewise.
10822 (visium_legitimate_address_p): Likewise.
10823 (visium_legitimize_address): Likewise.
10824 (visium_secondary_reload): Likewise.
10825 (visium_register_move_cost): Likewise.
10826 (visium_memory_move_cost): Likewise.
10827 (prepare_move_operands): Likewise.
10828 (ok_for_simple_move_operands): Likewise.
10829 (ok_for_simple_move_strict_operands): Likewise.
10830 (ok_for_simple_arith_logic_operands): Likewise.
10831 (visium_function_value_1): Likewise.
10832 (rtx_ok_for_offset_p): Likewise.
10833 (visium_legitimize_reload_address): Likewise.
10834 (visium_split_double_move): Likewise.
10835 (visium_expand_copysign): Likewise.
10836 (visium_expand_int_cstore): Likewise.
10837 (visium_expand_fp_cstore): Likewise.
10838 (visium_split_cstore): Likewise.
10839 (visium_select_cc_mode): Likewise.
10840 (visium_split_cbranch): Likewise.
10841 (output_cbranch): Likewise.
10842 (visium_print_operand_address): Likewise.
10843 * expmed.c (flip_storage_order): Likewise.
10844 * expmed.h (emit_cstore): Likewise.
10845 (flip_storage_order): Likewise.
10846 * genrecog.c (validate_pattern): Likewise.
10847 * hsa-gen.c (gen_hsa_addr): Likewise.
10848 * internal-fn.c (expand_arith_overflow): Likewise.
10849 * ira-color.c (allocno_copy_cost_saving): Likewise.
10850 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10851 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
10852 (process_invariant_for_inheritance): Likewise.
10853 * lra-eliminations.c (move_plus_up): Likewise.
10854 * omp-low.c (lower_oacc_reductions): Likewise.
10855 * simplify-rtx.c (simplify_subreg): Likewise.
10856 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
10857 (TARGET_CHKP_BOUND_MODE): Likewise..
10858 * targhooks.c (default_chkp_bound_mode): Likewise.
10859 (default_setup_incoming_vararg_bounds): Likewise.
10860 * targhooks.h (default_chkp_bound_mode): Likewise.
10861 (default_setup_incoming_vararg_bounds): Likewise.
10862 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
10863 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
10864 (have_whole_vector_shift): Likewise.
10865 * tree-vect-stmts.c (vectorizable_load): Likewise.
10866 * doc/tm.texi: Regenerate.
10868 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
10870 Graceful degrade if Binutils PR21472 is not available.
10873 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
10874 .rodata in flash test fails.
10875 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
10876 * confgure: Regenerate.
10877 * config.in: Regenerate.
10878 * config/avr/avr.c (avr_asm_named_section)
10879 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
10880 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
10881 (avr_asm_init_sections): Same.
10883 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10885 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
10886 (fma<VH:mode>4_intrinsic): Likewise.
10887 (*fmsub<VCVTF:mode>4): Likewise.
10888 (*fmsub<VH:mode>4_intrinsic): Likewise.
10890 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
10893 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
10894 Don't depend on "optimize > 0".
10895 (out_movhi_r_mr, out_movqi_mr_r): Same.
10896 (out_movhi_mr_r, out_movqi_r_mr): Same.
10897 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
10898 io_address_operand on "optimize > 0".
10900 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10902 * tree-loop-distribution.c: Add general explanantion on the pass.
10903 (generate_loops_for_partition): Mark distributed loop.
10904 (pg_add_dependence_edges): New parameter. Handle alias data
10905 dependence specially and record it in the parameter if asked.
10906 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
10907 (init_partition_graph_vertices, add_partition_graph_edge): New.
10908 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
10909 (free_partition_graph_vdata, build_partition_graph): New.
10910 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
10911 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
10912 (data_ref_segment_size, latch_dominated_by_data_ref): New.
10913 (compute_alias_check_pairs, version_loop_by_alias_check): New.
10914 (version_for_distribution_p, finalize_partitions): New.
10915 (distribute_loop): Handle alias data dependence specially. Factor
10916 out loop fusion code as functions and call these functions.
10918 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10920 * tree-loop-distribution.c (classify_partition): New parameter and
10921 better handle reduction statement.
10922 (rdg_build_partitions): Revise comment.
10923 (distribute_loop): Compute statements in all partitions and pass it
10924 to classify_partition.
10926 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10928 * tree-loop-distribution.c (enum partition_type): New.
10929 (struct partition): New field type.
10930 (partition_merge_into): Add parameter. Update partition type.
10931 (data_dep_in_cycle_p, update_type_for_merge): New functions.
10932 (build_rdg_partition_for_vertex): Compute partition type.
10933 (rdg_build_partitions): Dump partition type.
10934 (distribute_loop): Update calls to partition_merge_into.
10936 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10938 * tree-loop-distribution.c (struct ddr_hasher): New.
10939 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
10941 (classify_partition): Call get_data_dependence.
10942 (pg_add_dependence_edges): Ditto.
10943 (distribute_loop): Release data dependence hash table.
10945 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10947 * tree-loop-distribution.c (ref_base_address): Delete.
10948 (similar_memory_accesses): Rename ...
10949 (share_memory_accesses): ... to this. Check if partitions access
10950 the same memory reference.
10951 (distribute_loop): Call share_memory_accesses.
10953 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10955 * tree-loop-distribution.c (struct partition): New field recording
10956 its data reference.
10957 (partition_alloc, partition_free): Init and release data refs.
10958 (partition_merge_into): Merge data refs.
10959 (build_rdg_partition_for_vertex): Collect data refs for partition.
10960 (pg_add_dependence_edges): Change parameters from vector to bitmap.
10962 (distribute_loop): Remve data refs from vertice data of partition
10965 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10967 * tree-loop-distribution.c (params.h): Include header file.
10968 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
10969 (datarefs_vec): New global var.
10970 (create_rdg_vertices): Use datarefs_vec directly.
10971 (free_rdg): Don't free data references.
10972 (build_rdg): Update use. Don't free data references.
10973 (distribute_loop): Compute global variable for data references.
10974 Bail out if there are too many data references.
10976 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10978 * tree-loop-distribution.c (loop_nest): New global var.
10979 (build_rdg): Use loop directly, rather than loop nest.
10980 (pg_add_dependence_edges): Remove loop nest parameter. Use global
10982 (distribute_loop): Compute global variable loop nest. Update use.
10984 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10986 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
10987 (partition_merge_into): New parameter. Dump reason for fusion.
10988 (distribute_loop): Update use of partition_merge_into.
10990 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10992 * tree-loop-distribution.c (bb_top_order_index): New.
10993 (bb_top_order_index_size, bb_top_order_cmp): New.
10994 (stmts_from_loop): Use topological order.
10995 (pass_loop_distribution::execute): Compute and release topological
10996 order for basic blocks.
10998 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11000 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
11003 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11005 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
11006 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
11007 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
11008 * internal-fn.def (LOOP_DIST_ALIAS): New.
11009 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
11010 (fold_loop_internal_call): ... this.
11011 (vect_loop_dist_alias_call): New function.
11012 (set_uid_loop_bbs): Call fold_loop_internal_call.
11013 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
11016 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
11019 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
11020 Require dead FLAGS_REG at the beginning of a peephole.
11022 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
11025 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
11026 arguments in the call to __builtin_ia32_sbb_u32.
11027 (_subborrow_u64): Swap _X and _Y arguments in the call to
11028 __builtin_ia32_sbb_u64.
11030 2017-07-04 Jakub Jelinek <jakub@redhat.com>
11033 * tree-vrp.c (compare_assert_loc): Turn into a function template
11034 with stable template parameter. Only test if a->e is NULL,
11035 !a->e == !b->e has been verified already. Use e == NULL or
11036 e != NULL instead of e or ! e tests. If stable is true, don't use
11037 iterative_hash_expr, on the other side allow a or b or both NULL
11038 and sort the NULLs last.
11039 (process_assert_insertions): Sort using compare_assert_loc<false>
11040 instead of compare_assert_loc, later sort using
11041 compare_assert_loc<true> before calling process_assert_insertions_for
11042 in a loop. Use break instead of continue once seen NULL pointer.
11044 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
11046 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
11047 Cortex-R7 and Cortex-R8 processors.
11049 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
11051 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
11052 uninitialized while src is not.
11054 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
11056 * common/config/arm/arm-common.c: Adjust include path for
11058 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
11059 (arm-cpu.h): Create in build directory. Adjust dependency rules.
11060 (arm-cpu-data.h): Likewise.
11061 (arm-cpu-cdata.h): Likewise.
11062 * config/arm/arm-cpu.h: Delete.
11063 * config/arm/arm-cpu-cdata.h: Delete.
11064 * config/arm/arm-cpu-data.h: Delete.
11066 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
11068 * config/arm/arm-cpus.in (cortex-a55): New.
11069 (cortex-a75): Likewise.
11070 (cortex-a75.cortex-a55): Likewise.
11071 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
11073 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
11074 * config/arm/arm-cpu-cdata.h: Regenerate.
11075 * config/arm/arm-cpu-data.h: Regenerate.
11076 * config/arm/arm-cpu.h: Regenerate.
11077 * config/arm/arm-tables.opt: Regenerate.
11078 * config/arm/arm-tune.md: Regenerate.
11080 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
11082 * haifa-sched.c (sched_create_recovery_edges): Update profile.
11084 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
11086 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
11089 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
11091 PR tree-optimization/81292
11092 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
11093 full_string_p, also call adjust_related_strinfos if the adjustment
11094 is simple, otherwise invalidate related strinfos.
11096 2017-07-04 Martin Liska <mliska@suse.cz>
11099 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
11100 newly created variable as DECL_IGNORED_P.
11102 2017-07-04 Martin Liska <mliska@suse.cz>
11105 * ipa-inline.c (inline_small_functions):
11106 Use xstrdup_for_dump.
11108 2017-07-04 Tom de Vries <tom@codesourcery.com>
11110 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
11112 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
11115 * config/darwin.c (darwin_function_switched_text_sections):
11118 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
11120 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
11122 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
11124 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
11126 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11128 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
11129 min_profitable_iters, and th as inclusive lower bounds.
11130 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
11131 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
11132 for min_profitable_iters and min_profitable_estimate.
11133 (vect_transform_loop): Treat th as an inclusive lower bound.
11134 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
11136 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
11139 * config/darwin.c (darwin_function_switched_text_sections):
11140 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
11141 in two pieces, and suppress the use of buf.
11143 2017-07-03 Nathan Sidwell <nathan@acm.org>
11145 * hash-table.h (hash_table_mod1): Fix indentation.
11147 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11149 PR middle-end/81290
11150 * predict.c (force_edge_cold): Be more careful about propagation
11152 * profile-count.h (profile_probability::guessed,
11153 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
11155 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
11157 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
11159 * doc/invoke.texi (rcpc architecture extension): Document it.
11161 2017-07-03 Richard Biener <rguenther@suse.de>
11163 PR tree-optimization/60510
11164 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
11165 the scalar reduction PHI and use it.
11166 (vectorizable_reduction): Properly guard the single_defuse_cycle
11167 path for non-SLP reduction chains where we cannot use it.
11168 Rework reduc_def/index and vector type deduction. Rework
11169 vector operand gathering during reduction op code-gen.
11170 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
11171 chains dissolve the chain and leave it to non-SLP reduction
11174 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11176 * tree-data-ref.h (dr_alignment): Declare.
11177 * tree-data-ref.c (dr_alignment): New function.
11178 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
11179 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
11181 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
11183 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11185 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
11186 and base_misalignment fields.
11187 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
11188 * tree-data-ref.c: Include builtins.h.
11189 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
11190 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
11191 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
11192 * tree-vect-data-refs.c: Include tree-cfg.h.
11193 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
11194 fields instead of calculating an alignment here.
11195 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
11196 innermost_loop_behavior fields.
11198 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11200 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
11202 (DR_STEP_ALIGNMENT): New macro.
11203 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
11204 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
11205 (create_data_ref): Print it.
11206 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
11207 to tell whether the step preserves vector (mis)alignment.
11208 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
11209 Move the check for an integer step and generalise to all INTEGER_CST.
11210 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
11211 Print the outer step alignment.
11213 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11215 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
11216 with offset_alignment.
11217 (DR_ALIGNED_TO): Delete.
11218 (DR_OFFSET_ALIGNMENT): New macro.
11219 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
11220 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
11221 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
11222 (create_data_ref): Likewise.
11223 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
11224 (vect_analyze_data_refs): Likewise.
11225 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
11226 creating dummy innermost behavior.
11228 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11230 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
11231 with a "innermost_loop_behavior *" and refeence tree.
11232 * tree-data-ref.c (dr_analyze_innermost): Likewise.
11233 (create_data_ref): Update call accordingly.
11234 * tree-predcom.c (find_looparound_phi): Likewise.
11236 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11238 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
11239 fields with dr_wrt_vec_loop.
11240 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
11241 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
11242 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
11243 (vect_dr_behavior): New function.
11244 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
11245 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
11246 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
11247 track whether the step preserves the misalignment.
11248 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
11249 Use vect_dr_behavior.
11250 (vect_setup_realignment): Update call accordingly.
11251 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
11252 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
11253 call to vect_create_addr_base_for_vector_ref.
11254 (vect_create_cond_for_align_checks): Likewise.
11255 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
11256 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
11257 (vect_recog_mask_conversion_pattern): Likewise.
11258 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
11259 (new_stmt_vec_info): Remove redundant zeroing.
11261 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
11263 * common/config/arm/arm-common.c (arm_be8_option): New function.
11264 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
11265 (ISA_ARMv6): Add isa_bit_be8.
11266 * config/arm/arm.h (arm_be8_option): Add prototype.
11267 (BE8_SPEC_FUNCTION): New define.
11268 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
11269 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
11270 (mlittle-endian): Similarly.
11271 (mbe8, mbe32): New options.
11272 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
11273 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
11275 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11277 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
11279 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11281 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
11282 (cleanup_tree_cfg_bb): Use it.
11283 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
11285 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
11287 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11290 * loop-doloop.c (add_test): Update profile.
11292 2017-07-03 Martin Liska <mliska@suse.cz>
11295 * sanopt.c (rewrite_usage_of_param): New function.
11296 (sanitize_rewrite_addressable_params): Likewise.
11297 (pass_sanopt::execute): Call rewrite_usage_of_param.
11299 2017-07-03 Richard Biener <rguenther@suse.de>
11301 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
11302 back to using VIEW_CONVERT_EXPR.
11304 2017-07-03 Martin Liska <mliska@suse.cz>
11307 * doc/extend.texi: Document when a resolver function is
11308 generated for target_clones.
11310 2017-07-03 Martin Liska <mliska@suse.cz>
11312 * asan.c (asan_emit_stack_protection): Unpoison just red zones
11313 and shadow memory of auto variables which are subject of
11314 use-after-scope sanitization.
11315 (asan_expand_mark_ifn): Add do set only when is_poison.
11317 2016-07-03 Richard Biener <rguenther@suse.de>
11319 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
11321 (vect_force_simple_reduction): Record reduction def -> phi mapping.
11322 (vectorizable_reduction): Perform reduction PHI creation when
11323 visiting a reduction PHI and adjust and simplify code generation
11324 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
11325 (vect_transform_loop): Visit reduction PHIs.
11326 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
11327 defs into the SLP tree.
11328 (vect_build_slp_tree): Reduction defs terminate the recursion.
11329 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
11331 (vect_get_vec_defs_for_stmt_copy): Export.
11332 (vect_get_vec_defs): Likewise.
11333 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
11335 (vect_get_vec_defs_for_stmt_copy): Declare.
11336 (vect_get_vec_defs): Likewise.
11338 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11340 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
11341 parameter with a "loop" parameter and use it instead of the
11342 loop containing DR_STMT. Don't check simple_iv when doing
11343 BB analysis. Describe the two analysis modes in the comment.
11345 2017-07-03 Tom de Vries <tom@codesourcery.com>
11347 PR tree-optimization/69468
11348 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
11349 (find_same_succ_bb): Handle ignore_edge_flags.
11351 2017-07-03 Tom de Vries <tom@codesourcery.com>
11353 PR tree-optimization/81192
11354 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
11356 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
11358 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
11360 2017-07-03 Tom de Vries <tom@codesourcery.com>
11362 PR tree-optimization/81192
11363 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
11364 BB_SAME_SUCC (bb) == NULL.
11366 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11368 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
11371 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11373 * dumpfile.c: Include profile-count.h
11374 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
11376 (insert_cond_bb): Update profile.
11377 * tree-cfg.h (insert_cond_bb): Update prototype.
11378 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
11379 * tree-dump.c: Do not include tree-cfg.
11381 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11383 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
11385 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11387 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
11390 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11392 * tree-complex.c (expand_complex_div_wide): update profile.
11394 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11395 Alan Hayward <alan.hayward@arm.com>
11396 David Sherwood <david.sherwood@arm.com>
11398 * Makefile.in (MACHMODE_H): Remove insn-modes.h
11399 (CORETYPES_H): New define.
11400 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
11401 (insn-modes-inline.h, s-modes-inline-h): New rules.
11402 (generated_files): Add insn-modes-inline.h.
11403 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
11404 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
11405 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
11406 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
11407 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
11408 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
11409 (build/gencodes.o, build/genconditions.o): Likewise.
11410 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
11411 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
11412 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
11413 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
11414 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
11415 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
11416 * coretypes.h: Include everything up to real.h for generators.
11417 Include insn-modes.h first. Include wide-int-print.h after
11418 wide-int.h. Include insn-modes-inline.h and then machmode.h.
11419 * machmode.h: Don't include insn-modes.h here.
11420 * function-tests.c: Remove includes of signop.h, machmode.h,
11421 double-int.h and wide-int.h.
11423 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
11425 * optc-save-gen.awk: Likewise.
11426 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
11427 * godump.c: Remove include of wide-int-print.h.
11428 * pretty-print.h: Likewise.
11429 * wide-int-print.cc: Likewise.
11430 * wide-int.cc: Likewise.
11431 * hash-map-tests.c: Remove include of signop.h.
11432 * hash-set-tests.c: Likewise.
11433 * rtl-tests.c: Likewise.
11434 * mkconfig.sh: Remove include of machmode.h.
11435 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
11437 (emit_insn_modes_inline_h): ...this new function. Emit the code
11438 into an insn-modes-inline.h header file, adding appropriate
11439 include guards and end comments.
11440 (emit_insn_modes_c_header): Remove include of machmode.h.
11441 (emit_min_insn_modes_c_header): Include coretypes.h rather than
11443 (main): Handle -i flag and call emit_insn_modes_inline_h when
11446 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11448 * tree-ssa-strlen.c (strinfo): Rename the length field to
11449 nonzero_chars. Add a full_string_p field.
11450 (compare_nonzero_chars, zero_length_string_p): New functions.
11451 (get_addr_stridx): Add an offset_out parameter.
11452 Use compare_nonzero_chars.
11453 (get_stridx): Update accordingly. Use compare_nonzero_chars.
11454 (new_strinfo): Update after above changes to strinfo.
11455 (set_endptr_and_length): Set full_string_p.
11456 (get_string_length): Update after above changes to strinfo.
11457 (unshare_strinfo): Update call to new_strinfo.
11458 (maybe_invalidate): Likewise.
11459 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
11460 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
11461 as a uhwi instead of an shwi. Update after above changes to
11462 strinfo and new_strinfo.
11463 (zero_length_string): Assert that chainsi contains full strings.
11464 Use zero_length_string_p. Update call to new_strinfo.
11465 (adjust_related_strinfos): Update after above changes to strinfo.
11466 Copy full_string_p from origsi.
11467 (adjust_last_stmt): Use zero_length_string_p.
11468 (handle_builtin_strlen): Update after above changes to strinfo and
11469 new_strinfo. Install the lhs as the string length if the previous
11470 entry didn't describe a full string.
11471 (handle_builtin_strchr): Update after above changes to strinfo
11473 (handle_builtin_strcpy): Likewise.
11474 (handle_builtin_strcat): Likewise.
11475 (handle_builtin_malloc): Likewise.
11476 (handle_pointer_plus): Likewise.
11477 (handle_builtin_memcpy): Likewise. Track nonzero characters
11478 that aren't necessarily followed by a nul terminator.
11479 (handle_char_store): Likewise.
11481 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11483 PR tree-optimization/80769
11484 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
11485 for malloc and calloc. Document the new invariant that all related
11486 strinfos have delayed lengths or none do.
11487 (verify_related_strinfos): Move earlier in file.
11488 (set_endptr_and_length): New function, split out from...
11489 (get_string_length): ...here. Also set the lengths of related
11491 (zero_length_string): Assert that chainsi has known (rather than
11493 (adjust_related_strinfos): Likewise.
11495 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11497 PR tree-optimization/81136
11498 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
11499 assert that two references with the same misalignment have the same
11500 compile-time misalignment if those compile-time misalignments
11503 2017-07-01 Andi Kleen <ak@linux.intel.com>
11505 * print-tree.c (print_node): Print all attributes.
11507 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11509 * cfg.c (scale_bbs_frequencies): New function.
11510 * cfg.h (scale_bbs_frequencies): Declare it.
11511 * cfgloopanal.c (single_likely_exit): Cleanup.
11512 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
11514 (scale_loop_profile): Likewise.
11515 (loop_version): Likewise.
11516 (create_empty_loop_on_edge): Update.
11517 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
11518 scale_loop_frequencies, scale_loop_profile, loopify,
11519 loop_version): Update prototypes.
11520 * modulo-sched.c (sms_schedule): Update.
11521 * predict.c (unlikely_executed_edge_p): Also check probability.
11522 (probably_never_executed_edge_p): Fix typo.
11523 * tree-if-conv.c (version_loop_for_if_conversion): Update.
11524 * tree-parloops.c (gen_parallel_loop): Update.
11525 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
11526 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11527 * tree-ssa-loop-split.c (split_loop): Update.
11528 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
11529 * tree-vect-loop-manip.c (vect_do_peeling): Update.
11530 (vect_loop_versioning): Update.
11531 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11533 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11535 * trans-mem.c (split_bb_make_tm_edge): Update profile.
11537 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11539 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
11540 to keep profile consistent.
11542 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11544 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
11545 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
11546 * profile-count.h (max_safe_multiplier): Make unsigned.
11547 (profile_count::guessed_zero): New.
11549 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11551 * bb-reorder.c (fix_up_crossing_landing_pad,
11552 fix_crossing_conditional_branches): Use make_single_succ_edge
11553 to keep profile consistent.
11555 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11557 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
11560 2017-07-01 Jakub Jelinek <jakub@redhat.com>
11563 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
11564 the right scopes, make sure cond_jump isn't preserved between multiple
11565 iterations. Search for fallthru edge whenever there are 3+ edges and
11566 use find_fallthru_edge for it.
11568 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11570 Patch by Alexander Monakov <amonakov@ispras.ru>
11571 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
11572 probabilities consistently.
11574 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11576 * pa.c (pa_expand_compare_and_swap_loop): Update call of
11577 emit_cmp_and_jump_insns.
11579 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11582 * tree-inline.c (expand_call_inline): Combine profile statuses.
11584 2017-06-30 Andrew Pinski <apinski@cavium.com>
11586 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
11587 fold_stmt returned true.
11589 2017-06-30 Nathan Sidwell <nathan@acm.org>
11591 * ggc.h (empty_string): Delete.
11592 * cfgexpand.c (expand_asm_stmt): Use plain "".
11593 * optabs.c (expand_asm_memory_barrier): Likewise.
11594 * stringpool.c (empty_string): Delete.
11595 (digit_vector, digit_string): Delete.
11596 (ggc_alloc_string): Use plain "", don't optimize single digit
11597 strings. Use ggc_alloc_atomic.
11599 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
11601 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
11602 comparison set and one other set, use the cost of the non-comparison
11605 2017-06-30 Nathan Sidwell <nathan@acm.org>
11607 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
11610 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
11612 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
11613 loops. Remove now unneeded calls to gimple_switch_set_label() that
11614 just set removed labels to NULL_TREE.
11616 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
11618 * tree-ssanames.c (set_range_info_raw): Abstract from ...
11619 (set_range_info): ...here. Only call set_range_info_raw if domain
11621 (set_nonzero_bits): Call set_range_info_raw.
11622 * tree-ssanames.h (set_range_info_raw): New.
11624 2017-06-30 Jakub Jelinek <jakub@redhat.com>
11627 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
11628 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
11629 of nonimmediate_operand and <store_mask_constraint> instead of m
11630 for the input operand. For V8FI iterator, always split if input
11631 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
11632 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
11633 <store_mask_predicate> instead of register_operand and
11634 <store_mask_constraint> instead of v for the input operand. Make
11635 sure both operands aren't MEMs for if not <mask_applied>.
11637 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
11639 * lto-wrapper.c (copy_file) Close both file descriptors before
11642 2017-06-30 Martin Liska <mliska@suse.cz>
11645 * multiple_target.c (create_dispatcher_calls): Make ifunc
11646 also for function that don't have calls or are not referenced.
11648 2017-06-30 Richard Biener <rguenther@suse.de>
11650 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
11651 analyze the first scalar stmt. Move vector type computation
11652 for the BB case here from ...
11653 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
11654 live operation processing in the SLP case properly.
11656 2017-06-30 Richard Biener <rguenther@suse.de>
11658 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
11660 2017-06-30 Martin Liska <mliska@suse.cz>
11663 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
11664 before BUILT_IN_UNWIND_RESUME when ASAN is used.
11666 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
11668 * doc/invoke.texi (AArch64): Add missing options and remove redundant
11671 2017-06-30 Richard Biener <rguenther@suse.de>
11673 PR tree-optimization/81249
11674 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
11675 condition reduction result to original scalar type.
11677 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11679 * profile-count.h (enum profile_quality): Fix typos and whitespace
11682 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11684 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
11685 type for branch probabilities.
11687 2017-06-29 Julian Brown <julian@codesourcery.com>
11688 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11690 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
11691 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
11692 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
11693 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
11695 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11697 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
11698 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
11699 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
11700 CC usage from generic code to here.
11701 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
11702 CC usage into the target macros.
11704 2017-06-29 Maya Rashish <coypu@sdf.org>
11706 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
11709 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11711 * arm/arm-builtins.c: Include profile-count.h
11712 * except.c (sjlj_emit_function_enter): Use
11713 profile_probability::unlikely.
11715 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11717 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
11718 and tocrel_offset be pointer args rather than implicitly using
11720 (legitimate_constant_pool_address_p, rs6000_emit_move,
11721 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
11722 tocrel_offset and use in toc_relative_expr_p call.
11723 (print_operand, print_operand_address): Use static tocrel_base_oac
11724 and tocrel_offset_oac.
11725 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
11728 2017-06-29 Maya Rashish <coypu@sdf.org>
11730 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
11732 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
11734 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
11735 objects, take into account only the alignment of 'op0' and 'mode1' if
11738 2017-06-29 Steve Ellcey <sellcey@cavium.com>
11740 * ccmp.c (ccmp_tree_comparison_p): New function.
11741 (ccmp_candidate_p): Update to use above function.
11742 (get_compare_parts): New function.
11743 (expand_ccmp_next): Update to use new functions.
11744 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
11746 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
11747 take mode as argument.
11748 * ccmp.h (expand_ccmp_expr): Add mode as argument.
11749 * expr.c (expand_expr_real_1): Pass mode as argument.
11751 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
11753 * combine.c (combine_instructions): Print insns to dump_file, together
11756 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11758 * asan.c (asan_emit_stack_protection): Update.
11759 (create_cond_insert_point): Update.
11760 * auto-profile.c (afdo_propagate_circuit): Update.
11761 * basic-block.h (struct edge_def): Turn probability to
11762 profile_probability.
11763 (EDGE_FREQUENCY): Update.
11764 * bb-reorder.c (find_traces_1_round): Update.
11765 (better_edge_p): Update.
11766 (sanitize_hot_paths): Update.
11767 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
11768 (make_single_succ_edge): Update.
11769 (check_bb_profile): Update.
11770 (dump_edge_info): Update.
11771 (update_bb_profile_for_threading): Update.
11772 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
11773 probabilitycount to 0.
11774 * cfgbuild.c (compute_outgoing_frequencies): Update.
11775 * cfgcleanup.c (try_forward_edges): Update.
11776 (outgoing_edges_match): Update.
11777 (try_crossjump_to_edge): Update.
11778 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
11779 (expand_gimple_tailcall): Update.
11780 (construct_init_block): Use make_single_succ_edge.
11781 (construct_exit_block): Use make_single_succ_edge.
11782 * cfghooks.c (verify_flow_info): Update.
11783 (redirect_edge_succ_nodup): Update.
11784 (split_edge): Update.
11785 (account_profile_record): Update.
11786 * cfgloopanal.c (single_likely_exit): Update.
11787 * cfgloopmanip.c (scale_loop_profile): Update.
11788 (set_zero_probability): Remove.
11789 (duplicate_loop_to_header_edge): Update.
11790 * cfgloopmanip.h (loop_version): Update prototype.
11791 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
11792 (force_nonfallthru_and_redirect): Update.
11793 (update_br_prob_note): Update.
11794 (rtl_verify_edges): Update.
11795 (purge_dead_edges): Update.
11796 (rtl_lv_add_condition_to_bb): Update.
11797 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
11798 * cgraphunit.c (init_lowered_empty_function): Update.
11799 (cgraph_node::expand_thunk): Update.
11800 * cilk-common.c: Include profile-count.h
11801 * dojump.c (inv): Remove.
11802 (jumpifnot): Update.
11803 (jumpifnot_1): Update.
11804 (do_jump_1): Update.
11806 (do_jump_by_parts_greater_rtx): Update.
11807 (do_compare_rtx_and_jump): Update.
11808 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
11809 do_jump_1. do_compare_rtx_and_jump): Update prototype.
11810 * dwarf2cfi.c: Include profile-count.h
11811 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
11812 (sjlj_emit_dispatch_table): Likewise.
11813 * explow.c: Include profile-count.h
11814 * expmed.c (emit_store_flag_force): Update.
11815 (do_cmp_and_jump): Update.
11816 * expr.c (compare_by_pieces_d::generate): Update.
11817 (compare_by_pieces_d::finish_mode): Update.
11818 (emit_block_move_via_loop): Update.
11819 (store_expr_with_bounds): Update.
11820 (store_constructor): Update.
11821 (expand_expr_real_2): Update.
11822 (expand_expr_real_1): Update.
11823 * expr.h (try_casesi, try_tablejump): Update prototypes.
11824 * gimple-pretty-print.c (dump_probability): Update.
11825 (dump_profile): New.
11826 (dump_gimple_label): Update.
11827 (dump_gimple_bb_header): Update.
11828 * graph.c (draw_cfg_node_succ_edges): Update.
11829 * hsa-gen.c (convert_switch_statements): Update.
11830 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
11831 (find_if_case_1): Update.
11832 (find_if_case_2): Update.
11833 * internal-fn.c (expand_arith_overflow_result_store): Update.
11834 (expand_addsub_overflow): Update.
11835 (expand_neg_overflow): Update.
11836 (expand_mul_overflow): Update.
11837 (expand_vector_ubsan_overflow): Update.
11838 * ipa-cp.c (good_cloning_opportunity_p): Update.
11839 * ipa-split.c (split_function): Use make_single_succ_edge.
11840 * ipa-utils.c (ipa_merge_profiles): Update.
11841 * loop-doloop.c (add_test): Update.
11842 (doloop_modify): Update.
11843 * loop-unroll.c (compare_and_jump_seq): Update.
11844 (unroll_loop_runtime_iterations): Update.
11845 * lra-constraints.c (lra_inheritance): Update.
11846 * lto-streamer-in.c (input_cfg): Update.
11847 * lto-streamer-out.c (output_cfg): Update.
11848 * mcf.c (adjust_cfg_counts): Update.
11849 * modulo-sched.c (sms_schedule): Update.
11850 * omp-expand.c (expand_omp_for_init_counts): Update.
11851 (extract_omp_for_update_vars): Update.
11852 (expand_omp_ordered_sink): Update.
11853 (expand_omp_for_ordered_loops): Update.
11854 (expand_omp_for_generic): Update.
11855 (expand_omp_for_static_nochunk): Update.
11856 (expand_omp_for_static_chunk): Update.
11857 (expand_cilk_for): Update.
11858 (expand_omp_simd): Update.
11859 (expand_omp_taskloop_for_outer): Update.
11860 (expand_omp_taskloop_for_inner): Update.
11861 * omp-simd-clone.c (simd_clone_adjust): Update.
11862 * optabs.c (expand_doubleword_shift): Update.
11863 (expand_abs): Update.
11864 (emit_cmp_and_jump_insn_1): Update.
11865 (expand_compare_and_swap_loop): Update.
11866 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
11867 * predict.c (predictable_edge_p): Update.
11868 (edge_probability_reliable_p): Update.
11869 (set_even_probabilities): Update.
11870 (combine_predictions_for_insn): Update.
11871 (combine_predictions_for_bb): Update.
11872 (propagate_freq): Update.
11873 (estimate_bb_frequencies): Update.
11874 (force_edge_cold): Update.
11875 * profile-count.c (profile_count::dump): Add missing space into dump.
11876 (profile_count::debug): Add newline.
11877 (profile_count::differs_from_p): Explicitly convert to unsigned.
11878 (profile_count::stream_in): Update.
11879 (profile_probability::dump): New member function.
11880 (profile_probability::debug): New member function.
11881 (profile_probability::differs_from_p): New member function.
11882 (profile_probability::differs_lot_from_p): New member function.
11883 (profile_probability::stream_in): New member function.
11884 (profile_probability::stream_out): New member function.
11885 * profile-count.h (profile_count_quality): Rename to ...
11886 (profile_quality): ... this one.
11887 (profile_probability): New.
11888 (profile_count): Update.
11889 * profile.c (compute_branch_probabilities): Update.
11890 * recog.c (peep2_attempt): Update.
11891 * sched-ebb.c (schedule_ebbs): Update.
11892 * sched-rgn.c (find_single_block_region): Update.
11893 (compute_dom_prob_ps): Update.
11894 (schedule_region): Update.
11895 * sel-sched-ir.c (compute_succs_info): Update.
11896 * stmt.c (struct case_node): Update.
11897 (do_jump_if_equal): Update.
11898 (get_outgoing_edge_probs): Update.
11899 (conditional_probability): Update.
11900 (emit_case_dispatch_table): Update.
11901 (expand_case): Update.
11902 (expand_sjlj_dispatch_table): Update.
11903 (emit_case_nodes): Update.
11904 * targhooks.c: Update.
11905 * tracer.c (better_p): Update.
11906 (find_best_successor): Update.
11907 * trans-mem.c (expand_transaction): Update.
11908 * tree-call-cdce.c: Update.
11909 * tree-cfg.c (gimple_split_edge): Upate.
11910 (move_sese_region_to_fn): Upate.
11911 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
11912 * tree-eh.c (lower_resx): Upate.
11913 (cleanup_empty_eh_move_lp): Upate.
11914 * tree-if-conv.c (version_loop_for_if_conversion): Update.
11915 * tree-inline.c (copy_edges_for_bb): Update.
11916 (copy_cfg_body): Update.
11917 * tree-parloops.c (gen_parallel_loop): Update.
11918 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
11919 (gimple_gen_time_profiler): Update.
11920 * tree-ssa-dce.c (remove_dead_stmt): Update.
11921 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11922 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11923 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11924 (unloop_loops): Update.
11925 (try_peel_loop): Update.
11926 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11927 * tree-ssa-loop-split.c (connect_loops): Update.
11928 (split_loop): Update.
11929 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
11930 (hoist_guard): Update.
11931 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11932 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11933 (value_replacement): Update.
11934 * tree-ssa-reassoc.c (branch_fixup): Update.
11935 * tree-ssa-tail-merge.c (replace_block_by): Update.
11936 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11937 (create_edge_and_update_destination_phis): Update.
11938 (compute_path_counts): Update.
11939 (recompute_probabilities): Update.
11940 (update_joiner_offpath_counts): Update.
11941 (freqs_to_counts_path): Update.
11942 (duplicate_thread_path): Update.
11943 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11944 (struct switch_conv_info): Update.
11945 (gen_inbound_check): Update.
11946 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11947 (vect_do_peeling): Update.
11948 (vect_loop_versioning): Update.
11949 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11950 (optimize_mask_stores): Update.
11951 * ubsan.c (ubsan_expand_null_ifn): Update.
11952 * value-prof.c (gimple_divmod_fixed_value): Update.
11953 (gimple_divmod_fixed_value_transform): Update.
11954 (gimple_mod_pow2): Update.
11955 (gimple_mod_pow2_value_transform): Update.
11956 (gimple_mod_subtract): Update.
11957 (gimple_mod_subtract_transform): Update.
11958 (gimple_ic): Update.
11959 (gimple_stringop_fixed_value): Update.
11960 (gimple_stringops_transform): Update.
11961 * value-prof.h: Update.
11963 2017-06-29 Carl Love <cel@us.ibm.com>
11965 * config/rs6000/rs6000-c.c: Add support for built-in functions
11966 vector signed int vec_signed (vector float);
11967 vector signed long long vec_signed (vector double);
11968 vector signed int vec_signed2 (vector double, vector double);
11969 vector signed int vec_signede (vector double);
11970 vector signed int vec_signedo (vector double);
11971 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
11972 instruction generator.
11973 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
11974 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
11975 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
11977 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
11978 vunsignede_v2df): Add define_expands.
11979 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
11980 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
11981 VEC_UNSIGNEDO): Add definitions.
11982 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
11983 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
11984 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
11985 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
11986 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
11987 * config/rs6000/altivec.h (vec_signed, vec_signed2,
11988 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
11989 vec_unsignede, vec_unsignedo): Add builtin defines.
11990 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
11992 * doc/extend.texi: Update the built-in documentation file for the
11993 new built-in functions.
11995 2017-06-29 Richard Biener <rguenther@suse.de>
11997 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
11998 reduction chains to LOOP_VINFO_REDUCTIONS.
11999 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
12000 SLP reductions after processing reduction chains.
12002 2017-06-29 Nathan Sidwell <nathan@acm.org>
12004 * builtins.c (fold_builtin_FUNCTION): Use
12005 lang_hooks.decl_printable_name.
12007 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
12009 PR middle-end/81194
12010 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
12011 with only one label.
12012 * stmt.c (expand_case): Assert NCASES is greater than one.
12014 2017-06-29 Richard Biener <rguenther@suse.de>
12016 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
12018 (group_case_labels): Likewise.
12019 (find_taken_edge): Push sanity checking on val to workers...
12020 (find_taken_edge_cond_expr): ... here
12021 (find_taken_edge_switch_expr): ... and here, handle cases
12022 with just a default label.
12023 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
12024 (group_case_labels): Likewise.
12025 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
12026 group_case_labels does anything cleanup the CFG again.
12028 2017-06-29 Bin Cheng <bin.cheng@arm.com>
12030 PR tree-optimization/81196
12031 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
12032 exit condition comparing two IVs.
12034 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
12036 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
12037 profile to the dummy entry at the end of the list of architectures.
12038 * config/arm/arm-cpu-cdata.h: Regenerated.
12040 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12041 Michael Collison <michael.collison@arm.com>
12044 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
12046 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
12047 (*aarch64_reg_<mode>3_minus_mask): New pattern.
12048 (*aarch64_<optab>_reg_di3_mask2): New pattern.
12049 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
12050 of shift when the shift amount is masked with constant equal to
12051 the size of the mode.
12052 * config/aarch64/predicates.md (subreg_lowpart_operator): New
12055 2017-06-29 Martin Liska <mliska@suse.cz>
12057 * config/i386/i386.opt: Change range from [1,5] to [0,5].
12059 2017-06-29 Yury Gribov <tetra2005@gmail.com>
12062 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
12065 (edge_growth_cache_entry::edge_growth_cache_entry): New
12067 (reset_edge_growth_cache): Update to use constructor.
12069 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12071 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
12072 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
12073 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
12075 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
12077 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
12078 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
12080 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
12082 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
12083 (*-linux-uclibc*): Add t-uclibc tmake_file.
12084 * config/t-musl: New.
12085 * config/t-uclibc: New.
12087 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
12089 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
12091 (gen_comm_data): Emit architectural setting of arch_prof.
12092 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
12094 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
12095 (armv8-m.base, armv8-m.main): Likewise.
12096 * arm-protos.h (arm_build_target): Add profile field.
12097 (arch_option): Likewise.
12098 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
12100 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
12101 arm_active_target.profile.
12103 2017-06-28 Richard Biener <rguenther@suse.de>
12105 PR middle-end/81227
12106 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
12107 TYPE_OVERFLOW_WRAPS.
12108 * match.pd (negate_expr_p): Likewise.
12109 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
12110 fold_build2, not fold_binary.
12112 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
12114 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
12115 Convert memory address to Pmode.
12116 (aarch64_print_operand): Assert MEM operands are always Pmode.
12118 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
12121 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
12122 Remove redundant if.
12123 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
12124 * config/arm/aarch-common-protos.h
12125 (aarch_forward_to_shift_is_not_shifted_re): Remove.
12126 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
12128 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
12131 * multiple_target.c (create_dispatcher_calls): Set the default
12132 clone to be static, not public.
12134 2017-06-28 Richard Biener <rguenther@suse.de>
12136 * tree-vect-loop.c (vectorizable_reduction): Move special
12137 cond reduction IV var creation ...
12138 (vect_create_epilog_for_reduction): ... here. Remove induction_index
12139 parameter. Use STMT_VINFO_VECTYPE.
12140 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
12143 2017-06-28 Martin Liska <mliska@suse.cz>
12146 * ipa-visibility.c (non_local_p): Handle visibility.
12148 2017-06-28 Martin Liska <mliska@suse.cz>
12151 * common.opt: Add IntegerRange to various options.
12152 * opt-functions.awk (integer_range_info): New function.
12153 * optc-gen.awk: Add integer_range_info to cl_options struct.
12154 * opts-common.c (decode_cmdline_option): Handle
12155 CL_ERR_INT_RANGE_ARG.
12156 (cmdline_handle_error): Likewise.
12157 * opts.c (print_filtered_help): Show valid interval in
12158 when --help is provided.
12159 * opts.h (struct cl_option): Add range_min and range_max fields.
12160 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
12162 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
12164 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
12165 (x * C EQ/NE y * C): New transformation.
12167 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
12169 * genmultilib (combination_space): Accept '+' in option names.
12171 2017-06-28 Martin Liska <mliska@suse.cz>
12174 * asan.c (instrument_derefs): Bail out inner references
12175 that are hard register variables.
12177 2017-06-28 Jakub Jelinek <jakub@redhat.com>
12180 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
12181 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
12183 2017-06-28 Richard Biener <rguenther@suse.de>
12185 * tree-vectorizer.h (vect_get_vec_defs): Remove.
12186 (vect_get_slp_defs): Adjust.
12187 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
12189 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
12191 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
12192 get_initial_defs_for_reduction instead of vect_get_vec_defs.
12193 (vectorizable_reduction): Adjust.
12194 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
12196 (vect_get_slp_defs): Likewise.
12197 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
12198 (vectorizable_bswap): Adjust.
12199 (vectorizable_call): Likewise.
12200 (vectorizable_conversion): Likewise.
12201 (vectorizable_assignment): Likewise.
12202 (vectorizable_shift): Likewise.
12203 (vectorizable_operation): Likewise.
12204 (vectorizable_store): Likewise.
12205 (vectorizable_condition): Likewise.
12206 (vectorizable_comparison): Likewise.
12208 2017-06-28 Michael Collison <michael.collison@arm.com>
12211 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
12213 (arm_gen_movmemqi): Removed unused variable 'i'.
12214 Convert 'for' loop into 'while' loop.
12215 (arm_expand_prologue): Remove last unnecessary set of insn.
12216 (thumb_pop): Remove unused variable 'pushed_words'.
12217 (thumb_exit): Remove last unnecessary set of regs_to_pop.
12219 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12221 * config/s390/predicates.md: Use s390_rel_address_ok_p.
12222 * config/s390/s390-protos.h: Add prototype of
12223 s390_rel_address_ok_p.
12224 * config/s390/s390.c (s390_got_symbol): New function.
12225 (s390_rel_address_ok_p): New function.
12226 (legitimize_pic_address): Use s390_rel_address_ok_p.
12227 (s390_load_got): Use s390_got_symbol.
12228 (s390_option_override): Issue error if
12229 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
12230 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
12232 * config/s390/s390.opt: New option mpic-data-is-text-relative.
12234 2017-06-27 Andrew Pinski <apinski@cavium.com>
12236 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
12237 (X * copysign (1.0, X)): New pattern.
12238 (X * copysign (1.0, -X)): New pattern.
12239 (copysign (-1.0, CST)): New pattern.
12241 2017-06-27 Joseph Myers <joseph@codesourcery.com>
12243 * genmultilib (combination_space): Remove variable.
12244 Validate reuse rules against regular expression for any sequence
12245 of multilib options in any order.
12247 2017-06-27 Michael Collison <michael.collison@arm.com>
12249 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
12250 call aarch64_split_simd_combine.
12251 * (aarch64_combine_internal<mode>): Delete pattern.
12252 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
12253 Allow register and subreg operands.
12255 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12257 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
12258 specific need, just fallback on defaults.
12259 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
12261 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12262 Olivier Hainque <hainque@adacore.com>
12264 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
12266 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
12267 targets. Pick a default if no particular attempt applied.
12268 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
12271 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12273 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
12274 (x86_64-wrs-vxworks7): Likewise.
12276 2017-06-27 Marek Polacek <polacek@redhat.com>
12279 * ubsan.c (instrument_null): Check get_base_address's result for null.
12281 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
12283 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
12285 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
12287 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
12288 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
12289 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
12290 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
12291 New function types.
12292 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
12293 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
12294 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
12295 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
12296 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
12297 BUILT_IN_FEUPDATEENV): New builtins.
12298 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
12299 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
12300 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
12301 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
12303 (builtin_structptr_types): Adjust size.
12304 * tree.c (builtin_structptr_types): Add four entries.
12306 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12307 Olivier Hainque <hainque@adacore.com>
12309 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
12310 (TLS_SYM): New local macro, forcing reference to __tls__ on
12311 link command lines for VxWorks 7 RTPs, triggering initialization
12313 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
12314 OS features TLS support, true for RTPs on VxWorks 7.
12315 * config/vxworks.c (vxworks_override_options): Setup emutls
12318 2017-06-27 Jakub Jelinek <jakub@redhat.com>
12320 * predict.c (test_prediction_value_range): Use -1U instead of -1
12321 to avoid narrowing conversion warning.
12322 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
12323 to avoid narrowing conversion warning.
12324 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
12326 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
12328 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12330 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
12331 64bit configurations.
12332 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
12333 (SIZE_TYPE): Likewise.
12334 * config/vxworks.c (vxworks_emutls_var_fields): Use
12335 long_unsigned_type_node instead of unsigned_type_node as the offset
12336 field type, which is "pointer" mode in emutls.c.
12338 2017-06-27 Jakub Jelinek <jakub@redhat.com>
12341 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
12343 PR middle-end/81207
12344 * gimple-fold.c (replace_call_with_call_and_fold): Handle
12345 gimple_vuse copying separately from gimple_vdef copying.
12347 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12349 * value-prof.c (free_hist): Remove call to memset and the enclosing if
12352 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
12353 Olivier Hainque <hainque@adacore.com>
12355 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
12356 for all vxworks7 targets.
12357 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
12358 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
12359 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
12360 variations for VX6/VX7 and 32/64bits later on in ...
12361 (VXWORKS_LIB_SPEC): Leverage new macros.
12362 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
12363 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
12365 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
12367 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
12368 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
12370 2017-06-26 Carl Love <cel@us.ibm.com>
12372 * config/rs6000/rs6000-c.c: Add support for built-in functions
12373 vector bool char vec_reve (vector bool char);
12374 vector signed char vec_reve (vector signed char);
12375 vector unsigned char vec_reve (vector unsigned char);
12376 vector bool int vec_reve (vector bool int);
12377 vector signed int vec_reve (vector signed int);
12378 vector unsigned int vec_reve (vector unsigned int);
12379 vector bool long long vec_reve (vector bool long long);
12380 vector signed long long vec_reve (vector signed long long);
12381 vector unsigned long long vec_reve (vector unsigned long long);
12382 vector bool short vec_reve (vector bool short);
12383 vector signed short vec_reve (vector signed short);
12384 vector double vec_reve (vector double);
12385 vector float vec_reve (vector float);
12386 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
12387 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
12388 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
12389 (altivec_vreve): New pattern.
12390 * config/rs6000/altivec.h (vec_reve): New define.
12391 * doc/extend.texi (vec_rev): Update the built-in documentation file
12392 for the new built-in functions.
12394 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12396 PR tree-optimization/71815
12397 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
12399 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
12401 (slsr_process_phi): Likewise.
12402 (replace_uncond_cands_and_profitable_phis): Don't replace a
12403 multiply candidate with a stride of 1 (copy or cast).
12404 (phi_incr_cost): Call uses_consumed_by_stmt rather than
12406 (lowest_cost_path): Likewise.
12407 (total_savings): Likewise.
12409 2017-06-26 Richard Biener <rguenther@suse.de>
12412 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
12413 Use def_builtin_pure for all gather builtins.
12415 2017-06-26 Richard Biener <rguenther@suse.de>
12417 PR tree-optimization/81203
12418 * tree-tailcall.c (find_tail_calls): Do not move stmts into
12419 non-dominating BBs.
12421 2017-06-26 Marek Polacek <polacek@redhat.com>
12424 * doc/invoke.texi: Document -Wmultistatement-macros.
12426 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
12428 * doc/sourcebuild.texi (ARM-specific attributes): Document new
12429 arm_neon_ok_no_float_abi effective target.
12431 2017-06-26 Richard Biener <rguenther@suse.de>
12433 PR tree-optimization/80928
12434 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
12435 (copy_bbs): Set BB_DUPLICATED flag early.
12436 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
12438 (execute_on_shrinking_pred): Likewise.
12439 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
12440 BB_DUPLICATED blocks.
12441 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
12442 iterate over all PHIs considering removal of *gsi.
12444 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
12446 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
12449 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12451 * config/rs6000/rs6000-string.c: (expand_block_clear,
12452 do_load_for_compare, select_block_compare_mode,
12453 compute_current_alignment, expand_block_compare,
12454 expand_strncmp_align_check, expand_strn_compare,
12455 expand_block_move, rs6000_output_load_multiple)
12456 Move functions related to string/block move/compare
12457 to a separate file.
12458 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
12459 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
12460 for this function which is now used in two files.
12461 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
12462 * config.gcc: Add rs6000-string.o to extra_objs for
12463 targets powerpc*-*-* and rs6000*-*-*.
12465 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
12468 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
12469 32-bit, since indexed is not valid for DImode.
12470 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
12471 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
12472 (define_peephole2 for Altivec d-form load): Add 32-bit support.
12473 (define_peephole2 for Altivec d-form store): Likewise.
12476 * multiple_target.c (create_dispatcher_calls): Only create the
12477 dispatcher call if the function is the default clone of a
12478 versioned function.
12480 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
12482 PR middle-end/80902
12483 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
12484 a call, force the call to not be a tail call.
12486 2017-06-23 Jeff Law <law@redhat.com>
12488 * doc/contrib.texi: Add entry for Steven Pemberton's work on
12491 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
12493 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
12494 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
12495 handling for early expansion of vector shifts (sl,sr,sra,rl).
12496 (builtin_function_type): Add vector shift right instructions
12497 to the unsigned argument list.
12499 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
12501 rtl-optimizatoin/79286
12502 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
12503 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
12504 trap. PIC register plus a const unspec without offset can never trap.
12506 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
12508 * tree.h (builtin_structptr_type): New type.
12509 (builtin_structptr_types): Declare new array.
12510 * tree.c (builtin_structptr_types): New array.
12511 (free_lang_data, build_common_tree_nodes): Use it.
12513 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
12516 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
12519 2017-06-22 Matt Turner <mattst88@gmail.com>
12521 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
12522 Lake models to skylake case. Assume skylake for unknown
12523 models with clflushopt.
12525 2017-06-22 Jeff Law <law@redhat.com>
12527 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
12528 frame sizes that do not satisfy aarch64_uimm12_shift.
12530 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
12532 * profile-count.h (apply_probability,
12533 apply_scale, probability_in): Fix checks for zero.
12535 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12537 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
12538 * doc/cppdiropts.texi (-I @var{dir}): Document it.
12540 2016-06-22 Richard Biener <rguenther@suse.de>
12542 * tree-vect-loop.c (vect_model_reduction_cost): Handle
12543 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
12544 REDUC_MAX_EXPR support.
12545 (vectorizable_reduction): Likewise.
12546 (vect_create_epilog_for_reduction): Likewise.
12548 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
12550 * match.pd (A / (1 << B) -> A >> B): New.
12551 * generic-match-head.c: Include optabs-tree.h.
12552 * gimple-match-head.c: Likewise.
12553 * optabs-tree.h (target_supports_op_p): New.
12554 * optabs-tree.c (target_supports_op_p): New.
12556 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12558 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
12559 $gcc_cv_ld --help output.
12560 (gcc_cv_ld_demangle): Likewise.
12561 (gcc_cv_ld_eh_frame_hdr): Likewise.
12562 (gcc_cv_ld_pie): Likewise.
12563 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
12564 (gcc_cv_ld_buildid): Likewise.
12565 (gcc_cv_ld_sysroot): Likewise.
12566 (ld_bndplt_support): Likewise.
12567 (ld_pushpopstate_support): Likewise.
12568 * configure: Regenerate.
12569 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
12571 2017-06-21 Jakub Jelinek <jakub@redhat.com>
12574 * config/i386/sse.md (round<mode>2): Renumber match_dup and
12575 operands indexes to avoid gap between operands and match_dups.
12577 2017-06-21 Andrew Pinski <apinski@cavium.com>
12579 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
12580 Increment Arith_shift and Arith_shift_reg by 1.
12581 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
12583 * config/aarch64/aarch64.c (thunderx_tunings): Enable
12584 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
12585 (aarch64_strip_extend): Add new argument and test for it.
12586 (aarch64_cheap_mult_shift_p): New function.
12587 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
12588 add a cost if it is true.
12589 Update calls to aarch64_strip_extend.
12590 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
12592 2017-06-21 Andrew Pinski <apinski@cavium.com>
12594 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
12596 (thunderxt88): Likewise.
12597 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
12598 (thunderx_prefetch_tune): New variable.
12599 (thunderx2t99_prefetch_tune): Update for the correct values.
12600 (thunderxt88_tunings): New variable.
12601 (thunderx_tunings): Use thunderx_prefetch_tune instead of
12602 generic_prefetch_tune.
12603 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
12605 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12607 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
12608 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
12609 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
12610 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
12611 (aarch64_atomic_cas<mode>, GPI): Likewise.
12613 2017-06-21 Martin Liska <mliska@suse.cz>
12615 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
12616 statements on cold and hot labels.
12617 * predict.c (tree_estimate_probability_bb): Remove the
12618 prediction from this place.
12620 2017-06-21 Martin Liska <mliska@suse.cz>
12622 PR tree-optimization/79489
12623 * gimplify.c (maybe_add_early_return_predict_stmt): New
12625 (gimplify_return_expr): Call the function.
12626 * predict.c (tree_estimate_probability_bb): Remove handling
12628 * predict.def: Update comment about early return predictor.
12629 * gimple-predict.h (is_gimple_predict): New function.
12630 * predict.def: Change default value of early return to 66.
12631 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
12633 * passes.def: Put pass_strip_predict_hints to the beginning of
12636 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
12638 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
12639 FUNCTION_DECL declarations.
12640 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
12642 (dwaf2out_decl): Likewise.
12643 * godump.c (go_early_global_decl): Skip call to the real debug hook
12644 for FUNCTION_DECL declarations.
12645 * passes.c (rest_of_decl_compilation): Skip call to the
12646 early_global_decl debug hook for FUNCTION_DECL declarations, unless
12647 -fdump-go-spec is passed.
12649 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
12651 * config/i386/i386.c (struct builtin_isa): New field pure_p.
12652 Reorder for compactness.
12653 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
12654 (def_builtin_pure, def_builtin_pure2): New functions.
12655 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
12657 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
12659 * match.pd (nop_convert): New predicate.
12660 ((A +- CST1) +- CST2): Allow some NOP conversions.
12662 2017-06-21 Jakub Jelinek <jakub@redhat.com>
12665 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
12666 with ctors/dtors if GOVD_SHARED is set.
12668 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12670 * config/aarch64/aarch64.md (movti_aarch64):
12671 Emit mov rather than orr.
12672 (movtf_aarch64): Likewise.
12673 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
12674 Emit mov rather than orr.
12676 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12678 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
12679 Swap alternatives, make integer dup more expensive.
12681 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12683 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
12684 Return true for non-tls symbols.
12686 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
12688 * config/aarch64/aarch64-cores.def (cortex-a55): New.
12689 (cortex-a75): Likewise.
12690 (cortex-a75.cortex-a55): Likewise.
12691 * config/aarch64/aarch64-tune.md: Regenerate.
12692 * doc/invoke.texi (-mtune): Document new values for -mtune.
12694 2017-06-21 Tom de Vries <tom@codesourcery.com>
12696 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
12697 stack_size feature.
12698 (Effective-Target Keywords, Other attributes): Suggest using
12699 dg-add-options stack_size feature to get stack limit in stack_size
12700 effective target documentation.
12702 2017-06-21 Julian Brown <julian@codesourcery.com>
12703 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12705 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
12706 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
12707 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
12709 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
12710 attribute type list for neon_multiply.
12711 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
12712 attribute type list for neon_multiply.
12713 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
12714 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
12715 attribute type list for neon_multiply.
12716 * config/arm/types.md (crypto_pmull): Add.
12717 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
12718 attribute type list.
12720 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
12722 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
12725 2017-06-20 Jakub Jelinek <jakub@redhat.com>
12727 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
12728 to make sure not to dereference a NULL cost_classes_ptr pointer.
12730 2017-06-20 Carl Love <cel@us.ibm.com>
12732 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12733 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12734 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12735 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12736 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12737 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
12738 VMULOSW): New enum "unspec" values.
12739 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
12740 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
12741 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
12742 altivec_vmulosw): New patterns.
12743 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12744 VMULOSW): Add definitions.
12746 2017-06-20 Julia Koval <julia.koval@intel.com>
12748 * config/i386/i386.c: Fix rounding expand for new pattern.
12749 * config/i386/subst.md: Fix pattern (parallel -> unspec).
12751 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12753 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
12754 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
12756 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12758 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
12761 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12763 * config/aarch64/aarch64-cores.def: Rearrange to sort by
12764 architecture, then by implementer ID.
12765 * config/aarch64/aarch64-tune.md: Regenerate.
12767 2017-06-20 Richard Biener <rguenther@suse.de>
12769 PR middle-end/81097
12770 * fold-const.c (split_tree): Fold to type before negating.
12772 2017-06-20 David Malcolm <dmalcolm@redhat.com>
12774 * diagnostic-show-locus.c
12775 (selftest::test_fixit_deletion_affecting_newline): New function.
12776 (selftest::diagnostic_show_locus_c_tests): Call it.
12778 2017-06-20 Andreas Schwab <schwab@suse.de>
12781 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
12784 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
12786 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
12787 __ARM_FEATURE_COPROC according to support.
12789 2017-06-20 Jakub Jelinek <jakub@redhat.com>
12791 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
12792 Rewritten to avoid overflow for > 32-bit pointers.
12795 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
12796 by removing enum keyword.
12797 (ubsan_type_descriptor): Likewise. Formatting fix.
12800 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
12801 splitter): Require TARGET_SSE2 in the condition.
12803 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
12806 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
12807 for doing vector set of SFmode on ISA 3.0.
12808 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
12809 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
12811 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
12812 SFmode value into a V4SF variable that was extracted from another
12813 V4SF variable without converting the element to double precision
12814 and back to single precision vector format.
12815 (vsx_insert_extract_v4sf_p9_2): Likewise.
12817 2017-06-19 Jakub Jelinek <jakub@redhat.com>
12819 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
12820 in UWHI to avoid undefined overflow.
12823 * ubsan.h (enum ubsan_encode_value_phase): New.
12824 (ubsan_encode_value): Change second argument to
12825 enum ubsan_encode_value_phase with default value of
12826 UBSAN_ENCODE_VALUE_GENERIC.
12827 * ubsan.c (ubsan_encode_value): Change second argument to
12828 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
12829 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
12830 create_tmp_var_raw instead of create_tmp_var and use a
12832 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
12833 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
12834 ubsan_encode_value callers.
12837 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
12838 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
12839 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
12841 2017-06-19 Richard Biener <rguenther@suse.de>
12843 PR middle-end/81118
12844 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
12845 estimates if we changed anything.
12847 2017-06-19 Richard Biener <rguenther@suse.de>
12849 PR tree-optimization/80887
12850 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
12851 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
12852 simplified lookups, then reset mprts_hook.
12853 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
12855 (try_to_simplify): Likewise.
12857 2017-06-19 Martin Liska <mliska@suse.cz>
12860 * gimplify.c (gimplify_switch_expr):
12861 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
12863 2017-06-19 Martin Liska <mliska@suse.cz>
12865 * doc/install.texi: Document that PGO runs in 4 stages.
12867 2017-06-19 Martin Liska <mliska@suse.cz>
12870 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
12871 to dispatcher function name.
12872 * multiple_target.c (replace_function_decl): New function.
12873 (create_dispatcher_calls): Redirect both edges and references.
12875 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
12877 * profile-count.c (profile_count::dump): Dump quality.
12878 (profile_count::differs_from_p): Update for unsigned val.
12879 * profile-count.h (profile_count_quality): New enum.
12880 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
12882 2017-06-19 Richard Biener <rguenther@suse.de>
12884 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
12885 struct function as arg.
12886 (estimate_numbers_of_iterations): Export overload with loop arg.
12887 (free_numbers_of_iterations_estimates_loop): Use an overload of
12888 free_numbers_of_iterations_estimates instead.
12889 * tree-cfg.c (remove_bb): Adjust.
12890 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
12891 * tree-parloops.c (gen_parallel_loop): Likewise.
12892 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
12894 (tree_unroll_loops_completely): Likewise.
12895 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
12896 Use an overload instead and export.
12897 (estimated_loop_iterations): Adjust.
12898 (max_loop_iterations): Likewise.
12899 (likely_max_loop_iterations): Likewise.
12900 (estimate_numbers_of_iterations): Take struct function as arg
12902 (loop_exits_before_overflow): Adjust.
12903 (free_numbers_of_iterations_estimates_loop): Use an overload.
12904 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
12905 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
12907 2017-06-19 Richard Biener <rguenther@suse.de>
12910 * ipa-prop.c (find_constructor_constant_at_offset): Handle
12911 RANGE_EXPR conservatively.
12913 2017-06-16 Carl Love <cel@us.ibm.com>
12915 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12916 definitions for vec_float, vec_float2, vec_floato,
12917 vec_floate built-ins.
12918 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
12919 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
12921 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
12922 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
12923 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
12924 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
12925 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
12926 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
12927 vec_floato): Add builtin defines.
12928 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
12929 Update the built-in documentation file for the new built-in
12932 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12934 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
12935 (mthumb): Mark as the negative of -marm.
12937 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12939 * doc/invoke.texi (ARM Options, -mcpu): Document supported
12941 (ARM Options, -mtune): Document that this accepts the same
12942 extension options as -mcpu.
12943 (ARM Options, -mfpu): Document addition of -mfpu=auto.
12945 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12947 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
12948 permitted extensions.
12950 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12952 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
12953 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
12954 (armv8-m.main): Add option +nodsp.
12955 * config/arm/arm-cpu-cdata.h: Regenerated.
12957 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12959 * config/arm/t-fuchsia: New file.
12960 * config.gcc (arm*-*-fuchsia*): Use it.
12962 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12964 * config/arm/t-symbian: Rewrite for new option infrastructure.
12966 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12968 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
12969 (MULTILIB_REQUIRED): Likewise.
12971 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12973 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
12974 (MULTILIB_RESUE): Likewise.
12975 (MULTILIB_MATCHES): Likewise.
12976 (MULTLIB_REQUIRED): Likewise.
12978 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12980 * config/arm/t-rtems: Rewrite for new option framework.
12982 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12984 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
12985 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
12986 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
12987 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
12988 * config/arm/t-multilib: ... here.
12989 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
12990 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
12991 armv7-a and armv8*-a when A-profile libraries have not been built.
12992 * config/arm/t-rmprofile: Rewrite.
12994 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12996 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
12997 with a backslash. Remove the backslash after substituting unescaped
12999 * doc/fragments.texi (MULTILIB_REUSE): Document it.
13001 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13003 * config.gcc: (arm*-*-*): When building a-profile libraries, force
13004 the driver to pass through the default setting of -mfloat-abi.
13005 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
13007 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
13008 (all_feat_combs): New rule.
13009 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
13011 * config/arm/t-aprofile: Rewrite.
13013 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13015 * config/arm/arm.h (FPUTYPE_AUTO): Define.
13016 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
13017 fpu is not specified by the user/command-line.
13018 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
13019 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
13020 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
13021 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
13022 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
13023 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
13025 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13027 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
13028 * config/arm/t-arm-elf: Rewritten.
13030 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13032 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
13033 have some floating-point instructions.
13034 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
13035 (TARGET_MAYBE_HARD_FLOAT): New macro.
13036 * config/arm/arm-builtins.c (arm_init_builtins): Use
13037 TARGET_MAYBE_HARD_FLOAT.
13038 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
13040 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13042 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
13043 (configargs.h): Include it.
13044 (arm_print_hint_for_fpu_option): New function.
13045 (arm_parse_fpu_option): New function.
13046 (candidate_extension): New class.
13047 (arm_canon_for_multilib): New function.
13048 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
13049 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
13050 (ARCH_CANONICAL_SPECS): New macro.
13051 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
13053 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13055 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
13056 are set after handling multilib fragments. Set target_cpu_default2
13059 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13061 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
13063 (arm*-*-*): Set target_cpu_default2 to a quoted string.
13064 * config/arm/parsecpu.awk (check_cpu): Validate any extension
13066 (check_arch): Likewise.
13067 * config/arm/arm.c (arm_configure_build_target): Handle
13068 TARGET_CPU_DEFAULT being a string constant. Scan any feature
13069 options in the default.
13071 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13073 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
13074 when an option is an alias of another.
13075 * config/arm/parsecpu.awk (optalias): New parser token.
13076 (gen_comm_data): Mark non-alias options as such. Emit entries
13077 for extension aliases.
13078 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
13079 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
13080 (armv6kz, armv6zk, armv6t2): Likewise.
13081 (armv7): Make vfpv3-d16 an alias.
13082 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
13084 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
13085 Sort in canonical order.
13086 (armv8-a): Sort in canonical order.
13087 (armv8.1-a, armv8.2-a): Likewise.
13088 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
13090 (cortex-a9): Sort in canonical order.
13091 * config/arm/arm.c (selftests.h): Include it.
13092 (arm_test_cpu_arch_data): New function.
13093 (arm_run_self_tests): New function.
13094 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
13095 (targetm): Move declaration to the end of the file.
13096 * arm-cpu-cdata.h: Regenerated.
13098 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13100 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
13101 call to target_mode_check describing the type of option passed.
13102 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
13103 (arm_target_thumb_only): Use arm_parse_arch_option_name or
13104 arm_parse_cpu_option_name to match parameters against list of
13106 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
13107 arm_arch_core_flags data structure.
13108 * config/arm/arm-cpu_cdata.h: Regenerated.
13110 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13112 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
13114 (arm_print_hint_for_cpu_option): Likewise.
13115 (arm_print_hint_for_arch_option): Likewise.
13116 (arm_parse_cpu_option_name): Likewise.
13117 (arm_parse_arch_option_name): Likewise.
13118 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
13119 of entries in the all_fpus list.
13120 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
13121 (arm_parse_cpu_option_name): Declare.
13122 (arm_parse_arch_option_name): Declare.
13123 (arm_parse_option_features): Declare.
13124 (arm_intialize_isa): Declare.
13125 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
13127 (gen_comm_data): ... here. Make definitions non-static.
13128 * config/arm/arm-cpu-data.h: Regenerated.
13129 * config/arm/arm-cpu-cdata.h: Regenerated.
13131 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13133 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
13134 (cpu_arch_extension): New structure.
13135 (cpu_arch_option, arch_option, cpu_option): New structures.
13136 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
13137 architecture types.
13138 (gen_data): Generate new format data tables.
13139 * config/arm/arm.c (cpu_tune): New structure.
13140 (cpu_option, processors): Delete.
13141 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
13142 (arm_print_hint_for_cpu_option): ... this and ...
13143 (arm_print_hint_for_arch_option): ... this.
13144 (arm_parse_arch_cpu_name): Delete. Replace with ...
13145 (arm_parse_cpu_option_name): ... this and ...
13146 (arm_parse_arch_option_name): ... this.
13147 (arm_unrecognized_feature): Change type of target parameter to
13149 (arm_parse_arch_cpu_features): Delete. Replace with ...
13150 (arm_parse_option_features): ... this.
13151 (arm_configure_build_target): Rework to use new configuration data
13153 (arm_print_tune_info): Rework for new configuration data tables.
13154 * config/arm/arm-cpu-data.h: Regenerated.
13155 * config/arm/arm-cpu.h: Regenerated.
13157 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13159 * Makefile.in (OBJS): Move sbitmap.o from here ...
13160 (OBJS-libcommon): ... to here.
13162 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13164 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
13165 (ISA_ALL_CRYPTO): New macro.
13166 (ISA_ALL_SIMD): New macro
13167 (ISA_ALL_FP): New macro.
13168 * config/arm/arm.c (fpu_bitlist): Update initializer.
13169 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
13171 (arm9e): Add fpu. Add option for nofp
13172 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
13173 (arm926ej-s, arm1026ej-s): Likewise.
13174 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
13175 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
13176 neon-fp16, neon-vfpv4, nofp and nosimd.
13177 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
13178 (cortex-a8): Add fpu. Add option for nofp.
13179 (cortex-a9): Add fpu. Add options for nosimd and nofp.
13180 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
13181 (cortex-r4f): Add fpu.
13182 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
13183 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
13185 (cortex-r8): Likewise.
13186 (cortex-m4): Add fpu. Add option for nofp.
13187 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
13188 (cortex-a17.cortex-a7): Likewise.
13189 (cortex-a32): Add fpu. Add options for crypto and nofp.
13190 (cortex-a35, cortex-a53): Likewise.
13191 (cortex-a57): Add fpu. Add option for crypto.
13192 (cortex-a72, cortex-a73): Likewise.
13193 (exynos-m1): Likewise.
13194 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
13195 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
13196 (cortex-m33): Add fpu. Add option for nofp.
13197 * config/arm/arm-cpu-cdata.h: Regenerated
13198 * config/arm/arm-cpu-data.h: Regenerated.
13200 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13202 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
13203 (armv5te, armv5tej): Likewise.
13204 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
13205 (armv7): Add options fp and vfpv3-d16.
13206 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
13207 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
13209 (armv7ve): Likewise.
13210 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
13211 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
13212 (armv8-a): Add nocrypto option.
13213 (armv8.1-a, armv8.2-a): Likewise.
13214 (armv8-m.main): add options fp, fp.dp and nofp.
13216 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13218 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
13220 (armv8-a+crc): Delete.
13221 (armv8.1-a): Add options simd, crypto and nofp.
13222 (armv8.2-a): Add options fp16, simd, crypto and nofp.
13223 (armv8.2-a+fp16): Delete.
13224 (armv8-m.main): Add option dsp.
13225 (armv8-m.main+dsp): Delete.
13226 (cortex-a8): Add fpu. Add nofp option.
13227 (cortex-a9): Add fpu. Add nofp and nosimd options.
13228 * config/arm/parsecpu.awk (gen_data): Generate option tables and
13229 link to main cpu and architecture data structures.
13230 (gen_comm_data): Only put isa attributes from the main architecture
13232 (option): New statement for architecture and CPU entries.
13233 * arm.c (struct cpu_option): New structure.
13234 (struct processors): Add entry for options.
13235 (arm_unrecognized_feature): New function.
13236 (arm_parse_arch_cpu_name): Ignore any characters after the first
13238 (arm_parse_arch_cpu_feature): New function.
13239 (arm_configure_build_target): Separate out any CPU and architecture
13240 features and parse separately. Don't error out if -mfpu=auto is
13241 used with only an architecture string.
13242 (arm_print_asm_arch_directives): New function.
13243 (arm_file_start): Call it.
13244 * config/arm/arm-cpu-cdata.h: Regenerated.
13245 * config/arm/arm-cpu-data.h: Likewise.
13246 * config/arm/arm-tables.opt: Likewise.
13248 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13250 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
13251 assembler when it is not -mfpu=auto.
13253 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13255 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
13256 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
13257 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
13258 (ASM_CPU_SPEC): Rewrite.
13259 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
13260 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
13261 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
13262 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
13263 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
13264 copied string is NUL-terminated. Also strip any characters prefixed
13266 (arm_rewrite_selected_arch): New function.
13267 (arm_rewrite_march): New function.
13269 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13271 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
13272 (x_arm_cpu_string, x_arm_tune_string): Likewise.
13273 (march, mcpu, mtune): Convert to string-based options.
13274 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
13275 (arm_parse_arch_cpu_name): New function.
13276 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
13277 identify selected architecture or CPU.
13278 (arm_option_save): New function.
13279 (TARGET_OPTION_SAVE): Redefine.
13280 (arm_option_restore): Restore string options.
13281 (arm_option_print): Print string options.
13283 2017-06-16 Martin Sebor <msebor@redhat.com>
13285 PR tree-optimization/80933
13286 PR tree-optimization/80934
13287 * builtins.c (fold_builtin_3): Do not handle bcmp here.
13288 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
13289 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
13290 (gimple_fold_builtin): Call them.
13292 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13294 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
13295 as unlikely; update profile.
13297 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13299 * predict.c (force_edge_cold): Handle declaring edges impossible
13302 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13304 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
13306 (try_unroll_loop_completely): Fix reporting.
13308 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13310 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
13312 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
13315 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
13316 if given a non-constant argument for an intrinsic which requires a
13319 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13321 * profile.c (compare_freqs): New function.
13322 (branch_prob): Sort edge list.
13323 (find_spanning_tree): Assume that the list is priority sorted.
13325 2017-06-16 Richard Biener <rguenther@suse.de>
13327 PR tree-optimization/81090
13328 * passes.def (pass_record_bounds): Remove.
13329 * tree-pass.h (make_pass_record_bounds): Likewise.
13330 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
13331 make_pass_record_bounds): Likewise.
13332 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
13333 not free niter estimates at the beginning but at the end.
13334 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
13336 2017-06-16 Richard Biener <rguenther@suse.de>
13338 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
13339 initializer to workaround ICE in host GCC 4.8.
13341 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13343 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
13345 (clone_inlined_nodes): Update.
13347 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13349 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
13350 prefetch settings, and enable prefetching by default at -O3.
13352 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13354 * config/aarch64/aarch64.c (aarch64_override_options_internal):
13355 Set flag_prefetch_loop_arrays according to tuning data.
13357 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13359 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
13360 New tune structure.
13361 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
13362 [Unrelated to main purpose of the patch] Place the pointer field last
13363 to enable type checking errors when tune structure are wrongly merged.
13364 * config/aarch64/aarch64.c (generic_prefetch_tune,)
13365 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
13366 (thunderx2t99_prefetch_tune): New tune constants.
13367 (tune_params *_tunings): Update all tunings (no functional change).
13368 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
13369 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
13370 from tunings structures.
13372 2017-06-16 Jakub Jelinek <jakub@redhat.com>
13375 * ubsan.c (instrument_null): Add T argument, use it instead
13376 of computing it based on IS_LHS.
13377 (instrument_object_size): Likewise.
13378 (pass_ubsan::execute): Adjust instrument_null and
13379 instrument_object_size callers to pass gimple_get_lhs or
13380 gimple_assign_rhs1 result to it. Use instrument_null instead of
13381 calling get_base_address and instrument_mem_ref. Handle
13382 aggregate call arguments for object-size sanitization.
13384 2017-06-16 Yury Gribov <tetra2005@gmail.com>
13386 PR tree-optimization/81089
13387 * tree-vrp.c (is_masked_range_test): Validate operands of
13390 2017-06-15 Martin Sebor <msebor@redhat.com>
13393 * dumpfile.c (dump_register): Avoid calling memset to initialize
13394 a class with a default ctor.
13395 * gcc.c (struct compiler): Remove const qualification.
13396 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
13397 * hash-table.h: Ditto.
13398 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
13400 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
13401 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
13403 * params.h (struct param_info): Make struct members non-const.
13404 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
13405 with default initialization.
13406 * vec.h (vec_copy_construct, vec_default_construct): New helper
13408 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
13409 with vec_copy_construct.
13410 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
13411 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
13412 * doc/invoke.texi (-Wclass-memaccess): Document.
13414 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13416 * emit-rtl.h (is_leaf): Update comment about local
13417 register allocator.
13419 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13422 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
13423 for a variable to have a section before checking if the section has a
13425 Set section to.persistent if persistent attribute is set.
13426 Warn if .persistent attribute is used on an automatic variable.
13428 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
13430 PR rtl-optimization/80474
13431 * reorg.c (update_block): Do not ignore instructions in a delay slot.
13433 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
13435 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
13438 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
13440 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
13441 (casesi): Emit bounds checking as RTL.
13442 (casesi_internal_mips16_<mode>): Remove bounds checking.
13444 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
13446 * config/xtensa/xtensa.c (xtensa_option_override): Append
13447 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
13448 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
13449 xtensa_doloop_hooks): Define unconditionally.
13450 (xtensa_reorg_loops): Only call reorg_loops in the presence of
13452 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
13453 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
13454 for it in xtensa_option_override.
13455 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
13456 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
13458 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
13460 * doc/cppopts.texi: Document '-' special value to -MF.
13462 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
13464 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
13465 (cortex_a53_fconst): Likewise.
13466 (cortex_a53_fpmul): Likewise.
13467 (cortex_a53_f_load_64): Likewise.
13468 (cortex_a53_f_load_many): Likewise.
13469 (cortex_a53_advsimd_alu): Likewise.
13470 (cortex_a53_advsimd_alu_q): Likewise.
13471 (cortex_a53_advsimd_mul): Likewise.
13472 (cortex_a53_advsimd_mul_q): Likewise.
13473 (fpmac bypass): Add new bypass for fpmac-fpmac case.
13474 Add missing fmul, r2f_cvt and fconst cases.
13476 2017-06-14 Richard Biener <rguenther@suse.de>
13478 PR middle-end/81088
13479 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
13481 (fold_binary_loc): When associating do not treat pre-existing
13482 TREE_OVERFLOW on literal constants as a reason to allow
13483 TREE_OVERFLOW on associated literal constants.
13485 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13487 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
13488 (MASK_FEATURES): New macro.
13489 * config/sparc/sparc.c (sparc_option_override): Remove the special
13490 handling of -mfpu and generalize it to all MASK_FEATURES switches.
13492 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13494 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
13495 a division of 0 if non-call exceptions are enabled.
13497 2017-06-14 Andrew Pinski <apinski@cavium.com>
13498 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13501 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
13502 Improve vector initialization code gen for only variable case.
13504 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13506 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
13508 2017-06-14 Richard Biener <rguenther@suse.de>
13510 PR tree-optimization/81083
13511 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
13514 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13516 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
13517 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
13518 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
13519 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
13520 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
13521 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
13523 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13525 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
13526 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
13528 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13530 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
13532 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13534 * config/rs6000/t-rtems: Don't handle SPE.
13536 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13538 * config/rs6000/t-linux: Don't handle SPE.
13540 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13542 * config/rs6000/eabispe.h: Delete file.
13544 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13546 * config/rs6000/t-spe: Delete file.
13548 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13550 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
13551 (rs6000_legitimate_offset_address_p): Return false for anything in
13552 V2SImode or V2SFmode.
13554 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13556 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
13557 except V2SF and V2SI. Rearrange the vector modes, and add comments.
13558 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
13560 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
13561 (rs6000_legitimate_offset_address_p): Ditto.
13562 (rs6000_emit_move): Ditto.
13563 (rs6000_init_builtins): Remove V4HI_type_node.
13565 2017-06-13 Martin Liska <mliska@suse.cz>
13568 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
13569 (gate_asan): Likewise.
13570 * asan.h (asan_no_sanitize_address_p): Remove the function.
13571 (sanitize_flags_p): New function.
13572 * builtins.def: Fix coding style.
13573 * common.opt: Use renamed enum value.
13574 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
13575 * doc/extend.texi: Document no_sanitize attribute.
13576 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
13577 to SANITIZE_UNDEFINED_NONDEFAULT.
13578 * gcc.c (sanitize_spec_function): Use the renamed enum value.
13579 * gimple-fold.c (optimize_atomic_compare_exchange_p):
13580 Use sanitize_flags_p.
13581 * gimplify.c (gimplify_function_tree): Likewise.
13582 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
13583 * opts.c (parse_no_sanitize_attribute): New function.
13584 (common_handle_option): Use renamed enum value.
13585 * opts.h (parse_no_sanitize_attribute): Declare.
13586 * tree.c (sanitize_flags_p): New function.
13587 * tree.h: Declared here.
13588 * tsan.c: Use sanitize_flags_p.
13589 * ubsan.c (ubsan_expand_null_ifn): Likewise.
13590 (instrument_mem_ref): Likewise.
13591 (instrument_bool_enum_load): Likewise.
13592 (do_ubsan_in_current_function): Remove the function.
13593 (pass_ubsan::execute): Use sanitize_flags_p.
13594 * ubsan.h: Remove do_ubsan_in_current_function
13595 * tree-cfg.c (print_no_sanitize_attr_value): New function.
13596 (dump_function_to_file): Use it here.
13598 2017-06-13 Martin Jambor <mjambor@suse.cz>
13600 PR tree-optimization/80803
13601 PR tree-optimization/81063
13602 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
13603 (propagate_subaccesses_across_link): Enqueue subtree whenever
13604 necessary instead of relying on the caller.
13606 2017-06-13 Martin Jambor <mjambor@suse.cz>
13608 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
13609 that have a first_link.
13610 (sort_and_splice_var_accesses): Do not check first_link before
13612 (subtree_mark_written_and_enqueue): Likewise.
13613 (propagate_all_subaccesses): Likewise and do not stop at first
13614 parent with a first_link.
13616 2017-06-13 Martin Jambor <mjambor@suse.cz>
13618 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
13621 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13623 * match.pd: New pattern.
13625 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13627 * tree-vrp.c (is_masked_range_test): New function.
13628 (register_edge_assert_for): Determine ranges for
13631 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13633 PR tree-optimization/67328
13634 * fold-const.c (maskable_range_p): New function.
13635 (build_range_check): Generate bittests if possible.
13637 2017-06-13 Martin Liska <mliska@suse.cz>
13639 * gimple-pretty-print.c (dump_probability): Add new argument.
13640 (dump_edge_probability): Dump both probability and count.
13641 (dump_gimple_label): Likewise.
13642 (dump_gimple_bb_header): Likewise.
13644 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
13647 * config/avr/avr-devices.c: Fix indentation.
13648 * config/avr/gen-avr-mmcu-specs.c: Dito.
13650 2017-06-13 Richard Biener <rguenther@suse.de>
13652 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
13653 instead get vector type from stmt_info.
13654 (vectorizable_reduction): Adjust. Remove dead code.
13656 2017-06-13 Richard Biener <rguenther@suse.de>
13658 PR middle-end/81065
13659 * fold-const.c (extract_muldiv_1): Remove bogus distribution
13660 case of C * (x * C2 + C3).
13661 (fold_addr_of_array_ref_difference): Properly fold index difference.
13663 2017-06-12 David S. Miller <davem@davemloft.net>
13666 * config/sparc/sparc.md (return expander): Emit frame blockage if
13667 function uses alloca.
13669 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
13671 * combine.c (make_field_assignment): Check len rather than the mode
13672 precision when calling force_to_mode.
13674 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
13676 Support multilibs and devices that see flash in RAM address range.
13679 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
13680 (avr_mcu_t) <flash_pm_offset>: New field.
13681 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
13682 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
13683 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
13684 (AVR_TINY_PM_OFFSET): Remove macro.
13685 * config/avr/avr.opt (-mshort-calls): New option.
13686 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
13687 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
13688 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
13689 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
13690 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
13691 instead of avr_arch->have_jmp_call.
13692 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
13693 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
13694 avr_arch->flash_pm_offset to define.
13695 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
13696 new field flash_pm_offset. Add entry for avrxmega3.
13697 (avr_texinfo): Add entry for avrxmega3.
13698 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
13699 attiny212, attiny214,
13700 attiny412, attiny414, attiny416, attiny417,
13701 attiny814, attiny816, attiny817,
13702 attiny1614, attiny1616, attiny1617,
13703 attiny3214, attiny3216, attiny3217.
13704 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
13705 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
13706 (avr_print_operand_address) [AVR_TINY]: Same.
13707 (avr_asm_init_sections) <readonly_data_section>: Only patch
13708 callback if avr_arch->flash_pm_offset = 0.
13709 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
13710 for rodata if avr_arch->flash_pm_offset != 0.
13711 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
13712 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
13713 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
13714 (m_options): Append opt_rcall.
13715 (m_dirnames): Append dir_rcall.
13716 * config/avr/t-multilib: Regenerate.
13718 * configure.ac [target=avr]: Check whether avrxmega3 default
13719 linker description file works as needed.
13720 * configure: Regenerate.
13721 * doc/avr-mmcu.texi: Regenerate.
13722 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
13723 <__AVR_ARCH__>: Document avrxmega3 and 103.
13724 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
13725 <__AVR_SHORT_CALLS__>: Document it.
13726 <__AVR_PM_BASE_ADDRESS__>: Document it.
13727 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
13728 (AVR Variable Attributes) <progmem>: Document this is
13729 not needed for avrxmega3.
13730 (AVR Named Address Spaces) <__flash>: Dito.
13732 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
13734 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
13736 2017-06-12 Doug Rupp <rupp@adacore.com>
13738 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
13739 Append vxworks-stdint.h to the tm_file list.
13740 * config/vxworks-stdint.h: New file.
13742 2017-06-12 Martin Liska <mliska@suse.cz>
13744 PR tree-optimization/81041
13745 * tree-profile.c (gimple_gen_ic_func_profiler):
13746 Create an extra BB in profile-generate
13747 (gimple_gen_time_profiler): Likewise.
13749 2017-06-12 Jakub Jelinek <jakub@redhat.com>
13751 PR tree-optimization/81003
13752 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
13753 (update_range_test): Use it instead of force_gimple_operand_gsi.
13755 2017-06-12 Richard Biener <rguenther@suse.de>
13757 PR tree-optimization/81053
13758 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
13759 with backedge value not defined in loop. Simplify def stmt
13762 2017-06-11 Tom de Vries <tom@codesourcery.com>
13765 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
13767 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
13768 nvptx_cannot_force_const_mem.
13770 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13772 * opts.c (finish_options): Move test for flag_split_stack after
13773 it has been initialized.
13775 2017-06-11 Jason Merrill <jason@redhat.com>
13777 * tree.h (id_equal): New.
13778 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
13779 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
13780 instead of strcmp of IDENTIFIER_POINTER.
13782 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13784 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
13785 (mark_all_inlined_calls_cdtor): Fix formating.
13786 (inline_transform): Rescale profile before inlining.
13788 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13790 * cgraph.h (cgraph_edge::clone): Update prototype.
13791 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
13792 (cgraph_node::create_clone): Update.
13793 (cgraph_node::create_version_clone): Update.
13794 * tree-inline.c (copy_bb): Update.
13795 (expand_call_inline): Update.
13797 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
13799 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
13800 factored out from ...
13801 (rs6000_emit_prologue): ... here.
13803 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
13805 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
13806 factored out from ...
13807 (rs6000_emit_prologue): ... here.
13809 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13811 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
13813 (handle_missing_profiles): Fix computation of tp_first_run.
13814 (counts_to_freqs): Do not touch freqs when count is 0.
13816 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13818 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
13821 2017-06-10 Tom de Vries <tom@codesourcery.com>
13823 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
13824 attributes): Document signal effective target.
13826 2017-06-10 Tom de Vries <tom@codesourcery.com>
13828 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
13829 Document effective target stack_size.
13831 2017-06-09 David Malcolm <dmalcolm@redhat.com>
13833 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
13834 to the edit_context if they can be auto-applied.
13836 2017-06-9 Ian Lance Taylor <iant@golang.org>
13838 * opts.c (finish_options): If -fsplit-stack, disable implicit
13839 -forder-blocks-and-partition.
13840 * doc/invoke.texi (Optimize Options): Document that when using
13841 -fsplit-stack -forder-blocks-and-partition is not implicitly
13844 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13846 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
13847 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
13848 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
13849 * builtins.def (abort, trap, unreachable): Declare cold.
13850 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
13851 * tree-core.h (ECF_COLD): New.
13852 * tree.c (set_call_expr_flags): Handle ECF_COLD.
13853 (build_common_builtin_nodes): Mark unreachable and abort as cold.
13855 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13857 * predict.c (unlikely_executed_stmt_p): Cleanup.
13859 2017-06-09 Richard Biener <rguenther@suse.de>
13861 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
13862 model if the ref is always written to.
13864 2017-06-09 Tamar Christina <tamar.christina@arm.com>
13866 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
13868 2017-06-09 Tamar Christina <tamar.christina@arm.com>
13870 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
13873 2017-06-09 Tom de Vries <tom@codesourcery.com>
13876 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
13877 "target cannot support label values" when encountering LABEL_REF.
13879 2017-06-09 Martin Liska <mliska@suse.cz>
13881 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
13882 (gimple_gen_ic_func_profiler): Emit direct comparison
13883 of __gcov_indirect_call_callee with NULL.
13884 (gimple_gen_time_profiler): Change probability from
13885 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
13887 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13889 * profile.c (edge_gcov_counts): Turn to pointer.
13890 (compute_branch_probabilities, compute_branch_probabilities): Update.
13891 (branch_prob): Do not clear edge_gcov_count.
13892 * profile.h (edge_gcov_counts): Turn to pointer.
13893 (edge_gcov_count): Update.
13895 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13897 * gimple.h (gimple_check_failed): Mark cold.
13899 2017-06-09 Richard Biener <rguenther@suse.de>
13901 PR tree-optimization/66623
13902 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
13903 refactor check_reduction into two parts, properly computing
13904 whether we have to check reduction validity for outer loop
13907 2017-06-09 Richard Biener <rguenther@suse.de>
13909 PR tree-optimization/79483
13910 * graphite-scop-detection.c (order): New global.
13911 (get_order): Compute bb to order mapping that satisfies code
13912 generation constraints.
13913 (cmp_pbbs): New helper.
13914 (build_scops): Start domwalk at entry block, sort generated
13917 2017-06-09 Richard Biener <rguenther@suse.de>
13919 PR middle-end/81007
13920 * ipa-polymorphic-call.c
13921 (ipa_polymorphic_call_context::restrict_to_inner_class):
13922 Skip FIELD_DECLs with error_mark_node type.
13923 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
13926 2017-06-09 Martin Liska <mliska@suse.cz>
13928 * predict.c (struct branch_predictor): New struct.
13929 (test_prediction_value_range): New test.
13930 (predict_c_tests): New function.
13931 * selftest-run-tests.c (selftest::run_tests): Run the function.
13932 * selftest.h: Declare new tests.
13934 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
13937 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
13938 gen_add3_insn did not fail.
13939 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
13940 r0, construct that number in a temporary reg and add that reg to r0.
13941 If asked to put the result in r0 as well, fail.
13943 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
13945 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13946 for early expansion of vec_eqv.
13948 2017-06-08 Jakub Jelinek <jakub@redhat.com>
13950 PR middle-end/81005
13951 * ubsan.c (instrument_null): Avoid pointless code temporary.
13952 (pass_ubsan::execute): Instrument aggregate arguments of calls.
13954 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
13958 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
13961 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
13964 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13966 * predict.c (unlikely_executed_edge_p): Move ahead.
13967 (probably_never_executed_edge_p): Use it.
13969 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
13971 PR middle-end/79988
13972 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
13973 gimple_call_builtin_p call.
13975 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13977 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
13978 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
13979 rtl_check_failed_type2, rtl_check_failed_code1,
13980 rtl_check_failed_code2, rtl_check_failed_code_mode,
13981 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
13982 rtvec_check_failed_bounds, rtl_check_failed_flag,
13983 _fatal_insn_not_found, _fatal_insn): Likewise.
13984 * tree.h (tree_contains_struct_check_failed,
13985 tree_check_failed, tree_not_check_failed,
13986 tree_class_check_failed, tree_range_check_failed,
13987 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
13988 tree_vec_elt_check_failed, phi_node_elt_check_failed,
13989 tree_operand_check_failed, omp_clause_check_failed,
13990 omp_clause_operand_check_failed, omp_clause_range_check_failed):
13993 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13995 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
13996 flag_branch_probabilities.
13997 * ipa-inline.c (edge_badness): Likewise.
13998 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
13999 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
14000 * predict.c (maybe_hot_frequency_p): Likewise.
14001 (probably_never_executed): Likewise.
14002 * sched-ebb.c (schedule_ebbs): Likewise.
14003 * sched-rgn.c (find_single_block_region): Likewise.
14004 * tracer.c (tail_duplicate): Likewise.
14006 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14008 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
14009 longer requires x_flag_profile_use.
14011 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14013 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
14014 instead of flag_reorder_blocks_and_partition.
14015 * dbxout.c (dbxout_function_end): Likewise.
14016 * dwarf2out.c (gen_subprogram_die): Likewise.
14017 * haifa-sched.c (sched_create_recovery_edges): Likewise.
14018 * hw-doloop.c (reorg_loops): Likewise.
14019 * varasm.c (assemble_start_function,
14020 assemble_end_function): Likewise.
14021 (decide_function_section): Do not check for
14022 flag_reorder_blocks_and_partition.
14024 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
14026 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
14028 (chkp_get_hard_register_fake_addr_expr): Ditto.
14029 (chkp_build_addr_expr): Add check for hard reg case.
14030 (chkp_parse_array_and_component_ref): Ditto.
14031 (chkp_find_bounds_1): Ditto.
14032 (chkp_process_stmt): Don't generate bounds store for
14035 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14037 * predict.c (maybe_hot_bb_p): Do not check profile status.
14038 (maybe_hot_edge_p): Likewise.
14039 (probably_never_executed): Check for zero counts even if profile
14041 (unlikely_executed_edge_p): New function.
14042 (unlikely_executed_stmt_p): New function.
14043 (unlikely_executed_bb_p): New function.
14044 (set_even_probabilities): Use unlikely predicates.
14045 (combine_predictions_for_bb): Likewise.
14046 (predict_paths_for_bb): Likewise.
14047 (predict_paths_leading_to_edge): Likewise.
14048 (determine_unlikely_bbs): New function.
14049 (estimate_bb_frequencies): Use it.
14050 (compute_function_frequency): Use zero counts even if profile is
14052 * profile-count.h: Fix typo.
14054 2017-08-08 Julia Koval <julia.koval@intel.com>
14056 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
14057 _mm512_mask_cvtsepi16_storeu_epi8,
14058 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
14059 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
14060 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
14061 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
14062 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
14063 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
14064 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
14065 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
14066 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
14067 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
14068 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
14069 __builtin_ia32_pmovuswb256mem_mask,
14070 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
14071 __builtin_ia32_pmovwb512mem_mask): New builtins.
14073 2017-08-08 Julia Koval <julia.koval@intel.com>
14075 PR target/73350,80862
14076 * config/i386/subst.md (round): Fix round pattern.
14077 * config/i386/i386.c (ix86_erase_embedded_rounding):
14078 Fix erasing rounding for the fixed pattern.
14080 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14082 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
14084 2017-06-08 Martin Liska <mliska@suse.cz>
14086 PR gcov-profile/80911
14087 * gcov.c (block_info::block_info): New constructor.
14089 2017-06-07 Carl Love <cel@us.ibm.com>
14091 * config/rs6000/rs6000-c: The return type of the following
14092 built-in functions was implemented as int not long long. Fix sign
14093 of return value for the unsigned version of vec_mulo and vec_mule.
14094 vector unsigned long long vec_bperm (vector unsigned long long,
14095 vector unsigned char)
14096 vector signed long long vec_mule (vector signed int,
14098 vector unsigned long long vec_mule (vector unsigned int,
14099 vector unsigned int)
14100 vector signed long long vec_mulo (vector signed int,
14102 vector unsigned long long vec_mulo (vector unsigned int,
14103 vector unsigned int)
14104 * doc/extend.texi: Fix the documentation for the built-in
14107 2017-06-07 Carl Love <cel@us.ibm.com>
14110 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
14113 2017-06-07 Carl Love <cel@us.ibm.com>
14115 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
14116 support, Generate doublehv for signed int/float for BE case only.
14118 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
14120 * doc/invoke.texi (mcx16): Rewrite.
14122 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14124 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
14125 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
14126 *mov<mode>_softfloat, and an anonymous splitter): Use
14127 nonimmediate_operand instead of rs6000_nonimmediate_operand.
14129 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14131 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
14133 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
14134 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
14135 (rs6000_debug_reg_global): Adjust.
14136 (rs6000_init_hard_regno_mode_ok): Adjust.
14137 (rs6000_dbx_register_number): Adjust.
14138 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
14139 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
14140 Remove SPE_ACC and SPEFSCR.
14141 (REG_ALLOC_ORDER): Ditto.
14142 (FRAME_POINTER_REGNUM): Change to 111.
14143 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
14144 (REG_CLASS_NAMES): Ditto.
14145 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
14146 (REGISTER_NAMES): Ditto.
14147 (ADDITIONAL_REG_NAMES): Ditto.
14148 (rs6000_reg_names): Ditto.
14149 * config/rs6000/rs6000.md: Renumber some register number
14152 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14154 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
14156 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
14157 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
14159 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
14160 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
14161 Delete the SPE high registers.
14162 (REG_ALLOC_ORDER): Ditto.
14163 (enum reg_class): Remove SPE_HIGH_REGS.
14164 (REG_CLASS_NAMES): Ditto.
14165 (REG_CLASS_CONTENTS): Delete the SPE high registers.
14166 (REGISTER_NAMES): Ditto.
14167 (rs6000_reg_names): Ditto.
14168 * doc/tm.texi.in: Remove SPE as example.
14169 * doc/tm.texi: Regenerate.
14171 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14173 * config/rs6000/8540.md (ppc8540_brinc): Delete.
14174 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
14175 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
14176 * config/rs6000/rs6000.md (type): Remove "brinc".
14178 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14180 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
14181 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
14182 * config/rs6000/linuxspe.h: Delete file.
14183 * config/rs6000/rs6000.md: Don't include spe.md.
14184 * config/rs6000/spe.h: Delete file.
14185 * config/rs6000/spe.md: Delete file.
14186 * config/rs6000/t-rs6000: Remove spe.md.
14188 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14190 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
14191 (reg_or_none500mem_operand): Delete.
14192 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
14193 instead of reg_or_none500mem_operand.
14195 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14197 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
14198 handling of SPE flags.
14199 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
14201 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14203 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
14205 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
14206 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
14207 paired_divv2sf3): Similar.
14208 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
14209 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
14210 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
14212 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
14213 Rename the paired_* instruction patterns.
14214 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
14216 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
14217 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
14218 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
14219 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
14220 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
14221 PAIRED_VECTOR_MODE.
14222 (struct machine_function): Delete field spe_insn_chain_scanned_p.
14223 (spe_func_has_64bit_regs_p): Delete.
14224 (spe_expand_predicate_builtin): Delete.
14225 (spe_expand_evsel_builtin): Delete.
14226 (TARGET_DWARF_REGISTER_SPAN): Do not define.
14227 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
14228 (invalid_e500_subreg): Delete.
14229 (rs6000_legitimize_address): Always force_reg op2 as well, for
14230 paired single memory accesses.
14231 (rs6000_member_type_forces_blk): Delete.
14232 (rs6000_spe_function_arg): Delete.
14233 (rs6000_expand_unop_builtin): Delete SPE handling.
14234 (rs6000_expand_binop_builtin): Ditto.
14235 (spe_expand_stv_builtin): Delete.
14236 (bdesc_2arg_spe): Delete.
14237 (spe_expand_builtin): Delete.
14238 (spe_expand_predicate_builtin): Delete.
14239 (spe_expand_evsel_builtin): Delete.
14240 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
14241 (spe_init_builtins): Delete.
14242 (spe_func_has_64bit_regs_p): Delete.
14243 (savres_routine_name): Delete "info" parameter. Adjust callers.
14244 (rs6000_emit_stack_reset): Ditto.
14245 (rs6000_dwarf_register_span): Delete.
14246 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
14247 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
14248 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
14250 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
14252 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
14253 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
14254 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
14255 mulv2sf3, divv2sf3): Delete expanders.
14257 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14259 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
14261 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14263 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
14264 * config/rs6000/rs6000.c: Ditto.
14266 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14268 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
14269 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
14270 comparison_operator.
14272 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14274 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
14275 * config/rs6000/rs6000.opt: Ditto.
14276 * config/rs6000/t-rtems: Ditto.
14278 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14280 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
14281 TARGET_E500_SINGLE by 0, simplify.
14282 * config/rs6000/rs6000.c: Ditto.
14283 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
14284 (spe_build_register_parallel): Delete.
14285 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
14286 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
14287 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
14288 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
14289 (E500_CONVERT): Delete.
14290 * config/rs6000/spe.md: Remove many patterns and all define_constants.
14292 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14294 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
14295 * config/rs6000/dfp.md: Ditto.
14296 (negdd2, *negdd2_fpr): Merge.
14297 (absdd2, *absdd2_fpr): Merge.
14298 (negtd2, *negtd2_fpr): Merge.
14299 (abstd2, *abstd2_fpr): Merge.
14300 * config/rs6000/e500.h: Delete file.
14301 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
14302 TARGET_FPRS by 1 and simplify.
14303 * config/rs6000/rs6000-c.c: Ditto.
14304 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
14305 TARGET_DF_SPE by 0.
14306 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
14308 * config/rs6000/rs6000.md: Ditto.
14309 (floatdidf2, *floatdidf2_fpr): Merge.
14310 (move_from_CR_gt_bit): Delete.
14311 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
14312 (E500_CR_IOR_COMPARE): Delete.
14313 (All patterns that require !TARGET_FPRS): Delete.
14314 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
14316 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14318 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
14320 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14322 * graphds.c (add_edge): Intitialize edge's attached data.
14323 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
14324 pointer parameter. Call pointed function on each edge during
14325 graph traversing. Skip traversing the edge when the function
14327 (graphds_dfs, graphds_scc): Ditto.
14328 (for_each_edge): New parameter. Pass the new parameter to callback
14330 * graphds.h (skip_edge_callback): New function pointer type.
14331 (graphds_dfs, graphds_scc): New function pointer parameter.
14332 (graphds_edge_callback, for_each_edge): New parameter.
14334 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14336 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
14337 out code checking if runtime alias check is possible to below ...
14338 Call the new function.
14339 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
14340 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
14342 2017-06-07 Marek Polacek <polacek@redhat.com>
14345 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
14346 TYPE_OVERFLOW_WRAPS check.
14348 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14350 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
14351 if versioning is required.
14352 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
14353 peeling with the check for versioning.
14355 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14357 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
14358 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
14359 Set true to new parameter if new ssa variable is defined.
14360 (vect_gen_vector_loop_niters): Refactor. Set range information
14361 for the new vector loop bound variable.
14362 (vect_do_peeling): Ditto.
14364 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14366 * tree-affine.c (ssa.h): Include header file.
14367 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
14368 has wrapping overflow behavior.
14370 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14372 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
14374 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14376 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
14377 (tree_to_aff_combination): ... here.
14379 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14381 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
14382 reg_pressure model function.
14383 (ivopts_global_cost_for_size): Delete.
14384 (determine_set_costs, iv_ca_recount_cost): Call new model function
14385 ivopts_estimate_reg_pressure.
14387 2017-06-07 Tamar Christina <tamar.christina@arm.com>
14389 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
14390 expensive than udiv. Remove floating point cases from mod.
14392 2017-06-07 Tamar Christina <tamar.christina@arm.com>
14394 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
14395 Increase idiv cost.
14397 2017-06-07 Tamar Christina <tamar.christina@arm.com>
14399 * config/aarch64/aarch64.md
14400 (copysignsf3): Fix mask generation.
14402 2017-06-07 Jakub Jelinek <jakub@redhat.com>
14404 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
14406 (class dump_manager): Add register_dumps method.
14407 * dumpfile.c: Include langhooks.h.
14408 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
14409 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
14410 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
14411 (dump_manager::dump_register): Start with 512 entries instead of 32.
14412 (dump_manager::register_dumps): New method.
14413 * toplev.c (general_init): Instead of invoking register_dumps
14414 langhook, invoke register_dumps method on the dump manager.
14415 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
14418 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
14420 * doc/md.texi: Clarify the restrictions on a define_insn condition.
14421 Say that # requires an associated define_split to exist, and that
14422 the define_split must be suitable for use after register allocation.
14424 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14426 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
14427 (compute_outgoing_frequencies): Also initialize zero counts.
14428 (find_many_sub_basic_blocks): Do not produce uninitialized profile
14429 around loops; preserve more of profile when nothing changes.
14431 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
14433 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
14435 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
14436 * config/arm/arm-cpu-cdata.h: Regenerate.
14437 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
14438 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
14439 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
14440 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
14441 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
14443 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
14445 * config/arm/t-rmprofile: Likewise.
14446 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
14448 2017-06-06 David S. Miller <davem@davemloft.net>
14451 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
14452 blockage if function uses alloca.
14454 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14456 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
14457 New "uid" fields to hold pretty-print IDs of group and ref.
14458 Memory references are now identified as <group_id>:<ref_id>
14459 instead of using [random] addresses.
14460 (dump_mem_details): Simplify, no functional change.
14461 (dump_mem_ref): Simplify and make output more concise.
14462 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
14463 (find_or_create_group): Initialize group uid.
14464 (record_ref): Initialize ref uid. Improve debug output.
14465 (prune_group_by_reuse, should_issue_prefetch_p,)
14466 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
14467 (mark_nontemporal_store, determine_loop_nest_reuse):
14468 Improve debug output.
14470 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14472 * dbgcnt.def (prefetch): New debug counter.
14473 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
14474 (schedule_prefetches): Stop issueing prefetches if debug counter
14477 2017-06-06 Tom de Vries <tom@codesourcery.com>
14479 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
14480 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
14482 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14484 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
14485 Use aarch64_reg_or_zero predicate for operand 4.
14486 (aarch64_compare_and_swap<mode> define_insn_and_split):
14487 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
14488 (aarch64_store_exclusive<mode>): Likewise for operand 2.
14490 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
14492 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
14493 (arm_compute_save_core_reg_mask): This.
14494 (thumb1_compute_save_reg_mask): Rename into ...
14495 (thumb1_compute_save_core_reg_mask): This.
14496 (arm_compute_save_reg0_reg12_mask): Adapt comment.
14497 (arm_compute_frame_layout): Likewise.
14499 2017-06-06 Richard Biener <rguenther@suse.de>
14501 PR tree-optimization/80974
14502 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
14503 keep or clear leaders SSA info.
14505 2017-06-06 Tom de Vries <tom@codesourcery.com>
14507 * config/nvptx/nvptx.c (split_mode_p): New function.
14508 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
14510 2017-06-06 Tom de Vries <tom@codesourcery.com>
14512 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
14514 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14517 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
14520 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14522 * shrink-wrap.c (handle_simple_exit): Update profile.
14523 (try_shrink_wrapping): Upate profile.
14525 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14527 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
14528 (tree_guess_outgoing_edge_probabilities): New.
14529 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
14530 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
14532 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14534 * ipa-split.c (split_function): Initialize return bb profile.
14536 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14538 * profile.c (compute_branch_probabilities): Also initialize
14539 EXIT_BLOCK profile.
14541 2017-06-06 Richard Biener <rguenther@suse.de>
14543 PR tree-optimization/80928
14544 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
14545 (vect_analyze_loop_operations): Properly guard analysis for
14547 (vect_transform_loop): Likewise.
14548 (vect_analyze_loop_2): Also reset SLP type on PHIs.
14549 (vect_model_induction_cost): Do not cost for pure SLP.
14550 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
14551 of induction in inner loop vectorization.
14552 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
14553 (vect_get_and_check_slp_defs): Handle vect_induction_def.
14554 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
14556 (vect_analyze_slp_cost_1): Cost induction.
14557 (vect_detect_hybrid_slp_stmts): Handle PHIs.
14558 (vect_get_slp_vect_defs): Likewise.
14559 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
14560 (vect_transform_stmt): Handle SLP reductions.
14561 * tree-vectorizer.h (vectorizable_induction): Adjust.
14563 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
14565 * config/rs6000/rs6000.c (make_resolver_func): Update
14566 init_lowered_empty_function call.
14568 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
14570 * doc/invoke.texi: Document the -fprofile-abs-path option.
14571 * common.opt (fprofile-abs-path): New option.
14572 * gcov-io.h (gcov_write_filename): Declare.
14573 * gcov-io.c (gcov_write_filename): New function.
14574 * coverage.c (coverage_begin_function): Use gcov_write_filename.
14575 * profile.c (output_location): Likewise.
14577 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14579 * shring-wrap.c: Revert accidental commit.
14581 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
14583 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
14585 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14587 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
14589 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
14590 profile in callgraph edge.
14591 * profile-count.h (apply_probability): If THIS is 0, then result is 0
14592 (apply_scale): Likewise.
14593 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
14594 Also scale profile when inlining function with zero profile.
14595 (initialize_cfun): Update exit block profile even when it is zero.
14596 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
14597 when profile is read.
14599 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
14601 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
14602 (CLONE_*): New constants to define the processors we can generate
14603 code for with the target_clone attribute.
14604 (rs6000_clone_map): New array to identify which clone processors
14605 the current program is running on.
14606 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
14607 target_clone attribute.
14608 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
14609 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
14610 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
14611 (cpu_expand_builtin): Add support for target_clone attribute.
14612 (rs6000_valid_attribute_p): Allow "default" attribute.
14613 (get_decl_name): New debug function to simplify printing the
14614 current function name in debugging statements.
14615 (rs6000_clone_priority): New functions to support the target_clone
14616 attribute, and be able to generate code to switch between ISA 2.05
14617 through ISA 3.0 (power6 through power9).
14618 (rs6000_compare_version_priority): Likewise.
14619 (rs6000_get_function_versions_dispatcher): Likewise.
14620 (make_resolver_func): Likewise.
14621 (add_condition_to_bb): Likewise.
14622 (dispatch_function_versions): Likewise.
14623 (rs6000_generate_version_dispatcher_body): Likewise.
14624 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
14625 (fusion_gpr_load_p): Fix a spacing issue.
14626 * doc/extend.texi (Common Function Attributes): Document that the
14627 PowerPC supports the target_clone attribute.
14629 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
14631 * config/arm/arm.h: explain F symbol found in description of ARM
14632 register allocation in its legend.
14634 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14636 * config/mips/frame-header-opt.c: Include profile-count.h.
14637 * config/riscv/riscv.c: Include profile-count.h
14639 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14641 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
14643 (sm_set_flag_if_changed): Add bbs field.
14644 (execute_sm_if_changed_flag_set): Pass BBS.
14645 (execute_sm): Update.
14647 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14649 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
14652 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14654 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
14655 (peephole2): New peephole2 to emit the above.
14656 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
14658 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14660 * config/aarch64/aarch64.c (define_peephole2 above
14661 *sub_<shift>_<mode>): New peephole.
14663 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14665 * config/i386/i386.c (make_resolver_func): Update.
14666 * Makefile.in: Add profile-count.h and profile-count.o
14667 * auto-profile.c (afdo_indirect_call): Update to new API.
14668 (afdo_set_bb_count): Update.
14669 (afdo_propagate_edge): Update.
14670 (afdo_propagate_circuit): Update.
14671 (afdo_calculate_branch_prob): Update.
14672 (afdo_annotate_cfg): Update.
14673 * basic-block.h: Include profile-count.h
14674 (struct edge_def): Turn count to profile_count.
14675 (struct basic_block_def): Likewie.
14676 (REG_BR_PROB_BASE): Move to profile-count.h
14677 (RDIV): Move to profile-count.h
14678 * bb-reorder.c (max_entry_count): Turn to profile_count.
14679 (find_traces): Update.
14680 (rotate_loop):Update.
14681 (connect_traces):Update.
14682 (sanitize_hot_paths):Update.
14683 * bt-load.c (migrate_btr_defs): Update.
14684 * cfg.c (RDIV): Remove.
14685 (init_flow): Use alloc_block.
14686 (alloc_block): Uninitialize count.
14687 (unchecked_make_edge): Uninitialize count.
14688 (check_bb_profile): Update.
14689 (dump_edge_info): Update.
14690 (dump_bb_info): Update.
14691 (update_bb_profile_for_threading): Update.
14692 (scale_bbs_frequencies_int): Update.
14693 (scale_bbs_frequencies_gcov_type): Update.
14694 (scale_bbs_frequencies_profile_count): New.
14695 * cfg.h (update_bb_profile_for_threading): Update.
14696 (scale_bbs_frequencies_profile_count): Declare.
14697 * cfgbuild.c (compute_outgoing_frequencies): Update.
14698 (find_many_sub_basic_blocks): Update.
14699 * cfgcleanup.c (try_forward_edges): Update.
14700 (try_crossjump_to_edge): Update.
14701 * cfgexpand.c (expand_gimple_tailcall): Update.
14702 (construct_exit_block): Update.
14703 * cfghooks.c (verify_flow_info): Update.
14704 (dump_bb_for_graph): Update.
14705 (split_edge): Update.
14706 (make_forwarder_block): Update.
14707 (duplicate_block): Update.
14708 (account_profile_record): Update.
14709 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
14710 (get_estimated_loop_iterations): Update.
14711 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
14712 (single_likely_exit): Update.
14713 * cfgloopmanip.c (scale_loop_profile): Update.
14715 (set_zero_probability): Update.
14716 (lv_adjust_loop_entry_edge): Update.
14717 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
14718 (purge_dead_edges): Update.
14719 (rtl_account_profile_record): Update.
14720 * cgraph.c (cgraph_node::create): Uninitialize count.
14721 (symbol_table::create_edge): Uninitialize count.
14722 (cgraph_update_edges_for_call_stmt_node): Update.
14723 (cgraph_edge::dump_edge_flags): Update.
14724 (cgraph_node::dump): Update.
14725 (cgraph_edge::maybe_hot_p): Update.
14726 * cgraph.h: Include profile-count.h
14727 (create_clone), create_edge, create_indirect_edge): Update.
14728 (cgraph_node): Turn count to profile_count.
14729 (cgraph_edge0: Likewise.
14730 (make_speculative, clone): Update.
14731 (create_edge): Update.
14732 (init_lowered_empty_function): Update.
14733 * cgraphclones.c (cgraph_edge::clone): Update.
14734 (duplicate_thunk_for_node): Update.
14735 (cgraph_node::create_clone): Update.
14736 * cgraphunit.c (cgraph_node::analyze): Update.
14737 (cgraph_node::expand_thunk): Update.
14738 * final.c (dump_basic_block_info): Update.
14739 * gimple-streamer-in.c (input_bb): Update.
14740 * gimple-streamer-out.c (output_bb): Update.
14741 * graphite.c (print_global_statistics): Update.
14742 (print_graphite_scop_statistics): Update.
14743 * hsa-brig.c: Include basic-block.h.
14744 * hsa-dump.c: Include basic-block.h.
14745 * hsa-gen.c (T sum_slice): Update.
14746 (convert_switch_statements):Update.
14747 * hsa-regalloc.c: Include basic-block.h.
14748 * ipa-chkp.c (chkp_produce_thunks): Update.
14749 * ipa-cp.c (struct caller_statistics): Update.
14750 (init_caller_stats): Update.
14751 (gather_caller_stats): Update.
14752 (ipcp_cloning_candidate_p): Update.
14753 (good_cloning_opportunity_p): Update.
14754 (get_info_about_necessary_edges): Update.
14755 (dump_profile_updates): Update.
14756 (update_profiling_info): Update.
14757 (update_specialized_profile): Update.
14758 (perhaps_add_new_callers): Update.
14759 (decide_about_value): Update.
14760 (ipa_cp_c_finalize): Update.
14761 * ipa-devirt.c (struct odr_type_warn_count): Update.
14762 (struct decl_warn_count): Update.
14763 (struct final_warning_record): Update.
14764 (possible_polymorphic_call_targets): Update.
14765 (ipa_devirt): Update.
14766 * ipa-fnsummary.c (redirect_to_unreachable): Update.
14767 * ipa-icf.c (sem_function::merge): Update.
14768 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
14769 * ipa-inline.c (compute_uninlined_call_time): Update.
14770 (compute_inlined_call_time): Update.
14771 (want_inline_small_function_p): Update.
14772 (want_inline_self_recursive_call_p): Update.
14773 (edge_badness): Update.
14774 (lookup_recursive_calls): Update.
14775 (recursive_inlining): Update.
14776 (inline_small_functions): Update.
14777 (dump_overall_stats): Update.
14778 (dump_inline_stats): Update.
14779 * ipa-profile.c (ipa_profile_generate_summary): Update.
14780 (ipa_propagate_frequency): Update.
14781 (ipa_profile): Update.
14782 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14783 * ipa-utils.c (ipa_merge_profiles): Update.
14784 * loop-doloop.c (doloop_modify): Update.
14785 * loop-unroll.c (report_unroll): Update.
14786 (unroll_loop_runtime_iterations): Update.
14787 * lto-cgraph.c (lto_output_edge): Update.
14788 (lto_output_node): Update.
14789 (input_node): Update.
14790 (input_edge): Update.
14791 (merge_profile_summaries): Update.
14792 * lto-streamer-in.c (input_cfg): Update.
14793 * lto-streamer-out.c (output_cfg): Update.
14794 * mcf.c (create_fixup_graph): Update.
14795 (adjust_cfg_counts): Update.
14796 (sum_edge_counts): Update.
14797 * modulo-sched.c (sms_schedule): Update.
14798 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
14799 * predict.c (maybe_hot_count_p): Update.
14800 (probably_never_executed): Update.
14801 (dump_prediction): Update.
14802 (combine_predictions_for_bb): Update.
14803 (propagate_freq): Update.
14804 (handle_missing_profiles): Update.
14805 (counts_to_freqs): Update.
14806 (rebuild_frequencies): Update.
14807 (force_edge_cold): Update.
14808 * predict.h: Include profile-count.h
14809 (maybe_hot_count_p, counts_to_freqs): UPdate.
14810 * print-rtl-function.c: Do not include cfg.h
14811 * print-rtl.c: Include basic-block.h
14812 * profile-count.c: New file.
14813 * profile-count.h: New file.
14814 * profile.c (is_edge_inconsistent): Update.
14815 (correct_negative_edge_counts): Update.
14816 (is_inconsistent): Update.
14817 (set_bb_counts): Update.
14818 (read_profile_edge_counts): Update.
14819 (compute_frequency_overlap): Update.
14820 (compute_branch_probabilities): Update; Initialize and deinitialize
14822 (branch_prob): Update.
14823 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
14824 (edge_gcov_count): New.
14825 (bb_gcov_count): New.
14826 * shrink-wrap.c (try_shrink_wrapping): Update.
14827 * tracer.c (better_p): Update.
14828 * trans-mem.c (expand_transaction): Update.
14829 (ipa_tm_insert_irr_call): Update.
14830 (ipa_tm_insert_gettmclone_call): Update.
14831 * tree-call-cdce.c: Update.
14832 * tree-cfg.c (gimple_duplicate_sese_region): Update.
14833 (gimple_duplicate_sese_tail): Update.
14834 (gimple_account_profile_record): Update.
14835 (execute_fixup_cfg): Update.
14836 * tree-inline.c (copy_bb): Update.
14837 (copy_edges_for_bb): Update.
14838 (initialize_cfun): Update.
14839 (freqs_to_counts): Update.
14840 (copy_cfg_body): Update.
14841 (expand_call_inline): Update.
14842 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
14843 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
14844 (try_unroll_loop_completely): Update.
14845 (try_peel_loop): Update.
14846 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14847 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
14848 * tree-ssa-loop-split.c (connect_loops): Update.
14849 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
14850 * tree-ssa-reassoc.c (branch_fixup): Update.
14851 * tree-ssa-tail-merge.c (replace_block_by): Update.
14852 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
14853 (compute_path_counts): Update.
14854 (update_profile): Update.
14855 (recompute_probabilities): Update.
14856 (update_joiner_offpath_counts): Update.
14857 (estimated_freqs_path): Update.
14858 (freqs_to_counts_path): Update.
14859 (clear_counts_path): Update.
14860 (ssa_fix_duplicate_block_edges): Update.
14861 (duplicate_thread_path): Update.
14862 * tree-switch-conversion.c (case_bit_test_cmp): Update.
14863 (struct switch_conv_info): Update.
14864 * tree-tailcall.c (decrease_profile): Update.
14865 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
14866 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14867 * value-prof.c (check_counter): Update.
14868 (gimple_divmod_fixed_value): Update.
14869 (gimple_mod_pow2): Update.
14870 (gimple_mod_subtract): Update.
14871 (gimple_ic_transform): Update.
14872 (gimple_stringop_fixed_value): Update.
14873 * value-prof.h (gimple_ic): Update.
14875 2017-06-02 Carl Love <cel@us.ibm.com>
14877 * config/rs6000/rs6000-c: Add support for built-in functions
14878 vector double vec_doublee (vector signed int);
14879 vector double vec_doublee (vector unsigned int);
14880 vector double vec_doublee (vector float);
14881 vector double vec_doubleh (vector signed int);
14882 vector double vec_doubleh (vector unsigned int);
14883 vector double vec_doubleh (vector float);
14884 vector double vec_doublel (vector signed int);
14885 vector double vec_doublel (vector unsigned int);
14886 vector double vec_doublel (vector float);
14887 vector double vec_doubleo (vector signed int);
14888 vector double vec_doubleo (vector unsigned int);
14889 vector double vec_doubleo (vector float);.
14890 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
14891 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
14893 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
14894 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
14895 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
14897 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
14898 vec_doublel, vec_doubleh.
14899 * doc/extend.texi: Update the built-in documentation file for the
14900 new built-in functions.
14902 2017-06-02 David Malcolm <dmalcolm@redhat.com>
14905 * ipa-inline-analysis.c (free_growth_caches): Set
14906 edge_removal_hook_holder to NULL after removing it.
14908 2017-06-02 Sudakshina Das <sudi.das@arm.com>
14910 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
14911 comparision with zero.
14913 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
14914 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14915 for early expansion of vec_min and vec_max builtins.
14916 (builtin_function_type): Add min/max unsigned variants to those
14917 identified as having unsigned arguments.
14919 2017-06-02 Olivier Hainque <hainque@adacore.com>
14921 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
14923 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14925 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
14926 Use VALL_F16 iterator rather than VALL.
14928 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14930 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
14931 Emit CBNZ inside loop when doing a strong exchange and comparing
14932 against zero. Generate the CC flags after the loop.
14934 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
14936 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
14937 (dl_section_ref): New.
14938 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
14939 On AIX, append an expression to subtract the size of the
14940 section length to dl_section_ref.
14942 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
14944 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14945 for early expansion of vector absolute builtins.
14947 2017-06-02 Richard Biener <rguenther@suse.de>
14949 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
14950 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
14952 2017-06-02 Richard Biener <rguenther@suse.de>
14954 PR tree-optimization/80948
14955 * tree-tailcall.c (find_tail_calls): Track stmts to move in
14956 stmt order as well.
14958 2017-06-02 Richard Biener <rguenther@suse.de>
14960 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
14962 * tree-vect-stmts.c (process_use): Do not mark backedge defs
14963 for inductions as relevant.
14965 2017-06-02 Richard Biener <rguenther@suse.de>
14967 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
14968 (vectorizable_induction): ... this. Remove dead code.
14970 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
14972 * builtins. (expand_builtin_alloca): Remove second parameter and
14973 infer its value from the first parameter instead.
14974 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
14976 2017-06-02 Jakub Jelinek <jakub@redhat.com>
14978 PR rtl-optimization/80903
14979 * loop-doloop.c (add_test): Unshare sequence.
14981 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
14983 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
14985 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
14987 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
14989 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
14990 xlogue_layout::get_instance, logue_layout::xlogue_layout,
14991 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
14992 (xlogue_layout::get_stub_rtx): Make static.
14993 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
14994 (xlogue_layout::compute_stub_managed_regs): Rename to...
14995 (xlogue_layout::count_stub_managed_regs): ...this.
14996 (xlogue_layout::is_stub_managed_reg): New function.
14997 (xlogue_layout::m_stub_names): Rename to...
14998 (xlogue_layout::s_stub_names): ...this, make static.
14999 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
15000 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
15001 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
15002 xlogue_layout::s_stub_names): Instantiate statics.
15003 (stub_managed_regs): Remove.
15004 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
15005 (disable_call_ms2sysv_xlogues): Rename to...
15006 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
15007 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
15009 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
15010 change after reload_completed.
15011 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
15013 (ix86_expand_prologue): Likewise.
15014 (ix86_expand_epilogue): Likewise.
15015 (ix86_expand_split_stack_prologue): Likewise.
15016 (ix86_compute_frame_layout): Remove frame parameter ...
15017 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
15018 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
15020 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
15021 (ix86_frame): Move from here ...
15022 * config/i386/i386.h (ix86_frame): ... to here.
15023 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
15024 complete ix86_frame data structure instead. Remove some_ld_name.
15026 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
15028 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
15029 symbols that hold a DECL_VALUE_EXPR.
15031 2017-06-01 Martin Jambor <mjambor@suse.cz>
15033 PR tree-optimization/80898
15034 * tree-sra.c (process_subtree_disqualification): Removed.
15035 (disqualify_candidate): Do not acll
15036 process_subtree_disqualification.
15037 (subtree_mark_written_and_enqueue): New function.
15038 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
15039 RHS has been disqualified and re-queue LHS if necessary. Apart
15040 from that, ignore disqualified RHS.
15042 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15044 * config/s390/s390.c (s390_emit_epilogue): Disable early return
15045 address fetch for z10 or later.
15047 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15049 * config/arc/arc.md (tst_movb): Add guard when splitting.
15051 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15053 * config/arc/arc.c (arc_can_eliminate): Test against
15054 arc_frame_pointer_needed.
15056 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15058 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
15059 to prevent store reordering.
15060 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
15061 (type): Add block type.
15062 (stack_tie): Define special instruction to be used in
15065 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15067 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
15068 constraint. It is not valid for the pattern.
15069 (noncommutative_binary_comparison): Likewise.
15071 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15073 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
15076 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15078 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
15079 be used by the reg-alloc.
15081 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15083 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
15084 reg-alloc when having mul64 or mul32x16 instructions.
15085 (mulsidi3): Likewise.
15086 (umulsidi3): Likewise.
15087 (mulsi32x16): New pattern.
15088 (mulsi64): Likewise.
15089 (mulsidi64): Likewise.
15090 (umulsidi64): Likewise.
15091 (MUL32x16_REG): Define.
15092 (mul64_600): Use MUL32x16_REG.
15093 (mac64_600): Likewise.
15094 (umul64_600): Likewise.
15095 (umac64_600): Likewise.
15097 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15099 * config/arc/arc.md (mulsi3_700): Make it commutative.
15101 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
15103 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
15105 (*sign_extendsidi2_insn): Likewise for movstosw.
15107 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
15109 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
15110 the type of the input discriminant value. Convert the
15111 discriminant value of signedness vary.
15113 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
15115 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
15116 Add to -Wall section.
15118 2017-06-01 Richard Biener <rguenther@suse.de>
15120 PR middle-end/66313
15121 * fold-const.c (fold_plusminus_mult_expr): If the factored
15122 factor may be zero use a wrapping type for the inner operation.
15123 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
15124 and handle moved defs.
15125 (process_assignment): Properly guard the unary op case. Return a
15126 tri-state indicating that moving the stmt before the call may allow
15127 to continue. Pass through to_move.
15128 (find_tail_calls): Handle moving unrelated defs before
15131 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
15134 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
15135 splitter result in the canonical way.
15137 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
15139 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
15140 also for 32bit target. Update insn attributes.
15141 (zero-extendsidi2 splitter): Allow all registers for operand 1.
15143 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
15145 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
15146 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
15147 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
15148 (_mm_maskz_min_ss): New intrinsics.
15150 2017-05-31 Martin Liska <mliska@suse.cz>
15152 * tree-vect-loop.c (vect_create_epilog_for_reduction):
15153 Change comment style to one we normally use.
15154 (vectorizable_reduction): Likewise.
15155 (vectorizable_induction): Likewise.
15156 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15157 (vectorizable_call): Likewise.
15158 (vectorizable_simd_clone_call): Likewise.
15159 (vectorizable_conversion): Likewise.
15160 (vectorizable_assignment): Likewise.
15161 (vectorizable_shift): Likewise.
15162 (vectorizable_operation): Likewise.
15163 (vectorizable_store): Likewise.
15164 (vectorizable_load): Likewise.
15165 * tree-vectorizer.h: Likewise.
15167 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
15169 * passes.c (emergency_dump_function): New.
15170 * tree-pass.h (emergency_dump_function): Declare.
15171 * plugin.c (plugins_internal_error_function): Remove.
15172 * plugin.h (plugins_internal_error_function): Remove declaration.
15173 * toplev.c (internal_error_function): New static function. Use it...
15174 (general_init): ...here.
15176 2017-05-31 Graham Markall <graham.markall@embecosm.com>
15178 * config/arc/arc.c (arc_print_operand): Handle constant operands.
15179 (arc_rtx_costs): Add costs for new patterns.
15180 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
15181 * config/arc/predicates.md: Add _1_2_3_operand predicate.
15183 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
15185 * tree-ssa-strlen.c (get_next_strinfo): New function.
15186 (get_stridx_plus_constant): Use it.
15187 (zero_length_string): Likewise.
15188 (adjust_related_strinfos): Likewise.
15189 (adjust_last_stmt): Likewise.
15191 2017-05-31 Richard Biener <rguenther@suse.de>
15194 * config/i386/i386.c (ix86_expand_builtin): Remove assert
15195 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
15197 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
15199 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
15200 loop_vinfo argument and use of dependence distance vectors.
15201 Check instead whether the two references differ only in their
15202 initial value and assume that they have the same alignment if the
15203 difference is a multiple of the vector alignment.
15204 (vect_analyze_data_refs_alignment): Update call accordingly.
15206 2017-05-31 Martin Liska <mliska@suse.cz>
15209 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
15211 2017-05-31 Bin Cheng <bin.cheng@arm.com>
15213 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
15214 (create_intersect_range_checks): Move from ...
15215 * tree-data-ref.c (create_intersect_range_checks_index)
15216 (create_intersect_range_checks): ... to here.
15217 (create_runtime_alias_checks): New function factored from ...
15218 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
15219 here. Call above function.
15220 * tree-data-ref.h (create_runtime_alias_checks): New function.
15222 2017-05-31 Bin Cheng <bin.cheng@arm.com>
15224 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
15225 segment length for dr_b and compute it in wide_int.
15227 2017-05-31 Richard Biener <rguenther@suse.de>
15229 PR tree-optimization/80906
15230 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
15231 and pass through iv_map.
15232 (copy_bb_and_scalar_dependences): Adjust.
15233 (translate_pending_phi_nodes): Likewise.
15234 (copy_loop_close_phi_args): Handle code-generating IVs instead
15237 2017-05-30 David Malcolm <dmalcolm@redhat.com>
15239 * diagnostic-color.c (color_dict): Add "type-diff".
15240 (parse_gcc_colors): Update comment.
15241 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
15242 -fdiagnostics-show-template-tree and -fno-elide-type.
15243 (GCC_COLORS): Add type-diff to example.
15245 (-fdiagnostics-show-template-tree): New.
15246 (-fno-elide-type): New.
15247 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
15248 the pp_format_decoder callback. Call any m_format_postprocessor's
15250 (pretty_printer::pretty_printer): Initialize
15251 m_format_postprocessor.
15252 (pretty_printer::~pretty_printer): Delete any
15253 m_format_postprocessor.
15254 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
15255 (class format_postprocessor): New class.
15256 (struct pretty_printer::format_decoder): Document the new parameters.
15257 (struct pretty_printer::m_format_postprocessor): New field.
15258 * tree-diagnostic.c (default_tree_printer): Update for new
15259 bool and const char ** params.
15260 * tree-diagnostic.h (default_tree_printer): Likewise.
15262 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
15264 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
15265 (lwa_operand): Delete rs6000_gen_cell_microcode test.
15266 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
15267 rs6000_gen_cell_microcode code.
15268 (rs6000_final_prescan_insn): Delete.
15269 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
15270 "warn-cell-microcode" entries.
15271 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
15272 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
15273 throughout. Change cc_reg_not_micro_cr0_operand to
15274 cc_reg_not_cr0_operand throughout.
15275 (*extendhi<mode>2_noload): Delete.
15276 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
15277 (mwarn-cell-microcode): Delete.
15278 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
15279 -mgen-cell-microcode and -mwarn-cell-microcode.
15281 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
15284 * config/i386/constraints.md (Yd): New constraint.
15286 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
15287 and (?Yd, r) alternatives. Update insn attributes.
15288 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
15289 and (?*Yd, r) alternatives. Update insn attributes.
15290 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
15292 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
15294 * gimplify.c (gimplify_modify_expr): Don't create a
15295 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
15298 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
15300 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
15302 2017-05-30 Richard Biener <rguenther@suse.de>
15304 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
15305 and reduc_def fields.
15306 (STMT_VINFO_REDUC_TYPE): New define.
15307 (STMT_VINFO_REDUC_DEF): Likewise.
15308 (vect_force_simple_reduction): Adjust prototype.
15309 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
15310 (vect_is_simple_reduction): Remove check_reduction argument.
15311 (vect_force_simple_reduction): Adjust and set
15312 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
15313 (vectorizable_reduction): Do not re-do reduction analysis
15314 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
15315 * tree-parloops.c (gather_scalar_reductions): Adjust.
15317 2017-05-30 Richard Biener <rguenther@suse.de>
15319 PR middle-end/80901
15320 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
15323 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
15325 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
15326 Introduce unknown_misalignment parameter and remove vf.
15327 (vect_peeling_hash_get_lowest_cost):
15328 Pass unknown_misalignment parameter.
15329 (vect_enhance_data_refs_alignment):
15330 Fix unsupportable data ref treatment.
15332 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15334 * tree-vect-data-refs.c (vect_get_data_access_cost):
15335 Workaround for SLP handling.
15336 (vect_enhance_data_refs_alignment):
15337 Compute costs for doing no peeling at all, compare to the best
15338 peeling costs so far and avoid peeling if cheaper.
15340 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15342 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
15343 Return peeling info and set costs to zero for unlimited cost
15345 (vect_enhance_data_refs_alignment): Also inspect all datarefs
15346 with unknown misalignment. Compute and costs for unknown
15347 misalignment, compare them to the costs for known misalignment
15348 and choose the cheapest for peeling.
15350 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15352 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
15353 (vect_get_peeling_costs_all_drs): Create function.
15354 (vect_peeling_hash_get_lowest_cost):
15355 Use vect_get_peeling_costs_all_drs.
15356 (vect_peeling_supportable): Create function.
15357 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
15359 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15361 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
15362 DR_HAS_NEGATIVE_STEP.
15363 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
15364 (vect_enhance_data_refs_alignment): Use.
15365 (vect_duplicate_ssa_name_ptr_info): Use.
15366 * tree-vectorizer.h (dr_misalignment): Use.
15367 (known_alignment_for_access_p): Use.
15369 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15372 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
15374 (has_section_name): New function.
15376 2017-05-30 Martin Liska <mliska@suse.cz>
15379 * auto-profile.c (get_function_decl_from_block): Fix
15382 2017-05-30 Richard Biener <rguenther@suse.de>
15384 PR middle-end/80876
15385 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
15387 2017-05-30 Martin Liska <mliska@suse.cz>
15389 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
15390 * dumpfile.h (struct dump_file_info): Remove ctors.
15392 2017-05-30 Martin Liska <mliska@suse.cz>
15394 * predict.def: Fix GNU coding style.
15396 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
15398 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
15399 Mark 'to' argument with ATTRIBUTE_UNUSED.
15401 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
15403 * config/xtensa/xtensa.c (xtensa_emit_call): Use
15404 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
15405 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
15408 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
15410 * doc/install.texi (Options specification): Restore entry of
15411 --enable-sjlj-exceptions.
15413 2017-05-27 Michael Eager <eager@eagercon.com>
15416 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
15418 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
15420 * config/microblaze/microblaze.h
15421 (FIXED_REGISTERS): Update in macro.
15422 (CALL_USED_REGISTERS): Update in macro.
15424 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
15426 * doc/install.texi: Add links to macOS binary distributions.
15428 2017-05-27 Jakub Jelinek <jakub@redhat.com>
15432 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15434 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
15436 2017-05-26 Martin Liska <mliska@suse.cz>
15438 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
15440 2017-05-26 Martin Liska <mliska@suse.cz>
15442 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
15443 always leading ';; '.
15444 (dump_bb_info): Likewise.
15445 (brief_dump_cfg): Likewise.
15446 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
15447 * dumpfile.c: Remove usage of TDF_VERBOSE.
15448 * dumpfile.h (enum dump_kind): Likewise.
15449 (dump_gimple_bb_header): Do not use TDF_COMMENT.
15450 * print-tree.c (debug_verbose): Remove.
15451 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
15452 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
15453 * tree-diagnostic.c (default_tree_printer): Replace
15454 TDF_DIAGNOSTIC with TDF_SLIM.
15456 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15458 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
15459 in parameter loop, rather than loop_vinfo.
15460 (create_intersect_range_checks): Ditto.
15461 (vect_create_cond_for_alias_checks): Update call to above functions.
15463 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15465 PR tree-optimization/80815
15466 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
15467 for merging runtime alias checks. Handle negative DR_STEPs.
15469 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15471 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
15473 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
15474 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
15475 out code pruning runtime alias checks.
15476 * tree-data-ref.c (prune_runtime_alias_test_list): New function
15477 factored out from above.
15478 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
15480 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
15482 (prune_runtime_alias_test_list): New decalaration.
15484 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15486 * tree-vect-data-refs.c (compare_tree): Rename and move ...
15487 * tree-data-ref.c (data_ref_compare_tree): ... to here.
15488 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
15489 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
15490 (operator==, comp_dr_with_seg_len_pair): Ditto.
15491 (vect_prune_runtime_alias_test_list): Ditto.
15493 2017-05-26 Martin Liska <mliska@suse.cz>
15496 * params.def: Bound partial-inlining-entry-probability param.
15498 2017-05-26 Marek Polacek <polacek@redhat.com>
15501 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
15504 2017-05-26 Richard Biener <rguenther@suse.de>
15506 PR tree-optimization/80842
15507 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
15510 2017-05-26 Richard Biener <rguenther@suse.de>
15512 PR tree-optimization/80844
15513 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
15515 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
15517 * doc/md.texi (Machine Constraints): Update x86 family
15518 machine constraints section to match 'config/i386/constraints.md'.
15520 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
15522 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
15524 2017-05-25 Nathan Sidwell <nathan@acm.org>
15526 * doc/invoke.texi (--enable-languages): Update documentation.
15528 2017-05-25 Martin Liska <mliska@suse.cz>
15530 * dumpfile.c: Add TDF_FOLDING.
15531 * dumpfile.h (enum dump_kind): Likewise.
15532 * genmatch.c (dt_simplify::gen_1): Use it.
15534 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15536 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
15538 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15540 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
15541 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
15543 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15545 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
15546 * match.pd (X == C): Rewrite it here.
15547 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
15548 with_certain_nonzero_bits2): New predicates.
15549 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
15551 2017-05-24 Nathan Sidwell <nathan@acm.org>
15553 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
15556 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
15559 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
15561 * config/powerpcspe: New port. Files are copied from the rs6000
15562 port, with "rs6000" in filenames replaced by "powerpcspe".
15564 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
15566 PR rtl-optimization/80754
15567 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
15569 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
15571 * config/sparc/sparc.md (length): Return the correct value for -mflat
15572 sibcalls to match output_sibcall.
15574 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
15578 * config/rs6000/rs6000.c (struct machine_function): Add new field
15580 (rs6000_get_separate_components): Init that field, use it.
15581 (rs6000_components_for_bb): Use the field.
15583 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
15585 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
15587 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
15589 PR middle-end/80823
15590 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
15592 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15595 * config/s390/s390.c (s390_check_qrst_address): Check incoming
15596 address against address_operand predicate.
15597 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
15599 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
15601 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
15602 parameters passed indirectly.
15604 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
15606 * config/i386/i386.md (*movdi_internal): Remove SSE4
15607 alternative 18 (?r, *v). Update insn attributes.
15608 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
15609 Update insn attributes.
15610 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
15611 Update insn attributes.
15612 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
15613 alternative 1 (r, v). Remove isa attribute.
15614 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
15615 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
15616 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
15618 2017-05-23 Tom de Vries <tom@codesourcery.com>
15620 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
15623 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15625 * cgraphunit.c (symbol_table::process_new_functions): Update.
15626 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
15627 (inline_generate_summary): Rename to ...
15628 (ipa_fn_summary_generate): ... this one.
15629 (inline_read_summary): Rename to ...
15630 (ipa_fn_summary_read): ... this one.
15631 (inline_write_summary): Rename to ...
15632 (ipa_fn_summary_write): ... this one.
15633 (inline_free_summary): Rename to ...
15634 (ipa_free_fn_summary): ... this one.
15635 (pass_data_local_fn_summary, pass_local_fn_summary,
15636 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
15637 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
15638 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
15639 make_pass_ipa_fn_summary): New.
15640 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
15641 inline_write_summary, inline_free_summary): Remove.
15642 (ipa_free_fn_summary) : New.
15643 * ipa-inline.c (ipa_inline): Update.
15644 (pass_ipa_inline): Do not generate summaries.
15645 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
15647 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
15648 and add pass_ipa_fn_summary.
15649 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
15651 (make_pass_inline_parameters): Remove.
15653 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
15655 * omp-low.c (struct omp_context): Remove "default_kind" member.
15658 * omp-offload.c (execute_oacc_device_lower): Remove the
15659 parallelism dimensions function attributes for unparallelized
15660 OpenACC kernels constructs.
15662 2017-05-23 Martin Liska <mliska@suse.cz>
15664 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
15666 (cgraph_edge::make_speculative): Likewise.
15667 (cgraph_edge::resolve_speculation): Likewise.
15668 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
15669 (cgraph_node::dump): Likewise.
15670 * cgraph.h: Likewise.
15671 * cgraphunit.c (analyze_functions): Likewise.
15672 (symbol_table::compile): Likewise.
15673 * ipa-cp.c (print_all_lattices): Likewise.
15674 (determine_versionability): Likewise.
15675 (initialize_node_lattices): Likewise.
15676 (ipcp_verify_propagated_values): Likewise.
15677 (estimate_local_effects): Likewise.
15678 (update_profiling_info): Likewise.
15679 (create_specialized_node): Likewise.
15680 (perhaps_add_new_callers): Likewise.
15681 (decide_about_value): Likewise.
15682 (decide_whether_version_node): Likewise.
15683 (identify_dead_nodes): Likewise.
15684 (ipcp_store_bits_results): Likewise.
15685 * ipa-devirt.c (dump_targets): Likewise.
15686 (ipa_devirt): Likewise.
15687 * ipa-icf.c (sem_item::dump): Likewise.
15688 (sem_function::equals): Likewise.
15689 (sem_variable::equals): Likewise.
15690 (sem_item_optimizer::read_section): Likewise.
15691 (sem_item_optimizer::execute): Likewise.
15692 (congruence_class::dump): Likewise.
15693 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
15694 (dump_inline_summary): Likewise.
15695 (estimate_node_size_and_time): Likewise.
15696 (inline_analyze_function): Likewise.
15697 * ipa-inline-transform.c (inline_call): Likewise.
15698 * ipa-inline.c (report_inline_failed_reason): Likewise.
15699 (want_early_inline_function_p): Likewise.
15700 (edge_badness): Likewise.
15701 (update_edge_key): Likewise.
15702 (inline_small_functions): Likewise.
15703 * ipa-profile.c (ipa_profile): Likewise.
15704 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
15705 (ipa_make_edge_direct_to_target): Likewise.
15706 (remove_described_reference): Likewise.
15707 (ipa_impossible_devirt_target): Likewise.
15708 (propagate_controlled_uses): Likewise.
15709 (ipa_print_node_params): Likewise.
15710 (ipcp_transform_function): Likewise.
15711 * ipa-pure-const.c (pure_const_read_summary): Likewise.
15712 (propagate_pure_const): Likewise.
15713 * ipa-reference.c (generate_summary): Likewise.
15714 (read_write_all_from_decl): Likewise.
15715 (propagate): Likewise.
15716 (ipa_reference_read_optimization_summary): Likewise.
15717 * ipa-utils.c (ipa_merge_profiles): Likewise.
15718 * ipa.c (walk_polymorphic_call_targets): Likewise.
15719 (symbol_table::remove_unreachable_nodes): Likewise.
15720 (ipa_single_use): Likewise.
15721 * passes.c (execute_todo): Likewise.
15722 * predict.c (drop_profile): Likewise.
15723 * symtab.c (symtab_node::get_dump_name): New function.
15724 (symtab_node::dump_name): Likewise.
15725 (symtab_node::dump_asm_name): Likewise.
15726 (symtab_node::dump_references): Likewise.
15727 (symtab_node::dump_referring): Likewise.
15728 (symtab_node::dump_base): Likewise.
15729 (symtab_node::debug_symtab): Likewise.
15730 * tree-sra.c (convert_callers_for_node): Likewise.
15731 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
15732 * value-prof.c (init_node_map): Likewise.
15734 2017-05-23 Martin Liska <mliska@suse.cz>
15736 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
15737 and symtab_node::debug_symtab to symbol_table::debug.
15738 * cgraphunit.c (analyze_functions): Use the renamed function.
15739 (symbol_table::compile): Likewise.
15740 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
15741 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
15742 * passes.c (execute_todo): Likewise.
15743 * symtab.c (symbol_table::dump): New function.
15744 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
15746 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15748 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
15749 that nonconst implies exec.
15751 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15753 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
15754 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
15755 (inline_edge_summary_vec): Turn into ...
15756 (ipa_call_summaries): ... this one.
15757 (redirect_to_unreachable, edge_set_predicate,
15758 evaluate_properties_for_edge, inline_summary_alloc,
15759 reset_ipa_call_summary, reset_inline_summary,
15760 inline_summary_t::duplicate): Update.
15761 (inline_edge_duplication_hook): Turn to ...
15762 (ipa_call_summary_t::duplicate): ... this one.
15763 (inline_edge_removal_hook): Turn to ...
15764 (ipa_call_summary_t::remove): ... this one.
15765 (dump_inline_edge_summary): Turn to ...
15766 (dump_ipa_call_summary): ... this one.
15767 (estimate_function_body_sizes): Update.
15768 (inline_update_callee_summaries): Update.
15769 (remap_edge_change_prob): Update.
15770 (remap_edge_summaries): Update.
15771 (inline_merge_summary): Update.
15772 (do_estimate_edge_time): Update.
15773 (inline_generate_summary): Update.
15774 (inline_read_section): Update.
15775 (inline_read_summary): Update.
15776 (inline_free_summary): Update.
15777 * ipa-inline.c (can_inline_edge_p): Update.
15778 (compute_inlined_call_time): Update.
15779 (want_inline_small_function_p): Update.
15780 (edge_badness): Update.
15781 (early_inliner): Update.
15782 * ipa-inline.h (inline_edge_summary): Turn to ...
15783 (ipa_call_summary): ... this one.
15784 (ipa_call_summary_t): New class.
15785 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
15786 (ipa_call_summaries): New.
15787 (inline_edge_summary): Remove.
15788 (estimate_edge_growth): Update.
15789 * ipa-profile.c (ipa_propagate_frequency_1): Update.
15790 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15791 * ipa-split.c (execute_split_functions): Update.
15792 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
15794 2017-05-23 Tom de Vries <tom@codesourcery.com>
15796 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
15797 attributes): Document rdrand effective target.
15799 2017-05-23 Tom de Vries <tom@codesourcery.com>
15801 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
15802 attributes): Sort alphabetically.
15804 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
15806 * config/avr/genmultilib.awk: Use gsub instead of gensub.
15808 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
15811 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
15812 V2DF/V2DI splat into two separate patterns, one that handles
15813 registers, and the other that only handles memory. Drop support
15814 for splatting from a GPR on ISA 2.07 and then splitting the
15815 splat into direct move and splat.
15816 (vsx_splat_<mode>_reg): Likewise.
15817 (vsx_splat_<mode>_mem): Likewise.
15819 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
15821 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
15823 2017-05-22 Jakub Jelinek <jakub@redhat.com>
15825 PR middle-end/80809
15826 * omp-low.c (finish_taskreg_remap): New function.
15827 (finish_taskreg_scan): If unit size of ctx->record_type
15828 is non-constant, unshare the size expression and replace
15829 decls in it with possible outer var refs.
15831 PR middle-end/80809
15832 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
15833 GOVD_SHARED rather than GOVD_PRIVATE with it.
15834 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
15835 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
15837 PR middle-end/80853
15838 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
15839 as last argument to build_outer_var_ref for pointer bases of array
15840 section reductions.
15842 2017-05-19 Martin Sebor <msebor@redhat.com>
15844 * print-tree.c (print_node): Print DECL_READ_P flag.
15846 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15848 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
15849 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
15850 * cgraph.c: Likewise.
15851 * cgraphunit.c: Likewise.
15852 * gengtype.c: Likewise.
15853 * ipa-cp.c: Likewise.
15854 * ipa-devirt.c: Likewise.
15855 * ipa-icf.c: Likewise.
15856 * ipa-predicate.c: Likewise.
15857 * ipa-profile.c: Likewise.
15858 * ipa-prop.c: Likewise.
15859 * ipa-split.c: Likewise.
15861 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
15862 edge_predicate_pool, dump_inline_hints,
15863 inline_summary::account_size_time, redirect_to_unreachable,
15864 edge_set_predicate, set_hint_predicate,
15865 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
15866 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
15867 inline_summary_t::remove, remap_hint_predicate_after_duplication,
15868 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
15869 ipa_call_summary_t::remove, initialize_growth_caches,
15870 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
15871 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
15872 mark_modified, unmodified_parm_1, unmodified_parm,
15873 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
15874 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
15875 compute_bb_predicates, will_be_nonconstant_expr_predicate,
15876 will_be_nonconstant_predicate, record_modified_bb_info,
15877 get_minimal_bb, record_modified, param_change_prob,
15878 phi_result_unknown_predicate, predicate_for_phi_result,
15879 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
15880 estimate_function_body_sizes, compute_inline_parameters,
15881 compute_inline_parameters_for_curren, pass_data_inline_parameters,
15882 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
15883 inline_update_callee_summaries, remap_edge_change_prob,
15884 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
15885 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
15886 inline_analyze_function, inline_summary_t::insert,
15887 inline_generate_summary, read_ipa_call_summary, inline_read_section,
15888 inline_read_summary, write_ipa_call_summary, inline_write_summary,
15889 inline_free_summary): Move to ipa-fnsummary.h
15890 (predicate_t): Remove.
15891 * ipa-fnsummary.c: New file.
15892 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
15893 (enum inline_hints_vals, inline_hints, agg_position_info,
15894 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
15895 inline_summaries, ipa_call_summary, ipa_call_summary_t,
15896 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
15897 dump_inline_summary, dump_inline_hints, inline_generate_summary,
15898 inline_read_summary, inline_write_summary, inline_free_summary,
15899 inline_analyze_function, initialize_inline_failed,
15900 inline_merge_summary, inline_update_overall_summary,
15901 compute_inline_parameters): Move to ipa-fnsummary.h
15902 * ipa-fnsummary.h: New file.
15903 * ipa-inline-transform.h: Include ipa-inline.h.
15904 * ipa-inline.c: LIkewise.
15906 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15908 * ipa-inline.c (edge_badness): Use inlined_time instead of
15909 inline_summaries->get.
15911 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15913 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
15915 2017-05-22 Nathan Sidwell <nathan@acm.org>
15917 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
15918 (fdump-lang): Document 'raw' option.
15919 * dumpfile.h (TDI_tu): Delete.
15920 * dumpfile.c (dump_files): Remove translation-unit.
15921 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
15923 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
15925 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
15926 command option from $(AWK) call.
15927 * config/avr/genmultilib.awk: Simplify and rewrite so that it
15928 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
15929 [FORMAT]: Remove handling of variable.
15930 * config/avr/t-multilib: Regenerate.
15932 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15934 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
15936 (dump_inline_summary): Do not print self_time.
15937 (estimate_function_body_sizes): Do not set self_time.
15938 (compute_inline_parameters): Likewise.
15939 (inline_read_section, inline_write_summary): Do not stream self_time.
15940 * ipa-inline.h (inline_summary): Drop self_time.
15942 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15944 * ipa-inline-analysis.c (account_size_time): Rename to ...
15945 (inline_summary::account_size_time): ... this one.
15946 (reset_ipa_call_summary): Turn to ...
15947 (ipa_call_summary::reset): ... this one.
15948 (reset_inline_summary): Turn to ...
15949 (inline_summary::reset): ... this one.
15950 (inline_summary_t::remove): Update.
15951 (inline_summary_t::duplicate): Update.
15952 (ipa_call_summary_t::remove): Update.
15953 (dump_inline_summary): Update.
15954 (estimate_function_body_sizes): Update.
15955 (compute_inline_parameters): Update.
15956 (estimate_node_size_and_time): Update.
15957 (inline_merge_summary): Update.
15958 (inline_update_overall_summary): Update.
15959 (inline_read_section): Update.
15960 (inline_write_summary): Update.
15961 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
15962 add account_size_time and reset member functions.
15963 (ipa_call_summary): Add reset function.
15964 * ipa-predicate.h (predicate::operator &): Constify.
15966 2017-05-22 Richard Biener <rguenther@suse.de>
15968 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
15970 2017-05-19 Jason Merrill <jason@redhat.com>
15972 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
15974 2017-05-19 Marek Polacek <polacek@redhat.com>
15977 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
15978 TYPE_OVERFLOW_WRAPS checks.
15980 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
15982 * tree-core.h (enum omp_clause_default_kind): Add
15983 "OMP_CLAUSE_DEFAULT_PRESENT".
15984 * tree-pretty-print.c (dump_omp_clause): Handle it.
15985 * gimplify.c (enum gimplify_omp_var_data): Add
15986 "GOVD_MAP_FORCE_PRESENT".
15987 (gimplify_adjust_omp_clauses_1): Map it to
15988 "GOMP_MAP_FORCE_PRESENT".
15989 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
15991 * gimplify.c (oacc_default_clause): Clarify.
15993 2017-05-19 Nathan Sidwell <nathan@acm.org>
15995 LANG_HOOK_REGISTER_DUMPS
15996 * toplev.c (general_init): Call register dump lang hook.
15997 * doc/invoke.texi: Document -fdump-lang option family.
15998 * dumpfile.c (dump_files): Remove class dump here.
15999 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
16000 * dumpfile.h (tree_dump_index): Remove TDI_class.
16001 * langhooks-def.h (lhd_register_dumps): Declare.
16002 (LANG_HOOKS_REGISTER_DUMPS): Define.
16003 (LANG_HOOKS_INITIALIZER): Add it.
16004 * langhooks.c (lhd_register_dumps): Define.
16005 * langhooks.h (struct lang_hooks): Add register_dumps.
16007 2017-05-19 Nathan Sidwell <nathan@acm.org>
16009 * context.h (context::set_passes): New.
16010 * context.c (context::context): Do not create pass manager.
16011 * toplev.c (general_init): Create pass manager here.
16013 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
16015 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
16016 use this splitter if two add or or instructions would also work for
16017 the constant we want to generate.
16019 2017-05-19 Richard Biener <rguenther@suse.de>
16022 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
16023 predicate evaluation.
16025 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
16027 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
16029 * ipa-inline.c (want_inline_small_function_p): Do not cast to
16032 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
16034 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
16035 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
16036 (inline_edge_summary_vec): Turn into ...
16037 (ipa_call_summaries): ... this one.
16038 (redirect_to_unreachable, edge_set_predicate,
16039 evaluate_properties_for_edge, inline_summary_alloc,
16040 reset_ipa_call_summary, reset_inline_summary,
16041 inline_summary_t::duplicate): Update.
16042 (inline_edge_duplication_hook): Turn to ...
16043 (ipa_call_summary_t::duplicate): ... this one.
16044 (inline_edge_removal_hook): Turn to ...
16045 (ipa_call_summary_t::remove): ... this one.
16046 (dump_inline_edge_summary): Turn to ...
16047 (dump_ipa_call_summary): ... this one.
16048 (estimate_function_body_sizes): Update.
16049 (inline_update_callee_summaries): Update.
16050 (remap_edge_change_prob): Update.
16051 (remap_edge_summaries): Update.
16052 (inline_merge_summary): Update.
16053 (do_estimate_edge_time): Update.
16054 (inline_generate_summary): Update.
16055 (inline_read_section): Update.
16056 (inline_read_summary): Update.
16057 (inline_free_summary): Update.
16058 * ipa-inline.c (can_inline_edge_p): Update.
16059 (compute_inlined_call_time): Update.
16060 (want_inline_small_function_p): Update.
16061 (edge_badness): Update.
16062 (early_inliner): Update.
16063 * ipa-inline.h (inline_edge_summary): Turn to ...
16064 (ipa_call_summary): ... this one.
16065 (ipa_call_summary_t): New class.
16066 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
16067 (ipa_call_summaries): New.
16068 (inline_edge_summary): Remove.
16069 (estimate_edge_growth): Update.
16070 * ipa-profile.c (ipa_propagate_frequency_1): Update.
16071 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
16072 * ipa-split.c (execute_split_functions): Update.
16073 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
16075 2017-05-19 Richard Biener <rguenther@suse.de>
16077 PR middle-end/80764
16078 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
16080 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
16082 * config/rs6000/rs6000.c (struct machine_function): Add field
16083 fpr_is_wrapped_separately.
16084 (rs6000_get_separate_components): Use 64 components. Handle the
16085 new FPR components.
16086 (rs6000_components_for_bb): Handle the FPR components.
16087 (rs6000_emit_prologue_components): Handle the FPR components.
16088 (rs6000_emit_epilogue_components): Handle the FPR components.
16089 (rs6000_set_handled_components): Handle the FPR components.
16090 (rs6000_emit_prologue): Don't output prologue code for those FPRs
16091 that are already separately shrink-wrapped.
16092 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
16093 that are already separately shrink-wrapped.
16095 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
16098 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
16101 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
16102 (define_peephole2 for Altivec d-form load): Add peepholes to catch
16103 cases where the register allocator uses a move and an offsettable
16104 memory operation to/from a FPR register on ISA 2.06/2.07.
16105 (define_peephole2 for Altivec d-form store): Likewise.
16107 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
16110 * config/i386/mmx.md (*mov<mode>_internal): Enable
16111 alternatives 11, 12, 13 and 14 also for 32bit targets.
16112 Remove alternatives 15, 16, 17 and 18.
16113 * config/i386/sse.md (vec_concatv2di): Change
16114 alternative (!x, *y) to (x, ?!*Yn).
16116 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
16118 * dumpfile.h (enum dump_kind): Remove stray comma.
16120 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
16122 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
16123 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
16124 predicate::num_conditions
16125 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
16126 (CHANGED): turn into predicate::changed.
16127 (agg_position_info): Move to ipa-predicate.h
16128 (add_condition, predicate::add_clause, predicate::operator &=,
16129 predicate::or_with, predicate::evaluate, predicate::probability,
16130 dump_condition, dump_clause, predicate::dump,
16131 predicate::remap_after_duplication, predicate::remap_after_inlining,
16132 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
16133 (evaluate_conditions_for_known_args): Update.
16134 (set_cond_stmt_execution_predicate): Update.
16135 * ipa-inline.h: Include ipa-predicate.h
16136 (condition, inline_param_summary, conditions, agg_position_info,
16137 predicate): Move to ipa-predicate.h
16138 * ipa-predicate.c: New file.
16139 * ipa-predicate.h: New file.
16141 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
16143 * final.c (leaf_function_p): Check we are not in a sequence.
16145 2017-05-18 Martin Liska <mliska@suse.cz>
16147 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
16148 * dumpfile.c (dump_register): Use new enum dump_kind.
16149 (get_dump_file_name): Likewise.
16150 (dump_enable_all): Likewise.
16151 (dump_switch_p_1): Likewise.
16152 (enable_rtl_dump_file): Remove usage of TDF_RTL.
16153 * dumpfile.h (enum dump_kind): New enum type.
16154 (struct dump_file_info): Create constructor and
16155 format fields and comments.
16156 * passes.c (pass_manager::register_one_dump_file):
16158 * statistics.c (statistics_early_init): Likewise.
16159 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
16160 TDF_TREE with TDF_SLIM.
16161 (gather_memory_references_ref): Likewise.
16163 2017-05-18 Martin Liska <mliska@suse.cz>
16165 * vec.h (struct vnull): Use it.
16167 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
16169 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
16170 (true_predicate, false_predicate, true_predicate_p,
16171 false_predicate_p): Remove.
16172 (single_cond_predicate, not_inlined_predicate): Turn to member function
16174 (add_condition): Update.
16175 (add_clause): Turn to...
16176 (predicate::add_clause): ... this one; update; allow passing NULL
16178 (and_predicates): Turn to ...
16179 (predicate::operator &=): ... this one.
16180 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
16181 (or_predicates): Turn to ...
16182 (predicate::or_with): ... this one.
16183 (evaluate_predicate): Turn to ...
16184 (predicate::evaluate): ... this one.
16185 (predicate_probability): Turn to ...
16186 (predicate::probability): ... this one.
16187 (dump_condition): Update.
16188 (dump_predicate): Turn to ...
16189 (predicate::dump): ... this one.
16190 (account_size_time): Update.
16191 (edge_set_predicate): Update.
16192 (set_hint_predicate): UPdate.
16193 (evaluate_conditions_for_known_args): Update.
16194 (evaluate_properties_for_edge): Update.
16195 (remap_predicate_after_duplication): Turn to...
16196 (predicate::remap_after_duplication): ... this one.
16197 (remap_hint_predicate_after_duplication): Update.
16198 (inline_summary_t::duplicate): UPdate.
16199 (dump_inline_edge_summary): Update.
16200 (dump_inline_summary): Update.
16201 (set_cond_stmt_execution_predicate): Update.
16202 (set_switch_stmt_execution_predicate): Update.
16203 (compute_bb_predicates): Update.
16204 (will_be_nonconstant_expr_predicate): Update.
16205 (will_be_nonconstant_predicate): Update.
16206 (phi_result_unknown_predicate): Update.
16207 (predicate_for_phi_result): Update.
16208 (array_index_predicate): Update.
16209 (estimate_function_body_sizes): Update.
16210 (estimate_node_size_and_time): Update.
16211 (estimate_ipcp_clone_size_and_time): Update.
16212 (remap_predicate): Rename to ...
16213 (predicate::remap_after_inlining): ... this one.
16214 (remap_hint_predicate): Update.
16215 (inline_merge_summary): Update.
16216 (inline_update_overall_summary): Update.
16217 (estimate_size_after_inlining): Update.
16218 (read_predicate): Rename to ...
16219 (predicate::stream_in): ... this one.
16220 (read_inline_edge_summary): Update.
16221 (write_predicate): Rename to ...
16222 (predicate::stream_out): ... this one.
16223 (write_inline_edge_summary): Update.
16224 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
16225 (clause_t): Turn to uint32_t
16226 (predicate): Turn to class; implement constructor and operators
16228 (size_time_entry): Update.
16229 (inline_summary): Update.
16230 (inline_edge_summary): Update.
16232 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
16234 * fold-const.c (fold_binary_loc): Move transformation...
16235 * match.pd (C - X CMP X): ... here.
16237 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
16239 * config/sparc/sparc.c (sparc_option_override): Set function
16240 alignment for -mcpu=niagara7 to 64 to match the I$ line.
16241 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
16243 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
16245 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
16247 2017-05-18 Marek Polacek <polacek@redhat.com>
16250 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
16251 (pass_ubsan::execute): Call gimple_assign_single_p instead of
16252 gimple_assign_load_p.
16254 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
16256 PR middle-end/80692
16257 * real.c (do_compare): Give decimal_do_compare preference over
16258 comparing just the signs.
16260 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
16262 * doc/md.texi (Canonicalization of Instructions): Describe the
16263 canonical form of instructions that inherently set a condition
16266 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
16268 PR middle-end/80775
16269 * tree-cfg.c: Move deletion of unreachable case statements to after
16270 the merging of consecutive case labels.
16272 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
16274 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
16275 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
16276 restoring of callee-saved registers.
16278 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
16280 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
16281 * config/visium/visium.c (single_set_and_flags): Likewise.
16282 * config/visium/visium.md (Substitutions): Likewise.
16284 2017-05-17 Martin Liska <mliska@suse.cz>
16286 * cfg.c: Introduce dump_flags_t type and
16287 use it instead of int type.
16289 * cfghooks.c: Likewise.
16290 * cfghooks.h (struct cfg_hooks): Likewise.
16291 * cfgrtl.c: Likewise.
16292 * cfgrtl.h: Likewise.
16293 * cgraph.c (cgraph_node::get_body): Likewise.
16294 * coretypes.h: Likewise.
16295 * domwalk.c: Likewise.
16296 * domwalk.h: Likewise.
16297 * dumpfile.c (struct dump_option_value_info): Likewise.
16298 (dump_enable_all): Likewise.
16299 (dump_switch_p_1): Likewise.
16300 (opt_info_switch_p): Likewise.
16301 * dumpfile.h (enum tree_dump_index): Likewise.
16302 (struct dump_file_info): Likewise.
16303 * genemit.c: Likewise.
16304 * generic-match-head.c: Likewise.
16305 * gengtype.c (open_base_files): Likewise.
16306 * gimple-pretty-print.c: Likewise.
16307 * gimple-pretty-print.h: Likewise.
16308 * graph.c (print_graph_cfg): Likewise.
16309 * graphite-scop-detection.c (dot_all_sese): Likewise.
16310 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
16311 * loop-unroll.c (report_unroll): Likewise.
16312 * passes.c (pass_manager::register_one_dump_file): Likewise.
16313 * print-tree.c: Likewise.
16314 * statistics.c: Likewise.
16315 * tree-cfg.c: Likewise.
16316 * tree-cfg.h: Likewise.
16317 * tree-dfa.c: Likewise.
16318 * tree-dfa.h: Likewise.
16319 * tree-dump.c (dump_function): Likewise.
16320 * tree-dump.h (struct dump_info): Likewise.
16321 * tree-pretty-print.c: Likewise.
16322 * tree-pretty-print.h: Likewise.
16323 * tree-ssa-live.c: Likewise.
16324 * tree-ssa-live.h: Likewise.
16325 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
16326 * tree-vect-loop.c: Likewise.
16327 * tree-vect-slp.c: Likewise.
16329 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
16330 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16332 PR tree-optimization/80457
16333 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
16334 of all arguments to a statement as scalar_to_vec operations.
16335 (vectorizable_call): Adjust call to vect_model_simple_cost for
16337 (vectorizable_conversion): Likewise.
16338 (vectorizable_assignment): Likewise.
16339 (vectorizable_shift): Likewise.
16340 (vectorizable_operation): Likewise.
16341 (vectorizable_comparison): Likewise.
16342 (vect_is_simple_cond): Record the def types for operands.
16343 (vectorizable_condition): Likewise, call vect_model_simple_cost.
16344 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
16345 for statement argument count.
16347 2017-05-16 Carl Love <cel@us.ibm.com>
16349 * config/rs6000/rs6000-c: Add support for built-in functions
16350 vector unsigned long long vec_bperm (vector unsigned long long,
16351 vector unsigned char)
16352 vector signed long long vec_mule (vector signed int,
16354 vector unsigned long long vec_mule (vector unsigned int,
16355 vector unsigned int)
16356 vector signed long long vec_mulo (vector signed int,
16358 vector unsigned long long vec_mulo (vector unsigned int,
16359 vector unsigned int)
16360 vector signed char vec_sldw (vector signed char,
16361 vector signed char,
16363 vector unsigned char vec_sldw (vector unsigned char,
16364 vector unsigned char,
16366 vector signed short vec_sldw (vector signed short,
16367 vector signed short,
16369 vector unsigned short vec_sldw (vector unsigned short,
16370 vector unsigned short,
16372 vector signed int vec_sldw (vector signed int,
16375 vector unsigned int vec_sldw (vector unsigned int,
16376 vector unsigned int,
16378 vector signed long long vec_sldw (vector signed long long,
16379 vector signed long long,
16381 vector unsigned long long vec_sldw (vector unsigned long long,
16382 vector unsigned long long,
16384 * config/rs6000/rs6000-c: Add support for built-in functions
16385 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
16386 * config/rs6000/altivec.h: Add defintion for vec_sldw.
16387 * doc/extend.texi: Update the built-in documentation for the
16388 new built-in functions.
16390 2017-05-16 Marek Polacek <polacek@redhat.com>
16394 * tree.c (save_expr): Don't fold the expression.
16396 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
16398 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
16399 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
16400 and (?*y,m). Update insn attributes.
16402 2017-05-16 Martin Liska <mliska@suse.cz>
16404 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
16405 flags argument of print_gimple_stmt, print_gimple_expr,
16406 print_generic_stmt and print_generic_expr.
16407 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
16408 * coretypes.h: Likewise.
16409 * except.c (dump_eh_tree): Likewise.
16410 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
16411 * gimple-pretty-print.h: Likewise.
16412 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
16413 (backprop::push_to_worklist): Likewise.
16414 (backprop::pop_from_worklist): Likewise.
16415 (backprop::process_use): Likewise.
16416 (backprop::intersect_uses): Likewise.
16417 (note_replacement): Likewise.
16418 * gimple-ssa-store-merging.c
16419 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
16420 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
16421 (pass_store_merging::execute): Likewise.
16422 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
16423 (ssa_base_cand_dump_callback): Likewise.
16424 (dump_incr_vec): Likewise.
16425 (replace_refs): Likewise.
16426 (replace_mult_candidate): Likewise.
16427 (create_add_on_incoming_edge): Likewise.
16428 (create_phi_basis): Likewise.
16429 (insert_initializers): Likewise.
16430 (all_phi_incrs_profitable): Likewise.
16431 (introduce_cast_before_cand): Likewise.
16432 (replace_one_candidate): Likewise.
16433 * gimplify.c (gimplify_expr): Likewise.
16434 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
16435 (set_rename): Likewise.
16436 (rename_uses): Likewise.
16437 (copy_loop_phi_nodes): Likewise.
16438 (add_close_phis_to_merge_points): Likewise.
16439 (copy_loop_close_phi_args): Likewise.
16440 (copy_cond_phi_args): Likewise.
16441 (graphite_copy_stmts_from_block): Likewise.
16442 (translate_pending_phi_nodes): Likewise.
16443 * graphite-poly.c (print_pdr): Likewise.
16444 (dump_gbb_cases): Likewise.
16445 (dump_gbb_conditions): Likewise.
16446 (print_scop_params): Likewise.
16447 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
16448 (build_cross_bb_scalars_use): Likewise.
16449 (gather_bbs::before_dom_children): Likewise.
16450 * hsa-dump.c (dump_hsa_immed): Likewise.
16451 * ipa-cp.c (print_ipcp_constant_value): Likewise.
16452 (get_replacement_map): Likewise.
16453 * ipa-inline-analysis.c (dump_condition): Likewise.
16454 (estimate_function_body_sizes): Likewise.
16455 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
16456 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
16457 * ipa-prop.c (ipa_dump_param): Likewise.
16458 (ipa_print_node_jump_functions_for_edge): Likewise.
16459 (ipa_modify_call_arguments): Likewise.
16460 (ipa_modify_expr): Likewise.
16461 (ipa_dump_param_adjustments): Likewise.
16462 (ipa_dump_agg_replacement_values): Likewise.
16463 (ipcp_modif_dom_walker::before_dom_children): Likewise.
16464 * ipa-pure-const.c (check_stmt): Likewise.
16465 (pass_nothrow::execute): Likewise.
16466 * ipa-split.c (execute_split_functions): Likewise.
16467 * omp-offload.c (dump_oacc_loop_part): Likewise.
16468 (dump_oacc_loop): Likewise.
16469 * trans-mem.c (tm_log_emit): Likewise.
16470 (tm_memopt_accumulate_memops): Likewise.
16471 (dump_tm_memopt_set): Likewise.
16472 (dump_tm_memopt_transform): Likewise.
16473 * tree-cfg.c (gimple_verify_flow_info): Likewise.
16474 (print_loop): Likewise.
16475 * tree-chkp-opt.c (chkp_print_addr): Likewise.
16476 (chkp_gather_checks_info): Likewise.
16477 (chkp_get_check_result): Likewise.
16478 (chkp_remove_check_if_pass): Likewise.
16479 (chkp_use_outer_bounds_if_possible): Likewise.
16480 (chkp_reduce_bounds_lifetime): Likewise.
16481 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
16482 (chkp_mark_completed_bounds): Likewise.
16483 (chkp_register_incomplete_bounds): Likewise.
16484 (chkp_mark_invalid_bounds): Likewise.
16485 (chkp_maybe_copy_and_register_bounds): Likewise.
16486 (chkp_build_returned_bound): Likewise.
16487 (chkp_get_bound_for_parm): Likewise.
16488 (chkp_build_bndldx): Likewise.
16489 (chkp_get_bounds_by_definition): Likewise.
16490 (chkp_generate_extern_var_bounds): Likewise.
16491 (chkp_get_bounds_for_decl_addr): Likewise.
16492 * tree-chrec.c (chrec_apply): Likewise.
16493 * tree-data-ref.c (dump_data_reference): Likewise.
16494 (dump_subscript): Likewise.
16495 (dump_data_dependence_relation): Likewise.
16496 (analyze_overlapping_iterations): Likewise.
16497 * tree-inline.c (expand_call_inline): Likewise.
16498 (tree_function_versioning): Likewise.
16499 * tree-into-ssa.c (dump_defs_stack): Likewise.
16500 (dump_currdefs): Likewise.
16501 (dump_names_replaced_by): Likewise.
16502 (dump_update_ssa): Likewise.
16503 (update_ssa): Likewise.
16504 * tree-object-size.c (pass_object_sizes::execute): Likewise.
16505 * tree-parloops.c (build_new_reduction): Likewise.
16506 (try_create_reduction_list): Likewise.
16507 (ref_conflicts_with_region): Likewise.
16508 (oacc_entry_exit_ok_1): Likewise.
16509 (oacc_entry_exit_single_gang): Likewise.
16510 * tree-pretty-print.h: Likewise.
16511 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
16512 (get_scalar_evolution): Likewise.
16513 (add_to_evolution): Likewise.
16514 (get_loop_exit_condition): Likewise.
16515 (analyze_evolution_in_loop): Likewise.
16516 (analyze_initial_condition): Likewise.
16517 (analyze_scalar_evolution): Likewise.
16518 (instantiate_scev): Likewise.
16519 (number_of_latch_executions): Likewise.
16520 (gather_chrec_stats): Likewise.
16521 (final_value_replacement_loop): Likewise.
16522 (scev_const_prop): Likewise.
16523 * tree-sra.c (dump_access): Likewise.
16524 (disqualify_candidate): Likewise.
16525 (create_access): Likewise.
16526 (reject): Likewise.
16527 (maybe_add_sra_candidate): Likewise.
16528 (create_access_replacement): Likewise.
16529 (analyze_access_subtree): Likewise.
16530 (analyze_all_variable_accesses): Likewise.
16531 (sra_modify_assign): Likewise.
16532 (initialize_constant_pool_replacements): Likewise.
16533 (find_param_candidates): Likewise.
16534 (decide_one_param_reduction): Likewise.
16535 (replace_removed_params_ssa_names): Likewise.
16536 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
16537 * tree-ssa-copy.c (dump_copy_of): Likewise.
16538 (copy_prop_visit_cond_stmt): Likewise.
16539 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
16540 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
16541 (record_equivalences_from_stmt): Likewise.
16542 * tree-ssa-dse.c (compute_trims): Likewise.
16543 (delete_dead_call): Likewise.
16544 (delete_dead_assignment): Likewise.
16545 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
16546 (forward_propagate_into_cond): Likewise.
16547 (pass_forwprop::execute): Likewise.
16548 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16549 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
16551 (move_computations_worker): Likewise.
16552 (execute_sm): Likewise.
16553 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
16554 (remove_exits_and_undefined_stmts): Likewise.
16555 (remove_redundant_iv_tests): Likewise.
16556 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
16557 (adjust_iv_update_pos): Likewise.
16558 * tree-ssa-math-opts.c (bswap_replace): Likewise.
16559 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
16560 (value_replacement): Likewise.
16561 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
16562 * tree-ssa-pre.c (print_pre_expr): Likewise.
16563 (get_representative_for): Likewise.
16564 (create_expression_by_pieces): Likewise.
16565 (insert_into_preds_of_block): Likewise.
16566 (eliminate_insert): Likewise.
16567 (eliminate_dom_walker::before_dom_children): Likewise.
16568 (eliminate): Likewise.
16569 (remove_dead_inserted_code): Likewise.
16570 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
16571 * tree-ssa-reassoc.c (get_rank): Likewise.
16572 (eliminate_duplicate_pair): Likewise.
16573 (eliminate_plus_minus_pair): Likewise.
16574 (eliminate_not_pairs): Likewise.
16575 (undistribute_ops_list): Likewise.
16576 (eliminate_redundant_comparison): Likewise.
16577 (update_range_test): Likewise.
16578 (optimize_range_tests_var_bound): Likewise.
16579 (optimize_vec_cond_expr): Likewise.
16580 (rewrite_expr_tree): Likewise.
16581 (rewrite_expr_tree_parallel): Likewise.
16582 (linearize_expr): Likewise.
16583 (break_up_subtract): Likewise.
16584 (linearize_expr_tree): Likewise.
16585 (attempt_builtin_powi): Likewise.
16586 (attempt_builtin_copysign): Likewise.
16587 (transform_stmt_to_copy): Likewise.
16588 (transform_stmt_to_multiply): Likewise.
16589 (dump_ops_vector): Likewise.
16590 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
16591 (print_scc): Likewise.
16592 (set_ssa_val_to): Likewise.
16593 (visit_reference_op_store): Likewise.
16594 (visit_use): Likewise.
16595 (sccvn_dom_walker::before_dom_children): Likewise.
16596 (run_scc_vn): Likewise.
16597 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
16599 (expr_hash_elt::print): Likewise.
16600 (const_and_copies::pop_to_marker): Likewise.
16601 (const_and_copies::record_const_or_copy_raw): Likewise.
16602 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
16603 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
16604 (dump_predicates): Likewise.
16605 (find_uninit_use): Likewise.
16606 (warn_uninitialized_phi): Likewise.
16607 (pass_late_warn_uninitialized::execute): Likewise.
16608 * tree-ssa.c (verify_vssa): Likewise.
16609 (verify_ssa): Likewise.
16610 (maybe_optimize_var): Likewise.
16611 * tree-vrp.c (dump_value_range): Likewise.
16612 (dump_all_value_ranges): Likewise.
16613 (dump_asserts_for): Likewise.
16614 (register_edge_assert_for_2): Likewise.
16615 (vrp_visit_cond_stmt): Likewise.
16616 (vrp_visit_switch_stmt): Likewise.
16617 (vrp_visit_stmt): Likewise.
16618 (vrp_visit_phi_node): Likewise.
16619 (simplify_cond_using_ranges_1): Likewise.
16620 (fold_predicate_in): Likewise.
16621 (evrp_dom_walker::before_dom_children): Likewise.
16622 (evrp_dom_walker::push_value_range): Likewise.
16623 (evrp_dom_walker::pop_value_range): Likewise.
16624 (execute_early_vrp): Likewise.
16626 2017-05-16 Richard Biener <rguenther@suse.de>
16628 * dwarf2out.c (loc_list_from_tree_1): Do not create
16629 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
16631 2017-05-16 Richard Biener <rguenther@suse.de>
16633 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
16635 (note_variable_value_in_expr): If we resolved the decl ref
16636 do not push to the stack.
16638 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
16640 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
16641 operations in fast-math mode.
16642 (vaddq_f16): Likewise.
16643 (vmul_f16): Likewise.
16644 (vmulq_f16): Likewise.
16645 (vsub_f16): Likewise.
16646 (vsubq_f16): Likewise.
16647 * config/arm/neon.md (add<mode>3): New.
16649 (fma:<VH:mode>3): New. Also remove outdated comment.
16652 2017-05-16 Martin Liska <mliska@suse.cz>
16656 * ipa-devirt.c (warn_types_mismatch): Fix typo.
16657 (odr_types_equivalent_p): Likewise.
16659 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
16661 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
16663 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
16666 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
16667 non-interunit SSE move alternatives with '?'.
16668 (zero-extendsidi peephole2): New peephole to skip intermediate
16669 general register in SSE zero-extend sequence.
16671 2017-05-15 Jeff Law <law@redhat.com>
16673 * reorg.c (relax_delay_slots): Create a new variable to hold
16674 the temporary target rather than clobbering TARGET_LABEL.
16676 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
16677 missing argument to extract_bit_field call.
16678 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
16680 2017-05-15 Martin Liska <mliska@suse.cz>
16683 * gcc.c (process_command): Do not allow empty argument of -o option.
16685 2017-05-15 Renlin Li <renlin.li@arm.com>
16687 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
16688 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
16689 * config/aarch64/constraints.md (Usf): Add long call check.
16690 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
16691 (call_value): Likewise.
16692 (sibcall): Likewise.
16693 (sibcall_value): Likewise.
16695 (call_value_insn): New.
16696 (sibcall_insn): Update rtx pattern.
16697 (sibcall_value_insn): Likewise.
16698 (call_internal): Remove.
16699 (call_value_internal): Likewise.
16700 (sibcall_internal): Likewise.
16701 (sibcall_value_internal): Likewise.
16702 (call_reg): Likewise.
16703 (call_symbol): Likewise.
16704 (call_value_reg): Likewise.
16705 (call_value_symbol): Likewise.
16707 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
16710 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
16712 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
16714 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
16715 compatible with CCGOCmode and with CCZmode.
16717 2017-05-14 Martin Sebor <msebor@redhat.com>
16719 PR middle-end/77671
16720 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
16721 (gimple_fold_builtin_snprintf): Same.
16722 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
16723 (gimple_fold_builtin_snprintf): Same.
16724 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
16725 of character types.
16726 (is_call_safe): New function.
16727 (try_substitute_return_value): Call it.
16728 (try_simplify_call): New function.
16729 (pass_sprintf_length::handle_gimple_call): Call it.
16731 2017-05-14 Martin Sebor <msebor@redhat.com>
16733 PR middle-end/80669
16734 * builtins.c (expand_builtin_stpncpy): Simplify.
16736 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
16738 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
16739 * config/i386/i386.h
16740 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
16741 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
16742 (struct machine_function): Add new members call_ms2sysv,
16743 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
16744 (struct machine_frame_state): New fields sp_realigned and
16745 sp_realigned_offset.
16746 * config/i386/i386.c
16747 (enum xlogue_stub): New enum.
16748 (enum xlogue_stub_sets): New enum.
16749 (class xlogue_layout): New class.
16750 (struct ix86_frame): New fields stack_realign_allocate_offset,
16751 stack_realign_offset and outlined_save_offset. Modify comments to
16752 detail stack layout when using out-of-line stubs.
16753 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
16754 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
16755 -mcall-ms2sysv-xlogues.
16756 (stub_managed_regs): New static variable.
16757 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
16758 registers managed by out-of-line stub.
16759 (disable_call_ms2sysv_xlogues): New function.
16760 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
16761 m->call_ms2sysv when appropriate and compute frame layout for
16763 (sp_valid_at, fp_valid_at): New inline functions.
16764 (choose_basereg): New function.
16765 (choose_baseaddr): Add align parameter, use choose_basereg and modify
16767 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
16768 Use align parameter of choose_baseaddr to generated aligned SSE movs
16770 (pro_epilogue_adjust_stack): Modify to track
16771 machine_frame_state::sp_realigned.
16772 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
16773 (ix86_nsaved_sseregs): Likewise.
16774 (ix86_emit_save_regs): Likewise.
16775 (ix86_emit_save_regs_using_mov): Likewise.
16776 (ix86_emit_save_sse_regs_using_mov): Likewise.
16777 (get_scratch_register_on_entry): Likewise.
16778 (gen_frame_set): New function.
16779 (gen_frame_load): Likewise.
16780 (gen_frame_store): Likewise.
16781 (emit_outlined_ms2sysv_save): Likewise.
16782 (emit_outlined_ms2sysv_restore): Likewise.
16783 (ix86_expand_prologue): Modify stack re-alignment code and call
16784 emit_outlined_ms2sysv_save when appropriate.
16785 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
16786 parameter rtx_insn *insn, which allows the function to be used to only
16787 generate the notes.
16788 (ix86_expand_epilogue): Modify validity checks of frame and stack
16789 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
16790 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
16791 * config/i386/predicates.md
16792 (save_multiple): New predicate.
16793 (restore_multiple): Likewise.
16794 * config/i386/sse.md
16795 (save_multiple<mode>): New pattern.
16796 (save_multiple_realign<mode>): Likewise.
16797 (restore_multiple<mode>): Likewise.
16798 (restore_multiple_and_return<mode>): Likewise.
16799 (restore_multiple_leave_return<mode>): Likewise.
16800 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
16802 2017-05-14 Julia Koval <julia.koval@intel.com>
16804 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
16805 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
16806 (__builtin_ia32_xsetbv): New builtins.
16807 * config/i386/i386.c (ix86_expand_special_args_builtin):
16809 (ix86_expand_builtin): Special expand for new intrinsics.
16810 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
16811 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
16812 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
16814 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16816 * cfganal.c (inverted_post_order_compute): Change argument type
16818 * cfganal.h (inverted_post_order_compute): Adjust prototype.
16819 * df-core.c (rest_of_handle_df_initialize): Adjust.
16820 (rest_of_handle_df_finish): Likewise.
16821 (df_analyze_1): Likewise.
16822 (df_analyze): Likewise.
16823 (loop_inverted_post_order_compute): Change argument to be a vec *.
16824 (df_analyze_loop): Adjust.
16825 (df_get_n_blocks): Likewise.
16826 (df_get_postorder): Likewise.
16827 * df.h (struct df_d): Change field to be a vec.
16828 * lcm.c (compute_laterin): Adjust.
16829 (compute_available): Likewise.
16830 * lra-lives.c (lra_create_live_ranges_1): Likewise.
16831 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
16832 * tree-ssa-pre.c (compute_antic): Likewise.
16834 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16836 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
16837 (depth_first_search::depth_first_search): Change structure init
16838 function to this constructor.
16839 (depth_first_search::add_bb): Rename function to this member.
16840 (depth_first_search::execute): Likewise.
16841 (flow_dfs_compute_reverse_finish): Adjust.
16843 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16845 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
16846 (longest_simple_path): Likewise.
16847 * shrink-wrap.c (spread_components): Likewise.
16848 (disqualify_problematic_components): Likewise.
16849 (emit_common_heads_for_components): Likewise.
16850 (emit_common_tails_for_components): Likewise.
16851 (insert_prologue_epilogue_for_components): Likewise.
16853 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16855 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
16858 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16860 * df-core.c (df_set_blocks): Start using auto_bitmap.
16861 (df_compact_blocks): Likewise.
16862 * df-problems.c (df_rd_confluence_n): Likewise.
16863 * df-scan.c (df_insn_rescan_all): Likewise.
16864 (df_process_deferred_rescans): Likewise.
16865 (df_update_entry_block_defs): Likewise.
16866 (df_update_exit_block_uses): Likewise.
16867 (df_entry_block_bitmap_verify): Likewise.
16868 (df_exit_block_bitmap_verify): Likewise.
16869 (df_scan_verify): Likewise.
16870 * lra-constraints.c (lra_constraints): Likewise.
16871 (undo_optional_reloads): Likewise.
16872 (lra_undo_inheritance): Likewise.
16873 * lra-remat.c (calculate_gen_cands): Likewise.
16874 (do_remat): Likewise.
16875 * lra-spills.c (assign_spill_hard_regs): Likewise.
16876 (spill_pseudos): Likewise.
16877 * tree-ssa-pre.c (bitmap_set_and): Likewise.
16878 (bitmap_set_subtract_values): Likewise.
16880 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16882 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
16883 management with auto_bitmap.
16884 (fix_inter_tick): Likewise.
16885 (fix_recovery_deps): Likewise.
16886 * ira.c (add_store_equivs): Likewise.
16887 (find_moveable_pseudos): Likewise.
16888 (split_live_ranges_for_shrink_wrap): Likewise.
16889 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
16890 (rtx_reuse_manager::seen_def_p): Likewise.
16891 (rtx_reuse_manager::set_seen_def): Likewise.
16892 * print-rtl.h (class rtx_reuse_manager): Likewise.
16894 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16896 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
16898 (migrate_btr_def): Likewise.
16899 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
16900 * df-core.c (loop_post_order_compute): Likewise.
16901 (loop_inverted_post_order_compute): Likewise.
16902 * hsa-common.h: Likewise.
16903 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
16904 * init-regs.c (initialize_uninitialized_regs): Likewise.
16905 * ipa-inline.c (resolve_noninline_speculation): Likewise.
16906 (inline_small_functions): Likewise.
16907 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
16908 * ira.c (combine_and_move_insns): Likewise.
16909 (build_insn_chain): Likewise.
16910 * loop-invariant.c (find_invariants): Likewise.
16911 * lower-subreg.c (propagate_pseudo_copies): Likewise.
16912 * predict.c (tree_predict_by_opcode): Likewise.
16913 (predict_paths_leading_to): Likewise.
16914 (predict_paths_leading_to_edge): Likewise.
16915 (estimate_loops_at_level): Likewise.
16916 (estimate_loops): Likewise.
16917 * shrink-wrap.c (try_shrink_wrapping): Likewise.
16918 (spread_components): Likewise.
16919 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
16920 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
16921 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
16922 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
16923 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
16924 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
16925 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
16926 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
16927 (mark_threaded_blocks): Likewise.
16928 (thread_through_all_blocks): Likewise.
16929 * tree-ssa.c (verify_ssa): Likewise.
16930 (execute_update_addresses_taken): Likewise.
16931 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
16933 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16935 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
16937 (post_order_compute): Likewise.
16938 (inverted_post_order_compute): Likewise.
16939 (pre_and_rev_post_order_compute_fn): Likewise.
16941 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16943 * genrecog.c (int_set::int_set): Explicitly construct our
16944 auto_vec base class.
16945 * vec.h (auto_vec::auto_vec): New constructor.
16947 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16949 * bitmap.h (class auto_bitmap): New constructor taking
16950 bitmap_obstack * argument.
16952 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16954 * bitmap.h (class auto_bitmap): Change type of m_bits to
16955 bitmap_head, and adjust ctor / dtor and member operators.
16957 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
16959 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
16960 when returned register mode doesn't match original mode.
16962 2017-05-12 Jeff Law <law@redhat.com>
16963 Jakub Jelinek <jakub@redhat.com>
16965 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
16966 we look for cc setter after the compare-elim changes.
16967 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
16968 within the vector to match what compare-elim now expects.
16969 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
16970 (xorsi3_flags, one_cmplsi2_flags): Likewise.
16972 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
16973 after the compare-elim changes.
16974 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
16975 the vector to match what compare-elim now expects.
16976 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
16977 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
16978 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
16979 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
16980 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
16982 * config/visium/visium.c (single_set_and_flags): Fix where
16983 we look for cc setter after the compare-elim changes.
16984 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
16985 with the vector to match what compare-elim now expects.
16986 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
16987 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
16988 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
16989 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
16990 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
16991 (neg<mode>2_insn_set_overflow): Likewise.
16993 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
16995 PR middle-end/79794
16996 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
16997 maybe_expand_insn call, set ops[0].target. If still set after call,
16998 set alt_rtl. Add extra arg to recursive calls.
16999 (extract_bit_field): Add alt_rtl argument. Pass to
17001 * expmed.h (extract_bit_field): Fix prototype.
17002 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
17003 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
17004 to extract_bit_field_calls.
17005 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
17006 Pass alt_rtl to extract_bit_field calls.
17007 * calls.c (store_unaligned_arguments_into_psuedos)
17008 load_register_parameters): Pass extra NULL to extract_bit_field calls.
17009 * optabs.c (maybe_legitimize_operand): Clear op->target when call
17011 * optabs.h (struct expand_operand): Add target bitfield.
17013 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
17015 * compare-elim.c (try_eliminate_compare): Canonicalize
17016 operation with embedded compare to
17017 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
17018 (set (reg) (operation)].
17020 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
17022 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
17025 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
17026 cost of adding a carry flag for ADC instruction.
17027 [case MINUS]: Ignore the cost of subtracting a carry flag
17028 for SBB instruction.
17030 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
17032 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
17034 * config/rs6000/bmiintrin.h: New file.
17035 * config/rs6000/bmi2intrin.h: New file.
17036 * config/rs6000/x86intrin.h: New file.
17038 2017-05-12 Jeff Law <law@redhat.com>
17040 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
17043 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
17045 PR middle-end/80707
17046 * tree-cfg.c: Remove cfg edges of unreachable case statements.
17048 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17050 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
17051 early expansion of vector divide builtins.
17052 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
17053 builtins identified as having unsigned arguments.
17055 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17057 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
17058 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
17059 expansion of vector logical operations (and, andc, or, xor,
17062 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17064 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
17065 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
17067 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17069 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
17070 early GIMPLE expansion of vector multiplies.
17072 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17074 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
17075 TARGET_HAVE_MOVT conditional.
17076 (movt splitter): Likewise.
17078 2017-05-12 Richard Biener <rguenther@suse.de>
17080 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
17081 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17082 Fold all stmts not inplace.
17084 2017-05-12 Richard Biener <rguenther@suse.de>
17086 PR tree-optimization/80713
17087 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
17088 inserted_exprs bit for not removed stmts.
17090 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
17092 PR middle-end/69921
17093 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
17094 parallelized" attribute for parallelized OpenACC kernels.
17095 * omp-offload.c (execute_oacc_device_lower): Use it.
17097 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
17098 Set "oacc kernels" attribute.
17099 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
17100 parameter. Adjust all users.
17101 (oacc_fn_attrib_kernels_p): Remove function.
17102 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
17103 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
17104 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
17105 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
17106 assert "oacc kernels" attribute is set.
17108 2017-05-11 Carl Love <cel@us.ibm.com>
17110 * config/rs6000/rs6000-c: Add support for built-in functions
17111 vector unsigned char vec_popcnt (vector signed char)
17112 vector unsigned char vec_popcnt (vector unsigned char)
17113 vector unsigned short vec_popcnt (vector signed short)
17114 vector unsigned short vec_popcnt (vector unsigned short)
17115 vector unsigned int vec_popcnt (vector signed int)
17116 vector unsigned int vec_popcnt (vector unsigned int)
17117 vector unsigned long long vec_popcnt (vector signed long long)
17118 vector unsigned long long vec_popcnt (vector unsigned long long)
17119 vector signed long long vec_slo (vector signed long long,
17120 vector signed char)
17121 vector signed long long vec_slo (vector signed long long,
17122 vector unsigned char)
17123 vector unsigned long long vec_slo (vector unsigned long long,
17124 vector signed char)
17125 vector unsigned long long vec_slo (vector unsigned long long,
17126 vector unsigned char)
17127 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
17128 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
17129 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
17130 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
17131 * doc/extend.texi: Update the built-in documentation file for the
17132 new built-in functions.
17134 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
17136 * attribs.h (sorted_attr_string): Move machine independent
17137 functions for target clone support from the i386 port to common
17138 code. Rename ix86_function_versions to common_function_versions.
17139 Rename make_name to make_unique_name.
17140 (common_function_versions): Likewise.
17141 (make_unique_name): Likewise.
17142 (make_dispatcher_decl): Likewise.
17143 (is_function_default_version): Likewise.
17144 * attribs.c (attr_strcmp): Likewise.
17145 (sorted_attr_string): Likewise.
17146 (common_function_versions): Likewise.
17147 (make_unique_name): Likewise.
17148 (make_dispatcher_decl): Likewise.
17149 (is_function_default_version): Likewise.
17150 * config/i386/i386.c (attr_strcmp): Likewise.
17151 (sorted_attr_string): Likewise.
17152 (ix86_function_versions): Likewise.
17153 (make_name): Likewise.
17154 (make_dispatcher_decl): Likewise.
17155 (is_function_default_version): Likewise.
17156 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
17158 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17161 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
17162 Account for direct move costs for vec_construct of integer
17165 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
17168 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
17169 (UNSPEC_STX_ATOMIC): Ditto.
17170 (loaddi_via_sse): New insn.
17171 (storedi_via_sse): Ditto.
17172 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
17173 Update corresponding peephole2 patterns.
17174 (atomic_storedi_fpu): Ditto.
17176 2017-05-11 Julia Koval <julia.koval@intel.com>
17178 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
17179 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
17181 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
17182 (__builtin_ia32_rsqrt14ss_mask): New builtins.
17183 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
17185 2017-05-11 Nathan Sidwell <nathan@acm.org>
17187 * graphite-poly.c: Include dumpfile.h.
17189 * dumpfle.h (dump_function): Declare here ...
17190 * tree-dump.h (dump_function): ... not here.
17191 * dumpfile.c: #include tree-cfg.h.
17192 (dump_function): Move here from ...
17193 * tree-dump.c (dump_function): ... here.
17194 * gimplify.c: #include splay-tree.h, not tree-dump.h.
17195 * graphite-poly.c: Don't include tree-dump.h.
17196 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
17197 * print-tree.c: Likewise.
17198 * stor-layout.c: Likewise.
17199 * tree-nested.c: Likewise.
17201 * dumpfile.c (dump_start): Use TDF_FLAGS.
17202 (dump_enable_all): Fix TDF_KIND check thinko.
17204 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
17206 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17207 array entries to represent two legal parameterizations of the
17208 overloaded __builtin_cmpb function, as represented by the
17209 P6_OV_BUILTIN_CMPB constant.
17210 (altivec_resolve_overloaded_builtin): Add special case handling
17211 for the __builtin_cmpb function, as represented by the
17212 P6_OV_BUILTIN_CMPB constant.
17213 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
17214 (BU_P6_64BIT_2): New macro.
17215 (BU_P6_OVERLOAD_2): New macro
17216 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
17217 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
17218 (CMPB): Add overload support to represent both 32-bit and 64-bit
17219 compare-bytes function.
17220 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
17221 support for TARGET_CMPB.
17222 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
17223 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17224 documentation of the __builtin_cmpb overloaded built-in function.
17226 2017-05-11 Richard Biener <rguenther@suse.de>
17228 PR tree-optimization/80705
17229 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
17230 bases are not vectorizable.
17232 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17234 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
17235 when counting register pressure.
17237 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17239 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
17240 (iv_ca_more_deps): Renamed to ...
17241 (iv_ca_compare_deps): ... this.
17242 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
17244 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17246 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
17248 (determine_group_iv_costs): ... here.
17249 (find_inv_vars_cb): Record inv var if it's not recorded before.
17251 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17253 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
17254 (get_shiftadd_cost): Ditto.
17256 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17258 * tree-ssa-address.c: Include header file.
17259 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
17261 (add_to_parts): Refactor.
17262 (addr_to_parts): New parameter. Update use of move_hint_to_base.
17263 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
17266 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17268 PR tree-optimization/53090
17269 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
17271 (extract_cond_operands): Detect condition with IV on both sides
17272 and return COMP_IV_EXPR_2.
17273 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
17274 (rewrite_use_compare): Simplify by removing call to function
17275 extract_cond_operands.
17277 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17279 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
17280 (extract_cond_operands): Detect condition comparing against non-
17281 invariant bound and return appropriate enum value.
17282 (find_interesting_uses_cond): Update use of extract_cond_operands.
17283 Handle its return value accordingly.
17284 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
17286 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17288 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
17289 nonlinear iv_use computation in loop invariant sensitive way.
17291 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17293 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
17294 (find_iv_candidates): Call relate_compare_use_with_all_cands.
17296 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17298 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
17299 (dump_cand): Support iv_cand.inv_exprs.
17300 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
17302 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
17305 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17307 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
17309 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
17310 as local function. Include necessary header files.
17311 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
17313 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17315 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
17317 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17319 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
17320 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
17321 RSHIFT_EXPR and BIT_NOT_EXPR.
17323 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17325 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
17326 (adjust_setup_cost): New parameter supporting round up adjustment.
17327 (struct address_cost_data): Delete.
17328 (force_expr_to_var_cost): Don't bound cost with spill_cost.
17329 (split_address_cost, ptr_difference_cost): Delete.
17330 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
17331 (struct ainc_cost_data): New struct.
17332 (get_address_cost_ainc): New function.
17333 (get_address_cost, get_computation_cost): Reimplement.
17334 (determine_group_iv_cost_address): Record inv_expr for all uses of
17336 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
17337 (iv_ca_has_deps): Reimplemented to ...
17338 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
17340 (iv_ca_extend): Call iv_ca_more_deps.
17342 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17344 * tree-ssa-address.c (struct mem_address): Move to header file.
17345 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
17346 * tree-ssa-address.h (struct mem_address): Move from C file.
17347 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
17349 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17351 * tree-affine.h (aff_combination_type): New interface.
17352 (aff_combination_zero_p): Remove static.
17353 (aff_combination_const_p): New interface.
17354 (aff_combination_singleton_var_p): New interfaces.
17356 2017-05-11 Richard Biener <rguenther@suse.de>
17358 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17359 Skip unreachable blocks and destinations.
17360 (eliminate): Move stmt removal and fixup ...
17361 (fini_eliminate): ... here. Skip inserted exprs.
17362 (pass_pre::execute): Move fini_pre after fini_eliminate.
17363 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
17364 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
17365 PRE to get rid of dead code that has invalid SSA form and
17366 split critical edges again.
17368 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17370 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17372 2017-05-11 Richard Biener <rguenther@suse.de>
17374 * passes.c (execute_function_todo): Verify loops if they are
17375 said to be up-to-date.
17376 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
17377 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
17379 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
17382 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
17383 handle calling assemble_external ourself.
17386 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
17387 modes with zero size. Enhance comment.
17389 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17391 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
17392 built-ins for vec_xl and vec_xst with short and char pointer
17395 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
17397 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
17398 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
17399 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
17400 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
17401 (_mm_maskz_min_round_ss): New intrinsics.
17402 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
17403 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
17404 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
17405 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
17406 (__builtin_ia32_minss_mask_round): New builtins.
17407 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17408 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
17409 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
17411 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
17412 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
17414 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
17417 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
17419 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
17420 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
17421 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
17422 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
17423 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
17424 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
17425 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
17426 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17427 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
17428 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
17429 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
17430 (__builtin_ia32_mulss_mask_round): New builtins.
17431 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17432 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
17433 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
17435 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
17436 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
17438 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
17441 2017-05-10 Julia Koval <julia.koval@intel.com>
17443 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
17444 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
17445 (_mm256_setr_m128i): New intrinsics.
17447 2017-05-10 Julia Koval <julia.koval@intel.com>
17449 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
17450 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
17451 (_mm_maskz_rcp14_ss): New intrinsics.
17452 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
17453 (__builtin_ia32_rcp14ss_mask): New builtins.
17454 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
17456 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
17458 PR tree-optimization/51513
17459 * tree-cfg.c (gimple_seq_unreachable_p): New function.
17460 (assert_unreachable_fallthru_edge_p): Use it.
17461 (group_case_labels_stmt): Likewise.
17462 * tree-cfg.h: Prototype it.
17463 * stmt.c: Include cfghooks.h and tree-cfg.h.
17464 (emit_case_dispatch_table) <gap_label>: New local variable.
17465 Use it to fill dispatch table gaps.
17466 Test for default_label before updating probabilities.
17467 (expand_case) <default_label>: Remove unneeded initialization.
17468 Test for unreachable default case statement and remove its edge.
17469 Set default_label accordingly.
17470 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
17472 2017-05-10 Carl Love <cel@us.ibm.com>
17474 * config/rs6000/rs6000-c: Add support for built-in functions
17475 vector signed char vec_neg (vector signed char)
17476 vector signed short int vec_neg (vector short int)
17477 vector signed int vec_neg (vector signed int)
17478 vector signed long long vec_neg (vector signed long long)
17479 vector float vec_neg (vector float)
17480 vector double vec_neg (vector double)
17481 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
17483 * config/rs6000/altivec.h: Add define for vec_neg
17484 * doc/extend.texi: Update the built-in documentation for the
17485 new built-in functions.
17487 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17489 PR tree-optimization/77644
17490 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
17492 2017-05-10 Nathan Sidwell <nathan@acm.org>
17494 * dumpfile.h (TDI_lang_all): New.
17495 (TDF_KIND): New. Renumber others
17496 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
17498 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
17500 (get_dump_file_name): Adjust suffix generation.
17501 (dump_enable_all): Use TDF_KIND.
17502 * doc/invoke.texi (-fdump-lang-all): Document.
17504 * dumpfile.h: Tabify.
17506 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
17509 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
17510 Move member access before delete.
17512 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
17514 * tree-inline.c (expand_call_inline): Split block at stmt
17517 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
17520 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
17521 are now unused after splitting mov{sf,sd}_hardfloat.
17522 (f32_lr2): Likewise.
17523 (f32_lm): Likewise.
17524 (f32_lm2): Likewise.
17525 (f32_li): Likewise.
17526 (f32_li2): Likewise.
17527 (f32_lv): Likewise.
17528 (f32_sr): Likewise.
17529 (f32_sr2): Likewise.
17530 (f32_sm): Likewise.
17531 (f32_sm2): Likewise.
17532 (f32_si): Likewise.
17533 (f32_si2): Likewise.
17534 (f32_sv): Likewise.
17535 (f32_dm): Likewise.
17536 (f32_vsx): Likewise.
17537 (f32_av): Likewise.
17538 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
17539 For movsf, order stores so the VSX stores occur before the GPR
17540 store which encourages the register allocator to use a traditional
17541 FPR instead of a GPR. For movsd, order the stores so that the GPR
17542 store comes before the VSX stores to allow the power6 to work.
17543 This is due to the power6 not having a 32-bit integer store
17544 instruction from a FPR.
17545 (movsf_hardfloat): Likewise.
17546 (movsd_hardfloat): Likewise.
17548 2017-05-09 Martin Sebor <msebor@redhat.com>
17550 PR translation/80280
17551 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
17554 PR translation/80280
17555 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
17556 data member added in r247778.
17557 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
17559 2017-05-09 Nathan Sidwell <nathan@acm.org>
17561 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
17563 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
17566 2017-05-09 Marek Polacek <polacek@redhat.com>
17568 * doc/invoke.texi: Fix typo.
17570 2017-05-09 Richard Biener <rguenther@suse.de>
17572 * tree-vrp.c (vrp_val_is_max): Adjust comment.
17573 (vrp_val_is_min): Likewise.
17574 (set_value_range_to_value): Likewise.
17575 (set_value_range_to_nonnegative): Likewise.
17576 (gimple_assign_nonzero_p): Likewise.
17577 (gimple_stmt_nonzero_p): Likewise.
17578 (vrp_int_const_binop): Likewise. Remove unreachable case.
17579 (adjust_range_with_scev): Adjust comments.
17580 (compare_range_with_value): Likewise.
17581 (extract_range_from_phi_node): Likewise.
17582 (test_for_singularity): Likewise.
17584 2017-05-09 Richard Biener <rguenther@suse.de>
17586 * tree-vrp.c (get_single_symbol): Add assert that we don't
17587 get overflowed constants as invariant part.
17588 (compare_values_warnv): Add comment before the TREE_NO_WARNING
17589 checks. Use wi::cmp instead of recursing for integer constants.
17590 (compare_values): Just ignore whether we assumed undefined
17591 overflow instead of failing the compare.
17592 (extract_range_for_var_from_comparison_expr): Add comment before the
17593 TREE_NO_WARNING sets.
17594 (test_for_singularity): Likewise.
17595 (extract_range_from_comparison): Do not disable optimization
17596 when we assumed undefined overflow.
17597 (extract_range_basic): Remove init of unused var.
17599 2017-05-09 Richard Biener <rguenther@suse.de>
17601 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
17602 (extract_range_from_multiplicative_op_1): Adjust.
17603 (extract_range_from_binary_expr_1): Use int_const_binop.
17605 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
17608 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
17609 rs6000_store_data_bypass_p in seven define_bypass directives and
17610 in several comments.
17611 * config/rs6000/rs6000-protos.h: Add prototype for
17612 rs6000_store_data_bypass_p function.
17613 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
17614 function implements slightly different (rs6000-specific) semantics
17615 than store_data_bypass_p, returning false rather than aborting
17616 with assertion error when arguments do not satisfy the
17617 requirements of store data bypass.
17618 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
17619 rs6000_store_data_bypass_p.
17621 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
17623 * config/xtensa/xtensa-protos.h
17624 (xtensa_initial_elimination_offset): New declaration.
17625 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
17626 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
17627 macro definition, add case for FRAME_POINTER_REGNUM when
17628 FRAME_GROWS_DOWNWARD.
17629 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
17630 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
17631 xtensa_initial_elimination_offset.
17633 2017-05-08 Nathan Sidwell <nathan@acm.org>
17635 * doc/invoke.texi: Alphabetize -fdump options.
17637 2017-05-08 Martin Sebor <msebor@redhat.com>
17639 PR translation/80280
17640 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
17642 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
17644 * target.def (compute_frame_layout): New optional target hook.
17645 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
17646 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
17647 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
17649 * reload1.c (verify_initial_elim_offsets): Likewise.
17650 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
17651 (use_simple_return_p): Call arm_compute_frame_layout if needed.
17652 (arm_get_frame_offsets): Split up into this ...
17653 (arm_compute_frame_layout): ... and this function.
17655 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
17657 * config/aarch64/constraints.md (Usa): New constraint.
17658 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
17660 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
17662 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
17663 with_multilib_list after it has been checked.
17665 2017-05-08 Richard Biener <rguenther@suse.de>
17667 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
17668 (bitmap_set_subtract_values): Likewise.
17670 2017-05-08 Richard Biener <rguenther@suse.de>
17672 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
17673 (gimple_assign_nonzero): ... this and remove strict_overflow_p
17675 (gimple_stmt_nonzero_warnv_p): Rename to ...
17676 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
17678 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
17679 (extract_range_basic): Adjust, do not disable propagation on
17680 strict overflow sensitive simplification.
17681 (vrp_visit_cond_stmt): Likewise.
17683 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
17685 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
17686 body size unconditionally.
17688 2017-05-07 Jeff Law <law@redhat.com>
17691 2017-05-06 Jeff Law <law@redhat.com>
17692 PR tree-optimization/78496
17693 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
17696 PR tree-optimization/78496
17697 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
17698 (simplify_stmt_using_ranges): Call it.
17699 (vrp_dom_walker::before_dom_children): Extract equivalences
17700 from an ASSERT_EXPR with an equality comparison against a
17703 2017-05-06 Jeff Law <law@redhat.com>
17705 PR tree-optimization/78496
17706 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
17709 PR tree-optimization/78496
17710 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
17711 (simplify_stmt_using_ranges): Call it.
17712 (vrp_dom_walker::before_dom_children): Extract equivalences
17713 from an ASSERT_EXPR with an equality comparison against a
17716 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
17718 * lra-constraints.c (lra_copy_reg_equiv): New function.
17719 (split_reg): Use it to copy equivalence information from the
17720 original register to the spill register.
17722 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
17724 PR rtl-optimization/75964
17725 * simplify-rtx.c (simplify_const_relational_operation): Remove
17726 invalid handling of comparisons of integer ABS.
17728 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
17730 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
17731 initialize to zero.
17732 (init_regs): Remove declaration.
17733 (function_arg_advance_32): Initialize error_p as boolean variable.
17735 2017-05-05 Nathan Sidwell <nathan@acm.org>
17737 * store-motion.c (remove_reachable_equiv_notes): Reformat long
17738 lines. Use for (;;).
17740 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17742 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
17743 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
17744 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
17745 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
17746 VF=2 that require versioning.
17748 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17750 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
17753 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17755 * diagnostic.h (diagnostic_override_option_index): Convert from
17756 macro to inline function.
17758 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17760 * diagnostic.c (last_module_changed_p): New function.
17761 (set_last_module): New function.
17762 (diagnostic_report_current_module): Convert macro usage to
17763 the above functions.
17764 * diagnostic.h (diagnostic_context::last_module): Strengthen
17765 from const line_map * to const line_map_ordinary *.
17766 (diagnostic_last_module_changed): Delete macro.
17767 (diagnostic_set_last_module): Delete macro.
17769 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17771 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
17772 with diagnostic_report_diagnostic.
17773 (diagnostic_n_impl_richloc): Likewise.
17774 * diagnostic.h (report_diagnostic): Delete macro.
17775 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
17776 with diagnostic_report_diagnostic.
17777 * substring-locations.c (format_warning_va): Likewise.
17779 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17781 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
17782 save/restor of format_spec. Move option-printing code to...
17783 (print_option_information): ...this new function, and
17784 reimplement by simply printing to the pretty_printer,
17785 rather than appending to the format string.
17787 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17789 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
17790 handling logic into...
17791 (update_effective_level_from_pragmas): ...this new function.
17793 2017-05-04 Andrew Waterman <andrew@sifive.com>
17795 * config/riscv/riscv.opt (mstrict-align): New option.
17796 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
17797 (SLOW_UNALIGNED_ACCESS): Define.
17798 (riscv_slow_unaligned_access): Declare.
17799 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
17801 (riscv_slow_unaligned_access): New variable.
17802 (rocket_tune_info): Set slow_unaligned_access to true.
17803 (optimize_size_tune_info): Set slow_unaligned_access to false.
17804 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
17805 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
17806 (riscv_option_override): Set riscv_slow_unaligned_access.
17807 * doc/invoke.texi: Add -mstrict-align to RISC-V.
17809 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
17811 * config/riscv/riscv.md: Unify indentation.
17813 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
17818 * config/rs6000/rs6000.md (u code attribute): Add FIX and
17820 (extendsi<mode>2): Add support for doing sign extension via
17821 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
17822 don't have ISA 3.0 instructions.
17823 (extendsi<mode>2 splitter): Likewise.
17824 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
17825 generate the normal insns since SImode can now go in vector
17826 registers. Disallow the special UNSPECs needed for previous
17827 machines to hide SImode being used. Add new insns
17828 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
17829 (fix_trunc<mode>si2_stfiwx): Likewise.
17830 (fix_trunc<mode>si2_internal): Likewise.
17831 (fixuns_trunc<mode>si2): Likewise.
17832 (fixuns_trunc<mode>si2_stfiwx): Likewise.
17833 (fctiw<u>z_<mode>_smallint): Likewise.
17834 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
17835 of floating point to 32-bit integer from doing a direct move to
17836 the GPR registers to do a store.
17837 (fctiwz_<mode>): Break long line.
17839 2017-05-05 Bin Cheng <bin.cheng@arm.com>
17841 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
17842 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
17843 (addr_list, addr_offset_valid_p): New.
17844 (split_address_groups): Check offset validity with above function.
17845 (gt-tree-ssa-loop-ivopts.h): Include header file.
17847 2017-05-05 Nathan Sidwell <nathan@acm.org>
17849 * config.gcc (arm*-*-*): Add missing 'fi'.
17851 2017-05-05 Steve Ellcey <sellcey@cavium.com>
17853 * doc/invoke.texi (-fopt-info): Explicitly say order of options
17854 included in -fopt-info does not matter.
17855 * doc/optinfo.texi (-fopt-info): Fix description of default
17856 behavour. Explicitly say order of options included in -fopt-info
17859 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
17861 * config.gcc: Allow combinations of aprofile and rmprofile values for
17862 --with-multilib-list.
17863 * config/arm/t-multilib: New file.
17864 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
17865 variables. Remove setting of ISA and floating-point ABI in
17866 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
17867 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
17868 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
17869 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
17871 * config/arm/t-rmprofile: Likewise except for the matches changes.
17872 * doc/install.texi (--with-multilib-list): Document the combination of
17873 aprofile and rmprofile values and warn about pitfalls in doing that.
17875 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
17877 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
17878 (movdi_aarch64): Likewise.
17880 2017-05-05 Jakub Jelinek <jakub@redhat.com>
17882 PR tree-optimization/80632
17883 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
17885 (build_arrays): Initialize it for virtual phis.
17886 (fix_phi_nodes): Use it for virtual phis.
17888 PR tree-optimization/80558
17889 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
17890 [x, y] op z into [x op, y op z] for op & or | if conditions
17893 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
17894 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17897 * config/arm/arm.md (use_literal_pool): Remove.
17898 (64-bit immediate split): No longer takes cost into consideration
17899 if arm_disable_literal_pool is enabled.
17900 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
17901 used when arm_disable_literal_pool is enabled.
17902 (arm_max_const_double_inline_cost): Remove use of
17903 arm_disable_literal_pool.
17904 (push_minipool_fix): Add assert.
17905 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
17906 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
17907 (no_literal_pool_sf_immediate): New.
17909 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17911 PR tree-optimization/80613
17912 * tree-ssa-dce.c (propagate_necessity): Remove cases for
17913 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
17915 2017-05-05 Richard Biener <rguenther@suse.de>
17917 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
17919 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
17921 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
17922 of this flag from insn conditions due to removal from r247495.
17924 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
17926 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
17928 (arm_early_store_addr_dep_ptr): Likewise.
17929 * config/arm/aarch-common-protos.h
17930 (arm_early_load_addr_dep_ptr): Add prototype.
17931 (arm_early_store_addr_dep_ptr): Likewise.
17932 * config/arm/cortex-a53.md: Add new bypasses.
17934 2017-05-05 Jakub Jelinek <jakub@redhat.com>
17936 * tree.c (next_type_uid): Change type to unsigned.
17937 (type_hash_canon): Decrement back next_type_uid if
17938 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
17939 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
17942 2017-05-04 Martin Sebor <msebor@redhat.com>
17944 * builtins.c: Fix a trivial typo in a comment.
17946 PR middle-end/79234
17947 * builtins.c (check_sizes): Adjust to handle reading past the end.
17948 Avoid printing excessive upper bound of ranges. Use %E to print
17949 tree nodes instead of converting them to %wu.
17950 (expand_builtin_memchr): New function.
17951 (compute_dest_size): Rename...
17952 (compute_objsize): ...to this.
17953 (expand_builtin_memcpy): Adjust.
17954 (expand_builtin_mempcpy): Adjust.
17955 (expand_builtin_strcat): Adjust.
17956 (expand_builtin_strcpy): Adjust.
17957 (check_strncat_sizes): Adjust.
17958 (expand_builtin_strncat): Adjust.
17959 (expand_builtin_strncpy): Adjust and simplify.
17960 (expand_builtin_memset): Adjust.
17961 (expand_builtin_bzero): Adjust.
17962 (expand_builtin_memcmp): Adjust.
17963 (expand_builtin): Handle memcmp.
17964 (maybe_emit_chk_warning): Check strncat just once.
17966 2017-05-04 Martin Sebor <msebor@redhat.com>
17968 PR preprocessor/79214
17969 PR middle-end/79222
17970 PR middle-end/79223
17971 * builtins.c (check_sizes): Add inlining context and issue
17972 warnings even when -Wno-system-headers is set.
17973 (check_strncat_sizes): Same.
17974 (expand_builtin_strncat): Same.
17975 (expand_builtin_memmove): New function.
17976 (expand_builtin_stpncpy): Same.
17977 (expand_builtin): Handle memmove and stpncpy.
17979 2017-05-04 Bin Cheng <bin.cheng@arm.com>
17981 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
17982 which is not used any more.
17984 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17986 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
17988 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17990 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
17991 (cortexa53_tunings): Likewise.
17992 (cortexa57_tunings): Likewise.
17993 (cortexa72_tunings): Likewise.
17994 (cortexa73_tunings): Likewise.
17996 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17998 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
17999 Set loop alignment to 8.
18001 2017-05-04 Martin Sebor <msebor@redhat.com>
18003 PR translation/80280
18004 * builtins.c (expand_builtin_object_size): Add missing quoting to
18005 %D and like directives.
18006 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
18007 (hsa_type_for_tree_type): Same.
18008 (verify_function_arguments): Same.
18009 * symtab.c (symbol_table::change_decl_assembler_name): Same.
18010 * varasm.c (get_section): Same.
18013 2017-05-04 Martin Sebor <msebor@redhat.com>
18015 PR translation/80280
18016 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
18018 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
18020 * config/aarch64/aarch64.c (generic_addrcost_table):
18021 Change HI/TI mode setting.
18023 2017-05-04 Martin Jambor <mjambor@suse.cz>
18025 PR tree-optimization/80622
18026 * tree-sra.c (comes_initialized_p): New function.
18027 (build_accesses_from_assign): Only set write lazily when
18028 comes_initialized_p is false.
18029 (analyze_access_subtree): Use comes_initialized_p.
18030 (propagate_subaccesses_across_link): Assert !comes_initialized_p
18031 instead of testing for PARM_DECL.
18033 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18035 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
18036 constraint on operand 0 to allow more general addressing modes.
18037 Adjust output template.
18038 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
18040 * config/aarch64/aarch64-protos.h
18041 (aarch64_address_valid_for_prefetch_p): Declare prototype.
18042 * config/aarch64/constraints.md (Dp): New address constraint.
18043 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
18046 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
18048 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
18049 update use of estimate_ipcp_clone_size_and_time.
18050 (estimate_local_effects): Update use of
18051 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
18052 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
18053 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
18054 Return nonspecialized time.
18056 2017-05-04 Richard Biener <rguenther@suse.de>
18058 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
18059 for the last VUSE which def dominates the PHI. Directly call
18061 (get_continuation_for_phi_1): Remove.
18063 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
18065 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
18066 to explain the use of truncating division. Cap the number of
18067 iterations to the maximum given by nb_iterations_upper_bound,
18070 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18072 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
18073 * configure: Regenerate.
18074 * config.in: Regenerate.
18075 * config/i386/driver-mingw32.c: new file.
18076 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
18077 * config.host: Link driver-mingw32.o on MinGW host.
18078 * doc/install.texi: Document new --enable-mingw-wildcard configure
18081 2017-05-04 Marek Polacek <polacek@redhat.com>
18083 PR tree-optimization/80612
18084 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
18086 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18087 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
18089 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
18090 (movt splitter): Likewise.
18091 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
18092 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
18093 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
18094 block for Thumb-1 with MOVT.
18095 (thumb2_legitimate_address_p): Move code block ...
18096 (can_avoid_literal_pool_for_label_p): ... into this new function.
18097 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
18099 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
18100 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
18101 "M-profile targets with the MOVT instruction".
18103 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18105 * config/arm/arm-builtins.c (arm_init_builtins): Rename
18106 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
18107 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
18109 2017-05-04 Martin Liska <mliska@suse.cz>
18111 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
18112 variable cond_code.
18114 2017-05-04 Richard Biener <rguenther@suse.de>
18116 * tree.c (array_at_struct_end_p): Handle arrays at struct
18117 end with flexarrays more conservatively. Refactor and treat
18118 arrays of arrays or aggregates more strict. Fix
18119 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
18120 * tree.c (array_at_struct_end_p): Adjust prototype.
18121 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
18122 * gimple-fold.c (get_range_strlen): Likewise.
18123 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
18125 2017-05-04 Richard Biener <rguenther@suse.de>
18127 PR tree-optimization/31130
18128 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
18130 (supports_overflow_infinity): Likewise.
18131 (is_negative_overflow_infinity): Likewise.
18132 (is_positive_overflow_infinity): Likewise.
18133 (is_overflow_infinity): Likewise.
18134 (stmt_overflow_infinity): Likewise.
18135 (overflow_infinity_range_p): Likewise.
18136 (usable_range_p): Remove as always returning true.
18137 (make_overflow_infinity): Remove.
18138 (negative_overflow_infinity): Likewise.
18139 (positive_overflow_infinity): Likewise.
18140 (avoid_overflow_infinity): Likewise.
18141 (set_value_range): Adjust accordingly.
18142 (set_value_range_to_nonnegative): Likewise, remove now unused
18143 overflow_infinity arg.
18144 (vrp_operand_equal_p): Adjust.
18145 (update_value_range): Likewise.
18146 (range_int_cst_singleton_p): Likewise.
18147 (operand_less_p): Likewise.
18148 (compare_values_warnv): Likewise.
18149 (extract_range_for_var_from_comparison_expr): Likewise.
18150 (vrp_int_const_binop): Likewise.
18151 (zero_nonzero_bits_from_vr): Likewise.
18152 (extract_range_from_multiplicative_op_1): Likewise.
18153 (extract_range_from_binary_expr_1): Likewise.
18154 (extract_range_from_unary_expr): Likewise.
18155 (extract_range_from_comparison): Likewise.
18156 (extract_range_basic): Likewise.
18157 (adjust_range_with_scev): Likewise.
18158 (compare_ranges): Likewise.
18159 (compare_range_with_value): Likewise.
18160 (dump_value_range): Likewise.
18161 (test_for_singularity): Likewise, remove strict_overflow_p parameter
18163 (simplify_cond_using_ranges): Adjust.
18165 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
18167 * brig-builtins.def: Added a builtin for class_f64.
18168 * builtin-types.def: Added a builtin type needed by class_f64.
18170 2017-05-03 Jason Merrill <jason@redhat.com>
18172 * timevar.def: Add TV_CONSTEXPR.
18174 2017-05-03 David Malcolm <dmalcolm@redhat.com>
18176 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
18178 2017-05-03 Martin Jambor <mjambor@suse.cz>
18180 * ipa-prop.c (ipa_update_after_lto_read): Removed.
18181 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
18182 * ipa-cp.c (ipcp_propagate_stage): Do not call
18183 ipa_update_after_lto_read.
18184 * ipa-inline.c (ipa_inline): Likewise.
18186 2017-05-03 Martin Jambor <mjambor@suse.cz>
18188 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
18189 tag. Added a default constructor and a destructor.
18190 (ipa_edge_args_sum_t): New class;
18191 (ipa_edge_args_sum): Declare.
18192 (ipa_edge_args_vector): Remove declaration.
18193 (IPA_EDGE_REF): Use ipa_edge_args_sum.
18194 (ipa_free_edge_args_substructures): Remove declaration.
18195 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
18196 (ipa_edge_args_info_available_for_edge_p): Likewise.
18197 * ipa-prop.c (ipa_edge_args_vector): Removed.
18198 (edge_removal_hook_holder): Likewise.
18199 (edge_duplication_hook_holder): Likewise.
18200 (ipa_edge_args_sum): New variable.
18201 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
18202 ipa_edge_args_vector.
18203 (ipa_free_edge_args_substructures): Likewise.
18204 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
18205 ipa_edge_args_vector.
18206 (ipa_edge_removal_hook): Turned into method
18207 ipa_edge_args_sum_t::remove.
18208 (ipa_edge_duplication_hook): Turned into method
18209 ipa_edge_args_sum_t::duplicate.
18210 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
18211 registering edge hooks.
18212 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
18213 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
18214 ipa_edge_args_sum instead of ipa_edge_args_vector.
18215 * ipa-profile.c (ipa_profile): Likewise.
18217 2017-05-03 Martin Jambor <mjambor@suse.cz>
18219 * symbol-summary.h (function_summary): New method exists.
18220 (function_summary::symtab_removal): Deallocate through release.
18221 (call_summary): New class.
18222 (gt_ggc_mx): New overload.
18223 (gt_pch_nx): Likewise.
18224 (gt_pch_nx): Likewise.
18226 2017-05-03 Jeff Law <law@redhat.com>
18228 PR tree-optimization/78496
18229 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
18230 from simplify_cond_using_ranges. Split off code to walk
18231 backwards through casts into ...
18232 (simplify_cond_using_ranges_2): New function.
18233 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
18234 (execute_vrp): After identifying jump threads, call
18235 simplify_cond_using_ranges_2.
18237 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
18240 * ipa-inline.h (inline_summary): Add ctor.
18241 (create_ggc): Do not use ggc_cleared_alloc.
18243 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
18244 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18246 * gcc.c (handle_braces): Support escaping in switch matching
18248 * doc/invoke.texi (Spec Files): Document it.
18249 Remove superfluous @code markup in items.
18251 2017-05-03 David Malcolm <dmalcolm@redhat.com>
18253 * diagnostic-show-locus.c (struct column_range): New struct.
18254 (get_affected_columns): New function.
18255 (get_printed_columns): New function.
18256 (struct correction): New struct.
18257 (correction::ensure_capacity): New function.
18258 (correction::ensure_terminated): New function.
18259 (struct line_corrections): New struct.
18260 (line_corrections::~line_corrections): New dtor.
18261 (line_corrections::add_hint): New function.
18262 (layout::print_trailing_fixits): Reimplement in terms of the new
18264 (selftest::test_overlapped_fixit_printing): New function.
18265 (selftest::diagnostic_show_locus_c_tests): Call it.
18267 2017-05-03 Nathan Sidwell <nathan@acm.org>
18269 Canonicalize canonical type hashing
18270 * tree.h (type_hash_canon_hash): Declare.
18271 * tree.c (type_hash_list, attribute_hash_list): Move into
18272 type_hash_canon_hash.
18273 (build_type_attribute_qual_variant): Break out hash code calc into
18274 type_hash_canon_hash.
18275 (type_hash_canon_hash): New. Generic type hash computation.
18276 (build_range_type_1, build_array_type_1, build_function_type,
18277 build_method_type_directly, build_offset_type, build_complex_type,
18278 make_vector_type): Call it.
18280 2017-05-03 Richard Biener <rguenther@suse.de>
18282 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
18283 When all DRs have unknown misaligned do not always peel
18284 when there is a store but apply the same costing model as if
18285 there were only loads.
18287 2017-05-03 Richard Biener <rguenther@suse.de>
18290 PR tree-optimization/80492
18291 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
18292 compare_base_decls returning dont-know properly.
18294 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
18296 * config/arm/iterators.md (CCSI): New mode iterator.
18297 (arch): New mode attribute.
18298 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
18299 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
18300 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
18301 code iterator for success result mode.
18302 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
18303 the corresponding new insn generators.
18305 2017-05-03 Bin Cheng <bin.cheng@arm.com>
18308 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18309 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18311 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
18313 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
18314 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
18315 (DDR_A): Wrap DDR argument in brackets.
18316 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
18317 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
18318 (DDR_REVERSED_P): Likewise.
18320 2017-05-03 Jakub Jelinek <jakub@redhat.com>
18322 PR tree-optimization/79472
18323 * tree-switch-conversion.c (struct switch_conv_info): Add
18324 contiguous_range and default_case_nonstandard fields.
18325 (collect_switch_conv_info): Compute contiguous_range and
18326 default_case_nonstandard fields, don't clear final_bb if
18327 contiguous_range and only the default case doesn't have the required
18329 (check_all_empty_except_final): Set default_case_nonstandard instead
18330 of failing if contiguous_range and the default case doesn't have empty
18332 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
18333 and only the default case doesn't have the required constants. Skip
18335 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
18336 if default_case_nonstandard.
18337 (build_constructors): Build constant 1 just once. Assert that default
18338 values aren't inserted in between cases if contiguous_range. Skip
18340 (build_arrays): Skip virtual phis.
18341 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
18342 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
18343 Handle virtual phis.
18344 (gen_inbound_check): Handle default_case_nonstandard case.
18345 (process_switch): Adjust check_final_bb caller. Call
18346 gather_default_values with the first non-default case instead of
18347 default case if default_case_nonstandard.
18349 2017-05-02 Nathan Sidwell <nathan@acm.org>
18351 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
18352 check. Fix formatting.
18354 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
18356 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
18357 errors when comparing specialized and unspecialized times.
18359 2017-05-02 David Malcolm <dmalcolm@redhat.com>
18361 * diagnostic-show-locus.c
18362 (layout::should_print_annotation_line_p): Make private.
18363 (layout::print_annotation_line): Make private.
18364 (layout::annotation_line_showed_range_p): Make private.
18365 (layout::show_ruler): Make private.
18366 (layout::print_source_line): Make private. Pass in line and
18367 line_width, rather than calling location_get_source_line. Drop
18369 (layout::print_leading_fixits): New method.
18370 (layout::print_any_fixits): Rename to...
18371 (layout::print_trailing_fixits): ...this, and make private.
18372 Don't print newline fixits.
18373 (diagnostic_show_locus): Move logic for printing one row into...
18374 (layout::print_line): ...this new function. Move the
18375 location_get_source_line call and error-handling from
18376 print_source_line to here. Call print_leading_fixits, and rename
18377 print_any_fixits to print_trailing_fixits.
18378 (selftest::test_fixit_insert_containing_newline): Update now that
18379 newlines are partially supported.
18380 (selftest::test_fixit_insert_containing_newline_2): New test.
18381 (selftest::test_fixit_replace_containing_newline): Update comments.
18382 (selftest::diagnostic_show_locus_c_tests): Call the new test.
18383 * edit-context.c (class added_line): New class.
18384 (class edited_line): Describe newline handling in comment.
18385 (edited_line::actually_edited_p): New method.
18386 (edited_line::print_content): Delete redundant decl.
18387 (edited_line::m_predecessors): New field.
18388 (edited_file::print_content): Call edited_line::print_content.
18389 (edited_file::print_diff): Update to support newlines.
18390 (edited_file::print_diff_hunk): Likewise.
18391 (edited_file::print_run_of_changed_lines): New function.
18392 (edited_file::print_diff_line): Convert to...
18393 (print_diff_line): ...this.
18394 (edited_file::get_effective_line_count): New function.
18395 (edited_line::edited_line): Initialize new field m_predecessors.
18396 (edited_line::~edited_line): Clean up m_predecessors.
18397 (edited_line::apply_fixit): Handle newlines.
18398 (edited_line::get_effective_line_count): New function.
18399 (edited_line::print_content): New function.
18400 (edited_line::print_diff_lines): New function.
18401 (selftest::test_applying_fixits_insert_containing_newline): New
18403 (selftest::test_applying_fixits_replace_containing_newline): New
18405 (selftest::insert_line): New function.
18406 (selftest::test_applying_fixits_multiple_lines): Add example of
18408 (selftest::edit_context_c_tests): Call the new tests.
18410 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18412 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
18413 parameter cand. Update dump information.
18414 (get_computation_cost): Update uses.
18416 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18418 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
18419 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
18420 (get_computation_at, rewrite_use_address): Update use of
18421 get_computation_aff.
18423 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18425 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
18426 (get_computation): Delete.
18427 (get_computation_cost): Implement like get_computation_cost_at.
18428 Use get_computation_at.
18429 (get_computation_cost_at): Delete.
18430 (rewrite_use_nonlinear_expr): Use get_computation_at.
18431 (rewrite_use_compare, remove_unused_ivs): Ditto.
18433 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18435 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
18437 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18439 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
18440 (ivopts_global_cost_for_size): Rename parameter and update uses.
18441 (iv_ca_recount_cost): Update uses.
18442 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
18443 candidates seperately in n_invs and n_cands.
18444 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
18446 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18448 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
18449 (find_inv_vars_cb): New.
18450 (find_depends): Renamed to ...
18451 (find_inv_vars): ... this.
18452 (add_candidate_1, force_var_cost): Call find_inv_vars.
18453 (split_address_cost, determine_group_iv_cost_cond): Ditto.
18455 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18457 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
18458 inv_vars. Add inv_exprs.
18459 (struct iv_cand): Rename depends_on to inv_vars.
18460 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
18461 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
18462 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
18463 (dump_cand): Dump inv_vars.
18464 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
18465 (record_invariant, find_depends, add_candidate_1): Ditto.
18466 (set_group_iv_cost, force_var_cost): Ditto.
18467 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
18468 (get_computation_cost_at, get_computation_cost): Ditto.
18469 (determine_group_iv_cost_generic): Ditto.
18470 (determine_group_iv_cost_address): Ditto.
18471 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
18472 (determine_group_iv_costs): Ditto.
18473 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
18474 (iv_ca_set_remove_invariants): Renamed to ...
18475 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
18476 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
18477 (iv_ca_set_add_invariants): Renamed to ...
18478 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
18479 (iv_ca_set_cp): Use iv_ca_set_add_invs.
18480 (iv_ca_has_deps): Support inv_vars and inv_exprs.
18481 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
18482 (create_new_ivs): Remove useless dump.
18484 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18486 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
18488 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
18489 (iv_ca_set_no_cp, create_new_iv): Ditto.
18491 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18493 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18495 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
18497 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
18498 function tree_check2.
18500 2017-05-02 Martin Liska <mliska@suse.cz>
18502 * doc/gcov.texi: Add missing preposition.
18503 * gcov.c (function_info::function_info): Properly fill up
18504 all member variables.
18506 2017-05-02 Tamar Christina <tamar.christina@arm.com>
18508 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
18510 2017-05-02 Tamar Christina <tamar.christina@arm.com>
18512 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
18514 2017-05-02 Martin Liska <mliska@suse.cz>
18517 * lto-streamer-in.c (lto_read_tree_1): Remove
18518 LTO_STREAMER_DEBUG.
18519 * lto-streamer.c (struct tree_hash_entry): Likewise.
18520 (struct tree_entry_hasher): Likewise.
18521 (tree_entry_hasher::hash): Likewise.
18522 (tree_entry_hasher::equal): Likewise.
18523 (lto_streamer_init): Likewise.
18524 (lto_orig_address_map): Likewise.
18525 (lto_orig_address_get): Likewise.
18526 (lto_orig_address_remove): Likewise.
18527 * lto-streamer.h: Likewise.
18528 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
18529 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
18531 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
18533 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
18534 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
18535 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
18536 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
18537 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
18538 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
18539 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
18540 (mm_maskz_sub_ss): New intrinsics.
18541 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18542 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
18543 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
18544 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
18545 (__builtin_ia32_subss_mask_round): New builtins.
18546 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18547 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18548 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
18550 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
18551 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
18553 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
18556 2017-05-02 Martin Jambor <mjambor@suse.cz>
18558 PR tree-optimization/78687
18559 * tree-sra.c (access): New field parent.
18560 (process_subtree_disqualification): New function.
18561 (disqualify_candidate): Call it.
18562 (build_accesses_from_assign): Reset write flag if creating an
18564 (build_access_subtree): Fill in parent field and also prpagate
18565 down grp_write flag.
18566 (create_artificial_child_access): New parameter set_grp_write, set
18567 grp_write to its value.
18568 (propagate_subaccesses_across_link): Also propagate grp_write flag
18570 (propagate_all_subaccesses): Push the closest parent back to work
18571 queue if add_access_to_work_queue returned true.
18573 2017-05-02 Richard Biener <rguenther@suse.de>
18575 * common.opt (fstrict-overflow): Alias negative to fwrapv.
18576 * doc/invoke.texi (fstrict-overflow): Remove all traces of
18577 -fstrict-overflow documentation.
18578 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
18579 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
18580 flag_strict_overflow.
18581 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
18582 * lto-opts.c (lto_write_options): Do not stream it.
18583 * lto-wrapper.c (merge_and_complain): Do not handle it.
18584 * opts.c (default_options_table): Do not set -fstrict-overflow.
18585 (finish_options): Likewise do not clear it when sanitizing.
18586 * simplify-rtx.c (simplify_const_relational_operation): Do not
18587 test flag_strict_overflow.
18589 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
18591 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
18592 using enabled attribute.
18593 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
18594 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
18595 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
18596 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
18597 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
18598 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
18599 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
18600 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
18601 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
18602 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
18604 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
18606 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
18608 2017-05-02 Richard Biener <rguenther@suse.de>
18610 PR tree-optimization/80591
18612 2017-04-10 Richard Biener <rguenther@suse.de>
18614 * tree-ssa-structalias.c (find_func_aliases): Properly handle
18617 2017-05-02 Richard Biener <rguenther@suse.de>
18619 PR tree-optimization/80549
18620 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
18621 (cleanup_tree_cfg_noloop): Create forwarders to known loop
18622 headers if they do not have a preheader.
18624 2017-05-02 Martin Liska <mliska@suse.cz>
18627 * common.opt: Fix typo.
18628 * doc/invoke.texi: Likewise.
18630 2017-05-01 Jan Beulich <jbeulich@suse.com>
18632 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
18633 swapping, add (x,x,m,x,n) alternative.
18635 2017-05-01 Nathan Sidwell <nathan@acm.org>
18637 * calls.c (combine_pending_stack_adjustment_and_call): Remove
18638 unnecessary unadjusted_alignment check.
18640 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
18643 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
18645 * gimplify.c (gimplify_cilk_detach): New function.
18646 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
18647 * tree-core.h: Document EXPR_CILK_SPAWN.
18648 * tree.h (EXPR_CILK_SPAWN): Define.
18650 2017-05-01 David Malcolm <dmalcolm@redhat.com>
18652 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
18653 to use new fixit_hint representation, using the "replace" logic.
18654 (get_line_span_for_fixit_hint): Likewise.
18655 (layout::print_any_fixits): Likewise.
18656 (selftest::test_one_liner_many_fixits): Rename to...
18657 (selftest::test_one_liner_many_fixits_1): ...this, and update
18658 comment and expected output to reflect that the multiple fix-it
18659 hints are now consolidated into one insertion.
18660 (selftest::test_one_liner_many_fixits_2): New test.
18661 (selftest::test_diagnostic_show_locus_one_liner): Update for
18663 (selftest::test_fixit_consolidation): Update for fix-it API
18665 * diagnostic.c (print_parseable_fixits): Likewise.
18666 * edit-context.c (edited_line::m_line_events): Convert from
18667 auto_vec <line_event *> to auto_vec <line_event>.
18668 (class line_event): Convert from abstract base class to a concrete
18669 class, taking over the role of replace_event.
18670 (class insert_event): Delete.
18671 (class replace_event): Rename to class line_event. Convert to
18673 (edit_context::add_fixits): Reimplement.
18674 (edit_context::apply_insert): Delete.
18675 (edit_context::apply_replace): Rename to...
18676 (edit_context::apply_fixit): ...this. Convert to half-open range.
18677 (edited_file::apply_insert): Delete.
18678 (edited_file::apply_replace): Rename to...
18679 (edited_file::apply_fixit): ...this.
18680 (edited_line::~edited_line): Drop deletion of events.
18681 (edited_line::apply_insert): Delete.
18682 (edited_line::apply_replace): Rename to...
18683 (edited_line::apply_fixit): ...this. Convert to half-open range.
18684 Update for change to type of m_line_events.
18685 * edit-context.h (edit_context::apply_insert): Delete.
18686 (edit_context::apply_replace): Rename to...
18687 (edit_context::apply_fixit): ...this.
18689 2017-05-01 Martin Sebor <msebor@redhat.com>
18691 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
18694 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
18697 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
18698 __get_cpuid_max returns 0.
18699 (__get_cpuid_count): Ditto.
18701 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
18703 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
18704 replacement expression is another instance of one of its arguments.
18706 2017-05-01 Jakub Jelinek <jakub@redhat.com>
18709 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
18710 check for stack push/pop autoinc.
18711 * config/i386/i386.c (ix86_agi_dependent): Return false
18712 if the only reason why modified_in_p returned true is that
18713 addr is SP based and set_insn is a push or pop.
18715 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
18717 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
18720 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
18723 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
18724 (account_size_time): Use two predicates - exec_pred and
18726 (evaluate_conditions_for_known_args): Compute both clause and
18728 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
18729 (inline_summary_t::duplicate): Update.
18730 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
18732 (compute_inline_parameters): Likewise.
18733 (estimate_edge_size_and_time): Update caluclation of time.
18734 (estimate_node_size_and_time): Compute both time and nonspecialized
18736 (estimate_ipcp_clone_size_and_time): Update.
18737 (inline_merge_summary): Update.
18738 (do_estimate_edge_time): Update.
18739 (do_estimate_edge_size): Update.
18740 (do_estimate_edge_hints): Update.
18741 (inline_read_section, inline_write_summary): Stream both new predicates.
18742 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
18744 (compute_inlined_call_time): Cleanup.
18745 (big_speedup_p): Update.
18746 (edge_badness): Update.
18747 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
18748 (size_time_entry): Replace predicate by exec_predicate and
18749 nonconst_predicate.
18750 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
18751 (estimate_edge_time): Return also nonspec_time.
18752 (reset_edge_growth_cache): Update.
18754 2017-04-29 Jakub Jelinek <jakub@redhat.com>
18756 PR rtl-optimization/80491
18757 * ifcvt.c (noce_process_if_block): When looking for x setter
18758 with missing else_bb, don't check only the insn right before
18759 cond_earliest, but look for the last insn that x is modified in
18760 within the same bb.
18762 PR rtl-optimization/80491
18763 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
18765 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
18767 PR tree-optimization/80487
18768 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
18770 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18772 PR tree-optimization/79697
18773 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
18774 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
18775 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
18777 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
18778 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
18780 2017-04-28 Martin Sebor <msebor@redhat.com>
18782 PR tree-optimization/80523
18783 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
18784 (init_target_to_host_charmap, target_to_host, target_strtol10): New
18786 (maybe_warn, format_directive, parse_directive): Use new functions.
18787 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
18789 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
18791 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
18793 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
18795 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
18796 target_header_dir): Set correctly.
18797 * configure: Regenerated.
18798 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
18799 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
18800 instead of SYSTEM_HEADER_DIR.
18802 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
18804 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
18805 (estimate_local_effects): Likewise.
18806 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
18807 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
18808 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
18809 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
18810 do_estimate_edge_time, estimate_edge_time): Likewise.
18811 * ipa-inline-analysis.c (estimate_node_size_and_time,
18812 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
18813 (estimate_time_after_inlining): Remove.
18815 2017-04-28 Martin Liska <mliska@suse.cz>
18817 * doc/gcov.texi: Enhance documentation of gcov.
18819 2017-04-28 Martin Liska <mliska@suse.cz>
18821 * doc/gcov.texi: Sort options in alphabetic order.
18822 * doc/gcov-dump.texi: Likewise.
18823 * doc/gcov-tool.texi: Likewise.
18824 * gcov.c (print_usage): Likewise.
18825 * gcov-dump.c (print_usage): Likewise.
18826 * gcov-tool.c (print_merge_usage_message): Likewise.
18827 (print_rewrite_usage_message): Likewise.
18828 (print_overlap_usage_message): Likewise.
18830 2017-04-28 Martin Liska <mliska@suse.cz>
18832 PR gcov-profile/53915
18833 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
18835 2017-04-28 Martin Liska <mliska@suse.cz>
18837 PR gcov-profile/79891
18838 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
18839 is marked by compiler as living on a line.
18840 (get_cycles_count): Remove usage of the union.
18841 (output_intermediate_file): Likewise.
18842 (find_source): Fix GNU coding style.
18843 (accumulate_line_counts): Remove old non-all block mode.
18844 (output_lines): Remove usage of the union.
18845 * profile.c (output_location): Include all BBs, even if
18846 belonging to a same line (and file) as a previous BB.
18848 2017-04-28 Martin Liska <mliska@suse.cz>
18850 * gcov.c (process_args): Handle new argument 'w'.
18851 (read_graph_file): Assign ID to BBs.
18852 (output_branch_count): Display BB # if verbose flag is set.
18853 (output_lines): Likewise for arcs.
18854 (print_usage): Add '--verbose' option help.
18855 * doc/gcov.texi: Document --verbose (-w) option.
18857 2017-04-28 Martin Liska <mliska@suse.cz>
18859 * gcov.c (struct block_location_info): New struct.
18860 (process_file): Fill up the new structure.
18861 (read_graph_file): Replace usage of encoding by the newly added
18863 (add_line_counts): Likewise.
18864 (accumulate_line_counts): Remove usage of the union.
18865 (function_info::function_info): New function.
18866 (function_info::~function_info): Likewise.
18867 (process_file): Call delete instead of release_function.
18868 (release_function): Release the function.
18869 (release_structures): Call delete instead of release_function.
18870 (solve_flow_graph): Replace usage of num_blocks.
18871 (find_exception_blocks): Likewise.
18872 (output_lines): Fix GNU coding style.
18874 2017-04-28 Martin Liska <mliska@suse.cz>
18877 * coverage.c (coverage_remove_note_file): New function.
18878 * coverage.h: Declare the function.
18879 * toplev.c (finalize): Clean if an error has been seen.
18881 2017-04-28 Martin Liska <mliska@suse.cz>
18883 PR gcov-profile/80031
18884 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
18885 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
18886 * gcov.c (read_graph_file): Read just number of blocks.
18887 * profile.c (branch_prob): Do not stream 0 flags per a basic
18890 2017-04-28 Martin Liska <mliska@suse.cz>
18892 * gcov-dump.c (tag_*): Add new argument to declarations.
18893 (dump_gcov_file): Likewise.
18894 (tag_blocks): Add and use new argument depth.
18895 (tag_arcs): Likewise.
18896 (tag_lines): Likewise.
18897 (tag_counters): Likewise.
18898 (tag_summary): Likewise.
18899 (dump_working_sets): Use depth to do a proper indentation.
18901 2017-04-28 Jakub Jelinek <jakub@redhat.com>
18904 * cgraph.h (symtab_node::debug_symtab): No longer inline.
18905 * symtab.c (symtab_node::debug_symtab): Move definition here.
18907 2017-04-28 Richard Biener <rguenther@suse.de>
18909 * lto-streamer.h (LTO_major_version): Bump to 7.
18911 2017-04-28 Richard Biener <rguenther@suse.de>
18913 * tree-vrp.c (assert_info): New struct.
18914 (add_assert_info): New helper.
18915 (register_edge_assert_for_2): Refactor to add asserts to a vector
18917 (register_edge_assert_for_1): Likewise.
18918 (register_edge_assert_for): Likewise.
18919 (finish_register_edge_assert_for): New helper actually registering
18920 asserts where live on edge.
18921 (find_conditional_asserts): Adjust.
18922 (find_switch_asserts): Likewise.
18923 (evrp_dom_walker::try_find_new_range): Generalize.
18924 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
18926 2017-04-27 Marek Polacek <polacek@redhat.com>
18929 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
18930 arg10 and arg11 to itype.
18932 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
18934 * doc/extend.texi (Object Size Checking): Improve grammar.
18936 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
18939 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
18940 that the logic for permitting reciprocal estimates matches that
18943 2017-04-27 Jakub Jelinek <jakub@redhat.com>
18946 * tree.c (type_cache_hasher::equal): Only compare
18947 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
18948 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
18949 non-aggregate element types.
18950 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
18951 about the flag on ARRAY_TYPEs in the comment, formatting fix.
18953 2017-04-27 Richard Biener <rguenther@suse.de>
18955 PR middle-end/80533
18956 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
18957 stripping ARRAY_REFs from MEM_EXPR make sure we're not
18958 keeping a reference to a trailing array.
18960 2017-04-27 Richard Biener <rguenther@suse.de>
18962 PR middle-end/80539
18963 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
18964 being in loop-closed SSA form conservatively.
18965 (chrec_fold_multiply_poly_poly): Likewise.
18967 2017-04-27 Tamar Christina <tamar.christina@arm.com>
18969 PR middle-end/79665
18970 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
18971 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
18973 2017-04-27 Jakub Jelinek <jakub@redhat.com>
18976 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
18977 (aarch64_function_arg_alignment): Return unsigned int again, but still
18978 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
18979 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
18980 Don't emit -Wpsabi note.
18981 (aarch64_function_arg_boundary): Likewise.
18982 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
18985 2017-04-26 Nathan Sidwell <nathan@acm.org>
18987 * tree.h (crc32_unsigned_n): Declare.
18988 (crc32_unsigned, crc32_unsigned): Make inline.
18989 * tree.c (crc32_unsigned_bits): Replace with ...
18990 (crc32_unsigned_n): ... this.
18991 (crc32_unsigned, crc32_byte): Remove.
18992 (crc32_string): Remove unnecessary braces.
18994 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
18996 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
18997 * ipa-inline-analysis.c (MAX_TIME): Remove.
18998 (account_size_time): Use sreal for time.
18999 (dump_inline_summary): Update.
19000 (estimate_function_body_sizes): Update.
19001 (estimate_edge_size_and_time): Update.
19002 (estimate_calls_size_and_time): Update.
19003 (estimate_node_size_and_time): Update.
19004 (inline_merge_summary): Update.
19005 (inline_update_overall_summary): Update.
19006 (estimate_time_after_inlining): Update.
19007 (inline_read_section): Update.
19008 (inline_write_summary): Update.
19009 * ipa-inline.c (compute_uninlined_call_time): Update.
19010 (compute_inlined_call_time): Update.
19011 (recursive_inlining): Update.
19012 (inline_small_functions): Update.
19013 (dump_overall_stats): Update.
19014 * ipa-inline.h: Include sreal.h.
19015 (size_time_entry): Turn time to sreal.
19016 (inline_summary): Turn self_time nad time to sreal.
19018 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
19020 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
19022 (sreal::stream_out, sreal::stream_in): New.
19023 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
19025 2017-04-25 Jakub Jelinek <jakub@redhat.com>
19027 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
19030 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
19033 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
19035 (dimode_scalar_chain::compute_convert_gain): Ditto.
19036 (dimode_scalar_chain::make_vector_copies): Ditto.
19037 (dimode_scalar_chain::convert_reg): Ditto.
19038 (dimode_scalar_chain::convert_insn): Ditto.
19039 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
19040 (VI248_AVX512BW_1): New mode iterator.
19041 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
19042 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
19045 2017-04-25 Martin Sebor <msebor@redhat.com>
19047 PR tree-optimization/80497
19048 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
19049 constants are representable in HOST_WIDE_INT.
19050 (parse_directive): Ditto.
19052 2017-04-25 Martin Sebor <msebor@redhat.com>
19055 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
19056 (new_zero_array): Adjust signature.
19057 (dom_info::dom_init): Used unsigned rather that size_t.
19058 (dom_info::dom_info): Same.
19060 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19061 Jakub Jelinek <jakub@redhat.com>
19064 * config/arm/arm.c: Include gimple.h.
19065 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
19066 returns negative, increment ncrn only if it returned positive.
19067 (arm_needs_doubleword_align): Return int instead of bool,
19068 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
19069 members, but if there is any such non-FIELD_DECL
19070 > PARM_BOUNDARY aligned decl, return -1 instead of false.
19071 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
19072 returns negative, increment nregs only if it returned positive.
19073 (arm_setup_incoming_varargs): Likewise.
19074 (arm_function_arg_boundary): Emit -Wpsabi note if
19075 arm_needs_doubleword_align returns negative, return
19076 DOUBLEWORD_ALIGNMENT only if it returned positive.
19078 2017-04-25 Marek Polacek <polacek@redhat.com>
19081 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
19082 first argument to type.
19084 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
19087 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
19088 type checks to test for compatibility instead of equality.
19090 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19091 Jakub Jelinek <jakub@redhat.com>
19094 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
19096 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
19097 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
19098 the alignment computation, but return their maximum in warn_alignment.
19099 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
19100 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
19102 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
19103 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
19106 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19108 * config/arc/simdext.md (dmpyh): Fix typo.
19110 2017-04-25 Richard Biener <rguenther@suse.de>
19112 PR tree-optimization/80492
19113 * alias.c (compare_base_decls): Handle registers with asm
19114 specification conservatively.
19115 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
19116 compare_base_decls returning dont-know properly.
19118 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19120 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
19121 (legitimate_offset_address_p): New function.
19122 (arc_legitimate_address_p): Use above function.
19124 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19126 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
19128 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19130 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
19131 ACCH registers whenever they are available.
19133 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19135 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
19136 double regs fix when not used.
19138 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19140 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
19142 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
19143 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
19145 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19147 * config/arc/arc.c (arc_output_addsi): Check for h-register class
19148 when emitting short ADD instructions.
19150 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19152 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
19154 (cmpsi_cc_c_insn): Likewise.
19155 (cbranchsi4_scratch): Compute proper instruction length using
19156 compact_hreg_operand.
19157 * config/arc/predicates.md (compact_hreg_operand): New predicate.
19159 2017-04-25 Richard Biener <rguenther@suse.de>
19161 PR middle-end/80509
19162 * passes.c (pass_manager::pass_manager): Initialize
19163 m_name_to_pass_map.
19165 2017-04-25 Richard Biener <rguenther@suse.de>
19167 PR tree-optimization/79201
19168 * tree-ssa-sink.c (statement_sink_location): Handle calls.
19170 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19173 * config/s390/vector.md: Split MEM->GPR vector moves for
19174 non-s_operand addresses.
19176 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19179 * config/s390/predicates.md (reload_const_wide_int_operand): New
19181 * config/s390/s390.md ("movti"): Remove d/P alternative.
19182 ("movti_bigconst"): New pattern definition.
19184 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
19187 * s390-protos.h (s390_expand_cs_hqi): Removed.
19188 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
19189 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
19190 modes as well as CCZ1mode and CCZmode.
19191 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
19192 signature of s390_emit_compare_and_swap.
19193 (s390_expand_cs_hqi): Likewise, make static.
19194 (s390_expand_cs_tdsi): Generate an explicit compare before trying
19195 compare-and-swap, in some cases.
19196 (s390_expand_cs): Wrapper function.
19197 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
19199 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
19200 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
19201 patterns for small and large integers. Forbid symref memory operands.
19202 Move expander to s390.c. Require cc register.
19203 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
19204 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
19205 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
19206 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
19207 symref memory operands. Remove CC mode and call s390_match_ccmode
19209 ("atomic_exchange<mode>"): Allow and implement all integer modes.
19211 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
19213 * config/s390/s390.md (define_peephole2): New peephole to help
19214 combining the load-and-test pattern with volatile memory.
19216 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
19218 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
19219 with CCZmode for TARGET_Z196.
19221 2017-04-25 Jakub Jelinek <jakub@redhat.com>
19223 PR rtl-optimization/80501
19224 * combine.c (make_compound_operation_int): Set subreg_code to SET
19225 even for AND with mask of the sign bit of mode.
19227 PR rtl-optimization/80500
19228 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
19229 sum's initial value.
19231 2017-04-25 Julian Brown <julian@codesourcery.com>
19232 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19234 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
19236 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
19238 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
19240 2017-04-25 Julian Brown <julian@codesourcery.com>
19241 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19243 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
19244 (thunderx2t99_sha): New Reservation.
19246 2017-04-25 Julian Brown <julian@codesourcery.com>
19247 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19249 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
19250 type for 1-element load.
19252 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
19254 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
19256 2017-04-24 Martin Jambor <mjambor@suse.cz>
19258 PR tree-optimization/80293
19259 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
19260 char arrays not totally scalarizable if it is false.
19261 (analyze_all_variable_accesses): Pass correct value in the new
19262 parameter. Add a statistics counter.
19264 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
19266 PR middle-end/79931
19267 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
19269 2017-04-24 Richard Biener <rguenther@suse.de>
19271 PR tree-optimization/80494
19272 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
19273 out for complex types.
19275 2017-04-24 Richard Biener <rguenther@suse.de>
19277 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
19278 * tree-ssa-sccvn.c (print_scc): Print SCC size.
19279 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
19280 (DFS): Adjust and never fail.
19281 (sccvn_dom_walker::fail): Remove.
19282 (sccvn_dom_walker::before_dom_children): Adjust.
19283 (run_scc_vn): Likewise and never fail.
19284 * tree-ssa-pre.c (pass_pre::execute): Adjust.
19285 (pass_fre::execute): Likewise.
19287 2017-04-24 Richard Biener <rguenther@suse.de>
19289 PR tree-optimization/79725
19290 * tree-ssa-sink.c (statement_sink_location): Return whether
19291 failure reason was zero uses. Move that check later.
19292 (sink_code_in_bb): Deal with zero uses by removing the stmt
19295 2017-04-24 Richard Biener <rguenther@suse.de>
19298 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
19299 pointer-based references.
19301 2017-04-24 Richard Biener <rguenther@suse.de>
19304 * pass_manager.h (pass_manager::operator new): Remove.
19305 (pass_manager::operator delete): Likewise.
19306 * passes.c (pass_manager::operator new): Remove.
19307 (pass_manager::operator delete): Likewise.
19308 (pass_manager::pass_manager): Zero individual pass members.
19310 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
19313 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
19314 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
19315 Check "XEXP (src, 1)" operand here.
19316 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
19317 Check "XEXP (src, 1)" operand here.
19318 (dimode_scalar_chain::make_vector_copies): Detect count register
19319 of a shift instruction. Zero extend count register from QImode
19320 to DImode to satisfy vector shift pattern count operand predicate.
19321 Substitute vector shift count operand with a DImode copy.
19322 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
19325 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
19327 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
19328 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
19329 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
19330 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
19331 (UNSPEC_NOREX_MEM): Remove definition.
19333 2017-04-21 Richard Biener <rguenther@suse.de>
19335 PR tree-optimization/79547
19336 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
19337 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
19338 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
19339 without any constraints.
19341 2017-04-21 Richard Biener <rguenther@suse.de>
19343 PR tree-optimization/78847
19344 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
19346 2017-04-21 Richard Biener <rguenther@suse.de>
19348 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
19349 (build_distinct_type_copy): Likewise.
19350 (build_variant_type_copy): Likewise.
19351 * tree.c (build_qualified_type): Pass down mem-stat info.
19352 (build_distinct_type_copy): Likewise.
19353 (build_variant_type_copy): Likewise.
19355 2017-04-21 Richard Biener <rguenther@suse.de>
19357 PR tree-optimization/80237
19358 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
19360 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
19361 for a simplified result.
19363 2016-04-21 Richard Biener <rguenther@suse.de>
19365 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
19366 sth as strict as a simple_iv but a chrec without symbols and an
19367 operand defined in the loop we are peeling (and not some subloop).
19368 (propagate_constants_for_unrolling): Propagate all constants.
19370 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
19373 * config/i386/i386.c (print_reg): Remove assert for disalowed
19374 regno values, call output_operand_lossage instead.
19376 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
19379 * config/i386/constraints.md (Yc): New register constraint.
19380 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
19381 Use Yc constraint for alternative 2 of operand 0. Remove
19382 preferred_for_speed attribute.
19384 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
19386 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
19387 lastprivate clauses in SIMT case.
19389 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
19391 * doc/invoke.texi (-Wextra-semi): Document new warning option.
19393 2017-04-20 Richard Biener <rguenther@suse.de>
19395 PR tree-optimization/57796
19396 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
19397 as N scalar stores.
19398 (vect_model_load_cost): Cost gathers as N scalar loads.
19400 2017-04-20 Richard Biener <rguenther@suse.de>
19402 * ggc-page.c (ggc_allocated_p): Rename to ...
19403 (safe_lookup_page_table_entry): ... this and return the lookup
19405 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
19407 2017-04-20 Richard Biener <rguenther@suse.de>
19409 PR tree-optimization/80453
19410 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
19411 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
19412 from the conditions.
19413 (vn_phi_eq): Pass them down.
19414 (vn_phi_lookup): Record them.
19415 (vn_phi_insert): Likewise.
19417 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
19419 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
19420 uninitialized variable warning to avoid buffer overrun.
19422 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
19425 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
19426 is suppressed for '{ 0 }' in C.
19428 2017-04-20 Jakub Jelinek <jakub@redhat.com>
19430 * BASE-VER: Set to 8.0.0.
19432 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
19434 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
19435 priority .init_array and .fini_array section with SECTION_NOTYPE
19438 2017-04-20 Jakub Jelinek <jakub@redhat.com>
19440 PR middle-end/80423
19441 * tree.h (build_array_type): Add typeless_storage default argument.
19442 * tree.c (type_cache_hasher::equal): Also compare
19443 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
19444 (build_array_type): Add typeless_storage argument, set
19445 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
19447 (build_nonshared_array_type): Adjust build_array_type_1 caller.
19448 (build_array_type): Likewise. Add typeless_storage argument.
19450 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
19451 Jakub Jelinek <jakub@redhat.com>
19453 PR tree-optimization/80426
19454 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
19455 operation on symbolic operands, also compute the overflow for the
19456 invariant part when the operation degenerates into a negation.
19458 2017-04-19 Jakub Jelinek <jakub@redhat.com>
19461 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
19462 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
19465 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
19467 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
19470 * config/avr/avr.c (tree.h): Include it.
19471 (cgraph.h): Include it.
19472 (avr_encode_section_info): Don't warn for uninitialized progmem
19473 variable if it's just an alias.
19475 2017-04-19 Richard Biener <rguenther@suse.de>
19478 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
19479 when needed by AutoPGO.
19481 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
19484 * doc/lto.texi: Remove an extra 'that'.
19486 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
19488 PR rtl-optimization/80429
19489 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
19490 are only used in debug insns.
19492 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
19493 Vladimir Makarov <vmakarov@redhat.com>
19495 * config/sparc/predicates.md (input_operand): Add comment. Return
19496 true for any memory operand when LRA is in progress.
19497 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
19499 2017-04-18 Jeff Law <law@redhat.com>
19502 * mips.md ({return,simple_return}_internal): Do not overwrite
19505 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19507 PR tree-optimization/80443
19508 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
19509 instead of adding 1, subtract -1 and similarly instead of subtracting
19512 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
19514 PR rtl-optimization/80357
19515 * haifa-sched.c (tmp_bitmap): New variable.
19516 (model_recompute): Handle duplicate use records.
19517 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
19518 (free_global_sched_pressure_data): Free it.
19520 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
19523 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
19524 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
19525 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
19526 instead of SYSTEM_HEADER_DIR.
19528 2017-04-18 Jeff Law <law@redhat.com>
19530 PR middle-end/80422
19531 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
19532 predecessors after walking up the insn chain.
19534 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19537 * dwarf2out.c (modified_type_die): Try harder not to emit internal
19538 sizetype type into debug info.
19540 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
19543 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
19544 unneeded test for TARGET_UPPER_REGS_SF.
19545 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
19547 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19550 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
19551 instead of gsi_after_labels.
19553 2017-04-18 Jeff Law <law@redhat.com>
19555 * regcprop.c (maybe_mode_change): Avoid creating copies of the
19559 2017-04-13 Jeff Law <law@redhat.com>
19560 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
19561 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
19563 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
19566 * config/avr/avr.c (intl.h): Include it.
19567 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
19569 2017-04-18 Martin Liska <mliska@suse.cz>
19571 PR gcov-profile/78783
19572 * gcov-tool.c (gcov_output_files): Validate that destination
19573 file is either removed by the tool or by a user.
19575 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
19576 Guy Benyei <guybe@mellanox.com>
19578 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
19579 block, and do not negate it, the stored id is already negative.
19581 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
19583 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
19585 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
19588 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
19589 masks of options that should be turned off if the VSX vector
19590 options are turned off.
19591 (OTHER_P8_VECTOR_MASKS): Likewise.
19592 (OTHER_VSX_VECTOR_MASKS): Likewise.
19593 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
19594 rs6000_disable_incompatible_switches to validate no type switches
19596 (rs6000_incompatible_switch): New function to disallow turning on
19597 other vector options if -mno-vsx, -mno-power8-vector, or
19598 -mno-power9-vector are specified.
19600 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19602 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
19604 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19606 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
19607 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
19608 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
19609 (ARG_POINTER_CFA_OFFSET): Likewise.
19611 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19613 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
19614 conditions to take advantage of various optimizations.
19616 2017-04-13 Jeff Law <law@redhat.com>
19618 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
19619 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
19620 (zero_extendsidi2_dext): Likewise.
19622 2017-04-13 Jakub Jelinek <jakub@redhat.com>
19625 * fold-const.c (fold_ternary_loc): Revert
19626 use op0 instead of fold_convert_loc (loc, type, arg0) part of
19629 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
19631 PR rtl-optimization/80343
19632 * lra-remat.c (update_scratch_ops): Assign original hard reg to
19633 new scratch pseudo.
19635 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
19638 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
19639 to ubsan_encode_value.
19641 2017-04-13 Jeff Law <law@redhat.com>
19643 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
19644 appearing in DEBUG_INSNs.
19646 2017-04-13 Martin Liska <mliska@suse.cz>
19648 PR gcov-profile/80413
19649 * gcov-io.c (gcov_write_string): Copy to buffer just when
19650 allocated size is greater than zero.
19652 2017-04-13 Jakub Jelinek <jakub@redhat.com>
19655 * dwarf2out.c (decls_for_scope): Ignore declarations of
19656 current_function_decl in BLOCK_NONLOCALIZED_VARS.
19658 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
19661 * ipa-visibility.c (non_local_p): Fix typos.
19662 (localize_node): When localizing symbol in same comdat group,
19663 dissolve the group only when we know external symbols are going
19665 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
19667 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19669 PR tree-optimization/79390
19670 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
19671 order does not result in usable sequence, retry with reversed operand
19677 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
19678 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
19679 op0 instead of fold_convert_loc (loc, type, arg0).
19681 2017-04-12 Jeff Law <law@redhat.com>
19683 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
19684 has a delay slot in the generated code.
19686 * config/cris/cris.md (cris_preferred_reload_class): Return
19687 GENNONACR_REGS rather than GENERAL_REGS.
19689 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19692 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
19693 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
19694 signedness of the result type.
19696 2017-04-12 Richard Biener <rguenther@suse.de>
19697 Jeff Law <law@redhat.com>
19699 PR tree-optimization/80359
19700 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
19701 trim stores to TARGET_MEM_REFs.
19703 2017-04-12 Richard Biener <rguenther@suse.de>
19705 PR tree-optimization/79390
19706 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
19707 threading case even more.
19709 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
19712 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
19713 for quad_address_p for TImode, instead of just not indexed_address.
19715 2017-04-12 Richard Biener <rguenther@suse.de>
19716 Bernd Edlinger <bernd.edlinger@hotmail.de>
19718 PR middle-end/79671
19719 * alias.c (component_uses_parent_alias_set_from): Handle
19720 TYPE_TYPELESS_STORAGE.
19721 (get_alias_set): Likewise.
19722 * tree-core.h (tree_type_common): Add typeless_storage flag.
19723 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
19724 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
19725 for types containing members with TYPE_TYPELESS_STORAGE.
19726 (place_field): Likewise.
19727 (layout_type): Likewise for ARRAY_TYPE.
19728 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
19729 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
19730 TYPE_TYPELESS_STORAGE.
19731 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
19733 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19736 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
19737 first argument to type.
19739 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19743 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
19744 CONST0_RTX (mode) rather than const0_rtx where appropriate.
19745 (rs6000_expand_binop_builtin): Likewise.
19746 (rs6000_expand_ternop_builtin): Likewise; also add missing
19747 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
19748 vshasigma built-ins.
19749 * doc/extend.texi: Document that vec_xxpermdi's third argument
19750 must be a constant.
19752 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
19754 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
19755 Use shift_const cost parameter when calculating gain of STV shifts.
19757 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
19759 PR rtl-optimization/70478
19760 * lra-constraints.c (process_alt_operands): Check memory for
19761 disfavoring memory insn operand.
19763 2017-04-11 Jakub Jelinek <jakub@redhat.com>
19765 PR middle-end/80100
19766 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
19767 left shift in unsigned HOST_WIDE_INT type.
19769 PR rtl-optimization/80385
19770 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
19771 (not (neg X)) into (plus X -1) for complex or non-integral modes.
19774 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
19775 if they have any depend clauses.
19777 2017-04-11 Martin Liska <mliska@suse.cz>
19780 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
19781 * ipa-split.c (split_function): Create a local comdat symbol
19782 if caller is in a comdat group.
19784 2017-04-11 Martin Liska <mliska@suse.cz>
19787 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
19790 2017-04-11 Martin Sebor <msebor@redhat.com>
19792 PR middle-end/80364
19793 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
19794 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
19796 (directive::set_width, directive::set_precision, format_character):
19798 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
19801 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
19804 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
19805 conflict, set target->arch_name instead of target->cpu_name.
19807 2017-04-11 Richard Biener <rguenther@suse.de>
19809 PR tree-optimization/80374
19810 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
19811 build_zero_cst, remove fold_convertible_p check again.
19813 2017-04-11 Martin Liska <mliska@suse.cz>
19816 * ubsan.c (instrument_object_size): Do not instrument register
19819 2017-04-11 Jakub Jelinek <jakub@redhat.com>
19822 * config/i386/i386-builtin-types.def
19823 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
19824 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
19825 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
19826 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
19827 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
19828 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
19829 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
19830 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
19831 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
19832 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
19833 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
19834 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
19835 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
19836 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
19837 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
19838 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
19839 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
19840 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
19841 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
19842 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
19843 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
19844 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
19845 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
19846 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
19847 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
19848 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
19849 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
19850 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
19851 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
19852 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
19853 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
19854 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
19855 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
19856 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
19857 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
19858 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
19859 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
19860 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
19861 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
19862 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
19863 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
19864 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
19865 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
19866 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
19867 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
19868 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
19870 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
19871 flag to second_arg_count, handle 4 argument function type _COUNT
19872 aliases, handle second_arg_count on second argument rather than last.
19874 2017-04-10 Jeff Law <law@redhat.com>
19876 PR tree-optimization/80374
19877 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
19878 record anything if we can not convert integer_zero_node to the
19881 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
19884 * config/rs6000/rs6000.c (rs6000_option_override_internal):
19885 Enhance special handling given to the TARGET_P9_MINMAX option in
19886 relation to certain other options.
19888 2017-04-10 Bin Cheng <bin.cheng@arm.com>
19890 PR tree-optimization/80153
19891 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
19892 remove POINTER_PLUS_EXPR's base part directly, rather than through
19895 2017-04-10 Richard Biener <rguenther@suse.de>
19896 Bin Cheng <bin.cheng@arm.com>
19898 PR tree-optimization/80153
19899 * tree-affine.c (aff_combination_to_tree): Get base pointer from
19900 the first element of pointer type aff_tree. Build result expr in
19902 (add_elt_to_tree): Convert to type unconditionally. Remove other
19903 fold_convert calls.
19904 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
19905 (rewrite_use_nonlinear_expr): Check invariant using iv information.
19907 2017-04-10 Richard Biener <rguenther@suse.de>
19909 * tree-ssa-structalias.c (find_func_aliases): Properly handle
19912 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
19914 PR rtl-optimization/70478
19915 * lra-constraints.c (curr_small_class_check): New.
19916 (update_and_check_small_class_inputs): New.
19917 (process_alt_operands): Update curr_small_class_check. Disfavor
19918 alternative insn memory operands. Check available regs for small
19921 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
19924 * config/mips/mips.opt (-mvirt): Update description.
19925 * doc/invoke.texi (-mvirt): Likewise.
19927 2017-04-10 Richard Biener <rguenther@suse.de>
19929 PR middle-end/80362
19930 * fold-const.c (fold_binary_loc): Look at unstripped ops when
19931 looking for NEGATE_EXPR in -A / -B to A / B folding.
19933 2017-04-10 Martin Liska <mliska@suse.cz>
19935 PR gcov-profile/80224
19936 * gcov.c (print_usage): Fix usage string.
19937 (get_gcov_intermediate_filename): Remove.
19938 (output_gcov_file): Use both for normal and intermediate format.
19939 (generate_results): Do not initialize special file for
19940 intermediate format.
19942 2017-04-10 Richard Biener <rguenther@suse.de>
19944 PR tree-optimization/80304
19945 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
19948 2017-04-10 Nathan Sidwell <nathan@acm.org>
19951 * config/rs6000/rs6000.c (rs6000_vector_type): New.
19952 (rs6000_init_builtins): Use it.
19954 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19956 * config/arm/arm.md (<mrc>): Add mode to SET source.
19957 (<mrrc>): Likewise.
19959 2017-04-10 Richard Biener <rguenther@suse.de>
19961 PR middle-end/80344
19962 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
19964 2017-04-10 Jakub Jelinek <jakub@redhat.com>
19967 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
19968 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
19969 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
19970 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
19971 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
19972 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
19973 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
19974 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
19975 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
19976 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
19977 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
19978 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
19979 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
19980 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
19981 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
19982 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
19983 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
19984 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
19985 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
19986 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
19987 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
19988 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
19989 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
19991 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
19993 PR rtl-optimization/70478
19994 * lra-constraints.c: Reverse the last patch.
19996 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
19998 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
19999 Add comment for WCHAR_T.
20001 2017-04-08 Martin Liska <mliska@suse.cz>
20004 2017-04-07 Martin Liska <mliska@suse.cz>
20007 * ipa-split.c (split_function): Add function part to a same comdat
20010 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20013 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
20015 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
20017 * rs6000/rs6000.c (vec_load_pendulum): Rename...
20018 (vec_pairing): ...to this.
20019 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
20020 (rs6000_sched_init): Adjust for name change.
20021 (struct rs6000_sched_context): Likewise.
20022 (rs6000_init_sched_context): Likewise.
20023 (rs6000_set_sched_context): Likewise.
20025 2017-04-07 Jakub Jelinek <jakub@redhat.com>
20031 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
20033 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
20034 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
20035 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
20037 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
20039 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
20041 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
20043 PR rtl-optimization/70703
20044 * ira-color.c (update_conflict_hard_regno_costs): Use
20045 int64_t instead of HOST_WIDE_INT.
20047 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
20049 PR rtl-optimization/70478
20050 * lra-constraints.c (process_alt_operands): Disfavor alternative
20051 insn memory operands.
20053 2017-04-07 Jeff Law <law@redhat.com>
20055 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
20056 CALL and NOTE_INSN_CALL_ARG_LOCATION.
20058 2017-04-07 Martin Liska <mliska@suse.cz>
20061 * config/aarch64/aarch64.c (aarch64_process_target_attr):
20062 Show error message instead of an ICE.
20064 2017-04-07 Martin Liska <mliska@suse.cz>
20067 * ipa-split.c (split_function): Add function part to a same comdat
20070 2017-04-07 Richard Biener <rguenther@suse.de>
20072 PR middle-end/80341
20073 * tree.c (get_unwidened): Also handle ! for_type case for
20075 * convert.c (do_narrow): Split out from ...
20076 (convert_to_integer_1): ... here. Do not pass final truncation
20077 type to get_unwidened for TRUNC_DIV_EXPR.
20079 2017-04-07 Richard Biener <rguenther@suse.de>
20081 * tree-affine.c (wide_int_ext_for_comb): Take type rather
20083 (aff_combination_const): Adjust.
20084 (aff_combination_scale): Likewise.
20085 (aff_combination_add_elt): Likewise.
20086 (aff_combination_add_cst): Likewise.
20087 (aff_combination_convert): Likewise.
20088 (add_elt_to_tree): Likewise. Remove unused argument.
20089 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
20091 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
20093 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
20095 * config/arm/arm.c (arm_default_short_enums): Use
20096 ARM_DEFAULT_SHORT_ENUMS.
20097 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
20099 2017-04-06 Jakub Jelinek <jakub@redhat.com>
20102 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
20103 members with redundant out-of-class redeclaration.
20105 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
20108 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
20109 * config/i386/i386.md (*zero_extendsidi2):
20110 Add (?*x,*x) and (?*v,*v) alternatives.
20112 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
20115 * config/i386/i386.c (ix86_expand_builtin)
20116 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
20117 mode from insn data. Convert operands to insn operand mode.
20118 Copy operands that don't satisfy insn predicate to a register.
20120 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
20122 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
20125 2017-04-06 Richard Biener <rguenther@suse.de>
20127 PR tree-optimization/80334
20128 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
20129 preserve alignment of accesses.
20131 2017-04-06 Richard Biener <rguenther@suse.de>
20133 PR tree-optimization/80262
20134 * tree-sra.c (build_ref_for_offset): Preserve address-space
20136 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
20137 Drop useless address-space information on MEM_REF offsets.
20139 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
20141 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
20143 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
20145 PR rtl-optimization/70703
20146 * ira-color.c (update_conflict_hard_regno_costs): Use
20147 HOST_WIDE_INT instead of long.
20149 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
20152 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
20153 not defined for x86_64 target. Add -mmmx target option when __SSE2__
20155 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
20156 for x86_64 target. Handle -m3dnowa option.
20158 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
20160 PR rtl-optimization/70703
20161 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
20162 (update_conflict_hard_regno_costs): Use long instead of unsigned
20163 arithmetic for cost calculation.
20165 2017-04-05 Jakub Jelinek <jakub@redhat.com>
20166 Bernd Edlinger <bernd.edlinger@hotmail.de>
20169 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
20172 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
20175 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
20176 ptr_mode with Pmode throughout.
20177 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
20178 into probe_stack_range and use DImode.
20180 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
20183 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
20184 call_eh_return is true.
20186 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20188 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
20189 Initialize last_match_fntype_index.
20191 2017-04-05 Jakub Jelinek <jakub@redhat.com>
20194 * tree-nvr.c: Include internal-fn.h.
20195 (pass_return_slot::execute): Ignore internal calls without
20198 2017-04-04 Jakub Jelinek <jakub@redhat.com>
20199 Richard Biener <rguenther@suse.de>
20202 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
20203 captures used multiple times, except for the last use.
20204 * generic-match-head.c: Include gimplify.h.
20206 2017-04-04 Jakub Jelinek <jakub@redhat.com>
20208 PR tree-optimization/79390
20209 * target.h (struct noce_if_info): Declare.
20210 * targhooks.h (default_noce_conversion_profitable_p): Declare.
20211 * target.def (noce_conversion_profitable_p): New target hook.
20212 * ifcvt.h (struct noce_if_info): New type, moved from ...
20213 * ifcvt.c (struct noce_if_info): ... here.
20214 (noce_conversion_profitable_p): Renamed to ...
20215 (default_noce_conversion_profitable_p): ... this. No longer
20217 (noce_try_store_flag_constants, noce_try_addcc,
20218 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
20219 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
20220 instead of noce_conversion_profitable_p.
20221 * config/i386/i386.c: Include ifcvt.h.
20222 (ix86_option_override_internal): Don't override
20223 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
20224 (ix86_noce_conversion_profitable_p): New function.
20225 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
20226 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
20227 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
20228 * doc/tm.texi: Regenerated.
20230 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20232 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
20235 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20238 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
20239 instructions for small multiply cores.
20241 2017-04-04 Jeff Law <law@redhat.com>
20243 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
20245 (mips_expand_vec_perm_const): Initialize elements in orig_perm
20246 that are not set by the loop over the elements.
20248 2017-04-04 Jakub Jelinek <jakub@redhat.com>
20251 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
20252 int mode, convert_modes it to mode as unsigned, otherwise use
20253 lowpart_subreg to mode rather than SImode.
20254 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
20255 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
20256 Use DImode instead of SImode for the shift count operand.
20257 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
20260 2017-04-04 Richard Biener <rguenther@suse.de>
20262 PR middle-end/80281
20263 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
20264 arithmetic done for the negate or the plus. Simplify.
20265 (A - (-B) -> A + B): Likewise.
20266 * fold-const.c (split_tree): Make sure to not negate pointers.
20268 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
20270 PR rtl-optimization/60818
20271 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
20272 a compare of comparisons with the thing compared if this results
20273 in a different machine mode.
20275 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
20277 * alias.c (base_alias_check): Fix typo in comment.
20278 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
20279 * cgraphunit.c (symbol_table::compile): Likewise.
20280 * collect2.c (maybe_run_lto_and_relink): Likewise.
20281 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
20282 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
20283 * config/avr/avr.c (avr_map_op_t): Likewise.
20284 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
20285 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
20286 * config/epiphany/epiphany.md (movcc): Likewise.
20287 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
20288 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
20290 * config/mips/mips.c (mips_save_restore_reg): Likewise.
20291 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
20292 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
20293 * config/sh/sh.c (sh_rtx_costs): Likewise.
20294 * fold-const.c (fold_truth_andor): Likewise.
20295 * genautomata.c (collapse_flag): Likewise.
20296 * gengtype.h (struct type::u::s): Likewise.
20297 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
20298 * input.c (FORMAT_AMOUNT): Likewise.
20299 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
20300 (known_aggs_to_agg_replacement_list): Likewise.
20301 * ipa-inline-analysis.c: Likewise.
20302 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
20303 * ipa-polymorphic-call.c
20304 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
20305 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
20306 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
20308 * modulo-sched.c (apply_reg_moves): Likewise.
20309 * omp-expand.c (build_omp_regions_1): Likewise.
20310 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
20311 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
20312 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
20313 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
20314 * value-prof.c: Likewise.
20315 * var-tracking.c (val_reset): Likewise.
20317 2017-04-03 Richard Biener <rguenther@suse.de>
20319 PR tree-optimization/80275
20320 * fold-const.c (split_address_to_core_and_offset): Handle
20323 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
20325 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
20326 descriptors is at least equal to that of functions.
20328 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
20330 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
20332 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
20335 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
20336 (mov<IMOD4:mode>): New expander.
20337 (*mov<IMOD4:mode>_internal): New insn and split pattern.
20339 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
20341 PR rtl-optimization/79405
20342 * fwprop.c (propagations_left): New variable.
20343 (forward_propagate_into): Decrement it.
20344 (fwprop_init): Initialize it.
20345 (fw_prop): If the variable has reached zero, stop propagating.
20346 (fwprop_addr): Ditto.
20348 2017-03-31 Jakub Jelinek <jakub@redhat.com>
20351 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
20352 a FUNCTION_DECL, pass it as decl instead of origin to
20355 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
20357 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
20360 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
20363 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
20364 TARGET_VSX_SMALL_INTEGER.
20366 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20368 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20369 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
20371 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
20373 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
20374 extraction from odd-numbered MSA register.
20376 2017-03-31 Jakub Jelinek <jakub@redhat.com>
20378 PR middle-end/80173
20379 * expmed.c (store_bit_field_1): Don't attempt to create
20380 a word subreg out of hard registers wider than word if they
20381 have HARD_REGNO_NREGS of 1 for their mode.
20383 PR middle-end/80163
20384 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
20385 conversions to integer types wider than word and pointer.
20388 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
20389 (rtx_equal_for_cselib_p): Pass 0 to it.
20390 * cselib.c (cselib_hasher::equal): Likewise.
20391 (rtx_equal_for_cselib_1): Add depth argument. If depth
20392 is 128, don't look up VALUE locs and punt. Increment
20393 depth in recursive calls when walking VALUE locs.
20395 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
20397 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
20398 (make_gcov_file_name): Use the canonical path name for generating
20400 (read_line): Fix handling of files with ascii null bytes.
20402 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
20404 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
20405 to initialise a vector register instead
20406 of using a const_int.
20408 2017-03-30 Jakub Jelinek <jakub@redhat.com>
20410 PR translation/80189
20411 * gimplify.c (omp_default_clause): Use %qs instead of %s in
20412 diagnostic messages.
20414 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
20417 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
20418 (dfp_diex_<mode>): Update mode of operand 1.
20419 * doc/extend.texi (dxex, dxexq): Document change to return type.
20420 (diex, diexq): Document change to argument type.
20422 2017-03-30 Martin Jambor <mjambor@suse.cz>
20425 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
20426 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
20427 it reflects the signature changes performed at the callee side.
20428 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
20429 to cgraph_build_function_type_skip_args.
20430 (build_function_decl_skip_args): Adjust call to the above function.
20432 2017-03-30 Jakub Jelinek <jakub@redhat.com>
20435 * config/i386/sse.md
20436 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
20437 register as dest whenever it is a MEM not rtx_equal_p to the
20438 corresponding dup operand, and when forcing into reg move the
20439 reg into the memory afterwards.
20440 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
20441 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
20442 for the force_reg mode.
20443 (avx512vl_vextractf128<mode>): Use register as dest either
20444 always when a MEM, or when it is a MEM not rtx_equal_p to the
20445 corresponding dup operand, or even not when it is a CONST_VECTOR
20446 depending on the mode and lo vs. hi.
20447 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
20449 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
20450 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
20451 Likewise. Require that operands[2] is even.
20452 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
20453 Remove extraneous parens. Require that operands[2] is a multiple
20455 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
20456 operands[0] is a MEM if <mask_applied>, the predicates/constraints
20457 disallow memory then.
20459 2017-03-30 Richard Biener <rguenther@suse.de>
20461 PR tree-optimization/77498
20462 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
20463 to non-constants over backedges.
20465 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
20467 PR rtl-optimization/80233
20468 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
20469 as last_combined_insn. Do not test for BARRIER_P separately.
20471 2017-03-29 Andreas Schwab <schwab@suse.de>
20474 * calls.c (prepare_call_address): Convert funexp to Pmode before
20475 copying to temp reg.
20477 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20479 PR tree-optimization/80158
20480 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
20481 Handle possible future case of more than one alternate
20483 (replace_rhs_if_not_dup): Likewise.
20484 (replace_one_candidate): Likewise.
20486 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
20488 PR rtl-optimization/80193
20489 * ira.c (ira): Do not check allocation for LRA.
20491 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
20493 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
20494 (nvptx_output_simt_exit): Declare.
20495 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
20496 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
20497 (init_softstack_frame): Move initialization of crtl->is_leaf to...
20498 (nvptx_declare_function_name): ...here. Emit declaration of local
20499 memory space buffer for omp_simt_enter insn.
20500 (nvptx_output_unisimt_switch): New.
20501 (nvptx_output_softstack_switch): New.
20502 (nvptx_output_simt_enter): New.
20503 (nvptx_output_simt_exit): New.
20504 * config/nvptx/nvptx.h (struct machine_function): New fields
20505 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
20506 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
20507 (UNSPECV_SIMT_EXIT): Ditto.
20508 (omp_simt_enter_insn): New insn.
20509 (omp_simt_enter): New expansion.
20510 (omp_simt_exit): New insn.
20511 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
20513 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
20514 (expand_GOMP_SIMT_ENTER_ALLOC): New.
20515 (expand_GOMP_SIMT_EXIT): New.
20516 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
20517 (GOMP_SIMT_ENTER_ALLOC): Ditto.
20518 (GOMP_SIMT_EXIT): Ditto.
20519 * target-insns.def (omp_simt_enter): New insn.
20520 (omp_simt_exit): Ditto.
20521 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
20523 (lower_rec_simd_input_clauses): Implement SIMT privatization.
20524 (lower_rec_input_clauses): Likewise.
20525 (lower_lastprivate_clauses): Handle SIMT privatization.
20527 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
20528 (ompdevlow_adjust_simt_enter): New.
20529 (find_simtpriv_var_op): New.
20530 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
20531 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
20533 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
20534 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
20535 (copy_decl_for_dup_finish): Ditto.
20537 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
20539 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
20542 * config/i386/i386.c (ix86_option_override_internal): Always
20543 allow -mpreferred-stack-boundary=3 for 64-bit targets.
20545 2017-03-28 Bin Cheng <bin.cheng@arm.com>
20547 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
20549 2017-03-28 Bin Cheng <bin.cheng@arm.com>
20551 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
20552 mark new edge's irreducible flag accordign to it.
20553 (vect_do_peeling): Check loop preheader edge's irreducible flag
20554 and pass it to function slpeel_add_loop_guard.
20556 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
20558 PR tree-optimization/80218
20559 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
20560 Update block frequencies and counts.
20562 2017-03-28 Richard Biener <rguenther@suse.de>
20564 PR tree-optimization/78644
20565 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
20566 of a simplification result we may not use it at all.
20568 2017-03-28 Richard Biener <rguenther@suse.de>
20571 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
20572 without arguments, generate default definition of a SSA name.
20574 2017-03-28 Richard Biener <rguenther@suse.de>
20576 PR middle-end/80222
20577 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
20578 TYPE_REF_CAN_ALIAS_ALL references.
20579 * fold-const.c (fold_indirect_ref_1): Likewise.
20581 2017-03-28 Martin Liska <mliska@suse.cz>
20584 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
20585 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
20587 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
20588 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
20590 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
20591 (EXTRA_SPECS): Define.
20592 (SUBTARGET_EXTRA_SPECS): Likewise.
20593 (SUBTARGET_CPP_SPEC): Likewise.
20594 * config/arc/elf.h (EXTRA_SPECS): Renamed to
20595 SUBTARGET_EXTRA_SPECS.
20596 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
20598 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
20600 * config/arc/simdext.md (vst64_insn): Update pattern.
20601 (vld32wh_insn): Likewise.
20602 (vld32wl_insn): Likewise.
20603 (vld64_insn): Likewise.
20604 (vld32_insn): Likewise.
20606 2017-03-28 Marek Polacek <polacek@redhat.com>
20609 * fold-const.c (fold_comparison): Use protected_set_expr_location
20610 instead of SET_EXPR_LOCATION.
20612 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
20614 * tree.c (add_expr): Avoid name lookup warning.
20616 2017-03-27 Jeff Law <law@redhat.com>
20618 PR tree-optimization/80216
20619 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
20620 function name. Limit recursion depth.
20621 (record_temporary_equivalences): Corresponding changes.
20623 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
20625 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
20628 2017-03-27 Jakub Jelinek <jakub@redhat.com>
20631 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
20633 * cfgcleanup.c (reg_note_cfa_p): New array.
20634 (insns_have_identical_cfa_notes): New function.
20635 (old_insns_match_p): Don't cross-jump in between /f
20636 and non-/f instructions. If both i1 and i2 are frame related,
20637 verify all CFA notes, their order and content.
20639 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
20642 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
20643 HImode and SImode with zero extend to DImode to one insn.
20644 (bswap<mode>2_extenddi): Likewise.
20645 (bswapsi2_extenddi): Likewise.
20646 (bswaphi2_extendsi): Likewise.
20647 (bswaphi2): Combine bswap HImode and SImode into one insn.
20648 Separate memory insns from swapping register.
20649 (bswapsi2): Likewise.
20650 (bswap<mode>2): Likewise.
20651 (bswaphi2_internal): Delete, no longer used.
20652 (bswapsi2_internal): Likewise.
20653 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
20654 store, and gpr<-gpr swap insns.
20655 (bswap<mode>2_store): Likewise.
20656 (bswaphi2_reg): Register only splitter, combine with the splitter.
20657 (bswaphi2 splitter): Likewise.
20658 (bswapsi2_reg): Likewise.
20659 (bswapsi2 splitter): Likewise.
20660 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
20661 the insns into load, store, and register/register insns.
20662 (bswapdi2_ldbrx): Likewise.
20663 (bswapdi2_load): Likewise.
20664 (bswapdi2_store): Likewise.
20665 (bswapdi2_reg): Likewise.
20667 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
20669 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
20670 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
20672 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
20675 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
20677 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
20678 special handling for target option conflicts between dform
20679 options (-mpower9-dform, -mpower9-dform-vector,
20680 -mpower9-dform-scalar) and -mno-direct-move.
20682 2017-03-27 Richard Biener <rguenther@suse.de>
20684 PR tree-optimization/80181
20685 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
20687 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20689 * config/arc/predicates.md (move_double_src_operand): Replace the
20690 call to move_double_src_operand with a call to address_operand.
20692 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20694 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
20695 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
20696 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
20698 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20700 * config/arc/predicates.md (long_immediate_loadstore_operand):
20701 Consider scaled addresses cases.
20703 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20705 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
20706 restored when in interrupt.
20707 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
20708 doesn't have delay slot.
20710 2017-03-27 Richard Biener <rguenther@suse.de>
20713 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
20714 inlined thunk clones.
20716 2017-03-27 Jakub Jelinek <jakub@redhat.com>
20719 * asan.c (instrument_derefs): Copy over last operand from
20720 original COMPONENT_REF to the new COMPONENT_REF with
20721 DECL_BIT_FIELD_REPRESENTATIVE.
20722 * ubsan.c (instrument_object_size): Likewise.
20724 2017-03-27 Richard Biener <rguenther@suse.de>
20726 PR tree-optimization/80170
20727 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
20728 sure DR/SCEV didnt fold in constants we do not see when looking
20729 at the reference base alignment.
20731 2017-03-27 Richard Biener <rguenther@suse.de>
20733 PR middle-end/80171
20734 * gimple-fold.c (fold_ctor_reference): Properly guard against
20735 NULL return value from canonicalize_constructor_val.
20737 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
20740 * config/i386/i386.c (ix86_expand_builtin)
20741 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
20742 flags reg setting and flags reg using instructions.
20743 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
20744 clobbering instructions to zero extend op2.
20746 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
20748 * doc/install.texi (Configuration) <--with-aix-soname>:
20749 Update link to AIX ld.
20751 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
20753 PR rtl-optimization/80160
20754 PR rtl-optimization/80159
20755 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
20756 reg_alternate_class into account.
20758 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
20761 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
20762 to consider in curr_insn_transform.
20764 2017-03-24 Jakub Jelinek <jakub@redhat.com>
20766 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
20767 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
20768 and emit_mode_inner.
20770 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20772 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
20773 argument to the overloaded builtin variants. Use the new flag to
20774 deprecate certain builtin variants.
20775 * config/s390/s390-builtin-types.def: Add new builtin types.
20776 * config/s390/s390-builtins.h: Support new flags field for
20777 overloaded builtins.
20778 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
20779 (s390_macro_to_expand): Enable vector float data type.
20780 (s390_cpu_cpp_builtins_internal): Indicate support of the new
20781 builtins by incrementing the __VEC__ version number.
20782 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
20784 (s390_resolve_overloaded_builtin): Emit error messages depending
20785 on the builtin flags.
20786 * config/s390/s390.c (s390_expand_builtin): Support additional
20787 flags argument. Change error message to match the messages
20788 emitted in s390-c.c.
20789 * config/s390/s390.md: New UNSPEC_* constants.
20790 (op_type): Add new instruction types.
20791 * config/s390/vecintrin.h: Add new builtins and test data class
20793 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
20794 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
20795 (VEC_INEXACT, VEC_NOINEXACT): New constants.
20796 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
20797 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
20798 ("vec_mergel<mode>"): V_HW -> VEC_HW.
20800 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
20801 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
20802 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
20803 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
20805 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
20806 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
20807 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
20808 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
20810 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
20811 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
20812 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
20813 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
20814 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
20815 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
20816 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
20818 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
20819 ("vec_scatter_element<V_HW_4:mode>_DI")
20820 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
20821 ("vec_fpint<mode>", "vflls")
20822 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
20823 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
20824 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
20825 ("*vec_cmphe<mode>_cc"): ... these.
20827 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
20828 mode constant instead of magic value.
20830 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20832 * config/s390/s390.c (s390_expand_vec_compare): Support other
20833 vector floating point modes than just V2DF.
20834 (s390_expand_vcond): Likewise.
20835 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
20836 (s390_cannot_change_mode_class): Prevent mode changes between TF
20837 and V1TF in vector registers.
20838 * config/s390/s390.md (DF, SF): New mode attributes.
20839 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
20840 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
20841 SFmode support for VRs.
20842 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
20844 (VFT, VF_HW): New mode iterators.
20845 (vw, sdx): New mode attributes.
20846 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
20847 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
20848 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
20849 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
20850 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
20851 also the new vector floating point modes. Renaming to ...
20853 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
20854 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
20855 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
20856 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
20857 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
20858 ("vec_unordered<mode>"): ... these.
20860 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
20861 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
20862 ("*vec_extendv2df"): New insn definitions.
20864 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20866 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
20867 ("mulditi3_2", "*muldi3_sign"): New patterns.
20868 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
20869 rename the pattern definition.
20871 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20873 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
20875 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
20877 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20879 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
20880 instruction if possible.
20881 * config/s390/vector.md (vec_halfnumelts): New mode
20883 ("*vec_vllezlf<mode>"): New pattern.
20885 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20887 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
20888 ("popcountv4si2", "popcountv2di2"): Rename to ...
20889 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
20890 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
20892 ("popcount<mode>2_vxe"): New pattern.
20894 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20896 * common/config/s390/s390-common.c (processor_flags_table): Add
20898 * config.gcc: Add arch12.
20899 * config/s390/driver-native.c (s390_host_detect_local_cpu):
20900 Default to arch12 for unknown CPU model numbers.
20901 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
20902 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
20903 PROCESSOR_max sanity check.
20904 * config/s390/s390-opts.h (enum processor_type): Add
20906 * config/s390/s390.c (processor_table): Add arch12.
20907 (s390_expand_builtin): Add check for B_VXE flag.
20908 (s390_issue_rate): Add PROCESSOR_ARCH12.
20909 (s390_get_sched_attrmask): Likewise.
20910 (s390_get_unit_mask): Likewise.
20911 (s390_sched_score): Enable z13 scheduling for arch12.
20912 (s390_sched_reorder): Likewise.
20913 (s390_sched_variable_issue): Likewise.
20914 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
20916 (s390_tune_attr): Use z13 scheduling also for arch12.
20917 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
20918 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
20919 (TARGET_VXE_P): New macros.
20920 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
20921 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
20922 * config/s390/s390.opt: Add arch12 as processor_type.
20924 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20926 * config/s390/s390.md
20927 ("fixuns_truncdddi2", "fixuns_trunctddi2")
20928 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
20929 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
20931 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
20932 Rename expanders to ...
20934 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
20935 ("fixuns_truncdddi2_emu"): ... these.
20937 ("fixuns_trunc<mode>si2_emu"): New expander.
20939 ("*fixuns_truncdfdi2_z13"): Rename to ...
20940 ("*fixuns_truncdfdi2_vx"): ... this.
20942 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20944 * config/s390/2964.md: Remove the single element vector compare
20945 instructions which are no longer used.
20946 * config/s390/s390.c (s390_select_ccmode): Remove handling of
20948 (s390_canonicalize_comparison): Remove handling of DFmode
20950 (s390_expand_vec_compare_scalar): Remove function.
20951 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
20952 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
20954 ("*cmp<mode>_ccs"): Add wfcdb instruction.
20956 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20958 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
20960 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
20961 will anyway by matched by mov<mode>_64dfp.
20963 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20965 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
20966 vlef/vstef. Add missing operand to vleif.
20968 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20970 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
20971 pair for all vector types with 64 bit elements.
20972 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
20973 * config/s390/vector.md (V_HW_64): ... here.
20974 (V_128_NOSINGLE): New mode iterator.
20975 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
20976 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
20977 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
20978 ("*vec_load_pairv2di"): Change to ...
20979 ("*vec_load_pair<mode>"): ... this one.
20981 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20983 * config/s390/constraints.md: Add comments.
20984 (jKK): Reject element sizes > 8 bytes.
20985 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
20987 * config/s390/s390.md: Add the s_operand checks formerly in
20988 s390_split_ok_p to various splitters where they are still
20990 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
20991 for 128 bit vectors. Plus two splitters.
20993 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20995 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
20998 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21001 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
21002 error if the boundary argument is not constant.
21004 2017-03-24 Jakub Jelinek <jakub@redhat.com>
21006 PR rtl-optimization/80112
21007 * loop-doloop.c (doloop_condition_get): Don't check condition
21008 if cmp isn't SET with IF_THEN_ELSE src.
21010 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21012 PR tree-optimization/80158
21013 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
21014 replacing a candidate statement, also replace it for the
21015 candidate's alternate interpretation.
21016 (replace_rhs_if_not_dup): Likewise.
21017 (replace_one_candidate): Likewise.
21019 2017-03-24 Richard Biener <rguenther@suse.de>
21021 PR tree-optimization/80167
21022 * graphite-isl-ast-to-gimple.c
21023 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
21025 (translate_isl_ast_to_gimple::get_rename): Likewise.
21027 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
21029 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
21030 handling of certain combinations of target options, including the
21031 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
21032 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
21034 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21037 * config/arm/arm.md (*load_multiple): Add reload_completed to
21038 matching condition.
21040 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21041 Richard Biener <rguenth@suse.de>
21043 PR tree-optimization/79908
21044 PR tree-optimization/80136
21045 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
21046 been cast away, gimplify_and_add suffices.
21048 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
21050 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
21052 2017-03-23 Richard Biener <rguenther@suse.de>
21054 PR tree-optimization/80032
21055 * gimplify.c (gimple_push_cleanup): Forced unconditional
21056 cleanups still have to go to the conditional_cleanups
21059 2017-03-22 Jakub Jelinek <jakub@redhat.com>
21061 PR tree-optimization/80072
21062 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
21064 (next_operand_entry_id): Change type to unsigned int.
21065 (sort_by_operand_rank): Make sure to return the right return value
21066 even if unsigned fields are bigger than INT_MAX.
21067 (struct oecount): Change cnt and id type to unsigned int.
21068 (oecount_hasher::equal): Formatting fix.
21069 (oecount_cmp): Make sure to return the right return value
21070 even if unsigned fields are bigger than INT_MAX.
21071 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
21074 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
21075 TREE_READONLY on result if writing it more than once.
21078 * doc/invoke.texi (-fsanitize=thread): Document that with
21079 -fnon-call-exceptions atomics are not able to throw
21083 * tsan.c: Include tree-eh.h.
21084 (instrument_builtin_call): Call maybe_clean_eh_stmt or
21085 maybe_clean_or_replace_eh_stmt where needed.
21086 (instrument_memory_accesses): Add cfg_changed argument.
21087 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
21088 if it returned true.
21089 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
21091 PR rtl-optimization/63191
21092 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
21093 wrapper function, moved the whole old content into ...
21094 (ix86_delegitimize_address_1): ... this. New inline function.
21095 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
21096 true as last argument instead of ix86_delegitimize_address.
21098 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
21100 * config/aarch64/aarch64.c (generic_branch_cost): Copy
21101 cortexa57_branch_cost.
21103 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
21105 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
21107 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21110 * doc/md.texi (Constraints): Document wA constraint.
21111 * config/rs6000/constraints.md (wA): New.
21112 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
21113 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
21114 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
21115 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
21117 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
21120 * gimplify.c (is_oacc_declared): New function.
21121 (oacc_default_clause): Use it to set default flags for acc declared
21122 variables inside parallel regions.
21123 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
21124 declared variables.
21125 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
21126 declare attribute to any decl as necessary.
21128 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
21131 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
21132 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
21133 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
21134 (arm_arch_lpae): This.
21135 * config/arm/arm.c (arm_arch7ve): Rename into ...
21136 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
21137 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
21140 2017-03-22 Martin Liska <mliska@suse.cz>
21143 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
21144 error message instead of an ICE.
21146 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21148 * doc/extend.texi (6.11 Additional Floating Types): Revise.
21150 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
21152 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
21154 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
21157 2017-03-21 Martin Sebor <msebor@redhat.com>
21159 * doc/extend.texi: Use "cannot" instead of "can't."
21160 * doc/hostconfig.texi: Same.
21161 * doc/install.texi: Same.
21162 * doc/invoke.texi: Same.
21163 * doc/loop.texi: Same.
21164 * doc/md.texi: Same.
21165 * doc/objc.texi: Same.
21166 * doc/rtl.texi: Same.
21167 * doc/tm.texi: Same.
21168 * doc/tm.texi.in: Same.
21169 * doc/trouble.texi: Same.
21171 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
21174 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
21175 (collect_checksum_attributes): Set it.
21176 (die_checksum_ordered): Use it.
21178 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21180 PR tree-optimization/79908
21181 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
21182 change: For a VA_ARG whose LHS has been cast away, use
21183 force_gimple_operand to construct the side effects.
21185 2017-03-21 David Malcolm <dmalcolm@redhat.com>
21187 PR translation/80001
21188 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
21189 more amenable to translation.
21190 (oacc_loop_auto_partitions): Likewise.
21192 2017-03-21 Marek Polacek <polacek@redhat.com>
21193 Martin Sebor <msebor@redhat.com>
21195 PR tree-optimization/80109
21196 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
21197 on INTEGRAL_TYPE_P.
21199 2017-03-21 Jakub Jelinek <jakub@redhat.com>
21200 Segher Boessenkool <segher@kernel.crashing.org>
21203 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
21204 check reg_used_between_p between insn and one of succ or succ2
21205 depending on if succ is artificial insn not inserted into insn
21208 2017-03-21 Martin Liska <mliska@suse.cz>
21210 PR gcov-profile/80081
21211 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
21212 * doc/gcc.texi: Include gcov-dump stuff.
21213 * doc/gcov-dump.texi: New file.
21215 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
21217 PR rtl-optimization/79150
21218 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
21219 conditional jump, if the jump is the last insn of the loop.
21221 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21222 Richard Biener <rguenth@suse.de>
21224 PR tree-optimization/79908
21225 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
21226 been cast away, use force_gimple_operand to construct the side
21229 2017-03-21 Martin Liska <mliska@suse.cz>
21231 PR libfortran/79956
21232 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
21235 2017-03-21 Brad Spengler <spender@grsecurity.net>
21238 * plugin.c (htab_hash_plugin): New function.
21239 (add_new_plugin): Use it and adjust.
21240 (parse_plugin_arg_opt): Adjust.
21241 (init_one_plugin): Likewise.
21243 2017-03-21 Richard Biener <rguenther@suse.de>
21245 PR tree-optimization/80032
21246 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
21247 if set force the cleanup to happen unconditionally.
21248 (gimplify_target_expr): Push inserted clobbers with force_uncond
21249 to avoid them being removed by control-dependent DCE.
21251 2017-03-21 Richard Biener <rguenther@suse.de>
21253 PR tree-optimization/80122
21254 * tree-inline.c (copy_bb): Do not expans va-arg packs or
21255 va_arg_pack_len when the inlined call stmt requires pack
21257 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
21259 2017-03-21 Jakub Jelinek <jakub@redhat.com>
21262 * tsan.c (instrument_builtin_call): If the memory model argument
21263 is not a constant, assume it is valid.
21266 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
21269 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
21271 PR rtl-optimization/79910
21272 * combine.c (can_combine_p): Do not allow combining an I0 or I1
21273 if its dest is used by an insn before I2 (other than the combined
21274 insns themselves, which are properly handled already).
21276 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
21279 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
21281 * combine.c (record_used_regs): New static function.
21282 (try_combine): Handle situations where there is an additional
21283 instruction between I2 and I3 which needs to have a LOG_LINK
21287 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
21289 * combine.c (try_combine): Delete redundant i1 test. Call
21290 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
21292 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21295 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
21296 alternatives 13/14.
21298 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21300 PR tree-optimization/80054
21301 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
21302 the optimization if a PHI or any of its arguments is not dominated
21303 by the candidate's basis. Use gphi* rather than gimple* as
21305 (replace_profitable_candidates): Clean up a gimple* variable that
21306 should be a gphi* variable.
21308 2017-03-20 Martin Sebor <msebor@redhat.com>
21311 * doc/extend.texi (attribute constructor): Document present limitation.
21313 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
21316 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
21317 __POWER9_VECTOR__ #ifdef control, change template definition to
21318 use Power9-specific built-in function.
21319 (vec_any_eq): Likewise.
21320 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
21321 to control outcomes from this test.
21322 (vector_ae_<mode>p): For VEC_F modes, likewise.
21324 2017-03-20 Ian Lance Taylor <iant@google.com>
21326 * config/i386/i386.c (ix86_function_regparm): Save an extra
21327 register for -fsplit-stack with DECL_STATIC_CHAIN.
21329 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
21332 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
21333 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
21335 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
21337 * config/riscv/riscv.c (riscv_print_operand): Use "fence
21339 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
21342 2017-03-20 Marek Polacek <polacek@redhat.com>
21345 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
21347 2017-03-20 Richard Biener <rguenther@suse.de>
21349 PR tree-optimization/80113
21350 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
21351 allocate extra SSA name for PHI def.
21352 (add_close_phis_to_outer_loops): Likewise.
21353 (add_close_phis_to_merge_points): Likewise.
21354 (copy_loop_close_phi_args): Likewise.
21355 (copy_cond_phi_nodes): Likewise.
21357 2017-03-20 Martin Liska <mliska@suse.cz>
21359 PR middle-end/79753
21360 * tree-chkp.c (chkp_build_returned_bound): Do not build
21361 returned bounds for a LHS that's not a BOUNDED_P type.
21363 2017-03-20 Martin Liska <mliska@suse.cz>
21367 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
21368 COMPLEX_CST and VECTOR_CST.
21370 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21373 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
21374 target operand. A new splitter adds the clobber statement in case
21375 the target operand is dead anyway.
21377 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
21379 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
21380 to age-old versions of binutils and glibc.
21382 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
21384 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
21386 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
21388 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
21390 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
21392 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
21393 requirement for binutils 2.13.
21395 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
21397 * combine.c (try_combine): Delete redundant i1 test. Call
21398 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
21400 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
21402 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
21403 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
21405 <riscv64-*-elf>: Re-arrange section
21406 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
21407 <riscv32-*-linux>: Likewise.
21408 <riscv64-*-elf>: Likewise
21409 <riscv64-*-linux>: Likewise.
21411 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
21414 * aarch64.opt(verbose-cost-dump): Fix typo.
21416 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
21419 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
21420 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
21422 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
21424 * reload.c (find_reloads): When reloading a nonoffsettable address,
21425 use RELOAD_OTHER for it and its address reloads.
21427 PR rtl-optimization/79910
21428 * combine.c (record_used_regs): New static function.
21429 (try_combine): Handle situations where there is an additional
21430 instruction between I2 and I3 which needs to have a LOG_LINK
21433 2017-03-17 Jeff Law <law@redhat.com>
21435 PR tree-optimization/71437
21436 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
21437 conditional in the hash table first.
21438 (vrp_dom_walker::before_dom_children): Extract condition from
21439 ASSERT_EXPR. Record condition, its inverion and any implied
21440 conditions as well.
21442 2017-03-17 Marek Polacek <polacek@redhat.com>
21443 Markus Trippelsdorf <markus@trippelsdorf.de>
21445 PR tree-optimization/80079
21446 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
21449 2017-03-17 Richard Biener <rguenther@suse.de>
21451 PR middle-end/80075
21452 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
21453 Properly verify the LHS before the RHS possibly claims to be
21455 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
21458 2017-03-17 Martin Jambor <mjambor@suse.cz>
21460 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
21461 (List of -O2 options): Likewise.
21462 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
21465 2017-03-17 Tom de Vries <tom@codesourcery.com>
21467 * gcov-dump.c (print_usage): Print bug_report_url.
21469 2017-03-17 Richard Biener <rguenther@suse.de>
21471 PR middle-end/80050
21472 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
21473 (parser::peek): Likewise.
21475 2017-03-17 Richard Biener <rguenther@suse.de>
21477 PR tree-optimization/80048
21478 * sese.c (free_sese_info): Properly release rename_map and
21479 copied_bb_map elements.
21481 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
21483 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
21484 Add linked-list forward and backlinks. Insert on
21485 construction, remove on destruction.
21486 (class pass_store_merging): Add m_stores_head field.
21487 (pass_store_merging::terminate_and_process_all_chains):
21488 Iterate over m_stores_head list.
21489 (pass_store_merging::terminate_all_aliasing_chains):
21491 (pass_store_merging::execute): Check for debug stmts first.
21492 Push new chains onto the m_stores_head stack.
21494 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
21497 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
21498 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
21499 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
21501 2017-03-16 Jeff Law <law@redhat.com>
21503 PR tree-optimization/71437
21504 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
21505 member function. Implementation moved into after_dom_children
21506 member function and into the threader's thread_outgoing_edges
21508 (dom_opt_dom_walker::after_dom_children): Simplify by moving
21509 some code into new thread_outgoing_edges.
21510 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
21511 definition. Simplify marker handling (do it here). Assume we always
21512 have the available expression and the const/copies tables.
21513 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
21515 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
21516 * tree-vrp.c (equiv_stack): No longer file scoped.
21517 (vrp_dom_walker): New class.
21518 (vrp_dom_walker::before_dom_children): New member function.
21519 (vrp_dom_walker::after_dom_children): Likewise.
21520 (identify_jump_threads): Setup domwalker. Use it rather than
21521 walking edges in a random order by hand. Simplify setup/finalization.
21522 (finalize_jump_threads): Remove.
21523 (vrp_finalize): Do not call identify_jump_threads here.
21524 (execute_vrp): Do it here instead and call thread_through_all_blocks
21527 PR tree-optimization/71437
21528 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
21530 (simplify_stmt_for_jump_threading): Add basic_block argument. All
21532 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
21533 (dom_opt_dom_walker::thread_across_edge): Remove
21534 handle_dominating_asserts argument. All callers changed.
21535 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
21536 changes. Remove calls to lhs_of_dominating_assert. Other
21537 uses of handle_dominating_asserts turn into unconditional code
21538 (simplify_control_stmt_condition_1): Likewise.
21539 (simplify_control_stmt_condition): Likewise.
21540 (thread_through_normal_block, thread_across_edge): Likewise.
21541 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
21542 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
21543 object if it is not an SSA_NAME.
21544 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
21545 before calling into the VRP specific simplifiers.
21546 (identify_jump_threads): Remove handle_dominating_asserts
21549 2017-03-16 Jakub Jelinek <jakub@redhat.com>
21552 * tree-diagnostic.c (default_tree_printer): No longer static.
21553 * tree-diagnostic.h (default_tree_printer): New prototype.
21555 2017-03-16 Tamar Christina <tamar.christina@arm.com>
21557 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
21558 Change ins into fmov.
21560 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21562 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
21563 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
21564 Use h_con constraint for operand 1.
21565 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
21566 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
21568 2017-03-15 Jeff Law <law@redhat.com>
21570 PR tree-optimization/71437
21571 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
21572 (record_temporary_equivalences): Use it.
21574 PR tree-optimization/71437
21575 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
21576 tree-ssa-scopedtables.
21577 (lookup_avail_expr, build_and_record_new_cond): Likewise.
21578 (record_conditions, record_cond, vuse_eq): Likewise.
21579 (record_edge_info): Adjust to API tweak of record_conditions.
21580 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
21581 (record_temporary_equivalences, optimize_stmt): Likewise.
21582 (eliminate_redundant_computations): Likewise.
21583 (record_equivalences_from_stmt): Likewise.
21584 * tree-ssa-scopedtables.c: Include options.h and params.h.
21585 (vuse_eq): New function, moved from tree-ssa-dom.c
21586 (build_and_record_new_cond): Likewise.
21587 (record_conditions): Likewise. Accept vector of conditions rather
21588 than edge_equivalence structure for first argument.
21589 for the first argument.
21590 (avail_exprs_stack::lookup_avail_expr): New member function, moved
21591 from tree-ssa-dom.c.
21592 (avail_exprs_stack::record_cond): Likewise.
21593 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
21594 from tree-ssa-dom.c.
21595 (avail_exprs_stack): Add new member functions lookup_avail_expr
21597 (record_conditions): Declare.
21599 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
21602 * lra-constraints.c (process_alt_operands): Increase reject for
21603 reloading an input/output operand.
21605 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
21608 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
21609 insns to convert from signed/unsigned char/short to IEEE 128-bit
21611 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
21613 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
21616 * config/i386/i386.c (ix86_vector_duplicate_value): Create
21617 subreg of inner mode for values already in registers.
21619 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
21621 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
21622 iteration reg is used after the loop.
21624 2017-03-14 Martin Sebor <msebor@redhat.com>
21626 PR tree-optimization/79800
21627 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
21628 precision in negative-positive range.
21629 (format_floating): Call non-const overload with adjusted precision.
21631 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
21634 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
21637 2017-03-14 Martin Sebor <msebor@redhat.com>
21639 PR middle-end/80020
21640 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
21641 * builtins.def (aligned_alloc): Use it.
21644 * Makefile.in (GTFILES): Add calls.c.
21645 * calls.c: Include "gt-calls.h".
21647 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
21649 PR rtl-optimization/79728
21650 * regs.h (struct target_regs): New field
21651 x_contains_allocatable_regs_of_mode.
21652 (contains_allocatable_regs_of_mode): New macro.
21653 * reginfo.c (init_reg_sets_1): Initialize it, and change
21654 contains_reg_of_mode so it includes global regs as well.
21655 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
21656 rather than contains_regs_of_mode.
21658 2017-03-14 Martin Liska <mliska@suse.cz>
21660 * doc/invoke.texi: Document options that can't be combined with
21661 -fcheck-pointer-bounds.
21663 2017-03-14 Martin Liska <mliska@suse.cz>
21665 PR middle-end/79831
21666 * doc/invoke.texi (-Wchkp): Document the option.
21668 2017-03-14 Martin Liska <mliska@suse.cz>
21670 * Makefile.in: Install gcov-dump.
21672 2017-03-14 Martin Liska <mliska@suse.cz>
21674 * multiple_target.c (expand_target_clones): Bail out for
21675 an invalid attribute.
21677 2017-03-14 Richard Biener <rguenther@suse.de>
21679 * alias.c (struct alias_set_entry): Pack properly.
21680 * cfgloop.h (struct loop): Likewise.
21681 * cse.c (struct set): Likewise.
21682 * ipa-utils.c (struct searchc_env): Likewise.
21683 * loop-invariant.c (struct invariant): Likewise.
21684 * lra-remat.c (struct cand): Likewise.
21685 * recog.c (struct change_t): Likewise.
21686 * rtl.h (struct address_info): Likewise.
21687 * symbol-summary.h (function_summary): Likewise.
21688 * tree-loop-distribution.c (struct partition): Likewise.
21689 * tree-object-size.c (struct object_size_info): Likewise.
21690 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
21691 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
21692 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
21693 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
21694 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
21695 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
21696 (struct _stmt_vec_info): Likewise.
21698 2017-03-14 Martin Liska <mliska@suse.cz>
21701 * multiple_target.c (create_dispatcher_calls): Check that
21702 a target can create a function dispatcher.
21704 2017-03-14 Martin Liska <mliska@suse.cz>
21707 * multiple_target.c (expand_target_clones): Drop local.local
21708 flag for default implementation.
21710 2017-03-14 Richard Biener <rguenther@suse.de>
21712 PR tree-optimization/80030
21713 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
21715 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
21717 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
21718 gcc_fallthrough() instead of __attribute__((fallthrough));
21720 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
21722 * doc/gcc.texi: Remove "up" link to (DIR).
21723 * doc/gccint.texi: Ditto.
21725 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
21727 * doc/install.texi (Specific) <avr>: Remove reference to
21730 2017-03-13 Jeff Law <law@redhat.com>
21732 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
21733 attribute rather than comments.
21735 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
21736 match_scratch operand is highest.
21738 2017-03-13 Martin Liska <mliska@suse.cz>
21740 PR middle-end/78339
21741 * ipa-pure-const.c (warn_function_noreturn): If the declarations
21742 is a CHKP clone, use original declaration.
21744 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21746 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
21747 (arc_conditional_register_usage): Use a different allocation order
21748 when optimizing for size.
21749 * common/config/arc/arc-common.c (arc_option_optimization_table):
21750 Section anchors default on when optimizing for size.
21752 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21754 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
21756 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21758 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
21759 * config/arc/arc.md (cpu_facility): Add cd variant.
21760 (*movqi_insn): Add code density variant.
21761 (*movhi_insn): Likewise.
21762 (*movqi_insn): Likewise.
21763 (*addsi3_mixed): Likewise.
21764 (subsi3_insn): Likewise.
21766 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21768 * config/arc/arc.md (movsi_cond_exec): Update constraint.
21770 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21772 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
21773 expressions with MINUS and UNARY ops.
21775 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21778 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
21780 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
21781 between vec_select and vector argument.
21782 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
21783 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
21784 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
21785 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
21786 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
21787 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
21789 2017-03-13 Richard Biener <rguenther@suse.de>
21792 * params.def (vect-max-peeling-for-alignment): Fix typo.
21794 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21796 * doc/install.texi (Specific) <mips-*-*>: Remove description of
21797 issue that only occurred with binutils below 2.18.
21799 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21801 * doc/install.texi (Specific) <cris-axis-elf>: No longer
21802 refer to binutils 2.11/2.12 minimum.
21804 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21806 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
21807 ftp.kernel.org and simplify binutils requirement.
21809 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
21811 * doc/invoke.texi (Warning Options): Fix spelling of link-time
21813 (Optimize Options): Ditto. Also remove redundancy.
21815 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21817 PR translation/79848
21818 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
21820 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
21821 to G_ to avoid double translation.
21823 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21825 PR translation/79923
21826 * auto-profile.c (get_combined_location): Convert leading
21827 character of diagnostics to lower case and remove trailing period.
21828 (read_profile): Likewise for various diagnostics.
21829 * config/arm/arm.c (arm_option_override): Remove trailing period
21830 from various diagnostics.
21831 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
21832 (msp430_expand_delay_cycles): Likewise.
21834 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21837 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
21838 full command-line argument, rather than just "str".
21839 (aarch64_validate_march): Likewise.
21840 (aarch64_validate_mtune): Likewise.
21842 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
21844 PR rtl-optimization/78911
21845 * lra-assigns.c (must_not_spill_p): New function.
21846 (spill_for): Use it.
21848 2017-03-10 Jakub Jelinek <jakub@redhat.com>
21850 PR tree-optimization/79981
21851 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
21852 ATOMIC_COMPARE_EXCHANGE ifn result.
21853 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
21854 IFN_ATOMIC_COMPARE_EXCHANGE.
21856 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21859 * opts.c (parse_sanitizer_options): Add missing question mark to
21860 "did you mean" message.
21862 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21864 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
21866 (VMULEUH_UNS): Likewise.
21867 (VMULOUB_UNS): Likewise.
21868 (VMULOUH_UNS): Likewise.
21869 * config/rs6000/rs6000.c (builtin_function_type): Remove
21870 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
21872 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21875 * read-rtl-function.c (function_reader::read_rtx_operand): Update
21876 x with result of extra_parsing_for_operand_code_0.
21877 (function_reader::extra_parsing_for_operand_code_0): Convert
21878 return type from void to rtx, returning x. When reading
21879 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
21880 larger size containing struct block_symbol.
21882 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
21884 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
21885 -mfloat128-hardware without -m64.
21887 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
21890 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
21891 entries to the case statement that marks unsigned arguments to
21892 overloaded functions.
21894 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21896 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
21897 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
21899 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
21902 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
21903 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
21905 2017-03-10 Martin Liska <mliska@suse.cz>
21909 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
21910 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
21913 2017-03-10 Olivier Hainque <hainque@adacore.com>
21915 * tree-switch-conversion (array_value_type): Start by resetting
21916 candidate type to it's main variant.
21918 2017-03-10 Jakub Jelinek <jakub@redhat.com>
21920 PR rtl-optimization/79909
21921 * combine.c (try_combine): Use simplify_replace_rtx on individual
21922 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
21923 of the whole CALL_INSN_FUNCTION_USAGE.
21925 PR tree-optimization/79972
21926 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
21927 get_range_info on SSA_NAMEs. Formatting fixes.
21929 2017-03-10 Richard Biener <rguenther@suse.de>
21930 Jakub Jelinek <jakub@redhat.com>
21932 PR tree-optimization/77975
21933 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
21934 edge to be constant.
21935 (get_val_for): For constant x return it. Formatting fix.
21936 (loop_niter_by_eval): Avoid pointless looping if the next iteration
21937 would use the same bases as the current one.
21939 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21941 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
21942 instead of vec_select for V1TImode.
21943 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
21945 (VSX_LE_128): Add V1TI to this mode iterator.
21946 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
21947 (*vsx_le_perm_store_<mode>): Likewise.
21948 (pre-reload splitter for VSX stores): Likewise.
21949 (post-reload splitter for VSX stores): Likewise.
21950 (*vsx_xxpermdi2_le_<mode>): Likewise.
21951 (*vsx_lxvd2x2_le_<mode>): Likewise.
21952 (*vsx_stxvd2x2_le_<mode>): Likewise.
21954 2017-03-09 Michael Eager <eager@eagercon.com>
21956 Correct failures with --enable-checking=yes,rtl.
21958 * config/microblaze/microblaze.c (microblaze_expand_shift):
21959 Replace GET_CODE test with CONST_INT_P and INTVAL test with
21960 test for const0_rtx.
21961 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
21962 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
21964 2017-03-09 Richard Biener <rguenther@suse.de>
21966 PR tree-optimization/79977
21967 * graphite-scop-detection.c (scop_detection::merge_sese):
21968 Handle the case of extra exits to blocks dominating the entry.
21970 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
21972 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21975 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
21977 PR rtl-optimization/79949
21978 * lra-constraints.c (process_alt_operands): Check memory when
21979 trying to predict a cycle. Print about the overall increase.
21981 2017-03-09 Richard Biener <rguenther@suse.de>
21983 PR middle-end/79971
21984 * gimple-expr.c (useless_type_conversion_p): Preserve
21985 TYPE_SATURATING for fixed-point types.
21987 2017-03-09 Richard Biener <rguenther@suse.de>
21990 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
21991 alignment of BLKmode params.
21993 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21996 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
21997 (VALL_NO_V2Q): Likewise.
21999 * config/aarch64/aarch64-simd.md
22000 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
22002 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
22003 VALL_NO_V2Q mode iterator.
22004 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
22006 2017-03-09 Martin Liska <mliska@suse.cz>
22008 PR tree-optimization/79631
22009 * tree-chkp-opt.c (chkp_is_constant_addr): Call
22010 tree_int_cst_sign_bit just for INTEGER constants.
22012 2017-03-09 Martin Liska <mliska@suse.cz>
22016 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
22019 2017-03-09 Marek Polacek <polacek@redhat.com>
22022 * tree.c (inchash::add_expr): Handle TREE_VEC.
22024 2017-03-09 Martin Liska <mliska@suse.cz>
22027 (chkp_narrow_size_and_offset): New function.
22028 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
22029 (void chkp_parse_bit_field_ref): New function.
22030 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
22031 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
22033 2017-03-09 Martin Liska <mliska@suse.cz>
22036 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
22037 (chkp_find_bounds_1): Remove gcc_unreachable.
22039 2017-03-09 Jakub Jelinek <jakub@redhat.com>
22042 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
22043 BUILT_IN_SYNC*, determine the access type from the size suffix and
22044 always build a MEM_REF with that type. Handle forgotten
22045 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
22048 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
22049 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
22050 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
22051 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
22052 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
22053 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
22054 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
22055 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
22056 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
22057 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
22058 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
22059 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
22060 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
22061 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
22062 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
22063 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
22064 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
22065 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
22066 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
22067 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
22068 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
22069 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
22070 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
22071 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
22072 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
22073 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
22074 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
22075 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
22076 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
22077 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
22078 definitions outside of __OPTIMIZE__ guarded section.
22081 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
22082 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
22083 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
22084 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
22087 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22089 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
22090 ("vfenez<mode>"): Add missing constraints.
22092 2017-03-08 Martin Sebor <msebor@redhat.com>
22095 * config/nds32/nds32.c (nds32_option_override):
22096 Fix misspelled diagnostic.
22098 2017-03-08 Jakub Jelinek <jakub@redhat.com>
22101 * gimplify.c (gimplify_omp_for): Replace index var in outer
22102 taskloop statement with an artificial variable and add
22103 OMP_CLAUSE_PRIVATE clause for it.
22105 2017-03-08 Richard Biener <rguenther@suse.de>
22107 PR tree-optimization/79955
22108 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
22109 for accesses that are completely outside of the variable.
22111 2017-03-08 Andrew Haley <aph@redhat.com>
22113 PR tree-optimization/79943
22114 * tree-ssa-loop-split.c (compute_new_first_bound): When
22115 calculating the new upper bound, (END-BEG) should be added, not
22118 2017-03-08 Jakub Jelinek <jakub@redhat.com>
22120 * config/avr/avr.md (setmemhi): Make sure match_dup
22121 operand number comes before match_scratch.
22123 2017-03-08 Richard Biener <rguenther@suse.de>
22125 PR tree-optimization/79920
22126 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
22127 with ncopies == 1 to ...
22128 (vect_transform_slp_perm_load): ... here. Properly compute
22129 all element loads by iterating VF times over the group. Do
22130 not handle ncopies (computed in a broken way) in
22131 vect_create_mask_and_perm.
22133 2017-03-08 Jakub Jelinek <jakub@redhat.com>
22136 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
22137 is a uniform vector, use uniform_vector_p return value instead of
22138 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
22140 2017-03-07 Marek Polacek <polacek@redhat.com>
22142 PR middle-end/79809
22143 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
22144 (alloca_call_type): Likewise.
22146 2017-03-07 Martin Liska <mliska@suse.cz>
22148 * gcov.c (process_args): Put comment to correct location.
22150 2017-03-07 Martin Liska <mliska@suse.cz>
22152 PR middle-end/68270
22153 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
22154 Use array_at_struct_end_p instead of DECL_CHAIN (field).
22155 (chkp_narrow_bounds_for_field): Likewise.
22156 (chkp_parse_array_and_component_ref): Pass one more argument to
22159 2017-03-07 Richard Biener <rguenther@suse.de>
22161 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
22164 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
22166 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
22167 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
22169 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22172 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
22173 to end of description.
22174 (PARAM_MAX_STORES_TO_MERGE): Likewise.
22176 2017-03-07 Jakub Jelinek <jakub@redhat.com>
22178 PR rtl-optimization/79901
22179 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
22181 (*avx512f_<code><mode>3<mask_name>): ... this.
22182 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
22183 iterator instead of VI8_AVX2_AVX512BW.
22185 PR rtl-optimization/79901
22186 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
22187 min/max expander, expand it using expand_vec_cond_expr.
22190 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
22193 2017-03-06 Jakub Jelinek <jakub@redhat.com>
22196 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
22197 to void * for PCH reasons.
22198 * dwarf2out.c (output_loc_operands, output_die): Cast
22199 v.val_vec.array to unsigned char *.
22201 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
22204 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
22207 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
22209 PR rtl-optimization/79571
22210 * lra-constraints.c (process_alt_operands): Calculate static
22211 reject and subtract it from overall when only addresses will be
22214 2017-03-06 Julia Koval <julia.koval@intel.com>
22217 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
22218 incoming stack boundary to 128 for 64-bit targets.
22220 2017-03-06 Richard Biener <rguenther@suse.de>
22222 PR tree-optimization/79894
22223 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
22224 to NULL after folding it.
22226 2017-03-06 Richard Biener <rguenther@suse.de>
22228 PR tree-optimization/79824
22229 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
22230 check disabling peeling for gaps.
22232 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
22234 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
22235 attributes): Document gettimeofday.
22237 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
22239 * config/s390/s390.c (s390_option_override_internal): Set
22240 PARAM_MIN_VECT_LOOP_BOUND
22242 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
22244 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
22245 * config/s390/s390.md: Likewise.
22247 2017-03-06 Jakub Jelinek <jakub@redhat.com>
22250 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
22251 (<avx2_avx512>_perm<mode>): Rename to ...
22252 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
22254 (<avx512>_perm<mode>_mask): Rename to ...
22255 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
22257 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
22258 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
22259 instead of VI8F_256_512.
22260 (avx512f_perm<mode>): New define_expand.
22261 (avx512f_perm<mode>_mask): Likewise.
22262 (avx512f_perm<mode>_1<mask_name>): New define_insn.
22263 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
22265 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22267 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
22268 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
22270 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
22272 2017-03-06 Martin Liska <mliska@suse.cz>
22275 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
22276 when having a SSA NAME w/o VAR_DECL assigned to it.
22278 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22280 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
22281 msa_dpsub_<su>_d): Fix MODE for vec_select.
22283 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22285 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
22287 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
22289 2017-03-06 Richard Biener <rguenther@suse.de>
22291 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
22292 * plugin.c (register_plugin_info): Likewise.
22293 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
22295 2017-03-05 Jakub Jelinek <jakub@redhat.com>
22297 * config/i386/sse.md (sse_storehps, sse_storelps,
22298 avx_<castmode><avxsizesuffix>_<castmode>,
22299 avx512f_<castmode><avxsizesuffix>_<castmode>,
22300 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
22301 in condition that at least one operand is not a MEM.
22303 2017-03-03 Jakub Jelinek <jakub@redhat.com>
22305 PR middle-end/79805
22306 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
22307 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
22309 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
22310 gimple_call_nothrow_p flag based on whether original builtin can throw.
22311 If it can, emit following stmts on the fallthrough edge.
22312 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
22313 don't create new bb if inserting just debug stmts on the edge, try to
22314 insert them on the fallthru bb or just reset debug stmts.
22316 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
22319 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
22320 restore recog_data (including the operand rtxes inside it) around
22321 the call to get_insn_template.
22323 2017-03-03 Martin Sebor <msebor@redhat.com>
22325 PR tree-optimization/79699
22326 * context.c (context::~context): Free MPFR caches to avoid
22327 a memory leak on program exit.
22329 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22331 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
22332 Use wide_int::ulow () instead of .elt (0).
22334 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
22336 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
22337 (*pushxf): Limit oF constraint to 32bit targets and add oC
22338 constraint for 64bit targets.
22339 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
22340 (*pushdf): Change rmF constraint to rmC.
22342 2017-03-03 Martin Liska <mliska@suse.cz>
22344 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
22345 Remove unused variable.
22347 2017-03-03 Jakub Jelinek <jakub@redhat.com>
22350 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
22351 is a memory operand, increase num_memory.
22352 (ix86_expand_args_builtin): Likewise.
22354 2017-03-03 Jan Hubicka <jh@suse.cz>
22357 * ipa-devirt.c (maybe_record_node): Properly handle
22358 __cxa_pure_virtual visibility.
22360 2017-03-03 Martin Liska <mliska@suse.cz>
22362 PR tree-optimization/79803
22363 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
22365 (pass_loop_prefetch::execute): Disabled optimization if an
22366 assumption about L1 cache size is not met.
22368 2017-03-03 Martin Liska <mliska@suse.cz>
22370 PR rtl-optimization/79574
22371 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
22372 (hash_scan_set): Likewise.
22373 (dump_hash_table): Likewise.
22374 (hoist_code): Likewise.
22376 2017-03-03 Richard Biener <rguenther@suse.de>
22378 * fixed-value.c (fixed_from_string): Restore use of elt (1)
22379 in place of uhigh ().
22380 (fixed_convert_from_real): Likewise.
22382 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
22385 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
22387 2017-03-03 Richard Biener <rguenther@suse.de>
22389 PR middle-end/79818
22390 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
22391 TYPE_OVERFLOW_UNDEFINED check.
22393 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22395 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
22397 (vector_ae_<mode>_p): Likewise.
22398 (vector_nez_<mode>_p): Likewise.
22399 (vector_ne_v2di_p): Likewise.
22400 (vector_ae_v2di_p): Likewise.
22401 (vector_ne_<mode>_p): Likewise.
22402 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
22404 (vsx_tsqrt<mode>2_fe): Likewise.
22406 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
22409 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
22411 2017-03-02 Jakub Jelinek <jakub@redhat.com>
22413 PR rtl-optimization/79780
22414 * cprop.c (one_cprop_pass): When second and further conditional trap
22415 in a single basic block is turned into an unconditional trap, turn it
22416 into a deleted note to avoid RTL verification failures.
22418 2017-03-02 Richard Biener <rguenther@suse.de>
22420 * fold-const.c (const_binop): Use ulow () instead of elt (0).
22422 2017-03-02 Richard Biener <rguenther@suse.de>
22424 PR tree-optimization/79345
22426 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
22427 param and abort the walk, returning -1 if it is hit.
22428 (walk_aliased_vdefs): Take a limit param and pass it on.
22429 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
22430 defaulting to 0 and return a signed int.
22431 * tree-ssa-uninit.c (struct check_defs_data): New struct.
22432 (check_defs): New helper.
22433 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
22434 about uninitialized memory.
22435 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
22436 bogus uninitialized warning.
22437 (fixed_convert_from_real): Likewise.
22439 2017-03-02 Bin Cheng <bin.cheng@arm.com>
22441 PR tree-optimization/66768
22442 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
22443 iv_use if base object can't be determined.
22445 2017-03-02 Jakub Jelinek <jakub@redhat.com>
22447 PR tree-optimization/79345
22448 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
22449 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
22450 (get_pattern_stats): Initialize it.
22451 * genemit.c (gen_expand): Verify match_scratch numbers come after
22452 match_operand/match_dup numbers.
22453 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
22454 match_scratch numbers.
22455 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
22457 * config/s390/s390.md (trunctdsd2): Likewise.
22459 2017-03-02 Richard Biener <rguenther@suse.de>
22461 * wide-int.h (wide_int_storage::operator=): Implement in terms
22464 2017-03-02 Richard Biener <rguenther@suse.de>
22466 PR tree-optimization/79777
22467 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
22468 the to insert expression to sth existing.
22470 2017-03-01 Martin Sebor <msebor@redhat.com>
22472 PR middle-end/79692
22473 * gimple-ssa-sprintf.c
22474 (directive::known_width_and_precision): New function.
22475 (format_integer): Use it.
22476 (get_mpfr_format_length): Consider the full range of precision
22477 when computing %g output with the # flag. Set the likely byte
22478 count to 3 rather than 1 when precision is indeterminate.
22479 (format_floating): Correct the lower bound of precision.
22481 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22483 * doc/invoke.texi: Document default code model for 64-bit Linux.
22485 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22488 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
22489 udiv rather than div since input pattern is unsigned.
22491 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
22493 * config/i386/i386.c (print_reg): Warn for values of
22494 unsupported size in integer register.
22496 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
22499 * config/rs6000/predicates.md (current_file_function_operand): Do
22500 not allow self calls to be local if the function is replaceable.
22502 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
22505 * config/rs6000/altivec.h (vec_ctz and others): Change the
22506 preprocessor macro that controls conditional compilation from
22507 _ARCH_PWR9 to __POWER9_VECTOR__.
22508 (vec_all_ne): Change parameterization of __altivec_scalar_pred
22509 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
22510 control (instead of _ARCH_PWR9 control) so that template
22511 definition uses power9-specific function.
22512 (vec_any_eq): Likewise.
22513 (vec_all_ne): Change macro definition to use a power9-specific
22514 expansion under #ifdef __POWER9_VECTOR__ control (instead of
22515 _ARCH_PWR9 control).
22516 (vec_any_eq) Likewise.
22517 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
22518 expansion for CMPNEF to remove support for xvcmpnesp instruction.
22519 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
22520 support for xvcmpnedp instruction.
22521 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
22522 macro expansion so that Power9 implementation of vec_all_ne does
22523 not use the AltiVec predicate framework.
22524 (VCMPNEH_P): Likewise.
22525 (VCMPNEW_P): Likewise.
22526 (VCMPNED_P): Likewise.
22527 (VCMPNEFP_P): Likewise.
22528 (VCMPNEDP_P): Likewise.
22529 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
22530 implementation of vec_any_eq to not use AltiVec predicate
22532 (VCMPAEH_P): Likewise.
22533 (VCMPAEW_P): Likewise.
22534 (VCMPAED_P): Likewise.
22535 (VCMPAEFP_P): Likewise.
22536 (VCMPAEDP_P): Likewise.
22537 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
22538 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
22539 not use the AltiVec predicate framework.
22540 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
22541 of vec_any_eq to not use AltiVec predicate framework.
22542 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
22543 support for predefined __POWER9_VECTOR__ macro to indicate that
22544 Power9 instruction selection is enabled.
22545 (altivec_overloaded_builtins): Remove extraneous
22546 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
22547 function argument types RS6000_BTI_bool_V16QI and
22548 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
22549 entry for overloaded function argument types RS6000_BTI_bool_V4SI
22550 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
22551 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
22552 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
22553 Power9 for implementations of vec_cmpne. Change the signature for
22554 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
22555 (representing vec_all_ne) to remove the previously described first
22556 argument of type RS6000_BTI_INTSI, as this was an artifact of
22557 reliance on the AltiVec predicate framework, which is no longer
22558 used in the implementation of these functions. Add
22559 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
22560 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
22561 since, unlike the AltiVec predicate framework implementation, we
22562 do not share function descriptors between vec_alle and vec_anyeq.
22563 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
22564 set of modes that receive special treatment even when
22565 TARGET_P9_VECTOR is true. The special treatment emits code that
22566 does not depend on Power9 instructions.
22567 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
22568 define_expand to not rely on AltiVec predicate framework.
22569 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
22571 (vector_ne_v2di_p): Change this define_expand to not rely on
22572 AltiVec predicate framework.
22573 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
22575 (vector_ne_<mode>_p): Change this define_expand to not rely on
22576 AltiVec predicate framework.
22577 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
22579 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
22580 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
22581 define_insn pattern.
22582 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
22583 define_insn pattern because the xvcmpne<VSs>. instruction is not
22585 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
22586 instruction is not supported.
22588 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22590 * config/nvptx/nvptx.c: Include intl.h.
22592 2017-03-01 Martin Jambor <mjambor@suse.cz>
22595 * ipa-prop.h (ipa_bits): Removed field known.
22596 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
22597 to pointers. Adjusted their comments to warn about their sharing.
22598 (ipcp_transformation_summary): Change bits to a vector of pointers.
22599 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
22600 (ipa_get_ipa_bits_for_value): Declare.
22601 * tree-vrp.h (value_range): Mark as GTY((for_user)).
22602 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
22603 (ipa_bits_hash_table): Likewise.
22604 (ipa_vr_ggc_hash_traits): Likewise.
22605 (ipa_vr_hash_table): Likewise.
22606 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
22607 being pointers and vr_known being removed.
22608 (ipa_set_jf_unknown): Likewise.
22609 (ipa_get_ipa_bits_for_value): New function.
22610 (ipa_set_jfunc_bits): Likewise.
22611 (ipa_get_value_range): New overloaded functions.
22612 (ipa_set_jfunc_vr): Likewise.
22613 (ipa_compute_jump_functions_for_edge): Use the above functions to
22614 construct bits and vr parts of jump functions.
22615 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
22616 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
22618 (ipcp_grow_transformations_if_necessary): Also allocate
22619 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
22621 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
22622 them. Fix too long lines.
22623 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
22624 vr_known being removed.
22625 (ipa_read_jump_function): Use new setter functions to construct bits
22626 and vr parts of jump functions or set them to NULL.
22627 (write_ipcp_transformation_info): Adjust for bits being pointers.
22628 (read_ipcp_transformation_info): Likewise.
22629 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
22631 Include gt-ipa-prop.h.
22632 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
22634 (ipcp_store_bits_results): Likewise.
22635 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
22636 Do not write to existing jump functions but use a temporary instead.
22638 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22641 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
22642 attempt to use its first operand as BIT_FIELD_REF base.
22644 2017-03-01 Richard Biener <rguenther@suse.de>
22646 PR middle-end/79721
22647 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
22648 interpolating formula in wrapping arithmetic.
22649 (chrec_apply): Convert chrec_evaluate return value to wanted type.
22651 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22653 PR tree-optimization/79734
22654 * tree-vect-generic.c (expand_vector_condition): Optimize
22655 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
22656 Handle VEC_COND_EXPR where comparison has different inner width from
22657 type's inner width.
22659 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
22661 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
22662 markup, and similar issues. Remove @opindex entries for things
22663 that aren't options. Add missing -mmpy-option entries.
22665 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22667 PR tree-optimization/79737
22668 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
22669 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
22670 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
22671 instead of byte_size. Formatting fix.
22672 (shift_bytes_in_array_right): Formatting fix.
22674 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
22677 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
22678 condition on optimize for the leaf function test.
22680 2017-02-28 Martin Liska <mliska@suse.cz>
22683 * read-rtl-function.c (function_reader::handle_unknown_directive):
22684 Bail out when one uses -flto.
22686 2017-02-28 Martin Liska <mliska@suse.cz>
22688 * common.opt: Replace space with tabular for options of <number>
22690 * config/i386/i386.opt: Show <number> value for
22691 -mlarge-data-threshold.
22692 * opts.c (print_filtered_help): Do not display number in hexadecimal
22695 2017-02-28 Martin Liska <mliska@suse.cz>
22697 * common.opt: Fix --help=option -Q for options which are of
22700 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
22702 * config/i386/i386.c (print_reg): Error out for values
22703 of 8-bit size in invalid integer register.
22705 2017-02-28 Martin Sebor <msebor@redhat.com>
22707 PR tree-optimization/79691
22708 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
22710 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22713 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
22714 gcc_unreachable with output_operand_lossage.
22716 2017-02-28 Richard Biener <rguenther@suse.de>
22718 PR tree-optimization/79740
22719 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
22721 (visit_nary_op): Insert the nary into the hashtable if we
22722 pattern-matched sth.
22723 * tree-ssa-pre.c (eliminate_insert): Robustify.
22725 2017-02-28 Richard Biener <rguenther@suse.de>
22727 PR middle-end/79731
22728 * fold-const.c (decode_field_reference): Reject out-of-bound
22731 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22733 * config/i386/i386.c: Include intl.h.
22734 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
22735 instead of just cond ? "..." : "...".
22736 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
22737 * coverage.c (read_counts_file): Likewise.
22738 * omp-offload.c: Include intl.h.
22739 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
22740 of just cond ? "..." : "...".
22741 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
22742 of just cond ? "..." : "...".
22744 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
22747 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
22749 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
22750 'tune for' CPU name.
22751 * config/arm/arm-cpu-data.h: Regenerated.
22753 2017-02-28 Richard Biener <rguenther@suse.de>
22755 PR tree-optimization/79732
22756 * tree-inline.c (expand_call_inline): Do not shadow var.
22758 2017-02-28 Richard Biener <rguenther@suse.de>
22760 PR tree-optimization/79723
22761 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
22762 address-space properly.
22764 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
22766 * doc/optinfo.texi (Optimization groups): Fix option used for
22768 * doc/invoke.texi (-fopt-info): Document "omp".
22769 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
22770 (OPTGROUP_ALL): Add OPTGROUP_OMP.
22771 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
22772 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
22773 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
22775 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
22777 * dumpfile.c (optgroup_options): Instead of "openmp", associate
22778 OPTGROUP_OMP with "omp".
22780 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
22783 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
22784 for arithmetic shift of unsigned V2DI.
22786 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
22788 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
22789 arc/linux.h headers.
22790 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
22791 (LINK_SPEC): Likewise.
22792 (ARC_TLS_EXTRA_START_SPEC): Likewise.
22793 (EXTRA_SPECS): Likewise.
22794 (STARTFILE_SPEC): Likewise.
22795 (ENDFILE_SPEC): Likewise.
22796 (LIB_SPEC): Likewise.
22797 (TARGET_SDATA_DEFAULT): Likewise.
22798 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
22799 (MULTILIB_DEFAULTS): Likewise.
22800 (DWARF2_UNWIND_INFO): Likewise.
22801 * config/arc/big.h: New file.
22802 * config/arc/elf.h: Likewise.
22803 * config/arc/linux.h: Likewise.
22804 * config/arc/t-uClibc: Remove.
22806 2017-02-27 Bin Cheng <bin.cheng@arm.com>
22808 PR tree-optimization/77536
22809 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
22810 (tree_transform_and_unroll_loop): Use above function to compute the
22811 estimated niter of unrolled loop and use it when scaling profile.
22812 Also use count info rather than frequency if it's non-zero.
22813 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
22814 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
22815 (vect_transform_loop): Call above function.
22817 2017-02-27 Richard Biener <rguenther@suse.de>
22819 PR tree-optimization/45397
22820 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
22821 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
22822 (visit_nary_op): Add pattern matching for CSEing sign-changed
22823 or truncated operations with wider ones.
22825 2017-02-27 Richard Biener <rguenther@suse.de>
22827 PR tree-optimization/79690
22828 * tree-vect-stmts.c (vectorizable_store): Use vector type
22829 built from the DR with address-space.
22831 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
22833 * doc/invoke.texi (Optimize Options): Refine the description
22834 of asan-use-after-return.
22836 2017-02-25 Alan Modra <amodra@gmail.com>
22838 PR rtl-optimization/79584
22839 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
22840 base, not ad->base_term, the reg within base. Remove assertion
22841 that ad->base == ad->base_term. Replace gen_int_mode using
22842 bogus mode with const0_rtx.
22844 2017-02-25 Jakub Jelinek <jakub@redhat.com>
22846 PR middle-end/79396
22847 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
22848 FMA_EXPR like tcc_binary or tcc_unary.
22850 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
22853 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
22855 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
22856 (output_loc_operands): Handle DW_OP_call_ref and
22857 DW_OP_GNU_variable_value.
22858 (struct variable_value_struct): New type.
22859 (struct variable_value_hasher): Likewise.
22860 (variable_value_hash): New variable.
22861 (string_types): Remove.
22862 (copy_loc_descr): New function.
22863 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
22864 (prepend_loc_descr_to_each): New function.
22865 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
22866 instead of add_loc_descr_to_each if the first argument is single
22867 location list and the second has multiple.
22868 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
22869 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
22870 when looking for variable value which doesn't have other location info.
22871 (loc_list_from_tree): Formatting fix.
22872 (gen_array_type_die): Simplify DW_AT_string_length handling.
22873 (adjust_string_types): Remove.
22874 (gen_subprogram_die): Don't call adjust_string_types nor test/set
22875 string_types. Call resolve_variable_values.
22876 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
22877 (resolve_addr_in_expr): Likewise. Add A argument.
22878 (copy_deref_exprloc): Remove deref argument. Adjust for the
22879 original expression being DW_OP_GNU_variable_value with optionally
22880 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
22881 optionally after it.
22882 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
22883 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
22884 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
22885 (variable_value_hasher::hash, variable_value_hasher::equal): New
22887 (resolve_variable_value_in_expr, resolve_variable_value,
22888 resolve_variable_values, note_variable_value_in_expr,
22889 note_variable_value): New functions.
22890 (dwarf2out_early_finish): Call note_variable_value on all toplevel
22893 2017-02-24 Jakub Jelinek <jakub@redhat.com>
22896 * opts.h (handle_generated_option): Add GENERATED_P argument.
22897 * opts-common.c (handle_option): Adjust function comment.
22898 (handle_generated_option): Add GENERATED_P argument, pass it to
22900 (control_warning_option): Pass false to handle_generated_option
22902 * opts.c (maybe_default_option): Pass true to handle_generated_option
22904 * optc-gen.awk: Likewise.
22906 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22908 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
22909 a REG, look at the REG it is a SUBREG of.
22910 (splitter for cmpeqsi_t): Ditto.
22912 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22914 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
22915 the special USEs with the pattern of the insn, not the insn itself.
22917 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
22920 * doc/invoke.texi: Document -mload-store-pairs.
22922 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22923 Sandra Loosemore <sandra@codesourcery.com>
22925 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
22926 argument isn't a CONST_INT.
22927 (nios2_alternate_compare_const): Assert op is a CONST_INT.
22928 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
22929 (nios2_validate_compare): Bypass alternate compare logic if *op2
22930 is not a CONST_INT.
22931 (ldstwm_operation_p): Return false if first_base is not a REG or
22932 if first_offset is not a CONST_INT.
22934 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22936 * config/cris/cris.md: Use correct operand in a define_peephole2.
22938 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22940 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
22942 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22944 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
22945 this_insn if it is an INSN or JUMP_INSN.
22946 (force_offsettable): Look at base, not at addr.
22947 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
22948 on things that aren't necessarily CONST_INTs.
22950 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
22952 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
22953 -mfpmath=sse is the default also for x86-32 targets with SSE2
22954 instruction set when @option{-ffast-math} is enabled
22956 2017-02-24 Jeff Law <law@redhat.com>
22958 PR rtl-optimizatoin/79286
22959 * ira.c (update_equiv_regs): Drop may_trap_p exception to
22962 2017-02-24 Richard Biener <rguenther@suse.de>
22964 PR tree-optimization/79389
22965 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
22968 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
22970 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
22971 function comment to reflect reality.
22972 (loop_exits_before_overflow): Fix typo in function description.
22974 2017-02-24 Richard Biener <rguenther@suse.de>
22976 PR tree-optimization/79389
22977 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
22978 properly that a threading opportunity exists. Detect conditional
22979 copy/constant propagation opportunities.
22981 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
22983 * config/visium/visium.md (type): Add trap.
22984 (b): New mode attribute.
22985 (*btst): Rename into...
22986 (*btst<mode>): ...this and adjust.
22987 (*cbranchsi4_btst_insn): Rename into...
22988 (*cbranch<mode>4_btst_insn): ...this and adjust.
22989 (trap): New define_insn.
22991 2017-02-23 Jakub Jelinek <jakub@redhat.com>
22993 PR tree-optimization/79389
22994 * ifcvt.c (struct noce_if_info): Add rev_cond field.
22995 (noce_reversed_cond_code): New function.
22996 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
22997 reversed_comparison_code. Formatting fix.
22998 (noce_try_store_flag): Test rev_cond != NULL in addition to
22999 reversed_comparison_code.
23000 (noce_try_store_flag_constants): Likewise.
23001 (noce_try_store_flag_mask): Likewise.
23002 (noce_try_addcc): Use rev_cond if non-NULL instead of
23003 reversed_comparison_code.
23004 (noce_try_cmove_arith): Likewise. Formatting fixes.
23005 (noce_try_minmax, noce_try_abs): Clear rev_cond.
23006 (noce_find_if_block): Initialize rev_cond.
23007 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
23008 instead of false as last argument never attempt to reverse it
23011 2017-02-23 Bin Cheng <bin.cheng@arm.com>
23013 PR tree-optimization/79663
23014 * tree-predcom.c (combine_chains): Process refs in reverse order
23015 only for ZERO length chains, and add explaining comment.
23017 2017-02-23 Jeff Law <law@redhat.com>
23019 PR tree-optimization/79578
23020 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
23021 in call to operand_equal_p.
23023 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
23026 * config/i386/cpuid.h: Fix another undefined behavior.
23028 2017-02-23 Richard Biener <rguenther@suse.de>
23030 PR tree-optimization/79683
23031 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
23032 vector types for data-refs.
23034 2017-02-23 Martin Liska <mliska@suse.cz>
23036 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
23038 2017-02-23 Jakub Jelinek <jakub@redhat.com>
23040 PR middle-end/79665
23041 * internal-fn.c (get_range_pos_neg): Moved to ...
23042 * tree.c (get_range_pos_neg): ... here. No longer static.
23043 * tree.h (get_range_pos_neg): New prototype.
23044 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
23045 are known to be in between 0 and signed maximum inclusive, try to
23046 expand both unsigned and signed divmod and use the cheaper one from
23049 2017-02-22 Jeff Law <law@redhat.com>
23051 PR tree-optimization/79578
23052 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
23053 to compare base operands.
23055 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
23058 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
23059 gpc_reg_operand instead of fpr_reg_operand.
23061 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
23063 * config/mips/mips.c (mips_return_in_memory): Force FP
23064 vector types to be returned in memory for o32 ABI.
23066 2017-02-22 Jakub Jelinek <jakub@redhat.com>
23068 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
23069 instead of DW_TAG_member for static data member declarations and don't
23070 set no_linkage_name for static inline data members.
23071 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
23074 2017-02-22 Martin Liska <mliska@suse.cz>
23076 * doc/invoke.texi: Replace inequality signs with square brackets
23079 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23081 PR tree-optimization/68644
23082 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
23084 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
23087 * lra-constraints.c (simplify_operand_subreg): Handle
23088 WORD_REGISTER_OPERATIONS targets.
23090 2017-02-22 Jakub Jelinek <jakub@redhat.com>
23093 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
23094 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
23095 elimination by swapping fld*.
23097 2017-02-22 Richard Biener <rguenther@suse.de>
23099 PR tree-optimization/79673
23100 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
23101 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
23102 irrelevant address-space qualifiers and avoiding a
23103 ADDR_SPACE_CONVERT_EXPR from fold_convert.
23105 2017-02-22 Richard Biener <rguenther@suse.de>
23107 PR tree-optimization/79666
23108 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
23109 to not symbolically negate if that may introduce undefined
23112 2017-02-22 Martin Liska <mliska@suse.cz>
23115 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
23116 * data-streamer-out.c (streamer_write_gcov_count_stream):
23118 * value-prof.c (stream_out_histogram_value): Make assert more
23119 precise based on type of counter.
23121 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
23124 * config/i386/i386.md (standard_x87sse_constant_load splitter):
23125 Use nonimmediate_operand instead of memory_operand for operand 1.
23126 (float-extend standard_x87sse_constant_load splitter): Ditto.
23128 2017-02-21 Jeff Law <law@redhat.com>
23130 PR tree-optimization/79621
23131 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
23132 blocks with edges to themselves.
23134 2017-02-21 Jakub Jelinek <jakub@redhat.com>
23137 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
23138 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
23139 Use gimple_call_builtin_p.
23142 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
23143 on temporarily removed DEBUG_INSNs.
23145 PR tree-optimization/79649
23146 * tree-loop-distribution.c (classify_partition): Give up on
23147 non-generic address space loads/stores.
23149 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
23151 * doc/loop.texi (Loop manipulation): Remove nonexistent
23152 tree_ssa_loop_version from the documentation.
23153 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
23155 2017-02-21 Jakub Jelinek <jakub@redhat.com>
23158 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
23159 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
23160 * config/rs6000/rs6000.c: Include except.h.
23161 (rs6000_expand_split_stack_prologue): Call
23162 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
23164 2017-02-21 Martin Jambor <mjambor@suse.cz>
23167 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
23168 have been analyzed.
23170 2017-02-21 Martin Jambor <mjambor@suse.cz>
23172 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
23173 for backward compatibility only.
23174 * doc/invoke.texi (Option Summary): Remove all references to
23175 -fipa-cp-alignment.
23177 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
23181 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23183 * lra-constraints.c (curr_insn_transform): Handle
23184 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
23186 2017-02-21 Martin Liska <mliska@suse.cz>
23188 * config/i386/i386.opt: Replace -masm-dialect with -masm.
23190 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
23192 PR translation/79638
23193 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
23195 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
23198 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
23199 (arm_function_ok_for_sibcall): Return false for an indirect call by
23200 descriptor if all the argument registers are used.
23201 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
23202 alignment of the function.
23204 2017-02-21 Jakub Jelinek <jakub@redhat.com>
23206 PR tree-optimization/61441
23207 * simplify-rtx.c (simplify_const_unary_operation): For
23208 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
23209 the sNaN unmodified.
23211 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
23213 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
23214 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
23215 instead of SYSTEM_HEADER_DIR.
23217 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
23218 Martin Liška <mliska@suse.cz>
23220 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
23221 Fix typos and grammar, use active voice, and clarify.
23223 2017-02-20 Marek Polacek <polacek@redhat.com>
23225 PR middle-end/79537
23226 * gimplify.c (gimplify_expr): Handle unused *&&L;.
23229 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
23231 2017-02-20 Jakub Jelinek <jakub@redhat.com>
23234 * config/i386/i386.c (ix86_expand_builtin): Handle
23235 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
23236 ix86_builtins_isa[fcode].isa as a requirement of those
23237 flags and any other flag in the bitmask.
23238 (ix86_init_mmx_sse_builtins): Use 0 instead of
23239 ~OPTION_MASK_ISA_64BIT as mask.
23240 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
23241 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
23242 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
23243 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
23245 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23248 * lra-constraints.c (split_reg): Check requested split mode
23249 is supported by the register.
23251 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23253 * lra-constraints.c (simplify_operand_subreg): Remove early
23256 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23259 * lra-constraints.c (curr_insn_transform): Tighten condition
23260 for converting SUBREG reloads from OP_OUT to OP_INOUT.
23262 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23265 * lra-constraints.c (curr_insn_transform): Handle
23266 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
23268 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
23271 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
23273 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
23275 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
23278 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
23281 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23283 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
23286 2017-02-18 Jakub Jelinek <jakub@redhat.com>
23288 * final.c (last_columnnum, override_columnnum): New variables.
23289 (final_start_function): Set last_columnnum, pass it to begin_prologue
23290 hook and pass 0 to dwarf2out_begin_prologue.
23291 (final_scan_insn): Update override_columnnum. Pass last_columnnum
23292 to source_line debug hook.
23293 (notice_source_line): Compute last_columnnum and for debug_column_info
23294 return true on column changes.
23295 * debug.h (struct gcc_debug_hooks): Add column argument to
23296 source_line and begin_prologue hooks.
23297 (debug_nothing_int_charstar_int_bool): Remove prototype.
23298 (debug_nothing_int_int_charstar,
23299 debug_nothing_int_int_charstar_int_bool): New prototypes.
23300 (dwarf2out_begin_prologue): Add column argument.
23301 * debug.c (do_nothing_debug_hooks): Adjust source_line and
23302 begin_prologue hooks.
23303 (debug_nothing_int_charstar_int_bool): Remove.
23304 (debug_nothing_int_int_charstar,
23305 debug_nothing_int_int_charstar_int_bool): New functions.
23306 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
23307 through to dwarf2out_source_line.
23308 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
23309 (dwarf2out_source_line): Add column argument, emit it if requested.
23310 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
23312 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
23313 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
23314 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
23315 through to dwarf2out_begin_prologue.
23316 (vmsdbgout_source_line): Add column argument, pass it through to
23317 dwarf2out_source_line.
23318 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
23319 dbxout_source_line caller.
23320 (dbxout_source_line): Add column argument.
23322 * common.opt (gno-column-info, gcolumn-info): New options.
23323 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
23324 (check_die): Also test for multiple DW_AT_decl_column attributes.
23325 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
23326 DW_AT_decl_column if requested.
23327 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
23329 (gen_variable_die): Likewise.
23330 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
23331 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
23334 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
23335 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
23336 (ix86_handle_option): Handle OPT_m3dnowa.
23337 * doc/invoke.texi (-m3dnowa): Document.
23338 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
23339 -m3dnowa instead of -m3dnow -march=athlon.
23342 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
23343 instead of gcc_assert for K, r and R code checks. Formatting fixes.
23345 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23348 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
23349 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
23350 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
23351 generator for vsx_xxpermdi_<mode>_be.
23352 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
23353 force big-endian semantics.
23354 (vsx_xxpermdi_<mode>_be): New define_expand with same
23355 implementation as previous version of vsx_xxpermdi_<mode>.
23357 2017-02-17 Jakub Jelinek <jakub@redhat.com>
23359 PR tree-optimization/79327
23360 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
23361 variable, its initialization and use.
23363 2017-02-17 Julia Koval <julia.koval@intel.com>
23365 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
23366 (OPTION_MASK_ISA_PKU_UNSET): New.
23367 (ix86_handle_option): Handle -mrdpid.
23368 * config/i386/cpuid.h (bit_RDPID): New.
23369 * config/i386/driver-i386.c (host_detect_local_cpu):
23370 Detect RDPID feature.
23371 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
23372 * config/i386/i386-c.c (ix86_target_macros_internal):
23374 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
23375 (ix86_valid_target_attribute_inner_p): Add "rdpid".
23376 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
23377 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
23378 * config/i386/i386.md (define_insn "rdpid"): New.
23379 * config/i386/i386.opt Add -mrdpid.
23380 * config/i386/immintrin.h (_rdpid_u32): New.
23382 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
23384 PR rtl-optimization/79541
23385 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
23386 instead of transforming it into USE.
23388 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
23390 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
23391 If HONOR_SNANS (SFmode) force the input to a register.
23392 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
23393 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
23394 an frsp or similar insn.
23396 2017-02-17 Martin Liska <mliska@suse.cz>
23398 PR rtl-optimization/79577
23399 * params.def (selsched-max-sched-times): Increase minimum to 1.
23401 2017-02-17 Martin Liska <mliska@suse.cz>
23403 PR rtl-optimization/79574
23404 * gcse.c (want_to_gcse_p): Prevent integer overflow.
23406 2017-02-17 Martin Liska <mliska@suse.cz>
23408 PR tree-optimization/79529
23409 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
23410 ssa_defined_default_def_p to handle cases which are implicitly
23412 * tree-ssa.c (ssa_defined_default_def_p): New function.
23413 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
23414 which are implicitly defined.
23415 * tree-ssa.h (ssa_defined_default_def_p): Declare.
23417 2017-02-17 Richard Biener <rguenther@suse.de>
23419 PR middle-end/79576
23420 * params.def (max-ssa-name-query-depth): Limit to 10.
23422 2017-02-17 Richard Biener <rguenther@suse.de>
23424 PR tree-optimization/79552
23425 * tree-ssa-structalias.c (visit_loadstore): Properly verify
23428 2017-02-17 Richard Biener <rguenther@suse.de>
23431 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
23433 2017-02-17 Marek Polacek <polacek@redhat.com>
23435 PR middle-end/79536
23436 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
23437 (fold_negate_expr): New wrapper.
23439 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
23441 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
23442 Correct terminology and de-emphasize pre-standard behavior.
23444 2017-02-16 Alan Modra <amodra@gmail.com>
23446 PR rtl-optimization/79286
23447 * ira.c (def_dominates_uses): New function.
23448 (update_equiv_regs): Don't create an equivalence for insns that
23449 may trap where the register def does not dominate the use.
23451 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
23453 PR rtl-optimization/78127
23454 * lra.c (lra): Call lra_eliminate before finish the loop after
23457 2017-02-16 Richard Biener <rguenther@suse.de>
23459 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
23461 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
23462 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
23463 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
23464 (isl_val_int_from_wi): New function.
23465 (extract_affine_gmp): Rename to ...
23466 (extract_affine_wi): ... this, take a widest_int.
23467 (extract_affine_int): Just wrap extract_affine_wi.
23468 (add_param_constraints): Use isl_val_int_from_wi.
23469 (add_loop_constraints): Likewise, and extract_affine_wi.
23471 2017-02-15 Jeff Law <law@redhat.com>
23473 PR middle-end/79521
23474 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
23475 ira_init_register_move_cost_if_necessary.
23477 2017-02-15 Martin Sebor <msebor@redhat.com>
23479 PR middle-end/32003
23480 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
23481 removed in a prior commit.
23483 2017-02-15 Bin Cheng <bin.cheng@arm.com>
23485 PR tree-optimization/79347
23486 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
23487 counters during peeling.
23489 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
23491 * Makefile.in (site.exp): Remove "set ISLVER".
23493 2017-02-15 Jakub Jelinek <jakub@redhat.com>
23496 * real.c (real_from_integer): Call real_convert even for decimal.
23498 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
23501 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
23503 2017-02-14 Andrew Pinski <apinski@cavium.com>
23505 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
23506 cores and change the partno/implementer to be correct.
23507 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
23508 the 'B" as the implementer.
23509 * config/aarch64/aarch64-tune.md: Regenerate.
23511 2017-02-14 Carl Love <cel@us.ibm.com>
23513 * config/rs6000/rs6000.c: Add case statement entry to make the
23514 xvcvuxdsp built-in argument unsigned.
23515 * config/rs6000/vsx.md: Fix the source and return operand types so they
23516 match the instruction definitions from the ISA document. Fix typo
23517 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
23520 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
23523 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
23524 member early_clobber_alts.
23525 * lra-lives.c (reg_early_clobber_p): New.
23526 (process_bb_lives): Use it.
23527 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
23528 (debug_operand_data): Initialize early_clobber_alts.
23529 (setup_operand_alternative): Set up early_clobber_alts.
23530 (collect_non_operand_hard_regs): Ditto. Pass early clobber
23531 alternatives to new_insn_reg.
23532 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
23534 (lra_update_insn_regno_info): Pass the new arg.
23536 2017-02-14 Jakub Jelinek <jakub@redhat.com>
23538 PR middle-end/79505
23539 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
23540 (new_oacc_loop_raw): Don't clear already cleared fields.
23543 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
23544 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
23545 _mm512_prefetch_i64gather_ps): New inline functions and macros.
23547 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
23550 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
23552 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
23555 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
23556 the extra instruction to the right place to store 128-bit constant
23559 2017-02-14 Martin Sebor <msebor@redhat.com>
23561 PR middle-end/79448
23562 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
23563 warning for strings of unknown length.
23565 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
23567 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
23569 2017-02-14 Jeff Law <law@redhat.com>
23572 * ira-costs.c (scan_one_insn): Initialize register move costs
23573 for pseudos seen in USE/CLOBBER insns.
23575 PR tree-optimization/79095
23576 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
23577 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
23578 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
23579 if the operands are known to be not equal, then the resulting range
23581 (intersect_ranges): If the new range is ~[0,0] and the old range is
23582 wide, then prefer ~[0,0].
23583 * tree-vrp.c (overflow_comparison_p_1): New function.
23584 (overflow_comparison_p): New function.
23585 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
23586 if NAME is used in an overflow test.
23587 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
23588 overflow check that can be expressed as an equality test, then adjust
23589 ops to be that equality test.
23591 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23593 * config/s390/s390-builtin-types.def: Remove flags argument.
23594 * config/s390/s390.c (s390_init_builtins): Likewise.
23596 2017-02-14 Martin Liska <mliska@suse.cz>
23598 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
23599 vector. Fix trailing white spaces.
23601 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
23603 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
23606 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23608 PR rtl-optimization/68664
23609 * config/arm/arm.c (arm_sched_can_speculate_insn):
23610 New function. Declare prototype.
23611 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
23613 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23615 PR rtl-optimization/68664
23616 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
23618 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
23620 2017-02-14 Amit Pawar <amit.pawar@amd.com>
23622 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
23623 max skip bytes for function, loop and jump.
23625 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23627 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
23628 ABS_EXPR for gimple dump.
23630 2017-02-14 Jakub Jelinek <jakub@redhat.com>
23633 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
23635 PR tree-optimization/79408
23636 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
23637 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
23638 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
23639 also if rhs1 is INTEGER_CST.
23641 2017-02-14 Richard Biener <rguenther@suse.de>
23643 PR middle-end/79432
23644 * tree-into-ssa.c (insert_phi_nodes): When the function can
23645 have abnormal edges rewrite SSA names with broken use-def
23646 dominance out of SSA and register them for PHI insertion.
23648 2017-02-13 Martin Sebor <msebor@redhat.com>
23650 PR middle-end/79496
23651 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
23652 clearing info.nowrite flag when snprintf size argument is a range.
23654 2017-02-13 Jakub Jelinek <jakub@redhat.com>
23656 * cprop.c (cprop_jump): Add missing space in string literal.
23657 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
23658 (get_constraint_for_component_ref): Likewise.
23659 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
23660 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
23661 * lra-constraints.c (process_alt_operands): Likewise.
23662 * ipa-inline.c (inline_small_functions): Likewise.
23663 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
23664 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23665 * trans-mem.c (diagnose_tm_1_op): Likewise.
23666 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
23667 (grid_parallel_clauses_gridifiable): Likewise.
23669 * config/nvptx/mkoffload.c (process): Add space in between
23672 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
23673 "MOD4_SSE_REGS" and "ALL_REGS".
23675 * spellcheck.c (test_data): Add , in between "foo" and "food".
23677 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23680 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
23681 boundary crossing check and subsequent code generation agree.
23683 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23685 * config/aarch64/aarch64.c (has_memory_op): Delete.
23686 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
23689 2017-02-13 Jakub Jelinek <jakub@redhat.com>
23691 PR rtl-optimization/79388
23692 PR rtl-optimization/79450
23693 * combine.c (distribute_notes): When removing TEM_INSN for which
23694 corresponding dest has last value recorded, invalidate that last
23697 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23699 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
23700 of explicit '@'. Add missing assembly comment marker on branch costs
23703 2017-02-13 Nathan Sidwell <nathan@acm.org>
23705 * gengtype-lex.l (<in_struct>): Add '/'.
23707 2017-02-13 Martin Liska <mliska@suse.cz>
23710 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
23712 2017-02-13 Richard Biener <rguenther@suse.de>
23714 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
23716 * configure: Re-generate.
23717 * config.in: Likewise.
23718 * graphite-dependences.c: Simplify as if
23719 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
23720 * graphite-isl-ast-to-gimple.c: Likewise.
23721 * graphite-optimize-isl.c: Likewise.
23722 * graphite-poly.c: Likewise.
23723 * graphite-sese-to-poly.c: Likewise.
23724 * graphite.h: Likewise.
23725 * toplev.c: Include isl/version.h and use isl_version () for
23726 printing the ISL version.
23727 * doc/install.texi: Update ISL requirement.
23729 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
23731 * doc/standards.texi (Standards): Update reference to
23734 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
23736 * doc/extend.texi (Named Address Spaces): sourceware.org now
23738 * doc/install.texi (Binaries): Ditto.
23741 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
23743 * doc/cpp.texi: Replace "stringify"/"stringification" with C
23744 standard terminology "stringize"/"stringizing" throughout.
23745 * doc/cppinternals.texi: Likewise.
23747 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
23749 * doc/extend.texi: Fix some spelling mistakes and typos.
23750 * doc/invoke.texi: Likewise.
23752 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
23755 * params.def (inline-min-speedup) Change from 10 to 8.
23757 2017-02-11 Jakub Jelinek <jakub@redhat.com>
23759 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
23762 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
23765 * ipa-inline-analysis.c (get_minimal_bb): New function.
23766 (record_modified): Use it.
23767 (remap_edge_change_prob): Handle also ancestor functions.
23769 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
23771 * doc/contrib.texi (Contributors): Remove broken link into
23772 the Mauve CVS repository.
23774 2017-02-11 Jakub Jelinek <jakub@redhat.com>
23776 PR middle-end/79454
23777 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
23778 result computation whenever lhs doesn't have vector mode, not
23779 just when it has BLKmode.
23781 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
23783 * doc/makefile.texi (profiledbootstrap): Refer to the
23784 installation instructions only in textual form.
23786 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23789 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
23791 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
23793 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
23794 (Specific): Update mingw-w64 reference.
23796 (Specific): Remove broken link to Renesas RX processor.
23798 2017-02-10 Richard Biener <rguenther@suse.de>
23800 * toplev.c (process_options): Do not mention obsolete graphite
23801 options when printing sorry message about missing graphite support.
23802 Mention -floop-nest-optimize.
23804 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
23806 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
23807 (vtst_p16): Likewise.
23808 (vtstq_p8): Likewise.
23809 (vtstq_p16): Likewise.
23811 (vtstq_p64): Likewise.
23812 * config/arm/arm_neon.h (vgetq_lane_p64): New.
23813 (vset_lane_p64): New.
23814 (vsetq_lane_p64): New.
23816 2017-02-10 Jakub Jelinek <jakub@redhat.com>
23818 PR tree-optimization/79411
23819 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
23820 stmt operands are SSA_NAMEs used in abnormal phis.
23821 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
23824 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
23827 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
23830 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
23832 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
23834 2017-02-09 Jakub Jelinek <jakub@redhat.com>
23836 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
23840 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
23841 not arbitrary TREE_CONSTANT.
23844 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
23845 "omp declare target link" attribute unless is_global_var.
23846 * omp-offload.c (find_link_var_op): Likewise.
23848 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
23849 Chung-Lin Tang <cltang@codesourcery.com>
23851 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
23853 (gimplify_adjust_omp_clauses): Don't delete TILE.
23854 (gimplify_omp_for): Deal with TILE.
23855 * internal-fn.c (expand_GOACC_TILE): New function.
23856 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
23858 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
23859 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
23861 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
23862 avoid DIV for outermost collapse var.
23863 (expand_oacc_for): Insert tile element loop as needed. Adjust.
23864 Remove out of date comments, fix whitespace.
23865 * omp-general.c (omp_extract_for_data): Deal with tiling.
23866 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
23867 adjust OLF_DIM_BASE value.
23868 (struct omp_for_data): Add tiling field.
23869 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
23870 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
23871 for auto loops. Remove default auto determining, moved to
23872 oacc_loop_fixed_partitions.
23873 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
23874 stmts, add e_mask field.
23875 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
23876 (oacc_thread_numbers): Use oacc_dim_call.
23877 (oacc_xform_tile): New.
23878 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
23879 (finish_oacc_loop): Adjust for ifns vector.
23880 (oacc_loop_discover_walk): Append loop abstraction sites to list,
23881 add case for GOACC_TILE fns.
23882 (oacc_loop_xform_loop): Delete.
23883 (oacc_loop_process): Iterate over call list directly, and add
23884 handling for GOACC_TILE fns.
23885 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
23887 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
23888 vector partitioning to outer loops. Assign 2 partitions to loops
23889 when available. Add TILE handling.
23890 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
23891 (execite_oacc_device_lower): Process GOACC_TILE fns,
23892 ignore unknown specs.
23893 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
23894 * tree.c (omp_clause_num_ops): Adjust TILE ops.
23895 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
23897 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
23899 * configure.ac (ACX_BUGURL): Update.
23900 * configure: Regenerate.
23902 2017-02-09 Richard Biener <rguenther@suse.de>
23904 PR tree-optimization/69823
23905 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
23906 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
23908 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
23910 * config/arc/arc-c.def: Add __NPS400__ definition.
23911 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
23912 (TARGET_NPS400): Define.
23914 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
23916 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
23918 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
23919 pointer, arch_info.
23920 (arc_cpu_types): Fill the arch_info field with a pointer into the
23921 arc_arch_types table.
23922 (arc_selected_cpu): Declare.
23923 * config/arc/arc.c (arc_selected_cpu): Make global.
23924 (arc_selected_arch): Delete.
23925 (arc_base_cpu): Delete.
23926 (arc_override_options): Remove references to deleted variables,
23927 update access to arch information.
23928 (ARC_OPT): Update access to arch information.
23929 (ARC_OPTX): Likewise.
23930 * config/arc/arc.h (arc_base_cpu): Remove declaration.
23931 (TARGET_ARC600): Update access to arch information.
23932 (TARGET_ARC601): Likewise.
23933 (TARGET_ARC700): Likewise.
23934 (TARGET_EM): Likewise.
23935 (TARGET_HS): Likewise.
23936 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
23939 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
23942 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
23943 condition/operands for integer GE/LE/GEU/LEU operations.
23945 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
23947 PR translation/79397
23948 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
23951 2017-02-08 Martin Jambor <mjambor@suse.cz>
23954 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
23955 whether allocation happened.
23956 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
23957 nothing was allocated.
23959 2017-02-08 Jakub Jelinek <jakub@redhat.com>
23961 PR tree-optimization/79408
23962 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
23963 constant, but SSA_NAME with a known integer range, use the minimum
23964 of that range instead of op1 to determine if modulo can be replaced
23965 with its first operand.
23967 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23969 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
23971 2017-02-08 Richard Biener <rguenther@suse.de>
23973 PR tree-optimization/71824
23974 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
23975 Check all loops contained in the merged region.
23977 2017-02-07 Andrew Pinski <apinski@cavium.com>
23979 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
23981 2017-02-07 Andrew Pinski <apinski@cavium.com>
23983 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
23984 (thunderxt88): Likewise.
23985 (thunderxt81): Disable LSE and change v8.1 to v8.
23986 (thunderxt83): Likewise.
23988 2017-02-07 Jakub Jelinek <jakub@redhat.com>
23989 Richard Biener <rguenther@suse.de>
23991 PR middle-end/79399
23992 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
23993 type from int to size_t.
23994 * ira-costs.c (struct_costs_size): Change type from int to size_t.
23996 2017-02-07 Jakub Jelinek <jakub@redhat.com>
23998 PR rtl-optimization/79386
23999 * cprop.c (bypass_conditional_jumps): Initialize
24000 bypass_last_basic_block already before splitting bbs after
24001 unconditional traps...
24002 (bypass_conditional_jumps): ... rather than here.
24005 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
24006 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
24007 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
24008 fix -masm=intel patterns.
24010 2017-02-07 Richard Biener <rguenther@suse.de>
24012 PR tree-optimization/79256
24013 PR middle-end/79278
24014 * builtins.c (get_object_alignment_2): Use min_align_of_type
24015 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
24016 and ADJUST_FIELD_ALIGN.
24018 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
24020 * doc/tm.texi: Regenerate.
24021 * stor-layout.c (layout_decl): Adjust.
24022 (update_alignment_for_field): Likewise.
24023 (place_field): Likewise.
24024 (min_align_of_type): Likewise.
24025 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
24026 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
24027 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
24028 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
24029 * config/frv/frv.c (frv_adjust_field_align): Likewise.
24030 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
24031 * config/i386/i386.c (x86_field_alignment): Likewise.
24032 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
24033 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
24034 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
24035 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
24036 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
24037 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
24041 2017-01-30 Richard Biener <rguenther@suse.de>
24043 PR tree-optimization/79256
24044 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
24045 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
24048 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
24050 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
24051 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
24052 builtins to SImode and emit a zero-extend, if necessary.
24054 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
24056 * docs/invoke.texi (RISC-V Options): Alphabetize.
24058 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
24060 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
24063 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
24065 * config/riscv/riscv.c: New file.
24066 * common/config/riscv/riscv-common.c: Likewise.
24067 * config.gcc: Likewise.
24068 * config/riscv/constraints.md: Likewise.
24069 * config/riscv/elf.h: Likewise.
24070 * config/riscv/generic.md: Likewise.
24071 * config/riscv/linux.h: Likewise.
24072 * config/riscv/multilib-generator: Likewise.
24073 * config/riscv/peephole.md: Likewise.
24074 * config/riscv/pic.md: Likewise.
24075 * config/riscv/predicates.md: Likewise.
24076 * config/riscv/riscv-builtins.c: Likewise.
24077 * config/riscv/riscv-c.c: Likewise.
24078 * config/riscv/riscv-ftypes.def: Likewise.
24079 * config/riscv/riscv-modes.def: Likewise.
24080 * config/riscv/riscv-opts.h: Likewise.
24081 * config/riscv/riscv-protos.h: Likewise.
24082 * config/riscv/riscv.h: Likewise.
24083 * config/riscv/riscv.md: Likewise.
24084 * config/riscv/riscv.opt: Likewise.
24085 * config/riscv/sync.md: Likewise.
24086 * config/riscv/t-elf-multilib: Likewise.
24087 * config/riscv/t-linux: Likewise.
24088 * config/riscv/t-linux-multilib: Likewise.
24089 * config/riscv/t-riscv: Likewise.
24090 * configure.ac: Likewise.
24091 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
24092 Waterman as RISC-V maintainers.
24093 * doc/install.texi: Add RISC-V entries.
24094 * doc/invoke.texi: Add RISC-V options section.
24095 * doc/md.texi: Add RISC-V constraints section.
24096 * configure: Regenerated.
24098 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
24101 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
24102 false values to be constant vectors with all 0 or all 1 bits set.
24103 (vcondu<mode><mode>): Likewise.
24104 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
24106 (fpmask_comparison_operator): Update comment.
24107 (vecint_comparison_operator): New predicate.
24108 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
24109 vector conditionals when the true and false values are constant
24110 vectors with all 0 bits or all 1 bits set.
24112 2017-02-06 Martin Sebor <msebor@redhat.com>
24114 PR tree-optimization/79376
24115 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
24117 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
24119 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
24120 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
24121 to simplify split condition.
24123 2017-02-06 Jakub Jelinek <jakub@redhat.com>
24125 * omp-expand.c (oxpand_omp_atomic_fetch_op,
24126 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
24129 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
24131 PR rtl-optimization/68664
24132 * target.def (can_speculate_insn): New hook.
24133 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
24134 * doc/tm.texi: Regenerate.
24135 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
24136 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
24137 (rs6000_sched_can_speculate_insn): New function.
24139 2017-02-06 Jakub Jelinek <jakub@redhat.com>
24141 PR tree-optimization/79284
24142 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
24143 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
24144 vectorizable_mask_load_store, vectorizable_operation,
24145 vect_is_simple_cond, get_same_sized_vectype): Use it instead
24146 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
24147 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
24148 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
24149 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
24150 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
24151 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
24152 is_gimple_assign (stmt). Replace another such test with
24153 is_gimple_assign (stmt).
24155 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
24158 * config/avr/avr.c (rtl-iter.h): Include it.
24159 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
24160 (avr_legitimate_combined_insn): ...and implementation.
24162 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
24164 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
24165 * config/s390/s390.c (s390_const_operand_ok)
24166 (s390_canonicalize_comparison, s390_extract_part)
24167 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
24168 (s390_contiguous_bitmask_p, s390_rtx_costs)
24169 (legitimize_pic_address): Likewise.
24170 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
24171 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
24172 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
24173 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
24174 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
24176 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
24178 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
24179 REGNO($0) == REGNO($1).
24181 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24183 * config/s390/linux.h(SIZE_TYPE): Add comment.
24185 2017-02-06 Julian Brown <julian@codesourcery.com>
24186 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24187 Virendra Pathak <virendra.pathak@broadcom.com>
24189 * config/aarch64/aarch64-cores.def: Change the scheduler
24191 * config/aarch64/aarch64.md: Include thunderx2t99.md.
24192 * config/aarch64/thunderx2t99.md: New file.
24194 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
24196 * doc/standards.texi (Go Language): Update link to language
24199 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
24201 * tree-eh.c (lower_resx): Sanitize profile.
24202 (cleanup_empty_eh_move_lp): Likewise.
24204 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
24207 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
24209 * cfgloopmanip.h (loop_version): Update prototype.
24210 * modulo-sched.c (sms_schedule): Update call of loop_version.
24211 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
24212 * tree-parloops.c (gen_parallel_loop): Likewise.
24213 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
24214 * tree-ssa-loop-split.c (split_loop): Likewise.
24215 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
24216 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24218 2017-02-05 Martin Liska <mliska@suse.cz>
24221 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
24223 (print_operand_address): Initialize a struct to zero.
24225 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
24227 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
24228 garbage collector only in textual form.
24230 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
24232 * doc/extend.texi (x86 specific memory model extensions for
24233 transactional memory): Simplify a phrase.
24235 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
24238 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
24239 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
24240 (atomic_storedi_1): Likewise.
24242 2017-02-04 Jakub Jelinek <jakub@redhat.com>
24244 PR tree-optimization/79338
24245 * tree-parloops.c (gather_scalar_reductions): Don't call
24246 vect_analyze_loop_form for loop->inner before destroying loop's
24249 2017-02-03 Martin Sebor <msebor@redhat.com>
24251 PR tree-optimization/79327
24252 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
24253 when precision has resulted in leading zeros.
24254 (format_integer): Adjust the likely counter to assume an unknown
24255 argument that may be zero is non-zero.
24257 2017-02-03 Jason Merrill <jason@redhat.com>
24260 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
24261 avoid copying non-taken branch.
24263 2017-02-03 Jakub Jelinek <jakub@redhat.com>
24265 PR tree-optimization/79340
24266 * tree-vect-loop.c (vectorizable_reduction): Release
24267 vec_defs elements after safe_splicing them into other vectors.
24270 PR tree-optimization/79327
24271 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
24272 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
24274 (format_integer): Use wide_int_to_tree instead of build_int_cst
24275 + to_?hwi. If argmin is NULL, just set argmin and argmax to
24276 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
24277 of shortest and longest sequence.
24279 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
24281 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
24282 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
24284 2017-02-03 Walter Lee <walt@tilera.com>
24287 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
24288 after initial stackframe link reg save.
24289 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
24291 2017-02-03 Jakub Jelinek <jakub@redhat.com>
24294 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
24295 wu for stxssp alternative.
24297 2017-02-03 Martin Sebor <msebor@redhat.com>
24299 PR tree-optimization/79352
24300 * gimple-fold.c (get_range_strlen): Add argument.
24301 (get_range_strlen): Change return type to bool.
24302 (get_maxval_strlen): Pass in a dummy argument.
24303 * gimple-fold.h (get_range_strlen): Change return type to bool.
24304 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
24305 * tree.h (array_at_struct_end_p): Add argument.
24306 * tree.c (array_at_struct_end_p): Handle it.
24308 2017-02-03 Martin Liska <mliska@suse.cz>
24311 * multiple_target.c (create_dispatcher_calls): Redirect edge
24312 from a caller of a dispatcher.
24313 (expand_target_clones): Make the clones local.
24314 (ipa_target_clone): Do both target clones and resolvers.
24315 (ipa_dispatcher_calls): Remove the pass.
24316 (pass_dispatcher_calls::gate): Likewise.
24317 (make_pass_dispatcher_calls): Likewise.
24318 * passes.def (pass_target_clone): Put as very first IPA early
24321 2017-02-03 Martin Liska <mliska@suse.cz>
24323 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
24324 in case of a function with ifunc attribute.
24326 2017-02-03 Martin Liska <mliska@suse.cz>
24328 * cgraph.c (cgraph_node::dump): Dump function version info.
24329 * symtab.c (symtab_node::dump_base): Add missing new line.
24331 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
24333 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
24334 (ifcombine_ifandif): Use it.
24336 2017-02-03 Martin Liska <mliska@suse.cz>
24338 * doc/invoke.texi: Document default value for
24339 use-after-scope-direct-emission-threshold.
24341 2017-02-03 Martin Liska <mliska@suse.cz>
24343 PR tree-optimization/79339
24344 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
24345 (format_floating): Likewise.
24347 2017-02-03 Martin Liska <mliska@suse.cz>
24350 * ipa-prop.c (ipa_node_params_t::insert): Remove current
24352 (ipa_node_params_t::remove): Likewise.
24353 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
24354 initialization from removed ipa_node_params_t::insert.
24355 (ipa_node_params::~ipa_node_params): Move from removed
24356 ipa_node_params_t::release.
24357 * symbol-summary.h (symbol_summary::m_released): New member.
24358 Do not release a summary twice. Do not allow to call finalizer
24359 for types of a summary that live in GGC memory.
24361 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24363 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
24366 2017-02-02 Martin Sebor <msebor@redhat.com>
24368 PR middle-end/79275
24369 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
24370 (format_string): Tighten up the range of output for non-constant
24371 strings and correct the expected range for wide non-constant strings.
24373 2017-02-02 Martin Sebor <msebor@redhat.com>
24375 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
24377 PR middle-end/32003
24378 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
24380 (-fdump-tree-@var): Add to index and document how to come up
24381 with pass-specific option and dump file names.
24382 (-fdump-passes): Clarify where to look for output.
24384 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
24386 PR middle-end/77445
24387 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
24388 statistics of the analyzed path; allow threading for speed when
24389 any of BBs along the path are optimized for speed.
24391 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
24393 PR middle-end/78468
24394 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
24395 settings of the virtual registers.
24398 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
24400 * explow.c (get_dynamic_stack_size): Take known alignment of stack
24401 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
24404 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24406 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
24407 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
24409 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24411 * config/s390/s390.md: Add missing comments with the expanded
24413 * config/s390/vector.md: Likewise.
24414 * config/s390/vx-builtins.md: Likewise.
24416 2017-02-02 Jakub Jelinek <jakub@redhat.com>
24419 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
24420 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
24421 conditions on a single line.
24423 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24425 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
24426 __S390_VX__ to __VX__.
24428 2017-02-01 Andrew Pinski <apinski@cavium.com>
24430 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
24431 stmt_info to record_stmt_cost.
24432 (vect_get_known_peeling_cost): Pass stmt_info if known to
24434 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
24435 cpu_vector_cost field into
24436 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
24437 field into vec_int_stmt_cost and vec_fp_stmt_cost.
24438 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
24439 splitting of scalar_stmt_cost and vec_stmt_cost.
24440 (thunderx_vector_cost): Likewise.
24441 (cortexa57_vector_cost): LIkewise.
24442 (exynosm1_vector_cost): Likewise.
24443 (xgene1_vector_cost): Likewise.
24444 (thunderx2t99_vector_cost): Improve after the splitting of the two
24446 (aarch64_builtin_vectorization_cost): Update for the splitting of
24447 scalar_stmt_cost and vec_stmt_cost.
24449 2017-02-01 Torvald Riegel <triegel@redhat.com>
24450 Richard Henderson <rth@redhat.com>
24452 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
24453 conditional on existance of a fast atomic load.
24454 * optabs-query.c (can_atomic_load_p): New function.
24455 * optabs-query.h (can_atomic_load_p): Declare it.
24456 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
24457 no fast atomic load is available for the particular size of access.
24458 (expand_atomic_compare_and_swap): Likewise.
24459 (expand_atomic_load): Likewise.
24460 (expand_atomic_store): Likewise.
24461 (expand_atomic_fetch_op): Likewise.
24462 * testsuite/lib/target-supports.exp
24463 (check_effective_target_sync_int_128): Remove x86 because it provides
24464 no fast atomic load.
24465 (check_effective_target_sync_int_128_runtime): Likewise.
24467 2017-02-01 Richard Biener <rguenther@suse.de>
24469 * graphite.c: Include tree-vectorizer.h for find_loop_location.
24470 (graphite_transform_loops): Provide opt-info for optimized nests.
24471 * tree-parloop.c (parallelize_loops): Provide opt-info for
24472 parallelized loops.
24474 2017-02-01 Richard Biener <rguenther@suse.de>
24476 PR middle-end/79315
24477 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
24478 was not set before.
24480 2017-02-01 Richard Biener <rguenther@suse.de>
24482 PR tree-optimization/71824
24483 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
24484 Verify the loops are valid in the merged SESE region.
24485 (scop_detection::can_represent_loop_1): Check analyzing the
24486 evolution of the number of iterations in the region succeeds.
24488 2017-01-31 Ian Lance Taylor <iant@golang.org>
24490 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
24491 REG_ARGS_SIZE note to 32-bit push insns and call insn.
24493 2017-01-31 David Malcolm <dmalcolm@redhat.com>
24495 PR preprocessor/79210
24496 * input.c (get_substring_ranges_for_loc): Replace line_width
24497 assertion with error-handling.
24499 2017-01-31 Richard Biener <rguenther@suse.de>
24501 PR tree-optimization/77318
24502 * graphite-sese-to-poly.c (extract_affine): Fix assert.
24503 (create_pw_aff_from_tree): Take loop parameter.
24504 (add_condition_to_pbb): Pass loop of the condition to
24505 create_pw_aff_from_tree.
24507 2017-01-31 Jakub Jelinek <jakub@redhat.com>
24509 * config/s390/s390.c (s390_asan_shadow_offset): New function.
24510 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
24512 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
24516 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
24518 (convert_int_to_float128): Likewise.
24519 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
24520 (convert_int_to_float128): Likewise.
24521 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
24522 (UNSPEC_IEEE128_CONVERT): Likewise.
24523 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
24524 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
24525 Use local variables for IBM extended format.
24526 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
24527 (fix_trunc<mode>si2_fprs): Likewise.
24528 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
24529 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
24530 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
24531 to know that we can now have integers of all sizes in vector
24533 (fix<uns>_<mode>di2_hw): Likewise.
24534 (float<uns>_<mode>si2_hw): Likewise.
24535 (fix_<mode>si2_hw): Likewise.
24536 (fixuns_<mode>si2_hw): Likewise.
24537 (float<uns>_<mode>di2_hw): Likewise.
24538 (float_<mode>di2_hw): Likewise.
24539 (float_<mode>si2_hw): Likewise.
24540 (floatuns_<mode>di2_hw): Likewise.
24541 (floatuns_<mode>si2_hw): Likewise.
24542 (xscvqp<su>wz_<mode>): Delete, no longer used.
24543 (xscvqp<su>dz_<mode>): Likewise.
24544 (xscv<su>dqp_<mode>): Likewise.
24545 (ieee128_mfvsrd_64bit): Likewise.
24546 (ieee128_mfvsrd_32bit): Likewise.
24547 (ieee128_mfvsrwz): Likewise.
24548 (ieee128_mtvsrw): Likewise.
24549 (ieee128_mtvsrd_64bit): Likewise.
24550 (ieee128_mtvsrd_32bit): Likewise.
24552 2017-01-31 Martin Liska <mliska@suse.cz>
24555 * ipa-prop.c (ipa_free_all_node_params): Call release method
24556 instead of ~sumbol_summary to not to trigger double times
24559 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
24561 PR tree-optimization/71691
24562 * bitmap.h (class auto_bitmap): New.
24563 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
24564 is_maybe_undefined instead of ssa_undefined_value_p.
24566 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24568 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
24569 __S390_ARCH_LEVEL__ to __ARCH__.
24571 2017-01-31 Jakub Jelinek <jakub@redhat.com>
24573 PR tree-optimization/79267
24574 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
24575 if should_remove_lhs_p is true.
24577 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
24580 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
24581 (add_alignment_attribute): New.
24582 (base_type_die): Add alignment attribute.
24583 (subrange_type_die): Likewise.
24584 (modified_type_die): Likewise.
24585 (gen_array_type_die): Likewise.
24586 (gen_descr_array_type_die: Likewise.
24587 (gen_enumeration_type_die): Likewise.
24588 (gen_subprogram_die): Likewise.
24589 (gen_variable_die): Likewise.
24590 (gen_field_die): Likewise.
24591 (gen_ptr_to_mbr_type_die): Likewise.
24592 (gen_struct_or_union_type_die): Likewise.
24593 (gen_subroutine_type_die): Likewise.
24594 (gen_typedef_die): Likewise.
24595 (base_type_cmp): Compare alignment attribute.
24597 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24600 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
24601 (setb_unsigned) New pattern for setb with CCUNS.
24602 * config/rs6000/rs6000.c (expand_block_compare): Use a different
24603 subfc./subfe sequence to avoid overflow problems. Generate a
24604 shorter sequence with cmpld/setb for power9.
24605 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
24606 for generating subfc. instruction.
24607 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
24608 now uses this instruction.
24610 2017-01-30 Ian Lance Taylor <iant@google.com>
24613 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
24614 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
24616 2017-01-30 Martin Sebor <msebor@redhat.com>
24618 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
24619 Move constant to the right of a relational operator.
24620 (get_mpfr_format_length, format_character, format_string): Ditto.
24621 (should_warn_p, maybe_warn): Same.
24623 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
24625 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
24628 * asan.c (get_translation_unit_decl): Remove function.
24629 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
24631 2017-01-30 Martin Liska <mliska@suse.cz>
24633 PR gcov-profile/79259
24634 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
24635 -fprofile-generate.
24637 2017-01-30 Martin Liska <mliska@suse.cz>
24640 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
24641 Initialize variables with NULL value.
24643 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
24646 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
24648 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
24650 2017-01-30 Richard Biener <rguenther@suse.de>
24652 PR tree-optimization/79276
24653 * tree-vrp.c (process_assert_insertions): Properly adjust common
24654 when removing a duplicate.
24656 * gcc.dg/torture/pr79276.c: New testcase.
24658 2017-01-30 Richard Biener <rguenther@suse.de>
24660 PR tree-optimization/79256
24661 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
24662 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
24664 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
24666 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
24669 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
24670 ("*r<noxa>sbg_<mode>_sll_bitmask")
24671 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
24672 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
24673 Use contiguous_bitmask_nowrap_operand.
24675 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24678 * config/rs6000/altivec.h (vec_xl): Revise #define.
24679 (vec_xst): Likewise.
24681 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
24683 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
24685 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
24687 PR rtl-optimization/79194
24688 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
24689 traps before call to bypass_conditional_jumps.
24691 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
24693 PR tree-optimization/71374
24694 * lra-constraints.c (check_conflict_input_operands): New.
24695 (match_reload): Use it.
24697 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
24700 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
24701 account to calculate conflict_set.
24703 2017-01-27 Bin Cheng <bin.cheng@arm.com>
24705 PR rtl-optimization/78559
24706 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
24707 other_insn in combine.
24709 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24711 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
24712 uint16_type_node for BT_UINT16.
24714 2017-01-27 David Malcolm <dmalcolm@redhat.com>
24716 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
24717 "RTL Tests" to menu.
24718 (GIMPLE Tests): New node.
24719 (RTL Tests): New node.
24721 2017-01-27 Richard Biener <rguenther@suse.de>
24723 PR tree-optimization/79245
24724 * tree-loop-distribution.c (distribute_loop): Apply cost
24725 modeling also to detected patterns.
24727 2017-01-27 Richard Biener <rguenther@suse.de>
24729 PR tree-optimization/71433
24730 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
24731 (compare_assert_loc): New function.
24732 (process_assert_insertions): Sort and optimize assert locations
24733 to remove duplicates and push down identical assertions on
24734 edges to their destination block.
24736 2017-01-27 Richard Biener <rguenther@suse.de>
24738 PR tree-optimization/79244
24739 * tree-vrp.c (remove_range_assertions): Forcefully propagate
24740 out SSA names even if abnormal.
24742 2017-01-27 Jakub Jelinek <jakub@redhat.com>
24744 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
24745 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
24746 instead of MPFR_RNDN.
24748 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
24751 * arm.c (arm_option_override): Don't call build_target_option_node
24752 until after doing all option overrides.
24753 (arm_valid_target_attribute_tree): Likewise.
24755 2017-01-27 Martin Liska <mliska@suse.cz>
24757 * doc/invoke.texi (-fprofile-arcs): Document profiling support
24758 for {cd}tors and C++ {cd}tors.
24760 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
24762 * config/s390/s390.md ("*setmem_long_and")
24763 ("*setmem_long_and_31z"): Use zero_extend instead of and.
24765 2017-01-26 Martin Sebor <msebor@redhat.com>
24767 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
24770 2017-01-26 Martin Sebor <msebor@redhat.com>
24772 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
24773 HAVE_DFmode before using XFmode or DFmode.
24774 (parse_directive): Avoid using the z length modifier to avoid
24775 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
24777 PR middle-end/78703
24778 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
24779 to accept adjustment as an array.
24780 (get_int_range): New function.
24781 (struct directive): Make width and prec arrays.
24782 (directive::set_width, directive::set_precision): Call get_int_range.
24783 (format_integer, format_floating): Handle width and precision ranges.
24784 (format_string, parse_directive): Same.
24786 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24789 * dwarf2out.c (generate_skeleton_bottom_up): For children with
24790 comdat_type_p set, just clone them, but keep the children in the
24794 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
24795 which have direct callers with -fvar-tracking-assignments enabled
24797 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
24798 inside of type units.
24800 2017-01-26 Martin Sebor <msebor@redhat.com>
24802 PR middle-end/78703
24803 * gimple-ssa-sprintf.c (struct result_range): Add likely and
24805 (struct format_result): Replace number_chars, number_chars_min,
24806 and number_chars_max with a single member of struct result_range.
24808 (format_result::operator+=): Adjust.
24809 (struct fmtresult): Remove bounded. Handle likely and unlikely
24811 (fmtresult::adjust_for_width_or_precision): New function.
24812 (fmtresult:type_max_digits): New function.
24813 (bytes_remaining): Handle likely and unlikely counters.
24814 (min_bytes_remaining): Remove.
24815 (format_percent): Simplify.
24816 (format_integer, format_floating): Set likely and unlikely counters.
24817 (get_string_length, format_character, format_string): Same.
24818 (format_plain, should_warn_p): New function.
24819 (maybe_warn): Call should_warn_p. Update diagnostic messages
24820 and handle those for all directives, including plain strings.
24821 (format_directive): Handle likely and unlikely counters.
24822 Remove unnecessary quoting from diagnostics. Add an informational
24824 (add_bytes): Remove.
24825 (pass_sprintf_length::compute_format_length): Simplify.
24826 (try_substitute_return_value): Handle likely and unlikely counters.
24828 2017-01-26 Carl Love <cel@us.ibm.com>
24830 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
24831 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
24833 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
24836 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
24837 endianess for subregs into account.
24838 * lra-constraints.c (lra_constraints): Do risky transformations
24839 always on the first iteration.
24840 * lra-lives.c (check_pseudos_live_through_calls): Add arg
24841 last_call_used_reg_set.
24842 (process_bb_lives): Define and use last_call_used_reg_set.
24843 * lra.c (lra): Always continue after lra_constraints on the first
24846 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
24848 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
24850 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
24852 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24854 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
24855 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
24856 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
24857 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
24858 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
24859 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
24860 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
24861 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
24862 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
24864 2017-01-26 Marek Polacek <polacek@redhat.com>
24867 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
24868 for the third operand.
24870 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24872 PR middle-end/79236
24873 * omp-low.c (struct omp_context): Add simt_stmt field.
24874 (scan_omp_for): Return omp_context *.
24875 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
24876 context to the _simt_ SIMD stmt.
24877 (lower_omp_for): For combined SIMD with sibling _simt_
24878 SIMD, make sure to use the same decls in _looptemp_
24879 clauses as in the sibling.
24881 2017-01-26 David Sherwood <david.sherwood@arm.com>
24883 PR middle-end/79212
24884 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
24887 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24890 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
24891 emit fld b; fld a; if possible.
24893 * brig-builtins.def: Update copyright years.
24894 * config/arm/arm_acle_builtins.def: Update copyright years.
24896 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
24899 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
24900 constraint instead of o for the stxsd instruction.
24902 2017-01-25 Carl Love <cel@us.ibm.com>
24904 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
24905 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
24907 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
24909 * doc/invoke.texi (C++ Dialect Options): Fix typo.
24911 2017-01-25 Richard Biener <rguenther@suse.de>
24913 PR tree-optimization/69264
24914 * target.def (vector_alignment_reachable): Improve documentation.
24915 * doc/tm.texi: Regenerate.
24916 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
24918 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
24919 earlier changes with respect to TYPE_USER_ALIGN.
24920 (vector_alignment_reachable_p): Likewise. Improve dumping.
24922 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24925 * config/arm/arm.md (xordi3): Force constant operand into a register
24928 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24930 * doc/invoke.texi (-fstore-merging): Correct default optimization
24931 levels at which it is enabled.
24932 (-O): Move -fstore-merging from list to...
24935 2017-01-25 Richard Biener <rguenther@suse.de>
24938 * omp-expand.c: Include debug.h.
24939 (expand_omp_taskreg): Make sure to generate early debug before
24940 outlining anything from a function.
24941 (expand_omp_target): Likewise.
24942 (grid_expand_target_grid_body): Likewise.
24944 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
24947 * asan.c (get_translation_unit_decl): New function.
24948 (asan_add_global): Extract modules file name from globals
24949 TRANSLATION_UNIT_DECL name.
24951 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
24954 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
24955 for long calls with APCS frame and VFP.
24957 2017-01-24 David Malcolm <dmalcolm@redhat.com>
24959 * cfg.c (original_copy_tables_initialized_p): New function.
24960 * cfg.h (original_copy_tables_initialized_p): New decl.
24961 * cfgrtl.c (relink_block_chain): Guard the call to
24962 free_original_copy_tables with a call to
24963 original_copy_tables_initialized_p.
24964 * cgraph.h (symtab_node::native_rtl_p): New decl.
24965 * cgraphunit.c (symtab_node::native_rtl_p): New function.
24966 (symtab_node::needed_p): Don't assert for early assembly output
24967 for __RTL functions.
24968 (cgraph_node::finalize_function): Set "force_output" for __RTL
24970 (cgraph_node::analyze): Bail out early for __RTL functions.
24971 (analyze_functions): Update assertion to support __RTL functions.
24972 (cgraph_node::expand): Bail out early for __RTL functions.
24973 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
24975 * function.h (struct function): Update comment for field
24977 * gimple-expr.c: Include "tree-pass.h".
24978 (gimple_has_body_p): Return false for __RTL functions.
24979 * Makefile.in (OBJS): Add run-rtl-passes.o.
24980 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
24982 (gcc::pass_manager::get_clean_slate): New accessor.
24983 * passes.c: Include "insn-addr.h".
24984 (should_skip_pass_p): Add logging. Update logic for running
24985 "expand" to be compatible with both __GIMPLE and __RTL. Guard
24986 property-provider override so it is only done for gimple passes.
24988 (skip_pass): New function.
24989 (execute_one_pass): Call skip_pass when skipping passes.
24990 * read-md.c (md_reader::read_char): Support filtering
24991 the input to a subset of line numbers.
24992 (md_reader::md_reader): Initialize fields
24993 m_first_line and m_last_line.
24994 (md_reader::read_file_fragment): New function.
24995 * read-md.h (md_reader::read_file_fragment): New decl.
24996 (md_reader::m_first_line): New field.
24997 (md_reader::m_last_line): New field.
24998 * read-rtl-function.c (function_reader::create_function): Only
24999 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
25000 curr_properties. Set DECL_INITIAL to a dummy block.
25001 (read_rtl_function_body_from_file_range): New function.
25002 * read-rtl-function.h (read_rtl_function_body_from_file_range):
25004 * run-rtl-passes.c: New file.
25005 * run-rtl-passes.h: New file.
25007 2017-01-24 Jeff Law <law@redhat.com>
25009 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
25012 2017-01-24 Bin Cheng <bin.cheng@arm.com>
25014 PR tree-optimization/79159
25015 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
25016 (record_nonwrapping_iv): Improve boundary using above function if no
25017 value range information.
25019 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
25020 Martin Jambor <mjambor@suse.cz>
25022 * brig-builtins.def: New file.
25023 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
25024 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
25025 (DEF_HSAIL_SAT_BUILTIN): Likewise.
25026 (DEF_HSAIL_INTR_BUILTIN): Likewise.
25027 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
25028 * builtin-types.def (BT_INT8): New.
25029 (BT_INT16): Likewise.
25030 (BT_UINT8): Likewise.
25031 (BT_UINT16): Likewise.
25032 (BT_FN_ULONG): Likewise.
25033 (BT_FN_UINT_INT): Likewise.
25034 (BT_FN_UINT_ULONG): Likewise.
25035 (BT_FN_UINT_LONG): Likewise.
25036 (BT_FN_UINT_PTR): Likewise.
25037 (BT_FN_ULONG_PTR): Likewise.
25038 (BT_FN_INT8_FLOAT): Likewise.
25039 (BT_FN_INT16_FLOAT): Likewise.
25040 (BT_FN_UINT32_FLOAT): Likewise.
25041 (BT_FN_UINT16_FLOAT): Likewise.
25042 (BT_FN_UINT8_FLOAT): Likewise.
25043 (BT_FN_UINT64_FLOAT): Likewise.
25044 (BT_FN_UINT16_UINT32): Likewise.
25045 (BT_FN_UINT32_UINT16): Likewise.
25046 (BT_FN_UINT16_UINT16_UINT16): Likewise.
25047 (BT_FN_INT_PTR_INT): Likewise.
25048 (BT_FN_UINT_PTR_UINT): Likewise.
25049 (BT_FN_LONG_PTR_LONG): Likewise.
25050 (BT_FN_ULONG_PTR_ULONG): Likewise.
25051 (BT_FN_VOID_UINT64_UINT64): Likewise.
25052 (BT_FN_UINT8_UINT8_UINT8): Likewise.
25053 (BT_FN_INT8_INT8_INT8): Likewise.
25054 (BT_FN_INT16_INT16_INT16): Likewise.
25055 (BT_FN_INT_INT_INT): Likewise.
25056 (BT_FN_UINT_FLOAT_UINT): Likewise.
25057 (BT_FN_FLOAT_UINT_UINT): Likewise.
25058 (BT_FN_ULONG_UINT_UINT): Likewise.
25059 (BT_FN_ULONG_UINT_PTR): Likewise.
25060 (BT_FN_ULONG_ULONG_ULONG): Likewise.
25061 (BT_FN_UINT_UINT_UINT): Likewise.
25062 (BT_FN_VOID_UINT_PTR): Likewise.
25063 (BT_FN_UINT_UINT_PTR: Likewise.
25064 (BT_FN_UINT32_UINT64_PTR): Likewise.
25065 (BT_FN_INT_INT_UINT_UINT): Likewise.
25066 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
25067 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
25068 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
25069 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
25070 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
25071 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
25072 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
25073 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
25074 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
25075 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
25076 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
25077 * doc/frontends.texi: List BRIG FE.
25078 * doc/install.texi (Testing): Add BRIG tesring requirements.
25079 * doc/invoke.texi (Overall Options): Mention BRIG.
25080 * doc/standards.texi (Standards): Doucment BRIG HSA version.
25082 2017-01-24 Richard Biener <rguenther@suse.de>
25084 PR translation/79208
25085 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
25087 2017-01-24 Martin Jambor <mjambor@suse.cz>
25090 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
25091 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
25092 and known_contexts.
25094 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
25096 PR middle-end/79123
25097 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
25098 casts from signed to unsigned really don't have a range.
25100 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
25102 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
25103 GMP_RNDx for compatiblity.
25105 2017-01-24 Martin Liska <mliska@suse.cz>
25108 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
25109 that would prevent us to call alloca with -1 as argument.
25111 2017-01-24 Jakub Jelinek <jakub@redhat.com>
25113 * dwarf2out.c (output_compilation_unit_header, output_file_names):
25114 Avoid -Wformat-security warning.
25116 2017-01-23 Andrew Pinski <apinski@cavium.com>
25118 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
25121 2017-01-23 Martin Sebor <msebor@redhat.com>
25123 PR middle-end/78703
25124 * gimple-ssa-sprintf.c (warn_level): New global.
25125 (format_integer): Use it here and throughout the rest of the file.
25126 Use the same switch to compute sign as base.
25127 (maybe_warn): New function.
25128 (format_directive): Factor out warnings into maybe_warn.
25129 Add debugging output. Use warn_level.
25130 (add_bytes): Use warn_level.
25131 (pass_sprintf_length::compute_format_length): Add debugging output.
25132 (try_substitute_return_value): Same.
25133 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
25135 PR middle-end/78703
25136 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
25137 (struct fmtresult, format_integer, format_floating): Adjust.
25138 (fmtresult::fmtresult): Set max correctly in two argument ctor.
25139 (get_string_length, format_string,format_directive): Same.
25140 (pass_sprintf_length::compute_format_length): Same.
25141 (try_substitute_return_value): Simplify slightly.
25143 PR middle-end/78703
25144 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
25145 (fmtresult::operator+=): Outlined.
25146 (struct fmtresult): Add ctors.
25147 (struct conversion_spec): Rename...
25148 (struct directive): ...to this. Add and remove data members.
25149 (directive::set_width, directive::set_precision): New functions.
25150 (format_percent): Use fmtresult ctor.
25151 (get_width_and_precision): Remove.
25152 (format_integer): Make naming changes. Avoid computing width and
25154 (format_floating): Same. Adjust indentation.
25155 (format_character, format_none): New functions.
25156 (format_string): Moved character handling to format_character.
25157 (format_directive): Remove arguments, change return type.
25158 (parse_directive): New function.
25159 (pass_sprintf_length::compute_format_length): Move directive
25160 parsing to parse_directive.
25162 2017-01-23 Jakub Jelinek <jakub@redhat.com>
25164 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
25165 (assign_assembler_name_if_needed): ... this.
25166 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
25167 (assign_assembler_name_if_needed): ... this.
25168 (free_lang_data_in_cgraph): Adjust callers.
25169 * cgraphunit.c (cgraph_node::analyze): Likewise.
25170 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
25173 2017-01-23 Richard Biener <rguenther@suse.de>
25175 PR tree-optimization/79088
25176 PR tree-optimization/79188
25177 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
25178 resetting loop bounds after last path deletion. Reset loop
25179 bounds of the target loop, make code match the comments.
25180 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
25181 Make sure loops need no fixups.
25183 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
25185 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
25186 exponent support with double type for first argument.
25187 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
25188 type returned by __builtin_vec_extract_sig,
25189 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
25190 functions from "vector int" to "vector unsigned int" or from
25191 "vector long long int" to "vector unsigned long long int".
25192 Changed type returned by __builtin_vec_extract_exp,
25193 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
25194 functions from "vector int" to "vector unsigned int" or from
25195 "vector long long int" to "vector unsigned long long int".
25196 Changed return type of __builtin_vec_test_data_class,
25197 __builtin_vec_test_data_class_sp, and
25198 __builtin_vec_test_data_class_dp from "vector int" to
25199 "vector bool int" or from "vector long long int" to "vector bool
25200 long long int" and changed second argument type from "unsigned
25201 int" to "int". Added new overloaded function forms "vector float
25202 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
25203 "vector float __builtin_vec_insert_exp_sp (vector float, vector
25204 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
25205 double, vector unsigned long long int)" and "vector double
25206 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
25207 long int)". Changed return type of
25208 __builtin_scalar_test_data_class and
25209 __builtin_scalar_test_data_class_sp and
25210 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
25211 int" and changed second argument from "unsigned int" to "int".
25212 Changed type returned by __builtin_scalar_test_neg,
25213 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
25214 from "int" to "bool int". Added new overloaded function form
25215 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
25216 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
25217 exponent double-precision with floating point first argument.
25218 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
25219 documentation of scalar_test_data_class, scalar_test_neg,
25220 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
25221 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
25222 vec_test_data_class built-in functions to reflect refinements in
25223 their type signatures.
25225 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
25227 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
25229 (aarch64_elf_asm_destructor): Likewise.
25231 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
25233 PR rtl-optimization/78634
25234 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
25235 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
25236 * ifcvt.c (noce_try_cmove): Add missing cost check.
25238 PR rtl-optimization/71724
25239 * combine.c (if_then_else_cond): Look for situations where it is
25240 beneficial to undo the work of one of the recursive calls.
25242 2017-01-23 Bin Cheng <bin.cheng@arm.com>
25244 PR tree-optimization/70754
25245 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
25246 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
25247 combined stmt before it if not NULL.
25248 (combine_chains): Process refs reversely and compute dominance point
25251 2017-01-23 Martin Liska <mliska@suse.cz>
25253 PR tree-optimization/79196
25254 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
25255 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
25257 (strlen_optimize_stmt): Call the renamed function.
25259 2017-01-23 Michael Matz <matz@suse.de>
25261 PR tree-optimization/78384
25262 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
25264 2017-01-23 Richard Biener <rguenther@suse.de>
25266 PR tree-optimization/79186
25267 * tree-vrp.c (register_new_assert_for): Make sure we've seen
25268 both incoming edges before moving an assert.
25270 2017-01-23 Martin Jambor <mjambor@suse.cz>
25272 * ipa-prop.c (load_from_param_1): Removed.
25273 (load_from_unmodified_param): Bits from load_from_param_1 put back
25275 (load_from_param): Removed.
25276 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
25277 with stmt. Reverted back to use of load_from_unmodified_param.
25279 2017-01-23 Martin Jambor <mjambor@suse.cz>
25282 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
25283 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
25284 field a pointer to garbage collected vector, mark lattices and
25285 ipcp_orig_node with GTY((skip)).
25286 (ipa_get_param_count): Adjust to descriptors being a pointer.
25287 (ipa_get_param): Likewise.
25288 (ipa_get_type): Likewise.
25289 (ipa_get_param_move_cost): Likewise.
25290 (ipa_set_param_used): Likewise.
25291 (ipa_get_controlled_uses): Likewise.
25292 (ipa_set_controlled_uses): Likewise.
25293 (ipa_is_param_used): Likewise.
25294 (ipa_node_params_t): Move into garbage collector. New methods insert
25296 (ipa_node_params_sum): Annotate wth GTY(()).
25297 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
25299 (ipa_load_from_parm_agg): Adjust declaration.
25300 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
25301 * ipa-profile.c (ipa_profile): Likewise.
25302 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
25303 (ipa_populate_param_decls): Make descriptors parameter garbage
25305 (ipa_dump_param): Adjust to descriptors being a pointer.
25306 (ipa_alloc_node_params): Likewise.
25307 (ipa_initialize_node_params): Likewise.
25308 (load_from_param_1): Make descriptors parameter garbage collected.
25309 (load_from_unmodified_param): Likewise.
25310 (load_from_param): Likewise.
25311 (ipa_load_from_parm_agg): Likewise.
25312 (ipa_node_params::~ipa_node_params): Removed.
25313 (ipa_free_all_node_params): Remove call to delete operator.
25314 (ipa_node_params_t::insert): New.
25315 (ipa_node_params_t::remove): Likewise.
25316 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
25317 copy known_csts and known_contexts vectors.
25318 (ipa_read_node_info): Adjust to descriptors being a pointer.
25319 (ipcp_modif_dom_walker): Make m_descriptors field garbage
25321 (ipcp_transform_function): Make descriptors variable garbage
25324 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
25326 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
25327 * config/i386/avx512dqintrin.h: Ditto.
25328 * config/i386/avx512fintrin.h: Ditto.
25329 * config/i386/i386.c: Handle new builtins.
25330 * config/i386/i386-builtin.def: Add new builtins.
25331 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
25332 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
25334 2017-01-23 Jakub Jelinek <jakub@redhat.com>
25335 Martin Liska <mliska@suse.cz>
25337 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
25338 * asan.c (asan_expand_poison_ifn): Support stores and use
25339 appropriate ASAN report function.
25340 * internal-fn.c (expand_ASAN_POISON_USE): New function.
25341 * internal-fn.def (ASAN_POISON_USE): Declare.
25342 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
25343 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
25344 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
25345 ASAN_POISON calls w/o LHS.
25346 * tree-ssa.c (execute_update_addresses_taken): Create clobber
25347 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
25348 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
25349 * gimplify.c (asan_poison_variables): Add attribute
25350 use_after_scope_memory to variables that really needs to live
25352 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
25353 having the attribute.
25355 2017-01-23 Martin Liska <mliska@suse.cz>
25357 * asan.c (create_asan_shadow_var): New function.
25358 (asan_expand_poison_ifn): Likewise.
25359 * asan.h (asan_expand_poison_ifn): New declaration.
25360 * internal-fn.c (expand_ASAN_POISON): Likewise.
25361 * internal-fn.def (ASAN_POISON): New builtin.
25362 * sanopt.c (pass_sanopt::execute): Expand
25363 asan_expand_poison_ifn.
25364 * tree-inline.c (copy_decl_for_dup_finish): Make function
25366 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
25367 * tree-ssa.c (is_asan_mark_p): New function.
25368 (execute_update_addresses_taken): Rewrite local variables
25369 (identified just by use-after-scope as addressable) into SSA.
25371 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
25373 * doc/install.texi (Specific): opensource.apple.com uses https
25374 now. Remove trailing slash.
25376 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
25378 * README.Portability: Remove note on an Irix compatibility issue.
25380 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
25382 * gcov.c (INCLUDE_ALGORITHM): Define.
25383 (INCLUDE_VECTOR): Define.
25384 No longer include <vector> and <algorithm> directly.
25386 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
25388 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
25390 * doc/invoke.texi (Code Gen Options): Ditto.
25392 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
25395 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
25397 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
25399 rtl-optimization/79125
25400 * cprop.c (local_cprop_pass): Handle cases where we make an
25401 unconditional trap.
25403 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
25407 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
25408 read from, for big endian.
25410 2017-01-20 Jiong Wang <jiong.wang@arm.com>
25412 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
25413 register pauth builtins for LP64 only.
25415 2017-01-20 Marek Polacek <polacek@redhat.com>
25418 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
25421 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
25423 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
25425 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
25426 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
25427 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
25429 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25432 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
25433 in big-endian mode when they are not a single duplicated value.
25435 2017-01-20 Richard Biener <rguenther@suse.de>
25437 * BASE-VER: Bump to 7.0.1.
25439 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
25441 * omp-low.c (omplow_simd_context): New struct. Use it...
25442 (lower_rec_simd_input_clauses): ...here and...
25443 (lower_rec_input_clauses): ...here to hold common data. Adjust all
25444 references to idx, lane, max_vf, is_simt.
25446 2017-01-20 Graham Markall <graham.markall@embecosm.com>
25448 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
25451 2017-01-20 Martin Jambor <mjambor@suse.cz>
25453 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
25454 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
25456 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
25457 (GTFILES): Rename hsa.c to hsa-common.c.
25458 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
25459 * hsa-dump.c: Likewise.
25460 * hsa-gen.c: Likewise.
25461 * hsa-regalloc.c: Likewise.
25462 * ipa-hsa.c: Likewise.
25463 * omp-expand.c: Likewise.
25464 * omp-low.c: Likewise.
25465 * toplev.c: Likewise.
25467 2017-01-20 Marek Polacek <polacek@redhat.com>
25470 * doc/invoke.texi: Document -Wduplicated-branches.
25471 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
25472 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
25473 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
25474 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
25475 return 0 only when not OEP_LEXICOGRAPHIC.
25476 (fold_build_cleanup_point_expr): Use the expression
25477 location when building CLEANUP_POINT_EXPR.
25478 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
25479 * tree.c (add_expr): Handle error_mark_node.
25481 2017-01-20 Martin Liska <mliska@suse.cz>
25484 * tree-profile.c (init_ic_make_global_vars): Do not call
25486 (gimple_init_gcov_profiler): Likewise.
25488 2017-01-20 Martin Liska <mliska@suse.cz>
25491 * cgraph.h (maybe_create_reference): Remove argument and
25493 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
25495 * ipa-cp.c (create_specialized_node): Likewise.
25496 * symtab.c (symtab_node::maybe_create_reference): Handle
25497 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
25499 2017-01-20 Martin Liska <mliska@suse.cz>
25501 * read-rtl-function.c (function_reader::create_function): Use
25502 build_decl instread of build_decl_stat.
25504 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
25506 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
25507 * config/i386/avx512dqintrin.h: Ditto.
25508 * config/i386/avx512fintrin.h: Ditto.
25509 * config/i386/i386-builtin-types.def: Add new types.
25510 * config/i386/i386.c: Handle new types.
25511 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
25512 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
25513 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
25514 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
25515 (__builtin_ia32_kshiftridi): New.
25516 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
25518 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
25522 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
25523 define to rs6000_init_stack_protect_guard.
25524 (rs6000_init_stack_protect_guard): New function.
25526 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
25527 Yunqiang Su <yunqiang.su@imgtec.com>
25529 * config.gcc (supported_defaults): Add madd4.
25530 (with_madd4): Add validation.
25531 (all_defaults): Add madd4.
25532 * config/mips/mips.opt (mmadd4): New option.
25533 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
25535 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
25537 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
25538 (ISA_HAS_FUSED_MADD4): Likewise.
25539 * doc/invoke.texi (-mmadd4): Document the new option.
25540 * doc/install.texi (--with-madd4): Document the new option.
25542 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25544 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
25545 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
25546 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
25547 (aarch64_init_pauth_hint_builtins): New.
25548 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
25549 (aarch64_expand_builtin): Expand new builtins.
25551 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25553 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
25554 * combine-stack-adj.c (no_unhandled_cfa): Handle
25555 REG_CFA_TOGGLE_RA_MANGLE.
25556 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
25557 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
25558 info for return address signing.
25559 (aarch64_expand_epilogue): Likewise.
25561 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25563 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
25564 * config/aarch64/aarch64-protos.h
25565 (aarch64_return_address_signing_enabled): New declaration.
25566 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
25568 (aarch64_expand_prologue): Sign return address before it's pushed onto
25570 (aarch64_expand_epilogue): Authenticate return address fetched from
25572 (aarch64_override_options): Sanity check for ILP32 and ISA level.
25573 (aarch64_attributes): New function attributes for "sign-return-address".
25574 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
25575 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
25576 ("*do_return"): Generate combined instructions according to key index.
25577 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
25578 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
25580 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
25581 * config/aarch64/aarch64.opt (msign-return-address=): New.
25582 * doc/extend.texi (AArch64 Function Attributes): Documents
25583 "sign-return-address=".
25584 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
25586 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
25588 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
25589 overall option summary.
25591 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25593 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
25594 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
25595 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
25596 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
25598 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
25600 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
25601 -mpower9-minmax by default for -mcpu=power9.
25602 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
25603 128-bit floating point.
25605 2017-01-20 Alan Modra <amodra@gmail.com>
25607 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
25608 optimizing for size.
25610 2017-01-20 Alan Modra <amodra@gmail.com>
25613 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
25614 for strcmp and strncmp from corresponding builtin decl.
25616 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
25618 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
25619 instead of i386/rtems-64.h.
25620 * config/i386/rtems-64.h: Remove.
25622 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
25626 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
25628 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
25630 2017-01-19 Tamar Christina <tamar.christina@arm.com>
25632 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
25633 Change int to HOST_WIDE_INT.
25634 * config/aarch64/aarch64-protos.h
25635 (aarch64_simd_gen_const_vector_dup): Likewise.
25636 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
25638 2017-01-19 David Malcolm <dmalcolm@redhat.com>
25640 * langhooks-def.h (lhd_type_for_size): New decl.
25641 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
25642 * langhooks.c (lhd_type_for_size): New function, taken from
25645 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
25647 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
25648 define_bypass for CR latency.
25649 (power9-cracked-alu): Update bypass latency and remove power9-branch.
25650 (power9-alu2): Add define_bypass for CR latency.
25652 (power9-mul): Update insn latency.
25653 (power9-mul-compare): Update insn latency, bypass latency and remove
25656 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25658 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
25660 * config/aarch64/aarch64.md
25661 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
25662 aarch64_nopcrelative_literal_loads.
25663 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
25665 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
25667 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
25668 TARGET_LOONGSON_3A.
25669 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
25671 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
25674 * config.gcc (supported_defaults): Add lxc1-sxc1.
25675 (with_lxc1_sxc1): Add validation.
25676 (all_defaults): Add lxc1-sxc1.
25677 * config/mips/mips.opt (mlxc1-sxc1): New option.
25678 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
25680 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
25681 __mips_no_lxc1_sxc1.
25682 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
25683 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
25684 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
25686 2017-01-19 Richard Biener <rguenther@suse.de>
25688 PR tree-optimization/72488
25689 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
25690 sure to restore SSA info.
25691 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
25693 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
25695 PR rtl-optimization/79121
25696 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
25697 of the inner type when shifting an extended value.
25699 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
25702 * symtab.c (symtab_node::equal_address_to): Fix comparing of
25703 interposable aliases.
25705 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
25708 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
25709 Use the evmergelohi instruction.
25710 (mov_si<mode>_e500_subreg4_2_le): Likewise.
25711 (mov_sitf_e500_subreg8_2_be): Likewise.
25712 (mov_sitf_e500_subreg12_2_le): Likewise.
25713 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
25714 (mov_si<mode>_e500_subreg4_2_be): Likewise.
25715 (mov_sitf_e500_subreg8_2_le): Likewise.
25716 (mov_sitf_e500_subreg12_2_be): Likewise.
25718 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25720 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
25721 attribute from vecsimple to vecperm.
25722 (altivec_vbpermq2): Likewise.
25724 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25727 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
25729 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25730 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
25731 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
25732 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
25733 case where N arg is SIZE_MAX.
25734 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
25735 (cmpstrsi): Add pattern.
25737 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25739 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25740 __builtin_vec_revb builtins.
25741 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
25742 built-in functions to support generation of the ISA 3.0 XXBR<x>
25743 vector byte reverse instructions.
25744 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
25745 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
25746 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
25747 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
25748 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
25749 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
25750 (P9V_BUILTIN_VEC_REVB): Likewise.
25751 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
25752 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
25753 (p9_xxbrq_v16qi): Likewise.
25754 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
25755 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
25756 (p9_xxbrh_v8hi): Likewise.
25757 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
25758 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
25759 vec_revb built-in functions.
25761 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
25763 PR rtl-optimization/78952
25764 * config/i386/i386.md (any_extract): New code iterator.
25765 (*insvqi_2): Use any_extract for source operand.
25766 (*insvqi_3): Use any_shiftrt for source operand.
25768 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
25770 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
25772 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
25774 2017-01-18 Matthias Klose <doko@ubuntu.com>
25776 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
25778 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25780 * config/rs6000/altivec.h (vec_bperm): Change #define.
25781 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
25782 (altivec_vbpermq2): New define_insn.
25783 (altivec_vbpermd): Likewise.
25784 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
25785 function interface.
25786 (VBPERMD): Likewise.
25787 (VBPERM): New polymorphic function interface.
25788 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
25789 Add entries for P9V_BUILTIN_VEC_VBPERM.
25790 * doc/extend.texi: Add interfaces for vec_bperm.
25792 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25794 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
25795 first letter of error messages.
25796 (s390_resolve_overloaded_builtin): Likewise.
25797 * config/s390/s390.c (s390_expand_builtin): Likewise.
25798 (s390_invalid_arg_for_unprototyped_fn): Likewise.
25799 (s390_valid_target_attribute_inner_p): Likewise.
25800 * config/s390/s390.md ("tabort"): Likewise.
25802 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
25804 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
25805 (ISA_AVOID_DIV_HILO): New macro.
25806 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
25807 (ISA_HAS_DDIV): Likewise.
25809 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25811 * doc/invoke.texi (fabi-version): Correct number of occurrences.
25813 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25815 * doc/invoke.texi (fabi-version): Spelling fix.
25817 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25820 * doc/invoke.texi (fabi-version): Mention mangling fix for
25823 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25826 * doc/invoke.texi (fabi-version): Document discriminator mangling.
25828 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
25831 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
25832 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
25834 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
25836 (stack_protect_test): Ditto.
25837 * config/rs6000/rs6000.opt (mstack-protector-guard=,
25838 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
25840 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
25841 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
25842 -mstack-protector-guard-offset=.
25843 (RS/6000 and PowerPC Options): Ditto.
25845 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
25847 * config/i386/i386.h (MASK_CLASS_P): New define.
25848 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
25849 there are no registers from different register sets also when
25850 mask registers are used. Update function comment.
25851 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
25852 to (*k/*r) and (*k/*km) alternatives.
25854 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
25856 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
25857 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
25858 (EH_RETURN_HANDLER_RTX): New define.
25859 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
25860 Force frame pointer in EH return functions.
25861 (aarch64_expand_epilogue): Add barrier for eh_return.
25862 (aarch64_final_eh_return_addr): Remove.
25863 (aarch64_eh_return_handler_rtx): New function.
25864 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
25866 (aarch64_eh_return_handler_rtx): New prototype.
25868 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25870 * config/rs6000/altivec.h (vec_rlmi): New #define.
25871 (vec_vrlnm): Likewise.
25872 (vec_rlnm): Likewise.
25873 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
25874 (UNSPEC_VRLNM): Likewise.
25875 (VIlong): New mode iterator.
25876 (altivec_vrl<VI_char>mi): New define_insn.
25877 (altivec_vrl<VI_char>nm): Likewise.
25878 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
25880 (VRLDNM): Likewise.
25881 (RLNM): New polymorphic function entry.
25882 (VRLWMI): New monomorphic function entry.
25883 (VRLDMI): Likewise.
25884 (RLMI): New polymorphic function entry.
25885 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
25886 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
25887 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
25890 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25893 * dwarf2out.c (field_byte_offset): Restore the
25894 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
25895 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
25896 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
25899 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
25902 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
25904 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25907 * dwarf2out.c (add_data_member_location_attribute): For constant
25908 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
25909 instead of DW_AT_data_member_location, DW_AT_bit_offset and
25910 DW_AT_byte_size attributes.
25912 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
25914 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
25915 after forcing to constant memory when the code model is medium.
25917 2017-01-17 Julia Koval <julia.koval@intel.com>
25920 * config/i386/avx512fintrin.h
25921 (_mm512_i32gather_ps): Change __addr type to void const*.
25922 (_mm512_mask_i32gather_ps): Ditto.
25923 (_mm512_i32gather_pd): Ditto.
25924 (_mm512_mask_i32gather_pd): Ditto.
25925 (_mm512_i64gather_ps): Ditto.
25926 (_mm512_mask_i64gather_ps): Ditto.
25927 (_mm512_i64gather_pd): Ditto.
25928 (_mm512_mask_i64gather_pd): Ditto.
25929 (_mm512_i32gather_epi32): Ditto.
25930 (_mm512_mask_i32gather_epi32): Ditto.
25931 (_mm512_i32gather_epi64): Ditto.
25932 (_mm512_mask_i32gather_epi64): Ditto.
25933 (_mm512_i64gather_epi32): Ditto.
25934 (_mm512_mask_i64gather_epi32): Ditto.
25935 (_mm512_i64gather_epi64): Ditto.
25936 (_mm512_mask_i64gather_epi64): Ditto.
25937 (_mm512_i32scatter_ps): Change __addr type to void*.
25938 (_mm512_mask_i32scatter_ps): Ditto.
25939 (_mm512_i32scatter_pd): Ditto.
25940 (_mm512_mask_i32scatter_pd): Ditto.
25941 (_mm512_i64scatter_ps): Ditto.
25942 (_mm512_mask_i64scatter_ps): Ditto.
25943 (_mm512_i64scatter_pd): Ditto.
25944 (_mm512_mask_i64scatter_pd): Ditto.
25945 (_mm512_i32scatter_epi32): Ditto.
25946 (_mm512_mask_i32scatter_epi32): Ditto.
25947 (_mm512_i32scatter_epi64): Ditto.
25948 (_mm512_mask_i32scatter_epi64): Ditto.
25949 (_mm512_i64scatter_epi32): Ditto.
25950 (_mm512_mask_i64scatter_epi32): Ditto.
25951 (_mm512_i64scatter_epi64): Ditto.
25952 (_mm512_mask_i64scatter_epi64): Ditto.
25953 * config/i386/avx512pfintrin.h
25954 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
25955 (_mm512_mask_prefetch_i32gather_ps): Ditto.
25956 (_mm512_mask_prefetch_i64gather_pd): Ditto.
25957 (_mm512_mask_prefetch_i64gather_ps): Ditto.
25958 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
25959 (_mm512_prefetch_i32scatter_ps): Ditto.
25960 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
25961 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
25962 (_mm512_prefetch_i64scatter_pd): Ditto.
25963 (_mm512_prefetch_i64scatter_ps): Ditto.
25964 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
25965 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
25966 * config/i386/avx512vlintrin.h
25967 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
25968 (_mm_mmask_i32gather_ps): Ditto.
25969 (_mm256_mmask_i32gather_pd): Ditto.
25970 (_mm_mmask_i32gather_pd): Ditto.
25971 (_mm256_mmask_i64gather_ps): Ditto.
25972 (_mm_mmask_i64gather_ps): Ditto.
25973 (_mm256_mmask_i64gather_pd): Ditto.
25974 (_mm_mmask_i64gather_pd): Ditto.
25975 (_mm256_mmask_i32gather_epi32): Ditto.
25976 (_mm_mmask_i32gather_epi32): Ditto.
25977 (_mm256_mmask_i32gather_epi64): Ditto.
25978 (_mm_mmask_i32gather_epi64): Ditto.
25979 (_mm256_mmask_i64gather_epi32): Ditto.
25980 (_mm_mmask_i64gather_epi32): Ditto.
25981 (_mm256_mmask_i64gather_epi64): Ditto.
25982 (_mm_mmask_i64gather_epi64): Ditto.
25983 (_mm256_i32scatter_ps): Change __addr type to void*.
25984 (_mm256_mask_i32scatter_ps): Ditto.
25985 (_mm_i32scatter_ps): Ditto.
25986 (_mm_mask_i32scatter_ps): Ditto.
25987 (_mm256_i32scatter_pd): Ditto.
25988 (_mm256_mask_i32scatter_pd): Ditto.
25989 (_mm_i32scatter_pd): Ditto.
25990 (_mm_mask_i32scatter_pd): Ditto.
25991 (_mm256_i64scatter_ps): Ditto.
25992 (_mm256_mask_i64scatter_ps): Ditto.
25993 (_mm_i64scatter_ps): Ditto.
25994 (_mm_mask_i64scatter_ps): Ditto.
25995 (_mm256_i64scatter_pd): Ditto.
25996 (_mm256_mask_i64scatter_pd): Ditto.
25997 (_mm_i64scatter_pd): Ditto.
25998 (_mm_mask_i64scatter_pd): Ditto.
25999 (_mm256_i32scatter_epi32): Ditto.
26000 (_mm256_mask_i32scatter_epi32): Ditto.
26001 (_mm_i32scatter_epi32): Ditto.
26002 (_mm_mask_i32scatter_epi32): Ditto.
26003 (_mm256_i32scatter_epi64): Ditto.
26004 (_mm256_mask_i32scatter_epi64): Ditto.
26005 (_mm_i32scatter_epi64): Ditto.
26006 (_mm_mask_i32scatter_epi64): Ditto.
26007 (_mm256_i64scatter_epi32): Ditto.
26008 (_mm256_mask_i64scatter_epi32): Ditto.
26009 (_mm_i64scatter_epi32): Ditto.
26010 (_mm_mask_i64scatter_epi32): Ditto.
26011 (_mm256_i64scatter_epi64): Ditto.
26012 (_mm256_mask_i64scatter_epi64): Ditto.
26013 (_mm_i64scatter_epi64): Ditto.
26014 (_mm_mask_i64scatter_epi64): Ditto.
26015 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
26016 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
26017 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
26018 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
26019 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
26020 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
26021 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
26022 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
26023 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
26024 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
26025 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
26026 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
26027 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
26028 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
26029 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
26030 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
26031 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
26032 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
26033 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
26034 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
26035 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
26036 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
26037 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
26038 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
26039 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
26040 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
26041 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
26042 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
26043 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
26044 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
26045 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
26046 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
26047 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
26048 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
26049 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
26050 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
26051 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
26052 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
26053 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
26054 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
26055 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
26056 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
26057 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
26058 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
26059 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
26060 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
26061 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
26062 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
26063 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
26064 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
26065 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
26066 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
26067 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
26068 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
26069 definitions accordingly.
26071 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
26072 Kuan-Lin Chen <kuanlinchentw@gmail.com>
26075 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
26078 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
26081 * ira-conflicts.c (ira_build_conflicts): Update total conflict
26082 hard regs for inner regno.
26084 2017-01-17 Martin Liska <mliska@suse.cz>
26087 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
26088 assumption and add comment.
26090 2017-01-17 Nathan Sidwell <nathan@acm.org>
26092 * ipa-visibility.c (localize_node): New function, broken out of ...
26093 (function_and_variable_visibility): ... here. Call it.
26095 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
26097 PR middle-end/77445
26098 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
26099 correctly set frequency of oudgoing edge.
26100 (duplicate_thread_path): Fix profile updating.
26102 2017-01-17 Jakub Jelinek <jakub@redhat.com>
26105 * configure.ac: Add GCC_BASE_VER.
26106 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
26107 version from BASE-VER file.
26108 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
26109 (gcc.o): Depend on $(BASEVER).
26110 * common.opt (dumpfullversion): New option.
26111 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
26112 * doc/invoke.texi: Document -dumpfullversion.
26113 * doc/install.texi: Document --with-gcc-major-version-only.
26114 * configure: Regenerated.
26116 2017-01-17 Richard Biener <rguenther@suse.de>
26118 PR tree-optimization/71433
26119 * tree-vrp.c (register_new_assert_for): Merge same asserts
26120 on all incoming edges.
26121 (process_assert_insertions_for): Handle insertions at the
26124 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
26126 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
26127 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
26129 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
26132 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
26135 2017-01-17 Alan Modra <amodra@gmail.com>
26138 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
26139 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
26140 symbolic stack limit when pic.
26142 2017-01-16 Martin Sebor <msebor@redhat.com>
26144 PR tree-optimization/78608
26145 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
26147 2017-01-16 Jeff Law <law@redhat.com>
26150 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
26151 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
26152 for several include directories that may be relative to sysroot.
26153 * config/i386/x-mingw32 (gplus_includedir): Define.
26154 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
26155 (native_system_includedir): Likewise.
26156 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
26157 override if TARGET_SYSTEM_ROOT is defined.
26158 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
26160 PR tree-optimization/79090
26161 PR tree-optimization/33562
26162 PR tree-optimization/61912
26163 PR tree-optimization/77485
26164 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
26165 and computed trims into the dump file.
26167 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
26169 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
26171 2017-01-16 Jakub Jelinek <jakub@redhat.com>
26174 * gimplify.c (gimplify_init_constructor): If want_value and
26175 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
26179 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
26180 sequence. Formatting fixes.
26181 (doloop_optimize): Formatting fixes.
26184 * gcc.c (debug_level_greater_than_spec_func): New function.
26185 (static_spec_functions): Add debug-level-gt spec function.
26186 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
26188 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
26189 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
26190 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
26191 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
26192 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
26193 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
26195 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
26197 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
26198 QImode fixups to general and mask registers only.
26200 2017-01-16 Carl Love <cel@us.ibm.com>
26202 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
26203 for built-in functions
26204 vector signed char vec_nabs (vector signed char)
26205 vector signed short vec_nabs (vector signed short)
26206 vector signed int vec_nabs (vector signed int)
26207 vector signed long long vec_nabs (vector signed long long)
26208 vector float vec_nabs (vector float)
26209 vector double vec_nabs (vector double)
26210 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
26212 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
26213 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
26214 * doc/extend.texi: Update the documentation file for the new built-in
26217 2017-01-16 Martin Sebor <msebor@redhat.com>
26219 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
26222 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26224 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
26225 UNSPEC_VSX__XXSPLTD to require special splat handling.
26227 2017-01-16 David Malcolm <dmalcolm@redhat.com>
26230 * system.h: Poison strndup.
26232 2017-01-16 Alan Modra <amodra@gmail.com>
26235 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
26238 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
26240 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
26242 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
26244 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
26245 call recog here. Assert that INSN_CODE (insn) is non-negative.
26247 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
26250 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
26252 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
26253 in the currently scheduled RTL fragment.
26255 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
26257 PR rtl-optimization/78751
26258 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
26261 2017-01-14 Jeff Law <law@redhat.com>
26263 PR tree-optimization/79090
26264 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
26265 variable length stores.
26266 (compute_trims): Delete dead assignment to *trim_tail.
26267 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
26270 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
26272 PR rtl-optimization/78626
26273 PR rtl-optimization/78727
26274 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
26275 of a block, and split such blocks after everything else is finished.
26277 2017-01-14 Alan Modra <amodra@gmail.com>
26280 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
26281 target legitimate_combined_insn.
26282 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
26283 (rs6000_legitimate_combined_insn): New function.
26284 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
26286 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
26287 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
26288 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
26290 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
26292 * doc/frontends.texi (G++ and GCC): Remove references to Java.
26294 2017-01-13 Jeff Law <law@redhat.com>
26296 PR tree-optimization/33562
26297 PR tree-optimization/61912
26298 PR tree-optimization/77485
26299 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
26301 (delete_dead_assignment): Likewise.
26302 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
26303 statement to delete_dead_call and delete_dead_assignment.
26305 2017-01-13 David Malcolm <dmalcolm@redhat.com>
26308 * substring-locations.c (format_warning_va): Strengthen case 1 so
26309 that both endpoints of the substring must be within the format
26310 range for just the substring to be printed.
26312 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
26314 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
26315 * config/i386/i386.c (ix86_target_string): Add missing options
26316 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
26317 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
26318 flags_other and ix86_target_other to flags2_other. Display unknown
26320 (ix86_valid_target_attribute_inner_p): Add missing options and
26321 reorder options by implied ISAs, as in ix86_target_string.
26323 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
26325 * hash-table.h (hash_table::too_empty_p): New function.
26326 (hash_table::expand): Use it.
26327 (hash_table::traverse): Likewise.
26328 (hash_table::empty_slot): Use sizeof (value_type) instead of
26329 sizeof (PTR) to convert bytes to elements. Shrink the table
26330 if the current size is excessive for the current number of
26333 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
26335 * ira-costs.c (record_reg_classes): Break from the inner loop
26336 early once alt_fail is known to be true. Update outer loop
26337 handling accordingly.
26339 2017-01-13 Jeff Law <law@redhat.com>
26341 * tree-ssa-dse.c (decrement_count): New function.
26342 (increment_start_addr, maybe_trim_memstar_call): Likewise.
26343 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
26344 when we know the partially dead statement is a mem* function.
26346 PR tree-optimization/61912
26347 PR tree-optimization/77485
26348 * tree-ssa-dse.c: Include expr.h.
26349 (maybe_trim_constructor_store): New function.
26350 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
26352 PR tree-optimization/33562
26353 PR tree-optimization/61912
26354 PR tree-optimization/77485
26355 * doc/invoke.texi: Document new dse-max-object-size param.
26356 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
26357 * tree-ssa-dse.c: Include params.h.
26358 (dse_store_status): New enum.
26359 (initialize_ao_ref_for_dse): New, partially extracted from
26361 (valid_ao_ref_for_dse, normalize_ref): New.
26362 (setup_live_bytes_from_ref, compute_trims): Likewise.
26363 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
26364 (maybe_trim_partially_dead_store): Likewise.
26365 (maybe_trim_complex_store): Likewise.
26366 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
26367 Track what bytes live from the original store. Return tri-state
26368 for dead, partially dead or live.
26369 (dse_dom_walker): Add constructor, destructor and new private members.
26370 (delete_dead_call, delete_dead_assignment): New extracted from
26372 (dse_optimize_stmt): Make a member of dse_dom_walker.
26373 Use initialize_ao_ref_for_dse.
26375 PR tree-optimization/33562
26376 PR tree-optimization/61912
26377 PR tree-optimization/77485
26378 * sbitmap.h (bitmap_count_bits): Prototype.
26379 (bitmap_clear_range, bitmap_set_range): Likewise.
26380 * sbitmap.c (bitmap_clear_range): New function.
26381 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
26383 2017-01-13 Martin Liska <mliska@suse.cz>
26386 * function.c (set_cfun): Add new argument force.
26387 * function.h (set_cfun): Likewise.
26388 * ipa-inline-transform.c (inline_call): Use the function when
26389 strict alising from is dropped for function we inline to.
26391 2017-01-13 Richard Biener <rguenther@suse.de>
26393 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
26394 for dumping GIMPLE INTEGER_CSTs.
26396 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26398 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
26399 to 201112L since C++17.
26401 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
26404 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
26405 if -fsanitize=kernel-address is present.
26407 2017-01-13 Richard Biener <rguenther@suse.de>
26409 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
26410 as _Literal ( type ) number in case usual suffixes do not
26411 preserve all information.
26413 2017-01-13 Richard Biener <rguenther@suse.de>
26415 PR tree-optimization/77283
26416 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
26417 and ssa-iterators.h.
26418 (is_feasible_trace): Implement a cost model based on joiner
26421 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
26424 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
26425 char or short to __float128/_Float128 directly.
26427 2017-01-12 Martin Sebor <msebor@redhat.com>
26429 to -Wformat-overflow.
26430 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
26431 (min_bytes_remaining): Same.
26432 (get_string_length): Same.
26433 (format_string): Same.
26434 (format_directive): Same.
26436 (pass_sprintf_length::handle_gimple_call): Same.
26438 2017-01-12 Jakub Jelinek <jakub@redhat.com>
26440 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
26441 info.nowrite calls with no lhs that can't throw. Return bool
26442 whether gsi_remove has been called or not.
26443 (pass_sprintf_length::handle_gimple_call): Return bool whether
26444 try_substitute_return_value called gsi_remove. Formatting fix.
26445 (pass_sprintf_length::execute): Don't use gsi_remove if
26446 handle_gimple_call returned true.
26449 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
26450 be removed due to side-effects, don't remove following barrier nor
26451 turn the successor edge into fallthru edge.
26453 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26456 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
26457 element-reversing loads and stores as not swappable.
26459 2017-01-12 Nathan Sidwell <nathan@acm.org>
26460 Nicolai Stange <nicstange@gmail.com>
26462 * combine.c (try_combine): Don't ignore result of overlap checking
26463 loop. Combine overlap & asm check into single loop.
26465 2017-01-12 Richard Biener <rguenther@suse.de>
26467 * tree-pretty-print.c (dump_generic_node): Provide -gimple
26468 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
26470 2017-01-12 Richard Biener <rguenther@suse.de>
26472 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
26473 and TS_TARGET_OPTION directly derive from TS_BASE.
26474 * tree-core.h (tree_optimization_option): Derive from tree_base.
26475 (tree_target_option): Likewise.
26477 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
26479 * config/i386/i386.c (memory_address_length): Increase len
26480 only when rip_relative_addr_p returns false.
26482 2017-01-11 Julia Koval <julia.koval@intel.com>
26484 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
26485 (OPTION_MASK_ISA_SGX_SET): New.
26486 (ix86_handle_option): Handle OPT_msgx.
26487 * config.gcc: Added sgxintrin.h.
26488 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
26489 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
26490 * config/i386/i386.c (ix86_target_string): Add -msgx.
26492 (ix86_option_override_internal): Handle new options.
26493 (ix86_valid_target_attribute_inner_p): Add sgx.
26494 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
26495 * config/i386/i386.opt: Add msgx.
26496 * config/i386/sgxintrin.h: New file.
26497 * config/i386/x86intrin.h: Add sgxintrin.h.
26499 2017-01-11 Jakub Jelinek <jakub@redhat.com>
26502 * fold-const.c (maybe_nonzero_address): Return 1 for function
26504 (tree_single_nonzero_warnv_p): Don't handle function local objects
26508 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
26511 2017-01-11 David Malcolm <dmalcolm@redhat.com>
26514 * opts.c: Include "spellcheck.h"
26515 (struct string_fragment): New struct.
26516 (struct edit_distance_traits<const string_fragment &>): New
26518 (get_closest_sanitizer_option): New function.
26519 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
26521 2017-01-11 Jakub Jelinek <jakub@redhat.com>
26523 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
26525 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
26526 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
26527 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
26528 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
26529 for initial die_offset if dwarf_split_debug_info.
26530 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
26531 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
26533 (output_skeleton_debug_sections): Formatting fix. Use
26534 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
26535 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
26537 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
26539 * config/arm/cortex-a53.md: Add bypasses for
26540 cortex_a53_r2f_cvt.
26541 (cortex_a53_r2f): Only use for transfers.
26542 (cortex_a53_f2r): Likewise.
26543 (cortex_a53_r2f_cvt): Add reservation for conversions.
26544 (cortex_a53_f2r_cvt): Likewise.
26546 2017-01-11 Tamar Christina <tamar.christina@arm.com>
26548 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
26549 to all inlined functions, change static to extern.
26551 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
26554 * config/arm/arm.c (legitimize_pic_address): Handle reference to
26556 (arm_assemble_integer): Likewise.
26558 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
26560 * config.gcc: Use new awk script to check CPU, FPU and architecture
26561 parameters for --with-... options.
26562 * config/arm/parsecpu.awk: New file
26563 * config/arm/arm-cpus.in: New file.
26564 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
26566 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
26568 * config/arm/t-arm: Update dependency rules.
26569 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
26570 of processing .def files.
26571 * config/arm/genopt.sh: Deleted.
26572 * config/arm/gentune.sh: Deleted.
26573 * config/arm/arm-cores.def: Deleted.
26574 * config/arm/arm-arches.def: Deleted.
26575 * config/arm/arm-fpus.def: Deleted.
26576 * config/arm/arm-tune.md: Regenerated.
26577 * config/arm/arm-tables.opt: Regenerated.
26578 * config/arm/arm-cpu.h: New generated file.
26579 * config/arm/arm-cpu-data.h: New generated file.
26580 * config/arm/arm-cpu-cdata.h: New generated file.
26582 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
26585 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
26587 (input_varpool_node): Unpack dynamically_initialized bit.
26589 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
26591 PR rtl-optimization/79032
26592 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
26593 the alignment of the adjusted memory reference against that of MODE,
26594 instead of the alignment of the original memory reference.
26596 2017-01-11 Martin Jambor <mjambor@suse.cz>
26598 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
26600 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
26601 decorated functions.
26603 2017-01-11 Richard Biener <rguenther@suse.de>
26605 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
26606 set range/nonnull info for PHI results. Do not set it on
26607 stmts marked for removal.
26609 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
26611 * expr.c (store_field): In the bitfield case, fetch the return value
26612 from the registers before applying a single big-endian adjustment.
26613 Always do a final load for a BLKmode value not larger than a word.
26615 2017-01-10 David Malcolm <dmalcolm@redhat.com>
26618 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
26619 that we correctly handle column numbers greater than
26620 LINE_MAP_MAX_COLUMN_NUMBER.
26622 2017-01-10 Martin Sebor <msebor@redhat.com>
26624 PR middle-end/78245
26625 * gimple-ssa-sprintf.c (get_destination_size): Call
26626 {init,fini}object_sizes.
26627 * tree-object-size.c (addr_object_size): Adjust.
26628 (pass_through_call): Adjust.
26629 (pass_object_sizes::execute): Adjust.
26630 * tree-object-size.h (fini_object_sizes): Declare.
26632 2017-01-10 Martin Sebor <msebor@redhat.com>
26634 PR tree-optimization/78775
26635 * builtins.c (get_size_range): Move...
26636 * calls.c: ...to here.
26637 (alloc_max_size): Accept zero argument.
26638 (operand_signed_p): Remove.
26639 (maybe_warn_alloc_args_overflow): Call get_size_range.
26640 * calls.h (get_size_range): Declare.
26642 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
26644 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
26645 from TI's devices.csv file as of September 2016.
26646 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
26648 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
26650 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
26651 * doc/invoke.texi: Likewise.
26652 * doc/md.texi: Likewise.
26653 * doc/objc.texi: Likewise.
26655 2017-01-10 Joshua Conner <joshconner@google.com>
26657 * config/arm/fuchsia-elf.h: New file.
26658 * config/fuchsia.h: New file.
26659 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
26660 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
26662 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
26664 2016-01-10 Richard Biener <rguenther@suse.de>
26666 PR tree-optimization/79034
26667 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
26668 Propagate out degenerate PHIs in the joiner.
26670 2017-01-10 Martin Liska <mliska@suse.cz>
26672 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
26673 (sort_congruence_classes_by_decl_uid): Likewise.
26674 (sort_congruence_class_groups_by_decl_uid): Likewise.
26675 (sem_item_optimizer::merge_classes): Sort class, groups in these
26676 classes and members in the groups by DECL_UID of declarations.
26677 This would make merge operations stable.
26679 2017-01-10 Martin Liska <mliska@suse.cz>
26681 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
26682 usage of m_classes_vec.
26683 (sem_item_optimizer::~sem_item_optimizer): Likewise.
26684 (sem_item_optimizer::get_group_by_hash): Likewise.
26685 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
26686 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
26687 (sem_item_optimizer::verify_classes): Likewise.
26688 (sem_item_optimizer::process_cong_reduction): Likewise.
26689 (sem_item_optimizer::dump_cong_classes): Likewise.
26690 (sem_item_optimizer::merge_classes): Likewise.
26691 * ipa-icf.h (congruence_class_hash): Rename from
26692 congruence_class_group_hash. Remove declaration of m_classes_vec.
26694 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
26696 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
26697 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
26698 * config.gcc: Add avx512vpopcntdqintrin.h.
26699 * config/i386/avx512vpopcntdqintrin.h: New.
26700 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
26701 * config/i386/i386-builtin-types.def: Add new types.
26702 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
26703 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
26704 __builtin_ia32_vpopcountq_v8di_mask): New.
26705 * config/i386/i386-c.c (ix86_target_macros_internal): Define
26706 __AVX512VPOPCNTDQ__.
26707 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
26708 (PTA_AVX512VPOPCNTDQ): Define.
26709 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
26710 TARGET_AVX512VPOPCNTDQ_P): Define.
26711 * config/i386/i386.opt: Add mavx512vpopcntdq.
26712 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
26713 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
26715 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26717 PR middle-end/77484
26718 * predict.def (PRED_CALL): Set to 67.
26720 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
26722 * expr.c (store_field): In the bitfield case, if the value comes from
26723 a function call and is of an aggregate type returned in registers, do
26724 not modify the field mode; extract the value in all cases if the mode
26725 is BLKmode and the size is not larger than a word.
26727 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
26730 * config/i386/cpuid.h: Fix undefined behavior.
26732 2017-01-04 Jeff Law <law@redhat.com>
26734 PR tree-optimization/79007
26735 PR tree-optimization/67955
26736 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
26737 conservative for pt.null when flag_non_call_exceptions is on.
26739 2017-01-09 Jakub Jelinek <jakub@redhat.com>
26741 PR translation/79019
26742 PR translation/79020
26743 * params.def (PARAM_INLINE_MIN_SPEEDUP,
26744 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
26745 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
26747 * config/avr/avr.opt (maccumulate-args): Likewise.
26748 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
26749 * common.opt (freport-bug): Likewise.
26750 * cif-code.def (CIF_FINAL_ERROR): Likewise.
26751 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
26752 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
26753 translatable string.
26754 * config/i386/i386.c (function_value_32): Likewise.
26755 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
26756 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
26758 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
26759 * common/config/msp430/msp430-common.c (msp430_handle_option):
26761 * symtab.c (symtab_node::verify_base): Likewise.
26762 * opts.c (set_debug_level): Likewise.
26763 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
26764 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
26765 missing whitespace to translatable strings.
26766 * config/avr/avr.md (bswapsi2): Fix typo in comment.
26767 * config/sh/superh.h: Likewise.
26768 * config/i386/xopintrin.h: Likewise.
26769 * config/i386/znver1.md: Likewise.
26770 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
26771 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
26772 * double-int.h (struct double_int): Likewise.
26773 * double-int.c (div_and_round_double): Likewise.
26774 * wide-int.cc: Likewise.
26775 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
26776 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
26777 * cfgcleanup.c (crossjumps_occured): Renamed to ...
26778 (crossjumps_occurred): ... this.
26779 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
26782 PR tree-optimization/78899
26783 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
26784 returning bool return struct loop *, NULL for failure and the new
26786 (versionable_outer_loop_p): Don't version outer loop if it has
26787 dont_vectorized bit set.
26788 (tree_if_conversion): When versioning outer loop, ensure
26789 tree_if_conversion is performed also on the inner loop of the
26790 non-vectorizable outer loop copy.
26791 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
26792 LOOP_VECTORIZED in inner loop of the scalar outer loop and
26793 prevent vectorization of it.
26794 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
26795 the outer loop vectorization of the non-scalar version is attempted
26796 before vectorization of the inner loop in scalar version. If
26797 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
26798 vectorization of its inner loop.
26799 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
26800 has 2 inner loops, rename also on edges from bb whose single pred
26801 is outer_loop->header. Fix typo in function comment.
26803 2017-01-09 Martin Sebor <msebor@redhat.com>
26806 * asan.c (asan_emit_stack_protection): Increase local buffer size
26807 to avoid snprintf truncation warning.
26809 2017-01-09 Andrew Pinski <apinski@cavium.com>
26811 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
26812 to reference thunderx2t99 for the tuning structure
26813 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
26815 (thunderx2t99_extra_costs): This.
26816 * config/aarch64/aarch64-tune.md: Regenerate.
26817 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
26818 (vulcan_addrcost_table): This.
26819 (vulcan_regmove_cost): Rename to ...
26820 (thunderx2t99_regmove_cost): This.
26821 (vulcan_vector_cost): Rename to ...
26822 (thunderx2t99_vector_cost): this.
26823 (vulcan_branch_cost): Rename to ...
26824 (thunderx2t99_branch_cost): This.
26825 (vulcan_tunings): Rename to ...
26826 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
26827 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
26829 2017-01-09 Martin Jambor <mjambor@suse.cz>
26833 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
26834 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
26835 (propagate_vr_accross_jump_function): Use the above function for all
26836 value range computations for pass-through jump functions and type
26837 converasion from explicit value range values.
26838 (ipcp_propagate_stage): Do not attempt to deduce types of formal
26839 parameters from TYPE_ARG_TYPES.
26840 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
26841 (ipa_write_node_info): Stream type of the actual argument.
26842 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
26844 2017-01-09 Martin Liska <mliska@suse.cz>
26847 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
26848 (lookup_compiler): Do not show error message with have_E.
26850 2017-01-09 Jakub Jelinek <jakub@redhat.com>
26852 PR tree-optimization/78938
26853 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
26854 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
26855 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
26856 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
26859 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26861 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
26864 2017-01-09 Richard Biener <rguenther@suse.de>
26866 PR tree-optimization/78997
26867 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
26868 name condition properly.
26870 2017-01-09 Richard Biener <rguenther@suse.de>
26873 * dwarf2out.c (is_cxx): New overload with context.
26874 (is_naming_typedef_decl): Use it.
26876 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26878 * invoke.texi (Option Summary): Correct spacing in option lists
26879 and add line breaks to fix over-long lines.
26881 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26883 PR middle-end/17660
26885 * extend.texi (Common Variable Attributes): Add xref to GCC
26886 Internals manual to explain mode attribute keywords.
26888 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26891 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
26892 and Preprocessor Options.
26893 (Options for Linking): Document -pthread here....
26894 (RS/6000 and PowerPC Options): ...not here.
26895 (Solaris 2 Options): ...or here.
26896 * doc/cppopts.texi: Document -pthread.
26898 2017-01-08 Martin Sebor <msebor@redhat.com>
26900 PR middle-end/77708
26901 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
26902 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
26903 New member functions.
26904 (format_directive): Used them.
26906 (pass_sprintf_length::handle_gimple_call): Same.
26907 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
26908 to avoid truncation for any argument.
26909 (extract_affine_mul): Same.
26910 * tree.c (get_file_function_name): Same.
26912 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26914 PR middle-end/77484
26915 * predict.def (PRED_INDIR_CALL): Set to 86.
26917 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26919 PR preprocessor/54124
26920 * doc/cppopts.texi: Reformat -d subtable to list the full name
26921 of the options. Add cross-reference to the docs for the general
26922 compiler -d options.
26923 * doc/invoke.texi (Developer Options): Add cross-reference to the
26924 preprocessor-specific -d option documentation.
26926 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26928 PR preprocessor/13498
26929 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
26930 redudant material, and reflect new command-line options.
26931 (System Headers): Likewise.
26933 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26935 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
26936 -isystem, and -idirafter. Copy-edit.
26937 * doc/cppopts.texi: Copy-edit. Remove contradiction about
26938 default for -ftrack-macro-expansion. Delete obsolete and
26939 badly-formatted implementation details about -fdebug-cpp output.
26940 * doc/cppwarnopts.texi: Copy-edit.
26942 2017-01-07 David Malcolm <dmalcolm@redhat.com>
26945 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
26946 that the transition from a max line width >= 1<<10 to narrower
26947 lines works correctly.
26949 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
26951 * doc/options.texi (PerFunction): New.
26952 * opt-functions.awk (switch_flags): Map both Optimization and
26953 PerFunction to CL_OPTIMIZATION.
26954 * opth-gen.awk: Test for PerFunction flag along with
26956 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
26957 it only when the latter is present. Skip those that don't in
26958 the hash function generator.
26959 * common.opt (fvar-tracking): Mark as PerFunction instead of
26961 (fvar-tracking-assignments): Likewise.
26962 (fvar-tracking-assignments-toggle): Likewise.
26963 (fvar-tracking-uninit): Likewise.
26965 2017-01-07 Jakub Jelinek <jakub@redhat.com>
26967 PR translation/79018
26968 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
26971 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
26974 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
26975 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
26976 TARGET_LONG_JUMP_TABLE_OFFSETS.
26977 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
26978 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
26979 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
26980 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
26981 * config/m68k/m68k.md (tablejump expander): Likewise.
26982 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
26983 TARGET_LONG_JUMP_TABLE_OFFSETS.
26984 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
26985 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
26987 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
26988 David Holsgrove <david.holsgrove@xilinx.com>
26990 * common/config/microblaze/microblaze-common.c
26991 (TARGET_EXCEPT_UNWIND_INFO): Remove.
26992 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
26994 * config/microblaze/microblaze.c (microblaze_must_save_register)
26995 (microblaze_expand_epilogue, microblaze_return_addr): Handle
26997 (microblaze_eh_return): New function.
26998 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
26999 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
27000 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
27001 * config/microblaze/microblaze.md (eh_return): New pattern.
27003 2017-01-06 Jakub Jelinek <jakub@redhat.com>
27005 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
27006 GCC_DIAGNOSTIC_STRINGIFY): Define.
27008 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
27010 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27012 * config/arm/arm.md (<mcrr>): New.
27014 * config/arm/arm.c (arm_arch5te): New.
27015 (arm_option_override): Set arm_arch5te.
27016 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
27018 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
27019 (arm_mcrr_qualifiers): ... this. New.
27020 (MRRC_QUALIFIERS): Define to...
27021 (arm_mrrc_qualifiers): ... this. New.
27022 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
27024 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
27025 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
27026 (MRRCI, mrrc, MRRC): New.
27027 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
27028 VUNSPEC_MRRC2): New.
27030 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27032 * config/arm/arm.md (<mcr>): New.
27034 * config/arm/arm.c (arm_coproc_builtin_available): Add
27035 support for mcr, mrc, mcr2 and mrc2.
27036 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
27037 (arm_mcr_qualifiers): ... this. New.
27038 (MRC_QUALIFIERS): Define to ...
27039 (arm_mrc_qualifiers): ... this. New.
27040 (MCR_QUALIFIERS): Define to ...
27041 (arm_mcr_qualifiers): ... this. New.
27042 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
27044 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
27045 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
27046 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
27047 VUNSPEC_MRC2): New.
27049 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27051 * config/arm/arm.md (*ldc): New.
27055 * config/arm/arm.c (arm_coproc_builtin_available): Add
27056 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
27057 (arm_coproc_ldc_stc_legitimate_address): New.
27058 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
27059 'qualifier_const_pointer'.
27060 (LDC_QUALIFIERS): Define to...
27061 (arm_ldc_qualifiers): ... this. New.
27062 (STC_QUALIFIERS): Define to...
27063 (arm_stc_qualifiers): ... this. New.
27064 * config/arm/arm-protos.h
27065 (arm_coproc_ldc_stc_legitimate_address): New.
27066 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
27067 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
27068 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
27069 stc2, stcl, stc2l): New.
27070 * config/arm/constraints.md (Uz): New.
27071 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
27072 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
27073 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
27074 VUNSPEC_STC2L): New.
27076 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27078 * config/arm/arm.md (<cdp>): New.
27079 * config/arm/arm.c (neon_const_bounds): Rename this ...
27080 (arm_const_bounds): ... this.
27081 (arm_coproc_builtin_available): New.
27082 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
27083 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
27084 (CDP_QUALIFIERS): Define to...
27085 (arm_cdp_qualifiers): ... this. New.
27087 (arm_expand_builtin_args): Add case for 6 arguments.
27088 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
27089 (arm_const_bounds): ... this.
27090 (arm_coproc_builtin_available): New.
27091 * config/arm/arm_acle.h (__arm_cdp): New.
27093 * config/arm/arm_acle_builtins.def (cdp): New.
27095 * config/arm/iterators.md (CDPI,CDP,cdp): New.
27096 * config/arm/neon.md: Rename all 'neon_const_bounds' to
27097 'arm_const_bounds'.
27098 * config/arm/types.md (coproc): New.
27099 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
27100 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
27101 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
27102 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
27104 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27106 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
27107 (UBINOP_QUALIFIERS): New.
27109 (acle_builtin_data): New. Change comment.
27110 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
27111 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
27112 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
27113 arm_acle_builtins.def.
27114 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
27115 (arm_init_acle_builtins): New.
27116 (CRC32_BUILTIN): Remove.
27117 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
27118 crc32cb, crc32ch and crc32cw.
27119 (arm_init_crc32_builtins): Remove.
27120 (arm_init_builtins): Use arm_init_acle_builtins rather
27121 than arm_init_crc32_builtins.
27122 (arm_expand_acle_builtin): New.
27123 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
27124 * config/arm/arm_acle_builtins.def: New.
27126 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27128 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
27129 (arm_builtin_datum): ... this.
27130 (arm_init_neon_builtin): Rename to ...
27131 (arm_init_builtin): ... this. Add a new parameters PREFIX
27132 and USE_SIG_IN_NAME.
27133 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
27134 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
27135 'arm_builtin_datum'.
27136 (arm_init_vfp_builtins): Likewise.
27137 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
27138 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
27139 (arm_expand_neon_args): Rename to ...
27140 (arm_expand_builtin_args): ... this. Rename builtin_arg
27141 enum values and differentiate between ARG_BUILTIN_MEMORY
27142 and ARG_BUILTIN_NEON_MEMORY.
27143 (arm_expand_neon_builtin_1): Rename to ...
27144 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
27145 values, arm_expand_builtin_args and add bool parameter NEON.
27146 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
27147 (arm_expand_vfp_builtin): Likewise.
27148 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
27150 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27152 PR middle-end/77484
27153 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
27154 * predict.c (tree_estimate_probability_bb): Reverse direction of
27155 polymorphic call predictor.
27157 2017-01-06 David Malcolm <dmalcolm@redhat.com>
27159 * passes.c (execute_one_pass): Split out pass-skipping logic into...
27160 (determine_pass_name_match): ...this new function and...
27161 (should_skip_pass_p): ...this new function.
27163 2017-01-06 Nathan Sidwell <nathan@acm.org>
27165 * ipa-visibility.c (function_and_variable_visibility): Reformat
27166 comments and long lines. Remove extrneous if.
27167 * symtab.c (symtab_node::make_decl_local): Fix code format.
27168 (symtab_node::set_section_for_node): Fix comment typo.
27170 2017-01-06 Martin Liska <mliska@suse.cz>
27173 * lra-constraints.c: Rename invariant to lra_invariant.
27174 * predict.c (set_even_probabilities): Initialize e to NULL.
27176 2017-01-05 Martin Sebor <msebor@redhat.com>
27178 PR tree-optimization/78910
27179 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
27180 (format_integer): Correct off-by-one error in the handling
27181 of precision with negative numbers in signed conversions..
27183 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
27185 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
27187 2017-01-05 Jakub Jelinek <jakub@redhat.com>
27189 PR tree-optimization/71016
27190 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
27191 factor_out_conditional_conversion. Formatting fix.
27192 (factor_out_conditional_conversion): Add cond_stmt argument.
27193 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
27194 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
27197 2017-01-05 David Malcolm <dmalcolm@redhat.com>
27199 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
27200 read-rtl-function.o, and selftest-rtl.o.
27201 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
27202 (selftest::aarch64_test_loading_full_dump): New function.
27203 (selftest::aarch64_run_selftests): New function.
27204 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
27205 selftest::aarch64_run_selftests.
27206 * config/i386/i386.c
27207 (selftest::ix86_test_loading_dump_fragment_1): New function.
27208 (selftest::ix86_test_loading_call_insn): New function.
27209 (selftest::ix86_test_loading_full_dump): New function.
27210 (selftest::ix86_test_loading_unspec): New function.
27211 (selftest::ix86_run_selftests): Call the new functions.
27212 * emit-rtl.c (maybe_set_max_label_num): New function.
27213 * emit-rtl.h (maybe_set_max_label_num): New decl.
27214 * function.c (instantiate_decls): Guard call to
27215 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
27216 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
27218 * gensupport.c (gen_reader::gen_reader): Pass "false"
27219 for new "compact" param of rtx_reader.
27220 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
27221 rather than an empty string for NULL strings.
27222 * read-md.c: Potentially include config.h rather than bconfig.h.
27223 Wrap include of errors.h with #ifdef GENERATOR_FILE.
27224 (have_error): New global, copied from errors.c.
27225 (md_reader::read_name): Rename to...
27226 (md_reader::read_name_1): ...this, adding "out_loc" param,
27227 and converting "missing name or number" to returning false, rather
27229 (md_reader::read_name): Reimplement in terms of read_name_1.
27230 (md_reader::read_name_or_nil): New function.
27231 (md_reader::read_string): Handle "(nil)" by returning NULL.
27232 (md_reader::md_reader): Add new param "compact".
27233 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
27234 (md_reader::read_file): New method.
27235 * read-md.h (md_reader::md_reader): Add new param "compact".
27236 (md_reader::read_file): New method.
27237 (md_reader::is_compact): New accessor.
27238 (md_reader::read_name): Convert return type from void to file_location.
27239 (md_reader::read_name_or_nil): New decl.
27240 (md_reader::read_name_1): New decl.
27241 (md_reader::m_compact): New field.
27242 (noop_reader::noop_reader): Pass "false" for new "compact" param
27244 (rtx_reader::rtx_reader): Add new "compact" param.
27245 (rtx_reader::read_rtx_operand): Make virtual and convert return
27246 type from void to rtx.
27247 (rtx_reader::read_until): New decl.
27248 (rtx_reader::handle_any_trailing_information): New virtual function.
27249 (rtx_reader::postprocess): New virtual function.
27250 (rtx_reader::finalize_string): New virtual function.
27251 (rtx_reader::m_in_call_function_usage): New field.
27252 (rtx_reader::m_reuse_rtx_by_id): New field.
27253 * read-rtl-function.c: New file.
27254 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
27255 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
27256 (selftest::verify_three_block_rtl_cfg): New decl.
27257 * read-rtl-function.h: New file.
27258 * read-rtl.c: Potentially include config.h rather than bconfig.h.
27259 For host, include function.h, memmodel.h, and emit-rtl.h.
27260 (one_time_initialization): New function.
27261 (struct compact_insn_name): New struct.
27262 (compact_insn_names): New array.
27263 (find_code): Handle insn codes in compact dumps.
27264 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
27265 (bind_subst_iter_and_attr): Likewise.
27266 (add_condition_to_string): Likewise.
27267 (add_condition_to_rtx): Likewise.
27268 (apply_attribute_uses): Likewise.
27269 (add_current_iterators): Likewise.
27270 (apply_iterators): Likewise.
27271 (initialize_iterators): Guard usage of apply_subst_iterator with
27272 #ifdef GENERATOR_FILE.
27273 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
27274 (md_reader::read_mapping): Likewise.
27275 (add_define_attr_for_define_subst): Likewise.
27276 (add_define_subst_attr): Likewise.
27277 (read_subst_mapping): Likewise.
27278 (check_code_iterator): Likewise.
27279 (rtx_reader::read_rtx): Likewise. Move one-time initialization
27281 (one_time_initialization): New function.
27282 (rtx_reader::read_until): New method.
27283 (read_flags): New function.
27284 (parse_reg_note_name): New function.
27285 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
27286 Handle reuse_rtx ids.
27287 Wrap iterator lookup within #ifdef GENERATOR_FILE.
27288 Add parsing support for RTL dumps, mirroring the special-cases in
27289 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
27290 values, and calling handle_any_trailing_information.
27291 (rtx_reader::read_rtx_operand): Convert return type from void
27292 to rtx, returning return_rtx. Handle case 'e'. Call
27293 finalize_string on XSTR and XTMPL fields.
27294 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
27295 "(nil)" values were omitted. Call the postprocess vfunc on the
27297 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
27298 class ctor. Initialize m_in_call_function_usage. Call
27299 one_time_initialization.
27300 * rtl-tests.c (selftest::test_uncond_jump): Call
27301 set_new_first_and_last_insn.
27302 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
27303 * selftest-rtl.c: New file.
27304 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
27305 (selftest::get_insn_by_uid): New decl.
27306 * selftest-run-tests.c (selftest::run_tests): Call
27307 read_rtl_function_c_tests.
27308 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
27309 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
27312 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
27314 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
27315 operands in a special way. Assert that pos+len <= mode precision.
27317 2017-01-05 Jakub Jelinek <jakub@redhat.com>
27319 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
27320 3 argument Alias with unlimited for the negative form.
27321 (fno-vect-cost-model): Removed.
27323 2017-01-05 Martin Liska <mliska@suse.cz>
27325 * hsa-gen.c (gen_hsa_divmod): New function.
27326 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
27328 2017-01-05 Martin Liska <mliska@suse.cz>
27331 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
27334 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27336 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
27337 small constant length operands.
27339 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27341 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
27342 between loop iterations.
27344 2017-01-05 Martin Liska <mliska@suse.cz>
27347 * gimplify.c (gimplify_decl_expr): Compare to
27348 asan_poisoned_variables instread of checking flags.
27349 (gimplify_target_expr): Likewise.
27350 (gimplify_expr): Likewise.
27351 (gimplify_function_tree): Conditionally initialize
27352 asan_poisoned_variables.
27354 2017-01-04 Jeff Law <law@redhat.com>
27356 PR tree-optimizatin/78812
27357 * rtl.h (contains_mem_rtx_p): Prototype.
27358 * ifcvt.c (containts_mem_rtx_p): Move from here to...
27359 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
27360 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
27361 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
27362 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
27364 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27366 * input.c (assert_char_at_range): Default-initialize actual_range.
27368 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27370 * df-scan.c (df_ref_create_structure): Make regno unsigned,
27371 to match the caller.
27373 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27375 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
27376 insns after final jump in test to emit dummy move.
27378 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27380 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
27381 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
27383 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27385 * multiple_target.c (create_dispatcher_calls): Init e_next.
27386 * tree-ssa-loop-split.c (split_loop): Init border.
27387 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
27390 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
27395 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
27396 (altivec_register_operand): Do not return true if the operand
27397 contains a SUBREG mixing SImode and SFmode.
27398 (vsx_register_operand): Likewise.
27399 (vsx_reg_sfsubreg_ok): New predicate.
27400 (vfloat_operand): Do not return true if the operand contains a
27401 SUBREG mixing SImode and SFmode.
27402 (vint_operand): Likewise.
27403 (vlogical_operand): Likewise.
27404 (gpc_reg_operand): Likewise.
27405 (int_reg_operand): Likewise.
27406 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
27407 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
27408 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
27410 (rs6000_emit_move_si_sf_subreg): New helper function.
27411 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
27412 fixup SUBREGs involving SImode and SFmode.
27413 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
27414 numbers for the new peephole2 optimization.
27415 (peephole2 for SFmode unions): New peephole2 to optimize cases in
27416 the GLIBC math library that do AND/IOR/XOR operations on single
27417 precision floating point.
27418 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
27419 target macros to say whether we need to avoid SUBREGs mixing
27421 (TARGET_ALLOW_SF_SUBREG): Likewise.
27422 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
27423 (UNSPEC_SI_FROM_SF): Likewise.
27424 (iorxor): Change spacing.
27425 (and_ior_xor): New iterator for AND, IOR, and XOR.
27426 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
27427 (movdi_from_sf_zero_ext): Likewise.
27428 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
27429 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
27430 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
27431 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
27432 (fms<mode>4): Likewise.
27433 (fnma<mode>4): Likewise.
27434 (fnms<mode>4): Likewise.
27435 (nfma<mode>4): Likewise.
27436 (nfms<mode>4): Likewise.
27438 2017-01-04 Marek Polacek <polacek@redhat.com>
27441 * doc/invoke.texi: Document -Wpointer-compare.
27443 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27445 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
27448 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
27449 descriptions for -gdwarf-5 and emit them as uleb128 instead of
27452 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
27455 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
27456 documentation of the powerpc_popcntb_ok attribute.
27457 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27458 code to issue warning messages if a requested CPU configuration is
27459 not supported by the binary (assembler and loader) toolchain.
27460 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
27461 made to define a built-in function that has been disabled.
27462 (paired_init_builtins): Add assertion to prevent ICE if attempt is
27463 made to define a built-in function that has been disabled.
27464 (altivec_init_builtins): Add comment explaining why definition
27465 of the DST built-in functions is not preceded by an assertion
27466 check. Add assertions to prevent ICE if attempts are made to
27467 define an altivec predicate or an abs* built-in function that has
27469 (htm_init_builtins): Add comment explaining why definition of the
27470 htm built-in functions is not preceded by an assertion check.
27472 2017-01-04 Jeff Law <law@redhat.com>
27474 PR tree-optimizatin/67955
27475 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
27476 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
27477 the points-to solution does not include pt_null. Use DECL_PT_UID
27480 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
27482 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
27483 Use gen_int_mode instead of gen_lopwart for const_int operands.
27485 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27487 PR tree-optimization/71563
27488 * match.pd: Simplify X << Y into X if Y is known to be 0 or
27489 out of range value - has low bits known to be zero.
27491 2017-01-04 Alan Modra <amodra@gmail.com>
27493 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
27494 * configure: Regenerate.
27495 * config.in: Regenerate.
27497 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27500 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
27501 a substring of the message, but strcmp with the whole message. Ifdef
27502 ENABLE_NLS, translate the message first using dgettext.
27504 2017-01-03 Jeff Law <law@redhat.com>
27506 PR tree-optimizatin/78856
27507 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
27508 (mark_threaded_blocks): Remove code to truncate thread paths that
27509 cross multiple loop headers. Instead invalidate the cached loop
27510 iteration information and handle case of a thread path walking
27511 into an irreducible region.
27513 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
27516 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
27517 assertions. Add support for doing the signbit if the IEEE 128-bit
27518 floating point value is in a GPR.
27519 * config/rs6000/rs6000.md (Fsignbit): Delete.
27520 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
27521 Update the length attribute if the value is in a GPR.
27522 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
27523 the sign or zero extension instruction, since the value is always 0/1.
27524 (signbit<mode>2_dm2): Delete using <Fsignbit>.
27527 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
27528 extracting SImode to a GPR register so that we can generate a
27529 store, limit the vector to be in a traditional Altivec register
27530 for the vextuwrx instruction.
27532 2017-01-03 Ian Lance Taylor <iant@google.com>
27534 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
27536 2017-01-03 Martin Sebor <msebor@redhat.com>
27538 PR tree-optimization/78696
27539 * gimple-ssa-sprintf.c (format_floating): Correct handling of
27540 precision. Use MPFR for %f for greater fidelity. Correct handling
27542 (pass_sprintf_length::compute_format_length): Set width and precision
27543 specified by asrerisk to void_node for vararg functions.
27544 (try_substitute_return_value): Adjust dump output.
27546 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
27548 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
27550 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
27552 * doc/invoke.texi (SPARC options): Document -mlra as the default.
27553 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
27554 -mlra/-mno-lra was passed to the compiler.
27556 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
27558 PR rtl-optimization/65618
27559 * emit-rtl.c (try_split): Move initialization of "before" and
27560 "after" to just before the call to emit_insn_after_setloc.
27562 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
27564 * doc/md.texi (Standard Names): Remove reference to Java frontend.
27566 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
27568 * dwarf2out.c (gen_enumeration_type_die): When
27569 -gno-strict-dwarf, add a DW_AT_encoding attribute.
27571 2017-01-03 Jakub Jelinek <jakub@redhat.com>
27573 PR tree-optimization/78965
27574 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
27575 Change first argument from const call_info & to call_info &. For %n
27576 set info.nowrite to false.
27578 PR middle-end/78901
27579 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
27580 possibly throwing calls.
27582 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
27583 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
27584 and fns handling, rather than in a separate case SSA_NAME.
27586 2017-01-02 Jeff Law <law@redhat.com>
27588 * config/darwin-driver.c (darwin_driver_init): Const-correctness
27589 fixes for first_period and second_period variables.
27591 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
27594 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
27595 (*insvqi_1): New insn pattern.
27596 (*insvqi_1_mem_rex64): Ditto.
27597 (*insvqi_2): Ditto.
27598 (*insvqi_3): Rename from *insvqi.
27600 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
27602 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
27604 * doc/cfg.texi (Edges): Remove reference to Java.
27605 (Maintaining the CFG): Ditto.
27607 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27609 PR middle-end/77674
27610 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
27611 transparent aliases.
27613 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27615 PR middle-end/77484
27616 * predict.def (PRED_CALL): Update hitrate.
27617 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
27618 * predict.c (tree_estimate_probability_bb): Split CALL predictor
27619 into direct/indirect/polymorphic variants.
27621 2017-01-01 Jakub Jelinek <jakub@redhat.com>
27623 Update copyright years.
27625 * gcc.c (process_command): Update copyright notice dates.
27626 * gcov-dump.c (print_version): Ditto.
27627 * gcov.c (print_version): Ditto.
27628 * gcov-tool.c (print_version): Ditto.
27629 * gengtype.c (create_file): Ditto.
27630 * doc/cpp.texi: Bump @copying's copyright year.
27631 * doc/cppinternals.texi: Ditto.
27632 * doc/gcc.texi: Ditto.
27633 * doc/gccint.texi: Ditto.
27634 * doc/gcov.texi: Ditto.
27635 * doc/install.texi: Ditto.
27636 * doc/invoke.texi: Ditto.
27638 Copyright (C) 2017 Free Software Foundation, Inc.
27640 Copying and distribution of this file, with or without modification,
27641 are permitted in any medium without royalty provided the copyright
27642 notice and this notice are preserved.