1 2017-12-12 Jeff Law <law@redhat.com>
3 PR tree-optimization/83410
4 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
5 threads when parallelizing loops.
7 2017-12-15 Jakub Jelinek <jakub@redhat.com>
9 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
11 * config/alpha/alpha.c (vms_attribute_table): Swap
12 affects_type_identity and handler fields, adjust comments.
13 * config/mips/mips.c (mips_attribute_table): Likewise.
14 * config/visium/visium.c (visium_attribute_table): Likewise.
15 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
16 * config/microblaze/microblaze.c (microblaze_attribute_table):
18 * config/spu/spu.c (spu_attribute_table): Likewise.
19 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
20 * config/arc/arc.c (arc_attribute_table): Likewise.
21 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
22 * config/v850/v850.c (v850_handle_interrupt_attribute,
23 v850_handle_data_area_attribute): Formatting fixes.
24 (v850_attribute_table): Swap affects_type_identity and handler
25 fields, adjust comments.
26 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
27 * config/arm/arm.c (arm_attribute_table): Likewise.
28 * config/avr/avr.c (avr_attribute_table): Likewise.
29 * config/s390/s390.c (s390_attribute_table): Likewise.
30 * config/sh/sh.c (sh_attribute_table): Likewise.
31 * config/i386/i386.c (ix86_handle_cconv_attribute,
32 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
33 (ix86_attribute_table): Swap affects_type_identity and handler
34 fields, adjust comments.
35 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
36 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
37 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
38 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
39 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
40 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
41 * config/rx/rx.c (rx_attribute_table): Likewise.
42 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
43 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
44 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
45 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
46 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
47 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
48 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
49 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
50 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
51 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
52 * doc/plugins.texi (user_attr): Likewise. Add NULL for
54 * attribs.c (empty_attribute_table): Swap affects_type_identity and
56 (register_scoped_attributes, decl_attributes): Formatting fixes.
58 PR tree-optimization/83269
59 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
60 subtraction in arg0's type if type is signed and arg0 is unsigned.
64 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
65 convert? on @0 instead of convert. Check type of @1, not @0.
66 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
67 convert? on @0 instead of convert. Check type of @1, not @0.
68 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
69 only optimize if either both @1 and @2 types are narrower
70 precision, or both are wider or equal precision, and in the former
71 case only if both have undefined overflow.
73 2017-12-15 Richard Biener <rguenther@suse.de>
76 * internal-fn.def (IFN_NOP): Add.
77 * internal-fn.c (expand_NOP): Do nothing.
78 * lto-streamer-in.c (input_function): Instead of removing
79 sanitizer calls replace them with IFN_NOP calls.
81 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
82 Alan Hayward <alan.hayward@arm.com>
83 David Sherwood <david.sherwood@arm.com>
85 * dse.c (store_info, read_info_type): Replace begin and end with
87 (print_range): New function.
88 (set_all_positions_unneeded, any_positions_needed_p)
89 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
91 (record_store): Likewise. Optimize the case in which all positions
93 (get_stored_val): Replace read_begin and read_end with read_offset
95 (replace_read): Update call accordingly.
97 2017-12-15 Bin Cheng <bin.cheng@arm.com>
99 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
100 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
101 (loop_cand::loop_cand): Initialize above members.
102 (loop_cand::supported_operations): Delete.
103 (loop_cand::can_interchange_p): Inline above function.
104 (loop_cand::classify_simple_reduction): Record number of constant
105 initialized simple reductions.
106 (should_interchange_loops): New parameters. Check stmt cost of loops
108 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
109 Update call to should_interchange_loops.
110 (should_interchange_loop_nest): Update call to
111 should_interchange_loops.
113 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
116 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
117 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
118 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
120 2017-12-15 Julia Koval <julia.koval@intel.com>
122 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
123 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi): New.
124 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
125 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
126 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
128 2017-12-15 Julia Koval <julia.koval@intel.com>
130 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
131 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
132 * config/i386/sse.md (vaesenc_<mode>): New pattern.
133 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
134 _mm_aesenc_epi128): New intrinsics.
136 2017-12-15 Julia Koval <julia.koval@intel.com>
138 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
139 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi): New.
140 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
141 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
142 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
144 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
146 * gimple-ssa-strength-reduction.c (analyze_increments):
147 Distinguish replacement costs for constant strides from those for
150 2017-12-14 Jakub Jelinek <jakub@redhat.com>
152 * var-tracking.c (variable_tracking_main_1): Formatting fix.
154 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
156 * doc/invoke.texi: Document -Wcast-function-type.
157 * recog.h (stored_funcptr): Change signature.
158 * tree-dump.c (dump_node): Avoid warning.
159 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
161 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
164 * gimple-fold.c (get_range_strlen): Add the handling of non-member
167 2017-12-14 David Malcolm <dmalcolm@redhat.com>
169 PR tree-optimization/83312
170 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
171 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
172 "val" param, and to cope with arbitrary basic blocks.
173 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
174 handle NULL_TREE for "val", dropping "bb" param.
175 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
176 drop "bb" param. Handle NULL_TREE for "val".
177 (find_case_label_for_value): Make "switch_stmt" param const.
178 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
180 (vrp_prop::check_all_array_refs): Reimplement as...
181 (check_array_bounds_dom_walker::before_dom_children): ...this new
182 vfunc. Replace linear search through BB block list, excluding
183 those with non-executable in-edges via dominator walk.
185 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
187 * config/arm/arm.opt (mverbose-cost-dump): New option.
188 * config/arm/arm.c (arm_rtx_costs): Use it.
190 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
193 * reload1.c (emit_input_reload_insns): Skip debug markers.
195 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
197 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
201 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
202 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
203 * config/sh/sh.c: Skip debug insns besides notes.
204 * config/sh/sh.md: Likewise.
205 * config/sh/sh_treg_combine.cc: Likewise.
206 * config/sh/sync.md: Likewise.
208 2017-12-14 Tom de Vries <tom@codesourcery.com>
210 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
211 Add item for weak_undefined.
213 2017-12-14 Richard Biener <rguenther@suse.de>
215 PR tree-optimization/67842
216 * sese.h (bb_in_region): Remove #if 0'ed code.
218 2017-12-14 Richard Biener <rguenther@suse.de>
220 PR tree-optimization/83326
221 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
222 may_be_zero parameter and handle it by not marking the first
223 peeled copy as not exiting the loop.
224 (try_peel_loop): Likewise.
225 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
226 to handle the case of constant or zero iterations and perform
227 loop header copying on-the-fly.
229 2017-12-14 Richard Biener <rguenther@suse.de>
231 PR tree-optimization/83418
232 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
233 Instead of asserting we don't get unfolded comparisons deal with
236 2017-12-14 Jakub Jelinek <jakub@redhat.com>
239 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
240 basic blocks. Assert debug bind insns don't appear outside of bbs,
241 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
244 PR tree-optimization/83198
245 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
246 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
247 value if arg is a REAL_CST with incompatible type.
249 2017-12-14 Sudakshina Das <sudi.das@arm.com>
250 Bin Cheng <bin.cheng@arm.com>
253 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
255 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
258 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
260 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
261 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
263 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
264 Alan Hayward <alan.hayward@arm.com>
265 David Sherwood <david.sherwood@arm.com>
267 * poly-int.h: New file.
268 * poly-int-types.h: Likewise.
269 * coretypes.h: Include them.
270 (POLY_INT_CONVERSION): Define.
271 * target.def (estimated_poly_value): New hook.
272 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
273 * doc/tm.texi: Regenerate.
274 * doc/poly-int.texi: New file.
275 * doc/gccint.texi: Include it.
276 * doc/rtl.texi: Describe restrictions on subreg modes.
277 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
278 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
279 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
280 * targhooks.h (default_estimated_poly_value): Declare.
281 * targhooks.c (default_estimated_poly_value): New function.
282 * target.h (estimated_poly_value): Likewise.
283 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
284 (wi::unary_traits): Delete.
285 (wi::binary_traits::signed_shift_result_type): Define for
286 offset_int << HOST_WIDE_INT, etc.
287 (generic_wide_int::operator <<=): Define for all types and use
288 wi::lshift instead of <<.
289 (wi::hwi_with_prec): Add a default constructor.
290 (wi::ints_for): New class.
291 (operator <<): Define for all wide-int types.
292 (operator /): New function.
293 (operator %): Likewise.
294 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
295 (ASSERT_MAYBE_NE_AT): New macros.
297 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
298 Dominik Vogt <vogt@linux.vnet.ibm.com>
301 * emit-rtl.c (init_emit): Remove ??? comment.
302 * explow.c (get_dynamic_stack_size): Take known alignment of stack
303 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
304 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
305 alignment of 3 virtual registers to BITS_PER_WORD.
307 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
309 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
311 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
312 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
314 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
317 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
319 2017-12-13 Jakub Jelinek <jakub@redhat.com>
321 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
322 with the exception of debug begin stmt markers appear before
326 * final.c (rest_of_handle_final): Call variable_tracking_main only
327 if !flag_var_tracking.
329 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
330 Jakub Jelinek <jakub@redhat.com>
334 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
335 labels debug stmts that can only appear after labels.
337 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
339 PR rtl-optimization/82398
340 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
341 EXPR_USEFULNESS in priority comparison.
343 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
345 PR rtl-optimization/83393
346 * combine.c (move_deaths): If reg_stat points to a too new insn in
347 last_death, do not use it: find the proper insn instead.
349 2017-12-12 Jeff Law <law@redhat.com>
351 PR tree-optimization/83298
352 PR tree-optimization/83362
353 PR tree-optimization/83383
354 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
355 push_value_range a public interface. Add new argument to
356 record_ranges_from_stmt.
357 * gimple-ssa-evrp-analyze.c
358 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
359 Update comments. Handle recording temporary equivalences.
360 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
361 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
362 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
363 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
364 gimple-ssa-evrp-analyze.h.
365 (record_temporary_equivalences_from_phis): Add new argument. When
366 the PHI arg is an SSA_NAME, set the result's range to the range
368 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
370 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
371 Pass it down to children as needed.
372 (thread_outgoing_edges): Likewise.
373 (thread_across_edge): Likewise. Push/pop range state as needed.
374 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
376 2017-12-12 Julia Koval <julia.koval@intel.com>
378 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
379 (PTA_CANNONLAKE): Remove PTA_CLWB.
381 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
384 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
385 (vcondv2div2df): Likewise.
386 (vconduv2dfv2di): Likewise.
388 2017-12-12 Jakub Jelinek <jakub@redhat.com>
390 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
391 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
392 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
393 ATTR_MATHFN_FPROUNDING_ERRNO.
395 2017-12-12 Richard Biener <rguenther@suse.de>
397 PR tree-optimization/83385
398 * graphite-scop-detection.c (get_order, order): Remove.
399 (bb_to_rpo): New global.
401 (build_scops): Sort pbbs in RPO order.
403 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
405 * combine.c (simplify_set): Do not transform subregs to zero_extends
406 if the destination mode is a vector mode.
408 2017-12-12 Jakub Jelinek <jakub@redhat.com>
410 PR tree-optimization/83359
411 * tree-cfg.h (fold_loop_internal_call): Declare.
412 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
413 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
414 (find_loop_dist_alias): New function.
415 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
416 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
419 PR tree-optimization/80631
420 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
421 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
422 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
423 hardcoding zero as the value if COND_EXPR is never true. For
424 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
425 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
426 hardcoding MAX_EXPR as the reduction operation.
427 (is_nonwrapping_integer_induction): Allow negative step.
428 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
429 vect_create_epilog_for_reduction, if no value is suitable, don't
430 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
432 2017-12-12 Richard Biener <rguenther@suse.de>
434 PR tree-optimization/81889
435 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
436 range info from the non-wrapping IV instead of just the range
439 2017-12-12 Julia Koval <julia.koval@intel.com>
441 * config.gcc: Add vaesintrin.h.
442 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
444 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
445 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
447 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
448 * config/i386/immintrin.h: Include vaesintrin.h.
449 * config/i386/sse.md (vaesdec_<mode>): New pattern.
450 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
451 _mm_aesdec_epi128): New intrinsics.
453 2017-12-12 Julia Koval <julia.koval@intel.com>
455 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
456 OPTION_MASK_ISA_VAES_UNSET): New.
457 (ix86_handle_option): Handle -mvaes.
458 * config/i386/cpuid.h: Define bit_VAES.
459 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
460 * config/i386/i386-c.c (__VAES__): New.
461 * config/i386/i386.c (ix86_target_string): Add -mvaes.
462 (ix86_valid_target_attribute_inner_p): Ditto.
463 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
464 * config/i386/i386.opt: Add -mvaes.
465 * doc/invoke.texi: Ditto.
467 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
469 * debug.h (gcc_debug_hooks): Add inline_entry.
470 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
471 * debug.c (do_nothing_debug_hooks): Likewise.
472 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
473 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
474 (dwarf2_lineno_debug_hooks): Likewise.
476 * common.opt (gstatement-frontiers): New, setting
477 debug_nonbind_markers_p.
478 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
479 * toplev.c (process_options): Autodetect value for debug statement
481 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
482 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
484 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
485 markers. Integrate source bind into debug stmt expand loop.
486 (pass_expand::execute): Check debug marker limit. Avoid deep
487 TER and expand debug locations for debug bind insns only.
488 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
490 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
491 * final.c (reemit_insn_block_notes): Take current block from
492 nonbind markers. Declare note where it's first set.
493 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
494 begin stmt markers if enabled.
495 (notice_source_line): Handle nonbind markers. Fail if their
496 location is unknown or that of builtins.
497 (rest_of_handle_final): Convert begin stmt markers to notes if
498 var-tracking didn't run.
499 (rest_of_clean_state): Skip begin stmt markers.
500 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
502 * function.c (allocate_struct_function): Set begin_stmt_markers.
503 * function.h (struct function): Add debug_marker_count counter
504 and debug_nonbind_markers flag.
505 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
506 * gimple-low.c (lower_function_body): Adjust
507 debug_nonbind_markers.
508 (lower_stmt): Drop or skip gimple debug stmts.
509 (lower_try_catch): Skip debug stmts.
510 * gimple.c (gimple_build_debug_begin_stmt): New.
511 (gimple_copy): Increment debug_marker_count if copying one.
512 * gimple.h (gimple_build_debug_begin_stmt): Declare.
513 * gimplify.c (rexpr_location): New.
514 (rexpr_has_location): New.
515 (warn_switch_unreachable_r): Handle gimple debug stmts.
516 (shortcut_cond_r): Call expr_location.
518 (find_goto_label): New.
519 (shortcut_cond_expr): Call expr_has_location, expr_location, and
521 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
523 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
524 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
525 (LANG_HOOKS_INITIALIZER): ... this.
526 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
527 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
529 (update_ebb_live_info): Skip debug insn markers.
530 * lra.c (debug_insn_static_data): Rename to...
531 (debug_bind_static_data): ... this.
532 (debug_marker_static_data): New.
533 (lra_set_insn_recog_data): Select one of the above depending
535 (lra_update_isn_regno_info): Don't assume debug insns have
537 (push_insns): Skip debug insns.
538 * lto-streamer-in.c (input_function): Drop debug stmts
539 depending on active options. Adjust debug_nonbind_markers.
540 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
541 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
542 begin stmt marker notes.
543 (print_insn): Likewise.
544 * recog.c (extract_insn): Recognize rtl for debug markers.
545 * rtl.def (DEBUG_MARKER): New.
546 * tree-inline.c: Include params.h.
547 (remap_gimple_stmt): Handle nonbind markers.
548 (maybe_move_debug_stmts_to_successors): Likewise.
549 (copy_debug_stmt): Likewise.
550 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
551 markers regardless of no side effects.
552 (tsi_link_before): Don't update container's side effects when adding
554 (tsi_link_after): Likewise.
555 (expr_first): Skip begin stmt markers.
556 (expr_last): Likewise.
557 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
558 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
559 Disregard nonbind markers.
560 * tree.c (make_node_stat): Don't set side effects for begin stmt
562 (build1_stat): Likewise.
563 * tree.def (DEBUG_BEGIN_STMT): New.
564 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
565 * var-tracking.c (delete_debug_insns): Renamed to...
566 (delete_vta_debug_insns): ... this.
567 (reemit_marker_as_note): New.
568 (vt_initialize): Reemit markers.
569 (delete_vta_debug_insns): Likewise.
570 (vt_debug_insns_local): Reemit or delete markers.
571 (variable_tracking_main_1): Likewise.
572 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
573 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
574 (gimple_debug_nonbind_marker_p): New.
575 (gimple_build_debug_bind): Adjust.
576 (gimple_build_debug_begin_stmt): New.
577 * doc/invoke.texi (max-debug-marker-count): New param.
578 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
579 (debug_parameter_ref, debug_marker): New.
580 (NOTE_INSN_BEGIN_STMT): New.
581 (DEBUG_INSN): Describe begin stmt markers.
583 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
584 without debug insns, we wouldn't, but clean up debug insns
585 after a control flow insn nevertheless.
587 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
588 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
589 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
590 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
591 (rtl_tidy_fallthru_edge): Likewise.
592 (rtl_verify_fallthru): Likewise.
593 (rtl_verify_bb_layout): Likewise.
594 (skip_insns_after_block): Likewise.
595 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
596 * dwarf2out.c: Include print-rtl.h.
597 (dwarf2out_next_real_insn): New.
598 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
599 Dump debug binds in asm comments.
600 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
601 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
602 callers to use gsi_start_nondebug_bb instead.
603 (gsi_after_labels): Skip gimple debug stmts.
604 (gsi_start_nondebug): New.
605 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
606 (proper_loop_form_for_interchange): Adjust.
607 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
608 * gimple.h (gimple_seq_last_nondebug_stmt): New.
609 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
610 (collect_fallthrough_labels): Likewise.
611 (should_warn_for_implicit_fallthrough): Likewise.
612 (warn_implicit_fallthrough_r): Likewise.
613 (expand_FALLTHROUGH_r): Likewise.
614 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
615 (graphite_copy_stmts_from_block): Skip nonbind markers.
616 * haifa-sched.c (sched_extend_bb): Skip debug insns.
617 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
618 * jump.c (clean_barriers): Skip debug insns.
619 * omp-expand.c (expand_parallel_call): Skip debug insns.
620 (expand_task_call): Likewise.
621 (remove_exit_barrier): Likewise.
622 (expand_omp_taskreg): Likewise.
623 (expand_omp_for_init_counts): Likewise.
624 (expand_omp_for_generic): Likewise.
625 (expand_omp_for_static_nochunk): Likewise.
626 (expand_omp_for_static_chunk): Likewise.
627 (expand_omp_simd): Likewise.
628 (expand_omp_taskloop_for_outer): Likewise.
629 (expand_omp_taskloop_for_inner): Likewise.
630 (expand_oacc_for): Likewise.
631 (expand_omp_sections): Likewise.
632 (expand_omp_single): Likewise.
633 (expand_omp_synch): Likewise.
634 (expand_omp_atomic_load): Likewise.
635 (expand_omp_atomic_store): Likewise.
636 (expand_omp_atomic_fetch_op): Likewise.
637 (expand_omp_atomic_pipeline): Likewise.
638 (expand_omp_atomic_mutex): Likewise.
639 (expand_omp_target): Likewise.
640 (grid_expand_omp_for_loop): Likewise.
641 (grid_expand_target_grid_body): Likewise.
642 (build_omp_regions_1): Likewise.
643 * omp-low.c (check_combined_parallel): Skip debug stmts.
644 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
645 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
646 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
648 * tree-cfg.c (make_blobs_1): Skip debug stmts.
649 (make_edges): Likewise.
650 (cleanup_dead_labels): Likewise.
651 (gimple_can_merge_blocks_p): Likewise.
652 (stmt_starts_bb_p): Likewise.
653 (gimple_block_label): Likewise.
654 (gimple_redirect_edge_and_branch): Likewise.
655 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
657 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
659 * tree-pretty-print (print_declaration): Omit initializer in slim
661 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
663 (eliminate_unnecessary_stmts): Stabilize block removal order.
664 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
665 * var-tracking.c (get_first_insn): New.
666 (vt_emit_notes): Call it.
667 (vt_initialize): Walk any insns before the first BB.
668 (delete_debug_insns): Likewise.
670 * gimple.h (enum gimple_debug_subcode): Add
671 GIMPLE_DEBUG_BEGIN_STMT.
672 (gimple_debug_begin_stmt_p): New.
673 (gimple_debug_nonbind_marker_p): New.
674 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
675 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
676 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
677 * insn-notes.def (BEGIN_STMT): New.
678 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
679 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
680 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
681 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
682 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
683 (INSN_DEBUG_MARKER_KIND): New.
684 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
685 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
686 (INSN_VAR_LOCATION_PTR): New.
687 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
689 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
690 deps only if debug bind insns are enabled.
691 (pass_expand::execute): Avoid deep TER and expand
692 debug locations for debug bind insns only.
693 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
694 debug stmts special handling down to debug bind stmts.
695 * combine.c (try_combine): Narrow debug insns special handling
696 down to debug bind insns.
697 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
698 Narrow debug insns preexisting special handling down to debug
700 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
701 handling down to debug bind insns.
702 * function.c (instantiate_virtual_regs): Skip debug markers,
703 adjust handling of debug binds.
704 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
705 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
706 * haifa-sched.c (schedule_insn): Narrow special handling of debug
707 insns to debug bind insns.
708 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
709 special handling of debug stmts to debug bind stmts.
710 * ipa-split.c (split_function): Likewise.
711 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
712 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
714 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
715 * regrename.c (build_def_use): Likewise.
716 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
717 (pass_cprop_hardreg): Narrow special casing of debug insns to
719 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
720 * reload1.c (reload): Likewise.
721 * sese.c (sese_insert_phis_for_liveouts): Narrow special
722 casing of debug stmts to debug bind stmts.
723 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
724 * ssa-iterators.h (num_imm_uses): Likewise.
725 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
726 debug stmts to debug bind stmts.
727 * tree-inline.c (tree_function_versioning): Narrow special casing
728 of debug stmts to debug bind stmts.
729 * tree-loop-distribution.c (generate_loops_for_partition):
730 Narrow special casing of debug stmts to debug bind stmts.
731 * tree-sra.c (analyze_access_subtree): Narrow special casing
732 of debug stmts to debug bind stmts.
733 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
734 stmts to debug bind stmts.
735 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
736 casing of debug stmts to debug bind stmts.
737 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
738 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
739 casing of debug stmts to debug bind stmts.
740 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
742 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
743 debug stmts to debug bind stmts.
744 (gimple_replace_ssa_lhs): Likewise.
745 (insert_debug_temp_for_var_def): Likewise.
746 (insert_debug_temps_for_defs): Likewise.
747 (reset_debug_uses): Likewise.
748 * tree-ssanames.c (release_ssa_name_fn): Likewise.
749 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
750 (adjust_debug_stmts): Likewise.
751 (adjust_phi_and_debug_stmts): Likewise.
752 (vect_do_peeling): Likewise.
753 * tree-vect-loop.c (vect_transform_loop): Likewise.
754 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
755 * var-tracking.c (adjust_mems): Narrow special casing of debug
756 insns to debug bind insns.
757 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
758 (compute_bb_dataflow, vt_find_locations): Likewise.
759 (vt_expand_loc, emit_notes_for_changes): Likewise.
760 (vt_init_cfa_base): Likewise.
761 (vt_emit_notes): Likewise.
762 (vt_initialize): Likewise.
763 (vt_finalize): Likewise.
765 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
766 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
767 (next_nonnote_nondebug_insn_bb): New.
768 (prev_nonnote_nondebug_insn_bb): New.
769 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
770 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
771 (prev_nonnote_nondebug_insn_bb): Declare.
772 (next_nonnote_nondebug_insn_bb): Declare.
773 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
774 * cfgrtl.c (get_last_bb_insn): Likewise.
775 * lra.c (push_insns): Likewise.
777 2017-12-11 David Malcolm <dmalcolm@redhat.com>
780 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
782 * selftest.c (selftest::test_runner::test_runner): New ctor.
783 (selftest::test_runner::~test_runner): New dtor.
784 * selftest.h (class selftest::test_runner): New class.
786 2017-12-11 Carl Love <cel@us.ibm.com>
788 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
789 vec_extract_fp32_from_shortl]): Add #defines.
790 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
791 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
792 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
793 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
794 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
795 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
796 * doc/extend.texi: Add documentation for the added builtins.
798 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
800 PR rtl-optimization/80693
801 PR rtl-optimization/81019
802 PR rtl-optimization/81020
803 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
804 are not mentioned in i3. Place the REG_UNUSED note on i2,
805 possibly modified to REG_DEAD, if it did not originate in i3.
807 2017-12-11 Jakub Jelinek <jakub@redhat.com>
809 * recog.c (store_data_bypass_p_1): New function.
810 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
811 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
813 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
815 PR rtl-optimization/83361
816 * ifcvt.c (if_convert): Call fixup_partitions.
818 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
820 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
821 early folding of splat_u{8,16,32}.
823 2017-12-11 Jakub Jelinek <jakub@redhat.com>
825 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
826 output_operand_lossage first argument with capital letter.
827 (aarch64_override_options): Don't start error and sorry first argument
830 2017-12-11 Andi Kleen <ak@linux.intel.com>
832 PR gcov-profile/83355
833 * auto-profile.c (string_table::get_index_by_decl): Don't
834 recurse when abstract origin points to itself.
836 2017-12-11 Bin Cheng <bin.cheng@arm.com>
838 PR tree-optimization/83320
839 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
840 (prune_datarefs_not_in_loop): Ditto.
842 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
844 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
846 2017-12-10 Jakub Jelinek <jakub@redhat.com>
848 PR tree-optimization/83337
849 * gimple-loop-interchange.cc (compute_access_stride): Handle
850 bitfield DRs properly.
852 2017-12-09 Jakub Jelinek <jakub@redhat.com>
854 PR tree-optimization/83338
855 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
856 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
857 vector of unsigned integers to vector of signed integers.
859 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
861 PR rtl-optimization/83317
862 * lra-constraints.c (process_address_1): Add insn code check.
864 2017-12-08 Michael Matz <matz@suse.de>
866 Fix PR tree-optimization/83323
867 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
868 head-controlled loops and loop BBs.
869 * common.opt (funroll-and-jam): Remove, instead ...
870 (floop-unroll-and-jam): ... reuse this option.
871 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
872 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
873 (-floop-unroll-and-jam): ... this option.
875 2017-12-08 Jakub Jelinek <jakub@redhat.com>
877 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
878 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
879 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
881 2017-12-08 Richard Biener <rguenther@suse.de>
884 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
885 handle accesses outside of zero-sized vars.
887 2017-12-08 Martin Jambor <mjambor@suse.cz>
889 PR tree-optimization/83141
890 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
891 test for MEM_REFs implicitely changing types with padding. Remove
893 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
895 2017-12-08 Jakub Jelinek <jakub@redhat.com>
897 * config/arc/arc.c (arc_attribute_table): Add exclusions to
899 * config/avr/avr.c (avr_attribute_table): Likewise.
900 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
901 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
902 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
903 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
904 of struct attribute_spec.
905 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
907 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
910 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
912 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
915 * profile-count.c (profile_count::from_gcov_type): Move from
916 profile-count.h; handle overflow.
917 * profile-count.h (profile_count::from_gcov_type): Move offline.
919 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
921 PR rtl-optimization/83304
922 * combine.c (move_deaths): If we do not know where a register died,
925 2017-12-08 Richard Biener <rguenther@suse.de>
927 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
928 Provide -fopt-info-loop feedback when we interchange in a nest.
930 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
932 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
933 for armv6 ARM CPU IDs.
935 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
937 * common/config/arm/arm-common.c: Include <algorithm>.
938 (INCLUDE_VECTOR): Define.
939 (compare_opt_names): New function.
940 (arm_rewrite_selected_arch): Only strip out extensions that can be
941 expressed through -mfpu. Sort the remaining extensions
944 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
946 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
947 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
948 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
949 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
951 2017-06-08 Tristan Gingold <gindold@adacore.com>
954 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
955 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
957 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
959 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
960 of struct attribute_spec.
962 2017-12-08 Julia Koval <julia.koval@intel.com>
964 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
965 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
966 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
967 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
968 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
969 _mm_maskz_dpwssds_epi32): Ditto.
971 2017-12-08 Richard Biener <rguenther@suse.de>
973 PR tree-optimization/81303
974 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
975 conditions try to create a comparison vector type matching
976 the data vector type.
977 (vectorizable_condition): Adjust.
978 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
979 Leave invariant conditions alone in case we can vectorize those.
981 2017-12-08 Julia Koval <julia.koval@intel.com>
983 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
984 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
985 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
986 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
987 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
989 2017-12-08 Julia Koval <julia.koval@intel.com>
991 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
992 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
993 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
994 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
995 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
996 _mm_maskz_dpbusds_epi32): New intrinsics.
998 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
1000 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
1002 (ld<bh>io_signed): Likewise.
1003 (st<bhw>io): Likewise.
1004 * config/nios2/predicates.md (ldstio_memory_operand): Allow
1005 SMALL_INT12 constant integer operand.
1007 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
1009 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
1011 (nios2_large_constant_allowed): ...this. Adjust uses.
1012 (nios2_plus_symbolic_constant_p): Rename to...
1013 (nios2_plus_large_constant_p): ...this. Adjust uses.
1014 (nios2_legitimate_address_p): Correct CONST_INT handling.
1015 (nios2_symbolic_memory_operand_p): Rename to...
1016 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
1017 (nios2_large_constant_p): Check for large constant integers too.
1018 (nios2_split_large_constant): Handle constant integers.
1019 (nios2_split_symbolic_memory_operand): Rename to...
1020 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
1021 (nios2_legitimize_constant_address): Handle constant integers.
1022 (r0rel_constant_p): Handle small constant integers.
1023 (nios2_print_operand_address): Handle r0-relative integer addresses.
1024 * config/nios2/nios2-protos.h: Adjust for renamed functions.
1025 * config/nios2/nios2.md: Adjust for renamed functions.
1027 2017-12-07 Andrew Waterman <andrew@sifive.com>
1029 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
1030 (TARGET_HAVE_SRODATA_SECTION): New define.
1031 (riscv_select_section): New function.
1033 2017-12-08 Joseph Myers <joseph@codesourcery.com>
1034 Alexander Monakov <amonakov@ispras.ru>
1035 Jakub Jelinek <jakub@redhat.com>
1038 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
1040 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1042 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
1043 a bool success value. Don't call output_operand_lossage here.
1044 (aarch64_print_ldpstp_address): Return a bool success value.
1045 (aarch64_print_operand_address): Call output_addr_const if
1046 aarch64_print_address_internal fails.
1047 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
1048 'y'; call output_operand_lossage instead. Call output_operand_lossage
1049 if aarch64_print_ldpstp_address fails.
1051 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1053 * tree-vector-builder.h
1054 (tree_vector_builder::binary_encoded_nelts): Declare.
1055 * tree-vector-builder.c
1056 (tree_vector_builder::binary_encoded_nelts): New function.
1057 * fold-const.c (negate_expr_p): Likewise.
1058 (operand_equal_p, fold_checksum_tree): Likewise.
1059 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
1060 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
1061 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
1062 (uniform_vector_p): Likewise.
1063 * varasm.c (const_hash_1, compare_constant): Likewise.
1064 * tree-ssa-ccp.c: Include tree-vector-builder.h.
1065 (valid_lattice_transition): Operate directly on the VECTOR_CST
1067 * ipa-icf.c: Include tree-vector-builder.h.
1068 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
1069 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
1071 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1073 * tree.c (build_vector): Delete.
1074 * tree.h (build_vector): Make static and move into the self-testing
1077 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1079 * vector-builder.h (vector_builder::derived): New const overload.
1080 (vector_builder::elt): New function.
1081 * tree-vector-builder.h (tree_vector_builder::type): New function.
1082 (tree_vector_builder::apply_step): Declare.
1083 * tree-vector-builder.c (tree_vector_builder::apply_step): New
1085 * gimple-fold.h (tree_vector_builder): Declare.
1086 (gimple_build_vector): Take a tree_vector_builder instead of a
1087 type and vector of elements.
1088 * gimple-fold.c (gimple_build_vector): Likewise.
1089 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
1091 (get_initial_defs_for_reduction): Likewise.
1092 (vectorizable_induction): Likewise.
1094 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1096 * tree-vector-builder.h
1097 (tree_vector_builder::new_binary_operation): Declare.
1098 * tree-vector-builder.c
1099 (tree_vector_builder::new_binary_operation): New function.
1100 * fold-const.c (fold_relational_const): Use it.
1101 (const_binop): Likewise. Check that both input vectors have
1102 the same number of elements, thus excluding things like WIDEN_SUM.
1103 Check whether it is possible to operate directly on the encodings
1106 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1108 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
1109 new_unary_operation, operating only on the encoded elements.
1110 (const_unop): Likewise.
1111 (exact_inverse): Likewise.
1112 (distributes_over_addition_p): New function.
1113 (const_binop): Use tree_vector_builder and new_unary_operation
1114 for combinations of VECTOR_CST and INTEGER_CST. Operate only
1115 on the encoded elements unless the encoding is strided and the
1116 operation does not distribute over addition.
1117 (fold_convert_const): Use tree_vector_builder and
1118 new_unary_operation. Operate only on the encoded elements
1119 for truncating integer conversions, or for non-stepped encodings.
1121 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1123 * config/sparc/sparc.c: Include tree-vector-builder.h.
1124 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
1125 * expmed.c: Include tree-vector-builder.h.
1126 (make_tree): Use tree_vector_builder instead of build_vector.
1127 * fold-const.c: Include tree-vector-builder.h.
1128 (const_binop): Use tree_vector_builder instead of build_vector.
1129 (const_unop): Likewise.
1130 (native_interpret_vector): Likewise.
1131 (fold_vec_perm): Likewise.
1132 (fold_ternary_loc): Likewise.
1133 * gimple-fold.c: Include tree-vector-builder.h.
1134 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
1136 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
1137 (simplify_vector_constructor): Use tree_vector_builder instead
1139 * tree-vect-generic.c: Include tree-vector-builder.h.
1140 (add_rshift): Use tree_vector_builder instead of build_vector.
1141 (expand_vector_divmod): Likewise.
1142 (optimize_vector_constructor): Likewise.
1143 * tree-vect-loop.c: Include tree-vector-builder.h.
1144 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
1145 of build_vector. Explicitly use a stepped encoding for
1147 * tree-vect-slp.c: Include tree-vector-builder.h.
1148 (vect_get_constant_vectors): Use tree_vector_builder instead
1150 (vect_transform_slp_perm_load): Likewise.
1151 (vect_schedule_slp_instance): Likewise.
1152 * tree-vect-stmts.c: Include tree-vector-builder.h.
1153 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
1154 (vect_gen_perm_mask_any): Likewise.
1155 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
1156 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
1158 (build_vector_from_val): Likewise. Explicitly use a duplicate
1161 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
1163 * doc/generic.texi (VECTOR_CST): Describe new representation of
1165 * vector-builder.h: New file.
1166 * tree-vector-builder.h: Likewise.
1167 * tree-vector-builder.c: Likewise.
1168 * Makefile.in (OBJS): Add tree-vector-builder.o.
1169 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
1170 * tree-core.h (tree_base): Add a vector_cst field to the u union.
1171 (tree_vector): Change the number of elements to
1172 vector_cst_encoded_nelts.
1173 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
1174 (VECTOR_CST_ELTS): Delete.
1175 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
1176 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
1177 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
1178 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
1179 (VECTOR_CST_ENCODED_ELT): Likewise.
1180 (vector_cst_encoded_nelts): New function.
1181 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
1182 VECTOR_CST_NELTS_PER_PATTERN as arguments.
1183 (vector_cst_int_elt, vector_cst_elt): Declare.
1184 * tree.c: Include tree-vector-builder.h.
1185 (tree_code_size): Abort if passed VECTOR_CST.
1186 (tree_size): Update for new VECTOR_CST layout.
1187 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
1188 VECTOR_CST_NELTS_PER_PATTERN as arguments.
1189 (build_vector): Use tree_vector_builder.
1190 (vector_cst_int_elt, vector_cst_elt): New functions.
1191 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
1192 encoded elements and then create the vector in the canonical form.
1193 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
1194 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
1195 (tree_c_tests): Call test_vector_cst_patterns.
1196 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
1198 (hash_tree): Likewise.
1199 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
1200 (streamer_write_tree_header): Likewise.
1201 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
1202 (streamer_alloc_tree): Likewise. Update call to make_vector.
1203 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
1205 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1207 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
1208 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
1209 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
1211 2017-12-07 Bin Cheng <bin.cheng@arm.com>
1212 Richard Biener <rguenther@suse.de>
1214 PR tree-optimization/81303
1215 * Makefile.in (gimple-loop-interchange.o): New object file.
1216 * common.opt (floop-interchange): Reuse the option from graphite.
1217 * doc/invoke.texi (-floop-interchange): Ditto. New document for
1218 -floop-interchange and mention it for -O3.
1219 * opts.c (default_options_table): Enable -floop-interchange at -O3.
1220 * gimple-loop-interchange.cc: New file.
1221 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
1222 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
1223 * passes.def (pass_linterchange): New pass.
1224 * timevar.def (TV_LINTERCHANGE): New time var.
1225 * tree-pass.h (make_pass_linterchange): New declaration.
1226 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
1227 interchange. Record IV before/after increment in new parameters.
1228 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
1229 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
1231 (check_reduction_path): ...New function here.
1232 * tree-vectorizer.h (check_reduction_path): New declaration.
1234 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
1237 PR rtl-optimization/80818
1238 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
1239 always early clobbered.
1240 * lra-lives.c (process_bb_lives): Check input hard regs for early
1241 clobbered non-operand hard reg.
1243 2017-12-07 Jakub Jelinek <jakub@redhat.com>
1246 * tree-cfg.c (verify_gimple_assign_binary): Don't require
1247 types_compatible_p, just that TYPE_MODE is the same.
1249 2017-12-07 Martin Sebor <msebor@redhat.com>
1252 * attribs.c (empty_attribute_table): Initialize new member of
1253 struct attribute_spec.
1254 (decl_attributes): Add argument. Handle mutually exclusive
1255 combinations of attributes.
1256 (selftests::test_attribute_exclusions): New function.
1257 (selftests::attribute_c_tests): Ditto.
1258 * attribs.h (decl_attributes): Add default argument.
1259 * selftest.h (attribute_c_tests): Declare.
1260 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
1261 * tree-core.h (attribute_spec::exclusions, exclude): New type and
1263 * doc/extend.texi (Common Function Attributes): Update const and pure.
1264 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
1265 of struct attribute_spec.
1266 * config/arc/arc.c (arc_attribute_table): Same.
1267 * config/arm/arm.c (arm_attribute_table): Same.
1268 * config/avr/avr.c ( avr_attribute_table): Same.
1269 * config/bfin/bfin.c (bfin_attribute_table): Same.
1270 * config/cr16/cr16.c (cr16_attribute_table): Same.
1271 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
1272 * config/h8300/h8300.c (h8300_attribute_table): Same.
1273 * config/i386/i386.c (ix86_attribute_table): Same.
1274 * config/ia64/ia64.c (ia64_attribute_table): Same.
1275 * config/m32c/m32c.c (m32c_attribute_table): Same.
1276 * config/m32r/m32r.c (m32r_attribute_table): Same.
1277 * config/m68k/m68k.c (m68k_attribute_table): Same.
1278 * config/mcore/mcore.c (mcore_attribute_table): Same.
1279 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
1280 * config/mips/mips.c (mips_attribute_table): Same.
1281 * config/msp430/msp430.c (msp430_attribute_table): Same.
1282 * config/nds32/nds32.c (nds32_attribute_table): Same.
1283 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
1284 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
1285 * config/rl78/rl78.c (rl78__attribute_table): Same.
1286 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
1287 * onfig/rx/rx.c (rx_attribute_table): Same.
1288 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
1289 * config/sh/sh.c (sh_attribute_table): Same.
1290 * config/sparc/sparc.c (sparc_attribute_table): Same.
1291 * config/spu/spu.c (spu_attribute_table): Same.
1292 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
1293 * config/v850/v850.c (v850_attribute_table): Same.
1294 * config/visium/visium.c (visium_attribute_table): Same.
1296 2017-12-07 Tamar Christina <tamar.christina@arm.com>
1299 * config/arm/arm.c (INCLUDE_STRING): Define.
1300 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
1301 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
1304 2017-12-07 Michael Matz <matz@suse.de>
1306 Add unroll and jam pass
1308 * gimple-loop-jam.c: New file.
1309 * Makefile.in (OBJS): Add gimple-loop-jam.o.
1310 * common.opt (funroll-and-jam): New option.
1311 * opts.c (default_options_table): Add unroll-and-jam at -O3.
1312 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
1313 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
1314 * passes.def: Add pass_loop_jam.
1315 * timevar.def (TV_LOOP_JAM): Add.
1316 * tree-pass.h (make_pass_loop_jam): Declare.
1317 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
1318 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
1319 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
1320 to flow_loop_tree_node_add.
1321 (duplicate_subloops, copy_loops_to): Append to sibling list.
1322 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
1323 * doc/invoke.texi (-funroll-and-jam): Document new option.
1324 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
1326 2017-12-07 Richard Biener <rguenther@suse.de>
1328 PR tree-optimization/83296
1329 PR tree-optimization/67769
1330 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
1331 flow sensitive info in an unrelated BB.
1332 (value_replacement): Use reset_flow_sensitive_info.
1333 (minmax_replacement): Reset flow sensitive info on the def
1334 we move. Do not reset flow sensitive info in the whole BB
1335 we move the stmt to.
1336 (abs_replacement): Likewise.
1338 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1341 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1342 rs6000_cpu to the given -mcpu=, or to the default processor.
1344 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1346 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
1347 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
1348 instead of rs6000_cpu_attr.
1349 (rs6000_adjust_cost): Ditto.
1350 (is_microcoded_insn): Ditto.
1351 (rs6000_adjust_priority): Ditto.
1352 (rs6000_issue_rate): Ditto.
1353 (rs6000_use_sched_lookahead): Ditto.
1354 (rs6000_use_sched_lookahead_guard): Ditto.
1355 (rs6000_sched_reorder): Ditto.
1356 (force_new_group): Ditto.
1357 * config/rs6000/rs6000.md (cpu attribute): Ditto.
1358 (group_ending_nop): Ditto.
1360 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1362 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
1363 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
1364 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
1365 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
1367 (rs6000_reassociation_width): Ditto.
1368 (rs6000_emit_epilogue): Ditto.
1369 (rs6000_adjust_cost): Ditto.
1370 (is_microcoded_insn): Ditto.
1371 (is_cracked_insn): Ditto.
1372 (rs6000_adjust_priority): Ditto.
1373 (rs6000_sched_reorder): Ditto.
1374 (rs6000_sched_reorder2): Ditto.
1375 (insn_must_be_first_in_group): Ditto.
1376 (insn_must_be_last_in_group): Ditto.
1377 (rs6000_register_move_cost): Ditto.
1378 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
1381 2017-12-07 Julia Koval <julia.koval@intel.com>
1383 * config.gcc: Add vaesintrin.h.
1384 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
1386 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
1387 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
1389 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1390 * config/i386/immintrin.h: Include vaesintrin.h.
1391 * config/i386/sse.md (vaesdec_<mode>): New pattern.
1392 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
1393 _mm_aesdec_epi128): New intrinsics.
1395 2017-12-06 David Malcolm <dmalcolm@redhat.com>
1397 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
1398 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
1399 c-family/c-spellcheck.cc.
1400 (best_macro_match::best_macro_match): Likewise.
1402 (struct edit_distance_traits<cpp_hashnode *>): Move to
1403 c-family/c-spellcheck.h.
1404 (class best_macro_match): Likewise.
1406 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1408 PR tree-optimization/83293
1409 * gimple-ssa-strength-reduction.c (insert_initializers): Use
1410 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
1411 might insert into empty bb.
1414 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
1415 simplify for plus with :c added, and pointer_plus without that.
1416 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
1417 with undefined overflow and the conversion is not widening,
1418 perform negation in utype and only convert to type afterwards.
1419 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
1420 simplify for plus with :c added, and pointer_plus without that.
1421 If type is integral with undefined overflow and the conversion is
1422 not widening, perform minus in utype and only convert to type
1423 afterwards. Move the last pointer_diff_expr simplify into the
1426 2017-12-06 Martin Sebor <msebor@redhat.com>
1428 PR tree-optimization/82646
1429 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
1430 strncpy, not maxlen.
1432 2017-12-06 Martin Sebor <msebor@redhat.com>
1434 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
1437 PR tree-optimization/83075
1438 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
1439 strncat/strncpy don't change length of source string.
1441 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
1444 2017-11-29 Martin Aberg <maberg@gaisler.com>
1446 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
1447 to prevent b2bst errata sequence.
1448 (sqrtdf2_fix): Likewise.
1450 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1452 PR tree-optimization/81945
1453 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
1454 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
1455 to dest_cfun has orig_loop_num set, either remap it to the new
1456 loop number if the loop got moved too, or clear it.
1458 2017-12-05 Steve Ellcey <sellcey@cavium.com>
1460 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
1462 (thunderx2t99_nothing): New insn reservation.
1463 (thunderx2t99_mrs): New insn reservation.
1464 (thunderx2t99_multiple): New insn reservation.
1465 (thunderx2t99_alu_basi): Add bfx to reservation.
1466 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
1468 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1471 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
1473 2017-12-05 Bin Cheng <bin.cheng@arm.com>
1475 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
1476 tree-ssa-pre.c::remove_dead_inserted_code.
1477 * tree-ssa-dce.h: New file.
1478 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
1479 (remove_dead_inserted_code): Move and rename to function
1480 tree-ssa-dce.c::simple_dce_from_worklist.
1481 (pass_pre::execute): Update use.
1483 2017-12-05 Richard Biener <rguenther@suse.de>
1485 PR tree-optimization/83277
1486 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
1487 to code-gen liveout vars.
1489 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
1491 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
1492 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
1493 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
1496 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
1498 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
1500 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
1501 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
1504 2017-12-05 Richard Biener <rguenther@suse.de>
1506 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
1507 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
1508 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
1509 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
1511 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1513 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
1515 2017-12-05 Martin Liska <mliska@suse.cz>
1516 Jakub Jelinek <jakub@redhat.com>
1518 * doc/invoke.texi: Document the options.
1519 * flag-types.h (enum sanitize_code): Add
1520 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1521 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
1522 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1523 * opts.c: Define new sanitizer options.
1524 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
1525 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
1527 2017-12-05 Julia Koval <julia.koval@intel.com>
1529 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
1530 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
1531 (ix86_handle_option): Handle -mavx512vnni.
1532 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
1533 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
1534 * config/i386/i386-c (__AVX512VNNI__): New.
1535 * config/i386/i386.c (ix86_target_string): Handle new option.
1536 (ix86_valid_target_attribute_inner_p): Handle new option.
1537 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
1538 * config/i386/i386.opt (mavx512vnni): New option.
1540 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1543 * athlon.md: Disable for generic.
1544 * haswell.md: Enable for generic.
1545 * i386.c (ix86_sched_init_global): Add core hooks for generic.
1546 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
1548 (ix86_adjust_cost): Move generic to haswell path.
1550 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
1552 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
1553 instead of MEM_P in a couple more places. Fix formatting issues.
1555 2017-12-04 Jim Wilson <jimw@sifive.com>
1557 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
1558 instead of GP_REG_LAST-1.
1559 (riscv_adjust_libcall_cfi_prologue): Likewise.
1560 (riscv_adjust_libcall_cri_epilogue): Likewise.
1561 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
1564 2017-12-04 Luis Machado <luis.machado@linaro.org>
1566 * ipa-pure-const.c (check_decl): Add missing newline.
1567 (state_from_flags): Likewise.
1569 2017-12-04 Jeff Law <law@redhat.com>
1571 PR tree-optimizatin/78496
1572 * gimple-ssa-evrp-analyze.h
1573 (evrp_range_analyzer::get_vr_values): Simplify.
1574 * gimple-ssa-evrp-analyze.c: Corresponding changes.
1575 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
1576 and gimple-ssa-evrp-analyze.h.
1577 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
1578 (simplify_stmt_for_jump_threading): Copy a blob of code from
1579 tree-vrp.c to use ranges to simplify statements.
1580 (dom_opt_dom_walker::before_dom_children): Call
1581 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
1582 (dom_opt_dom_walker::after_dom_children): Similarly for
1583 evrp_range_analyzer::leave.
1584 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
1587 * gimple-ssa-evrp-analyze.c
1588 (evrp_range_analyzer::extract_range_from_stmt): Always use
1589 vr_values::update_value_range so preexisting range info is
1590 medged with new range info, even if the new range is VR_VARYING.
1592 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1594 * combine.c: Adjust comment.
1595 (use_crosses_set_p): Delete.
1596 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
1597 (try_combine): Ditto.
1599 2017-12-04 Richard Biener <rguenther@suse.de>
1601 PR tree-optimization/83255
1602 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
1603 Re-add zero-iteration check.
1605 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1607 PR rtl-optimization/83245
1608 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
1609 hard registers as earlyclobber, also if not in an asm.
1611 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1615 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1618 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1619 rs6000_cpu based on cpu_index, not tune_index.
1621 2017-12-04 Richard Biener <rguenther@suse.de>
1623 PR tree-optimization/83238
1624 * graphite-scop-detection.c (scop_detection::merge_sese): Make
1625 code match comment, rejecting invalid SESE regions.
1627 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
1629 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
1630 require base operand is a REG_POINTER prior to reload on targets
1631 with non-equivalent space registers.
1633 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1635 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
1636 (update_specialized_profile): Fix updating of counts.
1637 (perhaps_add_new_callers): Likewise.
1639 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1642 * x86-tune.def: Remove obsolette FIXMEs.
1643 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
1644 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
1645 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
1647 (X86_TUNE_PAD_RETURNS): Disable for generic.
1649 2017-12-02 Jakub Jelinek <jakub@redhat.com>
1651 PR tree-optimization/83170
1652 PR tree-optimization/83241
1653 * gimple-ssa-store-merging.c
1654 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
1655 gimple_vuse (ins_stmt) in case it has changed.
1656 (imm_store_chain_info::output_merged_store): Likewise.
1658 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
1662 * tree-cfg.c (pass_warn_function_return::execute): Handle
1663 __builtin_ubsan_handle_missing_return like __builtin_unreachable
1664 with BUILTINS_LOCATION.
1668 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
1669 is BLKmode for vector field with vector raw mode, use TYPE_MODE
1670 instead of DECL_MODE.
1672 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
1673 last argument to rtx pointer.
1674 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
1675 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
1676 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
1677 depending on the chosen ISAs.
1678 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
1679 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
1680 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
1682 * config/i386/sse.md (mov<mode>_internal): Likewise.
1683 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
1685 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1687 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
1689 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1691 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
1692 parameters from prototype.
1693 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
1694 parameters. Don't print a comment.
1695 (emit_fusion_gpr_load): Adjust.
1696 (emit_fusion_load_store): Adjust.
1697 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
1698 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
1699 comment on the second line.
1701 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1704 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1705 rs6000_cpu based on cpu_index, not tune_index.
1707 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1709 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
1710 which_alternative instead of which_alternative + 1.
1711 (output_asm_insn): Print an extra tab if the template is short.
1713 2017-12-01 Jim Wilson <jimw@sifive.com>
1715 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
1717 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
1718 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
1720 * doc/tm.texi: Regenerate.
1722 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
1725 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
1726 whether we can allocate pseudos before trying to fix an address.
1727 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
1728 memory address is indexed or indirect.
1729 (floatuns_<mode>si2_hw2): Likewise.
1731 2017-12-01 Jason Merrill <jason@redhat.com>
1733 * Makefile.in (TAGS): Add c-family/*.cc.
1735 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
1737 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
1738 (call_value_insn): Likewise.
1739 (sibcall_insn): Likewise.
1740 (sibcall_value_insn): Likewise.
1741 (movsi_aarch64): Likewise.
1742 (movdi_aarch64): Likewise.
1743 (add_losym_): Likewise.
1744 (ldr_got_small_): Likewise.
1745 (ldr_got_small_sidi): Likewise.
1746 (ldr_got_small_28k_): Likewise.
1747 (ldr_got_small_28k_sidi): Likewise.
1748 * config/aarch64/aarch64.c (aarch64_print_address_internal):
1749 Move output_addr_const to symbolic case. Add error check.
1751 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1753 * config/s390/predicates.md (plus16_Q_operand): New predicate.
1754 * config/s390/s390.md: Disable MVC merging peephole if it would
1755 disable operand forwarding.
1756 (new peephole2): Split MVCs if it would turn them into up to 2
1759 2017-12-01 Richard Biener <rguenther@suse.de>
1761 PR tree-optimization/83232
1762 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
1763 detection of same access. Instead of breaking the group here
1764 do not consider the duplicate. Add comment explaining real fix.
1766 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
1768 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
1770 2017-12-01 Sudakshina Das <sudi.das@arm.com>
1772 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
1774 2017-12-01 Jakub Jelinek <jakub@redhat.com>
1776 * function.h (struct function): Remove cilk_frame_decl,
1777 is_cilk_function and calls_cilk_spawn fields.
1778 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
1780 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
1781 cilk_elemental field.
1782 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
1783 * target.def: Adjust comment.
1784 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
1785 Don't test cilk_elemental.
1787 PR tree-optimization/83233
1788 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
1789 bswap_stat name for the struct.
1792 * tree.h (SWITCH_BREAK_LABEL_P): Define.
1793 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
1794 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
1795 SWITCH_BREAK_LABEL_P set on the label.
1796 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
1797 added for default case if it was missing and not all cases covered.
1798 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
1799 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
1801 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
1802 to avoid -Wimplicit-fallthrough warning.
1803 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
1804 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
1806 PR tree-optimization/83221
1807 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
1809 (init_reassoc): Formatting fix.
1812 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
1813 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
1817 * tree-cfg.c: Include asan.h.
1818 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
1820 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1822 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
1824 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1826 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
1827 vpcmpeqd instruction.
1829 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1831 * config/i386/i386.c (standard_sse_constant_opcode): Fix
1832 registers type for 128bit mode.
1834 2017-11-30 Jakub Jelinek <jakub@redhat.com>
1836 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
1838 * gengtype-state.c (read_state_pair): Likewise.
1839 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
1840 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
1841 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
1842 * ifcvt.c (noce_try_store_flag_constants): Likewise.
1843 * tree-ssa-ccp.c (ccp_finalize): Likewise.
1844 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
1845 * builtins.c (fold_builtin_3): Likewise.
1846 * graphite-scop-detection.c
1847 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
1848 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
1851 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
1854 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
1855 parameter. Return true if ignore_decl_rtl_set_p is true and other
1856 conditions are satisfied.
1857 * asan.h (asan_protect_global): Add new parameter.
1858 * varasm.c (categorize_decl_for_section): Pass true as second parameter
1859 to asan_protect_global calls.
1861 2017-11-30 Jim Wilson <jimw@sifive.com>
1863 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
1864 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
1865 -mstrict-align, add info on default value. Delete redundant lines for
1866 -mabi. Add missing -mexplicit-relocs docs.
1868 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1870 * config/arc/arc.md (trap): New pattern.
1872 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1874 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
1875 instruction to end into a delay slot.
1876 * config/arc/arc.md (cond_delay_insn): Check if the instruction
1877 can be placed into a delay slot against reg_note.
1879 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1881 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
1882 labels number of usages.
1884 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1886 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
1888 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
1890 2017-11-30 Jakub Jelinek <jakub@redhat.com>
1893 * internal-fn.c (expand_mul_overflow): Optimize unsigned
1894 multiplication by power of 2 constant into two shifts + comparison.
1896 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
1899 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
1901 2017-11-30 Richard Biener <rguenther@suse.de>
1903 PR tree-optimization/83202
1904 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
1905 allow_peel argument and guard peeling.
1906 (canonicalize_loop_induction_variables): Likewise.
1907 (canonicalize_induction_variables): Pass false.
1908 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
1909 peeling from cunrolli.
1911 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1913 * combine.c (try_combine): Print a message to dump file whenever
1914 I0, I1, or I2 cannot be combined into I3.
1916 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1918 PR rtl-optimization/83156
1919 PR rtl-optimization/82621
1920 * combine.c (try_combine): Don't split an I2 if one of the dests is
1921 set again before I3. Allow unused dests.
1923 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1925 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
1927 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
1929 PR rtl-optimization/80818
1930 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
1931 recursively. Use insn code for clobber.
1932 (lra_set_insn_recog_data): Pass the new arg to
1933 collect_non_operand_hard_regs.
1934 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
1936 (lra_update_insn_regno_info): Pass insn to
1937 add_regs_to_insn_regno_info.
1939 2017-11-29 Jim Wilson <jimw@sifive.com>
1940 Andrew Waterman <andrew@sifive.com>
1942 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
1943 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
1944 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
1945 use SINGLE_SHIFT_COST.
1946 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
1947 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
1949 2017-11-29 Julia Koval <julia.koval@intel.com>
1951 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
1952 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
1953 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
1954 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
1955 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
1956 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
1957 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
1958 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
1959 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
1960 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
1961 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
1962 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
1963 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
1964 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
1965 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
1966 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
1967 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
1968 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
1969 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
1970 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
1971 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
1972 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
1973 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
1974 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
1975 __builtin_ia32_vpshldv_v2di_mask,
1976 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
1977 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
1978 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
1980 2017-11-29 Julia Koval <julia.koval@intel.com>
1982 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
1983 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
1984 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
1985 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
1986 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
1987 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
1988 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
1989 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
1990 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
1991 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
1992 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
1993 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
1994 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
1995 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
1996 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
1997 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
1998 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
1999 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
2000 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2001 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
2002 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
2004 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2006 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
2007 movsi_pic_gotdata_op instruction as a load for the UT699 errata
2010 2017-11-29 Martin Aberg <maberg@gaisler.com>
2012 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
2013 to prevent b2bst errata sequence.
2014 (sqrtdf2_fix): Likewise.
2016 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2018 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
2019 (div_sqrt_insn_p): New function.
2020 (sparc_do_work_around_errata): Insert NOP instructions to
2021 prevent sequences that could trigger the TN-0013 errata for
2022 certain LEON3 processors.
2023 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
2024 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
2025 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
2026 (in_branch_delay): Prevent div and sqrt in delay slot if
2028 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
2030 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2032 * config/sparc/sparc.c (atomic_insn_p): New function.
2033 (sparc_do_work_around_errata): Insert NOP instructions to
2034 prevent sequences that could trigger the TN-0010 errata for
2036 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
2037 instruction referable in atomic_insns_p.
2039 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2041 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
2042 (atomic_compare_and_swap_leon3_1): Likewise.
2045 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2047 * config/sparc/sparc.c (fpop_insn_p): New function.
2048 (sparc_do_work_around_errata): Insert NOP instructions to
2049 prevent sequences that could trigger the TN-0012 errata for
2051 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
2052 * config/sparc/sparc.md (fix_gr712rc): New attribute.
2053 (in_branch_annul_delay): Prevent floating-point instructions
2054 in delay slot of annulled integer branch.
2056 2017-11-29 Richard Biener <rguenther@suse.de>
2058 PR tree-optimization/83202
2059 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
2061 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
2062 nodes vectorized to the same stmts multiple times.
2063 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
2064 (vect_analyze_slp_instance): Adjust.
2065 (scalar_stmts_to_slp_tree_map_t): New typedef.
2066 (vect_schedule_slp_instance): Add a map recording the SLP node
2067 representing the vectorized stmts for a set of scalar stmts.
2068 Avoid code-generating redundancies.
2069 (vect_schedule_slp): Allocate map and pass it down.
2071 2017-11-29 Nathan Sidwell <nathan@acm.org>
2074 * tree.c (build_complex_type): Fix canonicalization. Only fill in
2077 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
2079 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
2080 cases for printing LDP/STP memory addresses.
2081 (aarch64_print_address_internal): Renamed from
2082 aarch64_print_operand_address, added parameter, add Pmode check.
2083 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
2084 (aarch64_print_operand_address): Indirect to
2085 aarch64_print_address_internal.
2086 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
2087 'y' operand output specifier.
2089 2017-11-29 Jakub Jelinek <jakub@redhat.com>
2092 * tree.c (build_simple_mem_ref_loc): Handle
2093 get_addr_base_and_unit_offset returning a MEM_REF.
2096 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
2099 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
2102 2017-11-28 David Malcolm <dmalcolm@redhat.com>
2104 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
2106 (layout::move_to_column): Likewise.
2108 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2111 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
2112 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
2114 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2115 Martin Jambor <mjambor@suse.cz>
2118 * tree.h (expr_type_first_operand_type_p): Declare
2119 * tree.c (expr_type_first_operand_type_p): New function.
2120 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
2121 (ipa_value_from_jfunc): Adjust declaration.
2122 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
2123 Use it as result type for arithmetics, unless it is NULL in which case
2124 be more conservative.
2125 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
2126 ipa_get_jf_pass_through_result.
2127 (propagate_vals_across_pass_through): Likewise.
2128 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
2129 is to propagate_vals_across_pass_through.
2130 (propagate_constants_across_call): Pass PARM_TYPE to
2131 propagate_scalar_across_jump_function.
2132 (find_more_scalar_values_for_callers_subset): Pass parameter type to
2133 ipa_value_from_jfunc.
2134 (cgraph_edge_brings_all_scalars_for_node): Likewise.
2135 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
2136 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
2137 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
2138 target_type, pass it to ipa_value_from_jfunc.
2139 (update_indirect_edges_after_inlining): Pass parameter type to
2140 try_make_edge_direct_simple_call.
2142 2017-11-28 Jeff Law <law@redhat.com>
2144 * gimple-ssa-evrp-analyze.c
2145 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
2146 refine ranges if scev_initialized_p returns true.
2147 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
2149 2017-11-28 Julia Koval <julia.koval@intel.com>
2151 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
2152 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
2153 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
2154 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
2155 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
2156 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
2157 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
2158 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
2159 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
2160 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
2161 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
2162 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
2163 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
2164 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
2165 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
2166 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
2167 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
2168 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
2169 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
2170 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
2171 __builtin_ia32_vpshrd_v2di_mask): New builtins.
2172 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
2174 2017-11-28 Julia Koval <julia.koval@intel.com>
2176 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
2177 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
2178 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
2179 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
2180 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
2181 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
2182 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
2183 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
2184 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
2185 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
2186 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
2187 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
2188 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
2189 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
2190 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
2191 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
2192 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
2193 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
2194 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
2195 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
2196 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
2197 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
2198 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
2199 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
2200 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
2201 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
2202 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
2203 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
2204 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
2205 __builtin_ia32_vpshld_v2di_mask): New builtins.
2206 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2207 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
2209 2017-11-28 Richard Biener <rguenther@suse.de>
2211 PR tree-optimization/80776
2212 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
2214 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
2216 (evrp_range_analyzer::record_ranges_from_incoming_edges):
2217 If the incoming edge is an effective fallthru because the other
2218 edge only reaches a __builtin_unreachable () then record ranges
2219 derived from the controlling condition in SSA info.
2220 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
2221 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
2223 2017-11-28 Olivier Hainque <hainque@adacore.com>
2225 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
2226 on mingw build hosts.
2228 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2231 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
2233 2017-11-28 Richard Biener <rguenther@suse.de>
2236 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
2237 copies generated from memcpy use a character array as reference
2240 2017-11-28 Julia Koval <julia.koval@intel.com>
2241 Sebastian Peryt <sebastian.peryt@intel.com>
2243 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
2244 c-family/c-cilkplus.o, c-family/array-notation-common.o,
2245 cilk-common.o, cilk.h, cilk-common.c): Remove.
2247 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
2248 * builtins.c (is_builtin_name): Remove cilkplus condition.
2249 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
2250 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
2251 cilk-builtins.def, cilkplus.def): Remove.
2252 * cif-code.def (CILK_SPAWN): Remove.
2253 * cilk-builtins.def: Delete.
2254 * cilk-common.c: Ditto.
2256 * cilkplus.def: Ditto.
2257 * config/darwin.h (fcilkplus): Delete.
2258 * cppbuiltin.c: Ditto.
2259 * doc/extend.texi: Remove cilkplus doc.
2260 * doc/generic.texi: Ditto.
2261 * doc/invoke.texi: Ditto.
2262 * doc/passes.texi: Ditto.
2263 * gcc.c (fcilkplus): Remove.
2264 * gengtype.c (cilk.h): Remove.
2265 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
2267 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
2269 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
2271 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
2272 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
2273 cilkplus conditions.
2274 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
2275 inline_read_section): Ditto.
2276 * ipa-inline-analysis.c (cilk.h): Remove.
2277 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
2278 * lto-wrapper.c (merge_and_complain, append_compiler_options,
2279 append_linker_options): Remove condition for fcilkplus.
2280 * lto/lto-lang.c (cilk.h): Remove.
2281 (lto_init): Remove condition for fcilkplus.
2282 * omp-expand.c (expand_cilk_for_call): Delete.
2283 (expand_omp_taskreg, expand_omp_for_static_chunk,
2284 expand_omp_for): Remove cilkplus
2286 (expand_cilk_for): Delete.
2287 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
2288 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
2289 execute_lower_omp, diagnose_sb_0): Ditto.
2290 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
2291 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
2292 * tree-nested.c: Ditto.
2293 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
2294 (dump_generic_node): Ditto.
2295 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
2296 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
2298 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
2300 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2302 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
2303 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
2304 for vector mode and !TARGET_SIMD.
2306 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2308 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
2310 * tree.h (SWITCH_LABELS): Remove.
2311 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
2312 assert SWITCH_BODY is non-NULL.
2313 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
2315 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
2317 PR tree-optimization/80788
2318 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
2319 has TREE_OVERFLOW set, call drop_tree_overflow.
2321 2017-11-28 Richard Biener <rguenther@suse.de>
2323 PR tree-optimization/83158
2324 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
2326 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
2329 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
2330 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
2332 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
2334 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
2335 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
2337 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2340 * varasm.c (compare_constant): Take the mode of the constants into
2341 account when comparing floating point constants.
2343 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
2345 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
2346 from explicit instantiation of debug_helper.
2347 * vec.h (DEFINE_DEBUG_VEC): Ditto.
2349 2017-11-27 Richard Biener <rguenther@suse.de>
2351 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
2354 2017-11-27 Richard Biener <rguenther@suse.de>
2356 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
2357 (type_hash_canon): Also clear int_cst_hash_table entry for
2359 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
2361 2017-11-27 Tamar Christina <tamar.christina@arm.com>
2363 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2364 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2365 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2367 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
2369 * hash-map.h (gt_cleare_cache): Avoid UB.
2371 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
2373 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
2374 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
2375 (decide_unroll_runtime_iterations): Likewise.
2376 (decide_unroll_stupid): Likewise.
2378 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2381 * config/i386/i386.md: Add a loop with incssp.
2383 2017-11-27 Martin Jambor <mjambor@suse.cz>
2385 PR tree-optimization/81248
2386 * tree-sra.c (splice_param_accesses): Remove size check.
2387 (decide_one_param_reduction): Fix size check.
2388 * gimple-pretty-print.c (dump_profile): Silence warning.
2389 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
2391 2017-11-27 Jakub Jelinek <jakub@redhat.com>
2394 * dbxout.c (lastlineno): New variable.
2395 (dbx_debug_hooks): Use dbxout_switch_text_section as
2396 switch_text_section debug hook.
2397 (dbxout_function_end): Switch to current_function_section
2398 rather than function_section. If crtl->has_bb_partition,
2399 output just one N_FUN, depending on in_cold_section_p.
2400 (dbxout_source_line): Remember last lineno in lastlineno.
2401 (dbxout_switch_text_section): New function.
2402 (dbxout_function_decl): Adjust dbxout_block caller.
2403 (dbx_block_with_cold_children): New function.
2404 (dbxout_block): Return true if any LBRAC/RBRAC have been
2405 emitted. Use dbx_block_with_cold_children at depth == 0
2406 in second partition. Add PARENT_BLOCKNUM argument, pass
2407 it optionally adjusted to children. Output LBRAC/RBRAC
2408 around recursive call only if the block is in the current
2409 partition, if not and anything was output, emit empty
2411 * final.c (final_scan_insn): Compute cold_function_name
2412 before calling switch_text_section debug hook. Call
2413 that hook even if dwarf2out_do_frame if not emitting
2417 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
2418 TREE_READONLY decls.
2420 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
2422 PR rtl-optimization/82488
2423 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
2425 2017-11-26 Julia Koval <julia.koval@intel.com>
2427 * config/i386/i386.c (processor_target_table): Add skylake_cost for
2429 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
2432 2017-11-26 Julia Koval <julia.koval@intel.com>
2434 * config/i386/driver-i386.c (host_detect_local_cpu):
2435 Detect skylake-avx512.
2437 2017-11-26 Julia Koval <julia.koval@intel.com>
2439 * config.gcc: Add -march=cannonlake.
2440 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
2441 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
2442 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
2443 (PTA_CANNONLAKE): New.
2444 (processor_target_table): Add cannonlake.
2445 (ix86_option_override_internal): Ditto.
2446 (fold_builtin_cpu): Ditto.
2447 (get_builtin_code_for_version): Handle cannonlake.
2448 (M_INTEL_COREI7_CANNONLAKE): New.
2449 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
2450 * doc/invoke.texi: Add -march=cannonlake.
2452 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
2454 * plugin.c (add_new_plugin): Use platform-specific library extensions.
2455 (try_init_one_plugin): Alternative implementation for MinGW.
2456 * Makefile.in (plugin_implib): New.
2457 (gengtype-lex.c): Fix broken AIX workaround.
2458 * configure: Regenerate.
2459 * doc/plugins.texi: Document support for MinGW.
2461 2017-11-25 Jakub Jelinek <jakub@redhat.com>
2463 PR rtl-optimization/81553
2464 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
2465 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
2466 is a shift where C1 has different mode than the whole shift, use C1's
2467 mode for MULT rather than the shift's mode.
2470 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
2471 builtins not enabled in the currently selected ISA.
2473 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
2475 PR tree-optimization/71026
2476 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
2477 (insert_reciprocals): Change to insert reciprocals before a division
2478 by a square and to insert the square of a reciprocal.
2479 (execute_cse_reciprocals_1): Change to consider division by a square.
2480 (register_division_in): Add importance parameter.
2482 2017-11-24 Richard Biener <rguenther@suse.de>
2484 PR tree-optimization/82402
2485 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
2486 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
2488 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
2490 * match.pd (0-ptr): New transformation.
2492 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
2495 * ipa-inline.c (inline_small_functions): Set current badnes correctly
2496 when skipping checking.
2498 2017-11-24 Richard Biener <rguenther@suse.de>
2500 PR tree-optimization/83128
2501 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
2502 (vn_reference_lookup_3): Likewise.
2504 2017-11-24 Jakub Jelinek <jakub@redhat.com>
2507 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
2508 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
2511 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
2512 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
2513 check for the latter with gimple_call_builtin_p. Do not handle
2514 BUILT_IN_STPNCPY_CHK which is not a pass through call.
2516 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
2518 * config/arm/arm_neon.h: Fix pragma GCC push_options before
2521 2017-11-23 Julia Koval <julia.koval@intel.com>
2523 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
2524 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
2525 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
2526 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
2527 _mm512_maskz_expandloadu_epi16): New intrinsics.
2528 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
2529 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
2530 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
2531 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
2532 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
2533 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
2534 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
2535 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
2536 _mm256_maskz_expandloadu_epi8): New intrinsics.
2537 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
2538 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
2539 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
2540 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
2541 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
2542 * config/i386/sse.md (VI248_VLBW): New iterator.
2543 (expand<mode>_mask, expand<mode>_maskz): New patterns.
2545 2017-11-23 Julia Koval <julia.koval@intel.com>
2547 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
2548 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
2549 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
2550 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
2551 _mm512_mask_compressstoreu_epi16): New.
2552 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
2553 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
2554 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
2555 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
2556 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
2557 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
2558 _mm256_mask_compressstoreu_epi8): New.
2559 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
2560 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
2561 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
2562 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
2563 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
2564 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
2565 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
2566 __builtin_ia32_compresshi128_mask,
2567 __builtin_ia32_compressstoreuqi512_mask,
2568 __builtin_ia32_compressstoreuhi512_mask,
2569 __builtin_ia32_compressstoreuqi256_mask,
2570 __builtin_ia32_compressstoreuqi128_mask,
2571 __builtin_ia32_compressstoreuhi256_mask,
2572 __builtin_ia32_compressstoreuhi128_mask): New builtins.
2573 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
2575 (ix86_expand_special_args_builtin): Handle new types.
2576 (s4fma_expand): Handle new builtin array.
2577 * config/i386/immintrin.h: Include new headers.
2578 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
2579 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
2581 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2583 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
2584 fixes. Declare temp and g variables at the top in order to avoid
2585 {} in most of the cases.
2587 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
2589 * match.pd (ptr-0): New transformation.
2591 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2593 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
2594 (struct addr_mode_cost_table): New.
2595 (struct tune_params): Add field addr_mode_costs.
2596 * config/arm/arm.c (generic_addr_mode_costs): New.
2597 (arm_slowmul_tune): Initialise addr_mode_costs field.
2598 (arm_fastmul_tune): Likewise.
2599 (arm_strongarm_tune): Likewise.
2600 (arm_xscale_tune): Likewise.
2601 (arm_9e_tune): Likewise.
2602 (arm_marvell_pj4_tune): Likewise.
2603 (arm_v6t2_tune): Likewise.
2604 (arm_cortex_tune): Likewise.
2605 (arm_cortex_a8_tune): Likewise.
2606 (arm_cortex_a7_tune): Likewise.
2607 (arm_cortex_a15_tune): Likewise.
2608 (arm_cortex_a35_tune): Likewise.
2609 (arm_cortex_a53_tune): Likewise.
2610 (arm_cortex_a57_tune): Likewise.
2611 (arm_exynosm1_tune): Likewise.
2612 (arm_xgene1_tune): Likewise.
2613 (arm_cortex_a5_tune): Likewise.
2614 (arm_cortex_a9_tune): Likewise.
2615 (arm_cortex_a12_tune): Likewise.
2616 (arm_cortex_a73_tune): Likewise.
2617 (arm_v7m_tune): Likewise.
2618 (arm_cortex_m7_tune): Likewise.
2619 (arm_v6m_tune): Likewise.
2620 (arm_fa726te_tune): Likewise.
2621 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
2623 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2625 * config/arm/arm.c (arm_mem_costs): New function.
2626 (arm_rtx_costs_internal): Use arm_mem_costs.
2628 2017-11-23 Mark Wielaard <mark@klomp.org>
2630 * dwarf2out.c (init_sections_and_labels): Use generation to create
2631 unique ranges_section_label and ranges_base_label. Return generation.
2632 (output_rnglists): Add generation argument. Use generation to create
2633 unique ranges labels.
2634 (dwarf2out_finish): Get generation from init_sections_and_labels
2635 and pass generation to output_rnglists.
2637 2017-11-23 Mike Stump <mikestump@comcast.net>
2638 Eric Botcazou <ebotcazou@adacore.com>
2640 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
2641 * cfgloop.h (struct loop): Add unroll field.
2642 * function.h (struct function): Add has_unroll bitfield.
2643 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
2644 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
2645 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
2646 (pass_rtl_unroll_loops::gate): Likewise.
2647 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
2648 for which loop->unroll==1.
2649 (decide_unroll_constant_iterations): Use note for consistency and
2650 take loop->unroll into account. Return early if loop->unroll is set.
2651 Fix thinko in existing test.
2652 (decide_unroll_runtime_iterations): Use note for consistency and
2653 take loop->unroll into account.
2654 (decide_unroll_stupid): Likewise.
2655 * lto-streamer-in.c (input_cfg): Read loop->unroll.
2656 * lto-streamer-out.c (output_cfg): Write loop->unroll.
2657 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
2659 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
2660 (print_loop): Print loop->unroll if set.
2661 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
2662 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
2663 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
2665 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
2666 loop->unroll is set and smaller than the trip count. Otherwise bypass
2667 entirely the heuristics if loop->unroll is set. Remove dead note.
2668 Fix off-by-one bug in other note.
2669 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
2670 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
2672 (tree_unroll_loops_completely): Make static.
2673 (pass_complete_unroll::execute): Use correct type for variable.
2674 (pass_complete_unrolli::execute): Fix formatting.
2675 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
2677 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
2679 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
2680 enable when TARGET_PREFER_AVX128 is set.
2682 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2684 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
2686 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
2687 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
2689 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2691 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
2692 frequencies. Fix estimation of aggregate parameters.
2694 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2696 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
2699 2017-11-23 Tom de Vries <tom@codesourcery.com>
2701 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
2703 2017-11-23 Tom de Vries <tom@codesourcery.com>
2705 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
2708 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
2711 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
2712 sibcall_value_pcrel_fdpic): Use local variable instead of
2714 (calli_tbr_rel): Add missing operand 2.
2715 (call_valuei_tbr_rel): Add missing operand 3.
2717 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2720 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
2721 bitpos/bitsize covering the whole destination, use store_expr only if
2722 the complex mode is the same. Otherwise, use expand_normal and if
2723 it returns CONCAT, subreg each part separately instead of trying to
2724 subreg the whole result.
2726 2017-11-23 Richard Biener <rguenther@suse.de>
2728 PR tree-optimization/23094
2729 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
2730 come here from walking over backedges in the first iteration.
2731 (vn_reference_lookup_3): Skip clobbers that store the same value.
2733 2017-11-23 Richard Biener <rguenther@suse.de>
2735 PR tree-optimization/81403
2736 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
2737 a block we need a leader relative to.
2738 (phi_translate_1): For nary processing require a leader from
2739 get_representative_for given we run expression simplification
2740 using match-and-simplify. Remove previous fix.
2742 2017-11-22 Jeff Law <law@redhat.com>
2744 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
2745 Use new method allocate_value_range rather than accessing the
2746 vrp_value_range_pool data member directly.
2747 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
2748 to use extract_range_from_stmt method to avoid need for
2749 extract_range_from_assignment method.
2750 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
2751 method rather than setting values_propgated data member directly.
2752 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
2753 and values propagated data members and extract_range_from_assignment
2754 method. Reorder private data members to conform to standards.
2755 Add new methods set_lattice_propagation_complete and
2756 allocate_value_range.
2758 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
2760 PR rtl-optimization/83030
2761 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
2762 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
2763 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
2764 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
2765 CROSSING_JUMP_P flag is already set before setting it.
2766 * cfgrtl.c (fixup_partition_crossing): Likewise.
2767 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
2770 2017-11-22 Jakub Jelinek <jakub@redhat.com>
2772 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
2773 Handle the case where both arguments are using gen_const_vec_series.
2775 2017-11-22 David Malcolm <dmalcolm@redhat.com>
2778 * pretty-print.c (pp_format): Move quoting implementation to
2779 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
2780 to pass address of "quote" local.
2781 (pp_begin_quote): New function.
2782 (pp_end_quote): New function.
2783 * pretty-print.h (printer_fn): Convert penultimate param from bool
2785 (pp_begin_quote): New decl.
2786 (pp_end_quote): New decl.
2787 * tree-diagnostic.c (default_tree_printer): Convert penultimate
2788 param from bool to bool *.
2789 * tree-diagnostic.h (default_tree_printer): Likewise.
2791 2017-11-22 Jeff Law <law@redhat.com>
2793 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
2794 Initialize vr_values.
2795 (evrp_range_analyzer::try_find_new_range): Call methods attached to
2796 vr_values via vr_values class instance rather than delegators.
2797 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
2798 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
2799 (evrp_range_analyzer::push_value_range): Likewise.
2800 (evrp_range_analyzer::pop_value_range): Likewise.
2801 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
2802 most delegators. Those remaining are exposed as public interfaces.
2803 Make vr_values a pointer and private.
2804 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
2806 (evrp_range_analyzer::get_vr_value): New method.
2807 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
2808 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
2809 (class evrp_dom_walker): Attach evrp_folder class, initialize
2810 it in the ctor. Remove temporary delegators.
2811 (evrp_dom_walker::before_dom_children): Call methods in attached
2812 evrp_range_analyzer class via class instance pointer. Use
2813 free value_range_constant_singleton to remove need for
2814 op_with_constant_singleton_value delegator method. Do not
2815 create a vrp_prop class instance for every call! Narrow
2816 scope of a couple variables.
2817 (evrp_dom_walker::cleanup): Call methods in attached
2818 evrp_range_analyzer class via class instance pointer.
2819 * vr-values.h (class vr_values): Privatize many methods and
2822 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2824 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
2825 * cfgexpand.c (expand_debug_expr): Remove handling for them.
2826 * expr.c (expand_expr_real_2): Likewise.
2827 * fold-const.c (const_unop): Likewise.
2828 * optabs-tree.c (optab_for_tree_code): Likewise.
2829 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2830 * tree-inline.c (estimate_operator_cost): Likewise.
2831 * tree-pretty-print.c (dump_generic_node): Likewise.
2832 (op_code_prio): Likewise.
2833 (op_symbol_code): Likewise.
2834 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
2835 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
2836 * internal-fn.c (direct_internal_fn_optab): New function.
2837 (direct_internal_fn_array, direct_internal_fn_supported_p
2838 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
2839 * fold-const-call.c (fold_const_reduction): New function.
2840 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
2842 * tree-vect-loop.c: Include internal-fn.h.
2843 (reduction_code_for_scalar_code): Rename to...
2844 (reduction_fn_for_scalar_code): ...this and return an internal
2846 (vect_model_reduction_cost): Take an internal_fn rather than
2848 (vect_create_epilog_for_reduction): Likewise. Build calls rather
2850 (vectorizable_reduction): Use internal functions rather than tree
2851 codes for the reduction operation. Update calls to the functions
2853 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
2854 Use calls to internal functions rather than REDUC tree codes.
2855 * config/aarch64/aarch64-simd.md: Update comment accordingly.
2857 2017-11-22 Olivier Hainque <hainque@adacore.com>
2859 * config/vxworks.c (vxworks_override_options): Pick default
2860 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
2861 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
2862 DWARF_GNAT_ENCODINGS_DEFAULT.
2863 * config/vxworksae.h: Likewise.
2865 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
2867 PR tree-optimization/83104
2868 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
2871 2017-11-22 Marek Polacek <polacek@redhat.com>
2872 H.J. Lu <hongjiu.lu@intel.com>
2873 Jason Merrill <jason@redhat.com>
2878 * calls.c (initialize_argument_information): Call
2879 warn_parameter_passing_abi target hook.
2880 (store_one_arg): Use 0 for empty record size. Don't push 0 size
2881 argument onto stack.
2882 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
2883 * common.opt: Update -fabi-version description.
2884 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
2885 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
2887 (ix86_is_empty_record): New function.
2888 (ix86_warn_parameter_passing_abi): New function.
2889 (TARGET_EMPTY_RECORD_P): Redefine.
2890 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
2891 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
2892 * doc/tm.texi: Regenerated.
2893 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
2894 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
2895 * dwarf2out.c (get_ultimate_context): Move to tree.c.
2896 * explow.c (hard_function_value): Call arg_int_size_in_bytes
2897 instead of int_size_in_bytes.
2898 * expr.c (copy_blkmode_to_reg): Likewise.
2899 * function.c (aggregate_value_p): Return 0 for empty types.
2900 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
2902 (locate_and_pad_parm): Call arg size_in_bytes instead
2904 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
2905 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
2906 * target.def (empty_record_p, warn_parameter_passing_abi): New target
2908 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
2909 (std_gimplify_va_arg_expr): Skip empty records. Call
2910 arg_size_in_bytes instead size_in_bytes.
2911 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
2912 * tree-core.h (tree_type_common): Add empty_flag.
2913 (tree_decl_common): Update comments.
2914 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
2916 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
2917 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
2919 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
2920 * tree.c (default_is_empty_type): New function.
2921 (default_is_empty_record): New function.
2922 (arg_int_size_in_bytes): New function.
2923 (arg_size_in_bytes): New function.
2924 (get_ultimate_context): New function.
2925 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
2926 TRANSLATION_UNIT_WARN_EMPTY_P.
2927 (default_is_empty_record, arg_int_size_in_bytes,
2928 arg_size_in_bytes, get_ultimate_context): Declare.
2930 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2932 * config/arm/arm.c (cmse_clear_registers): New function.
2933 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
2934 code by call to cmse_clear_registers.
2935 (cmse_nonsecure_entry_clear_before_return): Likewise.
2937 2017-11-22 Tamar Christina <tamar.christina@arm.com>
2939 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
2940 (vdot_s32, vdotq_s32): New.
2941 (vdot_lane_u32, vdotq_lane_u32): New.
2942 (vdot_lane_s32, vdotq_lane_s32): New.
2945 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2948 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
2949 for unsigned values with fewer HWIs than the precision.
2950 (test_overflow): New function.
2951 (wide_int_cc_tests): Call it.
2953 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2955 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
2956 has the same mode class as Pmode.
2958 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2960 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
2961 padding_bits_to_clear_ptr.
2962 (cmse_nonsecure_entry_clear_before_return): Likewise.
2964 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2966 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
2967 auto_sbitap instead of integer bitfield to control register needing
2970 2017-11-22 Jakub Jelinek <jakub@redhat.com>
2972 PR tree-optimization/83044
2973 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
2974 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
2975 that need the upper bound. Subtract offset from
2976 get_addr_base_and_unit_offset only if positive and subtract it
2977 before division by eltsize rather than after it.
2980 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
2981 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
2982 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
2986 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
2987 expand_binop with *mul_widen_optab, make sure at least one of the
2988 operands doesn't have VOIDmode.
2991 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
2993 PR rtl-optimization/82044
2994 PR tree-optimization/82042
2995 * dse.c (record_store): Check for overflow.
2996 (check_mem_read_rtx): Properly check for overflow if width == -1, call
2997 add_wild_read instead of clear_rhs_from_active_local_stores on
2998 overflow and log it into dump_file.
3000 2017-11-22 Richard Biener <rguenther@suse.de>
3002 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
3003 fake edges to exit when looking for a place to insert.
3004 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
3006 (insert_into_preds_of_block): Commit edge insertion immediately,
3007 assert that doesn't require new BBs.
3008 (fini_pre): Release expressions.
3009 (pass_pre::execute): Shuffle things around a bit, if the fn
3010 is too large do not compute AVAIL either as this is really the
3013 2017-11-22 Richard Biener <rguenther@suse.de>
3015 PR tree-optimization/83089
3016 * tree-if-conv.c (pass_if_conversion::execute): If anything
3017 changed reset SCEV and free the number of iteration estimates.
3019 2017-11-21 Martin Sebor <msebor@redhat.com>
3021 PR tree-optimization/82945
3022 * calls.h (warn_nonstring_bound): Remove unused function.
3024 2017-11-21 Martin Sebor <msebor@redhat.com>
3026 PR tree-optimization/82945
3027 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
3028 * calls.h (maybe_warn_nonstring_arg): Declare new function.
3029 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
3031 (initialize_argument_information): Call maybe_warn_nonstring_arg.
3032 * calls.h (get_attr_nonstring_decl): Declare new function.
3033 * doc/extend.texi (attribute nonstring): Update.
3034 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
3035 get_attr_nonstring_decl and handle it.
3036 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
3037 detection of nul-termination.
3038 (strlen_to_stridx): Change to a pointer.
3039 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
3040 (pass_strlen::execute): Same.
3042 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
3044 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
3045 for the new option -mprefer-vector-width=[none|128|256|512].
3046 * config/i386/i386.c (ix86_target_string): remove old style options
3047 -mprefer-avx256 and make -mprefer-avx128 as alias.
3048 (ix86_option_override_internal): Apply defaults for the
3049 -mprefer-vector-width=[128|256] option.
3050 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
3051 Implement macros to work with -mprefer-vector-width=.
3052 * config/i386/i386.opt: Implemented option
3053 -mprefer-vector-width=[none|128|256|512].
3054 * doc/invoke.texi: Documentation for
3055 -mprefer-vector-width=[none|128|256|512].
3057 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
3059 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
3061 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
3063 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
3065 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
3067 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
3069 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
3070 * expr.c (expand_expr_real_2): Likewise.
3071 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
3072 fold_binary_loc): Likewise.
3073 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
3074 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
3075 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
3076 MINUS_EXPR transformations.
3077 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
3078 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
3079 * tree-inline.c (estimate_operator_cost): Likewise.
3080 * tree-pretty-print.c (dump_generic_node, op_code_prio,
3081 op_symbol_code): Likewise.
3082 * tree-vect-stmts.c (vectorizable_operation): Likewise.
3083 * vr-values.c (extract_range_from_binary_expr): Likewise.
3084 * varasm.c (initializer_constant_valid_p_1): Likewise.
3085 * tree.def: New tree code POINTER_DIFF_EXPR.
3087 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
3089 * config/i386/i386.md (*bswap<mode>2_movbe): Add
3090 integer suffix to movbe mnemonic.
3091 (*bswaphi2_movbe): Ditto.
3092 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
3094 2017-11-21 Jakub Jelinek <jakub@redhat.com>
3097 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
3098 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
3099 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
3101 2017-11-21 Martin Liska <mliska@suse.cz>
3103 * tree-inline.c (expand_call_inline): Remove not needed
3106 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
3107 Jakub Jelinek <jakub@redhat.com>
3110 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
3111 Remove static keyword from f variable.
3113 2017-11-21 Jakub Jelinek <jakub@redhat.com>
3115 PR tree-optimization/83086
3116 * gimple-ssa-store-merging.c
3117 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
3118 rather than n.base_addr.
3120 2017-11-21 Martin Liska <mliska@suse.cz>
3122 PR rtl-optimization/82044
3123 PR tree-optimization/82042
3124 * dse.c (check_mem_read_rtx): Check for overflow.
3126 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
3128 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
3131 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
3133 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
3134 and its corresponding call arg location note.
3136 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
3137 Andrew Burgess <andrew.burgess@embecosm.com>
3139 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
3141 (arc_return_slot_offset): Likewise.
3142 (arc_eh_return_address_location): New declaration.
3143 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
3144 (MUST_SAVE_REGISTER): Add exception handler case.
3145 (MUST_SAVE_RETURN_ADDR): Likewise.
3146 (arc_frame_pointer_required): Likewise.
3147 (arc_frame_pointer_needed): New function.
3148 (arc_compute_frame_size): Changed.
3149 (arc_expand_prologue): Likewise.
3150 (arc_expand_epilogue): Likewise.
3151 (arc_initial_elimination_offset): Likewise.
3152 (arc_return_slot_offset): Delete.
3153 (arc_eh_return_address_location): New function.
3154 (arc_builtin_setjmp_frame_value): Likewise.
3155 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
3156 (EH_RETURN_STACKADJ_RTX): Define.
3157 (EH_RETURN_HANDLER_RTX): Likewise.
3158 * config/arc/arc.md (eh_return): Delete.
3160 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
3162 * print-tree.h (debug_vec_tree): Remove prototype.
3163 * gdbinit.in (pvt): Remove macro.
3165 2017-11-21 Jakub Jelinek <jakub@redhat.com>
3167 PR tree-optimization/83047
3168 * gimple-ssa-store-merging.c
3169 (imm_store_chain_info::output_merged_store): If the loads with the
3170 same vuse are in different basic blocks, for load_gsi pick a load
3171 location that is dominated by the other loads.
3174 * config/i386/i386.c (ix86_memmodel_check): Start
3175 -Winvalid-memory-model diagnostics with lowercase letter.
3178 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
3179 set in_cold_section_p to the partition containing loc_list->first.
3180 When seeing loc_list->last_before_switch node, update secname and
3181 perform range_across_switch second partition handling only after that.
3184 * run-rtl-passes.c: Include debug.h.
3185 (run_rtl_passes): Call debug_hooks->assembly_start.
3186 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
3190 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
3191 OPTAB_DIRECT in calls to expand_simple_binop.
3193 2017-11-20 David Malcolm <dmalcolm@redhat.com>
3196 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
3198 2017-11-20 David Malcolm <dmalcolm@redhat.com>
3201 * spellcheck.h (best_match::blithely_get_best_candidate): New
3204 2017-11-20 Jakub Jelinek <jakub@redhat.com>
3206 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
3207 with lower case letter.
3209 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
3211 * config/i386/i386.md (bswaphi2): New expander.
3212 (*bswaphi2_movbe): New insn pattern.
3213 (bswaphi -> rorhi pepehole2): New peephole pattern.
3215 2017-11-20 Jeff Law <law@redhat.com>
3217 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
3218 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
3219 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
3220 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
3221 gimple-ssa-evrp-analyze.h.
3223 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
3224 set BB_VISITED here.
3225 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
3227 2017-11-20 Bin Cheng <bin.cheng@arm.com>
3229 * tree-predcom.c: Add general comment on Store-Store chains.
3230 (split_data_refs_to_components): Postpone clearing eliminate_store_p
3232 (get_chain_last_ref_at): Rename into...
3233 (get_chain_last_write_at): ...this.
3234 (get_chain_last_write_before_load): New function.
3235 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
3236 CT_STORE_STORE when write reference is added.
3237 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
3238 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
3239 (initialize_root_vars_store_elim_1): Ditto.
3240 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
3241 definition is created.
3242 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
3243 chain by replacing it with dominant stored value.
3245 2017-11-20 Bin Cheng <bin.cheng@arm.com>
3247 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
3249 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
3251 * vr-values.c (extract_range_from_binary_expr): Use a full range
3254 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
3256 * config/arm/arm.md (R4_REGNUM): Define constant.
3257 (nonsecure_call_internal): Remove r4 clobber.
3258 (nonsecure_call_value_internal): Likewise.
3259 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
3260 clobber and resequence match_operands.
3261 (nonsecure_call_value_reg_thumb1_v5): Likewise.
3262 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
3263 (nonsecure_call_value_reg_thumb2): Likewise.
3265 2017-11-20 Jakub Jelinek <jakub@redhat.com>
3267 PR tree-optimization/78821
3268 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
3269 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
3270 to the address of the base rather than the base itself.
3271 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
3272 (find_bswap_or_nop_finalize): New function.
3273 (find_bswap_or_nop): Use it.
3274 (bswap_replace): Return a tree rather than bool, change first
3275 argument from gimple * to gimple_stmt_iterator, allow inserting
3276 into an empty sequence, allow ins_stmt to be NULL - then emit
3277 all stmts into gsi. Fix up MEM_REF address gimplification.
3278 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
3279 (struct store_immediate_info): Add N and INS_STMT non-static
3281 (store_immediate_info::store_immediate_info): Initialize them
3282 from newly added ctor args.
3283 (merged_store_group::apply_stores): Formatting fixes. Sort by
3285 (stmts_may_clobber_ref_p): For stores call also
3286 refs_anti_dependent_p.
3287 (gather_bswap_load_refs): New function.
3288 (imm_store_chain_info::try_coalesce_bswap): New method.
3289 (imm_store_chain_info::coalesce_immediate_stores): Use it.
3290 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
3291 (imm_store_chain_info::output_merged_store): Fail if number of
3292 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
3293 and NOP_EXPR rhs_code.
3294 (pass_store_merging::process_store): Compute n and ins_stmt, if
3295 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
3296 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
3298 (pass_store_merging::execute): Calculate dominators.
3300 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
3301 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
3302 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
3303 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
3304 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
3305 class pass_optimize_bswap, bswap_replace,
3306 pass_optimize_bswap::execute): Moved to ...
3307 * gimple-ssa-store-merging.c: ... this file.
3308 Include optabs-tree.h.
3309 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
3310 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
3311 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
3312 anonymous namespace, remove static keywords.
3313 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
3314 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
3316 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
3319 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
3321 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
3323 * vec.h (debug_helper): New function.
3324 (DEFINE_DEBUG_VEC): New macro.
3325 * hash-set.h (debug_helper): New function.
3326 (DEFINE_DEBUG_HASH_SET): New macro.
3327 * cfg.c (debug_slim (edge)): New function.
3328 Call DEFINE_DEBUG_VEC for edges.
3329 Call DEFINE_DEBUG_HASH_SET for edges.
3330 * cfghooks.c (debug_slim (basic_block)): New function.
3331 Call DEFINE_DEBUG_VEC for basic blocks.
3332 Call DEFINE_DEBUG_HASH_SET for basic blocks.
3333 * print-tree.c (debug_slim): New function to handle trees.
3334 Call DEFINE_DEBUG_VEC for trees.
3335 Call DEFINE_DEBUG_HASH_SET for trees.
3336 (debug (vec<tree, va_gc>) &): Remove.
3337 (debug (<vec<tree, va_gc>) *): Remove.
3338 * print-rtl.c (debug_slim): New function to handle const_rtx.
3339 Call DEFINE_DEBUG_VEC for rtx_def.
3340 Call DEFINE_DEBUG_VEC for rtx_insn.
3341 Call DEFINE_DEBUG_HASH_SET for rtx_def.
3342 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
3343 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
3344 (debug (vec<rtx_insn *> *ptr): Remove.
3345 (debug_insn_vector): Remove.
3346 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
3348 2017-11-20 Tom de Vries <tom@codesourcery.com>
3350 PR rtl-optimization/82020
3351 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
3352 IF_THEN_ELSE condition.
3354 2017-11-19 Jeff Law <law@redhat.com>
3356 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
3357 of degenerates resulting from ignoring an edge.
3359 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3362 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
3364 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3367 * profile-count.c (profile_count::to_sreal_scale): Fix return value
3368 for uninitialied counts.
3370 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3373 * tree-inline.c (estimate_num_insns): Set to 1 at least.
3375 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3378 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
3380 2017-11-19 Tom de Vries <tom@codesourcery.com>
3382 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
3385 2017-11-19 Tom de Vries <tom@codesourcery.com>
3387 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
3390 2017-11-19 Tom de Vries <tom@codesourcery.com>
3392 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
3395 2017-11-19 Tom de Vries <tom@codesourcery.com>
3397 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
3399 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
3401 2017-11-19 Tom de Vries <tom@codesourcery.com>
3403 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
3405 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
3406 semicolon after MCORE_EXPORT_NAME call.
3408 2017-11-19 Tom de Vries <tom@codesourcery.com>
3411 * vmsdbgout.c (vmsdbgout_early_finish): New function.
3412 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
3414 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
3416 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
3417 (cgraph_edge::clone): Cleanup updating of profile.
3418 * ipa-cp.c (update_profiling_info): Likewise.
3419 * ipa-inline-transform.c (inline_transform): Likewise.
3420 * ipa-inline.c (inline_small_functions): Add missing space to dump.
3421 * ipa-split.c (execute_split_functions): Do not split when function
3423 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
3424 * profile-count.c (profile_count::dump): Add global0.
3425 (profile_count::to_cgraph_frequency): Do not ICE when entry is
3427 (profile_count::to_sreal_scale): Likewise.
3428 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
3429 (profile_count::combine_with_ipa_count): New function.
3430 * profile-count.h (profile_guessed_global0adjusted): New.
3431 (profile_count::adjusted_zero): New.
3432 (profile_count::global0adjusted): New.
3433 (profile_count::combine_with_ipa_count): New.
3434 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
3435 correct profile of return block of split functions.
3436 (copy_cfg_body): Remove unused profile_count.
3437 (copy_body): Likewise.
3438 (expand_call_inline): Update.
3439 (tree_function_versioning): Update.
3441 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
3443 * hash-set.h (hash_set::empty): New.
3444 * tree-ssa-threadbackward.h: Delete.
3445 * tree-ssa-threadbackward.c (class thread_jumps): New.
3446 Move max_threaded_paths into class.
3447 (fsm_find_thread_path): Remove arguments that are now in class.
3448 (profitable_jump_thread_path): Rename to...
3449 (thread_jumps::profitable_jump_thread_path): ...this.
3450 (convert_and_register_jump_thread_path): Rename to...
3451 (thread_jumps::convert_and_register_current_path): ...this.
3452 (check_subpath_and_update_thread_path): Rename to...
3453 (thread_jumps::check_subpath_and_update_thread_path): ...this.
3454 (register_jump_thread_path_if_profitable): Rename to...
3455 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
3456 (handle_phi): Rename to...
3457 (thread_jumps::handle_phi): ...this.
3458 (handle_assignment): Rename to...
3459 (thread_jumps::handle_assignment): ...this.
3460 (fsm_find_control_statement_thread_paths): Rename to...
3461 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
3462 (find_jump_threads_backwards): Rename to...
3463 (thread_jumps::find_jump_threads_backwards): ...this.
3464 Initialize path local data.
3465 (pass_thread_jumps::execute): Call find_jump_threads_backwards
3466 from within thread_jumps class.
3467 (pass_early_thread_jumps::execute): Same.
3469 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3471 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
3473 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3475 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
3476 to canonical location.
3478 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3481 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
3483 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
3485 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
3487 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
3488 rcrt1.o%s/grcrt1.o%s for -static-pie.
3490 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3492 * i386.c (ix86_multiplication_cost, ix86_division_cost,
3493 ix86_shift_rotate_cost): Break out from ...
3494 (ix86_rtx_costs): ... here.
3495 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
3498 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3500 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
3501 when entry block was promoted unlikely.
3502 (estimate_bb_frequencies): Increase frequency scale.
3503 * profile-count.h (profile_count): Export precision info.
3505 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3507 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
3508 disturbin profile of entry block.
3510 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3512 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
3515 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3517 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
3518 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
3520 2017-11-17 Jeff Law <law@redhat.com>
3522 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
3523 from evrp_dom_walker class. Various methods moved into new class.
3524 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
3525 (evrp_range_analyzer::enter): New method.
3526 (evrp_range_analyzer::leave): New method.
3527 (evrp_dom_walker): Remove delegators no longer needed by this class.
3528 Replace vr_values data member with evrp_range_analyzer
3530 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
3531 method extracted from evrp_dom_walker::before_dom_children.
3532 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
3533 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
3535 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
3536 Add private copy constructor and move assignment operators.
3537 Privatize methods and class data where trivially possible.
3538 (evrp_dom_walker::cleanup): New function, extracted from
3539 execute_early_vrp. Simplify access to class data.
3541 * vr-values.h (get_output_for_vrp): Prototype.
3542 * vr-values.c (get_output_for_vrp): New function extracted from
3543 vrp_visit_assignment_or_call and extract_range_from_stmt.
3544 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
3546 2017-11-17 Luis Machado <luis.machado@linaro.org>
3548 * config/aarch64/aarch64.c
3549 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
3550 (qdf24xx_tunings) <autoprefetcher_model>: Set to
3551 tune_params::AUTOPREFETCHER_WEAK.
3553 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3556 * config/arm/arm.c (arm_valid_target_attribute_rec):
3557 Parse "arch=" and "+<ext>".
3558 (arm_valid_target_attribute_tree): Re-init global options.
3559 (arm_option_override): Make non-static.
3560 (arm_options_perform_arch_sanity_checks): Make errors fatal.
3561 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
3562 (__ARM_FEATURE_CRC32): Support undef.
3563 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
3564 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
3566 2017-11-17 David Malcolm <dmalcolm@redhat.com>
3568 * gdbinit.in (break-on-diagnostic): New command.
3570 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3572 * config/i386/i386.c (ix86_expand_epilogue): Change simple
3573 return to indirect jump for EH return if control-flow
3574 protection is enabled. Change explicit 'false' argument in
3575 pro_epilogue_adjust_stack with a value of
3577 * config/i386/i386.md (simple_return_indirect_internal):
3578 Remove SImode restriction to support 64-bit.
3580 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3582 * combine.c (added_notes_insn): New.
3583 (try_combine): Handle added_notes_insn like added_links_insn.
3584 Rewrite return value code.
3585 (distribute_notes): Set added_notes_insn to the earliest insn we added
3588 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3590 PR rtl-optimization/82621
3591 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
3592 dest of one of those SETs is unused.
3594 2017-11-17 Richard Biener <rguenther@suse.de>
3597 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
3598 * tree-pretty-print.c (dump_generic_node): Handle
3599 annot_expr_parallel_kind.
3600 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
3601 * gimplify.c (gimple_boolify): Likewise.
3603 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3605 * config.gcc (extra_headers): Add cet.h for x86 targets.
3606 * config/i386/cet.h: New file.
3607 * doc/install.texi: Add --enable-cet/--disable-cet.
3609 2017-11-17 Richard Biener <rguenther@suse.de>
3611 PR tree-optimization/83017
3612 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
3613 (gen_parallel_loop): Properly count iterations.
3614 (parallelize_loops): Handle loop->can_be_parallel independent
3615 of flag_loop_parallelize_all. Make static profitability test match
3617 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
3618 * invoke.texi (parloops-min-per-thread): Document.
3620 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
3622 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
3623 upstreaming review comments.
3625 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3627 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
3628 with fast unaligned access.
3629 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
3631 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
3633 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
3634 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
3636 2017-11-17 Richard Biener <rguenther@suse.de>
3638 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
3639 folding of references.
3641 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
3644 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
3645 of replacing call to strncmp with corresponding call to strcmp when
3648 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
3650 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
3651 option prefer-avx256 for skylake-avx512 configuration.
3652 * config/i386/i386.c (ix86_option_override_internal): Ditto.
3653 (get_builtin_code_for_version): Ditto.
3655 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3656 Monk Chiang <sh.chiang04@gmail.com>
3658 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
3659 (FIXED_REGISTERS): Reserve more register numbers.
3660 (CALL_USED_REGISTERS): Likewise.
3661 (REG_ALLOC_ORDER): Likewise.
3662 (REG_CLASS_CONTENTS): Likewise.
3663 (REGISTER_NAMES): Likewise.
3665 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3666 Kito Cheng <kito.cheng@gmail.com>
3668 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
3670 * config/nds32/iterators.md: Add vector mode iterators and attributes.
3672 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
3674 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
3675 parameter list for vec_splats.
3677 2017-11-16 Joseph Myers <joseph@codesourcery.com>
3679 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
3681 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
3683 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
3686 * doc/install.texi: Document incompatibility of Perl >=5.6.26
3687 with the required version of automake 1.11.6.
3689 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
3691 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
3693 (DU_C2_3_power9): Correct reservation combinations.
3694 (FP_DIV_power9, VEC_DIV_power9): New.
3695 (power9-alu): Split out rotate/shift...
3696 (power9-rot): ...to here, correct dispatch resource.
3697 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
3699 (power9-fp): Correct latency.
3700 (power9-sdiv): Add div/sqrt resource.
3701 (power9-ddiv): Correct latency, add div/sqrt resource.
3702 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
3703 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
3705 (power9-qpdiv, power9-qpmul): Adjust resource usage.
3707 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
3709 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
3710 switch statement mapping KF built-ins to TF built-ins if we don't
3711 have the proper ISA 3.0 assembler support.
3713 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3715 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
3716 (lower_emutls_function_body): Do not compute it.
3718 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3720 * ipa-split.c (split_bb_info): Turn time to sreal.
3721 (split_point): Likewise.
3722 (dump_split_point): Likewise.
3723 (fine_split_points): Likewise.
3724 (execute_split_functions): Only zero split_bbs; turn time to sreals.
3726 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3728 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
3731 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3733 * predict.c (combine_predictions_for_bb): Preserve zero predicted
3735 (expensive_function_p): Remove useless assert.
3736 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
3738 2017-11-16 Martin Sebor <msebor@redhat.com>
3740 PR tree-optimization/82588
3741 PR tree-optimization/82583
3742 * tree-vrp.c (check_array_ref): Handle flexible array members,
3743 string literals, and inner indices.
3744 (search_for_addr_array): Add detail to diagnostics.
3746 2017-11-16 Nathan Sidwell <nathan@acm.org>
3750 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
3751 (SET_DECL_ASSEMBLER_NAME): Forward to
3752 overwrite_decl_assembler_name.
3753 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
3754 (overwrite_decl_assembler_name): Declare.
3755 * tree.c (overwrite_decl_assembler_name): New.
3756 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
3757 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
3758 (LANG_HOOKS_INITIALIZER): Add it.
3759 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
3760 * langhooks.c (lhd_set_decl_assembler_name): Use
3761 SET_DECL_ASSEMBLER_NAME.
3762 (lhd_overwrite_decl_assembler_name): Default implementation.
3764 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
3765 Jackson Woodruff <jackson.woodruff@arm.com>
3767 PR tree-optimization/71026
3768 * match.pd: Canonicalize constant multiplies in division.
3770 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
3772 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
3773 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
3774 Remove OPT_fomit_frame_pointer entry.
3775 * common/config/arc/arc-common.c: Likewise.
3776 * common/config/arm/arm-common.c: Likewise.
3777 * common/config/avr/avr-common.c: Likewise.
3778 * common/config/c6x/c6x-common.c: Likewise.
3779 * common/config/cr16/cr16-common.c: Likewise.
3780 * common/config/cris/cris-common.c: Likewise.
3781 * common/config/epiphany/epiphany-common.c: Likewise.
3782 * common/config/fr30/fr30-common.c: Likewise.
3783 * common/config/frv/frv-common.c: Likewise.
3784 * common/config/ia64/ia64-common.c: Likewise.
3785 * common/config/iq2000/iq2000-common.c: Likewise.
3786 * common/config/lm32/lm32-common.c: Likewise.
3787 * common/config/m32r/m32r-common.c: Likewise.
3788 * common/config/mcore/mcore-common.c: Likewise.
3789 * common/config/microblaze/microblaze-common.c: Likewise.
3790 * common/config/mips/mips-common.c: Likewise.
3791 * common/config/mmix/mmix-common.c: Likewise.
3792 * common/config/mn10300/mn10300-common.c: Likewise.
3793 * common/config/nios2/nios2-common.c: Likewise.
3794 * common/config/pa/pa-common.c: Likewise.
3795 * common/config/pdp11/pdp11-common.c: Likewise.
3796 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
3797 * common/config/riscv/riscv-common.c: Likewise.
3798 * common/config/rs6000/rs6000-common.c: Likewise.
3799 * common/config/rx/rx-common.c: Likewise.
3800 * common/config/s390/s390-common.c: Likewise.
3801 * common/config/sh/sh-common.c: Likewise.
3802 * common/config/sparc/sparc-common.c: Likewise.
3803 * common/config/tilegx/tilegx-common.c: Likewise.
3804 * common/config/tilepro/tilepro-common.c: Likewise.
3805 * common/config/v850/v850-common.c: Likewise.
3806 * common/config/visium/visium-common.c: Likewise.
3807 * common/config/xstormy16/xstormy16-common.c: Likewise.
3808 * common/config/xtensa/xtensa-common.c: Likewise.
3809 * invoke.texi (-fomit-frame-pointer): Update documentation.
3811 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3813 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
3815 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3817 * cfg.c (scale_bbs_frequencies_int,
3818 cale_bbs_frequencies_gcov_type): Remove.
3819 * cfg.h (scale_bbs_frequencies_int,
3820 cale_bbs_frequencies_gcov_type): Remove.
3822 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3824 * tree-ssa-loop-manip.c
3825 (scale_dominated_blocks_in_loop): Update to profile counts.
3826 (tree_transform_and_unroll_loop): Likewise.
3828 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3830 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
3831 scale_bbs_frequencies_int.
3833 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3835 * final.c (compute_alignments): Use counts rather than frequencies.
3837 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3839 * cfgloopanal.c: Include sreal.h
3840 (average_num_loop_insns): Use counts and sreal for accounting.
3842 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3844 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
3847 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3848 Kito Cheng <kito.cheng@gmail.com>
3850 * config/nds32/constraints.md: Provide more constraints.
3851 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
3852 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
3853 support constraints usage.
3855 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3857 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
3859 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3860 Kito Cheng <kito.cheng@gmail.com>
3862 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
3863 * config/nds32/nds32.opt: Refine the layout.
3864 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
3865 TARGET_EXT_STRING): Support new options.
3866 * config/nds32/nds32.h: Likewise.
3867 * config/nds32/nds32.md: Likewise.
3868 * config/nds32/nds32-predicates.c: Likewise.
3869 * config/nds32/constraints.md: Likewise.
3870 * common/config/nds32/nds32-common.c: Likewise.
3872 2017-11-16 Julia Koval <julia.koval@intel.com>
3875 * config/i386/gfniintrin.h: Add sse check.
3876 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
3878 2017-11-16 Julia Koval <julia.koval@intel.com>
3880 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
3881 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
3882 (ix86_handle_option): Handle -mavx512vbmi2.
3883 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
3884 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3885 * config/i386/i386-c.c (__AVX512VBMI2__): New.
3886 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
3887 (ix86_valid_target_attribute_inner_p): Ditto.
3888 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
3889 * config/i386/i386.opt (mavx512vbmi2): New option.
3890 * doc/invoke.texi: Add new option.
3892 2017-11-16 Julia Koval <julia.koval@intel.com>
3894 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
3895 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
3896 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
3897 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
3898 _mm512_gf2p8mul_epi8): New intrinsics.
3899 * config/i386/i386-builtin-types.def
3900 (V64QI_FTYPE_V64QI_V64QI): New type.
3901 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
3902 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
3903 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
3904 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
3905 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
3906 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3908 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
3910 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
3911 explicitly as a stream of bytes.
3913 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3915 * config/rs6000/altivec.h (vec_xst_be): New #define.
3916 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
3917 and externalize from *altivec_vperm_<mode>_internal.
3918 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
3920 (XL_BE_V8HI): Likewise.
3921 (XL_BE_V4SI): Likewise.
3922 (XL_BE_V4SI): Likewise.
3923 (XL_BE_V2DI): Likewise.
3924 (XL_BE_V4SF): Likewise.
3925 (XL_BE_V2DF): Likewise.
3926 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
3927 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
3928 all array entries with these keys: VSX_BUILTIN_VEC_XL,
3929 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
3930 VSX_BUILTIN_VEC_XST_BE.
3931 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
3932 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
3934 (altivec_init_builtins): Replace conditional calls to def_builtin
3935 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
3936 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
3937 with unconditional calls. Remove calls to def_builtin for
3938 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
3939 __builtin_vec_xst_be.
3940 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
3941 to define_expand, and add alternate RTL generation for P8.
3942 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
3943 vsx_ld_elemrev_v8hi.
3944 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
3945 add alternate RTL generation for P8.
3946 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
3947 vsx_ld_elemrev_v16qi.
3948 (vsx_st_elemrev_v8hi): Convert define_insn
3949 to define_expand, and add alternate RTL generation for P8.
3950 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
3951 vsx_st_elemrev_v8hi.
3952 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
3953 add alternate RTL generation for P8.
3954 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
3955 vsx_st_elemrev_v16qi.
3957 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
3960 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
3961 TARGET_AVX512ER check.
3962 (ix86_option_override_internal): Set MASK_VZEROUPPER if
3963 neither -mzeroupper nor -mno-zeroupper is used and
3964 TARGET_EMIT_VZEROUPPER is set.
3965 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
3966 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
3968 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
3970 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3971 folding of vector compares.
3972 (fold_build_vec_cmp): New helper function.
3973 (fold_compare_helper): New helper function.
3974 (builtin_function_type): Add compare builtins to the list of functions
3975 having unsigned arguments. Cosmetic updates to comment indentation.
3976 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
3977 the not+eq combination.
3979 2017-11-15 Bin Cheng <bin.cheng@arm.com>
3981 PR tree-optimization/82726
3982 PR tree-optimization/70754
3983 * tree-predcom.c (order_drefs_by_pos): New function.
3984 (combine_chains): Move code setting has_max_use_after to...
3985 (try_combine_chains): ...here. New parameter. Sort combined chains
3986 according to position information.
3987 (tree_predictive_commoning_loop): Update call to above function.
3988 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
3990 2017-11-15 Bin Cheng <bin.cheng@arm.com>
3992 PR tree-optimization/82726
3994 2017-01-23 Bin Cheng <bin.cheng@arm.com>
3996 PR tree-optimization/70754
3997 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
3998 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
3999 combined stmt before it if not NULL.
4000 (combine_chains): Process refs reversely and compute dominance point
4004 2017-02-23 Bin Cheng <bin.cheng@arm.com>
4006 PR tree-optimization/79663
4007 * tree-predcom.c (combine_chains): Process refs in reverse order
4008 only for ZERO length chains, and add explaining comment.
4010 2017-11-15 Tamar Christina <tamar.christina@arm.com>
4012 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
4013 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
4014 * doc/invoke.texi (ARM Options): Add armv8.3-a.
4016 2017-11-15 Tamar Christina <tamar.christina@arm.com>
4018 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
4020 2017-11-15 Martin Liska <mliska@suse.cz>
4022 * tree-cfg.c (pass_warn_function_return::execute):
4023 Compare warn_return_type for greater than zero.
4025 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
4029 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
4030 to return true on Xeon and not on Xeon Phi.
4031 (ix86_check_avx256_register): Changed to ...
4032 (ix86_check_avx_upper_register): ... this. Add extra check for
4033 VALID_AVX512F_REG_OR_XI_MODE.
4034 (ix86_avx_u128_mode_needed): Changed
4035 ix86_check_avx256_register to ix86_check_avx_upper_register.
4036 (ix86_check_avx256_stores): Changed to ...
4037 (ix86_check_avx_upper_stores): ... this. Changed
4038 ix86_check_avx256_register to ix86_check_avx_upper_register.
4039 (ix86_avx_u128_mode_after): Changed
4040 avx_reg256_found to avx_upper_reg_found. Changed
4041 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
4042 (ix86_avx_u128_mode_entry): Changed
4043 ix86_check_avx256_register to ix86_check_avx_upper_register.
4044 (ix86_avx_u128_mode_exit): Ditto.
4045 * config/i386/i386.h: (host_detect_local_cpu): New define.
4047 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
4049 * config/arm/xgene1.md (xgene1): Split into automatons
4050 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
4051 (xgene1_f_load): Adjust reservations and/or types.
4052 (xgene1_f_store): Likewise.
4053 (xgene1_load_pair): Likewise.
4054 (xgene1_store_pair): Likewise.
4055 (xgene1_fp_load1): Likewise.
4056 (xgene1_load1): Likewise.
4057 (xgene1_store1): Likewise.
4058 (xgene1_move): Likewise.
4059 (xgene1_alu): Likewise.
4060 (xgene1_simd): Likewise.
4061 (xgene1_bfm): Likewise.
4062 (xgene1_neon_load1): Likewise.
4063 (xgene1_neon_store1): Likewise.
4064 (xgene1_neon_logic): Likewise.
4065 (xgene1_neon_st1): Likewise.
4066 (xgene1_neon_ld1r): Likewise.
4067 (xgene1_alu_cond): Added.
4068 (xgene1_shift_reg): Likwise.
4069 (xgene1_bfx): Likewise.
4070 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
4072 2017-11-15 Jakub Jelinek <jakub@redhat.com>
4075 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
4077 (can_widen_mult_without_libcall): New function.
4078 (expand_mul_overflow): If only checking unsigned mul overflow,
4079 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
4080 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
4081 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
4082 (expand_DIVMOD): Formatting fix.
4083 * expmed.h (expand_mult): Add NO_LIBCALL argument.
4084 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
4085 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
4087 PR tree-optimization/82977
4088 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
4089 constructed temporary to strlen_to_stridx.put.
4091 2017-11-15 Martin Liska <mliska@suse.cz>
4093 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
4094 * configure: Regenerate.
4096 2017-11-15 Martin Liska <mliska@suse.cz>
4099 * config/sh/sh-mem.cc: Use proper probability for
4102 2017-11-14 Jeff Law <law@redhat.com>
4104 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
4105 the red zone for stack_clash_protection_final_dynamic_probe targets
4106 when the total dynamic stack size is zero bytes.
4108 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
4109 blocks is post order.
4111 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
4113 * dumpfile.h (TDF_COMPARE_DEBUG): New.
4114 * final.c (rest_of_clean_state): Set it for the
4115 -fcompare-debug dump.
4116 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
4117 class when TDF_COMPARE_DEBUG is set.
4119 * dwarf2out.c (gen_producer_string): Discard
4122 2017-11-15 Joseph Myers <joseph@codesourcery.com>
4125 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
4126 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
4127 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
4128 (__TGMATH_REAL_2_3): Remove macros.
4129 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
4130 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
4132 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
4134 (remquo): Define using __TGMATH_REAL_3.
4136 2017-11-14 Jeff Law <law@redhat.com>
4138 * vr-values.c: New file with contents extracted from tree-vrp.c.
4139 * Makefile.in (OBJS): Add vr-values.o
4140 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
4141 (set_value_range, set_and_canonicalize_value_range): Likewise.
4142 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
4143 (value_range_constant_singleton, symbolic_range_p): Likewise.
4144 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
4145 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
4146 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
4147 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
4148 (find_case_label_range, find_case_label_index): Likewise.
4149 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
4150 (range_int_cst_singleton_p, value_inside_range): Likewise.
4151 (get_single_symbol): Likewise.
4152 (switch_update): Move structure definition here.
4153 (to_remove_edges, to_update_switch_stmts): Provide externs.
4154 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
4155 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
4156 (vrp_val_is_min, set_value_range): Likewise.
4157 (set_and_canonicalize_value_range, copy_value_range): Likewise.
4158 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
4159 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
4160 (range_is_nonnull, range_int_cst_p): Likewwise.
4161 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
4162 (get_single_symbol, operand_less_p): Likewise
4163 (compare_values_warnv, compare_values): Likewise.
4164 (value_inside_range, value_range_constant_singleton): Likewise.
4165 (zero_nonzero_bitgs_from_vr): Likewise.
4166 (extract_range_from_binary_expr_1): Likewise.
4167 (overflow_comparison_p): Likewise.
4168 (to_remove_edges, to_update_switch_stmts): Likewise.
4169 (find_case_label-index, find_case_label_range): Likewise.
4170 (switch_update, set_value_range_to_nonnegative): Remove.
4171 (set_value_range_to_truthvalue): Likewise.
4172 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
4173 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
4174 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
4175 (find_case_label_ranges, test_for_singularity): Likewise.
4176 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
4177 (x_vr_values): Move to its remaining use site.
4179 2017-11-10 Jeff Law <law@redhat.com>
4181 * vr-values.h (VR_INITIALIZER): Move #define here.
4182 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
4183 * Makefile.in (OBJS): Add tree-evrp.o
4184 * tree-vrp.h (assert_info): Move structure definition here.
4185 (set_value_range_to_varying): Prototype.
4186 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
4187 (infer_value_range, register_edge_assert_for): Likewise.
4188 (stmt_interesting_for_vrp): Likewise.
4189 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
4190 (set_value_range_to_varying): No longer static.
4191 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
4192 (infer_value_range, register_edge_assert_for): Likewise.
4194 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
4196 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
4197 generate the XXBRD instruction.
4199 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
4200 (rs6000_builtin_type_compatible): Treat _Float128 and long double
4201 as being compatible if -mabi=ieeelongdouble.
4202 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
4203 to setup float128 built-ins with hardware support.
4204 (BU_FLOAT128_HW_2): Likewise.
4205 (BU_FLOAT128_HW_3): Likewise.
4206 (BU_FLOAT128_HW_VSX_1): Likewise.
4207 (BU_FLOAT128_HW_VSX_2): Likewise.
4208 (scalar_extract_expq): Change float128 built-in functions to
4209 accommodate having both KFmode and TFmode functions. Use the
4210 KFmode variant as the default.
4211 (scalar_extract_sigq): Likewise.
4212 (scalar_test_neg_qp): Likewise.
4213 (scalar_insert_exp_q): Likewise.
4214 (scalar_insert_exp_qp): Likewise.
4215 (scalar_test_data_class_qp): Likewise.
4216 (sqrtf128_round_to_odd): Delete processing the round to odd
4217 built-in functions as special built-in functions, and define them
4218 as float128 built-ins. Use the KFmode variant as the default.
4219 (truncf128_round_to_odd): Likewise.
4220 (addf128_round_to_odd): Likewise.
4221 (subf128_round_to_odd): Likewise.
4222 (mulf128_round_to_odd): Likewise.
4223 (divf128_round_to_odd): Likewise.
4224 (fmaf128_round_to_odd): Likewise.
4225 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
4226 support for KFmode and TFmode xststdcqp calls.
4227 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
4228 point, switch the built-in handlers for the get/set float128
4229 exponent, get float128 mantissa, float128 test built-ins, and the
4230 float128 round to odd built-in functions. Eliminate creating the
4231 float128 round to odd built-in functions as special built-ins.
4232 (rs6000_init_builtins): Eliminate special creation of the float128
4233 round to odd built-in functions.
4234 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
4235 function insns to support both TFmode and KFmode variants.
4236 (xsxsigqp_<mode>): Likewise.
4237 (xsiexpqpf_<mode>): Likewise.
4238 (xsiexpqp_<mode>): Likewise.
4239 (xststdcqp_<mode>): Likewise.
4240 (xststdcnegqp_<mode>): Likewise.
4241 (xststdcqp_<mode>): Likewise.
4243 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4245 * tree-ssa-threadupdate.c (compute_path_counts): Remove
4246 unused path_in_freq_ptr parameter.
4247 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
4249 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4251 * ipa-inline.c (edge_badness): Dump sreal frequency.
4252 (compute_inlined_call_time): Match natural implementaiton ...
4253 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
4254 forgotten division by CGRAPH_FREQ_BASE.
4256 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4258 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
4259 Solaris 11. Update comment.
4260 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
4262 * config/sol2.h (STARTFILE_SPEC): Likewise.
4263 * configure: Regenerate.
4265 2017-11-14 Carl Love <cel@us.ibm.com>
4267 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
4268 le_ and be_ prefixes to swap* variables. Remove
4269 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
4272 2017-11-14 Jason Merrill <jason@redhat.com>
4274 Support GTY((cache)) on hash_map.
4275 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
4276 (ggc_cache_remove): Override it instead of ggc_mx.
4277 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
4278 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
4279 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
4280 (simple_cache_map_traits): Override maybe_mx.
4281 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
4282 (hash_map): Friend gt_cleare_cache.
4283 (gt_cleare_cache): New.
4284 * tree.h (tree_cache_traits): New hash_map traits class.
4285 (tree_cache_map): New typedef.
4287 2017-11-14 Richard Biener <rguenther@suse.de>
4289 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
4290 paramter and handling.
4291 (cleanup_control_flow_bb): Likewise.
4292 (cleanup_control_flow_pre): New helper performing a DFS walk
4293 to call cleanup_control_flow_bb in PRE order.
4294 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
4295 via cleanup_control_flow_pre.
4297 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
4299 * config/aarch64/aarch64-simd.md
4300 (aarch64_simd_bsl<mode>_internal): Remove DImode.
4301 (*aarch64_simd_bsl<mode>_alt): Likewise.
4302 (aarch64_simd_bsldi_internal): New.
4303 (aarch64_simd_bsldi_alt): Likewise.
4305 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4307 * tracer.c (better_p): Do not compare frequencies.
4308 * reg-stack.c (better_edge): Likewise.
4309 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
4312 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4314 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
4315 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
4316 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
4317 * lto-streamer-in.c (input_function): Use update_max_bb_count.
4318 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
4319 * predict.c (maybe_hot_frequency_p): Inline to ...
4320 (maybe_hot_count_p): ... here; rewrite to counts.
4321 (counts_to_freqs): Rename to ...
4322 (update_max_bb_count): ... this one.
4323 (expensive_function_p): Use counts.
4324 (estimate_bb_frequencies): Update.
4325 (rebuild_frequencies): Update.
4326 * predict.h (counts_to_freqs): Rename to ...
4327 (update_max_bb_count): ... this one.
4328 * profile.c (compute_branch_probabilities): Add debug info
4329 * tree-inline.c (expand_call_inline): Update debug info.
4330 (optimize_inline_calls): Use update_max_bb_count..
4331 (tree_function_versioning): Use update_max_bb_count..
4332 * value-prof.c (gimple_value_profile_transformations):
4333 Do not use update_max_bb_count.
4335 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4337 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
4338 always use frequencies.
4340 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4342 * bb-reorder.c: Remove frequencies from comments.
4343 (better_edge_p): Use profile counts.
4344 (find_traces): Dump profile counts.
4345 (rotate_loop): Use profile counts.
4346 (find_traces_1_round): Likewise.
4347 (connect_better_edge_p): Use counts instead of probabilities for
4349 (copy_bb_p): Drop early check for non-0 frequency.
4350 (sanitize_hot_paths): Update comments.
4352 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4354 * ipa-split.c (struct split_point): Add count.
4355 (consider_split): Do not compute incoming frequency; compute incoming
4356 count and store it to split_point.
4357 (split_function): Set count of the call to split part correctly.
4359 2017-11-13 Carl Love <cel@us.ibm.com>
4361 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
4363 2017-11-13 Tom Tromey <tom@tromey.com>
4365 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
4367 2017-11-13 Carl Love <cel@us.ibm.com>
4369 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
4370 Add support for builtins:
4371 unsigned int vec_first_{,miss}_match_{,or_eos}index,
4372 vector {un,}signed {char,int,short},
4373 vector {un,}signed {char,int,short}) arguments.
4374 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
4375 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
4376 Add BU_P9V_AV_2 expansions for the builtins.
4377 * config/rs6000/altivec.h (vec_first_match_index,
4378 vec_first_mismatch_index, vec_first_match_or_eos_index,
4379 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
4380 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
4381 new extern declaration.
4382 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
4383 * config/rs6000/vsx.md (first_match_index_<mode>,
4384 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
4385 first_mismatch_or_eos_index_<mode>): Add define expand.
4386 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
4387 * doc/extend.texi: Update the built-in documenation file for the new
4390 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
4392 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
4393 into the min/max operations for _Float<N> and _Float<N>X types.
4395 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4398 * dwarf2out.c (do_eh_frame): New static variable.
4399 (dwarf2out_begin_prologue): Set it.
4400 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
4402 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4404 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
4406 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
4408 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4411 * debug.h (dwarf2out_do_eh_frame): Declare.
4412 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
4413 (dwarf2out_do_frame): Use it.
4414 (dwarf2out_do_cfi_asm): Likewise.
4415 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
4416 (dwarf2out_assembly_start): Likewise.
4417 (dwarf2out_begin_prologue): Fix comment.
4418 * toplev.c (compile_file): Always call dwarf2out_frame_finish
4419 if the target needs either debug or unwind DWARF2 info.
4420 * lto-opts.c (lto_write_options): Do not save -fexceptions,
4421 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
4422 -ftrapping-math, -ftrapv and -fwrapv.
4424 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4426 * cgraph.c (cgraph_edge::sreal_frequency): New function.
4427 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
4428 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
4429 (estimate_edge_size_and_time): Likewise.
4430 (ipa_merge_fn_summary_after_inlining): Likewise.
4431 * ipa-inline.c (cgraph_freq_base_rec): Remove.
4432 (compute_uninlined_call_time): Use sreal_frequency.
4433 (compute_inlined_call_time): Likewise.
4434 (ipa_inline): Do not initialize cgraph_freq_base_rec.
4435 * profile-count.c: Include sreal.h.
4436 (profile_count::to_sreal_scale): New.
4437 * profile-count.h: Forward declare sreal.
4438 (profile_count::to_sreal_scale): Declare.
4440 2017-11-13 Nathan Sidwell <nathan@acm.org>
4442 * diagnostic.c (maybe_line_and_column): New.
4443 (diagnostic_get_location_text): Use it.
4444 (diagnostic_report_current_module): Likewise.
4445 (test_diagnostic_get_location_text): Add tests.
4447 2017-11-13 Luis Machado <luis.machado@linaro.org>
4449 * doc/md.texi (Specifying processor pipeline description): Fix
4450 incorrect latency for the div instruction example.
4452 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4454 PR tree-optimization/78821
4455 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
4456 that bit_not_p is the same.
4457 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
4458 (split_group): Count precisely bit_not_p bits in each statement.
4459 (invert_op): New function.
4460 (imm_store_chain_info::output_merged_store): Use invert_op to
4461 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
4462 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
4464 2017-11-13 Martin Liska <mliska@suse.cz>
4466 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
4467 (struct source_info): Likewise.
4468 (add_branch_counts): Likewise.
4469 (add_line_counts): Likewise.
4470 (function_summary): Likewise.
4471 (output_intermediate_line): Likewise.
4472 (generate_results): Likewise.
4474 2017-11-13 Martin Liska <mliska@suse.cz>
4476 * gcov.c (struct block_info): Remove typedef for block_t.
4477 (struct line_info): Likewise.
4478 (line_info::has_block): Likewise.
4479 (EXIT_BLOCK): Likewise.
4480 (unblock): Likewise.
4481 (circuit): Likewise.
4482 (get_cycles_count): Likewise.
4483 (process_file): Likewise.
4484 (read_graph_file): Likewise.
4485 (solve_flow_graph): Likewise.
4486 (find_exception_blocks): Likewise.
4487 (add_line_counts): Likewise.
4488 (accumulate_line_info): Likewise.
4489 (output_line_details): Likewise.
4491 2017-11-13 Martin Liska <mliska@suse.cz>
4493 * gcov.c (struct arc_info): Remove typedef for arc_t.
4494 (struct line_info): Likewise.
4495 (add_branch_counts): Likewise.
4496 (output_branch_count): Likewise.
4497 (function_info::~function_info): Likewise.
4498 (circuit): Likewise.
4499 (output_intermediate_line): Likewise.
4500 (read_graph_file): Likewise.
4501 (solve_flow_graph): Likewise.
4502 (find_exception_blocks): Likewise.
4503 (add_line_counts): Likewise.
4504 (accumulate_line_info): Likewise.
4505 (output_line_details): Likewise.
4506 (output_function_details): Likewise.
4508 2017-11-13 Martin Liska <mliska@suse.cz>
4510 * gcov.c (struct function_info): Remove typedef for function_t.
4511 (struct source_info): Likewise.
4512 (source_info::get_functions_at_location): Likewise.
4513 (solve_flow_graph): Likewise.
4514 (find_exception_blocks): Likewise.
4515 (add_line_counts): Likewise.
4516 (output_intermediate_file): Likewise.
4517 (process_file): Likewise.
4518 (generate_results): Likewise.
4519 (release_structures): Likewise.
4520 (read_graph_file): Likewise.
4521 (read_count_file): Likewise.
4522 (accumulate_line_counts): Likewise.
4523 (output_lines): Likewise.
4525 2017-11-13 Martin Liska <mliska@suse.cz>
4527 * gcov.c (function_info::function_info): Remove num_counts
4528 and add vector<gcov_type>.
4529 (function_info::~function_info): Use the vector.
4530 (process_file): Likewise.
4531 (read_graph_file): Likewise.
4532 (read_count_file): Likewise.
4533 (solve_flow_graph): Likewise.
4535 2017-11-13 Martin Liska <mliska@suse.cz>
4537 * gcov.c (function_info::is_artificial): New function.
4538 (process_file): Erase all artificial early.
4539 (generate_results): Skip as all artificial are already
4542 2017-11-13 Martin Liska <mliska@suse.cz>
4544 * gcov.c (read_graph_file): Store to global vector of functions.
4545 (read_count_file): Iterate the vector.
4546 (process_file): Likewise.
4547 (generate_results): Likewise.
4548 (release_structures): Likewise.
4550 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4552 PR tree-optimization/82954
4553 * gimple-ssa-store-merging.c
4554 (imm_store_chain_info::coalesce_immediate_stores): If
4555 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
4557 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
4559 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
4560 Upddate call to ENDIAN_LANE_N.
4561 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
4562 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
4563 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
4564 and use aarch64_endian_lane_rtx.
4565 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
4567 2017-11-12 Tom de Vries <tom@codesourcery.com>
4569 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
4571 2017-11-12 Tom de Vries <tom@codesourcery.com>
4573 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
4575 2017-11-12 Tom de Vries <tom@codesourcery.com>
4577 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
4579 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4580 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4581 * defaults.h (ASM_OUTPUT_LABELREF): Same.
4583 2017-11-11 Martin Sebor <msebor@redhat.com>
4586 * doc/extend.texi (attribute nonstring): Remove spurious argument.
4589 * prefic.c (translate_name): Replace strncpy with memcpy to
4590 avoid -Wstringop-truncation.
4592 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4594 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
4596 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4598 * predict.c (maybe_hot_frequency_p): Do not use cfun.
4600 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4602 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
4605 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
4607 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
4608 enable generating XXBRH if the value is in a vector register.
4609 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
4610 value is in a vector register.
4611 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
4612 register to register bswap64's instead of doing the GPR sequence
4613 used on previous machines.
4614 (bswapdi2_xxbrd): New insn.
4615 (bswapdi2_reg): Disallow on ISA 3.0.
4616 (register to register bswap64 splitter): Do not split the insn on
4617 ISA 3.0 systems that use XXBRD.
4619 2017-11-10 Martin Sebor <msebor@redhat.com>
4622 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
4624 (find_subframework_file): Same.
4626 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4628 * auto-profile.c (afdo_indirect_call): Drop frequency.
4629 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
4630 (cgraph_node::create_edge): Drop frequency argument.
4631 (cgraph_node::create_indirect_edge): Drop frequency argument.
4632 (cgraph_edge::make_speculative): Drop frequency arguments.
4633 (cgraph_edge::resolve_speculation): Do not update frequencies
4634 (cgraph_edge::dump_edge_flags): Do not dump frequency.
4635 (cgraph_node::dump): Check consistency in IPA mode.
4636 (cgraph_edge::maybe_hot_p): Use IPA counter.
4637 (cgraph_edge::verify_count_and_frequency): Rename to ...
4638 (cgraph_edge::verify_count): ... this one; drop frequency checking.
4639 (cgraph_node::verify_node): Update.
4640 * cgraph.h (struct cgraph_edge): Drop frequency.
4641 (cgraph_edge::frequency): New function.
4642 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
4644 (cgraph_edge::rebuild_edges): Likewise.
4645 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
4646 (duplicate_thunk_for_node): Do not pass frequency.
4647 (cgraph_node::create_clone): Scale only counts.
4648 (cgraph_node::create_virtual_clone): Do not pass frequency.
4649 (cgraph_node::create_edge_including_clones): Do not pass frequency.
4650 (cgraph_node::create_version_clone): Do not pass frequency.
4651 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
4652 (cgraph_node::expand_thunk): Do not pass frequency.
4653 (cgraph_node::create_wrapper): Do not pass frequency.
4654 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
4656 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
4657 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
4658 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
4659 (gather_caller_stats): Use frequency function.
4660 (ipcp_cloning_candidate_p): Use frequency function.
4661 (ipcp_propagate_stage): Use frequency function.
4662 (get_info_about_necessary_edges): Use frequency function.
4663 (update_profiling_info): Update only IPA profile.
4664 (update_specialized_profile): Use frequency functoin.
4665 (perhaps_add_new_callers): Update only IPA profile.
4666 * ipa-devirt.c (ipa_devirt): Use IPA profile.
4667 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
4668 (dump_ipa_call_summary): Use frequency function.
4669 (estimate_edge_size_and_time): Use frequency function.
4670 (ipa_merge_fn_summary_after_inlining): Use frequency function.
4671 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
4672 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
4673 (update_noncloned_counts): ... ths one; scale counts only.
4674 (clone_inlined_nodes): Do not scale frequency.
4675 (inline_call): Do not pass frequency.
4676 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
4677 (compute_inlined_call_time): Use IPA profile.
4678 (want_inline_small_function_p): Use IPA profile.
4679 (want_inline_self_recursive_call_p): Use IPA profile.
4680 (edge_badness): Use IPA profile.
4681 (lookup_recursive_calls): Use IPA profile.
4682 (recursive_inlining): Do not pass frequency.
4683 (resolve_noninline_speculation): Do not update frequency.
4684 (inline_small_functions): Collect max of IPA profile.
4685 (dump_overall_stats): Dump IPA porfile.
4686 (dump_inline_stats): Dump IPA porfile.
4687 (ipa_inline): Collect IPA stats.
4688 * ipa-inline.h (clone_inlined_nodes): Update prototype.
4689 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
4690 (ipa_propagate_frequency): Use frequency function.
4691 (ipa_profile): Cleanup.
4692 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
4693 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
4694 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
4695 (input_node): Do not stream frequency.
4696 (input_edge): Do not stream frequency.
4697 (merge_profile_summaries): Scale only IPA profiles.
4698 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
4699 * predict.c (drop_profile): Do not recompute frequency.
4700 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
4701 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
4702 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
4703 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
4704 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
4705 * tree-inline.c (copy_bb): Do not scale frequency.
4706 (expand_call_inline): Do not scale frequency.
4707 (tree_function_versioning): Do not scale frequency.
4708 * ubsan.c (ubsan_create_edge): Do not pass frequency.
4710 2017-11-10 Julia Koval <julia.koval@intel.com>
4712 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
4713 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
4714 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
4715 (_mm256_maskz_gf2p8affine_epi64_epi8)
4716 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
4717 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
4718 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
4719 (__builtin_ia32_vgf2p8affineqb_v32qi)
4720 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
4721 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
4723 2017-11-10 Tamar Christina <tamar.christina@arm.com>
4727 (arm_option_override): Refactor.
4728 (arm_option_reconfigure_globals): New.
4729 (arm_options_perform_arch_sanity_checks): New.
4730 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
4732 (arm_options_perform_arch_sanity_checks): Likewise
4734 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
4736 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
4737 (power9-qpmul): New.
4738 * rs6000/rs6000.md ("type" attr): Add qmul.
4739 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
4740 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
4741 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
4743 2017-11-10 Martin Sebor <msebor@redhat.com>
4746 * builtins.c (compute_objsize): Handle arrays that
4747 compute_builtin_object_size likes to fail for. Make extern.
4748 * builtins.h (compute_objsize): Declare.
4749 (check_strncpy_sizes): New function.
4750 (expand_builtin_strncpy): Call check_strncpy_sizes.
4751 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
4752 -Wstringop-truncation.
4753 (gimple_fold_builtin_strncat): Same.
4754 * gimple.c (gimple_build_call_from_tree): Set call location.
4755 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
4756 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
4757 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
4758 (handle_builtin_strlen): Use strlen_to_stridx.
4759 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
4761 Use strlen_to_stridx.
4762 (pass_strlen::execute): Release strlen_to_stridx.
4763 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
4764 (-Wstringop-truncation): Document new option.
4766 2017-11-10 Martin Liska <mliska@suse.cz>
4768 PR gcov-profile/82702
4769 * gcov.c (main): Handle intermediate files in a different
4771 (get_gcov_intermediate_filename): New function.
4772 (output_gcov_file): Remove support of intermediate files.
4773 (generate_results): Allocate intermediate file.
4774 (release_structures): Clean-up properly fn_end.
4775 (output_intermediate_file): Start iterating with line 1.
4777 2017-11-10 Jakub Jelinek <jakub@redhat.com>
4779 PR tree-optimization/82929
4780 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
4781 ops_swapped_p non-static data member.
4782 (store_immediate_info::store_immediate_info): Clear it.
4783 (imm_store_chain_info::coalesce_immediate_stores): If swapping
4784 ops set ops_swapped_p.
4785 (count_multiple_uses): Handle ops_swapped_p.
4787 2017-11-10 Martin Liska <mliska@suse.cz>
4789 * coverage.c (coverage_init): Stream information about
4790 support of has_unexecuted_blocks.
4791 * doc/gcov.texi: Document that.
4792 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
4793 * gcov.c (read_graph_file): Likewise.
4794 (output_line_beginning): Fix a small issue with
4797 2017-11-10 Bin Cheng <bin.cheng@arm.com>
4799 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
4800 reference of trivial component.
4802 2017-11-10 Jakub Jelinek <jakub@redhat.com>
4805 * gimple-ssa-store-merging.c
4806 (pass_store_merging::terminate_all_aliasing_chains): For
4807 gimple_store_p stmts also call refs_output_dependent_p.
4809 PR rtl-optimization/82913
4810 * compare-elim.c (try_merge_compare): Punt if def_insn is not
4813 2017-11-09 Jeff Law <law@redhat.com>
4815 * vr-values.h: New file with vr_values class.
4816 * tree-vrp.c: Include vr-values.h
4817 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
4818 data objects into the vr_values class.
4819 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
4820 (get_value_range): Make it a member function within vr_values class.
4821 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
4822 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
4823 (op_with_constant_singleton_value_range): Likewise.
4824 (extract_range_for_var_from_comparison_expr): Likewise.
4825 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
4826 (extract_range_from_binary_expr): Likewise.
4827 (extract_range_from_unary_expr): Likewise.
4828 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
4829 (check_for_binary_op_overflow, extract_range_basic): Likewise.
4830 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
4831 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
4832 (compare_name_with_value, compare_names): Likewise.
4833 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
4834 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
4835 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
4836 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
4837 (extract_range_from_phi_node): Likewise.
4838 (simplify_truth_ops_using_ranges): Likewise.
4839 (simplify_div_or_mod_using_ranges): Likewise.
4840 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
4841 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
4842 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
4843 (simplify_float_conversion_using_ranges): Likewise.
4844 (simplify_internal_call_using_ranges): Likewise.
4845 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
4846 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
4847 poitner via x_vr_values for calls into gimple folder.
4848 (vrp_initialize_lattice): Make this the vr_values ctor.
4849 (vrp_free_lattice): Make this the vr_values dtor.
4850 (set_vr_value): New function.
4851 (class vrp_prop): Add vr_values data member. Add various member
4852 functions as well as member functions that delegate to vr_values.
4853 (check_array_ref): Make a member function within vrp_prop class.
4854 (search_for_addr_array, vrp_initialize): Likewise.
4855 (vrp_finalize): Likewise. Revamp to avoid direct access to
4856 vr_value, values_propagated, etc.
4857 (check_array_bounds): Extract vrp_prop class instance pointer from
4858 walk info structure. Use it to call member functions.
4859 (check_all_array_refs): Make a member function within vrp_prop class.
4860 Smuggle class instance pointer via walk info structure.
4861 (x_vr_values): New local static.
4862 (vrp_valueize): Use x_vr_values to get class instance.
4863 (vr_valueize_1): Likewise.
4864 (class vrp_folder): Add vr_values data member. Add various member
4865 functions as well as member functions that delegate to vr_values.
4866 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
4867 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
4868 class instance from vr_values. Use it to call member functions.
4869 (vrp_dom_walker): Add vr_values data member.
4870 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
4871 instance via x_vr_values.
4872 (identify_jump_threads): Accept vr_values as argument. Store
4873 it into the walker structure.
4874 (evrp_dom_walker): Add vr_values class data member. Add various
4876 (evrp_dom_walker::try_find_new_range): Use vr_values data
4877 member to access the memory allocator.
4878 (evrp_dom_walker::before_dom_children): Store vr_values class
4879 instance into the vrp_folder class.
4880 (evrp_dom_walker::push_value_range): Rework to avoid direct
4881 access to num_vr_values and vr_value.
4882 (evrp_dom_walker::pop_value_range): Likewise.
4883 (execute_early_vrp): Remove call to vrp_initialize_lattice.
4884 Use vr_values to get to dump_all_value_ranges member function.
4885 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
4886 and simplify_cond_using_ranges_2 via vrp_prop class instance.
4887 Pass vr_values class instance down to identify_jump_threads.
4888 Remove call to vrp_free_lattice.
4889 (debug_all_value_ranges): Remove.
4891 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
4892 (vrp_folder): Likewise.
4894 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
4895 Get it from the existing bitmap instead.
4896 (vrp_intersect_ranges_1): Likewise.
4898 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4900 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
4902 (store_immediate_info::store_immediate_info): Add bitnotp argument,
4903 set bit_not_p to it.
4904 (imm_store_chain_info::coalesce_immediate_stores): Break group
4905 if bit_not_p is different.
4906 (count_multiple_uses, split_group,
4907 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
4908 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
4909 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
4910 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
4911 to store_immediate_info ctor.
4913 2017-11-09 Jim Wilson <jimw@sifive.com>
4915 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
4916 (scan_prog_file): Likewise.
4918 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
4920 * bb-reorder.c (max_entry_frequency): Remove.
4921 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
4922 connect_traces, push_to_next_round_p): Remove prototypes.
4923 (find_traces_1_round): Use counts only.
4924 (push_to_next_round_p): Likewise.
4925 (find_traces): Likewise.
4926 (rotate_loop): Likewise.
4927 (find_traces_1_round): Likewise.
4928 (connect_traces): Likewise.
4929 (edge_order): Likewise.
4931 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
4933 * config/arm/arm.c (output_return_instruction): Add comments to
4934 indicate requirement for cmse_nonsecure_entry return to account
4935 for the size of clearing instruction output here.
4936 (thumb_exit): Likewise.
4937 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
4938 return in hardfloat mode.
4940 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
4942 * config/rs6000/rs6000.c (machine_function): Add a bool,
4943 "toc_is_wrapped_separately".
4944 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
4945 if it wasn't explicitly set or unset, we are optimizing for speed, and
4946 doing separate shrink-wrapping.
4947 (rs6000_get_separate_components): Enable the TOC component if
4948 saving the TOC register in the prologue.
4949 (rs6000_components_for_bb): Handle the TOC component.
4950 (rs6000_emit_prologue_components): Store the TOC register where needed.
4951 (rs6000_set_handled_components): Mark TOC as handled, if handled.
4952 (rs6000_emit_prologue): Don't save the TOC if that is already done.
4954 2017-11-09 Martin Jambor <mjambor@suse.cz>
4956 * ipa-param-manipulation.c: New file.
4957 * ipa-param-manipulation.h: Likewise.
4958 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
4959 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
4960 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
4961 (ipa_parm_adjustment): Likewise.
4962 (ipa_parm_adjustment_vec): Likewise.
4963 (ipa_get_vector_of_formal_parms): Moved declaration to
4964 ipa-param-manipulation.h.
4965 (ipa_get_vector_of_formal_parm_types): Likewise.
4966 (ipa_modify_formal_parameters): Likewise.
4967 (ipa_modify_call_arguments): Likewise.
4968 (ipa_combine_adjustments): Likewise.
4969 (ipa_dump_param_adjustments): Likewise.
4970 (ipa_modify_expr): Likewise.
4971 (ipa_get_adjustment_candidate): Likewise.
4972 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
4973 ipa-param-manipulation.c.
4974 (ipa_get_vector_of_formal_parm_types): Likewise.
4975 (ipa_modify_formal_parameters): Likewise.
4976 (ipa_modify_call_arguments): Likewise.
4977 (ipa_modify_expr): Likewise.
4978 (get_ssa_base_param): Likewise.
4979 (ipa_get_adjustment_candidate): Likewise.
4980 (index_in_adjustments_multiple_times_p): Likewise.
4981 (ipa_combine_adjustments): Likewise.
4982 (ipa_dump_param_adjustments): Likewise.
4983 * tree-sra.c: Also include ipa-param-manipulation.h
4984 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
4987 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4988 Alan Hayward <alan.hayward@arm.com>
4989 David Sherwood <david.sherwood@arm.com>
4991 * doc/sourcebuild.texi (vect_masked_store): Document.
4993 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4994 Alan Hayward <alan.hayward@arm.com>
4995 David Sherwood <david.sherwood@arm.com>
4997 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
4999 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5000 Alan Hayward <alan.hayward@arm.com>
5001 David Sherwood <david.sherwood@arm.com>
5003 * doc/sourcebuild.texi (vect_variable_length): Document.
5005 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5006 Alan Hayward <alan.hayward@arm.com>
5007 David Sherwood <david.sherwood@arm.com>
5009 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
5011 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5012 Alan Hayward <alan.hayward@arm.com>
5013 David Sherwood <david.sherwood@arm.com>
5015 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
5017 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5018 Alan Hayward <alan.hayward@arm.com>
5019 David Sherwood <david.sherwood@arm.com>
5021 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
5022 previously undocumented selectors.
5023 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
5025 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5027 * doc/rtl.texi (const_vector): Say that elements can be
5028 const_wide_ints too.
5029 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
5030 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
5031 (gen_vec_duplicate): Use it instead of CONSTANT_P.
5032 * optabs.c (expand_vector_broadcast): Likewise.
5034 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5035 Alan Hayward <alan.hayward@arm.com>
5036 David Sherwood <david.sherwood@arm.com>
5038 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
5039 scaled index even if the unscaled address was invalid.
5040 Don't increase the complexity of using a scale in that case.
5042 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5043 Alan Hayward <alan.hayward@arm.com>
5044 David Sherwood <david.sherwood@arm.com>
5046 * doc/rtl.texi: Rewrite the subreg rules so that they partition
5047 the inner register into REGMODE_NATURAL_SIZE bytes rather than
5048 UNITS_PER_WORD bytes.
5049 * emit-rtl.c (validate_subreg): Divide subregs into blocks
5050 based on REGMODE_NATURAL_SIZE of the inner mode.
5051 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
5052 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
5053 * expmed.c (lowpart_bit_field_p): Divide the value up into
5054 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
5055 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
5056 whether something is likely to occupy more than one register.
5058 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
5061 * ipa-inline-transform.c (update_noncloned_frequencies): Use
5062 profile_count::adjust_for_ipa_scaling.
5063 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
5064 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
5066 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
5068 2017-11-09 Jakub Jelinek <jakub@redhat.com>
5070 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
5071 (split_group): Add total_orig and total_new arguments, estimate the
5072 number of statements related to the store group without store merging
5073 and with store merging.
5074 (imm_store_chain_info::output_merged_store): Adjust split_group
5075 callers, punt if estimated number of statements with store merging
5076 is not smaller than estimated number of statements without it.
5078 (handled_load): Remove has_single_use checks.
5079 (pass_store_merging::process_store): Likewise.
5081 2017-11-09 Richard Biener <rguenther@suse.de>
5083 PR tree-optimization/82902
5084 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
5086 2017-11-09 Martin Liska <mliska@suse.cz>
5089 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
5092 2017-11-09 Martin Liska <mliska@suse.cz>
5094 PR tree-optimization/82669
5095 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
5097 2017-11-09 Martin Liska <mliska@suse.cz>
5099 PR gcov-profile/48463
5100 * coverage.c (coverage_begin_function): Output also end locus
5101 of a function and information whether the function is
5103 * gcov-dump.c (tag_function): Parse and print the information.
5104 * gcov.c (INCLUDE_MAP): Add include.
5105 (INCLUDE_SET): Likewise.
5106 (struct line_info): Move earlier in the source file because
5107 of vector<line_info> in function_info structure.
5108 (line_info::line_info): Likewise.
5109 (line_info::has_block): Likewise.
5110 (struct source_info): Add new member index.
5111 (source_info::get_functions_at_location): New function.
5112 (function_info::group_line_p): New function.
5113 (output_intermediate_line): New function.
5114 (output_intermediate_file): Use the mentioned function.
5115 (struct function_start): New.
5116 (struct function_start_pair_hash): Likewise.
5117 (process_file): Add code that identifies group functions.
5118 Assign lines either to global or function scope.
5119 (generate_results): Skip artificial functions.
5120 (find_source): Assign index for each source file.
5121 (read_graph_file): Read new flag artificial and end_line.
5122 (add_line_counts): Assign it either to global of function scope.
5123 (accumulate_line_counts): Isolate core of the function to
5124 accumulate_line_info and call it for both function and global
5126 (accumulate_line_info): New function.
5127 (output_line_beginning): Fix GNU coding style.
5128 (print_source_line): New function.
5129 (output_line_details): Likewise.
5130 (output_function_details): Likewise.
5131 (output_lines): Iterate both source (global) scope and function
5133 (struct function_line_start_cmp): New class.
5134 * doc/gcov.texi: Reflect changes in documentation.
5136 2017-11-09 Jakub Jelinek <jakub@redhat.com>
5139 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
5140 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
5141 and similarly for not instead of neg.
5143 2017-11-08 Andi Kleen <ak@linux.intel.com>
5145 * config/i386/i386.opt: Add -mforce-indirect-call.
5146 * config/i386/predicates.md: Check for flag_force_indirect_call.
5147 * doc/invoke.texi: Document -mforce-indirect-call
5149 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
5151 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
5153 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
5154 config/riscv/riscv.c (predict.h): New include.
5155 (riscv_slow_unaligned_access_p): No longer static.
5156 (riscv_block_move_straight): Add require.
5157 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
5159 2017-11-08 Jakub Jelinek <jakub@redhat.com>
5162 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
5163 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
5164 nonimmediate_operand predicate for operand 1 instead of
5167 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5169 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
5171 * config/aarch64/constraints.md (Uml): New constraint.
5172 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
5175 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5177 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
5178 of two vec_duplicates into a vec_concat.
5180 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5182 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
5183 Simplify vec_merge of vec_duplicate and vec_concat.
5184 * config/aarch64/constraints.md (Utq): New constraint.
5185 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
5188 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5190 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
5191 Simplify vec_merge of vec_duplicate and const_vector.
5192 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
5194 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
5195 mode iterator. Update predicate on operand 1 to
5196 handle non-const_vec constants. Delete constraints.
5197 (*aarch64_combinez_be<mode>): Likewise for operand 2.
5199 2017-11-08 Jakub Jelinek <jakub@redhat.com>
5201 PR tree-optimization/78821
5202 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
5204 (store_operand_info::store_operand_info): Initialize it to false.
5205 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
5206 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
5207 store in the group, and if chain_info is non-NULL, to ignore altogether
5209 (compatible_load_p): Fail if bit_not_p does not match.
5210 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
5211 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
5212 (pass_store_merging::process_store): Adjust
5213 terminate_all_aliasing_chains calls to pass NULL in all current spots,
5214 call terminate_all_aliasing_chains newly when adding a store into
5215 a chain with non-NULL chain_info.
5217 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
5219 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
5221 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
5223 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
5225 (aarch64_layout_frame): Initialise emit_frame_chain.
5226 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
5227 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
5229 2017-11-08 Martin Liska <mliska@suse.cz>
5231 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
5232 of gimple_call_internal_p.
5234 2017-11-07 Tom de Vries <tom@codesourcery.com>
5236 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
5238 2017-11-07 Tom de Vries <tom@codesourcery.com>
5240 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
5243 2017-11-08 Martin Liska <mliska@suse.cz>
5246 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
5248 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
5250 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
5252 2017-11-07 Jakub Jelinek <jakub@redhat.com>
5255 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
5256 (*cmp<mode>_ccz_1): New insn with $k alternative.
5259 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
5260 RTX_COMM_COMPARE as commutative as well.
5261 (ix86_binary_operator_ok): Formatting fix.
5262 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
5263 *<code><mode>3<mask_name><round_saeonly_name>,
5264 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
5265 *<s>mul<mode>3_highpart<mask_name>,
5266 *vec_widen_umult_even_v16si<mask_name>,
5267 *vec_widen_umult_even_v8si<mask_name>,
5268 *vec_widen_umult_even_v4si<mask_name>,
5269 *vec_widen_smult_even_v16si<mask_name>,
5270 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
5271 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
5272 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
5273 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
5274 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
5275 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
5276 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
5277 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
5278 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
5279 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
5280 ix86_binary_operator_ok. Formatting fixes.
5281 (*<plusminus_insn><mode>3<mask_name><round_name>,
5282 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
5285 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
5287 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
5288 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
5289 eq and ne if TARGET_ISEL.
5290 (cmp): New code_iterator.
5291 (UNS, UNSU_, UNSIK): New code_attrs.
5292 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
5293 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
5295 ("eq<mode>3"): ... this.
5296 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
5298 ("ne<mode>3"): ... this.
5300 2017-11-07 Julia Koval <julia.koval@intel.com>
5303 * common/config/i386/i386-common.c
5304 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
5305 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
5306 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
5307 * config/i386/i386.opt: Ditto.
5308 * config/i386/i386.c (ix86_target_string): Ditto.
5309 (ix86_option_override_internal): Ditto.
5310 (ix86_init_mpx_builtins): Move MPX to args2.
5311 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
5312 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
5313 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
5314 __builtin_ia32_vgf2p8affineinvqb_v32qi,
5315 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
5316 __builtin_ia32_vgf2p8affineinvqb_v16qi,
5317 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
5319 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
5322 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
5323 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
5324 (zero-extendsidi peephole2): Remove peephole.
5326 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
5329 * stor-layout.c: Include attribs.h.
5330 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
5331 explicit lookup of "aligned" attribute.
5333 2017-11-07 Andrew Waterman <andrew@sifive.com>
5335 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
5336 (riscv_expand_block_move): Likewise.
5337 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
5339 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
5340 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
5341 * config/riscv/riscv.c (riscv_block_move_straight): New function.
5342 (riscv_adjust_block_mem): Likewise.
5343 (riscv_block_move_loop): Likewise.
5344 (riscv_expand_block_move): Likewise.
5345 * config/riscv/riscv.md (movmemsi): New pattern.
5347 2017-11-07 Michael Clark <michaeljclark@mac.com>
5349 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
5350 (MUSL_DYNAMIC_LINKER): Likewise.
5352 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
5354 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
5355 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
5358 2017-11-07 Richard Biener <rguenther@suse.de>
5360 * match.pd: Fix build.
5362 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
5363 Jackson Woodruff <jackson.woodruff@arm.com>
5365 PR tree-optimization/71026
5366 * match.pd: Canonicalize negate in division.
5368 2017-11-07 Sudakshina Das <sudi.das@arm.com>
5371 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
5373 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5375 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
5376 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
5378 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5380 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
5381 non-scalar integral types.
5382 * match.pd (negate_expr_p): Handle MINUS_EXPR.
5383 (-(A-B), -(~A)): New transformations.
5385 2017-11-07 Tom de Vries <tom@codesourcery.com>
5387 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
5388 semicolon after "do {} while (0)".
5389 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5390 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5391 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5392 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5393 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5395 2017-11-07 Tom de Vries <tom@codesourcery.com>
5397 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
5398 after "do {} while (0)".
5399 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5400 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5401 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5402 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5403 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5405 2017-11-07 Tom de Vries <tom@codesourcery.com>
5408 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
5410 (arm_rtx_costs_internal): Add missing semicolon after
5411 HANDLE_NARROW_SHIFT_ARITH call.
5413 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5415 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
5416 disable isel if it was not set explicitly.
5418 2017-11-06 James Bowman <james.bowman@ftdichip.com>
5420 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
5421 (add_type_attribute) likewise.
5423 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
5425 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
5426 of struct ix86_frame.
5427 (ix86_initial_elimination_offset): Likewise.
5428 (ix86_expand_split_stack_prologue): Likewise.
5430 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
5432 * tree-vrp.h (enum value_range_type): Update stale comment.
5434 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5435 Alan Hayward <alan.hayward@arm.com>
5436 David Sherwood <david.sherwood@arm.com>
5438 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
5439 (aarch64_expand_vec_perm_const): Take the number of units too.
5440 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
5441 (aarch64_expand_vec_perm_const): Likewise.
5442 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
5443 (vec_perm<mode>): Update accordingly.
5445 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5446 Alan Hayward <alan.hayward@arm.com>
5447 David Sherwood <david.sherwood@arm.com>
5449 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
5450 Take the number of units too.
5451 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
5452 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
5453 but check for a vector mode before rather than after the call.
5454 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
5455 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
5456 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
5457 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
5458 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
5459 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
5460 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
5461 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
5462 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
5463 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
5464 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
5465 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
5466 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
5467 (aarch64_sqdmull2_n<mode>): Update accordingly.
5469 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5470 Alan Hayward <alan.hayward@arm.com>
5471 David Sherwood <david.sherwood@arm.com>
5473 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
5474 the number of units too.
5475 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
5476 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
5477 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
5478 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
5479 (vec_store_lanesxi<mode>): Update accordingly.
5481 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5482 Alan Hayward <alan.hayward@arm.com>
5483 David Sherwood <david.sherwood@arm.com>
5485 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
5486 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
5487 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
5488 of units rather than the mode.
5489 * config/aarch64/iterators.md (nunits): New mode attribute.
5490 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
5491 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
5492 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
5493 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
5494 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
5495 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
5496 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
5497 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
5498 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
5499 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
5500 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
5501 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
5502 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
5503 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
5504 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
5505 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
5506 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
5507 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
5508 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
5509 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
5510 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
5511 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
5512 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
5513 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
5514 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
5515 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
5516 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
5517 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
5518 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
5519 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
5520 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
5521 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
5522 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
5523 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
5524 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
5525 (aarch64_simd_vec_setv2di): Likewise.
5527 2017-11-06 Carl Love <cel@us.ibm.com>
5529 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
5531 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
5532 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
5534 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
5536 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
5537 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
5538 Add power 8 macro expansions.
5539 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
5540 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
5541 power 8 instructions. (VSX_XXBR): Add iterator.
5543 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
5545 * config/arm/arm.md (predicable_short_it): Change default to "no",
5546 improve documentation, remove uses that are identical to the default.
5547 (enabled_for_depr_it): Rename to enabled_for_short_it.
5548 * gcc/config/arm/arm-fixed.md (predicable_short_it):
5549 Remove default uses.
5550 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
5551 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
5552 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
5553 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
5555 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
5558 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
5559 float128 helper macros, which are no longer used after deleting
5560 the old 'q' built-in functions, and moving the round to odd
5561 built-in functions to being special built-in functions.
5562 (BU_FLOAT128_2): Likewise.
5563 (BU_FLOAT128_1_HW): Likewise.
5564 (BU_FLOAT128_2_HW): Likewise.
5565 (BU_FLOAT128_3_HW): Likewise.
5566 (FABSQ): Delete old 'q' built-in functions.
5567 (COPYSIGNQ): Likewise.
5568 (SQRTF128_ODD): Move round to odd built-in functions to be
5569 special built-in functions, so that we can handle
5570 -mabi=ieeelongdouble.
5571 (TRUNCF128_ODD): Likewise.
5572 (ADDF128_ODD): Likewise.
5573 (SUBF128_ODD): Likewise.
5574 (MULF128_ODD): Likewise.
5575 (DIVF128_ODD): Likewise.
5576 (FMAF128_ODD): Likewise.
5577 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
5578 built-in names to 'f128'.
5579 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
5580 old 'q' built-in functions, as the machine independent code for
5581 'f128' built-in functions handles this.
5582 (rs6000_expand_builtin): Add expansion for float128 round to odd
5583 functions, keying off on -mabi=ieeelongdouble of whether to use
5584 the KFmode or TFmode variant.
5585 (rs6000_init_builtins): Initialize the _Float128 round to odd
5587 * doc/extend.texi (PowerPC Built-in Functions): Document the old
5588 _Float128 'q' built-in functions are now mapped into the new
5589 'f128' built-in functions.
5591 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
5593 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
5594 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
5596 2017-11-06 David Malcolm <dmalcolm@redhat.com>
5599 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
5600 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
5601 * toplev.c: Include "ipa-fnsummary.h".
5602 (toplev::finalize): Call ipa_fnsummary_c_finalize.
5604 2017-11-06 Jakub Jelinek <jakub@redhat.com>
5606 PR tree-optimization/82838
5607 * gimple-ssa-store-merging.c
5608 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
5609 on a separate gimple_seq which is then appended to seq.
5611 2017-11-06 Jeff Law <law@redhat.com>
5614 * config/i386/i386.c (PROBE_INTERVAL): Remove.
5615 (get_probe_interval): New functions.
5616 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
5617 (ix86_adjust_stack_and_probe): Likewise.
5618 (output_adjust_stack_and_probe): Likewise.
5619 (ix86_emit_probe_stack_range): Likewise.
5620 (ix86_expand_prologue): Likewise.
5622 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5624 PR tree-optimization/82816
5625 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
5626 if the modes of the two types are the same.
5627 (convert_plusminus_to_widen): Likewise.
5629 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5631 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
5633 (usadv16qi): New define_expand.
5634 (usadv8hi): New define_expand.
5636 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
5639 * ipa-inline-transform.c (update_noncloned_frequencies): Always
5641 (inline_transform): Likewise.
5642 * predict.c (counts_to_freqs): Remove useless conditional.
5643 * profile-count.h (profile_count::apply_scale): Move sanity check.
5644 * tree-inline.c (copy_bb): Always scale.
5645 (copy_cfg_body): Likewise.
5647 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
5650 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
5652 (*cmp_ite0): Add enabled_for_depr_it attribute.
5653 (*cmp_ite1): Likewise.
5655 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5657 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
5660 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5662 * tree-vrp.c (vrp_int_const_binop): Return true on success and
5663 return the value by pointer.
5664 (extract_range_from_multiplicative_op_1): Update accordingly.
5665 Return as soon as an operation fails.
5667 2017-11-05 Tom de Vries <tom@codesourcery.com>
5670 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
5671 (DEF_SANITIZER_BUILTIN): ... here.
5672 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
5673 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
5675 2017-11-05 Tom de Vries <tom@codesourcery.com>
5678 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
5680 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
5681 ASM_OUTPUT_BEFORE_CASE_LABEL call.
5682 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
5684 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5685 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5686 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5688 2017-11-05 Tom de Vries <tom@codesourcery.com>
5691 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
5694 2017-11-04 Michael Clark <michaeljclark@mac.com>
5696 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
5697 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
5699 (*addsi3_extended): Likewise.
5700 (*addsi3_extended2): Likewise.
5701 (<optab>si3): Likewise.
5702 (<optab>di3): Likewise.
5703 (<optab><mode>3): Likewise.
5704 (<*optabe>si3_internal): Likewise.
5705 (zero_extendqi<SUPERQI:mode>2): Likewise.
5706 (*add<mode>hi3): Likewise.
5707 (*xor<mode>hi3): Likewise.
5708 (<optab>di3): Likewise.
5709 (*<optab>si3_extend): Likewise.
5710 (*sge<u>_<X:mode><GPR:mode>): Likewise.
5711 (*slt<u>_<X:mode><GPR:mode>): Likewise.
5712 (*sle<u>_<X:mode><GPR:mode>): Likewise.
5714 2017-11-04 Andrew Waterman <andrew@sifive.com>
5716 * config/riscv/riscv.c (riscv_option_override): Conditionally set
5717 TARGET_STRICT_ALIGN based upon -mtune argument.
5719 2017-11-04 Andrew Waterman <andrew@sifive.com>
5721 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
5723 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
5725 * config/i386/i386.c (choose_basereg): Use optional scratch
5726 register and add assertion.
5727 (x86_emit_outlined_ms2sysv_save): Use scratch register when
5728 needed, and don't allocate stack.
5729 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
5730 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
5731 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
5733 2017-11-03 Jeff Law <law@redhat.com>
5735 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
5736 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
5737 to probe at the start of a noreturn function.
5739 2017-11-03 Jakub Jelinek <jakub@redhat.com>
5741 PR tree-optimization/78821
5742 * gimple-ssa-store-merging.c: Update the file comment.
5743 (MAX_STORE_ALIAS_CHECKS): Define.
5744 (struct store_operand_info): New type.
5745 (store_operand_info::store_operand_info): New constructor.
5746 (struct store_immediate_info): Add rhs_code and ops data members.
5747 (store_immediate_info::store_immediate_info): Add rhscode, op0r
5748 and op1r arguments to the ctor, initialize corresponding data members.
5749 (struct merged_store_group): Add load_align_base and load_align
5751 (merged_store_group::merged_store_group): Initialize them.
5752 (merged_store_group::do_merge): Update them.
5753 (merged_store_group::apply_stores): Pick the constant for
5754 encode_tree_to_bitpos from one of the two operands, or skip
5755 encode_tree_to_bitpos if neither operand is a constant.
5756 (class pass_store_merging): Add process_store method decl. Remove
5757 bool argument from terminate_all_aliasing_chains method decl.
5758 (pass_store_merging::terminate_all_aliasing_chains): Remove
5759 var_offset_p argument and corresponding handling.
5760 (stmts_may_clobber_ref_p): New function.
5761 (compatible_load_p): New function.
5762 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
5763 if there is overlap and rhs_code is not INTEGER_CST. For
5764 non-overlapping stores terminate group if rhs is not mergeable.
5765 (get_alias_type_for_stmts): Change first argument from
5766 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
5767 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
5768 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
5770 (get_location_for_stmts): Change first argument from
5771 auto_vec<gimple *> & to vec<gimple *> &.
5772 (struct split_store): Remove orig_stmts data member, add orig_stores.
5773 (split_store::split_store): Create orig_stores rather than orig_stmts.
5774 (find_constituent_stmts): Renamed to ...
5775 (find_constituent_stores): ... this. Change second argument from
5776 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
5777 to info structures rather than the statements.
5778 (split_group): Rename ALLOW_UNALIGNED argument to
5779 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
5780 it. Adjust find_constituent_stores caller.
5781 (imm_store_chain_info::output_merged_store): Handle rhs_code other
5782 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
5783 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
5784 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
5785 (mem_valid_for_store_merging): New function.
5786 (handled_load): New function.
5787 (pass_store_merging::process_store): New method.
5788 (pass_store_merging::execute): Use process_store method. Adjust
5789 terminate_all_aliasing_chains caller.
5791 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5793 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
5794 Return true for more constants, symbols and label references.
5795 (aarch64_valid_floating_const): Remove unused function.
5797 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5800 * config/aarch64/aarch64.c (aarch64_layout_frame):
5801 Undo forcing of LR at bottom of frame.
5803 2017-11-03 Jeff Law <law@redhat.com>
5806 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
5807 for int_registers_saved.
5809 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
5810 extracted from tree-ssa-dom.c.
5811 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
5812 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
5813 (record_equivalences_from_incoming_edge): Add additional argument
5814 to single_pred_edge_ignoring_loop_edges call.
5815 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
5816 (uncprop_dom_walker::before_dom_children): Add additional argument
5817 to single_pred_edge_ignoring_loop_edges call.
5818 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
5819 single_pred_edge_ignoring_loop_edges rather than open coding.
5820 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
5822 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
5824 * match.pd (-(-A)): Rewrite.
5826 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5828 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
5829 (rs6000_emit_int_cmove): New declaration.
5830 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
5831 (rs6000_emit_sISEL): Delete.
5832 (rs6000_emit_int_cmove): Make non-static.
5833 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
5834 instead of rs6000_emit_sISEL.
5836 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
5838 * asan.c (create_cond_insert_point): Maintain profile.
5839 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
5841 * basic-block.h (struct basic_block_def): Remove frequency.
5842 (EDGE_FREQUENCY): Use to_frequency
5843 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
5845 (find_traces): Update to use to_frequency.
5846 (find_traces_1_round): Likewise; use only IPA counts.
5847 (bb_to_key): Likewise.
5848 (connect_traces): Use IPA counts only.
5849 (copy_bb_p): Update to use to_frequency.
5850 (fix_up_crossing_landing_pad): Likewise.
5851 (sanitize_hot_paths): Likewise.
5852 * bt-load.c (basic_block_freq): Likewise.
5853 * cfg.c (init_flow): Set count_max to uninitialized.
5854 (check_bb_profile): Remove frequencies; check counts.
5855 (dump_bb_info): Do not dump frequencies.
5856 (update_bb_profile_for_threading): Update counts only.
5857 (scale_bbs_frequencies_int): Likewise.
5858 (MAX_SAFE_MULTIPLIER): Remove.
5859 (scale_bbs_frequencies_gcov_type): Update counts only.
5860 (scale_bbs_frequencies_profile_count): Update counts only.
5861 (scale_bbs_frequencies): Update counts only.
5862 * cfg.h (struct control_flow_graph): Add count-max.
5863 (update_bb_profile_for_threading): Update prototype.
5864 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
5865 (find_many_sub_basic_blocks): Likewise.
5866 * cfgcleanup.c (try_forward_edges): Likewise.
5867 (try_crossjump_to_edge): Likewise.
5868 * cfgexpand.c (expand_gimple_cond): Likewise.
5869 (expand_gimple_tailcall): Likewise.
5870 (construct_init_block): Likewise.
5871 (construct_exit_block): Likewise.
5872 * cfghooks.c (verify_flow_info): Check consistency of counts.
5873 (dump_bb_for_graph): Do not dump frequencies.
5874 (split_block_1): Do not update frequencies.
5875 (split_edge): Do not update frequencies.
5876 (make_forwarder_block): Do not update frequencies.
5877 (duplicate_block): Do not update frequencies.
5878 (account_profile_record): Do not update frequencies.
5879 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
5880 for global heuristics.
5881 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
5882 (expected_loop_iterations_unbounded): Use counts only.
5883 * cfgloopmanip.c (scale_loop_profile): Simplify.
5884 (create_empty_loop_on_edge): Simplify
5886 (duplicate_loop_to_header_edge): Simplify
5887 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
5888 (update_br_prob_note): Take care of removing note when profile
5890 (relink_block_chain): Do not dump frequency.
5891 (rtl_account_profile_record): Use to_frequency.
5892 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
5893 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
5894 (cgraph_update_edges_for_call_stmt_node): Likewise.
5895 (cgraph_edge::verify_count_and_frequency): Update.
5896 (cgraph_node::verify_node): Temporarily disable frequency verification.
5897 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
5898 to_cgraph_frequency.
5899 (cgraph_edge::rebuild_edges): Convert to ipa counts.
5900 * cgraphunit.c (init_lowered_empty_function): Do not initialize
5902 (cgraph_node::expand_thunk): Update profile.
5903 * except.c (dw2_build_landing_pads): Do not update frequency.
5904 * final.c (compute_alignments): Use to_frequency.
5905 (dump_basic_block_info): Do not dump frequency.
5906 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
5907 (dump_gimple_bb_header): Do not dump frequency.
5908 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
5910 * gimple-streamer-in.c (input_bb): Do not stream frequency.
5911 * gimple-streamer-out.c (output_bb): Do not stream frequency.
5912 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
5913 (init_before_recovery): Do not update frequency.
5914 (sched_create_recovery_edges): Do not update frequency.
5915 * hsa-gen.c (convert_switch_statements): Do not update frequency.
5916 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
5917 (ipa_cp_c_finalize): Set max_count to uninitialized.
5918 * ipa-fnsummary.c (get_minimal_bb): Use counts.
5919 (param_change_prob): Use counts.
5920 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
5922 * ipa-split.c (consider_split): Use to_frequency.
5923 (split_function): Use to_frequency.
5924 * ira-build.c (loop_compare_func): Likewise.
5925 (mark_loops_for_removal): Likewise.
5926 (mark_all_loops_for_removal): Likewise.
5927 * loop-doloop.c (doloop_modify): Do not update frequency.
5928 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
5930 * lto-streamer-in.c (input_function): Update count_max.
5931 * omp-expand.c (expand_omp_taskreg): Update count_max.
5932 * omp-simd-clone.c (simd_clone_adjust): Update profile.
5933 * predict.c (maybe_hot_frequency_p): Use to_frequency.
5934 (maybe_hot_count_p): Use ipa counts only.
5935 (maybe_hot_bb_p): Simplify.
5936 (maybe_hot_edge_p): Simplify.
5937 (probably_never_executed): Do not take frequency argument.
5938 (probably_never_executed_bb_p): Do not pass frequency.
5939 (probably_never_executed_edge_p): Likewise.
5940 (combine_predictions_for_bb): Check that profile is nonzero.
5941 (propagate_freq): Do not set frequency.
5942 (drop_profile): Simplify.
5943 (counts_to_freqs): Simplify.
5944 (expensive_function_p): Use to_frequency.
5945 (propagate_unlikely_bbs_forward): Simplify.
5946 (determine_unlikely_bbs): Simplify.
5947 (estimate_bb_frequencies): Add hack to silence graphite issues.
5948 (compute_function_frequency): Use ipa counts.
5949 (pass_profile::execute): Update.
5950 (rebuild_frequencies): Use counts only.
5951 (force_edge_cold): Use counts only.
5952 * profile-count.c (profile_count::dump): Dump new count types.
5953 (profile_count::differs_from_p): Check compatiblity.
5954 (profile_count::to_frequency): New function.
5955 (profile_count::to_cgraph_frequency): New function.
5956 * profile-count.h (struct function): Declare.
5957 (enum profile_quality): Add profile_guessed_local and
5958 profile_guessed_global0.
5959 (class profile_proability): Decrease number of bits to 29;
5960 update from_reg_br_prob_note and to_reg_br_prob_note.
5961 (class profile_count: Update comment; decrease number of bits
5962 to 61. Check compatibility.
5963 (profile_count::compatible_p): New private member function.
5964 (profile_count::ipa_p): New member function.
5965 (profile_count::operator<): Handle global zero correctly.
5966 (profile_count::operator>): Handle global zero correctly.
5967 (profile_count::operator<=): Handle global zero correctly.
5968 (profile_count::operator>=): Handle global zero correctly.
5969 (profile_count::nonzero_p): New member function.
5970 (profile_count::force_nonzero): New member function.
5971 (profile_count::max): New member function.
5972 (profile_count::apply_scale): Handle IPA scalling.
5973 (profile_count::guessed_local): New member function.
5974 (profile_count::global0): New member function.
5975 (profile_count::ipa): New member function.
5976 (profile_count::to_frequency): Declare.
5977 (profile_count::to_cgraph_frequency): Declare.
5978 * profile.c (OVERLAP_BASE): Delete.
5979 (compute_frequency_overlap): Delete.
5980 (compute_branch_probabilities): Do not use compute_frequency_overlap.
5981 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
5982 * sched-ebb.c (rank): Use counts only.
5983 * shrink-wrap.c (handle_simple_exit): Use counts only.
5984 (try_shrink_wrapping): Use counts only.
5985 (place_prologue_for_one_component): Use counts only.
5986 * tracer.c (find_best_predecessor): Use to_frequency.
5987 (find_trace): Use to_frequency.
5988 (tail_duplicate): Use to_frequency.
5989 * trans-mem.c (expand_transaction): Do not update frequency.
5990 * tree-call-cdce.c: Do not update frequency.
5991 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
5992 (gimple_merge_blocks): Likewise.
5993 (gimple_split_edge): Likewise.
5994 (gimple_duplicate_sese_region): Likewise.
5995 (gimple_duplicate_sese_tail): Likewise.
5996 (move_sese_region_to_fn): Likewise.
5997 (gimple_account_profile_record): Likewise.
5998 (insert_cond_bb): Likewise.
5999 * tree-complex.c (expand_complex_div_wide): Likewise.
6000 * tree-eh.c (lower_resx): Update profile.
6001 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
6003 (initialize_cfun): Do not initialize frequencies
6004 (freqs_to_counts): Delete.
6005 (copy_cfg_body): Ignore count parameter.
6006 (copy_body): Update.
6007 (expand_call_inline): Update count_max.
6008 (optimize_inline_calls): Update count_max.
6009 (tree_function_versioning): Update count_max.
6010 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
6011 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
6013 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
6014 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
6015 (try_peel_loop): Likewise.
6016 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
6018 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
6019 (tree_transform_and_unroll_loop): Do not use frequencies
6020 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
6021 Use reliable prediction only.
6022 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
6023 * tree-ssa-sink.c (select_best_block): Use to_frequency.
6024 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
6025 probability scaling.
6026 * tree-ssa-threadupdate.c (create_block_for_threading): Do
6027 not update frequency
6028 (any_remaining_duplicated_blocks): Likewise.
6029 (update_profile): Likewise.
6030 (estimated_freqs_path): Delete.
6031 (freqs_to_counts_path): Delete.
6032 (clear_counts_path): Delete.
6033 (ssa_fix_duplicate_block_edges): Likewise.
6034 (duplicate_thread_path): Likewise.
6035 * tree-switch-conversion.c (gen_inbound_check): Use counts.
6036 * tree-tailcall.c (decrease_profile): Do not update frequency.
6037 (eliminate_tail_call): Likewise.
6038 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
6039 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
6040 (optimize_mask_stores): Likewise.
6041 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
6042 * ubsan.c (ubsan_expand_null_ifn): Update profile.
6043 (ubsan_expand_ptr_ifn): Update profile.
6044 * value-prof.c (gimple_ic): Simplify.
6045 * value-prof.h (gimple_ic): Update prototype.
6046 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
6047 * ipa-inline.c (compute_uninlined_call_time): Be sure that
6049 (want_inline_self_recursive_call_p): Likewise.
6050 (resolve_noninline_speculation): Only cummulate defined counts.
6051 (inline_small_functions): Use nonzero_p.
6052 (ipa_inline): Do not access freed node.
6054 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
6056 * config/aarch64/aarch64.c (aarch64_override_options_internal):
6057 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
6059 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
6061 * config/riscv/riscv.c (riscv_legitimize_move): Handle
6062 non-legitimate address.
6064 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
6066 * config/rs6000/rs6000.md (*lt0_disi): Delete.
6067 (*lt0_<mode>di, *lt0_<mode>si): New.
6069 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
6071 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
6072 TARGET_PAIRED_FLOAT.
6074 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
6075 Jim Wilson <jim.wilson@linaro.org>
6077 * config/aarch64/aarch64-cores.def (saphira): New CPU.
6078 * config/aarch64/aarch64-tune.md: Regenerated.
6079 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
6080 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
6082 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
6084 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
6085 (arc_expand_prologue): Restore blink for millicode.
6086 * config/arc/linux.h (LINK_EH_SPEC): Defined.
6088 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
6091 * config/i386/i386.c (ix86_vector_duplicate_value): Use
6092 gen_vec_duplicate after forcing the scalar into a register.
6094 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
6096 * combine (try_combine): Print the insns input to try_combine to the
6099 2017-11-02 Steve Ellcey <sellcey@cavium.com>
6102 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
6103 Remove second argument from aarch64_process_target_attr call.
6104 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
6106 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
6108 (aarch64_handle_attr_arch): Remove second argument.
6109 (aarch64_handle_attr_cpu): Ditto.
6110 (aarch64_handle_attr_tune): Ditto.
6111 (aarch64_handle_attr_isa_flags): Ditto.
6112 (aarch64_process_one_target_attr): Ditto.
6113 (aarch64_process_target_attr): Ditto.
6114 (aarch64_option_valid_attribute_p): Remove second argument.
6115 on aarch64_process_target_attr call.
6117 2017-11-02 David Malcolm <dmalcolm@redhat.com>
6119 * diagnostic.c: Include "selftest-diagnostic.h".
6120 (selftest::assert_location_text): New function.
6121 (selftest::test_diagnostic_get_location_text): New function.
6122 (selftest::diagnostic_c_tests): Call it.
6124 2017-11-02 David Malcolm <dmalcolm@redhat.com>
6126 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
6127 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
6128 (class selftest::test_diagnostic_context): Move to...
6129 * selftest-diagnostic.c: New file.
6130 * selftest-diagnostic.h: New file.
6132 2017-11-02 James Bowman <james.bowman@ftdichip.com>
6134 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
6135 offset range for FT32B.
6136 * config/ft32/ft32.h: option "mcompress" enables relaxation.
6137 * config/ft32/ft32.md: Add TARGET_NOPM.
6138 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
6139 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
6141 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
6143 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
6145 2017-11-02 Jeff Law <law@redhat.com>
6147 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
6148 virtual keyword on FINAL OVERRIDE members.
6150 * tree-ssa-propagate.h (ssa_propagation_engine): Group
6151 virtuals together. Add virtual destructor.
6152 (substitute_and_fold_engine): Similarly.
6154 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
6156 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
6158 2017-11-02 Richard Biener <rguenther@suse.de>
6160 PR tree-optimization/82795
6161 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
6163 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6165 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
6167 (gcc_GAS_CHECK_FEATURE): Remove.
6168 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
6169 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
6170 * configure: Regenerate.
6172 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
6174 * config/arc/arc.c (hwloop_optimize): Account for empty
6177 2017-11-02 Richard Biener <rguenther@suse.de>
6180 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
6181 Truncate ARRAY_REF index and element size.
6183 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
6185 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
6187 2017-11-01 Jeff Law <law@redhat.com>
6189 * tree-ssa-ccp.c (ccp_folder): New class derived from
6190 substitute_and_fold_engine.
6191 (ccp_folder::get_value): New member function.
6192 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
6193 (ccp_fold_stmt): Remove prototype.
6194 (ccp_finalize): Call substitute_and_fold from the ccp_class.
6195 * tree-ssa-copy.c (copy_folder): New class derived from
6196 substitute_and_fold_engine.
6197 (copy_folder::get_value): Renamed from get_value.
6198 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
6199 * tree-vrp.c (vrp_folder): New class derived from
6200 substitute_and_fold_engine.
6201 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
6202 (vrp_folder::get_value): New member function.
6203 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
6204 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
6205 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
6206 provide a class interface to folder/substitute routines.
6207 (ssa_prop_fold_stmt_fn): Remove typedef.
6208 (ssa_prop_get_value_fn): Likewise.
6209 (subsitute_and_fold): Remove prototype.
6210 (replace_uses_in): Likewise.
6211 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
6212 Renamed from replace_uses_in. Call the virtual member function
6213 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
6214 (substitute_and_fold_dom_walker): Remove initialization of
6215 data member entries for calbacks. Add substitute_and_fold_engine
6216 member and initialize it.
6217 (substitute_and_fold_dom_walker::before_dom_children0: Use the
6218 member functions for get_value, replace_phi_args_in c
6219 replace_uses_in, and fold_stmt calls.
6220 (substitute_and_fold_engine::substitute_and_fold): Renamed from
6221 substitute_and_fold. Remove assert. Update ctor call.
6223 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
6224 (ssa_prop_visit_phi_fn): Likewise.
6225 (class ssa_propagation_engine): New class to provide an interface
6227 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
6229 (ssa_prop_visit_phi): Likewise.
6230 (ssa_propagation_engine::simulate_stmt): Moved into class.
6231 Call visit_phi/visit_stmt from the class rather than via
6232 file scoped static variables.
6233 (ssa_propagation_engine::simulate_block): Moved into class.
6234 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
6235 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
6236 set file scoped statics for the visit_stmt/visit_phi callbacks.
6237 * tree-complex.c (complex_propagate): New class derived from
6238 ssa_propagation_engine.
6239 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
6240 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
6241 (tree_lower_complex): Call ssa_propagate via the complex_propagate
6243 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
6244 ssa_propagation_engine.
6245 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
6246 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
6247 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
6248 * tree-ssa-copy.c (copy_prop): New class derived from
6249 ssa_propagation_engine.
6250 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
6251 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
6252 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
6253 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
6254 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
6255 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
6256 (execute_vrp): Call ssa_propagate from the vrp_prop class.
6258 2017-11-01 Jakub Jelinek <jakub@redhat.com>
6260 PR rtl-optimization/82778
6261 PR rtl-optimization/82597
6262 * compare-elim.c (struct comparison): Add in_a_setter field.
6263 (find_comparison_dom_walker::before_dom_children): Remove killed
6264 bitmap and df_simulate_find_defs call, instead walk the defs.
6265 Compute last_setter and initialize in_a_setter. Merge definitions
6266 with first initialization for a few variables.
6267 (try_validate_parallel): Use insn_invalid_p instead of
6268 recog_memoized. Return insn rather than just the pattern.
6269 (try_merge_compare): Fix up comment. Don't uselessly test if
6270 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
6272 (execute_compare_elim_after_reload): Remove df_chain_add_problem
6275 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6276 Alan Hayward <alan.hayward@arm.com>
6277 David Sherwood <david.sherwood@arm.com>
6279 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
6280 aarch64_hard_regno_nregs to get the number of registers
6283 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6284 Alan Hayward <alan.hayward@arm.com>
6285 David Sherwood <david.sherwood@arm.com>
6287 * config/aarch64/constraints.md (Upl): Rename to...
6289 * config/aarch64/aarch64.md
6290 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
6293 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6294 Alan Hayward <alan.hayward@arm.com>
6295 David Sherwood <david.sherwood@arm.com>
6297 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
6298 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
6301 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6302 Alan Hayward <alan.hayward@arm.com>
6303 David Sherwood <david.sherwood@arm.com>
6305 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
6306 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
6307 (aarch64_evpc_dup): Generate rtl direcly, rather than using
6309 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
6310 of a single element.
6311 * config/aarch64/iterators.md: Add a comment above the permute
6312 unspecs to say that they are generated directly by
6313 aarch64_expand_vec_perm_const.
6314 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
6316 2017-11-01 Nathan Sidwell <nathan@acm.org>
6318 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
6320 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
6322 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
6323 and medany code models, and describe what they do.
6325 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6327 Revert accidental duplicate:
6329 * combine.c (can_change_dest_mode): Reject changes in
6330 REGMODE_NATURAL_SIZE.
6332 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
6334 PR rtl-optimization/64682
6335 PR rtl-optimization/69567
6336 PR rtl-optimization/69737
6337 PR rtl-optimization/82683
6338 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
6339 register mentioned in the note, drop the note, unless it came from I3,
6340 in which case it should go to I3 again.
6342 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6344 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
6345 and return false if not.
6346 (clear_bytes_written_by, live_bytes_read): Update accordingly.
6348 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6350 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
6351 range is known to be empty.
6353 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6354 Alan Hayward <alan.hayward@arm.com>
6355 David Sherwood <david.sherwood@arm.com>
6357 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
6358 and CONST_VECTOR_NUNITS instead of computing the number of units from
6359 the byte sizes of the vector and element.
6360 (simplify_binary_operation_1): Likewise.
6361 (simplify_const_binary_operation): Likewise.
6362 (simplify_ternary_operation): Likewise.
6364 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6365 Alan Hayward <alan.hayward@arm.com>
6366 David Sherwood <david.sherwood@arm.com>
6368 * var-tracking.c (INT_MEM_OFFSET): Replace with...
6369 (int_mem_offset): ...this new function.
6370 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
6371 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
6372 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
6375 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6376 Alan Hayward <alan.hayward@arm.com>
6377 David Sherwood <david.sherwood@arm.com>
6379 * lower-subreg.c (interesting_mode_p): New function.
6380 (compute_costs, find_decomposable_subregs, decompose_register)
6381 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
6382 (resolve_clobber, dump_choices): Use it.
6384 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6385 Alan Hayward <alan.hayward@arm.com>
6386 David Sherwood <david.sherwood@arm.com>
6388 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
6390 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6391 Alan Hayward <alan.hayward@arm.com>
6392 David Sherwood <david.sherwood@arm.com>
6394 * alias.c (find_base_value, find_base_term): Only process integer
6395 truncations. Check the precision rather than the size.
6397 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6398 Alan Hayward <alan.hayward@arm.com>
6399 David Sherwood <david.sherwood@arm.com>
6401 * machmode.h (is_narrower_int_mode): New function
6402 * optabs.c (expand_float, expand_fix): Use it.
6403 * dwarf2out.c (rotate_loc_descriptor): Likewise.
6405 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6406 Alan Hayward <alan.hayward@arm.com>
6407 David Sherwood <david.sherwood@arm.com>
6409 * rtl.h (narrower_subreg_mode): New function.
6410 * ira-color.c (update_costs_from_allocno): Use it.
6412 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6413 Alan Hayward <alan.hayward@arm.com>
6414 David Sherwood <david.sherwood@arm.com>
6416 * optabs-query.h (convert_optab_p): New function, split out from...
6417 (convert_optab_handler): ...here.
6418 (widening_optab_handler): Delete.
6419 (find_widening_optab_handler): Remove permit_non_widening parameter.
6420 (find_widening_optab_handler_and_mode): Likewise. Provide an
6421 override that operates on mode class wrappers.
6422 * optabs-query.c (widening_optab_handler): Delete.
6423 (find_widening_optab_handler_and_mode): Remove permit_non_widening
6424 parameter. Assert that the two modes are the same class and that
6425 the "from" mode is narrower than the "to" mode. Use
6426 convert_optab_handler instead of widening_optab_handler.
6427 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
6428 instead of widening_optab_handler.
6429 * expr.c (expand_expr_real_2): Update calls to
6430 find_widening_optab_handler.
6431 * optabs.c (expand_widen_pattern_expr): Likewise.
6432 (expand_binop_directly): Take the insn_code as a parameter.
6433 (expand_binop): Only call find_widening_optab_handler for
6434 conversion optabs; use optab_handler otherwise. Update calls
6435 to find_widening_optab_handler and expand_binop_directly.
6436 Use convert_optab_handler instead of widening_optab_handler.
6437 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
6438 find_widening_optab_handler and use scalar_mode rather than
6440 (convert_plusminus_to_widen): Likewise.
6442 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6443 Alan Hayward <alan.hayward@arm.com>
6444 David Sherwood <david.sherwood@arm.com>
6446 * machmode.h (fixed_size_mode): New class.
6447 * rtl.h (get_pool_mode): Return fixed_size_mode.
6448 * gengtype.c (main): Add fixed_size_mode.
6449 * target.def (get_raw_result_mode): Return a fixed_size_mode.
6450 (get_raw_arg_mode): Likewise.
6451 * doc/tm.texi: Regenerate.
6452 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
6453 * targhooks.c (default_get_reg_raw_mode): Likewise.
6454 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
6455 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
6456 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
6457 (msp430_get_raw_result_mode): Likewise.
6458 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
6459 * dbxout.c (dbxout_parms): Require fixed-size modes.
6460 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
6461 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
6462 * omp-low.c (lower_oacc_reductions): Likewise.
6463 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
6464 (simplify_subreg): Update accordingly.
6465 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
6466 (force_const_mem): Update accordingly. Return NULL_RTX for modes
6467 that aren't fixed-size.
6468 (get_pool_mode): Return a fixed_size_mode.
6469 (output_constant_pool_2): Take a fixed_size_mode.
6471 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6472 Alan Hayward <alan.hayward@arm.com>
6473 David Sherwood <david.sherwood@arm.com>
6475 * doc/rtl.texi (vec_series): Document.
6476 (const): Say that the operand can be a vec_series.
6477 * rtl.def (VEC_SERIES): New rtx code.
6478 * rtl.h (const_vec_series_p_1): Declare.
6479 (const_vec_series_p): New function.
6480 * emit-rtl.h (gen_const_vec_series): Declare.
6481 (gen_vec_series): Likewise.
6482 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
6483 (gen_vec_series): Likewise.
6484 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
6485 * simplify-rtx.c (simplify_unary_operation): Handle negations
6487 (simplify_binary_operation_series): New function.
6488 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
6489 (test_vector_ops_series): New function.
6490 (test_vector_ops): Call it.
6491 * config/powerpcspe/altivec.md (altivec_lvsl): Use
6492 gen_const_vec_series.
6493 (altivec_lvsr): Likewise.
6494 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
6496 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6497 Alan Hayward <alan.hayward@arm.com>
6498 David Sherwood <david.sherwood@arm.com>
6500 * doc/rtl.texi (const): Update description of address constants.
6501 Say that vector constants are allowed too.
6502 * common.md (E, F): Use CONSTANT_P instead of checking for
6504 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
6505 checking for CONST_VECTOR.
6506 * expmed.c (make_tree): Use build_vector_from_val for a CONST
6508 * expr.c (expand_expr_real_2): Check for vector modes instead
6509 of checking for CONST_VECTOR.
6510 * rtl.h (const_vec_p): New function.
6511 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
6512 (unwrap_const_vec_duplicate): Handle them here too.
6514 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6515 David Malcolm <dmalcolm@redhat.com>
6516 Alan Hayward <alan.hayward@arm.com>
6517 David Sherwood <david.sherwood@arm.com>
6519 * rtl.h (vec_duplicate_p): New function.
6520 * selftest-rtl.c (assert_rtx_eq_at): New function.
6521 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
6522 (assert_rtx_eq_at): Declare.
6523 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
6524 * selftest-run-tests.c (selftest::run_tests): Call it.
6525 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
6526 (simplify_unary_operation_1): Recursively handle vector duplicates.
6527 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
6529 (simplify_subreg): Handle subregs of vector duplicates.
6530 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
6531 (selftest::simplify_rtx_c_tests): New functions.
6533 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6534 Alan Hayward <alan.hayward@arm.com>
6535 David Sherwood <david.sherwood@arm.com>
6537 * emit-rtl.h (gen_const_vec_duplicate): Declare.
6538 (gen_vec_duplicate): Likewise.
6539 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
6541 (gen_const_vector): ...here.
6542 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
6543 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
6544 whose elements are all equal.
6545 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
6546 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
6547 (simplify_relational_operation): Likewise.
6548 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
6550 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
6551 (aarch64_expand_vector_init): Likewise.
6552 * config/arm/arm.c (neon_vdup_constant): Likewise.
6553 (neon_expand_vector_init): Likewise.
6554 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
6555 (arm_block_set_unaligned_vect): Likewise.
6556 (arm_block_set_aligned_vect): Likewise.
6557 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
6558 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
6559 (expand_vec_perm_even_odd_pack): Likewise.
6560 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
6561 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
6562 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
6563 gen_const_vec_duplicate.
6564 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
6565 * config/mips/mips.c (mips_gen_const_int_vector): Use
6566 gen_const_vec_duplicate.
6567 (mips_expand_vector_init): Use CONST0_RTX.
6568 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
6569 (define_split): Use gen_const_vec_duplicate.
6570 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
6571 (define_split): Use gen_const_vec_duplicate.
6572 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
6573 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
6574 * config/spu/spu.c (spu_const): Likewise.
6576 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6577 Alan Hayward <alan.hayward@arm.com>
6578 David Sherwood <david.sherwood@arm.com>
6580 * combine.c (can_change_dest_mode): Reject changes in
6581 REGMODE_NATURAL_SIZE.
6583 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
6585 * configure.ac (--enable-libssp): New.
6586 (gcc_cv_libc_provides_ssp): Check for explicit setting before
6587 trying to determine target-specific default. Adjust indentation.
6588 * configure: Regenerated.
6589 * doc/install.texi (Configuration): Expand --disable-libssp
6592 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
6594 config/i386/i386.c (ix86_expand_epilogue): Correct stack
6597 2017-10-31 Martin Jambor <mjambor@suse.cz>
6600 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
6602 2017-10-31 David Malcolm <dmalcolm@redhat.com>
6604 * auto-profile.c (autofdo_source_profile::read): Use
6605 UNKNOWN_LOCATION rather than 0.
6606 * diagnostic-core.h (warning_at_rich_loc): Rename to...
6607 (warning_at): ...this overload.
6608 (warning_at_rich_loc_n): Rename to...
6609 (warning_n): ...this overload.
6610 (error_at_rich_loc): Rename to...
6611 (error_at): ...this overload.
6612 (pedwarn_at_rich_loc): Rename to...
6613 (pedwarn): ...this overload.
6614 (permerror_at_rich_loc): Rename to...
6615 (permerror): ...this overload.
6616 (inform_at_rich_loc): Rename to...
6617 (inform): ...this overload.
6618 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
6619 (diagnostic_n_impl_richloc): Rename to...
6620 (diagnostic_n_impl): ...this rich_location *-based decl.
6621 (inform_at_rich_loc): Rename to...
6622 (inform): ...this, and add an assertion.
6623 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
6624 (warning_at_rich_loc): Rename to...
6625 (warning_at): ...this, and add an assertion.
6626 (warning_at_rich_loc_n): Rename to...
6627 (warning_n): ...this, and add an assertion.
6628 (warning_n): Update location_t-based implementation for removal of
6629 location_t-based diagnostic_n_impl.
6630 (pedwarn_at_rich_loc): Rename to...
6631 (pedwarn): ...this, and add an assertion.
6632 (permerror_at_rich_loc): Rename to...
6633 (permerror): ...this, and add an assertion.
6634 (error_n): Update for removal of location_t-based diagnostic_n_impl.
6635 (error_at_rich_loc): Rename to...
6636 (error_at): ...this, and add an assertion.
6637 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
6638 (driver::do_spec_on_infiles): Likewise.
6639 * substring-locations.c (format_warning_va): Update for renaming
6640 of inform_at_rich_loc.
6642 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
6644 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
6645 _Float<N>X built-in functions so that the variant without the
6646 "__builtin_" prefix is only enabled for the GNU C and Objective C
6647 languages when they are in non-strict ANSI/ISO mode.
6648 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
6649 * target.def (floatn_builtin_p): Add a target hook to control
6650 whether _Float<N> and _Float<N>X built-in functions without the
6651 "__builtin_" prefix are enabled, and return true for C and
6652 Objective C in the default hook. Include langhooks.h in
6654 * targhooks.h (default_floatn_builtin_p): Likewise.
6655 * targhooks.c (default_floatn_builtin_p): Likewise.
6656 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
6657 floatn_builtin_p target hook.
6658 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
6660 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
6661 Eric Botcazou <ebotcazou@adacore.com>
6663 PR rtl-optimization/81803
6664 * lra-constraints.c (curr_insn_transform): Also reload the whole
6665 register for a strict subreg no wider than a word if this is for
6666 a WORD_REGISTER_OPERATIONS target.
6668 2017-10-31 Jason Merrill <jason@redhat.com>
6670 * gdbinit.in: Skip over inlines from timevar.h.
6672 2017-10-31 Martin Liska <mliska@suse.cz>
6674 * doc/gcov.texi: Document new option.
6675 * gcov.c (print_usage): Likewise print it.
6676 (process_args): Support the argument.
6677 (format_count): New function.
6678 (format_gcov): Use the function.
6680 2017-10-31 Martin Liska <mliska@suse.cz>
6682 * gcov.c (struct name_map): do not use typedef.
6683 Define operator== and operator<.
6684 (name_search): Remove.
6685 (name_sort): Remove.
6686 (main): Do not allocate names.
6687 (process_file): Add vertical space.
6688 (generate_results): Use std::find.
6689 (release_structures): Do not release memory.
6690 (find_source): Use std::find.
6692 2017-10-31 Martin Liska <mliska@suse.cz>
6694 * gcov.c (struct line_info): Remove it's typedef.
6695 (line_info::line_info): Add proper ctor.
6696 (line_info::has_block): Do not use a typedef.
6697 (struct source_info): Do not use typedef.
6698 (circuit): Likewise.
6699 (get_cycles_count): Likewise.
6700 (output_intermediate_file): Iterate via vector iterator.
6701 (add_line_counts): Use std::vector methods.
6702 (accumulate_line_counts): Likewise.
6703 (output_lines): Likewise.
6705 2017-10-31 Martin Liska <mliska@suse.cz>
6707 * gcov.c (struct source_info): Remove typedef.
6708 (source_info::source_info): Add proper ctor.
6709 (accumulate_line_counts): Use struct, not it's typedef.
6710 (output_gcov_file): Likewise.
6711 (output_lines): Likewise.
6712 (main): Do not allocate an array.
6713 (output_intermediate_file): Use size of vector container.
6714 (process_file): Resize the vector.
6715 (generate_results): Do not preallocate, use newly added vector
6717 (release_structures): Do not release sources.
6718 (find_source): Use vector methods.
6719 (add_line_counts): Do not use typedef.
6721 2017-10-31 Martin Liska <mliska@suse.cz>
6723 * doc/gcov.texi: Document that.
6724 * gcov.c (add_line_counts): Mark lines with a non-executed
6726 (output_line_beginning): Handle such lines.
6727 (output_lines): Pass new argument.
6728 (output_intermediate_file): Print it in intermediate format.
6730 2017-10-31 Martin Liska <mliska@suse.cz>
6732 * color-macros.h: New file.
6733 * diagnostic-color.c: Factor out color related to macros to
6735 * doc/gcov.texi: Document -k option.
6736 * gcov.c (INCLUDE_STRING): Include string.h.
6737 (print_usage): Add -k option.
6738 (process_args): Parse it.
6739 (pad_count_string): New function.
6740 (output_line_beginning): Likewise.
6741 (DEFAULT_LINE_START): New macro.
6742 (output_lines): Support color output.
6744 2017-10-31 Martin Liska <mliska@suse.cz>
6746 PR gcov-profile/82633
6747 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
6748 their interaction with GCOV infrastructure.
6749 * configure.ac: Add -fkeep-{inline,static}-functions to
6751 * configure: Regenerate.
6753 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
6756 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
6758 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
6761 * config/rs6000/rs6000.md (allocate_stack): Force update interval
6762 into a register if it does not fit into an immediate offset field.
6764 2017-10-31 Olivier Hainque <hainque@adacore.com>
6766 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
6768 2017-10-31 Julia Koval <julia.koval@intel.com>
6770 * config.gcc: Add gfniintrin.h.
6771 * config/i386/gfniintrin.h: New.
6772 * config/i386/i386-builtin-types.def
6773 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
6774 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
6775 __builtin_ia32_vgf2p8affineinvqb_v32qi,
6776 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
6777 __builtin_ia32_vgf2p8affineinvqb_v16qi,
6778 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
6779 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
6780 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
6781 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
6782 V64QI_FTYPE_V64QI_V64QI_INT): New types.
6783 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
6784 * config/i386/immintrin.h: Include gfniintrin.h.
6785 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
6787 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
6789 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
6791 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
6793 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
6794 (arm_ashldi3_1bit): Remove pattern.
6795 (ashrdi3): Remove shift by 1 expansion.
6796 (arm_ashrdi3_1bit): Remove pattern.
6797 (lshrdi3): Remove shift by 1 expansion.
6798 (arm_lshrdi3_1bit): Remove pattern.
6799 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
6800 cost of ashldi3 by 1.
6801 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
6802 (<shift>di3_neon): Likewise.
6804 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6806 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
6807 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
6808 and (*aarch64_simd_mov<VQ:mode>).
6809 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
6810 pattern alternative.
6811 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
6812 attributes to match pattern alternative.
6814 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
6816 * config.gcc (powerpc*-*-*): Add emmintrin.h.
6817 * config/rs6000/emmintrin.h: New file.
6818 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
6820 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
6822 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
6823 * (movdi_vfp_cortexa8): Remove pattern.
6825 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6827 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
6829 (Specific, alpha*-dec-osf5.1): Remove.
6830 (Specific, mips-sgi-irix5): Remove.
6831 (Specific, mips-sgi-irix6): Remove.
6833 2017-10-30 Jakub Jelinek <jakub@redhat.com>
6836 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
6837 arguments to clear_bit_region_be.
6839 2017-10-30 Jim Wilson <wilson@tuliptree.org>
6841 * gimplify.c: Include memmodel.h.
6843 2017-10-30 Martin Jambor <mjambor@suse.cz>
6845 * omp-grid.c (grid_attempt_target_gridification): Also insert a
6846 condition whether loop should be executed at all.
6848 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
6850 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6851 gimple folding of vec_madd() intrinsics.
6852 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
6853 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
6854 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
6856 2017-10-30 Richard Biener <rguenther@suse.de>
6858 PR tree-optimization/82762
6860 2017-10-23 Richard Biener <rguenther@suse.de>
6862 PR tree-optimization/82129
6864 2017-08-01 Richard Biener <rguenther@suse.de>
6866 PR tree-optimization/81181
6867 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6868 (compute_antic): ... end of iteration here.
6870 2017-10-30 Joseph Myers <joseph@codesourcery.com>
6872 * doc/invoke.texi (C Dialect Options): Document -std=c17,
6873 -std=iso9899:2017 and -std=gnu17.
6874 * doc/standards.texi (C Language): Document C17 support.
6875 * doc/cpp.texi (Overview): Mention -std=c17.
6876 (Standard Predefined Macros): Document C11 and C17 values of
6877 __STDC_VERSION__. Do not refer to C99 support as incomplete.
6878 * doc/extend.texi (Inline): Do not list individual options for
6879 standards newer than C99.
6880 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
6882 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
6885 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
6887 * asan.c (asan_finish_file): Align asan globals array by shadow
6890 2017-10-30 Jakub Jelinek <jakub@redhat.com>
6893 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
6894 (struct store_immediate_info): Add bitregion_start and bitregion_end
6896 (store_immediate_info::store_immediate_info): Add brs and bre
6897 arguments and initialize bitregion_{start,end} from those.
6898 (struct merged_store_group): Add bitregion_start, bitregion_end,
6899 align_base and mask fields. Drop unnecessary struct keyword from
6900 struct store_immediate_info. Add do_merge method.
6901 (clear_bit_region_be): Use memset instead of loop storing zeros.
6902 (merged_store_group::do_merge): New method.
6903 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
6904 stores as long as the surrounding bitregions have no gaps.
6905 (merged_store_group::merge_overlapping): Use do_merge.
6906 (merged_store_group::apply_stores): Test that bitregion_{start,end}
6907 is byte aligned, rather than requiring that start and width are
6908 byte aligned. Drop unnecessary struct keyword from
6909 struct store_immediate_info. Allocate and populate also mask array.
6910 Make start of the arrays relative to bitregion_start rather than
6911 start and size them according to bitregion_{end,start} difference.
6912 (struct imm_store_chain_info): Drop unnecessary struct keyword from
6913 struct store_immediate_info.
6914 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
6915 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
6916 struct keyword from struct store_immediate_info.
6917 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
6918 between stores as long as the surrounding bitregions have no gaps.
6920 (struct split_store): Add orig non-static data member.
6921 (split_store::split_store): Initialize orig to false.
6922 (find_constituent_stmts): Return store_immediate_info *, non-NULL
6923 if there is exactly a single original stmt. Change stmts argument
6924 to pointer from reference, if NULL, don't push anything to it. Add
6925 first argument, use it to optimize skipping over orig stmts that
6926 are known to be before bitpos already. Simplify.
6927 (split_group): Return unsigned int count how many stores are or
6928 would be needed rather than a bool. Add allow_unaligned argument.
6929 Change split_stores argument from reference to pointer, if NULL,
6930 only do a dry run computing how many stores would be produced.
6931 Rewritten algorithm to use both alignment and misalign if
6932 !allow_unaligned and handle bitfield stores with gaps.
6933 (imm_store_chain_info::output_merged_store): Set start_byte_pos
6934 from bitregion_start instead of start. Compute allow_unaligned
6935 here, if true, do 2 split_group dry runs to compute which one
6936 produces fewer stores and prefer aligned if equal. Punt if
6937 new count is bigger or equal than original before emitting any
6938 statements, rather than during that. Remove no longer needed
6939 new_ssa_names tracking. Replace num_stmts with
6940 split_stores.length (). Use 32-bit stack allocated entries
6941 in split_stores auto_vec. Try to reuse original store lhs/rhs1
6942 if possible. Handle bitfields with gaps.
6943 (pass_store_merging::execute): Ignore bitsize == 0 stores.
6944 Compute bitregion_{start,end} for the stores and construct
6945 store_immediate_info with that. Formatting fixes.
6947 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
6950 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
6951 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
6953 2017-10-29 Jim Wilson <wilson@tuliptree.org>
6955 * gimplify.c: Include tm_p.h.
6957 * common.opt (gcoff): Re-add as ignored option.
6958 (gcoff1, gcoff2, gcoff3): Likewise.
6960 * Makefile.in (OBJS): Delete sdbout.o.
6961 (GTFILES): Delete $(srcdir)/sdbout.c.
6962 * debug.h: Delete sdb_debug_hooks.
6963 * final.c: Delete sdbout.h include.
6964 (final_scan_insn): Delete SDB_DEBUG check.
6965 (rest_of_clean_state): Likewise.
6966 * output.h: Delete sdb_begin_function_line.
6969 * toplev.c: Delete sdbout.h include.
6970 (process_options): Delete SDB_DEBUG check.
6971 * tree-core.h (tree_type_common): Delete pointer field of
6973 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
6974 TYPE_SYMTAB_POINTER.
6975 * tree.h (TYPE_SYMTAB_POINTER): Delete.
6976 (TYPE_SYMTAB_IS_POINTER): Delete.
6977 (TYPE_SYMTAB_IS_DIE): Renumber.
6978 * xcoffout.c: Refer to former sdbout.c file.
6979 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
6981 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
6982 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
6983 (Debugging Options): Delete -gcoff.
6984 (-gstabs): Delete SDB reference.
6986 (-gcoff@var{level}): Delete.
6987 * doc/passes.texi (Debugging information output): Delete SDB and
6988 sdbout.c references.
6989 * doc/tm.texi: Regenerate.
6990 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
6991 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
6993 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
6994 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
6995 (SDB_DEBUGGING_INFO): Delete.
6996 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
6997 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
6998 * target.def (output_source_filename): Delete COFF reference.
7000 * common.opt (gcoff): Delete.
7001 (gxcoff+): Update Negative chain.
7002 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
7004 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
7005 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
7006 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
7008 (expand_function_start): Change sdb reference to past tense.
7009 (expand_function_end): Change sdb reference to past tense.
7010 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
7011 * opts.c (debug_type_names): Delete coff entry.
7012 (common_handle_option): Delete OPT_gcoff case.
7013 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
7015 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
7016 * config/cris/cris.h: Delete SDB reference in comment.
7017 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
7018 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
7019 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
7020 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
7022 (ix86_expand_prologue): Likewise.
7023 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
7024 * config/ia64/ia64.h: Likewise.
7025 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
7026 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
7028 * config/mmix/mmix.h: Likewise.
7029 * config/nds32/nds32.c: Likewise.
7030 * config/stormy/storym16.h: Likewise.
7031 * config/visium/visium.h: Likewise.
7032 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
7034 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
7036 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
7037 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
7038 FRAME_POINTER_REGNUM point at high end of local var area.
7040 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
7042 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
7043 Move comment around. Do not reset best_edge for a copiable
7044 destination if the copy would cause a partition change.
7045 (better_edge_p): Remove redundant check.
7047 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
7049 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
7051 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
7053 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
7054 for math functions that have _Float<N> and _Float<N>X variants.
7055 (mathfn_built_in_2): Add support for math functions that have
7056 _Float<N> and _Float<N>X variants.
7057 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
7058 (expand_builtin_mathfn_ternary): Add support for fma with
7059 _Float<N> and _Float<N>X variants.
7060 (expand_builtin): Likewise.
7061 (fold_builtin_3): Likewise.
7062 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
7063 create math function _Float<N> and _Float<N>X variants as external
7065 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
7066 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
7067 the __builtin_ prefix and if not strict ansi, without the prefix.
7068 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
7069 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
7070 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
7071 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
7072 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
7073 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
7074 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
7075 function signatures for fma _Float<N> and _Float<N>X variants.
7076 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
7077 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
7078 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
7079 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
7080 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
7081 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
7082 * gencfn-macros.c (print_case_cfn): Add support for math functions
7083 that have _Float<N> and _Float<N>X variants.
7084 (print_define_operator_list): Likewise.
7085 (fltfn_suffixes): Likewise.
7087 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
7088 for math functions that have _Float<N> and _Float<N>X variants.
7089 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
7090 and _Float<N>X variants.
7091 (COPYSIGN): Likewise.
7094 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
7095 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
7097 (integer_valued_read_call_p): Likewise.
7098 * fold-const-call.c (fold_const_call_ss): Likewise.
7099 (fold_const_call_sss): Add support for copysign, fmin, and fmax
7100 _Float<N> and _Float<N>X variants.
7101 (fold_const_call_ssss): Add support for fma _Float<N> and
7102 _Float<N>X variants.
7103 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
7104 support for copysign and fma _Float<N> and _Float<N>X variants.
7105 (backprop::process_builtin_call_use): Likewise.
7106 * tree-call-cdce.c (can_test_argument_range); Add support for
7107 sqrt _Float<N> and _Float<N>X variants.
7108 (edom_only_function): Likewise.
7109 (get_no_error_domain): Likewise.
7110 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
7111 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
7112 copysign _Float<N> and _Float<N>X variants.
7113 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
7114 handled by machine independent code.
7115 (FMAF128): Likewise.
7116 * doc/cpp.texi (Common Predefined Macros): Document defining
7117 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
7118 fma _Float<N> and _Float<N>X variants.
7120 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
7123 * config/i386/i386-modes.def (CCFPU): Remove definition.
7124 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
7125 (ix86_cc_modes_compatible): Ditto.
7126 (ix86_expand_carry_flag_compare): Ditto.
7127 (ix86_expand_int_movcc): Ditto.
7128 (ix86_expand_int_addcc): Ditto.
7129 (ix86_reverse_condition): Ditto.
7130 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
7131 Return true/false for unordered/ordered fp comparisons.
7132 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
7133 (ix86_prepare_fp_compare_args): Update for rename.
7134 (ix86_expand_fp_compare): Update for rename. Generate unordered
7135 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
7136 (ix86_expand_sse_compare_and_jump): Ditto.
7137 * config/i386/predicates.md (fcmov_comparison_operator):
7138 Remove CCFPU mode handling.
7139 (ix86_comparison_operator): Ditto.
7140 (ix86_carry_flag_operator): Ditto.
7141 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
7142 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
7143 (*cmpu<mode>_cc_i387): Ditto.
7144 (FPCMP): Remove mode iterator.
7145 (unord): Remove mode attribute.
7146 (unord_subst): New define_subst transformation
7147 (unord): New define_subst attribute.
7149 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
7150 (*cmpi<unord>xf_i387): Ditto.
7151 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
7152 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
7153 using unord_subst transformation.
7154 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
7155 (round_saeonly): Also handle CCFP mode.
7156 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
7157 Remove UNSPEC_SAHF unspec handling.
7159 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
7161 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
7163 2017-10-27 Jeff Law <law@redhat.com>
7165 * gimple-ssa-sprintf.c: Include domwalk.h.
7166 (class sprintf_dom_walker): New class, derived from dom_walker.
7167 (sprintf_dom_walker::before_dom_children): New function.
7168 (struct call_info): Moved into sprintf_dom_walker class
7169 (compute_formath_length, handle_gimple_call): Likewise.
7170 (sprintf_length::execute): Call the dominator walker rather
7171 than walking the statements.
7173 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
7174 gimple statement locations.
7175 (check_array_bounds): Corresponding changes. Get the statement's
7176 location directly from wi->stmt.
7178 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
7181 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
7183 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
7185 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
7186 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
7188 2017-10-27 Jakub Jelinek <jakub@redhat.com>
7191 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
7192 * config/i386/i386.c (maybe_get_pool_constant): Removed.
7193 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
7194 maybe_get_pool_constant.
7195 * config/i386/predicates.md (zero_extended_scalar_load_operand):
7198 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7200 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
7201 2.26 caveat. Update gas and gld versions.
7202 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
7205 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7207 * cgraph.h (set_malloc_flag): Declare.
7208 * cgraph.c (set_malloc_flag_1): New function.
7209 (set_malloc_flag): Likewise.
7210 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
7211 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
7213 (read_ipa_call_summary): Add support for reading is_return_callee.
7214 (write_ipa_call_summary): Stream is_return_callee.
7215 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
7216 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
7217 ipa-prop.h, ipa-fnsummary.h.
7218 (pure_const_names): Change to static.
7219 (malloc_state_e): Define.
7220 (malloc_state_names): Define.
7221 (funct_state_d): Add field malloc_state.
7222 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
7223 (check_retval_uses): New function.
7224 (malloc_candidate_p): Likewise.
7225 (analyze_function): Add support for malloc attribute.
7226 (pure_const_write_summary): Stream malloc_state.
7227 (pure_const_read_summary): Add support for reading malloc_state.
7228 (dump_malloc_lattice): New function.
7229 (propagate_malloc): New function.
7230 (warn_function_malloc): New function.
7231 (ipa_pure_const::execute): Call propagate_malloc and
7232 ipa_free_fn_summary.
7233 (pass_local_pure_const::execute): Add support for malloc attribute.
7234 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
7235 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
7237 2017-10-27 Martin Liska <mliska@suse.cz>
7239 PR gcov-profile/82457
7240 * doc/invoke.texi: Document that one needs a non-strict ISO mode
7241 for fork-like functions to be properly instrumented.
7243 2017-10-27 Richard Biener <rguenther@suse.de>
7246 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
7247 info when we redirected EH.
7249 2017-10-26 Michael Collison <michael.collison@arm.com>
7251 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
7253 (<optab>_trunchf<GPI:mode>2: New pattern.
7254 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
7255 * config/aarch64/iterators.md (wv): New mode attribute.
7256 (vf, VF): New mode attributes.
7257 (vgp, VGP): New mode attributes.
7258 (s): Update attribute with SImode and DImode prefixes.
7260 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
7262 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
7263 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
7264 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
7265 (nios2_option_overide): Initialize it. Don't allow R0-relative
7266 addressing with PIC.
7267 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
7268 (nios2_symbolic_constant_p): Likewise.
7269 (nios2_legitimate_address_p): Likewise.
7270 (nios2_r0rel_section_name_p): New.
7271 (nios2_symbol_ref_in_r0rel_data_p): New.
7272 (nios2_emit_move_sequence): Handle r0rel_constant_p.
7273 (r0rel_constant_p): New.
7274 (nios2_print_operand_address): Handle r0rel_constant_p.
7275 (nios2_cdx_narrow_form_p): Likewise.
7276 * config/nios2/nios2.opt (mr0rel-sec=): New option.
7277 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
7278 (Nios II Options): Document -mr0rel-sec.
7280 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
7282 * config/nios2/nios2.c: Include xregex.h.
7283 (nios2_gprel_sec_regex): New.
7284 (nios2_option_overide): Initialize it. Don't allow GP-relative
7285 addressing with PIC.
7286 (nios2_small_section_name_p): Check for regex match.
7287 * config/nios2/nios2.opt (mgprel-sec=): New option.
7288 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
7289 (Nios II Options): Document -mgprel-sec.
7291 2017-10-26 Jim Wilson <wilson@tuliptree.org>
7293 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
7295 2017-10-26 Tom de Vries <tom@codesourcery.com>
7297 PR tree-optimization/82707
7298 * gimple.c (gimple_copy): Fix unsharing of
7299 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
7301 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
7303 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
7304 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
7305 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
7306 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
7307 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
7308 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
7309 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
7310 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
7311 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
7312 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
7313 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
7314 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
7315 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
7316 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
7317 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
7318 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
7319 _mm512_mask_cmpunord_ps_mask): New intrinsics.
7321 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7323 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
7325 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
7326 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
7327 warning to rs6000.c. Remove the Undocumented flag, since it has
7329 (-mabi=ibmlongdouble): Likewise.
7330 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
7331 already set, set the default format for long double.
7332 (rs6000_debug_reg_global): Print whether long double is IBM or
7334 (rs6000_option_override_internal): Rework setting long double
7335 format. Only warn if the user is changing the long double default
7336 and they did not use -Wno-psabi.
7337 * doc/invoke.texi (PowerPC options): Update the documentation for
7338 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
7340 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7341 Alan Hayward <alan.hayward@arm.com>
7342 David Sherwood <david.sherwood@arm.com>
7344 * rtl.h (wider_subreg_mode): New function.
7345 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
7346 rather than an unsigned int *.
7347 * ira-color.c (regno_max_ref_width): Replace with...
7348 (regno_max_ref_mode): ...this new variable.
7349 (coalesced_pseudo_reg_slot_compare): Update accordingly.
7350 Use wider_subreg_mode.
7351 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
7352 rather than an unsigned int *.
7353 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
7354 (process_alt_operands): Likewise.
7355 (invariant_p): Likewise.
7356 * lra-spills.c (assign_mem_slot): Likewise.
7357 (add_pseudo_to_slot): Likewise.
7358 * lra.c (collect_non_operand_hard_regs): Likewise.
7359 (add_regs_to_insn_regno_info): Likewise.
7360 * reload1.c (regno_max_ref_width): Replace with...
7361 (regno_max_ref_mode): ...this new variable.
7362 (reload): Update accordingly. Update call to
7363 ira_sort_regnos_for_alter_reg.
7364 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
7365 (init_eliminable_invariants): Update to use regno_max_ref_mode.
7366 (scan_paradoxical_subregs): Likewise.
7368 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7370 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
7371 (aarch64_frame): Add emit_frame_chain boolean.
7372 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7373 Move eh_return case to aarch64_layout_frame.
7374 (aarch64_layout_frame): Initialize emit_frame_chain.
7375 (aarch64_expand_prologue): Use emit_frame_chain.
7377 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7379 * config/aarch64/aarch64.c (aarch64_layout_frame):
7380 Ensure LR is always stored at the bottom of the callee-saves.
7381 Remove rarely used frame layout which saves callee-saves at top of
7382 frame, so the store of LR can be used as a valid probe in all cases.
7384 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7386 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
7387 Improve unaligned TImode/TFmode base/offset split.
7389 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7390 Alan Hayward <alan.hayward@arm.com>
7391 David Sherwood <david.sherwood@arm.com>
7393 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
7394 * combine.c (find_single_use_1): Likewise.
7395 (expand_field_assignment): Likewise.
7396 (move_deaths): Likewise.
7397 * lra-constraints.c (simplify_operand_subreg): Likewise.
7398 (curr_insn_transform): Likewise.
7399 * lra.c (collect_non_operand_hard_regs): Likewise.
7400 (add_regs_to_insn_regno_info): Likewise.
7401 * rtlanal.c (reg_referenced_p): Likewise.
7402 (covers_regno_no_parallel_p): Likewise.
7404 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7406 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
7407 Don't print any bits outside the precision of the value.
7408 * wide-int.cc (test_printing): Add some new tests.
7410 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7412 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
7413 supports -xbrace_comment option.
7414 * configure: Regenerate.
7415 * config.in: Regenerate.
7416 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
7417 (ASM_CPU_SPEC): Use it.
7419 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7421 * target.def (static_rtx_alignment): New hook.
7422 * targhooks.h (default_static_rtx_alignment): Declare.
7423 * targhooks.c (default_static_rtx_alignment): New function.
7424 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
7425 * doc/tm.texi: Regenerate.
7426 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
7427 instead of targetm.constant_alignment. Remove call to
7429 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7430 (cris_preferred_mininum_alignment): New function, split out from...
7431 (cris_constant_alignment): ...here.
7432 (cris_static_rtx_alignment): New function.
7433 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
7435 (ix86_constant_alignment): ...here.
7436 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7437 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7438 (mmix_static_rtx_alignment): New function.
7439 * config/spu/spu.c (spu_static_rtx_alignment): New function.
7440 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7442 2017-10-26 Tamar Christina <tamar.christina@arm.com>
7445 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
7446 Add flag_trapping_math and flag_fp_int_builtin_inexact.
7448 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
7450 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
7451 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
7452 mark as a sign-extending load.
7453 (local_pic_load_u): Define.
7455 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
7458 * fold-const.c (operand_equal_for_comparison_p): Also return true
7459 if ARG0 is a simple variant of ARG1 with narrower precision.
7460 (fold_ternary_loc): Always pass unstripped operands to the predicate.
7462 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
7464 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
7466 * i386.h (processor_costs): Add gather_static, gather_per_elt,
7467 scatter_static, scatter_per_elt.
7468 * x86-tune-costs.h: Add new cost entries.
7470 2017-10-25 Richard Biener <rguenther@suse.de>
7472 * tree-ssa-sccvn.h (vn_eliminate): Declare.
7473 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
7474 class pass_fre): Move to ...
7475 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
7476 class pass_fre): ... here and adjust for statistics.
7478 2017-10-25 Jakub Jelinek <jakub@redhat.com>
7481 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
7482 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
7483 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
7484 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
7487 2017-10-25 Richard Biener <rguenther@suse.de>
7489 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
7490 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
7491 eliminate_push_avail, eliminate_insert): Move inside...
7492 (class eliminate_dom_walker): ... this class in preparation
7494 (fini_eliminate): Remove by merging with ...
7495 (eliminate): ... this function. Adjust for class changes.
7496 (pass_pre::execute): Remove fini_eliminate call.
7497 (pass_fre::execute): Likewise.
7499 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7502 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
7503 (VPERMI2, VPERMI2I): New mode iterators.
7504 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
7505 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
7507 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
7508 mode iterator. Remove 3 old define_insn patterns.
7509 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
7510 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
7511 VPERMI2 mode iterator, remove the other two expanders.
7512 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
7513 to use VPERMI2 mode iterator, add another alternative for vpermi2*
7514 instructions, remove the other two patterns.
7515 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
7516 mode iterator, remove the other two patterns.
7517 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
7518 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
7519 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
7520 and adjust argument order accordingly.
7521 (ix86_expand_vec_perm): Adjust caller.
7522 (expand_vec_perm_1): Likewise.
7523 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
7524 (expand_vec_perm_vpermt2_vpshub2): ... this.
7525 (ix86_expand_vec_perm_const_1): Adjust caller.
7526 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
7529 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
7530 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
7531 (vec_shl_<mode>): Remove unused expander.
7532 (avx512bw_<shift_insn><mode>3): New define_insn.
7533 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
7534 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
7536 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
7539 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
7542 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7544 PR rtl-optimization/82396
7545 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
7546 (autopref_multipass_init): Simplify initialization.
7547 (autopref_rank_data): Simplify sort order.
7548 * gcc/sched-int.h (autopref_multipass_data_): Remove
7549 multi_mem_insn_p, min_offset and max_offset.
7551 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7554 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7555 Check special value of flag_omit_frame_pointer.
7556 (aarch64_can_eliminate): Likewise.
7557 (aarch64_override_options_after_change_1): Simplify handling of
7558 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
7560 2017-10-24 Richard Biener <rguenther@suse.de>
7562 PR tree-optimization/82697
7563 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
7564 zero for conditional load and unconditional store.
7566 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7568 * doc/install.texi: Document bootstrap-cet.
7570 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7573 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
7574 ENDBR instruction at function entrance if function is only
7577 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7580 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
7581 patterns to better describe from which operation the CF is computed.
7582 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
7583 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
7584 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
7585 is 0, use _0 suffixed expanders instead of emitting a comparison
7588 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
7590 * config/i386/i386.md(*movsf_internal, *movdf_internal):
7591 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
7593 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
7596 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
7597 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
7598 * loop-iv.c (iv_get_reaching_def): Likewise.
7599 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
7600 variable is promoted and the partition contains undefined values.
7602 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7604 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
7606 (nios2_address_cost): Define.
7607 (nios2_legitimize_address): Recognize (exp + constant) directly.
7608 (TARGET_ADDRESS_COST): Define.
7610 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7612 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
7613 (nios2_symbolic_memory_operand_p): Declare.
7614 (nios2_split_large_constant): Declare.
7615 (nios2_split_symbolic_memory_operand): Declare.
7616 * config/nios2/nios2.c: Adjust includes.
7617 (nios2_symbolic_constant_allowed): New.
7618 (nios2_symbolic_constant_p): New.
7619 (nios2_plus_symbolic_constant_p): New.
7620 (nios2_valid_addr_expr_p): Recognize addresses involving
7622 (nios2_legitimate_address_p): Likewise, also LO_SUM.
7623 (nios2_symbolic_memory_operand_p): New.
7624 (nios2_large_constant_p): New.
7625 (nios2_split_large_constant): New.
7626 (nios2_split_plus_large_constant): New.
7627 (nios2_split_symbolic_memory_operand): New.
7628 (nios2_legitimize_address): Code refactoring. Handle addresses
7629 involving symbolic constants.
7630 (nios2_emit_move_sequence): Likewise.
7631 (nios2_print_operand): Improve error output.
7632 (nios2_print_operand_address): Handle LO_SUM.
7633 (nios2_cdx_narrow_form_p): Likewise.
7634 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
7635 operands involving symbolic constants.
7636 (movhi_internal, movsi_internal): Likewise.
7637 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
7638 (extendhisi2, extendqi<mode>2): Likewise.
7640 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7642 * tree-pass.h (PROP_rtl_split_insns): Define.
7643 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
7645 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7647 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
7649 2017-10-23 Jakub Jelinek <jakub@redhat.com>
7652 * target.def (const_not_ok_for_debug_p): Default to
7653 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
7654 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
7655 * targhooks.c (default_const_not_ok_for_debug_p): New function.
7656 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
7657 which targetm.const_not_ok_for_debug_p returned true.
7658 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
7660 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
7662 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
7663 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
7664 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
7666 (ix86_const_not_ok_for_debug_p): New function.
7667 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
7668 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
7670 2017-10-23 David Malcolm <dmalcolm@redhat.com>
7673 * system.h: Conditionally include "unique-ptr.h" if
7674 INCLUDE_UNIQUE_PTR is defined.
7675 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
7676 of defining INCLUDE_UNIQUE_PTR before including "system.h".
7678 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
7680 * config/rl78/rl78.md: New define_expand "subdi3".
7682 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
7685 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
7686 DF_REF_INSN if DF_REF_INSN_INFO is false.
7688 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
7690 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
7691 xmm_move instead of sse_move.
7692 (sse_store_index): New function.
7693 (ix86_register_move_cost): Be more sensible about mismatch stall;
7694 model AVX moves correctly; make difference between sse->integer and
7696 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
7697 moves; make difference between SSE and AVX.
7698 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
7699 and zmm_move. Increase size of sse load and store tables;
7700 add unaligned load and store tables; add ssemmx_to_integer.
7701 * x86-tune-costs.h: Update all entries according to real
7702 move latencies from Agner Fog's manual and chip documentation.
7704 2017-10-23 Jakub Jelinek <jakub@redhat.com>
7707 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
7708 * config/i386/constraints.md (Wf): New constraint.
7709 * config/i386/i386.md (UNSPEC_SBB): New unspec.
7710 (cmp<dwi>_doubleword): Removed.
7711 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
7712 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
7713 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
7714 expand with cmp<dwi>_doubleword. For LTU and GEU use
7715 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
7717 * common.opt (gcolumn-info): Enable by default.
7718 * doc/invoke.texi (gcolumn-info): Document new default.
7720 2017-10-23 Richard Biener <rguenther@suse.de>
7722 PR tree-optimization/82672
7723 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
7724 Fold the stmt if we propagated into it.
7726 2017-10-23 Richard Biener <rguenther@suse.de>
7728 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
7729 (bitmap_remove_expr_from_set): ... this. All callers call this
7730 for non-constant values.
7731 (bitmap_set_subtract): Rename to...
7732 (bitmap_set_subtract_expressions): ... this. Adjust and
7734 (bitmap_set_contains_value): Remove superfluous check.
7735 (bitmap_set_replace_value): Inline into single caller ...
7736 (bitmap_value_replace_in_set): ... here and simplify.
7737 (dependent_clean): Merge into ...
7738 (clean): ... this using an overload. Adjust.
7739 (prune_clobbered_mems): Adjust.
7740 (compute_antic_aux): Likewise.
7741 (compute_partial_antic_aux): Likewise.
7743 2017-10-23 Richard Biener <rguenther@suse.de>
7745 PR tree-optimization/82129
7747 2017-08-01 Richard Biener <rguenther@suse.de>
7749 PR tree-optimization/81181
7750 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7751 (compute_antic): ... end of iteration here.
7753 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7755 * target.def (starting_frame_offset): New hook.
7756 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
7757 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
7758 * doc/tm.texi.in: Regenerate.
7759 * hooks.h (hook_hwi_void_0): Declare.
7760 * hooks.c (hook_hwi_void_0): New function.
7761 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
7762 STARTING_FRAME_OFFSET.
7763 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
7764 * reload1.c (reload): Likewise.
7765 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
7766 instead of STARTING_FRAME_OFFSET.
7767 * function.c (try_fit_stack_local): Likewise.
7768 (assign_stack_local_1): Likewise
7769 (instantiate_virtual_regs): Likewise.
7770 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
7771 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
7772 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
7773 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
7774 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
7775 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
7776 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
7777 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
7778 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
7779 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
7780 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
7781 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
7782 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
7783 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
7784 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
7785 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
7786 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
7787 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
7788 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
7789 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
7790 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
7791 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
7792 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
7793 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
7794 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
7795 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
7796 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
7797 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
7798 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
7799 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
7800 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
7801 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
7802 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
7803 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
7804 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
7805 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
7806 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
7807 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
7808 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
7809 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
7810 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
7811 * config/avr/avr.c (avr_starting_frame_offset): Make static and
7812 return a HOST_WIDE_INT.
7813 (avr_builtin_setjmp_frame_value): Use it instead of
7814 STARTING_FRAME_OFFSET.
7815 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7816 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
7817 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
7819 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7820 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
7821 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
7822 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7823 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
7824 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
7825 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7826 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
7827 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
7828 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7829 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
7830 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
7832 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7833 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
7834 * config/mips/mips.c (mips_compute_frame_info): Refer to
7835 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
7836 (mips_starting_frame_offset): New function.
7837 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7838 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
7839 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
7840 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
7841 and return a HOST_WIDE_INT.
7842 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7843 (mmix_initial_elimination_offset): Refer to
7844 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
7845 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
7846 * config/pa/pa.c (pa_starting_frame_offset): New function.
7847 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
7848 (pa_expand_prologue): Likewise.
7849 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7850 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
7851 !FRAME_GROWS_DOWNWARD handling to...
7852 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7853 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
7854 !FRAME_GROWS_DOWNWARD handling to...
7855 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7856 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
7857 !FRAME_GROWS_DOWNWARD handling to...
7858 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7859 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
7861 (rs6000_starting_frame_offset): New function.
7862 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
7863 !FRAME_GROWS_DOWNWARD handling to...
7864 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7865 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
7866 !FRAME_GROWS_DOWNWARD handling to...
7867 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7868 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
7869 !FRAME_GROWS_DOWNWARD handling to...
7870 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7871 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
7872 (rs6000_starting_frame_offset): New function.
7873 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
7874 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
7875 * config/vax/vax.c (vax_starting_frame_offset): New function.
7876 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
7877 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7878 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
7879 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
7880 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7881 * system.h (STARTING_FRAME_OFFSET): Poison.
7883 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7885 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
7886 SCALAR_TYPE_MODE instead of TYPE_MODE.
7888 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7889 Alan Hayward <alan.hayward@arm.com>
7890 David Sherwood <david.sherwood@arm.com>
7892 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
7894 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7895 Alan Hayward <alan.hayward@arm.com>
7896 David Sherwood <david.sherwood@arm.com>
7898 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
7900 2017-10-23 Richard Biener <rguenther@suse.de>
7902 PR tree-optimization/82129
7903 * tree-ssa-pre.c (bitmap_set_and): Remove.
7904 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
7905 canonicalizing expressions in the set to those with lowest
7906 ID rather than taking that from the first edge.
7908 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7910 * combine.c (rtx_equal_for_field_assignment_p): Use
7911 byte_lowpart_offset.
7913 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7914 Alan Hayward <alan.hayward@arm.com>
7915 David Sherwood <david.sherwood@arm.com>
7917 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
7918 to a SUBREG_PROMOTED_VAR.
7920 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7921 Alan Hayward <alan.hayward@arm.com>
7922 David Sherwood <david.sherwood@arm.com>
7924 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
7925 (expand_debug_source_expr): Likewise.
7926 * combine.c (combine_simplify_rtx): Likewise.
7927 * cse.c (fold_rtx): Likewise.
7928 * optabs.c (expand_float): Likewise.
7929 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7930 (simplify_binary_operation_1): Likewise.
7932 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7933 Alan Hayward <alan.hayward@arm.com>
7934 David Sherwood <david.sherwood@arm.com>
7936 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
7937 (record_promoted_value): Likewise.
7938 * expr.c (expand_expr_real_2): Likewise.
7939 * ree.c (update_reg_equal_equiv_notes): Likewise.
7940 (combine_set_extension): Likewise.
7941 * rtlanal.c (low_bitmask_len): Likewise.
7942 * simplify-rtx.c (neg_const_int): Likewise.
7943 (simplify_binary_operation_1): Likewise.
7945 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7946 Alan Hayward <alan.hayward@arm.com>
7947 David Sherwood <david.sherwood@arm.com>
7949 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
7950 * regcprop.c (maybe_mode_change): Likewise.
7951 * reload1.c (alter_reg): Likewise.
7953 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7955 * inchash.h (inchash::hash::add_wide_int): New function.
7956 * lto-streamer-out.c (hash_tree): Use it.
7958 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7960 * inchash.h (inchash::hash::add_wide_int): Rename to...
7961 (inchash::hash::add_hwi): ...this.
7962 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
7963 (polymorphic_call_target_hasher::hash): Likewise.
7964 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
7965 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
7966 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
7967 * lto-streamer-out.c (hash_tree): Likewise.
7968 * optc-save-gen.awk: Likewise.
7969 * tree.c (add_expr): Likewise.
7971 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
7974 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
7975 for ordered inequality comparisons even with TARGET_IEEE_FP.
7977 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
7980 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
7981 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
7982 Expand with cmp<dwi>_doubleword.
7984 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7986 * extend.texi: Add x86 specific to 'nocf_check' attribute.
7987 List CET intrinsics.
7988 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
7989 specific to -fcf-protection option.
7991 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7993 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
7994 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
7995 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
7996 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
7997 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
7998 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
7999 (extra_objs): Add cet.o for Linux/x86 targets.
8000 (tmake_file): Add i386/t-cet for Linux/x86 targets.
8001 * config/i386/cet.c: New file.
8002 * config/i386/cetintrin.h: Likewise.
8003 * config/i386/t-cet: Likewise.
8004 * config/i386/cpuid.h (bit_SHSTK): New.
8005 (bit_IBT): Likewise.
8006 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
8007 pass IBT and SHSTK bits.
8008 * config/i386/i386-builtin-types.def
8009 (VOID_FTYPE_UNSIGNED_PVOID): New.
8010 (VOID_FTYPE_UINT64_PVOID): Likewise.
8011 * config/i386/i386-builtin.def: Add CET intrinsics.
8012 * config/i386/i386-c.c (ix86_target_macros_internal): Add
8013 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
8014 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
8015 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
8017 * config/i386/i386.c (rest_of_insert_endbranch): New.
8018 (pass_data_insert_endbranch): Likewise.
8019 (pass_insert_endbranch): Likewise.
8020 (make_pass_insert_endbranch): Likewise.
8021 (ix86_notrack_prefixed_insn_p): Likewise.
8022 (ix86_target_string): Add -mibt, -mshstk flags.
8023 (ix86_option_override_internal): Add flag_cf_protection
8025 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
8026 (ix86_print_operand): Add 'notrack' prefix output.
8027 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
8028 (ix86_expand_builtin): Expand CET intrinsics.
8029 (x86_output_mi_thunk): Add 'endbranch' instruction.
8030 * config/i386/i386.h (TARGET_IBT): New.
8031 (TARGET_IBT_P): Likewise.
8032 (TARGET_SHSTK): Likewise.
8033 (TARGET_SHSTK_P): Likewise.
8034 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
8035 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
8036 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
8037 (builtin_setjmp_setup): New pattern.
8038 (builtin_longjmp): Likewise.
8039 (rdssp<mode>): Likewise.
8040 (incssp<mode>): Likewise.
8041 (saveprevssp): Likewise.
8042 (rstorssp): Likewise.
8043 (wrss<mode>): Likewise.
8044 (wruss<mode>): Likewise.
8045 (setssbsy): Likewise.
8046 (clrssbsy): Likewise.
8047 (nop_endbr): Likewise.
8048 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
8050 * config/i386/immintrin.h: Include <cetintrin.h>.
8051 * config/i386/linux-common.h
8052 (file_end_indicate_exec_stack_and_cet): New prototype.
8053 (TARGET_ASM_FILE_END): New.
8055 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
8057 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
8058 latencies instead of having separate table; make difference between
8059 integer and float costs.
8060 * i386.h (processor_costs): Remove scalar_stmt_cost,
8061 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
8062 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
8064 * x86-tune-costs.h: Remove entries which has been removed in
8065 procesor_costs from all tables; make cond_taken_branch_cost
8066 and cond_not_taken_branch_cost COST_N_INSNS based.
8068 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
8070 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
8072 2017-10-20 Jakub Jelinek <jakub@redhat.com>
8074 * config/i386/i386.md (isa): Remove fma_avx512f.
8075 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
8076 <avx512>_fmadd_<mode>_mask3<round_name>,
8077 <avx512>_fmsub_<mode>_mask<round_name>,
8078 <avx512>_fmsub_<mode>_mask3<round_name>,
8079 <avx512>_fnmadd_<mode>_mask<round_name>,
8080 <avx512>_fnmadd_<mode>_mask3<round_name>,
8081 <avx512>_fnmsub_<mode>_mask<round_name>,
8082 <avx512>_fnmsub_<mode>_mask3<round_name>,
8083 <avx512>_fmaddsub_<mode>_mask<round_name>,
8084 <avx512>_fmaddsub_<mode>_mask3<round_name>,
8085 <avx512>_fmsubadd_<mode>_mask<round_name>,
8086 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
8087 (*vec_widen_umult_even_v16si<mask_name>,
8088 *vec_widen_smult_even_v16si<mask_name>): Likewise.
8089 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
8091 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
8093 * extend.texi: Add 'nocf_check' documentation.
8094 * gimple.texi: Add second parameter to
8095 gimple_build_call_from_tree.
8096 * invoke.texi: Add -fcf-protection documentation.
8097 * rtl.texi: Add REG_CALL_NOTRACK documenation.
8099 2017-10-20 Richard Biener <rguenther@suse.de>
8101 PR tree-optimization/82473
8102 * tree-vect-loop.c (vectorizable_reduction): Properly get at
8103 the largest input type.
8105 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
8107 * c-attribs.c (handle_nocf_check_attribute): New function.
8108 (c_common_attribute_table): Add 'nocf_check' handling.
8109 * gimple-parser.c: Add second argument NULL to
8110 gimple_build_call_from_tree.
8111 * attrib.c (comp_type_attributes): Check nocf_check attribute.
8112 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
8114 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
8115 * common.opt: Add fcf-protection flag.
8116 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
8117 * flag-types.h: Add enum cf_protection_level.
8118 * gimple.c (gimple_build_call_from_tree): Add second parameter.
8119 Add 'nocf_check' attribute propagation to gimple call.
8120 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
8121 (gimple_build_call_from_tree): Update prototype.
8122 (gimple_call_nocf_check_p): New function.
8123 (gimple_call_set_nocf_check): Likewise.
8124 * gimplify.c: Add second argument to gimple_build_call_from_tree.
8125 * ipa-icf.c: Add nocf_check attribute in statement hash.
8126 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
8127 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
8128 * toplev.c (process_options): Add flag_cf_protection handling.
8130 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
8132 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
8134 2017-10-20 Richard Biener <rguenther@suse.de>
8136 PR tree-optimization/82603
8137 * tree-if-conv.c (predicate_mem_writes): Make sure to only
8138 remove false predicated stores.
8140 2017-10-20 Richard Biener <rguenther@suse.de>
8142 * graphite-isl-ast-to-gimple.c
8143 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
8144 Remove return value and simplify, dump copied stmt after lhs
8146 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
8147 Reduce dump verbosity.
8148 (gsi_insert_earliest): Likewise.
8149 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
8150 * graphite.c (print_global_statistics): Adjust dumping.
8151 (print_graphite_scop_statistics): Likewise.
8152 (print_graphite_statistics): Do not dump loops here.
8153 (graphite_transform_loops): But here.
8155 2017-10-20 Nicolas Roche <roche@adacore.com>
8157 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
8158 * configure: Regenerate.
8160 2017-10-20 Jakub Jelinek <jakub@redhat.com>
8163 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
8164 functions when optimizing replace GIMPLE_RETURN stmts with
8165 calls to __builtin_unreachable ().
8168 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
8169 for -fsanitize=thread link of executables.
8170 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
8171 link of executables.
8174 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
8176 (<shift_insn><mode>3<mask_name>): Change the last of the 3
8177 define_insns for logical vector shifts to use VI248_AVX512BW
8178 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
8179 condition, useless isa and prefix attributes. Change the first
8180 2 of these define_insns to ...
8181 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
8182 define_insn for avx512vl.
8183 (<shift_insn><mode>3): ... and this, new define_insn without
8184 masking for non-avx512vl.
8187 * config/i386/sse.md (*andnot<mode>3,
8188 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
8189 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
8190 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
8191 not applied use empty suffix even for TARGET_AVX512VL.
8192 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
8193 is applied, supply evex,evex or evex,evex,evex instead of just
8196 2017-10-20 Julia Koval <julia.koval@intel.com>
8198 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
8199 (OPTION_MASK_ISA_GFNI_UNSET): New.
8200 (ix86_handle_option): Handle OPT_mgfni.
8201 * config/i386/cpuid.h (bit_GFNI): New.
8202 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
8203 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
8204 * config/i386/i386.c (ix86_target_string): Add -mgfni.
8205 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
8206 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
8207 * config/i386/i386.opt: Add mgfni.
8209 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
8211 * config/msp430/msp430.c (msp430_option_override): Disable
8212 -fdelete-null-pointer-checks.
8213 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
8215 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
8217 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
8218 of x87 and SSE instructions.
8220 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
8222 * asan.c (create_cond_insert_point): Do not update edge count.
8223 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
8224 (afdo_propagate_circuit): Likewise.
8225 (afdo_calculate_branch_prob): Likewise.
8226 (afdo_annotate_cfg): Likewise.
8227 * basic-block.h (struct edge_def): Remove count.
8228 (edge_def::count): New accessor.
8229 * bb-reorder.c (rotate_loop): Update.
8230 (find_traces_1_round): Update.
8231 (connect_traces): Update.
8232 (sanitize_hot_paths): Update.
8233 * cfg.c (unchecked_make_edge): Update.
8234 (make_single_succ_edge): Update.
8235 (check_bb_profile): Update.
8236 (dump_edge_info): Update.
8237 (update_bb_profile_for_threading): Update.
8238 (scale_bbs_frequencies_int): Update.
8239 (scale_bbs_frequencies_gcov_type): Update.
8240 (scale_bbs_frequencies_profile_count): Update.
8241 (scale_bbs_frequencies): Update.
8242 * cfganal.c (connect_infinite_loops_to_exit): Update.
8243 * cfgbuild.c (compute_outgoing_frequencies): Update.
8244 (find_many_sub_basic_blocks): Update.
8245 * cfgcleanup.c (try_forward_edges): Update.
8246 (try_crossjump_to_edge): Update
8247 * cfgexpand.c (expand_gimple_cond): Update
8248 (expand_gimple_tailcall): Update
8249 (construct_exit_block): Update
8250 * cfghooks.c (verify_flow_info): Update
8251 (redirect_edge_succ_nodup): Update
8252 (split_edge): Update
8253 (make_forwarder_block): Update
8254 (duplicate_block): Update
8255 (account_profile_record): Update
8256 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
8257 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
8258 * cfgloopmanip.c (scale_loop_profile): Update.
8260 (lv_adjust_loop_entry_edge): Update.
8261 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
8262 (force_nonfallthru_and_redirect): Update.
8263 (purge_dead_edges): Update.
8264 (rtl_flow_call_edges_add): Update.
8265 * cgraphunit.c (init_lowered_empty_function): Update.
8266 (cgraph_node::expand_thunk): Update.
8267 * gimple-pretty-print.c (dump_probability): Update.
8268 (dump_edge_probability): Update.
8269 * gimple-ssa-isolate-paths.c (isolate_path): Update.
8270 * haifa-sched.c (sched_create_recovery_edges): Update.
8271 * hsa-gen.c (convert_switch_statements): Update.
8272 * ifcvt.c (dead_or_predicable): Update.
8273 * ipa-inline-transform.c (inline_transform): Update.
8274 * ipa-split.c (split_function): Update.
8275 * ipa-utils.c (ipa_merge_profiles): Update.
8276 * loop-doloop.c (add_test): Update.
8277 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
8278 * lto-streamer-in.c (input_cfg): Update.
8279 (input_function): Update.
8280 * lto-streamer-out.c (output_cfg): Update.
8281 * modulo-sched.c (sms_schedule): Update.
8282 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
8283 * predict.c (maybe_hot_edge_p): Update.
8284 (unlikely_executed_edge_p): Update.
8285 (probably_never_executed_edge_p): Update.
8286 (dump_prediction): Update.
8287 (drop_profile): Update.
8288 (propagate_unlikely_bbs_forward): Update.
8289 (determine_unlikely_bbs): Update.
8290 (force_edge_cold): Update.
8291 * profile.c (compute_branch_probabilities): Update.
8292 * reg-stack.c (better_edge): Update.
8293 * shrink-wrap.c (handle_simple_exit): Update.
8294 * tracer.c (better_p): Update.
8295 * trans-mem.c (expand_transaction): Update.
8296 (split_bb_make_tm_edge): Update.
8297 * tree-call-cdce.c: Update.
8298 * tree-cfg.c (gimple_find_sub_bbs): Update.
8299 (gimple_split_edge): Update.
8300 (gimple_duplicate_sese_region): Update.
8301 (gimple_duplicate_sese_tail): Update.
8302 (gimple_flow_call_edges_add): Update.
8303 (insert_cond_bb): Update.
8304 (execute_fixup_cfg): Update.
8305 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
8306 * tree-complex.c (expand_complex_div_wide): Update.
8307 * tree-eh.c (lower_resx): Update.
8308 (unsplit_eh): Update.
8309 (cleanup_empty_eh_move_lp): Update.
8310 * tree-inline.c (copy_edges_for_bb): Update.
8311 (freqs_to_counts): Update.
8312 (copy_cfg_body): Update.
8313 * tree-ssa-dce.c (remove_dead_stmt): Update.
8314 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
8315 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
8316 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
8317 (unloop_loops): Update.
8318 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
8319 * tree-ssa-loop-split.c (connect_loops): Update.
8320 (split_loop): Update.
8321 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
8322 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
8323 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
8324 * tree-ssa-reassoc.c (branch_fixup): Update.
8325 * tree-ssa-tail-merge.c (replace_block_by): Update.
8326 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
8327 (compute_path_counts): Update.
8328 (update_profile): Update.
8329 (recompute_probabilities): Update.
8330 (update_joiner_offpath_counts): Update.
8331 (estimated_freqs_path): Update.
8332 (freqs_to_counts_path): Update.
8333 (clear_counts_path): Update.
8334 (ssa_fix_duplicate_block_edges): Update.
8335 (duplicate_thread_path): Update.
8336 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
8337 (case_bit_test_cmp): Update.
8338 (collect_switch_conv_info): Update.
8339 (gen_inbound_check): Update.
8340 (do_jump_if_equal): Update.
8341 (emit_cmp_and_jump_insns): Update.
8342 * tree-tailcall.c (decrease_profile): Update.
8343 (eliminate_tail_call): Update.
8344 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
8345 (vect_do_peeling): Update.
8346 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
8347 * ubsan.c (ubsan_expand_null_ifn): Update.
8348 (ubsan_expand_ptr_ifn): Update.
8349 * value-prof.c (gimple_divmod_fixed_value): Update.
8350 (gimple_mod_pow2): Update.
8351 (gimple_mod_subtract): Update.
8352 (gimple_ic): Update.
8353 (gimple_stringop_fixed_value): Update.
8355 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
8358 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
8360 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
8362 PR rtl-optimization/82395
8363 * ira-color.c (allocno_priority_compare_func): Fix comparison step
8364 based on non_spilled_static_chain_regno_p.
8366 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
8368 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
8369 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
8370 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
8372 2017-10-19 Martin Sebor <msebor@redhat.com>
8374 PR tree-optimization/82596
8375 * tree.c (array_at_struct_end_p): Handle STRING_CST.
8377 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8379 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
8380 (get_mem_refs_of_builtin_call): Likewise.
8381 * builtins.c (expand_builtin_apply): Adjust call to
8382 allocate_dynamic_stack_space.
8383 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
8384 the third argument to allocate_dynamic_stack_space, otherwise -1.
8385 (expand_builtin): Deal with all alloca variants.
8386 (is_inexpensive_builtin): Likewise.
8387 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
8388 * calls.c (special_function_p): Deal with all alloca variants.
8389 (initialize_argument_information): Adjust call to
8390 allocate_dynamic_stack_space.
8391 (expand_call): Likewise.
8392 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
8393 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
8394 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
8395 use it for the stack usage computation.
8396 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
8397 * function.c (gimplify_parameters): Call build_alloca_call_expr.
8398 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
8399 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
8400 (in_loop_p): Remove first argument and useless check.
8401 (pass_walloca::execute): Remove useless test and adjust call to above.
8402 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
8403 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
8404 (gimplify_call_expr): Deal with all alloca variants.
8405 * hsa-gen.c (gen_hsa_alloca): Likewise.
8406 (gen_hsa_insns_for_call): Likewise.
8407 * ipa-pure-const.c (special_builtin_state): Likewise.
8408 * tree-chkp.c (chkp_build_returned_bound): Likewise.
8409 * tree-object-size.c (alloc_object_size): Likewise.
8410 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
8411 (call_may_clobber_ref_p_1): Likewise.
8412 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
8413 (ccp_fold_stmt): Likewise.
8414 (optimize_stack_restore): Likewise.
8415 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
8416 (mark_all_reaching_defs_necessary_1): Likewise.
8417 (propagate_necessity): Likewise.
8418 (eliminate_unnecessary_stmts): Likewise.
8419 * tree.c (build_common_builtin_nodes): Build
8420 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
8421 (build_alloca_call_expr): New function.
8422 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
8423 (CASE_BUILT_IN_ALLOCA): Likewise.
8424 (build_alloca_call_expr): Declare.
8425 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
8427 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8430 * dwarf2out.c (new_die_raw): New static inline function.
8431 (new_die): Use it to create the DIE.
8432 (add_AT_external_die_ref): Likewise.
8433 (clone_die): Likewise.
8434 (clone_as_declaration): Likewise.
8435 (dwarf2out_vms_debug_main_pointer): Likewise.
8436 (base_type_die): Likewise. Remove early return for corner cases.
8437 Do not call add_pubtype on the DIE here.
8438 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
8439 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
8440 typedefs for base types with DW_AT_endianity. Make sure a DIE with
8441 native order exists for base types, attach the DIE manually and call
8442 add_pubtype on it. Do not equate a reverse order DIE to the type.
8444 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
8446 * config/arm/arm.c (align_ok_ldrd_strd): New function.
8447 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
8449 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
8451 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8453 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
8454 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
8455 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
8456 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
8457 * opts.c (sanitizer_opts): Add builtin.
8458 * ubsan.c (instrument_builtin): New function.
8459 (pass_ubsan::execute): Call it.
8460 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
8461 * doc/invoke.texi: Document -fsanitize=builtin.
8463 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
8464 builtins, store max (log2 (align), 0) into uchar field instead of
8465 align into uptr field.
8466 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
8467 store uchar 0 field instead of uptr 0 field.
8468 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
8469 instead of passing one address of struct with 2 locations pass
8470 two addresses of structs with 1 location each.
8471 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
8472 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
8473 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
8474 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
8475 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
8476 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
8477 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
8478 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
8480 2017-10-19 Martin Liska <mliska@suse.cz>
8483 * file-find.c (remove_prefix): Remove.
8484 * file-find.h (remove_prefix): Likewise.
8485 * gcc-ar.c: Remove smartness of lookup.
8487 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
8489 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
8490 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
8491 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
8493 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8496 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
8497 (setcc + and to xor + setcc): New peephole2.
8499 2017-10-19 Tom de Vries <tom@codesourcery.com>
8501 * doc/sourcebuild.texi (Test Directives, Variants of
8502 dg-require-support): Add dg-require-stack-size.
8504 2017-10-19 Martin Liska <mliska@suse.cz>
8507 * gimplify.c (gimplify_decl_expr): Do not instrument variables
8508 that have a large alignment.
8509 (gimplify_target_expr): Likewise.
8511 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
8513 PR rtl-optimization/82602
8514 * ira.c (rtx_moveable_p): Return false for volatile asm.
8516 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
8519 * config/i386/i386-modes.def (CCGZ): New CC mode.
8520 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
8521 * config/i386/predicates.md (ix86_comparison_operator):
8523 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
8524 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
8525 with double-word subtraction.
8526 (put_condition_code): Handle CCGZmode.
8528 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
8530 * wide-int.cc (debug (const wide_int &)): New.
8531 (debug (const wide_int *)): New.
8532 (debug (const widest_int &)): New.
8533 (debug (const widest_int *)): New.
8535 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
8538 * lra-constraints.c (curr_insn_transform): Use non-input operand
8539 instead of output one for matched reload.
8541 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8543 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
8544 (tree-ssa-loop-ivopts.h): New header file.
8545 (struct builtin_info): New fields.
8546 (classify_builtin_1): Compute and record base and offset parts for
8547 memset builtin partition by calling strip_offset.
8548 (offset_cmp, fuse_memset_builtins): New functions.
8549 (finalize_partitions): Fuse adjacent memset partitions by calling
8551 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
8552 Expose the interface.
8553 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
8555 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8557 PR tree-optimization/82574
8558 * tree-loop-distribution.c (find_single_drs): New parameter. Check
8559 that data reference must be executed exactly once per iteration
8560 against the outermost loop in nest.
8561 (classify_partition): Update call to above function.
8563 2017-10-18 Richard Biener <rguenther@suse.de>
8565 PR tree-optimization/82591
8566 * graphite.c (graphite_transform_loops): Move code gen message
8568 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8569 Here. Handle scop_to_isl_ast failing.
8570 (scop_to_isl_ast): Limit the number of ISL operations.
8572 2017-10-18 Richard Biener <rguenther@suse.de>
8574 * graphite-isl-ast-to-gimple.c
8575 (translate_isl_ast_to_gimple::set_rename): Simplify.
8576 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
8577 (graphite_copy_stmts_from_block): ... here.
8578 (copy_bb_and_scalar_dependences): Simplify.
8579 (add_parameters_to_ivs_params): Canonicalize.
8580 (generate_entry_out_of_ssa_copies): Simplify.
8581 * graphite-sese-to-poly.c (extract_affine_name): Simplify
8582 by passing in ISL dimension.
8583 (parameter_index_in_region_1): Rename to ...
8584 (parameter_index_in_region): ... this.
8585 (extract_affine): Adjust assert, pass down parameter index.
8586 (add_param_constraints): Use range-info when available.
8587 (build_scop_context): Adjust.
8588 * sese.c (new_sese_info): Adjust.
8589 (free_sese_info): Likewise.
8590 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
8591 Remove unused typedefs.
8592 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
8594 2017-10-18 Martin Liska <mliska@suse.cz>
8596 * combine.c (simplify_compare_const): Add gcc_fallthrough.
8598 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8600 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
8601 (s390_sched_init): Do not reset s390_sched_state if we entered the
8602 current basic block via a fallthru edge and all others are unlikely.
8604 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8606 * config/s390/s390.c (NUM_SIDES): New variable.
8607 (LONGRUNNING_THRESHOLD): New variable.
8608 (LATENCY_FACTOR): New variable.
8609 (s390_sched_score): Decrease score for long-running instructions on
8611 (s390_sched_variable_issue): Perform bookkeeping for long-running
8614 2017-10-18 Richard Biener <rguenther@suse.de>
8616 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
8617 Simplify with removal of the parameter rename map.
8618 (set_rename): Likewise.
8619 (should_copy_to_new_region): Likewise.
8620 (graphite_copy_stmts_from_block): Likewise.
8621 (copy_bb_and_scalar_dependences): Remove initialization of
8622 unused copied_bb_map.
8624 (copy_internal_parameters): Likewise.
8625 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
8626 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
8627 Use INTEGRAL_TYPE_P.
8628 (parameter_index_in_region_1): Rename to ...
8629 (assign_parameter_index_in_region): ... this. Assert we have
8630 a parameter we handle.
8631 (scan_tree_for_params): Adjust.
8632 * sese.h (parameter_rename_map_t): Remove.
8633 (struct sese_info_t): Remove unused parameter_rename_map and
8634 copied_bb_map members.
8635 * sese.c (new_sese_info): Adjust.
8636 (free_sese_info): Likewise.
8638 2017-10-18 Martin Liska <mliska@suse.cz>
8641 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
8642 on an abnormal edge.
8644 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
8646 * doc/invoke.texi (ffunction-sections and fdata-sections):
8649 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8651 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
8652 the use statement can throw internally.
8654 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8656 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
8657 any RTX present on the RHS of a SET.
8658 * compare-elim.c (try_eliminate_compare): Restore comment.
8660 2017-10-17 Jakub Jelinek <jakub@redhat.com>
8662 * langhooks.h (struct lang_hooks): Document that tree_size langhook
8663 may be also called on tcc_type nodes.
8664 * langhooks.c (lhd_tree_size): Likewise.
8666 2017-10-17 David Malcolm <dmalcolm@redhat.com>
8668 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
8669 format_warning_at_substring.
8670 (maybe_warn): Convert source_range * param to a location_t. Pass
8671 UNKNOWN_LOCATION rather than NULL to fmtwarn.
8672 (format_directive): Remove code to extract source_ranges and
8673 source_range * in favor of just a location_t.
8674 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
8676 * substring-locations.c (format_warning_va): Convert
8677 source_range * param to a location_t.
8678 (format_warning_at_substring): Likewise.
8679 * substring-locations.h (format_warning_va): Likewise.
8680 (format_warning_at_substring): Likewise.
8682 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
8684 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
8686 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
8687 and scatter/gather ops.
8689 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
8690 vec_gather_load and vec_scatter_store.
8691 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
8692 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
8693 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
8694 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
8695 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
8696 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
8698 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
8700 * reg-stack.c (compare_for_stack_reg): Add bool argument.
8701 Detect FTST instruction and handle its register pops. Only pop
8702 second operand if can_pop_second_op is true.
8703 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
8704 set can_pop_second_op to false in the compare_for_stack_reg call.
8706 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
8707 output_fp_compare for stack register operands.
8708 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
8709 instructions here. Do not emit stack register pops here. Assert
8710 that FCOMPP pops next to top stack register. Rewrite function.
8712 2017-10-17 Nathan Sidwell <nathan@acm.org>
8715 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
8716 use DECL_ASSEMBLER_NAME_RAW.
8719 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
8722 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
8723 Wilco Dijkstra <wilco.dijkstra@arm.com>
8725 * builtins.c (expand_builtin_update_setjmp_buf): Add a
8726 converstion to Pmode from the buf_addr.
8728 2017-10-17 Richard Biener <rguenther@suse.de>
8730 * graphite-dependences.c (scop_get_reads_and_writes): Change
8731 output parameters to references.
8733 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
8735 PR 71026/tree-optimization
8736 * fold-const.c (distribute_real_division): Removed.
8737 (fold_binary_loc): Remove calls to distribute_real_divison.
8739 2017-10-17 Richard Biener <rguenther@suse.de>
8741 * graphite-scop-detection.c
8742 (scop_detection::stmt_has_simple_data_refs_p): Always use
8743 the full nest as region.
8744 (try_generate_gimple_bb): Likewise.
8745 * sese.c (scalar_evolution_in_region): Simplify now that
8746 SCEV can handle instantiation in regions.
8747 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
8748 in the non-loop part of a function if requested.
8750 2017-10-17 Richard Biener <rguenther@suse.de>
8752 PR tree-optimization/82563
8753 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
8755 (graphite_regenerate_ast_isl): Call it.
8756 * graphite-scop-detection.c (build_scops): Remove entry edge split.
8758 2017-10-17 Jakub Jelinek <jakub@redhat.com>
8760 PR tree-optimization/82549
8761 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
8762 Formatting fixes. Instead of calling make_bit_field_ref with negative
8765 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
8767 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
8768 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
8769 _mm_maskz_reduce_ss): New.
8770 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
8771 __builtin_ia32_reducess_mask): Ditto..
8772 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
8773 * config/i386/sse.md (reduces<mode>): Renamed to ...
8774 (reduces<mode><mask_scalar_name>): ... this.
8775 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
8777 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
8778 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
8780 2017-10-16 David Malcolm <dmalcolm@redhat.com>
8782 * Makefile.in (OBJS): Add unique-ptr-tests.o.
8783 * selftest-run-tests.c (selftest::run_tests): Call
8784 selftest::unique_ptr_tests_cc_tests.
8785 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
8786 * unique-ptr-tests.cc: New file.
8788 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
8791 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8793 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8794 (make_hard_regno_born, make_hard_regno_dead): Update
8795 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
8797 2017-10-16 Jeff Law <law@redhat.com>
8799 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
8801 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
8803 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
8805 2017-10-16 Olivier Hainque <hainque@adacore.com>
8807 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
8808 with_cpu if we were configured for an e500v2 target cpu name.
8810 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
8812 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
8813 * doc/invoke.texi: Document +nodsp as a valid extension for
8816 2017-10-16 Martin Liska <mliska@suse.cz>
8818 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
8819 (test_set_range): Likewise.
8820 (test_range_functions): Rename to ...
8821 (test_bit_in_range): ... this.
8822 (sbitmap_c_tests): Add new test.
8824 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8826 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
8828 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
8829 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
8831 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8833 * config/aarch64/aarch64-builtins.c
8834 (aarch64_types_quadopu_lane_qualifiers): New.
8835 (TYPES_QUADOPU_LANE): New.
8836 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
8837 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
8838 (aarch64_<sur>dot_laneq<vsi2qi>): New.
8839 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
8840 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
8841 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
8842 (Vdottype, DOTPROD): New.
8843 (sur): Add SDOT and UDOT.
8845 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8847 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
8848 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
8849 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
8851 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
8852 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
8853 Enable TARGET_DOTPROD.
8854 (cortex-a75.cortex-a55): Likewise.
8855 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
8857 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8859 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
8860 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
8862 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
8864 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
8865 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
8866 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
8867 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
8868 * config/arm/types.md (neon_dot, neon_dot_q): New.
8869 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
8871 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8873 * config/arm/arm.h (TARGET_DOTPROD): New.
8874 * config/arm/arm.c (arm_arch_dotprod): New.
8875 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
8876 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
8877 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
8878 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
8879 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
8880 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
8881 * doc/invoke.texi (armv8.2-a): Document dotprod
8883 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
8885 * i386.c (ix86_vec_cost): New function.
8886 (ix86_rtx_costs): Handle vector operations better.
8887 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
8888 * x86-tune-costs.h: Add new costs to all tables.
8890 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
8892 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
8894 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
8895 divsd, sqrtss and sqrtsd
8896 * x86-tune-costs.h: Add new entries to all costs.
8897 (znver1_cost): Fix to match real instruction latencies.
8899 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8900 Michael Collison <michael.collison@arm.com>
8902 * compare-elim.c: Include emit-rtl.h.
8903 (can_merge_compare_into_arith): New function.
8904 (try_validate_parallel): Likewise.
8905 (try_merge_compare): Likewise.
8906 (try_eliminate_compare): Call the above when no previous clobber
8908 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
8911 2017-10-14 Jakub Jelinek <jakub@redhat.com>
8915 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
8916 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
8920 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
8921 to be any operand_equal_p operands. For & (B - 1) require
8922 B to be power of 2. Recognize
8923 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
8925 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8928 * optabs.c (expand_memory_blockage): Fix call of
8929 targetm.have_memory_blockage.
8931 2017-10-14 Jakub Jelinek <jakub@redhat.com>
8934 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
8935 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
8936 objects to extra_objs instead of overwriting it.
8938 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8940 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
8941 Use any_fp_register_operand as operand[3] predicate. Simplify
8942 equality test for operands[2] and operands[4] memory location.
8943 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
8944 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
8945 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
8946 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
8947 any_fp_register_operand as operand[1] predicate. Simplify
8948 equality test for operands[0] and operands[3] memory location.
8949 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
8950 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
8951 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
8953 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8955 * target-insns.def: Add memory_blockage.
8956 * optabs.c (expand_memory_blockage): New function.
8957 (expand_asm_memory_barrier): Rename ...
8958 (expand_asm_memory_blockage): ... to this.
8959 (expand_mem_thread_fence): Call expand_memory_blockage
8960 instead of expand_asm_memory_barrier.
8961 (expand_mem_singnal_fence): Ditto.
8962 (expand_atomic_load): Ditto.
8963 (expand_atomic_store): Ditto.
8964 * doc/md.texi (Standard Pattern Names For Generation):
8965 Document memory_blockage instruction pattern.
8967 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
8969 * config/rl78/rl78.c (rl78_emit_libcall): New function.
8970 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
8971 * config/rl78/rl78.md: New define_expand "adddi3".
8973 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
8975 * cfghooks.c (verify_flow_info): Disable check that all probabilities
8978 2017-10-13 Jeff Law <law@redhat.com>
8980 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
8982 2017-10-13 Jakub Jelinek <jakub@redhat.com>
8985 * internal-fn.c (expand_mul_overflow): If both operands have
8986 the same highpart of -1 or 0 and the topmost bit of lowpart
8987 is different, overflow is if res <= 0 rather than res < 0.
8989 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
8991 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
8992 TARGET_P9_VECTOR code for unaligned_load case.
8994 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
8996 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
8998 2017-10-13 Nathan Sidwell <nathan@acm.org>
9000 * tree-core.h (tree_contains_struct): Make bool.
9001 * tree.c (tree_contains_struct): Likewise.
9002 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
9003 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
9004 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
9005 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
9006 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
9008 2017-10-13 Richard Biener <rguenther@suse.de>
9010 * graphite-isl-ast-to-gimple.c
9011 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
9012 parameters and dominance check.
9013 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
9014 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
9015 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
9016 Do not update SSA form here or do intermediate IL verification.
9017 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
9018 (graphite_initialize): Remove check on the number of loops in
9019 the function and inline into graphite_transform_loops.
9020 (graphite_finalize): Inline into graphite_transform_loops.
9021 (graphite_transform_loops): Perform SSA update and IL verification
9023 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
9025 2017-10-13 Richard Biener <rguenther@suse.de>
9027 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
9028 graphite_expression_type_precision): Avoid global constructor
9030 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
9031 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
9032 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
9033 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
9034 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
9035 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
9037 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
9040 * config/i386/i386.h (ix86_red_zone_size): New.
9041 * config/i386/i386.md (push peephole2s): Replace
9042 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
9044 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9045 Alan Hayward <alan.hayward@arm.com>
9046 David Sherwood <david.sherwood@arm.com>
9048 * combine.c (can_change_dest_mode): Reject changes in
9049 REGMODE_NATURAL_SIZE.
9051 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9052 Alan Hayward <alan.hayward@arm.com>
9053 David Sherwood <david.sherwood@arm.com>
9055 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
9056 (expand_debug_source_expr): Likewise.
9057 * combine.c (combine_simplify_rtx): Likewise.
9058 * cse.c (fold_rtx): Likewise.
9059 * fwprop.c (canonicalize_address): Likewise.
9060 * targhooks.c (default_shift_truncation_mask): Likewise.
9062 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9063 Alan Hayward <alan.hayward@arm.com>
9064 David Sherwood <david.sherwood@arm.com>
9066 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
9067 (widened_mode): Likewise.
9068 (expand_unop): Likewise.
9069 * ree.c (transform_ifelse): Likewise.
9070 (merge_def_and_ext): Likewise.
9071 (combine_reaching_defs): Likewise.
9072 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9074 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9075 Alan Hayward <alan.hayward@arm.com>
9076 David Sherwood <david.sherwood@arm.com>
9078 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
9079 * combine.c (gen_lowpart_for_combine): Likewise.
9080 * dwarf2out.c (rtl_for_decl_location): Likewise.
9081 * final.c (alter_subreg): Likewise.
9082 * rtlhooks.c (gen_lowpart_general): Likewise.
9083 (gen_lowpart_if_possible): Likewise.
9085 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9086 Alan Hayward <alan.hayward@arm.com>
9087 David Sherwood <david.sherwood@arm.com>
9089 * calls.c (expand_call): Use subreg_lowpart_offset.
9090 * cse.c (cse_insn): Likewise.
9091 * regcprop.c (copy_value): Likewise.
9092 (copyprop_hardreg_forward_1): Likewise.
9094 2017-10-13 Jakub Jelinek <jakub@redhat.com>
9097 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
9098 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
9099 =Q constraints to +Q and into insn condition add check
9100 that operands[0] and operands[1] are equal.
9101 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
9102 =Q constraints to +Q and into insn condition add check
9103 that operands[0] is equal to either operands[1] or operands[2].
9106 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
9107 instead of handling MINUS_EXPR twice (once for each argument),
9108 canonicalize operand order and handle just once, use rtype where
9109 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
9112 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
9113 any values of __C while still being pattern recognizable as a simple
9116 2017-10-13 Richard Biener <rguenther@suse.de>
9118 PR tree-optimization/82451
9120 2017-10-02 Richard Biener <rguenther@suse.de>
9122 PR tree-optimization/82355
9123 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
9124 a mapping for the enclosing loop but avoid generating one for
9126 (copy_bb_and_scalar_dependences): Remove premature codegen
9127 error on PHIs in blocks duplicated into multiple places.
9128 * graphite-scop-detection.c
9129 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
9130 in the region use it as loop and nest to analyze the DR in.
9131 (try_generate_gimple_bb): Likewise.
9132 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
9133 (add_loop_constraints): For blocks in a loop not in the region
9134 create a dimension with a single iteration.
9135 * sese.h (gbb_loop_at_index): Remove assert.
9137 * cfgloop.c (loop_preheader_edge): For the loop tree root
9138 return the single successor of the entry block.
9139 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
9140 Reset the SCEV hashtable and niters.
9141 * graphite-scop-detection.c
9142 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
9143 assert that we only have POLYNOMIAL_CHREC that vary in loops
9144 contained in the region.
9145 (scop_detection::graphite_can_represent_expr): Adjust.
9146 (scop_detection::stmt_has_simple_data_refs_p): For loops
9147 not in the region set loop to NULL. The nest is now the
9148 entry edge to the region.
9149 (try_generate_gimple_bb): Likewise.
9150 * sese.c (scalar_evolution_in_region): Adjust for
9151 instantiate_scev change.
9152 * tree-data-ref.h (graphite_find_data_references_in_stmt):
9153 Make nest parameter the edge into the region.
9154 (create_data_ref): Likewise.
9155 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
9156 entry edge into a region and adjust instantiate_scev calls.
9157 (create_data_ref): Likewise.
9158 (graphite_find_data_references_in_stmt): Likewise.
9159 (find_data_references_in_stmt): Pass the loop preheader edge
9160 from the nest argument.
9161 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
9162 parameter the edge into the region.
9163 (instantiate_parameters): Use the loop preheader edge as entry.
9164 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
9166 (get_instantiated_value_entry): Make instantiate_below parameter
9167 the edge into the region.
9168 (instantiate_scev_name): Likewise. Adjust dominance checks,
9169 when we cannot use loop-based instantiation instantiate by
9170 walking use-def chains.
9171 (instantiate_scev_poly): Adjust.
9172 (instantiate_scev_binary): Likewise.
9173 (instantiate_scev_convert): Likewise.
9174 (instantiate_scev_not): Likewise.
9175 (instantiate_array_ref): Remove.
9176 (instantiate_scev_3): Likewise.
9177 (instantiate_scev_2): Likewise.
9178 (instantiate_scev_1): Likewise.
9179 (instantiate_scev_r): Do not blindly handle N-operand trees.
9180 Do not instantiate array-refs. Handle all constants and invariants.
9181 (instantiate_scev): Make instantiate_below parameter
9182 the edge into the region.
9183 (resolve_mixers): Use the loop preheader edge for the region
9184 parameter to instantiate_scev_r.
9185 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
9187 2017-10-13 Richard Biener <rguenther@suse.de>
9189 PR tree-optimization/82525
9190 * graphite-isl-ast-to-gimple.c
9191 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
9193 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
9194 Fail code generation when we cannot represent the isl integer.
9195 (binary_op_to_tree): Elide modulo operations that are no-ops
9196 in the type we code generate. Remove now superfluous code
9199 2017-10-13 Richard Biener <rguenther@suse.de>
9201 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
9202 (scop_detection::harmful_loop_in_region): Remove premature
9203 IV type restriction.
9204 (scop_detection::graphite_can_represent_scev): We can handle
9205 pointer IVs just fine.
9207 2017-10-13 Alan Modra <amodra@gmail.com>
9209 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
9210 "Clobbers and Scratch Registers". Add paragraph on
9211 alternative to clobbers for scratch registers and OpenBLAS
9214 2017-10-13 Alan Modra <amodra@gmail.com>
9216 * doc/extend.texi (Clobbers): Correct vax example. Delete old
9217 example of a memory input for a string of known length. Move
9218 commentary out of table. Add a number of new examples
9219 covering array memory inputs.
9221 2017-10-12 Martin Liska <mliska@suse.cz>
9223 PR tree-optimization/82493
9224 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
9225 (test_range_functions): New function.
9226 (sbitmap_c_tests): Likewise.
9227 * selftest-run-tests.c (selftest::run_tests): Run new tests.
9228 * selftest.h (sbitmap_c_tests): New function.
9230 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
9232 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
9234 * config/rs6000/amo.h: Fix spacing issue.
9236 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9239 * config/i386/i386.md (*ashl<mode>3_mask_1,
9240 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
9241 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
9244 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
9246 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
9247 (profile_probability): Set max_probability
9248 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
9250 * profile-count.c (profile_probability::differs_from_p): Do not
9251 rely on max_probaiblity == 10000
9253 2017-10-12 Jeff Law <law@redhat.com>
9255 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
9258 2017-10-12 Martin Sebor <msebor@redhat.com>
9262 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
9263 (handle_alias_pairs): Call it.
9264 * common.opt (-Wattribute-alias): New option.
9265 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
9266 * doc/invoke.texi (-Wattribute-alias): Document.
9268 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
9271 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
9273 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9275 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9276 (make_hard_regno_born, make_hard_regno_dead): Update
9277 bb_killed_pseudos and bb_gen_pseudos.
9279 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
9281 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
9283 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
9285 * config/alpha/alpha.c (alpha_split_conditional_move):
9286 Use std::swap instead of manually swapping.
9287 (alpha_stdarg_optimize_hook): Ditto.
9288 (alpha_canonicalize_comparison): Ditto.
9290 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9292 * tree-loop-distribution.c (struct builtin_info): New struct.
9293 (struct partition): Refactor fields into struct builtin_info.
9294 (partition_free): Free struct builtin_info.
9295 (build_size_arg_loc, build_addr_arg_loc): Delete.
9296 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
9297 information from struct builtin_info.
9298 (find_single_drs): New function refactored from classify_partition.
9299 Also moved builtin validity checks to this function.
9300 (compute_access_range, alloc_builtin): New functions.
9301 (classify_builtin_st, classify_builtin_ldst): New functions.
9302 (classify_partition): Refactor code into functions find_single_drs,
9303 classify_builtin_st and classify_builtin_ldst.
9304 (distribute_loop): Don't do runtime alias check when distributing
9306 (find_seed_stmts_for_distribution): New function.
9307 (pass_loop_distribution::execute): Refactor code finding seed
9308 stmts into above function. Support distribution for the innermost
9309 two-level loop nest. Adjust dump information.
9311 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9313 * tree-loop-distribution.c: Adjust the general comment.
9314 (NUM_PARTITION_THRESHOLD): New macro.
9315 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
9316 (classify_partition): Skip builtin pattern of loop nest's inner loop.
9317 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
9318 in call to build_partition_graph.
9319 (finalize_partitions): New parameter. Make loop distribution more
9320 conservative by fusing more partitions.
9321 (distribute_loop): Don't do runtime alias check in case of loop nest
9323 (find_seed_stmts_for_distribution): New function.
9324 (prepare_perfect_loop_nest): New function.
9325 (pass_loop_distribution::execute): Refactor code finding seed stmts
9326 and loop nest into above functions. Support loop nest distribution.
9327 Adjust dump information accordingly.
9329 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9331 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
9332 and set PTYPE_SEQUENTIAL for merged partition.
9334 2017-10-12 Richard Biener <rguenther@suse.de>
9336 PR tree-optimization/69728
9338 2017-09-19 Richard Biener <rguenther@suse.de>
9340 PR tree-optimization/69728
9341 * graphite-sese-to-poly.c (schedule_error): New global.
9342 (add_loop_schedule): Handle empty domain by failing the
9344 (build_original_schedule): Handle schedule_error.
9346 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
9347 domain by returning an unchanged schedule.
9349 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9351 * genrecog.c (validate_pattern): For VEC_SELECT verify that
9352 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
9354 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
9356 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
9359 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9362 * expr.c (store_field): Don't optimize away bitsize == 0 store
9363 from CALL_EXPR with addressable return type.
9365 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9367 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
9368 * config/rs6000/rs6000.md (sel): Delete mode attribute.
9369 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
9370 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
9371 TARGET_ISEL instead of TARGET_ISEL<sel>.
9373 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
9375 * config/rs6000/rs6000.c
9376 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
9378 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9380 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
9382 (reg_or_cint_operand): Fix comment.
9383 (reg_or_zero_operand): New predicate.
9384 * config/rs6000/rs6000-protos.h (output_isel): Delete.
9385 * config/rs6000/rs6000.c (output_isel): Delete.
9386 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
9387 instead of reg_or_cint_operand. Output instruction directly (not via
9389 (isel_unsigned_<mode>): Ditto.
9390 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
9391 gpc_reg_operand. Add an instruction alternative for this. Output
9392 instruction directly.
9393 (*isel_reversed_unsigned_<mode>): Ditto.
9395 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9397 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
9398 (TARGET_CANONICALIZE_COMPARISON): Define.
9400 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
9403 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
9404 Check whether the dest is REG before adding REG_EQUIV note.
9406 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
9409 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9411 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9412 (make_hard_regno_born, make_hard_regno_dead): Update
9413 bb_killed_pseudos and bb_gen_pseudos.
9415 2017-10-11 Nathan Sidwell <nathan@acm.org>
9417 * incpath.h (enum incpath_kind): Name enum, prefix values.
9418 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
9419 * incpath.c (heads, tails): Use INC_MAX.
9420 (add_env_var_paths, add_standard_paths): Use incpath_kind.
9421 (merge_include_chains, split_quote_chain,
9422 register_include_chains): Update incpath_kind names.
9423 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
9424 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
9426 (add_framework_path, darwin_register_objc_includes): Likewise.
9427 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
9429 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9431 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
9432 Do not use float_operator operator predicate.
9433 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
9434 * config/i386/predicates.md (float_operator): Remove predicate.
9436 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9438 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
9439 (*jccxf_i387): Ditto.
9440 (*jcc<mode>_i387): Ditto.
9441 (*jccu<mode>_i387): Ditto.
9442 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
9443 (*jcc_*_i387 splitters): Remove.
9444 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
9445 * config/i386/i386.c (ix86_split_fp_branch): Remove.
9446 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
9449 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9451 * profile-count.h (slow_safe_scale_64bit): New function.
9452 (safe_scale_64bit): New inline.
9453 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
9454 * profile-count.c: Include wide-int.h
9455 (slow_safe_scale_64bit): New.
9457 2017-10-11 Nathan Sidwell <nathan@acm.org>
9459 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
9460 HAS_DECL_ASSEMBLER_NAME_P.
9461 * gimple-expr.c (gimple_decl_printable_name: Check
9462 HAS_DECL_ASSEMBLER_NAME_P too.
9463 * ipa-utils.h (type_in_anonymous_namespace_p): Check
9464 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
9465 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
9466 * passes.c (rest_of_decl_compilation): Check
9467 HAS_DECL_ASSEMBLER_NAME_P too.
9468 * recog.c (verify_changes): Likewise.
9469 * tree-pretty-print.c (dump_decl_name): Likewise.
9470 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
9472 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
9473 (SET_DECL_ASSEMBLER_NAME): Use it.
9474 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
9475 (COPY_DECL_ASSEMBLER_NAME): Likewise.
9476 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
9478 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9480 * config.gcc (i386, x86_64): Add extra objects.
9481 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
9482 (ix86_min_insn_size): Declare.
9483 (ix86_issue_rate): Declare.
9484 (ix86_adjust_cost): Declare.
9485 (ia32_multipass_dfa_lookahead): Declare.
9486 (ix86_macro_fusion_p): Declare.
9487 (ix86_macro_fusion_pair_p): Declare.
9488 (ix86_bd_has_dispatch): Declare.
9489 (ix86_bd_do_dispatch): Declare.
9490 (ix86_core2i7_init_hooks): Declare.
9491 (ix86_atom_sched_reorder): Declare.
9492 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
9493 (COSTS_N_BYTES): Move to x86-tune-costs.h.
9494 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
9495 (rip_relative_addr_p): Rename to ...
9496 (ix86_rip_relative_addr_p): ... this one; export.
9497 (memory_address_length): Update.
9498 (ix86_issue_rate): Move to x86-tune-sched.c.
9499 (ix86_flags_dependent): Move to x86-tune-sched.c.
9500 (ix86_agi_dependent): Move to x86-tune-sched.c.
9501 (exact_dependency_1): Move to x86-tune-sched.c.
9502 (exact_store_load_dependency): Move to x86-tune-sched.c.
9503 (ix86_adjust_cost): Move to x86-tune-sched.c.
9504 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
9505 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
9506 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
9507 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
9508 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
9509 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
9510 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
9511 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
9512 (min_insn_size): Rename to ...
9513 (ix86_min_insn_size): ... this one; export.
9514 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
9515 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
9516 (core2i7_first_cycle_multipass_backtrack): Move to
9517 x86-tune-sched-core.c.
9518 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
9519 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
9520 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
9521 (ix86_avoid_jump_mispredicts): Update.
9522 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
9523 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
9524 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
9525 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
9526 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
9527 (MAX_INSN): Move to ix86-tune-sched-bd.c.
9528 (MAX_IMM): Move to ix86-tune-sched-bd.c.
9529 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
9530 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
9531 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
9532 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
9533 (MAX_STORE): Move to ix86-tune-sched-bd.c.
9534 (BIG): Move to ix86-tune-sched-bd.c.
9535 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
9536 (enum insn_path): Move to ix86-tune-sched-bd.c.
9537 (get_mem_group): Move to ix86-tune-sched-bd.c.
9538 (is_cmp): Move to ix86-tune-sched-bd.c.
9539 (dispatch_violation): Move to ix86-tune-sched-bd.c.
9540 (is_branch): Move to ix86-tune-sched-bd.c.
9541 (is_prefetch): Move to ix86-tune-sched-bd.c.
9542 (init_window): Move to ix86-tune-sched-bd.c.
9543 (allocate_window): Move to ix86-tune-sched-bd.c.
9544 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
9545 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
9546 (process_end_window): Move to ix86-tune-sched-bd.c.
9547 (allocate_next_window): Move to ix86-tune-sched-bd.c.
9548 (find_constant): Move to ix86-tune-sched-bd.c.
9549 (get_num_immediates): Move to ix86-tune-sched-bd.c.
9550 (has_immediate): Move to ix86-tune-sched-bd.c.
9551 (get_insn_path): Move to ix86-tune-sched-bd.c.
9552 (get_insn_group): Move to ix86-tune-sched-bd.c.
9553 (count_num_restricted): Move to ix86-tune-sched-bd.c.
9554 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
9555 (add_insn_window): Move to ix86-tune-sched-bd.c.
9556 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
9557 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
9558 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
9559 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
9560 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
9561 (do_dispatch): Move to ix86-tune-sched-bd.c.
9562 (has_dispatch): Move to ix86-tune-sched-bd.c.
9563 * config/i386/t-i386: Add new object files.
9564 * config/i386/x86-tune-costs.h: New file.
9565 * config/i386/x86-tune-sched-atom.c: New file.
9566 * config/i386/x86-tune-sched-bd.c: New file.
9567 * config/i386/x86-tune-sched-core.c: New file.
9568 * config/i386/x86-tune-sched.c: New file.
9570 2017-10-11 Liu Hao <lh_mouse@126.com>
9572 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
9573 the generic version below instead.
9574 (should_colorize): Recognize Windows consoles as terminals
9576 * pretty-print.c [__MINGW32__] (write_all): New function.
9577 [__MINGW32__] (find_esc_head): Likewise.
9578 [__MINGW32__] (find_esc_terminator): Likewise.
9579 [__MINGW32__] (eat_esc_sequence): Likewise.
9580 [__MINGW32__] (mingw_ansi_fputs): New function that handles
9582 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
9585 2017-10-11 Richard Biener <rguenther@suse.de>
9587 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
9588 Properly call analyze_scalar_evolution with the loop of the stmt.
9590 2017-10-11 Richard Biener <rguenther@suse.de>
9592 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
9593 * tree-core.h (tree_base): Add chrec_var union member.
9594 * tree.h (CHREC_VAR): Remove.
9595 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
9596 * tree-chrec.h (build_polynomial_chrec): Adjust.
9597 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
9598 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
9600 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
9602 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
9603 * match.pd: ... here.
9604 ((T) X == (T) Y): Relax condition.
9606 2017-10-11 Bin Cheng <bin.cheng@arm.com>
9608 PR tree-optimization/82472
9609 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
9611 (break_alias_scc_partitions): Update postorder number.
9613 2017-10-11 Martin Liska <mliska@suse.cz>
9616 * opts.c (parse_no_sanitize_attribute): Do not use error_value
9618 * opts.h (parse_no_sanitize_attribute): Remove last argument.
9620 2017-10-11 Martin Liska <mliska@suse.cz>
9622 * print-rtl.c (print_insn): Move declaration of idbuf
9623 to same scope as name.
9625 2017-10-11 Martin Liska <mliska@suse.cz>
9630 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9633 2017-10-11 Martin Liska <mliska@suse.cz>
9636 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9639 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
9641 * config.gcc (default_gnu_indirect_function): Default to yes for
9642 arm*-*-linux* with glibc.
9644 2017-10-11 Richard Biener <rguenther@suse.de>
9646 * tree-scalar-evolution.c (get_scalar_evolution): Handle
9647 default-defs and types we do not want to analyze.
9648 (interpret_loop_phi): Replace unreachable code with an assert.
9649 (compute_scalar_evolution_in_loop): Remove and inline ...
9650 (analyze_scalar_evolution_1): ... here, replacing condition with
9651 what makes the intent clearer. Remove handling of cases
9652 get_scalar_evolution now handles.
9654 2017-10-10 Jim Wilson <wilson@tuliptree.org>
9656 PR rtl-optimization/81434
9657 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
9658 comment for main loop. In sched_group_found if, also add checks for
9659 pass and min_cost_group.
9661 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
9663 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
9664 (rs6000_insn_cost): New function.
9665 * config/rs6000/rs6000.md (cost): New attribute.
9667 2017-10-10 Jakub Jelinek <jakub@redhat.com>
9668 H.J. Lu <hongjiu.lu@intel.com>
9672 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
9673 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
9674 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
9675 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
9676 to OPTION_MASK_ISA_AVX512VL - builtins that have both
9677 OPTION_MASK_ISA_MMX and some other bit set require both
9678 mmx and the ISAs without the mmx bit.
9679 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
9680 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
9681 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
9682 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
9683 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
9684 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
9685 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
9686 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
9687 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
9688 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
9689 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
9690 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
9691 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
9692 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
9693 Add OPTION_MASK_ISA_MMX.
9695 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
9697 * config.gcc (armv7*-*-freebsd*): New target.
9698 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
9700 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
9702 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
9703 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
9706 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
9708 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
9709 a template parameter.
9710 (WIDE_INT_REF_FOR): Update accordingly.
9711 * tree.h (wi::int_traits <const_tree>): Delete.
9712 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
9713 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
9714 (wi::tree_to_wide_ref): New typedef.
9715 (wi::to_wide): New function.
9716 * calls.c (get_size_range): Use wi::to_wide when operating on
9718 * cgraph.c (cgraph_node::create_thunk): Likewise.
9719 * config/i386/i386.c (ix86_data_alignment): Likewise.
9720 (ix86_local_alignment): Likewise.
9721 * dbxout.c (stabstr_O): Likewise.
9722 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
9723 * expr.c (const_vector_from_tree): Likewise.
9724 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
9725 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
9726 (fold_negate_expr_1, int_const_binop_1, const_binop)
9727 (fold_convert_const_int_from_real, optimize_bit_field_compare)
9728 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
9729 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
9730 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
9731 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
9732 (fold_not_const, round_up_loc): Likewise.
9733 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
9734 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
9735 (alloca_call_type): Likewise.
9736 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
9737 * godump.c (go_output_typedef): Likewise.
9738 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
9739 * internal-fn.c (get_min_precision): Likewise.
9740 * ipa-cp.c (ipcp_store_vr_results): Likewise.
9741 * ipa-polymorphic-call.c
9742 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
9743 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
9744 (ipa_modify_call_arguments): Likewise.
9745 * match.pd: Likewise.
9746 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
9747 * print-tree.c (print_node_brief, print_node): Likewise.
9748 * stmt.c (expand_case): Likewise.
9749 * stor-layout.c (layout_type): Likewise.
9750 * tree-affine.c (tree_to_aff_combination): Likewise.
9751 * tree-cfg.c (group_case_labels_stmt): Likewise.
9752 * tree-data-ref.c (dr_analyze_indices): Likewise.
9753 (prune_runtime_alias_test_list): Likewise.
9754 * tree-dump.c (dequeue_and_dump): Likewise.
9755 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
9756 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
9757 * tree-pretty-print.c (dump_generic_node): Likewise.
9758 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
9759 (simple_iv_with_niters): Likewise.
9760 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
9761 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
9762 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
9763 * tree-ssa-loop-niter.c (split_to_var_and_offset)
9764 (refine_value_range_using_guard, number_of_iterations_ne_max)
9765 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
9766 (get_cst_init_from_scev, record_nonwrapping_iv)
9767 (scev_var_range_cant_overflow): Likewise.
9768 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
9769 * tree-ssa-pre.c (compute_avail): Likewise.
9770 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
9771 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
9772 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
9773 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
9774 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
9775 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
9776 (dump_case_nodes, try_switch_expansion): Likewise.
9777 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
9778 (vect_do_peeling): Likewise.
9779 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
9780 * tree-vect-stmts.c (vectorizable_load): Likewise.
9781 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
9782 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
9783 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
9784 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
9785 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
9786 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
9787 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
9788 (evrp_dom_walker::before_dom_children): Likewise.
9789 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
9790 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
9791 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
9792 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
9793 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
9794 (get_range_pos_neg): Likewise.
9795 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
9796 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
9797 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
9798 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
9799 * config/avr/avr.c (avr_fold_builtin): Likewise.
9800 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
9801 * config/msp430/msp430.c (msp430_attr): Likewise.
9802 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
9803 * config/powerpcspe/powerpcspe-c.c
9804 (altivec_resolve_overloaded_builtin): Likewise.
9805 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
9806 (rs6000_expand_ternop_builtin): Likewise.
9807 * config/rs6000/rs6000-c.c
9808 (altivec_resolve_overloaded_builtin): Likewise.
9809 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
9810 (rs6000_expand_ternop_builtin): Likewise.
9811 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
9813 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9815 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
9816 when copying loop nest with only one inner loop.
9818 2017-10-10 Richard Biener <rguenther@suse.de>
9820 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
9821 blocks if SCEV is active.
9822 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
9824 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
9825 (scev_initialize): Assert we are not yet initialized.
9827 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9829 * tree-loop-distribution.c (generate_loops_for_partition): Remove
9830 inner loop's exit stmt by making it always exit the loop, otherwise
9831 we would generate an infinite empty loop.
9833 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9835 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
9836 renaming variables in new preheader if it's deleted.
9838 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9840 * tree-loop-distribution.c (struct partition): Remove unused field
9841 loops of the structure.
9842 (partition_alloc, partition_free): Ditto.
9843 (build_rdg_partition_for_vertex): Ditto.
9845 2017-10-09 Jeff Law <law@redhat.com>
9847 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
9848 return type to match prototype and documentation.
9850 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9852 * config/rs6000/rs6000.c (processor_costs): Move to ...
9853 * config/rs6000/rs6000.h: ... here.
9854 (rs6000_cost): Declare.
9856 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
9858 * except.c (setjmp_fn): New global variable.
9859 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
9860 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
9861 if DONT_USE_BUILTIN_SETJMP is defined.
9863 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9865 * target.def (insn_cost): New hook.
9866 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
9867 * doc/tm.texi: Regenerate.
9868 * rtlanal.c (insn_cost): Use the new hook.
9870 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9872 * combine.c (combine_validate_cost): Compute the new insn_cost,
9873 not just pattern_cost.
9874 (try_combine): Adjust comment.
9876 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9878 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
9880 * combine.c (uid_insn_cost): Adjust comment.
9881 (combine_validate_cost): Adjust comment. Use pattern_cost instead
9883 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
9884 * dse.c (find_shift_sequence): Ditto.
9885 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
9886 (bb_valid_for_noce_process_p): Use pattern_cost.
9887 * rtl.h (insn_rtx_cost): Delete.
9888 (pattern_cost): New prototype.
9889 (insn_cost): New prototype.
9890 * rtlanal.c (insn_rtx_cost): Rename to...
9891 (pattern_cost): ... this.
9894 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
9896 * config/i386/i386.md (*jcc_2): Remove insn pattern.
9897 (*jcc<mode>_0_r_i387): Ditto.
9898 (*jccxf_r_i387): Ditto.
9899 (*jcc<mode>_r_i387): Ditto.
9900 (*jccu<mode>_r_i387): Ditto.
9901 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
9902 (*jcc): Rename from *jcc_1.
9904 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9906 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
9907 deferred rescans after the lvx/stvx recombination pre-pass.
9909 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
9911 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
9912 memory operation instruction support.
9913 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
9914 (rs6000-ibm-aix[789]*): Likewise.
9915 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
9916 Document new functions.
9918 2017-10-09 Richard Biener <rguenther@suse.de>
9920 PR tree-optimization/82397
9921 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
9922 equality only for semantically equal trees.
9924 2017-10-09 Richard Biener <rguenther@suse.de>
9926 PR tree-optimization/82449
9927 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
9928 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
9929 allow constant addresses.
9930 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
9933 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9935 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
9938 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9941 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
9944 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9947 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
9949 2017-10-09 Jakub Jelinek <jakub@redhat.com>
9952 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
9953 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
9955 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
9957 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
9958 (WI_BINARY_PREDICATE_RESULT): Likewise.
9959 (wi::binary_traits::operator_result): New type.
9960 (wi::binary_traits::predicate_result): Likewise.
9961 (generic_wide_int::operator~, unary generic_wide_int::operator-)
9962 (generic_wide_int::operator==, generic_wide_int::operator!=)
9963 (generic_wide_int::operator&, generic_wide_int::and_not)
9964 (generic_wide_int::operator|, generic_wide_int::or_not)
9965 (generic_wide_int::operator^, generic_wide_int::operator+
9966 (binary generic_wide_int::operator-, generic_wide_int::operator*):
9968 (operator~, unary operator-, operator==, operator!=, operator&)
9969 (operator|, operator^, operator+, binary operator-, operator*): New
9971 * expr.c (get_inner_reference): Use wi::bit_and_not.
9972 * fold-const.c (fold_binary_loc): Likewise.
9973 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
9974 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
9975 (bit_value_binop): Likewise.
9976 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
9977 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
9978 (extract_range_from_binary_expr_1): Likewise.
9979 (masked_increment): Likewise.
9980 (simplify_bit_ops_using_ranges): Likewise.
9982 2017-10-09 Martin Jambor <mjambor@suse.cz>
9985 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
9986 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
9987 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
9989 (hsa_fixup_mov_insn_type): New function.
9990 (hsa_op_with_type::get_in_type): Use it.
9991 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
9992 immediates in an assert.
9993 (hsa_op_with_type::extend_int_to_32bit): New method.
9994 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
9995 types. Convert to dest type if necessary.
9996 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
9997 (reg_for_gimple_ssa): Pass false as min32int to
9998 hsa_type_for_scalar_tree_type.
9999 (gen_hsa_addr): Fixup type when creating addresable temporary.
10000 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
10001 (gen_hsa_unary_operation): Extend operands and convert to dest type if
10002 necessary. Call hsa_fixup_mov_insn_type.
10003 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
10004 extend operands and convert to dest type if necessary.
10005 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
10006 to dest type if necessary.
10007 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
10008 if conversion nt necessary and size matches.
10009 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
10010 to dest type if necessary.
10011 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
10012 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
10014 (gen_hsa_clrsb): Likewise.
10015 (gen_hsa_ffs): Likewise.
10016 (gen_hsa_divmod): Extend operands and convert to dest type if
10018 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
10020 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
10022 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
10023 Remove empty default arguments. Use a brace block as output
10025 (conditional return): Ditto.
10027 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
10028 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
10030 (group_ending_nop): Ditto.
10031 (doloop_end): Ditto.
10032 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
10033 (splitters for those): Ditto.
10035 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
10037 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
10038 a conditional jump (and the compare for it) so that pc_rtx is the
10040 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
10041 for the deleted and renamed ctr<mode>_internal[234] patterns.
10042 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
10043 Delete second conditional return pattern.
10044 (ctr<mode>_internal2): Delete this second bdnz pattern.
10045 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
10046 (ctr<mode>_internal4): Delete this second bdz pattern.
10048 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
10050 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
10051 (always_initialized_rtx_for_ssa_name_p): New predicate.
10052 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
10053 (finish_out_of_ssa): Free new field of SA.
10054 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
10055 * tree-ssa-coalesce.c: Include tree-ssa.h.
10056 (get_parm_default_def_partitions): Remove extern keyword.
10057 (get_undefined_value_partitions): New function.
10058 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
10059 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
10060 uninitialized bits.
10061 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
10063 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
10065 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
10067 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
10069 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
10070 for targets that preffer 128bit.
10072 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
10074 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
10076 2017-10-08 Olivier Hainque <hainque@adacore.com>
10078 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
10079 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
10081 (thumb_set_return_address): Likewise.
10083 2017-10-08 Olivier Hainque <hainque@adacore.com>
10085 * common/config/arm/arm-common.c (arm_except_unwind_info):
10086 Handle DWARF2_UNWIND_INFO.
10088 2017-10-07 Michael Collison <michael.collison@arm.com>
10090 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
10093 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
10095 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
10096 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
10097 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
10098 defined, force the creation of a new block for a dispatch label.
10100 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
10102 * invoke.texi (Wsuggest-attribute=cold): Document.
10103 * common.opt (Wsuggest-attribute=cold): New
10104 * ipa-pure-const.c (warn_function_cold): New function.
10105 * predict.c (compute_function_frequency): Use it.
10106 * predict.h (warn_function_cold): Declare.
10108 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
10110 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
10113 2017-10-06 Martin Liska <mliska@suse.cz>
10115 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
10116 keyword for member functions.
10117 (struct sanopt_tree_couple): New struct.
10118 (struct sanopt_tree_couple_hash): New function.
10119 (struct sanopt_ctx): Add new hash_map.
10120 (has_dominating_ubsan_ptr_check): New function.
10121 (record_ubsan_ptr_check_stmt): Likewise.
10122 (maybe_optimize_ubsan_ptr_ifn): Likewise.
10123 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
10124 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
10126 2017-10-06 Sudakshina Das <sudi.das@arm.com>
10129 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
10130 aarch64_simd_valid_immediate on CONST_VECTORs.
10131 (aarch64_reg_or_bic_imm): Likewise.
10133 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
10135 PR rtl-optimization/82396
10136 * haifa-sched.c (ready_sort_real): Disable qsort checking.
10138 2017-10-06 Sebastian Pop <sebpop@gmail.com>
10140 * graphite-dependences.c (scop_get_reads): Move code to...
10141 (scop_get_must_writes): Move code to...
10142 (scop_get_may_writes): Move code to...
10143 (scop_get_reads_and_writes): ... here.
10144 (scop_get_dependences): Call scop_get_reads_and_writes.
10146 2017-10-06 Jakub Jelinek <jakub@redhat.com>
10148 PR tree-optimization/82434
10149 * fold-const.h (can_native_encode_type_p,
10150 can_native_encode_string_p): Remove.
10151 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
10152 don't encode anything, just return what would be otherwise returned.
10153 (native_encode_fixed, native_encode_complex, native_encode_vector):
10155 (native_encode_string): Likewise. Inline by hand
10156 can_native_encode_string_p.
10157 (can_native_encode_type_p): Remove.
10158 (can_native_encode_string_p): Remove.
10159 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
10160 STRING_CSTs using can_native_encode_string_p, test all
10161 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
10162 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
10163 argument from native_encode_expr.
10164 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
10165 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
10168 2017-10-06 Richard Biener <rguenther@suse.de>
10170 PR tree-optimization/82397
10171 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
10172 operand_equal_p but rely on data_ref_compare_tree for detecting
10174 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
10175 to match up with dr_group_sort_cmp.
10177 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10180 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
10182 * config/s390/s390-builtin-types.def: Regenerate.
10184 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10187 * config/s390/s390-builtin-types.def: Regenerate.
10188 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
10189 Change flag from B_VXE to B_VX.
10190 (s390_vec_min_dbl): Remove B_VXE flag.
10192 2017-10-06 Richard Biener <rguenther@suse.de>
10194 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
10195 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
10196 translate_isl_ast_to_gimple::is_valid_rename,
10197 translate_isl_ast_to_gimple::get_rename,
10198 translate_isl_ast_to_gimple::get_def_bb_for_const,
10199 translate_isl_ast_to_gimple::get_new_name,
10200 translate_isl_ast_to_gimple::collect_all_ssa_names,
10201 translate_isl_ast_to_gimple::copy_loop_phi_args,
10202 translate_isl_ast_to_gimple::collect_all_ssa_names,
10203 translate_isl_ast_to_gimple::copy_loop_phi_args,
10204 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
10205 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
10206 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
10207 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
10208 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
10209 translate_isl_ast_to_gimple::copy_cond_phi_args,
10210 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
10211 translate_isl_ast_to_gimple::edge_for_new_close_phis,
10212 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
10213 translate_isl_ast_to_gimple::rename_uses,
10214 translate_isl_ast_to_gimple::rename_all_uses): Remove.
10215 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
10216 (set_rename_for_each_def): Likewise.
10217 (graphite_copy_stmts_from_block): Handle debug stmt resetting
10218 here. Handle rewriting SCEV analyzable uses here.
10219 (copy_bb_and_scalar_dependences): Generate code for PHI
10221 (graphite_regenerate_ast_isl): Adjust.
10222 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
10223 (add_write, add_read): New functions.
10224 (build_cross_bb_scalars_def): Use it and simplify.
10225 (build_cross_bb_scalars_use): Likewise.
10226 (graphite_find_cross_bb_scalar_vars): Inline into...
10227 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
10228 simulating out-of-SSA. Compute liveout and add dependencies.
10229 (build_scops): Force an empty entry block.
10230 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
10232 (sese_build_liveouts): Declare.
10233 (sese_trivially_empty_bb_p): Likewise.
10234 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
10235 compute liveout and debug_liveout.
10236 (sese_bad_liveouts_use): Remove.
10237 (sese_reset_debug_liveouts_bb): Likewise.
10238 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
10239 (sese_build_liveouts): Build liveout and debug_liveout and store
10241 (new_sese_info): Adjust.
10242 (free_sese_info): Likewise.
10243 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
10244 do not build liveout here.
10245 (move_sese_in_condition): Adjust region entry.
10246 (scev_analyzable_p): Match up with chrec_apply requirements.
10247 (sese_trivially_empty_bb_p): New.
10248 * tree-into-ssa.c (get_reaching_def): Properly support generating
10249 default-defs for incremental rewrite of anonymous names.
10251 2017-10-06 Richard Biener <rguenther@suse.de>
10253 * graphite-sese-to-poly.c (extract_affine): For casts increasing
10254 precision do not perform modulo reduction.
10256 2017-10-06 Richard Biener <rguenther@suse.de>
10258 PR tree-optimization/82436
10259 * tree-vect-slp.c (vect_supported_load_permutation_p): More
10260 conservatively choose the vectorization factor when checking
10261 whether we can perform the required load permutation.
10262 (vect_transform_slp_perm_load): Assert when we may not fail.
10264 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
10266 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
10267 message for incompatible -msdata=* and -mcall-* options.
10269 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10271 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
10272 rate for post-reload scheduling.
10274 2017-10-05 Tamar Christina <tamar.christina@arm.com>
10276 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
10278 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10280 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
10281 to improve monte carlo in scimark.
10283 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10285 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
10286 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
10287 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
10288 pentium4_cost, nocona_cost): Set reassociation width to 1.
10289 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
10290 width to 2 for fp operations and 1 otherwise.
10291 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
10293 (atom_cost): Set reassociation width to 2.
10294 (slm_cost, generic_cost): Set fp reassociation width
10295 to 2 and 1 otherwise.
10296 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
10297 (core_cost): Set fp reassociation width to 4 and vector to 2.
10298 (ix86_reassociation_width): Rewrite using cost table; special case
10299 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
10300 and TARGET_AVX128_OPTIMAL.
10301 * config/i386/i386.h (processor_costs): Add
10302 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
10303 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
10304 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
10305 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
10306 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
10307 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
10309 2017-10-05 Nathan Sidwell <nathan@acm.org>
10311 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
10313 2017-10-05 Tamar Christina <tamar.christina@arm.com>
10315 * config/arm/arm.c (arm_test_fpu_data): New.
10316 (arm_run_selftests): Call arm_test_fpu_data.
10318 2017-10-04 Nathan Sidwell <nathan@acm.org>
10320 * toplev.c (toplev::main): Remove excess parens on pretty_printer
10322 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
10324 2017-10-04 Sudakshina Das <sudi.das@arm.com>
10326 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
10327 check type for aarch64_simd_valid_immediate.
10328 (aarch64_output_simd_mov_immediate): Update prototype.
10329 (aarch64_simd_valid_immediate): Update prototype.
10330 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
10331 support for ORR-immediate.
10332 (and<mode>3): modified pattern to add support for BIC-immediate.
10333 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
10334 now checks for valid immediate for BIC and ORR based on new enum
10336 (aarch64_output_simd_mov_immediate): Function now used to output
10337 BIC/ORR imm as well based on new enum argument.
10338 * config/aarch64/constraints.md (Do): New vector immediate constraint.
10340 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
10341 (aarch64_reg_or_bic_imm): Likewise.
10343 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10345 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
10346 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
10348 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
10352 PR rtl-optimization/82396
10353 * haifa-sched.c (autopref_multipass_init): Simplify
10355 (autopref_rank_data): Simplify sort order.
10356 * sched-int.h (autopref_multipass_data_): Remove
10357 multi_mem_insn_p, min_offset and max_offset.
10359 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10361 * doc/sourcebuild.texi: Document vect_peeling_profitable.
10363 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10365 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
10366 vect_doubleint_cvt.
10368 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10370 * doc/sourcebuild.texi: Document vect_long_mult.
10372 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
10374 PR tree-optimization/82413
10375 * fold-const.c (build_range_check): Use widest_int when comparing
10376 the maximum ETYPE value with HIGH.
10378 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
10380 PR rtl-optimization/82396
10381 * haifa-sched.c (autopref_multipass_init): Simplify
10383 (autopref_rank_data): Simplify sort order.
10384 * sched-int.h (autopref_multipass_data_): Remove
10385 multi_mem_insn_p, min_offset and max_offset.
10387 2017-10-04 Jakub Jelinek <jakub@redhat.com>
10389 PR tree-optimization/82381
10390 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
10391 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
10394 PR tree-optimization/82374
10395 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
10396 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
10397 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
10398 current_function_decl to the new decl.
10400 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
10402 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
10403 helper macro for IEEE float128 hardware built-in functions.
10404 (SQRTF128_ODD): Add built-in functions with the round-to-odd
10406 (TRUNCF128_ODD): Likewise.
10407 (ADDF128_ODD): Likewise.
10408 (SUBF128_ODD): Likewise.
10409 (MULF128_ODD): Likewise.
10410 (DIVF128_ODD): Likewise.
10411 (FMAF128_ODD): Likewise.
10412 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
10413 UNSPEC_TRUNC_ROUND_TO_ODD.
10414 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
10415 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
10416 floating point round to odd instructions.
10417 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
10418 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
10419 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
10420 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
10421 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
10422 (trunc<mode>sf2_hw): Change the truncate with round to odd
10423 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
10424 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
10425 to odd hardware instructions.
10426 (sub<mode>3_odd): Likewise.
10427 (mul<mode>3_odd): Likewise.
10428 (div<mode>3_odd): Likewise.
10429 (sqrt<mode>2_odd): Likewise.
10430 (fma<mode>4_odd): Likewise.
10431 (fms<mode>4_odd): Likewise.
10432 (nfma<mode>4_odd): Likewise.
10433 (nfms<mode>4_odd): Likewise.
10434 (trunc<mode>df2_odd): Change the truncate with round to odd
10435 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
10437 * doc/extend.texi (PowerPC built-in functions): Update documentation
10438 for existing IEEE float128-bit built-in functions. Add built-in
10439 functions that generate the IEEE 128-bit floating point round to
10442 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
10444 PR rtl-optimization/77729
10445 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
10446 to (X&(C1&~C2))|C2 transformations.
10448 2017-10-03 Martin Jambor <mjambor@suse.cz>
10450 PR tree-optimization/82363
10451 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
10452 mismatch, mark lacc written regardless of racc.
10454 2017-10-03 Jakub Jelinek <jakub@redhat.com>
10456 PR tree-optimization/82381
10457 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
10458 stmt_to_insert nor wheather SSA_NAMEs are default defs.
10459 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
10460 fallthrough into reassoc_stmt_dominates_stmt_p.
10463 * combine.c (combine_instructions): Don't combine in unreachable
10466 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
10469 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
10470 function to not use the have_cpu variable. Do not set cpu_index,
10471 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
10472 or the default cpu.
10473 (rs6000_valid_attribute_p): Remove duplicate initializations of
10474 old_optimize and func_optimize.
10475 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
10476 (rs6000_activate_target_options): Make global.
10477 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
10480 2017-10-02 Jakub Jelinek <jakub@redhat.com>
10482 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
10483 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
10484 Set *poffset to 0 and *psize and *pmax_size to -1 if
10485 *poffset + *psize overflows in HOST_WIDE_INT.
10487 PR tree-optimization/82387
10488 PR tree-optimization/82388
10489 PR tree-optimization/82389
10490 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
10491 instead of live_bytes non-NULL.
10493 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
10496 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
10499 2017-10-02 Richard Biener <rguenther@suse.de>
10501 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
10502 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
10503 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
10505 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10507 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
10508 requested precision matches the type's.
10509 * calls.c (alloc_max_size): Calculate the new candidate size as
10510 a widest_int and use wi::to_widest when comparing it with the
10511 current candidate size.
10512 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
10513 zero rather than integer_zero_node.
10514 * match.pd: Check for a no-op conversion before using wi::add
10515 rather than after. Use tree_to_uhwi when summing small shift
10516 counts into an unsigned int.
10518 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10519 Alan Hayward <alan.hayward@arm.com>
10520 David Sherwood <david.sherwood@arm.com>
10523 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
10524 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
10525 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
10526 POINTER_AND_FP_REGS.
10528 2017-10-02 Richard Biener <rguenther@suse.de>
10530 PR tree-optimization/82355
10531 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
10532 a mapping for the enclosing loop but avoid generating one for
10533 the loop tree root.
10534 (copy_bb_and_scalar_dependences): Remove premature codegen
10535 error on PHIs in blocks duplicated into multiple places.
10536 * graphite-scop-detection.c
10537 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
10538 in the region use it as loop and nest to analyze the DR in.
10539 (try_generate_gimple_bb): Likewise.
10540 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
10541 (add_loop_constraints): For blocks in a loop not in the region
10542 create a dimension with a single iteration.
10543 * sese.h (gbb_loop_at_index): Remove assert.
10545 2017-10-01 Kevin Buettner <kevinb@redhat.com>
10547 * omp-expand.c (adjust_context_scope): New function.
10548 (expand_parallel_call): Call adjust_context_scope.
10550 2017-10-01 Jeff Law <law@redhat.com>
10552 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
10553 dom_opt_dom_walker class with direct access to private members.
10554 Add comments. Call test_for_singularity.
10555 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
10556 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
10557 m_dummy_cond anymore.
10558 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
10560 (pass_dominator:execute): Build the dummy_cond here and pass it
10561 to the dom_opt_dom_walker ctor.
10562 (test_for_singularity): New function.
10564 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
10565 Maya Rashish <coypu@sdf.org>
10567 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
10568 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
10569 (alpha*-*-netbsd*) Use nbsd_tm_file.
10570 (arm*-*-netbsdelf*) Likewise.
10571 (i[34567]86-*-netbsdelf*) Likewise.
10572 (x86_64-*-netbsd*) Likewise.
10573 (mips*-*-netbsd*) Likewise.
10574 (powerpc-*-netbsd*) Likewise.
10575 (sh*-*-netbsd*) Likewise.
10576 (sparc-*-netbsdelf*) Likewise.
10577 (sparc64-*-netbsd*) Likewise.
10578 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
10580 (vax-*-netbsdelf*) Likewise.
10581 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
10582 (UINT_FAST8_TYPE) Likewise.
10583 (INT_FAST16_TYPE) Check CHAR_FAST16.
10584 (UINT_FAST16_TYPE) Likewise.
10586 2017-09-30 Jakub Jelinek <jakub@redhat.com>
10589 * config/i386/i386.md
10590 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
10591 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
10592 *divmodsi4_zext_2): New define_insn_and_split.
10593 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
10594 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
10595 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
10596 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
10597 New define_insn_and_split.
10598 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
10599 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
10600 operands[1] having DImode when mode is SImode.
10602 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
10603 always SImode for DIV and MOD in REG_EQUAL notes.
10605 2017-09-29 Yury Gribov <tetra2005@gmail.com>
10607 PR middle-end/82319
10608 * match.pd: Fix handling of NaNs in pattern.
10610 2017-09-29 Jeff Law <law@redhat.com>
10612 * sbitmap.c (bitmap_bit_in_range_p): New function.
10613 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
10614 * tree-ssa-dse.c (live_bytes_read): New function.
10615 (dse_classify_store): Ignore reads of dead bytes.
10617 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
10618 typos and whitespace errors.
10619 * config/i386/predicates.md (address_no_seg_operand): Likewise.
10620 * config/s390/s390.c (s390_emit_prologue): Likewise.
10622 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10625 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
10626 with a symbol for LRA.
10628 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10630 PR rtl-optimization/82338
10631 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
10633 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
10635 * genmodes.c (calc_wider_mode): Suppress qsort macro.
10636 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
10637 (qsort_chk): Declare.
10638 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
10639 (qsort_chk): New function.
10641 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10643 PR tree-optimization/82337
10644 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
10645 phi definition if the PHI result appears in an abnormal PHI.
10646 (find_basis_for_base_expr): Don't record a basis if the LHS of the
10647 basis appears in an abnormal PHI.
10649 2017-09-29 Richard Biener <rguenther@suse.de>
10651 * graphite-isl-ast-to-gimple.c
10652 (translate_isl_ast_to_gimple::set_codegen_error): New function.
10653 (binary_op_to_tree): Use it.
10654 (get_rename_from_scev): Likewise.
10655 (copy_loop_phi_nodes): Likewise.
10656 (copy_bb_and_scalar_dependences): Likewise.
10657 (translate_pending_phi_nodes): Likewise.
10659 2017-09-29 Jakub Jelinek <jakub@redhat.com>
10662 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
10663 for movabsq $(i32 << shift), r64.
10665 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
10667 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
10668 index when encoding %esp as %rsp to avoid 0x67 prefix.
10670 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
10672 * config/i386/i386.md (*movsf_internal, *movdf_internal):
10673 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
10675 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
10677 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
10678 Extensions with more than 16 double VFP registers.
10679 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
10680 to_clear_mask and all code related to it. Replace the remaining
10681 entry by a sbitmap and adapt code accordingly.
10683 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
10685 * brig-builtins.def: Change pure attributes to const.
10687 2017-09-28 Joseph Myers <joseph@codesourcery.com>
10689 * config.gcc (default_gnu_indirect_function): Default to yes for
10690 sparc*-*-linux* with glibc.
10692 2017-09-28 Joseph Myers <joseph@codesourcery.com>
10694 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
10695 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
10696 when creating .init_array and .fini_array sections with priority
10699 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
10702 * config/aarch64/aarch64.c
10703 (aarch64_builtin_support_vector_misalignment): Always return false
10704 when misalignment is unknown.
10706 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
10708 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
10709 this function to return false if the definition used by the swap
10710 instruction is artificial, or if the memory address from which the
10711 constant value is loaded is not represented by a base address held
10712 in a register or if the base address register is a frame or stack
10713 pointer. Additionally, return false if the base address of the
10714 loaded constant is a SYMBOL_REF but is not considered to be a
10716 (replace_swapped_load_constant): New function.
10717 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
10718 loaded constant vector with a load of a swapped constant vector.
10720 2017-09-27 Carl Love <cel@us.ibm.com>
10722 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
10723 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
10724 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
10727 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
10729 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
10730 first, always call autopref_rank_data otherwise.
10732 2017-09-27 Richard Biener <rguenther@suse.de>
10734 * graphite-scop-detection.c (find_scop_parameters): Move
10735 loop bound handling ...
10736 (gather_bbs::before_dom_children): ... here, avoiding the need
10737 to build scop_info->loop_nest.
10738 (record_loop_in_sese): Remove.
10739 * sese.h (sese_info_t::loop_nest): Remove.
10740 * sese.c (new_sese_info): Do not allocate loop_nest.
10741 (free_sese_info): Do not free loop_nest.
10743 2017-09-27 Jakub Jelinek <jakub@redhat.com>
10746 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
10747 lhs from calls if the lhs has addressable type.
10749 2017-09-27 Richard Biener <rguenther@suse.de>
10751 * graphite.h (scop::max_alias_set): New member.
10752 * graphite-scop-detection.c: Remove references to non-existing
10753 --param in comments.
10754 (build_alias_sets): Record the maximum alias set used for drs.
10755 (build_scops): Support zero as unlimited for
10756 --param graphite-max-arrays-per-scop.
10757 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
10758 and inline into ...
10759 (build_poly_sr_1): ... here. Compute alias set based on the
10760 maximum alias set used for drs rather than
10761 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
10763 2017-09-27 Richard Biener <rguenther@suse.de>
10765 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
10766 --param loop-block-tile-size=0 to disable tiling.
10768 2017-09-27 Richard Biener <rguenther@suse.de>
10770 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
10771 (graphite-max-nb-scop-params): Document special value zero.
10772 * domwalk.h (dom_walker::STOP): New symbolical constant.
10773 (dom_walker::dom_walker): Add optional parameter for bb to
10775 (dom_walker::~dom_walker): Declare.
10776 (dom_walker::before_dom_children): Document STOP return value.
10777 (dom_walker::m_user_bb_to_rpo): New member.
10778 (dom_walker::m_bb_to_rpo): Likewise.
10779 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
10780 mapping here if not provided by the user.
10781 (dom_walker::~dom_walker): Free bb to RPO mapping if not
10782 provided by the user.
10783 (dom_walker::STOP): Define.
10784 (dom_walker::walk): Do not compute bb to RPO mapping here.
10785 Support STOP return value from before_dom_children to stop
10787 * graphite-optimize-isl.c (optimize_isl): If the schedule
10788 is the same still generate code if -fgraphite-identity
10789 or -floop-parallelize-all are given.
10790 * graphite-scop-detection.c: Include cfganal.h.
10791 (gather_bbs::gather_bbs): Get and pass through bb to RPO
10793 (gather_bbs::before_dom_children): Return STOP for BBs
10795 (build_scops): Compute bb to RPO mapping and pass it to
10796 the domwalk. Treat --param graphite-max-nb-scop-params=0
10797 as not limiting the number of params.
10798 * graphite.c (graphite_initialize): Remove limit on the
10799 number of basic-blocks in a function.
10800 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
10801 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
10802 default value of 10.
10804 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10806 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
10807 Adjust code to eliminate needing to do the shift right 32-bits
10808 operation after XSCVDPSPN.
10810 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10812 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
10813 ((X / Y) != 0 -> X >= Y): Likewise.
10815 2017-09-26 Carl Love <cel@us.ibm.com>
10817 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
10818 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
10819 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
10820 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
10821 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
10822 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
10823 definitions and overloading.
10824 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
10825 statement for P9V_BUILTIN_XST_LEN_R.
10826 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
10827 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
10828 define_expand and define_insn for the instructions and builtins.
10829 * doc/extend.texi: Update the built-in documentation file for the new
10830 built-in functions.
10831 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
10832 define_insn for the instructions
10834 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
10837 * gcc/config/netbsd-protos.h: New file.
10838 * gcc/config/netbsd.c: New file.
10839 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
10840 * gcc/config/t-netbsd: New file.
10841 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
10842 (tmake_file) Add t-netbsd.
10843 (extra_objs) Add netbsd.o.
10845 2017-09-26 Janus Weil <janus@gcc.gnu.org>
10849 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
10851 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10853 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
10854 sign extension from a vector register to a GPR by doing a 32-bit
10855 direct move and then an EXTSW.
10856 (extendsi<mode>2 splitter): Likewise.
10857 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
10858 right or vector extract after doing XSCVDPSPN. Use
10859 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
10861 (movdi_from_sf_zero_ext): Likewise.
10862 (reload_gpr_from_vsxsf): Likewise.
10863 (p8_mfvsrd_4_disf): Delete, no longer used.
10864 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
10865 then needing to move the SFmode to a GPR to use the XSCVDPSP
10866 instruction instead of FRSP and XSCVDPSPN.
10867 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
10868 it is adjacent to the other XSCVSPDP insns.
10869 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
10870 SFmode to be in traditional Altivec registers.
10871 (vsx_xscvdpspn): Eliminate useless alternative constraint.
10872 (vsx_xscvspdpn): Likewise.
10873 (vsx_xscvspdpn_scalar): Likewise.
10875 2017-09-26 Martin Jambor <mjambor@suse.cz>
10877 * tree-sra.c (compare_access_positions): Put integral types first,
10878 stabilize sorting of integral types, remove conditions putting
10879 non-full-precision integers last.
10880 (sort_and_splice_var_accesses): Disable scalarization if a
10881 non-integert would be represented by a non-full-precision integer.
10883 2017-09-26 Joseph Myers <joseph@codesourcery.com>
10885 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
10886 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
10887 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
10888 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
10889 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
10890 conditionals inside the function instead of around it. Call
10891 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
10892 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
10894 2017-09-26 Richard Biener <rguenther@suse.de>
10896 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
10898 (scop_detection::build_scop_breadth): ... this. Removed.
10899 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
10900 (scop_detection::harmful_stmt_in_bb): Likewise.
10901 (scop_detection::graphite_can_represent_stmt): Likewise.
10902 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
10903 (scop_detection::can_represent_loop): Remove recursion, fold in ...
10904 (scop_detection::can_represent_loop_1): ... this. Removed.
10905 (scop_detection::harmful_loop_in_region): Simplify after inlining
10906 the above and remove more quadraticness.
10907 (build_scops): Adjust.
10908 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
10911 2017-09-26 Jakub Jelinek <jakub@redhat.com>
10914 * config/i386/i386.c (ix86_print_operand_address_as): Only test
10915 REGNO (base) == SP_REG if base is a REG.
10917 PR middle-end/35691
10918 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
10919 if it is different SSA_NAME.
10920 (optimize_range_tests_cmp_bitwise): New function.
10921 (optimize_range_tests): Call it.
10923 2017-09-26 Richard Biener <rguenther@suse.de>
10925 PR tree-optimization/82321
10926 * graphite.c (canonicalize_loop_closed_ssa): Properly check
10927 for the def being inside the loop.
10929 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10931 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
10933 * config/s390/s390-builtins.def: Fix constraint on op4.
10935 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10937 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
10938 independent expanders.
10939 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
10940 ("vec_ordered", "vec_unordered"): New expanders.
10942 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10944 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
10947 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10949 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
10950 vec_unpacks_lo_v16qi.
10951 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
10953 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10955 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
10956 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
10957 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
10959 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10961 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
10963 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
10965 ("*vec_slb<mode>"): New insn pattern.
10966 ("vec_shr_<mode>"): New expander.
10967 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
10968 and force the shift count operand to V16QImode.
10969 ("vec_srb<mode>"): Set shift count mode to V16QI.
10971 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10973 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
10974 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
10975 ("vec_widen_smult_hi_<mode>"): New expander definitions.
10977 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
10980 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
10982 2017-09-26 Richard Biener <rguenther@suse.de>
10984 PR tree-optimization/82320
10985 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
10988 2017-09-25 Jeff Law <law@redhat.com>
10990 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
10991 prototype for new argument.
10992 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
10993 mostly extracted from rs6000_emit_allocate_stack.
10994 (rs6000_emit_probe_stack_range_stack_clash): New function.
10995 (rs6000_emit_allocate_stack): Call
10996 rs6000_emit_probe_stack_range_stack_clash as needed.
10997 (rs6000_emit_probe_stack_range): Add additional argument
10998 to call to gen_probe_stack_range{si,di}.
10999 (output_probe_stack_range): New.
11000 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
11001 (output_probe_stack_range_stack_clash): New.
11002 (rs6000_emit_prologue): Emit notes into dump file as requested.
11003 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
11004 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
11005 Add additional operand and pass it to output_probe_stack_range.
11007 2017-09-25 Bin Cheng <bin.cheng@arm.com>
11009 PR tree-optimization/82163
11010 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
11011 (checking_verify_loop_closed_ssa): New parameter.
11012 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
11013 (check_loop_closed_ssa_stmt): Delete.
11014 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
11015 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
11016 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
11019 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
11021 * brig-builtins.def: Treat HSAIL barrier builtins as
11022 setjmp/longjump style functions.
11024 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
11026 * target.def (constant_alignment): New hook.
11027 * defaults.h (CONSTANT_ALIGNMENT): Delete.
11028 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
11029 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
11030 * doc/tm.texi: Regenerate.
11031 * targhooks.h (default_constant_alignment): Declare.
11032 (constant_alignment_word_strings): Likewise.
11033 * targhooks.c (default_constant_alignment): New function.
11034 (constant_alignment_word_strings): Likewise.
11035 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
11036 instead of CONSTANT_ALIGNMENT.
11037 * varasm.c (align_variable, get_variable_align, build_constant_desc)
11038 (force_const_mem): Likewise.
11039 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
11040 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
11041 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
11042 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11043 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
11045 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
11046 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11047 constant_alignment_word_strings.
11048 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
11049 (CONSTANT_ALIGNMENT): Likewise.
11050 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11051 (arm_constant_alignment): New function.
11052 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
11053 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11054 constant_alignment_word_strings.
11055 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
11056 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11057 constant_alignment_word_strings.
11058 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
11059 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11060 (cris_constant_alignment): New function.
11061 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
11062 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11063 (epiphany_constant_alignment): New function.
11064 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
11065 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11066 constant_alignment_word_strings.
11067 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
11068 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11069 constant_alignment_word_strings.
11070 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
11071 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11072 constant_alignment_word_strings.
11073 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
11074 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
11075 * config/i386/i386.c (ix86_constant_alignment): Make static.
11076 Use the same interface as the target hook.
11077 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11078 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
11079 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11080 constant_alignment_word_strings.
11081 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
11082 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
11083 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11084 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
11085 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11086 constant_alignment_word_strings.
11087 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
11088 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11089 constant_alignment_word_strings.
11090 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
11091 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11092 constant_alignment_word_strings.
11093 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
11094 * config/microblaze/microblaze.c (microblaze_constant_alignment):
11096 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11097 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
11098 * config/mips/mips.c (mips_constant_alignment): New function.
11099 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11100 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
11101 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
11102 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11103 (mmix_constant_alignment): Make static. Use the same interface
11104 as the target hook.
11105 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
11106 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11107 constant_alignment_word_strings.
11108 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
11109 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11110 constant_alignment_word_strings.
11111 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
11112 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11113 constant_alignment_word_strings.
11114 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
11115 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11116 (rs6000_constant_alignment): New function.
11117 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
11118 * config/riscv/riscv.c (riscv_constant_alignment): New function.
11119 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11120 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
11121 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11122 (rs6000_constant_alignment): New function.
11123 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
11124 * config/s390/s390.c (s390_constant_alignment): New function.
11125 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11126 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
11127 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11128 constant_alignment_word_strings.
11129 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
11130 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11131 (sparc_constant_alignment): New function.
11132 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
11133 * config/spu/spu.c (spu_constant_alignment): New function.
11134 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11135 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
11136 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11137 constant_alignment_word_strings.
11138 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
11139 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11140 constant_alignment_word_strings.
11141 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
11142 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11143 constant_alignment_word_strings.
11144 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
11145 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11146 (visium_constant_alignment): New function.
11147 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
11148 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11149 (xtensa_constant_alignment): New function.
11150 * system.h (CONSTANT_ALIGNMENT): Poison.
11152 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
11154 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11155 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
11156 (rs6000_builtin_valid_without_lhs): New helper function.
11157 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11158 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
11160 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
11162 * target.h (vec_perm_indices): Use unsigned short rather than
11164 (auto_vec_perm_indices): Likewise.
11165 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
11166 Use unsigned int rather than unsigned char.
11167 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
11169 2017-09-25 Richard Biener <rguenther@suse.de>
11171 * cfgloop.h (sort_sibling_loops): Declare.
11172 * cfgloop.c (sort_sibling_loops_cmp): New helper.
11173 (sort_sibling_loops): New function sorting the sibling loop list
11175 * graphite.c (graphite_transform_loops): Sort sibling loops.
11177 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
11179 * target.def (vec_perm_const_ok): Change sel parameter to
11181 * optabs-query.c (can_vec_perm_p): Update accordingly.
11182 * doc/tm.texi: Regenerate.
11183 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
11184 auto_vec_perm_indices and remove separate nelt field.
11185 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
11186 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
11187 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
11188 (aarch64_expand_vec_perm_const): Update accordingly.
11189 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
11190 to vec_perm_indices.
11191 * config/arm/arm.c (expand_vec_perm_d): Change perm to
11192 auto_vec_perm_indices and remove separate nelt field.
11193 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
11194 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
11195 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
11197 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
11198 to vec_perm_indices.
11199 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
11200 sel to vec_perm_indices.
11201 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
11202 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
11203 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
11205 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
11208 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
11211 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
11212 on the FUNCTION_DECL function context if it has a DIE that is a
11215 2017-09-25 Richard Biener <rguenther@suse.de>
11217 PR tree-optimization/82285
11218 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
11221 2017-09-25 Tom de Vries <tom@codesourcery.com>
11225 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
11228 2017-09-25 Richard Biener <rguenther@suse.de>
11230 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
11231 ISL errors other than isl_error_quota happen. Dump if the
11232 schedule is the same.
11233 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
11234 errors instead of aborting inside ISL.
11236 2017-09-25 Iain Sandoe <iain@codesourcery.com>
11239 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
11240 of libgcc_eh for m64.
11241 * config/i386/darwin64.h: Likewise.
11243 2017-09-25 Richard Biener <rguenther@suse.de>
11245 PR middle-end/82144
11246 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
11247 attribute for incomplete types nor twice for complete ones.
11249 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
11252 * config/i386/i386.c (ix86_print_operand_address_as): Encode
11253 %esp as %rsp to avoid 0x67 prefix if there is no index or base
11256 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
11259 * config/i386/i386.opt (mprefer-avx256): Use
11260 ix86_target_flags variable.
11261 * config/i386/i386.c (ix86_target_string): Move
11262 -mprefer-avx256 to flag2_opts.
11264 2017-09-22 Jakub Jelinek <jakub@redhat.com>
11266 PR middle-end/35691
11267 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
11268 and x != -1 | y != -1 into (x & y) != -1.
11270 2017-09-22 Steve Ellcey <sellcey@cavium.com>
11272 * config.gcc: Add new case statement to set
11273 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
11274 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
11275 s390x-*-linux* case statements. Added aarch64 to the list of
11276 supported architectures.
11278 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11280 PR tree-optimization/82289
11281 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
11282 STMT_VINFO_RELEVANT_P.
11284 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11285 Alan Hayward <alan.hayward@arm.com>
11286 David Sherwood <david.sherwood@arm.com>
11288 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
11289 for VR_RANGE only; don't allow VR_ANTI_RANGE.
11290 (extract_range_from_binary_expr_1): Don't call
11291 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
11293 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11294 Alan Hayward <alan.hayward@arm.com>
11295 David Sherwood <david.sherwood@arm.com>
11297 * target.def (preferred_vector_alignment): New hook.
11298 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
11300 * doc/tm.texi: Regenerate.
11301 * targhooks.h (default_preferred_vector_alignment): Declare.
11302 * targhooks.c (default_preferred_vector_alignment): New function.
11303 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
11305 (DR_TARGET_ALIGNMENT): New macro.
11306 (aligned_access_p): Update commentary.
11307 (vect_known_alignment_in_bytes): New function.
11308 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
11310 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
11311 Calculate the misalignment based on the target alignment rather than
11313 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
11314 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
11315 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
11316 the target alignment, rather than masking the element misalignment
11317 with the number of elements in a vector. Also use the target
11318 alignment when calculating the maximum number of peels.
11319 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
11320 instead of TYPE_ALIGN_UNIT.
11321 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
11322 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
11323 (vect_create_addr_base_for_vector_ref): Update call accordingly.
11324 (vect_create_data_ref_ptr): Likewise.
11325 (vect_setup_realignment): Realign by ANDing with
11326 -DR_TARGET_MISALIGNMENT.
11327 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
11328 the number of peels based on DR_TARGET_ALIGNMENT.
11329 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
11330 with the guaranteed alignment boundary when deciding whether
11332 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
11333 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
11334 (ensure_base_align): Remove stmt_info parameter. Get the
11335 target base alignment from DR_TARGET_ALIGNMENT.
11336 (vectorizable_store): Update call accordingly. Interpret
11337 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
11339 (vectorizable_load): Likewise.
11341 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11342 Alan Hayward <alan.hayward@arm.com>
11343 David Sherwood <david.sherwood@arm.com>
11345 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
11346 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
11347 (vect_enhance_data_refs_alignment): Likewise.
11349 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
11351 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
11352 error. Only quit immediately if parsing is complete.
11353 (BEGIN): Initialize fatal_err and parse_done.
11354 (begin fpu, end fpu): Check number of arguments.
11355 (begin arch, end arch): Likewise.
11356 (begin cpu, end cpu): Likewise.
11357 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
11358 (optalias): Likewise.
11360 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
11362 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
11363 * config/arm/arm-isa.h: Delete. Move definitions to ...
11364 * arm-cpus.in: ... here. Use new feature and fgroup values.
11365 * config/arm/arm.c (arm_option_override): Use lower case for feature
11367 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
11368 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
11369 * config/arm/parsecpu.awk (END): Add new command 'isa'.
11371 (print_isa_bits_for): New function.
11372 (gen_isa): New function.
11373 (gen_comm_data): Use print_isa_bits_for.
11374 (define feature): New keyword.
11375 (define fgroup): New keyword.
11376 * config/arm/t-arm (TM_H): Remove.
11377 (GTM_H): Add arm-isa.h.
11378 (arm-isa.h): Add rule to generate file.
11379 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
11380 case for feature bit names.
11382 2017-09-22 Richard Biener <rguenther@suse.de>
11384 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
11386 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
11387 print of no dependency loops ...
11388 * graphite.c (graphite_transform_loops): ... here.
11389 (canonicalize_loop_closed_ssa_form): Work from inner to outer
11391 (same_close_phi_node, remove_duplicate_close_phi,
11392 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
11393 (canonicalize_loop_closed_ssa): ... here and simplify.
11394 * graphite-optimize-isl.c: Include tree-vectorizer.h.
11395 (optimize_isl): Use dump_printf_loc to tell when we stopped
11396 optimizing because of an ISL timeout.
11398 2017-09-22 Richard Biener <rguenther@suse.de>
11400 PR tree-optimization/82291
11401 * tree-if-conv.c (predicate_mem_writes): Make sure to
11402 remove writes in blocks predicated with false.
11404 2017-09-22 Richard Biener <rguenther@suse.de>
11406 * sese.c: Include cfganal.h.
11407 (if_region_set_false_region): Remove.
11408 (create_if_region_on_edge): Likewise.
11409 (move_sese_in_condition): Re-implement without destroying
11412 2017-09-22 Richard Biener <rguenther@suse.de>
11414 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
11415 Verify both BBs contain loop PHI nodes before dispatching to
11416 copy_loop_phi_args.
11417 (graphite_regenerate_ast_isl): Do not recompute dominators,
11418 do not verify three times. Restructure for clarity.
11419 * graphite-scop-detection.c (same_close_phi_node,
11420 remove_duplicate_close_phi, make_close_phi_nodes_unique,
11421 defined_in_loop_p, canonicalize_loop_closed_ssa,
11422 canonicalize_loop_closed_ssa_form): Simplify, remove excess
11423 checking and SSA rewrite, move to ...
11424 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
11425 (graphite_initialize): Do not pass in ctx, do not reset the
11426 SCEV cache, compute only dominators.
11427 (graphite_transform_loops): Allocate ISL ctx after
11428 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
11429 Maintain post-dominators only around build_scops.
11430 * sese.c (if_region_set_false_region): Make static. Free
11431 and recompute dominators.
11432 (move_sese_in_condition): Assert we don't get called with
11433 post-dominators computed.
11434 * sese.h (if_region_set_false_region): Remove.
11436 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
11438 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
11439 mode attribute for TARGET_AVX512VL.
11441 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
11443 * config/i386/i386.opt (mprefer-avx256): New option.
11444 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
11446 (ix86_preferred_simd_mode): Return 256-bit AVX modes
11447 for TARGET_PREFER_AVX256.
11448 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
11450 2017-09-21 Jeff Law <law@redhat.com>
11452 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
11453 Fix dump output if the only stack space is for pushed registers.
11455 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11457 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
11460 2017-09-21 Martin Sebor <msebor@redhat.com>
11463 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
11464 code (in C++) or code that triggers warnings.
11466 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
11468 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
11470 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
11472 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
11473 * sched-rgn.c: Ditto.
11474 * sel-sched-ir.c: Ditto.
11476 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
11478 * toplev.h (set_random_seed): Adjust return type.
11479 * toplev.c (init_local_tick): Move eager initialization of random_seed
11480 to get_random_seed. Adjust comment.
11481 (init_random_seed): Inline to get_random_seed, delete.
11482 (get_random_seed): Initialize random_seed lazily.
11483 (set_random_seed): Do not return previous value.
11484 (print_switch_value): Do not call get_random_seed.
11486 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
11488 * cgraph.c (delete_function_version): New, broken out from...
11489 (cgraph_node::delete_function_version): ...here. Rename to
11490 cgraph_node::delete_function_version_by_decl. Update all uses.
11491 (cgraph_node::remove): Call delete_function_version.
11493 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11496 * tree-inline.c (expand_call_inline): Emit clobber stmts for
11497 VAR_DECLs to which addressable non-volatile parameters are mapped
11498 and for id->retvar after the return value assignment. Clear
11499 id->retval and id->retbnd after inlining.
11501 2017-09-21 Richard Biener <rguenther@suse.de>
11503 PR tree-optimization/82276
11504 PR tree-optimization/82244
11505 * tree-vrp.c (build_assert_expr_for): Set
11506 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
11508 (remove_range_assertions): Revert earlier change.
11510 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
11513 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
11515 2017-09-21 Richard Biener <rguenther@suse.de>
11517 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11518 Restore valid IL after code generation errors.
11519 * graphite.c (graphite_transform_loops): Diagnose code
11520 generation issues as MSG_MISSED_OPTIMIZATION and continue
11521 with processing SCOPs.
11523 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11524 Alan Hayward <alan.hayward@arm.com>
11525 David Sherwood <david.sherwood@arm.com>
11527 * calls.c (compute_argument_addresses): Use simplify_gen_binary
11528 rather than choosing between plus_constant and gen_rtx_<CODE>.
11529 * expr.c (emit_push_insn): Likewise.
11530 (expand_expr_real_2): Likewise.
11532 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11533 Alan Hayward <alan.hayward@arm.com>
11534 David Sherwood <david.sherwood@arm.com>
11536 * loop-unroll.c (split_iv): Call copy_rtx on the step.
11538 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11539 Alan Hayward <alan.hayward@arm.com>
11540 David Sherwood <david.sherwood@arm.com>
11542 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
11543 calling tree_to_uhwi.
11545 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11546 Alan Hayward <alan.hayward@arm.com>
11547 David Sherwood <david.sherwood@arm.com>
11549 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
11550 INTEGER_CST rather than a negative test for ADDR_EXPR.
11552 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11553 Alan Hayward <alan.hayward@arm.com>
11554 David Sherwood <david.sherwood@arm.com>
11556 * tree-vrp.c (extract_range_from_binary_expr_1): Check
11557 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
11559 2017-09-21 Richard Biener <rguenther@suse.de>
11561 PR tree-optimization/71351
11562 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
11563 graphite_create_new_loop_guard): Remove, fold remaining parts
11565 (translate_isl_ast_node_for): ... here and simplify.
11567 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11570 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
11571 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
11572 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
11573 alternative always use QI mode, for -Os imov (=R,R) alternative
11574 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
11577 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11578 Jeff Law <law@redhat.com>
11580 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
11581 (allocate_stack_space): New function, partially extracted from
11582 s390_emit_prologue.
11583 (s390_emit_prologue): Track offset to most recent stack probe.
11584 Code to allocate space moved into allocate_stack_space.
11585 Dump actions when no stack is allocated.
11586 (s390_prologue_plus_offset): New function.
11587 (s390_emit_stack_probe): Likewise.
11589 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
11591 * common.opt (Wa, Wl, Wp, g, gz=): Add
11593 (gno-column-info): Remove.
11594 (gcolumn-info): Drop RejectNegative.
11595 (gno-): New prefix.
11596 (gno-record-gcc-switches): Remove.
11597 (grecord-gcc-switches): Drop RejectNegative.
11598 (gno-split-dwarf): Remove.
11599 (gsplit-dwarf): Drop RejectNegative.
11600 (gno-strict-dwarf): Remove.
11601 (gstrict-dwarf): Drop RejectNegative.
11602 * config/darwin.opt (gfull, gused): Add RejectNegative.
11603 * dwarf2out.c (gen_producer_string): Drop
11604 gno-record-gcc-switches handler.
11605 * optc-gen.awk: Add g to prefixes with negative forms.
11606 * opts-common.c (remapping_prefix_p): New.
11607 (find_opt): Check it.
11608 (generate_canonical_option): Test g prefix.
11609 (option_map): Add -gno- mapping.
11610 (add_misspelling_candidates): Check remapping_prefix_p.
11612 2017-09-20 Jeff Law <law@redhat.com>
11614 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
11615 thinko in stack clash protection support.
11617 * explow.c (compute_stack_clash_protection_loop_data): Use
11618 CONST_INT_P instead of explicit test. Verify object is a
11619 CONST_INT_P before looking at INTVAL.
11620 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
11621 instead of explicit test.
11623 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
11626 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
11627 address instead of to r1 and r11.
11629 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
11631 * config.gcc: Support "knm".
11632 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
11633 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
11635 * config/i386/i386.c (m_KNM): Define.
11636 (processor_target_table): Add "knm".
11638 (ix86_option_override_internal): Add "knm".
11639 (ix86_issue_rate): Add PROCESSOR_KNM.
11640 (ix86_adjust_cost): Ditto.
11641 (ia32_multipass_dfa_lookahead): Ditto.
11642 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
11643 (fold_builtin_cpu): Add M_INTEL_KNM.
11644 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
11645 (processor_type): Add PROCESSOR_KNM.
11646 * config/i386/x86-tune.def: Add m_KNM.
11647 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
11649 2017-09-20 Richard Biener <rguenther@suse.de>
11651 PR tree-optimization/80213
11652 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
11653 are allowed in empty BBs as well.
11654 (canonicalize_loop_closed_ssa): Also look for other complex
11656 (scop_detection::get_sese): Include the loop-closed PHI block
11658 (scop_detection::merge_sese): Remove code adding extra blocks.
11659 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
11660 (build_scops): Assert the final returned scop is invalid.
11662 2017-09-20 Richard Biener <rguenther@suse.de>
11664 PR tree-optimization/82264
11665 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
11667 (vn_phi_lookup): Likewise.
11668 (vn_phi_insert): Likewise.
11670 2017-09-20 Jakub Jelinek <jakub@redhat.com>
11672 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
11673 that fits into uhwi or shwi, add DW_AT_const_value regardless
11674 of early_dwarf without going through RTL, using add_AT_unsigned
11677 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
11678 (DEBUG_LTO_ABBREV_SECTION): Likewise.
11679 (DEBUG_LTO_MACINFO_SECTION): Likewise.
11680 (DEBUG_MACRO_SECTION): Likewise.
11681 (DEBUG_LTO_MACRO_SECTION): Likewise.
11682 (DEBUG_STR_DWO_SECTION): Likewise.
11683 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
11684 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
11685 (DEBUG_LTO_DWO_LINE_SECTION): Define.
11686 (DEBUG_LTO_LINE_STR_SECTION): Define.
11687 (init_sections_and_labels): Initialize debug_line_str_section
11688 variable. Initialize debug_loc_section for -gdwarf-5 to
11689 DEBUG_LOCLISTS_SECTION. Formatting fixes.
11691 2017-09-20 Richard Biener <rguenther@suse.de>
11693 * graphite-sese-to-poly.c (extract_affine): Properly handle
11694 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
11696 2017-09-20 Richard Biener <rguenther@suse.de>
11698 PR tree-optimization/81373
11699 * graphite-scop-detection.c (build_cross_bb_scalars_def):
11700 Force SESE live-out defs to be handled even if they are
11703 2017-09-19 Jeff Law <law@redhat.com>
11705 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
11706 nothing for stack adjustments with REG_STACK_CHECK.
11707 * sched-deps.c (parse_add_or_inc): Reject insns with
11708 REG_STACK_CHECK from dependency breaking.
11709 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
11710 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
11711 * reg-notes.def (STACK_CHECK): New note.
11713 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
11714 (ix86_expand_prologue): Dump stack clash info as needed.
11715 Call ix86_adjust_stack_and_probe_stack_clash as needed.
11717 * function.c (dump_stack_clash_frame_info): New function.
11718 * function.h (dump_stack_clash_frame_info): Prototype.
11719 (enum stack_clash_probes): New enum.
11721 * config/alpha/alpha.c (alpha_expand_prologue): Also check
11722 flag_stack_clash_protection.
11723 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
11724 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
11725 (arm_frame_pointer_required): Likewise.
11726 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
11727 (ia64_expand_prologue): Likewise.
11728 * config/mips/mips.c (mips_expand_prologue): Likewise.
11729 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
11730 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
11731 (sparc_flat_expand_prologue): Likewise.
11732 * config/spu/spu.c (spu_expand_prologue): Likewise.
11734 * explow.c: Include "params.h".
11735 (anti_adjust_stack_and_probe_stack_clash): New function.
11736 (get_stack_check_protect): Likewise.
11737 (compute_stack_clash_protection_loop_data): Likewise.
11738 (emit_stack_clash_protection_loop_start): Likewise.
11739 (emit_stack_clash_protection_loop_end): Likewise.
11740 (allocate_dynamic_stack_space): Use get_stack_check_protect.
11741 Use anti_adjust_stack_and_probe_stack_clash.
11742 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
11743 (emit_stack_clash_protection_loop_start): Likewise.
11744 (emit_stack_clash_protection_loop_end): Likewise.
11745 * rtl.h (get_stack_check_protect): Prototype.
11746 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
11747 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
11748 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
11750 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
11752 * doc/tm.texi: Rebuilt.
11753 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
11754 get_stack_check_protect.
11755 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
11756 * config/arm/arm.c (arm_expand_prologue): Likewise.
11757 (arm_frame_pointer_required): Likewise.
11758 * config/i386/i386.c (ix86_expand_prologue): Likewise.
11759 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
11760 * config/mips/mips.c (mips_expand_prologue): Likewise.
11761 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
11762 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
11763 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
11764 (sparc_flat_expand_prologue): Likewise.
11766 * common.opt (-fstack-clash-protection): New option.
11767 * flag-types.h (enum stack_check_type): Note difference between
11768 -fstack-check= and -fstack-clash-protection.
11769 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
11770 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
11771 * toplev.c (process_options): Issue warnings/errors for cases
11772 not handled with -fstack-clash-protection.
11773 * doc/invoke.texi (-fstack-clash-protection): Document new option.
11774 (-fstack-check): Note additional problem with -fstack-check=generic.
11775 Note that -fstack-check is primarily for Ada and refer users
11776 to -fstack-clash-protection for stack-clash-protection.
11777 Document new params for stack clash protection.
11779 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
11781 * config/i386/i386.c (ix86_split_long_move): Do not handle
11782 address used for LEA in a special way.
11784 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
11786 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
11788 2017-09-19 Martin Sebor <msebor@redhat.com>
11791 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
11792 of incompatible types.
11794 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
11796 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11797 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
11798 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11799 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
11801 2017-09-19 Richard Biener <rguenther@suse.de>
11803 PR tree-optimization/82244
11804 * tree-vrp.c (remove_range_assertions): Do not propagate
11805 a constant to abnormals but replace the assert with a copy.
11807 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
11809 PR rtl-optimization/57878
11810 PR rtl-optimization/68988
11811 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
11812 avoidance test involving non_reload_pseudos. Move frequency test
11813 below the general fragmentation avoidance test.
11815 2017-09-19 Richard Biener <rguenther@suse.de>
11817 PR tree-optimization/69728
11818 * graphite-sese-to-poly.c (schedule_error): New global.
11819 (add_loop_schedule): Handle empty domain by failing the
11821 (build_original_schedule): Handle schedule_error.
11823 2017-09-19 Richard Biener <rguenther@suse.de>
11825 * graphite-scop-detection.c (scop_detection::can_represent_loop):
11826 Do not iterate to sibling loops but only to siblings of inner
11829 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
11832 * config/m68k/m68k.md (moveq feeding equality comparison): Check
11833 that the registers are different.
11835 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
11837 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
11838 to processor_model and "amdfam17h" to arch_names_table.
11839 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
11841 2017-09-18 Jakub Jelinek <jakub@redhat.com>
11844 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
11846 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11847 Alan Hayward <alan.hayward@arm.com>
11848 David Sherwood <david.sherwood@arm.com>
11850 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
11852 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
11854 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
11855 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
11856 vect_schedule_slp_instance.
11857 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
11858 Update call to vect_slp_analyze_node_operations. Simplify return
11860 (vect_slp_analyze_bb_1): Update call accordingly.
11861 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
11862 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
11863 (vect_schedule_slp): Update call accordingly.
11865 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11866 Alan Hayward <alan.hayward@arm.com>
11867 David Sherwood <david.sherwood@arm.com>
11869 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
11870 with types that aren't in fact scalar.
11872 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11874 * tree-vect-slp.c (vect_record_max_nunits): New function,
11876 (vect_build_slp_tree_1): ...here.
11877 (vect_build_slp_tree_2): Call it for phis too.
11879 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11881 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
11882 to vect_get_vec_def_for_operand when getting the mask operand.
11884 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11885 Alan Hayward <alan.hayward@arm.com>
11886 David Sherwood <david.sherwood@arm.com>
11888 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
11891 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11892 Alan Hayward <alan.hayward@arm.com>
11893 David Sherwood <david.sherwood@arm.com>
11895 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
11896 calculation for vector booleans.
11898 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11899 Alan Hayward <alan.hayward@arm.com>
11900 David Sherwood <david.sherwood@arm.com>
11902 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
11904 (vect_transform_stmt): ...here.
11905 (vect_analyze_stmt): Use it instead of calling
11906 vectorizable_live_operation directly.
11908 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
11910 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
11911 non-SIMT targets in acc vector loops.
11913 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
11915 * configure.ac: Add arc and check if assembler supports gdwarf2.
11916 * configure: Regenerate.
11918 2017-09-18 Richard Biener <rguenther@suse.de>
11920 PR tree-optimization/82220
11921 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
11922 epilogue niters from the min_profitable_iters compute.
11924 2017-09-18 Jakub Jelinek <jakub@redhat.com>
11927 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
11928 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
11929 (ix86_init_pic_reg): Revert 2017-09-01 changes.
11931 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
11934 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
11935 switching to a new text section.
11937 2017-09-18 Richard Biener <rguenther@suse.de>
11939 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
11941 (build_alias_set): Reject aliases with no access function.
11943 2017-09-18 Richard Biener <rguenther@suse.de>
11945 PR tree-optimization/79622
11946 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
11948 (build_cross_bb_scalars_use): Likewise.
11950 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
11952 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
11954 2017-09-18 Alan Modra <amodra@gmail.com>
11957 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
11958 stack_pointer_rtx for count 0. Update comments. Break up
11959 large rtl expression.
11961 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
11963 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
11964 Increase to 20 bytes.
11965 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
11966 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
11967 or avx version of the stub.
11969 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
11972 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
11973 compute the minimum stack alignment. Also update preferred stack
11974 boundary for leaf functions.
11976 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
11978 PR tree-optimization/82228
11979 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
11982 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
11984 * common/config/nds32/nds32-common.c
11985 (nds32_option_optimization_table): Refine formatting.
11986 (nds32_option_optimization_table): Use -fsched-pressure and
11987 -fomit-frame-pointer for specific optimization level.
11989 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
11991 * config/nds32/nds32.c: Refine formatting and comments.
11992 * config/nds32/nds32.h: Likewise.
11993 * config/nds32/nds32.md: Likewise.
11994 * config/nds32/nds32-cost.c: Likewise.
11995 * config/nds32/nds32-isr.c: Likewise.
11996 * config/nds32/nds32-md-auxiliary.c: Likewise.
11997 * config/nds32/nds32-multiple.md: Likewise.
11998 * config/nds32/nds32-predicates.c: Likewise.
12000 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
12001 Jakub Jelinek <jakub@redhat.com>
12003 Add support for -std=c++2a.
12004 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
12006 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
12008 2017-09-15 Steve Ellcey <sellcey@cavium.com>
12011 * doc/extend.texi (Common Function Attributes): Add
12012 references to ARM, AArch64, and S/390 specific attributes.
12013 (Function Specific Option Pragmas): Add AArch64 and S/390
12014 to list of back ends that support the target pragma.
12016 2017-09-15 Nathan Sidwell <nathan@acm.org>
12018 * doc/standards.texi: Fix C++17 description. Update URLs for
12021 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
12023 * common.opt (Wcast-align=strict): New warning option.
12024 * doc/invoke.texi: Document -Wcast-align=strict.
12026 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
12028 * cgraph.h (cgraph_thunk_info): Add comments.
12029 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
12030 assert for VIRTUAL_* arguments stricter.
12032 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
12034 PR tree-optimization/71026
12035 * match.pd: Move RDIV patterns from fold-const.c
12036 * fold-const.c (distribute_real_division): Removed.
12037 (fold_binary_loc): Remove calls to distribute_real_divison.
12039 2017-09-15 Jakub Jelinek <jakub@redhat.com>
12041 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
12042 c++1z and gnu++1z as deprecated. Change other references to
12043 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
12044 Change -Wc++1z-compat to -Wc++17-compat.
12045 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
12046 * dwarf2out.c (highest_c_language): Handle C++17.
12047 (gen_compile_unit_die): Likewise.
12049 2017-09-15 Jakub Jelinek <jakub@redhat.com>
12051 PR rtl-optimization/82192
12052 * combine.c (make_extraction): Don't look through non-paradoxical
12053 SUBREGs or TRUNCATE if pos + len is or might be bigger than
12056 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
12057 Alan Hayward <alan.hayward@arm.com>
12058 David Sherwood <david.sherwood@arm.com>
12060 * target.def (function_arg_offset): New hook.
12061 * targhooks.h (default_function_arg_offset): Declare.
12062 * targhooks.c (default_function_arg_offset): New function.
12063 * function.c (locate_and_pad_parm): Use
12064 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
12065 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
12066 (TARGET_FUNCTION_ARG_OFFSET): ...this.
12067 * doc/tm.texi: Regenerate.
12068 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
12069 * config/spu/spu.c (spu_function_arg_offset): New function.
12070 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
12071 * system.h (FUNCTION_ARG_OFFSET): Poison.
12073 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
12074 Alan Hayard <alan.hayward@arm.com>
12075 David Sherwood <david.sherwood@arm.com>
12077 * target.def (truly_noop_truncation): New hook.
12078 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
12079 than TRULY_NOOP_TRUNCATION.
12080 * hooks.h (hook_bool_uint_uint_true): Declare.
12081 * hooks.c (hook_bool_uint_uint_true): New function.
12082 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
12083 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
12084 * doc/tm.texi: Regenerate.
12085 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
12086 rather than TRULY_NOOP_TRUNCATION in comments.
12087 (simplify_comparison): Likewise.
12088 (record_truncated_value): Likewise.
12089 * expmed.c (extract_bit_field_1): Likewise.
12090 (extract_split_bit_field): Likewise.
12091 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
12092 instead of TRULY_NOOP_TRUNCATION.
12093 * function.c (assign_parm_setup_block): Likewise.
12094 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
12095 * rtlhooks.c: Include target.h.
12096 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
12097 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
12098 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
12099 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
12100 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
12101 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
12102 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
12103 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
12104 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
12105 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
12106 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
12107 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
12108 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
12109 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
12110 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
12111 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
12112 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
12113 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
12114 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
12115 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
12116 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
12117 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
12118 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
12119 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
12120 * config/mips/mips.c (mips_truly_noop_truncation): New function.
12121 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
12122 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
12123 rather than TRULY_NOOP_TRUNCATION in comments.
12124 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
12125 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
12126 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
12127 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
12128 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
12129 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
12130 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
12131 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
12132 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
12133 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
12134 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
12135 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
12136 rather than TRULY_NOOP_TRUNCATION in comments.
12137 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
12138 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
12139 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
12140 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
12141 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
12142 TRULY_NOOP_TRUNCATION condition.
12143 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
12144 (TRULY_NOOP_TRUNCATION): Delete.
12145 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
12146 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
12147 * config/spu/spu.c (spu_truly_noop_truncation): New function.
12148 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
12149 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
12150 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
12151 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
12152 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
12153 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
12154 rather than TRULY_NOOP_TRUNCATION in comments.
12155 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
12156 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
12157 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
12158 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
12159 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
12160 * system.h (TRULY_NOOP_TRUNCATION): Poison.
12162 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
12165 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
12166 (*cmp_ior): Likewise.
12167 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
12168 (*ior_scc_scc_cmp): Likewise.
12169 (*and_scc_scc): Likewise.
12170 (*and_scc_scc_cmp): Likewise.
12172 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
12173 Alan Hayard <alan.hayward@arm.com>
12174 David Sherwood <david.sherwood@arm.com>
12176 * target.def (can_change_mode_class): New hook.
12177 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
12178 (hard_regno_nregs): Likewise.
12179 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
12180 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
12181 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
12182 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
12183 (LOAD_EXTEND_OP): Update accordingly.
12184 * doc/tm.texi: Regenerate.
12185 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
12186 CANNOT_CHANGE_MODE_CLASS.
12187 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
12188 (REG_CAN_CHANGE_MODE_P): ...this new macro.
12189 * combine.c (simplify_set): Update accordingly.
12190 * emit-rtl.c (validate_subreg): Likewise.
12191 * recog.c (general_operand): Likewise.
12192 * regcprop.c (mode_change_ok): Likewise.
12193 * reload1.c (choose_reload_regs): Likewise.
12194 (inherit_piecemeal_p): Likewise.
12195 * rtlanal.c (simplify_subreg_regno): Likewise.
12196 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
12197 instead of CANNOT_CHANGE_MODE_CLASS.
12198 (reload_cse_simplify_operands): Likewise.
12199 * reload.c (push_reload): Use targetm.can_change_mode_class
12200 instead of CANNOT_CHANGE_MODE_CLASS.
12201 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
12202 REG_CANNOT_CHANGE_MODE_P.
12203 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12204 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
12205 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12206 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12207 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12208 (arm_can_change_mode_class): New function.
12209 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
12210 than CANNOT_CHANGE_MODE_CLASS in comments.
12211 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12212 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
12213 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
12214 (ix86_can_change_mode_class): ...this new function, inverting the
12215 sense of the return value.
12216 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12217 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12218 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12219 (ia64_can_change_mode_class): New function.
12220 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12221 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
12222 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
12223 (m32c_can_change_mode_class): ...this new function, inverting the
12224 sense of the return value.
12225 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12226 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12227 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
12228 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
12229 (mips_can_change_mode_class): ...this new function, inverting the
12230 sense of the return value.
12231 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12232 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12233 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12234 (msp430_can_change_mode_class): New function.
12235 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12236 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
12237 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12238 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12239 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12240 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
12241 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12242 (pa_cannot_change_mode_class): Replace with...
12243 (pa_can_change_mode_class): ...this new function, inverting the
12244 sense of the return value.
12245 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
12246 than CANNOT_CHANGE_MODE_CLASS in comments.
12247 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12248 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
12249 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12250 (pdp11_cannot_change_mode_class): Replace with...
12251 (pdp11_can_change_mode_class): ...this new function, inverting the
12252 sense of the return value.
12253 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12254 * config/powerpcspe/powerpcspe-protos.h
12255 (rs6000_cannot_change_mode_class_ptr): Delete.
12256 * config/powerpcspe/powerpcspe.c
12257 (rs6000_cannot_change_mode_class_ptr): Delete.
12258 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12259 (rs6000_option_override_internal): Assign to
12260 targetm.can_change_mode_class instead of
12261 rs6000_cannot_change_mode_class_ptr.
12262 (rs6000_cannot_change_mode_class): Replace with...
12263 (rs6000_can_change_mode_class): ...this new function, inverting the
12264 sense of the return value.
12265 (rs6000_debug_cannot_change_mode_class): Replace with...
12266 (rs6000_debug_can_change_mode_class): ...this new function.
12267 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12268 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
12269 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12270 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12271 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
12273 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
12274 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12275 (rs6000_option_override_internal): Assign to
12276 targetm.can_change_mode_class instead of
12277 rs6000_cannot_change_mode_class_ptr.
12278 (rs6000_cannot_change_mode_class): Replace with...
12279 (rs6000_can_change_mode_class): ...this new function, inverting the
12280 sense of the return value.
12281 (rs6000_debug_cannot_change_mode_class): Replace with...
12282 (rs6000_debug_can_change_mode_class): ...this new function.
12283 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12284 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
12285 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
12286 (s390_can_change_mode_class): ...this new function, inverting the
12287 sense of the return value.
12288 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12289 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12290 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
12291 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12292 (sh_cannot_change_mode_class): Replace with...
12293 (sh_can_change_mode_class): ...this new function, inverting the
12294 sense of the return value.
12295 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12296 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12297 (sparc_can_change_mode_class): New function.
12298 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12299 * config/spu/spu.c (spu_can_change_mode_class): New function.
12300 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12301 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12302 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12303 (visium_can_change_mode_class): New function.
12304 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
12306 2017-09-15 Richard Biener <rguenther@suse.de>
12308 PR tree-optimization/82217
12309 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
12310 but not undefined case.
12312 2017-09-15 Jakub Jelinek <jakub@redhat.com>
12315 * postreload.c (reload_cse_simplify_operands): Skip
12316 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
12318 2017-09-15 Richard Biener <rguenther@suse.de>
12320 PR tree-optimization/68823
12321 * graphite-scop-detection.c (build_alias_set): If we have a
12322 possible dependence check whether we can handle them by just
12323 looking at the DRs DR_ACCESS_FNs.
12324 (build_scops): If build_alias_set fails, fail the SCOP.
12326 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
12328 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
12329 to support float128 built-in functions that require the ISA 3.0
12331 (BU_FLOAT128_3_HW): Likewise.
12332 (SQRTF128): Add support for the IEEE 128-bit square root and fma
12333 built-in functions.
12334 (FMAF128): Likewise.
12336 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
12337 support for built-in functions that need the ISA 3.0 IEEE 128-bit
12338 floating point instructions.
12339 (rs6000_invalid_builtin): Likewise.
12340 (rs6000_builtin_mask_names): Likewise.
12341 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
12342 (RS6000_BTM_FLOAT128_HW): Likewise.
12343 (RS6000_BTM_COMMON): Likewise.
12344 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
12346 * doc/extend.texi (RS/6000 built-in functions): Document the
12347 IEEE 128-bit floating point square root and fused multiply-add
12348 built-in functions.
12350 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
12352 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
12353 reg (r2) isn't in the set of registers defined in the prologue.
12355 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12356 Alan Hayward <alan.hayward@arm.com>
12357 David Sherwood <david.sherwood@arm.com>
12359 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
12360 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
12361 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
12362 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
12363 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
12365 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
12366 max_vectorization_factor.
12367 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
12369 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12370 Alan Hayward <alan.hayward@arm.com>
12371 David Sherwood <david.sherwood@arm.com>
12373 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
12374 (vect_worthwhile_without_simd_p): Declare.
12375 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
12376 (vectorizable_reduction): Use it.
12377 * tree-vect-stmts.c (vectorizable_shift): Likewise.
12378 (vectorizable_operation): Likewise.
12380 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12381 Alan Hayward <alan.hayward@arm.com>
12382 David Sherwood <david.sherwood@arm.com>
12384 * tree-vectorizer.h (vect_get_num_copies): New function.
12385 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
12386 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12387 (vectorizable_induction): Likewise.
12388 (vectorizable_live_operation): Likewise.
12389 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12390 (vectorizable_bswap): Likewise.
12391 (vectorizable_call): Likewise.
12392 (vectorizable_conversion): Likewise.
12393 (vectorizable_assignment): Likewise.
12394 (vectorizable_shift): Likewise.
12395 (vectorizable_operation): Likewise.
12396 (vectorizable_store): Likewise.
12397 (vectorizable_load): Likewise.
12398 (vectorizable_condition): Likewise.
12399 (vectorizable_comparison): Likewise.
12400 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
12402 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12403 Alan Hayward <alan.hayward@arm.com>
12404 David Sherwood <david.sherwood@arm.com>
12406 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
12407 of vect_init_vector.
12409 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12410 Alan Hayward <alan.hayward@arm.com>
12411 David Sherwood <david.sherwood@arm.com>
12413 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
12414 an inline wrapper that provides a location.
12415 (gimple_build_vector): Likewise.
12416 * gimple-fold.c (gimple_build_vector_from_val): New function.
12417 (gimple_build_vector): Likewise.
12418 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
12419 functions to build the initial value. Always return a gimple value.
12420 (get_initial_defs_for_reduction): Likewise. Only compute
12422 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
12423 vect_init_vector on the results from get_initial_def(s)_for_reduction.
12424 (vectorizable_induction): Use gimple_build_vector rather than
12427 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12428 Alan Hayward <alan.hayward@arm.com>
12429 David Sherwood <david.sherwood@arm.com>
12431 * target.h (vec_perm_indices): New typedef.
12432 (auto_vec_perm_indices): Likewise.
12433 * optabs-query.h: Include target.h
12434 (can_vec_perm_p): Take a vec_perm_indices *.
12435 * optabs-query.c (can_vec_perm_p): Likewise.
12436 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
12437 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12438 * tree-vect-generic.c (lower_vec_perm): Likewise.
12439 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
12440 (vect_grouped_load_supported): Likewise.
12441 (vect_shift_permute_load_chain): Likewise.
12442 (vect_permute_store_chain): Use auto_vec_perm_indices.
12443 (vect_permute_load_chain): Likewise.
12444 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
12445 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
12446 Update uses of can_vec_perm_p.
12447 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
12448 mode with a number of elements. Take a vec_perm_indices *.
12449 (vect_create_epilog_for_reduction): Update accordingly.
12450 Use auto_vec_perm_indices.
12451 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
12452 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
12453 (vect_transform_slp_perm_load): Likewise.
12454 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
12455 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
12456 (vect_gen_perm_mask_checked): Likewise.
12457 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
12458 (vect_gen_perm_mask_checked): Likewise.
12459 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
12460 (vectorizable_store): Likewise.
12461 (vectorizable_load): Likewise.
12462 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
12463 (vectorizable_bswap): Likewise.
12465 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12466 Alan Hayward <alan.hayward@arm.com>
12467 David Sherwood <david.sherwood@arm.com>
12469 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
12470 * tree.c (build_vector): Likewise.
12471 (build_vector_from_ctor): Update accordingly.
12472 (build_vector_from_val): Likewise.
12473 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
12474 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12475 * tree-vect-generic.c (add_rshift): Likewise.
12476 (expand_vector_divmod): Likewise.
12477 (optimize_vector_constructor): Likewise.
12478 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
12479 (vect_transform_slp_perm_load): Likewise.
12480 (vect_schedule_slp_instance): Likewise.
12481 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
12482 (vectorizable_call): Likewise.
12483 (vect_gen_perm_mask_any): Likewise. Add elements in order.
12484 * expmed.c (make_tree): Likewise.
12485 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
12486 a vector passed to build_vector.
12487 (fold_convert_const): Likewise.
12488 (exact_inverse): Likewise.
12489 (fold_ternary_loc): Likewise.
12490 (fold_relational_const): Likewise.
12491 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
12492 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
12493 (const_unop): Likewise. Store the reduction accumulator in a
12494 variable rather than an array.
12495 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
12496 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
12497 the new vector, rather than constructing it after the input arrays.
12498 (native_interpret_vector): Use auto_vec<tree> when building
12499 a vector passed to build_vector. Add elements in order.
12500 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
12501 auto_vec<tree> when building a vector passed to build_vector.
12502 (vect_create_epilog_for_reduction): Likewise.
12503 (vectorizable_induction): Likewise.
12504 (get_initial_def_for_reduction): Likewise. Fix indentation of
12506 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
12508 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
12509 passed to build_vector.
12511 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12512 Alan Hayward <alan.hayward@arm.com>
12513 David Sherwood <david.sherwood@arm.com>
12515 * tree-core.h (tree_base::u): Add an "nelts" field.
12516 (tree_vector): Use VECTOR_CST_NELTS as the length.
12517 * tree.c (tree_size): Likewise.
12518 (make_vector): Initialize VECTOR_CST_NELTS.
12519 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
12520 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
12521 TYPE_VECTOR_SUBPARTS.
12522 * expr.c (const_vector_mask_from_tree): Consistently use "units"
12523 as the number of units, setting it from VECTOR_CST_NELTS.
12524 (const_vector_from_tree): Likewise.
12525 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
12526 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
12527 (fold_negate_expr_1): Likewise.
12528 (fold_convert_const): Likewise.
12529 (const_binop): Likewise. Differentiate the number of output and
12531 (const_unop): Likewise.
12532 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
12533 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
12534 in cases that did the opposite.
12536 2017-09-14 Richard Biener <rguenther@suse.de>
12538 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
12541 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
12543 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
12545 2017-09-14 Jakub Jelinek <jakub@redhat.com>
12548 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
12549 if and where to split a bb, except for splitting before debug insn
12550 sequences followed by non-label real insn. Delete debug insns
12551 in between basic blocks.
12553 * combine.c (make_compound_operation_int): Formatting fixes.
12555 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
12556 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12557 * config/netbsd.h (LINK_EH_SPEC): Likewise.
12558 * config/sol2.h (LINK_EH_SPEC): Likewise.
12559 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12560 * config/s390/linux.h (LINK_SPEC): Likewise.
12561 * config/freebsd.h (LINK_EH_SPEC): Likewise.
12562 * config/openbsd.h (LINK_EH_SPEC): Likewise.
12563 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12564 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
12565 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
12566 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12567 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
12568 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
12570 2017-09-13 Jakub Jelinek <jakub@redhat.com>
12572 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
12574 (ENDFILE_LINUX_SPEC): Likewise.
12575 (LINK_EH_SPEC): Likewise.
12576 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
12577 (LINK_OS_LINUX_SPEC32): Likewise.
12578 (LINK_OS_LINUX_SPEC64): Likewise.
12579 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
12580 (LINK_OS_LINUX_SPEC): Likewise.
12582 2017-09-13 Martin Liska <mliska@suse.cz>
12584 PR middle-end/82154
12585 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
12586 CASE_HIGH is NULL_TREE.
12588 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12589 Alan Hayward <alan.hayward@arm.com>
12590 David Sherwood <david.sherwood@arm.com>
12592 * target.def (secondary_memory_needed): New hook.
12593 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
12594 instead of SECONDARY_MEMORY_NEEDED.
12595 (secondary_memory_needed_mode): Likewise.
12596 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
12597 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
12598 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
12599 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
12600 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
12601 * doc/tm.texi: Regenerate.
12602 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
12603 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
12604 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12605 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
12606 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
12607 * config/i386/i386.c (inline_secondary_memory_needed): Put the
12608 mode argument first and change the reg_class arguments to reg_class_t.
12609 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
12610 Make static. Update the call to inline_secondary_memory_needed.
12611 (ix86_register_move_cost): Update the call to
12612 inline_secondary_memory_needed.
12613 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12614 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
12616 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
12617 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12619 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
12620 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
12621 * config/mips/mips.c (mips_secondary_memory_needed): Make static
12622 and match hook interface. Add comment from mips.h.
12623 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12624 * config/mmix/mmix.md (truncdfsf2): Refer to
12625 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12627 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
12628 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
12629 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12630 (pa_secondary_memory_needed): New function.
12631 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
12632 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
12633 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12634 (pdp11_secondary_memory_needed): Make static and match hook interface.
12635 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
12636 * config/powerpcspe/powerpcspe-protos.h
12637 (rs6000_secondary_memory_needed_ptr): Delete.
12638 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
12640 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12641 (rs6000_option_override_internal): Assign to
12642 targetm.secondary_memory_needed rather than
12643 rs6000_secondary_memory_needed_ptr.
12644 (rs6000_secondary_memory_needed): Match hook interface.
12645 (rs6000_debug_secondary_memory_needed): Likewise.
12646 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
12647 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
12648 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
12649 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12650 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
12651 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
12653 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
12654 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12655 (rs6000_option_override_internal): Assign to
12656 targetm.secondary_memory_needed rather than
12657 rs6000_secondary_memory_needed_ptr.
12658 (rs6000_secondary_memory_needed): Match hook interface.
12659 (rs6000_debug_secondary_memory_needed): Likewise.
12660 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
12661 * config/s390/s390.c (s390_secondary_memory_needed): New function.
12662 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12663 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
12664 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12665 (sparc_secondary_memory_needed): New function.
12666 * lra-constraints.c (check_and_process_move): Refer to
12667 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12669 (curr_insn_transform): Likewise.
12670 (process_alt_operands): Use targetm.secondary_memory_needed
12671 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12672 (check_secondary_memory_needed_p): Likewise.
12673 (choose_split_class): Likewise.
12674 * reload.c: Unconditionally include code that was previously
12675 conditional on SECONDARY_MEMORY_NEEDED.
12676 (push_secondary_reload): Use targetm.secondary_memory_needed
12677 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12678 (push_reload): Likewise.
12679 * reload1.c: Unconditionally include code that was previously
12680 conditional on SECONDARY_MEMORY_NEEDED.
12681 (choose_reload_regs): Use targetm.secondary_memory_needed
12682 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12683 (gen_reload): Likewise.
12684 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
12686 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12687 Alan Hayward <alan.hayward@arm.com>
12688 David Sherwood <david.sherwood@arm.com>
12690 * target.def (secondary_memory_needed_mode): New hook:
12691 * targhooks.c (default_secondary_memory_needed_mode): Declare.
12692 * targhooks.h (default_secondary_memory_needed_mode): New function.
12693 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
12694 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
12695 * doc/tm.texi: Regenerate.
12696 * lra-constraints.c (check_and_process_move): Use
12697 targetm.secondary_memory_needed_mode instead of
12698 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
12699 (curr_insn_transform): Likewise.
12700 * reload.c (get_secondary_mem): Likewise.
12701 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12702 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
12704 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12705 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12706 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
12707 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12708 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
12710 * config/powerpcspe/powerpcspe-protos.h
12711 (rs6000_secondary_memory_needed_mode): Delete.
12712 * config/powerpcspe/powerpcspe.c
12713 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12714 (rs6000_secondary_memory_needed_mode): Make static.
12715 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12716 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
12718 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
12720 (rs6000_secondary_memory_needed_mode): Make static.
12721 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12722 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
12723 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12724 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12725 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
12727 (sparc_secondary_memory_needed_mode): New function.
12728 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
12730 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
12732 * config/aarch64/constraints.md (Umq): New constraint.
12733 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
12735 (mov<mode>): Update condition.
12737 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12739 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
12740 when bitposition is the same.
12742 2017-09-13 Richard Biener <rguenther@suse.de>
12744 * dwarf2out.c (output_die_symbol): Remove.
12745 (output_die): Do not output a DIEs symbol.
12747 2017-09-13 Richard Biener <rguenther@suse.de>
12749 PR middle-end/82128
12750 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
12751 default-def to avoid breaking iterator update with the weird
12752 interaction with cgraph_update_edges_for_call_stmt_node.
12754 2017-09-13 Richard Biener <rguenther@suse.de>
12756 * tree-cfg.c (verify_gimple_assign_binary): Add verification
12757 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
12758 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
12759 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
12761 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
12763 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
12764 Disable pc relative literal load irrespective of
12765 TARGET_FIX_ERR_A53_84341 for default.
12767 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
12769 * config/sparc/sparc.c (output_return): Output the source location of
12770 the insn in the delay slot, if any.
12771 (output_sibcall): Likewise.
12773 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
12776 * common.opt (-static-pie): New alias.
12777 (shared): Negate static-pie.
12778 (-no-pie): Update help text.
12780 (static-pie): New option.
12781 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
12782 -static-pie support.
12783 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
12784 (LINK_EH_SPEC): Likewise.
12785 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12786 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
12787 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
12788 * gcc.c (LINK_COMMAND_SPEC): Likewise.
12789 (init_gcc_specs): Likewise.
12790 (init_spec): Likewise.
12791 (display_help): Update help message for -pie.
12792 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
12795 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
12797 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
12798 (movdi_aarch64): Likewise.
12799 (movti_aarch64): Likewise.
12801 2017-09-12 Simon Wright <simon@pushface.org>
12804 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
12805 calculation of the minor version, always output as 0.
12807 2017-09-12 Jakub Jelinek <jakub@redhat.com>
12810 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
12811 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
12812 on it early, rather than manual conversion late. For
12813 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
12814 instead of performing manual conversion.
12816 2017-09-12 Carl Love <cel@us.ibm.com>
12818 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
12819 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
12821 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12822 VMULOSW): Add definitions.
12823 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12824 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12825 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12826 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12827 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12829 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
12831 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
12833 (movti_aarch64): Likewise.
12834 (movdf_aarch64): Likewise.
12835 (movtf_aarch64): Likewise.
12836 (load_pairdi): Likewise.
12837 (store_pairdi): Likewise.
12838 (load_pairdf): Likewise.
12839 (store_pairdf): Likewise.
12840 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
12841 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
12842 (ldr_got_small_<mode>): Likewise.
12843 (ldr_got_small_28k_<mode>): Likewise.
12844 (ldr_got_tiny): Likewise.
12845 * config/aarch64/iterators.md (ldst_sz): New.
12846 (ldpstp_sz): Likewise.
12847 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
12849 (thunderx_load): Split load_8 to load_16.
12850 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
12852 (thunderx2t99_storepair_basic): Split store_8 to store_16.
12853 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
12854 (xgene1_store_pair): Split store_8 to store_16.
12855 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
12856 (falkor_st_0_st_sd): Split store_8 to store_16.
12858 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
12860 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
12861 and store1/2/3/4 to store_4/8/12/16.
12862 * config/aarch64/aarch64.md: Update for rename.
12863 * config/arm/arm.md: Likewise.: Likewise.
12864 * config/arm/arm.c: Likewise.
12865 * config/arm/thumb1.md: Likewise.
12866 * config/arm/thumb2.md: Likewise.
12867 * config/arm/vfp.md: Likewise.
12868 * config/arm/arm-generic.md: Likewise.
12869 * config/arm/arm1020e.md: Likewise.
12870 * config/arm/arm1026ejs.md: Likewise.
12871 * config/arm/arm1136jfs.md: Likewise.
12872 * config/arm/arm926ejs.md: Likewise.
12873 * config/arm/cortex-a15.md: Likewise.
12874 * config/arm/cortex-a17.md: Likewise.
12875 * config/arm/cortex-a5.md: Likewise.
12876 * config/arm/cortex-a53.md: Likewise.
12877 * config/arm/cortex-a57.md: Likewise.
12878 * config/arm/cortex-a7.md: Likewise.
12879 * config/arm/cortex-a8.md: Likewise.
12880 * config/arm/cortex-a9.md: Likewise.
12881 * config/arm/cortex-m4.md: Likewise.
12882 * config/arm/cortex-m7.md: Likewise.
12883 * config/arm/cortex-r4.md: Likewise.
12884 * config/arm/exynos-m1.md: Likewise.
12885 * config/arm/fa526.md: Likewise.
12886 * config/arm/fa606te.md: Likewise.
12887 * config/arm/fa626te.md: Likewise.
12888 * config/arm/fa726te.md: Likewise.
12889 * config/arm/fmp626.md: Likewise.
12890 * config/arm/iwmmxt.md: Likewise.
12891 * config/arm/ldmstm.md: Likewise.
12892 * config/arm/marvell-pj4.md: Likewise.
12893 * config/arm/xgene1.md: Likewise.
12894 * config/aarch64/thunderx.md: Likewise.
12895 * config/aarch64/thunderx2t99.md: Likewise.
12896 * config/aarch64/falkor.md: Likewise.
12898 2017-09-12 Martin Liska <mliska@suse.cz>
12900 * attribs.c (private_lookup_attribute): New function.
12901 * attribs.h (private_lookup_attribute): Declared here.
12902 (lookup_attribute): Called from this place.
12904 2017-09-12 Richard Biener <rguenther@suse.de>
12906 PR tree-optimization/82157
12907 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
12908 stmts with side-effects.
12910 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12911 Alan Hayward <alan.hayward@arm.com>
12912 David Sherwood <david.sherwood@arm.com>
12914 * target.def (hard_regno_nregs): New hook.
12915 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
12916 * targhooks.h (default_hard_regno_nregs): Declare.
12917 * targhooks.c (default_hard_regno_nregs): New function.
12918 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
12919 (TARGET_HARD_REGNO_NREGS): ...this hook.
12920 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
12921 (CLASS_MAX_NREGS): Likewise.
12922 * doc/tm.texi: Regenerate.
12923 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
12924 instead of HARD_REGNO_NREGS.
12925 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
12926 HARD_REGNO_NREGS in the comment.
12927 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
12928 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
12929 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
12930 Return an unsigned int.
12931 (TARGET_HARD_REGNO_NREGS): Redefine.
12932 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
12933 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
12934 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
12935 (arc_hard_regno_nregs): New function.
12936 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
12937 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
12938 (arm_hard_regno_nregs): New function.
12939 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
12940 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
12941 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
12942 (TARGET_HARD_REGNO_NREGS): Redefine.
12943 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
12944 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
12945 (HARD_REGNO_NREGS): Delete.
12946 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
12947 (cr16_hard_regno_nregs): New function.
12948 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
12949 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
12950 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
12951 (cris_hard_regno_nregs): New function.
12952 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
12953 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
12954 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
12955 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
12956 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
12957 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
12958 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
12959 (frv_hard_regno_nregs): Make static. Take and return an
12961 (frv_class_max_nregs): Remove outdated copy of documentation.
12962 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
12963 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
12964 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
12965 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
12966 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
12967 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
12968 (TARGET_HARD_REGNO_NREGS): Redefine.
12969 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
12970 (CLASS_MAX_NREGS): Update comment.
12971 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
12972 (ia64_hard_regno_nregs): New function.
12973 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
12974 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
12975 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
12976 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
12977 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
12979 (m32c_hard_regno_nregs): Likewise. Make static.
12980 (TARGET_HARD_REGNO_NREGS): Redefine.
12981 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
12982 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
12983 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
12984 (m68k_hard_regno_nregs): New function.
12985 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
12986 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
12987 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
12988 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
12989 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
12990 Take and return an unsigned int.
12991 (TARGET_HARD_REGNO_NREGS): Redefine.
12992 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
12993 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
12994 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
12995 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
12996 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
12997 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
12998 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
12999 (msp430_hard_regno_nregs): Make static. Take and return an
13001 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
13002 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
13003 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
13004 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
13005 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
13006 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
13007 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
13008 (TARGET_HARD_REGNO_NREGS): Redefine.
13009 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
13010 (PA_HARD_REGNO_NREGS): ...this.
13011 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
13012 (PA_HARD_REGNO_NREGS): ...this.
13013 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
13014 (pa_hard_regno_nregs): New function.
13015 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
13016 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
13017 (pdp11_hard_regno_nregs): New function.
13018 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
13019 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
13020 (rs6000_hard_regno_nregs_hook): New function.
13021 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
13022 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
13023 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
13024 Take and return an unsigned int. Move earlier in file.
13025 (TARGET_HARD_REGNO_NREGS): Redefine.
13026 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
13027 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
13028 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
13029 (rl78_hard_regno_nregs): Make static. Take and return an
13031 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
13032 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
13033 (rs6000_hard_regno_nregs_hook): New function.
13034 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
13035 * config/rx/rx.c (rx_hard_regno_nregs): New function.
13036 (TARGET_HARD_REGNO_NREGS): Redefine.
13037 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
13038 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
13039 instead of HARD_REGNO_NREGS.
13040 (s390_hard_regno_nregs): New function.
13041 (s390_hard_regno_mode_ok): Add comment from s390.h.
13042 (TARGET_HARD_REGNO_NREGS): Redefine.
13043 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
13044 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
13045 (sh_hard_regno_nregs): New function.
13046 (sh_pass_in_reg_p): Use it.
13047 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
13048 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
13049 (sparc_hard_regno_nregs): New function.
13050 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
13051 * config/spu/spu.c (spu_hard_regno_nregs): New function.
13052 (spu_function_arg_advance): Use it, supplying a valid register number.
13053 (TARGET_HARD_REGNO_NREGS): Redefine.
13054 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
13055 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
13056 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
13057 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
13058 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
13059 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
13060 (CLASS_MAX_NREGS): Remove copy of old documentation.
13061 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
13062 (visium_hard_regno_nregs): New function.
13063 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
13064 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
13065 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
13066 xtensa_hard_regno_nregs): New function.
13067 * system.h (HARD_REGNO_NREGS): Poison.
13069 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13071 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
13072 hard_regno_nregs instead of HARD_REGNO_NREGS.
13073 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
13074 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
13075 (c6x_expand_epilogue): Likewise.
13076 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
13077 (frv_read_iacc_argument): Likewise.
13078 * config/sh/sh.c: Include regs.h.
13079 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
13080 (regs_used): Likewise.
13081 (output_stack_adjust): Likewise.
13082 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
13083 * expmed.c: Include regs.h.
13084 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
13085 * ree.c: Include regs.h.
13086 (combine_reaching_defs): Use hard_regno_nregs instead of
13088 (add_removable_extension): Likewise.
13090 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13092 * regs.h (hard_regno_nregs): Turn into a function.
13093 (end_hard_regno): Update accordingly.
13094 * caller-save.c (setup_save_areas): Likewise.
13095 (save_call_clobbered_regs): Likewise.
13096 (replace_reg_with_saved_mem): Likewise.
13097 (insert_restore): Likewise.
13098 (insert_save): Likewise.
13099 * combine.c (can_change_dest_mode): Likewise.
13100 (move_deaths): Likewise.
13101 (distribute_notes): Likewise.
13102 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
13103 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
13104 (rs6000_split_multireg_move): Likewise.
13105 (rs6000_register_move_cost): Likewise.
13106 (rs6000_memory_move_cost): Likewise.
13107 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
13108 (rs6000_split_multireg_move): Likewise.
13109 (rs6000_register_move_cost): Likewise.
13110 (rs6000_memory_move_cost): Likewise.
13111 * cselib.c (cselib_reset_table): Likewise.
13112 (cselib_lookup_1): Likewise.
13113 * emit-rtl.c (set_mode_and_regno): Likewise.
13114 * function.c (aggregate_value_p): Likewise.
13115 * ira-color.c (setup_profitable_hard_regs): Likewise.
13116 (check_hard_reg_p): Likewise.
13117 (calculate_saved_nregs): Likewise.
13118 (assign_hard_reg): Likewise.
13119 (improve_allocation): Likewise.
13120 (calculate_spill_cost): Likewise.
13121 * ira-emit.c (modify_move_list): Likewise.
13122 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
13123 (ira_hard_reg_in_set_p): Likewise.
13124 * ira.c (setup_reg_mode_hard_regset): Likewise.
13125 (clarify_prohibited_class_mode_regs): Likewise.
13126 (check_allocation): Likewise.
13127 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13128 (lra_setup_reg_renumber): Likewise.
13129 (setup_try_hard_regno_pseudos): Likewise.
13130 (spill_for): Likewise.
13131 (assign_hard_regno): Likewise.
13132 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
13133 * lra-constraints.c (in_class_p): Likewise.
13134 (lra_constraint_offset): Likewise.
13135 (simplify_operand_subreg): Likewise.
13136 (lra_constraints): Likewise.
13137 (split_reg): Likewise.
13138 (split_if_necessary): Likewise.
13139 (invariant_p): Likewise.
13140 (inherit_in_ebb): Likewise.
13141 * lra-lives.c (process_bb_lives): Likewise.
13142 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
13143 (get_hard_regs): Likewise.
13144 (do_remat): Likewise.
13145 * lra-spills.c (assign_spill_hard_regs): Likewise.
13146 * mode-switching.c (create_pre_exit): Likewise.
13147 * postreload.c (reload_combine_recognize_pattern): Likewise.
13148 * recog.c (peep2_find_free_register): Likewise.
13149 * regcprop.c (kill_value_regno): Likewise.
13150 (set_value_regno): Likewise.
13151 (copy_value): Likewise.
13152 (maybe_mode_change): Likewise.
13153 (find_oldest_value_reg): Likewise.
13154 (copyprop_hardreg_forward_1): Likewise.
13155 * regrename.c (check_new_reg_p): Likewise.
13156 (regrename_do_replace): Likewise.
13157 * reload.c (push_reload): Likewise.
13158 (combine_reloads): Likewise.
13159 (find_dummy_reload): Likewise.
13160 (operands_match_p): Likewise.
13161 (find_reloads): Likewise.
13162 (find_equiv_reg): Likewise.
13163 (reload_adjust_reg_for_mode): Likewise.
13164 * reload1.c (count_pseudo): Likewise.
13165 (count_spilled_pseudo): Likewise.
13166 (find_reg): Likewise.
13167 (clear_reload_reg_in_use): Likewise.
13168 (free_for_value_p): Likewise.
13169 (allocate_reload_reg): Likewise.
13170 (choose_reload_regs): Likewise.
13171 (reload_adjust_reg_for_temp): Likewise.
13172 (emit_reload_insns): Likewise.
13173 (delete_output_reload): Likewise.
13174 * rtlanal.c (subreg_get_info): Likewise.
13175 * sched-deps.c (sched_analyze_reg): Likewise.
13176 * sel-sched.c (init_regs_for_mode): Likewise.
13177 (mark_unavailable_hard_regs): Likewise.
13178 (choose_best_reg_1): Likewise.
13179 (verify_target_availability): Likewise.
13180 * valtrack.c (dead_debug_insert_temp): Likewise.
13181 * var-tracking.c (track_loc_p): Likewise.
13182 (emit_note_insn_var_location): Likewise.
13183 * varasm.c (make_decl_rtl): Likewise.
13184 * reginfo.c (choose_hard_reg_mode): Likewise.
13185 (init_reg_modes_target): Refer directly to
13186 this_target_regs->x_hard_regno_nregs.
13188 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13190 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
13191 instead of hard_regno_nregs.
13193 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13195 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
13196 end_hard_regno instead of hard_regno_nregs.
13197 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
13198 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
13199 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
13200 * ira-color.c (improve_allocation): Likewise.
13201 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13202 * lra-lives.c (mark_regno_live): Likewise.
13203 (mark_regno_dead): Likewise.
13204 * lra-remat.c (operand_to_remat): Likewise.
13205 * lra.c (collect_non_operand_hard_regs): Likewise.
13206 * postreload.c (reload_combine_note_store): Likewise.
13207 (move2add_valid_value_p): Likewise.
13208 * reload.c (regno_clobbered_p): Likewise.
13210 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13212 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
13214 * config/v850/v850.c (v850_reorg): Likewise.
13215 * reload.c (refers_to_regno_for_reload_p): Likewise.
13216 (find_equiv_reg): Likewise.
13217 * reload1.c (reload_reg_reaches_end_p): Likewise.
13219 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13221 * caller-save.c (add_used_regs): Use REG_NREGS instead of
13223 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
13224 * config/arm/arm.c (output_move_neon): Likewise.
13225 (arm_attr_length_move_neon): Likewise.
13226 (neon_split_vcombine): Likewise.
13227 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
13228 (c6x_mark_reg_written): Likewise.
13229 (c6x_dwarf_register_span): Likewise.
13230 * config/i386/i386.c (ix86_save_reg): Likewise.
13231 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
13232 (rws_access_reg): Likewise.
13233 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
13234 * mode-switching.c (create_pre_exit): Likewise.
13235 * ree.c (combine_reaching_defs): Likewise.
13236 (add_removable_extension): Likewise.
13237 * regcprop.c (find_oldest_value_reg): Likewise.
13238 (copyprop_hardreg_forward_1): Likewise.
13239 * reload.c (reload_inner_reg_of_subreg): Likewise.
13240 (push_reload): Likewise.
13241 (combine_reloads): Likewise.
13242 (find_dummy_reload): Likewise.
13243 (reload_adjust_reg_for_mode): Likewise.
13244 * reload1.c (find_reload_regs): Likewise.
13245 (forget_old_reloads_1): Likewise.
13246 (reload_reg_free_for_value_p): Likewise.
13247 (reload_adjust_reg_for_temp): Likewise.
13248 (emit_reload_insns): Likewise.
13249 (delete_output_reload): Likewise.
13250 * sel-sched.c (choose_best_reg_1): Likewise.
13251 (choose_best_pseudo_reg): Likewise.
13253 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13254 Alan Hayward <alan.hayward@arm.com>
13255 David Sherwood <david.sherwood@arm.com>
13257 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
13258 * target.def (slow_unaligned_access): New hook.
13259 * targhooks.h (default_slow_unaligned_access): Declare.
13260 * targhooks.c (default_slow_unaligned_access): New function.
13261 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
13262 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
13263 * doc/tm.texi: Regenerate.
13264 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
13265 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
13266 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
13268 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
13269 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
13271 (rs6000_slow_unaligned_access): New function.
13272 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
13273 (expand_block_compare): Likewise.
13274 (expand_strn_compare): Likewise.
13275 (rs6000_rtx_costs): Likewise.
13276 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
13277 (riscv_slow_unaligned_access): Likewise.
13278 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
13279 (riscv_slow_unaligned_access_p): ...this and make static.
13280 (riscv_option_override): Update accordingly.
13281 (riscv_slow_unaligned_access): New function.
13282 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
13283 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
13284 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
13285 (rs6000_slow_unaligned_access): New function.
13286 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
13287 (rs6000_rtx_costs): Likewise.
13288 * config/rs6000/rs6000-string.c (expand_block_compare)
13289 (expand_strn_compare): Use targetm.slow_unaligned_access instead
13290 of SLOW_UNALIGNED_ACCESS.
13291 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
13292 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
13293 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
13294 of SLOW_UNALIGNED_ACCESS.
13295 * expmed.c (simple_mem_bitfield_p): Likewise.
13296 * expr.c (alignment_for_piecewise_move): Likewise.
13297 (emit_group_load_1): Likewise.
13298 (emit_group_store): Likewise.
13299 (copy_blkmode_from_reg): Likewise.
13300 (emit_push_insn): Likewise.
13301 (expand_assignment): Likewise.
13302 (store_field): Likewise.
13303 (expand_expr_real_1): Likewise.
13304 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
13305 * lra-constraints.c (simplify_operand_subreg): Likewise.
13306 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
13307 * gimple-ssa-store-merging.c: Likewise in block comment at start
13309 * tree-ssa-strlen.c: Include target.h.
13310 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
13311 of SLOW_UNALIGNED_ACCESS.
13312 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
13314 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13316 PR rtl-optimization/82185
13317 * expmed.c (emit_store_flag_int): Only test tem if it has been
13320 2017-09-12 Richard Biener <rguenther@suse.de>
13322 PR middle-end/82149
13323 * match.pd ((FTYPE) N CMP CST): Fix typo.
13325 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
13327 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
13329 (mips_near_type_p): Add 'short_call' attribute as a synonym
13331 * doc/extend.texi (short_call): Document new function attribute.
13333 2017-09-12 Jakub Jelinek <jakub@redhat.com>
13336 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
13337 assertion check that in the condition.
13338 (get_atomic_generic_size): Likewise. Before testing if parameter
13339 has pointer type, if it has array type, call for C++
13340 default_conversion to perform array-to-pointer conversion.
13342 2017-09-12 Richard Biener <rguenther@suse.de>
13344 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
13345 for operations we cannot scalarize.
13347 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
13349 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
13350 vectors heap vectors. Clean up comments.
13351 Make visited_bbs a reference.
13352 (profitable_jump_thread_path): Make GC
13353 vectors heap vectors. Clean up comments.
13355 (convert_and_register_jump_thread_path): Make GC vectors heap
13357 (check_subpath_and_update_thread_path): Same. Clean up comments.
13358 Make visited_bbs a reference.
13359 (handle_phi): Abstract common code to to
13360 register_jump_thread_path_if_profitable.
13361 Rename VAR_BB to DEF_BB.
13363 Make GC vectors heap vectors.
13364 Make visited_bbs a reference.
13365 (handle_assignment): Same.
13366 (register_jump_thread_path_if_profitable): New.
13367 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
13369 Make GC vectors heap vectors. Clean up comments.
13370 Make visited_bbs a reference.
13371 (find_jump_threads_backwards): Make visited_bbs live in the stack.
13372 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
13375 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
13378 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
13379 words of E_DImode object are reachable by xtensa_uimm8x4 access.
13381 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
13383 Revert r251800 and r251799.
13385 2017-09-11 Martin Jambor <mjambor@suse.cz>
13388 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
13389 arguments in advance.
13390 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
13391 use it to find predecessor edges.
13392 (naive_outof_ssa): Collect vector of predecessors.
13394 2017-09-08 Jason Merrill <jason@redhat.com>
13396 PR c++/70029 - ICE with ref-qualifier and -flto
13397 * langhooks.h (struct lang_hooks_for_types): Add
13398 copy_lang_qualifiers.
13399 * attribs.c (build_type_attribute_qual_variant): Use it.
13400 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
13402 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
13403 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
13405 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
13408 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
13409 (*mulsi3_sp64): New instruction.
13410 (mulsi3): New expander.
13412 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
13414 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
13416 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13418 * sancov.c: Include memmodel.h.
13420 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
13423 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
13426 2017-09-07 Carl Love <cel@us.ibm.com>
13428 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
13429 the sldi instruction.
13431 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
13433 * sancov.c: Include tm_p.h.
13435 2017-09-07 Jakub Jelinek <jakub@redhat.com>
13438 * output.h (switch_to_other_text_partition): New declaration.
13439 * varasm.c (switch_to_other_text_partition): New function.
13440 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
13441 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
13442 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
13443 to the other text partition before emitting LCL label and switch back
13444 after emitting the word after it.
13446 2017-09-07 Richard Biener <rguenther@suse.de>
13448 * passes.def (pass_split_crit_edges): Remove instance before PRE.
13449 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
13450 critical edges here, after loop init.
13451 (pass_data_pre): Remove PROP_no_crit_edges flags.
13452 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
13453 for valueization of call args to avoid leaking VN_TOP.
13454 (visit_use): Assert we do not visit default defs.
13455 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
13456 Use error_mark_node to more easily detect leaking VN_TOP.
13457 All default-defs are varying, not VN_TOP. Mark them visited.
13458 (run_scc_vn): Make code match comment.
13460 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13462 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
13463 OPTION_MASK_FLOAT128_KEYWORD.
13464 (POWERPC_MASKS): Likewise.
13465 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
13466 support for the -mfloat128-type option, and make -mfloat128
13467 default on PowerPC Linux systems. Define or undefine
13468 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
13469 Define __float128 to be __ieee128 if IEEE 128-bit support is
13470 enabled, or undefine it.
13471 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
13472 Delete defining __FLOAT128_TYPE__.
13473 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
13474 -mfloat128-type option and make -mfloat128 default on PowerPC
13476 (TARGET_FLOAT128_TYPE): Likewise.
13477 (-mfloat128-type): Likewise.
13478 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13479 Delete the -mfloat128-type option and make -mfloat128 default on
13480 PowerPC Linux systems. Always use __ieee128 to be the keyword for
13481 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
13482 128-bit floating point is enabled. Change tests from using
13483 -mfloat128-type to -mfloat128.
13484 (rs6000_mangle_type): Use the correct mangling for the __float128
13485 type even if normal long double is restricted to 64-bits.
13486 (floatn_mode): Enable the _Float128 type by default on VSX Linux
13488 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
13489 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
13490 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
13492 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
13493 documentation for -mfloat128.
13495 2017-09-06 Olivier Hainque <hainque@adacore.com>
13497 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
13499 2017-09-06 Wish Wu <wishwu007@gmail.com>
13500 Jakub Jelinek <jakub@redhat.com>
13502 * asan.c (initialize_sanitizer_builtins): Add
13503 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
13504 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
13505 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
13506 BT_FN_VOID_UINT64_PTR variables.
13507 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
13508 (BT_FN_VOID_UINT16_UINT16): Likewise.
13509 (BT_FN_VOID_UINT32_UINT32): Likewise.
13510 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
13511 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
13512 (BT_FN_VOID_UINT64_PTR): Likewise.
13513 * common.opt (flag_sanitize_coverage): New variable.
13514 (fsanitize-coverage=trace-pc): Remove.
13515 (fsanitize-coverage=): Add.
13516 * flag-types.h (enum sanitize_coverage_code): New enum.
13517 * fold-const.c (fold_range_test): Disable non-short-circuit
13518 optimization if flag_sanitize_coverage.
13519 (fold_truth_andor): Likewise.
13520 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
13521 * opts.c (COVERAGE_SANITIZER_OPT): Define.
13522 (coverage_sanitizer_opts): New array.
13523 (get_closest_sanitizer_option): Add OPTS argument, handle also
13524 OPT_fsanitize_coverage_.
13525 (parse_sanitizer_options): Adjusted to also handle
13526 OPT_fsanitize_coverage_.
13527 (common_handle_option): Add OPT_fsanitize_coverage_.
13528 * sancov.c (instrument_comparison, instrument_switch): New function.
13529 (sancov_pass): Add trace-cmp support.
13530 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
13531 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
13532 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
13533 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
13534 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
13535 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
13536 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
13537 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
13538 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
13539 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
13541 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13543 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13544 error. Only quit immediately if parsing is complete.
13545 (BEGIN): Initialize fatal_err and parse_done.
13546 (begin fpu, end fpu): Check number of arguments.
13547 (begin arch, end arch): Likewise.
13548 (begin cpu, end cpu): Likewise.
13549 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13550 (optalias): Likewise.
13552 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13554 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13555 * config/arm/arm-isa.h: Delete. Move definitions to ...
13556 * arm-cpus.in: ... here. Use new feature and fgroup values.
13557 * config/arm/arm.c (arm_option_override): Use lower case for feature
13559 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13560 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13561 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13563 (print_isa_bits_for): New function.
13564 (gen_isa): New function.
13565 (gen_comm_data): Use print_isa_bits_for.
13566 (define feature): New keyword.
13567 (define fgroup): New keyword.
13568 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
13569 (arm-isa.h): Add rule to generate file.
13570 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13571 case for feature bit names.
13573 2017-09-06 Richard Biener <rguenther@suse.de>
13575 * tree-ssa-pre.c (NECESSARY): Remove.
13576 (create_expression_by_pieces): Do not touch pass-local flags.
13577 (insert_into_preds_of_block): Likewise.
13578 (do_pre_regular_insertion): Likewise.
13579 (eliminate_insert): Likewise.
13580 (eliminate_dom_walker::before_dom_children): Likewise.
13581 (fini_eliminate): Do not look at inserted_exprs.
13582 (mark_operand_necessary): Remove.
13583 (remove_dead_inserted_code): Replace with simple work-list
13584 algorithm based on inserted_exprs and SSA uses.
13585 (pass_pre::execute): Re-order fini_eliminate and
13586 remove_dead_inserted_code.
13588 2017-09-06 Olivier Hainque <hainque@adacore.com>
13590 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
13591 for VxWorks 7. Adjust surrounding comments.
13593 2017-09-06 Richard Biener <rguenther@suse.de>
13595 * gimple-ssa-strength-reduction.c
13596 (find_candidates_dom_walker::before_dom_children): Also allow
13599 2017-09-06 Richard Biener <rguenther@suse.de>
13601 PR tree-optimization/82108
13602 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
13603 for gap in the non-permutation SLP case.
13605 2017-09-06 Martin Jambor <mjambor@suse.cz>
13607 PR tree-optimization/82078
13608 * tree-sra.c (sort_and_splice_var_accesses): Move call to
13609 add_access_to_work_queue...
13610 (build_accesses_from_assign): ...here.
13611 (propagate_all_subaccesses): Make sure racc is the group
13612 representative, if there is one.
13614 2017-09-06 Jakub Jelinek <jakub@redhat.com>
13616 PR middle-end/82095
13617 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
13620 2017-09-06 Richard Biener <rguenther@suse.de>
13622 * gimple-ssa-strength-reduction.c
13623 (find_candidates_dom_walker::before_doom_children): Use a
13624 type and not a mode check.
13626 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
13629 * config/arm/predicates.md (arm_general_adddi_operand): Create new
13631 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
13633 2017-09-05 Jeff Law <law@redhat.com>
13635 PR tree-optimization/64910
13636 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
13637 cases where we have 3 or more operands.
13639 2017-09-05 Jakub Jelinek <jakub@redhat.com>
13641 PR middle-end/81768
13642 * omp-low.c (lower_omp_for): Recompute tree invariant if
13643 gimple_omp_for_initial/final is ADDR_EXPR.
13645 PR middle-end/81768
13646 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
13647 into gimple val before gimplification fo the COND_EXPR.
13649 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
13651 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
13652 REGION_COPY argument.
13653 (thread_through_all_blocks): Remove unused argument to
13654 duplicate_thread_path.
13656 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13657 Alan Hayward <alan.hayward@arm.com>
13658 David Sherwood <david.sherwood@arm.com>
13660 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
13661 Take a scalar_mode rather than a machine_mode.
13662 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13663 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
13664 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13665 (aarch64_gen_adjusted_ldpstp): Likewise.
13666 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
13668 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13669 Alan Hayward <alan.hayward@arm.com>
13670 David Sherwood <david.sherwood@arm.com>
13672 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
13673 Take a scalar_int_mode instead of a machine_mode.
13674 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13675 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13676 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13677 (aarch64_simd_attr_length_rglist): Delete.
13678 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
13679 a scalar_int_mode instead of a machine_mode.
13680 (aarch64_add_offset): Likewise.
13681 (aarch64_internal_mov_immediate): Likewise
13682 (aarch64_add_constant_internal): Likewise.
13683 (aarch64_add_constant): Likewise.
13684 (aarch64_movw_imm): Likewise.
13685 (aarch64_rtx_arith_op_extract_p): Likewise.
13686 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13687 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13688 Remove assert that the mode isn't a vector.
13689 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13690 (aarch64_expand_mov_immediate): Update calls after above changes.
13691 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
13692 (aarch64_and_bitmask_imm): Check for scalar integer modes.
13693 (aarch64_move_imm): Likewise.
13694 (aarch64_can_const_movi_rtx_p): Likewise.
13695 (aarch64_strip_extend): Likewise.
13696 (aarch64_extr_rtx_p): Likewise.
13697 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
13698 a CONST_INT when the mode parameter is VOIDmode.
13699 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
13701 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13703 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
13704 * stor-layout.c (bitwise_mode_for_mode): Likewise.
13705 (bitwise_type_for_mode): Update accordingly.
13707 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13709 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
13710 * stor-layout.c (mode_for_size_tree): Likewise.
13711 (mode_for_array): Update accordingly.
13712 (layout_decl): Likewise.
13713 (compute_record_mode): Likewise. Only set the mode once.
13715 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13717 * target.def (get_mask_mode): Change return type to opt_mode.
13719 * doc/tm.texi: Regenerate.
13720 * targhooks.h (default_get_mask_mode): Return an opt_mode.
13721 * targhooks.c (default_get_mask_mode): Likewise.
13722 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
13723 * optabs-query.c (can_vec_mask_load_store_p): Update use of
13724 targetm.get_mask_mode.
13725 * tree.c (build_truth_vector_type): Likewise.
13727 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13729 * machmode.h (mode_for_vector): Return an opt_mode.
13730 * stor-layout.c (mode_for_vector): Likewise.
13731 (mode_for_int_vector): Update accordingly.
13732 (layout_type): Likewise.
13733 * config/i386/i386.c (emit_memmov): Likewise.
13734 (ix86_expand_set_or_movmem): Likewise.
13735 (ix86_expand_vector_init): Likewise.
13736 (ix86_get_mask_mode): Likewise.
13737 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
13739 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
13740 * expmed.c (extract_bit_field_1): Likewise.
13741 * expr.c (expand_expr_real_2): Likewise.
13742 * optabs-query.c (can_vec_perm_p): Likewise.
13743 (can_vec_mask_load_store_p): Likewise.
13744 * optabs.c (expand_vec_perm): Likewise.
13745 * targhooks.c (default_get_mask_mode): Likewise.
13746 * tree-vect-stmts.c (vectorizable_store): Likewise.
13747 (vectorizable_load): Likewise.
13748 (get_vectype_for_scalar_type_and_size): Likewise.
13750 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13752 * machmode.h (mode_for_int_vector): New function.
13753 * stor-layout.c (mode_for_int_vector): Likewise.
13754 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
13755 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
13756 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
13757 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
13758 (s390_expand_vcond): Likewise.
13760 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13762 * machmode.h (opt_machine_mode): New type.
13763 (opt_mode<T>): Allow construction from anything that can be
13765 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
13766 (mode_for_size): Return an opt_machine_mode.
13767 * stor-layout.c (mode_for_size): Likewise.
13768 (mode_for_size_tree): Update call accordingly.
13769 (bitwise_mode_for_mode): Likewise.
13770 (make_fract_type): Likewise.
13771 (make_accum_type): Likewise.
13772 * caller-save.c (replace_reg_with_saved_mem): Update call
13774 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13775 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13776 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13777 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13778 * expmed.c (extract_bit_field_1): Likewise.
13779 * reload.c (get_secondary_mem): Likewise.
13780 * varasm.c (assemble_integer): Likewise.
13781 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
13784 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13786 * machmode.h (decimal_float_mode_for_size): New function.
13787 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
13788 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
13789 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
13790 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
13791 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
13792 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
13794 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13796 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
13797 (get_builtin_sync_mode): Likewise.
13798 (expand_ifn_atomic_compare_exchange): Likewise.
13799 (expand_builtin_atomic_clear): Likewise.
13800 (expand_builtin_atomic_test_and_set): Likewise.
13801 (fold_builtin_atomic_always_lock_free): Likewise.
13802 * calls.c (compute_argument_addresses): Likewise.
13803 (emit_library_call_value_1): Likewise.
13804 (store_one_arg): Likewise.
13805 * combine.c (combine_instructions): Likewise.
13806 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
13807 * config/arm/arm.c (arm_function_value): Likewise.
13808 (aapcs_allocate_return_reg): Likewise.
13809 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
13810 * config/i386/i386.c (construct_container): Likewise.
13811 (ix86_gimplify_va_arg): Likewise.
13812 (ix86_expand_sse_cmp): Likewise.
13813 (emit_memmov): Likewise.
13814 (emit_memset): Likewise.
13815 (expand_small_movmem_or_setmem): Likewise.
13816 (ix86_expand_pextr): Likewise.
13817 (ix86_expand_pinsr): Likewise.
13818 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
13819 * config/microblaze/microblaze.c (microblaze_block_move_straight):
13821 * config/mips/mips.c (mips_function_value_1) Likewise.
13822 (mips_block_move_straight): Likewise.
13823 (mips_expand_ins_as_unaligned_store): Likewise.
13824 * config/powerpcspe/powerpcspe.c
13825 (rs6000_darwin64_record_arg_advance_flush): Likewise.
13826 (rs6000_darwin64_record_arg_flush): Likewise.
13827 * config/rs6000/rs6000.c
13828 (rs6000_darwin64_record_arg_advance_flush): Likewise.
13829 (rs6000_darwin64_record_arg_flush): Likewise.
13830 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
13831 (sparc_function_value_1): Likewise.
13832 * config/spu/spu.c (adjust_operand): Likewise.
13833 (spu_emit_branch_or_set): Likewise.
13834 (arith_immediate_p): Likewise.
13835 * emit-rtl.c (gen_lowpart_common): Likewise.
13836 * expr.c (expand_expr_real_1): Likewise.
13837 * function.c (assign_parm_setup_block): Likewise.
13838 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
13839 * reload1.c (alter_reg): Likewise.
13840 * stor-layout.c (mode_for_vector): Likewise.
13841 (layout_type): Likewise.
13843 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13845 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
13846 (spu_convert_move): Likewise.
13847 * lower-subreg.c (resolve_simple_move): Likewise.
13849 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13852 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
13853 define_insn to a define_expand.
13854 (altivec_vsum2sws_direct): New define_insn.
13855 (altivec_vsumsws): Convert from a define_insn to a define_expand.
13857 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
13859 * config/arm/arm.c (arm_option_params_internal): Improve setting of
13862 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
13867 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
13868 realign stack if stack alignment needed is less than incoming
13871 2017-09-05 Marek Polacek <polacek@redhat.com>
13874 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
13877 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
13879 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
13881 2017-09-05 Richard Biener <rguenther@suse.de>
13883 PR tree-optimization/82084
13884 * fold-const.c (can_native_encode_string_p): Handle wide characters.
13886 2017-09-05 Richard Biener <rguenther@suse.de>
13888 PR tree-optimization/82102
13889 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
13891 2017-09-05 Martin Liska <mliska@suse.cz>
13893 PR tree-optimization/82032
13894 * tree-cfg.c (generate_range_test): New function.
13895 * tree-cfg.h (generate_range_test): Declared here.
13896 * tree-cfgcleanup.c (convert_single_case_switch): New function.
13897 (cleanup_control_expr_graph): Use it.
13898 * tree-switch-conversion.c (try_switch_expansion): Remove
13900 (emit_case_nodes): Use generate_range_test.
13902 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
13905 * config/i386/i386.md (*<btsc><mode>_mask): Add
13906 TARGET_USE_BT to insn constraint.
13907 (*btr<mode>_mask): Ditto.
13909 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
13911 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
13912 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
13914 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
13917 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
13918 TARGET_NEON and TARGET_IWMMXT.
13919 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
13920 TARGET_NEON and TARGET_IWMMXT.
13921 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
13923 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
13925 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
13926 (ix86_rewrite_tls_address): Ditto.
13927 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
13928 (ix86_rewrite_tls_address_1): Ditto.
13929 (ix86_rewrite_tls_address): Ditto.
13930 * config/i386/predicates.md (tls_address_pattern): New predicate.
13931 * config/i386/i386.md (TLS address splitter): New splitter.
13933 2017-09-04 Richard Biener <rguenther@suse.de>
13935 PR tree-optimization/82084
13936 * fold-const.h (can_native_encode_string_p): Declare.
13937 * fold-const.c (can_native_encode_string_p): Factor out from ...
13938 (native_encode_string): ... here.
13939 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
13940 vectorizing stores from constants we later cannot handle.
13942 2017-09-04 Marek Polacek <polacek@redhat.com>
13945 * doc/invoke.texi: Update -Wtautological-compare documentation.
13947 2017-09-04 Jeff Law <law@redhat.com>
13949 PR tree-optimization/64910
13950 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
13951 swap the first and last operand if the last is a constant.
13953 2017-09-04 Marek Polacek <polacek@redhat.com>
13956 * convert.c (do_narrow): When sanitizing signed integer overflows,
13957 bail out for signed types.
13958 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
13960 2017-09-04 Richard Biener <rguenther@suse.de>
13962 PR tree-optimization/82060
13963 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13964 Move devirtualization after stmt folding and before EH/AB/noreturn
13965 cleanup to get the stmt refs canonicalized. Use a bool instead
13966 of gimple_modified_p since that doesn't work for NOPs. Schedule
13967 NOPs generated by folding for removal.
13969 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13970 Alan Hayward <alan.hayward@arm.com>
13971 David Sherwood <david.sherwood@arm.com>
13973 * coretypes.h (pad_direction): New enum.
13974 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
13975 (FUNCTION_ARG_PADDING): Likewise.
13976 * target.def (function_arg_padding): New hook.
13977 * targhooks.h (default_function_arg_padding): Declare.
13978 * targhooks.c (default_function_arg_padding): New function.
13979 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
13980 (TARGET_FUNCTION_ARG_PADDING): ...this.
13981 * doc/tm.texi: Regenerate.
13982 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
13983 instead of direction.
13984 (compute_argument_addresses): Likewise.
13985 (load_register_parameters): Likewise.
13986 (emit_library_call_value_1): Likewise.
13987 (store_one_arg): Use targetm.calls.function_arg_padding instead
13988 of FUNCTION_ARG_PADDING.
13989 (must_pass_in_stack_var_size_or_pad): Likewise.
13990 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
13991 (emit_group_store): Likewise.
13992 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
13993 instead of FUNCTION_ARG_PADDING.
13994 (emit_push_insn): Likewise, and propagate enum change throughout
13996 * function.h (direction): Delete.
13997 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
13999 * function.c (assign_parm_find_stack_rtl): Likewise.
14000 (assign_parm_setup_block_p): Likewise.
14001 (assign_parm_setup_block): Likewise.
14002 (gimplify_parameters): Likewise.
14003 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
14004 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
14006 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
14007 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
14008 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
14009 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
14010 (aarch64_function_arg_padding): ...this new function.
14011 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
14012 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14013 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
14014 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
14015 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
14016 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14017 (arm_pad_arg_upward): Replace with...
14018 (arm_function_arg_padding): ...this new function.
14019 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
14021 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
14022 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
14023 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14024 (ia64_hpux_function_arg_padding): Replace with...
14025 (ia64_function_arg_padding): ...this new function. Use pad_direction
14026 instead of direction. Check for TARGET_HPUX.
14027 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
14028 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14029 (iq2000_function_arg_padding): New function.
14030 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
14031 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
14032 (mips_function_arg_padding): ...this new function.
14033 (mips_pad_reg_upward): Update accordingly.
14034 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14035 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
14036 targetm.calls.function_arg_padding.
14037 (FUNCTION_ARG_PADDING): Delete.
14038 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
14039 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
14040 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
14041 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
14042 (nios2_block_reg_padding): Return pad_direction instead of direction.
14043 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
14044 instead of direction.
14045 (nios2_function_arg_padding): Likewise. Make static.
14046 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14047 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
14048 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
14049 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
14050 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14051 (pa_function_arg_padding): Make static. Return pad_direction instead
14053 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
14054 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
14055 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
14056 instead of direction. Use targetm.calls.function_arg_padding.
14057 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
14058 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
14059 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
14060 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
14061 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
14063 (function_arg_padding): Rename to...
14064 (rs6000_function_arg_padding): ...this. Make static. Return
14065 pad_direction instead of direction.
14066 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
14067 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
14068 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
14069 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
14070 instead of direction. Use targetm.calls.function_arg_padding.
14071 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
14072 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
14073 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
14074 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
14075 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14076 (function_arg_padding): Rename to...
14077 (rs6000_function_arg_padding): ...this. Make static. Return
14078 pad_direction instead of direction.
14079 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
14080 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
14081 * config/s390/s390.c (s390_function_arg_padding): New function.
14082 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14083 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
14084 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
14085 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14086 (function_arg_padding): Rename to...
14087 (sparc_function_arg_padding): ...this. Make static. Return
14088 pad_direction instead of direction.
14089 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
14090 * config/spu/spu.c (spu_function_arg_padding): New function.
14091 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14092 * system.h (FUNCTION_ARG_PADDING): Poison.
14094 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14095 Alan Hayward <alan.hayward@arm.com>
14096 David Sherwood <david.sherwood@arm.com>
14098 * target.def (modes_tieable_p): New hook.
14099 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
14100 (TARGET_MODES_TIEABLE_P): ...this.
14101 * doc/tm.texi.in: Regenerate.
14102 * hooks.h (hook_bool_mode_mode_true): Declare.
14103 * hooks.c (hook_bool_mode_mode_true): New function.
14104 * combine.c (subst): Use targetm.modes_tieable_p instead of
14106 * dse.c (find_shift_sequence): Likewise.
14107 * expmed.c (extract_low_bits): Likewise.
14108 * lower-subreg.c: Include target.h.
14109 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
14111 * rtlanal.c (rtx_cost): Likewise.
14112 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
14113 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
14114 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
14115 (TARGET_MODES_TIEABLE_P): Redefine.
14116 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
14117 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
14118 (TARGET_MODES_TIEABLE_P): Redefine.
14119 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
14120 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
14121 (arc_modes_tieable_p): New function.
14122 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
14123 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
14124 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
14125 (arm_modes_tieable_p): Make static.
14126 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
14127 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
14128 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
14129 (TARGET_MODES_TIEABLE_P): Redefine.
14130 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
14131 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
14132 (TARGET_MODES_TIEABLE_P): Redefine.
14133 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
14134 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
14135 (cr16_modes_tieable_p): New function.
14136 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
14137 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
14138 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
14139 (TRULY_NOOP_TRUNCATION): Update comment.
14140 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
14141 (TRULY_NOOP_TRUNCATION): Update comment.
14142 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
14143 (frv_modes_tieable_p): New function.
14144 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
14145 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
14146 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
14147 (TARGET_MODES_TIEABLE_P): Redefine.
14148 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
14149 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
14150 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
14151 (TARGET_MODES_TIEABLE_P): Redefine.
14152 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
14153 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
14154 (ia64_modes_tieable_p): New function.
14155 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
14156 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
14157 (iq2000_modes_tieable_p): New function.
14158 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
14159 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
14160 (lm32_modes_tieable_p): New function.
14161 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
14162 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
14163 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
14164 (TARGET_MODES_TIEABLE_P): Redefine.
14165 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
14166 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
14167 (m32r_modes_tieable_p): New function.
14168 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
14169 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
14170 (m68k_modes_tieable_p): New function.
14171 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
14172 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
14173 (mcore_modes_tieable_p): New function.
14174 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
14175 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
14177 (TARGET_MODES_TIEABLE_P): Redefine.
14178 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
14179 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
14180 * config/mips/mips.c (mips_modes_tieable_p): Make static.
14181 (TARGET_MODES_TIEABLE_P): Redefine.
14182 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
14183 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
14184 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
14185 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
14186 (mn10300_modes_tieable_p): ...this and make static.
14187 (TARGET_MODES_TIEABLE_P): Redefine.
14188 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
14189 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
14190 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
14191 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
14192 (msp430_modes_tieable_p): Make static.
14193 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
14194 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
14195 (TARGET_MODES_TIEABLE_P): Redefine.
14196 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
14197 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
14198 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
14199 (TARGET_MODES_TIEABLE_P): Redefine.
14200 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
14201 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
14202 * config/pa/pa.c (pa_modes_tieable_p): Make static.
14203 (TARGET_MODES_TIEABLE_P): Redefine.
14204 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
14205 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
14206 (pdp11_modes_tieable_p): New function.
14207 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
14208 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
14209 (rs6000_modes_tieable_p): New function.
14210 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
14211 * config/powerpcspe/powerpcspe.md: Update comment.
14212 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
14213 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
14214 (TARGET_MODES_TIEABLE_P): Redefine.
14215 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
14216 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
14217 (rl78_modes_tieable_p): New function.
14218 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
14219 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
14220 (rs6000_modes_tieable_p): New function.
14221 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
14222 * config/rs6000/rs6000.md: Update comment.
14223 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
14224 * config/rx/rx.c (rx_modes_tieable_p): New function.
14225 (TARGET_MODES_TIEABLE_P): Redefine.
14226 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
14227 * config/s390/s390.c (s390_modes_tieable_p): New function.
14228 (TARGET_MODES_TIEABLE_P): Redefine.
14229 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
14230 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
14231 (sh_modes_tieable_p): New function.
14232 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
14233 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
14234 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
14235 (sparc_modes_tieable_p): Make static.
14236 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
14237 * config/spu/spu.c (spu_modes_tieable_p): New function.
14238 (TARGET_MODES_TIEABLE_P): Redefine.
14239 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
14240 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
14241 (TARGET_MODES_TIEABLE_P): Redefine.
14242 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
14243 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
14244 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
14245 * config/v850/v850.c (v850_modes_tieable_p): New function.
14246 (TARGET_MODES_TIEABLE_P): Redefine.
14247 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
14248 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
14249 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
14250 (visium_modes_tieable_p): New function.
14251 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
14252 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
14253 (xtensa_modes_tieable_p): New function.
14254 * system.h (MODES_TIEABLE_P): Poison.
14256 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14257 Alan Hayward <alan.hayward@arm.com>
14258 David Sherwood <david.sherwood@arm.com>
14260 * target.def (hard_regno_mode_ok): New hook.
14261 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
14262 (TARGET_HARD_REGNO_MODE_OK): ...this.
14263 * doc/tm.texi.in: Regenerate.
14264 * hooks.h (hook_bool_uint_mode_true): Declare.
14265 * hooks.c (hook_bool_uint_mode_true): New function.
14266 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
14267 HARD_REGNO_MODE_OK.
14268 * genpreds.c (write_insn_preds_c): Add an include of target.h.
14269 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
14270 instead of HARD_REGNO_MODE_OK.
14271 * caller-save.c: Include target.h.
14272 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
14273 HARD_REGNO_MODE_OK.
14274 * combine.c (can_combine_p): Likewise.
14275 (combinable_i3pat): Likewise.
14276 (can_change_dest_mode): Likewise.
14277 * expr.c (init_expr_target): Likewise.
14278 (convert_move): Likewise.
14279 (convert_modes): Likewise.
14280 * ira.c (setup_prohibited_class_mode_regs): Likewise.
14281 (setup_prohibited_mode_move_regs): Likewise.
14282 * ira.h (target_ira): Likewise.
14283 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14284 * lra-constraints.c (process_alt_operands): Likewise.
14285 (split_reg): Likewise.
14286 * recog.c (peep2_find_free_register): Likewise.
14287 * ree.c (combine_reaching_defs): Likewise.
14288 * regcprop.c (maybe_mode_change): Likewise.
14289 * reginfo.c (init_reg_sets_1): Likewise.
14290 (choose_hard_reg_mode): Likewise.
14291 (simplifiable_subregs): Likewise.
14292 * regrename.c (check_new_reg_p): Likewise.
14293 * reload.c (find_valid_class): Likewise.
14294 (find_valid_class_1): Likewise.
14295 (reload_inner_reg_of_subreg): Likewise.
14296 (push_reload): Likewise.
14297 (combine_reloads): Likewise.
14298 (find_dummy_reload): Likewise.
14299 (find_reloads): Likewise.
14300 * reload1.c (find_reg): Likewise.
14301 (set_reload_reg): Likewise.
14302 (allocate_reload_reg): Likewise.
14303 (choose_reload_regs): Likewise.
14304 (reload_adjust_reg_for_temp): Likewise.
14305 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
14306 (simplify_subreg_regno): Likewise.
14307 * sel-sched.c (init_regs_for_mode): Likewise.
14308 * varasm.c (make_decl_rtl): Likewise.
14309 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
14310 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
14311 HARD_REGNO_MODE_OK.
14312 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
14313 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
14314 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14315 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
14316 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
14317 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14318 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
14319 (arc_mode_class): Delete.
14320 (HARD_REGNO_MODE_OK): Delete.
14321 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14322 (arc_hard_regno_mode_ok): Rename old array to...
14323 (arc_hard_regno_mode_ok_modes): ...this.
14324 (arc_conditional_register_usage): Update accordingly.
14325 (arc_mode_class): Make static.
14326 (arc_hard_regno_mode_ok): New function.
14327 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
14328 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
14329 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14330 (arm_hard_regno_mode_ok): Make static.
14331 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
14332 HARD_REGNO_MODE_OK.
14333 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
14334 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
14335 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
14337 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14338 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
14339 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
14340 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
14341 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
14342 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14343 * config/bfin/predicates.md (valid_reg_operand): Use
14344 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
14345 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
14346 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
14347 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14348 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
14349 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
14350 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14351 (cr16_hard_regno_mode_ok): Make static and return a bool.
14352 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
14353 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14354 (cris_hard_regno_mode_ok): New function.
14355 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
14356 (epiphany_mode_class): Delete.
14357 (HARD_REGNO_MODE_OK): Delete.
14358 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
14359 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14360 (hard_regno_mode_ok): Rename to...
14361 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
14362 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
14363 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
14364 HARD_REGNO_MODE_OK.
14365 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
14366 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
14367 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14368 (frv_hard_regno_mode_ok): Make static and return a bool.
14369 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
14370 HARD_REGNO_MODE_OK.
14371 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
14372 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
14373 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
14374 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
14376 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14377 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
14378 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
14379 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
14381 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14382 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
14383 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14384 (ia64_hard_regno_mode_ok): New function.
14385 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
14386 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14387 (iq2000_hard_regno_mode_ok): New function.
14388 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
14389 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14390 (lm32_hard_regno_mode_ok): New function.
14391 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
14392 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
14393 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
14394 instead of HARD_REGNO_MODE_OK.
14395 (m32c_hard_regno_ok): Rename to...
14396 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
14397 (m32c_cannot_change_mode_class): Update accordingly.
14398 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14399 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
14400 (m32r_mode_class): Delete.
14401 (HARD_REGNO_MODE_OK): Delete.
14402 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14403 (m32r_hard_regno_mode_ok): Rename to...
14404 (m32r_hard_regno_modes): ...this.
14405 (m32r_mode_class): Make static.
14406 (m32r_hard_regno_mode_ok): New function.
14407 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
14408 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
14409 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14410 (m68k_hard_regno_mode_ok): Make static.
14411 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
14412 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14413 (mcore_hard_regno_mode_ok): New function.
14414 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
14415 (HARD_REGNO_MODE_OK): Delete.
14416 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
14418 (microblaze_hard_regno_mode_ok_p): ...this and make static.
14419 (microblaze_hard_regno_mode_ok): New function.
14420 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14421 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
14422 (mips_hard_regno_mode_ok): Delete.
14423 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
14424 (mips_hard_regno_mode_ok_p): ...this and make static.
14425 (mips_hard_regno_mode_ok_p): Rename to...
14426 (mips_hard_regno_mode_ok_uncached): ...this.
14427 (mips_hard_regno_mode_ok): New function.
14428 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
14429 of HARD_REGNO_MODE_OK.
14430 (mips_option_override): Update after above name changes.
14431 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14432 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
14433 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
14434 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
14435 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
14436 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14437 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
14438 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
14439 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
14440 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14441 (msp430_hard_regno_mode_ok): Make static and return a bool.
14442 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
14443 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
14444 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
14446 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14447 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
14448 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
14449 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
14450 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
14451 (PA_HARD_REGNO_MODE_OK): ...this
14452 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
14453 (PA_HARD_REGNO_MODE_OK): ...this.
14454 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14455 (pa_hard_regno_mode_ok): New function.
14456 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
14457 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14458 (pdp11_hard_regno_mode_ok): New function.
14459 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
14460 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
14462 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
14464 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14465 (rs6000_hard_regno_mode_ok): Rename to...
14466 (rs6000_hard_regno_mode_ok_uncached): ...this.
14467 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14468 (rs6000_hard_regno_mode_ok): New function.
14469 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
14470 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
14471 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
14472 (riscv_hard_regno_mode_ok): ...this and make static.
14473 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14474 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
14475 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
14476 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14477 (rl78_hard_regno_mode_ok): Make static and return bool.
14478 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
14479 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
14481 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
14482 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14483 (rs6000_hard_regno_mode_ok): Rename to...
14484 (rs6000_hard_regno_mode_ok_uncached): ...this.
14485 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14486 (rs6000_hard_regno_mode_ok): New function.
14487 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
14488 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
14489 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14490 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
14491 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
14492 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
14493 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14494 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
14495 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
14496 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14497 (sh_hard_regno_mode_ok): Make static.
14498 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
14499 instead of HARD_REGNO_MODE_OK.
14500 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
14501 (sparc_mode_class): Delete.
14502 (HARD_REGNO_MODE_OK): Delete.
14503 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14504 (hard_regno_mode_classes): Make static.
14505 (sparc_mode_class): Likewise.
14506 (sparc_hard_regno_mode_ok): New function.
14507 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
14508 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
14509 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
14511 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14512 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
14513 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
14514 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
14515 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
14516 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14517 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
14518 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
14519 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14520 (visium_hard_regno_mode_ok): New function.
14521 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
14522 instead of HARD_REGNO_MODE_OK.
14523 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
14524 (HARD_REGNO_MODE_OK): Delete.
14525 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
14526 (xtensa_hard_regno_mode_ok_p): ...this and make static.
14527 (xtensa_option_override): Update accordingly.
14528 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14529 (xtensa_hard_regno_mode_ok): New function.
14530 * system.h (HARD_REGNO_MODE_OK): Poison.
14532 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14533 Alan Hayward <alan.hayward@arm.com>
14534 David Sherwood <david.sherwood@arm.com>
14536 * target.def (hard_regno_call_part_clobbered): New hook.
14537 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
14538 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
14539 * doc/tm.texi: Regenerate.
14540 * hooks.h (hook_bool_uint_mode_false): Declare.
14541 * hooks.c (hook_bool_uint_mode_false): New function.
14542 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14543 * cselib.c (cselib_process_insn): Use
14544 targetm.hard_regno_call_part_clobbered instead of
14545 HARD_REGNO_CALL_PART_CLOBBERED.
14546 * ira-conflicts.c (ira_build_conflicts): Likewise.
14547 * ira-costs.c (ira_tune_allocno_costs): Likewise.
14548 * lra-constraints.c (need_for_call_save_p): Likewise.
14549 * lra-lives.c: Include target.h.
14550 (check_pseudos_live_through_calls): Use
14551 targetm.hard_regno_call_part_clobbered instead of
14552 HARD_REGNO_CALL_PART_CLOBBERED.
14553 * regcprop.c: Include target.h.
14554 (copyprop_hardreg_forward_1): Use
14555 targetm.hard_regno_call_part_clobbered instead of
14556 HARD_REGNO_CALL_PART_CLOBBERED.
14557 * reginfo.c (choose_hard_reg_mode): Likewise.
14558 * regrename.c (check_new_reg_p): Likewise.
14559 * reload.c (find_equiv_reg): Likewise.
14560 * reload1.c (emit_reload_insns): Likewise.
14561 * sched-deps.c (deps_analyze_insn): Likewise.
14562 * sel-sched.c (init_regs_for_mode): Likewise.
14563 (mark_unavailable_hard_regs): Likewise.
14564 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
14565 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14566 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
14568 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14569 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14570 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
14572 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
14574 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14575 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14576 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
14578 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14579 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14580 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
14582 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14583 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
14585 * config/powerpcspe/powerpcspe.c
14586 (rs6000_hard_regno_call_part_clobbered): New function.
14587 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14588 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14589 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
14591 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14592 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14593 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
14595 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14596 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14597 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
14599 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14600 Alan Hayward <alan.hayward@arm.com>
14601 David Sherwood <david.sherwood@arm.com>
14603 * rtl.h (subreg_memory_offset): Declare.
14604 * emit-rtl.c (subreg_memory_offset): New function.
14605 * expmed.c (store_bit_field_1): Use it.
14606 * expr.c (undefined_operand_subword_p): Likewise.
14607 * simplify-rtx.c (simplify_subreg): Likewise.
14609 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
14611 PR rtl-optimization/57448
14614 * optabs.c (expand_atomic_load): Place compiler memory barriers if
14615 using atomic_load pattern.
14616 (expand_atomic_store): Likewise.
14618 2017-09-04 Jakub Jelinek <jakub@redhat.com>
14621 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
14622 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
14623 handling. Use replace_call_with_value with NULL instead of
14624 gsi_replace, unlink_stmt_vdef and release_defs.
14626 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
14629 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
14631 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14634 * rtl.h (emit_library_call_value_1): Declare.
14635 (emit_library_call): Replace declaration with a series of overloads.
14636 Remove the parameter count argument.
14637 (emit_library_call_value): Likewise.
14638 * calls.c (emit_library_call_value_1): Make global. Replace varargs
14639 with an "rtx_mode_t *".
14640 (emit_library_call_value): Delete.
14641 (emit_library_call): Likewise.
14642 * asan.c (asan_emit_stack_protection): Update calls accordingly.
14643 (asan_emit_allocas_unpoison): Likewise.
14644 * builtins.c (expand_builtin_powi): Likewise.
14645 (expand_asan_emit_allocas_unpoison): Likewise.
14646 * cfgexpand.c (expand_main_function): Likewise.
14647 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
14648 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
14649 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
14650 * config/arm/arm.c (arm_trampoline_init): Likewise.
14651 (arm_call_tls_get_addr): Likewise.
14652 (arm_expand_divmod_libfunc): Likewise.
14653 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
14654 (smulsi3_highpart): Likewise.
14655 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
14656 (c6x_expand_compare): Likewise.
14657 (c6x_expand_movmem): Likewise.
14658 * config/frv/frv.c (frv_trampoline_init): Likewise.
14659 * config/i386/i386.c (ix86_trampoline_init): Likewise.
14660 (ix86_expand_divmod_libfunc): Likewise.
14661 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
14662 (ia64_expand_compare): Likewise.
14663 (ia64_profile_hook): Likewise.
14664 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
14665 (nonlocal_goto): Likewise.
14666 (restore_stack_nonlocal): Likewise.
14667 * config/m32r/m32r.c (block_move_call): Likewise.
14668 (m32r_trampoline_init): Likewise.
14669 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
14670 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
14671 (m68k_call_m68k_read_tp): Likewise.
14672 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
14673 (microblaze_expand_divide): Likewise.
14674 * config/mips/mips.h (mips_args): Likewise.
14675 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
14676 (MIPS_ICACHE_SYNC): Likewise.
14677 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
14678 (nios2_trampoline_init): Likewise.
14679 * config/pa/pa.c (hppa_tls_call): Likewise.
14680 (pa_trampoline_init): Likewise.
14681 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
14682 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
14683 (expand_strn_compare): Likewise.
14684 (rs6000_generate_compare): Likewise.
14685 (rs6000_expand_float128_convert): Likewise.
14686 (output_profile_hook): Likewise.
14687 (rs6000_trampoline_init): Likewise.
14688 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
14689 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
14690 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
14691 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
14692 (rs6000_generate_compare): Likewise.
14693 (rs6000_expand_float128_convert): Likewise.
14694 (output_profile_hook): Likewise.
14695 (rs6000_trampoline_init): Likewise.
14696 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
14697 * config/sh/sh.c (sh_trampoline_init): Likewise.
14698 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
14699 (sparc_emit_float_lib_cmp): Likewise.
14700 (sparc32_initialize_trampoline): Likewise.
14701 (sparc64_initialize_trampoline): Likewise.
14702 (sparc_profile_hook): Likewise.
14703 * config/spu/spu.c (ea_load_store): Likewise.
14704 * config/spu/spu.md (floatunssidf2): Likewise.
14705 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
14706 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
14707 * config/visium/visium.c (expand_block_move_4): Likewise.
14708 (expand_block_move_2): Likewise.
14709 (expand_block_move_1): Likewise.
14710 (expand_block_set_4): Likewise.
14711 (expand_block_set_2): Likewise.
14712 (expand_block_set_1): Likewise.
14713 (visium_trampoline_init): Likewise.
14714 (visium_profile_hook): Likewise.
14715 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
14716 (xtensa_setup_frame_addresses): Likewise.
14717 (xtensa_trampoline_init): Likewise.
14718 * except.c (sjlj_emit_function_enter): Likewise.
14719 (sjlj_emit_function_exit): Likewise.
14720 * explow.c (allocate_dynamic_stack_space): Likewise.
14721 (probe_stack_range): Likewise.
14722 * expr.c (convert_mode_scalar): Likewise.
14723 * optabs.c (expand_binop): Likewise.
14724 (expand_twoval_binop_libfunc): Likewise.
14725 (expand_unop): Likewise.
14726 (prepare_cmp_insn): Likewise.
14727 (prepare_float_lib_cmp): Likewise.
14728 (expand_float): Likewise.
14729 (expand_fix): Likewise.
14730 (expand_fixed_convert): Likewise.
14731 (maybe_emit_sync_lock_test_and_set): Likewise.
14732 (expand_atomic_compare_and_swap): Likewise.
14733 (expand_mem_thread_fence): Likewise.
14734 (expand_atomic_fetch_op): Likewise.
14736 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
14738 * doc/generic.texi (OpenACC): Adjust URL.
14739 * doc/invoke.texi (C Dialect Options): Ditto.
14741 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
14743 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
14744 predicate for operand 1. Add (m,<S>) constraint.
14745 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
14746 Prevent memory operand 1 with register operand 2.
14748 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
14750 PR rtl-optimization/82024
14751 * combine.c (try_combine): If the combination result is a PARALLEL,
14752 and we only need to retain the SET in there that would be placed
14753 at I2, check that we can place that at I3 instead, before doing so.
14755 2017-09-01 Jakub Jelinek <jakub@redhat.com>
14758 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
14760 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
14761 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
14764 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
14765 Jeff Law <law@redhat.com>
14767 * varasm.c (bss_initializer_p): Do not put constants into .bss
14768 (categorize_decl_for_section): Handle bss_initializer_p returning
14769 false when DECL_INITIAL is NULL.
14771 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14774 * config/s390/s390.c (s390_can_inline_p): New function.
14776 2017-09-01 Jeff Law <law@redhat.com>
14778 PR tree-optimization/82052
14779 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
14780 Always initialize the returned slot after a hash table miss
14781 when INSERT is true.
14783 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
14785 * config/s390/s390.md (mem_signal_fence): Remove.
14786 * doc/md.texi (mem_signal_fence): Remove.
14787 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
14789 * target-insns.def (mem_signal_fence): Remove.
14791 2017-09-01 Jakub Jelinek <jakub@redhat.com>
14794 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
14797 * asan.c (create_odr_indicator): Strip name encoding from assembler
14798 name before appending it after __odr_asan_.
14800 2017-09-01 Martin Liska <mliska@suse.cz>
14802 PR tree-optimization/82059
14803 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
14804 frequency only when an edge is redirected.
14806 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14808 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
14809 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
14810 (arc_conditional_register_usage): Remove ARC600 lp_count
14812 (arc_file_start): Emit Tag_ARC_CPU_variation.
14813 (arc_can_use_doloop_p): New conditions to use ZOLs.
14814 (hwloop_fail): New function.
14815 (hwloop_optimize): Likewise.
14816 (hwloop_pattern_reg): Likewise.
14817 (arc_doloop_hooks): New struct, to be used with reorg_loops.
14818 (arc_reorg_loops): New function, calls reorg_loops.
14819 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
14820 (arc600_corereg_hazard): Remove ZOL checking, case handled by
14822 (arc_loop_hazard): Remove function, functionality moved into
14824 (arc_hazard): Remove arc_loop_hazard call.
14825 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
14826 into hwloop_optimize.
14827 (arc_label_align): Remove ZOL handling.
14828 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
14829 * config/arc/arc.md (doloop_begin): Remove pattern.
14830 (doloop_begin_i): Likewise.
14831 (doloop_end_i): Likewise.
14832 (doloop_fallback): Likewise.
14833 (doloop_fallback_m): Likewise.
14834 (doloop_end): Reimplement expand.
14835 (arc_lp): New pattern for LP instruction.
14836 (loop_end): New pattern.
14837 (loop_fail): Likewise.
14838 (decrement_and_branch_until_zero): Likewise.
14839 * config/arc/arc.opt (mlpc-width): New option.
14840 * doc/invoke.texi (mlpc-width): Document option.
14842 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14844 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
14845 (arc_ccfsm_advance): Fix checking for delay slots.
14846 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
14848 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14850 * config/arc/arc.md (movqi_insn): Add stores to save constant long
14852 (movhi_insn): Update store instruction constraint which are saving
14853 6-bit short immediates.
14854 (movsi_insn): Consider also short scaled load operations.
14855 (zero_extendhisi2_i): Use Usd constraint instead of T.
14856 (extendhisi2_i): Add q constraint.
14857 (arc_clzsi2): Add type and length attributes.
14858 (arc_ctzsi2): Likewise.
14859 * config/arc/constraints.md (Usc): Update constraint, the
14860 assembler can parse two relocations for a single instruction.
14862 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14864 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
14865 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
14867 2017-08-31 Olivier Hainque <hainque@adacore.com>
14869 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
14870 match as powerpc-wrs-vxworks*.
14872 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
14874 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
14875 register constraint for by-element operand.
14876 (aarch64_mls_elt_merge<mode>): Likewise.
14878 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14880 * config/arc/arc.c (arc_can_follow_jump): Check for short
14883 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14885 * config.gcc: Use g.opt for arc.
14886 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
14887 functionality moved to ...
14888 (legitimate_scaled_address_p): New function, ...here.
14889 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
14890 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
14891 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
14893 (arc_override_options): Handle G option.
14894 (arc_output_pic_addr_const): Correct function definition.
14895 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
14896 (arc_decl_anon_ns_mem_p): Delete.
14897 (arc_in_small_data_p): Overhaul this function to take into
14898 consideration the value given via G option.
14899 (arc_rewrite_small_data_1): Renamed and corrected old
14900 arc_rewrite_small_data function.
14901 (arc_rewrite_small_data): New function.
14902 (small_data_pattern): Don't use pic_offset_table_rtx.
14903 * config/arc/arc.h (CC1_SPEC): Recognize G option.
14904 * config/arc/simdext.md (movmisalignv2hi): Use
14905 prepare_move_operands function.
14907 (movmisalign*): Likewise.
14908 * doc/invoke.texi (ARC options): Document -G option.
14910 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14912 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
14914 * config/arc/arc.c (arc_print_operand): Output scalled address for
14915 sdata whenever is possible.
14916 (arc_in_small_data_p): Allow sdata for 64bit datum when double
14917 load/stores are available.
14918 (compact_sda_memory_operand): Check for the alignment required by
14919 code density instructions.
14920 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
14922 * config/arc/constraints.md (Usd): Update constraint.
14923 (Us0): New constraint.
14924 (Usc): Update constraint.
14926 2017-08-31 Richard Biener <rguenther@suse.de>
14928 PR middle-end/82054
14929 * dwarf2out.c (dwarf2out_early_global_decl): Process each
14930 function only once.
14932 2017-08-31 Tamar Christina <tamar.christina@arm.com>
14934 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
14935 Resize type_signature.
14937 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14938 Alan Hayward <alan.hayward@arm.com>
14939 David Sherwood <david.sherwood@arm.com>
14941 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
14942 subregs whose inner modes can be stored in GPRs.
14943 (aarch64_classify_index): Likewise.
14945 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14946 Alan Hayward <alan.hayward@arm.com>
14947 David Sherwood <david.sherwood@arm.com>
14949 * config/aarch64/iterators.md (V_cmp_result): Rename to...
14950 (V_INT_EQUIV): ...this.
14951 (v_cmp_result): Rename to...
14952 (v_int_equiv): ...this.
14953 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
14954 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
14955 (copysign<mode>3): Likewise.
14956 (aarch64_simd_bsl<mode>_internal): Likewise.
14957 (aarch64_simd_bsl<mode>): Likewise.
14958 (vec_cmp<mode><mode>): Likewise.
14959 (vcond<mode><mode>): Likewise.
14960 (vcond<v_cmp_mixed><mode>): Likewise.
14961 (vcondu<mode><v_cmp_mixed>): Likewise.
14962 (aarch64_cm<optab><mode>): Likewise.
14963 (aarch64_cmtst<mode>): Likewise.
14964 (aarch64_fac<optab><mode>): Likewise.
14965 (vec_perm_const<mode>): Likewise.
14966 (vcond_mask_<mode><v_cmp_result>): Rename to...
14967 (vcond_mask_<mode><v_int_equiv>): ...this.
14968 (vec_cmp<mode><v_cmp_result>): Rename to...
14969 (vec_cmp<mode><v_int_equiv>): ...this.
14971 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14972 Alan Hayward <alan.hayward@arm.com>
14973 David Sherwood <david.sherwood@arm.com>
14975 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
14977 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
14978 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
14979 (UNSPEC_LD4_DREG): New unspecs.
14980 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
14981 (aarch64_ld2<mode>_dreg_be): Replace with...
14982 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
14984 (aarch64_ld3<mode>_dreg_le)
14985 (aarch64_ld3<mode>_dreg_be): Replace with...
14986 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
14988 (aarch64_ld4<mode>_dreg_le)
14989 (aarch64_ld4<mode>_dreg_be): Replace with...
14990 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
14993 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14995 PR tree-optimization/81987
14996 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
14997 insert an initializer in a location not dominated by the stride
15000 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
15002 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
15003 on the entire header of the finally block in the fallthru case.
15005 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
15007 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
15009 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
15011 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
15012 rs6000_emit_move_from_cr and call renamed function.
15013 (rs6000_emit_prologue): Call renamed functions.
15014 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
15015 movesi_from_cr, remove volatile CRs.
15017 2017-08-30 Jon Beniston <jon@beniston.com>
15018 Richard Biener <rguenther@suse.de>
15020 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
15021 of VECTOR_MODE_P check.
15022 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
15023 element vector types.
15025 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15027 * df.h (df_read_modify_subreg_p): Remove in favor of...
15028 * rtl.h (read_modify_subreg_p): ...this new function. Take a
15029 const_rtx instead of an rtx.
15030 * cprop.c (local_cprop_find_used_regs): Update accordingly.
15031 * df-problems.c (df_word_lr_mark_ref): Likewise.
15032 * ira-lives.c (mark_pseudo_reg_live): Likewise.
15033 (mark_pseudo_reg_dead): Likewise.
15034 (mark_ref_dead): Likewise.
15035 * reginfo.c (init_subregs_of_mode): Likewise.
15036 * sched-deps.c (sched_analyze_1): Likewise.
15037 * df-scan.c (df_def_record_1): Likewise.
15038 (df_uses_record): Likewise.
15039 (df_read_modify_subreg_p): Remove in favor of...
15040 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
15041 const_rtx instead of an rtx.
15043 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15044 Alan Hayward <alan.hayward@arm.com>
15045 David Sherwood <david.sherwood@arm.com>
15047 * rtl.h (partial_subreg_p): New function.
15048 * caller-save.c (save_call_clobbered_regs): Use it.
15049 * calls.c (expand_call): Likewise.
15050 * combine.c (combinable_i3pat): Likewise.
15051 (simplify_set): Likewise.
15052 (make_extraction): Likewise.
15053 (make_compound_operation_int): Likewise.
15054 (gen_lowpart_or_truncate): Likewise.
15055 (force_to_mode): Likewise.
15056 (make_field_assignment): Likewise.
15057 (reg_truncated_to_mode): Likewise.
15058 (record_truncated_value): Likewise.
15059 (move_deaths): Likewise.
15060 * cse.c (record_jump_cond): Likewise.
15061 (cse_insn): Likewise.
15062 * cselib.c (cselib_lookup_1): Likewise.
15063 * expmed.c (extract_bit_field_using_extv): Likewise.
15064 * function.c (assign_parm_setup_reg): Likewise.
15065 * ifcvt.c (noce_convert_multiple_sets): Likewise.
15066 * ira-build.c (create_insn_allocnos): Likewise.
15067 * lra-coalesce.c (merge_pseudos): Likewise.
15068 * lra-constraints.c (match_reload): Likewise.
15069 (simplify_operand_subreg): Likewise.
15070 (curr_insn_transform): Likewise.
15071 * lra-lives.c (process_bb_lives): Likewise.
15072 * lra.c (new_insn_reg): Likewise.
15073 (lra_substitute_pseudo): Likewise.
15074 * regcprop.c (mode_change_ok): Likewise.
15075 (maybe_mode_change): Likewise.
15076 (copyprop_hardreg_forward_1): Likewise.
15077 * reload.c (push_reload): Likewise.
15078 (find_reloads): Likewise.
15079 (find_reloads_subreg_address): Likewise.
15080 * reload1.c (alter_reg): Likewise.
15081 (eliminate_regs_1): Likewise.
15082 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15084 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
15086 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
15087 back to if statements, including unpack.
15089 2017-08-30 Martin Liska <mliska@suse.cz>
15091 PR inline-asm/82001
15092 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
15094 (func_checker::compare_asm_inputs_outputs): ... this function.
15095 (func_checker::compare_gimple_asm): Use the function to compare
15096 also ASM constrains.
15097 * ipa-icf-gimple.h: Rename the function.
15099 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15100 Alan Hayward <alan.hayward@arm.com>
15101 David Sherwood <david.sherwood@arm.com>
15103 * coretypes.h (complex_mode): New type.
15104 * gdbhooks.py (build_pretty_printer): Handle it.
15105 * machmode.h (complex_mode): New class.
15106 (complex_mode::includes_p): New function.
15107 (is_complex_int_mode): Likewise.
15108 (is_complex_float_mode): Likewise.
15109 * genmodes.c (get_mode_class): Handle complex mode classes.
15110 * function.c (expand_function_end): Use is_complex_int_mode.
15112 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15113 Alan Hayward <alan.hayward@arm.com>
15114 David Sherwood <david.sherwood@arm.com>
15116 * coretypes.h (scalar_mode_pod): New typedef.
15117 * gdbhooks.py (build_pretty_printer): Handle it.
15118 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
15119 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
15120 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
15121 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
15122 as_a <scalar_mode>.
15124 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15125 Alan Hayward <alan.hayward@arm.com>
15126 David Sherwood <david.sherwood@arm.com>
15128 * machmode.h (mode_for_vector): Take a scalar_mode instead
15130 * stor-layout.c (mode_for_vector): Likewise.
15131 * explow.c (promote_mode): Use as_a <scalar_mode>.
15132 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
15134 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15135 Alan Hayward <alan.hayward@arm.com>
15136 David Sherwood <david.sherwood@arm.com>
15138 * target.def (preferred_simd_mode): Take a scalar_mode
15139 instead of a machine_mode.
15140 * targhooks.h (default_preferred_simd_mode): Likewise.
15141 * targhooks.c (default_preferred_simd_mode): Likewise.
15142 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
15143 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
15144 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
15145 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
15146 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
15147 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
15148 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
15149 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
15151 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
15152 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
15153 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
15154 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
15155 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
15156 * doc/tm.texi: Regenerate.
15157 * optabs-query.c (can_vec_mask_load_store_p): Return false for
15160 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15161 Alan Hayward <alan.hayward@arm.com>
15162 David Sherwood <david.sherwood@arm.com>
15164 * target.def (scalar_mode_supported_p): Take a scalar_mode
15165 instead of a machine_mode.
15166 * targhooks.h (default_scalar_mode_supported_p): Likewise.
15167 * targhooks.c (default_scalar_mode_supported_p): Likewise.
15168 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
15169 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
15170 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
15171 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
15172 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
15173 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
15174 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
15175 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
15176 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
15177 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
15178 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
15179 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
15181 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
15182 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
15183 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
15184 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
15185 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
15187 * doc/tm.texi: Regenerate.
15189 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15190 Alan Hayward <alan.hayward@arm.com>
15191 David Sherwood <david.sherwood@arm.com>
15193 * coretypes.h (opt_scalar_mode): New typedef.
15194 * gdbhooks.py (build_pretty_printers): Handle it.
15195 * machmode.h (mode_iterator::get_2xwider): Add overload for
15197 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
15199 * expr.c (convert_mode_scalar): Likewise.
15200 * omp-low.c (omp_clause_aligned_alignment): Likewise.
15201 * optabs.c (expand_float): Likewise.
15202 (expand_fix): Likewise.
15203 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15205 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15206 Alan Hayward <alan.hayward@arm.com>
15207 David Sherwood <david.sherwood@arm.com>
15209 * optabs.c (expand_float): Explicitly check for scalars before
15210 using a branching expansion.
15211 (expand_fix): Likewise.
15213 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15214 Alan Hayward <alan.hayward@arm.com>
15215 David Sherwood <david.sherwood@arm.com>
15217 * expr.c (convert_mode): Split scalar handling out into...
15218 (convert_mode_scalar): ...this new function. Treat the modes
15221 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15222 Alan Hayward <alan.hayward@arm.com>
15223 David Sherwood <david.sherwood@arm.com>
15225 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
15227 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
15229 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15230 Alan Hayward <alan.hayward@arm.com>
15231 David Sherwood <david.sherwood@arm.com>
15233 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
15234 rather than a machine_mode.
15235 (fixed_from_string): Likewise.
15236 (fixed_convert): Likewise.
15237 (fixed_convert_from_int): Likewise.
15238 (fixed_convert_from_real): Likewise.
15239 (real_convert_from_fixed): Likewise.
15240 * fixed-value.c (fixed_from_double_int): Likewise.
15241 (fixed_from_string): Likewise.
15242 (fixed_convert): Likewise.
15243 (fixed_convert_from_int): Likewise.
15244 (fixed_convert_from_real): Likewise.
15245 (real_convert_from_fixed): Likewise.
15246 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
15248 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15249 Alan Hayward <alan.hayward@arm.com>
15250 David Sherwood <david.sherwood@arm.com>
15252 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
15253 of separate mode class checks. Do not allow vector modes here.
15254 (immed_wide_int_const): Use as_a <scalar_mode>.
15255 * explow.c (trunc_int_for_mode): Likewise.
15256 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
15257 (wi::shwi): Likewise.
15258 (wi::min_value): Likewise.
15259 (wi::max_value): Likewise.
15260 * dwarf2out.c (loc_descriptor): Likewise.
15261 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
15262 for CONST_WIDE_INT.
15264 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15265 Alan Hayward <alan.hayward@arm.com>
15266 David Sherwood <david.sherwood@arm.com>
15268 * tree.h (SCALAR_TYPE_MODE): New macro.
15269 * expr.c (expand_expr_addr_expr_1): Use it.
15270 (expand_expr_real_2): Likewise.
15271 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
15272 (fold_convert_const_fixed_from_int): Likewise.
15273 (fold_convert_const_fixed_from_real): Likewise.
15274 (native_encode_fixed): Likewise
15275 (native_encode_complex): Likewise
15276 (native_encode_vector): Likewise.
15277 (native_interpret_fixed): Likewise.
15278 (native_interpret_real): Likewise.
15279 (native_interpret_complex): Likewise.
15280 (native_interpret_vector): Likewise.
15281 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
15282 (simd_clone_adjust_argument_types): Likewise.
15283 (simd_clone_init_simd_arrays): Likewise.
15284 (simd_clone_adjust): Likewise.
15285 * stor-layout.c (layout_type): Likewise.
15286 * tree.c (build_minus_one_cst): Likewise.
15287 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
15288 * tree-inline.c (estimate_move_cost): Likewise.
15289 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
15290 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
15291 (vectorizable_reduction): Likewise.
15292 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
15293 (vect_recog_mixed_size_cond_pattern): Likewise.
15294 (check_bool_pattern): Likewise.
15295 (adjust_bool_pattern): Likewise.
15296 (search_type_for_mask_1): Likewise.
15297 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
15298 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15299 (vectorizable_load): Likewise.
15300 (vectorizable_store): Likewise.
15301 * ubsan.c (ubsan_encode_value): Likewise.
15302 * varasm.c (output_constant): Likewise.
15304 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15305 Alan Hayward <alan.hayward@arm.com>
15306 David Sherwood <david.sherwood@arm.com>
15308 * coretypes.h (scalar_mode): New class.
15309 * machmode.h (scalar_mode): Likewise.
15310 (scalar_mode::includes_p): New function.
15311 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
15312 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
15313 * genmodes.c (get_mode_class): Handle remaining scalar modes.
15314 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
15315 * expmed.c (store_bit_field_1): Likewise.
15316 (extract_bit_field_1): Likewise.
15317 * expr.c (write_complex_part): Likewise.
15318 (read_complex_part): Likewise.
15319 (emit_move_complex_push): Likewise.
15320 (expand_expr_real_2): Likewise.
15321 * function.c (assign_parm_setup_reg): Likewise.
15322 (assign_parms_unsplit_complex): Likewise.
15323 * optabs.c (expand_binop): Likewise.
15324 * rtlanal.c (subreg_get_info): Likewise.
15325 * simplify-rtx.c (simplify_immed_subreg): Likewise.
15326 * varasm.c (output_constant_pool_2): Likewise.
15328 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15329 Alan Hayward <alan.hayward@arm.com>
15330 David Sherwood <david.sherwood@arm.com>
15332 * expmed.c (extract_high_half): Use scalar_int_mode and remove
15334 (expmed_mult_highpart_optab): Likewise.
15335 (expmed_mult_highpart): Likewise.
15337 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15338 Alan Hayward <alan.hayward@arm.com>
15339 David Sherwood <david.sherwood@arm.com>
15341 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
15342 instead of a machine_mode.
15343 (builtin_memset_read_str): Likewise.
15344 * builtins.c (c_readstr): Likewise.
15345 (builtin_memcpy_read_str): Likewise.
15346 (builtin_strncpy_read_str): Likewise.
15347 (builtin_memset_read_str): Likewise.
15348 (builtin_memset_gen_str): Likewise.
15349 (expand_builtin_signbit): Use scalar_int_mode for local variables.
15350 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
15351 instead of a machine_mode.
15352 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
15354 (make_extraction): Likewise.
15355 (try_widen_shift_mode): Take and return scalar_int_modes instead
15357 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
15358 a scalar_int_mode instead of a machine_mode.
15359 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
15360 (avr_addr_space_pointer_mode): Likewise.
15361 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
15362 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
15363 (msp430_unwind_word_mode): Likewise.
15364 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
15365 (spu_addr_space_pointer_mode): Likewise.
15366 (spu_addr_space_address_mode): Likewise.
15367 (spu_libgcc_cmp_return_mode): Likewise.
15368 (spu_libgcc_shift_count_mode): Likewise.
15369 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
15370 (rl78_addr_space_pointer_mode): Likewise.
15371 (fl78_unwind_word_mode): Likewise.
15372 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
15374 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
15375 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
15376 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
15377 (mips_valid_pointer_mode): Likewise.
15378 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
15379 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
15380 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
15382 (ft32_addr_space_address_mode): Likewise.
15383 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
15384 scalar_int_mode instead of a machine_mode.
15385 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
15387 (m32c_addr_space_address_mode): Likewise.
15388 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
15389 (rs6000_eh_return_filter_mode): Likewise.
15390 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
15391 (rs6000_eh_return_filter_mode): Likewise.
15392 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
15393 (s390_libgcc_shift_count_mode): Likewise.
15394 (s390_unwind_word_mode): Likewise.
15395 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
15397 * target.def (mode_rep_extended): Likewise.
15398 (valid_pointer_mode): Likewise.
15399 (addr_space.valid_pointer_mode): Likewise.
15400 (eh_return_filter_mode): Return a scalar_int_mode rather than
15402 (libgcc_cmp_return_mode): Likewise.
15403 (libgcc_shift_count_mode): Likewise.
15404 (unwind_word_mode): Likewise.
15405 (addr_space.pointer_mode): Likewise.
15406 (addr_space.address_mode): Likewise.
15407 * doc/tm.texi: Regenerate.
15408 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
15410 (do_jump): Use scalar_int_mode for local variables.
15411 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
15412 rather than a machine_mode.
15413 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
15414 (scompare_loc_descriptor_wide): Likewise.
15415 (scompare_loc_descriptor_narrow): Likewise.
15416 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
15418 * except.c (sjlj_emit_dispatch_table): Likewise.
15419 (expand_builtin_eh_copy_values): Likewise.
15420 * explow.c (convert_memory_address_addr_space_1): Likewise.
15421 Take a scalar_int_mode rather than a machine_mode.
15422 (convert_memory_address_addr_space): Take a scalar_int_mode rather
15423 than a machine_mode.
15424 (memory_address_addr_space): Use scalar_int_mode for local variables.
15425 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
15426 rather than a machine_mode.
15427 * expmed.c (mask_rtx): Likewise.
15428 (init_expmed_one_conv): Likewise.
15429 (expand_mult_highpart_adjust): Likewise.
15430 (extract_high_half): Likewise.
15431 (expmed_mult_highpart_optab): Likewise.
15432 (expmed_mult_highpart): Likewise.
15433 (expand_smod_pow2): Likewise.
15434 (expand_sdiv_pow2): Likewise.
15435 (emit_store_flag_int): Likewise.
15436 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
15438 (extract_low_bits): Likewise.
15439 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
15441 * expr.c (pieces_addr::adjust): Likewise.
15442 (can_store_by_pieces): Likewise.
15443 (store_by_pieces): Likewise.
15444 (clear_by_pieces_1): Likewise.
15445 (expand_expr_addr_expr_1): Likewise.
15446 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
15447 (expand_expr_real_1): Likewise.
15448 (try_casesi): Likewise.
15449 * final.c (shorten_branches): Likewise.
15450 * fold-const.c (fold_convert_const_int_from_fixed): Change the
15451 type of "mode" to machine_mode.
15452 * internal-fn.c (expand_arith_overflow_result_store): Take a
15453 scalar_int_mode rather than a machine_mode.
15454 (expand_mul_overflow): Use scalar_int_mode for local variables.
15455 * loop-doloop.c (doloop_modify): Likewise.
15456 (doloop_optimize): Likewise.
15457 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
15458 than a machine_mode.
15459 (expand_doubleword_shift_condmove): Likewise.
15460 (expand_doubleword_shift): Likewise.
15461 (expand_doubleword_clz): Likewise.
15462 (expand_doubleword_popcount): Likewise.
15463 (expand_doubleword_parity): Likewise.
15464 (expand_absneg_bit): Use scalar_int_mode for local variables.
15465 (prepare_float_lib_cmp): Likewise.
15466 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
15467 rather than a machine_mode.
15468 (convert_memory_address_addr_space): Likewise.
15469 (get_mode_bounds): Likewise.
15470 (get_address_mode): Return a scalar_int_mode rather than a
15472 * rtlanal.c (get_address_mode): Likewise.
15473 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
15474 than a machine_mode.
15475 * targhooks.c (default_mode_rep_extended): Likewise.
15476 (default_valid_pointer_mode): Likewise.
15477 (default_addr_space_valid_pointer_mode): Likewise.
15478 (default_eh_return_filter_mode): Return a scalar_int_mode rather
15479 than a machine_mode.
15480 (default_libgcc_cmp_return_mode): Likewise.
15481 (default_libgcc_shift_count_mode): Likewise.
15482 (default_unwind_word_mode): Likewise.
15483 (default_addr_space_pointer_mode): Likewise.
15484 (default_addr_space_address_mode): Likewise.
15485 * targhooks.h (default_eh_return_filter_mode): Likewise.
15486 (default_libgcc_cmp_return_mode): Likewise.
15487 (default_libgcc_shift_count_mode): Likewise.
15488 (default_unwind_word_mode): Likewise.
15489 (default_addr_space_pointer_mode): Likewise.
15490 (default_addr_space_address_mode): Likewise.
15491 (default_mode_rep_extended): Take a scalar_int_mode rather than
15493 (default_valid_pointer_mode): Likewise.
15494 (default_addr_space_valid_pointer_mode): Likewise.
15495 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
15497 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
15498 rather than a machine_mode.
15499 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
15500 for local variables.
15501 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15502 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
15503 than a machine_mode.
15505 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15506 Alan Hayward <alan.hayward@arm.com>
15507 David Sherwood <david.sherwood@arm.com>
15509 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
15510 the mode argument to scalar_int_mode.
15511 (do_jump_by_parts_zero_rtx): Likewise.
15512 (do_jump_by_parts_equality_rtx): Likewise.
15513 (do_jump_by_parts_greater): Take a mode argument.
15514 (do_jump_by_parts_equality): Likewise.
15515 (do_jump_1): Update calls accordingly.
15517 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15518 Alan Hayward <alan.hayward@arm.com>
15519 David Sherwood <david.sherwood@arm.com>
15521 * is-a.h (safe_dyn_cast): New function.
15522 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
15523 (jump_table_for_label): Likewise.
15524 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
15525 instead of an rtx_insn *.
15526 (shorten_branches): Use dyn_cast instead of LABEL_P and
15527 JUMP_TABLE_DATA_P. Use jump_table_for_label and
15528 rtx_jump_table_data::get_data_mode.
15529 (final_scan_insn): Likewise.
15531 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15532 Alan Hayward <alan.hayward@arm.com>
15533 David Sherwood <david.sherwood@arm.com>
15535 * combine.c (try_combine): Use is_a <scalar_int_mode> when
15536 trying to combine a full-register integer set with a subreg
15539 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15540 Alan Hayward <alan.hayward@arm.com>
15541 David Sherwood <david.sherwood@arm.com>
15543 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
15544 that is always either address_mode or pointer_mode.
15546 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15547 Alan Hayward <alan.hayward@arm.com>
15548 David Sherwood <david.sherwood@arm.com>
15550 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
15551 when the two are known to be equal.
15553 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15554 Alan Hayward <alan.hayward@arm.com>
15555 David Sherwood <david.sherwood@arm.com>
15557 * simplify-rtx.c (simplify_const_unary_operation): Use
15558 is_a <scalar_int_mode> instead of checking for a nonzero
15559 precision. Forcibly convert op_mode to a scalar_int_mode
15560 in that case. More clearly differentiate the operand and
15561 result modes and use the former when deciding what the value
15562 of a count-bits operation should be. Use is_int_mode instead
15563 of checking for a MODE_INT. Remove redundant check for whether
15564 this mode has a zero precision.
15566 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15567 Alan Hayward <alan.hayward@arm.com>
15568 David Sherwood <david.sherwood@arm.com>
15570 * optabs.c (widen_leading): Change the type of the mode argument
15571 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
15572 (widen_bswap): Likewise.
15573 (expand_parity): Likewise.
15574 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
15575 (expand_ffs): Likewise.
15576 (epand_unop): Check for scalar integer modes before calling the
15579 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15580 Alan Hayward <alan.hayward@arm.com>
15581 David Sherwood <david.sherwood@arm.com>
15583 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
15585 (expand_expr_real_1): Update call accordingly.
15587 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15588 Alan Hayward <alan.hayward@arm.com>
15589 David Sherwood <david.sherwood@arm.com>
15591 * expmed.c (store_bit_field_using_insv): Add op0_mode and
15592 value_mode arguments. Use scalar_int_mode internally.
15593 (store_bit_field_1): Rename the new integer mode from imode
15594 to op0_mode and use it instead of GET_MODE (op0). Update calls
15595 to store_split_bit_field, store_bit_field_using_insv and
15596 store_fixed_bit_field.
15597 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
15598 Use scalar_int_mode internally. Use a bit count rather than a mode
15599 when calculating the largest bit size for get_best_mode.
15600 Update calls to store_split_bit_field and store_fixed_bit_field_1.
15601 (store_fixed_bit_field_1): Add mode and value_mode arguments.
15602 Remove assertion that OP0 has a scalar integer mode.
15603 (store_split_bit_field): Add op0_mode and value_mode arguments.
15604 Update calls to extract_fixed_bit_field.
15605 (extract_bit_field_using_extv): Add an op0_mode argument.
15606 Use scalar_int_mode internally.
15607 (extract_bit_field_1): Rename the new integer mode from imode to
15608 op0_mode and use it instead of GET_MODE (op0). Update calls to
15609 extract_split_bit_field, extract_bit_field_using_extv and
15610 extract_fixed_bit_field.
15611 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
15612 to extract_split_bit_field and extract_fixed_bit_field_1.
15613 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
15614 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
15615 on the target mode.
15616 (extract_split_bit_field): Add an op0_mode argument. Update call
15617 to extract_fixed_bit_field.
15619 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15620 Alan Hayward <alan.hayward@arm.com>
15621 David Sherwood <david.sherwood@arm.com>
15623 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
15624 * explow.c (hard_function_value): Likewise.
15625 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
15626 convert_to_mode call outside the loop.
15627 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
15628 for the mode iterator. Require the mode specified by max_pieces
15630 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
15632 (copy_blkmode_to_reg): Likewise.
15633 (set_storage_via_setmem): Likewise.
15634 * optabs.c (prepare_cmp_insn): Likewise.
15635 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15636 * stor-layout.c (finish_bitfield_representative): Likewise.
15638 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15639 Alan Hayward <alan.hayward@arm.com>
15640 David Sherwood <david.sherwood@arm.com>
15642 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
15643 * expr.c (convert_move): Use them.
15644 (convert_modes): Likewise.
15645 (store_expr_with_bounds): Likewise.
15647 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15648 Alan Hayward <alan.hayward@arm.com>
15649 David Sherwood <david.sherwood@arm.com>
15651 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
15652 parameter for the mode of "x". Remove the "known_x", "known_mode"
15653 and "known_ret" arguments. Change the type of the mode argument
15654 to scalar_int_mode.
15655 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
15656 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
15657 (reg_num_sign_bit_copies_for_combine): Likewise.
15658 * rtlanal.c (nonzero_bits1): Likewise.
15659 (num_sign_bit_copies1): Likewise.
15660 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
15661 (reg_num_sign_bit_copies_general): Likewise.
15662 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
15663 (reg_nonzero_bits_general): Likewise.
15665 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15666 Alan Hayward <alan.hayward@arm.com>
15667 David Sherwood <david.sherwood@arm.com>
15669 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
15670 than in subroutines. Return 1 for non-integer modes.
15671 (cached_num_sign_bit_copies): Change the type of the mode parameter
15672 to scalar_int_mode.
15673 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
15674 classes. Handle CONST_INT_P first and then check whether X also
15675 has a scalar integer mode. Check the same thing for inner registers
15676 of a SUBREG and for values that are being extended or truncated.
15678 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15679 Alan Hayward <alan.hayward@arm.com>
15680 David Sherwood <david.sherwood@arm.com>
15682 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
15683 in subroutines. Return the mode mask for non-integer modes.
15684 (cached_nonzero_bits): Change the type of the mode parameter
15685 to scalar_int_mode.
15686 (nonzero_bits1): Likewise. Remove early exit for other mode
15687 classes. Handle CONST_INT_P first and then check whether X
15688 also has a scalar integer mode.
15690 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15691 Alan Hayward <alan.hayward@arm.com>
15692 David Sherwood <david.sherwood@arm.com>
15694 * expr.c (widest_int_mode_for_size): Make the comment match the code.
15695 Return a scalar_int_mode and assert that the size is greater than
15697 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
15698 (op_by_pieces_d::op_by_pieces_d): Likewise.
15699 (op_by_pieces_d::run): Likewise.
15700 (can_store_by_pieces): Likewise.
15702 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15703 Alan Hayward <alan.hayward@arm.com>
15704 David Sherwood <david.sherwood@arm.com>
15706 * combine.c (extract_left_shift): Add a mode argument and update
15708 (make_compound_operation_int): Change the type of the mode parameter
15709 to scalar_int_mode and update the call to extract_left_shift.
15711 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15712 Alan Hayward <alan.hayward@arm.com>
15713 David Sherwood <david.sherwood@arm.com>
15715 * combine.c (simplify_and_const_int): Change the type of the mode
15716 parameter to scalar_int_mode.
15717 (simplify_and_const_int_1): Likewise. Update recursive call.
15719 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15720 Alan Hayward <alan.hayward@arm.com>
15721 David Sherwood <david.sherwood@arm.com>
15723 * combine.c (simplify_compare_const): Check that the mode is a
15724 scalar_int_mode (rather than VOIDmode) before testing its
15726 (simplify_comparison): Move COMPARISON_P handling out of the
15727 loop and restrict the latter part of the loop to scalar_int_modes.
15728 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
15729 and when considering SUBREG_REGs. Use is_int_mode instead of
15730 checking GET_MODE_CLASS against MODE_INT.
15732 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15733 Alan Hayward <alan.hayward@arm.com>
15734 David Sherwood <david.sherwood@arm.com>
15736 * combine.c (try_widen_shift_mode): Move check for equal modes to...
15737 (simplify_shift_const_1): ...here. Use scalar_int_mode for
15738 shift_unit_mode and for modes involved in scalar shifts.
15740 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15741 Alan Hayward <alan.hayward@arm.com>
15742 David Sherwood <david.sherwood@arm.com>
15744 * combine.c (force_int_to_mode): New function, split out from...
15745 (force_to_mode): ...here. Keep xmode up-to-date and use it
15746 instead of GET_MODE (x).
15748 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15749 Alan Hayward <alan.hayward@arm.com>
15750 David Sherwood <david.sherwood@arm.com>
15752 * optabs-query.h (extraction_insn::struct_mode): Change type to
15753 opt_scalar_int_mode and update comment.
15754 (extraction_insn::field_mode): Change type to scalar_int_mode.
15755 (extraction_insn::pos_mode): Likewise.
15756 * combine.c (make_extraction): Update accordingly.
15757 * optabs-query.c (get_traditional_extraction_insn): Likewise.
15758 (get_optab_extraction_insn): Likewise.
15759 * recog.c (simplify_while_replacing): Likewise.
15760 * expmed.c (narrow_bit_field_mem): Change the type of the mode
15761 parameter to opt_scalar_int_mode.
15763 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15764 Alan Hayward <alan.hayward@arm.com>
15765 David Sherwood <david.sherwood@arm.com>
15767 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
15768 to a scalar_int_mode instead of a machine_mode.
15769 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
15770 (get_best_mode): Return a boolean and use a pointer argument to store
15771 the selected mode. Replace the limit mode parameter with a bit limit.
15772 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
15773 for the values returned by bit_field_mode_iterator::next_mode.
15774 (store_bit_field): Update call to get_best_mode.
15775 (store_fixed_bit_field): Likewise.
15776 (extract_fixed_bit_field): Likewise.
15777 * expr.c (optimize_bitfield_assignment_op): Likewise.
15778 * fold-const.c (optimize_bit_field_compare): Likewise.
15779 (fold_truth_andor_1): Likewise.
15780 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
15781 Update for new type of m_mode.
15782 (get_best_mode): As above.
15784 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15785 Alan Hayward <alan.hayward@arm.com>
15786 David Sherwood <david.sherwood@arm.com>
15788 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
15789 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
15790 (store_bit_field): Check is_a <scalar_int_mode> before calling
15791 strict_volatile_bitfield_p.
15792 (extract_bit_field): Likewise.
15794 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15795 Alan Hayward <alan.hayward@arm.com>
15796 David Sherwood <david.sherwood@arm.com>
15798 * target.def (cstore_mode): Return a scalar_int_mode.
15799 * doc/tm.texi: Regenerate.
15800 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
15801 * targhooks.h (default_cstore_mode): Likewise.
15802 * targhooks.c (default_cstore_mode): Likewise, using a forced
15804 * expmed.c (emit_cstore): Expect the target of the cstore to be
15807 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15808 Alan Hayward <alan.hayward@arm.com>
15809 David Sherwood <david.sherwood@arm.com>
15811 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
15813 (niter_desc): Likewise mode.
15814 (iv_analyze): Add a mode parameter.
15816 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
15817 and change its type to scalar_int_mode.
15818 * loop-iv.c: Update commentary at head of file.
15819 (iv_constant): Pass the mode paraeter before the rtx it describes
15820 and change its type to scalar_int_mode. Remove VOIDmode handling.
15821 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
15822 (iv_extend): Likewise.
15823 (shorten_into_mode): Likewise.
15824 (iv_add): Use scalar_int_mode.
15825 (iv_mult): Likewise.
15826 (iv_shift): Likewise.
15827 (canonicalize_iv_subregs): Likewise.
15828 (get_biv_step_1): Pass the outer_mode parameter before the rtx
15829 it describes and change its mode to scalar_int_mode. Also change
15830 the type of the returned inner_mode to scalar_int_mode.
15831 (get_biv_step): Likewise, turning outer_mode from a pointer
15832 into a direct parameter. Update call to get_biv_step_1.
15833 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
15834 iv_constant and get_biv_step.
15835 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
15836 and change its type to scalar_int_mode. Don't initialise iv->mode
15837 to VOIDmode and remove later checks for its still being VOIDmode.
15838 Update calls to iv_analyze_op and iv_analyze_expr. Check
15839 is_a <scalar_int_mode> when changing the mode under consideration.
15840 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
15841 Update call to iv_analyze_expr.
15842 (iv_analyze_op): Add a mode parameter. Reject subregs whose
15843 inner register is not also a scalar_int_mode. Update call to
15845 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
15846 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
15847 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
15848 separate mode class checks. Update calls to iv_analyze. Remove
15849 fix-up of VOIDmodes after iv_analyze_biv.
15850 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
15851 don't have a scalar_int_mode. Update call to biv_p.
15853 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15854 Alan Hayward <alan.hayward@arm.com>
15855 David Sherwood <david.sherwood@arm.com>
15857 * cfgexpand.c (convert_debug_memory_address): Use
15858 as_a <scalar_int_mode>.
15859 * combine.c (expand_compound_operation): Likewise.
15860 (make_extraction): Likewise.
15861 (change_zero_ext): Likewise.
15862 (simplify_comparison): Likewise.
15863 * cse.c (cse_insn): Likewise.
15864 * dwarf2out.c (minmax_loc_descriptor): Likewise.
15865 (mem_loc_descriptor): Likewise.
15866 (loc_descriptor): Likewise.
15867 * expmed.c (init_expmed_one_mode): Likewise.
15868 (synth_mult): Likewise.
15869 (emit_store_flag_1): Likewise.
15870 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
15871 of a comparison with size.
15872 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
15873 (reduce_to_bit_field_precision): Likewise.
15874 * function.c (expand_function_end): Likewise.
15875 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
15876 * loop-doloop.c (doloop_modify): Likewise.
15877 * optabs.c (expand_binop): Likewise.
15878 (expand_unop): Likewise.
15879 (expand_copysign_absneg): Likewise.
15880 (prepare_cmp_insn): Likewise.
15881 (maybe_legitimize_operand): Likewise.
15882 * recog.c (const_scalar_int_operand): Likewise.
15883 * rtlanal.c (get_address_mode): Likewise.
15884 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15885 (simplify_cond_clz_ctz): Likewise.
15886 * tree-nested.c (get_nl_goto_field): Likewise.
15887 * tree.c (build_vector_type_for_mode): Likewise.
15888 * var-tracking.c (use_narrower_mode): Likewise.
15890 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15891 Alan Hayward <alan.hayward@arm.com>
15892 David Sherwood <david.sherwood@arm.com>
15894 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
15895 * builtins.c (expand_builtin_signbit): Use it.
15896 * cfgexpand.c (expand_debug_expr): Likewise.
15897 * dojump.c (do_jump): Likewise.
15898 (do_compare_and_jump): Likewise.
15899 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
15900 * expmed.c (make_tree): Likewise.
15901 * expr.c (expand_expr_real_2): Likewise.
15902 (expand_expr_real_1): Likewise.
15903 (try_casesi): Likewise.
15904 * fold-const-call.c (fold_const_call_ss): Likewise.
15905 * fold-const.c (unextend): Likewise.
15906 (extract_muldiv_1): Likewise.
15907 (fold_single_bit_test): Likewise.
15908 (native_encode_int): Likewise.
15909 (native_encode_string): Likewise.
15910 (native_interpret_int): Likewise.
15911 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
15912 * internal-fn.c (expand_addsub_overflow): Likewise.
15913 (expand_neg_overflow): Likewise.
15914 (expand_mul_overflow): Likewise.
15915 (expand_arith_overflow): Likewise.
15916 * match.pd: Likewise.
15917 * stor-layout.c (layout_type): Likewise.
15918 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
15919 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
15920 * tree-ssanames.c (get_range_info): Likewise.
15921 * tree-switch-conversion.c (array_value_type) Likewise.
15922 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
15923 (vect_recog_divmod_pattern): Likewise.
15924 (vect_recog_mixed_size_cond_pattern): Likewise.
15925 * tree-vrp.c (extract_range_basic): Likewise.
15926 (simplify_float_conversion_using_ranges): Likewise.
15927 * tree.c (int_fits_type_p): Likewise.
15928 * ubsan.c (instrument_bool_enum_load): Likewise.
15929 * varasm.c (mergeable_string_section): Likewise.
15930 (narrowing_initializer_constant_valid_p): Likewise.
15931 (output_constant): Likewise.
15933 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15934 Alan Hayward <alan.hayward@arm.com>
15935 David Sherwood <david.sherwood@arm.com>
15937 * machmode.h (NARROWEST_INT_MODE): New macro.
15938 * expr.c (alignment_for_piecewise_move): Use it instead of
15939 GET_CLASS_NARROWEST_MODE (MODE_INT).
15940 (push_block): Likewise.
15941 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
15943 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15945 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15946 Alan Hayward <alan.hayward@arm.com>
15947 David Sherwood <david.sherwood@arm.com>
15949 * postreload.c (move2add_valid_value_p): Change the type of the
15950 mode parameter to scalar_int_mode.
15951 (move2add_use_add2_insn): Add a mode parameter and use it instead
15953 (move2add_use_add3_insn): Likewise.
15954 (reload_cse_move2add): Update accordingly.
15956 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15957 Alan Hayward <alan.hayward@arm.com>
15958 David Sherwood <david.sherwood@arm.com>
15960 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
15962 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
15963 * optabs.c (expand_unop): Likewise.
15965 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15966 Alan Hayward <alan.hayward@arm.com>
15967 David Sherwood <david.sherwood@arm.com>
15969 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
15970 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
15971 (popcount_loc_descriptor): Likewise.
15972 (bswap_loc_descriptor): Likewise.
15973 (rotate_loc_descriptor): Likewise.
15974 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
15975 calling the functions above.
15977 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15978 Alan Hayward <alan.hayward@arm.com>
15979 David Sherwood <david.sherwood@arm.com>
15981 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
15983 (try_combine): Likewise.
15984 (simplify_if_then_else): Likewise.
15985 * cse.c (cse_insn): Likewise.
15986 * dwarf2out.c (mem_loc_descriptor): Likewise.
15987 * emit-rtl.c (gen_lowpart_common): Likewise.
15988 * simplify-rtx.c (simplify_truncation): Likewise.
15989 (simplify_binary_operation_1): Likewise.
15990 (simplify_const_relational_operation): Likewise.
15991 (simplify_ternary_operation): Likewise.
15992 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
15994 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15995 Alan Hayward <alan.hayward@arm.com>
15996 David Sherwood <david.sherwood@arm.com>
15998 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
15999 * reload.c (push_reload): Likewise.
16000 (find_reloads): Likewise.
16002 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16003 Alan Hayward <alan.hayward@arm.com>
16004 David Sherwood <david.sherwood@arm.com>
16006 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
16007 (make_compound_operation_int): Likewise.
16008 (change_zero_ext): Likewise.
16009 * expr.c (convert_move): Likewise.
16010 (convert_modes): Likewise.
16011 * fwprop.c (forward_propagate_subreg): Likewise.
16012 * loop-iv.c (get_biv_step_1): Likewise.
16013 * optabs.c (widen_operand): Likewise.
16014 * postreload.c (move2add_valid_value_p): Likewise.
16015 * recog.c (simplify_while_replacing): Likewise.
16016 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16017 (simplify_binary_operation_1): Likewise. Remove redundant
16018 mode equality check.
16020 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16021 Alan Hayward <alan.hayward@arm.com>
16022 David Sherwood <david.sherwood@arm.com>
16024 * combine.c (combine_simplify_rtx): Add checks for
16025 is_a <scalar_int_mode>.
16026 (simplify_if_then_else): Likewise.
16027 (make_field_assignment): Likewise.
16028 (simplify_comparison): Likewise.
16029 * ifcvt.c (noce_try_bitop): Likewise.
16030 * loop-invariant.c (canonicalize_address_mult): Likewise.
16031 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16033 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16034 Alan Hayward <alan.hayward@arm.com>
16035 David Sherwood <david.sherwood@arm.com>
16037 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
16038 is_a <scalar_int_mode> instead of != BLKmode.
16040 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16041 Alan Hayward <alan.hayward@arm.com>
16042 David Sherwood <david.sherwood@arm.com>
16044 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
16045 instead of != VOIDmode.
16046 * combine.c (if_then_else_cond): Likewise.
16047 (change_zero_ext): Likewise.
16048 * dwarf2out.c (mem_loc_descriptor): Likewise.
16049 (loc_descriptor): Likewise.
16050 * rtlanal.c (canonicalize_condition): Likewise.
16051 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
16053 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16054 Alan Hayward <alan.hayward@arm.com>
16055 David Sherwood <david.sherwood@arm.com>
16057 * simplify-rtx.c (simplify_binary_operation_1): Use
16058 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
16060 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16061 Alan Hayward <alan.hayward@arm.com>
16062 David Sherwood <david.sherwood@arm.com>
16064 * wide-int.h (int_traits<unsigned char>) New class.
16065 (int_traits<unsigned short>) Likewise.
16066 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
16067 Use GET_MODE_UNIT_PRECISION and remove redundant test for
16069 * combine.c (set_nonzero_bits_and_sign_copies): Use
16070 is_a <scalar_int_mode>.
16071 (find_split_point): Likewise.
16072 (combine_simplify_rtx): Likewise.
16073 (simplify_logical): Likewise.
16074 (expand_compound_operation): Likewise.
16075 (expand_field_assignment): Likewise.
16076 (make_compound_operation): Likewise.
16077 (extended_count): Likewise.
16078 (change_zero_ext): Likewise.
16079 (simplify_comparison): Likewise.
16080 * dwarf2out.c (scompare_loc_descriptor): Likewise.
16081 (ucompare_loc_descriptor): Likewise.
16082 (minmax_loc_descriptor): Likewise.
16083 (mem_loc_descriptor): Likewise.
16084 (loc_descriptor): Likewise.
16085 * expmed.c (init_expmed_one_mode): Likewise.
16086 * lra-constraints.c (lra_constraint_offset): Likewise.
16087 * optabs.c (prepare_libcall_arg): Likewise.
16088 * postreload.c (move2add_note_store): Likewise.
16089 * reload.c (operands_match_p): Likewise.
16090 * rtl.h (load_extend_op): Likewise.
16091 * rtlhooks.c (gen_lowpart_general): Likewise.
16092 * simplify-rtx.c (simplify_truncation): Likewise.
16093 (simplify_unary_operation_1): Likewise.
16094 (simplify_binary_operation_1): Likewise.
16095 (simplify_const_binary_operation): Likewise.
16096 (simplify_const_relational_operation): Likewise.
16097 (simplify_subreg): Likewise.
16098 * stor-layout.c (bitwise_mode_for_mode): Likewise.
16099 * var-tracking.c (adjust_mems): Likewise.
16100 (prepare_call_arguments): Likewise.
16102 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16103 Alan Hayward <alan.hayward@arm.com>
16104 David Sherwood <david.sherwood@arm.com>
16106 * machmode.h (is_int_mode): New fuction.
16107 * combine.c (find_split_point): Use it.
16108 (combine_simplify_rtx): Likewise.
16109 (simplify_if_then_else): Likewise.
16110 (simplify_set): Likewise.
16111 (simplify_shift_const_1): Likewise.
16112 (simplify_comparison): Likewise.
16113 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
16114 * cse.c (notreg_cost): Likewise.
16115 (cse_insn): Likewise.
16116 * cselib.c (cselib_lookup_1): Likewise.
16117 * dojump.c (do_jump_1): Likewise.
16118 (do_compare_rtx_and_jump): Likewise.
16119 * dse.c (get_call_args): Likewise.
16120 * dwarf2out.c (rtl_for_decl_init): Likewise.
16121 (native_encode_initializer): Likewise.
16122 * expmed.c (emit_store_flag_1): Likewise.
16123 (emit_store_flag): Likewise.
16124 * expr.c (convert_modes): Likewise.
16125 (store_field): Likewise.
16126 (expand_expr_real_1): Likewise.
16127 * fold-const.c (fold_read_from_constant_string): Likewise.
16128 * gimple-ssa-sprintf.c (get_format_string): Likewise.
16129 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
16130 * optabs.c (expand_binop): Likewise.
16131 (expand_unop): Likewise.
16132 (expand_abs_nojump): Likewise.
16133 (expand_one_cmpl_abs_nojump): Likewise.
16134 * simplify-rtx.c (mode_signbit_p): Likewise.
16135 (val_signbit_p): Likewise.
16136 (val_signbit_known_set_p): Likewise.
16137 (val_signbit_known_clear_p): Likewise.
16138 (simplify_relational_operation_1): Likewise.
16139 * tree.c (vector_type_mode): Likewise.
16141 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16142 Alan Hayward <alan.hayward@arm.com>
16143 David Sherwood <david.sherwood@arm.com>
16145 * machmode.h (smallest_mode_for_size): Fix formatting.
16146 (smallest_int_mode_for_size): New function.
16147 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
16148 instead of smallest_mode_for_size.
16149 * combine.c (make_extraction): Likewise.
16150 * config/arc/arc.c (arc_expand_movmem): Likewise.
16151 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
16152 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
16153 * config/s390/s390.c (s390_expand_insv): Likewise.
16154 * config/sparc/sparc.c (assign_int_registers): Likewise.
16155 * config/spu/spu.c (spu_function_value): Likewise.
16156 (spu_function_arg): Likewise.
16157 * coverage.c (get_gcov_type): Likewise.
16158 (get_gcov_unsigned_t): Likewise.
16159 * dse.c (find_shift_sequence): Likewise.
16160 * expmed.c (store_bit_field_1): Likewise.
16161 * expr.c (convert_move): Likewise.
16162 (store_field): Likewise.
16163 * internal-fn.c (expand_arith_overflow): Likewise.
16164 * optabs-query.c (get_best_extraction_insn): Likewise.
16165 * optabs.c (expand_twoval_binop_libfunc): Likewise.
16166 * stor-layout.c (layout_type): Likewise.
16167 (initialize_sizetypes): Likewise.
16168 * targhooks.c (default_get_mask_mode): Likewise.
16169 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
16171 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16172 Alan Hayward <alan.hayward@arm.com>
16173 David Sherwood <david.sherwood@arm.com>
16175 * machmode.h (opt_mode::else_blk): New function.
16176 (int_mode_for_mode): Declare.
16177 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
16178 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
16180 * cfgexpand.c (expand_debug_expr): Likewise.
16181 * combine.c (gen_lowpart_or_truncate): Likewise.
16182 (gen_lowpart_for_combine): Likewise.
16183 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
16184 * config/avr/avr.c (avr_to_int_mode): Likewise.
16185 (avr_out_plus_1): Likewise.
16186 (avr_out_plus): Likewise.
16187 (avr_out_round): Likewise.
16188 * config/i386/i386.c (ix86_split_to_parts): Likewise.
16189 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
16190 (s390_expand_vcond): Likewise.
16191 * config/spu/spu.c (spu_split_immediate): Likewise.
16192 (spu_expand_mov): Likewise.
16193 * dse.c (get_stored_val): Likewise.
16194 * expmed.c (store_bit_field_1): Likewise.
16195 (convert_extracted_bit_field): Use int_mode_for_mode instead of
16197 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
16198 (extract_low_bits): Likewise.
16199 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
16200 handling rather than repeating the check.
16201 (emit_group_store): Likewise.
16202 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
16203 * optabs.c (expand_absneg_bit): Likewise.
16204 (expand_copysign_absneg): Likewise.
16205 (expand_copysign_bit): Likewise.
16206 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
16207 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
16208 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
16209 * var-tracking.c (prepare_call_arguments): Likewise.
16210 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
16211 int_mode_for_mode instead of mode_for_size.
16212 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
16214 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16215 Alan Hayward <alan.hayward@arm.com>
16216 David Sherwood <david.sherwood@arm.com>
16218 * machmode.h (int_mode_for_size): New function.
16219 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
16220 instead of mode_for_size.
16221 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
16223 * combine.c (expand_field_assignment): Use int_mode_for_size
16224 instead of mode_for_size.
16225 (make_extraction): Likewise.
16226 (simplify_shift_const_1): Likewise.
16227 (simplify_comparison): Likewise.
16228 * dojump.c (do_jump): Likewise.
16229 * dwarf2out.c (mem_loc_descriptor): Likewise.
16230 * emit-rtl.c (init_derived_machine_modes): Likewise.
16231 * expmed.c (flip_storage_order): Likewise.
16232 (convert_extracted_bit_field): Likewise.
16233 * expr.c (copy_blkmode_from_reg): Likewise.
16234 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
16235 * internal-fn.c (expand_mul_overflow): Likewise.
16236 * lower-subreg.c (simple_move): Likewise.
16237 * optabs-libfuncs.c (init_optabs): Likewise.
16238 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16239 * tree.c (vector_type_mode): Likewise.
16240 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
16241 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
16242 * tree-vect-generic.c (expand_vector_parallel): Likewise.
16243 * tree-vect-stmts.c (vectorizable_load): Likewise.
16244 (vectorizable_store): Likewise.
16246 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16247 Alan Hayward <alan.hayward@arm.com>
16248 David Sherwood <david.sherwood@arm.com>
16250 * coretypes.h (pod_mode): New type.
16251 (scalar_int_mode_pod): New typedef.
16252 * machmode.h (pod_mode): New class.
16253 (int_n_data_t::m): Change type to scalar_int_mode_pod.
16254 * genmodes.c (emit_mode_int_n): Update accordingly.
16255 * lower-subreg.h (target_lower_subreg): Change type to
16256 scalar_int_mode_pod.
16257 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
16258 scalar_int_mode_pod.
16260 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16261 Alan Hayward <alan.hayward@arm.com>
16262 David Sherwood <david.sherwood@arm.com>
16264 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
16265 machine_mode to scalar_int_mode.
16266 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
16267 (rs6000_option_override_internal): Remove cast to int.
16268 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
16269 machine_mode to scalar_int_mode.
16270 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
16271 (rs6000_option_override_internal): Remove cast to int.
16272 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
16273 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
16275 * config/s390/s390.c (s390_expand_builtin): Likewise.
16276 * coretypes.h (scalar_int_mode): New type.
16277 (opt_scalar_int_mode): New typedef.
16278 * machmode.h (scalar_int_mode): New class.
16279 (scalar_int_mode::includes_p): New function.
16280 (byte_mode): Change type to scalar_int_mode.
16281 (word_mode): Likewise.
16282 (ptr_mode): Likewise.
16283 * emit-rtl.c (byte_mode): Likewise.
16284 (word_mode): Likewise.
16285 (ptr_mode): Likewise.
16286 (init_derived_machine_modes): Update accordingly.
16287 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
16288 and MODE_PARTIAL_INT.
16289 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
16290 opt_scalar_int_mode.
16292 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16293 Alan Hayward <alan.hayward@arm.com>
16294 David Sherwood <david.sherwood@arm.com>
16296 * target.def (libgcc_floating_mode_supported_p): Take a
16298 * doc/tm.texi: Regenerate.
16299 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
16301 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
16302 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
16305 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16306 Alan Hayward <alan.hayward@arm.com>
16307 David Sherwood <david.sherwood@arm.com>
16309 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
16310 * doc/tm.texi: Regenerate.
16311 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
16312 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
16313 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
16314 * targhooks.h (default_floatn_mode): Likewise.
16315 * targhooks.c (default_floatn_mode): Likewise.
16316 * tree.c (build_common_tree_nodes): Update accordingly.
16318 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16319 Alan Hayward <alan.hayward@arm.com>
16320 David Sherwood <david.sherwood@arm.com>
16322 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
16323 (mode_iterator::iterate_p): Likewise.
16324 (mode_iterator::get_wider): Likewise.
16325 * expr.c (init_expr_target): Use opt_scalar_float_mode.
16327 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16328 Alan Hayward <alan.hayward@arm.com>
16329 David Sherwood <david.sherwood@arm.com>
16331 * coretypes.h (opt_scalar_float_mode): New typedef.
16332 * machmode.h (float_mode_for_size): New function.
16333 * emit-rtl.c (double_mode): Delete.
16334 (init_emit_once): Use float_mode_for_size.
16335 * stor-layout.c (layout_type): Likewise.
16336 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
16338 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16339 Alan Hayward <alan.hayward@arm.com>
16340 David Sherwood <david.sherwood@arm.com>
16342 * output.h (assemble_real): Take a scalar_float_mode.
16343 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
16344 * config/arm/arm.md (consttable_4): Likewise.
16345 (consttable_8): Likewise.
16346 (consttable_16): Likewise.
16347 * config/mips/mips.md (consttable_float): Likewise.
16348 * config/s390/s390.c (s390_output_pool_entry): Likewise.
16349 * varasm.c (assemble_real): Take a scalar_float_mode.
16350 (output_constant_pool_2): Update accordingly.
16351 (output_constant): Likewise.
16353 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16354 Alan Hayward <alan.hayward@arm.com>
16355 David Sherwood <david.sherwood@arm.com>
16357 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
16358 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
16359 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
16360 (native_encode_real): Likewise.
16361 (native_interpret_real): Likewise.
16362 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
16363 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16365 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16366 Alan Hayward <alan.hayward@arm.com>
16367 David Sherwood <david.sherwood@arm.com>
16369 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
16370 <scalar_float_mode>. Simplify.
16371 (gen_extend_conv_libfunc): Likewise.
16373 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16374 Alan Hayward <alan.hayward@arm.com>
16375 David Sherwood <david.sherwood@arm.com>
16377 * coretypes.h (scalar_float_mode): New type.
16378 * machmode.h (mode_traits::from_int): Use machine_mode if
16379 USE_ENUM_MODES is defined.
16380 (is_a): New function.
16382 (dyn_cast): Likewise.
16383 (scalar_float_mode): New class.
16384 (scalar_float_mode::includes_p): New function.
16385 (is_float_mode): Likewise.
16386 * gdbhooks.py (MachineModePrinter): New class.
16387 (build_pretty_printer): Use it for scalar_float_mode.
16388 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
16389 (format_helper::format_helper): Turn into a template.
16390 * genmodes.c (get_mode_class): New function.
16391 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
16392 or machine_mode if none.
16393 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
16394 as_a <scalar_float_mode>.
16395 * dwarf2out.c (mem_loc_descriptor): Likewise.
16396 (insert_float): Likewise.
16397 (add_const_value_attribute): Likewise.
16398 * simplify-rtx.c (simplify_immed_subreg): Likewise.
16399 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
16400 (expand_unop): Update accordingly.
16401 (expand_abs_nojump): Likewise.
16402 (expand_copysign_absneg): Take a scalar_float_mode.
16403 (expand_copysign_bit): Likewise.
16404 (expand_copysign): Update accordingly.
16406 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16407 Alan Hayward <alan.hayward@arm.com>
16408 David Sherwood <david.sherwood@arm.com>
16410 * coretypes.h (opt_mode): New class.
16411 * machmode.h (opt_mode): Likewise.
16412 (opt_mode::else_void): New function.
16413 (opt_mode::require): Likewise.
16414 (opt_mode::exists): Likewise.
16415 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
16416 (GET_MODE_2XWIDER_MODE): Likewise.
16417 (mode_iterator::get_wider): Update accordingly.
16418 (mode_iterator::get_2xwider): Likewise.
16419 (mode_iterator::get_known_wider): Likewise, turning into a template.
16420 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
16421 forcing a wider mode to exist.
16422 * config/cr16/cr16.h (LONG_REG_P): Likewise.
16423 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16424 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
16425 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
16426 * lower-subreg.c (init_lower_subreg): Likewise.
16427 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
16428 on the final iteration.
16429 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
16430 a wider mode exists before asking for a move pattern.
16431 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
16432 forcing a wider mode to exist.
16433 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
16434 returning false if no such mode exists.
16435 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
16436 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
16437 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
16438 Avoid checking for a MODE_INT if we already know the mode is not a
16440 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
16441 forcing a wider mode to exist.
16442 (expmed_mult_highpart_optab): Likewise.
16443 (expmed_mult_highpart): Likewise.
16444 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
16446 * lto-streamer-in.c (lto_input_mode_table): Likewise.
16447 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
16448 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16449 * internal-fn.c (expand_mul_overflow): Update use of
16450 GET_MODE_2XWIDER_MODE.
16451 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16452 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
16453 GET_MODE_WIDER_MODE.
16454 (convert_plusminus_to_widen): Likewise.
16455 * tree-switch-conversion.c (array_value_type): Likewise.
16456 * var-tracking.c (emit_note_insn_var_location): Likewise.
16457 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16458 Return false inside rather than outside the loop if no wider mode
16460 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
16461 and GET_MODE_2XWIDER_MODE
16462 (can_compare_p): Use else_void.
16463 * gdbhooks.py (OptMachineModePrinter): New class.
16464 (build_pretty_printer): Use it for opt_mode.
16466 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16467 Alan Hayward <alan.hayward@arm.com>
16468 David Sherwood <david.sherwood@arm.com>
16470 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
16471 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
16473 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16474 Alan Hayward <alan.hayward@arm.com>
16475 David Sherwood <david.sherwood@arm.com>
16477 * machmode.h (mode_traits): New structure.
16478 (get_narrowest_mode): New function.
16479 (mode_iterator::start): Likewise.
16480 (mode_iterator::iterate_p): Likewise.
16481 (mode_iterator::get_wider): Likewise.
16482 (mode_iterator::get_known_wider): Likewise.
16483 (mode_iterator::get_2xwider): Likewise.
16484 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
16485 (FOR_EACH_MODE): Likewise.
16486 (FOR_EACH_MODE_FROM): Likewise.
16487 (FOR_EACH_MODE_UNTIL): Likewise.
16488 (FOR_EACH_WIDER_MODE): Likewise.
16489 (FOR_EACH_2XWIDER_MODE): Likewise.
16490 * builtins.c (expand_builtin_strlen): Use new mode iterators.
16491 * combine.c (simplify_comparison): Likewise
16492 * config/i386/i386.c (type_natural_mode): Likewise.
16493 * cse.c (cse_insn): Likewise.
16494 * dse.c (find_shift_sequence): Likewise.
16495 * emit-rtl.c (init_derived_machine_modes): Likewise.
16496 (init_emit_once): Likewise.
16497 * explow.c (hard_function_value): Likewise.
16498 * expmed.c (extract_fixed_bit_field_1): Likewise.
16499 (extract_bit_field_1): Likewise.
16500 (expand_divmod): Likewise.
16501 (emit_store_flag_1): Likewise.
16502 * expr.c (init_expr_target): Likewise.
16503 (convert_move): Likewise.
16504 (alignment_for_piecewise_move): Likewise.
16505 (widest_int_mode_for_size): Likewise.
16506 (emit_block_move_via_movmem): Likewise.
16507 (copy_blkmode_to_reg): Likewise.
16508 (set_storage_via_setmem): Likewise.
16509 (compress_float_constant): Likewise.
16510 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16511 * optabs-query.c (get_best_extraction_insn): Likewise.
16512 * optabs.c (expand_binop): Likewise.
16513 (expand_twoval_unop): Likewise.
16514 (expand_twoval_binop): Likewise.
16515 (widen_leading): Likewise.
16516 (widen_bswap): Likewise.
16517 (expand_parity): Likewise.
16518 (expand_unop): Likewise.
16519 (prepare_cmp_insn): Likewise.
16520 (prepare_float_lib_cmp): Likewise.
16521 (expand_float): Likewise.
16522 (expand_fix): Likewise.
16523 (expand_sfix_optab): Likewise.
16524 * postreload.c (move2add_use_add2_insn): Likewise.
16525 * reg-stack.c (reg_to_stack): Likewise.
16526 * reginfo.c (choose_hard_reg_mode): Likewise.
16527 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16528 * stor-layout.c (mode_for_size): Likewise.
16529 (smallest_mode_for_size): Likewise.
16530 (mode_for_vector): Likewise.
16531 (finish_bitfield_representative): Likewise.
16532 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
16533 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
16534 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16535 * var-tracking.c (prepare_call_arguments): Likewise.
16537 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16538 Alan Hayward <alan.hayward@arm.com>
16539 David Sherwood <david.sherwood@arm.com>
16541 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
16542 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
16543 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
16544 * machmode.h (mode_size): Move earlier in file.
16545 (mode_precision): Likewise.
16546 (mode_inner): Likewise.
16547 (mode_nunits): Likewise.
16548 (mode_unit_size): Likewise.
16549 (unit_unit_precision): Likewise.
16550 (mode_wider): Likewise.
16551 (mode_2xwider): Likewise.
16552 (machine_mode): New class.
16553 (mode_to_bytes): New function.
16554 (mode_to_bits): Likewise.
16555 (mode_to_precision): Likewise.
16556 (mode_to_inner): Likewise.
16557 (mode_to_unit_size): Likewise.
16558 (mode_to_unit_precision): Likewise.
16559 (mode_to_nunits): Likewise.
16560 (GET_MODE_SIZE): Use mode_to_bytes.
16561 (GET_MODE_BITSIZE): Use mode_to_bits.
16562 (GET_MODE_PRECISION): Use mode_to_precision.
16563 (GET_MODE_INNER): Use mode_to_inner.
16564 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
16565 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
16566 (GET_MODE_NUNITS): Use mode_to_nunits.
16567 * system.h (ALWAYS_INLINE): New macro.
16568 * config/powerpcspe/powerpcspe-c.c
16569 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
16570 int for arg1_mode and arg2_mode.
16572 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16573 Alan Hayward <alan.hayward@arm.com>
16574 David Sherwood <david.sherwood@arm.com>
16576 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
16577 Prefix mode names with E_ in case statements.
16578 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16579 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
16580 (aarch64_split_simd_move): Likewise.
16581 (aarch64_gen_storewb_pair): Likewise.
16582 (aarch64_gen_loadwb_pair): Likewise.
16583 (aarch64_gen_store_pair): Likewise.
16584 (aarch64_gen_load_pair): Likewise.
16585 (aarch64_get_condition_code_1): Likewise.
16586 (aarch64_constant_pool_reload_icode): Likewise.
16587 (get_rsqrte_type): Likewise.
16588 (get_rsqrts_type): Likewise.
16589 (get_recpe_type): Likewise.
16590 (get_recps_type): Likewise.
16591 (aarch64_gimplify_va_arg_expr): Likewise.
16592 (aarch64_simd_container_mode): Likewise.
16593 (aarch64_emit_load_exclusive): Likewise.
16594 (aarch64_emit_store_exclusive): Likewise.
16595 (aarch64_expand_compare_and_swap): Likewise.
16596 (aarch64_gen_atomic_cas): Likewise.
16597 (aarch64_emit_bic): Likewise.
16598 (aarch64_emit_atomic_swap): Likewise.
16599 (aarch64_emit_atomic_load_op): Likewise.
16600 (aarch64_evpc_trn): Likewise.
16601 (aarch64_evpc_uzp): Likewise.
16602 (aarch64_evpc_zip): Likewise.
16603 (aarch64_evpc_ext): Likewise.
16604 (aarch64_evpc_rev): Likewise.
16605 (aarch64_evpc_dup): Likewise.
16606 (aarch64_gen_ccmp_first): Likewise.
16607 (aarch64_gen_ccmp_next): Likewise.
16608 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
16609 (alpha_emit_xfloating_libcall): Likewise.
16610 (emit_insxl): Likewise.
16611 (alpha_arg_type): Likewise.
16612 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
16613 (arc_preferred_simd_mode): Likewise.
16614 (arc_secondary_reload): Likewise.
16615 (get_arc_condition_code): Likewise.
16616 (arc_print_operand): Likewise.
16617 (arc_legitimate_constant_p): Likewise.
16618 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16619 * config/arc/arc.md (casesi_load): Likewise.
16620 (casesi_compact_jump): Likewise.
16621 * config/arc/predicates.md (proper_comparison_operator): Likewise.
16622 (cc_use_register): Likewise.
16623 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16624 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
16625 (arm_init_iwmmxt_builtins): Likewise.
16626 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
16627 (neon_expand_vector_init): Likewise.
16628 (arm_attr_length_move_neon): Likewise.
16629 (maybe_get_arm_condition_code): Likewise.
16630 (arm_emit_vector_const): Likewise.
16631 (arm_preferred_simd_mode): Likewise.
16632 (arm_output_iwmmxt_tinsr): Likewise.
16633 (thumb1_output_casesi): Likewise.
16634 (thumb2_output_casesi): Likewise.
16635 (arm_emit_load_exclusive): Likewise.
16636 (arm_emit_store_exclusive): Likewise.
16637 (arm_expand_compare_and_swap): Likewise.
16638 (arm_evpc_neon_vuzp): Likewise.
16639 (arm_evpc_neon_vzip): Likewise.
16640 (arm_evpc_neon_vrev): Likewise.
16641 (arm_evpc_neon_vtrn): Likewise.
16642 (arm_evpc_neon_vext): Likewise.
16643 (arm_validize_comparison): Likewise.
16644 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
16645 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
16646 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
16647 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
16648 (c6x_preferred_simd_mode): Likewise.
16649 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
16650 (epiphany_rtx_costs): Likewise.
16651 * config/epiphany/predicates.md (proper_comparison_operator):
16653 * config/frv/frv.c (condexec_memory_operand): Likewise.
16654 (frv_emit_move): Likewise.
16655 (output_move_single): Likewise.
16656 (output_condmove_single): Likewise.
16657 (frv_hard_regno_mode_ok): Likewise.
16658 (frv_matching_accg_mode): Likewise.
16659 * config/h8300/h8300.c (split_adds_subs): Likewise.
16660 (h8300_rtx_costs): Likewise.
16661 (h8300_print_operand): Likewise.
16662 (compute_mov_length): Likewise.
16663 (output_logical_op): Likewise.
16664 (compute_logical_op_length): Likewise.
16665 (compute_logical_op_cc): Likewise.
16666 (h8300_shift_needs_scratch_p): Likewise.
16667 (output_a_shift): Likewise.
16668 (compute_a_shift_length): Likewise.
16669 (compute_a_shift_cc): Likewise.
16670 (expand_a_rotate): Likewise.
16671 (output_a_rotate): Likewise.
16672 * config/i386/i386.c (classify_argument): Likewise.
16673 (function_arg_advance_32): Likewise.
16674 (function_arg_32): Likewise.
16675 (function_arg_64): Likewise.
16676 (function_value_64): Likewise.
16677 (ix86_gimplify_va_arg): Likewise.
16678 (ix86_legitimate_constant_p): Likewise.
16679 (put_condition_code): Likewise.
16680 (split_double_mode): Likewise.
16681 (ix86_avx256_split_vector_move_misalign): Likewise.
16682 (ix86_expand_vector_logical_operator): Likewise.
16683 (ix86_split_idivmod): Likewise.
16684 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
16685 (ix86_build_const_vector): Likewise.
16686 (ix86_build_signbit_mask): Likewise.
16687 (ix86_match_ccmode): Likewise.
16688 (ix86_cc_modes_compatible): Likewise.
16689 (ix86_expand_branch): Likewise.
16690 (ix86_expand_sse_cmp): Likewise.
16691 (ix86_expand_sse_movcc): Likewise.
16692 (ix86_expand_int_sse_cmp): Likewise.
16693 (ix86_expand_vec_perm_vpermi2): Likewise.
16694 (ix86_expand_vec_perm): Likewise.
16695 (ix86_expand_sse_unpack): Likewise.
16696 (ix86_expand_int_addcc): Likewise.
16697 (ix86_split_to_parts): Likewise.
16698 (ix86_vectorize_builtin_gather): Likewise.
16699 (ix86_vectorize_builtin_scatter): Likewise.
16700 (avx_vpermilp_parallel): Likewise.
16701 (inline_memory_move_cost): Likewise.
16702 (ix86_tieable_integer_mode_p): Likewise.
16703 (x86_maybe_negate_const_int): Likewise.
16704 (ix86_expand_vector_init_duplicate): Likewise.
16705 (ix86_expand_vector_init_one_nonzero): Likewise.
16706 (ix86_expand_vector_init_one_var): Likewise.
16707 (ix86_expand_vector_init_concat): Likewise.
16708 (ix86_expand_vector_init_interleave): Likewise.
16709 (ix86_expand_vector_init_general): Likewise.
16710 (ix86_expand_vector_set): Likewise.
16711 (ix86_expand_vector_extract): Likewise.
16712 (emit_reduc_half): Likewise.
16713 (ix86_emit_i387_round): Likewise.
16714 (ix86_mangle_type): Likewise.
16715 (ix86_expand_round_sse4): Likewise.
16716 (expand_vec_perm_blend): Likewise.
16717 (canonicalize_vector_int_perm): Likewise.
16718 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
16719 (expand_vec_perm_1): Likewise.
16720 (expand_vec_perm_interleave3): Likewise.
16721 (expand_vec_perm_even_odd_pack): Likewise.
16722 (expand_vec_perm_even_odd_1): Likewise.
16723 (expand_vec_perm_broadcast_1): Likewise.
16724 (ix86_vectorize_vec_perm_const_ok): Likewise.
16725 (ix86_expand_vecop_qihi): Likewise.
16726 (ix86_expand_mul_widen_hilo): Likewise.
16727 (ix86_expand_sse2_abs): Likewise.
16728 (ix86_expand_pextr): Likewise.
16729 (ix86_expand_pinsr): Likewise.
16730 (ix86_preferred_simd_mode): Likewise.
16731 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
16732 * config/i386/sse.md (*andnot<mode>3): Likewise.
16733 (<mask_codefor><code><mode>3<mask_name>): Likewise.
16734 (*<code><mode>3): Likewise.
16735 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
16736 (ia64_expand_atomic_op): Likewise.
16737 (ia64_arg_type): Likewise.
16738 (ia64_mode_to_int): Likewise.
16739 (ia64_scalar_mode_supported_p): Likewise.
16740 (ia64_vector_mode_supported_p): Likewise.
16741 (expand_vec_perm_broadcast): Likewise.
16742 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
16743 (iq2000_function_arg_advance): Likewise.
16744 (iq2000_function_arg): Likewise.
16745 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
16746 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
16747 (m68k_libcall_value): Likewise.
16748 (m68k_function_value): Likewise.
16749 (sched_attr_op_type): Likewise.
16750 * config/mcore/mcore.c (mcore_output_move): Likewise.
16751 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
16753 (microblaze_function_arg): Likewise.
16754 * config/mips/mips.c (mips16_build_call_stub): Likewise.
16755 (mips_print_operand): Likewise.
16756 (mips_mode_ok_for_mov_fmt_p): Likewise.
16757 (mips_vector_mode_supported_p): Likewise.
16758 (mips_preferred_simd_mode): Likewise.
16759 (mips_expand_vpc_loongson_even_odd): Likewise.
16760 (mips_expand_vec_unpack): Likewise.
16761 (mips_expand_vi_broadcast): Likewise.
16762 (mips_expand_vector_init): Likewise.
16763 (mips_expand_vec_reduc): Likewise.
16764 (mips_expand_msa_cmp): Likewise.
16765 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
16766 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
16767 (cc_flags_for_mode): Likewise.
16768 * config/msp430/msp430.c (msp430_print_operand): Likewise.
16769 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
16770 (nds32_output_casesi_pc_relative): Likewise.
16771 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16772 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
16773 (nvptx_gen_unpack): Likewise.
16774 (nvptx_gen_pack): Likewise.
16775 (nvptx_gen_shuffle): Likewise.
16776 (nvptx_gen_wcast): Likewise.
16777 (nvptx_preferred_simd_mode): Likewise.
16778 * config/pa/pa.c (pa_secondary_reload): Likewise.
16779 * config/pa/predicates.md (base14_operand): Likewise.
16780 * config/powerpcspe/powerpcspe-c.c
16781 (altivec_resolve_overloaded_builtin): Likewise.
16782 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
16784 (rs6000_preferred_simd_mode): Likewise.
16785 (output_vec_const_move): Likewise.
16786 (rs6000_expand_vector_extract): Likewise.
16787 (rs6000_split_vec_extract_var): Likewise.
16788 (reg_offset_addressing_ok_p): Likewise.
16789 (rs6000_legitimate_offset_address_p): Likewise.
16790 (rs6000_legitimize_address): Likewise.
16791 (rs6000_emit_set_const): Likewise.
16792 (rs6000_const_vec): Likewise.
16793 (rs6000_emit_move): Likewise.
16794 (spe_build_register_parallel): Likewise.
16795 (rs6000_darwin64_record_arg_recurse): Likewise.
16796 (swap_selector_for_mode): Likewise.
16797 (spe_init_builtins): Likewise.
16798 (paired_init_builtins): Likewise.
16799 (altivec_init_builtins): Likewise.
16800 (do_load_for_compare): Likewise.
16801 (rs6000_generate_compare): Likewise.
16802 (rs6000_expand_float128_convert): Likewise.
16803 (emit_load_locked): Likewise.
16804 (emit_store_conditional): Likewise.
16805 (rs6000_output_function_epilogue): Likewise.
16806 (rs6000_handle_altivec_attribute): Likewise.
16807 (rs6000_function_value): Likewise.
16808 (emit_fusion_gpr_load): Likewise.
16809 (emit_fusion_p9_load): Likewise.
16810 (emit_fusion_p9_store): Likewise.
16811 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
16812 (fusion_gpr_mem_load): Likewise.
16813 (fusion_addis_mem_combo_load): Likewise.
16814 (fusion_addis_mem_combo_store): Likewise.
16815 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
16816 (fusion_gpr_mem_load): Likewise.
16817 (fusion_addis_mem_combo_load): Likewise.
16818 (fusion_addis_mem_combo_store): Likewise.
16819 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16821 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
16822 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16823 (rs6000_preferred_simd_mode): Likewise.
16824 (output_vec_const_move): Likewise.
16825 (rs6000_expand_vector_extract): Likewise.
16826 (rs6000_split_vec_extract_var): Likewise.
16827 (reg_offset_addressing_ok_p): Likewise.
16828 (rs6000_legitimate_offset_address_p): Likewise.
16829 (rs6000_legitimize_address): Likewise.
16830 (rs6000_emit_set_const): Likewise.
16831 (rs6000_const_vec): Likewise.
16832 (rs6000_emit_move): Likewise.
16833 (rs6000_darwin64_record_arg_recurse): Likewise.
16834 (swap_selector_for_mode): Likewise.
16835 (paired_init_builtins): Likewise.
16836 (altivec_init_builtins): Likewise.
16837 (rs6000_expand_float128_convert): Likewise.
16838 (emit_load_locked): Likewise.
16839 (emit_store_conditional): Likewise.
16840 (rs6000_output_function_epilogue): Likewise.
16841 (rs6000_handle_altivec_attribute): Likewise.
16842 (rs6000_function_value): Likewise.
16843 (emit_fusion_gpr_load): Likewise.
16844 (emit_fusion_p9_load): Likewise.
16845 (emit_fusion_p9_store): Likewise.
16846 * config/rx/rx.c (rx_gen_move_template): Likewise.
16847 (flags_from_mode): Likewise.
16848 * config/s390/predicates.md (s390_alc_comparison): Likewise.
16849 (s390_slb_comparison): Likewise.
16850 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
16851 (s390_vector_mode_supported_p): Likewise.
16852 (s390_cc_modes_compatible): Likewise.
16853 (s390_match_ccmode_set): Likewise.
16854 (s390_canonicalize_comparison): Likewise.
16855 (s390_emit_compare_and_swap): Likewise.
16856 (s390_branch_condition_mask): Likewise.
16857 (s390_rtx_costs): Likewise.
16858 (s390_secondary_reload): Likewise.
16859 (__SECONDARY_RELOAD_CASE): Likewise.
16860 (s390_expand_cs): Likewise.
16861 (s390_preferred_simd_mode): Likewise.
16862 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
16863 * config/sh/sh.c (sh_print_operand): Likewise.
16864 (dump_table): Likewise.
16865 (sh_secondary_reload): Likewise.
16866 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16867 * config/sh/sh.md (casesi_worker_1): Likewise.
16868 (casesi_worker_2): Likewise.
16869 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
16870 (fcc_comparison_operator): Likewise.
16871 * config/sparc/sparc.c (sparc_expand_move): Likewise.
16872 (emit_soft_tfmode_cvt): Likewise.
16873 (sparc_preferred_simd_mode): Likewise.
16874 (output_cbranch): Likewise.
16875 (sparc_print_operand): Likewise.
16876 (sparc_expand_vec_perm_bmask): Likewise.
16877 (vector_init_bshuffle): Likewise.
16878 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
16879 (spu_vector_mode_supported_p): Likewise.
16880 (spu_expand_insv): Likewise.
16881 (spu_emit_branch_or_set): Likewise.
16882 (spu_handle_vector_attribute): Likewise.
16883 (spu_builtin_splats): Likewise.
16884 (spu_builtin_extract): Likewise.
16885 (spu_builtin_promote): Likewise.
16886 (spu_expand_sign_extend): Likewise.
16887 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
16888 (tilegx_simd_int): Likewise.
16889 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
16890 (tilepro_simd_int): Likewise.
16891 * config/v850/v850.c (const_double_split): Likewise.
16892 (v850_print_operand): Likewise.
16893 (ep_memory_offset): Likewise.
16894 * config/vax/vax.c (vax_rtx_costs): Likewise.
16895 (vax_output_int_move): Likewise.
16896 (vax_output_int_add): Likewise.
16897 (vax_output_int_subtract): Likewise.
16898 * config/visium/predicates.md (visium_branch_operator): Likewise.
16899 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
16900 (visium_print_operand_address): Likewise.
16901 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16902 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
16903 (xtensa_expand_conditional_branch): Likewise.
16904 (xtensa_copy_incoming_a7): Likewise.
16905 (xtensa_output_literal): Likewise.
16906 * dfp.c (decimal_real_maxval): Likewise.
16907 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
16909 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16910 Alan Hayward <alan.hayward@arm.com>
16911 David Sherwood <david.sherwood@arm.com>
16913 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
16914 (mode_nunits_inline): Likewise.
16915 (mode_inner_inline): Likewise.
16916 (mode_unit_size_inline): Likewise.
16917 (mode_unit_precision_inline): Likewise.
16918 (emit_insn_modes_h): Likewise. Also emit a #define of the
16920 (emit_mode_wider): Add an E_ prefix to mode names.
16921 (emit_mode_complex): Likewise.
16922 (emit_mode_inner): Likewise.
16923 (emit_mode_adjustments): Likewise.
16924 (emit_mode_int_n): Likewise.
16925 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
16926 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
16927 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
16928 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
16929 (CRC32_BUILTIN, ENTRY): Likewise.
16930 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
16931 (aarch64_pop_regs): Likewise.
16932 (aarch64_process_components): Likewise.
16933 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
16934 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
16935 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
16936 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
16937 * config/arm/arm.c (arm_init_libfuncs): Likewise.
16938 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
16940 * config/i386/i386-builtin.def (pcmpestr): Likewise.
16941 (pcmpistr): Likewise.
16942 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
16943 * config/mmix/mmix.c (mmix_output_condition): Likewise.
16944 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
16946 * config/rl78/rl78.c (mduc_regs): Likewise.
16947 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16948 (htm_expand_builtin): Likewise.
16949 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
16950 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
16951 * config/xtensa/xtensa.c (print_operand): Likewise.
16952 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
16953 (NUM_MODE_VECTOR_INT): Likewise.
16954 * genoutput.c (null_operand): Likewise.
16955 (output_operand_data): Likewise.
16956 * genrecog.c (print_parameter_value): Likewise.
16957 * lra.c (debug_operand_data): Likewise.
16959 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16961 * dwarf2out.c (scompare_loc_descriptor_wide)
16962 (scompare_loc_descriptor_narrow): New functions, split out from...
16963 (scompare_loc_descriptor): ...here.
16964 * expmed.c (emit_store_flag_int): New function, split out from...
16965 (emit_store_flag): ...here.
16967 2017-08-30 Richard Biener <rguenther@suse.de>
16969 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
16970 (dwarf2out_early_finish): Move setting of AT_pubnames from
16971 early debug output to early finish.
16973 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16975 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
16976 and -mdata-region to the assembler.
16978 2017-08-30 Richard Biener <rguenther@suse.de>
16980 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
16982 (gen_subprogram_die): Add DW_AT_object_pointer only early.
16983 (dwarf2out_early_global_decl): Only generate a DIE for the
16984 abstract origin if it doesn't already exist or is a declaration DIE.
16985 (resolve_addr): Do not add the linkage name twice when
16986 generating a stub DIE for the DW_TAG_GNU_call_site target.
16988 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16990 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16991 Use machine_mode rather than int for arg1_mode.
16993 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
16996 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
16997 that the second argument of the built-in functions to unpack
16998 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
16999 switch statement instead a lot of if statements.
17000 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
17001 Allow 64-bit values to be in Altivec registers as well as
17002 traditional floating point registers.
17003 (pack<mode>, FMOVE128_VSX iterator): Likewise.
17005 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
17007 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
17008 MAX_REGS_PER_ADDRESS == 1.
17010 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
17012 * config/i386/i386.opt (flag_fentry): Do not init to -1.
17013 * config/i386/i386.c (ix86_option_override_internal): Simplify
17014 setting of opts->x_flag_entry.
17016 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17017 Jakub Jelinek <jakub@redhat.com>
17018 Richard Biener <rguenther@suse.de>
17020 PR tree-optimization/81503
17021 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
17022 folded constant fits in the target type; reorder tests for clarity.
17024 2017-08-29 Martin Liska <mliska@suse.cz>
17026 * passes.def: Include pass_lower_switch.
17027 * stmt.c (dump_case_nodes): Remove and move to
17028 tree-switch-conversion.
17029 (case_values_threshold): Likewise.
17030 (expand_switch_as_decision_tree_p): Likewise.
17031 (emit_case_decision_tree): Likewise.
17032 (expand_case): Likewise.
17033 (balance_case_nodes): Likewise.
17034 (node_has_low_bound): Likewise.
17035 (node_has_high_bound): Likewise.
17036 (node_is_bounded): Likewise.
17037 (emit_case_nodes): Likewise.
17038 (struct simple_case_node): New struct.
17039 (add_case_node): Remove.
17040 (emit_case_dispatch_table): Use vector instead of case_list.
17041 (reset_out_edges_aux): Remove.
17042 (compute_cases_per_edge): Likewise.
17043 (expand_case): Build list of simple_case_node.
17044 (expand_sjlj_dispatch_table): Use it.
17045 * tree-switch-conversion.c (struct case_node): Moved from
17046 stmt.c and adjusted.
17047 (emit_case_nodes): Likewise.
17048 (node_has_low_bound): Likewise.
17049 (node_has_high_bound): Likewise.
17050 (node_is_bounded): Likewise.
17051 (case_values_threshold): Likewise.
17052 (reset_out_edges_aux): Likewise.
17053 (compute_cases_per_edge): Likewise.
17054 (add_case_node): Likewise.
17055 (dump_case_nodes): Likewise.
17056 (balance_case_nodes): Likewise.
17057 (expand_switch_as_decision_tree_p): Likewise.
17058 (emit_jump): Likewise.
17059 (emit_case_decision_tree): Likewise.
17060 (try_switch_expansion): Likewise.
17061 (do_jump_if_equal): Likewise.
17062 (emit_cmp_and_jump_insns): Likewise.
17063 (fix_phi_operands_for_edge): New function.
17064 (record_phi_operand_mapping): Likewise.
17065 (class pass_lower_switch): New pass.
17066 (pass_lower_switch::execute): New function.
17067 (make_pass_lower_switch): Likewise.
17068 (conditional_probability):
17069 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
17070 * tree-pass.h: Add make_pass_lower_switch.
17072 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17075 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
17078 2017-08-29 Richard Biener <rguenther@suse.de>
17080 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
17081 we do not add a DW_AT_inline attribute twice.
17082 (gen_subprogram_die): Remove code setting DW_AT_inline on
17083 DECL_ABSTRACT_P nodes.
17085 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
17087 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
17088 calls to internal functions.
17089 (gimplify_modify_expr): Likewise.
17090 * tree-call-cdce.c (use_internal_fn): Likewise.
17091 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
17092 (convert_to_divmod): Set the nothrow flag.
17093 * tree-if-conv.c (predicate_mem_writes): Likewise.
17094 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
17095 (vectorizable_call): Likewise.
17096 (vectorizable_store): Likewise.
17097 (vectorizable_load): Likewise.
17098 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
17099 (vect_recog_mask_conversion_pattern): Likewise.
17101 2017-08-29 Martin Liska <mliska@suse.cz>
17104 * gcc.c (driver_handle_option): Add new argument.
17105 * opts-common.c (handle_option): Pass
17106 target_option_override_hook.
17107 * opts-global.c (lang_handle_option): Add new option.
17108 (set_default_handlers): Add new argument.
17109 (decode_options): Likewise.
17110 * opts.c (target_handle_option): Likewise.
17111 (common_handle_option): Call target_option_override_hook.
17112 * opts.h (struct cl_option_handler_func): Add hook for
17113 target option override.
17114 (struct cl_option_handlers): Likewise.
17115 (set_default_handlers): Add new argument.
17116 (decode_options): Likewise.
17117 (common_handle_option): Likewise.
17118 (target_handle_option): Likewise.
17119 * toplev.c (toplev::main): Pass targetm.target_option.override
17122 2017-08-29 Richard Biener <rguenther@suse.de>
17123 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
17125 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
17126 life to the active subtree.
17128 2017-08-28 Jeff Law <law@redhat.com>
17130 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
17131 derive_equivalences.
17132 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
17133 Code moved into....
17134 (edge_info::derive_equivalences): New private member function
17136 * tree-ssa-dom.c (class edge_info): Changed from a struct
17137 to a class. Add ctor/dtor, methods and data members.
17138 (edge_info::edge_info): Renamed from allocate_edge_info.
17139 Initialize additional members.
17140 (edge_info::~edge_info): New.
17141 (free_dom_edge_info): Delete the edge info.
17142 (record_edge_info): Use new class & associated member functions.
17143 Tighten forms for testing for edge equivalences.
17144 (record_temporary_equivalences): Iterate over the simple
17145 equivalences rather than assuming there's only one per edge.
17146 (cprop_into_successor_phis): Iterate over the simple
17147 equivalences rather than assuming there's only one per edge.
17148 (optimize_stmt): Use operand_equal_p rather than pointer
17149 equality for mini-DSE code.
17151 2017-08-28 Nathan Sidwell <nathan@acm.org>
17153 * gcc.c (execute): Fold SIGPIPE handling into switch
17154 statement. Adjust internal error message.
17156 2017-08-28 Richard Biener <rguenther@suse.de>
17159 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
17160 Do nothing for removed DIEs.
17162 2017-08-28 Richard Biener <rguenther@suse.de>
17164 PR tree-optimization/81977
17165 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
17168 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
17171 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
17172 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
17173 using targetm.gen_mem_thread_fence.
17175 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
17178 * config/i386/i386.md (*<btsc><mode>): Change operand 2
17179 predicate to register_operand. Reorder operands.
17180 (*btr<mode>): Ditto.
17181 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
17182 (*btr<mode>_mask): Ditto.
17184 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
17186 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
17187 * config/rs6000/xmmintrin.h: New file.
17188 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
17190 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17193 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
17194 parameter and_insn and return it.
17195 (recombine_lvx_pattern): Insert a copy to ensure availability of
17196 the base register of the copied masking operation at the point of
17197 the instruction replacement.
17198 (recombine_stvx_pattern): Likewise.
17200 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
17202 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
17203 undocumented switches.
17204 (-mpower9-dform-vector): Likewise.
17205 (-mpower9-dform): Likewise.
17206 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
17207 comments to delete references to -mpower9-dform* switches.
17208 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
17209 Delete reference to -mpower9-dform* switches, test for
17210 -mpower9-vector instead.
17211 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
17212 (OTHER_P9_VECTOR_MASKS): Likewise.
17213 (POWERPC_MASKS): Likewise.
17214 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
17215 tests against -mpower9-dform* to -mpower9-vector. Delete code
17216 that checked for -mpower9-dform* consistancy with other options.
17217 Add test for -mpower9-misc to enable other power9 switches.
17218 (rs6000_init_hard_regno_mode_ok): Likewise.
17219 (rs6000_option_override_internal): Likewise.
17220 (rs6000_emit_prologue): Likewise.
17221 (rs6000_emit_epilogue): Likewise.
17222 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
17223 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
17224 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
17226 (emit_fusion_p9_store): Likewise.
17227 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
17228 resetting these macros if the assembler does not support ISA 3.0
17230 (TARGET_P9_DFORM_VECTOR): Likewise.
17231 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
17232 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
17234 2017-08-25 Alan Modra <amodra@gmail.com>
17236 PR rtl-optimization/81747
17237 * cse.c (cse_extended_basic_block): Don't attempt to record
17238 equivalences for degenerate conditional branches that branch
17239 to their fall-through.
17241 2017-08-24 Martin Sebor <msebor@redhat.com>
17243 PR middle-end/81908
17244 * gimple-fold.c (size_must_be_zero_p): New function.
17245 (gimple_fold_builtin_memory_op): Call it.
17247 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
17249 * config/rs6000/mm_malloc.h: New file.
17251 2017-08-24 Bin Cheng <bin.cheng@arm.com>
17253 PR tree-optimization/81913
17254 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
17255 analysis when either IVs in condition can wrap.
17257 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
17259 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
17260 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
17262 2017-08-24 Richard Biener <rguenther@suse.de>
17265 * targhooks.c (default_target_can_inline_p): Properly
17266 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
17267 is present and always compare.
17268 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
17269 infer -mfpmath=sse from TARGET_SSE_P.
17270 (ix86_can_inline_p): Properly use target_option_default_node when
17271 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
17273 2017-08-24 Richard Biener <rguenther@suse.de>
17276 * dwarf2out.c (output_die): Handle flag_generate_offload like
17278 (output_comp_unit): Likewise.
17279 (gen_array_type_die): Likewise.
17280 (dwarf2out_early_finish): Likewise.
17281 (note_variable_value_in_expr): Likewise.
17282 (dwarf2out_finish): Likewise. Adjust assert.
17283 * cgraphunit.c (symbol_table::compile): Move setting of
17284 flag_generate_offload earlier ...
17285 (symbol_table::finalize_compilation_unit): ... here, before
17286 early debug finalization.
17288 2017-08-24 Richard Biener <rguenther@suse.de>
17290 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
17291 and ipa-fnsummary.h.
17292 (ix86_can_inline_p): When ix86_fpmath flags do not match
17293 check whether the callee uses FP math at all.
17295 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
17297 PR middle-end/81931
17298 * tree-ssanames.c (get_nonzero_bits): Use element_precision
17299 instead of TYPE_PRECISION.
17301 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
17302 Alan Hayward <alan.hayward@arm.com>
17303 David Sherwood <david.sherwood@arm.com>
17305 * combine.c (make_extraction): Use subreg_offset_from_lsb.
17307 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
17309 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
17311 (ix86_frame::stack_realign_allocate): New field.
17312 (struct machine_frame_state): Modify comments.
17313 (machine_frame_state::sp_realigned_fp_end): New field.
17314 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
17315 layout calculation.
17316 (sp_valid_at): Add assertion to assure no attempt to access invalid
17317 offset of a realigned stack.
17318 (fp_valid_at): Likewise.
17319 (choose_baseaddr): Modify comments.
17320 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
17321 ix86_expand_prologue.
17322 (ix86_expand_prologue): Modify stack realignment and allocation.
17323 (ix86_expand_epilogue): Modify comments.
17324 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
17325 avx2_runtime, avx512f, and avx512f_runtime.
17327 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
17329 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
17330 (mstackrealign): Do not init to -1.
17331 * config/i386/i386.c (ix86_option_override_internal):
17332 Check opts_set, not opts when setting default value of
17333 opts->x_ix86_force_align_arg_pointer.
17335 2017-08-23 Richard Biener <rguenther@suse.de>
17337 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
17338 lang_hooks.decl_printable_name.
17339 * print-rtl-function.c (print_rtx_function): Likewise.
17340 * tree-pretty-print.c (dump_function_header): Likewise.
17342 2017-08-23 Richard Biener <rguenther@suse.de>
17345 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
17346 -g0 at compile-time.
17348 2017-08-23 Tamar Christina <tamar.christina@arm.com>
17350 PR middle-end/19706
17351 * doc/sourcebuild.texi (Other hardware attributes):
17354 2017-08-23 Tamar Christina <tamar.christina@arm.com>
17356 PR middle-end/19706
17357 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
17358 Fix single-use check.
17360 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
17362 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
17364 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
17366 * doc/install.texi: Modify to add more details on running selected
17369 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
17371 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
17372 is combined with -mabi=ms.
17373 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
17376 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17378 PR tree-optimization/81488
17379 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
17380 and cached_basis fields.
17381 (MAX_SPREAD): New constant.
17382 (alloc_cand_and_find_basis): Initialize new fields.
17383 (clear_visited): New function.
17384 (create_phi_basis_1): Rename from create_phi_basis, set visited
17385 and cached_basis fields.
17386 (create_phi_basis): New wrapper function.
17387 (phi_add_costs_1): Rename from phi_add_costs, add spread
17388 parameter, set visited field, short-circuit when limits reached.
17389 (phi_add_costs): New wrapper function.
17390 (record_phi_increments_1): Rename from record_phi_increments, set
17392 (record_phi_increments): New wrapper function.
17393 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
17394 (phi_incr_cost): New wrapper function.
17395 (all_phi_incrs_profitable_1): Rename from
17396 all_phi_incrs_profitable, set visited field.
17397 (all_phi_incrs_profitable): New wrapper function.
17399 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17400 Alan Hayward <alan.hayward@arm.com>
17401 David Sherwood <david.sherwood@arm.com>
17403 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
17404 that takes the outer and inner modes.
17405 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
17406 comparison as the canonical test for a paradoxical subreg.
17407 * combine.c (simplify_set): Use paradoxical_subreg_p.
17408 (make_extraction): Likewise.
17409 (force_to_mode): Likewise.
17410 (rtx_equal_for_field_assignment_p): Likewise.
17411 (gen_lowpart_for_combine): Likewise.
17412 (simplify_comparison): Likewise.
17413 * cse.c (equiv_constant): Likewise.
17414 * expmed.c (store_bit_field_1): Likewise.
17415 * final.c (alter_subreg): Likewise.
17416 * fwprop.c (propagate_rtx): Likewise.
17417 (forward_propagate_subreg): Likewise.
17418 * ira-conflicts.c (ira_build_conflicts): Likewise.
17419 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
17420 * lra-constraints.c (curr_insn_transform): Likewise.
17421 (split_reg): Likewise.
17422 * lra-eliminations.c (move_plus_up): Likewise.
17423 (lra_eliminate_regs_1): Likewise.
17424 * recog.c (general_operand): Likewise.
17425 * ree.c (combine_reaching_defs): Likewise.
17426 * reload.c (push_reload): Likewise.
17427 (find_reloads): Likewise.
17428 * reload1.c (elimination_effects): Likewise.
17429 (compute_reload_subreg_offset): Likewise.
17430 (choose_reload_regs): Likewise.
17431 * rtlanal.c (subreg_lsb_1): Likewise.
17432 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17433 (simplify_subreg): Likewise.
17434 * var-tracking.c (track_loc_p): Likewise.
17435 * emit-rtl.c (byte_lowpart_offset): Likewise.
17436 (paradoxical_subreg_p): Delete out-of-line definition.
17438 2017-08-22 Jeff Law <law@redhat.com>
17440 PR tree-optimization/81741
17441 PR tree-optimization/71947
17442 * tree-ssa-dom.c: Include tree-inline.h.
17443 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
17444 equivalences if one is more expensive to compute than the other.
17445 * tree-ssa-scopedtables.h (class const_or_copies): Make
17446 record_const_or_copy_raw method private.
17447 (class avail_exprs_stack): New method simplify_binary_operation.
17448 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
17449 avail_exprs_stack::simplify_binary_operation as needed.
17450 (avail_exprs_stack::simplify_binary_operation): New function.
17452 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
17454 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
17455 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
17456 (DOT_SYMBOLS): Likewise.
17457 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
17458 (RELOCATABLE_NEEDS_FIXUP): Likewise.
17459 (RS6000_ABI_NAME): Likewise.
17460 (TARGET_CMODEL): Likewise.
17461 (TOC_SECTION_ASM_OP): Likewise.
17462 (SET_CMODEL): New macro.
17463 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
17465 2017-08-22 Richard Biener <rguenther@suse.de>
17467 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
17468 to allow for free-lang-data replacements similar to verify_type_variant.
17470 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17471 Alan Hayward <alan.hayward@arm.com>
17472 David Sherwood <david.sherwood@arm.com>
17474 * config/aarch64/aarch64.md (casesi): Use DImode rather than
17475 VOIDmode for the LABEL_REF.
17477 2017-08-22 Richard Biener <rguenther@suse.de>
17479 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
17481 2017-08-22 Richard Biener <rguenther@suse.de>
17483 * common.opt (feliminate-dwarf2-dups): Ignore.
17484 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
17485 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
17486 same_die_p_wrap, compute_section_prefix,
17487 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
17488 (comdat_symbol_id, comdat_symbol_number): Likewise.
17489 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
17491 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
17492 (output_die): Mark unreachable path unreachable.
17493 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
17494 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
17495 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
17496 (dwarf2out_early_finish): Likewise.
17498 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
17500 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
17502 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
17505 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
17506 not VAR_P. Filter attribute warnings with OPT_Wattributes.
17507 (avr_attribute_table) <io, io_low, address>: Initialize
17508 .decl_required with true.
17510 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
17512 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
17513 undocumented debugging options.
17514 (-mvsx-scalar-double): Likewise.
17515 (-mallow-df-permute): Likewise.
17516 (-mvectorize-builtins): Likewise.
17517 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17518 (rs6000_builtin_vectorized_function): Likewise.
17519 (rs6000_builtin_md_vectorized_function): Likewise.
17520 (rs6000_opt_vars): Likewise.
17522 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
17525 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
17526 (*btrq_imm): Rename from *btrq.
17527 (*btcq_imm): Rename from *btcq.
17528 (btsc): New code attribute.
17529 (*<btsc><mode>): New insn pattern.
17530 (*btr<mode>): Ditto.
17531 (*<btsc><mode>_mask): New insn_and_split pattern.
17532 (*btr<mode>_mask): Ditto.
17534 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17535 Alan Hayward <alan.hayward@arm.com>
17536 David Sherwood <david.sherwood@arm.com>
17538 * function.c (pad_below): Simplify padding calculation.
17540 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17541 Alan Hayward <alan.hayward@arm.com>
17542 David Sherwood <david.sherwood@arm.com>
17544 * target.def (function_prologue): Remove frame size argument.
17545 (function_epilogue): Likewise.
17546 * doc/tm.texi: Regenerate.
17547 * final.c (final_start_function): Update call to function_prologue.
17548 (final_end_function): Update call to function_epilogue.
17549 (default_function_pro_epilogue): Remove frame size argument.
17550 * output.h (default_function_pro_epilogue): Likewise.
17551 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
17552 (arm_output_function_prologue): Likewise.
17553 * config/frv/frv.c (frv_function_prologue): Likewise.
17554 (frv_function_epilogue): Likewise.
17555 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
17556 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
17557 (ia64_output_function_epilogue): Likewise.
17558 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
17559 (m32r_output_function_epilogue): Likewise.
17560 * config/microblaze/microblaze.c (microblaze_function_prologue)
17561 (microblaze_function_epilogue): Likewise.
17562 * config/mips/mips.c (mips_output_function_prologue): Likewise.
17563 (mips_output_function_epilogue): Likewise.
17564 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
17565 (mmix_target_asm_function_epilogue): Likewise.
17566 * config/msp430/msp430.c (msp430_start_function): Likewise.
17567 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
17568 (nds32_asm_function_epilogue): Likewise.
17569 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
17570 * config/pa/pa.c (pa_output_function_prologue): Likewise.
17571 (pa_output_function_epilogue): Likewise.
17572 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
17573 (rs6000_output_function_epilogue): Likewise.
17574 * config/rl78/rl78.c (rl78_start_function): Likewise.
17575 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
17576 (rs6000_output_function_epilogue): Likewise.
17577 * config/rx/rx.c (rx_output_function_prologue): Likewise.
17578 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
17579 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
17580 (sparc_asm_function_epilogue): Likewise.
17582 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17584 * tree.h (type_has_mode_precision_p): New function.
17585 * convert.c (convert_to_integer_1): Use it.
17586 * expr.c (expand_expr_real_2): Likewise.
17587 (expand_expr_real_1): Likewise.
17588 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
17589 * match.pd: Likewise.
17590 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
17591 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
17592 * tree-tailcall.c (process_assignment): Likewise.
17593 * tree-vect-loop.c (vectorizable_reduction): Likewise.
17594 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
17595 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
17596 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17597 (vectorizable_assignment): Likewise.
17598 (vectorizable_shift): Likewise.
17599 (vectorizable_operation): Likewise.
17600 * tree-vrp.c (register_edge_assert_for_2): Likewise.
17602 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
17604 * match.pd: Add pow (C, x) simplification.
17606 2017-08-21 Richard Biener <rguenther@suse.de>
17608 PR tree-optimization/81900
17609 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
17610 for blocks with abnormal predecessors.
17611 (compute_antic): Do not set visited flag prematurely.
17613 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
17616 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
17618 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17620 * stor-layout.h (vector_type_mode): Move to...
17621 * tree.h (vector_type_mode): ...here.
17622 * stor-layout.c (vector_type_mode): Move to...
17623 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
17625 2017-08-21 Richard Biener <rguenther@suse.de>
17627 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
17628 register_external_die hooks.
17629 (debug_false_tree_charstarstar_uhwistar): Declare.
17630 (debug_nothing_tree_charstar_uhwi): Likewise.
17631 * debug.c (do_nothing_debug_hooks): Adjust.
17632 (debug_false_tree_charstarstar_uhwistar): New do nothing.
17633 (debug_nothing_tree_charstar_uhwi): Likewise.
17634 * dbxout.c (dbx_debug_hooks): Adjust.
17635 (xcoff_debug_hooks): Likewise.
17636 * sdbout.c (sdb_debug_hooks): Likewise.
17637 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
17638 * dwarf2out.c (macinfo_label_base): New global.
17639 (dwarf2out_register_external_die): New function for the
17640 register_external_die hook.
17641 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
17642 (dwarf2_debug_hooks): Use them.
17643 (dwarf2_lineno_debug_hooks): Adjust.
17644 (struct die_struct): Add with_offset flag.
17645 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
17646 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
17647 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
17648 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
17649 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
17650 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
17651 defining section names for the early LTO debug variants.
17652 (reset_indirect_string): New helper.
17653 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
17654 (print_dw_val): Add support for offsetted symbol references.
17655 (get_ultimate_context): Split out from is_cxx.
17656 (is_cxx): Use get_ultimate_context.
17657 (is_fortran): Add decl overload.
17658 (compute_comp_unit_symbol): Split out worker from
17659 compute_section_prefix.
17660 (compute_section_prefix): Call compute_comp_unit_symbol and
17661 set comdat_type_p here.
17662 (output_die): Skip DIE symbol output for the LTO added one.
17663 Handle DIE symbol references with offset.
17664 (output_comp_unit): Guard section name mangling properly.
17665 For LTO debug sections emit a symbol at the section beginning
17666 which we use to refer to its DIEs.
17667 (add_abstract_origin_attribute): For DIEs registered via
17668 dwarf2out_register_external_die directly refer to the early
17669 DIE rather than indirectly through the shadow one we created.
17670 Remove obsolete call to dwarf2out_abstract_function for
17671 non-function/block origins.
17672 (gen_array_type_die): When generating early LTO debug do
17673 not emit DW_AT_string_length.
17674 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
17675 late when in LTO. As suggested place a gcc_unreachable for
17676 the DECL_ABSTRACT_P case.
17677 (gen_subprogram_die): Avoid another specification DIE
17678 for early built declarations/definitions for the late LTO case.
17679 (gen_variable_die): Add type references for late duplicated VLA dies
17681 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
17682 we have the abstract instance already.
17683 (process_scope_var): Adjust decl DIE contexts in LTO which
17684 first puts them in limbo.
17685 (gen_decl_die): Do not generate type DIEs late apart from
17686 types for VLAs or for decls we do not yet have a DIE. Do not
17687 call dwarf2out_abstract_function late.
17688 (dwarf2out_early_global_decl): Make sure to create DIEs
17689 for abstract instances of a decl first.
17690 (dwarf2out_late_global_decl): Adjust comment.
17691 (output_macinfo_op): With multiple macro sections use
17692 macinfo_label_base to distinguish labels.
17693 (output_macinfo): Likewise. Update macinfo_label_base.
17694 Pass in the line info label.
17695 (note_variable_value_in_expr): When generating LTO resolve
17696 all variable values here by generating DIEs as needed.
17697 (init_sections_and_labels): Add early LTO debug flag parameter
17698 and generate different sections and names if set. Add generation
17699 counter for the labels so we can have multiple of them.
17700 (reset_dies): Helper to allow DIEs to be output multiple times.
17701 (dwarf2out_finish): When outputting DIEs to the fat part of an
17702 LTO object first reset DIEs.
17703 (dwarf2out_early_finish): Output early DIEs when generating LTO.
17704 (modified_type_die): Check for decl_ultimate_origin being self
17706 (gen_type_die_with_usage): Likewise.
17707 (gen_typedef_die): Allow decl_ultimate_origin being self.
17708 (set_decl_abstract_flags): Remove.
17709 (set_block_abstract_flags): Likewise.
17710 (dwarf2out_abstract_function): Treat the early generated DIEs
17711 as the abstract copy and only add DW_AT_inline and
17712 DW_AT_artificial here and call set_decl_origin_self.
17713 If the DIE has an abstract origin don't do anything.
17714 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
17715 if we have none yet (Go fails to build one, PR78628).
17716 (variably_modified_type_p): Prevent endless recursion for Ada
17717 cyclic pointer types.
17718 * lto-streamer-in.c: Include debug.h.
17719 (dref_queue): New global.
17720 (lto_read_tree_1): Stream in DIE references.
17721 (lto_input_tree): Register DIE references.
17722 (input_function): Stream DECL_DEBUG_ARGS.
17723 * lto-streamer-out.c: Include debug.h.
17724 (lto_write_tree_1): Output DIE references.
17725 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
17726 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
17727 (output_function): Stream DECL_DEBUG_ARGS.
17728 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
17729 Stream DECL_ABSTRACT_ORIGIN.
17730 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
17731 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
17732 DECL_CONTEXT for file-scope decls.
17733 * lto-streamer.h (struct dref_entry): Declare.
17734 (dref_queue): Likewise.
17735 * cfgexpand.c (pass_expand::execute): Do not call the
17736 outlining_inline_function hook here.
17737 * lto-wrapper.c (debug_obj): New global.
17738 (tool_cleanup): Unlink it if required.
17739 (debug_objcopy): New function.
17740 (run_gcc): Handle early debug sections in the IL files by
17741 extracting them to separate files, partially linkin them and
17742 feeding the result back as result to the linker.
17743 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
17744 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
17745 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
17746 sections into a separate segment.
17747 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
17749 (darwin_asm_dwarf_section): Likewise.
17750 (darwin_asm_output_dwarf_offset): Likewise.
17751 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
17753 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17754 Alan Hayward <alan.hayward@arm.com>
17755 David Sherwood <david.sherwood@arm.com>
17757 * read-md.h (md_reader::record_potential_iterator_use): Replace
17758 pointer argument with an rtx and an index.
17759 * read-rtl.c (iterator_group::apply_iterator): Likewise.
17760 (apply_mode_iterator): Likewise.
17761 (apply_code_iterator): Likewise.
17762 (apply_int_iterator): Likewise.
17763 (apply_subst_iterator): Likewise.
17764 (record_iterator_use): Likewise.
17765 (record_attribute_use): Likewise.
17766 (md_reader::record_potential_iterator_use): Likewise. Update calls
17767 to record_iterator_use and apply_iterator.
17768 (iterator_use): Replace ptr with x and index.
17769 (attribute_use): Likewise.
17770 (apply_attribute_uses): Update calls to apply_iterator.
17771 (apply_iterators): Likewise. Update initialization of iterator_use.
17772 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
17773 and record_potential_iterator_use.
17774 (rtx_reader::read_rtx_operand): Likewise.
17776 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17777 Alan Hayward <alan.hayward@arm.com>
17778 David Sherwood <david.sherwood@arm.com>
17780 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
17783 2017-08-21 Richard Biener <rguenther@suse.de>
17785 PR middle-end/81884
17786 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
17787 at struct end conservatively when comparing common bases.
17789 2017-08-21 Richard Biener <rguenther@suse.de>
17791 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
17792 (mem_ref_in_stmt): Remove.
17793 (determine_max_movement): Use ref index to get at the reference.
17794 (invariantness_dom_walker::before_dom_children): Deal with
17795 lim data already initialized.
17796 (gather_mem_refs_stmt): Initialize lim data and record ref index.
17798 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
17800 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
17801 (TARGET_ISA_ROUND): Ditto.
17802 (TARGET_ROUND): Ditto.
17803 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
17804 * config/i386/i386.md: Ditto.
17805 * config/i386/sse.md: Ditto.
17806 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
17807 with OPTION_MASK_ISA_SSE4_1.
17809 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
17812 * doc/extend.texi (x86 Built-in Functions): Correct the name of
17813 __builtin_ia32_lzcnt_u16.
17815 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
17818 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
17819 (rs6000_set_current_function): Rewrite function to use it.
17821 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
17824 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
17825 and TYPE_WARN_IF_NOT_ALIGN.
17826 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
17827 (handle_warn_if_not_align): New.
17828 (place_union_field): Call handle_warn_if_not_align.
17829 (place_field): Call handle_warn_if_not_align.
17830 Copy TYPE_WARN_IF_NOT_ALIGN.
17831 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
17832 (layout_type): Likewise.
17833 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
17835 (tree_decl_common): Add warn_if_not_align.
17836 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
17837 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
17838 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
17839 (DECL_WARN_IF_NOT_ALIGN): Likewise.
17840 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
17841 * doc/extend.texi: Document warn_if_not_aligned attribute.
17842 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
17844 2017-08-17 Martin Liska <mliska@suse.cz>
17847 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
17848 (get_data_dependence): Use it as pointer type.
17849 (distribute_loop): Likewise.
17851 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17853 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
17854 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
17855 (p8_vmrgow_<mode>_direct): New define_insn.
17856 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
17857 handle endianness for vmrgew and vmrgow permute patterns.
17859 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
17861 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
17862 * config/rs6000/rs6000-cpus.def: Remove comment.
17863 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
17864 (POWERPC_MASKS): Likewise.
17865 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
17866 use of TARGET_VSX_TIMODE.
17867 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
17868 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
17869 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
17870 (rs6000_option_override_internal): Remove dead code.
17871 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
17872 (rs6000_legitimize_reload_address): Likewise.
17873 (rs6000_legitimate_address_p): Likewise.
17874 (rs6000_opt_masks): Delete "vsx-timode".
17875 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
17876 from function comment.
17877 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
17878 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
17879 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
17881 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
17882 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
17883 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
17884 (VSX_TI): Likewise.
17886 (define_peephole2): Likewise.
17888 2017-08-17 Martin Sebor <msebor@redhat.com>
17891 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
17892 past the end of an array.
17893 (test_pp_format): Add test cases.
17895 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
17897 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
17898 missing ECF_NOTHROW flags.
17900 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
17903 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
17904 operands residing in integer registers.
17905 (*vsx_le_perm_load_<mode>): Likewise.
17906 (*vsx_le_perm_store_<mode>): Likewise.
17907 (define_peephole2): Add peepholes to optimize the above.
17909 2017-08-17 Marek Polacek <polacek@redhat.com>
17911 PR middle-end/81814
17912 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
17913 to mimic what shorten_compare did. Change the return type to bool.
17914 (fold_cond_expr_with_comparison): Update call to
17915 operand_equal_for_comparison_p.
17916 (fold_ternary_loc): Likewise.
17918 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
17920 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
17922 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
17924 2017-08-17 Richard Biener <rguenther@suse.de>
17926 * tree-ssa-structalias.c (solve_graph): When propagating
17927 to successors update the graphs succ edges and avoid duplicate work.
17929 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
17932 * config/i386/i386.c (ix86_option_override_internal): Save target
17933 specific options after ix86_stack_protector_guard_reg was changed.
17935 2017-08-17 Richard Biener <rguenther@suse.de>
17937 PR tree-optimization/81827
17938 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
17939 (new_var_info): Initialize it conservatively.
17940 (get_call_vi): Mark register vars.
17941 (new_scalar_tmp_constraint_exp): Likewise.
17942 (handle_rhs_call): Likewise.
17943 (handle_const_call): Likewise.
17944 (create_function_info_for): Likewise.
17945 (solve_constraints): Sort varinfos to separate register from
17946 non-register vars to pack points-to solution bitmaps during
17949 2017-08-17 Marek Polacek <polacek@redhat.com>
17951 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
17953 2017-08-17 Richard Biener <rguenther@suse.de>
17955 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
17956 to true when overflow is undefined and we saturated the result.
17958 2017-08-17 Alan Modra <amodra@gmail.com>
17961 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
17962 Don't use store multiple if only one reg needs saving.
17963 (interesting_frame_related_regno): New function.
17964 (rs6000_frame_related): Don't emit frame info for regs that
17966 (rs6000_emit_epilogue): Likewise.
17968 2017-08-16 Nathan Sidwell <nathan@acm.org>
17970 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
17971 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
17972 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
17973 * tree.c (free_lang_data_in_type): Use else-if chain. Always
17974 clear TYPE_LANG_1. Remove obsolete member-function stripping.
17975 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
17976 (verify_type): Adjust for TYPE_BINFO move.
17977 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
17978 process TYPE_BINFO directly.
17979 (hash_tree): Likewise.
17980 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17982 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17985 2017-08-16 David Malcolm <dmalcolm@redhat.com>
17987 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
17989 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
17992 * config/i386/i386.md (*anddi_1_btr): Change predicates of
17993 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
17994 Add ix86_binary_operator_ok to insn constraint.
17995 (*iordi_1_bts): Ditto.
17996 (*xordi_1_btc): Ditto.
17997 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
17998 Update corresponding peephole2 pattern.
18002 2017-08-16 Bin Cheng <bin.cheng@arm.com>
18004 PR tree-optimization/81832
18005 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
18006 copy loop header which has IFN_LOOP_DIST_ALIAS call.
18008 2017-08-16 Marek Polacek <polacek@redhat.com>
18011 * fold-const.c (fold_indirect_ref_1): Restore original behavior
18012 regarding size_zero_node.
18014 2017-08-16 Martin Liska <mliska@suse.cz>
18017 * config.gcc: Respect previously set extra_objs in case
18020 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
18022 PR tree-optimization/81835
18023 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
18024 the phi SSA_NAME. Check that the condition in a COND_EXPR does
18025 not depend on the phi.
18027 2017-08-16 Alan Modra <amodra@gmail.com>
18029 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
18032 2017-08-16 Alan Modra <amodra@gmail.com>
18034 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
18035 (save_reg_p): ..into this. Update all callers.
18036 (first_reg_to_save): Simplify.
18038 2017-08-16 Alan Modra <amodra@gmail.com>
18040 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
18043 2017-08-15 Joseph Myers <joseph@codesourcery.com>
18047 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
18048 constant count if that count is less than 32.
18050 2017-08-15 Nathan Sidwell <nathan@acm.org>
18052 * gcc.c (execute): Emit friendlier message if inferior is killed
18053 by an external cause.
18055 2017-08-15 Richard Biener <rguenther@suse.de>
18057 PR tree-optimization/81790
18058 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
18059 CONSTRUCTORs from simplifying and VN.
18061 2017-08-14 Martin Sebor <msebor@redhat.com>
18063 * builtin-attrs.def: Add comments.
18065 2017-08-14 Martin Sebor <msebor@redhat.com>
18068 * doc/extend.texi (attribute nonstring): Document new attribute.
18070 2017-08-14 Martin Sebor <msebor@redhat.com>
18073 * tree-diagnostic.c (default_tree_printer): Handle %G.
18074 * gimple-pretty-print.h (percent_G_format): Declare new function.
18075 * gimple-pretty-print.c (percent_G_format): Define.
18076 * tree-pretty-print.c (percent_K_format): Add argument.
18078 2017-08-14 Martin Sebor <msebor@redhat.com>
18080 PR translation/79998
18081 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
18082 Remove a stray space.
18084 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
18087 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
18088 (*iordi_1_bts): Ditto.
18089 (*xordi_1_btc): Ditto.
18091 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18094 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
18095 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18097 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
18098 quoted strings, and make more translator-friendly.
18099 (darwin_rs6000_override_options): Likewise.
18100 (rs6000_option_override_internal): Likewise.
18101 (rs6000_return_in_memory): Fix overlong line.
18102 (init_cmulative_args): Use quoted strings, and make more
18103 translator-friendly.
18104 (rs6000_pass_by_reference): Fix overlong line.
18105 (def_builtin): Use quoted strings.
18106 (altivec_expand_predicate_builtin): Use quoted strings, and make
18107 more translator-friendly.
18108 (htm_expand_builtin): Use quoted strings.
18109 (cpu_expand_builtin): Use quoted strings, and make more
18110 translator-friendly.
18111 (altivec_expand_builtin): Likewise.
18112 (paired_expand_predicate_builtin): Likewise.
18113 (rs6000_invalid_builtin): Likewise.
18114 (builtin_function_type): Use quoted strings.
18115 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
18116 more translator-friendly.
18117 (rs6000_trampoline_init): Likewise.
18118 (rs6000_handle_altivec_attribute): Likewise.
18119 (rs6000_inner_target_options): Use quoted strings.
18120 (rs6000_disable_incompatible_switches): Likewise.
18121 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
18122 strings, and make more translator-friendly.
18123 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
18125 2017-08-14 Bin Cheng <bin.cheng@arm.com>
18127 PR tree-optimization/81799
18128 * tree-loop-distribution.c (version_loop_by_alias_check): Force
18129 cond_expr to simple gimple operand.
18131 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
18133 PR middle-end/46932
18134 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
18136 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
18140 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
18141 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
18142 TARGET_GASISR_PROLOGUES.
18143 * config/avr/avr.c (avr_option_override): Same.
18144 (avr_pass_pre_proep::execute): Same.
18146 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
18149 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
18150 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
18152 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
18154 * config/i386/i386.md (*load_tp_<mode>): Redefine as
18155 define_insn_and_split. Split to a memory load from 0 in
18156 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
18157 using PTR mode iterator.
18158 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
18159 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
18160 (*add_tp_<mode>): Redefine as define_insn_and_split.
18161 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
18162 address space. Merge with *add_tp_x32 using PTR mode iterator.
18163 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
18164 Split to an add with a memory load from 0 in
18165 DEFAULT_TLS_SEG_REG address space.
18167 2017-08-12 Andrew Pinski <apinski@cavium.com>
18169 * config/aarch64/aarch64-option-extensions.def (rdma):
18170 Fix feature string to what Linux prints out in /proc/cpuinfo.
18172 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
18175 * dwarf2out.c (modified_type_die): For C typedef types that have
18176 an ultimate origin, process the ultimate origin instead of the
18178 (gen_typedef_die): Assert that input DECLs have no ultimate
18180 (gen_type_die_with_usage): For typedef variants that have an
18181 ultimate origin, just call gen_decl_die on the original DECL.
18182 (process_scope_var): Avoid creating DIEs for local typedefs and
18183 concrete static variables.
18185 2017-08-12 Alan Modra <amodra@gmail.com>
18189 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
18190 match gnu-user.h startfile.
18191 (ENDFILE_LINUX_SPEC): Similarly.
18193 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
18196 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
18198 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
18200 2017-08-11 Tamar Christina <tamar.christina@arm.com>
18201 * config/aarch64/aarch64.md (mov<mode>): Change.
18202 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
18203 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
18204 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
18206 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
18208 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
18209 for storage order barriers.
18211 2017-08-11 Martin Liska <mliska@suse.cz>
18214 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
18215 variables of void type.
18217 2017-08-11 Martin Liska <mliska@suse.cz>
18219 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
18220 TARGET_SUPPORTS_ALIASES.
18221 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
18222 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
18223 (optimize_weakref): Likewise.
18224 * symtab.c (symtab_node::noninterposable_alias): Likewise.
18225 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
18226 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
18228 2017-08-11 Martin Liska <mliska@suse.cz>
18231 * config/i386/i386.c (make_resolver_func): Do complete
18232 refactoring of the function.
18234 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
18237 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
18238 * config/i386/i386.c (ix86_stack_protect_guard): Use
18239 ix86_stack_protect_guard_symbol_str to generate varible declaration.
18240 * doc/invoke.texi (x86 Options): Document
18241 -mstack-protector-guard-symbol= option.
18243 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
18245 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
18246 * config/i386/i386.c (ix86_split_stack_guard): New function.
18247 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
18248 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
18249 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
18250 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
18251 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
18252 (split_stack_space_check): Call ix86_split_stack_guard.
18254 2017-08-10 Martin Sebor <msebor@redhat.com>
18256 * print-tree.c (print_node): Print location using the established
18258 Replace spaces with colons.
18259 (debug_raw, debug): Ditto.
18261 2017-08-10 Martin Sebor <msebor@redhat.com>
18264 * pretty-print.c (pp_format): Correct the handling of %s precision.
18266 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
18269 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
18271 (ix86_finalize_stack_frame_flags): This. Also clear
18272 frame_pointer_needed if -fno-omit-frame-pointer is used without
18274 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
18275 with ix86_finalize_stack_frame_flags.
18276 (ix86_expand_epilogue): Likewise.
18277 (ix86_expand_split_stack_prologue): Likewise.
18278 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
18280 2017-08-10 Martin Liska <mliska@suse.cz>
18283 * c-attribs.c (handle_target_attribute):
18284 Report warning for an empty string argument of target attribute.
18286 2017-08-09 Jakub Jelinek <jakub@redhat.com>
18289 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
18291 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
18292 or DECL_NONLOCAL labels.
18293 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
18294 or DECL_NONLOCAL labels here.
18296 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
18298 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
18299 to indicate when early gimple folding has been disabled.
18300 (rs6000_gimple_fold_builtin): Add debug content.
18301 (rs6000_invalid_builtin): Fix whitespace.
18302 (rs6000_expand_builtin): Fix whitespace.
18303 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
18305 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
18308 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
18309 SAVE_MULTIPLE if not all the registers that saves, should be saved.
18311 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
18313 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
18314 (qdf24xx): Likewise.
18315 * config/aarch64/aarch64.md: Include falkor.md.
18316 * config/aarch64/falkor.md: New.
18318 2017-08-09 Marek Polacek <polacek@redhat.com>
18321 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
18322 * diagnostic.c (emit_diagnostic): Add a comment.
18323 (emit_diagnostic_valist): New function.
18325 2017-08-09 Marek Polacek <polacek@redhat.com>
18328 * input.c (make_location): New overload.
18329 * input.h (make_location): Declare.
18331 2017-08-08 Alan Modra <amodra@gmail.com>
18332 H.J. Lu <hongjiu.lu@intel.com>
18335 * gcc.c (NO_PIE_SPEC): Delete.
18336 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
18338 (LINK_PIE_SPEC): ..to here.
18339 (LINK_COMMAND_SPEC): Support -no-pie.
18340 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
18341 chain of crtbegin*.o selection, update for PIE_SPEC changes and
18343 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
18344 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
18345 (ENDFILE_CRTEND_SPEC): Similarly.
18347 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
18350 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
18351 (mstack-protector-guard-offset=): Ditto.
18352 * config/i386/i386.c (ix86_option_override): Handle
18353 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
18355 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
18356 ix86_stack_protect_guard_offset variables.
18357 (TARGET_STACK_PROTECT_GUARD): Always define.
18358 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
18359 and -mstack-protector-guard-offset= options.
18361 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18363 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
18364 boundary case for the last candidate.
18366 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18368 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
18369 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
18371 2017-08-08 Tamar Christina <tamar.christina@arm.com>
18373 PR middle-end/19706
18374 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
18375 * config/aarch64/aarch64-builtins.c
18376 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
18377 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
18378 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
18380 2017-08-08 Tamar Christina <tamar.christina@arm.com>
18381 Andrew Pinski <pinskia@gmail.com>
18383 PR middle-end/19706
18384 * internal-fn.def (XORSIGN): New.
18385 * optabs.def (xorsign_optab): New.
18386 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
18387 (convert_expand_mult_copysign): New.
18388 (pass_optimize_widening_mul::execute): Call
18389 convert_expand_mult_copysign.
18391 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18393 PR tree-optimization/81354
18394 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
18395 Insert on edges rather than explicitly creating landing pads.
18396 (analyze_candidates_and_replace): Commit edge inserts.
18398 2017-08-08 Richard Biener <rguenther@suse.de>
18400 PR middle-end/81719
18401 * tree-ssa-loop-niter.c: Include tree-dfa.h.
18402 (expand_simple_operations): Also look through ADDR_EXPRs with
18403 MEM_REF bases treating them as POINTER_PLUS_EXPR.
18405 2017-08-08 Richard Biener <rguenther@suse.de>
18407 PR tree-optimization/81723
18408 * tree-vect-slp.c (struct bst_traits): New hash traits.
18409 (bst_fail): New global.
18410 (vect_build_slp_tree_2): New worker, split out from ...
18411 (vect_build_slp_tree): ... this now wrapping it with using
18412 bst_fail set to cache SLP tree build fails. Properly handle
18414 (vect_analyze_slp_instance): Allocate and free bst_fail.
18416 2017-08-08 Martin Liska <mliska@suse.cz>
18419 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
18420 LABEL_DECLs that can be from a different function.
18422 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18424 PR tree-optimization/81744
18425 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
18426 loop's number of iterations.
18428 2017-08-08 Martin Liska <mliska@suse.cz>
18430 * asan.c: Include header files.
18431 * attribs.c (build_decl_attribute_variant): New function moved
18433 (build_type_attribute_qual_variant): Likewise.
18434 (cmp_attrib_identifiers): Likewise.
18435 (simple_cst_list_equal): Likewise.
18436 (omp_declare_simd_clauses_equal): Likewise.
18437 (attribute_value_equal): Likewise.
18438 (comp_type_attributes): Likewise.
18439 (build_type_attribute_variant): Likewise.
18440 (lookup_ident_attribute): Likewise.
18441 (remove_attribute): Likewise.
18442 (merge_attributes): Likewise.
18443 (merge_type_attributes): Likewise.
18444 (merge_decl_attributes): Likewise.
18445 (merge_dllimport_decl_attributes): Likewise.
18446 (handle_dll_attribute): Likewise.
18447 (attribute_list_equal): Likewise.
18448 (attribute_list_contained): Likewise.
18449 * attribs.h (lookup_attribute): New function moved from tree.[ch].
18450 (lookup_attribute_by_prefix): Likewise.
18451 * bb-reorder.c: Include header files.
18452 * builtins.c: Likewise.
18453 * calls.c: Likewise.
18454 * cfgexpand.c: Likewise.
18455 * cgraph.c: Likewise.
18456 * cgraphunit.c: Likewise.
18457 * convert.c: Likewise.
18458 * dwarf2out.c: Likewise.
18459 * final.c: Likewise.
18460 * fold-const.c: Likewise.
18461 * function.c: Likewise.
18462 * gimple-expr.c: Likewise.
18463 * gimple-fold.c: Likewise.
18464 * gimple-pretty-print.c: Likewise.
18465 * gimple.c: Likewise.
18466 * gimplify.c: Likewise.
18467 * hsa-common.c: Likewise.
18468 * hsa-gen.c: Likewise.
18469 * internal-fn.c: Likewise.
18470 * ipa-chkp.c: Likewise.
18471 * ipa-cp.c: Likewise.
18472 * ipa-devirt.c: Likewise.
18473 * ipa-fnsummary.c: Likewise.
18474 * ipa-inline.c: Likewise.
18475 * ipa-visibility.c: Likewise.
18477 * lto-cgraph.c: Likewise.
18478 * omp-expand.c: Likewise.
18479 * omp-general.c: Likewise.
18480 * omp-low.c: Likewise.
18481 * omp-offload.c: Likewise.
18482 * omp-simd-clone.c: Likewise.
18483 * opts-global.c: Likewise.
18484 * passes.c: Likewise.
18485 * predict.c: Likewise.
18486 * sancov.c: Likewise.
18487 * sanopt.c: Likewise.
18488 * symtab.c: Likewise.
18489 * toplev.c: Likewise.
18490 * trans-mem.c: Likewise.
18491 * tree-chkp.c: Likewise.
18492 * tree-eh.c: Likewise.
18493 * tree-into-ssa.c: Likewise.
18494 * tree-object-size.c: Likewise.
18495 * tree-parloops.c: Likewise.
18496 * tree-profile.c: Likewise.
18497 * tree-ssa-ccp.c: Likewise.
18498 * tree-ssa-live.c: Likewise.
18499 * tree-ssa-loop.c: Likewise.
18500 * tree-ssa-sccvn.c: Likewise.
18501 * tree-ssa-structalias.c: Likewise.
18502 * tree-ssa.c: Likewise.
18503 * tree-streamer-in.c: Likewise.
18504 * tree-vectorizer.c: Likewise.
18505 * tree-vrp.c: Likewise.
18506 * tsan.c: Likewise.
18507 * ubsan.c: Likewise.
18508 * varasm.c: Likewise.
18509 * varpool.c: Likewise.
18510 * tree.c: Remove functions moved to attribs.[ch].
18511 * tree.h: Likewise.
18512 * config/aarch64/aarch64.c: Add attrs.h header file.
18513 * config/alpha/alpha.c: Likewise.
18514 * config/arc/arc.c: Likewise.
18515 * config/arm/arm.c: Likewise.
18516 * config/avr/avr.c: Likewise.
18517 * config/bfin/bfin.c: Likewise.
18518 * config/c6x/c6x.c: Likewise.
18519 * config/cr16/cr16.c: Likewise.
18520 * config/cris/cris.c: Likewise.
18521 * config/darwin.c: Likewise.
18522 * config/epiphany/epiphany.c: Likewise.
18523 * config/fr30/fr30.c: Likewise.
18524 * config/frv/frv.c: Likewise.
18525 * config/ft32/ft32.c: Likewise.
18526 * config/h8300/h8300.c: Likewise.
18527 * config/i386/winnt.c: Likewise.
18528 * config/ia64/ia64.c: Likewise.
18529 * config/iq2000/iq2000.c: Likewise.
18530 * config/lm32/lm32.c: Likewise.
18531 * config/m32c/m32c.c: Likewise.
18532 * config/m32r/m32r.c: Likewise.
18533 * config/m68k/m68k.c: Likewise.
18534 * config/mcore/mcore.c: Likewise.
18535 * config/microblaze/microblaze.c: Likewise.
18536 * config/mips/mips.c: Likewise.
18537 * config/mmix/mmix.c: Likewise.
18538 * config/mn10300/mn10300.c: Likewise.
18539 * config/moxie/moxie.c: Likewise.
18540 * config/msp430/msp430.c: Likewise.
18541 * config/nds32/nds32-isr.c: Likewise.
18542 * config/nds32/nds32.c: Likewise.
18543 * config/nios2/nios2.c: Likewise.
18544 * config/nvptx/nvptx.c: Likewise.
18545 * config/pa/pa.c: Likewise.
18546 * config/pdp11/pdp11.c: Likewise.
18547 * config/powerpcspe/powerpcspe.c: Likewise.
18548 * config/riscv/riscv.c: Likewise.
18549 * config/rl78/rl78.c: Likewise.
18550 * config/rx/rx.c: Likewise.
18551 * config/s390/s390.c: Likewise.
18552 * config/sh/sh.c: Likewise.
18553 * config/sol2.c: Likewise.
18554 * config/sparc/sparc.c: Likewise.
18555 * config/spu/spu.c: Likewise.
18556 * config/stormy16/stormy16.c: Likewise.
18557 * config/tilegx/tilegx.c: Likewise.
18558 * config/tilepro/tilepro.c: Likewise.
18559 * config/v850/v850.c: Likewise.
18560 * config/vax/vax.c: Likewise.
18561 * config/visium/visium.c: Likewise.
18562 * config/xtensa/xtensa.c: Likewise.
18564 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18567 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
18568 constraints since the -mupper-regs-* switches have been
18570 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
18571 into a vector from a double word element that was extracted from
18572 another vector, and eliminate extra XXPERMDI instructions.
18573 (vsx_concat_<mode>_2): Likewise.
18574 (vsx_concat_<mode>_3): Likewise.
18575 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
18576 concat to allow optimizing inserts from previous extracts.
18578 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
18580 * config/i386/i386.c (ix86_stack_protect_guard): Generate
18581 memory reference to a SSP offset in TLS address space.
18582 (ix86_print_operand) <case '@'>: Remove.
18583 (ix86_print_operand_punct_valid_p): Remove '@' code.
18584 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
18585 UNSPEC_SP_TLS_TEST.
18586 (stack_tls_protect_set_<mode>): Remove.
18587 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
18588 (stack_tls_protect_test_<mode>): Remove.
18589 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
18591 2017-08-07 Olivier Hainque <hainque@adacore.com>
18594 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
18596 2017-08-07 Douglas Rupp <rupp@adacore.com>
18598 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
18599 variable was referenced as multidir in command.
18601 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18604 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
18607 2017-08-07 Martin Liska <mliska@suse.cz>
18609 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
18610 * config/rl78/rl78.c: Add include of attribs.h.
18611 * config/sh/sh.c: Likewise.
18612 * config/v850/v850.c: Likewise.
18614 2017-08-07 Tom de Vries <tom@codesourcery.com>
18616 PR middle-end/78266
18617 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
18619 2017-08-07 Martin Liska <mliska@suse.cz>
18621 * config/mips/mips.c: Include attribs.h.
18623 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
18626 * doc/invoke.texi: Document change in behvaior for -Ofast for
18629 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
18631 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
18633 (aarch64_pop_regs): Likewise.
18634 (aarch64_gen_load_pair): Likewise.
18635 (aarch64_save_callee_saves): Likewise.
18636 (aarch64_restore_callee_saves): Likewise.
18638 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18640 * config/i386/i386.c: Revert the last change.
18642 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18645 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
18647 (ix86_finalize_stack_frame_flags): This. Also clear
18648 frame_pointer_needed if -fno-omit-frame-pointer is used without
18650 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
18651 with ix86_finalize_stack_frame_flags.
18652 (ix86_expand_epilogue): Likewise.
18653 (ix86_expand_split_stack_prologue): Likewise.
18655 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18658 * config/i386/i386.c (get_builtin_code_for_version): Set priority
18659 to P_AES for Westmere.
18661 2017-08-07 Jonathan Yong <10walls@gmail.com>
18663 * config/i386/mingw.opt (fset-stack-executable): Removed.
18664 * config/i386/cygming.opt (fset-stack-executable): Moved
18666 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
18668 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
18670 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
18672 2017-08-07 Marek Polacek <polacek@redhat.com>
18674 PR middle-end/81737
18675 * fold-const.c (fold_indirect_ref_1): Check type_domain.
18677 2017-08-07 Martin Liska <mliska@suse.cz>
18679 * attribs.h (canonicalize_attr_name): New function.
18680 (cmp_attribs): Move from c-format.c and adjusted.
18681 (is_attribute_p): Moved from tree.h.
18682 * tree-inline.c: Add new includes.
18683 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
18684 (private_is_attribute_p): Remove.
18685 (private_lookup_attribute): Likewise.
18686 (private_lookup_attribute_by_prefix): Simplify.
18687 (remove_attribute): Use is_attribute_p.
18688 * tree.h: Remove removed declarations.
18690 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18692 PR middle-end/81698
18693 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
18694 instead of computing it in the function. Formatting fix.
18695 (expand_case): Don't rely on default_edge being the first edge,
18696 clear it if removing it, pass default_edge to
18697 emit_case_dispatch_table.
18698 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
18701 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
18703 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
18704 insn in the function, emit NOP after the insn.
18706 2017-08-06 Tom de Vries <tom@codesourcery.com>
18708 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
18711 2017-08-06 Tom de Vries <tom@codesourcery.com>
18713 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
18716 2017-08-04 Yury Gribov <tetra2005@gmail.com>
18718 PR tree-optimization/57371
18719 * match.pd: New pattern.
18721 2017-08-04 Marek Polacek <polacek@redhat.com>
18723 PR middle-end/81695
18724 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
18725 perform the computation in offset_int.
18727 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18729 PR tree-optimization/81136
18730 * tree-vectorizer.h: Include tree-hash-traits.h.
18731 (vec_base_alignments): New typedef.
18732 (vec_info): Add a base_alignments field.
18733 (vect_record_base_alignments): Declare.
18734 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
18736 (DR_IS_CONDITIONAL_IN_STMT): New macro.
18737 (create_data_ref): Add an is_conditional_in_stmt argument.
18738 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
18739 the is_conditional_in_stmt field.
18740 (data_ref_loc): Add an is_conditional_in_stmt field.
18741 (get_references_in_stmt): Set the is_conditional_in_stmt field.
18742 (find_data_references_in_stmt): Update call to create_data_ref.
18743 (graphite_find_data_references_in_stmt): Likewise.
18744 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
18745 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
18746 (vect_record_base_alignment): New function.
18747 (vect_record_base_alignments): Likewise.
18748 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
18749 for nested statements even if we fail to compute a misalignment.
18750 Use pooled base alignments for unconditional references.
18751 (vect_find_same_alignment_drs): Compare base addresses instead
18753 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
18754 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
18756 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18758 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
18759 Add an explicit name for the enum. Use auto_vec for slp_instances
18760 and grouped_stores.
18761 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
18763 (_bb_vec_info): Add a constructor and destructor.
18764 (vinfo_for_stmt): Return NULL for uids of -1 as well.
18765 (destroy_loop_vec_info): Delete.
18766 (vect_destroy_datarefs): Likewise.
18767 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
18768 (vec_info::vec_info): New function.
18769 (vec_info::~vec_info): Likewise.
18770 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
18771 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
18772 destroy_loop_vec_info.
18773 * tree-vect-loop.c (new_loop_vec_info): Replace with...
18774 (_loop_vec_info::_loop_vec_info): ...this.
18775 (destroy_loop_vec_info): Replace with...
18776 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
18777 the stmt_vec_infos. Leave handling of vec_info information to its
18778 destructor. Remove explicit vector releases.
18779 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
18780 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
18781 * tree-vect-slp.c (new_bb_vec_info): Replace with...
18782 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
18783 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
18784 (destroy_bb_vec_info): Replace with...
18785 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
18786 information to its destructor.
18787 (vect_slp_analyze_bb_1): Use new and delete instead of
18788 new_bb_vec_info and destroy_bb_vec_info.
18789 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
18792 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18794 * tree-data-ref.h (subscript): Add access_fn field.
18795 (data_dependence_relation): Add could_be_independent_p.
18796 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
18797 (same_access_functions): Move to tree-data-ref.c.
18798 * tree-data-ref.c (ref_contains_union_access_p): New function.
18799 (access_fn_component_p): Likewise.
18800 (access_fn_components_comparable_p): Likewise.
18801 (dr_analyze_indices): Add a reference to access_fn_component_p.
18802 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
18804 (constant_access_functions): Likewise.
18805 (add_other_self_distances): Likewise.
18806 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
18807 (initialize_data_dependence_relation): Use XCNEW and remove
18808 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
18809 of access functions that have the same type. Allow the
18810 subsequence to end with different bases in some circumstances.
18811 Record the chosen access functions in SUB_ACCESS_FN.
18812 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
18813 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
18814 (subscript_dependence_tester_1): Likewise dra and drb.
18815 (build_classic_dist_vector): Update calls accordingly.
18816 (subscript_dependence_tester): Likewise.
18817 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
18818 DDR_COULD_BE_INDEPENDENT_P.
18819 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
18820 comp_alias_ddrs instead of may_alias_ddrs.
18821 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
18823 (vect_analyze_data_ref_dependence): Use it if
18824 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
18825 distance vectors if that fails.
18826 (dependence_distance_ge_vf): New function.
18827 (vect_prune_runtime_alias_test_list): Use it. Don't clear
18828 LOOP_VINFO_MAY_ALIAS_DDRS.
18830 2017-08-04 Richard Biener <rguenther@suse.de>
18832 PR middle-end/81705
18833 * fold-const.c (fold_binary_loc): Properly restrict
18834 minus_var0 && minus_var1 case when associating undefined overflow
18837 2017-08-04 Tom de Vries <tom@codesourcery.com>
18839 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
18841 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18843 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18844 Don't start diagnostic messages with a capital letter.
18845 * config/rs6000/rs6000.c (rs6000_option_override_internal):
18847 (rs6000_invalid_builtin): Likewise.
18848 (rs6000_trampoline_init): Likewise.
18850 2017-08-03 Jakub Jelinek <jakub@redhat.com>
18853 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
18854 after setting changeable df flags.
18856 2017-08-03 Richard Biener <rguenther@suse.de>
18858 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
18859 up if the use is in USE - X.
18861 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
18863 * toplev.c (dumpfile.h): New include.
18864 (internal_error_reentered): New static function. Use it...
18865 (internal_error_function): ...here to handle reentered internal_error.
18867 2017-08-03 Richard Biener <rguenther@suse.de>
18869 PR middle-end/81148
18870 * fold-const.c (split_tree): Add minus_var and minus_con
18871 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
18872 here but always use minus_*.
18873 (associate_trees): Assert we never associate with MINUS_EXPR
18874 and NULL first operand. Do not recurse for PLUS_EXPR operands
18875 when associating as MINUS_EXPR either.
18876 (fold_binary_loc): Track minus_var and minus_con.
18878 2017-08-03 Tom de Vries <tom@codesourcery.com>
18881 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
18882 ACCEL_COMPILER, apply finish_options on
18883 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
18885 2017-08-03 Tom de Vries <tom@codesourcery.com>
18888 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
18889 function_entry_patch_area_size > 0.
18891 2017-08-03 Jakub Jelinek <jakub@redhat.com>
18894 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
18895 instead of 10??LU, perform unit multiplication in wide_int,
18896 don't change alloc_object_size_limit if the limit is larger
18899 PR tree-optimization/81655
18900 PR tree-optimization/81588
18901 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
18902 the case when ranges[i].low and high are 1 for unsigned type with
18905 PR middle-end/81052
18906 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
18907 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
18909 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18911 * tree-vrp.h: Add include guard.
18913 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
18916 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
18917 (ud2): New insn pattern.
18918 * config/i386/i386.c (ix86_expand_epilogue):
18919 For naked functions, generate ud2 instead of trap insn.
18921 2017-08-02 Marek Polacek <polacek@redhat.com>
18924 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
18926 2017-08-02 Tom de Vries <tom@codesourcery.com>
18927 Cesar Philippidis <cesar@codesourcery.com>
18929 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
18930 Add missing edge probabilities.
18932 2017-08-02 Tamar Christina <tamar.christina@arm.com>
18934 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
18935 Correct endianness.
18937 2017-08-02 Jakub Jelinek <jakub@redhat.com>
18939 PR middle-end/79499
18940 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
18941 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
18942 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
18943 prologue_seq sequences - if any.
18945 2017-08-02 Richard Biener <rguenther@suse.de>
18947 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
18948 via vectors if supported, integer extracts via punning if supported
18949 or otherwise vector extracts.
18951 2017-08-02 Richard Biener <rguenther@suse.de>
18953 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
18955 (bitmap_insert_into_set): ... this.
18957 2017-08-02 Richard Biener <rguenther@suse.de>
18959 PR tree-optimization/81633
18961 2015-08-17 Alan Hayward <alan.hayward@arm.com>
18963 PR tree-optimization/71752
18964 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
18966 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
18968 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
18969 (machine_function::call_ms2sysv_pad_out): Remove field.
18970 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
18971 (ix86_compute_frame_layout): Likewise.
18973 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
18976 * config/i386/i386.c (ix86_set_func_type): Disallow naked
18977 attribute with interrupt attribute.
18979 2017-08-01 Andrew Pinski <apinski@cavium.com>
18981 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
18982 BIT_INSERT_EXPR's operand 1
18983 to see if the types precision matches.
18985 2017-08-01 Martin Liska <mliska@suse.cz>
18987 PR middle-end/70140
18988 * builtins.c (expand_builtin_memcpy_args): Remove.
18989 (expand_builtin_memcpy): Call newly added function
18990 expand_builtin_memory_copy_args.
18991 (expand_builtin_memcpy_with_bounds): Likewise.
18992 (expand_builtin_mempcpy): Remove last argument.
18993 (expand_builtin_mempcpy_with_bounds): Likewise.
18994 (expand_builtin_memory_copy_args): New function created from
18995 expand_builtin_mempcpy_args with small modifications.
18996 (expand_builtin_mempcpy_args): Remove.
18997 (expand_builtin_stpcpy): Remove unused argument.
18998 (expand_builtin): Likewise.
18999 (expand_builtin_with_bounds): Likewise.
19001 2017-08-01 Martin Liska <mliska@suse.cz>
19004 Make mempcpy more optimal (PR middle-end/70140).
19006 2017-08-01 Jakub Jelinek <jakub@redhat.com>
19009 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
19010 __builtin_vec_cmpne verify both arguments are compatible vectors
19011 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
19012 verify arg1_type is a pointer or array type. For __builtin_vec_st,
19013 move computation of aligned to after checking the argument types.
19017 * config/rs6000/vsx.md (vextract_fp_from_shorth,
19018 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
19021 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
19022 Doug Rupp <rupp@adacore.com>
19023 Olivier Hainque <hainque@adacore.com>
19025 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
19026 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
19028 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
19029 for TARGET_OS_CPP_BUILTIN.
19030 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
19031 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
19033 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
19034 passing required abi options to the assembler for EABI configurations.
19035 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
19036 of .text.hot and .text.unlikely sections for kernel modules when
19037 using ARM style exceptions.
19038 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
19039 options. Add EXTRA_CC1_SPEC.
19040 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
19042 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
19044 (ARM_TARGET2_DWARF_FORMAT): Define.
19045 * config/arm/t-vxworks: Adjust multilib control to removal of the
19046 Diab command line options.
19048 2017-08-01 Martin Liska <mliska@suse.cz>
19050 PR gcov-profile/81561
19051 * gcov.c (unblock): Make unblocking safe as we need to preserve
19052 index correspondence of blocks and block_lists.
19054 2017-08-01 Richard Biener <rguenther@suse.de>
19056 PR tree-optimization/81181
19057 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
19058 (compute_antic): ... end of iteration here.
19060 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
19062 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
19063 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
19064 (ftree-slp-vectorize): Likewise.
19065 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
19066 can no longer be set independent of flag_tree_loop_vectorize.
19067 * omp-general.c (emp_max_vf): Likewise.
19068 * opts.c (enable_fdo_optimizations): Remove references to
19069 flag_tree_vectorize, these are now implicit.
19070 (common_handle_option): Remove handling for OPT_ftree_vectorize,
19071 and leave it for the options machinery.
19073 2017-08-01 Martin Liska <mliska@suse.cz>
19075 PR middle-end/70140
19076 * builtins.c (expand_builtin_memcpy_args): Remove.
19077 (expand_builtin_memcpy): Call newly added function
19078 expand_builtin_memory_copy_args.
19079 (expand_builtin_memcpy_with_bounds): Likewise.
19080 (expand_builtin_mempcpy): Remove last argument.
19081 (expand_builtin_mempcpy_with_bounds): Likewise.
19082 (expand_builtin_memory_copy_args): New function created from
19083 expand_builtin_mempcpy_args with small modifications.
19084 (expand_builtin_mempcpy_args): Remove.
19085 (expand_builtin_stpcpy): Remove unused argument.
19086 (expand_builtin): Likewise.
19087 (expand_builtin_with_bounds): Likewise.
19089 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
19092 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
19093 print "ds:" only for immediates in generic address space.
19095 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
19098 * config/i386/i386.c (ix86_funciton_naked): New prototype.
19099 (ix86_function_ok_for_sibcall): Return false for naked functions.
19101 2017-08-01 Richard Biener <rguenther@suse.de>
19103 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
19104 (compute_antic): Seed worklist with exit block predecessors.
19105 * cfganal.c (dfs_find_deadend): For a cycle return the source
19106 of the edge closing it.
19108 2017-08-01 Tamar Christina <tamar.christina@arm.com>
19110 * config/aarch64/aarch64.c
19111 (aarch64_can_const_movi_rtx_p): Move 0 check.
19113 2017-08-01 Bin Cheng <bin.cheng@arm.com>
19115 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
19116 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
19118 * match.pd: Ditto in address comparison pattern.
19120 2017-08-01 Bin Cheng <bin.cheng@arm.com>
19122 PR tree-optimization/81627
19123 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
19124 closed ssa form for store-store chain.
19126 2017-08-01 Bin Cheng <bin.cheng@arm.com>
19128 PR tree-optimization/81620
19129 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
19130 for store-store chain.
19132 2017-08-01 Jakub Jelinek <jakub@redhat.com>
19134 PR tree-optimization/81588
19135 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
19136 ranges[i].in_p, invert comparison code ccode. For >/>=,
19137 swap rhs1 and rhs2 and comparison code unconditionally,
19138 for </<= don't do that. Don't swap rhs1/rhs2 again if
19139 ranges[i].in_p, instead invert comparison code ccode if
19140 opcode or oe->rank is BIT_IOR_EXPR.
19143 * optabs.def (vec_extract_optab, vec_init_optab): Change from
19144 a direct optab to conversion optab.
19145 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
19146 with GET_MODE_INNER as last argument instead of optab_handler.
19147 * expmed.c (extract_bit_field_1): Likewise. Use vector from
19148 vector extraction if possible and optab is available.
19149 * expr.c (store_constructor): Use convert_optab_handler instead
19150 of optab_handler. Use vector initialization from smaller
19151 vectors if possible and optab is available.
19152 * tree-vect-stmts.c (vectorizable_load): Likewise.
19153 * doc/md.texi (vec_extract, vec_init): Document that the optabs
19154 now have two modes.
19155 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
19156 of vec_init from half-sized vectors with the same element mode.
19157 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
19158 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
19159 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
19160 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
19161 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
19162 after mode in gen_vec_extract* calls.
19163 (vec_extract<mode>): Renamed to ...
19164 (vec_extract<mode><ssescalarmodelower>): ... this.
19165 (vec_extract<mode><ssehalfvecmodelower>): New expander.
19166 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
19167 element mode after mode in gen_vec_init* calls.
19168 (VEC_INIT_HALF_MODE): New mode iterator.
19169 (vec_init<mode>): Renamed to ...
19170 (vec_init<mode><ssescalarmodelower>): ... this.
19171 (vec_init<mode><ssehalfvecmodelower>): New expander.
19172 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
19173 (vec_extractv2sfsf): ... this.
19174 (vec_initv2sf): Renamed to ...
19175 (vec_initv2sfsf): ... this.
19176 (vec_extractv2si): Renamed to ...
19177 (vec_extractv2sisi): ... this.
19178 (vec_initv2si): Renamed to ...
19179 (vec_initv2sisi): ... this.
19180 (vec_extractv4hi): Renamed to ...
19181 (vec_extractv4hihi): ... this.
19182 (vec_initv4hi): Renamed to ...
19183 (vec_initv4hihi): ... this.
19184 (vec_extractv8qi): Renamed to ...
19185 (vec_extractv8qiqi): ... this.
19186 (vec_initv8qi): Renamed to ...
19187 (vec_initv8qiqi): ... this.
19188 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
19189 (vec_init<mode>): Renamed to ...
19190 (vec_init<mode><VEC_base_l>): ... this.
19191 (vec_extract<mode>): Renamed to ...
19192 (vec_extract<mode><VEC_base_l>): ... this.
19193 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
19194 (vec_initv2sfsf): ... this.
19195 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
19196 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
19197 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
19198 element mode after mode in gen_vec_init* calls.
19199 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
19200 (vec_init<mode><Vel>): ... this.
19201 (vec_extract<mode>): Renamed to ...
19202 (vec_extract<mode><Vel>): ... this.
19203 * config/aarch64/iterators.md (Vel): New mode attribute.
19204 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
19205 Add element mode after mode in gen_vec_extract* calls.
19206 * config/s390/vector.md (non_vec_l): New mode attribute.
19207 (vec_extract<mode>): Renamed to ...
19208 (vec_extract<mode><non_vec_l>): ... this.
19209 (vec_init<mode>): Renamed to ...
19210 (vec_init<mode><non_vec_l>): ... this.
19211 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
19212 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
19214 * config/arm/iterators.md (V_elem_l): New mode attribute.
19215 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
19216 (vec_extract<mode><V_elem_l>): ... this.
19217 (vec_extractv2di): Renamed to ...
19218 (vec_extractv2didi): ... this.
19219 (vec_init<mode>): Renamed to ...
19220 (vec_init<mode><V_elem_l>): ... this.
19221 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
19222 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
19223 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
19224 Add element mode after gen_vec_extract* calls.
19225 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
19226 (vec_init<mode><unitmode>): ... this.
19227 (vec_extract<mode>): Renamed to ...
19228 (vec_extract<mode><unitmode>): ... this.
19229 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
19230 (vec_init<mode><unitmode>): ... this.
19231 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
19232 (vec_initv2sfsf): ... this.
19233 (vec_extractv2sf): Renamed to ...
19234 (vec_extractv2sfsf): ... this.
19235 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
19236 Add element mode after gen_vec_extract* calls.
19237 * config/mips/mips.md (unitmode): New mode iterator.
19238 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
19239 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
19240 * config/spu/spu.md (inner_l): New mode attribute.
19241 (vec_init<mode>): Renamed to ...
19242 (vec_init<mode><inner_l>): ... this.
19243 (vec_extract<mode>): Renamed to ...
19244 (vec_extract<mode><inner_l>): ... this.
19245 * config/sparc/sparc.md (veltmode): New mode iterator.
19246 (vec_init<VMALL:mode>): Renamed to ...
19247 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
19248 * config/ia64/vect.md (vec_initv2si): Renamed to ...
19249 (vec_initv2sisi): ... this.
19250 (vec_initv2sf): Renamed to ...
19251 (vec_initv2sfsf): ... this.
19252 (vec_extractv2sf): Renamed to ...
19253 (vec_extractv2sfsf): ... this.
19254 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
19255 (vec_init<mode>): Renamed to ...
19256 (vec_init<mode><VEC_base_l>): ... this.
19257 (vec_extract<mode>): Renamed to ...
19258 (vec_extract<mode><VEC_base_l>): ... this.
19259 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
19260 (vec_initv2sfsf): ... this.
19261 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
19262 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
19263 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
19264 gen_vec_init* calls.
19266 2017-08-01 Richard Biener <rguenther@suse.de>
19268 PR tree-optimization/81297
19269 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
19270 TREE_OVERFLOW from INTEGER_CSTs.
19272 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
19274 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
19276 2017-07-31 Carl Love <cel@us.ibm.com>
19278 * config/rs6000/rs6000-c: Add support for built-in functions
19279 vector signed char vec_xl_be (signed long long, signed char *);
19280 vector unsigned char vec_xl_be (signed long long, unsigned char *);
19281 vector signed int vec_xl_be (signed long long, signed int *);
19282 vector unsigned int vec_xl_be (signed long long, unsigned int *);
19283 vector signed long long vec_xl_be (signed long long, signed long long *);
19284 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
19285 vector signed short vec_xl_be (signed long long, signed short *);
19286 vector unsigned short vec_xl_be (signed long long, unsigned short *);
19287 vector double vec_xl_be (signed long long, double *);
19288 vector float vec_xl_be (signed long long, float *);
19289 * config/rs6000/altivec.h (vec_xl_be): Add #define.
19290 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
19291 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
19293 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
19294 (altivec_expand_builtin): Add switch statement to call
19295 altivec_expand_xl_be for each builtin.
19296 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
19297 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
19298 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
19299 __builtin_vsx_le_be_v16qi.
19300 * doc/extend.texi: Update the built-in documentation file for the
19301 new built-in functions.
19303 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
19306 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
19308 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
19310 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19312 * config.gcc: Add z14.
19313 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
19314 CPU model numbers for z13s and z14.
19315 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
19317 * config/s390/s390-opts.h (enum processor_type): Rename
19318 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
19319 * config/s390/s390.c (processor_table): Add field for CPU name to
19320 be passed to Binutils.
19321 (s390_asm_output_machine_for_arch): Use the new field in
19322 processor_table for Binutils.
19323 (s390_expand_builtin): Replace arch12 with z14.
19324 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
19325 (s390_get_sched_attrmask): Likewise.
19326 (s390_get_unit_mask): Likewise.
19327 * config/s390/s390.opt: Add z14 to processor_type enum.
19329 2017-07-31 Martin Jambor <mjambor@suse.cz>
19332 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
19333 regardless of optimization level.
19335 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
19336 Martin Liska <mliska@suse.cz>
19338 * predict.def: Remove old comment and adjust probability.
19339 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
19340 PREDICT statements.
19342 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
19345 * config/i386/i386.c (ix86_function_naked): New function.
19346 (ix86_can_use_return_insn_p): Return false for naked functions.
19347 (ix86_expand_prologue): Skip prologue for naked functions.
19348 (ix86_expand_epilogue): Skip epilogue for naked functions
19349 and emit trap instruction.
19350 (ix86_warn_func_return): New function.
19351 (ix86_attribute_table): Add "naked" attribute specification.
19352 (TARGET_WARN_FUNC_RETURN): Define.
19353 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
19355 2017-07-31 Martin Liska <mliska@suse.cz>
19357 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
19358 (dump_gimple_bb_header): Always dump BB info.
19359 (pp_cfg_jump): Do not append info about BB when dumping a jump.
19361 2017-07-31 Martin Liska <mliska@suse.cz>
19364 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
19365 also with current_function_decl non-null equality.
19367 2017-07-31 Jakub Jelinek <jakub@redhat.com>
19370 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
19371 change type to the element type, instead add eltype variable and
19372 use it where we are interested in the element type.
19374 PR tree-optimization/81603
19375 * ipa-polymorphic-call.c
19376 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
19377 offset arithmetic in offset_int, bail out if the resulting bit offset
19378 doesn't fit into shwi.
19380 2017-07-31 Martin Liska <mliska@suse.cz>
19382 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
19383 (gimplify_save_expr): Fix comment.
19385 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
19388 * config/i386/i386.c (ix86_function_arg): Update arguments for
19390 (ix86_compute_frame_layout): Set the initial stack offset to
19391 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
19392 INCOMING_FRAME_SP_OFFSET.
19393 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
19394 stack before exception handler returns.
19395 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
19396 the 'ERROR_CODE' for exception handler.
19398 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
19400 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
19401 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
19402 (ASM_OUTPUT_REG_POP): Ditto.
19403 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
19404 instead of asm_fprintf to output pure string.
19406 2017-07-29 Jakub Jelinek <jakub@redhat.com>
19408 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
19409 to imported_module_or_decl hook.
19410 (debug_nothing_tree_tree_tree_bool): Remove.
19411 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
19412 * debug.c (do_nothing_debug_hooks): Use
19413 debug_nothing_tree_tree_tree_bool_bool instead of
19414 debug_nothing_tree_tree_tree_bool.
19415 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19416 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
19417 * sdbout.c (sdb_debug_hooks): Likewise.
19418 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
19419 (gen_namespace_die): Add DW_AT_export_symbols attribute if
19421 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
19422 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
19423 attribute, don't add anything.
19425 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19427 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
19428 (fold_build2_stat_loc): Likewise.
19429 (fold_build3_stat_loc): Likewise.
19430 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
19431 (fold_build1_loc): Remove macro.
19432 (fold_build2_loc): Likewise.
19433 (fold_build3_loc): Likewise.
19435 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19437 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
19438 (gimple_build_debug_bind_source_stat): Likewise.
19439 * gimple.h (gimple_build_debug_bind): Remove macro.
19440 (gimple_build_debug_bind_source): Likewise.
19442 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19444 * bitmap.c (bitmap_alloc): Adjust.
19445 (bitmap_gc_alloc): Likewise.
19446 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
19448 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19450 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
19451 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
19452 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
19453 (bitmap_gc_alloc_stat): Likewise.
19454 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
19456 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19458 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
19459 * rtl.h (shallow_copy_rtx): Remove macro.
19461 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19463 * emit-rtl.c (gen_raw_REG): Adjust.
19464 * gengenrtl.c (gendef): Likewise.
19465 * rtl.c (rtx_alloc_stat): Remove _stat from name.
19466 * rtl.h (rtx_alloc): Remove macro.
19468 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19470 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
19471 (build_tree_list_stat): Likewise.
19472 * tree.h (build_tree_list): Remove macro.
19473 (build_tree_list_vec): Likewise.
19475 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19477 * tree.c (make_vector_stat): Remove _stat from name.
19478 (build_vector_stat): Likewise.
19479 * tree.h (make_vector_stat): Remove macro.
19480 (build_vector_stat): Likewise.
19482 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19484 * tree.h (build_var_debug_value): Remove prototype.
19486 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19488 * tree.c (tree_cons_stat): Remove _stat from name.
19489 * tree.h (tree_cons): Remove macro.
19491 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19493 * tree.c (build_vl_exp_stat): Remove _stat from name.
19494 * tree.h (build_vl_exp): Remove macro.
19496 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19498 * tree.c (build_decl_stat): Remove _stat from name.
19499 * tree.h (build_decl): Remove macro.
19501 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19503 * gimple.c (gimple_build_with_ops_stat): Adjust.
19504 (gimple_alloc_stat): Remove _stat from name.
19505 * gimple.h (gimple_alloc): Remove macro.
19507 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19509 * tree.c (make_tree_vec_stat): Remove _stat from name.
19510 (grow_tree_vec_stat): Likewise.
19511 * tree.h (make_tree_vec_stat): Adjust prototype.
19512 (grow_tree_vec_stat): Likewise.
19513 (make_tree_vec): Remove macro.
19514 (grow_tree_vec): Likewise.
19516 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19518 * fold-const.c (fold_build1_stat_loc): Adjust.
19519 (fold_build2_stat_loc): Likewise.
19520 (fold_build3_stat_loc): Likewise.
19521 * tree.c (build0_stat): Remove _stat from name.
19522 (build1_stat): Likewise.
19523 (build2_stat): Likewise.
19524 (build3_stat): Likewise.
19525 (build4_stat): Likewise.
19526 (build5_stat): Likewise.
19527 * tree.h (build1_loc): Remove macro, and rename _stat function
19529 (build2_loc): Likewise.
19530 (build3_loc): Likewise.
19531 (build4_loc): Likewise.
19532 (build5_loc): Likewise.
19534 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19536 * tree.c (make_int_cst_stat): Remove _stat from name.
19537 * tree.h (make_int_cst_stat): Adjust prototype.
19538 (make_int_cst): Remove macro.
19540 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19542 * tree.c (make_tre_binfo_stat): Remove _stat from name.
19543 * tree.h (make_tree_binfo_stat): Adjust prototype.
19544 (make_tree_binfo): Remove.
19546 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19548 * tree.c (copy_node_stat): Rename to copy_node.
19549 (build_distinct_type_copy): Adjust.
19550 * tree.h (copy_node_stat): Adjust prototype.
19551 (copy_node): Remove macro.
19553 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19555 * tree.c (make_node_stat): rename to make_node.
19556 (build_tree_list_stat): Adjust.
19557 (build0_stat): Likewise.
19558 (build2_stat): Likewise.
19559 (build3_stat): Likewise.
19560 (build4_stat): Likewise.
19561 (build5_stat): Likewise.
19562 (build_decl_stat): Likewise.
19563 * tree.h (make_node_stat): Adjust prototype.
19564 (make_node): remove macro.
19566 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
19568 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
19569 (PPC_FEATURE2_SCV): Likewise.
19570 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
19572 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19574 * config/aarch64/aarch64.c
19575 (aarch64_internal_mov_immediate): Add new special pattern.
19576 * config/aarch64/aarch64.md (*movdi_aarch64):
19577 Add reg/32bit const mov case.
19579 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19580 Richard Sandiford <richard.sandiford@linaro.org>
19582 * config/aarch64/aarch64.md (mov<mode>): Generalize.
19583 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
19584 Add integer and movi cases.
19585 (movi-split-hf-df-sf split, fp16): New.
19586 (enabled): Added TARGET_FP_F16INST.
19587 * config/aarch64/iterators.md (GPF_HF): New.
19588 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
19590 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19592 * config/aarch64/aarch64.c
19593 (aarch64_simd_container_mode): Add prototype.
19594 (aarch64_expand_mov_immediate): Add HI support.
19595 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
19596 (aarch64_can_const_movi_rtx_p): New.
19597 (aarch64_preferred_reload_class):
19598 Remove restrictions of using FP registers for certain SIMD operations.
19599 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
19600 (aarch64_valid_floating_const): Add integer move validation.
19601 (aarch64_simd_imm_scalar_p): Remove.
19602 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
19603 (aarch64_legitimate_constant_p): Expand list of supported cases.
19604 * config/aarch64/aarch64-protos.h
19605 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
19606 (aarch64_reinterpret_float_as_int): New.
19607 (aarch64_simd_imm_scalar_p): Remove.
19608 * config/aarch64/constraints.md (Uvi): New.
19609 (Dd): Split into Ds and new Dd.
19610 * config/aarch64/aarch64.md (*movsi_aarch64):
19612 (*movdi_aarch64): Add SIMD mov case.
19614 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19616 * tree-predcom.c: (struct chain): Handle store-store chain in which
19617 stores for elimination only store loop invariant values.
19618 (execute_pred_commoning_chain): Ditto.
19619 (prepare_initializers_chain_store_elim): Ditto.
19620 (prepare_finalizers): Ditto.
19621 (is_inv_store_elimination_chain): New function.
19622 (initialize_root_vars_store_elim_1): New function.
19624 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19626 * tree-predcom.c: Revise general description of the pass.
19627 (enum chain_type): New enum type for store elimination.
19628 (struct chain): New field supporting store elimination.
19629 (struct component): Ditto.
19630 (dump_chain): Dump store-stores chain.
19631 (release_chain): Release resources.
19632 (split_data_refs_to_components): Compute and create component
19633 contains only stores for elimination.
19634 (get_chain_last_ref_at): New function.
19635 (make_invariant_chain): Initialization.
19636 (make_rooted_chain): Specify chain type in parameter and record it.
19637 (add_looparound_copies): Skip for store-stores chain.
19638 (determine_roots_comp): Compute type of chain and pass it to
19640 (initialize_root_vars_store_elim_2): New function.
19641 (finalize_eliminated_stores): New function.
19642 (remove_stmt): Handle store for elimination.
19643 (execute_pred_commoning_chain): Execute predictive commoning on
19644 store-store chains.
19645 (determine_unroll_factor): Skip unroll for store-stores chain.
19646 (prepare_initializers_chain_store_elim): New function.
19647 (prepare_initializers_chain): Hanlde store-store chain.
19648 (prepare_finalizers_chain, prepare_finalizers): New function.
19649 (tree_predictive_commoning_loop): Return integer value indicating
19650 if loop is unrolled or lcssa form is corrupted.
19651 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
19653 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19655 * tree-predcom.c (initialize_root): Delete.
19656 (execute_pred_commoning_chain): Initialize root vars and replace
19657 reference of non-combined chain directly, rather than call above
19660 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19662 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
19663 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
19665 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19667 * tree-predcom.c (struct chain): New field init_seq.
19668 (release_chain): Release init_seq.
19669 (prepare_initializers_chain): Record intialization stmts in above
19671 (insert_init_seqs): New function.
19672 (tree_predictive_commoning_loop): Call insert_init_seqs.
19674 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19676 * tree-predcom.c (determine_roots_comp): Skip trivial components.
19678 2017-07-28 Richard Biener <rguenther@suse.de>
19680 * match.pd: Remove superfluous :c.
19681 * genmatch.c (simplify::id): Add member.
19682 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
19684 (current_id): New global.
19685 (dt_node::parent): Move from ...
19686 (dt_operand::parent): ... here. Add for_id member.
19687 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
19688 (decision_tree::find_node): Relax order requirement when
19689 merging DT_TRUE nodes to ones inbetween the current simplify
19690 and the one we try to merge with. Add diagnostic whenever
19691 we need to enforce pattern order by not merging.
19692 (decision_tree::insert): Set current_id.
19693 (decision_tree::print_node): Dump parent node and for_id.
19694 (parser::last_id): Add member.
19695 (parser::push_simplify): Assign unique id.
19696 (parser::parser): Initialize last_id.
19698 2017-07-28 Martin Liska <mliska@suse.cz>
19701 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
19702 gimple_build_debug_bind.
19704 2017-07-28 Richard Biener <rguenther@suse.de>
19706 PR tree-optimization/81502
19707 * match.pd: Add pattern combining BIT_INSERT_EXPR with
19709 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
19711 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
19712 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
19713 for BIT_FIELD_REF args.
19714 * fold-const.c (make_bit_field_ref): Likewise.
19715 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
19717 2017-07-28 Jakub Jelinek <jakub@redhat.com>
19720 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
19721 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
19722 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
19723 Or it into SANITIZER_UNDEFINED.
19724 * ubsan.c: Include gimple-fold.h and varasm.h.
19725 (ubsan_expand_ptr_ifn): New function.
19726 (instrument_pointer_overflow): New function.
19727 (maybe_instrument_pointer_overflow): New function.
19728 (instrument_object_size): Formatting fix.
19729 (pass_ubsan::execute): Call instrument_pointer_overflow
19730 and maybe_instrument_pointer_overflow.
19731 * internal-fn.c (expand_UBSAN_PTR): New function.
19732 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
19733 * sanitizer.def (__ubsan_handle_pointer_overflow,
19734 __ubsan_handle_pointer_overflow_abort): New builtins.
19735 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
19736 * internal-fn.def (UBSAN_PTR): New internal function.
19737 * opts.c (sanitizer_opts): Add pointer-overflow.
19738 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
19739 * fold-const.c (build_range_check): Compute pointer range check in
19740 integral type if pointer arithmetics would be needed. Formatting
19743 2017-07-28 Martin Liska <mliska@suse.cz>
19746 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
19747 parameters that are of a variable-length.
19749 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
19751 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
19753 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
19754 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
19755 (CRT_CALL_STATIC_FUNCTION): Likewise.
19756 (ASM_DEFAULT_SPEC): New define.
19757 (ASM_SPEC32): Likewise.
19758 (ASM_SPEC64): Likewise.
19759 (ASM_SPEC_COMMON): Likewise.
19760 (ASM_SPEC): Likewise.
19761 (INVALID_64BIT): Likewise.
19762 (LINK_OS_DEFAULT_SPEC): Likewise.
19763 (LINK_OS_SPEC32): Likewise.
19764 (LINK_OS_SPEC64): Likewise.
19765 (POWERPC_LINUX): Likewise.
19766 (PTRDIFF_TYPE): Likewise.
19767 (RESTORE_FP_PREFIX): Likewise.
19768 (RESTORE_FP_SUFFIX): Likewise.
19769 (SAVE_FP_PREFIX): Likewise.
19770 (SAVE_FP_SUFFIX): Likewise.
19771 (SIZE_TYPE): Likewise.
19772 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
19773 (TARGET_64BIT): Likewise.
19774 (TARGET_64BIT): Likewise.
19775 (TARGET_AIX): Likewise.
19776 (WCHAR_TYPE_SIZE): Likewise.
19777 (WCHAR_TYPE): Undefine.
19778 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
19779 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
19780 (CPP_OS_RTEMS_SPEC): Delete.
19781 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
19782 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
19784 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
19786 2017-07-28 Jakub Jelinek <jakub@redhat.com>
19788 PR tree-optimization/81578
19789 * tree-parloops.c (build_new_reduction): Bail out if
19790 reduction_code isn't one of the standard OpenMP reductions.
19791 Move the details printing after that decision.
19793 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19795 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
19796 related to reload_in_progress.
19797 (splat_input_operand): Likewise.
19798 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
19800 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
19802 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
19803 (TARGET_INSTANTIATE_DECLS): Likewise.
19804 (legitimate_indexed_address_p): Delete reload_in_progress code.
19805 (rs6000_debug_legitimate_address_p): Likewise.
19806 (rs6000_eliminate_indexed_memrefs): Likewise.
19807 (rs6000_emit_le_vsx_store): Likewise.
19808 (rs6000_emit_move_si_sf_subreg): Likewise.
19809 (rs6000_emit_move): Likewise.
19810 (register_to_reg_type): Likewise.
19811 (rs6000_pre_atomic_barrier): Likewise.
19812 (rs6000_machopic_legitimize_pic_address): Likewise.
19813 (rs6000_allocate_stack_temp): Likewise.
19814 (rs6000_address_for_fpconvert): Likewise.
19815 (rs6000_address_for_altivec): Likewise.
19816 (rs6000_secondary_memory_needed_rtx): Delete function.
19817 (rs6000_check_sdmode): Likewise.
19818 (rs6000_alloc_sdmode_stack_slot): Likewise.
19819 (rs6000_instantiate_decls): Likewise.
19820 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
19821 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
19822 Delete reload_in_progress.
19823 (*vec_reload_and_plus_<mptrsize>): Likewise.
19824 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
19825 (vsx_div_v2di): Likewise.
19826 (vsx_udiv_v2di): Likewise.
19828 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19830 * config/rs6000/rs6000.opt (mlra): Replace with stub.
19831 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
19832 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
19833 (rs6000_debug_reg_global): Delete print of LRA status.
19834 (rs6000_option_override_internal): Delete dead LRA related code.
19835 (rs6000_lra_p): Delete function.
19836 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
19838 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
19840 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
19841 * config/riscv/rtems.h: New file.
19843 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19844 Sudakshina Das <sudi.das@arm.com>
19846 * config/aarch64/aarch64.md
19847 (define_split for and<mode>3nr_compare): Move
19848 non aarch64_logical_operand to a register.
19849 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
19850 register immediate operand to a register.
19851 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
19853 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19855 PR middle-end/81564
19856 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
19858 2017-07-27 Richard Biener <rguenther@suse.de>
19860 PR tree-optimization/81573
19861 PR tree-optimization/81494
19862 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
19863 multi defuse cycle case.
19865 2017-07-27 Richard Biener <rguenther@suse.de>
19867 PR tree-optimization/81571
19868 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
19871 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
19873 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
19874 earlier and only if MASK_FPU is set. Adjust formatting.
19876 2017-07-27 Martin Liska <mliska@suse.cz>
19878 * opt-functions.awk: Add validation of value of Init.
19879 * optc-gen.awk: Pass new argument.
19881 2017-07-27 Martin Liska <mliska@suse.cz>
19883 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
19884 Fix wrong condition.
19886 2017-07-27 Martin Liska <mliska@suse.cz>
19888 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
19889 BBs and edges seen by autoFDO.
19891 2017-07-27 Richard Biener <rguenther@suse.de>
19893 PR tree-optimization/81502
19894 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
19895 with incompatible but same sized type.
19896 (execute_update_addresses_taken): Likewise.
19898 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
19900 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
19901 flag_tree_loop_vectorize rather than flag_tree_vectorize.
19903 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19906 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
19907 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
19908 Change s_operand to memory_operand.
19910 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
19912 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
19913 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
19914 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
19915 Emit instructions rather than returning an expression. Handle TFmode
19916 and KFmode by casting to TImode.
19917 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
19918 (rs6000_emit_le_vsx_store): Likewise.
19919 * config/rs6000/vsx.md (VSX_TI): New iterator.
19920 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
19921 (*vsx_le_undo_permute_<mode>): Likewise.
19922 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
19923 emit the split sequence.
19924 (*vsx_le_perm_store_<mode>): Likewise.
19926 2017-07-27 Jakub Jelinek <jakub@redhat.com>
19928 PR tree-optimization/81555
19929 PR tree-optimization/81556
19930 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
19931 if true, force CHANGED for the recursive invocation.
19932 (reassociate_bb): Remember original length of ops array, pass
19933 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
19935 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
19936 attributes for noipa attribute. For naked attribute use
19937 lookup_attribute first before lookup_attribute_spec.
19938 * final.c (rest_of_handle_final): Disable IPA RA for functions with
19940 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
19941 for functions with noipa attribute.
19942 (cgraph_externally_visible_p): Return true for functions with noipa
19944 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
19945 for functions with noipa attribute.
19946 * doc/extend.texi: Document noipa function attribute.
19947 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
19948 also for functions with noipa attribute.
19949 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
19951 2017-07-26 Andrew Pinski <apinski@cavium.com>
19953 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
19954 vec_unalign_load_cost and vec_unalign_store_cost.
19956 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
19958 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
19959 -mvsx-small-integer option.
19960 (ISA_3_0_MASKS_IEEE): Likewise.
19961 (OTHER_VSX_VECTOR_MASKS): Likewise.
19962 (POWERPC_MASKS): Likewise.
19963 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
19964 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
19965 code, only testing for DImode being allowed in non-VSX floating
19967 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
19968 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
19970 (rs6000_option_override_internal): Delete -mvsx-small-integer.
19971 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
19972 TARGET_P8_VECTOR test.
19973 (rs6000_secondary_reload_simple_move): Likewise.
19974 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
19975 since TARGET_P9_VECTOR was already tested.
19976 (rs6000_opt_masks): Remove -mvsx-small-integer.
19977 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
19978 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19980 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
19981 test for TARGET_VEXTRACTUB was used, and that uses
19983 (p9 extract splitter): Likewise.
19984 (vsx_extract_<mode>_di_p9): Likewise.
19985 (vsx_extract_<mode>_store_p9): Likewise.
19986 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
19987 for TARGET_P9_VECTOR was used. Delete code that is now dead with
19988 the elimination of TARGET_VSX_SMALL_INTEGER.
19989 (vsx_extract_<mode>_p8): Likewise.
19990 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
19991 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
19992 (vsx_set_<mode>_p9): Likewise.
19993 (vsx_set_v4sf_p9): Likewise.
19994 (vsx_set_v4sf_p9_zero): Likewise.
19995 (vsx_insert_extract_v4sf_p9): Likewise.
19996 (vsx_insert_extract_v4sf_p9_2): Likewise.
19997 * config/rs6000/rs6000.md (sign extend splitter): Change
19998 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
19999 (floatsi<mode>2_lfiwax_mem): Likewise.
20000 (floatunssi<mode>2_lfiwzx_mem): Likewise.
20001 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
20002 since a test for TARGET_P9_VECTOR was used.
20003 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
20004 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
20005 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
20006 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
20007 TARGET_P8_VECTOR test.
20008 (fix_trunc<mode>si2_stfiwx): Likewise.
20009 (fix_trunc<mode>si2_internal): Likewise.
20010 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
20011 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
20013 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
20014 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
20015 TARGET_P8_VECTOR test.
20016 (fixuns_trunc<mode>si2_stfiwx): Likewise.
20017 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
20018 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
20020 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
20021 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
20022 since a test for TARGET_P9_VECTOR was used.
20023 (splitter for loading small constants): Likewise.
20025 2017-07-26 Andrew Pinski <apinski@cavium.com>
20027 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
20030 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
20033 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
20034 (fp_valid_at): Likewise.
20036 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
20038 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
20039 (qdf24xx_addrcost_table): Likewise.
20040 (cortexa57_tunings): Update to use generic_branch_cost.
20041 (cortexa72_tunings): Likewise.
20042 (cortexa73_tunings): Likewise.
20043 (qdf24xx_tunings): Likewise.
20045 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
20047 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
20048 (thunderx2t99_branch_cost): Likewise.
20049 (cortexa35_tunings): Update to use generic_branch_cost.
20050 (cortexa53_tunings): Likewise.
20051 (cortexa57_tunings): Likewise.
20052 (cortexa72_tunings): Likewise.
20053 (cortexa73_tunings): Likewise.
20054 (thunderx2t99_tunings): Likewise.
20056 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
20058 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
20059 (sparc_option_override): Honour MASK_FSMULD.
20060 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
20061 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
20062 * config/sparc/sparc.opt (mfsmuld): New option.
20063 * doc/invoke.texi (mfsmuld): Document option.
20065 2017-07-26 Marek Polacek <polacek@redhat.com>
20067 PR middle-end/70992
20068 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
20070 2017-07-26 Richard Biener <rguenther@suse.de>
20072 * gimple-match-head.c (do_valueize): Return OP if valueize
20074 (get_def): New helper to get at the def stmt of a SSA name
20075 if valueize allows.
20076 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
20077 do_valueize to get at the def stmt.
20078 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
20080 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
20082 PR middle-end/46932
20083 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
20085 2017-07-26 Martin Liska <mliska@suse.cz>
20088 * function.c (expand_function_start): Make expansion of
20089 nonlocal_goto_save_area after parm_birth_insn.
20091 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
20093 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
20094 from all CPU target flags enable members.
20096 2017-07-26 Richard Biener <rguenther@suse.de>
20098 * genmatch.c (dt_simplify::gen): Make iterator vars const.
20099 (decision_tree::gen): Make 'type' const.
20100 (write_predicate): Likewise.
20102 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
20104 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
20105 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
20106 (rs6000_option_override_internal): Likewise.
20107 (rs6000_expand_vector_set): Likewise.
20108 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
20109 (TARGET_UPPER_REGS_SF): Likewise.
20110 (TARGET_UPPER_REGS_DI): Likewise.
20111 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
20112 (TARGET_DIRECT_MOVE_64BIT): Likewise.
20113 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
20114 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
20115 (Splitters for DI constants in Altivec registers): Likewise.
20116 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
20117 (vsx_set_v4sf_p9): Likewise.
20118 (vsx_set_v4sf_p9_zero): Likewise.
20119 (vsx_insert_extract_v4sf_p9): Likewise.
20120 (vsx_insert_extract_v4sf_p9_2): Likewise.
20122 2017-07-25 Carl Love <cel@us.ibm.com>
20124 * doc/extend.texi: Update the built-in documentation file for the
20125 existing built-in functions
20126 vector signed char vec_cnttz (vector signed char);
20127 vector unsigned char vec_cnttz (vector unsigned char);
20128 vector signed short vec_cnttz (vector signed short);
20129 vector unsigned short vec_cnttz (vector unsigned short);
20130 vector signed int vec_cnttz (vector signed int);
20131 vector unsigned int vec_cnttz (vector unsigned int);
20132 vector signed long long vec_cnttz (vector signed long long);
20133 vector unsigned long long vec_cnttz (vector unsigned long long);
20135 2017-07-25 Andrew Pinski <apinski@cavium.com>
20137 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
20138 accesses where the use is for the first operand of a BIT_INSERT.
20140 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
20143 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
20144 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
20146 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
20148 * config/i386/gstabs.h: Delete.
20149 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
20151 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
20153 * config/i386/i386.c (ix86_decompose_address): Do not check for
20154 register RTX when looking at index_reg or base_reg.
20155 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
20157 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
20159 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
20160 to update EH info here.
20162 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
20164 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
20166 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
20168 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
20170 2017-07-25 Torsten Duwe <duwe@suse.de>
20172 * common.opt: Introduce -fpatchable-function-entry
20173 command line option, and its variables function_entry_patch_area_size
20174 and function_entry_patch_area_start.
20175 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
20176 including a two-value parser.
20177 * target.def (print_patchable_function_entry): New target hook.
20178 * targhooks.h (default_print_patchable_function_entry): New function.
20179 * targhooks.c (default_print_patchable_function_entry): Likewise.
20180 * toplev.c (process_options): Switch off IPA-RA if
20181 patchable function entries are being generated.
20182 * varasm.c (assemble_start_function): Look at the
20183 patchable-function-entry command line switch and current
20184 function attributes and maybe generate NOP instructions by
20185 calling the print_patchable_function_entry hook.
20186 * doc/extend.texi: Document patchable_function_entry attribute.
20187 * doc/invoke.texi: Document -fpatchable_function_entry
20188 command line option.
20189 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
20191 * doc/tm.texi: Re-generate.
20193 2017-07-25 Jakub Jelinek <jakub@redhat.com>
20196 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
20197 TARGET_AVX512DQ rather than TARGET_AVX512BW.
20199 2017-07-25 Tamar Christina <tamar.christina@arm.com>
20201 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
20203 2017-07-25 Richard Biener <rguenther@suse.de>
20205 PR tree-optimization/81455
20206 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
20207 not walk in cycles when looking for guards.
20209 2017-07-25 Richard Biener <rguenther@suse.de>
20211 PR tree-optimization/81529
20212 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
20213 when optimizing backedge uses.
20215 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
20217 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
20219 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
20220 to dl_section_ref. On AIX, append an expression to subtract
20221 the size of the section length to dl_section_ref.
20223 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
20225 * configure.ac: If any of the config.* scripts fail, exit 1.
20226 * configure: Regenerate.
20228 2017-07-25 Richard Biener <rguenther@suse.de>
20230 PR middle-end/81546
20231 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
20232 of immediate uses, be more verbose on errors.
20234 2017-07-25 Richard Biener <rguenther@suse.de>
20236 PR tree-optimization/81510
20237 * tree-vect-loop.c (vect_is_simple_reduction): When the
20238 reduction stmt is not inside the loop bail out.
20240 2017-07-25 Richard Biener <rguenther@suse.de>
20242 PR tree-optimization/81303
20243 * tree-vect-loop-manip.c (vect_loop_versioning): Build
20244 profitability check against LOOP_VINFO_NITERSM1.
20246 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
20248 * domwalk.c (cmp_bb_postorder): Simplify.
20249 (sort_bbs_postorder): New function. Use it...
20250 (dom_walker::walk): ...here to optimize common cases.
20252 2017-07-25 Martin Liska <mliska@suse.cz>
20255 * ipa-visibility.c (function_and_variable_visibility): Make the
20256 redirection just on target that supports aliasing.
20257 Fix GNU coding style.
20259 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
20262 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
20264 * config/arm/rtems.h: Likewise.
20265 * config/bfin/rtems.h: Likewise.
20266 * config/i386/rtemself.h: Likewise.
20267 * config/lm32/rtems.h: Likewise.
20268 * config/m32c/rtems.h: Likewise.
20269 * config/m68k/rtemself.h: Likewise.
20270 * config/microblaze/rtems.h: Likewise.
20271 * config/mips/rtems.h: Likewise.
20272 * config/moxie/rtems.h: Likewise.
20273 * config/nios2/rtems.h: Likewise.
20274 * config/powerpcspe/rtems.h: Likewise.
20275 * config/rs6000/rtems.h: Likewise.
20276 * config/rtems.h: Likewise.
20277 * config/sh/rtems.h: Likewise.
20278 * config/sh/rtemself.h: Likewise.
20279 * config/sparc/rtemself.h: Likewise.
20281 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
20284 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
20285 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
20286 * tree-ssa-structalias.c (alias_get_name): Same.
20288 2017-07-25 Bin Cheng <bin.cheng@arm.com>
20291 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
20292 instructions if no du chain is found.
20294 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
20296 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
20298 2017-07-25 Richard Biener <rguenther@suse.de>
20300 PR middle-end/81505
20301 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
20304 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
20306 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
20307 upper-regs options.
20308 (ISA_2_7_MASKS_SERVER): Likewise.
20309 (ISA_3_0_MASKS_IEEE): Likewise.
20310 (OTHER_P8_VECTOR_MASKS): Likewise.
20311 (OTHER_VSX_VECTOR_MASKS): Likewise.
20312 (POWERPC_MASKS): Likewise.
20313 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
20314 duplicate list of options.
20315 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
20316 explicit -mupper-regs options.
20317 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
20318 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
20319 alias for -mupper-regs-df.
20320 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
20321 (rs6000_init_hard_regno_mode_ok): Likewise.
20322 (rs6000_option_override_internal): Likewise.
20323 (rs6000_opt_masks): Likewise.
20324 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
20325 options in terms of whether -mvsx or -mpower8-vector was used.
20326 (TARGET_UPPER_REGS_DI): Likewise.
20327 (TARGET_UPPER_REGS_SF): Likewise.
20328 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
20329 -mupper-regs-* options.
20331 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
20333 * passes.c (emergency_dump_function): Print some empty lines and a
20334 header before the RTL dump.
20336 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
20338 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
20340 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
20343 * config/aarch64/aarch64.c (aarch64_classify_symbol):
20344 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
20346 2017-07-24 Carl Love <cel@us.ibm.com>
20348 * config/rs6000/rs6000-c.c: Add support for built-in functions
20349 vector float vec_extract_fp32_from_shorth (vector unsigned short);
20350 vector float vec_extract_fp32_from_shortl (vector unsigned short);
20351 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
20352 vec_extract_fp_from_shortl): Add defines for the two builtins.
20353 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
20354 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
20356 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
20357 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
20358 * doc/extend.texi: Update the built-in documentation file for the
20359 new built-in function.
20361 2017-07-24 Jakub Jelinek <jakub@redhat.com>
20364 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
20366 * doc/generic.texi: Likewise.
20367 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
20368 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
20370 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
20372 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
20373 (aarch64_mls_elt_merge<mode>): Likewise.
20375 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
20377 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
20378 having __cxa_atexit.
20380 2017-07-23 Michael Collison <michael.collison@arm.com>
20382 * config/arm/arm.c (arm_option_override): Deprecate
20383 use of -mstructure-size-boundary.
20384 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
20385 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
20387 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20390 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
20391 Reduce cost estimate for direct moves.
20393 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
20396 * config/i386/i386.c (ix86_option_override_internal): Disable
20397 BMI, BMI2 and TBM instructions for -m16.
20399 2017-07-21 Carl Love <cel@us.ibm.com>
20401 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20402 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20403 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20404 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20405 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20406 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20407 VMULOSW): New enum "unspec" values.
20408 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20409 altivec_vmulosw): New patterns.
20410 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20411 VMULOSW): Add definitions.
20413 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
20415 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
20416 (qdf24xx): Likewise.
20417 * config/aarch64/aarch64-options-extensions.def (rdma); New.
20418 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
20419 (AARCH64_FL_V8_1): Renumber.
20420 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
20421 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
20422 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
20423 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
20424 rdma to feature modifiers list.
20426 2017-07-21 Yury Gribov <tetra2005@gmail.com>
20428 PR middle-end/56727
20429 * ipa-visibility (function_and_variable_visibility): Convert
20430 recursive PLT call to direct call if appropriate.
20432 2017-07-21 Andrew Pinski <apinski@cavium.com>
20434 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
20435 operand 1 to see if the types precision matches.
20436 * fold-const.c (operand_equal_p): Likewise.
20438 2017-07-21 Richard Biener <rguenther@suse.de>
20440 PR tree-optimization/81303
20441 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
20442 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
20443 (vect_peeling_hash_get_lowest_cost): Adjust.
20444 (vect_enhance_data_refs_alignment): Likewise. Use
20445 vect_get_peeling_costs_all_drs to compute the penalty for no
20446 peeling to match up costs.
20448 2017-07-21 Richard Biener <rguenther@suse.de>
20450 PR tree-optimization/81500
20451 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
20452 we didn't identify a reduction path.
20454 2017-07-21 Tom de Vries <tom@codesourcery.com>
20455 Cesar Philippidis <cesar@codesourcery.com>
20457 PR gcov-profile/81442
20458 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
20461 2017-07-21 Tom de Vries <tom@codesourcery.com>
20464 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
20466 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
20467 nvptx_override_options_after_change.
20469 2017-07-21 Ulrich Drepper <drepper@redhat.com>
20471 * dwarf2out.c (output_file_names): Avoid double testing for
20472 dwarf_version >= 5.
20474 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
20476 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
20478 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20480 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
20482 (try_crossjump_to_edge): Do not punt on partitioned functions.
20484 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20486 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
20487 Put all BBs reachable only via paths crossing cold region to cold
20489 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
20491 2016-07-21 Richard Biener <rguenther@suse.de>
20493 PR tree-optimization/81303
20494 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
20495 into account prologue and epilogue iterations when raising
20496 min_profitable_iters to sth at least covering one vector iteration.
20498 2017-07-21 Tamar Christina <tamar.christina@arm.com>
20500 * config/arm/arm.c (arm_test_cpu_arch_dat):
20503 2017-07-20 Nathan Sidwell <nathan@acm.org>
20505 Remove TYPE_METHODS.
20506 * tree.h (TYPE_METHODS): Delete.
20507 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
20508 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
20509 (dbxout_type_methods): Scan TYPE_FIELDS.
20510 (dbxout_type): Don't check TYPE_METHODS here.
20511 * function.c (use_register_for_decl): Always ignore register for
20512 class types when not optimizing.
20513 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
20514 * tree.c (free_lang_data_in_type): Stitch out member functions and
20515 templates from TYPE_FIELDS.
20516 (build_distinct_type_copy, verify_type_variant,
20517 verify_type): Member fns are on TYPE_FIELDS.
20518 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
20519 * tree-pretty-print.c (dump_generic_node): Likewise.
20521 2017-07-20 Jakub Jelinek <jakub@redhat.com>
20524 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
20525 V2TImode and V4TImode.
20526 (ix86_expand_vector_extract): Likewise.
20527 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
20528 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
20529 (ssescalarmode): Handle V4TImode and V2TImode.
20530 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
20531 (*vec_extractv2ti, *vec_extractv4ti): New insns.
20532 (VEXTRACTI128_MODE): New mode iterator.
20533 (splitter for *vec_extractv?ti first element): New.
20534 (VEC_INIT_MODE): New mode iterator.
20535 (vec_init<mode>): Consolidate 3 expanders into one using
20536 VEC_INIT_MODE mode iterator.
20538 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20540 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
20541 non_spilled_static_chain_regno_p.
20543 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20545 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
20547 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
20549 * bb-reorder.c (connect_traces): Allow copying of blocks within
20552 2017-07-20 Richard Biener <rguenther@suse.de>
20554 * gimple.h (gimple_phi_result): Add gphi * overload.
20555 (gimple_phi_result_ptr): Likewise.
20556 (gimple_phi_arg): Likewise. Adjust index assert to only
20557 allow actual argument accesses rather than all slots available
20559 (gimple_phi_arg_def): Add gphi * overload.
20560 * tree-phinodes.c (make_phi_node): Initialize only actual
20562 (resize_phi_node): Clear memory not covered by old node,
20563 do not initialize excess argument slots.
20564 (reserve_phi_args_for_new_edge): Initialize new argument slot
20567 2017-07-20 Bin Cheng <bin.cheng@arm.com>
20569 PR tree-optimization/81388
20571 2016-07-21 Bin Cheng <bin.cheng@arm.com>
20573 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
20574 by removing computation of may_be_zero.
20576 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20577 Tom de Vries <tom@codesourcery.com>
20579 PR middle-end/81030
20580 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
20581 when gimple level profile disagrees with what RTL expander did.
20583 2017-07-20 Richard Biener <rguenther@suse.de>
20585 PR tree-optimization/61171
20586 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
20587 (vect_analyze_stmt): Add slp instance parameter.
20588 (vectorizable_reduction): Likewise.
20589 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
20590 (vect_is_simple_reduction): Deal with chains not detected
20591 as SLP reduction chain, specifically not properly associated
20592 chains containing a mix of plus/minus.
20593 (get_reduction_op): Remove.
20594 (get_initial_defs_for_reduction): Simplify, pass in whether
20595 this is a reduction chain, pass in the SLP node for the PHIs.
20596 (vect_create_epilog_for_reduction): Get the SLP instance as
20598 (vectorizable_reduction): Get the SLP instance as arg.
20599 During analysis remember the SLP node with the PHIs in the
20600 instance. Simplify getting at the vectorized reduction PHIs.
20601 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
20602 through SLP instance.
20603 (vect_slp_analyze_operations): Likewise.
20604 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
20605 (vect_transform_stmt): Likewise.
20607 2017-07-20 Tom de Vries <tom@codesourcery.com>
20609 PR tree-optimization/81489
20610 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
20611 read of phi arg location to before loop that modifies phi.
20613 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20615 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
20618 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20620 PR middle-end/81331
20621 * except.c (execute): Fix ordering issue.
20623 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20625 PR rtl-optimization/81423
20626 * combine.c (make_compound_operation_int): Don't try to optimize
20627 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
20629 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20631 PR rtl-optimization/81423
20632 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
20633 with a constant that is -1 in the truncated to mode.
20635 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20637 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
20638 (determine_unlikely_bbs): ... here.
20639 * predict.h (propagate_unlikely_bbs_forward): Declare.
20640 * cfgexpand.c (pass_expand::execute): Use it.
20641 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
20643 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
20644 propagate_unlikely_bbs_forward.
20646 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20648 PR middle-end/81331
20649 * except.c (maybe_add_nop_after_section_switch): New function.
20652 2017-07-19 Tom de Vries <tom@codesourcery.com>
20654 * gimple.h (gimple_phi_set_arg): Make assert more strict.
20656 2017-07-19 Tom de Vries <tom@codesourcery.com>
20658 * gimple.h (gimple_phi_arg): Make assert more strict.
20660 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
20662 * config.gcc (powerpc*-*-*): Add mmintrin.h.
20663 * config/rs6000/mmintrin.h: New file.
20664 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
20666 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20668 PR tree-optimization/81346
20669 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
20671 2017-07-19 Tom de Vries <tom@codesourcery.com>
20673 * config/nvptx/nvptx.md (VECIM): Add V2DI.
20675 2017-07-19 Tom de Vries <tom@codesourcery.com>
20677 * config/nvptx/nvptx-modes.def: Add V2DImode.
20678 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
20679 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
20680 (nvptx_output_mov_insn): Handle lack of mov.b128.
20681 (nvptx_print_operand): Handle 'H' and 'L' codes.
20682 (nvptx_vector_mode_supported): Allow V2DImode.
20683 (nvptx_preferred_simd_mode): New function.
20684 (nvptx_data_alignment): New function.
20685 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
20686 nvptx_preferred_simd_mode.
20687 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
20689 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
20691 2017-07-19 Tom de Vries <tom@codesourcery.com>
20693 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
20694 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
20695 (nvptx_vector_mode_supported): New function. Allow V2SImode.
20696 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
20697 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
20698 (mov<VECIM>_insn): New define_insn.
20699 (define_expand "mov<VECIM>): New define_expand.
20701 2017-07-19 Tom de Vries <tom@codesourcery.com>
20703 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
20705 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20707 PR tree-optimization/81346
20708 * fold-const.h (fold_div_compare, range_check_type): Declare.
20709 * fold-const.c (range_check_type): New function.
20710 (build_range_check): Use range_check_type.
20711 (fold_div_compare): No longer static, rewritten into
20712 a match.pd helper function.
20713 (fold_comparison): Don't call fold_div_compare here.
20714 * match.pd (X / C1 op C2): New optimization using fold_div_compare
20715 as helper function.
20717 2017-07-19 Nathan Sidwell <nathan@acm.org>
20719 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
20720 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
20721 * tree.c (find_decls_types_r, verify_type): Use
20722 TYPE_{MIN,MAX}_VALUE_RAW.
20723 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
20724 (hash_tree): Likewise.
20725 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20727 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20730 2017-07-18 Tom de Vries <tom@codesourcery.com>
20732 PR middle-end/81464
20733 * omp-expand.c (expand_omp_for_static_chunk): Handle
20734 equal-argument loop exit phi.
20736 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
20739 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
20740 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
20741 operand 2 predicate.
20742 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
20743 operand 2 predicate.
20744 (ror,rol -> rorx splitters): Use const_int_operand as
20745 operand 2 predicate.
20747 2017-06-18 Richard Biener <rguenther@suse.de>
20749 PR tree-optimization/81410
20750 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
20751 the gap in the ! slp_perm SLP case after each group.
20753 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20755 PR middle-end/81463
20756 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
20759 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20761 PR middle-end/81462
20762 * predict.c (set_even_probabilities): Cleanup; do not affect
20763 probabilities that are already known.
20764 (combine_predictions_for_bb): Call even when count is set.
20766 2017-07-18 Nathan Sidwell <nathan@acm.org>
20768 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
20771 2017-07-18 Bin Cheng <bin.cheng@arm.com>
20774 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
20775 optimization for loop niter analysis.
20777 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
20780 * config/avr/avr.c (avr_optimize_casesi): Don't use
20781 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
20783 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
20785 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
20786 body_cost_vec from _vect_peel_extended_info.
20787 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
20788 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
20791 2017-07-18 Bin Cheng <bin.cheng@arm.com>
20793 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
20795 2017-07-18 Richard Biener <rguenther@suse.de>
20797 PR tree-optimization/80620
20798 PR tree-optimization/81403
20799 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
20800 info when re-using a VN table entry.
20802 2017-07-18 Richard Biener <rguenther@suse.de>
20804 PR tree-optimization/81418
20805 * tree-vect-loop.c (vectorizable_reduction): Properly compute
20806 vectype_in. Verify that with lane-reducing reduction operations
20807 we have a single def-use cycle.
20809 2017-07-17 Carl Love <cel@us.ibm.com>
20811 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
20813 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20814 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20815 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20816 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20817 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20818 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20819 VMULOSW): New enum "unspec" values.
20820 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20821 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20822 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20823 altivec_vmulosw): New patterns.
20824 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20825 VMULOSW): Add definitions.
20827 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
20829 * config/alpha/alpha.c: Include predict.h.
20831 2017-07-17 Yury Gribov <tetra2005@gmail.com>
20833 * tree-vrp.c (compare_assert_loc): Fix comparison function
20834 to return predictable results.
20836 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20838 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
20840 (subdi3): Likewise.
20841 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
20842 * doc/invoke.texi (mexpand-adddi): Update text.
20844 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20846 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
20847 that also clobbers the CC register. The old expand code is moved
20849 (*arc_clzsi2): ... here.
20850 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
20851 the CC register. The old expand code is moved to ...
20852 (arc_ctzsi2): ... here.
20854 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20856 * config/arc/arc.opt (mindexed-loads): Use initial value
20857 TARGET_INDEXED_LOADS_DEFAULT.
20858 (mauto-modify-reg): Use initial value
20859 TARGET_AUTO_MODIFY_REG_DEFAULT.
20860 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
20861 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
20862 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
20863 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
20865 2017-07-17 Martin Liska <mliska@suse.cz>
20868 * opts.c (finish_options): Do not allow -fgnu-tm
20869 w/ -fsanitize={kernel-,}address. Say sorry.
20871 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20874 * tree-loop-distribution.c (classify_partition): Only assert on
20875 numer of iterations.
20876 (merge_dep_scc_partitions): Delete prameter. Update function call.
20877 (distribute_loop): Remove code handling loop with unknown niters.
20878 (pass_loop_distribution::execute): Skip loop with unknown niters.
20880 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20883 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
20884 function sort_partitions_by_post_order.
20886 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20888 PR tree-optimization/81374
20889 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
20890 the max index of basic blocks, rather than number of basic blocks.
20892 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20894 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
20896 (arc_legitimate_pic_operand_p): Likewise.
20897 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
20899 (arc_needs_pcl_p): Likewise.
20900 (arc_legitimate_pc_offset_p): Likewise.
20901 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
20902 function is also used in constrains.md.
20903 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
20904 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
20905 PLUS. Only return true/false in known cases, otherwise assert.
20906 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
20907 is already called in arc_legitimate_constant_p.
20908 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
20910 (LEGITIMATE_PIC_OPERAND_P): Use
20911 arc_raw_symbolic_reference_mentioned_p function.
20912 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
20917 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20918 Andrew Burgess <andrew.burgess@embecosm.com>
20920 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
20921 (arc_return_address_register): New function.
20922 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
20923 (arc_handle_fndecl_attribute): Add naked attribute.
20924 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
20925 (TARGET_WARN_FUNC_RETURN): Likewise.
20926 (arc_allocate_stack_slots_for_args): New function.
20927 (arc_warn_func_return): Likewise.
20928 (machine_function): Change type fn_type.
20929 (arc_compute_function_type): Consider new naked function type,
20930 change function return type.
20931 (arc_must_save_register): Adapt to handle new
20932 arc_compute_function_type's return type.
20933 (arc_expand_prologue): Likewise.
20934 (arc_expand_epilogue): Likewise.
20935 (arc_return_address_regs): Delete.
20936 (arc_return_address_register): New function.
20937 (arc_epilogue_uses): Use above function.
20938 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
20939 (arc_function_type): Change encoding, add naked type.
20940 (ARC_INTERRUPT_P): Change to handle the new encoding.
20941 (ARC_FAST_INTERRUPT_P): Likewise.
20942 (ARC_NORMAL_P): Define.
20943 (ARC_NAKED_P): Likewise.
20944 (arc_compute_function_type): Delete prototype.
20945 * config/arc/arc.md (in_ret_delay_slot): Use
20946 arc_return_address_register function.
20947 (simple_return): Likewise.
20948 (p_return_i): Likewise.
20950 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20952 PR tree-optimization/81428
20953 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
20954 can't be built for those types.
20956 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
20958 Remove stuff dead since r239246.
20960 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
20961 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
20962 (avr_inform_devices): Remove dead stuff.
20964 2017-07-17 Tamar Christina <tamar.christina@arm.com>
20966 * config/arm/arm_neon.h: Fix softp typo.
20968 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20970 PR tree-optimization/81365
20971 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
20972 aggregate moves onto bb predecessor edges, make sure there are no
20973 loads that could alias the lhs in between the start of bb and the
20976 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
20979 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
20980 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
20981 [LSHIFTRT, outer_code = TRUNCATE]: Same.
20983 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20985 PR tree-optimization/81396
20986 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
20987 (init_symbolic_number): Initialize it to 1.
20988 (perform_symbolic_merge): Add n_ops from both operands into the new
20990 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
20991 without base_addr as useless if they need more than one operation.
20992 (bswap_replace): Handle !bswap case for NULL base_addr.
20994 2017-07-17 Tom de Vries <tom@codesourcery.com>
20997 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
21000 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
21002 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
21003 conditional builtin define __FIX_LEON3FT_B2BST.
21005 2017-07-17 Daniel Cederman <cederman@gaisler.com>
21007 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
21008 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
21011 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
21013 PR rtl-optimization/81424
21014 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
21015 to remove potential trapping from operands if -fnon-call-exceptions.
21017 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21019 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
21020 profile_proability for scalling.
21021 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
21023 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21025 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
21027 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21029 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
21030 fixpoint arithmetics.
21032 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21034 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
21035 fixpoint arithmetics.
21037 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21039 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
21040 fixpoint arithmetics.
21042 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21044 * profile-count.h (profile_probability::from_reg_br_prob_note,
21045 profile_probability::to_reg_br_prob_note): New functions.
21046 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
21047 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
21048 * predict.c (probability_reliable_p): Update.
21049 (edge_probability_reliable_p): Update.
21050 (br_prob_note_reliable_p): Update.
21051 (invert_br_probabilities): Update.
21052 (add_reg_br_prob_note): New function.
21053 (combine_predictions_for_insn): Update.
21054 * asan.c (asan_clear_shadow): Update.
21055 * cfgbuild.c (compute_outgoing_frequencies): Update.
21056 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
21057 (update_br_prob_note): Update.
21058 (rtl_verify_edges): Update.
21059 (purge_dead_edges): Update.
21060 (fixup_reorder_chain): Update.
21061 * emit-rtl.c (try_split): Update.
21062 * ifcvt.c (cond_exec_process_insns): Update.
21063 (cond_exec_process_if_block): Update.
21064 (dead_or_predicable): Update.
21065 * internal-fn.c (expand_addsub_overflow): Update.
21066 (expand_neg_overflow): Update.
21067 (expand_mul_overflow): Update.
21068 * loop-doloop.c (doloop_modify): Update.
21069 * loop-unroll.c (compare_and_jump_seq): Update.
21070 * optabs.c (emit_cmp_and_jump_insn_1): Update.
21071 * predict.h: Update.
21072 * reorg.c (mostly_true_jump): Update.
21074 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
21075 * config/alpha/alpha.c (emit_unlikely_jump): Update.
21076 * config/arc/arc.c: (emit_unlikely_jump): Update.
21077 * config/arm/arm.c: (emit_unlikely_jump): Update.
21078 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
21079 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
21080 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
21081 (ix86_print_operand): Update.
21082 (ix86_split_fp_branch): Update.
21083 (predict_jump): Update.
21084 * config/ia64/ia64.c (ia64_print_operand): Update.
21085 * config/mmix/mmix.c (mmix_print_operand): Update.
21086 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
21087 (rs6000_expand_split_stack_prologue): Update.
21088 * config/rs6000/rs6000.c: Update.
21089 * config/s390/s390.c (s390_expand_vec_strlen): Update.
21090 (s390_expand_vec_movstr): Update.
21091 (s390_expand_cs_tdsi): Update.
21092 (s390_expand_split_stack_prologue): Update.
21093 * config/sh/sh.c (sh_print_operand): Update.
21094 (expand_cbranchsi4): Update.
21095 (expand_cbranchdi4): Update.
21096 * config/sparc/sparc.c (output_v9branch): Update.
21097 * config/spu/spu.c (get_branch_target): Update.
21098 (ea_load_store_inline): Update.
21099 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
21100 * config/tilepro/tilepro.c: Update.
21102 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
21104 * gimplify.c (mostly_copy_tree_r): Revert latest change.
21105 (gimplify_save_expr): Likewise.
21107 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21109 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
21111 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21113 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
21115 * timevar.def (TV_IPA_FNSUMMARY): Define.
21117 2017-07-16 Daniel Cederman <cederman@gaisler.com>
21119 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
21120 to back store errata sensitive sequence from being generated.
21121 (sqrtdf2_fix): Likewise.
21123 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21125 * tree-ssa-threadupdate.c (compute_path_counts,
21126 update_joiner_offpath_counts): Use profile_probability.
21128 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
21131 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21133 * config/arm/arm-c.c (arm_cpu_builtins): Define
21134 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
21136 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
21138 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21139 array entries to represent __ieee128 versions of the
21140 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
21141 scalar_extract_sig, and scalar_insert_exp built-in functions.
21142 (altivec_resolve_overloaded_builtin): Add special case handling
21143 for the __builtin_scalar_insert_exp function, as represented by
21144 the P9V_BUILTIN_VEC_VSIEDP constant.
21145 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
21146 exponent support for __ieee128 argument.
21147 (VSESQP): Add scalar extract signature support for __ieee128
21149 (VSTDCNQP): Add scalar test negative support for __ieee128
21151 (VSIEQP): Add scalar insert exponent support for __int128 argument
21152 with __ieee128 result.
21153 (VSIEQPF): Add scalar insert exponent support for __ieee128
21154 argument with __ieee128 result.
21155 (VSTDCQP): Add scalar test data class support for __ieee128
21157 (VSTDCNQP): Add overload support for scalar test negative with
21158 __ieee128 argument.
21159 (VSTDCQP): Add overload support for scalar test data class
21160 __ieee128 argument.
21161 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
21162 UNSPEC_VSX_SXSIGDP.
21163 (UNSPEC_VSX_SIEXPQP): New constant.
21164 (xsxexpqp): New insn for VSX scalar extract exponent quad
21166 (xsxsigqp): New insn for VSX scalar extract significand quad
21168 (xsiexpqpf): New insn for VSX scalar insert exponent quad
21169 precision with floating point argument.
21170 (xststdcqp): New expand for VSX scalar test data class quad
21172 (xststdcnegqp): New expand for VSX scalar test negative quad
21174 (xststdcqp): New insn to match expansions for VSX scalar test data
21175 class quad precision and VSX scalar test negative quad precision.
21176 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
21177 special case operand checking to enforce that second operand of
21178 VSX scalar test data class with quad precision argument is a 7-bit
21180 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
21181 prototypes and descriptions of __ieee128 versions of
21182 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
21183 scalar_test_data_class, and scalar_test_neg built-in functions.
21185 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21187 PR tree-optimization/81162
21188 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
21189 replace a negate with an add.
21191 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
21193 * doc/invoke.texi (arm/-mcpu): Document +crypto.
21195 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21197 * config/arm/arm-c.c (arm_cpu_builtins): Define
21198 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
21200 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21202 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
21203 (armv8-r): Set ARM Cortex-R52 as default CPU.
21204 * config/arm/arm-tables.opt: Regenerate.
21205 * config/arm/arm-tune.md: Regenerate.
21206 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
21208 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
21209 extension for -mcpu=cortex-r52.
21211 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21213 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
21214 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
21215 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
21216 (fp-armv8): Define it as FP_ARMv8 only.
21217 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
21218 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
21220 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
21221 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
21222 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
21223 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
21224 than TARGET_FPU_ARMV8.
21225 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
21226 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
21227 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
21229 * config/arm/neon.md (neon_vrint): Likewise.
21230 (neon_vcvt): Likewise.
21231 (neon_<fmaxmin_op><mode>): Likewise.
21232 (<fmaxmin><mode>3): Likewise.
21233 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
21234 * config/arm/predicates.md (arm_cond_move_operator): Check against
21235 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
21237 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
21239 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
21240 to top of function.
21242 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21244 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
21245 loop in comment with memset.
21247 2017-07-14 Martin Liska <mliska@suse.cz>
21249 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
21250 * dwarf2out.c (is_java): Remove the function.
21251 (output_pubname): Remove usage of the function.
21252 (lower_bound_default): Remove usage of DW_LANG_Java.
21253 (gen_compile_unit_die): Likewise.
21254 * gcc.c: Remove compiler defaults for .java and .zip files.
21255 * gimple-expr.c (remove_suffix): Change as there's no longer
21256 extension than 4-letter one.
21257 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
21258 (gimplify_save_expr): Likewise.
21259 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
21260 as it's possible even for other languages than Java.
21261 * langhooks.h (struct lang_hooks): Remove Java from a comment.
21262 * lto-opts.c (lto_write_options): Remove reference to Java.
21263 * opts.c (strip_off_ending): Update file extension handling.
21264 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
21265 * tree-eh.c (lower_resx): Likewise.
21266 * tree.c (free_lang_data_in_type): Remove dead code.
21267 (find_decls_types_r): Likewise.
21268 (build_common_builtin_nodes): Remove Java from a comment.
21269 (verify_type): Remove dead code.
21270 * varasm.c (assemble_external): Remove Java from a comment.
21272 2017-07-14 Martin Liska <mliska@suse.cz>
21274 * opts.c (finish_options): Add quotes.
21275 (common_handle_option): Likewise.
21277 2017-07-14 Martin Liska <mliska@suse.cz>
21279 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
21280 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
21281 Remove N_SO_PASCAL.
21282 * dwarf2out.c (lower_bound_default): Do not handle
21284 (gen_compile_unit_die): Likewise.
21285 * gcc.c: Remove default extension binding for GNU Pascal.
21286 * stmt.c: Remove Pascal language from a comment.
21287 * xcoffout.c: Likewise.
21289 2017-07-13 David Malcolm <dmalcolm@redhat.com>
21292 * diagnostic-show-locus.c (fixit_cmp): New function.
21293 (layout::layout): Sort m_fixit_hints.
21294 (column_range::column_range): Assert that the values are valid.
21295 (struct char_span): New struct.
21296 (correction::overwrite): New method.
21297 (struct source_line): New struct.
21298 (line_corrections::add_hint): Add assertions. Reimplement memcpy
21299 calls in terms of classes source_line and char_span, and
21300 correction::overwrite.
21301 (selftest::test_overlapped_fixit_printing_2): New function.
21302 (selftest::diagnostic_show_locus_c_tests): Call it.
21304 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
21306 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
21307 early if there is no lhs.
21309 2017-07-13 Martin Liska <mliska@suse.cz>
21311 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
21312 (gen_reference_type_die): Likewise.
21313 * stor-layout.c: Remove Pascal-related comment.
21315 2017-07-13 Martin Liska <mliska@suse.cz>
21317 * opts.c (finish_options): Add quotes to error messages.
21318 (parse_sanitizer_options): Likewise.
21320 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
21322 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
21324 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
21326 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
21328 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
21330 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
21332 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
21334 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
21337 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
21338 provides the hardware capability bits, define the macro
21339 __BUILTIN_CPU_SUPPORTS__.
21340 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
21341 if GLIBC does not provide the hardware capability bits. Add a
21342 gcc_unreachable call if the built-in cpu function is neither
21343 __builtin_cpu_is nor __builtin_cpu_supports.
21344 (rs6000_get_function_versions_dispatcher): Change the warning
21345 that an old GLIBC is used which does not export the capability
21346 bits to be an error.
21347 * doc/extend.texi (target_clones attribute): Document the
21348 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
21349 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
21350 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
21351 the macros defined by GCC if the newer GLIBC is available.
21353 2017-07-12 Jeff Law <law@redhat.com>
21355 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
21356 remaining includes slightly.
21357 * config/riscv/riscv-builtins.c: Include profile-count.h.
21359 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21362 * config/avr/avr.c (avr_set_current_function): In diagnostic
21363 messages: Quote keywords and (parts of) identifiers.
21364 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
21367 2017-07-12 Carl Love <cel@us.ibm.com>
21369 * config/rs6000/rs6000-c.c: Add support for built-in functions
21370 vector bool char vec_revb (vector bool char);
21371 vector bool short vec_revb (vector short char);
21372 vector bool int vec_revb (vector bool int);
21373 vector bool long long vec_revb (vector bool long long);
21374 * doc/extend.texi: Update the built-in documentation file for the
21375 new built-in functions.
21377 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21379 * config/s390/s390.md: Remove movcc splitter.
21381 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21383 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
21384 load/store on condition.
21386 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21389 * config/avr/avr.c (avr_encode_section_info)
21390 [progmem && !TREE_READONLY]: Error if progmem object needs
21393 2017-07-11 Michael Collison <michael.collison@arm.com>
21395 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
21398 2017-07-11 Carl Love <cel@us.ibm.com>
21400 * config/rs6000/rs6000-c.c: Add support for builtins
21401 vector unsigned int vec_parity_lsbb (vector signed int);
21402 vector unsigned int vec_parity_lsbb (vector unsigned int);
21403 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
21404 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
21405 vector unsigned long long vec_parity_lsbb (vector signed long long);
21406 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
21407 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
21408 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
21409 * doc/extend.texi: Update the built-in documentation file for the
21410 new built-in functions.
21412 2017-07-11 David Malcolm <dmalcolm@redhat.com>
21414 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
21415 (layout::m_primary_loc): New field.
21416 (layout::layout): Initialize new field. Move location filtering
21417 logic from here to...
21418 (layout::maybe_add_location_range): ...this new method. Add
21419 support for filtering to just the lines already specified by other
21421 (layout::will_show_line_p): New method.
21422 (gcc_rich_location::add_location_if_nearby): New method.
21423 (selftest::test_add_location_if_nearby): New test function.
21424 (selftest::diagnostic_show_locus_c_tests): Call it.
21425 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
21428 2017-07-11 Tom de Vries <tom@codesourcery.com>
21430 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
21431 (bb_first_real_insn): New function.
21432 (nvptx_single): Add extra initialization of broadcasted condition
21435 2017-07-11 Nathan Sidwell <nathan@acm.org>
21437 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
21439 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
21441 * doc/extend.texi (AVR Function Attributes): Remove weblink to
21442 Binutils doc as TEXI will mess them up.
21443 * doc/invoke.texi (AVR Options): Same here.
21445 2017-07-11 Daniel Cederman <cederman@gaisler.com>
21447 * config/sparc/sparc.opt (mfix-ut700): New option.
21448 (mfix-gr712rc): Likewise.
21449 (sparc_fix_b2bst): New variable.
21450 * doc/invoke.texi (SPARC options): Document them.
21451 (ARM options): Fix warnings.
21452 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
21453 instructions to prevent sequences that can trigger the store-store
21454 errata for certain LEON3FT processors.
21455 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
21456 (sparc_option_override): Set sparc_fix_b2bst appropriately.
21457 * config/sparc/sparc.md (fix_b2bst): New attribute.
21458 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
21460 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
21463 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
21465 (*rsqrtsf2_sse): Ditto.
21467 (div<mode>3): Macroize insn from divdf3 and divsf3
21468 using MODEF mode iterator.
21470 2017-07-10 Martin Sebor <msebor@redhat.com>
21472 PR tree-optimization/80397
21473 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
21474 instead of testing for equality to INTEGER_TYPE.
21476 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
21478 * config.gcc: Remove uclibc from arc target spec.
21480 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
21482 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
21484 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21487 * lto-wrapper.c (remove_option): New function.
21488 (merge_and_complain): Merge PIC/PIE options more realistically.
21490 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21492 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
21496 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
21497 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
21498 * config.in: Regenerate.
21499 * configure: Regenerate.
21500 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
21501 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
21502 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
21503 (TARGET_GASISR_PROLOGUES): ...target mask.
21504 * common/config/avr/avr-common.c
21505 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
21506 Set -mgas-isr-prologues.
21507 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
21508 INSERT_PASS_BEFORE for it.
21509 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
21510 * config/avr/avr.c (avr_option_override)
21511 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
21512 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
21513 (avr_attribute_table) <no_gccisr>: Add new function attribute.
21514 (avr_set_current_function) <is_no_gccisr>: Init machine field.
21515 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
21517 (make_avr_pass_pre_proep): New function.
21518 (emit_push_sfr) <treg>: Add argument to function and use it
21519 instead of TMP_REG.
21520 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
21521 and set machine->gasisr.yes.
21522 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
21523 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
21524 __gcc_isr.n_pushed to .L__stack_usage.
21525 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
21526 (avr_asm_final_postscan_insn): ...this new static function.
21527 * config/avr/avr.h (machine_function)
21528 <is_no_gccisr, use_L__stack_usage>: New fields.
21529 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
21530 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
21531 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
21532 (gasisr, *gasisr): New expander and insn.
21533 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21534 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
21535 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
21537 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
21539 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
21542 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21544 Move jump-tables out of .text again.
21547 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
21548 (ASM_OUTPUT_ADDR_VEC): New function.
21549 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
21550 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
21551 INSN_ADDRESSes as asm comment.
21552 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
21553 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
21554 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
21555 * config/avr/avr.md (*tablejump): Adjust comment.
21556 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
21557 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
21559 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
21560 (avr_output_addr_vec): New proto.
21561 (avr_log_t) <insn_addresses>: New field.
21563 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
21566 * config/i386/i386.c (ix86_function_arg_advance): Set
21567 outgoing_args_on_stack to true if there are outgoing arguments
21569 (ix86_function_arg): Likewise.
21570 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
21571 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
21572 * config/i386/i386.h (machine_function): Add
21573 outgoing_args_on_stack.
21575 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
21577 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
21578 supporting pthreds.
21579 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
21581 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
21583 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
21584 (REAL_H): Remove $(MACHMODE_H).
21585 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
21587 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
21588 $(MACHMODE_H) and double-int.h.
21589 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
21591 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
21594 2017-07-07 Andrew Pinski <apinski@cavium.com>
21596 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
21597 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
21599 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21601 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
21602 Add warning if GCC was not configured to link against a GLIBC that
21603 exports the hardware capability bits.
21604 (make_resolver_func): Make resolver function private and not a
21605 COMDAT function. Create the name with clone_function_name instead
21606 of make_unique_name.
21609 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
21610 correct operand in doing the split.
21612 2017-07-07 Carl Love <cel@us.ibm.com>
21614 * config/rs6000/rs6000-c: Add support for built-in function
21615 vector unsigned short vec_pack_to_short_fp32 (vector float,
21617 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
21618 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
21619 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
21620 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
21621 (convert_4f32_8i16): Add define_expand.
21622 * doc/extend.texi: Update the built-in documentation file for the
21623 new built-in function.
21625 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21627 * config/sparc/m8.md: New file.
21628 * config/sparc/sparc.md: Include m8.md.
21630 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21632 * config/sparc/sparc.opt: New option -mvis4b.
21633 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
21634 (sparc_option_override): Handle VIS4B.
21635 (enum sparc_builtins): Define
21636 SPARC_BUILTIN_DICTUNPACK{8,16,32},
21637 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
21638 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
21639 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
21640 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
21641 (check_constant_argument): New function.
21642 (sparc_vis_init_builtins): Define builtins
21643 __builtin_vis_dictunpack{8,16,32},
21644 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
21645 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
21646 __builtin_vis_fpcmpde{8,16,32}shl and
21647 __builtin_vis_fpcmpur{8,16,32}shl.
21648 (sparc_expand_builtin): Check that the constant operands to
21649 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
21650 constant and in range.
21651 * config/sparc/sparc-c.c (sparc_target_macros): Handle
21653 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
21654 (SPARC_IMM5_P): Likewise.
21655 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
21656 (enabled): Handle vis4b.
21657 (UNSPEC_DICTUNPACK): New unspec.
21658 (UNSPEC_FPCMPSHL): Likewise.
21659 (UNSPEC_FPUCMPSHL): Likewise.
21660 (UNSPEC_FPCMPDESHL): Likewise.
21661 (UNSPEC_FPCMPURSHL): Likewise.
21662 (cpu_feature): New CPU feature `vis4b'.
21663 (dictunpack{8,16,32}): New insns.
21664 (FPCSMODE): New mode iterator.
21665 (fpcscond): New code iterator.
21666 (fpcsucond): Likewise.
21667 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
21668 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
21669 (fpcmpde{8,16,32}{si,di}shl): Likewise.
21670 (fpcmpur{8,16,32}{si,di}shl): Likewise.
21671 * config/sparc/constraints.md: Define constraints `q' for unsigned
21672 2-bit integer constants and `t' for unsigned 5-bit integer
21674 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
21676 (imm5_operand_dictunpack16): Likewise.
21677 (imm5_operand_dictunpack32): Likewise.
21678 (imm2_operand): Likewise.
21679 * doc/invoke.texi (SPARC Options): Document -mvis4b.
21680 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
21681 ditunpack* and fpcmp*shl builtins.
21683 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21685 * config.gcc: Handle m8 in --with-{cpu,tune} options.
21686 * config.in: Add HAVE_AS_SPARC6 define.
21687 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
21689 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
21691 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
21692 (CPP_CPU_SPEC): Handle m8.
21693 (ASM_CPU_SPEC): Likewise.
21694 * config/sparc/sparc-opts.h (enum processor_type): Add
21696 * config/sparc/sparc.c (m8_costs): New struct.
21697 (sparc_option_override): Handle TARGET_CPU_m8.
21698 (sparc32_initialize_trampoline): Likewise.
21699 (sparc64_initialize_trampoline): Likewise.
21700 (sparc_issue_rate): Likewise.
21701 (sparc_register_move_cost): Likewise.
21702 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
21703 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
21704 (ASM_CPU64_DEFAULT_SPEC): Likewise.
21705 (CPP_CPU_SPEC): Handle M8.
21706 (ASM_CPU_SPEC): Likewise.
21707 (AS_M8_FLAG): Define.
21708 * config/sparc/sparc.md: Add m8 to the cpu attribute.
21709 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
21710 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
21712 * configure: Regenerate.
21713 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
21716 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21718 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
21720 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
21721 ("*movdi_insn_sp32"): Do not set v3pipe.
21722 ("*movsi_insn"): Likewise.
21723 ("*movdi_insn_sp64"): Likewise.
21724 ("*movsf_insn"): Likewise.
21725 ("*movdf_insn_sp32"): Likewise.
21726 ("*movdf_insn_sp64"): Likewise.
21727 ("*zero_extendsidi2_insn_sp64"): Likewise.
21728 ("*sign_extendsidi2_insn"): Likewise.
21729 ("*mov<VM32:mode>_insn"): Likewise.
21730 ("*mov<VM64:mode>_insn_sp64"): Likewise.
21731 ("*mov<VM64:mode>_insn_sp32"): Likewise.
21732 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
21733 ("<vlop:code><VL:mode>3"): Likewise.
21734 ("*not_<vlop:code><VL:mode>3"): Likewise.
21735 ("*nand<VL:mode>_vis"): Likewise.
21736 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
21737 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
21738 ("one_cmpl<VL:mode>2"): Likewise.
21739 ("faligndata<VM64:mode>_vis"): Likewise.
21740 ("alignaddrsi_vis"): Likewise.
21741 ("alignaddrdi_vis"): Likweise.
21742 ("alignaddrlsi_vis"): Likewise.
21743 ("alignaddrldi_vis"): Likewise.
21744 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
21745 ("bmaskdi_vis"): Likewise.
21746 ("bmasksi_vis"): Likewise.
21747 ("bshuffle<VM64:mode>_vis"): Likewise.
21748 ("cmask8<P:mode>_vis"): Likewise.
21749 ("cmask16<P:mode>_vis"): Likewise.
21750 ("cmask32<P:mode>_vis"): Likewise.
21751 ("pdistn<P:mode>_vis"): Likewise.
21752 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
21754 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21756 * config/sparc/sparc.md ("subtype"): New insn attribute.
21757 ("*wrgsr_sp64"): Set insn subtype.
21758 ("*rdgsr_sp64"): Likewise.
21759 ("alignaddrsi_vis"): Likewise.
21760 ("alignaddrdi_vis"): Likewise.
21761 ("alignaddrlsi_vis"): Likewise.
21762 ("alignaddrldi_vis"): Likewise.
21763 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
21764 ("fexpand_vis"): Likewise.
21765 ("fpmerge_vis"): Likewise.
21766 ("faligndata<VM64:mode>_vis"): Likewise.
21767 ("bshuffle<VM64:mode>_vis"): Likewise.
21768 ("cmask8<P:mode>_vis"): Likewise.
21769 ("cmask16<P:mode>_vis"): Likewise.
21770 ("cmask32<P:mode>_vis"): Likewise.
21771 ("fchksm16_vis"): Likewise.
21772 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
21773 ("fmean16_vis"): Likewise.
21774 ("fp<plusminus_insn>64_vis"): Likewise.
21775 ("<plusminus_insn>v8qi3"): Likewise.
21776 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
21777 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
21778 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
21779 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
21780 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
21781 ("*movqi_insn"): Likewise.
21782 ("*movhi_insn"): Likewise.
21783 ("*movsi_insn"): Likewise.
21784 ("movsi_pic_gotdata_op"): Likewise.
21785 ("*movdi_insn_sp32"): Likewise.
21786 ("*movdi_insn_sp64"): Likewise.
21787 ("movdi_pic_gotdata_op"): Likewise.
21788 ("*movsf_insn"): Likewise.
21789 ("*movdf_insn_sp32"): Likewise.
21790 ("*movdf_insn_sp64"): Likewise.
21791 ("*zero_extendhisi2_insn"): Likewise.
21792 ("*zero_extendqihi2_insn"): Likewise.
21793 ("*zero_extendqisi2_insn"): Likewise.
21794 ("*zero_extendqidi2_insn"): Likewise.
21795 ("*zero_extendhidi2_insn"): Likewise.
21796 ("*zero_extendsidi2_insn_sp64"): Likewise.
21797 ("ldfsr"): Likewise.
21798 ("prefetch_64"): Likewise.
21799 ("prefetch_32"): Likewise.
21800 ("tie_ld32"): Likewise.
21801 ("tie_ld64"): Likewise.
21802 ("*tldo_ldub_sp32"): Likewise.
21803 ("*tldo_ldub1_sp32"): Likewise.
21804 ("*tldo_ldub2_sp32"): Likewise.
21805 ("*tldo_ldub_sp64"): Likewise.
21806 ("*tldo_ldub1_sp64"): Likewise.
21807 ("*tldo_ldub2_sp64"): Likewise.
21808 ("*tldo_ldub3_sp64"): Likewise.
21809 ("*tldo_lduh_sp32"): Likewise.
21810 ("*tldo_lduh1_sp32"): Likewise.
21811 ("*tldo_lduh_sp64"): Likewise.
21812 ("*tldo_lduh1_sp64"): Likewise.
21813 ("*tldo_lduh2_sp64"): Likewise.
21814 ("*tldo_lduw_sp32"): Likewise.
21815 ("*tldo_lduw_sp64"): Likewise.
21816 ("*tldo_lduw1_sp64"): Likewise.
21817 ("*tldo_ldx_sp64"): Likewise.
21818 ("*mov<VM32:mode>_insn"): Likewise.
21819 ("*mov<VM64:mode>_insn_sp64"): Likewise.
21820 ("*mov<VM64:mode>_insn_sp32"): Likewise.
21822 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21824 * config/sparc/sparc.md ("type"): New insn type viscmp.
21825 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
21827 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
21828 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
21829 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
21830 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
21832 ("n7_vis_logical_11cycle"): Likewise.
21833 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
21834 * config/sparc/niagara2.md ("niag3_vis": Likewise.
21835 * config/sparc/niagara.md ("niag_vis"): Likewise.
21836 * config/sparc/ultra3.md ("us3_fga"): Likewise.
21837 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
21839 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21841 * config/sparc/sparc.md: New instruction type `bmask'.
21842 (bmaskdi_vis): Use the `bmask' type.
21843 (bmasksi_vis): Likewise.
21844 * config/sparc/ultra3.md (us3_array): Likewise.
21845 * config/sparc/niagara7.md (n7_array): Likewise.
21846 * config/sparc/niagara4.md (n4_array): Likewise.
21847 * config/sparc/niagara2.md (niag2_vis): Likewise.
21848 (niag3_vis): Likewise.
21849 * config/sparc/niagara.md (niag_vis): Likewise.
21851 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21853 * ipa-comdats.c: Remove optimize check from gate.
21854 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
21855 for functions not optimized.
21856 (ipa_fn_summary_read): Skip optimize check.
21857 (ipa_fn_summary_write): Likewise.
21858 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
21860 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
21862 (can_inline_edge_p): Check flag_pcc_struct_return for match.
21863 (check_callers): Give up on caller which is not optimized.
21864 (inline_small_functions): Likewise.
21865 (ipa_inline): Do not give up when not optimizing.
21866 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
21867 away unoptimizes cdtors.
21868 (whole_program_function_and_variable_visibility): Do
21869 ipa_discover_readonly_nonaddressable_vars in LTO mode.
21870 * ipa.c (process_references): Do not check optimize.
21871 (symbol_table::remove_unreachable_nodes): Update optimize check.
21872 (set_writeonly_bit): Update optimize check.
21873 (pass_ipa_cdtor_merge::gate): Do not check optimize.
21874 (pass_ipa_single_use::gate): Remove.
21876 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21878 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
21879 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
21880 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
21881 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
21882 permute_load, permute_store, adjust_extract, adjust_splat,
21883 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
21884 replace_swap_with_copy, dump_swap_insn_table,
21885 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
21886 recombine_lvx_pattern, recombine_stvx_pattern,
21887 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
21888 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
21889 to file rs6000-p8swap.c.
21890 * config/rs6000/rs6000-p8swap.c: New file.
21891 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
21892 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
21893 and rs6000*-*-* targets.
21895 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21897 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
21899 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21901 * lto-wrapper.c (merge_and_complain): Do not merge
21902 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
21903 fsigned_zeros, ftrapping_math, fwrapv.
21904 (append_compiler_options): Do not track these options.
21905 (append_linker_options): Likewie
21907 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21909 * cgraphunit.c (cgraph_node::finalize_function): When
21910 !flag_toplevel_reorde set no_reorder flag.
21911 (varpool_node::finalize_decl): Likewise.
21912 (symbol_table::compile): Drop no toplevel reorder path.
21914 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21916 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
21917 edges; zero probability is not better than uninitialized.
21919 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
21921 * asan.h (asan_sanitize_allocas_p): Declare.
21922 * asan.c (asan_sanitize_allocas_p): New function.
21923 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
21924 (handle_builtin_alloca): Likewise.
21925 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
21926 if !asan_sanitize_allocas_p.
21927 * params.def (asan-instrument-allocas): Add new option.
21928 * params.h (ASAN_PROTECT_ALLOCAS): Define.
21929 * opts.c (common_handle_option): Disable allocas sanitization for
21932 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
21934 * asan.c: Include gimple-fold.h.
21935 (get_last_alloca_addr): New function.
21936 (handle_builtin_stackrestore): Likewise.
21937 (handle_builtin_alloca): Likewise.
21938 (asan_emit_allocas_unpoison): Likewise.
21939 (get_mem_refs_of_builtin_call): Add new parameter, remove const
21940 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
21941 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
21942 (instrument_builtin_call): Pass gimple iterator to
21943 get_mem_refs_of_builtin_call.
21944 (last_alloca_addr): New global.
21945 * asan.h (asan_emit_allocas_unpoison): Declare.
21946 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
21947 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
21948 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
21949 if function calls alloca.
21950 * gimple-fold.c (replace_call_with_value): Remove static keyword.
21951 * gimple-fold.h (replace_call_with_value): Declare.
21952 * internal-fn.c: Include asan.h.
21953 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
21954 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
21956 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21958 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
21959 (C_SELFTEST_FLAGS): New.
21960 (CPP_SELFTEST_FLAGS): New.
21961 (SELFTEST_DEPS): New, from deps of s-selftest.
21962 (C_SELFTEST_DEPS): New, from deps of s-selftest.
21963 (CPP_SELFTEST_DEPS): New.
21964 (selftest): Add dependency on s-selftest-c++.
21965 (s-selftest): Rename to...
21966 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
21967 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
21968 than SELFTEST_FLAGS.
21969 (selftest-gdb): Rename to...
21970 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
21972 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
21973 (selftest-valgrind): Rename to...
21974 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
21976 (selftest-valgrind): Reintroduce as an alias for
21977 selftest-c-valgrind.
21978 (s-selftest-c++): New.
21979 (selftest-c++-gdb): New.
21980 (selftest-c++-valgrind): New.
21982 2017-07-06 Olivier Hainque <hainque@adacore.com>
21984 * gcc.c (process_command): When deciding if undefined variables
21985 should be ignored when processing specs, accept "gcc -v" as well.
21987 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21989 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
21990 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
21992 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21994 * config/arm/arm-cpus.in (armv8-r): Add new entry.
21995 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
21996 * config/arm/arm-tables.opt: Regenerate.
21997 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
21999 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
22001 2017-07-06 Carl Love <cel@us.ibm.com>
22003 * ChangeLog: Clean up from mid air collision
22005 2017-07-06 Carl Love <cel@us.ibm.com>
22007 * config/rs6000/rs6000-c.c: Add support for built-in functions
22008 vector signed int vec_subc (vector signed int, vector signed int);
22009 vector signed __int128 vec_subc (vector signed __int128,
22010 vector signed __int128);
22011 vector unsigned __int128 vec_subc (vector unsigned __int128,
22012 vector unsigned __int128);
22013 vector signed int vec_sube (vector signed int, vector signed int,
22014 vector signed int);
22015 vector unsigned int vec_sube (vector unsigned int,
22016 vector unsigned int,
22017 vector unsigned int);
22018 vector signed __int128 vec_sube (vector signed __int128,
22019 vector signed __int128,
22020 vector signed__int128);
22021 vector unsigned __int128 vec_sube (vector unsigned __int128,
22022 vector unsigned __int128,
22023 vector unsigned __int128);
22024 vector signed int vec_subec (vector signed int, vector signed int,
22025 vector signed int);
22026 vector unsigned int vec_subec (vector unsigned int,
22027 vector unsigned int,
22028 vector unsigned int);
22029 vector signed __int128 vec_subec (vector signed __int128,
22030 vector signed __int128,
22031 vector signed__int128);
22032 vector unsigned __int128 vec_subec (vector unsigned __int128,
22033 vector unsigned __int128,
22034 vector unsigned __int128);
22035 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
22036 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
22037 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
22038 BU_ALTIVEC_OVERLOAD_X definitions.
22039 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
22040 * doc/extend.texi: Update the built-in documentation file for the new
22041 built-in functions.
22043 2017-07-06 David Malcolm <dmalcolm@redhat.com>
22046 * diagnostic-show-locus.c (layout::layout): Use start and finish
22047 spelling location for the start and finish of each range.
22048 * genmatch.c (linemap_client_expand_location_to_spelling_point):
22049 Add unused aspect param.
22050 * input.c (expand_location_1): Add "aspect" param, and use it
22051 to access the correct part of the location.
22052 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
22054 (expand_location_to_spelling_point): Likewise.
22055 (linemap_client_expand_location_to_spelling_point): Add "aspect"
22056 param, and pass it to expand_location_1.
22058 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
22060 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
22061 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
22062 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
22063 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
22064 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
22065 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
22066 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
22067 _mm_maskz_getmant_ss): New intrinsics.
22068 (__builtin_ia32_getexpss128_mask): Changed to ...
22069 __builtin_ia32_getexpss128_round ... this.
22070 (__builtin_ia32_getexpsd128_mask): Changed to ...
22071 __builtin_ia32_getexpsd128_round ... this.
22072 * config/i386/i386-builtin-types.def
22073 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
22074 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
22075 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
22076 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
22077 __builtin_ia32_getmantss_mask_round): New builtins.
22078 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
22079 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
22080 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
22081 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
22082 * config/i386/sse.md
22083 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
22084 avx512f_sgetexp<mode><mask_scalar_name>
22085 <round_saeonly_scalar_name> ... this.
22086 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
22087 %0, %1, %2<round_saeonly_op3>}): Changed to ...
22088 vgetexp<ssescalarmodesuffix>
22089 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22090 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
22091 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
22092 avx512f_vgetmant<mode><mask_scalar_name>
22093 <round_saeonly_scalar_name> ... this.
22094 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
22095 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
22096 vgetmant<ssescalarmodesuffix>
22097 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
22098 %0<mask_scalar_operand4>, %1, %2
22099 <round_saeonly_scalar_mask_op4>, %3} ... this.
22100 * config/i386/subst.md (mask_scalar_operand4,
22101 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
22102 round_saeonly_scalar_nimm_predicate): New subst attributes.
22104 2017-07-06 Julia Koval <julia.koval@intel.com>
22106 * config/i386/i386.c (ix86_erase_embedded_rounding):
22107 Remove code for old rounding pattern.
22109 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
22111 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
22113 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
22115 * doc/sourcebuild.texi (Test Directives, Variants of
22116 dg-require-support): Add documentation for dg-require-stack-check.
22118 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
22120 * config/i386/subst.md (mask_scalar, round_scalar,
22121 round_saeonly_scalar): New meta-templates.
22122 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
22123 round_scalar_mask_operand3, round_scalar_mask_op3,
22124 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
22125 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
22126 round_saeonly_scalar_constraint,
22127 round_saeonly_scalar_prefix): New subst attribute.
22128 * config/i386/sse.md
22129 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
22130 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
22131 <round_scalar_name> ... this.
22132 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
22133 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
22134 <round_scalar_name> ... this.
22135 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
22136 <sse>_vm<code><mode>3<mask_scalar_name>
22137 <round_saeonly_scalar_name> ... this.
22138 (v<plusminus_mnemonic><ssescalarmodesuffix>
22139 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
22140 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
22141 v<plusminus_mnemonic><ssescalarmodesuffix>
22142 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22143 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
22144 (v<multdiv_mnemonic><ssescalarmodesuffix>
22145 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
22146 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
22147 v<multdiv_mnemonic><ssescalarmodesuffix>
22148 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22149 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
22150 (v<maxmin_float><ssescalarmodesuffix>
22151 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
22152 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
22153 v<maxmin_float><ssescalarmodesuffix>
22154 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22155 %0<mask_scalar_operand3>, %1, %<iptr>2
22156 <round_saeonly_scalar_mask_op3>} ... this.
22158 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
22160 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
22161 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
22163 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
22164 Alan Hayward <alan.hayward@arm.com>
22165 David Sherwood <david.sherwood@arm.com>
22167 * combine.c (simplify_if_then_else): Remove "enum" before
22169 * compare-elim.c (can_eliminate_compare): Likewise.
22170 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
22172 (aarch64_lookup_simd_builtin_type): Likewise.
22173 (aarch64_simd_builtin_type): Likewise.
22174 (aarch64_init_simd_builtin_types): Likewise.
22175 (aarch64_simd_expand_args): Likewise.
22176 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
22178 (aarch64_reverse_mask): Likewise.
22179 (aarch64_simd_emit_reg_reg_move): Likewise.
22180 (aarch64_gen_adjusted_ldpstp): Likewise.
22181 (aarch64_ccmp_mode_to_code): Likewise.
22182 (aarch64_operands_ok_for_ldpstp): Likewise.
22183 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
22184 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
22186 (aarch64_min_divisions_for_recip_mul): Likewise.
22187 (aarch64_reassociation_width): Likewise.
22188 (aarch64_get_condition_code_1): Likewise.
22189 (aarch64_simd_emit_reg_reg_move): Likewise.
22190 (aarch64_simd_attr_length_rglist): Likewise.
22191 (aarch64_reverse_mask): Likewise.
22192 (aarch64_operands_ok_for_ldpstp): Likewise.
22193 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
22194 (aarch64_gen_adjusted_ldpstp): Likewise.
22195 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
22197 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
22198 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
22199 (arm_lookup_simd_builtin_type): Likewise.
22200 (arm_simd_builtin_type): Likewise.
22201 (arm_init_simd_builtin_types): Likewise.
22202 (arm_expand_builtin_args): Likewise.
22203 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
22204 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
22205 (ft32_setup_incoming_varargs): Likewise.
22206 (ft32_function_arg): Likewise.
22207 (ft32_function_arg_advance): Likewise.
22208 (ft32_pass_by_reference): Likewise.
22209 (ft32_arg_partial_bytes): Likewise.
22210 (ft32_valid_pointer_mode): Likewise.
22211 (ft32_addr_space_pointer_mode): Likewise.
22212 (ft32_addr_space_legitimate_address_p): Likewise.
22213 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
22215 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
22216 (ix86_emit_outlined_ms2sysv_restore): Likewise.
22217 (iamcu_alignment): Likewise.
22218 (canonicalize_vector_int_perm): Likewise.
22219 (ix86_noce_conversion_profitable_p): Likewise.
22220 (ix86_mpx_bound_mode): Likewise.
22221 (ix86_operands_ok_for_move_multiple): Likewise.
22222 * config/microblaze/microblaze-protos.h
22223 (microblaze_expand_conditional_branch_reg): Likewise.
22224 * config/microblaze/microblaze.c
22225 (microblaze_expand_conditional_branch_reg): Likewise.
22226 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
22228 (rs6000_reassociation_width): Likewise.
22229 (rs6000_invalid_binary_op): Likewise.
22230 (fusion_p9_p): Likewise.
22231 (emit_fusion_p9_load): Likewise.
22232 (emit_fusion_p9_store): Likewise.
22233 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
22235 (riscv_hard_regno_mode_ok_p): Likewise.
22236 (riscv_address_insns): Likewise.
22237 (riscv_split_symbol): Likewise.
22238 (riscv_legitimize_move): Likewise.
22239 (riscv_function_value): Likewise.
22240 (riscv_hard_regno_nregs): Likewise.
22241 (riscv_expand_builtin): Likewise.
22242 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
22243 (riscv_build_integer): Likewise.
22244 (riscv_split_integer): Likewise.
22245 (riscv_legitimate_constant_p): Likewise.
22246 (riscv_cannot_force_const_mem): Likewise.
22247 (riscv_regno_mode_ok_for_base_p): Likewise.
22248 (riscv_valid_base_register_p): Likewise.
22249 (riscv_valid_offset_p): Likewise.
22250 (riscv_valid_lo_sum_p): Likewise.
22251 (riscv_classify_address): Likewise.
22252 (riscv_legitimate_address_p): Likewise.
22253 (riscv_address_insns): Likewise.
22254 (riscv_load_store_insns): Likewise.
22255 (riscv_force_binary): Likewise.
22256 (riscv_split_symbol): Likewise.
22257 (riscv_force_address): Likewise.
22258 (riscv_legitimize_address): Likewise.
22259 (riscv_move_integer): Likewise.
22260 (riscv_legitimize_const_move): Likewise.
22261 (riscv_legitimize_move): Likewise.
22262 (riscv_address_cost): Likewise.
22263 (riscv_subword): Likewise.
22264 (riscv_output_move): Likewise.
22265 (riscv_canonicalize_int_order_test): Likewise.
22266 (riscv_emit_int_order_test): Likewise.
22267 (riscv_function_arg_boundary): Likewise.
22268 (riscv_pass_mode_in_fpr_p): Likewise.
22269 (riscv_pass_fpr_single): Likewise.
22270 (riscv_pass_fpr_pair): Likewise.
22271 (riscv_get_arg_info): Likewise.
22272 (riscv_function_arg): Likewise.
22273 (riscv_function_arg_advance): Likewise.
22274 (riscv_arg_partial_bytes): Likewise.
22275 (riscv_function_value): Likewise.
22276 (riscv_pass_by_reference): Likewise.
22277 (riscv_setup_incoming_varargs): Likewise.
22278 (riscv_print_operand): Likewise.
22279 (riscv_elf_select_rtx_section): Likewise.
22280 (riscv_save_restore_reg): Likewise.
22281 (riscv_for_each_saved_reg): Likewise.
22282 (riscv_register_move_cost): Likewise.
22283 (riscv_hard_regno_mode_ok_p): Likewise.
22284 (riscv_hard_regno_nregs): Likewise.
22285 (riscv_class_max_nregs): Likewise.
22286 (riscv_memory_move_cost): Likewise.
22287 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
22288 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
22289 (rl78_addr_space_address_mode): Likewise.
22290 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
22292 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
22293 (rs6000_reassociation_width): Likewise.
22294 (rs6000_invalid_binary_op): Likewise.
22295 (fusion_p9_p): Likewise.
22296 (emit_fusion_p9_load): Likewise.
22297 (emit_fusion_p9_store): Likewise.
22298 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
22299 (ok_for_simple_move_operands): Likewise.
22300 (ok_for_simple_move_strict_operands): Likewise.
22301 (ok_for_simple_arith_logic_operands): Likewise.
22302 (visium_legitimize_reload_address): Likewise.
22303 (visium_select_cc_mode): Likewise.
22304 (output_cbranch): Likewise.
22305 (visium_split_double_move): Likewise.
22306 (visium_expand_copysign): Likewise.
22307 (visium_expand_int_cstore): Likewise.
22308 (visium_expand_fp_cstore): Likewise.
22309 * config/visium/visium.c (visium_pass_by_reference): Likewise.
22310 (visium_function_arg): Likewise.
22311 (visium_function_arg_advance): Likewise.
22312 (visium_libcall_value): Likewise.
22313 (visium_setup_incoming_varargs): Likewise.
22314 (visium_legitimate_constant_p): Likewise.
22315 (visium_legitimate_address_p): Likewise.
22316 (visium_legitimize_address): Likewise.
22317 (visium_secondary_reload): Likewise.
22318 (visium_register_move_cost): Likewise.
22319 (visium_memory_move_cost): Likewise.
22320 (prepare_move_operands): Likewise.
22321 (ok_for_simple_move_operands): Likewise.
22322 (ok_for_simple_move_strict_operands): Likewise.
22323 (ok_for_simple_arith_logic_operands): Likewise.
22324 (visium_function_value_1): Likewise.
22325 (rtx_ok_for_offset_p): Likewise.
22326 (visium_legitimize_reload_address): Likewise.
22327 (visium_split_double_move): Likewise.
22328 (visium_expand_copysign): Likewise.
22329 (visium_expand_int_cstore): Likewise.
22330 (visium_expand_fp_cstore): Likewise.
22331 (visium_split_cstore): Likewise.
22332 (visium_select_cc_mode): Likewise.
22333 (visium_split_cbranch): Likewise.
22334 (output_cbranch): Likewise.
22335 (visium_print_operand_address): Likewise.
22336 * expmed.c (flip_storage_order): Likewise.
22337 * expmed.h (emit_cstore): Likewise.
22338 (flip_storage_order): Likewise.
22339 * genrecog.c (validate_pattern): Likewise.
22340 * hsa-gen.c (gen_hsa_addr): Likewise.
22341 * internal-fn.c (expand_arith_overflow): Likewise.
22342 * ira-color.c (allocno_copy_cost_saving): Likewise.
22343 * lra-assigns.c (find_hard_regno_for_1): Likewise.
22344 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
22345 (process_invariant_for_inheritance): Likewise.
22346 * lra-eliminations.c (move_plus_up): Likewise.
22347 * omp-low.c (lower_oacc_reductions): Likewise.
22348 * simplify-rtx.c (simplify_subreg): Likewise.
22349 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
22350 (TARGET_CHKP_BOUND_MODE): Likewise..
22351 * targhooks.c (default_chkp_bound_mode): Likewise.
22352 (default_setup_incoming_vararg_bounds): Likewise.
22353 * targhooks.h (default_chkp_bound_mode): Likewise.
22354 (default_setup_incoming_vararg_bounds): Likewise.
22355 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
22356 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
22357 (have_whole_vector_shift): Likewise.
22358 * tree-vect-stmts.c (vectorizable_load): Likewise.
22359 * doc/tm.texi: Regenerate.
22361 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
22363 Graceful degrade if Binutils PR21472 is not available.
22366 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
22367 .rodata in flash test fails.
22368 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
22369 * confgure: Regenerate.
22370 * config.in: Regenerate.
22371 * config/avr/avr.c (avr_asm_named_section)
22372 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
22373 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
22374 (avr_asm_init_sections): Same.
22376 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22378 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
22379 (fma<VH:mode>4_intrinsic): Likewise.
22380 (*fmsub<VCVTF:mode>4): Likewise.
22381 (*fmsub<VH:mode>4_intrinsic): Likewise.
22383 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
22386 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
22387 Don't depend on "optimize > 0".
22388 (out_movhi_r_mr, out_movqi_mr_r): Same.
22389 (out_movhi_mr_r, out_movqi_r_mr): Same.
22390 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
22391 io_address_operand on "optimize > 0".
22393 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22395 * tree-loop-distribution.c: Add general explanantion on the pass.
22396 (generate_loops_for_partition): Mark distributed loop.
22397 (pg_add_dependence_edges): New parameter. Handle alias data
22398 dependence specially and record it in the parameter if asked.
22399 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
22400 (init_partition_graph_vertices, add_partition_graph_edge): New.
22401 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
22402 (free_partition_graph_vdata, build_partition_graph): New.
22403 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
22404 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
22405 (data_ref_segment_size, latch_dominated_by_data_ref): New.
22406 (compute_alias_check_pairs, version_loop_by_alias_check): New.
22407 (version_for_distribution_p, finalize_partitions): New.
22408 (distribute_loop): Handle alias data dependence specially. Factor
22409 out loop fusion code as functions and call these functions.
22411 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22413 * tree-loop-distribution.c (classify_partition): New parameter and
22414 better handle reduction statement.
22415 (rdg_build_partitions): Revise comment.
22416 (distribute_loop): Compute statements in all partitions and pass it
22417 to classify_partition.
22419 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22421 * tree-loop-distribution.c (enum partition_type): New.
22422 (struct partition): New field type.
22423 (partition_merge_into): Add parameter. Update partition type.
22424 (data_dep_in_cycle_p, update_type_for_merge): New functions.
22425 (build_rdg_partition_for_vertex): Compute partition type.
22426 (rdg_build_partitions): Dump partition type.
22427 (distribute_loop): Update calls to partition_merge_into.
22429 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22431 * tree-loop-distribution.c (struct ddr_hasher): New.
22432 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
22434 (classify_partition): Call get_data_dependence.
22435 (pg_add_dependence_edges): Ditto.
22436 (distribute_loop): Release data dependence hash table.
22438 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22440 * tree-loop-distribution.c (ref_base_address): Delete.
22441 (similar_memory_accesses): Rename ...
22442 (share_memory_accesses): ... to this. Check if partitions access
22443 the same memory reference.
22444 (distribute_loop): Call share_memory_accesses.
22446 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22448 * tree-loop-distribution.c (struct partition): New field recording
22449 its data reference.
22450 (partition_alloc, partition_free): Init and release data refs.
22451 (partition_merge_into): Merge data refs.
22452 (build_rdg_partition_for_vertex): Collect data refs for partition.
22453 (pg_add_dependence_edges): Change parameters from vector to bitmap.
22455 (distribute_loop): Remve data refs from vertice data of partition
22458 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22460 * tree-loop-distribution.c (params.h): Include header file.
22461 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
22462 (datarefs_vec): New global var.
22463 (create_rdg_vertices): Use datarefs_vec directly.
22464 (free_rdg): Don't free data references.
22465 (build_rdg): Update use. Don't free data references.
22466 (distribute_loop): Compute global variable for data references.
22467 Bail out if there are too many data references.
22469 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22471 * tree-loop-distribution.c (loop_nest): New global var.
22472 (build_rdg): Use loop directly, rather than loop nest.
22473 (pg_add_dependence_edges): Remove loop nest parameter. Use global
22475 (distribute_loop): Compute global variable loop nest. Update use.
22477 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22479 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
22480 (partition_merge_into): New parameter. Dump reason for fusion.
22481 (distribute_loop): Update use of partition_merge_into.
22483 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22485 * tree-loop-distribution.c (bb_top_order_index): New.
22486 (bb_top_order_index_size, bb_top_order_cmp): New.
22487 (stmts_from_loop): Use topological order.
22488 (pass_loop_distribution::execute): Compute and release topological
22489 order for basic blocks.
22491 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22493 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
22496 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22498 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
22499 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
22500 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
22501 * internal-fn.def (LOOP_DIST_ALIAS): New.
22502 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
22503 (fold_loop_internal_call): ... this.
22504 (vect_loop_dist_alias_call): New function.
22505 (set_uid_loop_bbs): Call fold_loop_internal_call.
22506 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
22509 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22512 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
22513 Require dead FLAGS_REG at the beginning of a peephole.
22515 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22518 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
22519 arguments in the call to __builtin_ia32_sbb_u32.
22520 (_subborrow_u64): Swap _X and _Y arguments in the call to
22521 __builtin_ia32_sbb_u64.
22523 2017-07-04 Jakub Jelinek <jakub@redhat.com>
22526 * tree-vrp.c (compare_assert_loc): Turn into a function template
22527 with stable template parameter. Only test if a->e is NULL,
22528 !a->e == !b->e has been verified already. Use e == NULL or
22529 e != NULL instead of e or ! e tests. If stable is true, don't use
22530 iterative_hash_expr, on the other side allow a or b or both NULL
22531 and sort the NULLs last.
22532 (process_assert_insertions): Sort using compare_assert_loc<false>
22533 instead of compare_assert_loc, later sort using
22534 compare_assert_loc<true> before calling process_assert_insertions_for
22535 in a loop. Use break instead of continue once seen NULL pointer.
22537 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22539 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
22540 Cortex-R7 and Cortex-R8 processors.
22542 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22544 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
22545 uninitialized while src is not.
22547 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
22549 * common/config/arm/arm-common.c: Adjust include path for
22551 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
22552 (arm-cpu.h): Create in build directory. Adjust dependency rules.
22553 (arm-cpu-data.h): Likewise.
22554 (arm-cpu-cdata.h): Likewise.
22555 * config/arm/arm-cpu.h: Delete.
22556 * config/arm/arm-cpu-cdata.h: Delete.
22557 * config/arm/arm-cpu-data.h: Delete.
22559 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
22561 * config/arm/arm-cpus.in (cortex-a55): New.
22562 (cortex-a75): Likewise.
22563 (cortex-a75.cortex-a55): Likewise.
22564 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
22566 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
22567 * config/arm/arm-cpu-cdata.h: Regenerate.
22568 * config/arm/arm-cpu-data.h: Regenerate.
22569 * config/arm/arm-cpu.h: Regenerate.
22570 * config/arm/arm-tables.opt: Regenerate.
22571 * config/arm/arm-tune.md: Regenerate.
22573 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22575 * haifa-sched.c (sched_create_recovery_edges): Update profile.
22577 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22579 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
22582 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
22584 PR tree-optimization/81292
22585 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
22586 full_string_p, also call adjust_related_strinfos if the adjustment
22587 is simple, otherwise invalidate related strinfos.
22589 2017-07-04 Martin Liska <mliska@suse.cz>
22592 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
22593 newly created variable as DECL_IGNORED_P.
22595 2017-07-04 Martin Liska <mliska@suse.cz>
22598 * ipa-inline.c (inline_small_functions):
22599 Use xstrdup_for_dump.
22601 2017-07-04 Tom de Vries <tom@codesourcery.com>
22603 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
22605 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22608 * config/darwin.c (darwin_function_switched_text_sections):
22611 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
22613 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
22615 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22617 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
22619 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22621 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
22622 min_profitable_iters, and th as inclusive lower bounds.
22623 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
22624 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
22625 for min_profitable_iters and min_profitable_estimate.
22626 (vect_transform_loop): Treat th as an inclusive lower bound.
22627 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
22629 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22632 * config/darwin.c (darwin_function_switched_text_sections):
22633 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
22634 in two pieces, and suppress the use of buf.
22636 2017-07-03 Nathan Sidwell <nathan@acm.org>
22638 * hash-table.h (hash_table_mod1): Fix indentation.
22640 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22642 PR middle-end/81290
22643 * predict.c (force_edge_cold): Be more careful about propagation
22645 * profile-count.h (profile_probability::guessed,
22646 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
22648 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
22650 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
22652 * doc/invoke.texi (rcpc architecture extension): Document it.
22654 2017-07-03 Richard Biener <rguenther@suse.de>
22656 PR tree-optimization/60510
22657 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
22658 the scalar reduction PHI and use it.
22659 (vectorizable_reduction): Properly guard the single_defuse_cycle
22660 path for non-SLP reduction chains where we cannot use it.
22661 Rework reduc_def/index and vector type deduction. Rework
22662 vector operand gathering during reduction op code-gen.
22663 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
22664 chains dissolve the chain and leave it to non-SLP reduction
22667 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22669 * tree-data-ref.h (dr_alignment): Declare.
22670 * tree-data-ref.c (dr_alignment): New function.
22671 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
22672 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
22674 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
22676 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22678 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
22679 and base_misalignment fields.
22680 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
22681 * tree-data-ref.c: Include builtins.h.
22682 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
22683 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
22684 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
22685 * tree-vect-data-refs.c: Include tree-cfg.h.
22686 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
22687 fields instead of calculating an alignment here.
22688 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
22689 innermost_loop_behavior fields.
22691 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22693 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
22695 (DR_STEP_ALIGNMENT): New macro.
22696 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
22697 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
22698 (create_data_ref): Print it.
22699 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
22700 to tell whether the step preserves vector (mis)alignment.
22701 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
22702 Move the check for an integer step and generalise to all INTEGER_CST.
22703 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
22704 Print the outer step alignment.
22706 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22708 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
22709 with offset_alignment.
22710 (DR_ALIGNED_TO): Delete.
22711 (DR_OFFSET_ALIGNMENT): New macro.
22712 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
22713 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
22714 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
22715 (create_data_ref): Likewise.
22716 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
22717 (vect_analyze_data_refs): Likewise.
22718 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
22719 creating dummy innermost behavior.
22721 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22723 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
22724 with a "innermost_loop_behavior *" and refeence tree.
22725 * tree-data-ref.c (dr_analyze_innermost): Likewise.
22726 (create_data_ref): Update call accordingly.
22727 * tree-predcom.c (find_looparound_phi): Likewise.
22729 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22731 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
22732 fields with dr_wrt_vec_loop.
22733 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
22734 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
22735 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
22736 (vect_dr_behavior): New function.
22737 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
22738 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
22739 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
22740 track whether the step preserves the misalignment.
22741 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
22742 Use vect_dr_behavior.
22743 (vect_setup_realignment): Update call accordingly.
22744 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
22745 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
22746 call to vect_create_addr_base_for_vector_ref.
22747 (vect_create_cond_for_align_checks): Likewise.
22748 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
22749 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
22750 (vect_recog_mask_conversion_pattern): Likewise.
22751 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
22752 (new_stmt_vec_info): Remove redundant zeroing.
22754 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22756 * common/config/arm/arm-common.c (arm_be8_option): New function.
22757 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
22758 (ISA_ARMv6): Add isa_bit_be8.
22759 * config/arm/arm.h (arm_be8_option): Add prototype.
22760 (BE8_SPEC_FUNCTION): New define.
22761 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
22762 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
22763 (mlittle-endian): Similarly.
22764 (mbe8, mbe32): New options.
22765 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
22766 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
22768 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22770 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
22772 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22774 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
22775 (cleanup_tree_cfg_bb): Use it.
22776 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
22778 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
22780 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22783 * loop-doloop.c (add_test): Update profile.
22785 2017-07-03 Martin Liska <mliska@suse.cz>
22788 * sanopt.c (rewrite_usage_of_param): New function.
22789 (sanitize_rewrite_addressable_params): Likewise.
22790 (pass_sanopt::execute): Call rewrite_usage_of_param.
22792 2017-07-03 Richard Biener <rguenther@suse.de>
22794 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
22795 back to using VIEW_CONVERT_EXPR.
22797 2017-07-03 Martin Liska <mliska@suse.cz>
22800 * doc/extend.texi: Document when a resolver function is
22801 generated for target_clones.
22803 2017-07-03 Martin Liska <mliska@suse.cz>
22805 * asan.c (asan_emit_stack_protection): Unpoison just red zones
22806 and shadow memory of auto variables which are subject of
22807 use-after-scope sanitization.
22808 (asan_expand_mark_ifn): Add do set only when is_poison.
22810 2016-07-03 Richard Biener <rguenther@suse.de>
22812 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
22814 (vect_force_simple_reduction): Record reduction def -> phi mapping.
22815 (vectorizable_reduction): Perform reduction PHI creation when
22816 visiting a reduction PHI and adjust and simplify code generation
22817 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
22818 (vect_transform_loop): Visit reduction PHIs.
22819 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
22820 defs into the SLP tree.
22821 (vect_build_slp_tree): Reduction defs terminate the recursion.
22822 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
22824 (vect_get_vec_defs_for_stmt_copy): Export.
22825 (vect_get_vec_defs): Likewise.
22826 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
22828 (vect_get_vec_defs_for_stmt_copy): Declare.
22829 (vect_get_vec_defs): Likewise.
22831 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22833 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
22834 parameter with a "loop" parameter and use it instead of the
22835 loop containing DR_STMT. Don't check simple_iv when doing
22836 BB analysis. Describe the two analysis modes in the comment.
22838 2017-07-03 Tom de Vries <tom@codesourcery.com>
22840 PR tree-optimization/69468
22841 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
22842 (find_same_succ_bb): Handle ignore_edge_flags.
22844 2017-07-03 Tom de Vries <tom@codesourcery.com>
22846 PR tree-optimization/81192
22847 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
22849 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
22851 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
22853 2017-07-03 Tom de Vries <tom@codesourcery.com>
22855 PR tree-optimization/81192
22856 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
22857 BB_SAME_SUCC (bb) == NULL.
22859 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22861 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
22864 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22866 * dumpfile.c: Include profile-count.h
22867 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
22869 (insert_cond_bb): Update profile.
22870 * tree-cfg.h (insert_cond_bb): Update prototype.
22871 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
22872 * tree-dump.c: Do not include tree-cfg.
22874 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22876 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
22878 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22880 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
22883 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22885 * tree-complex.c (expand_complex_div_wide): update profile.
22887 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22888 Alan Hayward <alan.hayward@arm.com>
22889 David Sherwood <david.sherwood@arm.com>
22891 * Makefile.in (MACHMODE_H): Remove insn-modes.h
22892 (CORETYPES_H): New define.
22893 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
22894 (insn-modes-inline.h, s-modes-inline-h): New rules.
22895 (generated_files): Add insn-modes-inline.h.
22896 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
22897 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
22898 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
22899 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
22900 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
22901 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
22902 (build/gencodes.o, build/genconditions.o): Likewise.
22903 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
22904 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
22905 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
22906 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
22907 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
22908 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
22909 * coretypes.h: Include everything up to real.h for generators.
22910 Include insn-modes.h first. Include wide-int-print.h after
22911 wide-int.h. Include insn-modes-inline.h and then machmode.h.
22912 * machmode.h: Don't include insn-modes.h here.
22913 * function-tests.c: Remove includes of signop.h, machmode.h,
22914 double-int.h and wide-int.h.
22916 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
22918 * optc-save-gen.awk: Likewise.
22919 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
22920 * godump.c: Remove include of wide-int-print.h.
22921 * pretty-print.h: Likewise.
22922 * wide-int-print.cc: Likewise.
22923 * wide-int.cc: Likewise.
22924 * hash-map-tests.c: Remove include of signop.h.
22925 * hash-set-tests.c: Likewise.
22926 * rtl-tests.c: Likewise.
22927 * mkconfig.sh: Remove include of machmode.h.
22928 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
22930 (emit_insn_modes_inline_h): ...this new function. Emit the code
22931 into an insn-modes-inline.h header file, adding appropriate
22932 include guards and end comments.
22933 (emit_insn_modes_c_header): Remove include of machmode.h.
22934 (emit_min_insn_modes_c_header): Include coretypes.h rather than
22936 (main): Handle -i flag and call emit_insn_modes_inline_h when
22939 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22941 * tree-ssa-strlen.c (strinfo): Rename the length field to
22942 nonzero_chars. Add a full_string_p field.
22943 (compare_nonzero_chars, zero_length_string_p): New functions.
22944 (get_addr_stridx): Add an offset_out parameter.
22945 Use compare_nonzero_chars.
22946 (get_stridx): Update accordingly. Use compare_nonzero_chars.
22947 (new_strinfo): Update after above changes to strinfo.
22948 (set_endptr_and_length): Set full_string_p.
22949 (get_string_length): Update after above changes to strinfo.
22950 (unshare_strinfo): Update call to new_strinfo.
22951 (maybe_invalidate): Likewise.
22952 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
22953 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
22954 as a uhwi instead of an shwi. Update after above changes to
22955 strinfo and new_strinfo.
22956 (zero_length_string): Assert that chainsi contains full strings.
22957 Use zero_length_string_p. Update call to new_strinfo.
22958 (adjust_related_strinfos): Update after above changes to strinfo.
22959 Copy full_string_p from origsi.
22960 (adjust_last_stmt): Use zero_length_string_p.
22961 (handle_builtin_strlen): Update after above changes to strinfo and
22962 new_strinfo. Install the lhs as the string length if the previous
22963 entry didn't describe a full string.
22964 (handle_builtin_strchr): Update after above changes to strinfo
22966 (handle_builtin_strcpy): Likewise.
22967 (handle_builtin_strcat): Likewise.
22968 (handle_builtin_malloc): Likewise.
22969 (handle_pointer_plus): Likewise.
22970 (handle_builtin_memcpy): Likewise. Track nonzero characters
22971 that aren't necessarily followed by a nul terminator.
22972 (handle_char_store): Likewise.
22974 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22976 PR tree-optimization/80769
22977 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
22978 for malloc and calloc. Document the new invariant that all related
22979 strinfos have delayed lengths or none do.
22980 (verify_related_strinfos): Move earlier in file.
22981 (set_endptr_and_length): New function, split out from...
22982 (get_string_length): ...here. Also set the lengths of related
22984 (zero_length_string): Assert that chainsi has known (rather than
22986 (adjust_related_strinfos): Likewise.
22988 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22990 PR tree-optimization/81136
22991 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
22992 assert that two references with the same misalignment have the same
22993 compile-time misalignment if those compile-time misalignments
22996 2017-07-01 Andi Kleen <ak@linux.intel.com>
22998 * print-tree.c (print_node): Print all attributes.
23000 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23002 * cfg.c (scale_bbs_frequencies): New function.
23003 * cfg.h (scale_bbs_frequencies): Declare it.
23004 * cfgloopanal.c (single_likely_exit): Cleanup.
23005 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
23007 (scale_loop_profile): Likewise.
23008 (loop_version): Likewise.
23009 (create_empty_loop_on_edge): Update.
23010 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
23011 scale_loop_frequencies, scale_loop_profile, loopify,
23012 loop_version): Update prototypes.
23013 * modulo-sched.c (sms_schedule): Update.
23014 * predict.c (unlikely_executed_edge_p): Also check probability.
23015 (probably_never_executed_edge_p): Fix typo.
23016 * tree-if-conv.c (version_loop_for_if_conversion): Update.
23017 * tree-parloops.c (gen_parallel_loop): Update.
23018 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
23019 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23020 * tree-ssa-loop-split.c (split_loop): Update.
23021 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
23022 * tree-vect-loop-manip.c (vect_do_peeling): Update.
23023 (vect_loop_versioning): Update.
23024 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23026 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23028 * trans-mem.c (split_bb_make_tm_edge): Update profile.
23030 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23032 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
23033 to keep profile consistent.
23035 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23037 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
23038 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
23039 * profile-count.h (max_safe_multiplier): Make unsigned.
23040 (profile_count::guessed_zero): New.
23042 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23044 * bb-reorder.c (fix_up_crossing_landing_pad,
23045 fix_crossing_conditional_branches): Use make_single_succ_edge
23046 to keep profile consistent.
23048 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23050 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
23053 2017-07-01 Jakub Jelinek <jakub@redhat.com>
23056 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
23057 the right scopes, make sure cond_jump isn't preserved between multiple
23058 iterations. Search for fallthru edge whenever there are 3+ edges and
23059 use find_fallthru_edge for it.
23061 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23063 Patch by Alexander Monakov <amonakov@ispras.ru>
23064 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
23065 probabilities consistently.
23067 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23069 * pa.c (pa_expand_compare_and_swap_loop): Update call of
23070 emit_cmp_and_jump_insns.
23072 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23075 * tree-inline.c (expand_call_inline): Combine profile statuses.
23077 2017-06-30 Andrew Pinski <apinski@cavium.com>
23079 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
23080 fold_stmt returned true.
23082 2017-06-30 Nathan Sidwell <nathan@acm.org>
23084 * ggc.h (empty_string): Delete.
23085 * cfgexpand.c (expand_asm_stmt): Use plain "".
23086 * optabs.c (expand_asm_memory_barrier): Likewise.
23087 * stringpool.c (empty_string): Delete.
23088 (digit_vector, digit_string): Delete.
23089 (ggc_alloc_string): Use plain "", don't optimize single digit
23090 strings. Use ggc_alloc_atomic.
23092 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
23094 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
23095 comparison set and one other set, use the cost of the non-comparison
23098 2017-06-30 Nathan Sidwell <nathan@acm.org>
23100 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
23103 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
23105 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
23106 loops. Remove now unneeded calls to gimple_switch_set_label() that
23107 just set removed labels to NULL_TREE.
23109 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
23111 * tree-ssanames.c (set_range_info_raw): Abstract from ...
23112 (set_range_info): ...here. Only call set_range_info_raw if domain
23114 (set_nonzero_bits): Call set_range_info_raw.
23115 * tree-ssanames.h (set_range_info_raw): New.
23117 2017-06-30 Jakub Jelinek <jakub@redhat.com>
23120 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
23121 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
23122 of nonimmediate_operand and <store_mask_constraint> instead of m
23123 for the input operand. For V8FI iterator, always split if input
23124 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
23125 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
23126 <store_mask_predicate> instead of register_operand and
23127 <store_mask_constraint> instead of v for the input operand. Make
23128 sure both operands aren't MEMs for if not <mask_applied>.
23130 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
23132 * lto-wrapper.c (copy_file) Close both file descriptors before
23135 2017-06-30 Martin Liska <mliska@suse.cz>
23138 * multiple_target.c (create_dispatcher_calls): Make ifunc
23139 also for function that don't have calls or are not referenced.
23141 2017-06-30 Richard Biener <rguenther@suse.de>
23143 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
23144 analyze the first scalar stmt. Move vector type computation
23145 for the BB case here from ...
23146 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
23147 live operation processing in the SLP case properly.
23149 2017-06-30 Richard Biener <rguenther@suse.de>
23151 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
23153 2017-06-30 Martin Liska <mliska@suse.cz>
23156 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
23157 before BUILT_IN_UNWIND_RESUME when ASAN is used.
23159 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
23161 * doc/invoke.texi (AArch64): Add missing options and remove redundant
23164 2017-06-30 Richard Biener <rguenther@suse.de>
23166 PR tree-optimization/81249
23167 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
23168 condition reduction result to original scalar type.
23170 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23172 * profile-count.h (enum profile_quality): Fix typos and whitespace
23175 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23177 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
23178 type for branch probabilities.
23180 2017-06-29 Julian Brown <julian@codesourcery.com>
23181 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23183 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
23184 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
23185 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
23186 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
23188 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23190 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
23191 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
23192 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
23193 CC usage from generic code to here.
23194 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
23195 CC usage into the target macros.
23197 2017-06-29 Maya Rashish <coypu@sdf.org>
23199 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
23202 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23204 * arm/arm-builtins.c: Include profile-count.h
23205 * except.c (sjlj_emit_function_enter): Use
23206 profile_probability::unlikely.
23208 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23210 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
23211 and tocrel_offset be pointer args rather than implicitly using
23213 (legitimate_constant_pool_address_p, rs6000_emit_move,
23214 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
23215 tocrel_offset and use in toc_relative_expr_p call.
23216 (print_operand, print_operand_address): Use static tocrel_base_oac
23217 and tocrel_offset_oac.
23218 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
23221 2017-06-29 Maya Rashish <coypu@sdf.org>
23223 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
23225 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
23227 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
23228 objects, take into account only the alignment of 'op0' and 'mode1' if
23231 2017-06-29 Steve Ellcey <sellcey@cavium.com>
23233 * ccmp.c (ccmp_tree_comparison_p): New function.
23234 (ccmp_candidate_p): Update to use above function.
23235 (get_compare_parts): New function.
23236 (expand_ccmp_next): Update to use new functions.
23237 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
23239 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
23240 take mode as argument.
23241 * ccmp.h (expand_ccmp_expr): Add mode as argument.
23242 * expr.c (expand_expr_real_1): Pass mode as argument.
23244 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
23246 * combine.c (combine_instructions): Print insns to dump_file, together
23249 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23251 * asan.c (asan_emit_stack_protection): Update.
23252 (create_cond_insert_point): Update.
23253 * auto-profile.c (afdo_propagate_circuit): Update.
23254 * basic-block.h (struct edge_def): Turn probability to
23255 profile_probability.
23256 (EDGE_FREQUENCY): Update.
23257 * bb-reorder.c (find_traces_1_round): Update.
23258 (better_edge_p): Update.
23259 (sanitize_hot_paths): Update.
23260 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
23261 (make_single_succ_edge): Update.
23262 (check_bb_profile): Update.
23263 (dump_edge_info): Update.
23264 (update_bb_profile_for_threading): Update.
23265 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
23266 probabilitycount to 0.
23267 * cfgbuild.c (compute_outgoing_frequencies): Update.
23268 * cfgcleanup.c (try_forward_edges): Update.
23269 (outgoing_edges_match): Update.
23270 (try_crossjump_to_edge): Update.
23271 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
23272 (expand_gimple_tailcall): Update.
23273 (construct_init_block): Use make_single_succ_edge.
23274 (construct_exit_block): Use make_single_succ_edge.
23275 * cfghooks.c (verify_flow_info): Update.
23276 (redirect_edge_succ_nodup): Update.
23277 (split_edge): Update.
23278 (account_profile_record): Update.
23279 * cfgloopanal.c (single_likely_exit): Update.
23280 * cfgloopmanip.c (scale_loop_profile): Update.
23281 (set_zero_probability): Remove.
23282 (duplicate_loop_to_header_edge): Update.
23283 * cfgloopmanip.h (loop_version): Update prototype.
23284 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
23285 (force_nonfallthru_and_redirect): Update.
23286 (update_br_prob_note): Update.
23287 (rtl_verify_edges): Update.
23288 (purge_dead_edges): Update.
23289 (rtl_lv_add_condition_to_bb): Update.
23290 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
23291 * cgraphunit.c (init_lowered_empty_function): Update.
23292 (cgraph_node::expand_thunk): Update.
23293 * cilk-common.c: Include profile-count.h
23294 * dojump.c (inv): Remove.
23295 (jumpifnot): Update.
23296 (jumpifnot_1): Update.
23297 (do_jump_1): Update.
23299 (do_jump_by_parts_greater_rtx): Update.
23300 (do_compare_rtx_and_jump): Update.
23301 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
23302 do_jump_1. do_compare_rtx_and_jump): Update prototype.
23303 * dwarf2cfi.c: Include profile-count.h
23304 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
23305 (sjlj_emit_dispatch_table): Likewise.
23306 * explow.c: Include profile-count.h
23307 * expmed.c (emit_store_flag_force): Update.
23308 (do_cmp_and_jump): Update.
23309 * expr.c (compare_by_pieces_d::generate): Update.
23310 (compare_by_pieces_d::finish_mode): Update.
23311 (emit_block_move_via_loop): Update.
23312 (store_expr_with_bounds): Update.
23313 (store_constructor): Update.
23314 (expand_expr_real_2): Update.
23315 (expand_expr_real_1): Update.
23316 * expr.h (try_casesi, try_tablejump): Update prototypes.
23317 * gimple-pretty-print.c (dump_probability): Update.
23318 (dump_profile): New.
23319 (dump_gimple_label): Update.
23320 (dump_gimple_bb_header): Update.
23321 * graph.c (draw_cfg_node_succ_edges): Update.
23322 * hsa-gen.c (convert_switch_statements): Update.
23323 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
23324 (find_if_case_1): Update.
23325 (find_if_case_2): Update.
23326 * internal-fn.c (expand_arith_overflow_result_store): Update.
23327 (expand_addsub_overflow): Update.
23328 (expand_neg_overflow): Update.
23329 (expand_mul_overflow): Update.
23330 (expand_vector_ubsan_overflow): Update.
23331 * ipa-cp.c (good_cloning_opportunity_p): Update.
23332 * ipa-split.c (split_function): Use make_single_succ_edge.
23333 * ipa-utils.c (ipa_merge_profiles): Update.
23334 * loop-doloop.c (add_test): Update.
23335 (doloop_modify): Update.
23336 * loop-unroll.c (compare_and_jump_seq): Update.
23337 (unroll_loop_runtime_iterations): Update.
23338 * lra-constraints.c (lra_inheritance): Update.
23339 * lto-streamer-in.c (input_cfg): Update.
23340 * lto-streamer-out.c (output_cfg): Update.
23341 * mcf.c (adjust_cfg_counts): Update.
23342 * modulo-sched.c (sms_schedule): Update.
23343 * omp-expand.c (expand_omp_for_init_counts): Update.
23344 (extract_omp_for_update_vars): Update.
23345 (expand_omp_ordered_sink): Update.
23346 (expand_omp_for_ordered_loops): Update.
23347 (expand_omp_for_generic): Update.
23348 (expand_omp_for_static_nochunk): Update.
23349 (expand_omp_for_static_chunk): Update.
23350 (expand_cilk_for): Update.
23351 (expand_omp_simd): Update.
23352 (expand_omp_taskloop_for_outer): Update.
23353 (expand_omp_taskloop_for_inner): Update.
23354 * omp-simd-clone.c (simd_clone_adjust): Update.
23355 * optabs.c (expand_doubleword_shift): Update.
23356 (expand_abs): Update.
23357 (emit_cmp_and_jump_insn_1): Update.
23358 (expand_compare_and_swap_loop): Update.
23359 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
23360 * predict.c (predictable_edge_p): Update.
23361 (edge_probability_reliable_p): Update.
23362 (set_even_probabilities): Update.
23363 (combine_predictions_for_insn): Update.
23364 (combine_predictions_for_bb): Update.
23365 (propagate_freq): Update.
23366 (estimate_bb_frequencies): Update.
23367 (force_edge_cold): Update.
23368 * profile-count.c (profile_count::dump): Add missing space into dump.
23369 (profile_count::debug): Add newline.
23370 (profile_count::differs_from_p): Explicitly convert to unsigned.
23371 (profile_count::stream_in): Update.
23372 (profile_probability::dump): New member function.
23373 (profile_probability::debug): New member function.
23374 (profile_probability::differs_from_p): New member function.
23375 (profile_probability::differs_lot_from_p): New member function.
23376 (profile_probability::stream_in): New member function.
23377 (profile_probability::stream_out): New member function.
23378 * profile-count.h (profile_count_quality): Rename to ...
23379 (profile_quality): ... this one.
23380 (profile_probability): New.
23381 (profile_count): Update.
23382 * profile.c (compute_branch_probabilities): Update.
23383 * recog.c (peep2_attempt): Update.
23384 * sched-ebb.c (schedule_ebbs): Update.
23385 * sched-rgn.c (find_single_block_region): Update.
23386 (compute_dom_prob_ps): Update.
23387 (schedule_region): Update.
23388 * sel-sched-ir.c (compute_succs_info): Update.
23389 * stmt.c (struct case_node): Update.
23390 (do_jump_if_equal): Update.
23391 (get_outgoing_edge_probs): Update.
23392 (conditional_probability): Update.
23393 (emit_case_dispatch_table): Update.
23394 (expand_case): Update.
23395 (expand_sjlj_dispatch_table): Update.
23396 (emit_case_nodes): Update.
23397 * targhooks.c: Update.
23398 * tracer.c (better_p): Update.
23399 (find_best_successor): Update.
23400 * trans-mem.c (expand_transaction): Update.
23401 * tree-call-cdce.c: Update.
23402 * tree-cfg.c (gimple_split_edge): Upate.
23403 (move_sese_region_to_fn): Upate.
23404 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
23405 * tree-eh.c (lower_resx): Upate.
23406 (cleanup_empty_eh_move_lp): Upate.
23407 * tree-if-conv.c (version_loop_for_if_conversion): Update.
23408 * tree-inline.c (copy_edges_for_bb): Update.
23409 (copy_cfg_body): Update.
23410 * tree-parloops.c (gen_parallel_loop): Update.
23411 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
23412 (gimple_gen_time_profiler): Update.
23413 * tree-ssa-dce.c (remove_dead_stmt): Update.
23414 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
23415 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
23416 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
23417 (unloop_loops): Update.
23418 (try_peel_loop): Update.
23419 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23420 * tree-ssa-loop-split.c (connect_loops): Update.
23421 (split_loop): Update.
23422 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
23423 (hoist_guard): Update.
23424 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
23425 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
23426 (value_replacement): Update.
23427 * tree-ssa-reassoc.c (branch_fixup): Update.
23428 * tree-ssa-tail-merge.c (replace_block_by): Update.
23429 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
23430 (create_edge_and_update_destination_phis): Update.
23431 (compute_path_counts): Update.
23432 (recompute_probabilities): Update.
23433 (update_joiner_offpath_counts): Update.
23434 (freqs_to_counts_path): Update.
23435 (duplicate_thread_path): Update.
23436 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
23437 (struct switch_conv_info): Update.
23438 (gen_inbound_check): Update.
23439 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
23440 (vect_do_peeling): Update.
23441 (vect_loop_versioning): Update.
23442 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23443 (optimize_mask_stores): Update.
23444 * ubsan.c (ubsan_expand_null_ifn): Update.
23445 * value-prof.c (gimple_divmod_fixed_value): Update.
23446 (gimple_divmod_fixed_value_transform): Update.
23447 (gimple_mod_pow2): Update.
23448 (gimple_mod_pow2_value_transform): Update.
23449 (gimple_mod_subtract): Update.
23450 (gimple_mod_subtract_transform): Update.
23451 (gimple_ic): Update.
23452 (gimple_stringop_fixed_value): Update.
23453 (gimple_stringops_transform): Update.
23454 * value-prof.h: Update.
23456 2017-06-29 Carl Love <cel@us.ibm.com>
23458 * config/rs6000/rs6000-c.c: Add support for built-in functions
23459 vector signed int vec_signed (vector float);
23460 vector signed long long vec_signed (vector double);
23461 vector signed int vec_signed2 (vector double, vector double);
23462 vector signed int vec_signede (vector double);
23463 vector signed int vec_signedo (vector double);
23464 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
23465 instruction generator.
23466 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23467 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
23468 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
23470 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
23471 vunsignede_v2df): Add define_expands.
23472 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
23473 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
23474 VEC_UNSIGNEDO): Add definitions.
23475 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23476 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
23477 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
23478 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
23479 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
23480 * config/rs6000/altivec.h (vec_signed, vec_signed2,
23481 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
23482 vec_unsignede, vec_unsignedo): Add builtin defines.
23483 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
23485 * doc/extend.texi: Update the built-in documentation file for the
23486 new built-in functions.
23488 2017-06-29 Richard Biener <rguenther@suse.de>
23490 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
23491 reduction chains to LOOP_VINFO_REDUCTIONS.
23492 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
23493 SLP reductions after processing reduction chains.
23495 2017-06-29 Nathan Sidwell <nathan@acm.org>
23497 * builtins.c (fold_builtin_FUNCTION): Use
23498 lang_hooks.decl_printable_name.
23500 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
23502 PR middle-end/81194
23503 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
23504 with only one label.
23505 * stmt.c (expand_case): Assert NCASES is greater than one.
23507 2017-06-29 Richard Biener <rguenther@suse.de>
23509 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
23511 (group_case_labels): Likewise.
23512 (find_taken_edge): Push sanity checking on val to workers...
23513 (find_taken_edge_cond_expr): ... here
23514 (find_taken_edge_switch_expr): ... and here, handle cases
23515 with just a default label.
23516 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
23517 (group_case_labels): Likewise.
23518 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
23519 group_case_labels does anything cleanup the CFG again.
23521 2017-06-29 Bin Cheng <bin.cheng@arm.com>
23523 PR tree-optimization/81196
23524 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
23525 exit condition comparing two IVs.
23527 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
23529 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
23530 profile to the dummy entry at the end of the list of architectures.
23531 * config/arm/arm-cpu-cdata.h: Regenerated.
23533 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23534 Michael Collison <michael.collison@arm.com>
23537 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
23539 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
23540 (*aarch64_reg_<mode>3_minus_mask): New pattern.
23541 (*aarch64_<optab>_reg_di3_mask2): New pattern.
23542 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
23543 of shift when the shift amount is masked with constant equal to
23544 the size of the mode.
23545 * config/aarch64/predicates.md (subreg_lowpart_operator): New
23548 2017-06-29 Martin Liska <mliska@suse.cz>
23550 * config/i386/i386.opt: Change range from [1,5] to [0,5].
23552 2017-06-29 Yury Gribov <tetra2005@gmail.com>
23555 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
23558 (edge_growth_cache_entry::edge_growth_cache_entry): New
23560 (reset_edge_growth_cache): Update to use constructor.
23562 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23564 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
23565 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23566 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23568 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
23570 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
23571 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
23573 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
23575 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
23576 (*-linux-uclibc*): Add t-uclibc tmake_file.
23577 * config/t-musl: New.
23578 * config/t-uclibc: New.
23580 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
23582 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
23584 (gen_comm_data): Emit architectural setting of arch_prof.
23585 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
23587 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
23588 (armv8-m.base, armv8-m.main): Likewise.
23589 * arm-protos.h (arm_build_target): Add profile field.
23590 (arch_option): Likewise.
23591 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
23593 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
23594 arm_active_target.profile.
23596 2017-06-28 Richard Biener <rguenther@suse.de>
23598 PR middle-end/81227
23599 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
23600 TYPE_OVERFLOW_WRAPS.
23601 * match.pd (negate_expr_p): Likewise.
23602 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
23603 fold_build2, not fold_binary.
23605 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23607 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
23608 Convert memory address to Pmode.
23609 (aarch64_print_operand): Assert MEM operands are always Pmode.
23611 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23614 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
23615 Remove redundant if.
23616 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
23617 * config/arm/aarch-common-protos.h
23618 (aarch_forward_to_shift_is_not_shifted_re): Remove.
23619 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
23621 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
23624 * multiple_target.c (create_dispatcher_calls): Set the default
23625 clone to be static, not public.
23627 2017-06-28 Richard Biener <rguenther@suse.de>
23629 * tree-vect-loop.c (vectorizable_reduction): Move special
23630 cond reduction IV var creation ...
23631 (vect_create_epilog_for_reduction): ... here. Remove induction_index
23632 parameter. Use STMT_VINFO_VECTYPE.
23633 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
23636 2017-06-28 Martin Liska <mliska@suse.cz>
23639 * ipa-visibility.c (non_local_p): Handle visibility.
23641 2017-06-28 Martin Liska <mliska@suse.cz>
23644 * common.opt: Add IntegerRange to various options.
23645 * opt-functions.awk (integer_range_info): New function.
23646 * optc-gen.awk: Add integer_range_info to cl_options struct.
23647 * opts-common.c (decode_cmdline_option): Handle
23648 CL_ERR_INT_RANGE_ARG.
23649 (cmdline_handle_error): Likewise.
23650 * opts.c (print_filtered_help): Show valid interval in
23651 when --help is provided.
23652 * opts.h (struct cl_option): Add range_min and range_max fields.
23653 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
23655 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
23657 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
23658 (x * C EQ/NE y * C): New transformation.
23660 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
23662 * genmultilib (combination_space): Accept '+' in option names.
23664 2017-06-28 Martin Liska <mliska@suse.cz>
23667 * asan.c (instrument_derefs): Bail out inner references
23668 that are hard register variables.
23670 2017-06-28 Jakub Jelinek <jakub@redhat.com>
23673 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
23674 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
23676 2017-06-28 Richard Biener <rguenther@suse.de>
23678 * tree-vectorizer.h (vect_get_vec_defs): Remove.
23679 (vect_get_slp_defs): Adjust.
23680 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
23682 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
23684 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
23685 get_initial_defs_for_reduction instead of vect_get_vec_defs.
23686 (vectorizable_reduction): Adjust.
23687 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
23689 (vect_get_slp_defs): Likewise.
23690 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
23691 (vectorizable_bswap): Adjust.
23692 (vectorizable_call): Likewise.
23693 (vectorizable_conversion): Likewise.
23694 (vectorizable_assignment): Likewise.
23695 (vectorizable_shift): Likewise.
23696 (vectorizable_operation): Likewise.
23697 (vectorizable_store): Likewise.
23698 (vectorizable_condition): Likewise.
23699 (vectorizable_comparison): Likewise.
23701 2017-06-28 Michael Collison <michael.collison@arm.com>
23704 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
23706 (arm_gen_movmemqi): Removed unused variable 'i'.
23707 Convert 'for' loop into 'while' loop.
23708 (arm_expand_prologue): Remove last unnecessary set of insn.
23709 (thumb_pop): Remove unused variable 'pushed_words'.
23710 (thumb_exit): Remove last unnecessary set of regs_to_pop.
23712 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23714 * config/s390/predicates.md: Use s390_rel_address_ok_p.
23715 * config/s390/s390-protos.h: Add prototype of
23716 s390_rel_address_ok_p.
23717 * config/s390/s390.c (s390_got_symbol): New function.
23718 (s390_rel_address_ok_p): New function.
23719 (legitimize_pic_address): Use s390_rel_address_ok_p.
23720 (s390_load_got): Use s390_got_symbol.
23721 (s390_option_override): Issue error if
23722 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
23723 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
23725 * config/s390/s390.opt: New option mpic-data-is-text-relative.
23727 2017-06-27 Andrew Pinski <apinski@cavium.com>
23729 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
23730 (X * copysign (1.0, X)): New pattern.
23731 (X * copysign (1.0, -X)): New pattern.
23732 (copysign (-1.0, CST)): New pattern.
23734 2017-06-27 Joseph Myers <joseph@codesourcery.com>
23736 * genmultilib (combination_space): Remove variable.
23737 Validate reuse rules against regular expression for any sequence
23738 of multilib options in any order.
23740 2017-06-27 Michael Collison <michael.collison@arm.com>
23742 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
23743 call aarch64_split_simd_combine.
23744 * (aarch64_combine_internal<mode>): Delete pattern.
23745 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
23746 Allow register and subreg operands.
23748 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23750 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
23751 specific need, just fallback on defaults.
23752 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
23754 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23755 Olivier Hainque <hainque@adacore.com>
23757 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
23759 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
23760 targets. Pick a default if no particular attempt applied.
23761 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
23764 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23766 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
23767 (x86_64-wrs-vxworks7): Likewise.
23769 2017-06-27 Marek Polacek <polacek@redhat.com>
23772 * ubsan.c (instrument_null): Check get_base_address's result for null.
23774 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
23776 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
23778 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
23780 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
23781 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
23782 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
23783 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
23784 New function types.
23785 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
23786 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
23787 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
23788 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
23789 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
23790 BUILT_IN_FEUPDATEENV): New builtins.
23791 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
23792 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
23793 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
23794 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
23796 (builtin_structptr_types): Adjust size.
23797 * tree.c (builtin_structptr_types): Add four entries.
23799 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23800 Olivier Hainque <hainque@adacore.com>
23802 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
23803 (TLS_SYM): New local macro, forcing reference to __tls__ on
23804 link command lines for VxWorks 7 RTPs, triggering initialization
23806 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
23807 OS features TLS support, true for RTPs on VxWorks 7.
23808 * config/vxworks.c (vxworks_override_options): Setup emutls
23811 2017-06-27 Jakub Jelinek <jakub@redhat.com>
23813 * predict.c (test_prediction_value_range): Use -1U instead of -1
23814 to avoid narrowing conversion warning.
23815 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
23816 to avoid narrowing conversion warning.
23817 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
23819 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
23821 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23823 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
23824 64bit configurations.
23825 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
23826 (SIZE_TYPE): Likewise.
23827 * config/vxworks.c (vxworks_emutls_var_fields): Use
23828 long_unsigned_type_node instead of unsigned_type_node as the offset
23829 field type, which is "pointer" mode in emutls.c.
23831 2017-06-27 Jakub Jelinek <jakub@redhat.com>
23834 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
23836 PR middle-end/81207
23837 * gimple-fold.c (replace_call_with_call_and_fold): Handle
23838 gimple_vuse copying separately from gimple_vdef copying.
23840 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23842 * value-prof.c (free_hist): Remove call to memset and the enclosing if
23845 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
23846 Olivier Hainque <hainque@adacore.com>
23848 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
23849 for all vxworks7 targets.
23850 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
23851 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
23852 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
23853 variations for VX6/VX7 and 32/64bits later on in ...
23854 (VXWORKS_LIB_SPEC): Leverage new macros.
23855 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
23856 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
23858 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
23860 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
23861 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
23863 2017-06-26 Carl Love <cel@us.ibm.com>
23865 * config/rs6000/rs6000-c.c: Add support for built-in functions
23866 vector bool char vec_reve (vector bool char);
23867 vector signed char vec_reve (vector signed char);
23868 vector unsigned char vec_reve (vector unsigned char);
23869 vector bool int vec_reve (vector bool int);
23870 vector signed int vec_reve (vector signed int);
23871 vector unsigned int vec_reve (vector unsigned int);
23872 vector bool long long vec_reve (vector bool long long);
23873 vector signed long long vec_reve (vector signed long long);
23874 vector unsigned long long vec_reve (vector unsigned long long);
23875 vector bool short vec_reve (vector bool short);
23876 vector signed short vec_reve (vector signed short);
23877 vector double vec_reve (vector double);
23878 vector float vec_reve (vector float);
23879 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
23880 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
23881 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
23882 (altivec_vreve): New pattern.
23883 * config/rs6000/altivec.h (vec_reve): New define.
23884 * doc/extend.texi (vec_rev): Update the built-in documentation file
23885 for the new built-in functions.
23887 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23889 PR tree-optimization/71815
23890 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
23892 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
23894 (slsr_process_phi): Likewise.
23895 (replace_uncond_cands_and_profitable_phis): Don't replace a
23896 multiply candidate with a stride of 1 (copy or cast).
23897 (phi_incr_cost): Call uses_consumed_by_stmt rather than
23899 (lowest_cost_path): Likewise.
23900 (total_savings): Likewise.
23902 2017-06-26 Richard Biener <rguenther@suse.de>
23905 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
23906 Use def_builtin_pure for all gather builtins.
23908 2017-06-26 Richard Biener <rguenther@suse.de>
23910 PR tree-optimization/81203
23911 * tree-tailcall.c (find_tail_calls): Do not move stmts into
23912 non-dominating BBs.
23914 2017-06-26 Marek Polacek <polacek@redhat.com>
23917 * doc/invoke.texi: Document -Wmultistatement-macros.
23919 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
23921 * doc/sourcebuild.texi (ARM-specific attributes): Document new
23922 arm_neon_ok_no_float_abi effective target.
23924 2017-06-26 Richard Biener <rguenther@suse.de>
23926 PR tree-optimization/80928
23927 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
23928 (copy_bbs): Set BB_DUPLICATED flag early.
23929 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
23931 (execute_on_shrinking_pred): Likewise.
23932 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
23933 BB_DUPLICATED blocks.
23934 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
23935 iterate over all PHIs considering removal of *gsi.
23937 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
23939 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
23942 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23944 * config/rs6000/rs6000-string.c: (expand_block_clear,
23945 do_load_for_compare, select_block_compare_mode,
23946 compute_current_alignment, expand_block_compare,
23947 expand_strncmp_align_check, expand_strn_compare,
23948 expand_block_move, rs6000_output_load_multiple)
23949 Move functions related to string/block move/compare
23950 to a separate file.
23951 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
23952 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
23953 for this function which is now used in two files.
23954 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
23955 * config.gcc: Add rs6000-string.o to extra_objs for
23956 targets powerpc*-*-* and rs6000*-*-*.
23958 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
23961 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
23962 32-bit, since indexed is not valid for DImode.
23963 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
23964 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
23965 (define_peephole2 for Altivec d-form load): Add 32-bit support.
23966 (define_peephole2 for Altivec d-form store): Likewise.
23969 * multiple_target.c (create_dispatcher_calls): Only create the
23970 dispatcher call if the function is the default clone of a
23971 versioned function.
23973 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
23975 PR middle-end/80902
23976 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
23977 a call, force the call to not be a tail call.
23979 2017-06-23 Jeff Law <law@redhat.com>
23981 * doc/contrib.texi: Add entry for Steven Pemberton's work on
23984 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
23986 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
23987 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
23988 handling for early expansion of vector shifts (sl,sr,sra,rl).
23989 (builtin_function_type): Add vector shift right instructions
23990 to the unsigned argument list.
23992 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
23994 rtl-optimizatoin/79286
23995 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
23996 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
23997 trap. PIC register plus a const unspec without offset can never trap.
23999 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
24001 * tree.h (builtin_structptr_type): New type.
24002 (builtin_structptr_types): Declare new array.
24003 * tree.c (builtin_structptr_types): New array.
24004 (free_lang_data, build_common_tree_nodes): Use it.
24006 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
24009 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
24012 2017-06-22 Matt Turner <mattst88@gmail.com>
24014 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
24015 Lake models to skylake case. Assume skylake for unknown
24016 models with clflushopt.
24018 2017-06-22 Jeff Law <law@redhat.com>
24020 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
24021 frame sizes that do not satisfy aarch64_uimm12_shift.
24023 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
24025 * profile-count.h (apply_probability,
24026 apply_scale, probability_in): Fix checks for zero.
24028 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24030 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
24031 * doc/cppdiropts.texi (-I @var{dir}): Document it.
24033 2016-06-22 Richard Biener <rguenther@suse.de>
24035 * tree-vect-loop.c (vect_model_reduction_cost): Handle
24036 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
24037 REDUC_MAX_EXPR support.
24038 (vectorizable_reduction): Likewise.
24039 (vect_create_epilog_for_reduction): Likewise.
24041 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
24043 * match.pd (A / (1 << B) -> A >> B): New.
24044 * generic-match-head.c: Include optabs-tree.h.
24045 * gimple-match-head.c: Likewise.
24046 * optabs-tree.h (target_supports_op_p): New.
24047 * optabs-tree.c (target_supports_op_p): New.
24049 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24051 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
24052 $gcc_cv_ld --help output.
24053 (gcc_cv_ld_demangle): Likewise.
24054 (gcc_cv_ld_eh_frame_hdr): Likewise.
24055 (gcc_cv_ld_pie): Likewise.
24056 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
24057 (gcc_cv_ld_buildid): Likewise.
24058 (gcc_cv_ld_sysroot): Likewise.
24059 (ld_bndplt_support): Likewise.
24060 (ld_pushpopstate_support): Likewise.
24061 * configure: Regenerate.
24062 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
24064 2017-06-21 Jakub Jelinek <jakub@redhat.com>
24067 * config/i386/sse.md (round<mode>2): Renumber match_dup and
24068 operands indexes to avoid gap between operands and match_dups.
24070 2017-06-21 Andrew Pinski <apinski@cavium.com>
24072 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
24073 Increment Arith_shift and Arith_shift_reg by 1.
24074 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
24076 * config/aarch64/aarch64.c (thunderx_tunings): Enable
24077 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
24078 (aarch64_strip_extend): Add new argument and test for it.
24079 (aarch64_cheap_mult_shift_p): New function.
24080 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
24081 add a cost if it is true.
24082 Update calls to aarch64_strip_extend.
24083 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
24085 2017-06-21 Andrew Pinski <apinski@cavium.com>
24087 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
24089 (thunderxt88): Likewise.
24090 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
24091 (thunderx_prefetch_tune): New variable.
24092 (thunderx2t99_prefetch_tune): Update for the correct values.
24093 (thunderxt88_tunings): New variable.
24094 (thunderx_tunings): Use thunderx_prefetch_tune instead of
24095 generic_prefetch_tune.
24096 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
24098 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24100 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
24101 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
24102 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
24103 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
24104 (aarch64_atomic_cas<mode>, GPI): Likewise.
24106 2017-06-21 Martin Liska <mliska@suse.cz>
24108 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
24109 statements on cold and hot labels.
24110 * predict.c (tree_estimate_probability_bb): Remove the
24111 prediction from this place.
24113 2017-06-21 Martin Liska <mliska@suse.cz>
24115 PR tree-optimization/79489
24116 * gimplify.c (maybe_add_early_return_predict_stmt): New
24118 (gimplify_return_expr): Call the function.
24119 * predict.c (tree_estimate_probability_bb): Remove handling
24121 * predict.def: Update comment about early return predictor.
24122 * gimple-predict.h (is_gimple_predict): New function.
24123 * predict.def: Change default value of early return to 66.
24124 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
24126 * passes.def: Put pass_strip_predict_hints to the beginning of
24129 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
24131 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
24132 FUNCTION_DECL declarations.
24133 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
24135 (dwaf2out_decl): Likewise.
24136 * godump.c (go_early_global_decl): Skip call to the real debug hook
24137 for FUNCTION_DECL declarations.
24138 * passes.c (rest_of_decl_compilation): Skip call to the
24139 early_global_decl debug hook for FUNCTION_DECL declarations, unless
24140 -fdump-go-spec is passed.
24142 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
24144 * config/i386/i386.c (struct builtin_isa): New field pure_p.
24145 Reorder for compactness.
24146 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
24147 (def_builtin_pure, def_builtin_pure2): New functions.
24148 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
24150 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
24152 * match.pd (nop_convert): New predicate.
24153 ((A +- CST1) +- CST2): Allow some NOP conversions.
24155 2017-06-21 Jakub Jelinek <jakub@redhat.com>
24158 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
24159 with ctors/dtors if GOVD_SHARED is set.
24161 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
24163 * config/aarch64/aarch64.md (movti_aarch64):
24164 Emit mov rather than orr.
24165 (movtf_aarch64): Likewise.
24166 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
24167 Emit mov rather than orr.
24169 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
24171 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
24172 Swap alternatives, make integer dup more expensive.
24174 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
24176 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
24177 Return true for non-tls symbols.
24179 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
24181 * config/aarch64/aarch64-cores.def (cortex-a55): New.
24182 (cortex-a75): Likewise.
24183 (cortex-a75.cortex-a55): Likewise.
24184 * config/aarch64/aarch64-tune.md: Regenerate.
24185 * doc/invoke.texi (-mtune): Document new values for -mtune.
24187 2017-06-21 Tom de Vries <tom@codesourcery.com>
24189 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
24190 stack_size feature.
24191 (Effective-Target Keywords, Other attributes): Suggest using
24192 dg-add-options stack_size feature to get stack limit in stack_size
24193 effective target documentation.
24195 2017-06-21 Julian Brown <julian@codesourcery.com>
24196 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24198 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
24199 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
24200 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
24202 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
24203 attribute type list for neon_multiply.
24204 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
24205 attribute type list for neon_multiply.
24206 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
24207 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
24208 attribute type list for neon_multiply.
24209 * config/arm/types.md (crypto_pmull): Add.
24210 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
24211 attribute type list.
24213 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
24215 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
24218 2017-06-20 Jakub Jelinek <jakub@redhat.com>
24220 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
24221 to make sure not to dereference a NULL cost_classes_ptr pointer.
24223 2017-06-20 Carl Love <cel@us.ibm.com>
24225 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24226 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
24227 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
24228 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
24229 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
24230 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
24231 VMULOSW): New enum "unspec" values.
24232 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
24233 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
24234 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
24235 altivec_vmulosw): New patterns.
24236 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
24237 VMULOSW): Add definitions.
24239 2017-06-20 Julia Koval <julia.koval@intel.com>
24241 * config/i386/i386.c: Fix rounding expand for new pattern.
24242 * config/i386/subst.md: Fix pattern (parallel -> unspec).
24244 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24246 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
24247 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
24249 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24251 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
24254 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24256 * config/aarch64/aarch64-cores.def: Rearrange to sort by
24257 architecture, then by implementer ID.
24258 * config/aarch64/aarch64-tune.md: Regenerate.
24260 2017-06-20 Richard Biener <rguenther@suse.de>
24262 PR middle-end/81097
24263 * fold-const.c (split_tree): Fold to type before negating.
24265 2017-06-20 David Malcolm <dmalcolm@redhat.com>
24267 * diagnostic-show-locus.c
24268 (selftest::test_fixit_deletion_affecting_newline): New function.
24269 (selftest::diagnostic_show_locus_c_tests): Call it.
24271 2017-06-20 Andreas Schwab <schwab@suse.de>
24274 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
24277 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24279 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
24280 __ARM_FEATURE_COPROC according to support.
24282 2017-06-20 Jakub Jelinek <jakub@redhat.com>
24284 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
24285 Rewritten to avoid overflow for > 32-bit pointers.
24288 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
24289 by removing enum keyword.
24290 (ubsan_type_descriptor): Likewise. Formatting fix.
24293 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
24294 splitter): Require TARGET_SSE2 in the condition.
24296 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
24299 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
24300 for doing vector set of SFmode on ISA 3.0.
24301 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
24302 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
24304 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
24305 SFmode value into a V4SF variable that was extracted from another
24306 V4SF variable without converting the element to double precision
24307 and back to single precision vector format.
24308 (vsx_insert_extract_v4sf_p9_2): Likewise.
24310 2017-06-19 Jakub Jelinek <jakub@redhat.com>
24312 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
24313 in UWHI to avoid undefined overflow.
24316 * ubsan.h (enum ubsan_encode_value_phase): New.
24317 (ubsan_encode_value): Change second argument to
24318 enum ubsan_encode_value_phase with default value of
24319 UBSAN_ENCODE_VALUE_GENERIC.
24320 * ubsan.c (ubsan_encode_value): Change second argument to
24321 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
24322 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
24323 create_tmp_var_raw instead of create_tmp_var and use a
24325 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
24326 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
24327 ubsan_encode_value callers.
24330 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
24331 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
24332 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
24334 2017-06-19 Richard Biener <rguenther@suse.de>
24336 PR middle-end/81118
24337 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
24338 estimates if we changed anything.
24340 2017-06-19 Richard Biener <rguenther@suse.de>
24342 PR tree-optimization/80887
24343 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
24344 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
24345 simplified lookups, then reset mprts_hook.
24346 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
24348 (try_to_simplify): Likewise.
24350 2017-06-19 Martin Liska <mliska@suse.cz>
24353 * gimplify.c (gimplify_switch_expr):
24354 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
24356 2017-06-19 Martin Liska <mliska@suse.cz>
24358 * doc/install.texi: Document that PGO runs in 4 stages.
24360 2017-06-19 Martin Liska <mliska@suse.cz>
24363 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
24364 to dispatcher function name.
24365 * multiple_target.c (replace_function_decl): New function.
24366 (create_dispatcher_calls): Redirect both edges and references.
24368 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
24370 * profile-count.c (profile_count::dump): Dump quality.
24371 (profile_count::differs_from_p): Update for unsigned val.
24372 * profile-count.h (profile_count_quality): New enum.
24373 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
24375 2017-06-19 Richard Biener <rguenther@suse.de>
24377 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
24378 struct function as arg.
24379 (estimate_numbers_of_iterations): Export overload with loop arg.
24380 (free_numbers_of_iterations_estimates_loop): Use an overload of
24381 free_numbers_of_iterations_estimates instead.
24382 * tree-cfg.c (remove_bb): Adjust.
24383 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
24384 * tree-parloops.c (gen_parallel_loop): Likewise.
24385 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
24387 (tree_unroll_loops_completely): Likewise.
24388 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
24389 Use an overload instead and export.
24390 (estimated_loop_iterations): Adjust.
24391 (max_loop_iterations): Likewise.
24392 (likely_max_loop_iterations): Likewise.
24393 (estimate_numbers_of_iterations): Take struct function as arg
24395 (loop_exits_before_overflow): Adjust.
24396 (free_numbers_of_iterations_estimates_loop): Use an overload.
24397 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
24398 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
24400 2017-06-19 Richard Biener <rguenther@suse.de>
24403 * ipa-prop.c (find_constructor_constant_at_offset): Handle
24404 RANGE_EXPR conservatively.
24406 2017-06-16 Carl Love <cel@us.ibm.com>
24408 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24409 definitions for vec_float, vec_float2, vec_floato,
24410 vec_floate built-ins.
24411 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
24412 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
24414 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
24415 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
24416 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
24417 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
24418 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
24419 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
24420 vec_floato): Add builtin defines.
24421 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
24422 Update the built-in documentation file for the new built-in
24425 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24427 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
24428 (mthumb): Mark as the negative of -marm.
24430 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24432 * doc/invoke.texi (ARM Options, -mcpu): Document supported
24434 (ARM Options, -mtune): Document that this accepts the same
24435 extension options as -mcpu.
24436 (ARM Options, -mfpu): Document addition of -mfpu=auto.
24438 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24440 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
24441 permitted extensions.
24443 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24445 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
24446 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
24447 (armv8-m.main): Add option +nodsp.
24448 * config/arm/arm-cpu-cdata.h: Regenerated.
24450 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24452 * config/arm/t-fuchsia: New file.
24453 * config.gcc (arm*-*-fuchsia*): Use it.
24455 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24457 * config/arm/t-symbian: Rewrite for new option infrastructure.
24459 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24461 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
24462 (MULTILIB_REQUIRED): Likewise.
24464 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24466 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
24467 (MULTILIB_RESUE): Likewise.
24468 (MULTILIB_MATCHES): Likewise.
24469 (MULTLIB_REQUIRED): Likewise.
24471 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24473 * config/arm/t-rtems: Rewrite for new option framework.
24475 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24477 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
24478 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
24479 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
24480 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
24481 * config/arm/t-multilib: ... here.
24482 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
24483 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
24484 armv7-a and armv8*-a when A-profile libraries have not been built.
24485 * config/arm/t-rmprofile: Rewrite.
24487 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24489 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
24490 with a backslash. Remove the backslash after substituting unescaped
24492 * doc/fragments.texi (MULTILIB_REUSE): Document it.
24494 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24496 * config.gcc: (arm*-*-*): When building a-profile libraries, force
24497 the driver to pass through the default setting of -mfloat-abi.
24498 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
24500 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
24501 (all_feat_combs): New rule.
24502 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
24504 * config/arm/t-aprofile: Rewrite.
24506 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24508 * config/arm/arm.h (FPUTYPE_AUTO): Define.
24509 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
24510 fpu is not specified by the user/command-line.
24511 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
24512 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
24513 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
24514 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
24515 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
24516 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
24518 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24520 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
24521 * config/arm/t-arm-elf: Rewritten.
24523 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24525 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
24526 have some floating-point instructions.
24527 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
24528 (TARGET_MAYBE_HARD_FLOAT): New macro.
24529 * config/arm/arm-builtins.c (arm_init_builtins): Use
24530 TARGET_MAYBE_HARD_FLOAT.
24531 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
24533 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24535 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
24536 (configargs.h): Include it.
24537 (arm_print_hint_for_fpu_option): New function.
24538 (arm_parse_fpu_option): New function.
24539 (candidate_extension): New class.
24540 (arm_canon_for_multilib): New function.
24541 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
24542 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
24543 (ARCH_CANONICAL_SPECS): New macro.
24544 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
24546 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24548 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
24549 are set after handling multilib fragments. Set target_cpu_default2
24552 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24554 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
24556 (arm*-*-*): Set target_cpu_default2 to a quoted string.
24557 * config/arm/parsecpu.awk (check_cpu): Validate any extension
24559 (check_arch): Likewise.
24560 * config/arm/arm.c (arm_configure_build_target): Handle
24561 TARGET_CPU_DEFAULT being a string constant. Scan any feature
24562 options in the default.
24564 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24566 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
24567 when an option is an alias of another.
24568 * config/arm/parsecpu.awk (optalias): New parser token.
24569 (gen_comm_data): Mark non-alias options as such. Emit entries
24570 for extension aliases.
24571 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
24572 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
24573 (armv6kz, armv6zk, armv6t2): Likewise.
24574 (armv7): Make vfpv3-d16 an alias.
24575 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
24577 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
24578 Sort in canonical order.
24579 (armv8-a): Sort in canonical order.
24580 (armv8.1-a, armv8.2-a): Likewise.
24581 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
24583 (cortex-a9): Sort in canonical order.
24584 * config/arm/arm.c (selftests.h): Include it.
24585 (arm_test_cpu_arch_data): New function.
24586 (arm_run_self_tests): New function.
24587 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
24588 (targetm): Move declaration to the end of the file.
24589 * arm-cpu-cdata.h: Regenerated.
24591 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24593 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
24594 call to target_mode_check describing the type of option passed.
24595 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
24596 (arm_target_thumb_only): Use arm_parse_arch_option_name or
24597 arm_parse_cpu_option_name to match parameters against list of
24599 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
24600 arm_arch_core_flags data structure.
24601 * config/arm/arm-cpu_cdata.h: Regenerated.
24603 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24605 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
24607 (arm_print_hint_for_cpu_option): Likewise.
24608 (arm_print_hint_for_arch_option): Likewise.
24609 (arm_parse_cpu_option_name): Likewise.
24610 (arm_parse_arch_option_name): Likewise.
24611 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
24612 of entries in the all_fpus list.
24613 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
24614 (arm_parse_cpu_option_name): Declare.
24615 (arm_parse_arch_option_name): Declare.
24616 (arm_parse_option_features): Declare.
24617 (arm_intialize_isa): Declare.
24618 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
24620 (gen_comm_data): ... here. Make definitions non-static.
24621 * config/arm/arm-cpu-data.h: Regenerated.
24622 * config/arm/arm-cpu-cdata.h: Regenerated.
24624 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24626 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
24627 (cpu_arch_extension): New structure.
24628 (cpu_arch_option, arch_option, cpu_option): New structures.
24629 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
24630 architecture types.
24631 (gen_data): Generate new format data tables.
24632 * config/arm/arm.c (cpu_tune): New structure.
24633 (cpu_option, processors): Delete.
24634 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
24635 (arm_print_hint_for_cpu_option): ... this and ...
24636 (arm_print_hint_for_arch_option): ... this.
24637 (arm_parse_arch_cpu_name): Delete. Replace with ...
24638 (arm_parse_cpu_option_name): ... this and ...
24639 (arm_parse_arch_option_name): ... this.
24640 (arm_unrecognized_feature): Change type of target parameter to
24642 (arm_parse_arch_cpu_features): Delete. Replace with ...
24643 (arm_parse_option_features): ... this.
24644 (arm_configure_build_target): Rework to use new configuration data
24646 (arm_print_tune_info): Rework for new configuration data tables.
24647 * config/arm/arm-cpu-data.h: Regenerated.
24648 * config/arm/arm-cpu.h: Regenerated.
24650 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24652 * Makefile.in (OBJS): Move sbitmap.o from here ...
24653 (OBJS-libcommon): ... to here.
24655 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24657 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
24658 (ISA_ALL_CRYPTO): New macro.
24659 (ISA_ALL_SIMD): New macro
24660 (ISA_ALL_FP): New macro.
24661 * config/arm/arm.c (fpu_bitlist): Update initializer.
24662 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
24664 (arm9e): Add fpu. Add option for nofp
24665 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
24666 (arm926ej-s, arm1026ej-s): Likewise.
24667 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
24668 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
24669 neon-fp16, neon-vfpv4, nofp and nosimd.
24670 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
24671 (cortex-a8): Add fpu. Add option for nofp.
24672 (cortex-a9): Add fpu. Add options for nosimd and nofp.
24673 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
24674 (cortex-r4f): Add fpu.
24675 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
24676 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
24678 (cortex-r8): Likewise.
24679 (cortex-m4): Add fpu. Add option for nofp.
24680 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
24681 (cortex-a17.cortex-a7): Likewise.
24682 (cortex-a32): Add fpu. Add options for crypto and nofp.
24683 (cortex-a35, cortex-a53): Likewise.
24684 (cortex-a57): Add fpu. Add option for crypto.
24685 (cortex-a72, cortex-a73): Likewise.
24686 (exynos-m1): Likewise.
24687 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
24688 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
24689 (cortex-m33): Add fpu. Add option for nofp.
24690 * config/arm/arm-cpu-cdata.h: Regenerated
24691 * config/arm/arm-cpu-data.h: Regenerated.
24693 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24695 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
24696 (armv5te, armv5tej): Likewise.
24697 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
24698 (armv7): Add options fp and vfpv3-d16.
24699 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
24700 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
24702 (armv7ve): Likewise.
24703 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
24704 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
24705 (armv8-a): Add nocrypto option.
24706 (armv8.1-a, armv8.2-a): Likewise.
24707 (armv8-m.main): add options fp, fp.dp and nofp.
24709 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24711 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
24713 (armv8-a+crc): Delete.
24714 (armv8.1-a): Add options simd, crypto and nofp.
24715 (armv8.2-a): Add options fp16, simd, crypto and nofp.
24716 (armv8.2-a+fp16): Delete.
24717 (armv8-m.main): Add option dsp.
24718 (armv8-m.main+dsp): Delete.
24719 (cortex-a8): Add fpu. Add nofp option.
24720 (cortex-a9): Add fpu. Add nofp and nosimd options.
24721 * config/arm/parsecpu.awk (gen_data): Generate option tables and
24722 link to main cpu and architecture data structures.
24723 (gen_comm_data): Only put isa attributes from the main architecture
24725 (option): New statement for architecture and CPU entries.
24726 * arm.c (struct cpu_option): New structure.
24727 (struct processors): Add entry for options.
24728 (arm_unrecognized_feature): New function.
24729 (arm_parse_arch_cpu_name): Ignore any characters after the first
24731 (arm_parse_arch_cpu_feature): New function.
24732 (arm_configure_build_target): Separate out any CPU and architecture
24733 features and parse separately. Don't error out if -mfpu=auto is
24734 used with only an architecture string.
24735 (arm_print_asm_arch_directives): New function.
24736 (arm_file_start): Call it.
24737 * config/arm/arm-cpu-cdata.h: Regenerated.
24738 * config/arm/arm-cpu-data.h: Likewise.
24739 * config/arm/arm-tables.opt: Likewise.
24741 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24743 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
24744 assembler when it is not -mfpu=auto.
24746 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24748 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
24749 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
24750 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
24751 (ASM_CPU_SPEC): Rewrite.
24752 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
24753 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
24754 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
24755 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
24756 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
24757 copied string is NUL-terminated. Also strip any characters prefixed
24759 (arm_rewrite_selected_arch): New function.
24760 (arm_rewrite_march): New function.
24762 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24764 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
24765 (x_arm_cpu_string, x_arm_tune_string): Likewise.
24766 (march, mcpu, mtune): Convert to string-based options.
24767 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
24768 (arm_parse_arch_cpu_name): New function.
24769 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
24770 identify selected architecture or CPU.
24771 (arm_option_save): New function.
24772 (TARGET_OPTION_SAVE): Redefine.
24773 (arm_option_restore): Restore string options.
24774 (arm_option_print): Print string options.
24776 2017-06-16 Martin Sebor <msebor@redhat.com>
24778 PR tree-optimization/80933
24779 PR tree-optimization/80934
24780 * builtins.c (fold_builtin_3): Do not handle bcmp here.
24781 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
24782 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
24783 (gimple_fold_builtin): Call them.
24785 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24787 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
24788 as unlikely; update profile.
24790 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24792 * predict.c (force_edge_cold): Handle declaring edges impossible
24795 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24797 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
24799 (try_unroll_loop_completely): Fix reporting.
24801 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24803 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
24805 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
24808 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
24809 if given a non-constant argument for an intrinsic which requires a
24812 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24814 * profile.c (compare_freqs): New function.
24815 (branch_prob): Sort edge list.
24816 (find_spanning_tree): Assume that the list is priority sorted.
24818 2017-06-16 Richard Biener <rguenther@suse.de>
24820 PR tree-optimization/81090
24821 * passes.def (pass_record_bounds): Remove.
24822 * tree-pass.h (make_pass_record_bounds): Likewise.
24823 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
24824 make_pass_record_bounds): Likewise.
24825 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
24826 not free niter estimates at the beginning but at the end.
24827 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
24829 2017-06-16 Richard Biener <rguenther@suse.de>
24831 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
24832 initializer to workaround ICE in host GCC 4.8.
24834 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24836 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
24838 (clone_inlined_nodes): Update.
24840 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24842 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
24843 prefetch settings, and enable prefetching by default at -O3.
24845 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24847 * config/aarch64/aarch64.c (aarch64_override_options_internal):
24848 Set flag_prefetch_loop_arrays according to tuning data.
24850 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24852 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
24853 New tune structure.
24854 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
24855 [Unrelated to main purpose of the patch] Place the pointer field last
24856 to enable type checking errors when tune structure are wrongly merged.
24857 * config/aarch64/aarch64.c (generic_prefetch_tune,)
24858 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
24859 (thunderx2t99_prefetch_tune): New tune constants.
24860 (tune_params *_tunings): Update all tunings (no functional change).
24861 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
24862 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
24863 from tunings structures.
24865 2017-06-16 Jakub Jelinek <jakub@redhat.com>
24868 * ubsan.c (instrument_null): Add T argument, use it instead
24869 of computing it based on IS_LHS.
24870 (instrument_object_size): Likewise.
24871 (pass_ubsan::execute): Adjust instrument_null and
24872 instrument_object_size callers to pass gimple_get_lhs or
24873 gimple_assign_rhs1 result to it. Use instrument_null instead of
24874 calling get_base_address and instrument_mem_ref. Handle
24875 aggregate call arguments for object-size sanitization.
24877 2017-06-16 Yury Gribov <tetra2005@gmail.com>
24879 PR tree-optimization/81089
24880 * tree-vrp.c (is_masked_range_test): Validate operands of
24883 2017-06-15 Martin Sebor <msebor@redhat.com>
24886 * dumpfile.c (dump_register): Avoid calling memset to initialize
24887 a class with a default ctor.
24888 * gcc.c (struct compiler): Remove const qualification.
24889 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
24890 * hash-table.h: Ditto.
24891 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
24893 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
24894 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
24896 * params.h (struct param_info): Make struct members non-const.
24897 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
24898 with default initialization.
24899 * vec.h (vec_copy_construct, vec_default_construct): New helper
24901 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
24902 with vec_copy_construct.
24903 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
24904 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
24905 * doc/invoke.texi (-Wclass-memaccess): Document.
24907 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24909 * emit-rtl.h (is_leaf): Update comment about local
24910 register allocator.
24912 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
24915 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
24916 for a variable to have a section before checking if the section has a
24918 Set section to.persistent if persistent attribute is set.
24919 Warn if .persistent attribute is used on an automatic variable.
24921 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
24923 PR rtl-optimization/80474
24924 * reorg.c (update_block): Do not ignore instructions in a delay slot.
24926 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
24928 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
24931 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
24933 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
24934 (casesi): Emit bounds checking as RTL.
24935 (casesi_internal_mips16_<mode>): Remove bounds checking.
24937 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
24939 * config/xtensa/xtensa.c (xtensa_option_override): Append
24940 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
24941 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
24942 xtensa_doloop_hooks): Define unconditionally.
24943 (xtensa_reorg_loops): Only call reorg_loops in the presence of
24945 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
24946 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
24947 for it in xtensa_option_override.
24948 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
24949 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
24951 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
24953 * doc/cppopts.texi: Document '-' special value to -MF.
24955 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
24957 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
24958 (cortex_a53_fconst): Likewise.
24959 (cortex_a53_fpmul): Likewise.
24960 (cortex_a53_f_load_64): Likewise.
24961 (cortex_a53_f_load_many): Likewise.
24962 (cortex_a53_advsimd_alu): Likewise.
24963 (cortex_a53_advsimd_alu_q): Likewise.
24964 (cortex_a53_advsimd_mul): Likewise.
24965 (cortex_a53_advsimd_mul_q): Likewise.
24966 (fpmac bypass): Add new bypass for fpmac-fpmac case.
24967 Add missing fmul, r2f_cvt and fconst cases.
24969 2017-06-14 Richard Biener <rguenther@suse.de>
24971 PR middle-end/81088
24972 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
24974 (fold_binary_loc): When associating do not treat pre-existing
24975 TREE_OVERFLOW on literal constants as a reason to allow
24976 TREE_OVERFLOW on associated literal constants.
24978 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24980 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
24981 (MASK_FEATURES): New macro.
24982 * config/sparc/sparc.c (sparc_option_override): Remove the special
24983 handling of -mfpu and generalize it to all MASK_FEATURES switches.
24985 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24987 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
24988 a division of 0 if non-call exceptions are enabled.
24990 2017-06-14 Andrew Pinski <apinski@cavium.com>
24991 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24994 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
24995 Improve vector initialization code gen for only variable case.
24997 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24999 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
25001 2017-06-14 Richard Biener <rguenther@suse.de>
25003 PR tree-optimization/81083
25004 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
25007 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25009 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
25010 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
25011 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
25012 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
25013 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
25014 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
25016 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25018 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
25019 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
25021 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25023 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
25025 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25027 * config/rs6000/t-rtems: Don't handle SPE.
25029 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25031 * config/rs6000/t-linux: Don't handle SPE.
25033 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25035 * config/rs6000/eabispe.h: Delete file.
25037 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25039 * config/rs6000/t-spe: Delete file.
25041 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25043 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
25044 (rs6000_legitimate_offset_address_p): Return false for anything in
25045 V2SImode or V2SFmode.
25047 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25049 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
25050 except V2SF and V2SI. Rearrange the vector modes, and add comments.
25051 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
25053 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
25054 (rs6000_legitimate_offset_address_p): Ditto.
25055 (rs6000_emit_move): Ditto.
25056 (rs6000_init_builtins): Remove V4HI_type_node.
25058 2017-06-13 Martin Liska <mliska@suse.cz>
25061 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
25062 (gate_asan): Likewise.
25063 * asan.h (asan_no_sanitize_address_p): Remove the function.
25064 (sanitize_flags_p): New function.
25065 * builtins.def: Fix coding style.
25066 * common.opt: Use renamed enum value.
25067 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
25068 * doc/extend.texi: Document no_sanitize attribute.
25069 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
25070 to SANITIZE_UNDEFINED_NONDEFAULT.
25071 * gcc.c (sanitize_spec_function): Use the renamed enum value.
25072 * gimple-fold.c (optimize_atomic_compare_exchange_p):
25073 Use sanitize_flags_p.
25074 * gimplify.c (gimplify_function_tree): Likewise.
25075 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
25076 * opts.c (parse_no_sanitize_attribute): New function.
25077 (common_handle_option): Use renamed enum value.
25078 * opts.h (parse_no_sanitize_attribute): Declare.
25079 * tree.c (sanitize_flags_p): New function.
25080 * tree.h: Declared here.
25081 * tsan.c: Use sanitize_flags_p.
25082 * ubsan.c (ubsan_expand_null_ifn): Likewise.
25083 (instrument_mem_ref): Likewise.
25084 (instrument_bool_enum_load): Likewise.
25085 (do_ubsan_in_current_function): Remove the function.
25086 (pass_ubsan::execute): Use sanitize_flags_p.
25087 * ubsan.h: Remove do_ubsan_in_current_function
25088 * tree-cfg.c (print_no_sanitize_attr_value): New function.
25089 (dump_function_to_file): Use it here.
25091 2017-06-13 Martin Jambor <mjambor@suse.cz>
25093 PR tree-optimization/80803
25094 PR tree-optimization/81063
25095 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
25096 (propagate_subaccesses_across_link): Enqueue subtree whenever
25097 necessary instead of relying on the caller.
25099 2017-06-13 Martin Jambor <mjambor@suse.cz>
25101 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
25102 that have a first_link.
25103 (sort_and_splice_var_accesses): Do not check first_link before
25105 (subtree_mark_written_and_enqueue): Likewise.
25106 (propagate_all_subaccesses): Likewise and do not stop at first
25107 parent with a first_link.
25109 2017-06-13 Martin Jambor <mjambor@suse.cz>
25111 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
25114 2017-06-13 Yury Gribov <tetra2005@gmail.com>
25116 * match.pd: New pattern.
25118 2017-06-13 Yury Gribov <tetra2005@gmail.com>
25120 * tree-vrp.c (is_masked_range_test): New function.
25121 (register_edge_assert_for): Determine ranges for
25124 2017-06-13 Yury Gribov <tetra2005@gmail.com>
25126 PR tree-optimization/67328
25127 * fold-const.c (maskable_range_p): New function.
25128 (build_range_check): Generate bittests if possible.
25130 2017-06-13 Martin Liska <mliska@suse.cz>
25132 * gimple-pretty-print.c (dump_probability): Add new argument.
25133 (dump_edge_probability): Dump both probability and count.
25134 (dump_gimple_label): Likewise.
25135 (dump_gimple_bb_header): Likewise.
25137 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
25140 * config/avr/avr-devices.c: Fix indentation.
25141 * config/avr/gen-avr-mmcu-specs.c: Dito.
25143 2017-06-13 Richard Biener <rguenther@suse.de>
25145 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
25146 instead get vector type from stmt_info.
25147 (vectorizable_reduction): Adjust. Remove dead code.
25149 2017-06-13 Richard Biener <rguenther@suse.de>
25151 PR middle-end/81065
25152 * fold-const.c (extract_muldiv_1): Remove bogus distribution
25153 case of C * (x * C2 + C3).
25154 (fold_addr_of_array_ref_difference): Properly fold index difference.
25156 2017-06-12 David S. Miller <davem@davemloft.net>
25159 * config/sparc/sparc.md (return expander): Emit frame blockage if
25160 function uses alloca.
25162 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
25164 * combine.c (make_field_assignment): Check len rather than the mode
25165 precision when calling force_to_mode.
25167 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
25169 Support multilibs and devices that see flash in RAM address range.
25172 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
25173 (avr_mcu_t) <flash_pm_offset>: New field.
25174 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
25175 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
25176 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
25177 (AVR_TINY_PM_OFFSET): Remove macro.
25178 * config/avr/avr.opt (-mshort-calls): New option.
25179 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
25180 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
25181 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
25182 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
25183 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
25184 instead of avr_arch->have_jmp_call.
25185 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
25186 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
25187 avr_arch->flash_pm_offset to define.
25188 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
25189 new field flash_pm_offset. Add entry for avrxmega3.
25190 (avr_texinfo): Add entry for avrxmega3.
25191 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
25192 attiny212, attiny214,
25193 attiny412, attiny414, attiny416, attiny417,
25194 attiny814, attiny816, attiny817,
25195 attiny1614, attiny1616, attiny1617,
25196 attiny3214, attiny3216, attiny3217.
25197 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
25198 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
25199 (avr_print_operand_address) [AVR_TINY]: Same.
25200 (avr_asm_init_sections) <readonly_data_section>: Only patch
25201 callback if avr_arch->flash_pm_offset = 0.
25202 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
25203 for rodata if avr_arch->flash_pm_offset != 0.
25204 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
25205 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
25206 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
25207 (m_options): Append opt_rcall.
25208 (m_dirnames): Append dir_rcall.
25209 * config/avr/t-multilib: Regenerate.
25211 * configure.ac [target=avr]: Check whether avrxmega3 default
25212 linker description file works as needed.
25213 * configure: Regenerate.
25214 * doc/avr-mmcu.texi: Regenerate.
25215 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
25216 <__AVR_ARCH__>: Document avrxmega3 and 103.
25217 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
25218 <__AVR_SHORT_CALLS__>: Document it.
25219 <__AVR_PM_BASE_ADDRESS__>: Document it.
25220 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
25221 (AVR Variable Attributes) <progmem>: Document this is
25222 not needed for avrxmega3.
25223 (AVR Named Address Spaces) <__flash>: Dito.
25225 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
25227 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
25229 2017-06-12 Doug Rupp <rupp@adacore.com>
25231 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
25232 Append vxworks-stdint.h to the tm_file list.
25233 * config/vxworks-stdint.h: New file.
25235 2017-06-12 Martin Liska <mliska@suse.cz>
25237 PR tree-optimization/81041
25238 * tree-profile.c (gimple_gen_ic_func_profiler):
25239 Create an extra BB in profile-generate
25240 (gimple_gen_time_profiler): Likewise.
25242 2017-06-12 Jakub Jelinek <jakub@redhat.com>
25244 PR tree-optimization/81003
25245 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
25246 (update_range_test): Use it instead of force_gimple_operand_gsi.
25248 2017-06-12 Richard Biener <rguenther@suse.de>
25250 PR tree-optimization/81053
25251 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
25252 with backedge value not defined in loop. Simplify def stmt
25255 2017-06-11 Tom de Vries <tom@codesourcery.com>
25258 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
25260 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
25261 nvptx_cannot_force_const_mem.
25263 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25265 * opts.c (finish_options): Move test for flag_split_stack after
25266 it has been initialized.
25268 2017-06-11 Jason Merrill <jason@redhat.com>
25270 * tree.h (id_equal): New.
25271 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
25272 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
25273 instead of strcmp of IDENTIFIER_POINTER.
25275 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25277 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
25278 (mark_all_inlined_calls_cdtor): Fix formating.
25279 (inline_transform): Rescale profile before inlining.
25281 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25283 * cgraph.h (cgraph_edge::clone): Update prototype.
25284 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
25285 (cgraph_node::create_clone): Update.
25286 (cgraph_node::create_version_clone): Update.
25287 * tree-inline.c (copy_bb): Update.
25288 (expand_call_inline): Update.
25290 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
25292 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
25293 factored out from ...
25294 (rs6000_emit_prologue): ... here.
25296 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
25298 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
25299 factored out from ...
25300 (rs6000_emit_prologue): ... here.
25302 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25304 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
25306 (handle_missing_profiles): Fix computation of tp_first_run.
25307 (counts_to_freqs): Do not touch freqs when count is 0.
25309 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25311 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
25314 2017-06-10 Tom de Vries <tom@codesourcery.com>
25316 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
25317 attributes): Document signal effective target.
25319 2017-06-10 Tom de Vries <tom@codesourcery.com>
25321 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
25322 Document effective target stack_size.
25324 2017-06-09 David Malcolm <dmalcolm@redhat.com>
25326 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
25327 to the edit_context if they can be auto-applied.
25329 2017-06-9 Ian Lance Taylor <iant@golang.org>
25331 * opts.c (finish_options): If -fsplit-stack, disable implicit
25332 -forder-blocks-and-partition.
25333 * doc/invoke.texi (Optimize Options): Document that when using
25334 -fsplit-stack -forder-blocks-and-partition is not implicitly
25337 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25339 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
25340 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
25341 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
25342 * builtins.def (abort, trap, unreachable): Declare cold.
25343 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
25344 * tree-core.h (ECF_COLD): New.
25345 * tree.c (set_call_expr_flags): Handle ECF_COLD.
25346 (build_common_builtin_nodes): Mark unreachable and abort as cold.
25348 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25350 * predict.c (unlikely_executed_stmt_p): Cleanup.
25352 2017-06-09 Richard Biener <rguenther@suse.de>
25354 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
25355 model if the ref is always written to.
25357 2017-06-09 Tamar Christina <tamar.christina@arm.com>
25359 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
25361 2017-06-09 Tamar Christina <tamar.christina@arm.com>
25363 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
25366 2017-06-09 Tom de Vries <tom@codesourcery.com>
25369 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
25370 "target cannot support label values" when encountering LABEL_REF.
25372 2017-06-09 Martin Liska <mliska@suse.cz>
25374 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
25375 (gimple_gen_ic_func_profiler): Emit direct comparison
25376 of __gcov_indirect_call_callee with NULL.
25377 (gimple_gen_time_profiler): Change probability from
25378 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
25380 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25382 * profile.c (edge_gcov_counts): Turn to pointer.
25383 (compute_branch_probabilities, compute_branch_probabilities): Update.
25384 (branch_prob): Do not clear edge_gcov_count.
25385 * profile.h (edge_gcov_counts): Turn to pointer.
25386 (edge_gcov_count): Update.
25388 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25390 * gimple.h (gimple_check_failed): Mark cold.
25392 2017-06-09 Richard Biener <rguenther@suse.de>
25394 PR tree-optimization/66623
25395 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
25396 refactor check_reduction into two parts, properly computing
25397 whether we have to check reduction validity for outer loop
25400 2017-06-09 Richard Biener <rguenther@suse.de>
25402 PR tree-optimization/79483
25403 * graphite-scop-detection.c (order): New global.
25404 (get_order): Compute bb to order mapping that satisfies code
25405 generation constraints.
25406 (cmp_pbbs): New helper.
25407 (build_scops): Start domwalk at entry block, sort generated
25410 2017-06-09 Richard Biener <rguenther@suse.de>
25412 PR middle-end/81007
25413 * ipa-polymorphic-call.c
25414 (ipa_polymorphic_call_context::restrict_to_inner_class):
25415 Skip FIELD_DECLs with error_mark_node type.
25416 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
25419 2017-06-09 Martin Liska <mliska@suse.cz>
25421 * predict.c (struct branch_predictor): New struct.
25422 (test_prediction_value_range): New test.
25423 (predict_c_tests): New function.
25424 * selftest-run-tests.c (selftest::run_tests): Run the function.
25425 * selftest.h: Declare new tests.
25427 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
25430 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
25431 gen_add3_insn did not fail.
25432 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
25433 r0, construct that number in a temporary reg and add that reg to r0.
25434 If asked to put the result in r0 as well, fail.
25436 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
25438 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25439 for early expansion of vec_eqv.
25441 2017-06-08 Jakub Jelinek <jakub@redhat.com>
25443 PR middle-end/81005
25444 * ubsan.c (instrument_null): Avoid pointless code temporary.
25445 (pass_ubsan::execute): Instrument aggregate arguments of calls.
25447 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
25451 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
25454 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
25457 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25459 * predict.c (unlikely_executed_edge_p): Move ahead.
25460 (probably_never_executed_edge_p): Use it.
25462 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25464 PR middle-end/79988
25465 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
25466 gimple_call_builtin_p call.
25468 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25470 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
25471 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
25472 rtl_check_failed_type2, rtl_check_failed_code1,
25473 rtl_check_failed_code2, rtl_check_failed_code_mode,
25474 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
25475 rtvec_check_failed_bounds, rtl_check_failed_flag,
25476 _fatal_insn_not_found, _fatal_insn): Likewise.
25477 * tree.h (tree_contains_struct_check_failed,
25478 tree_check_failed, tree_not_check_failed,
25479 tree_class_check_failed, tree_range_check_failed,
25480 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
25481 tree_vec_elt_check_failed, phi_node_elt_check_failed,
25482 tree_operand_check_failed, omp_clause_check_failed,
25483 omp_clause_operand_check_failed, omp_clause_range_check_failed):
25486 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25488 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
25489 flag_branch_probabilities.
25490 * ipa-inline.c (edge_badness): Likewise.
25491 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
25492 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
25493 * predict.c (maybe_hot_frequency_p): Likewise.
25494 (probably_never_executed): Likewise.
25495 * sched-ebb.c (schedule_ebbs): Likewise.
25496 * sched-rgn.c (find_single_block_region): Likewise.
25497 * tracer.c (tail_duplicate): Likewise.
25499 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25501 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
25502 longer requires x_flag_profile_use.
25504 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25506 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
25507 instead of flag_reorder_blocks_and_partition.
25508 * dbxout.c (dbxout_function_end): Likewise.
25509 * dwarf2out.c (gen_subprogram_die): Likewise.
25510 * haifa-sched.c (sched_create_recovery_edges): Likewise.
25511 * hw-doloop.c (reorg_loops): Likewise.
25512 * varasm.c (assemble_start_function,
25513 assemble_end_function): Likewise.
25514 (decide_function_section): Do not check for
25515 flag_reorder_blocks_and_partition.
25517 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25519 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
25521 (chkp_get_hard_register_fake_addr_expr): Ditto.
25522 (chkp_build_addr_expr): Add check for hard reg case.
25523 (chkp_parse_array_and_component_ref): Ditto.
25524 (chkp_find_bounds_1): Ditto.
25525 (chkp_process_stmt): Don't generate bounds store for
25528 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25530 * predict.c (maybe_hot_bb_p): Do not check profile status.
25531 (maybe_hot_edge_p): Likewise.
25532 (probably_never_executed): Check for zero counts even if profile
25534 (unlikely_executed_edge_p): New function.
25535 (unlikely_executed_stmt_p): New function.
25536 (unlikely_executed_bb_p): New function.
25537 (set_even_probabilities): Use unlikely predicates.
25538 (combine_predictions_for_bb): Likewise.
25539 (predict_paths_for_bb): Likewise.
25540 (predict_paths_leading_to_edge): Likewise.
25541 (determine_unlikely_bbs): New function.
25542 (estimate_bb_frequencies): Use it.
25543 (compute_function_frequency): Use zero counts even if profile is
25545 * profile-count.h: Fix typo.
25547 2017-08-08 Julia Koval <julia.koval@intel.com>
25549 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
25550 _mm512_mask_cvtsepi16_storeu_epi8,
25551 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
25552 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
25553 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
25554 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
25555 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
25556 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
25557 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
25558 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
25559 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
25560 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
25561 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
25562 __builtin_ia32_pmovuswb256mem_mask,
25563 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
25564 __builtin_ia32_pmovwb512mem_mask): New builtins.
25566 2017-08-08 Julia Koval <julia.koval@intel.com>
25568 PR target/73350,80862
25569 * config/i386/subst.md (round): Fix round pattern.
25570 * config/i386/i386.c (ix86_erase_embedded_rounding):
25571 Fix erasing rounding for the fixed pattern.
25573 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25575 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
25577 2017-06-08 Martin Liska <mliska@suse.cz>
25579 PR gcov-profile/80911
25580 * gcov.c (block_info::block_info): New constructor.
25582 2017-06-07 Carl Love <cel@us.ibm.com>
25584 * config/rs6000/rs6000-c: The return type of the following
25585 built-in functions was implemented as int not long long. Fix sign
25586 of return value for the unsigned version of vec_mulo and vec_mule.
25587 vector unsigned long long vec_bperm (vector unsigned long long,
25588 vector unsigned char)
25589 vector signed long long vec_mule (vector signed int,
25591 vector unsigned long long vec_mule (vector unsigned int,
25592 vector unsigned int)
25593 vector signed long long vec_mulo (vector signed int,
25595 vector unsigned long long vec_mulo (vector unsigned int,
25596 vector unsigned int)
25597 * doc/extend.texi: Fix the documentation for the built-in
25600 2017-06-07 Carl Love <cel@us.ibm.com>
25603 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
25606 2017-06-07 Carl Love <cel@us.ibm.com>
25608 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
25609 support, Generate doublehv for signed int/float for BE case only.
25611 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
25613 * doc/invoke.texi (mcx16): Rewrite.
25615 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25617 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
25618 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
25619 *mov<mode>_softfloat, and an anonymous splitter): Use
25620 nonimmediate_operand instead of rs6000_nonimmediate_operand.
25622 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25624 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
25626 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25627 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
25628 (rs6000_debug_reg_global): Adjust.
25629 (rs6000_init_hard_regno_mode_ok): Adjust.
25630 (rs6000_dbx_register_number): Adjust.
25631 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
25632 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25633 Remove SPE_ACC and SPEFSCR.
25634 (REG_ALLOC_ORDER): Ditto.
25635 (FRAME_POINTER_REGNUM): Change to 111.
25636 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
25637 (REG_CLASS_NAMES): Ditto.
25638 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
25639 (REGISTER_NAMES): Ditto.
25640 (ADDITIONAL_REG_NAMES): Ditto.
25641 (rs6000_reg_names): Ditto.
25642 * config/rs6000/rs6000.md: Renumber some register number
25645 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25647 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
25649 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25650 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
25652 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
25653 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25654 Delete the SPE high registers.
25655 (REG_ALLOC_ORDER): Ditto.
25656 (enum reg_class): Remove SPE_HIGH_REGS.
25657 (REG_CLASS_NAMES): Ditto.
25658 (REG_CLASS_CONTENTS): Delete the SPE high registers.
25659 (REGISTER_NAMES): Ditto.
25660 (rs6000_reg_names): Ditto.
25661 * doc/tm.texi.in: Remove SPE as example.
25662 * doc/tm.texi: Regenerate.
25664 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25666 * config/rs6000/8540.md (ppc8540_brinc): Delete.
25667 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
25668 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
25669 * config/rs6000/rs6000.md (type): Remove "brinc".
25671 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25673 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
25674 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
25675 * config/rs6000/linuxspe.h: Delete file.
25676 * config/rs6000/rs6000.md: Don't include spe.md.
25677 * config/rs6000/spe.h: Delete file.
25678 * config/rs6000/spe.md: Delete file.
25679 * config/rs6000/t-rs6000: Remove spe.md.
25681 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25683 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
25684 (reg_or_none500mem_operand): Delete.
25685 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
25686 instead of reg_or_none500mem_operand.
25688 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25690 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
25691 handling of SPE flags.
25692 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
25694 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25696 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
25698 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
25699 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
25700 paired_divv2sf3): Similar.
25701 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
25702 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
25703 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
25705 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
25706 Rename the paired_* instruction patterns.
25707 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
25709 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
25710 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
25711 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
25712 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
25713 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
25714 PAIRED_VECTOR_MODE.
25715 (struct machine_function): Delete field spe_insn_chain_scanned_p.
25716 (spe_func_has_64bit_regs_p): Delete.
25717 (spe_expand_predicate_builtin): Delete.
25718 (spe_expand_evsel_builtin): Delete.
25719 (TARGET_DWARF_REGISTER_SPAN): Do not define.
25720 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
25721 (invalid_e500_subreg): Delete.
25722 (rs6000_legitimize_address): Always force_reg op2 as well, for
25723 paired single memory accesses.
25724 (rs6000_member_type_forces_blk): Delete.
25725 (rs6000_spe_function_arg): Delete.
25726 (rs6000_expand_unop_builtin): Delete SPE handling.
25727 (rs6000_expand_binop_builtin): Ditto.
25728 (spe_expand_stv_builtin): Delete.
25729 (bdesc_2arg_spe): Delete.
25730 (spe_expand_builtin): Delete.
25731 (spe_expand_predicate_builtin): Delete.
25732 (spe_expand_evsel_builtin): Delete.
25733 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
25734 (spe_init_builtins): Delete.
25735 (spe_func_has_64bit_regs_p): Delete.
25736 (savres_routine_name): Delete "info" parameter. Adjust callers.
25737 (rs6000_emit_stack_reset): Ditto.
25738 (rs6000_dwarf_register_span): Delete.
25739 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
25740 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
25741 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
25743 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
25745 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
25746 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
25747 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
25748 mulv2sf3, divv2sf3): Delete expanders.
25750 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25752 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
25754 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25756 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
25757 * config/rs6000/rs6000.c: Ditto.
25759 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25761 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
25762 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
25763 comparison_operator.
25765 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25767 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
25768 * config/rs6000/rs6000.opt: Ditto.
25769 * config/rs6000/t-rtems: Ditto.
25771 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25773 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
25774 TARGET_E500_SINGLE by 0, simplify.
25775 * config/rs6000/rs6000.c: Ditto.
25776 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
25777 (spe_build_register_parallel): Delete.
25778 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
25779 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
25780 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
25781 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
25782 (E500_CONVERT): Delete.
25783 * config/rs6000/spe.md: Remove many patterns and all define_constants.
25785 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25787 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
25788 * config/rs6000/dfp.md: Ditto.
25789 (negdd2, *negdd2_fpr): Merge.
25790 (absdd2, *absdd2_fpr): Merge.
25791 (negtd2, *negtd2_fpr): Merge.
25792 (abstd2, *abstd2_fpr): Merge.
25793 * config/rs6000/e500.h: Delete file.
25794 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
25795 TARGET_FPRS by 1 and simplify.
25796 * config/rs6000/rs6000-c.c: Ditto.
25797 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
25798 TARGET_DF_SPE by 0.
25799 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
25801 * config/rs6000/rs6000.md: Ditto.
25802 (floatdidf2, *floatdidf2_fpr): Merge.
25803 (move_from_CR_gt_bit): Delete.
25804 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
25805 (E500_CR_IOR_COMPARE): Delete.
25806 (All patterns that require !TARGET_FPRS): Delete.
25807 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
25809 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25811 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
25813 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25815 * graphds.c (add_edge): Intitialize edge's attached data.
25816 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
25817 pointer parameter. Call pointed function on each edge during
25818 graph traversing. Skip traversing the edge when the function
25820 (graphds_dfs, graphds_scc): Ditto.
25821 (for_each_edge): New parameter. Pass the new parameter to callback
25823 * graphds.h (skip_edge_callback): New function pointer type.
25824 (graphds_dfs, graphds_scc): New function pointer parameter.
25825 (graphds_edge_callback, for_each_edge): New parameter.
25827 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25829 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
25830 out code checking if runtime alias check is possible to below ...
25831 Call the new function.
25832 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
25833 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
25835 2017-06-07 Marek Polacek <polacek@redhat.com>
25838 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
25839 TYPE_OVERFLOW_WRAPS check.
25841 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25843 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
25844 if versioning is required.
25845 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
25846 peeling with the check for versioning.
25848 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25850 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
25851 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
25852 Set true to new parameter if new ssa variable is defined.
25853 (vect_gen_vector_loop_niters): Refactor. Set range information
25854 for the new vector loop bound variable.
25855 (vect_do_peeling): Ditto.
25857 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25859 * tree-affine.c (ssa.h): Include header file.
25860 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
25861 has wrapping overflow behavior.
25863 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25865 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
25867 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25869 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
25870 (tree_to_aff_combination): ... here.
25872 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25874 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
25875 reg_pressure model function.
25876 (ivopts_global_cost_for_size): Delete.
25877 (determine_set_costs, iv_ca_recount_cost): Call new model function
25878 ivopts_estimate_reg_pressure.
25880 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25882 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
25883 expensive than udiv. Remove floating point cases from mod.
25885 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25887 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
25888 Increase idiv cost.
25890 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25892 * config/aarch64/aarch64.md
25893 (copysignsf3): Fix mask generation.
25895 2017-06-07 Jakub Jelinek <jakub@redhat.com>
25897 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
25899 (class dump_manager): Add register_dumps method.
25900 * dumpfile.c: Include langhooks.h.
25901 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
25902 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
25903 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
25904 (dump_manager::dump_register): Start with 512 entries instead of 32.
25905 (dump_manager::register_dumps): New method.
25906 * toplev.c (general_init): Instead of invoking register_dumps
25907 langhook, invoke register_dumps method on the dump manager.
25908 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
25911 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
25913 * doc/md.texi: Clarify the restrictions on a define_insn condition.
25914 Say that # requires an associated define_split to exist, and that
25915 the define_split must be suitable for use after register allocation.
25917 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25919 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
25920 (compute_outgoing_frequencies): Also initialize zero counts.
25921 (find_many_sub_basic_blocks): Do not produce uninitialized profile
25922 around loops; preserve more of profile when nothing changes.
25924 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
25926 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
25928 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
25929 * config/arm/arm-cpu-cdata.h: Regenerate.
25930 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
25931 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
25932 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
25933 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
25934 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
25936 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
25938 * config/arm/t-rmprofile: Likewise.
25939 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
25941 2017-06-06 David S. Miller <davem@davemloft.net>
25944 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
25945 blockage if function uses alloca.
25947 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25949 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
25950 New "uid" fields to hold pretty-print IDs of group and ref.
25951 Memory references are now identified as <group_id>:<ref_id>
25952 instead of using [random] addresses.
25953 (dump_mem_details): Simplify, no functional change.
25954 (dump_mem_ref): Simplify and make output more concise.
25955 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
25956 (find_or_create_group): Initialize group uid.
25957 (record_ref): Initialize ref uid. Improve debug output.
25958 (prune_group_by_reuse, should_issue_prefetch_p,)
25959 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
25960 (mark_nontemporal_store, determine_loop_nest_reuse):
25961 Improve debug output.
25963 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25965 * dbgcnt.def (prefetch): New debug counter.
25966 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
25967 (schedule_prefetches): Stop issueing prefetches if debug counter
25970 2017-06-06 Tom de Vries <tom@codesourcery.com>
25972 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
25973 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
25975 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25977 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
25978 Use aarch64_reg_or_zero predicate for operand 4.
25979 (aarch64_compare_and_swap<mode> define_insn_and_split):
25980 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
25981 (aarch64_store_exclusive<mode>): Likewise for operand 2.
25983 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
25985 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
25986 (arm_compute_save_core_reg_mask): This.
25987 (thumb1_compute_save_reg_mask): Rename into ...
25988 (thumb1_compute_save_core_reg_mask): This.
25989 (arm_compute_save_reg0_reg12_mask): Adapt comment.
25990 (arm_compute_frame_layout): Likewise.
25992 2017-06-06 Richard Biener <rguenther@suse.de>
25994 PR tree-optimization/80974
25995 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
25996 keep or clear leaders SSA info.
25998 2017-06-06 Tom de Vries <tom@codesourcery.com>
26000 * config/nvptx/nvptx.c (split_mode_p): New function.
26001 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
26003 2017-06-06 Tom de Vries <tom@codesourcery.com>
26005 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
26007 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26010 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
26013 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26015 * shrink-wrap.c (handle_simple_exit): Update profile.
26016 (try_shrink_wrapping): Upate profile.
26018 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26020 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
26021 (tree_guess_outgoing_edge_probabilities): New.
26022 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
26023 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
26025 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26027 * ipa-split.c (split_function): Initialize return bb profile.
26029 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26031 * profile.c (compute_branch_probabilities): Also initialize
26032 EXIT_BLOCK profile.
26034 2017-06-06 Richard Biener <rguenther@suse.de>
26036 PR tree-optimization/80928
26037 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
26038 (vect_analyze_loop_operations): Properly guard analysis for
26040 (vect_transform_loop): Likewise.
26041 (vect_analyze_loop_2): Also reset SLP type on PHIs.
26042 (vect_model_induction_cost): Do not cost for pure SLP.
26043 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
26044 of induction in inner loop vectorization.
26045 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
26046 (vect_get_and_check_slp_defs): Handle vect_induction_def.
26047 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
26049 (vect_analyze_slp_cost_1): Cost induction.
26050 (vect_detect_hybrid_slp_stmts): Handle PHIs.
26051 (vect_get_slp_vect_defs): Likewise.
26052 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
26053 (vect_transform_stmt): Handle SLP reductions.
26054 * tree-vectorizer.h (vectorizable_induction): Adjust.
26056 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
26058 * config/rs6000/rs6000.c (make_resolver_func): Update
26059 init_lowered_empty_function call.
26061 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
26063 * doc/invoke.texi: Document the -fprofile-abs-path option.
26064 * common.opt (fprofile-abs-path): New option.
26065 * gcov-io.h (gcov_write_filename): Declare.
26066 * gcov-io.c (gcov_write_filename): New function.
26067 * coverage.c (coverage_begin_function): Use gcov_write_filename.
26068 * profile.c (output_location): Likewise.
26070 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26072 * shring-wrap.c: Revert accidental commit.
26074 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
26076 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
26078 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26080 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
26082 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
26083 profile in callgraph edge.
26084 * profile-count.h (apply_probability): If THIS is 0, then result is 0
26085 (apply_scale): Likewise.
26086 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
26087 Also scale profile when inlining function with zero profile.
26088 (initialize_cfun): Update exit block profile even when it is zero.
26089 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
26090 when profile is read.
26092 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
26094 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
26095 (CLONE_*): New constants to define the processors we can generate
26096 code for with the target_clone attribute.
26097 (rs6000_clone_map): New array to identify which clone processors
26098 the current program is running on.
26099 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
26100 target_clone attribute.
26101 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
26102 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
26103 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
26104 (cpu_expand_builtin): Add support for target_clone attribute.
26105 (rs6000_valid_attribute_p): Allow "default" attribute.
26106 (get_decl_name): New debug function to simplify printing the
26107 current function name in debugging statements.
26108 (rs6000_clone_priority): New functions to support the target_clone
26109 attribute, and be able to generate code to switch between ISA 2.05
26110 through ISA 3.0 (power6 through power9).
26111 (rs6000_compare_version_priority): Likewise.
26112 (rs6000_get_function_versions_dispatcher): Likewise.
26113 (make_resolver_func): Likewise.
26114 (add_condition_to_bb): Likewise.
26115 (dispatch_function_versions): Likewise.
26116 (rs6000_generate_version_dispatcher_body): Likewise.
26117 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
26118 (fusion_gpr_load_p): Fix a spacing issue.
26119 * doc/extend.texi (Common Function Attributes): Document that the
26120 PowerPC supports the target_clone attribute.
26122 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
26124 * config/arm/arm.h: explain F symbol found in description of ARM
26125 register allocation in its legend.
26127 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26129 * config/mips/frame-header-opt.c: Include profile-count.h.
26130 * config/riscv/riscv.c: Include profile-count.h
26132 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26134 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
26136 (sm_set_flag_if_changed): Add bbs field.
26137 (execute_sm_if_changed_flag_set): Pass BBS.
26138 (execute_sm): Update.
26140 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26142 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
26145 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26147 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
26148 (peephole2): New peephole2 to emit the above.
26149 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
26151 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26153 * config/aarch64/aarch64.c (define_peephole2 above
26154 *sub_<shift>_<mode>): New peephole.
26156 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26158 * config/i386/i386.c (make_resolver_func): Update.
26159 * Makefile.in: Add profile-count.h and profile-count.o
26160 * auto-profile.c (afdo_indirect_call): Update to new API.
26161 (afdo_set_bb_count): Update.
26162 (afdo_propagate_edge): Update.
26163 (afdo_propagate_circuit): Update.
26164 (afdo_calculate_branch_prob): Update.
26165 (afdo_annotate_cfg): Update.
26166 * basic-block.h: Include profile-count.h
26167 (struct edge_def): Turn count to profile_count.
26168 (struct basic_block_def): Likewie.
26169 (REG_BR_PROB_BASE): Move to profile-count.h
26170 (RDIV): Move to profile-count.h
26171 * bb-reorder.c (max_entry_count): Turn to profile_count.
26172 (find_traces): Update.
26173 (rotate_loop):Update.
26174 (connect_traces):Update.
26175 (sanitize_hot_paths):Update.
26176 * bt-load.c (migrate_btr_defs): Update.
26177 * cfg.c (RDIV): Remove.
26178 (init_flow): Use alloc_block.
26179 (alloc_block): Uninitialize count.
26180 (unchecked_make_edge): Uninitialize count.
26181 (check_bb_profile): Update.
26182 (dump_edge_info): Update.
26183 (dump_bb_info): Update.
26184 (update_bb_profile_for_threading): Update.
26185 (scale_bbs_frequencies_int): Update.
26186 (scale_bbs_frequencies_gcov_type): Update.
26187 (scale_bbs_frequencies_profile_count): New.
26188 * cfg.h (update_bb_profile_for_threading): Update.
26189 (scale_bbs_frequencies_profile_count): Declare.
26190 * cfgbuild.c (compute_outgoing_frequencies): Update.
26191 (find_many_sub_basic_blocks): Update.
26192 * cfgcleanup.c (try_forward_edges): Update.
26193 (try_crossjump_to_edge): Update.
26194 * cfgexpand.c (expand_gimple_tailcall): Update.
26195 (construct_exit_block): Update.
26196 * cfghooks.c (verify_flow_info): Update.
26197 (dump_bb_for_graph): Update.
26198 (split_edge): Update.
26199 (make_forwarder_block): Update.
26200 (duplicate_block): Update.
26201 (account_profile_record): Update.
26202 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
26203 (get_estimated_loop_iterations): Update.
26204 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
26205 (single_likely_exit): Update.
26206 * cfgloopmanip.c (scale_loop_profile): Update.
26208 (set_zero_probability): Update.
26209 (lv_adjust_loop_entry_edge): Update.
26210 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
26211 (purge_dead_edges): Update.
26212 (rtl_account_profile_record): Update.
26213 * cgraph.c (cgraph_node::create): Uninitialize count.
26214 (symbol_table::create_edge): Uninitialize count.
26215 (cgraph_update_edges_for_call_stmt_node): Update.
26216 (cgraph_edge::dump_edge_flags): Update.
26217 (cgraph_node::dump): Update.
26218 (cgraph_edge::maybe_hot_p): Update.
26219 * cgraph.h: Include profile-count.h
26220 (create_clone), create_edge, create_indirect_edge): Update.
26221 (cgraph_node): Turn count to profile_count.
26222 (cgraph_edge0: Likewise.
26223 (make_speculative, clone): Update.
26224 (create_edge): Update.
26225 (init_lowered_empty_function): Update.
26226 * cgraphclones.c (cgraph_edge::clone): Update.
26227 (duplicate_thunk_for_node): Update.
26228 (cgraph_node::create_clone): Update.
26229 * cgraphunit.c (cgraph_node::analyze): Update.
26230 (cgraph_node::expand_thunk): Update.
26231 * final.c (dump_basic_block_info): Update.
26232 * gimple-streamer-in.c (input_bb): Update.
26233 * gimple-streamer-out.c (output_bb): Update.
26234 * graphite.c (print_global_statistics): Update.
26235 (print_graphite_scop_statistics): Update.
26236 * hsa-brig.c: Include basic-block.h.
26237 * hsa-dump.c: Include basic-block.h.
26238 * hsa-gen.c (T sum_slice): Update.
26239 (convert_switch_statements):Update.
26240 * hsa-regalloc.c: Include basic-block.h.
26241 * ipa-chkp.c (chkp_produce_thunks): Update.
26242 * ipa-cp.c (struct caller_statistics): Update.
26243 (init_caller_stats): Update.
26244 (gather_caller_stats): Update.
26245 (ipcp_cloning_candidate_p): Update.
26246 (good_cloning_opportunity_p): Update.
26247 (get_info_about_necessary_edges): Update.
26248 (dump_profile_updates): Update.
26249 (update_profiling_info): Update.
26250 (update_specialized_profile): Update.
26251 (perhaps_add_new_callers): Update.
26252 (decide_about_value): Update.
26253 (ipa_cp_c_finalize): Update.
26254 * ipa-devirt.c (struct odr_type_warn_count): Update.
26255 (struct decl_warn_count): Update.
26256 (struct final_warning_record): Update.
26257 (possible_polymorphic_call_targets): Update.
26258 (ipa_devirt): Update.
26259 * ipa-fnsummary.c (redirect_to_unreachable): Update.
26260 * ipa-icf.c (sem_function::merge): Update.
26261 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
26262 * ipa-inline.c (compute_uninlined_call_time): Update.
26263 (compute_inlined_call_time): Update.
26264 (want_inline_small_function_p): Update.
26265 (want_inline_self_recursive_call_p): Update.
26266 (edge_badness): Update.
26267 (lookup_recursive_calls): Update.
26268 (recursive_inlining): Update.
26269 (inline_small_functions): Update.
26270 (dump_overall_stats): Update.
26271 (dump_inline_stats): Update.
26272 * ipa-profile.c (ipa_profile_generate_summary): Update.
26273 (ipa_propagate_frequency): Update.
26274 (ipa_profile): Update.
26275 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26276 * ipa-utils.c (ipa_merge_profiles): Update.
26277 * loop-doloop.c (doloop_modify): Update.
26278 * loop-unroll.c (report_unroll): Update.
26279 (unroll_loop_runtime_iterations): Update.
26280 * lto-cgraph.c (lto_output_edge): Update.
26281 (lto_output_node): Update.
26282 (input_node): Update.
26283 (input_edge): Update.
26284 (merge_profile_summaries): Update.
26285 * lto-streamer-in.c (input_cfg): Update.
26286 * lto-streamer-out.c (output_cfg): Update.
26287 * mcf.c (create_fixup_graph): Update.
26288 (adjust_cfg_counts): Update.
26289 (sum_edge_counts): Update.
26290 * modulo-sched.c (sms_schedule): Update.
26291 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
26292 * predict.c (maybe_hot_count_p): Update.
26293 (probably_never_executed): Update.
26294 (dump_prediction): Update.
26295 (combine_predictions_for_bb): Update.
26296 (propagate_freq): Update.
26297 (handle_missing_profiles): Update.
26298 (counts_to_freqs): Update.
26299 (rebuild_frequencies): Update.
26300 (force_edge_cold): Update.
26301 * predict.h: Include profile-count.h
26302 (maybe_hot_count_p, counts_to_freqs): UPdate.
26303 * print-rtl-function.c: Do not include cfg.h
26304 * print-rtl.c: Include basic-block.h
26305 * profile-count.c: New file.
26306 * profile-count.h: New file.
26307 * profile.c (is_edge_inconsistent): Update.
26308 (correct_negative_edge_counts): Update.
26309 (is_inconsistent): Update.
26310 (set_bb_counts): Update.
26311 (read_profile_edge_counts): Update.
26312 (compute_frequency_overlap): Update.
26313 (compute_branch_probabilities): Update; Initialize and deinitialize
26315 (branch_prob): Update.
26316 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
26317 (edge_gcov_count): New.
26318 (bb_gcov_count): New.
26319 * shrink-wrap.c (try_shrink_wrapping): Update.
26320 * tracer.c (better_p): Update.
26321 * trans-mem.c (expand_transaction): Update.
26322 (ipa_tm_insert_irr_call): Update.
26323 (ipa_tm_insert_gettmclone_call): Update.
26324 * tree-call-cdce.c: Update.
26325 * tree-cfg.c (gimple_duplicate_sese_region): Update.
26326 (gimple_duplicate_sese_tail): Update.
26327 (gimple_account_profile_record): Update.
26328 (execute_fixup_cfg): Update.
26329 * tree-inline.c (copy_bb): Update.
26330 (copy_edges_for_bb): Update.
26331 (initialize_cfun): Update.
26332 (freqs_to_counts): Update.
26333 (copy_cfg_body): Update.
26334 (expand_call_inline): Update.
26335 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26336 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
26337 (try_unroll_loop_completely): Update.
26338 (try_peel_loop): Update.
26339 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26340 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
26341 * tree-ssa-loop-split.c (connect_loops): Update.
26342 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
26343 * tree-ssa-reassoc.c (branch_fixup): Update.
26344 * tree-ssa-tail-merge.c (replace_block_by): Update.
26345 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
26346 (compute_path_counts): Update.
26347 (update_profile): Update.
26348 (recompute_probabilities): Update.
26349 (update_joiner_offpath_counts): Update.
26350 (estimated_freqs_path): Update.
26351 (freqs_to_counts_path): Update.
26352 (clear_counts_path): Update.
26353 (ssa_fix_duplicate_block_edges): Update.
26354 (duplicate_thread_path): Update.
26355 * tree-switch-conversion.c (case_bit_test_cmp): Update.
26356 (struct switch_conv_info): Update.
26357 * tree-tailcall.c (decrease_profile): Update.
26358 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26359 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26360 * value-prof.c (check_counter): Update.
26361 (gimple_divmod_fixed_value): Update.
26362 (gimple_mod_pow2): Update.
26363 (gimple_mod_subtract): Update.
26364 (gimple_ic_transform): Update.
26365 (gimple_stringop_fixed_value): Update.
26366 * value-prof.h (gimple_ic): Update.
26368 2017-06-02 Carl Love <cel@us.ibm.com>
26370 * config/rs6000/rs6000-c: Add support for built-in functions
26371 vector double vec_doublee (vector signed int);
26372 vector double vec_doublee (vector unsigned int);
26373 vector double vec_doublee (vector float);
26374 vector double vec_doubleh (vector signed int);
26375 vector double vec_doubleh (vector unsigned int);
26376 vector double vec_doubleh (vector float);
26377 vector double vec_doublel (vector signed int);
26378 vector double vec_doublel (vector unsigned int);
26379 vector double vec_doublel (vector float);
26380 vector double vec_doubleo (vector signed int);
26381 vector double vec_doubleo (vector unsigned int);
26382 vector double vec_doubleo (vector float);.
26383 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
26384 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
26386 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
26387 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
26388 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
26390 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
26391 vec_doublel, vec_doubleh.
26392 * doc/extend.texi: Update the built-in documentation file for the
26393 new built-in functions.
26395 2017-06-02 David Malcolm <dmalcolm@redhat.com>
26398 * ipa-inline-analysis.c (free_growth_caches): Set
26399 edge_removal_hook_holder to NULL after removing it.
26401 2017-06-02 Sudakshina Das <sudi.das@arm.com>
26403 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
26404 comparision with zero.
26406 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26407 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26408 for early expansion of vec_min and vec_max builtins.
26409 (builtin_function_type): Add min/max unsigned variants to those
26410 identified as having unsigned arguments.
26412 2017-06-02 Olivier Hainque <hainque@adacore.com>
26414 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
26416 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26418 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
26419 Use VALL_F16 iterator rather than VALL.
26421 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26423 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
26424 Emit CBNZ inside loop when doing a strong exchange and comparing
26425 against zero. Generate the CC flags after the loop.
26427 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
26429 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
26430 (dl_section_ref): New.
26431 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
26432 On AIX, append an expression to subtract the size of the
26433 section length to dl_section_ref.
26435 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26437 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26438 for early expansion of vector absolute builtins.
26440 2017-06-02 Richard Biener <rguenther@suse.de>
26442 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
26443 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
26445 2017-06-02 Richard Biener <rguenther@suse.de>
26447 PR tree-optimization/80948
26448 * tree-tailcall.c (find_tail_calls): Track stmts to move in
26449 stmt order as well.
26451 2017-06-02 Richard Biener <rguenther@suse.de>
26453 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
26455 * tree-vect-stmts.c (process_use): Do not mark backedge defs
26456 for inductions as relevant.
26458 2017-06-02 Richard Biener <rguenther@suse.de>
26460 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
26461 (vectorizable_induction): ... this. Remove dead code.
26463 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
26465 * builtins. (expand_builtin_alloca): Remove second parameter and
26466 infer its value from the first parameter instead.
26467 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
26469 2017-06-02 Jakub Jelinek <jakub@redhat.com>
26471 PR rtl-optimization/80903
26472 * loop-doloop.c (add_test): Unshare sequence.
26474 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
26476 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
26478 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
26480 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
26482 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
26483 xlogue_layout::get_instance, logue_layout::xlogue_layout,
26484 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
26485 (xlogue_layout::get_stub_rtx): Make static.
26486 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
26487 (xlogue_layout::compute_stub_managed_regs): Rename to...
26488 (xlogue_layout::count_stub_managed_regs): ...this.
26489 (xlogue_layout::is_stub_managed_reg): New function.
26490 (xlogue_layout::m_stub_names): Rename to...
26491 (xlogue_layout::s_stub_names): ...this, make static.
26492 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
26493 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
26494 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
26495 xlogue_layout::s_stub_names): Instantiate statics.
26496 (stub_managed_regs): Remove.
26497 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
26498 (disable_call_ms2sysv_xlogues): Rename to...
26499 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
26500 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
26502 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
26503 change after reload_completed.
26504 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
26506 (ix86_expand_prologue): Likewise.
26507 (ix86_expand_epilogue): Likewise.
26508 (ix86_expand_split_stack_prologue): Likewise.
26509 (ix86_compute_frame_layout): Remove frame parameter ...
26510 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
26511 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
26513 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
26514 (ix86_frame): Move from here ...
26515 * config/i386/i386.h (ix86_frame): ... to here.
26516 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
26517 complete ix86_frame data structure instead. Remove some_ld_name.
26519 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26521 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
26522 symbols that hold a DECL_VALUE_EXPR.
26524 2017-06-01 Martin Jambor <mjambor@suse.cz>
26526 PR tree-optimization/80898
26527 * tree-sra.c (process_subtree_disqualification): Removed.
26528 (disqualify_candidate): Do not acll
26529 process_subtree_disqualification.
26530 (subtree_mark_written_and_enqueue): New function.
26531 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
26532 RHS has been disqualified and re-queue LHS if necessary. Apart
26533 from that, ignore disqualified RHS.
26535 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26537 * config/s390/s390.c (s390_emit_epilogue): Disable early return
26538 address fetch for z10 or later.
26540 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26542 * config/arc/arc.md (tst_movb): Add guard when splitting.
26544 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26546 * config/arc/arc.c (arc_can_eliminate): Test against
26547 arc_frame_pointer_needed.
26549 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26551 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
26552 to prevent store reordering.
26553 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
26554 (type): Add block type.
26555 (stack_tie): Define special instruction to be used in
26558 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26560 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
26561 constraint. It is not valid for the pattern.
26562 (noncommutative_binary_comparison): Likewise.
26564 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26566 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
26569 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26571 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
26572 be used by the reg-alloc.
26574 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26576 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
26577 reg-alloc when having mul64 or mul32x16 instructions.
26578 (mulsidi3): Likewise.
26579 (umulsidi3): Likewise.
26580 (mulsi32x16): New pattern.
26581 (mulsi64): Likewise.
26582 (mulsidi64): Likewise.
26583 (umulsidi64): Likewise.
26584 (MUL32x16_REG): Define.
26585 (mul64_600): Use MUL32x16_REG.
26586 (mac64_600): Likewise.
26587 (umul64_600): Likewise.
26588 (umac64_600): Likewise.
26590 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26592 * config/arc/arc.md (mulsi3_700): Make it commutative.
26594 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
26596 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
26598 (*sign_extendsidi2_insn): Likewise for movstosw.
26600 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26602 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
26603 the type of the input discriminant value. Convert the
26604 discriminant value of signedness vary.
26606 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
26608 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
26609 Add to -Wall section.
26611 2017-06-01 Richard Biener <rguenther@suse.de>
26613 PR middle-end/66313
26614 * fold-const.c (fold_plusminus_mult_expr): If the factored
26615 factor may be zero use a wrapping type for the inner operation.
26616 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
26617 and handle moved defs.
26618 (process_assignment): Properly guard the unary op case. Return a
26619 tri-state indicating that moving the stmt before the call may allow
26620 to continue. Pass through to_move.
26621 (find_tail_calls): Handle moving unrelated defs before
26624 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
26627 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
26628 splitter result in the canonical way.
26630 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
26632 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
26633 also for 32bit target. Update insn attributes.
26634 (zero-extendsidi2 splitter): Allow all registers for operand 1.
26636 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
26638 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
26639 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
26640 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
26641 (_mm_maskz_min_ss): New intrinsics.
26643 2017-05-31 Martin Liska <mliska@suse.cz>
26645 * tree-vect-loop.c (vect_create_epilog_for_reduction):
26646 Change comment style to one we normally use.
26647 (vectorizable_reduction): Likewise.
26648 (vectorizable_induction): Likewise.
26649 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
26650 (vectorizable_call): Likewise.
26651 (vectorizable_simd_clone_call): Likewise.
26652 (vectorizable_conversion): Likewise.
26653 (vectorizable_assignment): Likewise.
26654 (vectorizable_shift): Likewise.
26655 (vectorizable_operation): Likewise.
26656 (vectorizable_store): Likewise.
26657 (vectorizable_load): Likewise.
26658 * tree-vectorizer.h: Likewise.
26660 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
26662 * passes.c (emergency_dump_function): New.
26663 * tree-pass.h (emergency_dump_function): Declare.
26664 * plugin.c (plugins_internal_error_function): Remove.
26665 * plugin.h (plugins_internal_error_function): Remove declaration.
26666 * toplev.c (internal_error_function): New static function. Use it...
26667 (general_init): ...here.
26669 2017-05-31 Graham Markall <graham.markall@embecosm.com>
26671 * config/arc/arc.c (arc_print_operand): Handle constant operands.
26672 (arc_rtx_costs): Add costs for new patterns.
26673 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
26674 * config/arc/predicates.md: Add _1_2_3_operand predicate.
26676 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26678 * tree-ssa-strlen.c (get_next_strinfo): New function.
26679 (get_stridx_plus_constant): Use it.
26680 (zero_length_string): Likewise.
26681 (adjust_related_strinfos): Likewise.
26682 (adjust_last_stmt): Likewise.
26684 2017-05-31 Richard Biener <rguenther@suse.de>
26687 * config/i386/i386.c (ix86_expand_builtin): Remove assert
26688 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
26690 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26692 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
26693 loop_vinfo argument and use of dependence distance vectors.
26694 Check instead whether the two references differ only in their
26695 initial value and assume that they have the same alignment if the
26696 difference is a multiple of the vector alignment.
26697 (vect_analyze_data_refs_alignment): Update call accordingly.
26699 2017-05-31 Martin Liska <mliska@suse.cz>
26702 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
26704 2017-05-31 Bin Cheng <bin.cheng@arm.com>
26706 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
26707 (create_intersect_range_checks): Move from ...
26708 * tree-data-ref.c (create_intersect_range_checks_index)
26709 (create_intersect_range_checks): ... to here.
26710 (create_runtime_alias_checks): New function factored from ...
26711 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
26712 here. Call above function.
26713 * tree-data-ref.h (create_runtime_alias_checks): New function.
26715 2017-05-31 Bin Cheng <bin.cheng@arm.com>
26717 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
26718 segment length for dr_b and compute it in wide_int.
26720 2017-05-31 Richard Biener <rguenther@suse.de>
26722 PR tree-optimization/80906
26723 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
26724 and pass through iv_map.
26725 (copy_bb_and_scalar_dependences): Adjust.
26726 (translate_pending_phi_nodes): Likewise.
26727 (copy_loop_close_phi_args): Handle code-generating IVs instead
26730 2017-05-30 David Malcolm <dmalcolm@redhat.com>
26732 * diagnostic-color.c (color_dict): Add "type-diff".
26733 (parse_gcc_colors): Update comment.
26734 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
26735 -fdiagnostics-show-template-tree and -fno-elide-type.
26736 (GCC_COLORS): Add type-diff to example.
26738 (-fdiagnostics-show-template-tree): New.
26739 (-fno-elide-type): New.
26740 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
26741 the pp_format_decoder callback. Call any m_format_postprocessor's
26743 (pretty_printer::pretty_printer): Initialize
26744 m_format_postprocessor.
26745 (pretty_printer::~pretty_printer): Delete any
26746 m_format_postprocessor.
26747 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
26748 (class format_postprocessor): New class.
26749 (struct pretty_printer::format_decoder): Document the new parameters.
26750 (struct pretty_printer::m_format_postprocessor): New field.
26751 * tree-diagnostic.c (default_tree_printer): Update for new
26752 bool and const char ** params.
26753 * tree-diagnostic.h (default_tree_printer): Likewise.
26755 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
26757 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
26758 (lwa_operand): Delete rs6000_gen_cell_microcode test.
26759 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
26760 rs6000_gen_cell_microcode code.
26761 (rs6000_final_prescan_insn): Delete.
26762 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
26763 "warn-cell-microcode" entries.
26764 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
26765 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
26766 throughout. Change cc_reg_not_micro_cr0_operand to
26767 cc_reg_not_cr0_operand throughout.
26768 (*extendhi<mode>2_noload): Delete.
26769 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
26770 (mwarn-cell-microcode): Delete.
26771 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
26772 -mgen-cell-microcode and -mwarn-cell-microcode.
26774 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
26777 * config/i386/constraints.md (Yd): New constraint.
26779 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
26780 and (?Yd, r) alternatives. Update insn attributes.
26781 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
26782 and (?*Yd, r) alternatives. Update insn attributes.
26783 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
26785 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
26787 * gimplify.c (gimplify_modify_expr): Don't create a
26788 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
26791 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
26793 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
26795 2017-05-30 Richard Biener <rguenther@suse.de>
26797 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
26798 and reduc_def fields.
26799 (STMT_VINFO_REDUC_TYPE): New define.
26800 (STMT_VINFO_REDUC_DEF): Likewise.
26801 (vect_force_simple_reduction): Adjust prototype.
26802 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
26803 (vect_is_simple_reduction): Remove check_reduction argument.
26804 (vect_force_simple_reduction): Adjust and set
26805 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
26806 (vectorizable_reduction): Do not re-do reduction analysis
26807 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
26808 * tree-parloops.c (gather_scalar_reductions): Adjust.
26810 2017-05-30 Richard Biener <rguenther@suse.de>
26812 PR middle-end/80901
26813 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
26816 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
26818 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
26819 Introduce unknown_misalignment parameter and remove vf.
26820 (vect_peeling_hash_get_lowest_cost):
26821 Pass unknown_misalignment parameter.
26822 (vect_enhance_data_refs_alignment):
26823 Fix unsupportable data ref treatment.
26825 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26827 * tree-vect-data-refs.c (vect_get_data_access_cost):
26828 Workaround for SLP handling.
26829 (vect_enhance_data_refs_alignment):
26830 Compute costs for doing no peeling at all, compare to the best
26831 peeling costs so far and avoid peeling if cheaper.
26833 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26835 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
26836 Return peeling info and set costs to zero for unlimited cost
26838 (vect_enhance_data_refs_alignment): Also inspect all datarefs
26839 with unknown misalignment. Compute and costs for unknown
26840 misalignment, compare them to the costs for known misalignment
26841 and choose the cheapest for peeling.
26843 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26845 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
26846 (vect_get_peeling_costs_all_drs): Create function.
26847 (vect_peeling_hash_get_lowest_cost):
26848 Use vect_get_peeling_costs_all_drs.
26849 (vect_peeling_supportable): Create function.
26850 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
26852 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26854 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
26855 DR_HAS_NEGATIVE_STEP.
26856 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
26857 (vect_enhance_data_refs_alignment): Use.
26858 (vect_duplicate_ssa_name_ptr_info): Use.
26859 * tree-vectorizer.h (dr_misalignment): Use.
26860 (known_alignment_for_access_p): Use.
26862 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
26865 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
26867 (has_section_name): New function.
26869 2017-05-30 Martin Liska <mliska@suse.cz>
26872 * auto-profile.c (get_function_decl_from_block): Fix
26875 2017-05-30 Richard Biener <rguenther@suse.de>
26877 PR middle-end/80876
26878 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
26880 2017-05-30 Martin Liska <mliska@suse.cz>
26882 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
26883 * dumpfile.h (struct dump_file_info): Remove ctors.
26885 2017-05-30 Martin Liska <mliska@suse.cz>
26887 * predict.def: Fix GNU coding style.
26889 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
26891 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
26892 Mark 'to' argument with ATTRIBUTE_UNUSED.
26894 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
26896 * config/xtensa/xtensa.c (xtensa_emit_call): Use
26897 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
26898 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
26901 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
26903 * doc/install.texi (Options specification): Restore entry of
26904 --enable-sjlj-exceptions.
26906 2017-05-27 Michael Eager <eager@eagercon.com>
26909 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
26911 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
26913 * config/microblaze/microblaze.h
26914 (FIXED_REGISTERS): Update in macro.
26915 (CALL_USED_REGISTERS): Update in macro.
26917 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
26919 * doc/install.texi: Add links to macOS binary distributions.
26921 2017-05-27 Jakub Jelinek <jakub@redhat.com>
26925 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26927 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
26929 2017-05-26 Martin Liska <mliska@suse.cz>
26931 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
26933 2017-05-26 Martin Liska <mliska@suse.cz>
26935 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
26936 always leading ';; '.
26937 (dump_bb_info): Likewise.
26938 (brief_dump_cfg): Likewise.
26939 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
26940 * dumpfile.c: Remove usage of TDF_VERBOSE.
26941 * dumpfile.h (enum dump_kind): Likewise.
26942 (dump_gimple_bb_header): Do not use TDF_COMMENT.
26943 * print-tree.c (debug_verbose): Remove.
26944 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
26945 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
26946 * tree-diagnostic.c (default_tree_printer): Replace
26947 TDF_DIAGNOSTIC with TDF_SLIM.
26949 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26951 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
26952 in parameter loop, rather than loop_vinfo.
26953 (create_intersect_range_checks): Ditto.
26954 (vect_create_cond_for_alias_checks): Update call to above functions.
26956 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26958 PR tree-optimization/80815
26959 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
26960 for merging runtime alias checks. Handle negative DR_STEPs.
26962 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26964 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
26966 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
26967 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
26968 out code pruning runtime alias checks.
26969 * tree-data-ref.c (prune_runtime_alias_test_list): New function
26970 factored out from above.
26971 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
26973 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
26975 (prune_runtime_alias_test_list): New decalaration.
26977 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26979 * tree-vect-data-refs.c (compare_tree): Rename and move ...
26980 * tree-data-ref.c (data_ref_compare_tree): ... to here.
26981 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
26982 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
26983 (operator==, comp_dr_with_seg_len_pair): Ditto.
26984 (vect_prune_runtime_alias_test_list): Ditto.
26986 2017-05-26 Martin Liska <mliska@suse.cz>
26989 * params.def: Bound partial-inlining-entry-probability param.
26991 2017-05-26 Marek Polacek <polacek@redhat.com>
26994 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
26997 2017-05-26 Richard Biener <rguenther@suse.de>
26999 PR tree-optimization/80842
27000 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
27003 2017-05-26 Richard Biener <rguenther@suse.de>
27005 PR tree-optimization/80844
27006 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
27008 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
27010 * doc/md.texi (Machine Constraints): Update x86 family
27011 machine constraints section to match 'config/i386/constraints.md'.
27013 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
27015 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
27017 2017-05-25 Nathan Sidwell <nathan@acm.org>
27019 * doc/invoke.texi (--enable-languages): Update documentation.
27021 2017-05-25 Martin Liska <mliska@suse.cz>
27023 * dumpfile.c: Add TDF_FOLDING.
27024 * dumpfile.h (enum dump_kind): Likewise.
27025 * genmatch.c (dt_simplify::gen_1): Use it.
27027 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27029 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
27031 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27033 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
27034 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
27036 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27038 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
27039 * match.pd (X == C): Rewrite it here.
27040 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
27041 with_certain_nonzero_bits2): New predicates.
27042 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
27044 2017-05-24 Nathan Sidwell <nathan@acm.org>
27046 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
27049 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
27052 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
27054 * config/powerpcspe: New port. Files are copied from the rs6000
27055 port, with "rs6000" in filenames replaced by "powerpcspe".
27057 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
27059 PR rtl-optimization/80754
27060 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
27062 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
27064 * config/sparc/sparc.md (length): Return the correct value for -mflat
27065 sibcalls to match output_sibcall.
27067 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
27071 * config/rs6000/rs6000.c (struct machine_function): Add new field
27073 (rs6000_get_separate_components): Init that field, use it.
27074 (rs6000_components_for_bb): Use the field.
27076 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
27078 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
27080 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
27082 PR middle-end/80823
27083 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
27085 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27088 * config/s390/s390.c (s390_check_qrst_address): Check incoming
27089 address against address_operand predicate.
27090 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
27092 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
27094 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
27095 parameters passed indirectly.
27097 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
27099 * config/i386/i386.md (*movdi_internal): Remove SSE4
27100 alternative 18 (?r, *v). Update insn attributes.
27101 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
27102 Update insn attributes.
27103 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
27104 Update insn attributes.
27105 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
27106 alternative 1 (r, v). Remove isa attribute.
27107 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
27108 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
27109 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
27111 2017-05-23 Tom de Vries <tom@codesourcery.com>
27113 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
27116 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27118 * cgraphunit.c (symbol_table::process_new_functions): Update.
27119 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
27120 (inline_generate_summary): Rename to ...
27121 (ipa_fn_summary_generate): ... this one.
27122 (inline_read_summary): Rename to ...
27123 (ipa_fn_summary_read): ... this one.
27124 (inline_write_summary): Rename to ...
27125 (ipa_fn_summary_write): ... this one.
27126 (inline_free_summary): Rename to ...
27127 (ipa_free_fn_summary): ... this one.
27128 (pass_data_local_fn_summary, pass_local_fn_summary,
27129 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
27130 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
27131 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
27132 make_pass_ipa_fn_summary): New.
27133 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
27134 inline_write_summary, inline_free_summary): Remove.
27135 (ipa_free_fn_summary) : New.
27136 * ipa-inline.c (ipa_inline): Update.
27137 (pass_ipa_inline): Do not generate summaries.
27138 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
27140 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
27141 and add pass_ipa_fn_summary.
27142 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
27144 (make_pass_inline_parameters): Remove.
27146 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
27148 * omp-low.c (struct omp_context): Remove "default_kind" member.
27151 * omp-offload.c (execute_oacc_device_lower): Remove the
27152 parallelism dimensions function attributes for unparallelized
27153 OpenACC kernels constructs.
27155 2017-05-23 Martin Liska <mliska@suse.cz>
27157 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
27159 (cgraph_edge::make_speculative): Likewise.
27160 (cgraph_edge::resolve_speculation): Likewise.
27161 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
27162 (cgraph_node::dump): Likewise.
27163 * cgraph.h: Likewise.
27164 * cgraphunit.c (analyze_functions): Likewise.
27165 (symbol_table::compile): Likewise.
27166 * ipa-cp.c (print_all_lattices): Likewise.
27167 (determine_versionability): Likewise.
27168 (initialize_node_lattices): Likewise.
27169 (ipcp_verify_propagated_values): Likewise.
27170 (estimate_local_effects): Likewise.
27171 (update_profiling_info): Likewise.
27172 (create_specialized_node): Likewise.
27173 (perhaps_add_new_callers): Likewise.
27174 (decide_about_value): Likewise.
27175 (decide_whether_version_node): Likewise.
27176 (identify_dead_nodes): Likewise.
27177 (ipcp_store_bits_results): Likewise.
27178 * ipa-devirt.c (dump_targets): Likewise.
27179 (ipa_devirt): Likewise.
27180 * ipa-icf.c (sem_item::dump): Likewise.
27181 (sem_function::equals): Likewise.
27182 (sem_variable::equals): Likewise.
27183 (sem_item_optimizer::read_section): Likewise.
27184 (sem_item_optimizer::execute): Likewise.
27185 (congruence_class::dump): Likewise.
27186 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
27187 (dump_inline_summary): Likewise.
27188 (estimate_node_size_and_time): Likewise.
27189 (inline_analyze_function): Likewise.
27190 * ipa-inline-transform.c (inline_call): Likewise.
27191 * ipa-inline.c (report_inline_failed_reason): Likewise.
27192 (want_early_inline_function_p): Likewise.
27193 (edge_badness): Likewise.
27194 (update_edge_key): Likewise.
27195 (inline_small_functions): Likewise.
27196 * ipa-profile.c (ipa_profile): Likewise.
27197 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
27198 (ipa_make_edge_direct_to_target): Likewise.
27199 (remove_described_reference): Likewise.
27200 (ipa_impossible_devirt_target): Likewise.
27201 (propagate_controlled_uses): Likewise.
27202 (ipa_print_node_params): Likewise.
27203 (ipcp_transform_function): Likewise.
27204 * ipa-pure-const.c (pure_const_read_summary): Likewise.
27205 (propagate_pure_const): Likewise.
27206 * ipa-reference.c (generate_summary): Likewise.
27207 (read_write_all_from_decl): Likewise.
27208 (propagate): Likewise.
27209 (ipa_reference_read_optimization_summary): Likewise.
27210 * ipa-utils.c (ipa_merge_profiles): Likewise.
27211 * ipa.c (walk_polymorphic_call_targets): Likewise.
27212 (symbol_table::remove_unreachable_nodes): Likewise.
27213 (ipa_single_use): Likewise.
27214 * passes.c (execute_todo): Likewise.
27215 * predict.c (drop_profile): Likewise.
27216 * symtab.c (symtab_node::get_dump_name): New function.
27217 (symtab_node::dump_name): Likewise.
27218 (symtab_node::dump_asm_name): Likewise.
27219 (symtab_node::dump_references): Likewise.
27220 (symtab_node::dump_referring): Likewise.
27221 (symtab_node::dump_base): Likewise.
27222 (symtab_node::debug_symtab): Likewise.
27223 * tree-sra.c (convert_callers_for_node): Likewise.
27224 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
27225 * value-prof.c (init_node_map): Likewise.
27227 2017-05-23 Martin Liska <mliska@suse.cz>
27229 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
27230 and symtab_node::debug_symtab to symbol_table::debug.
27231 * cgraphunit.c (analyze_functions): Use the renamed function.
27232 (symbol_table::compile): Likewise.
27233 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
27234 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
27235 * passes.c (execute_todo): Likewise.
27236 * symtab.c (symbol_table::dump): New function.
27237 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
27239 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27241 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
27242 that nonconst implies exec.
27244 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27246 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
27247 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
27248 (inline_edge_summary_vec): Turn into ...
27249 (ipa_call_summaries): ... this one.
27250 (redirect_to_unreachable, edge_set_predicate,
27251 evaluate_properties_for_edge, inline_summary_alloc,
27252 reset_ipa_call_summary, reset_inline_summary,
27253 inline_summary_t::duplicate): Update.
27254 (inline_edge_duplication_hook): Turn to ...
27255 (ipa_call_summary_t::duplicate): ... this one.
27256 (inline_edge_removal_hook): Turn to ...
27257 (ipa_call_summary_t::remove): ... this one.
27258 (dump_inline_edge_summary): Turn to ...
27259 (dump_ipa_call_summary): ... this one.
27260 (estimate_function_body_sizes): Update.
27261 (inline_update_callee_summaries): Update.
27262 (remap_edge_change_prob): Update.
27263 (remap_edge_summaries): Update.
27264 (inline_merge_summary): Update.
27265 (do_estimate_edge_time): Update.
27266 (inline_generate_summary): Update.
27267 (inline_read_section): Update.
27268 (inline_read_summary): Update.
27269 (inline_free_summary): Update.
27270 * ipa-inline.c (can_inline_edge_p): Update.
27271 (compute_inlined_call_time): Update.
27272 (want_inline_small_function_p): Update.
27273 (edge_badness): Update.
27274 (early_inliner): Update.
27275 * ipa-inline.h (inline_edge_summary): Turn to ...
27276 (ipa_call_summary): ... this one.
27277 (ipa_call_summary_t): New class.
27278 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
27279 (ipa_call_summaries): New.
27280 (inline_edge_summary): Remove.
27281 (estimate_edge_growth): Update.
27282 * ipa-profile.c (ipa_propagate_frequency_1): Update.
27283 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27284 * ipa-split.c (execute_split_functions): Update.
27285 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
27287 2017-05-23 Tom de Vries <tom@codesourcery.com>
27289 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
27290 attributes): Document rdrand effective target.
27292 2017-05-23 Tom de Vries <tom@codesourcery.com>
27294 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
27295 attributes): Sort alphabetically.
27297 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
27299 * config/avr/genmultilib.awk: Use gsub instead of gensub.
27301 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
27304 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
27305 V2DF/V2DI splat into two separate patterns, one that handles
27306 registers, and the other that only handles memory. Drop support
27307 for splatting from a GPR on ISA 2.07 and then splitting the
27308 splat into direct move and splat.
27309 (vsx_splat_<mode>_reg): Likewise.
27310 (vsx_splat_<mode>_mem): Likewise.
27312 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
27314 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
27316 2017-05-22 Jakub Jelinek <jakub@redhat.com>
27318 PR middle-end/80809
27319 * omp-low.c (finish_taskreg_remap): New function.
27320 (finish_taskreg_scan): If unit size of ctx->record_type
27321 is non-constant, unshare the size expression and replace
27322 decls in it with possible outer var refs.
27324 PR middle-end/80809
27325 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
27326 GOVD_SHARED rather than GOVD_PRIVATE with it.
27327 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
27328 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
27330 PR middle-end/80853
27331 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
27332 as last argument to build_outer_var_ref for pointer bases of array
27333 section reductions.
27335 2017-05-19 Martin Sebor <msebor@redhat.com>
27337 * print-tree.c (print_node): Print DECL_READ_P flag.
27339 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27341 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
27342 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
27343 * cgraph.c: Likewise.
27344 * cgraphunit.c: Likewise.
27345 * gengtype.c: Likewise.
27346 * ipa-cp.c: Likewise.
27347 * ipa-devirt.c: Likewise.
27348 * ipa-icf.c: Likewise.
27349 * ipa-predicate.c: Likewise.
27350 * ipa-profile.c: Likewise.
27351 * ipa-prop.c: Likewise.
27352 * ipa-split.c: Likewise.
27354 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
27355 edge_predicate_pool, dump_inline_hints,
27356 inline_summary::account_size_time, redirect_to_unreachable,
27357 edge_set_predicate, set_hint_predicate,
27358 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
27359 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
27360 inline_summary_t::remove, remap_hint_predicate_after_duplication,
27361 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
27362 ipa_call_summary_t::remove, initialize_growth_caches,
27363 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
27364 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
27365 mark_modified, unmodified_parm_1, unmodified_parm,
27366 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
27367 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
27368 compute_bb_predicates, will_be_nonconstant_expr_predicate,
27369 will_be_nonconstant_predicate, record_modified_bb_info,
27370 get_minimal_bb, record_modified, param_change_prob,
27371 phi_result_unknown_predicate, predicate_for_phi_result,
27372 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
27373 estimate_function_body_sizes, compute_inline_parameters,
27374 compute_inline_parameters_for_curren, pass_data_inline_parameters,
27375 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
27376 inline_update_callee_summaries, remap_edge_change_prob,
27377 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
27378 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
27379 inline_analyze_function, inline_summary_t::insert,
27380 inline_generate_summary, read_ipa_call_summary, inline_read_section,
27381 inline_read_summary, write_ipa_call_summary, inline_write_summary,
27382 inline_free_summary): Move to ipa-fnsummary.h
27383 (predicate_t): Remove.
27384 * ipa-fnsummary.c: New file.
27385 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
27386 (enum inline_hints_vals, inline_hints, agg_position_info,
27387 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
27388 inline_summaries, ipa_call_summary, ipa_call_summary_t,
27389 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
27390 dump_inline_summary, dump_inline_hints, inline_generate_summary,
27391 inline_read_summary, inline_write_summary, inline_free_summary,
27392 inline_analyze_function, initialize_inline_failed,
27393 inline_merge_summary, inline_update_overall_summary,
27394 compute_inline_parameters): Move to ipa-fnsummary.h
27395 * ipa-fnsummary.h: New file.
27396 * ipa-inline-transform.h: Include ipa-inline.h.
27397 * ipa-inline.c: LIkewise.
27399 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27401 * ipa-inline.c (edge_badness): Use inlined_time instead of
27402 inline_summaries->get.
27404 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27406 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
27408 2017-05-22 Nathan Sidwell <nathan@acm.org>
27410 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
27411 (fdump-lang): Document 'raw' option.
27412 * dumpfile.h (TDI_tu): Delete.
27413 * dumpfile.c (dump_files): Remove translation-unit.
27414 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
27416 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
27418 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
27419 command option from $(AWK) call.
27420 * config/avr/genmultilib.awk: Simplify and rewrite so that it
27421 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
27422 [FORMAT]: Remove handling of variable.
27423 * config/avr/t-multilib: Regenerate.
27425 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27427 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
27429 (dump_inline_summary): Do not print self_time.
27430 (estimate_function_body_sizes): Do not set self_time.
27431 (compute_inline_parameters): Likewise.
27432 (inline_read_section, inline_write_summary): Do not stream self_time.
27433 * ipa-inline.h (inline_summary): Drop self_time.
27435 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27437 * ipa-inline-analysis.c (account_size_time): Rename to ...
27438 (inline_summary::account_size_time): ... this one.
27439 (reset_ipa_call_summary): Turn to ...
27440 (ipa_call_summary::reset): ... this one.
27441 (reset_inline_summary): Turn to ...
27442 (inline_summary::reset): ... this one.
27443 (inline_summary_t::remove): Update.
27444 (inline_summary_t::duplicate): Update.
27445 (ipa_call_summary_t::remove): Update.
27446 (dump_inline_summary): Update.
27447 (estimate_function_body_sizes): Update.
27448 (compute_inline_parameters): Update.
27449 (estimate_node_size_and_time): Update.
27450 (inline_merge_summary): Update.
27451 (inline_update_overall_summary): Update.
27452 (inline_read_section): Update.
27453 (inline_write_summary): Update.
27454 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
27455 add account_size_time and reset member functions.
27456 (ipa_call_summary): Add reset function.
27457 * ipa-predicate.h (predicate::operator &): Constify.
27459 2017-05-22 Richard Biener <rguenther@suse.de>
27461 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
27463 2017-05-19 Jason Merrill <jason@redhat.com>
27465 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
27467 2017-05-19 Marek Polacek <polacek@redhat.com>
27470 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
27471 TYPE_OVERFLOW_WRAPS checks.
27473 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
27475 * tree-core.h (enum omp_clause_default_kind): Add
27476 "OMP_CLAUSE_DEFAULT_PRESENT".
27477 * tree-pretty-print.c (dump_omp_clause): Handle it.
27478 * gimplify.c (enum gimplify_omp_var_data): Add
27479 "GOVD_MAP_FORCE_PRESENT".
27480 (gimplify_adjust_omp_clauses_1): Map it to
27481 "GOMP_MAP_FORCE_PRESENT".
27482 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
27484 * gimplify.c (oacc_default_clause): Clarify.
27486 2017-05-19 Nathan Sidwell <nathan@acm.org>
27488 LANG_HOOK_REGISTER_DUMPS
27489 * toplev.c (general_init): Call register dump lang hook.
27490 * doc/invoke.texi: Document -fdump-lang option family.
27491 * dumpfile.c (dump_files): Remove class dump here.
27492 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
27493 * dumpfile.h (tree_dump_index): Remove TDI_class.
27494 * langhooks-def.h (lhd_register_dumps): Declare.
27495 (LANG_HOOKS_REGISTER_DUMPS): Define.
27496 (LANG_HOOKS_INITIALIZER): Add it.
27497 * langhooks.c (lhd_register_dumps): Define.
27498 * langhooks.h (struct lang_hooks): Add register_dumps.
27500 2017-05-19 Nathan Sidwell <nathan@acm.org>
27502 * context.h (context::set_passes): New.
27503 * context.c (context::context): Do not create pass manager.
27504 * toplev.c (general_init): Create pass manager here.
27506 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
27508 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
27509 use this splitter if two add or or instructions would also work for
27510 the constant we want to generate.
27512 2017-05-19 Richard Biener <rguenther@suse.de>
27515 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
27516 predicate evaluation.
27518 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27520 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
27522 * ipa-inline.c (want_inline_small_function_p): Do not cast to
27525 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27527 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
27528 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
27529 (inline_edge_summary_vec): Turn into ...
27530 (ipa_call_summaries): ... this one.
27531 (redirect_to_unreachable, edge_set_predicate,
27532 evaluate_properties_for_edge, inline_summary_alloc,
27533 reset_ipa_call_summary, reset_inline_summary,
27534 inline_summary_t::duplicate): Update.
27535 (inline_edge_duplication_hook): Turn to ...
27536 (ipa_call_summary_t::duplicate): ... this one.
27537 (inline_edge_removal_hook): Turn to ...
27538 (ipa_call_summary_t::remove): ... this one.
27539 (dump_inline_edge_summary): Turn to ...
27540 (dump_ipa_call_summary): ... this one.
27541 (estimate_function_body_sizes): Update.
27542 (inline_update_callee_summaries): Update.
27543 (remap_edge_change_prob): Update.
27544 (remap_edge_summaries): Update.
27545 (inline_merge_summary): Update.
27546 (do_estimate_edge_time): Update.
27547 (inline_generate_summary): Update.
27548 (inline_read_section): Update.
27549 (inline_read_summary): Update.
27550 (inline_free_summary): Update.
27551 * ipa-inline.c (can_inline_edge_p): Update.
27552 (compute_inlined_call_time): Update.
27553 (want_inline_small_function_p): Update.
27554 (edge_badness): Update.
27555 (early_inliner): Update.
27556 * ipa-inline.h (inline_edge_summary): Turn to ...
27557 (ipa_call_summary): ... this one.
27558 (ipa_call_summary_t): New class.
27559 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
27560 (ipa_call_summaries): New.
27561 (inline_edge_summary): Remove.
27562 (estimate_edge_growth): Update.
27563 * ipa-profile.c (ipa_propagate_frequency_1): Update.
27564 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27565 * ipa-split.c (execute_split_functions): Update.
27566 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
27568 2017-05-19 Richard Biener <rguenther@suse.de>
27570 PR middle-end/80764
27571 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
27573 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
27575 * config/rs6000/rs6000.c (struct machine_function): Add field
27576 fpr_is_wrapped_separately.
27577 (rs6000_get_separate_components): Use 64 components. Handle the
27578 new FPR components.
27579 (rs6000_components_for_bb): Handle the FPR components.
27580 (rs6000_emit_prologue_components): Handle the FPR components.
27581 (rs6000_emit_epilogue_components): Handle the FPR components.
27582 (rs6000_set_handled_components): Handle the FPR components.
27583 (rs6000_emit_prologue): Don't output prologue code for those FPRs
27584 that are already separately shrink-wrapped.
27585 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
27586 that are already separately shrink-wrapped.
27588 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27591 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
27594 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
27595 (define_peephole2 for Altivec d-form load): Add peepholes to catch
27596 cases where the register allocator uses a move and an offsettable
27597 memory operation to/from a FPR register on ISA 2.06/2.07.
27598 (define_peephole2 for Altivec d-form store): Likewise.
27600 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
27603 * config/i386/mmx.md (*mov<mode>_internal): Enable
27604 alternatives 11, 12, 13 and 14 also for 32bit targets.
27605 Remove alternatives 15, 16, 17 and 18.
27606 * config/i386/sse.md (vec_concatv2di): Change
27607 alternative (!x, *y) to (x, ?!*Yn).
27609 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
27611 * dumpfile.h (enum dump_kind): Remove stray comma.
27613 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27615 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
27616 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
27617 predicate::num_conditions
27618 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
27619 (CHANGED): turn into predicate::changed.
27620 (agg_position_info): Move to ipa-predicate.h
27621 (add_condition, predicate::add_clause, predicate::operator &=,
27622 predicate::or_with, predicate::evaluate, predicate::probability,
27623 dump_condition, dump_clause, predicate::dump,
27624 predicate::remap_after_duplication, predicate::remap_after_inlining,
27625 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
27626 (evaluate_conditions_for_known_args): Update.
27627 (set_cond_stmt_execution_predicate): Update.
27628 * ipa-inline.h: Include ipa-predicate.h
27629 (condition, inline_param_summary, conditions, agg_position_info,
27630 predicate): Move to ipa-predicate.h
27631 * ipa-predicate.c: New file.
27632 * ipa-predicate.h: New file.
27634 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
27636 * final.c (leaf_function_p): Check we are not in a sequence.
27638 2017-05-18 Martin Liska <mliska@suse.cz>
27640 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
27641 * dumpfile.c (dump_register): Use new enum dump_kind.
27642 (get_dump_file_name): Likewise.
27643 (dump_enable_all): Likewise.
27644 (dump_switch_p_1): Likewise.
27645 (enable_rtl_dump_file): Remove usage of TDF_RTL.
27646 * dumpfile.h (enum dump_kind): New enum type.
27647 (struct dump_file_info): Create constructor and
27648 format fields and comments.
27649 * passes.c (pass_manager::register_one_dump_file):
27651 * statistics.c (statistics_early_init): Likewise.
27652 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
27653 TDF_TREE with TDF_SLIM.
27654 (gather_memory_references_ref): Likewise.
27656 2017-05-18 Martin Liska <mliska@suse.cz>
27658 * vec.h (struct vnull): Use it.
27660 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27662 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
27663 (true_predicate, false_predicate, true_predicate_p,
27664 false_predicate_p): Remove.
27665 (single_cond_predicate, not_inlined_predicate): Turn to member function
27667 (add_condition): Update.
27668 (add_clause): Turn to...
27669 (predicate::add_clause): ... this one; update; allow passing NULL
27671 (and_predicates): Turn to ...
27672 (predicate::operator &=): ... this one.
27673 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
27674 (or_predicates): Turn to ...
27675 (predicate::or_with): ... this one.
27676 (evaluate_predicate): Turn to ...
27677 (predicate::evaluate): ... this one.
27678 (predicate_probability): Turn to ...
27679 (predicate::probability): ... this one.
27680 (dump_condition): Update.
27681 (dump_predicate): Turn to ...
27682 (predicate::dump): ... this one.
27683 (account_size_time): Update.
27684 (edge_set_predicate): Update.
27685 (set_hint_predicate): UPdate.
27686 (evaluate_conditions_for_known_args): Update.
27687 (evaluate_properties_for_edge): Update.
27688 (remap_predicate_after_duplication): Turn to...
27689 (predicate::remap_after_duplication): ... this one.
27690 (remap_hint_predicate_after_duplication): Update.
27691 (inline_summary_t::duplicate): UPdate.
27692 (dump_inline_edge_summary): Update.
27693 (dump_inline_summary): Update.
27694 (set_cond_stmt_execution_predicate): Update.
27695 (set_switch_stmt_execution_predicate): Update.
27696 (compute_bb_predicates): Update.
27697 (will_be_nonconstant_expr_predicate): Update.
27698 (will_be_nonconstant_predicate): Update.
27699 (phi_result_unknown_predicate): Update.
27700 (predicate_for_phi_result): Update.
27701 (array_index_predicate): Update.
27702 (estimate_function_body_sizes): Update.
27703 (estimate_node_size_and_time): Update.
27704 (estimate_ipcp_clone_size_and_time): Update.
27705 (remap_predicate): Rename to ...
27706 (predicate::remap_after_inlining): ... this one.
27707 (remap_hint_predicate): Update.
27708 (inline_merge_summary): Update.
27709 (inline_update_overall_summary): Update.
27710 (estimate_size_after_inlining): Update.
27711 (read_predicate): Rename to ...
27712 (predicate::stream_in): ... this one.
27713 (read_inline_edge_summary): Update.
27714 (write_predicate): Rename to ...
27715 (predicate::stream_out): ... this one.
27716 (write_inline_edge_summary): Update.
27717 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
27718 (clause_t): Turn to uint32_t
27719 (predicate): Turn to class; implement constructor and operators
27721 (size_time_entry): Update.
27722 (inline_summary): Update.
27723 (inline_edge_summary): Update.
27725 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
27727 * fold-const.c (fold_binary_loc): Move transformation...
27728 * match.pd (C - X CMP X): ... here.
27730 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
27732 * config/sparc/sparc.c (sparc_option_override): Set function
27733 alignment for -mcpu=niagara7 to 64 to match the I$ line.
27734 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
27736 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
27738 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
27740 2017-05-18 Marek Polacek <polacek@redhat.com>
27743 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
27744 (pass_ubsan::execute): Call gimple_assign_single_p instead of
27745 gimple_assign_load_p.
27747 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
27749 PR middle-end/80692
27750 * real.c (do_compare): Give decimal_do_compare preference over
27751 comparing just the signs.
27753 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
27755 * doc/md.texi (Canonicalization of Instructions): Describe the
27756 canonical form of instructions that inherently set a condition
27759 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
27761 PR middle-end/80775
27762 * tree-cfg.c: Move deletion of unreachable case statements to after
27763 the merging of consecutive case labels.
27765 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
27767 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
27768 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
27769 restoring of callee-saved registers.
27771 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
27773 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
27774 * config/visium/visium.c (single_set_and_flags): Likewise.
27775 * config/visium/visium.md (Substitutions): Likewise.
27777 2017-05-17 Martin Liska <mliska@suse.cz>
27779 * cfg.c: Introduce dump_flags_t type and
27780 use it instead of int type.
27782 * cfghooks.c: Likewise.
27783 * cfghooks.h (struct cfg_hooks): Likewise.
27784 * cfgrtl.c: Likewise.
27785 * cfgrtl.h: Likewise.
27786 * cgraph.c (cgraph_node::get_body): Likewise.
27787 * coretypes.h: Likewise.
27788 * domwalk.c: Likewise.
27789 * domwalk.h: Likewise.
27790 * dumpfile.c (struct dump_option_value_info): Likewise.
27791 (dump_enable_all): Likewise.
27792 (dump_switch_p_1): Likewise.
27793 (opt_info_switch_p): Likewise.
27794 * dumpfile.h (enum tree_dump_index): Likewise.
27795 (struct dump_file_info): Likewise.
27796 * genemit.c: Likewise.
27797 * generic-match-head.c: Likewise.
27798 * gengtype.c (open_base_files): Likewise.
27799 * gimple-pretty-print.c: Likewise.
27800 * gimple-pretty-print.h: Likewise.
27801 * graph.c (print_graph_cfg): Likewise.
27802 * graphite-scop-detection.c (dot_all_sese): Likewise.
27803 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
27804 * loop-unroll.c (report_unroll): Likewise.
27805 * passes.c (pass_manager::register_one_dump_file): Likewise.
27806 * print-tree.c: Likewise.
27807 * statistics.c: Likewise.
27808 * tree-cfg.c: Likewise.
27809 * tree-cfg.h: Likewise.
27810 * tree-dfa.c: Likewise.
27811 * tree-dfa.h: Likewise.
27812 * tree-dump.c (dump_function): Likewise.
27813 * tree-dump.h (struct dump_info): Likewise.
27814 * tree-pretty-print.c: Likewise.
27815 * tree-pretty-print.h: Likewise.
27816 * tree-ssa-live.c: Likewise.
27817 * tree-ssa-live.h: Likewise.
27818 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
27819 * tree-vect-loop.c: Likewise.
27820 * tree-vect-slp.c: Likewise.
27822 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
27823 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27825 PR tree-optimization/80457
27826 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
27827 of all arguments to a statement as scalar_to_vec operations.
27828 (vectorizable_call): Adjust call to vect_model_simple_cost for
27830 (vectorizable_conversion): Likewise.
27831 (vectorizable_assignment): Likewise.
27832 (vectorizable_shift): Likewise.
27833 (vectorizable_operation): Likewise.
27834 (vectorizable_comparison): Likewise.
27835 (vect_is_simple_cond): Record the def types for operands.
27836 (vectorizable_condition): Likewise, call vect_model_simple_cost.
27837 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
27838 for statement argument count.
27840 2017-05-16 Carl Love <cel@us.ibm.com>
27842 * config/rs6000/rs6000-c: Add support for built-in functions
27843 vector unsigned long long vec_bperm (vector unsigned long long,
27844 vector unsigned char)
27845 vector signed long long vec_mule (vector signed int,
27847 vector unsigned long long vec_mule (vector unsigned int,
27848 vector unsigned int)
27849 vector signed long long vec_mulo (vector signed int,
27851 vector unsigned long long vec_mulo (vector unsigned int,
27852 vector unsigned int)
27853 vector signed char vec_sldw (vector signed char,
27854 vector signed char,
27856 vector unsigned char vec_sldw (vector unsigned char,
27857 vector unsigned char,
27859 vector signed short vec_sldw (vector signed short,
27860 vector signed short,
27862 vector unsigned short vec_sldw (vector unsigned short,
27863 vector unsigned short,
27865 vector signed int vec_sldw (vector signed int,
27868 vector unsigned int vec_sldw (vector unsigned int,
27869 vector unsigned int,
27871 vector signed long long vec_sldw (vector signed long long,
27872 vector signed long long,
27874 vector unsigned long long vec_sldw (vector unsigned long long,
27875 vector unsigned long long,
27877 * config/rs6000/rs6000-c: Add support for built-in functions
27878 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
27879 * config/rs6000/altivec.h: Add defintion for vec_sldw.
27880 * doc/extend.texi: Update the built-in documentation for the
27881 new built-in functions.
27883 2017-05-16 Marek Polacek <polacek@redhat.com>
27887 * tree.c (save_expr): Don't fold the expression.
27889 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
27891 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
27892 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
27893 and (?*y,m). Update insn attributes.
27895 2017-05-16 Martin Liska <mliska@suse.cz>
27897 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
27898 flags argument of print_gimple_stmt, print_gimple_expr,
27899 print_generic_stmt and print_generic_expr.
27900 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
27901 * coretypes.h: Likewise.
27902 * except.c (dump_eh_tree): Likewise.
27903 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
27904 * gimple-pretty-print.h: Likewise.
27905 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
27906 (backprop::push_to_worklist): Likewise.
27907 (backprop::pop_from_worklist): Likewise.
27908 (backprop::process_use): Likewise.
27909 (backprop::intersect_uses): Likewise.
27910 (note_replacement): Likewise.
27911 * gimple-ssa-store-merging.c
27912 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
27913 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
27914 (pass_store_merging::execute): Likewise.
27915 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
27916 (ssa_base_cand_dump_callback): Likewise.
27917 (dump_incr_vec): Likewise.
27918 (replace_refs): Likewise.
27919 (replace_mult_candidate): Likewise.
27920 (create_add_on_incoming_edge): Likewise.
27921 (create_phi_basis): Likewise.
27922 (insert_initializers): Likewise.
27923 (all_phi_incrs_profitable): Likewise.
27924 (introduce_cast_before_cand): Likewise.
27925 (replace_one_candidate): Likewise.
27926 * gimplify.c (gimplify_expr): Likewise.
27927 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
27928 (set_rename): Likewise.
27929 (rename_uses): Likewise.
27930 (copy_loop_phi_nodes): Likewise.
27931 (add_close_phis_to_merge_points): Likewise.
27932 (copy_loop_close_phi_args): Likewise.
27933 (copy_cond_phi_args): Likewise.
27934 (graphite_copy_stmts_from_block): Likewise.
27935 (translate_pending_phi_nodes): Likewise.
27936 * graphite-poly.c (print_pdr): Likewise.
27937 (dump_gbb_cases): Likewise.
27938 (dump_gbb_conditions): Likewise.
27939 (print_scop_params): Likewise.
27940 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
27941 (build_cross_bb_scalars_use): Likewise.
27942 (gather_bbs::before_dom_children): Likewise.
27943 * hsa-dump.c (dump_hsa_immed): Likewise.
27944 * ipa-cp.c (print_ipcp_constant_value): Likewise.
27945 (get_replacement_map): Likewise.
27946 * ipa-inline-analysis.c (dump_condition): Likewise.
27947 (estimate_function_body_sizes): Likewise.
27948 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
27949 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
27950 * ipa-prop.c (ipa_dump_param): Likewise.
27951 (ipa_print_node_jump_functions_for_edge): Likewise.
27952 (ipa_modify_call_arguments): Likewise.
27953 (ipa_modify_expr): Likewise.
27954 (ipa_dump_param_adjustments): Likewise.
27955 (ipa_dump_agg_replacement_values): Likewise.
27956 (ipcp_modif_dom_walker::before_dom_children): Likewise.
27957 * ipa-pure-const.c (check_stmt): Likewise.
27958 (pass_nothrow::execute): Likewise.
27959 * ipa-split.c (execute_split_functions): Likewise.
27960 * omp-offload.c (dump_oacc_loop_part): Likewise.
27961 (dump_oacc_loop): Likewise.
27962 * trans-mem.c (tm_log_emit): Likewise.
27963 (tm_memopt_accumulate_memops): Likewise.
27964 (dump_tm_memopt_set): Likewise.
27965 (dump_tm_memopt_transform): Likewise.
27966 * tree-cfg.c (gimple_verify_flow_info): Likewise.
27967 (print_loop): Likewise.
27968 * tree-chkp-opt.c (chkp_print_addr): Likewise.
27969 (chkp_gather_checks_info): Likewise.
27970 (chkp_get_check_result): Likewise.
27971 (chkp_remove_check_if_pass): Likewise.
27972 (chkp_use_outer_bounds_if_possible): Likewise.
27973 (chkp_reduce_bounds_lifetime): Likewise.
27974 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
27975 (chkp_mark_completed_bounds): Likewise.
27976 (chkp_register_incomplete_bounds): Likewise.
27977 (chkp_mark_invalid_bounds): Likewise.
27978 (chkp_maybe_copy_and_register_bounds): Likewise.
27979 (chkp_build_returned_bound): Likewise.
27980 (chkp_get_bound_for_parm): Likewise.
27981 (chkp_build_bndldx): Likewise.
27982 (chkp_get_bounds_by_definition): Likewise.
27983 (chkp_generate_extern_var_bounds): Likewise.
27984 (chkp_get_bounds_for_decl_addr): Likewise.
27985 * tree-chrec.c (chrec_apply): Likewise.
27986 * tree-data-ref.c (dump_data_reference): Likewise.
27987 (dump_subscript): Likewise.
27988 (dump_data_dependence_relation): Likewise.
27989 (analyze_overlapping_iterations): Likewise.
27990 * tree-inline.c (expand_call_inline): Likewise.
27991 (tree_function_versioning): Likewise.
27992 * tree-into-ssa.c (dump_defs_stack): Likewise.
27993 (dump_currdefs): Likewise.
27994 (dump_names_replaced_by): Likewise.
27995 (dump_update_ssa): Likewise.
27996 (update_ssa): Likewise.
27997 * tree-object-size.c (pass_object_sizes::execute): Likewise.
27998 * tree-parloops.c (build_new_reduction): Likewise.
27999 (try_create_reduction_list): Likewise.
28000 (ref_conflicts_with_region): Likewise.
28001 (oacc_entry_exit_ok_1): Likewise.
28002 (oacc_entry_exit_single_gang): Likewise.
28003 * tree-pretty-print.h: Likewise.
28004 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
28005 (get_scalar_evolution): Likewise.
28006 (add_to_evolution): Likewise.
28007 (get_loop_exit_condition): Likewise.
28008 (analyze_evolution_in_loop): Likewise.
28009 (analyze_initial_condition): Likewise.
28010 (analyze_scalar_evolution): Likewise.
28011 (instantiate_scev): Likewise.
28012 (number_of_latch_executions): Likewise.
28013 (gather_chrec_stats): Likewise.
28014 (final_value_replacement_loop): Likewise.
28015 (scev_const_prop): Likewise.
28016 * tree-sra.c (dump_access): Likewise.
28017 (disqualify_candidate): Likewise.
28018 (create_access): Likewise.
28019 (reject): Likewise.
28020 (maybe_add_sra_candidate): Likewise.
28021 (create_access_replacement): Likewise.
28022 (analyze_access_subtree): Likewise.
28023 (analyze_all_variable_accesses): Likewise.
28024 (sra_modify_assign): Likewise.
28025 (initialize_constant_pool_replacements): Likewise.
28026 (find_param_candidates): Likewise.
28027 (decide_one_param_reduction): Likewise.
28028 (replace_removed_params_ssa_names): Likewise.
28029 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
28030 * tree-ssa-copy.c (dump_copy_of): Likewise.
28031 (copy_prop_visit_cond_stmt): Likewise.
28032 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
28033 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
28034 (record_equivalences_from_stmt): Likewise.
28035 * tree-ssa-dse.c (compute_trims): Likewise.
28036 (delete_dead_call): Likewise.
28037 (delete_dead_assignment): Likewise.
28038 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
28039 (forward_propagate_into_cond): Likewise.
28040 (pass_forwprop::execute): Likewise.
28041 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
28042 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
28044 (move_computations_worker): Likewise.
28045 (execute_sm): Likewise.
28046 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
28047 (remove_exits_and_undefined_stmts): Likewise.
28048 (remove_redundant_iv_tests): Likewise.
28049 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
28050 (adjust_iv_update_pos): Likewise.
28051 * tree-ssa-math-opts.c (bswap_replace): Likewise.
28052 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
28053 (value_replacement): Likewise.
28054 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
28055 * tree-ssa-pre.c (print_pre_expr): Likewise.
28056 (get_representative_for): Likewise.
28057 (create_expression_by_pieces): Likewise.
28058 (insert_into_preds_of_block): Likewise.
28059 (eliminate_insert): Likewise.
28060 (eliminate_dom_walker::before_dom_children): Likewise.
28061 (eliminate): Likewise.
28062 (remove_dead_inserted_code): Likewise.
28063 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
28064 * tree-ssa-reassoc.c (get_rank): Likewise.
28065 (eliminate_duplicate_pair): Likewise.
28066 (eliminate_plus_minus_pair): Likewise.
28067 (eliminate_not_pairs): Likewise.
28068 (undistribute_ops_list): Likewise.
28069 (eliminate_redundant_comparison): Likewise.
28070 (update_range_test): Likewise.
28071 (optimize_range_tests_var_bound): Likewise.
28072 (optimize_vec_cond_expr): Likewise.
28073 (rewrite_expr_tree): Likewise.
28074 (rewrite_expr_tree_parallel): Likewise.
28075 (linearize_expr): Likewise.
28076 (break_up_subtract): Likewise.
28077 (linearize_expr_tree): Likewise.
28078 (attempt_builtin_powi): Likewise.
28079 (attempt_builtin_copysign): Likewise.
28080 (transform_stmt_to_copy): Likewise.
28081 (transform_stmt_to_multiply): Likewise.
28082 (dump_ops_vector): Likewise.
28083 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
28084 (print_scc): Likewise.
28085 (set_ssa_val_to): Likewise.
28086 (visit_reference_op_store): Likewise.
28087 (visit_use): Likewise.
28088 (sccvn_dom_walker::before_dom_children): Likewise.
28089 (run_scc_vn): Likewise.
28090 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
28092 (expr_hash_elt::print): Likewise.
28093 (const_and_copies::pop_to_marker): Likewise.
28094 (const_and_copies::record_const_or_copy_raw): Likewise.
28095 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
28096 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
28097 (dump_predicates): Likewise.
28098 (find_uninit_use): Likewise.
28099 (warn_uninitialized_phi): Likewise.
28100 (pass_late_warn_uninitialized::execute): Likewise.
28101 * tree-ssa.c (verify_vssa): Likewise.
28102 (verify_ssa): Likewise.
28103 (maybe_optimize_var): Likewise.
28104 * tree-vrp.c (dump_value_range): Likewise.
28105 (dump_all_value_ranges): Likewise.
28106 (dump_asserts_for): Likewise.
28107 (register_edge_assert_for_2): Likewise.
28108 (vrp_visit_cond_stmt): Likewise.
28109 (vrp_visit_switch_stmt): Likewise.
28110 (vrp_visit_stmt): Likewise.
28111 (vrp_visit_phi_node): Likewise.
28112 (simplify_cond_using_ranges_1): Likewise.
28113 (fold_predicate_in): Likewise.
28114 (evrp_dom_walker::before_dom_children): Likewise.
28115 (evrp_dom_walker::push_value_range): Likewise.
28116 (evrp_dom_walker::pop_value_range): Likewise.
28117 (execute_early_vrp): Likewise.
28119 2017-05-16 Richard Biener <rguenther@suse.de>
28121 * dwarf2out.c (loc_list_from_tree_1): Do not create
28122 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
28124 2017-05-16 Richard Biener <rguenther@suse.de>
28126 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
28128 (note_variable_value_in_expr): If we resolved the decl ref
28129 do not push to the stack.
28131 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
28133 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
28134 operations in fast-math mode.
28135 (vaddq_f16): Likewise.
28136 (vmul_f16): Likewise.
28137 (vmulq_f16): Likewise.
28138 (vsub_f16): Likewise.
28139 (vsubq_f16): Likewise.
28140 * config/arm/neon.md (add<mode>3): New.
28142 (fma:<VH:mode>3): New. Also remove outdated comment.
28145 2017-05-16 Martin Liska <mliska@suse.cz>
28149 * ipa-devirt.c (warn_types_mismatch): Fix typo.
28150 (odr_types_equivalent_p): Likewise.
28152 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
28154 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
28156 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
28159 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
28160 non-interunit SSE move alternatives with '?'.
28161 (zero-extendsidi peephole2): New peephole to skip intermediate
28162 general register in SSE zero-extend sequence.
28164 2017-05-15 Jeff Law <law@redhat.com>
28166 * reorg.c (relax_delay_slots): Create a new variable to hold
28167 the temporary target rather than clobbering TARGET_LABEL.
28169 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
28170 missing argument to extract_bit_field call.
28171 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
28173 2017-05-15 Martin Liska <mliska@suse.cz>
28176 * gcc.c (process_command): Do not allow empty argument of -o option.
28178 2017-05-15 Renlin Li <renlin.li@arm.com>
28180 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
28181 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
28182 * config/aarch64/constraints.md (Usf): Add long call check.
28183 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
28184 (call_value): Likewise.
28185 (sibcall): Likewise.
28186 (sibcall_value): Likewise.
28188 (call_value_insn): New.
28189 (sibcall_insn): Update rtx pattern.
28190 (sibcall_value_insn): Likewise.
28191 (call_internal): Remove.
28192 (call_value_internal): Likewise.
28193 (sibcall_internal): Likewise.
28194 (sibcall_value_internal): Likewise.
28195 (call_reg): Likewise.
28196 (call_symbol): Likewise.
28197 (call_value_reg): Likewise.
28198 (call_value_symbol): Likewise.
28200 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
28203 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
28205 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
28207 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
28208 compatible with CCGOCmode and with CCZmode.
28210 2017-05-14 Martin Sebor <msebor@redhat.com>
28212 PR middle-end/77671
28213 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
28214 (gimple_fold_builtin_snprintf): Same.
28215 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
28216 (gimple_fold_builtin_snprintf): Same.
28217 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
28218 of character types.
28219 (is_call_safe): New function.
28220 (try_substitute_return_value): Call it.
28221 (try_simplify_call): New function.
28222 (pass_sprintf_length::handle_gimple_call): Call it.
28224 2017-05-14 Martin Sebor <msebor@redhat.com>
28226 PR middle-end/80669
28227 * builtins.c (expand_builtin_stpncpy): Simplify.
28229 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
28231 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
28232 * config/i386/i386.h
28233 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
28234 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
28235 (struct machine_function): Add new members call_ms2sysv,
28236 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
28237 (struct machine_frame_state): New fields sp_realigned and
28238 sp_realigned_offset.
28239 * config/i386/i386.c
28240 (enum xlogue_stub): New enum.
28241 (enum xlogue_stub_sets): New enum.
28242 (class xlogue_layout): New class.
28243 (struct ix86_frame): New fields stack_realign_allocate_offset,
28244 stack_realign_offset and outlined_save_offset. Modify comments to
28245 detail stack layout when using out-of-line stubs.
28246 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
28247 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
28248 -mcall-ms2sysv-xlogues.
28249 (stub_managed_regs): New static variable.
28250 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
28251 registers managed by out-of-line stub.
28252 (disable_call_ms2sysv_xlogues): New function.
28253 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
28254 m->call_ms2sysv when appropriate and compute frame layout for
28256 (sp_valid_at, fp_valid_at): New inline functions.
28257 (choose_basereg): New function.
28258 (choose_baseaddr): Add align parameter, use choose_basereg and modify
28260 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
28261 Use align parameter of choose_baseaddr to generated aligned SSE movs
28263 (pro_epilogue_adjust_stack): Modify to track
28264 machine_frame_state::sp_realigned.
28265 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
28266 (ix86_nsaved_sseregs): Likewise.
28267 (ix86_emit_save_regs): Likewise.
28268 (ix86_emit_save_regs_using_mov): Likewise.
28269 (ix86_emit_save_sse_regs_using_mov): Likewise.
28270 (get_scratch_register_on_entry): Likewise.
28271 (gen_frame_set): New function.
28272 (gen_frame_load): Likewise.
28273 (gen_frame_store): Likewise.
28274 (emit_outlined_ms2sysv_save): Likewise.
28275 (emit_outlined_ms2sysv_restore): Likewise.
28276 (ix86_expand_prologue): Modify stack re-alignment code and call
28277 emit_outlined_ms2sysv_save when appropriate.
28278 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
28279 parameter rtx_insn *insn, which allows the function to be used to only
28280 generate the notes.
28281 (ix86_expand_epilogue): Modify validity checks of frame and stack
28282 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
28283 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
28284 * config/i386/predicates.md
28285 (save_multiple): New predicate.
28286 (restore_multiple): Likewise.
28287 * config/i386/sse.md
28288 (save_multiple<mode>): New pattern.
28289 (save_multiple_realign<mode>): Likewise.
28290 (restore_multiple<mode>): Likewise.
28291 (restore_multiple_and_return<mode>): Likewise.
28292 (restore_multiple_leave_return<mode>): Likewise.
28293 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
28295 2017-05-14 Julia Koval <julia.koval@intel.com>
28297 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
28298 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
28299 (__builtin_ia32_xsetbv): New builtins.
28300 * config/i386/i386.c (ix86_expand_special_args_builtin):
28302 (ix86_expand_builtin): Special expand for new intrinsics.
28303 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
28304 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
28305 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
28307 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28309 * cfganal.c (inverted_post_order_compute): Change argument type
28311 * cfganal.h (inverted_post_order_compute): Adjust prototype.
28312 * df-core.c (rest_of_handle_df_initialize): Adjust.
28313 (rest_of_handle_df_finish): Likewise.
28314 (df_analyze_1): Likewise.
28315 (df_analyze): Likewise.
28316 (loop_inverted_post_order_compute): Change argument to be a vec *.
28317 (df_analyze_loop): Adjust.
28318 (df_get_n_blocks): Likewise.
28319 (df_get_postorder): Likewise.
28320 * df.h (struct df_d): Change field to be a vec.
28321 * lcm.c (compute_laterin): Adjust.
28322 (compute_available): Likewise.
28323 * lra-lives.c (lra_create_live_ranges_1): Likewise.
28324 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
28325 * tree-ssa-pre.c (compute_antic): Likewise.
28327 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28329 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
28330 (depth_first_search::depth_first_search): Change structure init
28331 function to this constructor.
28332 (depth_first_search::add_bb): Rename function to this member.
28333 (depth_first_search::execute): Likewise.
28334 (flow_dfs_compute_reverse_finish): Adjust.
28336 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28338 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
28339 (longest_simple_path): Likewise.
28340 * shrink-wrap.c (spread_components): Likewise.
28341 (disqualify_problematic_components): Likewise.
28342 (emit_common_heads_for_components): Likewise.
28343 (emit_common_tails_for_components): Likewise.
28344 (insert_prologue_epilogue_for_components): Likewise.
28346 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28348 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
28351 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28353 * df-core.c (df_set_blocks): Start using auto_bitmap.
28354 (df_compact_blocks): Likewise.
28355 * df-problems.c (df_rd_confluence_n): Likewise.
28356 * df-scan.c (df_insn_rescan_all): Likewise.
28357 (df_process_deferred_rescans): Likewise.
28358 (df_update_entry_block_defs): Likewise.
28359 (df_update_exit_block_uses): Likewise.
28360 (df_entry_block_bitmap_verify): Likewise.
28361 (df_exit_block_bitmap_verify): Likewise.
28362 (df_scan_verify): Likewise.
28363 * lra-constraints.c (lra_constraints): Likewise.
28364 (undo_optional_reloads): Likewise.
28365 (lra_undo_inheritance): Likewise.
28366 * lra-remat.c (calculate_gen_cands): Likewise.
28367 (do_remat): Likewise.
28368 * lra-spills.c (assign_spill_hard_regs): Likewise.
28369 (spill_pseudos): Likewise.
28370 * tree-ssa-pre.c (bitmap_set_and): Likewise.
28371 (bitmap_set_subtract_values): Likewise.
28373 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28375 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
28376 management with auto_bitmap.
28377 (fix_inter_tick): Likewise.
28378 (fix_recovery_deps): Likewise.
28379 * ira.c (add_store_equivs): Likewise.
28380 (find_moveable_pseudos): Likewise.
28381 (split_live_ranges_for_shrink_wrap): Likewise.
28382 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
28383 (rtx_reuse_manager::seen_def_p): Likewise.
28384 (rtx_reuse_manager::set_seen_def): Likewise.
28385 * print-rtl.h (class rtx_reuse_manager): Likewise.
28387 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28389 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
28391 (migrate_btr_def): Likewise.
28392 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
28393 * df-core.c (loop_post_order_compute): Likewise.
28394 (loop_inverted_post_order_compute): Likewise.
28395 * hsa-common.h: Likewise.
28396 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
28397 * init-regs.c (initialize_uninitialized_regs): Likewise.
28398 * ipa-inline.c (resolve_noninline_speculation): Likewise.
28399 (inline_small_functions): Likewise.
28400 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
28401 * ira.c (combine_and_move_insns): Likewise.
28402 (build_insn_chain): Likewise.
28403 * loop-invariant.c (find_invariants): Likewise.
28404 * lower-subreg.c (propagate_pseudo_copies): Likewise.
28405 * predict.c (tree_predict_by_opcode): Likewise.
28406 (predict_paths_leading_to): Likewise.
28407 (predict_paths_leading_to_edge): Likewise.
28408 (estimate_loops_at_level): Likewise.
28409 (estimate_loops): Likewise.
28410 * shrink-wrap.c (try_shrink_wrapping): Likewise.
28411 (spread_components): Likewise.
28412 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
28413 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
28414 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
28415 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
28416 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
28417 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
28418 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
28419 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
28420 (mark_threaded_blocks): Likewise.
28421 (thread_through_all_blocks): Likewise.
28422 * tree-ssa.c (verify_ssa): Likewise.
28423 (execute_update_addresses_taken): Likewise.
28424 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
28426 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28428 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
28430 (post_order_compute): Likewise.
28431 (inverted_post_order_compute): Likewise.
28432 (pre_and_rev_post_order_compute_fn): Likewise.
28434 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28436 * genrecog.c (int_set::int_set): Explicitly construct our
28437 auto_vec base class.
28438 * vec.h (auto_vec::auto_vec): New constructor.
28440 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28442 * bitmap.h (class auto_bitmap): New constructor taking
28443 bitmap_obstack * argument.
28445 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28447 * bitmap.h (class auto_bitmap): Change type of m_bits to
28448 bitmap_head, and adjust ctor / dtor and member operators.
28450 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
28452 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
28453 when returned register mode doesn't match original mode.
28455 2017-05-12 Jeff Law <law@redhat.com>
28456 Jakub Jelinek <jakub@redhat.com>
28458 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
28459 we look for cc setter after the compare-elim changes.
28460 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
28461 within the vector to match what compare-elim now expects.
28462 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
28463 (xorsi3_flags, one_cmplsi2_flags): Likewise.
28465 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
28466 after the compare-elim changes.
28467 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
28468 the vector to match what compare-elim now expects.
28469 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
28470 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
28471 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
28472 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
28473 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
28475 * config/visium/visium.c (single_set_and_flags): Fix where
28476 we look for cc setter after the compare-elim changes.
28477 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
28478 with the vector to match what compare-elim now expects.
28479 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
28480 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
28481 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
28482 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
28483 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
28484 (neg<mode>2_insn_set_overflow): Likewise.
28486 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
28488 PR middle-end/79794
28489 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
28490 maybe_expand_insn call, set ops[0].target. If still set after call,
28491 set alt_rtl. Add extra arg to recursive calls.
28492 (extract_bit_field): Add alt_rtl argument. Pass to
28494 * expmed.h (extract_bit_field): Fix prototype.
28495 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
28496 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
28497 to extract_bit_field_calls.
28498 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
28499 Pass alt_rtl to extract_bit_field calls.
28500 * calls.c (store_unaligned_arguments_into_psuedos)
28501 load_register_parameters): Pass extra NULL to extract_bit_field calls.
28502 * optabs.c (maybe_legitimize_operand): Clear op->target when call
28504 * optabs.h (struct expand_operand): Add target bitfield.
28506 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28508 * compare-elim.c (try_eliminate_compare): Canonicalize
28509 operation with embedded compare to
28510 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
28511 (set (reg) (operation)].
28513 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
28515 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28518 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
28519 cost of adding a carry flag for ADC instruction.
28520 [case MINUS]: Ignore the cost of subtracting a carry flag
28521 for SBB instruction.
28523 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
28525 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
28527 * config/rs6000/bmiintrin.h: New file.
28528 * config/rs6000/bmi2intrin.h: New file.
28529 * config/rs6000/x86intrin.h: New file.
28531 2017-05-12 Jeff Law <law@redhat.com>
28533 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
28536 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
28538 PR middle-end/80707
28539 * tree-cfg.c: Remove cfg edges of unreachable case statements.
28541 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28543 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28544 early expansion of vector divide builtins.
28545 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
28546 builtins identified as having unsigned arguments.
28548 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28550 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
28551 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
28552 expansion of vector logical operations (and, andc, or, xor,
28555 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28557 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
28558 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
28560 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28562 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28563 early GIMPLE expansion of vector multiplies.
28565 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28567 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
28568 TARGET_HAVE_MOVT conditional.
28569 (movt splitter): Likewise.
28571 2017-05-12 Richard Biener <rguenther@suse.de>
28573 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
28574 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
28575 Fold all stmts not inplace.
28577 2017-05-12 Richard Biener <rguenther@suse.de>
28579 PR tree-optimization/80713
28580 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
28581 inserted_exprs bit for not removed stmts.
28583 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
28585 PR middle-end/69921
28586 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
28587 parallelized" attribute for parallelized OpenACC kernels.
28588 * omp-offload.c (execute_oacc_device_lower): Use it.
28590 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
28591 Set "oacc kernels" attribute.
28592 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
28593 parameter. Adjust all users.
28594 (oacc_fn_attrib_kernels_p): Remove function.
28595 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
28596 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
28597 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
28598 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
28599 assert "oacc kernels" attribute is set.
28601 2017-05-11 Carl Love <cel@us.ibm.com>
28603 * config/rs6000/rs6000-c: Add support for built-in functions
28604 vector unsigned char vec_popcnt (vector signed char)
28605 vector unsigned char vec_popcnt (vector unsigned char)
28606 vector unsigned short vec_popcnt (vector signed short)
28607 vector unsigned short vec_popcnt (vector unsigned short)
28608 vector unsigned int vec_popcnt (vector signed int)
28609 vector unsigned int vec_popcnt (vector unsigned int)
28610 vector unsigned long long vec_popcnt (vector signed long long)
28611 vector unsigned long long vec_popcnt (vector unsigned long long)
28612 vector signed long long vec_slo (vector signed long long,
28613 vector signed char)
28614 vector signed long long vec_slo (vector signed long long,
28615 vector unsigned char)
28616 vector unsigned long long vec_slo (vector unsigned long long,
28617 vector signed char)
28618 vector unsigned long long vec_slo (vector unsigned long long,
28619 vector unsigned char)
28620 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
28621 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
28622 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
28623 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
28624 * doc/extend.texi: Update the built-in documentation file for the
28625 new built-in functions.
28627 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
28629 * attribs.h (sorted_attr_string): Move machine independent
28630 functions for target clone support from the i386 port to common
28631 code. Rename ix86_function_versions to common_function_versions.
28632 Rename make_name to make_unique_name.
28633 (common_function_versions): Likewise.
28634 (make_unique_name): Likewise.
28635 (make_dispatcher_decl): Likewise.
28636 (is_function_default_version): Likewise.
28637 * attribs.c (attr_strcmp): Likewise.
28638 (sorted_attr_string): Likewise.
28639 (common_function_versions): Likewise.
28640 (make_unique_name): Likewise.
28641 (make_dispatcher_decl): Likewise.
28642 (is_function_default_version): Likewise.
28643 * config/i386/i386.c (attr_strcmp): Likewise.
28644 (sorted_attr_string): Likewise.
28645 (ix86_function_versions): Likewise.
28646 (make_name): Likewise.
28647 (make_dispatcher_decl): Likewise.
28648 (is_function_default_version): Likewise.
28649 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
28651 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28654 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
28655 Account for direct move costs for vec_construct of integer
28658 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
28661 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
28662 (UNSPEC_STX_ATOMIC): Ditto.
28663 (loaddi_via_sse): New insn.
28664 (storedi_via_sse): Ditto.
28665 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
28666 Update corresponding peephole2 patterns.
28667 (atomic_storedi_fpu): Ditto.
28669 2017-05-11 Julia Koval <julia.koval@intel.com>
28671 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
28672 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
28674 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
28675 (__builtin_ia32_rsqrt14ss_mask): New builtins.
28676 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
28678 2017-05-11 Nathan Sidwell <nathan@acm.org>
28680 * graphite-poly.c: Include dumpfile.h.
28682 * dumpfle.h (dump_function): Declare here ...
28683 * tree-dump.h (dump_function): ... not here.
28684 * dumpfile.c: #include tree-cfg.h.
28685 (dump_function): Move here from ...
28686 * tree-dump.c (dump_function): ... here.
28687 * gimplify.c: #include splay-tree.h, not tree-dump.h.
28688 * graphite-poly.c: Don't include tree-dump.h.
28689 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
28690 * print-tree.c: Likewise.
28691 * stor-layout.c: Likewise.
28692 * tree-nested.c: Likewise.
28694 * dumpfile.c (dump_start): Use TDF_FLAGS.
28695 (dump_enable_all): Fix TDF_KIND check thinko.
28697 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
28699 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
28700 array entries to represent two legal parameterizations of the
28701 overloaded __builtin_cmpb function, as represented by the
28702 P6_OV_BUILTIN_CMPB constant.
28703 (altivec_resolve_overloaded_builtin): Add special case handling
28704 for the __builtin_cmpb function, as represented by the
28705 P6_OV_BUILTIN_CMPB constant.
28706 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
28707 (BU_P6_64BIT_2): New macro.
28708 (BU_P6_OVERLOAD_2): New macro
28709 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
28710 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
28711 (CMPB): Add overload support to represent both 32-bit and 64-bit
28712 compare-bytes function.
28713 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
28714 support for TARGET_CMPB.
28715 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
28716 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28717 documentation of the __builtin_cmpb overloaded built-in function.
28719 2017-05-11 Richard Biener <rguenther@suse.de>
28721 PR tree-optimization/80705
28722 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
28723 bases are not vectorizable.
28725 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28727 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
28728 when counting register pressure.
28730 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28732 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
28733 (iv_ca_more_deps): Renamed to ...
28734 (iv_ca_compare_deps): ... this.
28735 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
28737 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28739 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
28741 (determine_group_iv_costs): ... here.
28742 (find_inv_vars_cb): Record inv var if it's not recorded before.
28744 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28746 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
28747 (get_shiftadd_cost): Ditto.
28749 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28751 * tree-ssa-address.c: Include header file.
28752 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
28754 (add_to_parts): Refactor.
28755 (addr_to_parts): New parameter. Update use of move_hint_to_base.
28756 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
28759 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28761 PR tree-optimization/53090
28762 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
28764 (extract_cond_operands): Detect condition with IV on both sides
28765 and return COMP_IV_EXPR_2.
28766 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
28767 (rewrite_use_compare): Simplify by removing call to function
28768 extract_cond_operands.
28770 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28772 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
28773 (extract_cond_operands): Detect condition comparing against non-
28774 invariant bound and return appropriate enum value.
28775 (find_interesting_uses_cond): Update use of extract_cond_operands.
28776 Handle its return value accordingly.
28777 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
28779 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28781 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
28782 nonlinear iv_use computation in loop invariant sensitive way.
28784 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28786 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
28787 (find_iv_candidates): Call relate_compare_use_with_all_cands.
28789 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28791 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
28792 (dump_cand): Support iv_cand.inv_exprs.
28793 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
28795 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
28798 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28800 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
28802 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
28803 as local function. Include necessary header files.
28804 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
28806 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28808 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
28810 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28812 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
28813 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
28814 RSHIFT_EXPR and BIT_NOT_EXPR.
28816 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28818 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
28819 (adjust_setup_cost): New parameter supporting round up adjustment.
28820 (struct address_cost_data): Delete.
28821 (force_expr_to_var_cost): Don't bound cost with spill_cost.
28822 (split_address_cost, ptr_difference_cost): Delete.
28823 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
28824 (struct ainc_cost_data): New struct.
28825 (get_address_cost_ainc): New function.
28826 (get_address_cost, get_computation_cost): Reimplement.
28827 (determine_group_iv_cost_address): Record inv_expr for all uses of
28829 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
28830 (iv_ca_has_deps): Reimplemented to ...
28831 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
28833 (iv_ca_extend): Call iv_ca_more_deps.
28835 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28837 * tree-ssa-address.c (struct mem_address): Move to header file.
28838 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
28839 * tree-ssa-address.h (struct mem_address): Move from C file.
28840 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
28842 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28844 * tree-affine.h (aff_combination_type): New interface.
28845 (aff_combination_zero_p): Remove static.
28846 (aff_combination_const_p): New interface.
28847 (aff_combination_singleton_var_p): New interfaces.
28849 2017-05-11 Richard Biener <rguenther@suse.de>
28851 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
28852 Skip unreachable blocks and destinations.
28853 (eliminate): Move stmt removal and fixup ...
28854 (fini_eliminate): ... here. Skip inserted exprs.
28855 (pass_pre::execute): Move fini_pre after fini_eliminate.
28856 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
28857 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
28858 PRE to get rid of dead code that has invalid SSA form and
28859 split critical edges again.
28861 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28863 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28865 2017-05-11 Richard Biener <rguenther@suse.de>
28867 * passes.c (execute_function_todo): Verify loops if they are
28868 said to be up-to-date.
28869 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
28870 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
28872 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
28875 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
28876 handle calling assemble_external ourself.
28879 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
28880 modes with zero size. Enhance comment.
28882 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28884 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
28885 built-ins for vec_xl and vec_xst with short and char pointer
28888 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
28890 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
28891 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
28892 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
28893 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
28894 (_mm_maskz_min_round_ss): New intrinsics.
28895 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
28896 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
28897 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
28898 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
28899 (__builtin_ia32_minss_mask_round): New builtins.
28900 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28901 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28902 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
28904 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
28905 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
28907 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
28910 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
28912 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
28913 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
28914 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
28915 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
28916 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
28917 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
28918 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
28919 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28920 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
28921 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
28922 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
28923 (__builtin_ia32_mulss_mask_round): New builtins.
28924 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28925 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28926 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
28928 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
28929 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
28931 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
28934 2017-05-10 Julia Koval <julia.koval@intel.com>
28936 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
28937 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
28938 (_mm256_setr_m128i): New intrinsics.
28940 2017-05-10 Julia Koval <julia.koval@intel.com>
28942 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
28943 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
28944 (_mm_maskz_rcp14_ss): New intrinsics.
28945 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
28946 (__builtin_ia32_rcp14ss_mask): New builtins.
28947 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
28949 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
28951 PR tree-optimization/51513
28952 * tree-cfg.c (gimple_seq_unreachable_p): New function.
28953 (assert_unreachable_fallthru_edge_p): Use it.
28954 (group_case_labels_stmt): Likewise.
28955 * tree-cfg.h: Prototype it.
28956 * stmt.c: Include cfghooks.h and tree-cfg.h.
28957 (emit_case_dispatch_table) <gap_label>: New local variable.
28958 Use it to fill dispatch table gaps.
28959 Test for default_label before updating probabilities.
28960 (expand_case) <default_label>: Remove unneeded initialization.
28961 Test for unreachable default case statement and remove its edge.
28962 Set default_label accordingly.
28963 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
28965 2017-05-10 Carl Love <cel@us.ibm.com>
28967 * config/rs6000/rs6000-c: Add support for built-in functions
28968 vector signed char vec_neg (vector signed char)
28969 vector signed short int vec_neg (vector short int)
28970 vector signed int vec_neg (vector signed int)
28971 vector signed long long vec_neg (vector signed long long)
28972 vector float vec_neg (vector float)
28973 vector double vec_neg (vector double)
28974 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
28976 * config/rs6000/altivec.h: Add define for vec_neg
28977 * doc/extend.texi: Update the built-in documentation for the
28978 new built-in functions.
28980 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28982 PR tree-optimization/77644
28983 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
28985 2017-05-10 Nathan Sidwell <nathan@acm.org>
28987 * dumpfile.h (TDI_lang_all): New.
28988 (TDF_KIND): New. Renumber others
28989 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
28991 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
28993 (get_dump_file_name): Adjust suffix generation.
28994 (dump_enable_all): Use TDF_KIND.
28995 * doc/invoke.texi (-fdump-lang-all): Document.
28997 * dumpfile.h: Tabify.
28999 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
29002 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
29003 Move member access before delete.
29005 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
29007 * tree-inline.c (expand_call_inline): Split block at stmt
29010 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
29013 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
29014 are now unused after splitting mov{sf,sd}_hardfloat.
29015 (f32_lr2): Likewise.
29016 (f32_lm): Likewise.
29017 (f32_lm2): Likewise.
29018 (f32_li): Likewise.
29019 (f32_li2): Likewise.
29020 (f32_lv): Likewise.
29021 (f32_sr): Likewise.
29022 (f32_sr2): Likewise.
29023 (f32_sm): Likewise.
29024 (f32_sm2): Likewise.
29025 (f32_si): Likewise.
29026 (f32_si2): Likewise.
29027 (f32_sv): Likewise.
29028 (f32_dm): Likewise.
29029 (f32_vsx): Likewise.
29030 (f32_av): Likewise.
29031 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
29032 For movsf, order stores so the VSX stores occur before the GPR
29033 store which encourages the register allocator to use a traditional
29034 FPR instead of a GPR. For movsd, order the stores so that the GPR
29035 store comes before the VSX stores to allow the power6 to work.
29036 This is due to the power6 not having a 32-bit integer store
29037 instruction from a FPR.
29038 (movsf_hardfloat): Likewise.
29039 (movsd_hardfloat): Likewise.
29041 2017-05-09 Martin Sebor <msebor@redhat.com>
29043 PR translation/80280
29044 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
29047 PR translation/80280
29048 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
29049 data member added in r247778.
29050 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
29052 2017-05-09 Nathan Sidwell <nathan@acm.org>
29054 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
29056 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
29059 2017-05-09 Marek Polacek <polacek@redhat.com>
29061 * doc/invoke.texi: Fix typo.
29063 2017-05-09 Richard Biener <rguenther@suse.de>
29065 * tree-vrp.c (vrp_val_is_max): Adjust comment.
29066 (vrp_val_is_min): Likewise.
29067 (set_value_range_to_value): Likewise.
29068 (set_value_range_to_nonnegative): Likewise.
29069 (gimple_assign_nonzero_p): Likewise.
29070 (gimple_stmt_nonzero_p): Likewise.
29071 (vrp_int_const_binop): Likewise. Remove unreachable case.
29072 (adjust_range_with_scev): Adjust comments.
29073 (compare_range_with_value): Likewise.
29074 (extract_range_from_phi_node): Likewise.
29075 (test_for_singularity): Likewise.
29077 2017-05-09 Richard Biener <rguenther@suse.de>
29079 * tree-vrp.c (get_single_symbol): Add assert that we don't
29080 get overflowed constants as invariant part.
29081 (compare_values_warnv): Add comment before the TREE_NO_WARNING
29082 checks. Use wi::cmp instead of recursing for integer constants.
29083 (compare_values): Just ignore whether we assumed undefined
29084 overflow instead of failing the compare.
29085 (extract_range_for_var_from_comparison_expr): Add comment before the
29086 TREE_NO_WARNING sets.
29087 (test_for_singularity): Likewise.
29088 (extract_range_from_comparison): Do not disable optimization
29089 when we assumed undefined overflow.
29090 (extract_range_basic): Remove init of unused var.
29092 2017-05-09 Richard Biener <rguenther@suse.de>
29094 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
29095 (extract_range_from_multiplicative_op_1): Adjust.
29096 (extract_range_from_binary_expr_1): Use int_const_binop.
29098 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
29101 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
29102 rs6000_store_data_bypass_p in seven define_bypass directives and
29103 in several comments.
29104 * config/rs6000/rs6000-protos.h: Add prototype for
29105 rs6000_store_data_bypass_p function.
29106 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
29107 function implements slightly different (rs6000-specific) semantics
29108 than store_data_bypass_p, returning false rather than aborting
29109 with assertion error when arguments do not satisfy the
29110 requirements of store data bypass.
29111 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
29112 rs6000_store_data_bypass_p.
29114 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
29116 * config/xtensa/xtensa-protos.h
29117 (xtensa_initial_elimination_offset): New declaration.
29118 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
29119 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
29120 macro definition, add case for FRAME_POINTER_REGNUM when
29121 FRAME_GROWS_DOWNWARD.
29122 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
29123 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
29124 xtensa_initial_elimination_offset.
29126 2017-05-08 Nathan Sidwell <nathan@acm.org>
29128 * doc/invoke.texi: Alphabetize -fdump options.
29130 2017-05-08 Martin Sebor <msebor@redhat.com>
29132 PR translation/80280
29133 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
29135 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
29137 * target.def (compute_frame_layout): New optional target hook.
29138 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
29139 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
29140 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
29142 * reload1.c (verify_initial_elim_offsets): Likewise.
29143 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
29144 (use_simple_return_p): Call arm_compute_frame_layout if needed.
29145 (arm_get_frame_offsets): Split up into this ...
29146 (arm_compute_frame_layout): ... and this function.
29148 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
29150 * config/aarch64/constraints.md (Usa): New constraint.
29151 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
29153 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
29155 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
29156 with_multilib_list after it has been checked.
29158 2017-05-08 Richard Biener <rguenther@suse.de>
29160 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
29161 (bitmap_set_subtract_values): Likewise.
29163 2017-05-08 Richard Biener <rguenther@suse.de>
29165 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
29166 (gimple_assign_nonzero): ... this and remove strict_overflow_p
29168 (gimple_stmt_nonzero_warnv_p): Rename to ...
29169 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
29171 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
29172 (extract_range_basic): Adjust, do not disable propagation on
29173 strict overflow sensitive simplification.
29174 (vrp_visit_cond_stmt): Likewise.
29176 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
29178 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
29179 body size unconditionally.
29181 2017-05-07 Jeff Law <law@redhat.com>
29184 2017-05-06 Jeff Law <law@redhat.com>
29185 PR tree-optimization/78496
29186 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
29189 PR tree-optimization/78496
29190 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
29191 (simplify_stmt_using_ranges): Call it.
29192 (vrp_dom_walker::before_dom_children): Extract equivalences
29193 from an ASSERT_EXPR with an equality comparison against a
29196 2017-05-06 Jeff Law <law@redhat.com>
29198 PR tree-optimization/78496
29199 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
29202 PR tree-optimization/78496
29203 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
29204 (simplify_stmt_using_ranges): Call it.
29205 (vrp_dom_walker::before_dom_children): Extract equivalences
29206 from an ASSERT_EXPR with an equality comparison against a
29209 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
29211 * lra-constraints.c (lra_copy_reg_equiv): New function.
29212 (split_reg): Use it to copy equivalence information from the
29213 original register to the spill register.
29215 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
29217 PR rtl-optimization/75964
29218 * simplify-rtx.c (simplify_const_relational_operation): Remove
29219 invalid handling of comparisons of integer ABS.
29221 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
29223 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
29224 initialize to zero.
29225 (init_regs): Remove declaration.
29226 (function_arg_advance_32): Initialize error_p as boolean variable.
29228 2017-05-05 Nathan Sidwell <nathan@acm.org>
29230 * store-motion.c (remove_reachable_equiv_notes): Reformat long
29231 lines. Use for (;;).
29233 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29235 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
29236 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
29237 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
29238 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
29239 VF=2 that require versioning.
29241 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29243 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
29246 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29248 * diagnostic.h (diagnostic_override_option_index): Convert from
29249 macro to inline function.
29251 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29253 * diagnostic.c (last_module_changed_p): New function.
29254 (set_last_module): New function.
29255 (diagnostic_report_current_module): Convert macro usage to
29256 the above functions.
29257 * diagnostic.h (diagnostic_context::last_module): Strengthen
29258 from const line_map * to const line_map_ordinary *.
29259 (diagnostic_last_module_changed): Delete macro.
29260 (diagnostic_set_last_module): Delete macro.
29262 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29264 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
29265 with diagnostic_report_diagnostic.
29266 (diagnostic_n_impl_richloc): Likewise.
29267 * diagnostic.h (report_diagnostic): Delete macro.
29268 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
29269 with diagnostic_report_diagnostic.
29270 * substring-locations.c (format_warning_va): Likewise.
29272 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29274 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
29275 save/restor of format_spec. Move option-printing code to...
29276 (print_option_information): ...this new function, and
29277 reimplement by simply printing to the pretty_printer,
29278 rather than appending to the format string.
29280 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29282 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
29283 handling logic into...
29284 (update_effective_level_from_pragmas): ...this new function.
29286 2017-05-04 Andrew Waterman <andrew@sifive.com>
29288 * config/riscv/riscv.opt (mstrict-align): New option.
29289 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
29290 (SLOW_UNALIGNED_ACCESS): Define.
29291 (riscv_slow_unaligned_access): Declare.
29292 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
29294 (riscv_slow_unaligned_access): New variable.
29295 (rocket_tune_info): Set slow_unaligned_access to true.
29296 (optimize_size_tune_info): Set slow_unaligned_access to false.
29297 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
29298 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
29299 (riscv_option_override): Set riscv_slow_unaligned_access.
29300 * doc/invoke.texi: Add -mstrict-align to RISC-V.
29302 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
29304 * config/riscv/riscv.md: Unify indentation.
29306 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29311 * config/rs6000/rs6000.md (u code attribute): Add FIX and
29313 (extendsi<mode>2): Add support for doing sign extension via
29314 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
29315 don't have ISA 3.0 instructions.
29316 (extendsi<mode>2 splitter): Likewise.
29317 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
29318 generate the normal insns since SImode can now go in vector
29319 registers. Disallow the special UNSPECs needed for previous
29320 machines to hide SImode being used. Add new insns
29321 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
29322 (fix_trunc<mode>si2_stfiwx): Likewise.
29323 (fix_trunc<mode>si2_internal): Likewise.
29324 (fixuns_trunc<mode>si2): Likewise.
29325 (fixuns_trunc<mode>si2_stfiwx): Likewise.
29326 (fctiw<u>z_<mode>_smallint): Likewise.
29327 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
29328 of floating point to 32-bit integer from doing a direct move to
29329 the GPR registers to do a store.
29330 (fctiwz_<mode>): Break long line.
29332 2017-05-05 Bin Cheng <bin.cheng@arm.com>
29334 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
29335 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
29336 (addr_list, addr_offset_valid_p): New.
29337 (split_address_groups): Check offset validity with above function.
29338 (gt-tree-ssa-loop-ivopts.h): Include header file.
29340 2017-05-05 Nathan Sidwell <nathan@acm.org>
29342 * config.gcc (arm*-*-*): Add missing 'fi'.
29344 2017-05-05 Steve Ellcey <sellcey@cavium.com>
29346 * doc/invoke.texi (-fopt-info): Explicitly say order of options
29347 included in -fopt-info does not matter.
29348 * doc/optinfo.texi (-fopt-info): Fix description of default
29349 behavour. Explicitly say order of options included in -fopt-info
29352 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29354 * config.gcc: Allow combinations of aprofile and rmprofile values for
29355 --with-multilib-list.
29356 * config/arm/t-multilib: New file.
29357 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
29358 variables. Remove setting of ISA and floating-point ABI in
29359 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
29360 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
29361 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
29362 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
29364 * config/arm/t-rmprofile: Likewise except for the matches changes.
29365 * doc/install.texi (--with-multilib-list): Document the combination of
29366 aprofile and rmprofile values and warn about pitfalls in doing that.
29368 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
29370 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
29371 (movdi_aarch64): Likewise.
29373 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29375 PR tree-optimization/80632
29376 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
29378 (build_arrays): Initialize it for virtual phis.
29379 (fix_phi_nodes): Use it for virtual phis.
29381 PR tree-optimization/80558
29382 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
29383 [x, y] op z into [x op, y op z] for op & or | if conditions
29386 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
29387 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29390 * config/arm/arm.md (use_literal_pool): Remove.
29391 (64-bit immediate split): No longer takes cost into consideration
29392 if arm_disable_literal_pool is enabled.
29393 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
29394 used when arm_disable_literal_pool is enabled.
29395 (arm_max_const_double_inline_cost): Remove use of
29396 arm_disable_literal_pool.
29397 (push_minipool_fix): Add assert.
29398 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
29399 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
29400 (no_literal_pool_sf_immediate): New.
29402 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29404 PR tree-optimization/80613
29405 * tree-ssa-dce.c (propagate_necessity): Remove cases for
29406 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
29408 2017-05-05 Richard Biener <rguenther@suse.de>
29410 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
29412 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
29414 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
29415 of this flag from insn conditions due to removal from r247495.
29417 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
29419 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
29421 (arm_early_store_addr_dep_ptr): Likewise.
29422 * config/arm/aarch-common-protos.h
29423 (arm_early_load_addr_dep_ptr): Add prototype.
29424 (arm_early_store_addr_dep_ptr): Likewise.
29425 * config/arm/cortex-a53.md: Add new bypasses.
29427 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29429 * tree.c (next_type_uid): Change type to unsigned.
29430 (type_hash_canon): Decrement back next_type_uid if
29431 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
29432 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
29435 2017-05-04 Martin Sebor <msebor@redhat.com>
29437 * builtins.c: Fix a trivial typo in a comment.
29439 PR middle-end/79234
29440 * builtins.c (check_sizes): Adjust to handle reading past the end.
29441 Avoid printing excessive upper bound of ranges. Use %E to print
29442 tree nodes instead of converting them to %wu.
29443 (expand_builtin_memchr): New function.
29444 (compute_dest_size): Rename...
29445 (compute_objsize): ...to this.
29446 (expand_builtin_memcpy): Adjust.
29447 (expand_builtin_mempcpy): Adjust.
29448 (expand_builtin_strcat): Adjust.
29449 (expand_builtin_strcpy): Adjust.
29450 (check_strncat_sizes): Adjust.
29451 (expand_builtin_strncat): Adjust.
29452 (expand_builtin_strncpy): Adjust and simplify.
29453 (expand_builtin_memset): Adjust.
29454 (expand_builtin_bzero): Adjust.
29455 (expand_builtin_memcmp): Adjust.
29456 (expand_builtin): Handle memcmp.
29457 (maybe_emit_chk_warning): Check strncat just once.
29459 2017-05-04 Martin Sebor <msebor@redhat.com>
29461 PR preprocessor/79214
29462 PR middle-end/79222
29463 PR middle-end/79223
29464 * builtins.c (check_sizes): Add inlining context and issue
29465 warnings even when -Wno-system-headers is set.
29466 (check_strncat_sizes): Same.
29467 (expand_builtin_strncat): Same.
29468 (expand_builtin_memmove): New function.
29469 (expand_builtin_stpncpy): Same.
29470 (expand_builtin): Handle memmove and stpncpy.
29472 2017-05-04 Bin Cheng <bin.cheng@arm.com>
29474 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
29475 which is not used any more.
29477 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29479 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
29481 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29483 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
29484 (cortexa53_tunings): Likewise.
29485 (cortexa57_tunings): Likewise.
29486 (cortexa72_tunings): Likewise.
29487 (cortexa73_tunings): Likewise.
29489 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29491 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
29492 Set loop alignment to 8.
29494 2017-05-04 Martin Sebor <msebor@redhat.com>
29496 PR translation/80280
29497 * builtins.c (expand_builtin_object_size): Add missing quoting to
29498 %D and like directives.
29499 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
29500 (hsa_type_for_tree_type): Same.
29501 (verify_function_arguments): Same.
29502 * symtab.c (symbol_table::change_decl_assembler_name): Same.
29503 * varasm.c (get_section): Same.
29506 2017-05-04 Martin Sebor <msebor@redhat.com>
29508 PR translation/80280
29509 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
29511 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29513 * config/aarch64/aarch64.c (generic_addrcost_table):
29514 Change HI/TI mode setting.
29516 2017-05-04 Martin Jambor <mjambor@suse.cz>
29518 PR tree-optimization/80622
29519 * tree-sra.c (comes_initialized_p): New function.
29520 (build_accesses_from_assign): Only set write lazily when
29521 comes_initialized_p is false.
29522 (analyze_access_subtree): Use comes_initialized_p.
29523 (propagate_subaccesses_across_link): Assert !comes_initialized_p
29524 instead of testing for PARM_DECL.
29526 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29528 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
29529 constraint on operand 0 to allow more general addressing modes.
29530 Adjust output template.
29531 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
29533 * config/aarch64/aarch64-protos.h
29534 (aarch64_address_valid_for_prefetch_p): Declare prototype.
29535 * config/aarch64/constraints.md (Dp): New address constraint.
29536 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
29539 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
29541 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
29542 update use of estimate_ipcp_clone_size_and_time.
29543 (estimate_local_effects): Update use of
29544 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
29545 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
29546 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
29547 Return nonspecialized time.
29549 2017-05-04 Richard Biener <rguenther@suse.de>
29551 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
29552 for the last VUSE which def dominates the PHI. Directly call
29554 (get_continuation_for_phi_1): Remove.
29556 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
29558 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
29559 to explain the use of truncating division. Cap the number of
29560 iterations to the maximum given by nb_iterations_upper_bound,
29563 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
29565 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
29566 * configure: Regenerate.
29567 * config.in: Regenerate.
29568 * config/i386/driver-mingw32.c: new file.
29569 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
29570 * config.host: Link driver-mingw32.o on MinGW host.
29571 * doc/install.texi: Document new --enable-mingw-wildcard configure
29574 2017-05-04 Marek Polacek <polacek@redhat.com>
29576 PR tree-optimization/80612
29577 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
29579 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29580 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
29582 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
29583 (movt splitter): Likewise.
29584 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
29585 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
29586 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
29587 block for Thumb-1 with MOVT.
29588 (thumb2_legitimate_address_p): Move code block ...
29589 (can_avoid_literal_pool_for_label_p): ... into this new function.
29590 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
29592 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
29593 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
29594 "M-profile targets with the MOVT instruction".
29596 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29598 * config/arm/arm-builtins.c (arm_init_builtins): Rename
29599 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
29600 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
29602 2017-05-04 Martin Liska <mliska@suse.cz>
29604 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
29605 variable cond_code.
29607 2017-05-04 Richard Biener <rguenther@suse.de>
29609 * tree.c (array_at_struct_end_p): Handle arrays at struct
29610 end with flexarrays more conservatively. Refactor and treat
29611 arrays of arrays or aggregates more strict. Fix
29612 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
29613 * tree.c (array_at_struct_end_p): Adjust prototype.
29614 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
29615 * gimple-fold.c (get_range_strlen): Likewise.
29616 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
29618 2017-05-04 Richard Biener <rguenther@suse.de>
29620 PR tree-optimization/31130
29621 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
29623 (supports_overflow_infinity): Likewise.
29624 (is_negative_overflow_infinity): Likewise.
29625 (is_positive_overflow_infinity): Likewise.
29626 (is_overflow_infinity): Likewise.
29627 (stmt_overflow_infinity): Likewise.
29628 (overflow_infinity_range_p): Likewise.
29629 (usable_range_p): Remove as always returning true.
29630 (make_overflow_infinity): Remove.
29631 (negative_overflow_infinity): Likewise.
29632 (positive_overflow_infinity): Likewise.
29633 (avoid_overflow_infinity): Likewise.
29634 (set_value_range): Adjust accordingly.
29635 (set_value_range_to_nonnegative): Likewise, remove now unused
29636 overflow_infinity arg.
29637 (vrp_operand_equal_p): Adjust.
29638 (update_value_range): Likewise.
29639 (range_int_cst_singleton_p): Likewise.
29640 (operand_less_p): Likewise.
29641 (compare_values_warnv): Likewise.
29642 (extract_range_for_var_from_comparison_expr): Likewise.
29643 (vrp_int_const_binop): Likewise.
29644 (zero_nonzero_bits_from_vr): Likewise.
29645 (extract_range_from_multiplicative_op_1): Likewise.
29646 (extract_range_from_binary_expr_1): Likewise.
29647 (extract_range_from_unary_expr): Likewise.
29648 (extract_range_from_comparison): Likewise.
29649 (extract_range_basic): Likewise.
29650 (adjust_range_with_scev): Likewise.
29651 (compare_ranges): Likewise.
29652 (compare_range_with_value): Likewise.
29653 (dump_value_range): Likewise.
29654 (test_for_singularity): Likewise, remove strict_overflow_p parameter
29656 (simplify_cond_using_ranges): Adjust.
29658 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
29660 * brig-builtins.def: Added a builtin for class_f64.
29661 * builtin-types.def: Added a builtin type needed by class_f64.
29663 2017-05-03 Jason Merrill <jason@redhat.com>
29665 * timevar.def: Add TV_CONSTEXPR.
29667 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29669 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
29671 2017-05-03 Martin Jambor <mjambor@suse.cz>
29673 * ipa-prop.c (ipa_update_after_lto_read): Removed.
29674 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
29675 * ipa-cp.c (ipcp_propagate_stage): Do not call
29676 ipa_update_after_lto_read.
29677 * ipa-inline.c (ipa_inline): Likewise.
29679 2017-05-03 Martin Jambor <mjambor@suse.cz>
29681 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
29682 tag. Added a default constructor and a destructor.
29683 (ipa_edge_args_sum_t): New class;
29684 (ipa_edge_args_sum): Declare.
29685 (ipa_edge_args_vector): Remove declaration.
29686 (IPA_EDGE_REF): Use ipa_edge_args_sum.
29687 (ipa_free_edge_args_substructures): Remove declaration.
29688 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
29689 (ipa_edge_args_info_available_for_edge_p): Likewise.
29690 * ipa-prop.c (ipa_edge_args_vector): Removed.
29691 (edge_removal_hook_holder): Likewise.
29692 (edge_duplication_hook_holder): Likewise.
29693 (ipa_edge_args_sum): New variable.
29694 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
29695 ipa_edge_args_vector.
29696 (ipa_free_edge_args_substructures): Likewise.
29697 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
29698 ipa_edge_args_vector.
29699 (ipa_edge_removal_hook): Turned into method
29700 ipa_edge_args_sum_t::remove.
29701 (ipa_edge_duplication_hook): Turned into method
29702 ipa_edge_args_sum_t::duplicate.
29703 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
29704 registering edge hooks.
29705 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
29706 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
29707 ipa_edge_args_sum instead of ipa_edge_args_vector.
29708 * ipa-profile.c (ipa_profile): Likewise.
29710 2017-05-03 Martin Jambor <mjambor@suse.cz>
29712 * symbol-summary.h (function_summary): New method exists.
29713 (function_summary::symtab_removal): Deallocate through release.
29714 (call_summary): New class.
29715 (gt_ggc_mx): New overload.
29716 (gt_pch_nx): Likewise.
29717 (gt_pch_nx): Likewise.
29719 2017-05-03 Jeff Law <law@redhat.com>
29721 PR tree-optimization/78496
29722 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
29723 from simplify_cond_using_ranges. Split off code to walk
29724 backwards through casts into ...
29725 (simplify_cond_using_ranges_2): New function.
29726 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
29727 (execute_vrp): After identifying jump threads, call
29728 simplify_cond_using_ranges_2.
29730 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
29733 * ipa-inline.h (inline_summary): Add ctor.
29734 (create_ggc): Do not use ggc_cleared_alloc.
29736 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
29737 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29739 * gcc.c (handle_braces): Support escaping in switch matching
29741 * doc/invoke.texi (Spec Files): Document it.
29742 Remove superfluous @code markup in items.
29744 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29746 * diagnostic-show-locus.c (struct column_range): New struct.
29747 (get_affected_columns): New function.
29748 (get_printed_columns): New function.
29749 (struct correction): New struct.
29750 (correction::ensure_capacity): New function.
29751 (correction::ensure_terminated): New function.
29752 (struct line_corrections): New struct.
29753 (line_corrections::~line_corrections): New dtor.
29754 (line_corrections::add_hint): New function.
29755 (layout::print_trailing_fixits): Reimplement in terms of the new
29757 (selftest::test_overlapped_fixit_printing): New function.
29758 (selftest::diagnostic_show_locus_c_tests): Call it.
29760 2017-05-03 Nathan Sidwell <nathan@acm.org>
29762 Canonicalize canonical type hashing
29763 * tree.h (type_hash_canon_hash): Declare.
29764 * tree.c (type_hash_list, attribute_hash_list): Move into
29765 type_hash_canon_hash.
29766 (build_type_attribute_qual_variant): Break out hash code calc into
29767 type_hash_canon_hash.
29768 (type_hash_canon_hash): New. Generic type hash computation.
29769 (build_range_type_1, build_array_type_1, build_function_type,
29770 build_method_type_directly, build_offset_type, build_complex_type,
29771 make_vector_type): Call it.
29773 2017-05-03 Richard Biener <rguenther@suse.de>
29775 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
29776 When all DRs have unknown misaligned do not always peel
29777 when there is a store but apply the same costing model as if
29778 there were only loads.
29780 2017-05-03 Richard Biener <rguenther@suse.de>
29783 PR tree-optimization/80492
29784 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
29785 compare_base_decls returning dont-know properly.
29787 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
29789 * config/arm/iterators.md (CCSI): New mode iterator.
29790 (arch): New mode attribute.
29791 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
29792 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
29793 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
29794 code iterator for success result mode.
29795 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
29796 the corresponding new insn generators.
29798 2017-05-03 Bin Cheng <bin.cheng@arm.com>
29801 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29802 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29804 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
29806 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
29807 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
29808 (DDR_A): Wrap DDR argument in brackets.
29809 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
29810 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
29811 (DDR_REVERSED_P): Likewise.
29813 2017-05-03 Jakub Jelinek <jakub@redhat.com>
29815 PR tree-optimization/79472
29816 * tree-switch-conversion.c (struct switch_conv_info): Add
29817 contiguous_range and default_case_nonstandard fields.
29818 (collect_switch_conv_info): Compute contiguous_range and
29819 default_case_nonstandard fields, don't clear final_bb if
29820 contiguous_range and only the default case doesn't have the required
29822 (check_all_empty_except_final): Set default_case_nonstandard instead
29823 of failing if contiguous_range and the default case doesn't have empty
29825 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
29826 and only the default case doesn't have the required constants. Skip
29828 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
29829 if default_case_nonstandard.
29830 (build_constructors): Build constant 1 just once. Assert that default
29831 values aren't inserted in between cases if contiguous_range. Skip
29833 (build_arrays): Skip virtual phis.
29834 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
29835 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
29836 Handle virtual phis.
29837 (gen_inbound_check): Handle default_case_nonstandard case.
29838 (process_switch): Adjust check_final_bb caller. Call
29839 gather_default_values with the first non-default case instead of
29840 default case if default_case_nonstandard.
29842 2017-05-02 Nathan Sidwell <nathan@acm.org>
29844 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
29845 check. Fix formatting.
29847 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
29849 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
29850 errors when comparing specialized and unspecialized times.
29852 2017-05-02 David Malcolm <dmalcolm@redhat.com>
29854 * diagnostic-show-locus.c
29855 (layout::should_print_annotation_line_p): Make private.
29856 (layout::print_annotation_line): Make private.
29857 (layout::annotation_line_showed_range_p): Make private.
29858 (layout::show_ruler): Make private.
29859 (layout::print_source_line): Make private. Pass in line and
29860 line_width, rather than calling location_get_source_line. Drop
29862 (layout::print_leading_fixits): New method.
29863 (layout::print_any_fixits): Rename to...
29864 (layout::print_trailing_fixits): ...this, and make private.
29865 Don't print newline fixits.
29866 (diagnostic_show_locus): Move logic for printing one row into...
29867 (layout::print_line): ...this new function. Move the
29868 location_get_source_line call and error-handling from
29869 print_source_line to here. Call print_leading_fixits, and rename
29870 print_any_fixits to print_trailing_fixits.
29871 (selftest::test_fixit_insert_containing_newline): Update now that
29872 newlines are partially supported.
29873 (selftest::test_fixit_insert_containing_newline_2): New test.
29874 (selftest::test_fixit_replace_containing_newline): Update comments.
29875 (selftest::diagnostic_show_locus_c_tests): Call the new test.
29876 * edit-context.c (class added_line): New class.
29877 (class edited_line): Describe newline handling in comment.
29878 (edited_line::actually_edited_p): New method.
29879 (edited_line::print_content): Delete redundant decl.
29880 (edited_line::m_predecessors): New field.
29881 (edited_file::print_content): Call edited_line::print_content.
29882 (edited_file::print_diff): Update to support newlines.
29883 (edited_file::print_diff_hunk): Likewise.
29884 (edited_file::print_run_of_changed_lines): New function.
29885 (edited_file::print_diff_line): Convert to...
29886 (print_diff_line): ...this.
29887 (edited_file::get_effective_line_count): New function.
29888 (edited_line::edited_line): Initialize new field m_predecessors.
29889 (edited_line::~edited_line): Clean up m_predecessors.
29890 (edited_line::apply_fixit): Handle newlines.
29891 (edited_line::get_effective_line_count): New function.
29892 (edited_line::print_content): New function.
29893 (edited_line::print_diff_lines): New function.
29894 (selftest::test_applying_fixits_insert_containing_newline): New
29896 (selftest::test_applying_fixits_replace_containing_newline): New
29898 (selftest::insert_line): New function.
29899 (selftest::test_applying_fixits_multiple_lines): Add example of
29901 (selftest::edit_context_c_tests): Call the new tests.
29903 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29905 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
29906 parameter cand. Update dump information.
29907 (get_computation_cost): Update uses.
29909 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29911 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
29912 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
29913 (get_computation_at, rewrite_use_address): Update use of
29914 get_computation_aff.
29916 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29918 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
29919 (get_computation): Delete.
29920 (get_computation_cost): Implement like get_computation_cost_at.
29921 Use get_computation_at.
29922 (get_computation_cost_at): Delete.
29923 (rewrite_use_nonlinear_expr): Use get_computation_at.
29924 (rewrite_use_compare, remove_unused_ivs): Ditto.
29926 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29928 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
29930 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29932 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
29933 (ivopts_global_cost_for_size): Rename parameter and update uses.
29934 (iv_ca_recount_cost): Update uses.
29935 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
29936 candidates seperately in n_invs and n_cands.
29937 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
29939 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29941 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
29942 (find_inv_vars_cb): New.
29943 (find_depends): Renamed to ...
29944 (find_inv_vars): ... this.
29945 (add_candidate_1, force_var_cost): Call find_inv_vars.
29946 (split_address_cost, determine_group_iv_cost_cond): Ditto.
29948 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29950 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
29951 inv_vars. Add inv_exprs.
29952 (struct iv_cand): Rename depends_on to inv_vars.
29953 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
29954 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
29955 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
29956 (dump_cand): Dump inv_vars.
29957 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
29958 (record_invariant, find_depends, add_candidate_1): Ditto.
29959 (set_group_iv_cost, force_var_cost): Ditto.
29960 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
29961 (get_computation_cost_at, get_computation_cost): Ditto.
29962 (determine_group_iv_cost_generic): Ditto.
29963 (determine_group_iv_cost_address): Ditto.
29964 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
29965 (determine_group_iv_costs): Ditto.
29966 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
29967 (iv_ca_set_remove_invariants): Renamed to ...
29968 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
29969 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
29970 (iv_ca_set_add_invariants): Renamed to ...
29971 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
29972 (iv_ca_set_cp): Use iv_ca_set_add_invs.
29973 (iv_ca_has_deps): Support inv_vars and inv_exprs.
29974 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
29975 (create_new_ivs): Remove useless dump.
29977 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29979 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
29981 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
29982 (iv_ca_set_no_cp, create_new_iv): Ditto.
29984 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29986 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29988 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
29990 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
29991 function tree_check2.
29993 2017-05-02 Martin Liska <mliska@suse.cz>
29995 * doc/gcov.texi: Add missing preposition.
29996 * gcov.c (function_info::function_info): Properly fill up
29997 all member variables.
29999 2017-05-02 Tamar Christina <tamar.christina@arm.com>
30001 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
30003 2017-05-02 Tamar Christina <tamar.christina@arm.com>
30005 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
30007 2017-05-02 Martin Liska <mliska@suse.cz>
30010 * lto-streamer-in.c (lto_read_tree_1): Remove
30011 LTO_STREAMER_DEBUG.
30012 * lto-streamer.c (struct tree_hash_entry): Likewise.
30013 (struct tree_entry_hasher): Likewise.
30014 (tree_entry_hasher::hash): Likewise.
30015 (tree_entry_hasher::equal): Likewise.
30016 (lto_streamer_init): Likewise.
30017 (lto_orig_address_map): Likewise.
30018 (lto_orig_address_get): Likewise.
30019 (lto_orig_address_remove): Likewise.
30020 * lto-streamer.h: Likewise.
30021 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
30022 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
30024 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
30026 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
30027 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
30028 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
30029 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
30030 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
30031 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
30032 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
30033 (mm_maskz_sub_ss): New intrinsics.
30034 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30035 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
30036 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
30037 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
30038 (__builtin_ia32_subss_mask_round): New builtins.
30039 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30040 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30041 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
30043 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
30044 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
30046 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
30049 2017-05-02 Martin Jambor <mjambor@suse.cz>
30051 PR tree-optimization/78687
30052 * tree-sra.c (access): New field parent.
30053 (process_subtree_disqualification): New function.
30054 (disqualify_candidate): Call it.
30055 (build_accesses_from_assign): Reset write flag if creating an
30057 (build_access_subtree): Fill in parent field and also prpagate
30058 down grp_write flag.
30059 (create_artificial_child_access): New parameter set_grp_write, set
30060 grp_write to its value.
30061 (propagate_subaccesses_across_link): Also propagate grp_write flag
30063 (propagate_all_subaccesses): Push the closest parent back to work
30064 queue if add_access_to_work_queue returned true.
30066 2017-05-02 Richard Biener <rguenther@suse.de>
30068 * common.opt (fstrict-overflow): Alias negative to fwrapv.
30069 * doc/invoke.texi (fstrict-overflow): Remove all traces of
30070 -fstrict-overflow documentation.
30071 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
30072 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
30073 flag_strict_overflow.
30074 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
30075 * lto-opts.c (lto_write_options): Do not stream it.
30076 * lto-wrapper.c (merge_and_complain): Do not handle it.
30077 * opts.c (default_options_table): Do not set -fstrict-overflow.
30078 (finish_options): Likewise do not clear it when sanitizing.
30079 * simplify-rtx.c (simplify_const_relational_operation): Do not
30080 test flag_strict_overflow.
30082 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
30084 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
30085 using enabled attribute.
30086 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
30087 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
30088 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
30089 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
30090 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
30091 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
30092 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
30093 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
30094 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
30095 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
30097 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
30099 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
30101 2017-05-02 Richard Biener <rguenther@suse.de>
30103 PR tree-optimization/80591
30105 2017-04-10 Richard Biener <rguenther@suse.de>
30107 * tree-ssa-structalias.c (find_func_aliases): Properly handle
30110 2017-05-02 Richard Biener <rguenther@suse.de>
30112 PR tree-optimization/80549
30113 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
30114 (cleanup_tree_cfg_noloop): Create forwarders to known loop
30115 headers if they do not have a preheader.
30117 2017-05-02 Martin Liska <mliska@suse.cz>
30120 * common.opt: Fix typo.
30121 * doc/invoke.texi: Likewise.
30123 2017-05-01 Jan Beulich <jbeulich@suse.com>
30125 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
30126 swapping, add (x,x,m,x,n) alternative.
30128 2017-05-01 Nathan Sidwell <nathan@acm.org>
30130 * calls.c (combine_pending_stack_adjustment_and_call): Remove
30131 unnecessary unadjusted_alignment check.
30133 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
30136 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
30138 * gimplify.c (gimplify_cilk_detach): New function.
30139 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
30140 * tree-core.h: Document EXPR_CILK_SPAWN.
30141 * tree.h (EXPR_CILK_SPAWN): Define.
30143 2017-05-01 David Malcolm <dmalcolm@redhat.com>
30145 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
30146 to use new fixit_hint representation, using the "replace" logic.
30147 (get_line_span_for_fixit_hint): Likewise.
30148 (layout::print_any_fixits): Likewise.
30149 (selftest::test_one_liner_many_fixits): Rename to...
30150 (selftest::test_one_liner_many_fixits_1): ...this, and update
30151 comment and expected output to reflect that the multiple fix-it
30152 hints are now consolidated into one insertion.
30153 (selftest::test_one_liner_many_fixits_2): New test.
30154 (selftest::test_diagnostic_show_locus_one_liner): Update for
30156 (selftest::test_fixit_consolidation): Update for fix-it API
30158 * diagnostic.c (print_parseable_fixits): Likewise.
30159 * edit-context.c (edited_line::m_line_events): Convert from
30160 auto_vec <line_event *> to auto_vec <line_event>.
30161 (class line_event): Convert from abstract base class to a concrete
30162 class, taking over the role of replace_event.
30163 (class insert_event): Delete.
30164 (class replace_event): Rename to class line_event. Convert to
30166 (edit_context::add_fixits): Reimplement.
30167 (edit_context::apply_insert): Delete.
30168 (edit_context::apply_replace): Rename to...
30169 (edit_context::apply_fixit): ...this. Convert to half-open range.
30170 (edited_file::apply_insert): Delete.
30171 (edited_file::apply_replace): Rename to...
30172 (edited_file::apply_fixit): ...this.
30173 (edited_line::~edited_line): Drop deletion of events.
30174 (edited_line::apply_insert): Delete.
30175 (edited_line::apply_replace): Rename to...
30176 (edited_line::apply_fixit): ...this. Convert to half-open range.
30177 Update for change to type of m_line_events.
30178 * edit-context.h (edit_context::apply_insert): Delete.
30179 (edit_context::apply_replace): Rename to...
30180 (edit_context::apply_fixit): ...this.
30182 2017-05-01 Martin Sebor <msebor@redhat.com>
30184 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
30187 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
30190 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
30191 __get_cpuid_max returns 0.
30192 (__get_cpuid_count): Ditto.
30194 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
30196 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
30197 replacement expression is another instance of one of its arguments.
30199 2017-05-01 Jakub Jelinek <jakub@redhat.com>
30202 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
30203 check for stack push/pop autoinc.
30204 * config/i386/i386.c (ix86_agi_dependent): Return false
30205 if the only reason why modified_in_p returned true is that
30206 addr is SP based and set_insn is a push or pop.
30208 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
30210 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
30213 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
30216 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
30217 (account_size_time): Use two predicates - exec_pred and
30219 (evaluate_conditions_for_known_args): Compute both clause and
30221 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
30222 (inline_summary_t::duplicate): Update.
30223 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
30225 (compute_inline_parameters): Likewise.
30226 (estimate_edge_size_and_time): Update caluclation of time.
30227 (estimate_node_size_and_time): Compute both time and nonspecialized
30229 (estimate_ipcp_clone_size_and_time): Update.
30230 (inline_merge_summary): Update.
30231 (do_estimate_edge_time): Update.
30232 (do_estimate_edge_size): Update.
30233 (do_estimate_edge_hints): Update.
30234 (inline_read_section, inline_write_summary): Stream both new predicates.
30235 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
30237 (compute_inlined_call_time): Cleanup.
30238 (big_speedup_p): Update.
30239 (edge_badness): Update.
30240 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
30241 (size_time_entry): Replace predicate by exec_predicate and
30242 nonconst_predicate.
30243 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
30244 (estimate_edge_time): Return also nonspec_time.
30245 (reset_edge_growth_cache): Update.
30247 2017-04-29 Jakub Jelinek <jakub@redhat.com>
30249 PR rtl-optimization/80491
30250 * ifcvt.c (noce_process_if_block): When looking for x setter
30251 with missing else_bb, don't check only the insn right before
30252 cond_earliest, but look for the last insn that x is modified in
30253 within the same bb.
30255 PR rtl-optimization/80491
30256 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
30258 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
30260 PR tree-optimization/80487
30261 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
30263 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30265 PR tree-optimization/79697
30266 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
30267 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
30268 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
30270 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
30271 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
30273 2017-04-28 Martin Sebor <msebor@redhat.com>
30275 PR tree-optimization/80523
30276 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
30277 (init_target_to_host_charmap, target_to_host, target_strtol10): New
30279 (maybe_warn, format_directive, parse_directive): Use new functions.
30280 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
30282 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
30284 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
30286 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
30288 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
30289 target_header_dir): Set correctly.
30290 * configure: Regenerated.
30291 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30292 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30293 instead of SYSTEM_HEADER_DIR.
30295 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
30297 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
30298 (estimate_local_effects): Likewise.
30299 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
30300 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
30301 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
30302 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
30303 do_estimate_edge_time, estimate_edge_time): Likewise.
30304 * ipa-inline-analysis.c (estimate_node_size_and_time,
30305 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
30306 (estimate_time_after_inlining): Remove.
30308 2017-04-28 Martin Liska <mliska@suse.cz>
30310 * doc/gcov.texi: Enhance documentation of gcov.
30312 2017-04-28 Martin Liska <mliska@suse.cz>
30314 * doc/gcov.texi: Sort options in alphabetic order.
30315 * doc/gcov-dump.texi: Likewise.
30316 * doc/gcov-tool.texi: Likewise.
30317 * gcov.c (print_usage): Likewise.
30318 * gcov-dump.c (print_usage): Likewise.
30319 * gcov-tool.c (print_merge_usage_message): Likewise.
30320 (print_rewrite_usage_message): Likewise.
30321 (print_overlap_usage_message): Likewise.
30323 2017-04-28 Martin Liska <mliska@suse.cz>
30325 PR gcov-profile/53915
30326 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
30328 2017-04-28 Martin Liska <mliska@suse.cz>
30330 PR gcov-profile/79891
30331 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
30332 is marked by compiler as living on a line.
30333 (get_cycles_count): Remove usage of the union.
30334 (output_intermediate_file): Likewise.
30335 (find_source): Fix GNU coding style.
30336 (accumulate_line_counts): Remove old non-all block mode.
30337 (output_lines): Remove usage of the union.
30338 * profile.c (output_location): Include all BBs, even if
30339 belonging to a same line (and file) as a previous BB.
30341 2017-04-28 Martin Liska <mliska@suse.cz>
30343 * gcov.c (process_args): Handle new argument 'w'.
30344 (read_graph_file): Assign ID to BBs.
30345 (output_branch_count): Display BB # if verbose flag is set.
30346 (output_lines): Likewise for arcs.
30347 (print_usage): Add '--verbose' option help.
30348 * doc/gcov.texi: Document --verbose (-w) option.
30350 2017-04-28 Martin Liska <mliska@suse.cz>
30352 * gcov.c (struct block_location_info): New struct.
30353 (process_file): Fill up the new structure.
30354 (read_graph_file): Replace usage of encoding by the newly added
30356 (add_line_counts): Likewise.
30357 (accumulate_line_counts): Remove usage of the union.
30358 (function_info::function_info): New function.
30359 (function_info::~function_info): Likewise.
30360 (process_file): Call delete instead of release_function.
30361 (release_function): Release the function.
30362 (release_structures): Call delete instead of release_function.
30363 (solve_flow_graph): Replace usage of num_blocks.
30364 (find_exception_blocks): Likewise.
30365 (output_lines): Fix GNU coding style.
30367 2017-04-28 Martin Liska <mliska@suse.cz>
30370 * coverage.c (coverage_remove_note_file): New function.
30371 * coverage.h: Declare the function.
30372 * toplev.c (finalize): Clean if an error has been seen.
30374 2017-04-28 Martin Liska <mliska@suse.cz>
30376 PR gcov-profile/80031
30377 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
30378 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
30379 * gcov.c (read_graph_file): Read just number of blocks.
30380 * profile.c (branch_prob): Do not stream 0 flags per a basic
30383 2017-04-28 Martin Liska <mliska@suse.cz>
30385 * gcov-dump.c (tag_*): Add new argument to declarations.
30386 (dump_gcov_file): Likewise.
30387 (tag_blocks): Add and use new argument depth.
30388 (tag_arcs): Likewise.
30389 (tag_lines): Likewise.
30390 (tag_counters): Likewise.
30391 (tag_summary): Likewise.
30392 (dump_working_sets): Use depth to do a proper indentation.
30394 2017-04-28 Jakub Jelinek <jakub@redhat.com>
30397 * cgraph.h (symtab_node::debug_symtab): No longer inline.
30398 * symtab.c (symtab_node::debug_symtab): Move definition here.
30400 2017-04-28 Richard Biener <rguenther@suse.de>
30402 * lto-streamer.h (LTO_major_version): Bump to 7.
30404 2017-04-28 Richard Biener <rguenther@suse.de>
30406 * tree-vrp.c (assert_info): New struct.
30407 (add_assert_info): New helper.
30408 (register_edge_assert_for_2): Refactor to add asserts to a vector
30410 (register_edge_assert_for_1): Likewise.
30411 (register_edge_assert_for): Likewise.
30412 (finish_register_edge_assert_for): New helper actually registering
30413 asserts where live on edge.
30414 (find_conditional_asserts): Adjust.
30415 (find_switch_asserts): Likewise.
30416 (evrp_dom_walker::try_find_new_range): Generalize.
30417 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
30419 2017-04-27 Marek Polacek <polacek@redhat.com>
30422 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
30423 arg10 and arg11 to itype.
30425 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
30427 * doc/extend.texi (Object Size Checking): Improve grammar.
30429 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
30432 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
30433 that the logic for permitting reciprocal estimates matches that
30436 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30439 * tree.c (type_cache_hasher::equal): Only compare
30440 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
30441 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
30442 non-aggregate element types.
30443 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
30444 about the flag on ARRAY_TYPEs in the comment, formatting fix.
30446 2017-04-27 Richard Biener <rguenther@suse.de>
30448 PR middle-end/80533
30449 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
30450 stripping ARRAY_REFs from MEM_EXPR make sure we're not
30451 keeping a reference to a trailing array.
30453 2017-04-27 Richard Biener <rguenther@suse.de>
30455 PR middle-end/80539
30456 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
30457 being in loop-closed SSA form conservatively.
30458 (chrec_fold_multiply_poly_poly): Likewise.
30460 2017-04-27 Tamar Christina <tamar.christina@arm.com>
30462 PR middle-end/79665
30463 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
30464 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
30466 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30469 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
30470 (aarch64_function_arg_alignment): Return unsigned int again, but still
30471 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
30472 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30473 Don't emit -Wpsabi note.
30474 (aarch64_function_arg_boundary): Likewise.
30475 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30478 2017-04-26 Nathan Sidwell <nathan@acm.org>
30480 * tree.h (crc32_unsigned_n): Declare.
30481 (crc32_unsigned, crc32_unsigned): Make inline.
30482 * tree.c (crc32_unsigned_bits): Replace with ...
30483 (crc32_unsigned_n): ... this.
30484 (crc32_unsigned, crc32_byte): Remove.
30485 (crc32_string): Remove unnecessary braces.
30487 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30489 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
30490 * ipa-inline-analysis.c (MAX_TIME): Remove.
30491 (account_size_time): Use sreal for time.
30492 (dump_inline_summary): Update.
30493 (estimate_function_body_sizes): Update.
30494 (estimate_edge_size_and_time): Update.
30495 (estimate_calls_size_and_time): Update.
30496 (estimate_node_size_and_time): Update.
30497 (inline_merge_summary): Update.
30498 (inline_update_overall_summary): Update.
30499 (estimate_time_after_inlining): Update.
30500 (inline_read_section): Update.
30501 (inline_write_summary): Update.
30502 * ipa-inline.c (compute_uninlined_call_time): Update.
30503 (compute_inlined_call_time): Update.
30504 (recursive_inlining): Update.
30505 (inline_small_functions): Update.
30506 (dump_overall_stats): Update.
30507 * ipa-inline.h: Include sreal.h.
30508 (size_time_entry): Turn time to sreal.
30509 (inline_summary): Turn self_time nad time to sreal.
30511 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30513 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
30515 (sreal::stream_out, sreal::stream_in): New.
30516 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
30518 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30520 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
30523 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
30526 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
30528 (dimode_scalar_chain::compute_convert_gain): Ditto.
30529 (dimode_scalar_chain::make_vector_copies): Ditto.
30530 (dimode_scalar_chain::convert_reg): Ditto.
30531 (dimode_scalar_chain::convert_insn): Ditto.
30532 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
30533 (VI248_AVX512BW_1): New mode iterator.
30534 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
30535 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
30538 2017-04-25 Martin Sebor <msebor@redhat.com>
30540 PR tree-optimization/80497
30541 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
30542 constants are representable in HOST_WIDE_INT.
30543 (parse_directive): Ditto.
30545 2017-04-25 Martin Sebor <msebor@redhat.com>
30548 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
30549 (new_zero_array): Adjust signature.
30550 (dom_info::dom_init): Used unsigned rather that size_t.
30551 (dom_info::dom_info): Same.
30553 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30554 Jakub Jelinek <jakub@redhat.com>
30557 * config/arm/arm.c: Include gimple.h.
30558 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30559 returns negative, increment ncrn only if it returned positive.
30560 (arm_needs_doubleword_align): Return int instead of bool,
30561 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
30562 members, but if there is any such non-FIELD_DECL
30563 > PARM_BOUNDARY aligned decl, return -1 instead of false.
30564 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30565 returns negative, increment nregs only if it returned positive.
30566 (arm_setup_incoming_varargs): Likewise.
30567 (arm_function_arg_boundary): Emit -Wpsabi note if
30568 arm_needs_doubleword_align returns negative, return
30569 DOUBLEWORD_ALIGNMENT only if it returned positive.
30571 2017-04-25 Marek Polacek <polacek@redhat.com>
30574 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
30575 first argument to type.
30577 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
30580 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
30581 type checks to test for compatibility instead of equality.
30583 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30584 Jakub Jelinek <jakub@redhat.com>
30587 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
30589 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
30590 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
30591 the alignment computation, but return their maximum in warn_alignment.
30592 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30593 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
30595 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
30596 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30599 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30601 * config/arc/simdext.md (dmpyh): Fix typo.
30603 2017-04-25 Richard Biener <rguenther@suse.de>
30605 PR tree-optimization/80492
30606 * alias.c (compare_base_decls): Handle registers with asm
30607 specification conservatively.
30608 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
30609 compare_base_decls returning dont-know properly.
30611 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30613 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
30614 (legitimate_offset_address_p): New function.
30615 (arc_legitimate_address_p): Use above function.
30617 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30619 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
30621 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30623 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
30624 ACCH registers whenever they are available.
30626 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30628 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
30629 double regs fix when not used.
30631 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30633 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
30635 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
30636 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
30638 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30640 * config/arc/arc.c (arc_output_addsi): Check for h-register class
30641 when emitting short ADD instructions.
30643 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30645 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
30647 (cmpsi_cc_c_insn): Likewise.
30648 (cbranchsi4_scratch): Compute proper instruction length using
30649 compact_hreg_operand.
30650 * config/arc/predicates.md (compact_hreg_operand): New predicate.
30652 2017-04-25 Richard Biener <rguenther@suse.de>
30654 PR middle-end/80509
30655 * passes.c (pass_manager::pass_manager): Initialize
30656 m_name_to_pass_map.
30658 2017-04-25 Richard Biener <rguenther@suse.de>
30660 PR tree-optimization/79201
30661 * tree-ssa-sink.c (statement_sink_location): Handle calls.
30663 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30666 * config/s390/vector.md: Split MEM->GPR vector moves for
30667 non-s_operand addresses.
30669 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30672 * config/s390/predicates.md (reload_const_wide_int_operand): New
30674 * config/s390/s390.md ("movti"): Remove d/P alternative.
30675 ("movti_bigconst"): New pattern definition.
30677 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30680 * s390-protos.h (s390_expand_cs_hqi): Removed.
30681 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
30682 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
30683 modes as well as CCZ1mode and CCZmode.
30684 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
30685 signature of s390_emit_compare_and_swap.
30686 (s390_expand_cs_hqi): Likewise, make static.
30687 (s390_expand_cs_tdsi): Generate an explicit compare before trying
30688 compare-and-swap, in some cases.
30689 (s390_expand_cs): Wrapper function.
30690 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
30692 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
30693 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
30694 patterns for small and large integers. Forbid symref memory operands.
30695 Move expander to s390.c. Require cc register.
30696 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
30697 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
30698 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
30699 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
30700 symref memory operands. Remove CC mode and call s390_match_ccmode
30702 ("atomic_exchange<mode>"): Allow and implement all integer modes.
30704 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30706 * config/s390/s390.md (define_peephole2): New peephole to help
30707 combining the load-and-test pattern with volatile memory.
30709 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30711 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
30712 with CCZmode for TARGET_Z196.
30714 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30716 PR rtl-optimization/80501
30717 * combine.c (make_compound_operation_int): Set subreg_code to SET
30718 even for AND with mask of the sign bit of mode.
30720 PR rtl-optimization/80500
30721 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
30722 sum's initial value.
30724 2017-04-25 Julian Brown <julian@codesourcery.com>
30725 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30727 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
30729 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
30731 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
30733 2017-04-25 Julian Brown <julian@codesourcery.com>
30734 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30736 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
30737 (thunderx2t99_sha): New Reservation.
30739 2017-04-25 Julian Brown <julian@codesourcery.com>
30740 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30742 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
30743 type for 1-element load.
30745 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
30747 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
30749 2017-04-24 Martin Jambor <mjambor@suse.cz>
30751 PR tree-optimization/80293
30752 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
30753 char arrays not totally scalarizable if it is false.
30754 (analyze_all_variable_accesses): Pass correct value in the new
30755 parameter. Add a statistics counter.
30757 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
30759 PR middle-end/79931
30760 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
30762 2017-04-24 Richard Biener <rguenther@suse.de>
30764 PR tree-optimization/80494
30765 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
30766 out for complex types.
30768 2017-04-24 Richard Biener <rguenther@suse.de>
30770 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
30771 * tree-ssa-sccvn.c (print_scc): Print SCC size.
30772 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
30773 (DFS): Adjust and never fail.
30774 (sccvn_dom_walker::fail): Remove.
30775 (sccvn_dom_walker::before_dom_children): Adjust.
30776 (run_scc_vn): Likewise and never fail.
30777 * tree-ssa-pre.c (pass_pre::execute): Adjust.
30778 (pass_fre::execute): Likewise.
30780 2017-04-24 Richard Biener <rguenther@suse.de>
30782 PR tree-optimization/79725
30783 * tree-ssa-sink.c (statement_sink_location): Return whether
30784 failure reason was zero uses. Move that check later.
30785 (sink_code_in_bb): Deal with zero uses by removing the stmt
30788 2017-04-24 Richard Biener <rguenther@suse.de>
30791 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
30792 pointer-based references.
30794 2017-04-24 Richard Biener <rguenther@suse.de>
30797 * pass_manager.h (pass_manager::operator new): Remove.
30798 (pass_manager::operator delete): Likewise.
30799 * passes.c (pass_manager::operator new): Remove.
30800 (pass_manager::operator delete): Likewise.
30801 (pass_manager::pass_manager): Zero individual pass members.
30803 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
30806 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
30807 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
30808 Check "XEXP (src, 1)" operand here.
30809 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
30810 Check "XEXP (src, 1)" operand here.
30811 (dimode_scalar_chain::make_vector_copies): Detect count register
30812 of a shift instruction. Zero extend count register from QImode
30813 to DImode to satisfy vector shift pattern count operand predicate.
30814 Substitute vector shift count operand with a DImode copy.
30815 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
30818 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
30820 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
30821 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
30822 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
30823 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
30824 (UNSPEC_NOREX_MEM): Remove definition.
30826 2017-04-21 Richard Biener <rguenther@suse.de>
30828 PR tree-optimization/79547
30829 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
30830 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
30831 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
30832 without any constraints.
30834 2017-04-21 Richard Biener <rguenther@suse.de>
30836 PR tree-optimization/78847
30837 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
30839 2017-04-21 Richard Biener <rguenther@suse.de>
30841 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
30842 (build_distinct_type_copy): Likewise.
30843 (build_variant_type_copy): Likewise.
30844 * tree.c (build_qualified_type): Pass down mem-stat info.
30845 (build_distinct_type_copy): Likewise.
30846 (build_variant_type_copy): Likewise.
30848 2017-04-21 Richard Biener <rguenther@suse.de>
30850 PR tree-optimization/80237
30851 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
30853 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
30854 for a simplified result.
30856 2016-04-21 Richard Biener <rguenther@suse.de>
30858 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
30859 sth as strict as a simple_iv but a chrec without symbols and an
30860 operand defined in the loop we are peeling (and not some subloop).
30861 (propagate_constants_for_unrolling): Propagate all constants.
30863 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
30866 * config/i386/i386.c (print_reg): Remove assert for disalowed
30867 regno values, call output_operand_lossage instead.
30869 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
30872 * config/i386/constraints.md (Yc): New register constraint.
30873 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
30874 Use Yc constraint for alternative 2 of operand 0. Remove
30875 preferred_for_speed attribute.
30877 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
30879 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
30880 lastprivate clauses in SIMT case.
30882 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
30884 * doc/invoke.texi (-Wextra-semi): Document new warning option.
30886 2017-04-20 Richard Biener <rguenther@suse.de>
30888 PR tree-optimization/57796
30889 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
30890 as N scalar stores.
30891 (vect_model_load_cost): Cost gathers as N scalar loads.
30893 2017-04-20 Richard Biener <rguenther@suse.de>
30895 * ggc-page.c (ggc_allocated_p): Rename to ...
30896 (safe_lookup_page_table_entry): ... this and return the lookup
30898 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
30900 2017-04-20 Richard Biener <rguenther@suse.de>
30902 PR tree-optimization/80453
30903 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
30904 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
30905 from the conditions.
30906 (vn_phi_eq): Pass them down.
30907 (vn_phi_lookup): Record them.
30908 (vn_phi_insert): Likewise.
30910 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
30912 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
30913 uninitialized variable warning to avoid buffer overrun.
30915 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
30918 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
30919 is suppressed for '{ 0 }' in C.
30921 2017-04-20 Jakub Jelinek <jakub@redhat.com>
30923 * BASE-VER: Set to 8.0.0.
30925 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
30927 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
30928 priority .init_array and .fini_array section with SECTION_NOTYPE
30931 2017-04-20 Jakub Jelinek <jakub@redhat.com>
30933 PR middle-end/80423
30934 * tree.h (build_array_type): Add typeless_storage default argument.
30935 * tree.c (type_cache_hasher::equal): Also compare
30936 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
30937 (build_array_type): Add typeless_storage argument, set
30938 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
30940 (build_nonshared_array_type): Adjust build_array_type_1 caller.
30941 (build_array_type): Likewise. Add typeless_storage argument.
30943 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
30944 Jakub Jelinek <jakub@redhat.com>
30946 PR tree-optimization/80426
30947 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
30948 operation on symbolic operands, also compute the overflow for the
30949 invariant part when the operation degenerates into a negation.
30951 2017-04-19 Jakub Jelinek <jakub@redhat.com>
30954 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
30955 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
30958 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
30960 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
30963 * config/avr/avr.c (tree.h): Include it.
30964 (cgraph.h): Include it.
30965 (avr_encode_section_info): Don't warn for uninitialized progmem
30966 variable if it's just an alias.
30968 2017-04-19 Richard Biener <rguenther@suse.de>
30971 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
30972 when needed by AutoPGO.
30974 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
30977 * doc/lto.texi: Remove an extra 'that'.
30979 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
30981 PR rtl-optimization/80429
30982 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
30983 are only used in debug insns.
30985 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
30986 Vladimir Makarov <vmakarov@redhat.com>
30988 * config/sparc/predicates.md (input_operand): Add comment. Return
30989 true for any memory operand when LRA is in progress.
30990 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
30992 2017-04-18 Jeff Law <law@redhat.com>
30995 * mips.md ({return,simple_return}_internal): Do not overwrite
30998 2017-04-18 Jakub Jelinek <jakub@redhat.com>
31000 PR tree-optimization/80443
31001 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
31002 instead of adding 1, subtract -1 and similarly instead of subtracting
31005 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
31007 PR rtl-optimization/80357
31008 * haifa-sched.c (tmp_bitmap): New variable.
31009 (model_recompute): Handle duplicate use records.
31010 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
31011 (free_global_sched_pressure_data): Free it.
31013 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
31016 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
31017 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31018 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31019 instead of SYSTEM_HEADER_DIR.
31021 2017-04-18 Jeff Law <law@redhat.com>
31023 PR middle-end/80422
31024 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
31025 predecessors after walking up the insn chain.
31027 2017-04-18 Jakub Jelinek <jakub@redhat.com>
31030 * dwarf2out.c (modified_type_die): Try harder not to emit internal
31031 sizetype type into debug info.
31033 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
31036 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
31037 unneeded test for TARGET_UPPER_REGS_SF.
31038 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
31040 2017-04-18 Jakub Jelinek <jakub@redhat.com>
31043 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
31044 instead of gsi_after_labels.
31046 2017-04-18 Jeff Law <law@redhat.com>
31048 * regcprop.c (maybe_mode_change): Avoid creating copies of the
31052 2017-04-13 Jeff Law <law@redhat.com>
31053 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
31054 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
31056 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
31059 * config/avr/avr.c (intl.h): Include it.
31060 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
31062 2017-04-18 Martin Liska <mliska@suse.cz>
31064 PR gcov-profile/78783
31065 * gcov-tool.c (gcov_output_files): Validate that destination
31066 file is either removed by the tool or by a user.
31068 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
31069 Guy Benyei <guybe@mellanox.com>
31071 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
31072 block, and do not negate it, the stored id is already negative.
31074 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
31076 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
31078 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
31081 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
31082 masks of options that should be turned off if the VSX vector
31083 options are turned off.
31084 (OTHER_P8_VECTOR_MASKS): Likewise.
31085 (OTHER_VSX_VECTOR_MASKS): Likewise.
31086 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
31087 rs6000_disable_incompatible_switches to validate no type switches
31089 (rs6000_incompatible_switch): New function to disallow turning on
31090 other vector options if -mno-vsx, -mno-power8-vector, or
31091 -mno-power9-vector are specified.
31093 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
31095 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
31097 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
31099 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
31100 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
31101 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
31102 (ARG_POINTER_CFA_OFFSET): Likewise.
31104 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
31106 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
31107 conditions to take advantage of various optimizations.
31109 2017-04-13 Jeff Law <law@redhat.com>
31111 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
31112 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
31113 (zero_extendsidi2_dext): Likewise.
31115 2017-04-13 Jakub Jelinek <jakub@redhat.com>
31118 * fold-const.c (fold_ternary_loc): Revert
31119 use op0 instead of fold_convert_loc (loc, type, arg0) part of
31122 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
31124 PR rtl-optimization/80343
31125 * lra-remat.c (update_scratch_ops): Assign original hard reg to
31126 new scratch pseudo.
31128 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
31131 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
31132 to ubsan_encode_value.
31134 2017-04-13 Jeff Law <law@redhat.com>
31136 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
31137 appearing in DEBUG_INSNs.
31139 2017-04-13 Martin Liska <mliska@suse.cz>
31141 PR gcov-profile/80413
31142 * gcov-io.c (gcov_write_string): Copy to buffer just when
31143 allocated size is greater than zero.
31145 2017-04-13 Jakub Jelinek <jakub@redhat.com>
31148 * dwarf2out.c (decls_for_scope): Ignore declarations of
31149 current_function_decl in BLOCK_NONLOCALIZED_VARS.
31151 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
31154 * ipa-visibility.c (non_local_p): Fix typos.
31155 (localize_node): When localizing symbol in same comdat group,
31156 dissolve the group only when we know external symbols are going
31158 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
31160 2017-04-12 Jakub Jelinek <jakub@redhat.com>
31162 PR tree-optimization/79390
31163 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
31164 order does not result in usable sequence, retry with reversed operand
31170 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
31171 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
31172 op0 instead of fold_convert_loc (loc, type, arg0).
31174 2017-04-12 Jeff Law <law@redhat.com>
31176 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
31177 has a delay slot in the generated code.
31179 * config/cris/cris.md (cris_preferred_reload_class): Return
31180 GENNONACR_REGS rather than GENERAL_REGS.
31182 2017-04-12 Jakub Jelinek <jakub@redhat.com>
31185 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
31186 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
31187 signedness of the result type.
31189 2017-04-12 Richard Biener <rguenther@suse.de>
31190 Jeff Law <law@redhat.com>
31192 PR tree-optimization/80359
31193 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
31194 trim stores to TARGET_MEM_REFs.
31196 2017-04-12 Richard Biener <rguenther@suse.de>
31198 PR tree-optimization/79390
31199 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
31200 threading case even more.
31202 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
31205 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
31206 for quad_address_p for TImode, instead of just not indexed_address.
31208 2017-04-12 Richard Biener <rguenther@suse.de>
31209 Bernd Edlinger <bernd.edlinger@hotmail.de>
31211 PR middle-end/79671
31212 * alias.c (component_uses_parent_alias_set_from): Handle
31213 TYPE_TYPELESS_STORAGE.
31214 (get_alias_set): Likewise.
31215 * tree-core.h (tree_type_common): Add typeless_storage flag.
31216 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
31217 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
31218 for types containing members with TYPE_TYPELESS_STORAGE.
31219 (place_field): Likewise.
31220 (layout_type): Likewise for ARRAY_TYPE.
31221 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
31222 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
31223 TYPE_TYPELESS_STORAGE.
31224 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
31226 2017-04-12 Jakub Jelinek <jakub@redhat.com>
31229 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
31230 first argument to type.
31232 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31236 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
31237 CONST0_RTX (mode) rather than const0_rtx where appropriate.
31238 (rs6000_expand_binop_builtin): Likewise.
31239 (rs6000_expand_ternop_builtin): Likewise; also add missing
31240 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
31241 vshasigma built-ins.
31242 * doc/extend.texi: Document that vec_xxpermdi's third argument
31243 must be a constant.
31245 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
31247 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
31248 Use shift_const cost parameter when calculating gain of STV shifts.
31250 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
31252 PR rtl-optimization/70478
31253 * lra-constraints.c (process_alt_operands): Check memory for
31254 disfavoring memory insn operand.
31256 2017-04-11 Jakub Jelinek <jakub@redhat.com>
31258 PR middle-end/80100
31259 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
31260 left shift in unsigned HOST_WIDE_INT type.
31262 PR rtl-optimization/80385
31263 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
31264 (not (neg X)) into (plus X -1) for complex or non-integral modes.
31267 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
31268 if they have any depend clauses.
31270 2017-04-11 Martin Liska <mliska@suse.cz>
31273 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
31274 * ipa-split.c (split_function): Create a local comdat symbol
31275 if caller is in a comdat group.
31277 2017-04-11 Martin Liska <mliska@suse.cz>
31280 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
31283 2017-04-11 Martin Sebor <msebor@redhat.com>
31285 PR middle-end/80364
31286 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
31287 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
31289 (directive::set_width, directive::set_precision, format_character):
31291 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
31294 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
31297 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
31298 conflict, set target->arch_name instead of target->cpu_name.
31300 2017-04-11 Richard Biener <rguenther@suse.de>
31302 PR tree-optimization/80374
31303 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
31304 build_zero_cst, remove fold_convertible_p check again.
31306 2017-04-11 Martin Liska <mliska@suse.cz>
31309 * ubsan.c (instrument_object_size): Do not instrument register
31312 2017-04-11 Jakub Jelinek <jakub@redhat.com>
31315 * config/i386/i386-builtin-types.def
31316 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
31317 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
31318 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
31319 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
31320 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
31321 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
31322 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
31323 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
31324 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
31325 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
31326 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
31327 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
31328 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
31329 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
31330 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
31331 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
31332 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
31333 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
31334 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
31335 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
31336 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
31337 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
31338 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
31339 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
31340 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
31341 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
31342 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
31343 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
31344 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
31345 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
31346 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
31347 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
31348 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
31349 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
31350 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
31351 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
31352 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
31353 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
31354 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
31355 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
31356 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
31357 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
31358 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
31359 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
31360 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
31361 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
31363 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
31364 flag to second_arg_count, handle 4 argument function type _COUNT
31365 aliases, handle second_arg_count on second argument rather than last.
31367 2017-04-10 Jeff Law <law@redhat.com>
31369 PR tree-optimization/80374
31370 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
31371 record anything if we can not convert integer_zero_node to the
31374 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
31377 * config/rs6000/rs6000.c (rs6000_option_override_internal):
31378 Enhance special handling given to the TARGET_P9_MINMAX option in
31379 relation to certain other options.
31381 2017-04-10 Bin Cheng <bin.cheng@arm.com>
31383 PR tree-optimization/80153
31384 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
31385 remove POINTER_PLUS_EXPR's base part directly, rather than through
31388 2017-04-10 Richard Biener <rguenther@suse.de>
31389 Bin Cheng <bin.cheng@arm.com>
31391 PR tree-optimization/80153
31392 * tree-affine.c (aff_combination_to_tree): Get base pointer from
31393 the first element of pointer type aff_tree. Build result expr in
31395 (add_elt_to_tree): Convert to type unconditionally. Remove other
31396 fold_convert calls.
31397 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
31398 (rewrite_use_nonlinear_expr): Check invariant using iv information.
31400 2017-04-10 Richard Biener <rguenther@suse.de>
31402 * tree-ssa-structalias.c (find_func_aliases): Properly handle
31405 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
31407 PR rtl-optimization/70478
31408 * lra-constraints.c (curr_small_class_check): New.
31409 (update_and_check_small_class_inputs): New.
31410 (process_alt_operands): Update curr_small_class_check. Disfavor
31411 alternative insn memory operands. Check available regs for small
31414 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
31417 * config/mips/mips.opt (-mvirt): Update description.
31418 * doc/invoke.texi (-mvirt): Likewise.
31420 2017-04-10 Richard Biener <rguenther@suse.de>
31422 PR middle-end/80362
31423 * fold-const.c (fold_binary_loc): Look at unstripped ops when
31424 looking for NEGATE_EXPR in -A / -B to A / B folding.
31426 2017-04-10 Martin Liska <mliska@suse.cz>
31428 PR gcov-profile/80224
31429 * gcov.c (print_usage): Fix usage string.
31430 (get_gcov_intermediate_filename): Remove.
31431 (output_gcov_file): Use both for normal and intermediate format.
31432 (generate_results): Do not initialize special file for
31433 intermediate format.
31435 2017-04-10 Richard Biener <rguenther@suse.de>
31437 PR tree-optimization/80304
31438 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
31441 2017-04-10 Nathan Sidwell <nathan@acm.org>
31444 * config/rs6000/rs6000.c (rs6000_vector_type): New.
31445 (rs6000_init_builtins): Use it.
31447 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31449 * config/arm/arm.md (<mrc>): Add mode to SET source.
31450 (<mrrc>): Likewise.
31452 2017-04-10 Richard Biener <rguenther@suse.de>
31454 PR middle-end/80344
31455 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
31457 2017-04-10 Jakub Jelinek <jakub@redhat.com>
31460 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
31461 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
31462 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
31463 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
31464 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
31465 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
31466 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
31467 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
31468 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
31469 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
31470 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
31471 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
31472 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
31473 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
31474 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
31475 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
31476 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
31477 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
31478 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
31479 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
31480 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
31481 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
31482 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
31484 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
31486 PR rtl-optimization/70478
31487 * lra-constraints.c: Reverse the last patch.
31489 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
31491 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
31492 Add comment for WCHAR_T.
31494 2017-04-08 Martin Liska <mliska@suse.cz>
31497 2017-04-07 Martin Liska <mliska@suse.cz>
31500 * ipa-split.c (split_function): Add function part to a same comdat
31503 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31506 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
31508 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
31510 * rs6000/rs6000.c (vec_load_pendulum): Rename...
31511 (vec_pairing): ...to this.
31512 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
31513 (rs6000_sched_init): Adjust for name change.
31514 (struct rs6000_sched_context): Likewise.
31515 (rs6000_init_sched_context): Likewise.
31516 (rs6000_set_sched_context): Likewise.
31518 2017-04-07 Jakub Jelinek <jakub@redhat.com>
31524 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
31526 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
31527 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
31528 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
31530 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
31532 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
31534 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31536 PR rtl-optimization/70703
31537 * ira-color.c (update_conflict_hard_regno_costs): Use
31538 int64_t instead of HOST_WIDE_INT.
31540 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31542 PR rtl-optimization/70478
31543 * lra-constraints.c (process_alt_operands): Disfavor alternative
31544 insn memory operands.
31546 2017-04-07 Jeff Law <law@redhat.com>
31548 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
31549 CALL and NOTE_INSN_CALL_ARG_LOCATION.
31551 2017-04-07 Martin Liska <mliska@suse.cz>
31554 * config/aarch64/aarch64.c (aarch64_process_target_attr):
31555 Show error message instead of an ICE.
31557 2017-04-07 Martin Liska <mliska@suse.cz>
31560 * ipa-split.c (split_function): Add function part to a same comdat
31563 2017-04-07 Richard Biener <rguenther@suse.de>
31565 PR middle-end/80341
31566 * tree.c (get_unwidened): Also handle ! for_type case for
31568 * convert.c (do_narrow): Split out from ...
31569 (convert_to_integer_1): ... here. Do not pass final truncation
31570 type to get_unwidened for TRUNC_DIV_EXPR.
31572 2017-04-07 Richard Biener <rguenther@suse.de>
31574 * tree-affine.c (wide_int_ext_for_comb): Take type rather
31576 (aff_combination_const): Adjust.
31577 (aff_combination_scale): Likewise.
31578 (aff_combination_add_elt): Likewise.
31579 (aff_combination_add_cst): Likewise.
31580 (aff_combination_convert): Likewise.
31581 (add_elt_to_tree): Likewise. Remove unused argument.
31582 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
31584 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
31586 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
31588 * config/arm/arm.c (arm_default_short_enums): Use
31589 ARM_DEFAULT_SHORT_ENUMS.
31590 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
31592 2017-04-06 Jakub Jelinek <jakub@redhat.com>
31595 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
31596 members with redundant out-of-class redeclaration.
31598 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31601 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
31602 * config/i386/i386.md (*zero_extendsidi2):
31603 Add (?*x,*x) and (?*v,*v) alternatives.
31605 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31608 * config/i386/i386.c (ix86_expand_builtin)
31609 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
31610 mode from insn data. Convert operands to insn operand mode.
31611 Copy operands that don't satisfy insn predicate to a register.
31613 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
31615 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
31618 2017-04-06 Richard Biener <rguenther@suse.de>
31620 PR tree-optimization/80334
31621 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
31622 preserve alignment of accesses.
31624 2017-04-06 Richard Biener <rguenther@suse.de>
31626 PR tree-optimization/80262
31627 * tree-sra.c (build_ref_for_offset): Preserve address-space
31629 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
31630 Drop useless address-space information on MEM_REF offsets.
31632 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
31634 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
31636 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31638 PR rtl-optimization/70703
31639 * ira-color.c (update_conflict_hard_regno_costs): Use
31640 HOST_WIDE_INT instead of long.
31642 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
31645 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
31646 not defined for x86_64 target. Add -mmmx target option when __SSE2__
31648 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
31649 for x86_64 target. Handle -m3dnowa option.
31651 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31653 PR rtl-optimization/70703
31654 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
31655 (update_conflict_hard_regno_costs): Use long instead of unsigned
31656 arithmetic for cost calculation.
31658 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31659 Bernd Edlinger <bernd.edlinger@hotmail.de>
31662 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
31665 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
31668 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
31669 ptr_mode with Pmode throughout.
31670 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
31671 into probe_stack_range and use DImode.
31673 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
31676 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
31677 call_eh_return is true.
31679 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31681 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
31682 Initialize last_match_fntype_index.
31684 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31687 * tree-nvr.c: Include internal-fn.h.
31688 (pass_return_slot::execute): Ignore internal calls without
31691 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31692 Richard Biener <rguenther@suse.de>
31695 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
31696 captures used multiple times, except for the last use.
31697 * generic-match-head.c: Include gimplify.h.
31699 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31701 PR tree-optimization/79390
31702 * target.h (struct noce_if_info): Declare.
31703 * targhooks.h (default_noce_conversion_profitable_p): Declare.
31704 * target.def (noce_conversion_profitable_p): New target hook.
31705 * ifcvt.h (struct noce_if_info): New type, moved from ...
31706 * ifcvt.c (struct noce_if_info): ... here.
31707 (noce_conversion_profitable_p): Renamed to ...
31708 (default_noce_conversion_profitable_p): ... this. No longer
31710 (noce_try_store_flag_constants, noce_try_addcc,
31711 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
31712 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
31713 instead of noce_conversion_profitable_p.
31714 * config/i386/i386.c: Include ifcvt.h.
31715 (ix86_option_override_internal): Don't override
31716 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
31717 (ix86_noce_conversion_profitable_p): New function.
31718 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
31719 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
31720 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
31721 * doc/tm.texi: Regenerated.
31723 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31725 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
31728 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31731 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
31732 instructions for small multiply cores.
31734 2017-04-04 Jeff Law <law@redhat.com>
31736 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
31738 (mips_expand_vec_perm_const): Initialize elements in orig_perm
31739 that are not set by the loop over the elements.
31741 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31744 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
31745 int mode, convert_modes it to mode as unsigned, otherwise use
31746 lowpart_subreg to mode rather than SImode.
31747 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
31748 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
31749 Use DImode instead of SImode for the shift count operand.
31750 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
31753 2017-04-04 Richard Biener <rguenther@suse.de>
31755 PR middle-end/80281
31756 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
31757 arithmetic done for the negate or the plus. Simplify.
31758 (A - (-B) -> A + B): Likewise.
31759 * fold-const.c (split_tree): Make sure to not negate pointers.
31761 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
31763 PR rtl-optimization/60818
31764 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
31765 a compare of comparisons with the thing compared if this results
31766 in a different machine mode.
31768 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
31770 * alias.c (base_alias_check): Fix typo in comment.
31771 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
31772 * cgraphunit.c (symbol_table::compile): Likewise.
31773 * collect2.c (maybe_run_lto_and_relink): Likewise.
31774 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
31775 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
31776 * config/avr/avr.c (avr_map_op_t): Likewise.
31777 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
31778 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
31779 * config/epiphany/epiphany.md (movcc): Likewise.
31780 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
31781 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
31783 * config/mips/mips.c (mips_save_restore_reg): Likewise.
31784 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
31785 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
31786 * config/sh/sh.c (sh_rtx_costs): Likewise.
31787 * fold-const.c (fold_truth_andor): Likewise.
31788 * genautomata.c (collapse_flag): Likewise.
31789 * gengtype.h (struct type::u::s): Likewise.
31790 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
31791 * input.c (FORMAT_AMOUNT): Likewise.
31792 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
31793 (known_aggs_to_agg_replacement_list): Likewise.
31794 * ipa-inline-analysis.c: Likewise.
31795 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
31796 * ipa-polymorphic-call.c
31797 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
31798 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
31799 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
31801 * modulo-sched.c (apply_reg_moves): Likewise.
31802 * omp-expand.c (build_omp_regions_1): Likewise.
31803 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
31804 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
31805 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
31806 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
31807 * value-prof.c: Likewise.
31808 * var-tracking.c (val_reset): Likewise.
31810 2017-04-03 Richard Biener <rguenther@suse.de>
31812 PR tree-optimization/80275
31813 * fold-const.c (split_address_to_core_and_offset): Handle
31816 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
31818 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
31819 descriptors is at least equal to that of functions.
31821 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
31823 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
31825 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
31828 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
31829 (mov<IMOD4:mode>): New expander.
31830 (*mov<IMOD4:mode>_internal): New insn and split pattern.
31832 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
31834 PR rtl-optimization/79405
31835 * fwprop.c (propagations_left): New variable.
31836 (forward_propagate_into): Decrement it.
31837 (fwprop_init): Initialize it.
31838 (fw_prop): If the variable has reached zero, stop propagating.
31839 (fwprop_addr): Ditto.
31841 2017-03-31 Jakub Jelinek <jakub@redhat.com>
31844 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
31845 a FUNCTION_DECL, pass it as decl instead of origin to
31848 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
31850 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
31853 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
31856 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
31857 TARGET_VSX_SMALL_INTEGER.
31859 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31861 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
31862 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
31864 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
31866 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
31867 extraction from odd-numbered MSA register.
31869 2017-03-31 Jakub Jelinek <jakub@redhat.com>
31871 PR middle-end/80173
31872 * expmed.c (store_bit_field_1): Don't attempt to create
31873 a word subreg out of hard registers wider than word if they
31874 have HARD_REGNO_NREGS of 1 for their mode.
31876 PR middle-end/80163
31877 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
31878 conversions to integer types wider than word and pointer.
31881 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
31882 (rtx_equal_for_cselib_p): Pass 0 to it.
31883 * cselib.c (cselib_hasher::equal): Likewise.
31884 (rtx_equal_for_cselib_1): Add depth argument. If depth
31885 is 128, don't look up VALUE locs and punt. Increment
31886 depth in recursive calls when walking VALUE locs.
31888 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
31890 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
31891 (make_gcov_file_name): Use the canonical path name for generating
31893 (read_line): Fix handling of files with ascii null bytes.
31895 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
31897 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
31898 to initialise a vector register instead
31899 of using a const_int.
31901 2017-03-30 Jakub Jelinek <jakub@redhat.com>
31903 PR translation/80189
31904 * gimplify.c (omp_default_clause): Use %qs instead of %s in
31905 diagnostic messages.
31907 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
31910 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
31911 (dfp_diex_<mode>): Update mode of operand 1.
31912 * doc/extend.texi (dxex, dxexq): Document change to return type.
31913 (diex, diexq): Document change to argument type.
31915 2017-03-30 Martin Jambor <mjambor@suse.cz>
31918 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
31919 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
31920 it reflects the signature changes performed at the callee side.
31921 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
31922 to cgraph_build_function_type_skip_args.
31923 (build_function_decl_skip_args): Adjust call to the above function.
31925 2017-03-30 Jakub Jelinek <jakub@redhat.com>
31928 * config/i386/sse.md
31929 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
31930 register as dest whenever it is a MEM not rtx_equal_p to the
31931 corresponding dup operand, and when forcing into reg move the
31932 reg into the memory afterwards.
31933 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
31934 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
31935 for the force_reg mode.
31936 (avx512vl_vextractf128<mode>): Use register as dest either
31937 always when a MEM, or when it is a MEM not rtx_equal_p to the
31938 corresponding dup operand, or even not when it is a CONST_VECTOR
31939 depending on the mode and lo vs. hi.
31940 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
31942 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
31943 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
31944 Likewise. Require that operands[2] is even.
31945 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
31946 Remove extraneous parens. Require that operands[2] is a multiple
31948 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
31949 operands[0] is a MEM if <mask_applied>, the predicates/constraints
31950 disallow memory then.
31952 2017-03-30 Richard Biener <rguenther@suse.de>
31954 PR tree-optimization/77498
31955 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
31956 to non-constants over backedges.
31958 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
31960 PR rtl-optimization/80233
31961 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
31962 as last_combined_insn. Do not test for BARRIER_P separately.
31964 2017-03-29 Andreas Schwab <schwab@suse.de>
31967 * calls.c (prepare_call_address): Convert funexp to Pmode before
31968 copying to temp reg.
31970 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31972 PR tree-optimization/80158
31973 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
31974 Handle possible future case of more than one alternate
31976 (replace_rhs_if_not_dup): Likewise.
31977 (replace_one_candidate): Likewise.
31979 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
31981 PR rtl-optimization/80193
31982 * ira.c (ira): Do not check allocation for LRA.
31984 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
31986 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
31987 (nvptx_output_simt_exit): Declare.
31988 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
31989 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
31990 (init_softstack_frame): Move initialization of crtl->is_leaf to...
31991 (nvptx_declare_function_name): ...here. Emit declaration of local
31992 memory space buffer for omp_simt_enter insn.
31993 (nvptx_output_unisimt_switch): New.
31994 (nvptx_output_softstack_switch): New.
31995 (nvptx_output_simt_enter): New.
31996 (nvptx_output_simt_exit): New.
31997 * config/nvptx/nvptx.h (struct machine_function): New fields
31998 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
31999 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
32000 (UNSPECV_SIMT_EXIT): Ditto.
32001 (omp_simt_enter_insn): New insn.
32002 (omp_simt_enter): New expansion.
32003 (omp_simt_exit): New insn.
32004 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
32006 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
32007 (expand_GOMP_SIMT_ENTER_ALLOC): New.
32008 (expand_GOMP_SIMT_EXIT): New.
32009 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
32010 (GOMP_SIMT_ENTER_ALLOC): Ditto.
32011 (GOMP_SIMT_EXIT): Ditto.
32012 * target-insns.def (omp_simt_enter): New insn.
32013 (omp_simt_exit): Ditto.
32014 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
32016 (lower_rec_simd_input_clauses): Implement SIMT privatization.
32017 (lower_rec_input_clauses): Likewise.
32018 (lower_lastprivate_clauses): Handle SIMT privatization.
32020 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
32021 (ompdevlow_adjust_simt_enter): New.
32022 (find_simtpriv_var_op): New.
32023 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
32024 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
32026 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
32027 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
32028 (copy_decl_for_dup_finish): Ditto.
32030 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
32032 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
32035 * config/i386/i386.c (ix86_option_override_internal): Always
32036 allow -mpreferred-stack-boundary=3 for 64-bit targets.
32038 2017-03-28 Bin Cheng <bin.cheng@arm.com>
32040 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
32042 2017-03-28 Bin Cheng <bin.cheng@arm.com>
32044 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
32045 mark new edge's irreducible flag accordign to it.
32046 (vect_do_peeling): Check loop preheader edge's irreducible flag
32047 and pass it to function slpeel_add_loop_guard.
32049 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
32051 PR tree-optimization/80218
32052 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
32053 Update block frequencies and counts.
32055 2017-03-28 Richard Biener <rguenther@suse.de>
32057 PR tree-optimization/78644
32058 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
32059 of a simplification result we may not use it at all.
32061 2017-03-28 Richard Biener <rguenther@suse.de>
32064 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
32065 without arguments, generate default definition of a SSA name.
32067 2017-03-28 Richard Biener <rguenther@suse.de>
32069 PR middle-end/80222
32070 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
32071 TYPE_REF_CAN_ALIAS_ALL references.
32072 * fold-const.c (fold_indirect_ref_1): Likewise.
32074 2017-03-28 Martin Liska <mliska@suse.cz>
32077 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
32078 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
32080 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
32081 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
32083 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
32084 (EXTRA_SPECS): Define.
32085 (SUBTARGET_EXTRA_SPECS): Likewise.
32086 (SUBTARGET_CPP_SPEC): Likewise.
32087 * config/arc/elf.h (EXTRA_SPECS): Renamed to
32088 SUBTARGET_EXTRA_SPECS.
32089 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
32091 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
32093 * config/arc/simdext.md (vst64_insn): Update pattern.
32094 (vld32wh_insn): Likewise.
32095 (vld32wl_insn): Likewise.
32096 (vld64_insn): Likewise.
32097 (vld32_insn): Likewise.
32099 2017-03-28 Marek Polacek <polacek@redhat.com>
32102 * fold-const.c (fold_comparison): Use protected_set_expr_location
32103 instead of SET_EXPR_LOCATION.
32105 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
32107 * tree.c (add_expr): Avoid name lookup warning.
32109 2017-03-27 Jeff Law <law@redhat.com>
32111 PR tree-optimization/80216
32112 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
32113 function name. Limit recursion depth.
32114 (record_temporary_equivalences): Corresponding changes.
32116 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
32118 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
32121 2017-03-27 Jakub Jelinek <jakub@redhat.com>
32124 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
32126 * cfgcleanup.c (reg_note_cfa_p): New array.
32127 (insns_have_identical_cfa_notes): New function.
32128 (old_insns_match_p): Don't cross-jump in between /f
32129 and non-/f instructions. If both i1 and i2 are frame related,
32130 verify all CFA notes, their order and content.
32132 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
32135 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
32136 HImode and SImode with zero extend to DImode to one insn.
32137 (bswap<mode>2_extenddi): Likewise.
32138 (bswapsi2_extenddi): Likewise.
32139 (bswaphi2_extendsi): Likewise.
32140 (bswaphi2): Combine bswap HImode and SImode into one insn.
32141 Separate memory insns from swapping register.
32142 (bswapsi2): Likewise.
32143 (bswap<mode>2): Likewise.
32144 (bswaphi2_internal): Delete, no longer used.
32145 (bswapsi2_internal): Likewise.
32146 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
32147 store, and gpr<-gpr swap insns.
32148 (bswap<mode>2_store): Likewise.
32149 (bswaphi2_reg): Register only splitter, combine with the splitter.
32150 (bswaphi2 splitter): Likewise.
32151 (bswapsi2_reg): Likewise.
32152 (bswapsi2 splitter): Likewise.
32153 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
32154 the insns into load, store, and register/register insns.
32155 (bswapdi2_ldbrx): Likewise.
32156 (bswapdi2_load): Likewise.
32157 (bswapdi2_store): Likewise.
32158 (bswapdi2_reg): Likewise.
32160 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
32162 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
32163 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
32165 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
32168 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
32170 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32171 special handling for target option conflicts between dform
32172 options (-mpower9-dform, -mpower9-dform-vector,
32173 -mpower9-dform-scalar) and -mno-direct-move.
32175 2017-03-27 Richard Biener <rguenther@suse.de>
32177 PR tree-optimization/80181
32178 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
32180 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32182 * config/arc/predicates.md (move_double_src_operand): Replace the
32183 call to move_double_src_operand with a call to address_operand.
32185 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32187 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
32188 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
32189 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
32191 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32193 * config/arc/predicates.md (long_immediate_loadstore_operand):
32194 Consider scaled addresses cases.
32196 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32198 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
32199 restored when in interrupt.
32200 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
32201 doesn't have delay slot.
32203 2017-03-27 Richard Biener <rguenther@suse.de>
32206 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
32207 inlined thunk clones.
32209 2017-03-27 Jakub Jelinek <jakub@redhat.com>
32212 * asan.c (instrument_derefs): Copy over last operand from
32213 original COMPONENT_REF to the new COMPONENT_REF with
32214 DECL_BIT_FIELD_REPRESENTATIVE.
32215 * ubsan.c (instrument_object_size): Likewise.
32217 2017-03-27 Richard Biener <rguenther@suse.de>
32219 PR tree-optimization/80170
32220 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
32221 sure DR/SCEV didnt fold in constants we do not see when looking
32222 at the reference base alignment.
32224 2017-03-27 Richard Biener <rguenther@suse.de>
32226 PR middle-end/80171
32227 * gimple-fold.c (fold_ctor_reference): Properly guard against
32228 NULL return value from canonicalize_constructor_val.
32230 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
32233 * config/i386/i386.c (ix86_expand_builtin)
32234 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
32235 flags reg setting and flags reg using instructions.
32236 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
32237 clobbering instructions to zero extend op2.
32239 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
32241 * doc/install.texi (Configuration) <--with-aix-soname>:
32242 Update link to AIX ld.
32244 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
32246 PR rtl-optimization/80160
32247 PR rtl-optimization/80159
32248 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
32249 reg_alternate_class into account.
32251 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
32254 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
32255 to consider in curr_insn_transform.
32257 2017-03-24 Jakub Jelinek <jakub@redhat.com>
32259 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
32260 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
32261 and emit_mode_inner.
32263 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32265 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
32266 argument to the overloaded builtin variants. Use the new flag to
32267 deprecate certain builtin variants.
32268 * config/s390/s390-builtin-types.def: Add new builtin types.
32269 * config/s390/s390-builtins.h: Support new flags field for
32270 overloaded builtins.
32271 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
32272 (s390_macro_to_expand): Enable vector float data type.
32273 (s390_cpu_cpp_builtins_internal): Indicate support of the new
32274 builtins by incrementing the __VEC__ version number.
32275 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
32277 (s390_resolve_overloaded_builtin): Emit error messages depending
32278 on the builtin flags.
32279 * config/s390/s390.c (s390_expand_builtin): Support additional
32280 flags argument. Change error message to match the messages
32281 emitted in s390-c.c.
32282 * config/s390/s390.md: New UNSPEC_* constants.
32283 (op_type): Add new instruction types.
32284 * config/s390/vecintrin.h: Add new builtins and test data class
32286 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
32287 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
32288 (VEC_INEXACT, VEC_NOINEXACT): New constants.
32289 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
32290 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
32291 ("vec_mergel<mode>"): V_HW -> VEC_HW.
32293 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
32294 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
32295 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
32296 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
32298 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
32299 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
32300 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
32301 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
32303 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
32304 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
32305 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
32306 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
32307 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
32308 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
32309 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
32311 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
32312 ("vec_scatter_element<V_HW_4:mode>_DI")
32313 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
32314 ("vec_fpint<mode>", "vflls")
32315 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
32316 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
32317 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
32318 ("*vec_cmphe<mode>_cc"): ... these.
32320 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
32321 mode constant instead of magic value.
32323 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32325 * config/s390/s390.c (s390_expand_vec_compare): Support other
32326 vector floating point modes than just V2DF.
32327 (s390_expand_vcond): Likewise.
32328 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
32329 (s390_cannot_change_mode_class): Prevent mode changes between TF
32330 and V1TF in vector registers.
32331 * config/s390/s390.md (DF, SF): New mode attributes.
32332 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
32333 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
32334 SFmode support for VRs.
32335 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
32337 (VFT, VF_HW): New mode iterators.
32338 (vw, sdx): New mode attributes.
32339 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
32340 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
32341 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
32342 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
32343 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
32344 also the new vector floating point modes. Renaming to ...
32346 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
32347 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
32348 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
32349 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
32350 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
32351 ("vec_unordered<mode>"): ... these.
32353 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
32354 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
32355 ("*vec_extendv2df"): New insn definitions.
32357 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32359 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
32360 ("mulditi3_2", "*muldi3_sign"): New patterns.
32361 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
32362 rename the pattern definition.
32364 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32366 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
32368 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
32370 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32372 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
32373 instruction if possible.
32374 * config/s390/vector.md (vec_halfnumelts): New mode
32376 ("*vec_vllezlf<mode>"): New pattern.
32378 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32380 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
32381 ("popcountv4si2", "popcountv2di2"): Rename to ...
32382 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
32383 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
32385 ("popcount<mode>2_vxe"): New pattern.
32387 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32389 * common/config/s390/s390-common.c (processor_flags_table): Add
32391 * config.gcc: Add arch12.
32392 * config/s390/driver-native.c (s390_host_detect_local_cpu):
32393 Default to arch12 for unknown CPU model numbers.
32394 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
32395 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
32396 PROCESSOR_max sanity check.
32397 * config/s390/s390-opts.h (enum processor_type): Add
32399 * config/s390/s390.c (processor_table): Add arch12.
32400 (s390_expand_builtin): Add check for B_VXE flag.
32401 (s390_issue_rate): Add PROCESSOR_ARCH12.
32402 (s390_get_sched_attrmask): Likewise.
32403 (s390_get_unit_mask): Likewise.
32404 (s390_sched_score): Enable z13 scheduling for arch12.
32405 (s390_sched_reorder): Likewise.
32406 (s390_sched_variable_issue): Likewise.
32407 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
32409 (s390_tune_attr): Use z13 scheduling also for arch12.
32410 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
32411 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
32412 (TARGET_VXE_P): New macros.
32413 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
32414 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
32415 * config/s390/s390.opt: Add arch12 as processor_type.
32417 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32419 * config/s390/s390.md
32420 ("fixuns_truncdddi2", "fixuns_trunctddi2")
32421 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
32422 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
32424 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
32425 Rename expanders to ...
32427 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
32428 ("fixuns_truncdddi2_emu"): ... these.
32430 ("fixuns_trunc<mode>si2_emu"): New expander.
32432 ("*fixuns_truncdfdi2_z13"): Rename to ...
32433 ("*fixuns_truncdfdi2_vx"): ... this.
32435 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32437 * config/s390/2964.md: Remove the single element vector compare
32438 instructions which are no longer used.
32439 * config/s390/s390.c (s390_select_ccmode): Remove handling of
32441 (s390_canonicalize_comparison): Remove handling of DFmode
32443 (s390_expand_vec_compare_scalar): Remove function.
32444 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
32445 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
32447 ("*cmp<mode>_ccs"): Add wfcdb instruction.
32449 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32451 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
32453 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
32454 will anyway by matched by mov<mode>_64dfp.
32456 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32458 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
32459 vlef/vstef. Add missing operand to vleif.
32461 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32463 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
32464 pair for all vector types with 64 bit elements.
32465 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
32466 * config/s390/vector.md (V_HW_64): ... here.
32467 (V_128_NOSINGLE): New mode iterator.
32468 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
32469 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
32470 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
32471 ("*vec_load_pairv2di"): Change to ...
32472 ("*vec_load_pair<mode>"): ... this one.
32474 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32476 * config/s390/constraints.md: Add comments.
32477 (jKK): Reject element sizes > 8 bytes.
32478 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
32480 * config/s390/s390.md: Add the s_operand checks formerly in
32481 s390_split_ok_p to various splitters where they are still
32483 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
32484 for 128 bit vectors. Plus two splitters.
32486 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32488 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
32491 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32494 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
32495 error if the boundary argument is not constant.
32497 2017-03-24 Jakub Jelinek <jakub@redhat.com>
32499 PR rtl-optimization/80112
32500 * loop-doloop.c (doloop_condition_get): Don't check condition
32501 if cmp isn't SET with IF_THEN_ELSE src.
32503 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32505 PR tree-optimization/80158
32506 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
32507 replacing a candidate statement, also replace it for the
32508 candidate's alternate interpretation.
32509 (replace_rhs_if_not_dup): Likewise.
32510 (replace_one_candidate): Likewise.
32512 2017-03-24 Richard Biener <rguenther@suse.de>
32514 PR tree-optimization/80167
32515 * graphite-isl-ast-to-gimple.c
32516 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
32518 (translate_isl_ast_to_gimple::get_rename): Likewise.
32520 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32522 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
32523 handling of certain combinations of target options, including the
32524 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
32525 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
32527 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32530 * config/arm/arm.md (*load_multiple): Add reload_completed to
32531 matching condition.
32533 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32534 Richard Biener <rguenth@suse.de>
32536 PR tree-optimization/79908
32537 PR tree-optimization/80136
32538 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32539 been cast away, gimplify_and_add suffices.
32541 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
32543 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
32545 2017-03-23 Richard Biener <rguenther@suse.de>
32547 PR tree-optimization/80032
32548 * gimplify.c (gimple_push_cleanup): Forced unconditional
32549 cleanups still have to go to the conditional_cleanups
32552 2017-03-22 Jakub Jelinek <jakub@redhat.com>
32554 PR tree-optimization/80072
32555 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
32557 (next_operand_entry_id): Change type to unsigned int.
32558 (sort_by_operand_rank): Make sure to return the right return value
32559 even if unsigned fields are bigger than INT_MAX.
32560 (struct oecount): Change cnt and id type to unsigned int.
32561 (oecount_hasher::equal): Formatting fix.
32562 (oecount_cmp): Make sure to return the right return value
32563 even if unsigned fields are bigger than INT_MAX.
32564 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
32567 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
32568 TREE_READONLY on result if writing it more than once.
32571 * doc/invoke.texi (-fsanitize=thread): Document that with
32572 -fnon-call-exceptions atomics are not able to throw
32576 * tsan.c: Include tree-eh.h.
32577 (instrument_builtin_call): Call maybe_clean_eh_stmt or
32578 maybe_clean_or_replace_eh_stmt where needed.
32579 (instrument_memory_accesses): Add cfg_changed argument.
32580 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
32581 if it returned true.
32582 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
32584 PR rtl-optimization/63191
32585 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
32586 wrapper function, moved the whole old content into ...
32587 (ix86_delegitimize_address_1): ... this. New inline function.
32588 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
32589 true as last argument instead of ix86_delegitimize_address.
32591 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32593 * config/aarch64/aarch64.c (generic_branch_cost): Copy
32594 cortexa57_branch_cost.
32596 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32598 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
32600 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32603 * doc/md.texi (Constraints): Document wA constraint.
32604 * config/rs6000/constraints.md (wA): New.
32605 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
32606 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
32607 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
32608 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
32610 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
32613 * gimplify.c (is_oacc_declared): New function.
32614 (oacc_default_clause): Use it to set default flags for acc declared
32615 variables inside parallel regions.
32616 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
32617 declared variables.
32618 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
32619 declare attribute to any decl as necessary.
32621 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
32624 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
32625 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
32626 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
32627 (arm_arch_lpae): This.
32628 * config/arm/arm.c (arm_arch7ve): Rename into ...
32629 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
32630 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
32633 2017-03-22 Martin Liska <mliska@suse.cz>
32636 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
32637 error message instead of an ICE.
32639 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32641 * doc/extend.texi (6.11 Additional Floating Types): Revise.
32643 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
32645 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
32647 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32650 2017-03-21 Martin Sebor <msebor@redhat.com>
32652 * doc/extend.texi: Use "cannot" instead of "can't."
32653 * doc/hostconfig.texi: Same.
32654 * doc/install.texi: Same.
32655 * doc/invoke.texi: Same.
32656 * doc/loop.texi: Same.
32657 * doc/md.texi: Same.
32658 * doc/objc.texi: Same.
32659 * doc/rtl.texi: Same.
32660 * doc/tm.texi: Same.
32661 * doc/tm.texi.in: Same.
32662 * doc/trouble.texi: Same.
32664 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
32667 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
32668 (collect_checksum_attributes): Set it.
32669 (die_checksum_ordered): Use it.
32671 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32673 PR tree-optimization/79908
32674 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
32675 change: For a VA_ARG whose LHS has been cast away, use
32676 force_gimple_operand to construct the side effects.
32678 2017-03-21 David Malcolm <dmalcolm@redhat.com>
32680 PR translation/80001
32681 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
32682 more amenable to translation.
32683 (oacc_loop_auto_partitions): Likewise.
32685 2017-03-21 Marek Polacek <polacek@redhat.com>
32686 Martin Sebor <msebor@redhat.com>
32688 PR tree-optimization/80109
32689 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
32690 on INTEGRAL_TYPE_P.
32692 2017-03-21 Jakub Jelinek <jakub@redhat.com>
32693 Segher Boessenkool <segher@kernel.crashing.org>
32696 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
32697 check reg_used_between_p between insn and one of succ or succ2
32698 depending on if succ is artificial insn not inserted into insn
32701 2017-03-21 Martin Liska <mliska@suse.cz>
32703 PR gcov-profile/80081
32704 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
32705 * doc/gcc.texi: Include gcov-dump stuff.
32706 * doc/gcov-dump.texi: New file.
32708 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
32710 PR rtl-optimization/79150
32711 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
32712 conditional jump, if the jump is the last insn of the loop.
32714 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32715 Richard Biener <rguenth@suse.de>
32717 PR tree-optimization/79908
32718 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32719 been cast away, use force_gimple_operand to construct the side
32722 2017-03-21 Martin Liska <mliska@suse.cz>
32724 PR libfortran/79956
32725 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
32728 2017-03-21 Brad Spengler <spender@grsecurity.net>
32731 * plugin.c (htab_hash_plugin): New function.
32732 (add_new_plugin): Use it and adjust.
32733 (parse_plugin_arg_opt): Adjust.
32734 (init_one_plugin): Likewise.
32736 2017-03-21 Richard Biener <rguenther@suse.de>
32738 PR tree-optimization/80032
32739 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
32740 if set force the cleanup to happen unconditionally.
32741 (gimplify_target_expr): Push inserted clobbers with force_uncond
32742 to avoid them being removed by control-dependent DCE.
32744 2017-03-21 Richard Biener <rguenther@suse.de>
32746 PR tree-optimization/80122
32747 * tree-inline.c (copy_bb): Do not expans va-arg packs or
32748 va_arg_pack_len when the inlined call stmt requires pack
32750 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
32752 2017-03-21 Jakub Jelinek <jakub@redhat.com>
32755 * tsan.c (instrument_builtin_call): If the memory model argument
32756 is not a constant, assume it is valid.
32759 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
32762 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
32764 PR rtl-optimization/79910
32765 * combine.c (can_combine_p): Do not allow combining an I0 or I1
32766 if its dest is used by an insn before I2 (other than the combined
32767 insns themselves, which are properly handled already).
32769 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
32772 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
32774 * combine.c (record_used_regs): New static function.
32775 (try_combine): Handle situations where there is an additional
32776 instruction between I2 and I3 which needs to have a LOG_LINK
32780 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
32782 * combine.c (try_combine): Delete redundant i1 test. Call
32783 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
32785 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32788 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
32789 alternatives 13/14.
32791 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32793 PR tree-optimization/80054
32794 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
32795 the optimization if a PHI or any of its arguments is not dominated
32796 by the candidate's basis. Use gphi* rather than gimple* as
32798 (replace_profitable_candidates): Clean up a gimple* variable that
32799 should be a gphi* variable.
32801 2017-03-20 Martin Sebor <msebor@redhat.com>
32804 * doc/extend.texi (attribute constructor): Document present limitation.
32806 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
32809 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
32810 __POWER9_VECTOR__ #ifdef control, change template definition to
32811 use Power9-specific built-in function.
32812 (vec_any_eq): Likewise.
32813 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
32814 to control outcomes from this test.
32815 (vector_ae_<mode>p): For VEC_F modes, likewise.
32817 2017-03-20 Ian Lance Taylor <iant@google.com>
32819 * config/i386/i386.c (ix86_function_regparm): Save an extra
32820 register for -fsplit-stack with DECL_STATIC_CHAIN.
32822 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
32825 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
32826 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
32828 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
32830 * config/riscv/riscv.c (riscv_print_operand): Use "fence
32832 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
32835 2017-03-20 Marek Polacek <polacek@redhat.com>
32838 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
32840 2017-03-20 Richard Biener <rguenther@suse.de>
32842 PR tree-optimization/80113
32843 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
32844 allocate extra SSA name for PHI def.
32845 (add_close_phis_to_outer_loops): Likewise.
32846 (add_close_phis_to_merge_points): Likewise.
32847 (copy_loop_close_phi_args): Likewise.
32848 (copy_cond_phi_nodes): Likewise.
32850 2017-03-20 Martin Liska <mliska@suse.cz>
32852 PR middle-end/79753
32853 * tree-chkp.c (chkp_build_returned_bound): Do not build
32854 returned bounds for a LHS that's not a BOUNDED_P type.
32856 2017-03-20 Martin Liska <mliska@suse.cz>
32860 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
32861 COMPLEX_CST and VECTOR_CST.
32863 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32866 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
32867 target operand. A new splitter adds the clobber statement in case
32868 the target operand is dead anyway.
32870 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
32872 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
32873 to age-old versions of binutils and glibc.
32875 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
32877 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
32879 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
32881 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
32883 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
32885 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
32886 requirement for binutils 2.13.
32888 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
32890 * combine.c (try_combine): Delete redundant i1 test. Call
32891 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
32893 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
32895 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
32896 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
32898 <riscv64-*-elf>: Re-arrange section
32899 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
32900 <riscv32-*-linux>: Likewise.
32901 <riscv64-*-elf>: Likewise
32902 <riscv64-*-linux>: Likewise.
32904 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
32907 * aarch64.opt(verbose-cost-dump): Fix typo.
32909 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
32912 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
32913 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
32915 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
32917 * reload.c (find_reloads): When reloading a nonoffsettable address,
32918 use RELOAD_OTHER for it and its address reloads.
32920 PR rtl-optimization/79910
32921 * combine.c (record_used_regs): New static function.
32922 (try_combine): Handle situations where there is an additional
32923 instruction between I2 and I3 which needs to have a LOG_LINK
32926 2017-03-17 Jeff Law <law@redhat.com>
32928 PR tree-optimization/71437
32929 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
32930 conditional in the hash table first.
32931 (vrp_dom_walker::before_dom_children): Extract condition from
32932 ASSERT_EXPR. Record condition, its inverion and any implied
32933 conditions as well.
32935 2017-03-17 Marek Polacek <polacek@redhat.com>
32936 Markus Trippelsdorf <markus@trippelsdorf.de>
32938 PR tree-optimization/80079
32939 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
32942 2017-03-17 Richard Biener <rguenther@suse.de>
32944 PR middle-end/80075
32945 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
32946 Properly verify the LHS before the RHS possibly claims to be
32948 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
32951 2017-03-17 Martin Jambor <mjambor@suse.cz>
32953 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
32954 (List of -O2 options): Likewise.
32955 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
32958 2017-03-17 Tom de Vries <tom@codesourcery.com>
32960 * gcov-dump.c (print_usage): Print bug_report_url.
32962 2017-03-17 Richard Biener <rguenther@suse.de>
32964 PR middle-end/80050
32965 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
32966 (parser::peek): Likewise.
32968 2017-03-17 Richard Biener <rguenther@suse.de>
32970 PR tree-optimization/80048
32971 * sese.c (free_sese_info): Properly release rename_map and
32972 copied_bb_map elements.
32974 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
32976 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
32977 Add linked-list forward and backlinks. Insert on
32978 construction, remove on destruction.
32979 (class pass_store_merging): Add m_stores_head field.
32980 (pass_store_merging::terminate_and_process_all_chains):
32981 Iterate over m_stores_head list.
32982 (pass_store_merging::terminate_all_aliasing_chains):
32984 (pass_store_merging::execute): Check for debug stmts first.
32985 Push new chains onto the m_stores_head stack.
32987 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
32990 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
32991 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
32992 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
32994 2017-03-16 Jeff Law <law@redhat.com>
32996 PR tree-optimization/71437
32997 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
32998 member function. Implementation moved into after_dom_children
32999 member function and into the threader's thread_outgoing_edges
33001 (dom_opt_dom_walker::after_dom_children): Simplify by moving
33002 some code into new thread_outgoing_edges.
33003 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
33004 definition. Simplify marker handling (do it here). Assume we always
33005 have the available expression and the const/copies tables.
33006 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
33008 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
33009 * tree-vrp.c (equiv_stack): No longer file scoped.
33010 (vrp_dom_walker): New class.
33011 (vrp_dom_walker::before_dom_children): New member function.
33012 (vrp_dom_walker::after_dom_children): Likewise.
33013 (identify_jump_threads): Setup domwalker. Use it rather than
33014 walking edges in a random order by hand. Simplify setup/finalization.
33015 (finalize_jump_threads): Remove.
33016 (vrp_finalize): Do not call identify_jump_threads here.
33017 (execute_vrp): Do it here instead and call thread_through_all_blocks
33020 PR tree-optimization/71437
33021 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
33023 (simplify_stmt_for_jump_threading): Add basic_block argument. All
33025 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
33026 (dom_opt_dom_walker::thread_across_edge): Remove
33027 handle_dominating_asserts argument. All callers changed.
33028 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
33029 changes. Remove calls to lhs_of_dominating_assert. Other
33030 uses of handle_dominating_asserts turn into unconditional code
33031 (simplify_control_stmt_condition_1): Likewise.
33032 (simplify_control_stmt_condition): Likewise.
33033 (thread_through_normal_block, thread_across_edge): Likewise.
33034 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
33035 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
33036 object if it is not an SSA_NAME.
33037 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
33038 before calling into the VRP specific simplifiers.
33039 (identify_jump_threads): Remove handle_dominating_asserts
33042 2017-03-16 Jakub Jelinek <jakub@redhat.com>
33045 * tree-diagnostic.c (default_tree_printer): No longer static.
33046 * tree-diagnostic.h (default_tree_printer): New prototype.
33048 2017-03-16 Tamar Christina <tamar.christina@arm.com>
33050 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
33051 Change ins into fmov.
33053 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33055 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
33056 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
33057 Use h_con constraint for operand 1.
33058 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
33059 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
33061 2017-03-15 Jeff Law <law@redhat.com>
33063 PR tree-optimization/71437
33064 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
33065 (record_temporary_equivalences): Use it.
33067 PR tree-optimization/71437
33068 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
33069 tree-ssa-scopedtables.
33070 (lookup_avail_expr, build_and_record_new_cond): Likewise.
33071 (record_conditions, record_cond, vuse_eq): Likewise.
33072 (record_edge_info): Adjust to API tweak of record_conditions.
33073 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
33074 (record_temporary_equivalences, optimize_stmt): Likewise.
33075 (eliminate_redundant_computations): Likewise.
33076 (record_equivalences_from_stmt): Likewise.
33077 * tree-ssa-scopedtables.c: Include options.h and params.h.
33078 (vuse_eq): New function, moved from tree-ssa-dom.c
33079 (build_and_record_new_cond): Likewise.
33080 (record_conditions): Likewise. Accept vector of conditions rather
33081 than edge_equivalence structure for first argument.
33082 for the first argument.
33083 (avail_exprs_stack::lookup_avail_expr): New member function, moved
33084 from tree-ssa-dom.c.
33085 (avail_exprs_stack::record_cond): Likewise.
33086 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
33087 from tree-ssa-dom.c.
33088 (avail_exprs_stack): Add new member functions lookup_avail_expr
33090 (record_conditions): Declare.
33092 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
33095 * lra-constraints.c (process_alt_operands): Increase reject for
33096 reloading an input/output operand.
33098 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
33101 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
33102 insns to convert from signed/unsigned char/short to IEEE 128-bit
33104 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
33106 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
33109 * config/i386/i386.c (ix86_vector_duplicate_value): Create
33110 subreg of inner mode for values already in registers.
33112 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
33114 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
33115 iteration reg is used after the loop.
33117 2017-03-14 Martin Sebor <msebor@redhat.com>
33119 PR tree-optimization/79800
33120 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
33121 precision in negative-positive range.
33122 (format_floating): Call non-const overload with adjusted precision.
33124 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33127 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
33130 2017-03-14 Martin Sebor <msebor@redhat.com>
33132 PR middle-end/80020
33133 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
33134 * builtins.def (aligned_alloc): Use it.
33137 * Makefile.in (GTFILES): Add calls.c.
33138 * calls.c: Include "gt-calls.h".
33140 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
33142 PR rtl-optimization/79728
33143 * regs.h (struct target_regs): New field
33144 x_contains_allocatable_regs_of_mode.
33145 (contains_allocatable_regs_of_mode): New macro.
33146 * reginfo.c (init_reg_sets_1): Initialize it, and change
33147 contains_reg_of_mode so it includes global regs as well.
33148 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
33149 rather than contains_regs_of_mode.
33151 2017-03-14 Martin Liska <mliska@suse.cz>
33153 * doc/invoke.texi: Document options that can't be combined with
33154 -fcheck-pointer-bounds.
33156 2017-03-14 Martin Liska <mliska@suse.cz>
33158 PR middle-end/79831
33159 * doc/invoke.texi (-Wchkp): Document the option.
33161 2017-03-14 Martin Liska <mliska@suse.cz>
33163 * Makefile.in: Install gcov-dump.
33165 2017-03-14 Martin Liska <mliska@suse.cz>
33167 * multiple_target.c (expand_target_clones): Bail out for
33168 an invalid attribute.
33170 2017-03-14 Richard Biener <rguenther@suse.de>
33172 * alias.c (struct alias_set_entry): Pack properly.
33173 * cfgloop.h (struct loop): Likewise.
33174 * cse.c (struct set): Likewise.
33175 * ipa-utils.c (struct searchc_env): Likewise.
33176 * loop-invariant.c (struct invariant): Likewise.
33177 * lra-remat.c (struct cand): Likewise.
33178 * recog.c (struct change_t): Likewise.
33179 * rtl.h (struct address_info): Likewise.
33180 * symbol-summary.h (function_summary): Likewise.
33181 * tree-loop-distribution.c (struct partition): Likewise.
33182 * tree-object-size.c (struct object_size_info): Likewise.
33183 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
33184 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
33185 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
33186 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
33187 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
33188 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
33189 (struct _stmt_vec_info): Likewise.
33191 2017-03-14 Martin Liska <mliska@suse.cz>
33194 * multiple_target.c (create_dispatcher_calls): Check that
33195 a target can create a function dispatcher.
33197 2017-03-14 Martin Liska <mliska@suse.cz>
33200 * multiple_target.c (expand_target_clones): Drop local.local
33201 flag for default implementation.
33203 2017-03-14 Richard Biener <rguenther@suse.de>
33205 PR tree-optimization/80030
33206 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
33208 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
33210 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
33211 gcc_fallthrough() instead of __attribute__((fallthrough));
33213 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
33215 * doc/gcc.texi: Remove "up" link to (DIR).
33216 * doc/gccint.texi: Ditto.
33218 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
33220 * doc/install.texi (Specific) <avr>: Remove reference to
33223 2017-03-13 Jeff Law <law@redhat.com>
33225 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
33226 attribute rather than comments.
33228 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
33229 match_scratch operand is highest.
33231 2017-03-13 Martin Liska <mliska@suse.cz>
33233 PR middle-end/78339
33234 * ipa-pure-const.c (warn_function_noreturn): If the declarations
33235 is a CHKP clone, use original declaration.
33237 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33239 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
33240 (arc_conditional_register_usage): Use a different allocation order
33241 when optimizing for size.
33242 * common/config/arc/arc-common.c (arc_option_optimization_table):
33243 Section anchors default on when optimizing for size.
33245 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33247 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
33249 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33251 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
33252 * config/arc/arc.md (cpu_facility): Add cd variant.
33253 (*movqi_insn): Add code density variant.
33254 (*movhi_insn): Likewise.
33255 (*movqi_insn): Likewise.
33256 (*addsi3_mixed): Likewise.
33257 (subsi3_insn): Likewise.
33259 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33261 * config/arc/arc.md (movsi_cond_exec): Update constraint.
33263 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33265 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
33266 expressions with MINUS and UNARY ops.
33268 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33271 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
33273 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
33274 between vec_select and vector argument.
33275 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
33276 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
33277 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
33278 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
33279 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
33280 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
33282 2017-03-13 Richard Biener <rguenther@suse.de>
33285 * params.def (vect-max-peeling-for-alignment): Fix typo.
33287 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33289 * doc/install.texi (Specific) <mips-*-*>: Remove description of
33290 issue that only occurred with binutils below 2.18.
33292 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33294 * doc/install.texi (Specific) <cris-axis-elf>: No longer
33295 refer to binutils 2.11/2.12 minimum.
33297 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33299 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
33300 ftp.kernel.org and simplify binutils requirement.
33302 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
33304 * doc/invoke.texi (Warning Options): Fix spelling of link-time
33306 (Optimize Options): Ditto. Also remove redundancy.
33308 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33310 PR translation/79848
33311 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
33313 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
33314 to G_ to avoid double translation.
33316 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33318 PR translation/79923
33319 * auto-profile.c (get_combined_location): Convert leading
33320 character of diagnostics to lower case and remove trailing period.
33321 (read_profile): Likewise for various diagnostics.
33322 * config/arm/arm.c (arm_option_override): Remove trailing period
33323 from various diagnostics.
33324 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
33325 (msp430_expand_delay_cycles): Likewise.
33327 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33330 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
33331 full command-line argument, rather than just "str".
33332 (aarch64_validate_march): Likewise.
33333 (aarch64_validate_mtune): Likewise.
33335 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
33337 PR rtl-optimization/78911
33338 * lra-assigns.c (must_not_spill_p): New function.
33339 (spill_for): Use it.
33341 2017-03-10 Jakub Jelinek <jakub@redhat.com>
33343 PR tree-optimization/79981
33344 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
33345 ATOMIC_COMPARE_EXCHANGE ifn result.
33346 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
33347 IFN_ATOMIC_COMPARE_EXCHANGE.
33349 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33352 * opts.c (parse_sanitizer_options): Add missing question mark to
33353 "did you mean" message.
33355 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33357 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
33359 (VMULEUH_UNS): Likewise.
33360 (VMULOUB_UNS): Likewise.
33361 (VMULOUH_UNS): Likewise.
33362 * config/rs6000/rs6000.c (builtin_function_type): Remove
33363 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
33365 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33368 * read-rtl-function.c (function_reader::read_rtx_operand): Update
33369 x with result of extra_parsing_for_operand_code_0.
33370 (function_reader::extra_parsing_for_operand_code_0): Convert
33371 return type from void to rtx, returning x. When reading
33372 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
33373 larger size containing struct block_symbol.
33375 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
33377 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
33378 -mfloat128-hardware without -m64.
33380 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
33383 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
33384 entries to the case statement that marks unsigned arguments to
33385 overloaded functions.
33387 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33389 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
33390 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
33392 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
33395 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
33396 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
33398 2017-03-10 Martin Liska <mliska@suse.cz>
33402 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
33403 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
33406 2017-03-10 Olivier Hainque <hainque@adacore.com>
33408 * tree-switch-conversion (array_value_type): Start by resetting
33409 candidate type to it's main variant.
33411 2017-03-10 Jakub Jelinek <jakub@redhat.com>
33413 PR rtl-optimization/79909
33414 * combine.c (try_combine): Use simplify_replace_rtx on individual
33415 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
33416 of the whole CALL_INSN_FUNCTION_USAGE.
33418 PR tree-optimization/79972
33419 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
33420 get_range_info on SSA_NAMEs. Formatting fixes.
33422 2017-03-10 Richard Biener <rguenther@suse.de>
33423 Jakub Jelinek <jakub@redhat.com>
33425 PR tree-optimization/77975
33426 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
33427 edge to be constant.
33428 (get_val_for): For constant x return it. Formatting fix.
33429 (loop_niter_by_eval): Avoid pointless looping if the next iteration
33430 would use the same bases as the current one.
33432 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33434 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
33435 instead of vec_select for V1TImode.
33436 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
33438 (VSX_LE_128): Add V1TI to this mode iterator.
33439 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
33440 (*vsx_le_perm_store_<mode>): Likewise.
33441 (pre-reload splitter for VSX stores): Likewise.
33442 (post-reload splitter for VSX stores): Likewise.
33443 (*vsx_xxpermdi2_le_<mode>): Likewise.
33444 (*vsx_lxvd2x2_le_<mode>): Likewise.
33445 (*vsx_stxvd2x2_le_<mode>): Likewise.
33447 2017-03-09 Michael Eager <eager@eagercon.com>
33449 Correct failures with --enable-checking=yes,rtl.
33451 * config/microblaze/microblaze.c (microblaze_expand_shift):
33452 Replace GET_CODE test with CONST_INT_P and INTVAL test with
33453 test for const0_rtx.
33454 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
33455 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
33457 2017-03-09 Richard Biener <rguenther@suse.de>
33459 PR tree-optimization/79977
33460 * graphite-scop-detection.c (scop_detection::merge_sese):
33461 Handle the case of extra exits to blocks dominating the entry.
33463 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
33465 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
33468 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
33470 PR rtl-optimization/79949
33471 * lra-constraints.c (process_alt_operands): Check memory when
33472 trying to predict a cycle. Print about the overall increase.
33474 2017-03-09 Richard Biener <rguenther@suse.de>
33476 PR middle-end/79971
33477 * gimple-expr.c (useless_type_conversion_p): Preserve
33478 TYPE_SATURATING for fixed-point types.
33480 2017-03-09 Richard Biener <rguenther@suse.de>
33483 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
33484 alignment of BLKmode params.
33486 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33489 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
33490 (VALL_NO_V2Q): Likewise.
33492 * config/aarch64/aarch64-simd.md
33493 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
33495 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
33496 VALL_NO_V2Q mode iterator.
33497 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
33499 2017-03-09 Martin Liska <mliska@suse.cz>
33501 PR tree-optimization/79631
33502 * tree-chkp-opt.c (chkp_is_constant_addr): Call
33503 tree_int_cst_sign_bit just for INTEGER constants.
33505 2017-03-09 Martin Liska <mliska@suse.cz>
33509 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
33512 2017-03-09 Marek Polacek <polacek@redhat.com>
33515 * tree.c (inchash::add_expr): Handle TREE_VEC.
33517 2017-03-09 Martin Liska <mliska@suse.cz>
33520 (chkp_narrow_size_and_offset): New function.
33521 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
33522 (void chkp_parse_bit_field_ref): New function.
33523 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
33524 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
33526 2017-03-09 Martin Liska <mliska@suse.cz>
33529 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
33530 (chkp_find_bounds_1): Remove gcc_unreachable.
33532 2017-03-09 Jakub Jelinek <jakub@redhat.com>
33535 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
33536 BUILT_IN_SYNC*, determine the access type from the size suffix and
33537 always build a MEM_REF with that type. Handle forgotten
33538 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
33541 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
33542 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
33543 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
33544 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
33545 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
33546 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
33547 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
33548 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
33549 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
33550 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
33551 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
33552 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
33553 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
33554 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
33555 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
33556 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
33557 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
33558 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
33559 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
33560 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
33561 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
33562 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
33563 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
33564 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
33565 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
33566 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
33567 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
33568 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
33569 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
33570 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
33571 definitions outside of __OPTIMIZE__ guarded section.
33574 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
33575 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
33576 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
33577 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
33580 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33582 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
33583 ("vfenez<mode>"): Add missing constraints.
33585 2017-03-08 Martin Sebor <msebor@redhat.com>
33588 * config/nds32/nds32.c (nds32_option_override):
33589 Fix misspelled diagnostic.
33591 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33594 * gimplify.c (gimplify_omp_for): Replace index var in outer
33595 taskloop statement with an artificial variable and add
33596 OMP_CLAUSE_PRIVATE clause for it.
33598 2017-03-08 Richard Biener <rguenther@suse.de>
33600 PR tree-optimization/79955
33601 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
33602 for accesses that are completely outside of the variable.
33604 2017-03-08 Andrew Haley <aph@redhat.com>
33606 PR tree-optimization/79943
33607 * tree-ssa-loop-split.c (compute_new_first_bound): When
33608 calculating the new upper bound, (END-BEG) should be added, not
33611 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33613 * config/avr/avr.md (setmemhi): Make sure match_dup
33614 operand number comes before match_scratch.
33616 2017-03-08 Richard Biener <rguenther@suse.de>
33618 PR tree-optimization/79920
33619 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
33620 with ncopies == 1 to ...
33621 (vect_transform_slp_perm_load): ... here. Properly compute
33622 all element loads by iterating VF times over the group. Do
33623 not handle ncopies (computed in a broken way) in
33624 vect_create_mask_and_perm.
33626 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33629 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
33630 is a uniform vector, use uniform_vector_p return value instead of
33631 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
33633 2017-03-07 Marek Polacek <polacek@redhat.com>
33635 PR middle-end/79809
33636 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
33637 (alloca_call_type): Likewise.
33639 2017-03-07 Martin Liska <mliska@suse.cz>
33641 * gcov.c (process_args): Put comment to correct location.
33643 2017-03-07 Martin Liska <mliska@suse.cz>
33645 PR middle-end/68270
33646 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
33647 Use array_at_struct_end_p instead of DECL_CHAIN (field).
33648 (chkp_narrow_bounds_for_field): Likewise.
33649 (chkp_parse_array_and_component_ref): Pass one more argument to
33652 2017-03-07 Richard Biener <rguenther@suse.de>
33654 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
33657 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
33659 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
33660 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
33662 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33665 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
33666 to end of description.
33667 (PARAM_MAX_STORES_TO_MERGE): Likewise.
33669 2017-03-07 Jakub Jelinek <jakub@redhat.com>
33671 PR rtl-optimization/79901
33672 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
33674 (*avx512f_<code><mode>3<mask_name>): ... this.
33675 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
33676 iterator instead of VI8_AVX2_AVX512BW.
33678 PR rtl-optimization/79901
33679 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
33680 min/max expander, expand it using expand_vec_cond_expr.
33683 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
33686 2017-03-06 Jakub Jelinek <jakub@redhat.com>
33689 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
33690 to void * for PCH reasons.
33691 * dwarf2out.c (output_loc_operands, output_die): Cast
33692 v.val_vec.array to unsigned char *.
33694 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
33697 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
33700 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
33702 PR rtl-optimization/79571
33703 * lra-constraints.c (process_alt_operands): Calculate static
33704 reject and subtract it from overall when only addresses will be
33707 2017-03-06 Julia Koval <julia.koval@intel.com>
33710 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
33711 incoming stack boundary to 128 for 64-bit targets.
33713 2017-03-06 Richard Biener <rguenther@suse.de>
33715 PR tree-optimization/79894
33716 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
33717 to NULL after folding it.
33719 2017-03-06 Richard Biener <rguenther@suse.de>
33721 PR tree-optimization/79824
33722 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
33723 check disabling peeling for gaps.
33725 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
33727 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
33728 attributes): Document gettimeofday.
33730 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
33732 * config/s390/s390.c (s390_option_override_internal): Set
33733 PARAM_MIN_VECT_LOOP_BOUND
33735 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
33737 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
33738 * config/s390/s390.md: Likewise.
33740 2017-03-06 Jakub Jelinek <jakub@redhat.com>
33743 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
33744 (<avx2_avx512>_perm<mode>): Rename to ...
33745 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
33747 (<avx512>_perm<mode>_mask): Rename to ...
33748 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
33750 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
33751 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
33752 instead of VI8F_256_512.
33753 (avx512f_perm<mode>): New define_expand.
33754 (avx512f_perm<mode>_mask): Likewise.
33755 (avx512f_perm<mode>_1<mask_name>): New define_insn.
33756 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
33758 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33760 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
33761 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
33763 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
33765 2017-03-06 Martin Liska <mliska@suse.cz>
33768 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
33769 when having a SSA NAME w/o VAR_DECL assigned to it.
33771 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33773 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
33774 msa_dpsub_<su>_d): Fix MODE for vec_select.
33776 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33778 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
33780 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
33782 2017-03-06 Richard Biener <rguenther@suse.de>
33784 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
33785 * plugin.c (register_plugin_info): Likewise.
33786 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
33788 2017-03-05 Jakub Jelinek <jakub@redhat.com>
33790 * config/i386/sse.md (sse_storehps, sse_storelps,
33791 avx_<castmode><avxsizesuffix>_<castmode>,
33792 avx512f_<castmode><avxsizesuffix>_<castmode>,
33793 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
33794 in condition that at least one operand is not a MEM.
33796 2017-03-03 Jakub Jelinek <jakub@redhat.com>
33798 PR middle-end/79805
33799 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
33800 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
33802 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
33803 gimple_call_nothrow_p flag based on whether original builtin can throw.
33804 If it can, emit following stmts on the fallthrough edge.
33805 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
33806 don't create new bb if inserting just debug stmts on the edge, try to
33807 insert them on the fallthru bb or just reset debug stmts.
33809 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
33812 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
33813 restore recog_data (including the operand rtxes inside it) around
33814 the call to get_insn_template.
33816 2017-03-03 Martin Sebor <msebor@redhat.com>
33818 PR tree-optimization/79699
33819 * context.c (context::~context): Free MPFR caches to avoid
33820 a memory leak on program exit.
33822 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33824 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
33825 Use wide_int::ulow () instead of .elt (0).
33827 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
33829 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
33830 (*pushxf): Limit oF constraint to 32bit targets and add oC
33831 constraint for 64bit targets.
33832 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
33833 (*pushdf): Change rmF constraint to rmC.
33835 2017-03-03 Martin Liska <mliska@suse.cz>
33837 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
33838 Remove unused variable.
33840 2017-03-03 Jakub Jelinek <jakub@redhat.com>
33843 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
33844 is a memory operand, increase num_memory.
33845 (ix86_expand_args_builtin): Likewise.
33847 2017-03-03 Jan Hubicka <jh@suse.cz>
33850 * ipa-devirt.c (maybe_record_node): Properly handle
33851 __cxa_pure_virtual visibility.
33853 2017-03-03 Martin Liska <mliska@suse.cz>
33855 PR tree-optimization/79803
33856 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
33858 (pass_loop_prefetch::execute): Disabled optimization if an
33859 assumption about L1 cache size is not met.
33861 2017-03-03 Martin Liska <mliska@suse.cz>
33863 PR rtl-optimization/79574
33864 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
33865 (hash_scan_set): Likewise.
33866 (dump_hash_table): Likewise.
33867 (hoist_code): Likewise.
33869 2017-03-03 Richard Biener <rguenther@suse.de>
33871 * fixed-value.c (fixed_from_string): Restore use of elt (1)
33872 in place of uhigh ().
33873 (fixed_convert_from_real): Likewise.
33875 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
33878 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
33880 2017-03-03 Richard Biener <rguenther@suse.de>
33882 PR middle-end/79818
33883 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
33884 TYPE_OVERFLOW_UNDEFINED check.
33886 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33888 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
33890 (vector_ae_<mode>_p): Likewise.
33891 (vector_nez_<mode>_p): Likewise.
33892 (vector_ne_v2di_p): Likewise.
33893 (vector_ae_v2di_p): Likewise.
33894 (vector_ne_<mode>_p): Likewise.
33895 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
33897 (vsx_tsqrt<mode>2_fe): Likewise.
33899 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
33902 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
33904 2017-03-02 Jakub Jelinek <jakub@redhat.com>
33906 PR rtl-optimization/79780
33907 * cprop.c (one_cprop_pass): When second and further conditional trap
33908 in a single basic block is turned into an unconditional trap, turn it
33909 into a deleted note to avoid RTL verification failures.
33911 2017-03-02 Richard Biener <rguenther@suse.de>
33913 * fold-const.c (const_binop): Use ulow () instead of elt (0).
33915 2017-03-02 Richard Biener <rguenther@suse.de>
33917 PR tree-optimization/79345
33919 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
33920 param and abort the walk, returning -1 if it is hit.
33921 (walk_aliased_vdefs): Take a limit param and pass it on.
33922 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
33923 defaulting to 0 and return a signed int.
33924 * tree-ssa-uninit.c (struct check_defs_data): New struct.
33925 (check_defs): New helper.
33926 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
33927 about uninitialized memory.
33928 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
33929 bogus uninitialized warning.
33930 (fixed_convert_from_real): Likewise.
33932 2017-03-02 Bin Cheng <bin.cheng@arm.com>
33934 PR tree-optimization/66768
33935 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
33936 iv_use if base object can't be determined.
33938 2017-03-02 Jakub Jelinek <jakub@redhat.com>
33940 PR tree-optimization/79345
33941 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
33942 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
33943 (get_pattern_stats): Initialize it.
33944 * genemit.c (gen_expand): Verify match_scratch numbers come after
33945 match_operand/match_dup numbers.
33946 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
33947 match_scratch numbers.
33948 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
33950 * config/s390/s390.md (trunctdsd2): Likewise.
33952 2017-03-02 Richard Biener <rguenther@suse.de>
33954 * wide-int.h (wide_int_storage::operator=): Implement in terms
33957 2017-03-02 Richard Biener <rguenther@suse.de>
33959 PR tree-optimization/79777
33960 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
33961 the to insert expression to sth existing.
33963 2017-03-01 Martin Sebor <msebor@redhat.com>
33965 PR middle-end/79692
33966 * gimple-ssa-sprintf.c
33967 (directive::known_width_and_precision): New function.
33968 (format_integer): Use it.
33969 (get_mpfr_format_length): Consider the full range of precision
33970 when computing %g output with the # flag. Set the likely byte
33971 count to 3 rather than 1 when precision is indeterminate.
33972 (format_floating): Correct the lower bound of precision.
33974 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33976 * doc/invoke.texi: Document default code model for 64-bit Linux.
33978 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33981 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
33982 udiv rather than div since input pattern is unsigned.
33984 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
33986 * config/i386/i386.c (print_reg): Warn for values of
33987 unsupported size in integer register.
33989 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
33992 * config/rs6000/predicates.md (current_file_function_operand): Do
33993 not allow self calls to be local if the function is replaceable.
33995 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
33998 * config/rs6000/altivec.h (vec_ctz and others): Change the
33999 preprocessor macro that controls conditional compilation from
34000 _ARCH_PWR9 to __POWER9_VECTOR__.
34001 (vec_all_ne): Change parameterization of __altivec_scalar_pred
34002 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
34003 control (instead of _ARCH_PWR9 control) so that template
34004 definition uses power9-specific function.
34005 (vec_any_eq): Likewise.
34006 (vec_all_ne): Change macro definition to use a power9-specific
34007 expansion under #ifdef __POWER9_VECTOR__ control (instead of
34008 _ARCH_PWR9 control).
34009 (vec_any_eq) Likewise.
34010 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
34011 expansion for CMPNEF to remove support for xvcmpnesp instruction.
34012 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
34013 support for xvcmpnedp instruction.
34014 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
34015 macro expansion so that Power9 implementation of vec_all_ne does
34016 not use the AltiVec predicate framework.
34017 (VCMPNEH_P): Likewise.
34018 (VCMPNEW_P): Likewise.
34019 (VCMPNED_P): Likewise.
34020 (VCMPNEFP_P): Likewise.
34021 (VCMPNEDP_P): Likewise.
34022 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
34023 implementation of vec_any_eq to not use AltiVec predicate
34025 (VCMPAEH_P): Likewise.
34026 (VCMPAEW_P): Likewise.
34027 (VCMPAED_P): Likewise.
34028 (VCMPAEFP_P): Likewise.
34029 (VCMPAEDP_P): Likewise.
34030 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
34031 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
34032 not use the AltiVec predicate framework.
34033 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
34034 of vec_any_eq to not use AltiVec predicate framework.
34035 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
34036 support for predefined __POWER9_VECTOR__ macro to indicate that
34037 Power9 instruction selection is enabled.
34038 (altivec_overloaded_builtins): Remove extraneous
34039 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
34040 function argument types RS6000_BTI_bool_V16QI and
34041 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
34042 entry for overloaded function argument types RS6000_BTI_bool_V4SI
34043 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
34044 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
34045 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
34046 Power9 for implementations of vec_cmpne. Change the signature for
34047 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
34048 (representing vec_all_ne) to remove the previously described first
34049 argument of type RS6000_BTI_INTSI, as this was an artifact of
34050 reliance on the AltiVec predicate framework, which is no longer
34051 used in the implementation of these functions. Add
34052 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
34053 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
34054 since, unlike the AltiVec predicate framework implementation, we
34055 do not share function descriptors between vec_alle and vec_anyeq.
34056 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
34057 set of modes that receive special treatment even when
34058 TARGET_P9_VECTOR is true. The special treatment emits code that
34059 does not depend on Power9 instructions.
34060 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
34061 define_expand to not rely on AltiVec predicate framework.
34062 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
34064 (vector_ne_v2di_p): Change this define_expand to not rely on
34065 AltiVec predicate framework.
34066 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
34068 (vector_ne_<mode>_p): Change this define_expand to not rely on
34069 AltiVec predicate framework.
34070 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
34072 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
34073 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
34074 define_insn pattern.
34075 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
34076 define_insn pattern because the xvcmpne<VSs>. instruction is not
34078 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
34079 instruction is not supported.
34081 2017-03-01 Jakub Jelinek <jakub@redhat.com>
34083 * config/nvptx/nvptx.c: Include intl.h.
34085 2017-03-01 Martin Jambor <mjambor@suse.cz>
34088 * ipa-prop.h (ipa_bits): Removed field known.
34089 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
34090 to pointers. Adjusted their comments to warn about their sharing.
34091 (ipcp_transformation_summary): Change bits to a vector of pointers.
34092 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
34093 (ipa_get_ipa_bits_for_value): Declare.
34094 * tree-vrp.h (value_range): Mark as GTY((for_user)).
34095 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
34096 (ipa_bits_hash_table): Likewise.
34097 (ipa_vr_ggc_hash_traits): Likewise.
34098 (ipa_vr_hash_table): Likewise.
34099 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
34100 being pointers and vr_known being removed.
34101 (ipa_set_jf_unknown): Likewise.
34102 (ipa_get_ipa_bits_for_value): New function.
34103 (ipa_set_jfunc_bits): Likewise.
34104 (ipa_get_value_range): New overloaded functions.
34105 (ipa_set_jfunc_vr): Likewise.
34106 (ipa_compute_jump_functions_for_edge): Use the above functions to
34107 construct bits and vr parts of jump functions.
34108 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
34109 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
34111 (ipcp_grow_transformations_if_necessary): Also allocate
34112 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
34114 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
34115 them. Fix too long lines.
34116 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
34117 vr_known being removed.
34118 (ipa_read_jump_function): Use new setter functions to construct bits
34119 and vr parts of jump functions or set them to NULL.
34120 (write_ipcp_transformation_info): Adjust for bits being pointers.
34121 (read_ipcp_transformation_info): Likewise.
34122 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
34124 Include gt-ipa-prop.h.
34125 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
34127 (ipcp_store_bits_results): Likewise.
34128 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
34129 Do not write to existing jump functions but use a temporary instead.
34131 2017-03-01 Jakub Jelinek <jakub@redhat.com>
34134 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
34135 attempt to use its first operand as BIT_FIELD_REF base.
34137 2017-03-01 Richard Biener <rguenther@suse.de>
34139 PR middle-end/79721
34140 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
34141 interpolating formula in wrapping arithmetic.
34142 (chrec_apply): Convert chrec_evaluate return value to wanted type.
34144 2017-03-01 Jakub Jelinek <jakub@redhat.com>
34146 PR tree-optimization/79734
34147 * tree-vect-generic.c (expand_vector_condition): Optimize
34148 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
34149 Handle VEC_COND_EXPR where comparison has different inner width from
34150 type's inner width.
34152 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
34154 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
34155 markup, and similar issues. Remove @opindex entries for things
34156 that aren't options. Add missing -mmpy-option entries.
34158 2017-02-28 Jakub Jelinek <jakub@redhat.com>
34160 PR tree-optimization/79737
34161 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
34162 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
34163 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
34164 instead of byte_size. Formatting fix.
34165 (shift_bytes_in_array_right): Formatting fix.
34167 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
34170 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
34171 condition on optimize for the leaf function test.
34173 2017-02-28 Martin Liska <mliska@suse.cz>
34176 * read-rtl-function.c (function_reader::handle_unknown_directive):
34177 Bail out when one uses -flto.
34179 2017-02-28 Martin Liska <mliska@suse.cz>
34181 * common.opt: Replace space with tabular for options of <number>
34183 * config/i386/i386.opt: Show <number> value for
34184 -mlarge-data-threshold.
34185 * opts.c (print_filtered_help): Do not display number in hexadecimal
34188 2017-02-28 Martin Liska <mliska@suse.cz>
34190 * common.opt: Fix --help=option -Q for options which are of
34193 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
34195 * config/i386/i386.c (print_reg): Error out for values
34196 of 8-bit size in invalid integer register.
34198 2017-02-28 Martin Sebor <msebor@redhat.com>
34200 PR tree-optimization/79691
34201 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
34203 2017-02-28 Jakub Jelinek <jakub@redhat.com>
34206 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
34207 gcc_unreachable with output_operand_lossage.
34209 2017-02-28 Richard Biener <rguenther@suse.de>
34211 PR tree-optimization/79740
34212 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
34214 (visit_nary_op): Insert the nary into the hashtable if we
34215 pattern-matched sth.
34216 * tree-ssa-pre.c (eliminate_insert): Robustify.
34218 2017-02-28 Richard Biener <rguenther@suse.de>
34220 PR middle-end/79731
34221 * fold-const.c (decode_field_reference): Reject out-of-bound
34224 2017-02-28 Jakub Jelinek <jakub@redhat.com>
34226 * config/i386/i386.c: Include intl.h.
34227 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
34228 instead of just cond ? "..." : "...".
34229 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
34230 * coverage.c (read_counts_file): Likewise.
34231 * omp-offload.c: Include intl.h.
34232 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
34233 of just cond ? "..." : "...".
34234 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
34235 of just cond ? "..." : "...".
34237 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
34240 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
34242 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
34243 'tune for' CPU name.
34244 * config/arm/arm-cpu-data.h: Regenerated.
34246 2017-02-28 Richard Biener <rguenther@suse.de>
34248 PR tree-optimization/79732
34249 * tree-inline.c (expand_call_inline): Do not shadow var.
34251 2017-02-28 Richard Biener <rguenther@suse.de>
34253 PR tree-optimization/79723
34254 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
34255 address-space properly.
34257 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
34259 * doc/optinfo.texi (Optimization groups): Fix option used for
34261 * doc/invoke.texi (-fopt-info): Document "omp".
34262 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
34263 (OPTGROUP_ALL): Add OPTGROUP_OMP.
34264 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
34265 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
34266 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
34268 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
34270 * dumpfile.c (optgroup_options): Instead of "openmp", associate
34271 OPTGROUP_OMP with "omp".
34273 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
34276 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
34277 for arithmetic shift of unsigned V2DI.
34279 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
34281 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
34282 arc/linux.h headers.
34283 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
34284 (LINK_SPEC): Likewise.
34285 (ARC_TLS_EXTRA_START_SPEC): Likewise.
34286 (EXTRA_SPECS): Likewise.
34287 (STARTFILE_SPEC): Likewise.
34288 (ENDFILE_SPEC): Likewise.
34289 (LIB_SPEC): Likewise.
34290 (TARGET_SDATA_DEFAULT): Likewise.
34291 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
34292 (MULTILIB_DEFAULTS): Likewise.
34293 (DWARF2_UNWIND_INFO): Likewise.
34294 * config/arc/big.h: New file.
34295 * config/arc/elf.h: Likewise.
34296 * config/arc/linux.h: Likewise.
34297 * config/arc/t-uClibc: Remove.
34299 2017-02-27 Bin Cheng <bin.cheng@arm.com>
34301 PR tree-optimization/77536
34302 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
34303 (tree_transform_and_unroll_loop): Use above function to compute the
34304 estimated niter of unrolled loop and use it when scaling profile.
34305 Also use count info rather than frequency if it's non-zero.
34306 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
34307 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
34308 (vect_transform_loop): Call above function.
34310 2017-02-27 Richard Biener <rguenther@suse.de>
34312 PR tree-optimization/45397
34313 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
34314 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
34315 (visit_nary_op): Add pattern matching for CSEing sign-changed
34316 or truncated operations with wider ones.
34318 2017-02-27 Richard Biener <rguenther@suse.de>
34320 PR tree-optimization/79690
34321 * tree-vect-stmts.c (vectorizable_store): Use vector type
34322 built from the DR with address-space.
34324 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
34326 * doc/invoke.texi (Optimize Options): Refine the description
34327 of asan-use-after-return.
34329 2017-02-25 Alan Modra <amodra@gmail.com>
34331 PR rtl-optimization/79584
34332 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
34333 base, not ad->base_term, the reg within base. Remove assertion
34334 that ad->base == ad->base_term. Replace gen_int_mode using
34335 bogus mode with const0_rtx.
34337 2017-02-25 Jakub Jelinek <jakub@redhat.com>
34339 PR middle-end/79396
34340 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
34341 FMA_EXPR like tcc_binary or tcc_unary.
34343 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
34346 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
34348 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
34349 (output_loc_operands): Handle DW_OP_call_ref and
34350 DW_OP_GNU_variable_value.
34351 (struct variable_value_struct): New type.
34352 (struct variable_value_hasher): Likewise.
34353 (variable_value_hash): New variable.
34354 (string_types): Remove.
34355 (copy_loc_descr): New function.
34356 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
34357 (prepend_loc_descr_to_each): New function.
34358 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
34359 instead of add_loc_descr_to_each if the first argument is single
34360 location list and the second has multiple.
34361 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
34362 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
34363 when looking for variable value which doesn't have other location info.
34364 (loc_list_from_tree): Formatting fix.
34365 (gen_array_type_die): Simplify DW_AT_string_length handling.
34366 (adjust_string_types): Remove.
34367 (gen_subprogram_die): Don't call adjust_string_types nor test/set
34368 string_types. Call resolve_variable_values.
34369 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
34370 (resolve_addr_in_expr): Likewise. Add A argument.
34371 (copy_deref_exprloc): Remove deref argument. Adjust for the
34372 original expression being DW_OP_GNU_variable_value with optionally
34373 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
34374 optionally after it.
34375 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
34376 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
34377 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
34378 (variable_value_hasher::hash, variable_value_hasher::equal): New
34380 (resolve_variable_value_in_expr, resolve_variable_value,
34381 resolve_variable_values, note_variable_value_in_expr,
34382 note_variable_value): New functions.
34383 (dwarf2out_early_finish): Call note_variable_value on all toplevel
34386 2017-02-24 Jakub Jelinek <jakub@redhat.com>
34389 * opts.h (handle_generated_option): Add GENERATED_P argument.
34390 * opts-common.c (handle_option): Adjust function comment.
34391 (handle_generated_option): Add GENERATED_P argument, pass it to
34393 (control_warning_option): Pass false to handle_generated_option
34395 * opts.c (maybe_default_option): Pass true to handle_generated_option
34397 * optc-gen.awk: Likewise.
34399 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34401 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
34402 a REG, look at the REG it is a SUBREG of.
34403 (splitter for cmpeqsi_t): Ditto.
34405 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34407 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
34408 the special USEs with the pattern of the insn, not the insn itself.
34410 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
34413 * doc/invoke.texi: Document -mload-store-pairs.
34415 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34416 Sandra Loosemore <sandra@codesourcery.com>
34418 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
34419 argument isn't a CONST_INT.
34420 (nios2_alternate_compare_const): Assert op is a CONST_INT.
34421 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
34422 (nios2_validate_compare): Bypass alternate compare logic if *op2
34423 is not a CONST_INT.
34424 (ldstwm_operation_p): Return false if first_base is not a REG or
34425 if first_offset is not a CONST_INT.
34427 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34429 * config/cris/cris.md: Use correct operand in a define_peephole2.
34431 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34433 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
34435 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34437 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
34438 this_insn if it is an INSN or JUMP_INSN.
34439 (force_offsettable): Look at base, not at addr.
34440 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
34441 on things that aren't necessarily CONST_INTs.
34443 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
34445 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
34446 -mfpmath=sse is the default also for x86-32 targets with SSE2
34447 instruction set when @option{-ffast-math} is enabled
34449 2017-02-24 Jeff Law <law@redhat.com>
34451 PR rtl-optimizatoin/79286
34452 * ira.c (update_equiv_regs): Drop may_trap_p exception to
34455 2017-02-24 Richard Biener <rguenther@suse.de>
34457 PR tree-optimization/79389
34458 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
34461 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
34463 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
34464 function comment to reflect reality.
34465 (loop_exits_before_overflow): Fix typo in function description.
34467 2017-02-24 Richard Biener <rguenther@suse.de>
34469 PR tree-optimization/79389
34470 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
34471 properly that a threading opportunity exists. Detect conditional
34472 copy/constant propagation opportunities.
34474 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
34476 * config/visium/visium.md (type): Add trap.
34477 (b): New mode attribute.
34478 (*btst): Rename into...
34479 (*btst<mode>): ...this and adjust.
34480 (*cbranchsi4_btst_insn): Rename into...
34481 (*cbranch<mode>4_btst_insn): ...this and adjust.
34482 (trap): New define_insn.
34484 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34486 PR tree-optimization/79389
34487 * ifcvt.c (struct noce_if_info): Add rev_cond field.
34488 (noce_reversed_cond_code): New function.
34489 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
34490 reversed_comparison_code. Formatting fix.
34491 (noce_try_store_flag): Test rev_cond != NULL in addition to
34492 reversed_comparison_code.
34493 (noce_try_store_flag_constants): Likewise.
34494 (noce_try_store_flag_mask): Likewise.
34495 (noce_try_addcc): Use rev_cond if non-NULL instead of
34496 reversed_comparison_code.
34497 (noce_try_cmove_arith): Likewise. Formatting fixes.
34498 (noce_try_minmax, noce_try_abs): Clear rev_cond.
34499 (noce_find_if_block): Initialize rev_cond.
34500 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
34501 instead of false as last argument never attempt to reverse it
34504 2017-02-23 Bin Cheng <bin.cheng@arm.com>
34506 PR tree-optimization/79663
34507 * tree-predcom.c (combine_chains): Process refs in reverse order
34508 only for ZERO length chains, and add explaining comment.
34510 2017-02-23 Jeff Law <law@redhat.com>
34512 PR tree-optimization/79578
34513 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
34514 in call to operand_equal_p.
34516 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
34519 * config/i386/cpuid.h: Fix another undefined behavior.
34521 2017-02-23 Richard Biener <rguenther@suse.de>
34523 PR tree-optimization/79683
34524 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
34525 vector types for data-refs.
34527 2017-02-23 Martin Liska <mliska@suse.cz>
34529 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
34531 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34533 PR middle-end/79665
34534 * internal-fn.c (get_range_pos_neg): Moved to ...
34535 * tree.c (get_range_pos_neg): ... here. No longer static.
34536 * tree.h (get_range_pos_neg): New prototype.
34537 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
34538 are known to be in between 0 and signed maximum inclusive, try to
34539 expand both unsigned and signed divmod and use the cheaper one from
34542 2017-02-22 Jeff Law <law@redhat.com>
34544 PR tree-optimization/79578
34545 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
34546 to compare base operands.
34548 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
34551 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
34552 gpc_reg_operand instead of fpr_reg_operand.
34554 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
34556 * config/mips/mips.c (mips_return_in_memory): Force FP
34557 vector types to be returned in memory for o32 ABI.
34559 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34561 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
34562 instead of DW_TAG_member for static data member declarations and don't
34563 set no_linkage_name for static inline data members.
34564 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
34567 2017-02-22 Martin Liska <mliska@suse.cz>
34569 * doc/invoke.texi: Replace inequality signs with square brackets
34572 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
34575 * lra-constraints.c (simplify_operand_subreg): Handle
34576 WORD_REGISTER_OPERATIONS targets.
34578 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34581 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
34582 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
34583 elimination by swapping fld*.
34585 2017-02-22 Richard Biener <rguenther@suse.de>
34587 PR tree-optimization/79673
34588 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
34589 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
34590 irrelevant address-space qualifiers and avoiding a
34591 ADDR_SPACE_CONVERT_EXPR from fold_convert.
34593 2017-02-22 Richard Biener <rguenther@suse.de>
34595 PR tree-optimization/79666
34596 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
34597 to not symbolically negate if that may introduce undefined
34600 2017-02-22 Martin Liska <mliska@suse.cz>
34603 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
34604 * data-streamer-out.c (streamer_write_gcov_count_stream):
34606 * value-prof.c (stream_out_histogram_value): Make assert more
34607 precise based on type of counter.
34609 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
34612 * config/i386/i386.md (standard_x87sse_constant_load splitter):
34613 Use nonimmediate_operand instead of memory_operand for operand 1.
34614 (float-extend standard_x87sse_constant_load splitter): Ditto.
34616 2017-02-21 Jeff Law <law@redhat.com>
34618 PR tree-optimization/79621
34619 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
34620 blocks with edges to themselves.
34622 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34625 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
34626 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
34627 Use gimple_call_builtin_p.
34630 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
34631 on temporarily removed DEBUG_INSNs.
34633 PR tree-optimization/79649
34634 * tree-loop-distribution.c (classify_partition): Give up on
34635 non-generic address space loads/stores.
34637 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
34639 * doc/loop.texi (Loop manipulation): Remove nonexistent
34640 tree_ssa_loop_version from the documentation.
34641 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
34643 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34646 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
34647 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
34648 * config/rs6000/rs6000.c: Include except.h.
34649 (rs6000_expand_split_stack_prologue): Call
34650 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
34652 2017-02-21 Martin Jambor <mjambor@suse.cz>
34655 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
34656 have been analyzed.
34658 2017-02-21 Martin Jambor <mjambor@suse.cz>
34660 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
34661 for backward compatibility only.
34662 * doc/invoke.texi (Option Summary): Remove all references to
34663 -fipa-cp-alignment.
34665 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
34669 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34671 * lra-constraints.c (curr_insn_transform): Handle
34672 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34674 2017-02-21 Martin Liska <mliska@suse.cz>
34676 * config/i386/i386.opt: Replace -masm-dialect with -masm.
34678 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
34680 PR translation/79638
34681 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
34683 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
34686 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
34687 (arm_function_ok_for_sibcall): Return false for an indirect call by
34688 descriptor if all the argument registers are used.
34689 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
34690 alignment of the function.
34692 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34694 PR tree-optimization/61441
34695 * simplify-rtx.c (simplify_const_unary_operation): For
34696 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
34697 the sNaN unmodified.
34699 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34701 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34702 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34703 instead of SYSTEM_HEADER_DIR.
34705 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
34706 Martin Liška <mliska@suse.cz>
34708 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
34709 Fix typos and grammar, use active voice, and clarify.
34711 2017-02-20 Marek Polacek <polacek@redhat.com>
34713 PR middle-end/79537
34714 * gimplify.c (gimplify_expr): Handle unused *&&L;.
34717 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
34719 2017-02-20 Jakub Jelinek <jakub@redhat.com>
34722 * config/i386/i386.c (ix86_expand_builtin): Handle
34723 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
34724 ix86_builtins_isa[fcode].isa as a requirement of those
34725 flags and any other flag in the bitmask.
34726 (ix86_init_mmx_sse_builtins): Use 0 instead of
34727 ~OPTION_MASK_ISA_64BIT as mask.
34728 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
34729 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
34730 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
34731 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
34733 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34736 * lra-constraints.c (split_reg): Check requested split mode
34737 is supported by the register.
34739 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34741 * lra-constraints.c (simplify_operand_subreg): Remove early
34744 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34747 * lra-constraints.c (curr_insn_transform): Tighten condition
34748 for converting SUBREG reloads from OP_OUT to OP_INOUT.
34750 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34753 * lra-constraints.c (curr_insn_transform): Handle
34754 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34756 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
34759 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
34761 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
34763 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
34766 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
34769 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
34771 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
34774 2017-02-18 Jakub Jelinek <jakub@redhat.com>
34776 * final.c (last_columnnum, override_columnnum): New variables.
34777 (final_start_function): Set last_columnnum, pass it to begin_prologue
34778 hook and pass 0 to dwarf2out_begin_prologue.
34779 (final_scan_insn): Update override_columnnum. Pass last_columnnum
34780 to source_line debug hook.
34781 (notice_source_line): Compute last_columnnum and for debug_column_info
34782 return true on column changes.
34783 * debug.h (struct gcc_debug_hooks): Add column argument to
34784 source_line and begin_prologue hooks.
34785 (debug_nothing_int_charstar_int_bool): Remove prototype.
34786 (debug_nothing_int_int_charstar,
34787 debug_nothing_int_int_charstar_int_bool): New prototypes.
34788 (dwarf2out_begin_prologue): Add column argument.
34789 * debug.c (do_nothing_debug_hooks): Adjust source_line and
34790 begin_prologue hooks.
34791 (debug_nothing_int_charstar_int_bool): Remove.
34792 (debug_nothing_int_int_charstar,
34793 debug_nothing_int_int_charstar_int_bool): New functions.
34794 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
34795 through to dwarf2out_source_line.
34796 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
34797 (dwarf2out_source_line): Add column argument, emit it if requested.
34798 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
34800 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
34801 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
34802 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
34803 through to dwarf2out_begin_prologue.
34804 (vmsdbgout_source_line): Add column argument, pass it through to
34805 dwarf2out_source_line.
34806 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
34807 dbxout_source_line caller.
34808 (dbxout_source_line): Add column argument.
34810 * common.opt (gno-column-info, gcolumn-info): New options.
34811 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
34812 (check_die): Also test for multiple DW_AT_decl_column attributes.
34813 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
34814 DW_AT_decl_column if requested.
34815 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
34817 (gen_variable_die): Likewise.
34818 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
34819 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
34822 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
34823 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
34824 (ix86_handle_option): Handle OPT_m3dnowa.
34825 * doc/invoke.texi (-m3dnowa): Document.
34826 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
34827 -m3dnowa instead of -m3dnow -march=athlon.
34830 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
34831 instead of gcc_assert for K, r and R code checks. Formatting fixes.
34833 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34836 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
34837 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
34838 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
34839 generator for vsx_xxpermdi_<mode>_be.
34840 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
34841 force big-endian semantics.
34842 (vsx_xxpermdi_<mode>_be): New define_expand with same
34843 implementation as previous version of vsx_xxpermdi_<mode>.
34845 2017-02-17 Jakub Jelinek <jakub@redhat.com>
34847 PR tree-optimization/79327
34848 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
34849 variable, its initialization and use.
34851 2017-02-17 Julia Koval <julia.koval@intel.com>
34853 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
34854 (OPTION_MASK_ISA_PKU_UNSET): New.
34855 (ix86_handle_option): Handle -mrdpid.
34856 * config/i386/cpuid.h (bit_RDPID): New.
34857 * config/i386/driver-i386.c (host_detect_local_cpu):
34858 Detect RDPID feature.
34859 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
34860 * config/i386/i386-c.c (ix86_target_macros_internal):
34862 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
34863 (ix86_valid_target_attribute_inner_p): Add "rdpid".
34864 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
34865 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
34866 * config/i386/i386.md (define_insn "rdpid"): New.
34867 * config/i386/i386.opt Add -mrdpid.
34868 * config/i386/immintrin.h (_rdpid_u32): New.
34870 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
34872 PR rtl-optimization/79541
34873 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
34874 instead of transforming it into USE.
34876 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
34878 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
34879 If HONOR_SNANS (SFmode) force the input to a register.
34880 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
34881 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
34882 an frsp or similar insn.
34884 2017-02-17 Martin Liska <mliska@suse.cz>
34886 PR rtl-optimization/79577
34887 * params.def (selsched-max-sched-times): Increase minimum to 1.
34889 2017-02-17 Martin Liska <mliska@suse.cz>
34891 PR rtl-optimization/79574
34892 * gcse.c (want_to_gcse_p): Prevent integer overflow.
34894 2017-02-17 Martin Liska <mliska@suse.cz>
34896 PR tree-optimization/79529
34897 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
34898 ssa_defined_default_def_p to handle cases which are implicitly
34900 * tree-ssa.c (ssa_defined_default_def_p): New function.
34901 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
34902 which are implicitly defined.
34903 * tree-ssa.h (ssa_defined_default_def_p): Declare.
34905 2017-02-17 Richard Biener <rguenther@suse.de>
34907 PR middle-end/79576
34908 * params.def (max-ssa-name-query-depth): Limit to 10.
34910 2017-02-17 Richard Biener <rguenther@suse.de>
34912 PR tree-optimization/79552
34913 * tree-ssa-structalias.c (visit_loadstore): Properly verify
34916 2017-02-17 Richard Biener <rguenther@suse.de>
34919 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
34921 2017-02-17 Marek Polacek <polacek@redhat.com>
34923 PR middle-end/79536
34924 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
34925 (fold_negate_expr): New wrapper.
34927 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
34929 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
34930 Correct terminology and de-emphasize pre-standard behavior.
34932 2017-02-16 Alan Modra <amodra@gmail.com>
34934 PR rtl-optimization/79286
34935 * ira.c (def_dominates_uses): New function.
34936 (update_equiv_regs): Don't create an equivalence for insns that
34937 may trap where the register def does not dominate the use.
34939 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
34941 PR rtl-optimization/78127
34942 * lra.c (lra): Call lra_eliminate before finish the loop after
34945 2017-02-16 Richard Biener <rguenther@suse.de>
34947 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
34949 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
34950 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
34951 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
34952 (isl_val_int_from_wi): New function.
34953 (extract_affine_gmp): Rename to ...
34954 (extract_affine_wi): ... this, take a widest_int.
34955 (extract_affine_int): Just wrap extract_affine_wi.
34956 (add_param_constraints): Use isl_val_int_from_wi.
34957 (add_loop_constraints): Likewise, and extract_affine_wi.
34959 2017-02-15 Jeff Law <law@redhat.com>
34961 PR middle-end/79521
34962 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
34963 ira_init_register_move_cost_if_necessary.
34965 2017-02-15 Martin Sebor <msebor@redhat.com>
34967 PR middle-end/32003
34968 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
34969 removed in a prior commit.
34971 2017-02-15 Bin Cheng <bin.cheng@arm.com>
34973 PR tree-optimization/79347
34974 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
34975 counters during peeling.
34977 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
34979 * Makefile.in (site.exp): Remove "set ISLVER".
34981 2017-02-15 Jakub Jelinek <jakub@redhat.com>
34984 * real.c (real_from_integer): Call real_convert even for decimal.
34986 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
34989 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
34991 2017-02-14 Andrew Pinski <apinski@cavium.com>
34993 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
34994 cores and change the partno/implementer to be correct.
34995 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
34996 the 'B" as the implementer.
34997 * config/aarch64/aarch64-tune.md: Regenerate.
34999 2017-02-14 Carl Love <cel@us.ibm.com>
35001 * config/rs6000/rs6000.c: Add case statement entry to make the
35002 xvcvuxdsp built-in argument unsigned.
35003 * config/rs6000/vsx.md: Fix the source and return operand types so they
35004 match the instruction definitions from the ISA document. Fix typo
35005 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
35008 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
35011 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
35012 member early_clobber_alts.
35013 * lra-lives.c (reg_early_clobber_p): New.
35014 (process_bb_lives): Use it.
35015 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
35016 (debug_operand_data): Initialize early_clobber_alts.
35017 (setup_operand_alternative): Set up early_clobber_alts.
35018 (collect_non_operand_hard_regs): Ditto. Pass early clobber
35019 alternatives to new_insn_reg.
35020 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
35022 (lra_update_insn_regno_info): Pass the new arg.
35024 2017-02-14 Jakub Jelinek <jakub@redhat.com>
35026 PR middle-end/79505
35027 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
35028 (new_oacc_loop_raw): Don't clear already cleared fields.
35031 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
35032 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
35033 _mm512_prefetch_i64gather_ps): New inline functions and macros.
35035 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
35038 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
35040 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
35043 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
35044 the extra instruction to the right place to store 128-bit constant
35047 2017-02-14 Martin Sebor <msebor@redhat.com>
35049 PR middle-end/79448
35050 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
35051 warning for strings of unknown length.
35053 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
35055 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
35057 2017-02-14 Jeff Law <law@redhat.com>
35060 * ira-costs.c (scan_one_insn): Initialize register move costs
35061 for pseudos seen in USE/CLOBBER insns.
35063 PR tree-optimization/79095
35064 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
35065 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
35066 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
35067 if the operands are known to be not equal, then the resulting range
35069 (intersect_ranges): If the new range is ~[0,0] and the old range is
35070 wide, then prefer ~[0,0].
35071 * tree-vrp.c (overflow_comparison_p_1): New function.
35072 (overflow_comparison_p): New function.
35073 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
35074 if NAME is used in an overflow test.
35075 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
35076 overflow check that can be expressed as an equality test, then adjust
35077 ops to be that equality test.
35079 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35081 * config/s390/s390-builtin-types.def: Remove flags argument.
35082 * config/s390/s390.c (s390_init_builtins): Likewise.
35084 2017-02-14 Martin Liska <mliska@suse.cz>
35086 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
35087 vector. Fix trailing white spaces.
35089 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
35091 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
35094 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35096 PR rtl-optimization/68664
35097 * config/arm/arm.c (arm_sched_can_speculate_insn):
35098 New function. Declare prototype.
35099 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
35101 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35103 PR rtl-optimization/68664
35104 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
35106 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
35108 2017-02-14 Amit Pawar <amit.pawar@amd.com>
35110 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
35111 max skip bytes for function, loop and jump.
35113 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
35115 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
35116 ABS_EXPR for gimple dump.
35118 2017-02-14 Jakub Jelinek <jakub@redhat.com>
35121 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
35123 PR tree-optimization/79408
35124 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
35125 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
35126 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
35127 also if rhs1 is INTEGER_CST.
35129 2017-02-14 Richard Biener <rguenther@suse.de>
35131 PR middle-end/79432
35132 * tree-into-ssa.c (insert_phi_nodes): When the function can
35133 have abnormal edges rewrite SSA names with broken use-def
35134 dominance out of SSA and register them for PHI insertion.
35136 2017-02-13 Martin Sebor <msebor@redhat.com>
35138 PR middle-end/79496
35139 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
35140 clearing info.nowrite flag when snprintf size argument is a range.
35142 2017-02-13 Jakub Jelinek <jakub@redhat.com>
35144 * cprop.c (cprop_jump): Add missing space in string literal.
35145 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
35146 (get_constraint_for_component_ref): Likewise.
35147 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
35148 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
35149 * lra-constraints.c (process_alt_operands): Likewise.
35150 * ipa-inline.c (inline_small_functions): Likewise.
35151 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
35152 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
35153 * trans-mem.c (diagnose_tm_1_op): Likewise.
35154 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
35155 (grid_parallel_clauses_gridifiable): Likewise.
35157 * config/nvptx/mkoffload.c (process): Add space in between
35160 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
35161 "MOD4_SSE_REGS" and "ALL_REGS".
35163 * spellcheck.c (test_data): Add , in between "foo" and "food".
35165 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35168 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
35169 boundary crossing check and subsequent code generation agree.
35171 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35173 * config/aarch64/aarch64.c (has_memory_op): Delete.
35174 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
35177 2017-02-13 Jakub Jelinek <jakub@redhat.com>
35179 PR rtl-optimization/79388
35180 PR rtl-optimization/79450
35181 * combine.c (distribute_notes): When removing TEM_INSN for which
35182 corresponding dest has last value recorded, invalidate that last
35185 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35187 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
35188 of explicit '@'. Add missing assembly comment marker on branch costs
35191 2017-02-13 Nathan Sidwell <nathan@acm.org>
35193 * gengtype-lex.l (<in_struct>): Add '/'.
35195 2017-02-13 Martin Liska <mliska@suse.cz>
35198 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
35200 2017-02-13 Richard Biener <rguenther@suse.de>
35202 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
35204 * configure: Re-generate.
35205 * config.in: Likewise.
35206 * graphite-dependences.c: Simplify as if
35207 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
35208 * graphite-isl-ast-to-gimple.c: Likewise.
35209 * graphite-optimize-isl.c: Likewise.
35210 * graphite-poly.c: Likewise.
35211 * graphite-sese-to-poly.c: Likewise.
35212 * graphite.h: Likewise.
35213 * toplev.c: Include isl/version.h and use isl_version () for
35214 printing the ISL version.
35215 * doc/install.texi: Update ISL requirement.
35217 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
35219 * doc/standards.texi (Standards): Update reference to
35222 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
35224 * doc/extend.texi (Named Address Spaces): sourceware.org now
35226 * doc/install.texi (Binaries): Ditto.
35229 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
35231 * doc/cpp.texi: Replace "stringify"/"stringification" with C
35232 standard terminology "stringize"/"stringizing" throughout.
35233 * doc/cppinternals.texi: Likewise.
35235 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
35237 * doc/extend.texi: Fix some spelling mistakes and typos.
35238 * doc/invoke.texi: Likewise.
35240 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
35243 * params.def (inline-min-speedup) Change from 10 to 8.
35245 2017-02-11 Jakub Jelinek <jakub@redhat.com>
35247 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
35250 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
35253 * ipa-inline-analysis.c (get_minimal_bb): New function.
35254 (record_modified): Use it.
35255 (remap_edge_change_prob): Handle also ancestor functions.
35257 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
35259 * doc/contrib.texi (Contributors): Remove broken link into
35260 the Mauve CVS repository.
35262 2017-02-11 Jakub Jelinek <jakub@redhat.com>
35264 PR middle-end/79454
35265 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
35266 result computation whenever lhs doesn't have vector mode, not
35267 just when it has BLKmode.
35269 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
35271 * doc/makefile.texi (profiledbootstrap): Refer to the
35272 installation instructions only in textual form.
35274 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35277 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
35279 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
35281 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
35282 (Specific): Update mingw-w64 reference.
35284 (Specific): Remove broken link to Renesas RX processor.
35286 2017-02-10 Richard Biener <rguenther@suse.de>
35288 * toplev.c (process_options): Do not mention obsolete graphite
35289 options when printing sorry message about missing graphite support.
35290 Mention -floop-nest-optimize.
35292 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
35294 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
35295 (vtst_p16): Likewise.
35296 (vtstq_p8): Likewise.
35297 (vtstq_p16): Likewise.
35299 (vtstq_p64): Likewise.
35300 * config/arm/arm_neon.h (vgetq_lane_p64): New.
35301 (vset_lane_p64): New.
35302 (vsetq_lane_p64): New.
35304 2017-02-10 Jakub Jelinek <jakub@redhat.com>
35306 PR tree-optimization/79411
35307 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
35308 stmt operands are SSA_NAMEs used in abnormal phis.
35309 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
35312 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
35315 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
35318 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
35320 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
35322 2017-02-09 Jakub Jelinek <jakub@redhat.com>
35324 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
35328 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
35329 not arbitrary TREE_CONSTANT.
35332 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
35333 "omp declare target link" attribute unless is_global_var.
35334 * omp-offload.c (find_link_var_op): Likewise.
35336 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
35337 Chung-Lin Tang <cltang@codesourcery.com>
35339 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
35341 (gimplify_adjust_omp_clauses): Don't delete TILE.
35342 (gimplify_omp_for): Deal with TILE.
35343 * internal-fn.c (expand_GOACC_TILE): New function.
35344 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
35346 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
35347 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
35349 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
35350 avoid DIV for outermost collapse var.
35351 (expand_oacc_for): Insert tile element loop as needed. Adjust.
35352 Remove out of date comments, fix whitespace.
35353 * omp-general.c (omp_extract_for_data): Deal with tiling.
35354 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
35355 adjust OLF_DIM_BASE value.
35356 (struct omp_for_data): Add tiling field.
35357 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
35358 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
35359 for auto loops. Remove default auto determining, moved to
35360 oacc_loop_fixed_partitions.
35361 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
35362 stmts, add e_mask field.
35363 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
35364 (oacc_thread_numbers): Use oacc_dim_call.
35365 (oacc_xform_tile): New.
35366 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
35367 (finish_oacc_loop): Adjust for ifns vector.
35368 (oacc_loop_discover_walk): Append loop abstraction sites to list,
35369 add case for GOACC_TILE fns.
35370 (oacc_loop_xform_loop): Delete.
35371 (oacc_loop_process): Iterate over call list directly, and add
35372 handling for GOACC_TILE fns.
35373 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
35375 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
35376 vector partitioning to outer loops. Assign 2 partitions to loops
35377 when available. Add TILE handling.
35378 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
35379 (execite_oacc_device_lower): Process GOACC_TILE fns,
35380 ignore unknown specs.
35381 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
35382 * tree.c (omp_clause_num_ops): Adjust TILE ops.
35383 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
35385 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
35387 * configure.ac (ACX_BUGURL): Update.
35388 * configure: Regenerate.
35390 2017-02-09 Richard Biener <rguenther@suse.de>
35392 PR tree-optimization/69823
35393 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
35394 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
35396 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35398 * config/arc/arc-c.def: Add __NPS400__ definition.
35399 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
35400 (TARGET_NPS400): Define.
35402 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35404 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
35406 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
35407 pointer, arch_info.
35408 (arc_cpu_types): Fill the arch_info field with a pointer into the
35409 arc_arch_types table.
35410 (arc_selected_cpu): Declare.
35411 * config/arc/arc.c (arc_selected_cpu): Make global.
35412 (arc_selected_arch): Delete.
35413 (arc_base_cpu): Delete.
35414 (arc_override_options): Remove references to deleted variables,
35415 update access to arch information.
35416 (ARC_OPT): Update access to arch information.
35417 (ARC_OPTX): Likewise.
35418 * config/arc/arc.h (arc_base_cpu): Remove declaration.
35419 (TARGET_ARC600): Update access to arch information.
35420 (TARGET_ARC601): Likewise.
35421 (TARGET_ARC700): Likewise.
35422 (TARGET_EM): Likewise.
35423 (TARGET_HS): Likewise.
35424 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
35427 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
35430 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
35431 condition/operands for integer GE/LE/GEU/LEU operations.
35433 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
35435 PR translation/79397
35436 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
35439 2017-02-08 Martin Jambor <mjambor@suse.cz>
35442 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
35443 whether allocation happened.
35444 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
35445 nothing was allocated.
35447 2017-02-08 Jakub Jelinek <jakub@redhat.com>
35449 PR tree-optimization/79408
35450 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
35451 constant, but SSA_NAME with a known integer range, use the minimum
35452 of that range instead of op1 to determine if modulo can be replaced
35453 with its first operand.
35455 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35457 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
35459 2017-02-08 Richard Biener <rguenther@suse.de>
35461 PR tree-optimization/71824
35462 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
35463 Check all loops contained in the merged region.
35465 2017-02-07 Andrew Pinski <apinski@cavium.com>
35467 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
35469 2017-02-07 Andrew Pinski <apinski@cavium.com>
35471 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
35472 (thunderxt88): Likewise.
35473 (thunderxt81): Disable LSE and change v8.1 to v8.
35474 (thunderxt83): Likewise.
35476 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35477 Richard Biener <rguenther@suse.de>
35479 PR middle-end/79399
35480 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
35481 type from int to size_t.
35482 * ira-costs.c (struct_costs_size): Change type from int to size_t.
35484 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35486 PR rtl-optimization/79386
35487 * cprop.c (bypass_conditional_jumps): Initialize
35488 bypass_last_basic_block already before splitting bbs after
35489 unconditional traps...
35490 (bypass_conditional_jumps): ... rather than here.
35493 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
35494 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
35495 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
35496 fix -masm=intel patterns.
35498 2017-02-07 Richard Biener <rguenther@suse.de>
35500 PR tree-optimization/79256
35501 PR middle-end/79278
35502 * builtins.c (get_object_alignment_2): Use min_align_of_type
35503 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
35504 and ADJUST_FIELD_ALIGN.
35506 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
35508 * doc/tm.texi: Regenerate.
35509 * stor-layout.c (layout_decl): Adjust.
35510 (update_alignment_for_field): Likewise.
35511 (place_field): Likewise.
35512 (min_align_of_type): Likewise.
35513 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
35514 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
35515 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
35516 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
35517 * config/frv/frv.c (frv_adjust_field_align): Likewise.
35518 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
35519 * config/i386/i386.c (x86_field_alignment): Likewise.
35520 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
35521 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
35522 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
35523 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
35524 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
35525 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
35529 2017-01-30 Richard Biener <rguenther@suse.de>
35531 PR tree-optimization/79256
35532 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35533 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35536 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
35538 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
35539 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
35540 builtins to SImode and emit a zero-extend, if necessary.
35542 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35544 * docs/invoke.texi (RISC-V Options): Alphabetize.
35546 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35548 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
35551 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35553 * config/riscv/riscv.c: New file.
35554 * common/config/riscv/riscv-common.c: Likewise.
35555 * config.gcc: Likewise.
35556 * config/riscv/constraints.md: Likewise.
35557 * config/riscv/elf.h: Likewise.
35558 * config/riscv/generic.md: Likewise.
35559 * config/riscv/linux.h: Likewise.
35560 * config/riscv/multilib-generator: Likewise.
35561 * config/riscv/peephole.md: Likewise.
35562 * config/riscv/pic.md: Likewise.
35563 * config/riscv/predicates.md: Likewise.
35564 * config/riscv/riscv-builtins.c: Likewise.
35565 * config/riscv/riscv-c.c: Likewise.
35566 * config/riscv/riscv-ftypes.def: Likewise.
35567 * config/riscv/riscv-modes.def: Likewise.
35568 * config/riscv/riscv-opts.h: Likewise.
35569 * config/riscv/riscv-protos.h: Likewise.
35570 * config/riscv/riscv.h: Likewise.
35571 * config/riscv/riscv.md: Likewise.
35572 * config/riscv/riscv.opt: Likewise.
35573 * config/riscv/sync.md: Likewise.
35574 * config/riscv/t-elf-multilib: Likewise.
35575 * config/riscv/t-linux: Likewise.
35576 * config/riscv/t-linux-multilib: Likewise.
35577 * config/riscv/t-riscv: Likewise.
35578 * configure.ac: Likewise.
35579 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
35580 Waterman as RISC-V maintainers.
35581 * doc/install.texi: Add RISC-V entries.
35582 * doc/invoke.texi: Add RISC-V options section.
35583 * doc/md.texi: Add RISC-V constraints section.
35584 * configure: Regenerated.
35586 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
35589 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
35590 false values to be constant vectors with all 0 or all 1 bits set.
35591 (vcondu<mode><mode>): Likewise.
35592 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
35594 (fpmask_comparison_operator): Update comment.
35595 (vecint_comparison_operator): New predicate.
35596 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
35597 vector conditionals when the true and false values are constant
35598 vectors with all 0 bits or all 1 bits set.
35600 2017-02-06 Martin Sebor <msebor@redhat.com>
35602 PR tree-optimization/79376
35603 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
35605 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
35607 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
35608 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
35609 to simplify split condition.
35611 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35613 * omp-expand.c (oxpand_omp_atomic_fetch_op,
35614 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
35617 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
35619 PR rtl-optimization/68664
35620 * target.def (can_speculate_insn): New hook.
35621 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
35622 * doc/tm.texi: Regenerate.
35623 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
35624 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
35625 (rs6000_sched_can_speculate_insn): New function.
35627 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35629 PR tree-optimization/79284
35630 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
35631 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
35632 vectorizable_mask_load_store, vectorizable_operation,
35633 vect_is_simple_cond, get_same_sized_vectype): Use it instead
35634 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
35635 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
35636 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
35637 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
35638 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
35639 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
35640 is_gimple_assign (stmt). Replace another such test with
35641 is_gimple_assign (stmt).
35643 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35646 * config/avr/avr.c (rtl-iter.h): Include it.
35647 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
35648 (avr_legitimate_combined_insn): ...and implementation.
35650 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
35652 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
35653 * config/s390/s390.c (s390_const_operand_ok)
35654 (s390_canonicalize_comparison, s390_extract_part)
35655 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
35656 (s390_contiguous_bitmask_p, s390_rtx_costs)
35657 (legitimize_pic_address): Likewise.
35658 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
35659 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
35660 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
35661 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
35662 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
35664 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35666 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
35667 REGNO($0) == REGNO($1).
35669 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35671 * config/s390/linux.h(SIZE_TYPE): Add comment.
35673 2017-02-06 Julian Brown <julian@codesourcery.com>
35674 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
35675 Virendra Pathak <virendra.pathak@broadcom.com>
35677 * config/aarch64/aarch64-cores.def: Change the scheduler
35679 * config/aarch64/aarch64.md: Include thunderx2t99.md.
35680 * config/aarch64/thunderx2t99.md: New file.
35682 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35684 * doc/standards.texi (Go Language): Update link to language
35687 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
35689 * tree-eh.c (lower_resx): Sanitize profile.
35690 (cleanup_empty_eh_move_lp): Likewise.
35692 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
35695 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
35697 * cfgloopmanip.h (loop_version): Update prototype.
35698 * modulo-sched.c (sms_schedule): Update call of loop_version.
35699 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
35700 * tree-parloops.c (gen_parallel_loop): Likewise.
35701 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
35702 * tree-ssa-loop-split.c (split_loop): Likewise.
35703 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
35704 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
35706 2017-02-05 Martin Liska <mliska@suse.cz>
35709 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
35711 (print_operand_address): Initialize a struct to zero.
35713 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35715 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
35716 garbage collector only in textual form.
35718 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35720 * doc/extend.texi (x86 specific memory model extensions for
35721 transactional memory): Simplify a phrase.
35723 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
35726 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
35727 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
35728 (atomic_storedi_1): Likewise.
35730 2017-02-04 Jakub Jelinek <jakub@redhat.com>
35732 PR tree-optimization/79338
35733 * tree-parloops.c (gather_scalar_reductions): Don't call
35734 vect_analyze_loop_form for loop->inner before destroying loop's
35737 2017-02-03 Martin Sebor <msebor@redhat.com>
35739 PR tree-optimization/79327
35740 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
35741 when precision has resulted in leading zeros.
35742 (format_integer): Adjust the likely counter to assume an unknown
35743 argument that may be zero is non-zero.
35745 2017-02-03 Jason Merrill <jason@redhat.com>
35748 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
35749 avoid copying non-taken branch.
35751 2017-02-03 Jakub Jelinek <jakub@redhat.com>
35753 PR tree-optimization/79340
35754 * tree-vect-loop.c (vectorizable_reduction): Release
35755 vec_defs elements after safe_splicing them into other vectors.
35758 PR tree-optimization/79327
35759 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
35760 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
35762 (format_integer): Use wide_int_to_tree instead of build_int_cst
35763 + to_?hwi. If argmin is NULL, just set argmin and argmax to
35764 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
35765 of shortest and longest sequence.
35767 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
35769 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
35770 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
35772 2017-02-03 Walter Lee <walt@tilera.com>
35775 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
35776 after initial stackframe link reg save.
35777 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
35779 2017-02-03 Jakub Jelinek <jakub@redhat.com>
35782 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
35783 wu for stxssp alternative.
35785 2017-02-03 Martin Sebor <msebor@redhat.com>
35787 PR tree-optimization/79352
35788 * gimple-fold.c (get_range_strlen): Add argument.
35789 (get_range_strlen): Change return type to bool.
35790 (get_maxval_strlen): Pass in a dummy argument.
35791 * gimple-fold.h (get_range_strlen): Change return type to bool.
35792 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
35793 * tree.h (array_at_struct_end_p): Add argument.
35794 * tree.c (array_at_struct_end_p): Handle it.
35796 2017-02-03 Martin Liska <mliska@suse.cz>
35799 * multiple_target.c (create_dispatcher_calls): Redirect edge
35800 from a caller of a dispatcher.
35801 (expand_target_clones): Make the clones local.
35802 (ipa_target_clone): Do both target clones and resolvers.
35803 (ipa_dispatcher_calls): Remove the pass.
35804 (pass_dispatcher_calls::gate): Likewise.
35805 (make_pass_dispatcher_calls): Likewise.
35806 * passes.def (pass_target_clone): Put as very first IPA early
35809 2017-02-03 Martin Liska <mliska@suse.cz>
35811 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
35812 in case of a function with ifunc attribute.
35814 2017-02-03 Martin Liska <mliska@suse.cz>
35816 * cgraph.c (cgraph_node::dump): Dump function version info.
35817 * symtab.c (symtab_node::dump_base): Add missing new line.
35819 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
35821 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
35822 (ifcombine_ifandif): Use it.
35824 2017-02-03 Martin Liska <mliska@suse.cz>
35826 * doc/invoke.texi: Document default value for
35827 use-after-scope-direct-emission-threshold.
35829 2017-02-03 Martin Liska <mliska@suse.cz>
35831 PR tree-optimization/79339
35832 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
35833 (format_floating): Likewise.
35835 2017-02-03 Martin Liska <mliska@suse.cz>
35838 * ipa-prop.c (ipa_node_params_t::insert): Remove current
35840 (ipa_node_params_t::remove): Likewise.
35841 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
35842 initialization from removed ipa_node_params_t::insert.
35843 (ipa_node_params::~ipa_node_params): Move from removed
35844 ipa_node_params_t::release.
35845 * symbol-summary.h (symbol_summary::m_released): New member.
35846 Do not release a summary twice. Do not allow to call finalizer
35847 for types of a summary that live in GGC memory.
35849 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
35851 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
35854 2017-02-02 Martin Sebor <msebor@redhat.com>
35856 PR middle-end/79275
35857 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
35858 (format_string): Tighten up the range of output for non-constant
35859 strings and correct the expected range for wide non-constant strings.
35861 2017-02-02 Martin Sebor <msebor@redhat.com>
35863 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
35865 PR middle-end/32003
35866 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
35868 (-fdump-tree-@var): Add to index and document how to come up
35869 with pass-specific option and dump file names.
35870 (-fdump-passes): Clarify where to look for output.
35872 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
35874 PR middle-end/77445
35875 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
35876 statistics of the analyzed path; allow threading for speed when
35877 any of BBs along the path are optimized for speed.
35879 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
35881 PR middle-end/78468
35882 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
35883 settings of the virtual registers.
35886 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
35888 * explow.c (get_dynamic_stack_size): Take known alignment of stack
35889 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
35892 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35894 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
35895 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
35897 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35899 * config/s390/s390.md: Add missing comments with the expanded
35901 * config/s390/vector.md: Likewise.
35902 * config/s390/vx-builtins.md: Likewise.
35904 2017-02-02 Jakub Jelinek <jakub@redhat.com>
35907 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
35908 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
35909 conditions on a single line.
35911 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35913 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
35914 __S390_VX__ to __VX__.
35916 2017-02-01 Andrew Pinski <apinski@cavium.com>
35918 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
35919 stmt_info to record_stmt_cost.
35920 (vect_get_known_peeling_cost): Pass stmt_info if known to
35922 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
35923 cpu_vector_cost field into
35924 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
35925 field into vec_int_stmt_cost and vec_fp_stmt_cost.
35926 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
35927 splitting of scalar_stmt_cost and vec_stmt_cost.
35928 (thunderx_vector_cost): Likewise.
35929 (cortexa57_vector_cost): LIkewise.
35930 (exynosm1_vector_cost): Likewise.
35931 (xgene1_vector_cost): Likewise.
35932 (thunderx2t99_vector_cost): Improve after the splitting of the two
35934 (aarch64_builtin_vectorization_cost): Update for the splitting of
35935 scalar_stmt_cost and vec_stmt_cost.
35937 2017-02-01 Torvald Riegel <triegel@redhat.com>
35938 Richard Henderson <rth@redhat.com>
35940 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
35941 conditional on existance of a fast atomic load.
35942 * optabs-query.c (can_atomic_load_p): New function.
35943 * optabs-query.h (can_atomic_load_p): Declare it.
35944 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
35945 no fast atomic load is available for the particular size of access.
35946 (expand_atomic_compare_and_swap): Likewise.
35947 (expand_atomic_load): Likewise.
35948 (expand_atomic_store): Likewise.
35949 (expand_atomic_fetch_op): Likewise.
35950 * testsuite/lib/target-supports.exp
35951 (check_effective_target_sync_int_128): Remove x86 because it provides
35952 no fast atomic load.
35953 (check_effective_target_sync_int_128_runtime): Likewise.
35955 2017-02-01 Richard Biener <rguenther@suse.de>
35957 * graphite.c: Include tree-vectorizer.h for find_loop_location.
35958 (graphite_transform_loops): Provide opt-info for optimized nests.
35959 * tree-parloop.c (parallelize_loops): Provide opt-info for
35960 parallelized loops.
35962 2017-02-01 Richard Biener <rguenther@suse.de>
35964 PR middle-end/79315
35965 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
35966 was not set before.
35968 2017-02-01 Richard Biener <rguenther@suse.de>
35970 PR tree-optimization/71824
35971 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
35972 Verify the loops are valid in the merged SESE region.
35973 (scop_detection::can_represent_loop_1): Check analyzing the
35974 evolution of the number of iterations in the region succeeds.
35976 2017-01-31 Ian Lance Taylor <iant@golang.org>
35978 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
35979 REG_ARGS_SIZE note to 32-bit push insns and call insn.
35981 2017-01-31 David Malcolm <dmalcolm@redhat.com>
35983 PR preprocessor/79210
35984 * input.c (get_substring_ranges_for_loc): Replace line_width
35985 assertion with error-handling.
35987 2017-01-31 Richard Biener <rguenther@suse.de>
35989 PR tree-optimization/77318
35990 * graphite-sese-to-poly.c (extract_affine): Fix assert.
35991 (create_pw_aff_from_tree): Take loop parameter.
35992 (add_condition_to_pbb): Pass loop of the condition to
35993 create_pw_aff_from_tree.
35995 2017-01-31 Jakub Jelinek <jakub@redhat.com>
35997 * config/s390/s390.c (s390_asan_shadow_offset): New function.
35998 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
36000 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
36004 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
36006 (convert_int_to_float128): Likewise.
36007 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
36008 (convert_int_to_float128): Likewise.
36009 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
36010 (UNSPEC_IEEE128_CONVERT): Likewise.
36011 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
36012 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
36013 Use local variables for IBM extended format.
36014 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
36015 (fix_trunc<mode>si2_fprs): Likewise.
36016 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
36017 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
36018 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
36019 to know that we can now have integers of all sizes in vector
36021 (fix<uns>_<mode>di2_hw): Likewise.
36022 (float<uns>_<mode>si2_hw): Likewise.
36023 (fix_<mode>si2_hw): Likewise.
36024 (fixuns_<mode>si2_hw): Likewise.
36025 (float<uns>_<mode>di2_hw): Likewise.
36026 (float_<mode>di2_hw): Likewise.
36027 (float_<mode>si2_hw): Likewise.
36028 (floatuns_<mode>di2_hw): Likewise.
36029 (floatuns_<mode>si2_hw): Likewise.
36030 (xscvqp<su>wz_<mode>): Delete, no longer used.
36031 (xscvqp<su>dz_<mode>): Likewise.
36032 (xscv<su>dqp_<mode>): Likewise.
36033 (ieee128_mfvsrd_64bit): Likewise.
36034 (ieee128_mfvsrd_32bit): Likewise.
36035 (ieee128_mfvsrwz): Likewise.
36036 (ieee128_mtvsrw): Likewise.
36037 (ieee128_mtvsrd_64bit): Likewise.
36038 (ieee128_mtvsrd_32bit): Likewise.
36040 2017-01-31 Martin Liska <mliska@suse.cz>
36043 * ipa-prop.c (ipa_free_all_node_params): Call release method
36044 instead of ~sumbol_summary to not to trigger double times
36047 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
36049 PR tree-optimization/71691
36050 * bitmap.h (class auto_bitmap): New.
36051 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
36052 is_maybe_undefined instead of ssa_undefined_value_p.
36054 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36056 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
36057 __S390_ARCH_LEVEL__ to __ARCH__.
36059 2017-01-31 Jakub Jelinek <jakub@redhat.com>
36061 PR tree-optimization/79267
36062 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
36063 if should_remove_lhs_p is true.
36065 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
36068 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
36069 (add_alignment_attribute): New.
36070 (base_type_die): Add alignment attribute.
36071 (subrange_type_die): Likewise.
36072 (modified_type_die): Likewise.
36073 (gen_array_type_die): Likewise.
36074 (gen_descr_array_type_die: Likewise.
36075 (gen_enumeration_type_die): Likewise.
36076 (gen_subprogram_die): Likewise.
36077 (gen_variable_die): Likewise.
36078 (gen_field_die): Likewise.
36079 (gen_ptr_to_mbr_type_die): Likewise.
36080 (gen_struct_or_union_type_die): Likewise.
36081 (gen_subroutine_type_die): Likewise.
36082 (gen_typedef_die): Likewise.
36083 (base_type_cmp): Compare alignment attribute.
36085 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36088 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
36089 (setb_unsigned) New pattern for setb with CCUNS.
36090 * config/rs6000/rs6000.c (expand_block_compare): Use a different
36091 subfc./subfe sequence to avoid overflow problems. Generate a
36092 shorter sequence with cmpld/setb for power9.
36093 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
36094 for generating subfc. instruction.
36095 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
36096 now uses this instruction.
36098 2017-01-30 Ian Lance Taylor <iant@google.com>
36101 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
36102 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
36104 2017-01-30 Martin Sebor <msebor@redhat.com>
36106 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
36107 Move constant to the right of a relational operator.
36108 (get_mpfr_format_length, format_character, format_string): Ditto.
36109 (should_warn_p, maybe_warn): Same.
36111 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
36113 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
36116 * asan.c (get_translation_unit_decl): Remove function.
36117 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
36119 2017-01-30 Martin Liska <mliska@suse.cz>
36121 PR gcov-profile/79259
36122 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
36123 -fprofile-generate.
36125 2017-01-30 Martin Liska <mliska@suse.cz>
36128 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
36129 Initialize variables with NULL value.
36131 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
36134 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
36136 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
36138 2017-01-30 Richard Biener <rguenther@suse.de>
36140 PR tree-optimization/79276
36141 * tree-vrp.c (process_assert_insertions): Properly adjust common
36142 when removing a duplicate.
36144 2017-01-30 Richard Biener <rguenther@suse.de>
36146 PR tree-optimization/79256
36147 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
36148 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
36150 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
36152 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
36155 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
36156 ("*r<noxa>sbg_<mode>_sll_bitmask")
36157 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
36158 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
36159 Use contiguous_bitmask_nowrap_operand.
36161 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36164 * config/rs6000/altivec.h (vec_xl): Revise #define.
36165 (vec_xst): Likewise.
36167 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
36169 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
36171 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
36173 PR rtl-optimization/79194
36174 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
36175 traps before call to bypass_conditional_jumps.
36177 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
36179 PR tree-optimization/71374
36180 * lra-constraints.c (check_conflict_input_operands): New.
36181 (match_reload): Use it.
36183 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
36186 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
36187 account to calculate conflict_set.
36189 2017-01-27 Bin Cheng <bin.cheng@arm.com>
36191 PR rtl-optimization/78559
36192 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
36193 other_insn in combine.
36195 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
36197 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
36198 uint16_type_node for BT_UINT16.
36200 2017-01-27 David Malcolm <dmalcolm@redhat.com>
36202 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
36203 "RTL Tests" to menu.
36204 (GIMPLE Tests): New node.
36205 (RTL Tests): New node.
36207 2017-01-27 Richard Biener <rguenther@suse.de>
36209 PR tree-optimization/79245
36210 * tree-loop-distribution.c (distribute_loop): Apply cost
36211 modeling also to detected patterns.
36213 2017-01-27 Richard Biener <rguenther@suse.de>
36215 PR tree-optimization/71433
36216 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
36217 (compare_assert_loc): New function.
36218 (process_assert_insertions): Sort and optimize assert locations
36219 to remove duplicates and push down identical assertions on
36220 edges to their destination block.
36222 2017-01-27 Richard Biener <rguenther@suse.de>
36224 PR tree-optimization/79244
36225 * tree-vrp.c (remove_range_assertions): Forcefully propagate
36226 out SSA names even if abnormal.
36228 2017-01-27 Jakub Jelinek <jakub@redhat.com>
36230 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
36231 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
36232 instead of MPFR_RNDN.
36234 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
36237 * arm.c (arm_option_override): Don't call build_target_option_node
36238 until after doing all option overrides.
36239 (arm_valid_target_attribute_tree): Likewise.
36241 2017-01-27 Martin Liska <mliska@suse.cz>
36243 * doc/invoke.texi (-fprofile-arcs): Document profiling support
36244 for {cd}tors and C++ {cd}tors.
36246 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
36248 * config/s390/s390.md ("*setmem_long_and")
36249 ("*setmem_long_and_31z"): Use zero_extend instead of and.
36251 2017-01-26 Martin Sebor <msebor@redhat.com>
36253 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
36256 2017-01-26 Martin Sebor <msebor@redhat.com>
36258 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
36259 HAVE_DFmode before using XFmode or DFmode.
36260 (parse_directive): Avoid using the z length modifier to avoid
36261 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
36263 PR middle-end/78703
36264 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
36265 to accept adjustment as an array.
36266 (get_int_range): New function.
36267 (struct directive): Make width and prec arrays.
36268 (directive::set_width, directive::set_precision): Call get_int_range.
36269 (format_integer, format_floating): Handle width and precision ranges.
36270 (format_string, parse_directive): Same.
36272 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36275 * dwarf2out.c (generate_skeleton_bottom_up): For children with
36276 comdat_type_p set, just clone them, but keep the children in the
36280 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
36281 which have direct callers with -fvar-tracking-assignments enabled
36283 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
36284 inside of type units.
36286 2017-01-26 Martin Sebor <msebor@redhat.com>
36288 PR middle-end/78703
36289 * gimple-ssa-sprintf.c (struct result_range): Add likely and
36291 (struct format_result): Replace number_chars, number_chars_min,
36292 and number_chars_max with a single member of struct result_range.
36294 (format_result::operator+=): Adjust.
36295 (struct fmtresult): Remove bounded. Handle likely and unlikely
36297 (fmtresult::adjust_for_width_or_precision): New function.
36298 (fmtresult:type_max_digits): New function.
36299 (bytes_remaining): Handle likely and unlikely counters.
36300 (min_bytes_remaining): Remove.
36301 (format_percent): Simplify.
36302 (format_integer, format_floating): Set likely and unlikely counters.
36303 (get_string_length, format_character, format_string): Same.
36304 (format_plain, should_warn_p): New function.
36305 (maybe_warn): Call should_warn_p. Update diagnostic messages
36306 and handle those for all directives, including plain strings.
36307 (format_directive): Handle likely and unlikely counters.
36308 Remove unnecessary quoting from diagnostics. Add an informational
36310 (add_bytes): Remove.
36311 (pass_sprintf_length::compute_format_length): Simplify.
36312 (try_substitute_return_value): Handle likely and unlikely counters.
36314 2017-01-26 Carl Love <cel@us.ibm.com>
36316 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
36317 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
36319 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
36322 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
36323 endianess for subregs into account.
36324 * lra-constraints.c (lra_constraints): Do risky transformations
36325 always on the first iteration.
36326 * lra-lives.c (check_pseudos_live_through_calls): Add arg
36327 last_call_used_reg_set.
36328 (process_bb_lives): Define and use last_call_used_reg_set.
36329 * lra.c (lra): Always continue after lra_constraints on the first
36332 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
36334 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
36336 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
36338 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36340 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
36341 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
36342 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
36343 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
36344 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
36345 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
36346 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
36347 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
36348 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
36350 2017-01-26 Marek Polacek <polacek@redhat.com>
36353 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
36354 for the third operand.
36356 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36358 PR middle-end/79236
36359 * omp-low.c (struct omp_context): Add simt_stmt field.
36360 (scan_omp_for): Return omp_context *.
36361 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
36362 context to the _simt_ SIMD stmt.
36363 (lower_omp_for): For combined SIMD with sibling _simt_
36364 SIMD, make sure to use the same decls in _looptemp_
36365 clauses as in the sibling.
36367 2017-01-26 David Sherwood <david.sherwood@arm.com>
36369 PR middle-end/79212
36370 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
36373 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36376 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
36377 emit fld b; fld a; if possible.
36379 * brig-builtins.def: Update copyright years.
36380 * config/arm/arm_acle_builtins.def: Update copyright years.
36382 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
36385 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
36386 constraint instead of o for the stxsd instruction.
36388 2017-01-25 Carl Love <cel@us.ibm.com>
36390 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
36391 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
36393 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
36395 * doc/invoke.texi (C++ Dialect Options): Fix typo.
36397 2017-01-25 Richard Biener <rguenther@suse.de>
36399 PR tree-optimization/69264
36400 * target.def (vector_alignment_reachable): Improve documentation.
36401 * doc/tm.texi: Regenerate.
36402 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
36404 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
36405 earlier changes with respect to TYPE_USER_ALIGN.
36406 (vector_alignment_reachable_p): Likewise. Improve dumping.
36408 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36411 * config/arm/arm.md (xordi3): Force constant operand into a register
36414 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36416 * doc/invoke.texi (-fstore-merging): Correct default optimization
36417 levels at which it is enabled.
36418 (-O): Move -fstore-merging from list to...
36421 2017-01-25 Richard Biener <rguenther@suse.de>
36424 * omp-expand.c: Include debug.h.
36425 (expand_omp_taskreg): Make sure to generate early debug before
36426 outlining anything from a function.
36427 (expand_omp_target): Likewise.
36428 (grid_expand_target_grid_body): Likewise.
36430 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
36433 * asan.c (get_translation_unit_decl): New function.
36434 (asan_add_global): Extract modules file name from globals
36435 TRANSLATION_UNIT_DECL name.
36437 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
36440 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
36441 for long calls with APCS frame and VFP.
36443 2017-01-24 David Malcolm <dmalcolm@redhat.com>
36445 * cfg.c (original_copy_tables_initialized_p): New function.
36446 * cfg.h (original_copy_tables_initialized_p): New decl.
36447 * cfgrtl.c (relink_block_chain): Guard the call to
36448 free_original_copy_tables with a call to
36449 original_copy_tables_initialized_p.
36450 * cgraph.h (symtab_node::native_rtl_p): New decl.
36451 * cgraphunit.c (symtab_node::native_rtl_p): New function.
36452 (symtab_node::needed_p): Don't assert for early assembly output
36453 for __RTL functions.
36454 (cgraph_node::finalize_function): Set "force_output" for __RTL
36456 (cgraph_node::analyze): Bail out early for __RTL functions.
36457 (analyze_functions): Update assertion to support __RTL functions.
36458 (cgraph_node::expand): Bail out early for __RTL functions.
36459 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
36461 * function.h (struct function): Update comment for field
36463 * gimple-expr.c: Include "tree-pass.h".
36464 (gimple_has_body_p): Return false for __RTL functions.
36465 * Makefile.in (OBJS): Add run-rtl-passes.o.
36466 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
36468 (gcc::pass_manager::get_clean_slate): New accessor.
36469 * passes.c: Include "insn-addr.h".
36470 (should_skip_pass_p): Add logging. Update logic for running
36471 "expand" to be compatible with both __GIMPLE and __RTL. Guard
36472 property-provider override so it is only done for gimple passes.
36474 (skip_pass): New function.
36475 (execute_one_pass): Call skip_pass when skipping passes.
36476 * read-md.c (md_reader::read_char): Support filtering
36477 the input to a subset of line numbers.
36478 (md_reader::md_reader): Initialize fields
36479 m_first_line and m_last_line.
36480 (md_reader::read_file_fragment): New function.
36481 * read-md.h (md_reader::read_file_fragment): New decl.
36482 (md_reader::m_first_line): New field.
36483 (md_reader::m_last_line): New field.
36484 * read-rtl-function.c (function_reader::create_function): Only
36485 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
36486 curr_properties. Set DECL_INITIAL to a dummy block.
36487 (read_rtl_function_body_from_file_range): New function.
36488 * read-rtl-function.h (read_rtl_function_body_from_file_range):
36490 * run-rtl-passes.c: New file.
36491 * run-rtl-passes.h: New file.
36493 2017-01-24 Jeff Law <law@redhat.com>
36495 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
36498 2017-01-24 Bin Cheng <bin.cheng@arm.com>
36500 PR tree-optimization/79159
36501 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
36502 (record_nonwrapping_iv): Improve boundary using above function if no
36503 value range information.
36505 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
36506 Martin Jambor <mjambor@suse.cz>
36508 * brig-builtins.def: New file.
36509 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
36510 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
36511 (DEF_HSAIL_SAT_BUILTIN): Likewise.
36512 (DEF_HSAIL_INTR_BUILTIN): Likewise.
36513 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
36514 * builtin-types.def (BT_INT8): New.
36515 (BT_INT16): Likewise.
36516 (BT_UINT8): Likewise.
36517 (BT_UINT16): Likewise.
36518 (BT_FN_ULONG): Likewise.
36519 (BT_FN_UINT_INT): Likewise.
36520 (BT_FN_UINT_ULONG): Likewise.
36521 (BT_FN_UINT_LONG): Likewise.
36522 (BT_FN_UINT_PTR): Likewise.
36523 (BT_FN_ULONG_PTR): Likewise.
36524 (BT_FN_INT8_FLOAT): Likewise.
36525 (BT_FN_INT16_FLOAT): Likewise.
36526 (BT_FN_UINT32_FLOAT): Likewise.
36527 (BT_FN_UINT16_FLOAT): Likewise.
36528 (BT_FN_UINT8_FLOAT): Likewise.
36529 (BT_FN_UINT64_FLOAT): Likewise.
36530 (BT_FN_UINT16_UINT32): Likewise.
36531 (BT_FN_UINT32_UINT16): Likewise.
36532 (BT_FN_UINT16_UINT16_UINT16): Likewise.
36533 (BT_FN_INT_PTR_INT): Likewise.
36534 (BT_FN_UINT_PTR_UINT): Likewise.
36535 (BT_FN_LONG_PTR_LONG): Likewise.
36536 (BT_FN_ULONG_PTR_ULONG): Likewise.
36537 (BT_FN_VOID_UINT64_UINT64): Likewise.
36538 (BT_FN_UINT8_UINT8_UINT8): Likewise.
36539 (BT_FN_INT8_INT8_INT8): Likewise.
36540 (BT_FN_INT16_INT16_INT16): Likewise.
36541 (BT_FN_INT_INT_INT): Likewise.
36542 (BT_FN_UINT_FLOAT_UINT): Likewise.
36543 (BT_FN_FLOAT_UINT_UINT): Likewise.
36544 (BT_FN_ULONG_UINT_UINT): Likewise.
36545 (BT_FN_ULONG_UINT_PTR): Likewise.
36546 (BT_FN_ULONG_ULONG_ULONG): Likewise.
36547 (BT_FN_UINT_UINT_UINT): Likewise.
36548 (BT_FN_VOID_UINT_PTR): Likewise.
36549 (BT_FN_UINT_UINT_PTR: Likewise.
36550 (BT_FN_UINT32_UINT64_PTR): Likewise.
36551 (BT_FN_INT_INT_UINT_UINT): Likewise.
36552 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
36553 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
36554 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
36555 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
36556 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
36557 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
36558 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
36559 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
36560 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
36561 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
36562 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
36563 * doc/frontends.texi: List BRIG FE.
36564 * doc/install.texi (Testing): Add BRIG tesring requirements.
36565 * doc/invoke.texi (Overall Options): Mention BRIG.
36566 * doc/standards.texi (Standards): Doucment BRIG HSA version.
36568 2017-01-24 Richard Biener <rguenther@suse.de>
36570 PR translation/79208
36571 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
36573 2017-01-24 Martin Jambor <mjambor@suse.cz>
36576 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
36577 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
36578 and known_contexts.
36580 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
36582 PR middle-end/79123
36583 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
36584 casts from signed to unsigned really don't have a range.
36586 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
36588 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
36589 GMP_RNDx for compatiblity.
36591 2017-01-24 Martin Liska <mliska@suse.cz>
36594 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
36595 that would prevent us to call alloca with -1 as argument.
36597 2017-01-24 Jakub Jelinek <jakub@redhat.com>
36599 * dwarf2out.c (output_compilation_unit_header, output_file_names):
36600 Avoid -Wformat-security warning.
36602 2017-01-23 Andrew Pinski <apinski@cavium.com>
36604 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
36607 2017-01-23 Martin Sebor <msebor@redhat.com>
36609 PR middle-end/78703
36610 * gimple-ssa-sprintf.c (warn_level): New global.
36611 (format_integer): Use it here and throughout the rest of the file.
36612 Use the same switch to compute sign as base.
36613 (maybe_warn): New function.
36614 (format_directive): Factor out warnings into maybe_warn.
36615 Add debugging output. Use warn_level.
36616 (add_bytes): Use warn_level.
36617 (pass_sprintf_length::compute_format_length): Add debugging output.
36618 (try_substitute_return_value): Same.
36619 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
36621 PR middle-end/78703
36622 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
36623 (struct fmtresult, format_integer, format_floating): Adjust.
36624 (fmtresult::fmtresult): Set max correctly in two argument ctor.
36625 (get_string_length, format_string,format_directive): Same.
36626 (pass_sprintf_length::compute_format_length): Same.
36627 (try_substitute_return_value): Simplify slightly.
36629 PR middle-end/78703
36630 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
36631 (fmtresult::operator+=): Outlined.
36632 (struct fmtresult): Add ctors.
36633 (struct conversion_spec): Rename...
36634 (struct directive): ...to this. Add and remove data members.
36635 (directive::set_width, directive::set_precision): New functions.
36636 (format_percent): Use fmtresult ctor.
36637 (get_width_and_precision): Remove.
36638 (format_integer): Make naming changes. Avoid computing width and
36640 (format_floating): Same. Adjust indentation.
36641 (format_character, format_none): New functions.
36642 (format_string): Moved character handling to format_character.
36643 (format_directive): Remove arguments, change return type.
36644 (parse_directive): New function.
36645 (pass_sprintf_length::compute_format_length): Move directive
36646 parsing to parse_directive.
36648 2017-01-23 Jakub Jelinek <jakub@redhat.com>
36650 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
36651 (assign_assembler_name_if_needed): ... this.
36652 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
36653 (assign_assembler_name_if_needed): ... this.
36654 (free_lang_data_in_cgraph): Adjust callers.
36655 * cgraphunit.c (cgraph_node::analyze): Likewise.
36656 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
36659 2017-01-23 Richard Biener <rguenther@suse.de>
36661 PR tree-optimization/79088
36662 PR tree-optimization/79188
36663 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
36664 resetting loop bounds after last path deletion. Reset loop
36665 bounds of the target loop, make code match the comments.
36666 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
36667 Make sure loops need no fixups.
36669 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
36671 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
36672 exponent support with double type for first argument.
36673 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
36674 type returned by __builtin_vec_extract_sig,
36675 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
36676 functions from "vector int" to "vector unsigned int" or from
36677 "vector long long int" to "vector unsigned long long int".
36678 Changed type returned by __builtin_vec_extract_exp,
36679 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
36680 functions from "vector int" to "vector unsigned int" or from
36681 "vector long long int" to "vector unsigned long long int".
36682 Changed return type of __builtin_vec_test_data_class,
36683 __builtin_vec_test_data_class_sp, and
36684 __builtin_vec_test_data_class_dp from "vector int" to
36685 "vector bool int" or from "vector long long int" to "vector bool
36686 long long int" and changed second argument type from "unsigned
36687 int" to "int". Added new overloaded function forms "vector float
36688 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
36689 "vector float __builtin_vec_insert_exp_sp (vector float, vector
36690 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
36691 double, vector unsigned long long int)" and "vector double
36692 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
36693 long int)". Changed return type of
36694 __builtin_scalar_test_data_class and
36695 __builtin_scalar_test_data_class_sp and
36696 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
36697 int" and changed second argument from "unsigned int" to "int".
36698 Changed type returned by __builtin_scalar_test_neg,
36699 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
36700 from "int" to "bool int". Added new overloaded function form
36701 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
36702 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
36703 exponent double-precision with floating point first argument.
36704 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
36705 documentation of scalar_test_data_class, scalar_test_neg,
36706 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
36707 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
36708 vec_test_data_class built-in functions to reflect refinements in
36709 their type signatures.
36711 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
36713 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
36715 (aarch64_elf_asm_destructor): Likewise.
36717 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
36719 PR rtl-optimization/78634
36720 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
36721 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
36722 * ifcvt.c (noce_try_cmove): Add missing cost check.
36724 PR rtl-optimization/71724
36725 * combine.c (if_then_else_cond): Look for situations where it is
36726 beneficial to undo the work of one of the recursive calls.
36728 2017-01-23 Bin Cheng <bin.cheng@arm.com>
36730 PR tree-optimization/70754
36731 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
36732 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
36733 combined stmt before it if not NULL.
36734 (combine_chains): Process refs reversely and compute dominance point
36737 2017-01-23 Martin Liska <mliska@suse.cz>
36739 PR tree-optimization/79196
36740 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
36741 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
36743 (strlen_optimize_stmt): Call the renamed function.
36745 2017-01-23 Michael Matz <matz@suse.de>
36747 PR tree-optimization/78384
36748 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
36750 2017-01-23 Richard Biener <rguenther@suse.de>
36752 PR tree-optimization/79186
36753 * tree-vrp.c (register_new_assert_for): Make sure we've seen
36754 both incoming edges before moving an assert.
36756 2017-01-23 Martin Jambor <mjambor@suse.cz>
36758 * ipa-prop.c (load_from_param_1): Removed.
36759 (load_from_unmodified_param): Bits from load_from_param_1 put back
36761 (load_from_param): Removed.
36762 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
36763 with stmt. Reverted back to use of load_from_unmodified_param.
36765 2017-01-23 Martin Jambor <mjambor@suse.cz>
36768 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
36769 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
36770 field a pointer to garbage collected vector, mark lattices and
36771 ipcp_orig_node with GTY((skip)).
36772 (ipa_get_param_count): Adjust to descriptors being a pointer.
36773 (ipa_get_param): Likewise.
36774 (ipa_get_type): Likewise.
36775 (ipa_get_param_move_cost): Likewise.
36776 (ipa_set_param_used): Likewise.
36777 (ipa_get_controlled_uses): Likewise.
36778 (ipa_set_controlled_uses): Likewise.
36779 (ipa_is_param_used): Likewise.
36780 (ipa_node_params_t): Move into garbage collector. New methods insert
36782 (ipa_node_params_sum): Annotate wth GTY(()).
36783 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
36785 (ipa_load_from_parm_agg): Adjust declaration.
36786 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
36787 * ipa-profile.c (ipa_profile): Likewise.
36788 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
36789 (ipa_populate_param_decls): Make descriptors parameter garbage
36791 (ipa_dump_param): Adjust to descriptors being a pointer.
36792 (ipa_alloc_node_params): Likewise.
36793 (ipa_initialize_node_params): Likewise.
36794 (load_from_param_1): Make descriptors parameter garbage collected.
36795 (load_from_unmodified_param): Likewise.
36796 (load_from_param): Likewise.
36797 (ipa_load_from_parm_agg): Likewise.
36798 (ipa_node_params::~ipa_node_params): Removed.
36799 (ipa_free_all_node_params): Remove call to delete operator.
36800 (ipa_node_params_t::insert): New.
36801 (ipa_node_params_t::remove): Likewise.
36802 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
36803 copy known_csts and known_contexts vectors.
36804 (ipa_read_node_info): Adjust to descriptors being a pointer.
36805 (ipcp_modif_dom_walker): Make m_descriptors field garbage
36807 (ipcp_transform_function): Make descriptors variable garbage
36810 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
36812 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
36813 * config/i386/avx512dqintrin.h: Ditto.
36814 * config/i386/avx512fintrin.h: Ditto.
36815 * config/i386/i386.c: Handle new builtins.
36816 * config/i386/i386-builtin.def: Add new builtins.
36817 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
36818 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
36820 2017-01-23 Jakub Jelinek <jakub@redhat.com>
36821 Martin Liska <mliska@suse.cz>
36823 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
36824 * asan.c (asan_expand_poison_ifn): Support stores and use
36825 appropriate ASAN report function.
36826 * internal-fn.c (expand_ASAN_POISON_USE): New function.
36827 * internal-fn.def (ASAN_POISON_USE): Declare.
36828 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
36829 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
36830 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
36831 ASAN_POISON calls w/o LHS.
36832 * tree-ssa.c (execute_update_addresses_taken): Create clobber
36833 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
36834 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
36835 * gimplify.c (asan_poison_variables): Add attribute
36836 use_after_scope_memory to variables that really needs to live
36838 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
36839 having the attribute.
36841 2017-01-23 Martin Liska <mliska@suse.cz>
36843 * asan.c (create_asan_shadow_var): New function.
36844 (asan_expand_poison_ifn): Likewise.
36845 * asan.h (asan_expand_poison_ifn): New declaration.
36846 * internal-fn.c (expand_ASAN_POISON): Likewise.
36847 * internal-fn.def (ASAN_POISON): New builtin.
36848 * sanopt.c (pass_sanopt::execute): Expand
36849 asan_expand_poison_ifn.
36850 * tree-inline.c (copy_decl_for_dup_finish): Make function
36852 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
36853 * tree-ssa.c (is_asan_mark_p): New function.
36854 (execute_update_addresses_taken): Rewrite local variables
36855 (identified just by use-after-scope as addressable) into SSA.
36857 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
36859 * doc/install.texi (Specific): opensource.apple.com uses https
36860 now. Remove trailing slash.
36862 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
36864 * README.Portability: Remove note on an Irix compatibility issue.
36866 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
36868 * gcov.c (INCLUDE_ALGORITHM): Define.
36869 (INCLUDE_VECTOR): Define.
36870 No longer include <vector> and <algorithm> directly.
36872 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
36874 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
36876 * doc/invoke.texi (Code Gen Options): Ditto.
36878 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
36881 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
36883 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
36885 rtl-optimization/79125
36886 * cprop.c (local_cprop_pass): Handle cases where we make an
36887 unconditional trap.
36889 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
36893 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
36894 read from, for big endian.
36896 2017-01-20 Jiong Wang <jiong.wang@arm.com>
36898 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
36899 register pauth builtins for LP64 only.
36901 2017-01-20 Marek Polacek <polacek@redhat.com>
36904 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
36907 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
36909 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
36911 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
36912 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
36913 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
36915 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36918 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
36919 in big-endian mode when they are not a single duplicated value.
36921 2017-01-20 Richard Biener <rguenther@suse.de>
36923 * BASE-VER: Bump to 7.0.1.
36925 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
36927 * omp-low.c (omplow_simd_context): New struct. Use it...
36928 (lower_rec_simd_input_clauses): ...here and...
36929 (lower_rec_input_clauses): ...here to hold common data. Adjust all
36930 references to idx, lane, max_vf, is_simt.
36932 2017-01-20 Graham Markall <graham.markall@embecosm.com>
36934 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
36937 2017-01-20 Martin Jambor <mjambor@suse.cz>
36939 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
36940 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
36942 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
36943 (GTFILES): Rename hsa.c to hsa-common.c.
36944 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
36945 * hsa-dump.c: Likewise.
36946 * hsa-gen.c: Likewise.
36947 * hsa-regalloc.c: Likewise.
36948 * ipa-hsa.c: Likewise.
36949 * omp-expand.c: Likewise.
36950 * omp-low.c: Likewise.
36951 * toplev.c: Likewise.
36953 2017-01-20 Marek Polacek <polacek@redhat.com>
36956 * doc/invoke.texi: Document -Wduplicated-branches.
36957 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
36958 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
36959 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
36960 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
36961 return 0 only when not OEP_LEXICOGRAPHIC.
36962 (fold_build_cleanup_point_expr): Use the expression
36963 location when building CLEANUP_POINT_EXPR.
36964 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
36965 * tree.c (add_expr): Handle error_mark_node.
36967 2017-01-20 Martin Liska <mliska@suse.cz>
36970 * tree-profile.c (init_ic_make_global_vars): Do not call
36972 (gimple_init_gcov_profiler): Likewise.
36974 2017-01-20 Martin Liska <mliska@suse.cz>
36977 * cgraph.h (maybe_create_reference): Remove argument and
36979 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
36981 * ipa-cp.c (create_specialized_node): Likewise.
36982 * symtab.c (symtab_node::maybe_create_reference): Handle
36983 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
36985 2017-01-20 Martin Liska <mliska@suse.cz>
36987 * read-rtl-function.c (function_reader::create_function): Use
36988 build_decl instread of build_decl_stat.
36990 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
36992 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
36993 * config/i386/avx512dqintrin.h: Ditto.
36994 * config/i386/avx512fintrin.h: Ditto.
36995 * config/i386/i386-builtin-types.def: Add new types.
36996 * config/i386/i386.c: Handle new types.
36997 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
36998 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
36999 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
37000 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
37001 (__builtin_ia32_kshiftridi): New.
37002 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
37004 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
37008 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
37009 define to rs6000_init_stack_protect_guard.
37010 (rs6000_init_stack_protect_guard): New function.
37012 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
37013 Yunqiang Su <yunqiang.su@imgtec.com>
37015 * config.gcc (supported_defaults): Add madd4.
37016 (with_madd4): Add validation.
37017 (all_defaults): Add madd4.
37018 * config/mips/mips.opt (mmadd4): New option.
37019 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
37021 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
37023 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
37024 (ISA_HAS_FUSED_MADD4): Likewise.
37025 * doc/invoke.texi (-mmadd4): Document the new option.
37026 * doc/install.texi (--with-madd4): Document the new option.
37028 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37030 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
37031 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
37032 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
37033 (aarch64_init_pauth_hint_builtins): New.
37034 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
37035 (aarch64_expand_builtin): Expand new builtins.
37037 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37039 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
37040 * combine-stack-adj.c (no_unhandled_cfa): Handle
37041 REG_CFA_TOGGLE_RA_MANGLE.
37042 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
37043 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
37044 info for return address signing.
37045 (aarch64_expand_epilogue): Likewise.
37047 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37049 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
37050 * config/aarch64/aarch64-protos.h
37051 (aarch64_return_address_signing_enabled): New declaration.
37052 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
37054 (aarch64_expand_prologue): Sign return address before it's pushed onto
37056 (aarch64_expand_epilogue): Authenticate return address fetched from
37058 (aarch64_override_options): Sanity check for ILP32 and ISA level.
37059 (aarch64_attributes): New function attributes for "sign-return-address".
37060 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
37061 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
37062 ("*do_return"): Generate combined instructions according to key index.
37063 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
37064 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
37066 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
37067 * config/aarch64/aarch64.opt (msign-return-address=): New.
37068 * doc/extend.texi (AArch64 Function Attributes): Documents
37069 "sign-return-address=".
37070 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
37072 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
37074 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
37075 overall option summary.
37077 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37079 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
37080 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
37081 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
37082 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
37084 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
37086 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
37087 -mpower9-minmax by default for -mcpu=power9.
37088 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
37089 128-bit floating point.
37091 2017-01-20 Alan Modra <amodra@gmail.com>
37093 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
37094 optimizing for size.
37096 2017-01-20 Alan Modra <amodra@gmail.com>
37099 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
37100 for strcmp and strncmp from corresponding builtin decl.
37102 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
37104 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
37105 instead of i386/rtems-64.h.
37106 * config/i386/rtems-64.h: Remove.
37108 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
37112 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
37114 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
37116 2017-01-19 Tamar Christina <tamar.christina@arm.com>
37118 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
37119 Change int to HOST_WIDE_INT.
37120 * config/aarch64/aarch64-protos.h
37121 (aarch64_simd_gen_const_vector_dup): Likewise.
37122 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
37124 2017-01-19 David Malcolm <dmalcolm@redhat.com>
37126 * langhooks-def.h (lhd_type_for_size): New decl.
37127 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
37128 * langhooks.c (lhd_type_for_size): New function, taken from
37131 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
37133 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
37134 define_bypass for CR latency.
37135 (power9-cracked-alu): Update bypass latency and remove power9-branch.
37136 (power9-alu2): Add define_bypass for CR latency.
37138 (power9-mul): Update insn latency.
37139 (power9-mul-compare): Update insn latency, bypass latency and remove
37142 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37144 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
37146 * config/aarch64/aarch64.md
37147 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
37148 aarch64_nopcrelative_literal_loads.
37149 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
37151 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
37153 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
37154 TARGET_LOONGSON_3A.
37155 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
37157 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
37160 * config.gcc (supported_defaults): Add lxc1-sxc1.
37161 (with_lxc1_sxc1): Add validation.
37162 (all_defaults): Add lxc1-sxc1.
37163 * config/mips/mips.opt (mlxc1-sxc1): New option.
37164 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
37166 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
37167 __mips_no_lxc1_sxc1.
37168 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
37169 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
37170 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
37172 2017-01-19 Richard Biener <rguenther@suse.de>
37174 PR tree-optimization/72488
37175 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
37176 sure to restore SSA info.
37177 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
37179 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
37181 PR rtl-optimization/79121
37182 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
37183 of the inner type when shifting an extended value.
37185 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
37188 * symtab.c (symtab_node::equal_address_to): Fix comparing of
37189 interposable aliases.
37191 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
37194 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
37195 Use the evmergelohi instruction.
37196 (mov_si<mode>_e500_subreg4_2_le): Likewise.
37197 (mov_sitf_e500_subreg8_2_be): Likewise.
37198 (mov_sitf_e500_subreg12_2_le): Likewise.
37199 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
37200 (mov_si<mode>_e500_subreg4_2_be): Likewise.
37201 (mov_sitf_e500_subreg8_2_le): Likewise.
37202 (mov_sitf_e500_subreg12_2_be): Likewise.
37204 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37206 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
37207 attribute from vecsimple to vecperm.
37208 (altivec_vbpermq2): Likewise.
37210 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37213 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
37215 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37216 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
37217 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
37218 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
37219 case where N arg is SIZE_MAX.
37220 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
37221 (cmpstrsi): Add pattern.
37223 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
37225 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
37226 __builtin_vec_revb builtins.
37227 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
37228 built-in functions to support generation of the ISA 3.0 XXBR<x>
37229 vector byte reverse instructions.
37230 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
37231 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
37232 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
37233 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
37234 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
37235 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
37236 (P9V_BUILTIN_VEC_REVB): Likewise.
37237 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
37238 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
37239 (p9_xxbrq_v16qi): Likewise.
37240 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
37241 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
37242 (p9_xxbrh_v8hi): Likewise.
37243 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
37244 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
37245 vec_revb built-in functions.
37247 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
37249 PR rtl-optimization/78952
37250 * config/i386/i386.md (any_extract): New code iterator.
37251 (*insvqi_2): Use any_extract for source operand.
37252 (*insvqi_3): Use any_shiftrt for source operand.
37254 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
37256 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
37258 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
37260 2017-01-18 Matthias Klose <doko@ubuntu.com>
37262 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
37264 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37266 * config/rs6000/altivec.h (vec_bperm): Change #define.
37267 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
37268 (altivec_vbpermq2): New define_insn.
37269 (altivec_vbpermd): Likewise.
37270 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
37271 function interface.
37272 (VBPERMD): Likewise.
37273 (VBPERM): New polymorphic function interface.
37274 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
37275 Add entries for P9V_BUILTIN_VEC_VBPERM.
37276 * doc/extend.texi: Add interfaces for vec_bperm.
37278 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37280 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
37281 first letter of error messages.
37282 (s390_resolve_overloaded_builtin): Likewise.
37283 * config/s390/s390.c (s390_expand_builtin): Likewise.
37284 (s390_invalid_arg_for_unprototyped_fn): Likewise.
37285 (s390_valid_target_attribute_inner_p): Likewise.
37286 * config/s390/s390.md ("tabort"): Likewise.
37288 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
37290 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
37291 (ISA_AVOID_DIV_HILO): New macro.
37292 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
37293 (ISA_HAS_DDIV): Likewise.
37295 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37297 * doc/invoke.texi (fabi-version): Correct number of occurrences.
37299 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37301 * doc/invoke.texi (fabi-version): Spelling fix.
37303 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37306 * doc/invoke.texi (fabi-version): Mention mangling fix for
37309 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37312 * doc/invoke.texi (fabi-version): Document discriminator mangling.
37314 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
37317 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
37318 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
37320 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
37322 (stack_protect_test): Ditto.
37323 * config/rs6000/rs6000.opt (mstack-protector-guard=,
37324 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
37326 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
37327 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
37328 -mstack-protector-guard-offset=.
37329 (RS/6000 and PowerPC Options): Ditto.
37331 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
37333 * config/i386/i386.h (MASK_CLASS_P): New define.
37334 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
37335 there are no registers from different register sets also when
37336 mask registers are used. Update function comment.
37337 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
37338 to (*k/*r) and (*k/*km) alternatives.
37340 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
37342 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
37343 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
37344 (EH_RETURN_HANDLER_RTX): New define.
37345 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
37346 Force frame pointer in EH return functions.
37347 (aarch64_expand_epilogue): Add barrier for eh_return.
37348 (aarch64_final_eh_return_addr): Remove.
37349 (aarch64_eh_return_handler_rtx): New function.
37350 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
37352 (aarch64_eh_return_handler_rtx): New prototype.
37354 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37356 * config/rs6000/altivec.h (vec_rlmi): New #define.
37357 (vec_vrlnm): Likewise.
37358 (vec_rlnm): Likewise.
37359 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
37360 (UNSPEC_VRLNM): Likewise.
37361 (VIlong): New mode iterator.
37362 (altivec_vrl<VI_char>mi): New define_insn.
37363 (altivec_vrl<VI_char>nm): Likewise.
37364 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
37366 (VRLDNM): Likewise.
37367 (RLNM): New polymorphic function entry.
37368 (VRLWMI): New monomorphic function entry.
37369 (VRLDMI): Likewise.
37370 (RLMI): New polymorphic function entry.
37371 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
37372 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
37373 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
37376 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37379 * dwarf2out.c (field_byte_offset): Restore the
37380 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
37381 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
37382 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
37385 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37388 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
37390 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37393 * dwarf2out.c (add_data_member_location_attribute): For constant
37394 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
37395 instead of DW_AT_data_member_location, DW_AT_bit_offset and
37396 DW_AT_byte_size attributes.
37398 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37400 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
37401 after forcing to constant memory when the code model is medium.
37403 2017-01-17 Julia Koval <julia.koval@intel.com>
37406 * config/i386/avx512fintrin.h
37407 (_mm512_i32gather_ps): Change __addr type to void const*.
37408 (_mm512_mask_i32gather_ps): Ditto.
37409 (_mm512_i32gather_pd): Ditto.
37410 (_mm512_mask_i32gather_pd): Ditto.
37411 (_mm512_i64gather_ps): Ditto.
37412 (_mm512_mask_i64gather_ps): Ditto.
37413 (_mm512_i64gather_pd): Ditto.
37414 (_mm512_mask_i64gather_pd): Ditto.
37415 (_mm512_i32gather_epi32): Ditto.
37416 (_mm512_mask_i32gather_epi32): Ditto.
37417 (_mm512_i32gather_epi64): Ditto.
37418 (_mm512_mask_i32gather_epi64): Ditto.
37419 (_mm512_i64gather_epi32): Ditto.
37420 (_mm512_mask_i64gather_epi32): Ditto.
37421 (_mm512_i64gather_epi64): Ditto.
37422 (_mm512_mask_i64gather_epi64): Ditto.
37423 (_mm512_i32scatter_ps): Change __addr type to void*.
37424 (_mm512_mask_i32scatter_ps): Ditto.
37425 (_mm512_i32scatter_pd): Ditto.
37426 (_mm512_mask_i32scatter_pd): Ditto.
37427 (_mm512_i64scatter_ps): Ditto.
37428 (_mm512_mask_i64scatter_ps): Ditto.
37429 (_mm512_i64scatter_pd): Ditto.
37430 (_mm512_mask_i64scatter_pd): Ditto.
37431 (_mm512_i32scatter_epi32): Ditto.
37432 (_mm512_mask_i32scatter_epi32): Ditto.
37433 (_mm512_i32scatter_epi64): Ditto.
37434 (_mm512_mask_i32scatter_epi64): Ditto.
37435 (_mm512_i64scatter_epi32): Ditto.
37436 (_mm512_mask_i64scatter_epi32): Ditto.
37437 (_mm512_i64scatter_epi64): Ditto.
37438 (_mm512_mask_i64scatter_epi64): Ditto.
37439 * config/i386/avx512pfintrin.h
37440 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
37441 (_mm512_mask_prefetch_i32gather_ps): Ditto.
37442 (_mm512_mask_prefetch_i64gather_pd): Ditto.
37443 (_mm512_mask_prefetch_i64gather_ps): Ditto.
37444 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
37445 (_mm512_prefetch_i32scatter_ps): Ditto.
37446 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
37447 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
37448 (_mm512_prefetch_i64scatter_pd): Ditto.
37449 (_mm512_prefetch_i64scatter_ps): Ditto.
37450 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
37451 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
37452 * config/i386/avx512vlintrin.h
37453 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
37454 (_mm_mmask_i32gather_ps): Ditto.
37455 (_mm256_mmask_i32gather_pd): Ditto.
37456 (_mm_mmask_i32gather_pd): Ditto.
37457 (_mm256_mmask_i64gather_ps): Ditto.
37458 (_mm_mmask_i64gather_ps): Ditto.
37459 (_mm256_mmask_i64gather_pd): Ditto.
37460 (_mm_mmask_i64gather_pd): Ditto.
37461 (_mm256_mmask_i32gather_epi32): Ditto.
37462 (_mm_mmask_i32gather_epi32): Ditto.
37463 (_mm256_mmask_i32gather_epi64): Ditto.
37464 (_mm_mmask_i32gather_epi64): Ditto.
37465 (_mm256_mmask_i64gather_epi32): Ditto.
37466 (_mm_mmask_i64gather_epi32): Ditto.
37467 (_mm256_mmask_i64gather_epi64): Ditto.
37468 (_mm_mmask_i64gather_epi64): Ditto.
37469 (_mm256_i32scatter_ps): Change __addr type to void*.
37470 (_mm256_mask_i32scatter_ps): Ditto.
37471 (_mm_i32scatter_ps): Ditto.
37472 (_mm_mask_i32scatter_ps): Ditto.
37473 (_mm256_i32scatter_pd): Ditto.
37474 (_mm256_mask_i32scatter_pd): Ditto.
37475 (_mm_i32scatter_pd): Ditto.
37476 (_mm_mask_i32scatter_pd): Ditto.
37477 (_mm256_i64scatter_ps): Ditto.
37478 (_mm256_mask_i64scatter_ps): Ditto.
37479 (_mm_i64scatter_ps): Ditto.
37480 (_mm_mask_i64scatter_ps): Ditto.
37481 (_mm256_i64scatter_pd): Ditto.
37482 (_mm256_mask_i64scatter_pd): Ditto.
37483 (_mm_i64scatter_pd): Ditto.
37484 (_mm_mask_i64scatter_pd): Ditto.
37485 (_mm256_i32scatter_epi32): Ditto.
37486 (_mm256_mask_i32scatter_epi32): Ditto.
37487 (_mm_i32scatter_epi32): Ditto.
37488 (_mm_mask_i32scatter_epi32): Ditto.
37489 (_mm256_i32scatter_epi64): Ditto.
37490 (_mm256_mask_i32scatter_epi64): Ditto.
37491 (_mm_i32scatter_epi64): Ditto.
37492 (_mm_mask_i32scatter_epi64): Ditto.
37493 (_mm256_i64scatter_epi32): Ditto.
37494 (_mm256_mask_i64scatter_epi32): Ditto.
37495 (_mm_i64scatter_epi32): Ditto.
37496 (_mm_mask_i64scatter_epi32): Ditto.
37497 (_mm256_i64scatter_epi64): Ditto.
37498 (_mm256_mask_i64scatter_epi64): Ditto.
37499 (_mm_i64scatter_epi64): Ditto.
37500 (_mm_mask_i64scatter_epi64): Ditto.
37501 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
37502 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
37503 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
37504 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
37505 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
37506 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
37507 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
37508 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
37509 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
37510 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
37511 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
37512 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
37513 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
37514 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
37515 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
37516 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
37517 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
37518 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
37519 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
37520 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
37521 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
37522 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
37523 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
37524 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
37525 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
37526 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
37527 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
37528 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
37529 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
37530 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
37531 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
37532 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
37533 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
37534 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
37535 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
37536 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
37537 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
37538 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
37539 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
37540 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
37541 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
37542 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
37543 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
37544 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
37545 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
37546 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
37547 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
37548 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
37549 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
37550 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
37551 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
37552 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
37553 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
37554 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
37555 definitions accordingly.
37557 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
37558 Kuan-Lin Chen <kuanlinchentw@gmail.com>
37561 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
37564 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
37567 * ira-conflicts.c (ira_build_conflicts): Update total conflict
37568 hard regs for inner regno.
37570 2017-01-17 Martin Liska <mliska@suse.cz>
37573 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
37574 assumption and add comment.
37576 2017-01-17 Nathan Sidwell <nathan@acm.org>
37578 * ipa-visibility.c (localize_node): New function, broken out of ...
37579 (function_and_variable_visibility): ... here. Call it.
37581 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
37583 PR middle-end/77445
37584 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
37585 correctly set frequency of oudgoing edge.
37586 (duplicate_thread_path): Fix profile updating.
37588 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37591 * configure.ac: Add GCC_BASE_VER.
37592 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
37593 version from BASE-VER file.
37594 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
37595 (gcc.o): Depend on $(BASEVER).
37596 * common.opt (dumpfullversion): New option.
37597 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
37598 * doc/invoke.texi: Document -dumpfullversion.
37599 * doc/install.texi: Document --with-gcc-major-version-only.
37600 * configure: Regenerated.
37602 2017-01-17 Richard Biener <rguenther@suse.de>
37604 PR tree-optimization/71433
37605 * tree-vrp.c (register_new_assert_for): Merge same asserts
37606 on all incoming edges.
37607 (process_assert_insertions_for): Handle insertions at the
37610 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
37612 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
37613 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
37615 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
37618 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
37621 2017-01-17 Alan Modra <amodra@gmail.com>
37624 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
37625 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
37626 symbolic stack limit when pic.
37628 2017-01-16 Martin Sebor <msebor@redhat.com>
37630 PR tree-optimization/78608
37631 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
37633 2017-01-16 Jeff Law <law@redhat.com>
37636 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
37637 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
37638 for several include directories that may be relative to sysroot.
37639 * config/i386/x-mingw32 (gplus_includedir): Define.
37640 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
37641 (native_system_includedir): Likewise.
37642 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
37643 override if TARGET_SYSTEM_ROOT is defined.
37644 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
37646 PR tree-optimization/79090
37647 PR tree-optimization/33562
37648 PR tree-optimization/61912
37649 PR tree-optimization/77485
37650 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
37651 and computed trims into the dump file.
37653 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
37655 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
37657 2017-01-16 Jakub Jelinek <jakub@redhat.com>
37660 * gimplify.c (gimplify_init_constructor): If want_value and
37661 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
37665 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
37666 sequence. Formatting fixes.
37667 (doloop_optimize): Formatting fixes.
37670 * gcc.c (debug_level_greater_than_spec_func): New function.
37671 (static_spec_functions): Add debug-level-gt spec function.
37672 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
37674 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37675 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37676 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
37677 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
37678 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
37679 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
37681 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
37683 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
37684 QImode fixups to general and mask registers only.
37686 2017-01-16 Carl Love <cel@us.ibm.com>
37688 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
37689 for built-in functions
37690 vector signed char vec_nabs (vector signed char)
37691 vector signed short vec_nabs (vector signed short)
37692 vector signed int vec_nabs (vector signed int)
37693 vector signed long long vec_nabs (vector signed long long)
37694 vector float vec_nabs (vector float)
37695 vector double vec_nabs (vector double)
37696 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
37698 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
37699 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
37700 * doc/extend.texi: Update the documentation file for the new built-in
37703 2017-01-16 Martin Sebor <msebor@redhat.com>
37705 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
37708 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37710 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
37711 UNSPEC_VSX__XXSPLTD to require special splat handling.
37713 2017-01-16 David Malcolm <dmalcolm@redhat.com>
37716 * system.h: Poison strndup.
37718 2017-01-16 Alan Modra <amodra@gmail.com>
37721 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
37724 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
37726 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
37728 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
37730 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
37731 call recog here. Assert that INSN_CODE (insn) is non-negative.
37733 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
37736 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
37738 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
37739 in the currently scheduled RTL fragment.
37741 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
37743 PR rtl-optimization/78751
37744 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
37747 2017-01-14 Jeff Law <law@redhat.com>
37749 PR tree-optimization/79090
37750 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
37751 variable length stores.
37752 (compute_trims): Delete dead assignment to *trim_tail.
37753 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
37756 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
37758 PR rtl-optimization/78626
37759 PR rtl-optimization/78727
37760 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
37761 of a block, and split such blocks after everything else is finished.
37763 2017-01-14 Alan Modra <amodra@gmail.com>
37766 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
37767 target legitimate_combined_insn.
37768 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
37769 (rs6000_legitimate_combined_insn): New function.
37770 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
37772 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
37773 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
37774 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
37776 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
37778 * doc/frontends.texi (G++ and GCC): Remove references to Java.
37780 2017-01-13 Jeff Law <law@redhat.com>
37782 PR tree-optimization/33562
37783 PR tree-optimization/61912
37784 PR tree-optimization/77485
37785 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
37787 (delete_dead_assignment): Likewise.
37788 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
37789 statement to delete_dead_call and delete_dead_assignment.
37791 2017-01-13 David Malcolm <dmalcolm@redhat.com>
37794 * substring-locations.c (format_warning_va): Strengthen case 1 so
37795 that both endpoints of the substring must be within the format
37796 range for just the substring to be printed.
37798 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
37800 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
37801 * config/i386/i386.c (ix86_target_string): Add missing options
37802 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
37803 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
37804 flags_other and ix86_target_other to flags2_other. Display unknown
37806 (ix86_valid_target_attribute_inner_p): Add missing options and
37807 reorder options by implied ISAs, as in ix86_target_string.
37809 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
37811 * hash-table.h (hash_table::too_empty_p): New function.
37812 (hash_table::expand): Use it.
37813 (hash_table::traverse): Likewise.
37814 (hash_table::empty_slot): Use sizeof (value_type) instead of
37815 sizeof (PTR) to convert bytes to elements. Shrink the table
37816 if the current size is excessive for the current number of
37819 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
37821 * ira-costs.c (record_reg_classes): Break from the inner loop
37822 early once alt_fail is known to be true. Update outer loop
37823 handling accordingly.
37825 2017-01-13 Jeff Law <law@redhat.com>
37827 * tree-ssa-dse.c (decrement_count): New function.
37828 (increment_start_addr, maybe_trim_memstar_call): Likewise.
37829 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
37830 when we know the partially dead statement is a mem* function.
37832 PR tree-optimization/61912
37833 PR tree-optimization/77485
37834 * tree-ssa-dse.c: Include expr.h.
37835 (maybe_trim_constructor_store): New function.
37836 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
37838 PR tree-optimization/33562
37839 PR tree-optimization/61912
37840 PR tree-optimization/77485
37841 * doc/invoke.texi: Document new dse-max-object-size param.
37842 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
37843 * tree-ssa-dse.c: Include params.h.
37844 (dse_store_status): New enum.
37845 (initialize_ao_ref_for_dse): New, partially extracted from
37847 (valid_ao_ref_for_dse, normalize_ref): New.
37848 (setup_live_bytes_from_ref, compute_trims): Likewise.
37849 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
37850 (maybe_trim_partially_dead_store): Likewise.
37851 (maybe_trim_complex_store): Likewise.
37852 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
37853 Track what bytes live from the original store. Return tri-state
37854 for dead, partially dead or live.
37855 (dse_dom_walker): Add constructor, destructor and new private members.
37856 (delete_dead_call, delete_dead_assignment): New extracted from
37858 (dse_optimize_stmt): Make a member of dse_dom_walker.
37859 Use initialize_ao_ref_for_dse.
37861 PR tree-optimization/33562
37862 PR tree-optimization/61912
37863 PR tree-optimization/77485
37864 * sbitmap.h (bitmap_count_bits): Prototype.
37865 (bitmap_clear_range, bitmap_set_range): Likewise.
37866 * sbitmap.c (bitmap_clear_range): New function.
37867 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
37869 2017-01-13 Martin Liska <mliska@suse.cz>
37872 * function.c (set_cfun): Add new argument force.
37873 * function.h (set_cfun): Likewise.
37874 * ipa-inline-transform.c (inline_call): Use the function when
37875 strict alising from is dropped for function we inline to.
37877 2017-01-13 Richard Biener <rguenther@suse.de>
37879 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
37880 for dumping GIMPLE INTEGER_CSTs.
37882 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
37884 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
37885 to 201112L since C++17.
37887 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
37890 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
37891 if -fsanitize=kernel-address is present.
37893 2017-01-13 Richard Biener <rguenther@suse.de>
37895 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
37896 as _Literal ( type ) number in case usual suffixes do not
37897 preserve all information.
37899 2017-01-13 Richard Biener <rguenther@suse.de>
37901 PR tree-optimization/77283
37902 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
37903 and ssa-iterators.h.
37904 (is_feasible_trace): Implement a cost model based on joiner
37907 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
37910 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
37911 char or short to __float128/_Float128 directly.
37913 2017-01-12 Martin Sebor <msebor@redhat.com>
37915 to -Wformat-overflow.
37916 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
37917 (min_bytes_remaining): Same.
37918 (get_string_length): Same.
37919 (format_string): Same.
37920 (format_directive): Same.
37922 (pass_sprintf_length::handle_gimple_call): Same.
37924 2017-01-12 Jakub Jelinek <jakub@redhat.com>
37926 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
37927 info.nowrite calls with no lhs that can't throw. Return bool
37928 whether gsi_remove has been called or not.
37929 (pass_sprintf_length::handle_gimple_call): Return bool whether
37930 try_substitute_return_value called gsi_remove. Formatting fix.
37931 (pass_sprintf_length::execute): Don't use gsi_remove if
37932 handle_gimple_call returned true.
37935 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
37936 be removed due to side-effects, don't remove following barrier nor
37937 turn the successor edge into fallthru edge.
37939 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37942 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
37943 element-reversing loads and stores as not swappable.
37945 2017-01-12 Nathan Sidwell <nathan@acm.org>
37946 Nicolai Stange <nicstange@gmail.com>
37948 * combine.c (try_combine): Don't ignore result of overlap checking
37949 loop. Combine overlap & asm check into single loop.
37951 2017-01-12 Richard Biener <rguenther@suse.de>
37953 * tree-pretty-print.c (dump_generic_node): Provide -gimple
37954 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
37956 2017-01-12 Richard Biener <rguenther@suse.de>
37958 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
37959 and TS_TARGET_OPTION directly derive from TS_BASE.
37960 * tree-core.h (tree_optimization_option): Derive from tree_base.
37961 (tree_target_option): Likewise.
37963 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
37965 * config/i386/i386.c (memory_address_length): Increase len
37966 only when rip_relative_addr_p returns false.
37968 2017-01-11 Julia Koval <julia.koval@intel.com>
37970 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
37971 (OPTION_MASK_ISA_SGX_SET): New.
37972 (ix86_handle_option): Handle OPT_msgx.
37973 * config.gcc: Added sgxintrin.h.
37974 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
37975 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
37976 * config/i386/i386.c (ix86_target_string): Add -msgx.
37978 (ix86_option_override_internal): Handle new options.
37979 (ix86_valid_target_attribute_inner_p): Add sgx.
37980 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
37981 * config/i386/i386.opt: Add msgx.
37982 * config/i386/sgxintrin.h: New file.
37983 * config/i386/x86intrin.h: Add sgxintrin.h.
37985 2017-01-11 Jakub Jelinek <jakub@redhat.com>
37988 * fold-const.c (maybe_nonzero_address): Return 1 for function
37990 (tree_single_nonzero_warnv_p): Don't handle function local objects
37994 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
37997 2017-01-11 David Malcolm <dmalcolm@redhat.com>
38000 * opts.c: Include "spellcheck.h"
38001 (struct string_fragment): New struct.
38002 (struct edit_distance_traits<const string_fragment &>): New
38004 (get_closest_sanitizer_option): New function.
38005 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
38007 2017-01-11 Jakub Jelinek <jakub@redhat.com>
38009 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
38011 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
38012 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
38013 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
38014 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
38015 for initial die_offset if dwarf_split_debug_info.
38016 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
38017 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
38019 (output_skeleton_debug_sections): Formatting fix. Use
38020 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
38021 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
38023 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
38025 * config/arm/cortex-a53.md: Add bypasses for
38026 cortex_a53_r2f_cvt.
38027 (cortex_a53_r2f): Only use for transfers.
38028 (cortex_a53_f2r): Likewise.
38029 (cortex_a53_r2f_cvt): Add reservation for conversions.
38030 (cortex_a53_f2r_cvt): Likewise.
38032 2017-01-11 Tamar Christina <tamar.christina@arm.com>
38034 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
38035 to all inlined functions, change static to extern.
38037 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
38040 * config/arm/arm.c (legitimize_pic_address): Handle reference to
38042 (arm_assemble_integer): Likewise.
38044 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
38046 * config.gcc: Use new awk script to check CPU, FPU and architecture
38047 parameters for --with-... options.
38048 * config/arm/parsecpu.awk: New file
38049 * config/arm/arm-cpus.in: New file.
38050 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
38052 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
38054 * config/arm/t-arm: Update dependency rules.
38055 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
38056 of processing .def files.
38057 * config/arm/genopt.sh: Deleted.
38058 * config/arm/gentune.sh: Deleted.
38059 * config/arm/arm-cores.def: Deleted.
38060 * config/arm/arm-arches.def: Deleted.
38061 * config/arm/arm-fpus.def: Deleted.
38062 * config/arm/arm-tune.md: Regenerated.
38063 * config/arm/arm-tables.opt: Regenerated.
38064 * config/arm/arm-cpu.h: New generated file.
38065 * config/arm/arm-cpu-data.h: New generated file.
38066 * config/arm/arm-cpu-cdata.h: New generated file.
38068 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
38071 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
38073 (input_varpool_node): Unpack dynamically_initialized bit.
38075 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
38077 PR rtl-optimization/79032
38078 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
38079 the alignment of the adjusted memory reference against that of MODE,
38080 instead of the alignment of the original memory reference.
38082 2017-01-11 Martin Jambor <mjambor@suse.cz>
38084 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
38086 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
38087 decorated functions.
38089 2017-01-11 Richard Biener <rguenther@suse.de>
38091 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
38092 set range/nonnull info for PHI results. Do not set it on
38093 stmts marked for removal.
38095 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
38097 * expr.c (store_field): In the bitfield case, fetch the return value
38098 from the registers before applying a single big-endian adjustment.
38099 Always do a final load for a BLKmode value not larger than a word.
38101 2017-01-10 David Malcolm <dmalcolm@redhat.com>
38104 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
38105 that we correctly handle column numbers greater than
38106 LINE_MAP_MAX_COLUMN_NUMBER.
38108 2017-01-10 Martin Sebor <msebor@redhat.com>
38110 PR middle-end/78245
38111 * gimple-ssa-sprintf.c (get_destination_size): Call
38112 {init,fini}object_sizes.
38113 * tree-object-size.c (addr_object_size): Adjust.
38114 (pass_through_call): Adjust.
38115 (pass_object_sizes::execute): Adjust.
38116 * tree-object-size.h (fini_object_sizes): Declare.
38118 2017-01-10 Martin Sebor <msebor@redhat.com>
38120 PR tree-optimization/78775
38121 * builtins.c (get_size_range): Move...
38122 * calls.c: ...to here.
38123 (alloc_max_size): Accept zero argument.
38124 (operand_signed_p): Remove.
38125 (maybe_warn_alloc_args_overflow): Call get_size_range.
38126 * calls.h (get_size_range): Declare.
38128 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
38130 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
38131 from TI's devices.csv file as of September 2016.
38132 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
38134 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
38136 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
38137 * doc/invoke.texi: Likewise.
38138 * doc/md.texi: Likewise.
38139 * doc/objc.texi: Likewise.
38141 2017-01-10 Joshua Conner <joshconner@google.com>
38143 * config/arm/fuchsia-elf.h: New file.
38144 * config/fuchsia.h: New file.
38145 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
38146 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
38148 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
38150 2016-01-10 Richard Biener <rguenther@suse.de>
38152 PR tree-optimization/79034
38153 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
38154 Propagate out degenerate PHIs in the joiner.
38156 2017-01-10 Martin Liska <mliska@suse.cz>
38158 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
38159 (sort_congruence_classes_by_decl_uid): Likewise.
38160 (sort_congruence_class_groups_by_decl_uid): Likewise.
38161 (sem_item_optimizer::merge_classes): Sort class, groups in these
38162 classes and members in the groups by DECL_UID of declarations.
38163 This would make merge operations stable.
38165 2017-01-10 Martin Liska <mliska@suse.cz>
38167 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
38168 usage of m_classes_vec.
38169 (sem_item_optimizer::~sem_item_optimizer): Likewise.
38170 (sem_item_optimizer::get_group_by_hash): Likewise.
38171 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
38172 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
38173 (sem_item_optimizer::verify_classes): Likewise.
38174 (sem_item_optimizer::process_cong_reduction): Likewise.
38175 (sem_item_optimizer::dump_cong_classes): Likewise.
38176 (sem_item_optimizer::merge_classes): Likewise.
38177 * ipa-icf.h (congruence_class_hash): Rename from
38178 congruence_class_group_hash. Remove declaration of m_classes_vec.
38180 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
38182 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
38183 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
38184 * config.gcc: Add avx512vpopcntdqintrin.h.
38185 * config/i386/avx512vpopcntdqintrin.h: New.
38186 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
38187 * config/i386/i386-builtin-types.def: Add new types.
38188 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
38189 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
38190 __builtin_ia32_vpopcountq_v8di_mask): New.
38191 * config/i386/i386-c.c (ix86_target_macros_internal): Define
38192 __AVX512VPOPCNTDQ__.
38193 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
38194 (PTA_AVX512VPOPCNTDQ): Define.
38195 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
38196 TARGET_AVX512VPOPCNTDQ_P): Define.
38197 * config/i386/i386.opt: Add mavx512vpopcntdq.
38198 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
38199 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
38201 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38203 PR middle-end/77484
38204 * predict.def (PRED_CALL): Set to 67.
38206 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
38208 * expr.c (store_field): In the bitfield case, if the value comes from
38209 a function call and is of an aggregate type returned in registers, do
38210 not modify the field mode; extract the value in all cases if the mode
38211 is BLKmode and the size is not larger than a word.
38213 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
38216 * config/i386/cpuid.h: Fix undefined behavior.
38218 2017-01-04 Jeff Law <law@redhat.com>
38220 PR tree-optimization/79007
38221 PR tree-optimization/67955
38222 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
38223 conservative for pt.null when flag_non_call_exceptions is on.
38225 2017-01-09 Jakub Jelinek <jakub@redhat.com>
38227 PR translation/79019
38228 PR translation/79020
38229 * params.def (PARAM_INLINE_MIN_SPEEDUP,
38230 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
38231 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
38233 * config/avr/avr.opt (maccumulate-args): Likewise.
38234 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
38235 * common.opt (freport-bug): Likewise.
38236 * cif-code.def (CIF_FINAL_ERROR): Likewise.
38237 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
38238 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
38239 translatable string.
38240 * config/i386/i386.c (function_value_32): Likewise.
38241 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
38242 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
38244 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
38245 * common/config/msp430/msp430-common.c (msp430_handle_option):
38247 * symtab.c (symtab_node::verify_base): Likewise.
38248 * opts.c (set_debug_level): Likewise.
38249 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
38250 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
38251 missing whitespace to translatable strings.
38252 * config/avr/avr.md (bswapsi2): Fix typo in comment.
38253 * config/sh/superh.h: Likewise.
38254 * config/i386/xopintrin.h: Likewise.
38255 * config/i386/znver1.md: Likewise.
38256 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
38257 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
38258 * double-int.h (struct double_int): Likewise.
38259 * double-int.c (div_and_round_double): Likewise.
38260 * wide-int.cc: Likewise.
38261 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
38262 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
38263 * cfgcleanup.c (crossjumps_occured): Renamed to ...
38264 (crossjumps_occurred): ... this.
38265 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
38268 PR tree-optimization/78899
38269 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
38270 returning bool return struct loop *, NULL for failure and the new
38272 (versionable_outer_loop_p): Don't version outer loop if it has
38273 dont_vectorized bit set.
38274 (tree_if_conversion): When versioning outer loop, ensure
38275 tree_if_conversion is performed also on the inner loop of the
38276 non-vectorizable outer loop copy.
38277 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
38278 LOOP_VECTORIZED in inner loop of the scalar outer loop and
38279 prevent vectorization of it.
38280 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
38281 the outer loop vectorization of the non-scalar version is attempted
38282 before vectorization of the inner loop in scalar version. If
38283 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
38284 vectorization of its inner loop.
38285 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
38286 has 2 inner loops, rename also on edges from bb whose single pred
38287 is outer_loop->header. Fix typo in function comment.
38289 2017-01-09 Martin Sebor <msebor@redhat.com>
38292 * asan.c (asan_emit_stack_protection): Increase local buffer size
38293 to avoid snprintf truncation warning.
38295 2017-01-09 Andrew Pinski <apinski@cavium.com>
38297 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
38298 to reference thunderx2t99 for the tuning structure
38299 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
38301 (thunderx2t99_extra_costs): This.
38302 * config/aarch64/aarch64-tune.md: Regenerate.
38303 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
38304 (vulcan_addrcost_table): This.
38305 (vulcan_regmove_cost): Rename to ...
38306 (thunderx2t99_regmove_cost): This.
38307 (vulcan_vector_cost): Rename to ...
38308 (thunderx2t99_vector_cost): this.
38309 (vulcan_branch_cost): Rename to ...
38310 (thunderx2t99_branch_cost): This.
38311 (vulcan_tunings): Rename to ...
38312 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
38313 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
38315 2017-01-09 Martin Jambor <mjambor@suse.cz>
38319 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
38320 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
38321 (propagate_vr_accross_jump_function): Use the above function for all
38322 value range computations for pass-through jump functions and type
38323 converasion from explicit value range values.
38324 (ipcp_propagate_stage): Do not attempt to deduce types of formal
38325 parameters from TYPE_ARG_TYPES.
38326 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
38327 (ipa_write_node_info): Stream type of the actual argument.
38328 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
38330 2017-01-09 Martin Liska <mliska@suse.cz>
38333 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
38334 (lookup_compiler): Do not show error message with have_E.
38336 2017-01-09 Jakub Jelinek <jakub@redhat.com>
38338 PR tree-optimization/78938
38339 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
38340 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
38341 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
38342 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
38345 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38347 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
38350 2017-01-09 Richard Biener <rguenther@suse.de>
38352 PR tree-optimization/78997
38353 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
38354 name condition properly.
38356 2017-01-09 Richard Biener <rguenther@suse.de>
38359 * dwarf2out.c (is_cxx): New overload with context.
38360 (is_naming_typedef_decl): Use it.
38362 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38364 * invoke.texi (Option Summary): Correct spacing in option lists
38365 and add line breaks to fix over-long lines.
38367 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38369 PR middle-end/17660
38371 * extend.texi (Common Variable Attributes): Add xref to GCC
38372 Internals manual to explain mode attribute keywords.
38374 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38377 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
38378 and Preprocessor Options.
38379 (Options for Linking): Document -pthread here....
38380 (RS/6000 and PowerPC Options): ...not here.
38381 (Solaris 2 Options): ...or here.
38382 * doc/cppopts.texi: Document -pthread.
38384 2017-01-08 Martin Sebor <msebor@redhat.com>
38386 PR middle-end/77708
38387 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
38388 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
38389 New member functions.
38390 (format_directive): Used them.
38392 (pass_sprintf_length::handle_gimple_call): Same.
38393 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
38394 to avoid truncation for any argument.
38395 (extract_affine_mul): Same.
38396 * tree.c (get_file_function_name): Same.
38398 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38400 PR middle-end/77484
38401 * predict.def (PRED_INDIR_CALL): Set to 86.
38403 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38405 PR preprocessor/54124
38406 * doc/cppopts.texi: Reformat -d subtable to list the full name
38407 of the options. Add cross-reference to the docs for the general
38408 compiler -d options.
38409 * doc/invoke.texi (Developer Options): Add cross-reference to the
38410 preprocessor-specific -d option documentation.
38412 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38414 PR preprocessor/13498
38415 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
38416 redudant material, and reflect new command-line options.
38417 (System Headers): Likewise.
38419 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38421 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
38422 -isystem, and -idirafter. Copy-edit.
38423 * doc/cppopts.texi: Copy-edit. Remove contradiction about
38424 default for -ftrack-macro-expansion. Delete obsolete and
38425 badly-formatted implementation details about -fdebug-cpp output.
38426 * doc/cppwarnopts.texi: Copy-edit.
38428 2017-01-07 David Malcolm <dmalcolm@redhat.com>
38431 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
38432 that the transition from a max line width >= 1<<10 to narrower
38433 lines works correctly.
38435 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
38437 * doc/options.texi (PerFunction): New.
38438 * opt-functions.awk (switch_flags): Map both Optimization and
38439 PerFunction to CL_OPTIMIZATION.
38440 * opth-gen.awk: Test for PerFunction flag along with
38442 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
38443 it only when the latter is present. Skip those that don't in
38444 the hash function generator.
38445 * common.opt (fvar-tracking): Mark as PerFunction instead of
38447 (fvar-tracking-assignments): Likewise.
38448 (fvar-tracking-assignments-toggle): Likewise.
38449 (fvar-tracking-uninit): Likewise.
38451 2017-01-07 Jakub Jelinek <jakub@redhat.com>
38453 PR translation/79018
38454 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
38457 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
38460 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
38461 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
38462 TARGET_LONG_JUMP_TABLE_OFFSETS.
38463 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
38464 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
38465 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
38466 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
38467 * config/m68k/m68k.md (tablejump expander): Likewise.
38468 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
38469 TARGET_LONG_JUMP_TABLE_OFFSETS.
38470 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
38471 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
38473 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
38474 David Holsgrove <david.holsgrove@xilinx.com>
38476 * common/config/microblaze/microblaze-common.c
38477 (TARGET_EXCEPT_UNWIND_INFO): Remove.
38478 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
38480 * config/microblaze/microblaze.c (microblaze_must_save_register)
38481 (microblaze_expand_epilogue, microblaze_return_addr): Handle
38483 (microblaze_eh_return): New function.
38484 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
38485 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
38486 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
38487 * config/microblaze/microblaze.md (eh_return): New pattern.
38489 2017-01-06 Jakub Jelinek <jakub@redhat.com>
38491 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
38492 GCC_DIAGNOSTIC_STRINGIFY): Define.
38494 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
38496 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38498 * config/arm/arm.md (<mcrr>): New.
38500 * config/arm/arm.c (arm_arch5te): New.
38501 (arm_option_override): Set arm_arch5te.
38502 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
38504 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
38505 (arm_mcrr_qualifiers): ... this. New.
38506 (MRRC_QUALIFIERS): Define to...
38507 (arm_mrrc_qualifiers): ... this. New.
38508 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
38510 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
38511 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
38512 (MRRCI, mrrc, MRRC): New.
38513 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
38514 VUNSPEC_MRRC2): New.
38516 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38518 * config/arm/arm.md (<mcr>): New.
38520 * config/arm/arm.c (arm_coproc_builtin_available): Add
38521 support for mcr, mrc, mcr2 and mrc2.
38522 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
38523 (arm_mcr_qualifiers): ... this. New.
38524 (MRC_QUALIFIERS): Define to ...
38525 (arm_mrc_qualifiers): ... this. New.
38526 (MCR_QUALIFIERS): Define to ...
38527 (arm_mcr_qualifiers): ... this. New.
38528 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
38530 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
38531 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
38532 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
38533 VUNSPEC_MRC2): New.
38535 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38537 * config/arm/arm.md (*ldc): New.
38541 * config/arm/arm.c (arm_coproc_builtin_available): Add
38542 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
38543 (arm_coproc_ldc_stc_legitimate_address): New.
38544 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
38545 'qualifier_const_pointer'.
38546 (LDC_QUALIFIERS): Define to...
38547 (arm_ldc_qualifiers): ... this. New.
38548 (STC_QUALIFIERS): Define to...
38549 (arm_stc_qualifiers): ... this. New.
38550 * config/arm/arm-protos.h
38551 (arm_coproc_ldc_stc_legitimate_address): New.
38552 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
38553 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
38554 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
38555 stc2, stcl, stc2l): New.
38556 * config/arm/constraints.md (Uz): New.
38557 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
38558 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
38559 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
38560 VUNSPEC_STC2L): New.
38562 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38564 * config/arm/arm.md (<cdp>): New.
38565 * config/arm/arm.c (neon_const_bounds): Rename this ...
38566 (arm_const_bounds): ... this.
38567 (arm_coproc_builtin_available): New.
38568 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
38569 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
38570 (CDP_QUALIFIERS): Define to...
38571 (arm_cdp_qualifiers): ... this. New.
38573 (arm_expand_builtin_args): Add case for 6 arguments.
38574 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
38575 (arm_const_bounds): ... this.
38576 (arm_coproc_builtin_available): New.
38577 * config/arm/arm_acle.h (__arm_cdp): New.
38579 * config/arm/arm_acle_builtins.def (cdp): New.
38581 * config/arm/iterators.md (CDPI,CDP,cdp): New.
38582 * config/arm/neon.md: Rename all 'neon_const_bounds' to
38583 'arm_const_bounds'.
38584 * config/arm/types.md (coproc): New.
38585 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
38586 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
38587 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
38588 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
38590 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38592 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
38593 (UBINOP_QUALIFIERS): New.
38595 (acle_builtin_data): New. Change comment.
38596 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
38597 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
38598 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
38599 arm_acle_builtins.def.
38600 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
38601 (arm_init_acle_builtins): New.
38602 (CRC32_BUILTIN): Remove.
38603 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
38604 crc32cb, crc32ch and crc32cw.
38605 (arm_init_crc32_builtins): Remove.
38606 (arm_init_builtins): Use arm_init_acle_builtins rather
38607 than arm_init_crc32_builtins.
38608 (arm_expand_acle_builtin): New.
38609 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
38610 * config/arm/arm_acle_builtins.def: New.
38612 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38614 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
38615 (arm_builtin_datum): ... this.
38616 (arm_init_neon_builtin): Rename to ...
38617 (arm_init_builtin): ... this. Add a new parameters PREFIX
38618 and USE_SIG_IN_NAME.
38619 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
38620 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
38621 'arm_builtin_datum'.
38622 (arm_init_vfp_builtins): Likewise.
38623 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
38624 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
38625 (arm_expand_neon_args): Rename to ...
38626 (arm_expand_builtin_args): ... this. Rename builtin_arg
38627 enum values and differentiate between ARG_BUILTIN_MEMORY
38628 and ARG_BUILTIN_NEON_MEMORY.
38629 (arm_expand_neon_builtin_1): Rename to ...
38630 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
38631 values, arm_expand_builtin_args and add bool parameter NEON.
38632 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
38633 (arm_expand_vfp_builtin): Likewise.
38634 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
38636 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38638 PR middle-end/77484
38639 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
38640 * predict.c (tree_estimate_probability_bb): Reverse direction of
38641 polymorphic call predictor.
38643 2017-01-06 David Malcolm <dmalcolm@redhat.com>
38645 * passes.c (execute_one_pass): Split out pass-skipping logic into...
38646 (determine_pass_name_match): ...this new function and...
38647 (should_skip_pass_p): ...this new function.
38649 2017-01-06 Nathan Sidwell <nathan@acm.org>
38651 * ipa-visibility.c (function_and_variable_visibility): Reformat
38652 comments and long lines. Remove extrneous if.
38653 * symtab.c (symtab_node::make_decl_local): Fix code format.
38654 (symtab_node::set_section_for_node): Fix comment typo.
38656 2017-01-06 Martin Liska <mliska@suse.cz>
38659 * lra-constraints.c: Rename invariant to lra_invariant.
38660 * predict.c (set_even_probabilities): Initialize e to NULL.
38662 2017-01-05 Martin Sebor <msebor@redhat.com>
38664 PR tree-optimization/78910
38665 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
38666 (format_integer): Correct off-by-one error in the handling
38667 of precision with negative numbers in signed conversions..
38669 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
38671 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
38673 2017-01-05 Jakub Jelinek <jakub@redhat.com>
38675 PR tree-optimization/71016
38676 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
38677 factor_out_conditional_conversion. Formatting fix.
38678 (factor_out_conditional_conversion): Add cond_stmt argument.
38679 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
38680 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
38683 2017-01-05 David Malcolm <dmalcolm@redhat.com>
38685 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
38686 read-rtl-function.o, and selftest-rtl.o.
38687 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
38688 (selftest::aarch64_test_loading_full_dump): New function.
38689 (selftest::aarch64_run_selftests): New function.
38690 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
38691 selftest::aarch64_run_selftests.
38692 * config/i386/i386.c
38693 (selftest::ix86_test_loading_dump_fragment_1): New function.
38694 (selftest::ix86_test_loading_call_insn): New function.
38695 (selftest::ix86_test_loading_full_dump): New function.
38696 (selftest::ix86_test_loading_unspec): New function.
38697 (selftest::ix86_run_selftests): Call the new functions.
38698 * emit-rtl.c (maybe_set_max_label_num): New function.
38699 * emit-rtl.h (maybe_set_max_label_num): New decl.
38700 * function.c (instantiate_decls): Guard call to
38701 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
38702 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
38704 * gensupport.c (gen_reader::gen_reader): Pass "false"
38705 for new "compact" param of rtx_reader.
38706 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
38707 rather than an empty string for NULL strings.
38708 * read-md.c: Potentially include config.h rather than bconfig.h.
38709 Wrap include of errors.h with #ifdef GENERATOR_FILE.
38710 (have_error): New global, copied from errors.c.
38711 (md_reader::read_name): Rename to...
38712 (md_reader::read_name_1): ...this, adding "out_loc" param,
38713 and converting "missing name or number" to returning false, rather
38715 (md_reader::read_name): Reimplement in terms of read_name_1.
38716 (md_reader::read_name_or_nil): New function.
38717 (md_reader::read_string): Handle "(nil)" by returning NULL.
38718 (md_reader::md_reader): Add new param "compact".
38719 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
38720 (md_reader::read_file): New method.
38721 * read-md.h (md_reader::md_reader): Add new param "compact".
38722 (md_reader::read_file): New method.
38723 (md_reader::is_compact): New accessor.
38724 (md_reader::read_name): Convert return type from void to file_location.
38725 (md_reader::read_name_or_nil): New decl.
38726 (md_reader::read_name_1): New decl.
38727 (md_reader::m_compact): New field.
38728 (noop_reader::noop_reader): Pass "false" for new "compact" param
38730 (rtx_reader::rtx_reader): Add new "compact" param.
38731 (rtx_reader::read_rtx_operand): Make virtual and convert return
38732 type from void to rtx.
38733 (rtx_reader::read_until): New decl.
38734 (rtx_reader::handle_any_trailing_information): New virtual function.
38735 (rtx_reader::postprocess): New virtual function.
38736 (rtx_reader::finalize_string): New virtual function.
38737 (rtx_reader::m_in_call_function_usage): New field.
38738 (rtx_reader::m_reuse_rtx_by_id): New field.
38739 * read-rtl-function.c: New file.
38740 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
38741 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
38742 (selftest::verify_three_block_rtl_cfg): New decl.
38743 * read-rtl-function.h: New file.
38744 * read-rtl.c: Potentially include config.h rather than bconfig.h.
38745 For host, include function.h, memmodel.h, and emit-rtl.h.
38746 (one_time_initialization): New function.
38747 (struct compact_insn_name): New struct.
38748 (compact_insn_names): New array.
38749 (find_code): Handle insn codes in compact dumps.
38750 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
38751 (bind_subst_iter_and_attr): Likewise.
38752 (add_condition_to_string): Likewise.
38753 (add_condition_to_rtx): Likewise.
38754 (apply_attribute_uses): Likewise.
38755 (add_current_iterators): Likewise.
38756 (apply_iterators): Likewise.
38757 (initialize_iterators): Guard usage of apply_subst_iterator with
38758 #ifdef GENERATOR_FILE.
38759 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
38760 (md_reader::read_mapping): Likewise.
38761 (add_define_attr_for_define_subst): Likewise.
38762 (add_define_subst_attr): Likewise.
38763 (read_subst_mapping): Likewise.
38764 (check_code_iterator): Likewise.
38765 (rtx_reader::read_rtx): Likewise. Move one-time initialization
38767 (one_time_initialization): New function.
38768 (rtx_reader::read_until): New method.
38769 (read_flags): New function.
38770 (parse_reg_note_name): New function.
38771 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
38772 Handle reuse_rtx ids.
38773 Wrap iterator lookup within #ifdef GENERATOR_FILE.
38774 Add parsing support for RTL dumps, mirroring the special-cases in
38775 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
38776 values, and calling handle_any_trailing_information.
38777 (rtx_reader::read_rtx_operand): Convert return type from void
38778 to rtx, returning return_rtx. Handle case 'e'. Call
38779 finalize_string on XSTR and XTMPL fields.
38780 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
38781 "(nil)" values were omitted. Call the postprocess vfunc on the
38783 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
38784 class ctor. Initialize m_in_call_function_usage. Call
38785 one_time_initialization.
38786 * rtl-tests.c (selftest::test_uncond_jump): Call
38787 set_new_first_and_last_insn.
38788 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
38789 * selftest-rtl.c: New file.
38790 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
38791 (selftest::get_insn_by_uid): New decl.
38792 * selftest-run-tests.c (selftest::run_tests): Call
38793 read_rtl_function_c_tests.
38794 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
38795 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
38798 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
38800 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
38801 operands in a special way. Assert that pos+len <= mode precision.
38803 2017-01-05 Jakub Jelinek <jakub@redhat.com>
38805 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
38806 3 argument Alias with unlimited for the negative form.
38807 (fno-vect-cost-model): Removed.
38809 2017-01-05 Martin Liska <mliska@suse.cz>
38811 * hsa-gen.c (gen_hsa_divmod): New function.
38812 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
38814 2017-01-05 Martin Liska <mliska@suse.cz>
38817 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
38820 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38822 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
38823 small constant length operands.
38825 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38827 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
38828 between loop iterations.
38830 2017-01-05 Martin Liska <mliska@suse.cz>
38833 * gimplify.c (gimplify_decl_expr): Compare to
38834 asan_poisoned_variables instread of checking flags.
38835 (gimplify_target_expr): Likewise.
38836 (gimplify_expr): Likewise.
38837 (gimplify_function_tree): Conditionally initialize
38838 asan_poisoned_variables.
38840 2017-01-04 Jeff Law <law@redhat.com>
38842 PR tree-optimizatin/78812
38843 * rtl.h (contains_mem_rtx_p): Prototype.
38844 * ifcvt.c (containts_mem_rtx_p): Move from here to...
38845 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
38846 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
38847 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
38848 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
38850 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38852 * input.c (assert_char_at_range): Default-initialize actual_range.
38854 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38856 * df-scan.c (df_ref_create_structure): Make regno unsigned,
38857 to match the caller.
38859 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38861 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
38862 insns after final jump in test to emit dummy move.
38864 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38866 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
38867 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
38869 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38871 * multiple_target.c (create_dispatcher_calls): Init e_next.
38872 * tree-ssa-loop-split.c (split_loop): Init border.
38873 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
38876 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
38881 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
38882 (altivec_register_operand): Do not return true if the operand
38883 contains a SUBREG mixing SImode and SFmode.
38884 (vsx_register_operand): Likewise.
38885 (vsx_reg_sfsubreg_ok): New predicate.
38886 (vfloat_operand): Do not return true if the operand contains a
38887 SUBREG mixing SImode and SFmode.
38888 (vint_operand): Likewise.
38889 (vlogical_operand): Likewise.
38890 (gpc_reg_operand): Likewise.
38891 (int_reg_operand): Likewise.
38892 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
38893 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
38894 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
38896 (rs6000_emit_move_si_sf_subreg): New helper function.
38897 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
38898 fixup SUBREGs involving SImode and SFmode.
38899 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
38900 numbers for the new peephole2 optimization.
38901 (peephole2 for SFmode unions): New peephole2 to optimize cases in
38902 the GLIBC math library that do AND/IOR/XOR operations on single
38903 precision floating point.
38904 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
38905 target macros to say whether we need to avoid SUBREGs mixing
38907 (TARGET_ALLOW_SF_SUBREG): Likewise.
38908 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
38909 (UNSPEC_SI_FROM_SF): Likewise.
38910 (iorxor): Change spacing.
38911 (and_ior_xor): New iterator for AND, IOR, and XOR.
38912 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
38913 (movdi_from_sf_zero_ext): Likewise.
38914 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
38915 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
38916 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
38917 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
38918 (fms<mode>4): Likewise.
38919 (fnma<mode>4): Likewise.
38920 (fnms<mode>4): Likewise.
38921 (nfma<mode>4): Likewise.
38922 (nfms<mode>4): Likewise.
38924 2017-01-04 Marek Polacek <polacek@redhat.com>
38927 * doc/invoke.texi: Document -Wpointer-compare.
38929 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38931 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
38934 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
38935 descriptions for -gdwarf-5 and emit them as uleb128 instead of
38938 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
38941 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
38942 documentation of the powerpc_popcntb_ok attribute.
38943 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
38944 code to issue warning messages if a requested CPU configuration is
38945 not supported by the binary (assembler and loader) toolchain.
38946 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
38947 made to define a built-in function that has been disabled.
38948 (paired_init_builtins): Add assertion to prevent ICE if attempt is
38949 made to define a built-in function that has been disabled.
38950 (altivec_init_builtins): Add comment explaining why definition
38951 of the DST built-in functions is not preceded by an assertion
38952 check. Add assertions to prevent ICE if attempts are made to
38953 define an altivec predicate or an abs* built-in function that has
38955 (htm_init_builtins): Add comment explaining why definition of the
38956 htm built-in functions is not preceded by an assertion check.
38958 2017-01-04 Jeff Law <law@redhat.com>
38960 PR tree-optimizatin/67955
38961 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
38962 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
38963 the points-to solution does not include pt_null. Use DECL_PT_UID
38966 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
38968 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
38969 Use gen_int_mode instead of gen_lopwart for const_int operands.
38971 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38973 PR tree-optimization/71563
38974 * match.pd: Simplify X << Y into X if Y is known to be 0 or
38975 out of range value - has low bits known to be zero.
38977 2017-01-04 Alan Modra <amodra@gmail.com>
38979 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
38980 * configure: Regenerate.
38981 * config.in: Regenerate.
38983 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38986 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
38987 a substring of the message, but strcmp with the whole message. Ifdef
38988 ENABLE_NLS, translate the message first using dgettext.
38990 2017-01-03 Jeff Law <law@redhat.com>
38992 PR tree-optimizatin/78856
38993 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
38994 (mark_threaded_blocks): Remove code to truncate thread paths that
38995 cross multiple loop headers. Instead invalidate the cached loop
38996 iteration information and handle case of a thread path walking
38997 into an irreducible region.
38999 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
39002 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
39003 assertions. Add support for doing the signbit if the IEEE 128-bit
39004 floating point value is in a GPR.
39005 * config/rs6000/rs6000.md (Fsignbit): Delete.
39006 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
39007 Update the length attribute if the value is in a GPR.
39008 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
39009 the sign or zero extension instruction, since the value is always 0/1.
39010 (signbit<mode>2_dm2): Delete using <Fsignbit>.
39013 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
39014 extracting SImode to a GPR register so that we can generate a
39015 store, limit the vector to be in a traditional Altivec register
39016 for the vextuwrx instruction.
39018 2017-01-03 Ian Lance Taylor <iant@google.com>
39020 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
39022 2017-01-03 Martin Sebor <msebor@redhat.com>
39024 PR tree-optimization/78696
39025 * gimple-ssa-sprintf.c (format_floating): Correct handling of
39026 precision. Use MPFR for %f for greater fidelity. Correct handling
39028 (pass_sprintf_length::compute_format_length): Set width and precision
39029 specified by asrerisk to void_node for vararg functions.
39030 (try_substitute_return_value): Adjust dump output.
39032 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
39034 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
39036 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
39038 * doc/invoke.texi (SPARC options): Document -mlra as the default.
39039 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
39040 -mlra/-mno-lra was passed to the compiler.
39042 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
39044 PR rtl-optimization/65618
39045 * emit-rtl.c (try_split): Move initialization of "before" and
39046 "after" to just before the call to emit_insn_after_setloc.
39048 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
39050 * doc/md.texi (Standard Names): Remove reference to Java frontend.
39052 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
39054 * dwarf2out.c (gen_enumeration_type_die): When
39055 -gno-strict-dwarf, add a DW_AT_encoding attribute.
39057 2017-01-03 Jakub Jelinek <jakub@redhat.com>
39059 PR tree-optimization/78965
39060 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
39061 Change first argument from const call_info & to call_info &. For %n
39062 set info.nowrite to false.
39064 PR middle-end/78901
39065 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
39066 possibly throwing calls.
39068 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
39069 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
39070 and fns handling, rather than in a separate case SSA_NAME.
39072 2017-01-02 Jeff Law <law@redhat.com>
39074 * config/darwin-driver.c (darwin_driver_init): Const-correctness
39075 fixes for first_period and second_period variables.
39077 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
39080 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
39081 (*insvqi_1): New insn pattern.
39082 (*insvqi_1_mem_rex64): Ditto.
39083 (*insvqi_2): Ditto.
39084 (*insvqi_3): Rename from *insvqi.
39086 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
39088 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
39090 * doc/cfg.texi (Edges): Remove reference to Java.
39091 (Maintaining the CFG): Ditto.
39093 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39095 PR middle-end/77674
39096 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
39097 transparent aliases.
39099 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39101 PR middle-end/77484
39102 * predict.def (PRED_CALL): Update hitrate.
39103 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
39104 * predict.c (tree_estimate_probability_bb): Split CALL predictor
39105 into direct/indirect/polymorphic variants.
39107 2017-01-01 Jakub Jelinek <jakub@redhat.com>
39109 Update copyright years.
39111 * gcc.c (process_command): Update copyright notice dates.
39112 * gcov-dump.c (print_version): Ditto.
39113 * gcov.c (print_version): Ditto.
39114 * gcov-tool.c (print_version): Ditto.
39115 * gengtype.c (create_file): Ditto.
39116 * doc/cpp.texi: Bump @copying's copyright year.
39117 * doc/cppinternals.texi: Ditto.
39118 * doc/gcc.texi: Ditto.
39119 * doc/gccint.texi: Ditto.
39120 * doc/gcov.texi: Ditto.
39121 * doc/install.texi: Ditto.
39122 * doc/invoke.texi: Ditto.
39124 Copyright (C) 2017 Free Software Foundation, Inc.
39126 Copying and distribution of this file, with or without modification,
39127 are permitted in any medium without royalty provided the copyright
39128 notice and this notice are preserved.