1 2018-10-25 Richard Biener <rguenther@suse.de>
3 * tree-if-conv.c: Include tree-ssa-sccvn.h.
4 (tree_if_conversion): Run CSE on the if-converted loop body.
6 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
9 * rtl.c (RTX_CODE_HWINT_P_1): New helper macro.
10 (RTX_CODE_HWINT_P): New macro.
11 (rtx_code_size): Use RTX_CODE_HWINT_P ().
13 2018-10-25 Jan Hubicka <jh@suse.cz>
15 * ipa-devirt.c (odr_types_equivalent_p): Do not ICE if one of types
18 2018-10-25 Richard Biener <rguenther@suse.de>
20 PR tree-optimization/87665
21 PR tree-optimization/87745
22 * tree-vectorizer.h (get_earlier_stmt): Remove.
23 (get_later_stmt): Pick up UID from the original non-pattern stmt.
25 2018-10-25 Sam Tebbs <sam.tebbs@arm.com>
27 * options.texi (Deprecated): Move list to Var section.
29 2018-10-24 Segher Boessenkool <segher@kernel.crashing.org>
31 PR rtl-optimization/87720
32 * combine.c (make_more_copies): Skip if the dest is pc_rtx.
34 2018-10-24 Alexandre Oliva <aoliva@redhat.com>
36 * gimple-ssa-isolate-paths.c
37 (find_implicit_erroneous_behavior): Do not change code if the
38 pass is running for warnings only.
39 (find_explicit_erroneous_behavior): Likewise.
41 2018-10-24 Michael Meissner <meissner@linux.ibm.com>
43 * config/rs6000/rs6000.c (TARGET_MANGLE_DECL_ASSEMBLER_NAME):
44 Define as rs6000_mangle_decl_assembler_name.
45 (rs6000_mangle_decl_assembler_name): If the user switched from IBM
46 long double to IEEE long double, switch the names of the long
47 double built-in functions to be <func>f128 instead of <func>l.
49 2018-10-24 Martin Sebor <msebor@redhat.com>
51 * doc/extend.texi (nonnull): List no-argument form. Reference
52 -fno-delete-null-pointer-checks and -fisolate-erroneous-paths-attribute.
54 2018-10-24 Richard Biener <rguenther@suse.de>
56 * tree-ssa-sccvn.c (do_rpo_vn): Free rpo_state.
58 2018-10-24 Martin Liska <mliska@suse.cz>
60 PR tree-optimization/84436
61 * tree-switch-conversion.c (switch_conversion::contains_same_values_p):
63 (switch_conversion::contains_linear_function_p): New.
64 (switch_conversion::build_one_array): Support linear
65 transformation on input.
66 * tree-switch-conversion.h (struct switch_conversion): Add
67 contains_linear_function_p declaration.
69 2018-10-24 Richard Biener <rguenther@suse.de>
71 * varasm.c (const_hash_1): Return hash of ADDR_EXPR
72 if its argument is CONSTANT_CLASS_P.
74 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
76 * ipa-utils.h (type_with_linkage_p): No longer check for TYPE_STUB_DECL;
77 it is wrong for forward declarations.
79 2018-10-24 Ilya Leoshkevich <iii@linux.ibm.com>
81 * config/s390/s390.c (s390_check_qrst_address): Add the missing
82 SYMBOL_REF_P () check.
84 2018-10-24 Richard Biener <rguenther@suse.de>
86 PR tree-optimization/87105
87 * tree-vect-data-refs.c (vect_analyze_group_access_1): Adjust
89 (vect_analyze_data_ref_accesses): Handle duplicate loads and
90 stores by splitting the affected group after the fact.
91 * tree-vect-slp.c (vect_build_slp_tree_2): Dump when we
92 fail the SLP build because of size constraints.
94 2018-10-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
96 * configure.ac (gcc_cv_ld_aligned_shf_merge): New test.
97 * configure: Regenerate.
98 * config.in: Regenerate.
99 * varasm.c (mergeable_string_section): Use readonly_data_section
100 if linker doesn't support SHF_MERGE with alignment > 8.
101 (mergeable_constant_section): Likewise.
103 2018-10-24 Richard Biener <rguenther@suse.de>
105 PR tree-optimization/84013
106 * tree-ssa-structalias.c (struct msdi_data): New struct for
107 marshalling data to walk_stmt_load_store_ops.
108 (maybe_set_dependence_info): Refactor as callback for
109 walk_stmt_load_store_ops.
110 (compute_dependence_clique): Set restrict info on all stmt kinds.
112 2018-10-24 Martin Liska <mliska@suse.cz>
114 * cgraph.c (cgraph_node::dump):
115 Remove reduntant dumps and make tp_first_run dump more compact.
117 2018-10-24 Richard Biener <rguenther@suse.de>
119 PR tree-optimization/87665
120 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Adjust
123 2018-10-12 Jeff Law <law@redhat.com>
125 * config/h8300/h8300.c (h8300_expand_prologue): Fix stm generation
128 2018-10-23 Richard Biener <rguenther@suse.de>
130 * tree-vrp.c (add_assert_info): Guard dump_printf with
132 * gimple-ssa-evrp-analyze.c
133 (evrp_range_analyzer::record_ranges_from_incoming_edge):
134 Use value_range::ignore_equivs_equal_p.
136 2018-10-23 Richard Biener <rguenther@suse.de>
138 PR tree-optimization/87105
139 PR tree-optimization/87608
140 * passes.def (pass_all_early_optimizations): Add early phi-opt
142 * tree-ssa-phiopt.c (value_replacement): Ignore NOPs and predicts in
143 addition to debug stmts.
144 (tree_ssa_phiopt_worker): Add early_p argument, do only min/max
145 and abs replacement early.
146 * tree-cfg.c (gimple_empty_block_p): Likewise.
148 2018-10-23 Richard Earnshaw <rearnsha@arm.com>
151 * config.gcc (arm*-*-netbsdelf*): Default to StrongARM if no CPU
152 specified to configure.
153 (arm*-*-*): Use ARM7TDMI as the target CPU if no default provided.
155 2018-10-23 Richard Biener <rguenther@suse.de>
157 PR tree-optimization/87700
158 * tree-ssa-copy.c (set_copy_of_val): Fix change detection logic.
160 2018-10-23 Jakub Jelinek <jakub@redhat.com>
163 * config/i386/avx512vlintrin.h (_mm_mask_mullo_epi32): Change type of
164 second argument from __mmask16 to __mmask8.
165 * config/i386/avx512vlbwintrin.h (_mm_mask_packus_epi32,
166 _mm_mask_packs_epi32): Likewise.
167 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i64scatter_ps):
169 (_mm512_mask_prefetch_i64scatter_pd): Likewise. Formatting fix.
171 2018-10-23 Richard Biener <rguenther@suse.de>
173 * tree-vect-stmts.c (vect_analyze_stmt): Fix typo in comment.
175 2018-10-23 Richard Biener <rguenther@suse.de>
177 PR tree-optimization/86144
178 * tree-vect-stmts.c (vect_analyze_stmt): Prefer -mveclibabi
181 2018-10-23 Richard Biener <rguenther@suse.de>
183 PR tree-optimization/87693
184 * tree-ssa-threadedge.c (thread_around_empty_blocks): Handle
185 the case we do not find the taken edge.
187 2018-10-22 Paul Koning <ni1d@arrl.net>
189 * symtab.c (symtab_node::increase_alignment): Correct max
192 2018-10-22 Yury Gribov <tetra2005@gmail.com>
194 PR tree-optimization/87633
195 * match.pd: Do not generate unordered integer comparisons.
197 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
199 PR rtl-optimization/87600
200 * combine.c: Add include of expr.h.
201 (cant_combine_insn_p): Do not combine moves from any hard non-fixed
202 register to a pseudo.
203 (make_more_copies): New function, add a copy to a new pseudo after
204 the moves from hard registers into pseudos.
205 (rest_of_handle_combine): Declare rebuild_jump_labels_after_combine
206 later. Call make_more_copies.
208 2018-10-22 Andrew Stubbs <ams@codesourcery.com>
210 * lra-constraints.c (process_alt_operands): New local array,
211 matching_early_clobber. Check matching_early_clobber before
212 decrementing reject, and set matching_early_clobber after.
214 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
217 * config/rs6000/rs6000.c (print_operand_address): For unexpected RTL
218 call output_addr_const and hope for the best.
220 2018-10-22 Richard Biener <rguenther@suse.de>
222 * gimple-ssa-evrp-analyze.c
223 (evrp_range_analyzer::record_ranges_from_incoming_edge): Be
224 smarter about what ranges to use.
225 * tree-vrp.c (add_assert_info): Dump here.
226 (register_edge_assert_for_2): Instead of here at multiple but
229 * gcc.dg/tree-ssa/evrp12.c: New testcase.
230 * gcc.dg/predict-6.c: Adjust.
231 * gcc.dg/tree-ssa/vrp33.c: Disable EVRP.
232 * gcc.dg/tree-ssa/vrp02.c: Likewise.
233 * gcc.dg/tree-ssa/cunroll-9.c: Likewise.
235 2018-10-22 Steven Bosscher <steven@gcc.gnu.org>
236 Richard Biener <rguenther@suse.de>
238 * bitmap.h: Update data structure documentation, including a
239 description of bitmap views as either linked-lists or splay trees.
240 (struct bitmap_element_def): Update comments for splay tree bitmaps.
241 (struct bitmap_head_def): Likewise.
242 (bitmap_list_view, bitmap_tree_view): New prototypes.
243 (bitmap_initialize_stat): Initialize a bitmap_head's indx and
245 (bmp_iter_set_init): Assert the iterated bitmaps are in list form.
246 (bmp_iter_and_init, bmp_iter_and_compl_init): Likewise.
247 * bitmap.c (bitmap_elem_to_freelist): Unregister overhead of a
248 released bitmap element here.
249 (bitmap_element_free): Remove.
250 (bitmap_elt_clear_from): Work on splay tree bitmaps.
251 (bitmap_list_link_element): Renamed from bitmap_element_link. Move
252 this function similar ones such that linked-list bitmap implementation
253 functions are grouped.
254 (bitmap_list_unlink_element): Renamed from bitmap_element_unlink,
255 and moved for grouping.
256 (bitmap_list_insert_element_after): Renamed from
257 bitmap_elt_insert_after, and moved for grouping.
258 (bitmap_list_find_element): New function spliced from bitmap_find_bit.
259 (bitmap_tree_link_left, bitmap_tree_link_right,
260 bitmap_tree_rotate_left, bitmap_tree_rotate_right, bitmap_tree_splay,
261 bitmap_tree_link_element, bitmap_tree_unlink_element,
262 bitmap_tree_find_element): New functions for splay-tree bitmap
264 (bitmap_element_link, bitmap_element_unlink, bitmap_elt_insert_after):
265 Renamed and moved, see above entries.
266 (bitmap_tree_listify_from): New function to convert part of a splay
267 tree bitmap to a linked-list bitmap.
268 (bitmap_list_view): Convert a splay tree bitmap to linked-list form.
269 (bitmap_tree_view): Convert a linked-list bitmap to splay tree form.
270 (bitmap_find_bit): Remove.
271 (bitmap_clear, bitmap_clear_bit, bitmap_set_bit,
272 bitmap_single_bit_set_p, bitmap_first_set_bit, bitmap_last_set_bit):
273 Handle splay tree bitmaps.
274 (bitmap_copy, bitmap_count_bits, bitmap_and, bitmap_and_into,
275 bitmap_elt_copy, bitmap_and_compl, bitmap_and_compl_into,
276 bitmap_compl_and_into, bitmap_elt_ior, bitmap_ior, bitmap_ior_into,
277 bitmap_xor, bitmap_xor_into, bitmap_equal_p, bitmap_intersect_p,
278 bitmap_intersect_compl_p, bitmap_ior_and_compl,
279 bitmap_ior_and_compl_into, bitmap_set_range, bitmap_clear_range,
280 bitmap_hash): Reject trying to act on splay tree bitmaps. Make
281 corresponding changes to use linked-list specific bitmap_element
282 manipulation functions as applicable for efficiency.
283 (bitmap_tree_to_vec): New function.
284 (debug_bitmap_elt_file): New function split out from ...
285 (debug_bitmap_file): ... here. Handle splay tree bitmaps.
286 (bitmap_print): Likewise.
288 PR tree-optimization/63155
289 * tree-ssa-propagate.c (ssa_prop_init): Use tree-view for the
291 * tree-ssa-coalesce.c (coalesce_ssa_name): Populate used_in_copies
294 2018-10-22 Martin Liska <mliska@suse.cz>
296 PR tree-optimization/87686
298 2018-08-29 Martin Liska <mliska@suse.cz>
300 * tree-switch-conversion.c (switch_conversion::expand):
301 Strenghten assumption about gswitch statements.
303 2018-10-22 Martin Liska <mliska@suse.cz>
305 * ipa-icf.c (sem_item::compare_attributes): Remove.
306 (sem_item::compare_referenced_symbol_properties): Use
307 attribute_list_equal instead.
308 (sem_function::equals_wpa): Likewise.
309 * ipa-icf.h: Remove compare_attributes.
311 2018-10-22 Richard Biener <rguenther@suse.de>
314 * mem-stats.h (mem_usage::operator==): Fix pasto.
316 2018-10-22 Richard Biener <rguenther@suse.de>
318 PR tree-optimization/87640
319 * tree-vrp.c (set_value_range_with_overflow): Decompose
321 (extract_range_from_binary_expr_1): Adjust.
323 2018-10-22 Martin Jambor <mjambor@suse.cz>
325 * tree-eh.h (stmt_could_throw_p): Add function parameter.
326 (stmt_can_throw_external): Likewise.
327 (stmt_can_throw_internal): Likewise.
328 * tree-eh.c (lower_eh_constructs_2): Pass cfun to stmt_could_throw_p.
329 (lower_eh_constructs_2): Likewise.
330 (stmt_could_throw_p): Add fun parameter, use it instead of cfun.
331 (stmt_can_throw_external): Likewise.
332 (stmt_can_throw_internal): Likewise.
333 (maybe_clean_eh_stmt_fn): Pass cfun to stmt_could_throw_p.
334 (maybe_clean_or_replace_eh_stmt): Pass cfun to stmt_could_throw_p.
335 (maybe_duplicate_eh_stmt_fn): Pass new_fun to stmt_could_throw_p.
336 (maybe_duplicate_eh_stmt): Pass cfun to stmt_could_throw_p.
337 (pass_lower_eh_dispatch::execute): Pass cfun to
338 stmt_can_throw_external.
339 (cleanup_empty_eh): Likewise.
340 (verify_eh_edges): Pass cfun to stmt_could_throw_p.
341 * cgraph.c (cgraph_edge::set_call_stmt): Pass a function to
342 stmt_can_throw_external instead of pushing it to cfun.
343 (symbol_table::create_edge): Likewise.
344 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Pass cfun to
345 stmt_can_throw_internal.
346 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Pass cfun
347 to stmt_could_throw_p.
348 * gimple-ssa-store-merging.c (handled_load): Pass cfun to
349 stmt_can_throw_internal.
350 (pass_store_merging::execute): Likewise.
351 * gimple-ssa-strength-reduction.c
352 (find_candidates_dom_walker::before_dom_children): Pass cfun to
354 * gimplify-me.c (gimple_regimplify_operands): Pass cfun to
355 stmt_can_throw_internal.
356 * ipa-pure-const.c (check_call): Pass cfun to stmt_could_throw_p and
357 to stmt_can_throw_external.
358 (check_stmt): Pass cfun to stmt_could_throw_p.
359 (check_stmt): Pass cfun to stmt_can_throw_external.
360 (pass_nothrow::execute): Likewise.
361 * trans-mem.c (expand_call_tm): Pass cfun to stmt_can_throw_internal.
362 * tree-cfg.c (is_ctrl_altering_stmt): Pass cfun to
363 stmt_can_throw_internal.
364 (verify_gimple_in_cfg): Pass cfun to stmt_could_throw_p.
365 (stmt_can_terminate_bb_p): Pass cfun to stmt_can_throw_external.
366 (gimple_purge_dead_eh_edges): Pass cfun to stmt_can_throw_internal.
367 * tree-complex.c (expand_complex_libcall): Pass cfun to
368 stmt_could_throw_p and to stmt_can_throw_internal.
369 (expand_complex_multiplication): Pass cfun to stmt_can_throw_internal.
370 * tree-inline.c (copy_edges_for_bb): Likewise.
371 (maybe_move_debug_stmts_to_successors): Likewise.
372 * tree-outof-ssa.c (ssa_is_replaceable_p): Pass cfun to
374 * tree-parloops.c (oacc_entry_exit_ok_1): Likewise.
375 * tree-sra.c (scan_function): Pass cfun to stmt_can_throw_external.
376 * tree-ssa-alias.c (stmt_kills_ref_p): Pass cfun to
377 stmt_can_throw_internal.
378 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Likewise.
379 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Pass cfun to
381 (mark_aliased_reaching_defs_necessary_1): Pass cfun to
382 stmt_can_throw_internal.
383 * tree-ssa-forwprop.c (pass_forwprop::execute): Likewise.
384 * tree-ssa-loop-im.c (movement_possibility): Pass cfun to
386 * tree-ssa-loop-ivopts.c (find_givs_in_stmt_scev): Likewise.
387 (add_autoinc_candidates): Pass cfun to stmt_can_throw_internal.
388 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
389 (convert_mult_to_fma_1): Likewise.
390 (convert_to_divmod): Likewise.
391 * tree-ssa-phiprop.c (propagate_with_phi): Likewise.
392 * tree-ssa-pre.c (compute_avail): Pass cfun to stmt_could_throw_p.
393 * tree-ssa-propagate.c
394 (substitute_and_fold_dom_walker::before_dom_children): Likewise.
395 * tree-ssa-reassoc.c (suitable_cond_bb): Likewise.
396 (maybe_optimize_range_tests): Likewise.
397 (linearize_expr_tree): Likewise.
398 (reassociate_bb): Likewise.
399 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
400 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Likewise.
401 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
402 (handle_char_store): Likewise.
403 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Pass cfun to
404 stmt_can_throw_internal.
405 * tree-vect-patterns.c (check_bool_pattern): Pass cfun to
407 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
408 (vectorizable_call): Pass cfun to stmt_can_throw_internal.
409 (vectorizable_simd_clone_call): Likewise.
410 * value-prof.c (gimple_ic): Pass cfun to stmt_could_throw_p.
411 (gimple_stringop_fixed_value): Likewise.
413 2018-10-22 Ilya Leoshkevich <iii@linux.ibm.com>
415 * config/s390/s390.c (s390_loadrelative_operand_p): Accept
416 literal pool references.
417 (s390_check_qrst_address): Adapt to the new behavior of
418 s390_loadrelative_operand_p ().
420 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
423 * config/i386/sse.md (*andnot<mode>3_bcst): New.
425 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
428 * config/i386/sse.md (*<code><mode>3_bcst): New.
430 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
433 * config/i386/sse.md (avx512bcst): Updated for V4SI, V2DI, V8SI,
434 V4DI, V16SI and V8DI.
435 (*sub<mode>3<mask_name>_bcst): New.
436 (*add<mode>3<mask_name>_bcst): Likewise.
438 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
441 * config/i386/avx512fintrin.h (_mm512_fnmsub_round_pd): Use
442 __builtin_ia32_vfnmsubpd512_mask.
443 (_mm512_mask_fnmsub_round_pd): Likewise.
444 (_mm512_fnmsub_pd): Likewise.
445 (_mm512_mask_fnmsub_pd): Likewise.
446 (_mm512_maskz_fnmsub_round_pd): Use
447 __builtin_ia32_vfnmsubpd512_maskz.
448 (_mm512_maskz_fnmsub_pd): Likewise.
449 (_mm512_fnmsub_round_ps): Use __builtin_ia32_vfnmsubps512_mask.
450 (_mm512_mask_fnmsub_round_ps): Likewise.
451 (_mm512_fnmsub_ps): Likewise.
452 (_mm512_mask_fnmsub_ps): Likewise.
453 (_mm512_maskz_fnmsub_round_ps): Use
454 __builtin_ia32_vfnmsubps512_maskz.
455 (_mm512_maskz_fnmsub_ps): Likewise.
456 * config/i386/avx512vlintrin.h (_mm256_mask_fnmsub_pd): Use
457 __builtin_ia32_vfnmsubpd256_mask.
458 (_mm256_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256_maskz.
459 (_mm_mask_fnmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
460 (_mm_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd128_maskz.
461 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
462 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
463 (_mm256_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_maskz.
464 (_mm_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_mask.
465 (_mm_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_maskz.
466 * config/i386/fmaintrin.h (_mm_fnmsub_pd): Use
467 __builtin_ia32_vfnmsubpd.
468 (_mm256_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256.
469 (_mm_fnmsub_ps): Use __builtin_ia32_vfnmsubps.
470 (_mm256_fnmsub_ps): Use __builtin_ia32_vfnmsubps256.
471 (_mm_fnmsub_sd): Use __builtin_ia32_vfnmsubsd3.
472 (_mm_fnmsub_ss): Use __builtin_ia32_vfnmsubss3.
473 * config/i386/i386-builtin.def: Add
474 __builtin_ia32_vfnmsubpd256_mask,
475 __builtin_ia32_vfnmsubpd256_maskz,
476 __builtin_ia32_vfnmsubpd128_mask,
477 __builtin_ia32_vfnmsubpd128_maskz,
478 __builtin_ia32_vfnmsubps256_mask,
479 __builtin_ia32_vfnmsubps256_maskz,
480 __builtin_ia32_vfnmsubps128_mask,
481 __builtin_ia32_vfnmsubps128_maskz,
482 __builtin_ia32_vfnmsubpd512_mask,
483 __builtin_ia32_vfnmsubpd512_maskz,
484 __builtin_ia32_vfnmsubps512_mask,
485 __builtin_ia32_vfnmsubps512_maskz, __builtin_ia32_vfnmsubss3,
486 __builtin_ia32_vfnmsubsd3, __builtin_ia32_vfnmsubps,
487 __builtin_ia32_vfnmsubpd, __builtin_ia32_vfnmsubps256 and.
488 __builtin_ia32_vfnmsubpd256.
489 * config/i386/sse.md (fma4i_fnmsub_<mode>): New.
490 (<avx512>_fnmsub_<mode>_maskz<round_expand_name>): Likewise.
491 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_1):
493 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_2):
495 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_3):
497 (fmai_vmfnmsub_<mode><round_name>): Likewise.
499 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
502 * config/i386/avx512fintrin.h (_mm512_fnmadd_round_pd): Use
503 __builtin_ia32_vfnmaddpd512_mask.
504 (_mm512_mask_fnmadd_round_pd): Likewise.
505 (_mm512_fnmadd_pd): Likewise.
506 (_mm512_mask_fnmadd_pd): Likewise.
507 (_mm512_maskz_fnmadd_round_pd): Use
508 __builtin_ia32_vfnmaddpd512_maskz.
509 (_mm512_maskz_fnmadd_pd): Likewise.
510 (_mm512_fnmadd_round_ps): Use __builtin_ia32_vfnmaddps512_mask.
511 (_mm512_mask_fnmadd_round_ps): Likewise.
512 (_mm512_fnmadd_ps): Likewise.
513 (_mm512_mask_fnmadd_ps): Likewise.
514 (_mm512_maskz_fnmadd_round_ps): Use
515 __builtin_ia32_vfnmaddps512_maskz.
516 (_mm512_maskz_fnmadd_ps): Likewise.
517 * config/i386/avx512vlintrin.h (_mm256_mask_fnmadd_pd): Use
518 __builtin_ia32_vfnmaddpd256_mask.
519 (_mm256_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256_maskz.
520 (_mm_mask_fnmadd_pd): Use __builtin_ia32_vfmaddpd128_mask
521 (_mm_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd128_maskz.
522 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
523 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
524 (_mm256_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_maskz.
525 (_mm_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_mask.
526 (_mm_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_maskz.
527 * config/i386/fmaintrin.h (_mm_fnmadd_pd): Use
528 __builtin_ia32_vfnmaddpd.
529 (_mm256_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256.
530 (_mm_fnmadd_ps): Use __builtin_ia32_vfnmaddps.
531 (_mm256_fnmadd_ps): Use __builtin_ia32_vfnmaddps256.
532 (_mm_fnmadd_sd): Use __builtin_ia32_vfnmaddsd3.
533 (_mm_fnmadd_ss): Use __builtin_ia32_vfnmaddss3.
534 * config/i386/i386-builtin.def: Add
535 __builtin_ia32_vfnmaddpd256_mask,
536 __builtin_ia32_vfnmaddpd256_maskz,
537 __builtin_ia32_vfnmaddpd128_mask,
538 __builtin_ia32_vfnmaddpd128_maskz,
539 __builtin_ia32_vfnmaddps256_mask,
540 __builtin_ia32_vfnmaddps256_maskz,
541 __builtin_ia32_vfnmaddps128_mask,
542 __builtin_ia32_vfnmaddps128_maskz,
543 __builtin_ia32_vfnmaddpd512_mask,
544 __builtin_ia32_vfnmaddpd512_maskz,
545 __builtin_ia32_vfnmaddps512_mask,
546 __builtin_ia32_vfnmaddps512_maskz, __builtin_ia32_vfnmaddss3,
547 __builtin_ia32_vfnmaddsd3, __builtin_ia32_vfnmaddps,
548 __builtin_ia32_vfnmaddpd, __builtin_ia32_vfnmaddps256 and.
549 __builtin_ia32_vfnmaddpd256.
550 * config/i386/sse.md (fma4i_fnmadd_<mode>): New.
551 (<avx512>_fnmadd_<mode>_maskz<round_expand_name>): Likewise.
552 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_1):
554 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_2):
556 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_3):
558 (fmai_vmfnmadd_<mode><round_name>): Likewise.
560 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
563 * config/i386/avx512fintrin.h (_mm512_fmsub_round_pd): Use
564 __builtin_ia32_vfmsubpd512_mask.
565 (_mm512_mask_fmsub_round_pd): Likewise.
566 (_mm512_fmsub_pd): Likewise.
567 (_mm512_mask_fmsub_pd): Likewise.
568 (_mm512_maskz_fmsub_round_pd): Use
569 __builtin_ia32_vfmsubpd512_maskz.
570 (_mm512_maskz_fmsub_pd): Likewise.
571 (_mm512_fmsub_round_ps): Use __builtin_ia32_vfmsubps512_mask.
572 (_mm512_mask_fmsub_round_ps): Likewise.
573 (_mm512_fmsub_ps): Likewise.
574 (_mm512_mask_fmsub_ps): Likewise.
575 (_mm512_maskz_fmsub_round_ps): Use
576 __builtin_ia32_vfmsubps512_maskz.
577 (_mm512_maskz_fmsub_ps): Likewise.
578 * config/i386/avx512vlintrin.h (_mm256_mask_fmsub_pd): Use
579 __builtin_ia32_vfmsubpd256_mask.
580 (_mm256_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd256_maskz.
581 (_mm_mask_fmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
582 (_mm_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd128_maskz.
583 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
584 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
585 (_mm256_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps256_maskz.
586 (_mm_mask_fmsub_ps): Use __builtin_ia32_vfmsubps128_mask.
587 (_mm_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps128_maskz.
588 * config/i386/fmaintrin.h (_mm_fmsub_pd): Use
589 __builtin_ia32_vfmsubpd.
590 (_mm256_fmsub_pd): Use __builtin_ia32_vfmsubpd256.
591 (_mm_fmsub_ps): Use __builtin_ia32_vfmsubps.
592 (_mm256_fmsub_ps): Use __builtin_ia32_vfmsubps256.
593 (_mm_fmsub_sd): Use __builtin_ia32_vfmsubsd3.
594 (_mm_fmsub_ss): Use __builtin_ia32_vfmsubss3.
595 * config/i386/i386-builtin.def: Add
596 __builtin_ia32_vfmsubpd256_mask,
597 __builtin_ia32_vfmsubpd256_maskz,
598 __builtin_ia32_vfmsubpd128_mask,
599 __builtin_ia32_vfmsubpd128_maskz,
600 __builtin_ia32_vfmsubps256_mask,
601 __builtin_ia32_vfmsubps256_maskz,
602 __builtin_ia32_vfmsubps128_mask,
603 __builtin_ia32_vfmsubps128_maskz,
604 __builtin_ia32_vfmsubpd512_mask,
605 __builtin_ia32_vfmsubpd512_maskz,
606 __builtin_ia32_vfmsubps512_mask,
607 __builtin_ia32_vfmsubps512_maskz, __builtin_ia32_vfmsubss3,
608 __builtin_ia32_vfmsubsd3, __builtin_ia32_vfmsubps,
609 __builtin_ia32_vfmsubpd, __builtin_ia32_vfmsubps256 and.
610 __builtin_ia32_vfmsubpd256.
611 * config/i386/sse.md (fma4i_fmsub_<mode>): New.
612 (<avx512>_fmsub_<mode>_maskz<round_expand_name>): Likewise.
613 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_1):
615 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_2):
617 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_3):
619 (fmai_vmfmsub_<mode><round_name>): Likewise.
621 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
623 * config/i386/sse.md (*<plusminus_insn><mode>3<mask_name>_bcst_1):
624 Remove plus. Renamed to ...
625 (*sub<mode>3<mask_name>_bcst): This.
626 (*add<mode>3<mask_name>_bcst_2): Renamede to ...
627 (*add<mode>3<mask_name>_bcst): This.
629 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
632 * config/i386/sse.md (*mul<mode>3<mask_name>_bcst): New.
634 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
637 * i386/avx512vlintrin.h (_mm256_or_epi32): New.
638 (_mm_or_epi32): Likewise.
639 (_mm256_xor_epi32): Likewise.
640 (_mm_xor_epi32): Likewise.
641 (_mm256_or_epi64): Likewise.
642 (_mm_or_epi64): Likewise.
643 (_mm256_xor_epi64): Likewise.
644 (_mm_xor_epi64): Likewise.
646 2018-10-20 H.J. Lu <hongjiu.lu@intel.com>
649 * config/i386/sse.md (*<avx512>_div<mode>3<mask_name>_bcst): New.
651 2018-10-20 Jakub Jelinek <jakub@redhat.com>
654 * varasm.c (decode_addr_const): Handle COMPOUND_LITERAL_EXPR.
656 2018-10-20 Andreas Schwab <schwab@linux-m68k.org>
658 * doc/ux.texi: Move @section directly after @node.
660 2018-10-19 Jakub Jelinek <jakub@redhat.com>
664 * omp-low.c (check_omp_nesting_restrictions): Diagnose ordered without
665 depend closely nested inside of loop with ordered clause with
668 2018-10-19 David Malcolm <dmalcolm@redhat.com>
670 * Makefile.in (TEXI_GCCINT_FILES): Add ux.texi.
671 * doc/gccint.texi: Include ux.texi and use it in top-level menu.
672 * doc/ux.texi: New file.
674 2018-10-19 Segher Boessenkool <segher@kernel.crashing.org>
676 * config/rs6000/rs6000.h (REG_ALLOC_ORDER): Move 68 (that is, CR0) to
677 be the first CR field allocated.
679 2018-10-19 Richard Biener <rguenther@suse.de>
682 * config/i386/i386.c (ix86_builtin_vectorization_cost): Use
683 TYPE_VECTOR_SUBPARTS and avoid relying on vector mode.
685 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
689 (*<plusminus_insn><mode>3<mask_name>_bcst_1): New.
690 (*add<mode>3<mask_name>_bcst_2): Likewise.
692 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
695 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
696 Replace nonimmediate_operand with register_operand.
697 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
699 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
702 2018-10-19 Ilya Leoshkevich <iii@linux.ibm.com>
704 PR rtl-optimization/87596
705 * lra-constraints.c (spill_hard_reg_in_range): Use INSN_P () +
706 lra_get_insn_recog_data () instead of lra_insn_recog_data[]
707 for instructions in FROM..TO range.
709 2018-10-19 Eric Botcazou <ebotcazou@adacore.com>
711 * cfgexpand.c (expand_one_var): Use specific wording in error message
712 for non-local frame variables.
713 * stor-layout.c (layout_decl): Do not issue a warning for them.
715 2018-10-19 Robin Dapp <rdapp@linux.ibm.com>
717 * haifa-sched.c (priority): Add force_recompute parameter.
718 (apply_replacement): Call priority () with force_recompute = true.
719 (restore_pattern): Likewise.
721 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
723 * simplify-rtx.c (simplify_subreg): Limit mask of vec_merge to
724 HOST_BITS_PER_WIDE_INT.
725 (test_vector_ops_duplicate): Likewise.
727 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
730 * config/i386/sse.md (VF_AVX512): New.
731 (avx512bcst): Likewise.
732 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
734 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
736 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
739 2018-10-18 Jonathan Wakely <jwakely@redhat.com>
741 * doc/invoke.texi (-dumpversion): Improve grammar.
742 (-dumpfullversion): Make more consistent with -dumpversion.
744 2018-10-18 Uros Bizjak <ubizjak@gmail.com>
746 * config/i386/i386.c (ix86_emit_fp_unordered_jump):
747 Set JUMP_LABEL to the jump insn.
748 (ix86_emit_i387_log1p): Use ix86_expand_branch to expand branch.
749 Predict emitted jump and add label to jump insn.
751 2018-10-18 David Malcolm <dmalcolm@redhat.com>
753 PR tree-optimization/87562
754 * input.c (get_substring_ranges_for_loc): Use
755 LRK_SPELLING_LOCATION rather than LRK_MACRO_EXPANSION_POINT when
756 getting the linemap for the endpoint. Verify that it's either
757 in the same linemap as the start point's spelling location, or
758 at least in the same file.
760 2018-10-18 Richard Biener <rguenther@suse.de>
762 * config/i386/i386.c (ix86_builtin_vectorization_cost): Do not
763 feed width-specific load/store costs through ix86_vec_cost.
764 * config/i386/x86-tune-costs.h (athlon_cost): Adjust.
766 (bdver_cost): Likewise.
767 (znver1_cost): Likewise.
768 (btver1_cost): Likewise.
769 (btver2_cost): Likewise.
771 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
773 * simplify-rtx.c (simplify_subreg): Call simplify_gen_subreg
774 to simplify subreg of vec_merge.
776 2018-10-18 Richard Biener <rguenther@suse.de>
778 * config/i386/i386.c: Fix costing of vector FMA.
780 2018-10-18 Richard Biener <rguenther@suse.de>
782 * config/i386/i386.c (ix86_vec_cost): Remove !parallel path
784 (ix86_builtin_vectorization_cost): For vec_construct properly
785 cost insertion into SSE regs.
786 (...): Adjust calls to ix86_vec_cost.
788 2018-10-18 Richard Biener <rguenther@suse.de>
792 2018-02-07 Richard Biener <rguenther@suse.de>
794 PR tree-optimization/84204
795 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
798 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
801 * simplify-rtx.c (simplify_subreg): Simplify subreg of vec_merge
803 (test_vector_ops_duplicate): Add test for a scalar subreg of a
804 VEC_MERGE of a VEC_DUPLICATE.
806 2018-10-17 Joseph Myers <joseph@codesourcery.com>
808 * doc/cpp.texi (__STDC_VERSION__): Document C2X handling.
809 * doc/invoke.texi (-std=c2x, -std=gnu2x): Document new options.
810 * doc/standards.texi (C Language): Document C2X.
811 * dwarf2out.c (highest_c_language), config/rl78/rl78.c
812 (rl78_option_override): Handle "GNU C2X" language name.
814 2018-10-17 Joseph Myers <joseph@codesourcery.com>
816 * doc/invoke.texi (-std=c17), doc/standards.texi (C Language):
817 Document C17 as published in 2018.
819 2018-10-17 Eric Botcazou <ebotcazou@adacore.com>
822 * fold-const.c (fold_truth_andor_1): If the right side is not constant,
823 bail out if both sides do not have the same storage order.
825 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
827 * bitmap.c (bitmap_head::dump): New.
828 * bitmap.h (bitmap_head): Add dump().
829 * gimple-ssa-evrp-analyze.c
830 (evrp_range_analyzer::try_find_new_range): Adjust for value_range API.
831 (evrp_range_analyzer::set_ssa_range_info): Same.
832 (evrp_range_analyzer::record_ranges_from_phis): Same.
833 (evrp_range_analyzer::record_ranges_from_stmt): Same.
834 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Same.
835 * gimple-ssa-sprintf.c (get_int_range): Same.
836 (format_integer): Same.
837 (sprintf_dom_walker::handle_gimple_call): Same.
838 * ipa-cp.c (ipcp_vr_lattice::meet_with_1): Same.
839 (ipcp_vr_lattice::top_p): Same.
840 (ipcp_vr_lattice::bottom_p): Same.
841 (ipcp_vr_lattice::set_to_bottom): Same.
842 (ipa_vr_operation_and_type_effects): Same.
843 (propagate_vr_across_jump_function): Same.
844 (ipcp_store_vr_results): Same.
845 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Same.
846 (ipa_print_node_jump_functions_for_edge): Same.
847 (ipa_get_value_range): Same.
848 (ipa_compute_jump_functions_for_edge): Same.
849 (ipa_write_jump_function): Same.
850 * tree-ssa-dom.c (simplify_stmt_for_jump_threading): Same.
851 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
853 * vr-values.c (set_value_range_to_nonnegative): Same.
854 (set_value_range_to_truthvalue): Same.
855 (vr_values::get_value_range): Same.
856 (vr_values::set_defs_to_varying): Same.
857 (vr_values::update_value_range): Same.
858 (symbolic_range_based_on_p): Same.
859 (vr_values::op_with_boolean_value_range_p): Same.
860 (vr_values::extract_range_for_var_from_comparison_expr): Same.
861 (vr_values::extract_range_from_ssa_name): Same.
862 (vr_values::extract_range_from_binary_expr): Same.
863 (vr_values::extract_range_from_unary_expr): Same.
864 (vr_values::extract_range_from_cond_expr): Same.
865 (vr_values::extract_range_from_comparison): Same.
866 (vr_values::check_for_binary_op_overflow): Same.
867 (vr_values::extract_range_basic): Same.
868 (vr_values::extract_range_from_assignment): Same.
869 (compare_ranges): Same.
870 (compare_range_with_value): Same.
871 (vr_values::adjust_range_with_scev): Same.
872 (vrp_valueize): Same.
873 (vrp_valueize_1): Same.
874 (vr_values::get_vr_for_comparison): Same.
875 (vr_values::compare_name_with_value): Same.
876 (vr_values::compare_names): Same.
877 (vr_values::vrp_evaluate_conditional): Same.
878 (find_case_label_ranges): Same.
879 (vr_values::vrp_visit_switch_stmt): Same.
880 (vr_values::extract_range_from_phi_node): Same.
881 (vr_values::simplify_div_or_mod_using_ranges): Same.
882 (vr_values::simplify_bit_ops_using_ranges): Same.
883 (test_for_singularity): Same.
884 (range_fits_type_p): Same.
885 (vr_values::simplify_cond_using_ranges_1): Same.
886 (vr_values::simplify_switch_using_ranges): Same.
887 (vr_values::simplify_float_conversion_using_ranges): Same.
888 (vr_values::two_valued_val_range_p): Same.
889 (vr_values::add_equivalence): Move to value_range::equiv_add.
890 * vr-values.h (vr_values::add_equivalence): Remove.
891 (VR_INITIALIZER): Remove.
892 * tree-vrp.c (value_range::set): New.
893 (value_range::equiv_add): New.
894 (value_range::value_range): New.
895 (value_range::deep_copy): New.
896 (value_range::check): New.
897 (value_range::equal_p): New.
898 (value_range::ignore_equivs_equal_p): New.
899 (value_range::operator==): New.
900 (value_range::operator!=): New.
901 (value_range::symbolic_p): New.
902 (value_range::numeric_p): New.
903 (value_range::set_undefined): New.
904 (value_range::set_varying): New.
905 (value_range::may_contain_p): New.
906 (value_range::equiv_clear): New.
907 (value_range::singleton_p): New.
908 (value_range::intersect): New.
909 (value_range::dump): New.
910 (value_range::set_and_canonicalize): New.
911 (set_value_range): Adjust for value_range API.
912 (set_value_range_to_undefined): Same.
913 (set_value_range_to_varying): Same.
914 (set_and_canonicalize_value_range): Same.
915 (set_value_range_to_nonnull): Same.
916 (set_value_range_to_null): Same.
917 (range_is_null): Same.
918 (range_is_nonnull): Same.
919 (range_int_cst_p): Same.
920 (range_int_cst_singleton_p): Same.
921 (symbolic_range_p): Same.
922 (range_includes_zero_p): Same.
923 (value_range_constant_singleton): Same.
924 (vrp_set_zero_nonzero_bits): Same.
925 (ranges_from_anti_range): Same.
926 (extract_range_into_wide_ints): Same.
927 (extract_range_from_multiplicative_op): Same.
928 (set_value_range_with_overflow): Same.
929 (extract_range_from_binary_expr_1): Same.
930 (extract_range_from_unary_expr): Same.
931 (dump_value_range): Same.
932 (debug_value_range): Same.
933 (vrp_prop::check_array_ref): Same.
934 (vrp_prop::check_mem_ref): Same.
935 (vrp_prop::vrp_initialize): Same.
936 (vrp_prop::visit_stmt): Same.
937 (intersect_ranges): Same.
938 (vrp_prop::visit_phi): Same.
939 (vrp_prop::vrp_finalize): Same.
940 (determine_value_range_1): Same.
941 (determine_value_range): Same.
942 (vrp_intersect_ranges_1): Rename to...
943 (vrp_intersect_1): this.
944 (vrp_intersect_ranges): Rename to...
945 (value_range::intersect_helper): ...this.
946 (vrp_meet_1): Rename to...
947 (value_range::union_helper): ...this.
948 (vrp_meet): Rename to...
949 (value_range::union_): ...this.
950 (copy_value_range): Remove.
951 * tree-vrp.h (struct value_range): Rewrite into a proper class.
952 (value_range::vrtype): New.
953 (value_range::type): New.
954 (value_range::equiv): New.
955 (value_range::min): New.
956 (value_range::max): New.
957 (value_range::varying_p): New.
958 (value_range::undefined_p): New.
959 (value_range::null_p): New.
960 (value_range::equiv_add): New.
961 (copy_value_range): Remove.
963 2018-10-17 David Malcolm <dmalcolm@redhat.com>
965 * Makefile.in (SELFTEST_TARGETS): New.
966 (selftest) Change from s-selftest-c to $(SELFTEST_TARGETS).
967 (C_SELFTEST_FLAGS, C_SELFTEST_DEPS, s-selftest-c, selftest-c-gdb)
968 (selftest-gdb, selftest-c-valgrind, selftest-valgrind): Move to
970 (CPP_SELFTEST_FLAGS, CPP_SELFTEST_DEPS, s-selftest-c++)
971 (selftest-c++-gdb, selftest-c++-valgrind): Move to
973 * configure: Regenerate.
974 * configure.ac (selftest_languages): New.
976 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
978 * tree-vrp.c (extract_range_from_multiplicative_op): Remove
979 overflow wraps argument.
980 (extract_range_from_binary_expr_1): Do not pass overflow wraps to
981 wide_int_range_multiplicative_op.
982 * wide-int-range.cc (wide_int_range_mult_wrapping): Remove
983 overflow wraps argument.
984 (wide_int_range_multiplicative_op): Same.
985 (wide_int_range_lshift): Same.
986 (wide_int_range_div): Same.
987 * wide-int-range.h (wide_int_range_multiplicative_op): Same.
988 (wide_int_range_lshift): Same.
989 (wide_int_range_div): Same.
991 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
993 * wide-int-range.h (wide_int_range_shift_undefined_p): Adjust to
994 use sign as argument.
995 * tree-vrp.c (extract_range_from_binary_expr_1): Pass sign to
996 wide_int_range_shift_undefined_p.
998 2018-10-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1000 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>):
1002 (@despeculate_copy<ALLI_TI:mode>): ... This.
1003 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): Remove
1006 2018-10-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1008 * config.gcc: Obsolete *-*-solaris2.10*.
1009 * doc/install.texi (Specific, *-*-solaris2*): Document it.
1011 2018-10-12 Jeff Law <law@redhat.com>
1013 * config/ft32/ft32.md (ft32_general_movsrc_operand): Disable
1014 reg + sym +- const_int addressing modes.
1016 2018-10-15 David Malcolm <dmalcolm@redhat.com>
1018 * common.opt (fdiagnostics-minimum-margin-width=): New option.
1019 * diagnostic-show-locus.c (layout::layout): Apply the minimum
1021 (layout::start_annotation_line): Only print up to 3 of the
1022 margin character, to avoid touching the left-hand side.
1023 (selftest::test_diagnostic_show_locus_fixit_lines): Update for
1024 minimum margin width, as set by test_diagnostic_context's ctor.
1025 (selftest::test_fixit_insert_containing_newline): Likewise.
1026 (selftest::test_fixit_insert_containing_newline_2): Likewise.
1027 (selftest::test_line_numbers_multiline_range): Clear
1028 dc.min_margin_width.
1029 * diagnostic.c (diagnostic_initialize): Initialize
1031 * diagnostic.h (struct diagnostic_context): Add field
1033 * doc/invoke.texi: Add -fdiagnostics-minimum-margin-width=.
1034 * opts.c (common_handle_option): Handle
1035 OPT_fdiagnostics_minimum_margin_width_.
1036 * selftest-diagnostic.c
1037 (selftest::test_diagnostic_context::test_diagnostic_context):
1038 Initialize min_margin_width to 6.
1039 * toplev.c (general_init): Initialize global_dc->min_margin_width.
1041 2018-10-15 David Malcolm <dmalcolm@redhat.com>
1043 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
1044 Fix usage of "error_at_rich_loc" in the comment.
1046 2018-10-15 Renlin Li <renlin.li@arm.com>
1049 * tree-vectorizer.c (try_vectorize_loop_1): Don't use
1050 if-conversioned loop when it contains ifn with types not
1051 supported by backend.
1052 * internal-fn.c (expand_direct_optab_fn): Add an assert.
1053 (direct_internal_fn_supported_p): New helper function.
1054 * internal-fn.h (direct_internal_fn_supported_p): Declare.
1056 2018-10-15 Jakub Jelinek <jakub@redhat.com>
1059 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
1060 Remove OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
1061 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
1063 2018-10-15 Bin Cheng <bin.cheng@linux.alibaba.com>
1065 PR tree-optimization/87022
1066 * tree-loop-distribution.c (pg_add_dependence_edges): Check all
1067 bits in dist vector rather than the first one.
1069 2018-10-15 Richard Biener <rguenther@suse.de>
1072 * tree-ssa-structalias.c (struct vls_data): Add escaped_p member.
1073 (visit_loadstore): When a used restrict tag escaped verify that
1074 the points-to solution of "other" pointers do not include
1076 (compute_dependence_clique): If a used restrict tag escaped
1077 communicated that down to visit_loadstore.
1079 2018-10-15 Andreas Krebbel <krebbel@linux.ibm.com>
1081 * config/s390/s390.c (s390_expand_vec_init): Force vector element
1082 into reg if it isn't a general operand.
1084 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
1087 * config/i386/sse.md (*vec_dupv2di): Add register source to
1090 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
1093 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
1094 Add OPTION_MASK_ISA_AVX512IFMA_UNSET,
1095 OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
1096 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
1098 2018-10-13 Eric Botcazou <ebotcazou@adacore.com>
1100 * dwarf2cfi.c (struct dw_trace_info): Add args_size_defined_for_eh.
1101 (notice_args_size): Set it in the current trace if no insn that can
1102 throw internally has been seen yet.
1103 (connect_traces): When connecting args_size between traces, allow the
1104 incoming values not to match if there is an insn setting it before the
1105 first insn that can throw internally; in that case, force the creation
1106 of a CFI note on this latter insn.
1108 2018-10-13 Jonathan Wakely <jwakely@redhat.com>
1110 * opt-problem.h (opt_wrapper): Use template-argument-list when naming
1111 the base class, because using the injected-class-name was not clearly
1112 specified until DR 176.
1114 2018-10-12 Paul Koning <ni1d@arrl.net>
1116 * config/pdp11/pdp11.md (doloop_end): New expander.
1117 (doloop_end_insn): renamed from "doloop_end".
1118 (addqi3): New pattern.
1119 (subqi3): New pattern.
1120 * config/pdp11/predicates.md (incdec_operand): New predicate.
1122 2018-10-12 Yury Gribov <tetra2005@gmail.com>
1125 * real.c (format_helper::can_represent_integral_type_p): New function
1126 * real.h (format_helper::can_represent_integral_type_p): Ditto.
1127 * match.pd: New pattern.
1129 2018-10-12 Alexandre Oliva <oliva@adacore.com>
1131 * configure.ac: Introduce --enable-large-address-aware
1132 to define MINGW_DEFAULT_LARGE_ADDR_AWARE.
1133 * doc/install.texi: Document it.
1134 * configure, config.in: Rebuilt.
1135 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Define,
1136 based on MINGW_DEFAULT_LARGE_ADDR_AWARE.
1137 (LINK_SPEC): Insert it.
1138 * config/i386/mingw-264.h: Likewise.
1140 * cygming.h (HAVE_GAS_ALIGNED_COMM): Fallback-define.
1142 2018-10-12 Peter Bergner <bergner@linux.ibm.com>
1144 PR rtl-optimization/87600
1145 * ira-lives (non_conflicting_reg_copy_p): Disable for non LRA targets.
1147 2018-10-12 Paul Koning <ni1d@arrl.net>
1149 * doc/md.texi (doloop_end): Document that the pattern code may
1150 need to check operand mode.
1152 2018-10-12 Wilco Dijkstra <wdijkstr@arm.com>
1154 * config/aarch64/aarch64.md (zero_extendsidi2_aarch64): Add alternatives
1155 to zero-extend between int and floating-point registers.
1156 (load_pair_zero_extendsidi2_aarch64): Add alternative for zero-extended
1157 ldp into floating-point registers. Add type and arch attributes.
1158 (zero_extend<SHORT:mode><GPI:mode>2_aarch64): Add arch attribute.
1159 Use f_loads for type attribute.
1161 2018-10-11 Martin Sebor <msebor@redhat.com>
1163 * doc/extend.texi (attribute packed): Correct typos.
1165 2018-10-11 Martin Sebor <msebor@redhat.com>
1167 * doc/extend.texi (attribute flatten): Mention interaction with
1170 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
1173 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
1174 Set new_decl virtual flag to zero.
1176 2018-10-11 Martin Sebor <msebor@redhat.com>
1179 * doc/extend.texi (attribute format_arg): Discuss using multiple
1180 attributes on a single function.
1182 2018-10-11 Giuliano Belinassi <giuliano.belinassi@usp.br>
1184 PR tree-optimization/86829
1185 * match.pd (sin (atan (x))): New simplification rules.
1186 (cos (atan (x))): Likewise.
1187 * real.c (build_sinatan_real): New function.
1188 * real.h (build_sinatan_real): Prototype.
1190 2018-10-11 Will Schmidt <will_schmidt@vnet.ibm.com>
1192 * config/rs6000/rs6000.c (map_to_integral_tree_type): New helper
1194 (fold_mergeeo_helper): New helper function.
1195 (rs6000_gimple_fold_builtin): Add hooks for vec_mergee and vec_mergeo
1196 intrinsics. Correct some whitespace indentation issues.
1198 2018-10-11 Wilco Dijkstra <wdijkstr@arm.com>
1201 * config/aarch64/aarch64.c (aarch64_mask_and_shift_for_ubfiz_p):
1202 Use HOST_WIDE_INT_1U for shift.
1204 2018-10-11 Doug Rupp <rupp@adacore.com>
1205 Olivier Hainque <hainque@adacore.com>
1207 * config/rs6000/vxworks.h (VXWORKS_RELAX_LINK_SPEC): New macro.
1208 Pass --relax to the linker for RTPs.
1209 (LINK_SPEC): Append VXWORKS_RELAX_LINK_SPEC.
1211 2018-10-11 Andrew Stubbs <ams@codesourcery.com>
1212 Jan Hubicka <jh@suse.cz>
1213 Martin Jambor <mjambor@suse.cz>
1215 * print-rtl.c (print_rtx_operand_codes_E_and_V): Print how many times
1216 the same elements are repeated rather than printing all of them.
1217 * read-rtl.c (rtx_reader::read_rtx_operand): Recognize and expand
1218 "repeated" elements.
1219 * read-rtl-function.c (test_loading_repeat): New function.
1220 (read_rtl_function_c_tests): Call test_loading_repeat.
1221 * rtl-tests.c (test_dumping_repeat): New function.
1222 (rtl_tests_c_tests): Call test_dumping_repeat.
1224 2018-10-11 Richard Biener <rguenther@suse.de>
1226 * config/i386/x86-tune-costs.h (bdver?_memcpy, bdver?_memset,
1227 bdver?_cost): Unify to ...
1228 (bdver_memcpy, bdver_memset, bdver_cost): ... this.
1229 * config/i386/i386.c (processor_cost_table): Adjust.
1231 2018-10-10 Eric Botcazou <ebotcazou@adacore.com>
1234 * cgraphunit.c (cgraph_node::expand_thunk): Force DECL_IGNORED_P on
1235 the thunk when expanding to GIMPLE.
1237 2018-10-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
1239 * varasm.c (mergeable_string_section): Don't try to move zero-length
1240 strings to the merge section.
1242 2018-10-10 Uros Bizjak <ubizjak@gmail.com>
1245 * config/i386/mmx.md (const_vector 0 -> mem splitter): New splitter.
1247 2018-10-10 Jakub Jelinek <jakub@redhat.com>
1250 * config/i386/i386-builtin.def (IX86_BUILTIN_RDPMC): Move from args set
1251 to special_args set.
1253 2018-10-10 Richard Biener <rguenther@suse.de>
1255 * config/i386/sse.md (reduc_plus_scal_v8df, reduc_plus_scal_v4df,
1256 reduc_plus_scal_v2df, reduc_plus_scal_v16sf, reduc_plus_scal_v8sf,
1257 reduc_plus_scal_v4sf): Merge into pattern reducing to half width
1258 and recursing and pattern terminating the recursion on SSE
1259 vector width using ix86_expand_reduc.
1260 (reduc_sminmax_scal_<mode>): Split into part reducing to half
1261 width and recursing and SSE2 vector variant doing the final
1262 reduction with ix86_expand_reduc.
1263 (reduc_uminmax_scal_<mode>): Likewise for the AVX512 variants
1264 with terminating the recursion at AVX level, splitting that
1267 2018-10-09 David Malcolm <dmalcolm@redhat.com>
1269 * genmatch.c (error_cb): Rename to...
1270 (diagnostic_cb): ...this, converting int params to enums.
1271 (fatal_at): Update for renaming.
1272 (warning_at): Likewise.
1274 * input.c (selftest::ebcdic_execution_charset::apply):
1275 Update for renaming of...
1276 (selftest::ebcdic_execution_charset::on_error): ...this, renaming
1278 (selftest::ebcdic_execution_charset::on_diagnostic): ...this,
1279 converting level and reason to enums.
1280 (class selftest::lexer_error_sink): Rename to...
1281 (class selftest::lexer_test_options): ...this, renaming field
1282 "m_errors" to "m_diagnostics".
1283 (selftest::lexer_test_options::apply): Update for renaming of...
1284 (selftest::lexer_test_options::on_error): ...this, renaming to...
1285 (selftest::lexer_test_options::on_diagnostic): ...this
1286 converting level and reason to enums.
1287 (selftest::test_lexer_string_locations_raw_string_unterminated):
1288 Update for renamings.
1289 * opth-gen.awk (struct cpp_reason_option_codes_t): Use enum for
1292 2018-10-09 Paul A. Clarke <pc@us.ibm.com>
1294 * config.gcc (powerpc*-*-*): Add pmmintrin.h to extra_headers.
1295 * config/rs6000/pmmintrin.h: New file.
1297 2018-10-09 Eric Botcazou <ebotcazou@adacore.com>
1299 PR tree-optimization/86659
1300 * gimple-match.h (gimple_match_op constructors): Initialize reverse.
1302 2018-10-09 Richard Biener <rguenther@suse.de>
1304 PR tree-optimization/63155
1305 * tree-ssa-structalias.c: Include tree-ssa.h.
1306 (get_constraint_for_ssa_var): For undefs return nothing_id.
1307 (find_func_aliases): Cleanup PHI handling.
1309 2018-10-09 Richard Biener <rguenther@suse.de>
1311 * tree-vectorizer.c (dump_stmt_cost): Fix cut&paste missing
1314 2018-10-09 Martin Liska <mliska@suse.cz>
1316 * asan.c (asan_emit_stack_protection): If a stack variable
1317 is located in a same file as current function, then emit
1318 line info into variable definition string.
1320 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
1322 * print-rtl.c (rtx_writer::print_rtx_operand_code_i): Print column
1325 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
1327 * cgraphunit.c (cgraph_node::expand_thunk): Do not force DECL_IGNORED_P
1330 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
1332 PR tree-optimization/86659
1333 * gimple-match.h (struct gimple_match_op): Add reverse field.
1334 (gimple_match_op::set_op): New overloaded method.
1335 * gimple-match-head.c (maybe_build_generic_op) <BIT_FIELD_REF>: Set
1336 the REF_REVERSE_STORAGE_ORDER flag on the value.
1337 (gimple_simplify) <GIMPLE_ASSIGN>: For BIT_FIELD_REF, propagate the
1338 REF_REVERSE_STORAGE_ORDER flag and avoid simplifying if it is set.
1340 2018-10-08 Richard Sandiford <richard.sandiford@arm.com>
1343 * gimple-ssa-backprop.c (backprop::intersect_uses): Use
1344 FOR_EACH_IMM_USE_FAST instead of FOR_EACH_IMM_USE_STMT.
1346 2018-10-08 H.J. Lu <hongjiu.lu@intel.com>
1349 * config/i386/avx512fintrin.h (_mm512_mask_fmaddsub_round_pd):
1350 Defined with __builtin_ia32_vfmaddsubpd512_mask.
1352 2018-10-08 Richard Biener <rguenther@suse.de>
1354 * config/i386/x86-tune-costs.h (znver1_cost): Make AVX256 vector loads
1355 cost the same as AVX128 ones.
1357 2018-10-08 Paul Koning <ni1d@arrl.net>
1359 * config/pdp11/pdp11-protos.h (output_block_move): Remove.
1360 (expand_block_move): New function.
1361 * config/pdp11/pdp11.c (output_block_move): Remove.
1362 (expand_block_move): New function.
1363 * config/pdp11/pdp11.h (MOVE_RATIO): New definition.
1364 * config/pdp11/pdp11.md (movmemhi): Use expand_block_move.
1365 (*movmemhi1): Remove.
1367 2018-10-08 Robin Dapp <rdapp@linux.ibm.com>
1369 * config/s390/2827.md: Increase latencies for some FP instructions.
1371 2018-10-08 Richard Biener <rguenther@suse.de>
1373 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
1375 * tree-vectorizer.c (dump_stmt_cost): Add cost param and dump it.
1376 * tree-vectorizer.h (dump_stmt_cost): Adjust.
1377 (add_stmt_cost): Dump return value of the hook.
1379 2018-10-08 Richard Biener <rguenther@suse.de>
1381 PR tree-optimization/63155
1382 * tree-ssa-propagate.c (add_ssa_edge): Do cheap check first.
1383 (ssa_propagation_engine::ssa_propagate): Remove redundant
1384 bitmap bit clearing.
1386 2018-10-05 Peter Bergner <bergner@linux.ibm.com>
1388 PR rtl-optimization/86939
1389 PR rtl-optimization/87479
1390 * ira.h (non_conflicting_reg_copy_p): New prototype.
1391 * ira-lives.c (ignore_reg_for_conflicts): New static variable.
1392 (make_hard_regno_dead): Don't add conflicts for register
1393 ignore_reg_for_conflicts.
1394 (make_object_dead): Likewise.
1395 (non_conflicting_reg_copy_p): New function.
1396 (process_bb_node_lives): Set ignore_reg_for_conflicts for copies.
1397 Remove special conflict handling of REAL_PIC_OFFSET_TABLE_REGNUM.
1398 * lra-lives.c (ignore_reg_for_conflicts): New static variable.
1399 (make_hard_regno_dead): Don't add conflicts for register
1400 ignore_reg_for_conflicts. Remove special conflict handling of
1401 REAL_PIC_OFFSET_TABLE_REGNUM. Remove now unused argument
1402 check_pic_pseudo_p and update callers.
1403 (mark_pseudo_dead): Don't add conflicts for register
1404 ignore_reg_for_conflicts.
1405 (process_bb_lives): Set ignore_reg_for_conflicts for copies.
1407 2018-10-05 Andrew Waterman <andrew@sifive.com>
1408 Jim Wilson <jimw@sifive.com>
1410 * config/riscv/riscv.md (f<quiet_pattern>_quiet<ANYF:mode><X:mode>4):
1411 Add define_expand. Add ! HONOR_SNANS check to current pattern. Add
1412 new pattern using HONOR_SNANS that emits one extra instruction.
1414 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
1416 * config/rs6000/rs6000.md (unnamed mfcr scc_comparison_operator
1417 patterns): Merge SI and DI patterns to a GPR pattern.
1418 (unnamed define_insn and define_split for record form of that): Merge
1419 to a single define_insn_and_split pattern.
1421 2018-10-05 David Malcolm <dmalcolm@redhat.com>
1424 * input.c (expand_location_to_spelling_point): Add param "aspect"
1425 and use rather than hardcoding LOCATION_ASPECT_CARET.
1426 (get_substring_ranges_for_loc): Handle the case of a single token
1427 within a macro expansion.
1428 * input.h (expand_location_to_spelling_point): Add "aspect" param,
1429 defaulting to LOCATION_ASPECT_CARET.
1431 2018-10-05 Paul Koning <ni1d@arrl.net>
1433 * config/pdp11/pdp11.c (TARGET_CXX_GUARD_TYPE): Define.
1434 (TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT): Define.
1435 (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
1436 (pdp11_guard_type): New function.
1438 2018-10-05 Paul Koning <ni1d@arrl.net>
1440 * config/pdp11/pdp11.h (FLOAT_TYPE_SIZE): Always 32.
1441 * config/pdp11/pdp11.opt (mfloat32): Remove.
1443 * doc/invoke.texi (pdp11 -mfloat32): Remove:
1444 (pdp11 -mfloat64): Remove.
1446 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
1448 * config/i386/i386.md (*cmpxf_cc_i387): Remove pattern.
1449 (*cmp<mode>_cc_i387): Ditto.
1450 (*cmpu<mode>_cc_i387): Ditto.
1451 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
1452 * config/i386/i386.c (ix86_expand_fp_compare): Remove
1454 <case IX86_FPCMP_SAHF>: Do not generate pattern with HImode clobber.
1455 Emit x86_sahf_1 pattern.
1456 (ix86_expand_compare): Update call to ix86_expand_fp_compare.
1457 (ix86_expand_carry_flag_compare): Ditto.
1459 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
1461 * config/i386/i386.md (*cmpxf_i387): Change operand 2 predicate
1462 to reg_or_0_operand. Add "C" constraint.
1463 (*cmpxf_cc_i387): Ditto.
1464 (*cmp<mode>_i387): Change operand 2 predicate
1465 to nonimm_or_0_operand. Add "C" constraint.
1466 (*cmp<mode>_cc_i387): Ditto.
1467 (*cmp<mode>_0_i387): Remove insn pattern.
1468 (*cmp<mode>_0_cc_i387): Ditto.
1470 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
1472 * config/i386/constraints.md ("C"): Do not depend on TARGET_SSE.
1473 * config/i386/predicates.md (nonimm_or_0_operand): Rename
1474 from vector_move_operand. Update all uses.
1476 2018-10-05 Martin Sebor <msebor@redhat.com>
1478 PR tree-optimization/87490
1479 * builtins.c (expand_builtin_strnlen): Handle a null data.decl
1482 2018-10-05 Richard Biener <rguenther@suse.de>
1484 PR tree-optimization/63155
1485 * tree-ssa-ccp.c (ccp_propagate::visit_phi): Avoid excess
1486 vertical space in dumpfiles.
1487 * tree-ssa-propagate.h
1488 (ssa_propagation_engine::process_ssa_edge_worklist): Remove.
1489 * tree-ssa-propagate.c (cfg_blocks_back): New global.
1490 (ssa_edge_worklist_back): Likewise.
1491 (curr_order): Likewise.
1492 (cfg_blocks_get): Remove abstraction.
1493 (cfg_blocks_add): Likewise.
1494 (cfg_blocks_empty_p): Likewise.
1495 (add_ssa_edge): Add to current or next worklist based on
1497 (add_control_edge): Likewise.
1498 (ssa_propagation_engine::process_ssa_edge_worklist): Fold
1500 (ssa_propagation_engine::ssa_propagate): ... here. Unify
1501 iteration from CFG and SSA edge worklist so we process
1502 everything in RPO order, prioritizing forward progress
1504 (ssa_prop_init): Allocate new worklists, do not dump
1506 (ssa_prop_fini): Free new worklists.
1508 2018-10-05 Richard Biener <rguenther@suse.de>
1510 * tree-core.h (tree_block::abstract_flag): Remove.
1511 (tree_block::block_num): Make full 32bits.
1512 * tree.def (BLOCK): Remove docs about BLOCK_ABSTRACT.
1513 * tree.h (BLOCK_ABSTRACT): Remove.
1514 * dwarf2out.c (gen_lexical_block_die): Remove dead code
1515 resulting from BLOCK_ABSTRACT being always false.
1516 (gen_inlined_subroutine_die): Likewise.
1517 (gen_block_die): Likewise.
1518 * tree.c (block_ultimate_origin): Likewise.
1519 * tree-pretty-print.c (dump_block_node): Remove code dealing
1520 with BLOCK_ABSTRACT.
1521 * tree-ssa-live.c (dump_scope_block): Likewise.
1522 * tree-streamer-in.c (unpack_ts_block_value_fields): Likewise.
1523 * tree-streamer-out.c (pack_ts_block_value_fields): Likewise.
1525 2018-10-05 Richard Biener <rguenther@suse.de>
1527 * config/i386/i386.c (ix86_add_stmt_cost): When scalar cost
1528 is asked for initialize mode to the component mode of the
1531 2018-10-05 H.J. Lu <hongjiu.lu@intel.com>
1534 * config/i386/gnu-user.h (ASM_SPEC): Don't pass -msse2avx to
1535 assembler for -mavx.
1536 * config/i386/gnu-user64.h (ASM_SPEC): Likewise.
1538 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
1541 * config/rs6000/rs6000-builtin.def (RS6000_BUILTIN_SET_FPSCR_DRN): Use
1543 * config/rs6000/rs6000.md (rs6000_set_fpscr_rn): Require the operand
1544 to be DImode. When using mffscrn, force the operand to a register.
1546 2018-10-04 Uros Bizjak <ubizjak@gmail.com>
1548 * config/i386/i386.md (*fop_<X87MODEF:mode>_2_i387): Macroize insn
1549 from *fop_<MODEF:mode>_2_i387 and *fop_xf_2_i387 using
1550 X87MODEF mode iterator.
1551 (*fop_<X87MODEF:mode>_3_i387): Macroize insn from
1552 *fop_<MODEF:mode>_3_i387 and *fop_xf_3_i387 using
1553 X87MODEF mode iterator.
1555 2018-10-04 Vinay Kumar <vinay.kumar@blackfigtech.com>
1557 * doc/invoke.texi (-Wno-prio-ctor-dtor): Document new warning
1558 -Wno-prio-ctor-dtor.
1560 2018-10-04 David Malcolm <dmalcolm@redhat.com>
1562 * Makefile.in (OBJS): Add opt-problem.o.
1563 * dump-context.h: Include "selftest.h.
1564 (selftest::temp_dump_context): New forward decl.
1565 (class dump_context): Make friend of class
1566 selftest::temp_dump_context.
1567 (dump_context::dump_loc_immediate): New decl.
1568 (class dump_pretty_printer): Move here from dumpfile.c.
1569 (class temp_dump_context): Move to namespace selftest.
1570 (temp_dump_context::temp_dump_context): Add param
1571 "forcibly_enable_dumping".
1572 (selftest::verify_dumped_text):
1573 (ASSERT_DUMPED_TEXT_EQ): Move here from dumpfile.c.
1574 (selftest::verify_item):
1575 (ASSERT_IS_TEXT): Move here from dumpfile.c.
1576 (ASSERT_IS_TREE): Likewise.
1577 (ASSERT_IS_GIMPLE): Likewise.
1578 * dumpfile.c (dump_context::dump_loc): Move immediate dumping
1580 (dump_context::dump_loc_immediate): ...this new function.
1581 (class dump_pretty_printer): Move to dump-context.h.
1582 (dump_switch_p_1): Don't enable MSG_PRIORITY_REEMITTED.
1583 (opt_info_switch_p_1): Enable MSG_PRIORITY_REEMITTED.
1584 (temp_dump_context::temp_dump_context): Move to "selftest"
1585 namespace. Add param "forcibly_enable_dumping", and use it to
1586 conditionalize the use of m_pp;
1587 (selftest::verify_dumped_text): Make non-static.
1588 (ASSERT_DUMPED_TEXT_EQ): Move to dump-context.h.
1589 (selftest::verify_item): Make non-static.
1590 (ASSERT_IS_TEXT): Move to dump-context.h.
1591 (ASSERT_IS_TREE): Likewise.
1592 (ASSERT_IS_GIMPLE): Likewise.
1593 (selftest::test_capture_of_dump_calls): Pass "true" for new
1594 param of temp_dump_context.
1595 * dumpfile.h (enum dump_flag): Add MSG_PRIORITY_REEMITTED, adding
1596 it to MSG_ALL_PRIORITIES. Update values of TDF_COMPARE_DEBUG and
1598 * opt-problem.cc: New file.
1599 * opt-problem.h: New file.
1600 * optinfo-emit-json.cc
1601 (selftest::test_building_json_from_dump_calls): Pass "true" for
1602 new param of temp_dump_context.
1603 * optinfo.cc (optinfo_kind_to_dump_flag): New function.
1604 (optinfo::emit_for_opt_problem): New function.
1605 (optinfo::emit): Clarity which emit_item is used.
1606 * optinfo.h (optinfo::get_dump_location): New accessor.
1607 (optinfo::emit_for_opt_problem): New decl.
1608 (optinfo::emit): Make const.
1609 * selftest-run-tests.c (selftest::run_tests): Call
1610 selftest::opt_problem_cc_tests.
1611 * selftest.h (selftest::opt_problem_cc_tests): New decl.
1612 * tree-data-ref.c (dr_analyze_innermost): Convert return type from
1613 bool to opt_result, converting fprintf messages to
1614 opt_result::failure_at calls. Add "stmt" param for use by the
1616 (create_data_ref): Pass "stmt" to the dr_analyze_innermost call.
1617 (runtime_alias_check_p): Convert return type from bool to
1618 opt_result, converting dump_printf calls to
1619 opt_result::failure_at, using the statement DDR_A for their
1621 (find_data_references_in_stmt): Convert return type from bool to
1622 opt_result, converting "return false" to opt_result::failure_at
1624 * tree-data-ref.h: Include "opt-problem.h".
1625 (dr_analyze_innermost): Convert return type from bool to opt_result,
1626 and add a const gimple * param.
1627 (find_data_references_in_stmt): Convert return type from bool to
1629 (runtime_alias_check_p): Likewise.
1630 * tree-predcom.c (find_looparound_phi): Pass "init_stmt" to
1631 dr_analyze_innermost.
1632 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test):
1633 Convert return type from bool to opt_result, adding a message for
1634 the PARAM_VECT_MAX_VERSION_FOR_ALIAS_CHECKS zero case.
1635 (vect_analyze_data_ref_dependence): Convert return type from bool
1636 to opt_result. Change sense of return type from "false"
1637 effectively meaning "no problems" to "false" meaning a problem,
1638 so that "return false" becomes "return opt_result::success".
1639 Convert "return true" calls to opt_result::failure_at, using
1640 the location of statement A rather than vect_location.
1641 (vect_analyze_data_ref_dependences): Convert return type from bool
1643 (verify_data_ref_alignment): Likewise, converting dump_printf_loc
1644 calls to opt_result::failure_at, using the stmt location rather
1646 (vect_verify_datarefs_alignment): Convert return type from bool
1648 (vect_enhance_data_refs_alignment): Likewise. Split local "stat"
1649 into multiple more-tightly-scoped copies.
1650 (vect_analyze_data_refs_alignment): Convert return type from bool
1652 (vect_analyze_data_ref_accesses): Likewise, converting a
1653 "return false" to a "return opt_result::failure_at", adding a
1655 (vect_prune_runtime_alias_test_list): Convert return type from
1656 bool to opt_result, converting dump_printf_loc to
1657 opt_result::failure_at. Add a %G to show the pertinent statement,
1658 and use the stmt's location rather than vect_location.
1659 (vect_find_stmt_data_reference): Convert return type from
1660 bool to opt_result, converting dump_printf_loc to
1661 opt_result::failure_at, using stmt's location.
1662 (vect_analyze_data_refs): Convert return type from bool to
1663 opt_result. Convert "return false" to "return
1664 opt_result::failure_at", adding messages as needed.
1665 * tree-vect-loop.c (vect_determine_vf_for_stmt_1): Convert return
1666 type from bool to opt_result.
1667 (vect_determine_vf_for_stmt): Likewise.
1668 (vect_determine_vectorization_factor): Likewise, converting
1669 dump_printf_loc to opt_result::failure_at, using location of phi
1670 rather than vect_location.
1671 (vect_analyze_loop_form_1): Convert return type from bool to
1672 opt_result, converting dump_printf_loc calls, retaining the use of
1674 (vect_analyze_loop_form): Convert return type from loop_vec_info
1675 to opt_loop_vec_info.
1676 (vect_analyze_loop_operations): Convert return type from bool to
1677 opt_result, converting dump_printf_loc calls, using the location
1678 of phi/stmt rather than vect_location where available. Convert
1679 various "return false" to "return opt_result::failure_at" with
1680 "unsupported phi" messages.
1681 (vect_get_datarefs_in_loop): Convert return type from bool to
1682 opt_result. Add a message for the
1683 PARAM_LOOP_MAX_DATAREFS_FOR_DATADEPS failure.
1684 (vect_analyze_loop_2): Convert return type from bool to
1685 opt_result. Ensure "ok" is set to a opt_result::failure_at before
1686 each "goto again;", adding new messages where needed.
1687 Add "unsupported grouped {store|load}" messages.
1688 (vect_analyze_loop): Convert return type from loop_vec_info to
1690 * tree-vect-slp.c (vect_analyze_slp): Convert return type from
1692 * tree-vect-stmts.c (process_use): Likewise, converting
1693 dump_printf_loc call and using stmt location, rather than
1695 (vect_mark_stmts_to_be_vectorized): Likeise.
1696 (vect_analyze_stmt): Likewise, adding a %G.
1697 (vect_get_vector_types_for_stmt): Convert return type from bool to
1698 opt_result, converting dump_printf_loc calls and using stmt
1699 location, rather than vect_location.
1700 (vect_get_mask_type_for_stmt): Convert return type from tree to
1701 opt_tree, converting dump_printf_loc calls and using stmt location.
1702 * tree-vectorizer.c: Include "opt-problem.h.
1703 (try_vectorize_loop_1): Flag "Analyzing loop at" dump message as
1704 MSG_PRIORITY_INTERNALS. Convert local "loop_vinfo" from
1705 loop_vec_info to opt_loop_vec_info. If if fails, and dumping is
1706 enabled, use it to report at the top level "couldn't vectorize
1707 loop" followed by the problem.
1708 * tree-vectorizer.h (opt_loop_vec_info): New typedef.
1709 (vect_mark_stmts_to_be_vectorized): Convert return type from bool
1711 (vect_analyze_stmt): Likewise.
1712 (vect_get_vector_types_for_stmt): Likewise.
1713 (tree vect_get_mask_type_for_stmt): Likewise.
1714 (vect_analyze_data_ref_dependences): Likewise.
1715 (vect_enhance_data_refs_alignment): Likewise.
1716 (vect_analyze_data_refs_alignment): Likewise.
1717 (vect_verify_datarefs_alignment): Likewise.
1718 (vect_analyze_data_ref_accesses): Likewise.
1719 (vect_prune_runtime_alias_test_list): Likewise.
1720 (vect_find_stmt_data_reference): Likewise.
1721 (vect_analyze_data_refs): Likewise.
1722 (vect_analyze_loop): Convert return type from loop_vec_info to
1724 (vect_analyze_loop_form): Likewise.
1725 (vect_analyze_slp): Convert return type from bool to opt_result.
1727 2018-10-04 David Malcolm <dmalcolm@redhat.com>
1729 * doc/invoke.texi (-fopt-info): Document new "internals"
1731 * dump-context.h (dump_context::apply_dump_filter_p): New decl.
1732 * dumpfile.c (dump_options): Update for renaming of MSG_ALL to
1734 (optinfo_verbosity_options): Add "internals".
1735 (kind_as_string): Update for renaming of MSG_ALL to MSG_ALL_KINDS.
1736 (dump_context::apply_dump_filter_p): New member function.
1737 (dump_context::dump_loc): Use apply_dump_filter_p rather than
1738 explicitly masking the dump_kind.
1739 (dump_context::begin_scope): Increment the scope depth first. Use
1740 apply_dump_filter_p rather than explicitly masking the dump_kind.
1741 (dump_context::emit_item): Use apply_dump_filter_p rather than
1742 explicitly masking the dump_kind.
1743 (dump_dec): Likewise.
1744 (dump_hex): Likewise.
1745 (dump_switch_p_1): Default to MSG_ALL_PRIORITIES.
1746 (opt_info_switch_p_1): Default to MSG_PRIORITY_USER_FACING.
1747 (opt_info_switch_p): Update handling of default
1748 MSG_OPTIMIZED_LOCATIONS to cope with default of
1749 MSG_PRIORITY_USER_FACING.
1750 (dump_basic_block): Use apply_dump_filter_p rather than explicitly
1751 masking the dump_kind.
1752 (selftest::test_capture_of_dump_calls): Update test_dump_context
1753 instances to use MSG_ALL_KINDS | MSG_PRIORITY_USER_FACING rather
1754 than MSG_ALL. Generalize scope test to be run at all four
1755 combinations of with/without MSG_PRIORITY_USER_FACING and
1756 MSG_PRIORITY_INTERNALS, adding examples of explicit priority
1757 for each of the two values.
1758 * dumpfile.h (enum dump_flag): Add comment about the MSG_* flags.
1759 Rename MSG_ALL to MSG_ALL_KINDS. Add MSG_PRIORITY_USER_FACING,
1760 MSG_PRIORITY_INTERNALS, and MSG_ALL_PRIORITIES, updating the
1761 values for TDF_COMPARE_DEBUG and TDF_ALL_VALUES.
1762 (AUTO_DUMP_SCOPE): Add a note to the comment about the interaction
1763 with MSG_PRIORITY_*.
1764 * tree-vect-loop-manip.c (vect_loop_versioning): Mark versioning
1765 dump messages as MSG_PRIORITY_USER_FACING.
1766 * tree-vectorizer.h (DUMP_VECT_SCOPE): Add a note to the comment
1767 about the interaction with MSG_PRIORITY_*.
1769 2018-10-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1771 * varasm.c (output_constant): Add new parameter merge_strings.
1772 Make strings properly zero terminated in merge string sections.
1773 (mergeable_string_section): Don't fail if the last char is non-zero.
1774 (assemble_variable_contents): Handle merge string sections.
1775 (assemble_variable): Likewise.
1776 (assemble_constant_contents): Likewise.
1777 (output_constant_def_contents): Likewise.
1778 (output_constructor_array_range,
1779 output_constructor_regular_field): Adjust call to output_constant.
1780 (output_object_block): Adjust call to assemble_constant_contents
1781 and assemble_variable_contents.
1783 2018-10-04 Martin Liska <mliska@suse.cz>
1786 * cgraphunit.c (process_function_and_variable_attributes):
1787 Warn about a function with alias attribute and a body.
1789 2018-10-04 Martin Liska <mliska@suse.cz>
1792 * multiple_target.c (redirect_to_specific_clone): New function.
1793 (ipa_target_clone): Use it.
1794 * tree-inline.c: Fix comment.
1796 2018-10-04 David Malcolm <dmalcolm@redhat.com>
1798 * dumpfile.c (gcc::dump_manager::dump_manager): Initialize new
1800 (gcc::dump_manager::~dump_manager): Free m_optinfo_filename.
1801 (gcc::dump_manager::register_pass): New member function, adapted
1802 from loop body in gcc::pass_manager::register_pass, adding a
1803 call to update_dfi_for_opt_info.
1804 (gcc::dump_manager::opt_info_enable_passes): Store the
1805 -fopt-info options into the new fields. Move the loop
1807 (gcc::dump_manager::update_dfi_for_opt_info): ...this new member
1809 * dumpfile.h (struct opt_pass): New forward decl.
1810 (gcc::dump_manager::register_pass): New decl.
1811 (gcc::dump_manager::update_dfi_for_opt_info): New decl.
1812 (class gcc::dump_manager): Add fields "m_optgroup_flags",
1813 "m_optinfo_flags", and "m_optinfo_filename".
1814 * passes.c (gcc::pass_manager::register_pass): Move all of the
1815 dump-handling code to gcc::dump_manager::register_pass.
1817 2018-10-04 Peter Bergner <bergner@linux.ibm.com>
1819 PR rtl-optimization/87466
1820 * target.def (setjmp_preserves_nonvolatile_regs_p): New target hook.
1821 * doc/tm.texi.in (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P): New hook.
1822 * doc/tm.texi: Regenerate.
1823 * ira-lives.c (process_bb_node_lives): Use the new target hook.
1824 * lra-lives.c (process_bb_lives): Likewise.
1825 * config/rs6000/rs6000.c (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P):
1828 2018-10-04 Tamar Christina <tamar.christina@arm.com>
1830 * params.c (add_params): Fix initialization.
1832 2018-10-04 Martin Liska <mliska@suse.cz>
1834 PR gcov-profile/84107
1835 * tree-profile.c (init_ic_make_global_vars):
1836 Remove ic_void_ptr_var and ic_gcov_type_ptr_var.
1837 Come up with new ic_tuple* variables. Emit
1838 __gcov_indirect_call{,_topn} variables.
1839 (gimple_gen_ic_profiler): Access the variable
1841 (gimple_gen_ic_func_profiler): Access
1842 __gcov_indirect_call.callee field.
1843 (gimple_init_gcov_profiler): Use ptr_type_node.
1844 * value-prof.c (gimple_ic): Use ptr_type_node.
1846 2018-10-04 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1848 PR tree-optimization/85787
1849 * ipa-pure-const.c (malloc_candidate_p_1): Move most of malloc_candidate_p
1850 into this function and add support for detecting multiple phis.
1851 (DUMP_AND_RETURN): Move from malloc_candidate_p into top-level macro.
1853 2018-10-04 Martin Liska <mliska@suse.cz>
1856 * ipa-inline.c (inline_to_all_callers_1):
1857 Call ultimate_alias_target for node being inlined.
1859 2018-10-03 Jeff Law <law@redhat.com>
1861 * gimple-ssa-sprintf.c (format_string): Do not hardcode size of
1863 * tree.c (get_typenode_from_name): Moved from fortran/trans-types.c.
1864 * tree.h (get_typenode_from_name): Prototype.
1866 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
1868 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
1869 Change operand 2 predicate to nonimmediate_operand.
1870 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
1872 2018-10-03 Martin Sebor <msebor@redhat.com>
1873 Jeff Law <law@redhat.com>
1875 * gimple-ssa-sprintf.c (struct fmtresult): Add new member and
1877 (get_string_length): Detect unterminated arrays.
1878 (format_string): Same.
1879 (format_directive): Warn about unterminated arrays.
1880 (handle_gimple_call): Mark statements with no_warning as needed.
1882 2018-10-03 Jim Wilson <jimw@sifive.com>
1884 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): For ABI_ILP32E,
1885 also define __riscv_abi_rve. Delete trailing white space.
1887 2018-10-03 Paul Koning <ni1d@arrl.net>
1889 Enable LRA register allocator for PDP11.
1890 * config/pdp11/constraints.md (Q): Use define_memory_constraint.
1893 * config/pdp11/pdp11.c (pdp11_lra_p): New function.
1894 * config/pdp11/pdp11.opt (-mlra): New option.
1895 * doc/invoke.texi (PDP-11 Options): Document -mlra.
1897 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
1899 * config/i386/i386.md (*<absneg:code>extendsfdf2): Remove.
1900 (*<absneg:code>extend<mode>xf2): Ditto.
1902 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
1904 PR tree-optimization/87415
1905 * tree-vrp.c (set_value_range_with_overflow): Special case one bit
1908 2018-10-02 Jeff Law <law@redhat.com>
1910 * gimple-fold.c (get_range_strlen): Only set *nonstr when
1911 an unterminated string is discovered. Bubble up range
1912 even for unterminated strings.
1913 (gimple_fold_builtin_strlen): Do not fold if get_range_strlen
1914 indicates the string was not terminated via NONSTR.
1916 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
1918 * tree-vrp.c (extract_range_from_unary_expr): Special case all
1919 pointer conversions.
1920 Do not do anything special for anti-ranges.
1922 2018-10-03 Jérôme Lambourg <lambourg@adacore.com>
1924 * config/arm/vxworks.h (ARM_TARGET2_DWARF_FORMAT): Adjust to
1925 DW_EH_PE_pcrel | DW_EH_PE_indirect for RTPs.
1927 2018-10-03 Martin Liska <mliska@suse.cz>
1929 PR gcov-profile/86109
1930 * coverage.c (coverage_begin_function): Do not
1931 mark lambdas as artificial.
1932 * tree-core.h (struct GTY): Remove tm_clone_flag
1933 and introduce new lambda_function.
1934 * tree.h (DECL_LAMBDA_FUNCTION): New macro.
1936 2018-10-02 Aaron Sawdey <acsawdey@linux.ibm.com>
1939 * config/rs6000/rs6000-string.c (expand_strn_compare): Check that both
1940 P8_VECTOR and VSX are enabled.
1942 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
1944 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
1945 0x3907 as CPU model number.
1947 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
1949 * common/config/s390/s390-common.c: Rename PF_ARCH12 to PF_Z14.
1950 * config/s390/s390.h (enum processor_flags): Rename PF_ARCH12 to
1951 PF_Z14. Rename TARGET_CPU_ARCH12 to TARGET_CPU_Z14,
1952 TARGET_CPU_ARCH12_P to TARGET_CPU_Z14_P, TARGET_ARCH12 to
1953 TARGET_Z14, and TARGET_ARCH12_P to TARGET_Z14_P.
1954 * config/s390/s390.md: Likewise. Rename also the cpu attribute
1955 value from arch12 to z14.
1957 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
1959 * config/i386/i386.md (fxam<mode>2_i387_with_temp): Remove.
1961 (isinf<mode>2): Ditto.
1963 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
1965 * config/i386/i386.c (ix86_emit_i387_round): Extend op1 to XFmode
1966 before emitting fxam. Perform calculations in XFmode.
1968 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
1970 * match.pd (((X /[ex] A) +- B) * A): New transformation.
1972 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
1975 * fold-const.c (fold_plusminus_mult_expr): Handle complex and vectors.
1976 * tree.c (signed_or_unsigned_type_for): Handle complex.
1978 2018-10-02 Jeff Law <law@redhat.com>
1980 * gimple-fold.c (get_range_strlen): Remove dead code.
1982 2018-10-02 Martin Sebor <msebor@redhat.com>
1983 Jeff Law <law@redhat.com>
1985 * builtins.c (unterminated_array): Add new arguments.
1986 If argument is not terminated, bubble up size and exact
1988 (expand_builtin_strnlen): Detect, avoid expanding
1989 and diagnose unterminated arrays.
1990 (c_strlen): Fill in offset of start of unterminated strings.
1991 * builtins.h (unterminated_array): Update prototype.
1993 2018-10-02 Richard Biener <rguenther@suse.de>
1995 * config/i386/sse.md (reduc_plus_scal_v4df): Avoid the use
1996 of haddv4df, first reduce to SSE width and exploit the fact
1997 that we only need element zero with the reduction result.
1998 (reduc_plus_scal_v2df): Likewise.
2000 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
2002 * dojump.h (do_jump): Delete.
2003 (do_jump_1): Likewise.
2004 (split_comparison): Move around.
2005 * dojump.c (do_jump): Make static.
2006 (do_jump_1): Likewise.
2007 (jumpifnot): Move around.
2008 (jumpifnot_1): Likewise.
2010 (jumpif_1): Likewise.
2011 * expr.c (expand_expr_real_1): Call jumpif[not] instead of do_jump.
2013 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
2015 * reorg.c (make_return_insns): Use emit_copy_of_insn_after for the
2016 insns in the delay slot and add_insn_after for the jump insn.
2018 2018-10-02 Richard Biener <rguenther@suse.de>
2020 * tree-inline.c (expand_call_inline): Use the location of
2021 the callee declaration for the inline-entry marker.
2022 * final.c (notice_source_line): Remove special-casing of
2023 NOTE_INSN_INLINE_ENTRY.
2025 2018-10-01 Carl Love <cel@us.ibm.com>
2028 * config/rs6000/rs6000-builtin.def (__builtin_mffsl): New.
2029 (__builtin_mtfsb0): New.
2030 (__builtin_mtfsb1): New.
2031 ( __builtin_set_fpscr_rn): New.
2032 (__builtin_set_fpscr_drn): New.
2033 * config/rs6000/rs6000.c (rs6000_expand_mtfsb_builtin): Add.
2034 (rs6000_expand_set_fpscr_rn_builtin): Add.
2035 (rs6000_expand_set_fpscr_drn_builtin): Add.
2036 (rs6000_expand_builtin): Add case statement entries for
2037 RS6000_BUILTIN_MTFSB0, RS6000_BUILTIN_MTFSB1,
2038 RS6000_BUILTIN_SET_FPSCR_RN, RS6000_BUILTIN_SET_FPSCR_DRN,
2039 RS6000_BUILTIN_MFFSL.
2040 (rs6000_init_builtins): Add ftype initialization and def_builtin
2041 calls for __builtin_mffsl, __builtin_mtfsb0, __builtin_mtfsb1,
2042 __builtin_set_fpscr_rn, __builtin_set_fpscr_drn.
2043 * config/rs6000.md (rs6000_mtfsb0, rs6000_mtfsb1, rs6000_mffscrn,
2044 rs6000_mffscdrn): Add define_insn.
2045 (rs6000_set_fpscr_rn, rs6000_set_fpscr_drn): Add define_expand.
2046 * doc/extend.texi: Add documentation for the builtins.
2048 2018-10-01 Richard Biener <rguenther@suse.de>
2050 PR tree-optimization/87465
2051 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Fix typo
2052 causing branch miscounts.
2054 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2056 * common/config/aarch64/aarch64-common.c (TARGET_OPTION_DEFAULT_PARAM,
2057 aarch64_option_default_param): New.
2058 (params.h): Include.
2059 (TARGET_OPTION_VALIDATE_PARAM, aarch64_option_validate_param): New.
2060 * config/aarch64/aarch64.c (aarch64_override_options_internal): Simplify
2061 stack-clash protection validation code.
2063 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2065 * params.c (validate_param): New.
2066 (add_params): Use it.
2067 (set_param_value): Refactor param validation into validate_param.
2068 (diagnostic.h): Include.
2069 * diagnostic.h (diagnostic_ready_p): New.
2071 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2073 * params.c (set_param_value):
2074 Add index of parameter being validated.
2075 * common/common-target.def (option_validate_param): New.
2076 * common/common-targhooks.h (default_option_validate_param): New.
2077 * common/common-targhooks.c (default_option_validate_param): New.
2078 * doc/tm.texi.in (TARGET_OPTION_VALIDATE_PARAM): New.
2079 * doc/tm.texi: Regenerate.
2081 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2084 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2085 Add validation for stack-clash parameters and set defaults.
2087 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2090 * configure.ac: Add stack-clash-protection-guard-size.
2091 * doc/install.texi: Document it.
2092 * config.in (DEFAULT_STK_CLASH_GUARD_SIZE): New.
2093 * params.def: Update comment for guard-size.
2094 (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE,
2095 PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Update description.
2096 * configure: Regenerate.
2098 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2101 * config/aarch64/aarch64.h (STACK_CLASH_MIN_BYTES_OUTGOING_ARGS,
2102 STACK_DYNAMIC_OFFSET): New.
2103 * config/aarch64/aarch64.c (aarch64_layout_frame):
2104 Update outgoing args size.
2105 (aarch64_stack_clash_protection_alloca_probe_range,
2106 TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
2108 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2111 * explow.c (anti_adjust_stack_and_probe_stack_clash): Support custom
2113 * target.def (stack_clash_protection_alloca_probe_range): New.
2114 (stack_clash_protection_final_dynamic_probe): Remove.
2115 * targhooks.h (default_stack_clash_protection_alloca_probe_range) New.
2116 (default_stack_clash_protection_final_dynamic_probe): Remove.
2117 * targhooks.c: Likewise.
2118 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
2119 (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE): Remove.
2120 * doc/tm.texi: Regenerate.
2122 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2125 * config/aarch64/aarch64-protos.h (aarch64_output_probe_sve_stack_clash): New.
2126 * config/aarch64/aarch64.c (aarch64_output_probe_sve_stack_clash,
2127 aarch64_clamp_to_uimm12_shift): New.
2128 (aarch64_allocate_and_probe_stack_space): Add SVE specific section.
2129 * config/aarch64/aarch64.md (probe_sve_stack_clash): New.
2131 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2134 * config/aarch64/aarch64.c (aarch64_layout_frame): Add assert.
2136 2018-10-01 Jeff Law <law@redhat.com>
2137 Richard Sandiford <richard.sandiford@linaro.org>
2138 Tamar Christina <tamar.christina@arm.com>
2141 * config/aarch64/aarch64.md
2142 (probe_stack_range): Add k (SP) constraint.
2143 * config/aarch64/aarch64.h (STACK_CLASH_CALLER_GUARD,
2144 STACK_CLASH_MAX_UNROLL_PAGES): New.
2145 * config/aarch64/aarch64.c (aarch64_output_probe_stack_range): Emit
2146 stack probes for stack clash.
2147 (aarch64_allocate_and_probe_stack_space): New.
2148 (aarch64_expand_prologue): Use it.
2149 (aarch64_expand_epilogue): Likewise and update IP regs re-use criteria.
2150 (aarch64_sub_sp): Add emit_move_imm optional param.
2152 2018-10-01 MCC CS <deswurstes@users.noreply.github.com>
2154 PR tree-optimization/87261
2155 * match.pd: Remove trailing whitespace.
2156 Add (x & y) | ~(x | y) -> ~(x ^ y),
2157 (~x | y) ^ (x ^ y) -> x | ~y and (x ^ y) | ~(x | y) -> ~(x & y)
2159 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
2161 * config/arc/arc.md (*add_n): Clean up pattern, update instruction
2163 (ashlsi3_insn): Update instruction constraints.
2164 (ashrsi3_insn): Likewise.
2165 (rotrsi3): Likewise.
2166 (add_shift): Likewise.
2167 * config/arc/constraints.md (Csz): New 32 bit constraint. It
2168 avoids placing in the limm field small constants which, otherwise,
2169 could end into a small instruction.
2171 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
2173 * config/arc/arc.md (maddsidi4_split): Don't use dmac if the
2174 destination register is not odd-even.
2175 (umaddsidi4_split): Likewise.
2177 2018-10-01 Richard Biener <rguenther@suse.de>
2179 * tree-inline.c (expand_call_inline): Store origin of fn
2180 in BLOCK_ABSTRACT_ORIGIN for the inline BLOCK.
2181 * tree.c (block_ultimate_origin): Simplify and do some
2184 2018-09-30 Uros Bizjak <ubizjak@gmail.com>
2186 * config/i386/mmx.md (EMMS): New int iterator.
2187 (emms): New int attribute.
2188 (mmx_<emms>): Macroize insn from *mmx_emms and *mmx_femms using
2189 EMMS int iterator. Explicitly declare clobbers.
2190 (mmx_emms): Remove expander.
2192 * config/i386/predicates.md (emms_operation): Remove predicate.
2193 (vzeroall_pattern): New predicate.
2194 (vzeroupper_pattern): Rename from vzeroupper_operation.
2195 * config/i386/i386.c (ix86_avx_u128_mode_after): Use
2196 vzeroupper_pattern and vzeroall_pattern predicates.
2198 2018-09-30 Peter Bergner <bergner@linux.ibm.com>
2200 PR rtl-optimization/86939
2201 * ira-lives.c (make_hard_regno_born): Rename from this...
2202 (make_hard_regno_live): ... to this. Remove update to conflict
2203 information. Update function comment.
2204 (make_hard_regno_dead): Add conflict information update. Update
2206 (make_object_born): Rename from this...
2207 (make_object_live): ... to this. Remove update to conflict information.
2208 Update function comment.
2209 (make_object_dead): Add conflict information update. Update function
2211 (mark_pseudo_regno_live): Call make_object_live.
2212 (mark_pseudo_regno_subword_live): Likewise.
2213 (mark_hard_reg_dead): Update function comment.
2214 (mark_hard_reg_live): Call make_hard_regno_live.
2215 (process_bb_node_lives): Likewise.
2216 * lra-lives.c (make_hard_regno_born): Rename from this...
2217 (make_hard_regno_live): ... to this. Remove update to conflict
2218 information. Remove now uneeded check_pic_pseudo_p argument.
2219 Update function comment.
2220 (make_hard_regno_dead): Add check_pic_pseudo_p argument and add update
2221 to conflict information. Update function comment.
2222 (mark_pseudo_live): Remove update to conflict information. Update
2224 (mark_pseudo_dead): Add conflict information update.
2225 (mark_regno_live): Call make_hard_regno_live.
2226 (mark_regno_dead): Call make_hard_regno_dead with new arguement.
2227 (process_bb_lives): Call make_hard_regno_live and make_hard_regno_dead.
2229 2018-09-29 H.J. Lu <hongjiu.lu@intel.com>
2232 * config/i386/i386.c (construct_container): Use TImode for
2233 BLKmode values in 2 integer registers.
2235 2018-09-29 Jeff Law <law@redhat.com>
2237 * builtins.c (unterminated_array): Pass in c_strlen_data * to
2238 c_strlen rather than just a tree *.
2239 (c_strlen): Change NONSTR argument to a c_strlen_data pointer.
2240 Update recursive calls appropriately. If caller did not provide a
2241 suitable data pointer, create a local one. When a non-terminated
2242 string is discovered, bubble up information about the string via the
2243 c_strlen_data object.
2244 * builtins.h (c_strlen): Update prototype.
2245 (c_strlen_data): New structure.
2246 * gimple-fold.c (get_range_strlen): Update calls to c_strlen.
2247 For a type 2 call, if c_strlen indicates a non-terminated string
2248 use the length of the non-terminated string.
2249 (gimple_fold_builtin_stpcpy): Update calls to c_strlen.
2251 2018-09-29 Jakub Jelinek <jakub@redhat.com>
2254 * config/i386/avx512fintrin.h (_mm512_abs_pd, _mm512_mask_abs_pd): Use
2255 __m512d type for __A argument rather than __m512.
2257 2018-09-28 John David Anglin <danglin@gcc.gnu.org>
2259 * match.pd (simple_comparison): Don't optimize if either operand is
2260 a function pointer when target needs function pointer canonicalization.
2262 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
2264 * config/rs6000/driver-rs6000.c (asm_names): Adjust the entries for
2265 power5 .. power9 to remove indirection.
2266 * config/rs6000/rs6000.h (ASM_CPU_POWER5_SPEC, ASM_CPU_POWER6_SPEC,
2267 ASM_CPU_POWER7_SPEC, ASM_CPU_POWER8_SPEC, ASM_CPU_POWER9_SPEC,
2268 ASM_CPU_476_SPEC): Delete.
2269 (ASM_CPU_SPEC): Adjust.
2270 (EXTRA_SPECS): Delete asm_cpu_power5, asm_cpu_power6, asm_cpu_power7,
2271 asm_cpu_power8, asm_cpu_power9, asm_cpu_476.
2273 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
2275 * config.in: Delete HAVE_AS_DCI.
2276 * config/powerpcspe/powerpcspe.h: Treat HAVE_AS_DCI as always true.
2277 * config/rs6000/rs6000.h: Ditto.
2278 * configure.ac: Delete HAVE_AS_DCI.
2279 * configure: Regenerate.
2281 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
2283 * config.in (HAVE_AS_LWSYNC): Delete.
2284 * config/powerpcspe/powerpcspe.h (TARGET_LWSYNC_INSTRUCTION): Delete.
2285 * config/powerpcspe/sync.md (*lwsync): Always generate lwsync, never
2287 * config/rs6000/rs6000.h (TARGET_LWSYNC_INSTRUCTION): Delete.
2288 * config/rs6000/sync.md (*lwsync): Always generate lwsync, never do it
2290 * configure.ac: Delete HAVE_AS_LWSYNC.
2291 * configure: Regenerate.
2293 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
2294 Pierre-Marie de Rodat <derodat@adacore.com>
2296 * calls.c (expand_call): Try to do a tail call for thunks at -O0 too.
2297 * cgraph.h (struct cgraph_thunk_info): Add indirect_offset.
2298 (cgraph_node::create_thunk): Add indirect_offset parameter.
2299 (thunk_adjust): Likewise.
2300 * cgraph.c (cgraph_node::create_thunk): Add indirect_offset parameter
2301 and initialize the corresponding field with it.
2302 (cgraph_node::dump): Dump indirect_offset field.
2303 * cgraphclones.c (duplicate_thunk_for_node): Deal with indirect_offset.
2304 * cgraphunit.c (cgraph_node::analyze): Be prepared for external thunks.
2305 (thunk_adjust): Add indirect_offset parameter and deal with it.
2306 (cgraph_node::expand_thunk): Deal with the indirect_offset field and
2307 pass it to thunk_adjust. Do not call the target hook if it's non-zero
2308 or if the thunk is external or local. Fix formatting. Do not chain
2309 the RESULT_DECL to BLOCK_VARS. Pass the static chain to the target,
2310 if any, in the GIMPLE representation.
2311 * ipa-icf.c (sem_function::equals_wpa): Deal with indirect_offset.
2312 * lto-cgraph.c (lto_output_node): Write indirect_offset field.
2313 (input_node): Read indirect_offset field.
2314 * tree-inline.c (expand_call_inline): Pass indirect_offset field in the
2315 call to thunk_adjust.
2316 * tree-nested.c (struct nesting_info): Add thunk_p field.
2317 (create_nesting_tree): Set it.
2318 (convert_all_function_calls): Copy static chain from targets to thunks.
2319 (finalize_nesting_tree_1): Return early for thunks.
2320 (unnest_nesting_tree_1): Do not finalize thunks.
2321 (gimplify_all_functions): Do not gimplify thunks.
2323 2018-09-28 David Malcolm <dmalcolm@redhat.com>
2325 * opt-suggestions.c (option_proposer::build_option_suggestions):
2326 Release "option_values".
2328 2018-09-28 David Malcolm <dmalcolm@redhat.com>
2330 * coverage.c (get_coverage_counts): Convert problem-reporting dump
2331 messages from MSG_OPTIMIZED_LOCATIONS to MSG_MISSED_OPTIMIZATION.
2332 * dumpfile.c (kind_as_string): New function.
2333 (dump_loc): Rather than a hardcoded prefix of "note: ", use
2334 kind_as_string to vary the prefix based on dump_kind.
2335 (selftest::test_capture_of_dump_calls): Update for above.
2337 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
2339 * config/i386/i386.h (SSE_REGNO): Fix check for FIRST_REX_SSE_REG.
2340 (GET_SSE_REGNO): Rename from SSE_REGNO. Update all uses for rename.
2342 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
2344 * config/i386/i386.h (CC_REGNO): Remove FPSR_REG.
2345 * config/i386/i386.c (ix86_fixed_condition_code_regs): Use
2346 INVALID_REGNUM instead of FPSR_REG.
2347 (ix86_md_asm_adjust): Do not clobber FPSR_REG.
2348 * config/i386/i386.md: Update comment of FP compares.
2349 (fldenv): Do not clobber FPSR_REG.
2351 2018-09-28 Richard Biener <rguenther@suse.de>
2353 * tree.h (BLOCK_ORIGIN): New.
2354 * omp-expand.c (grid_expand_target_grid_body): Assign
2355 BLOCK_ORIGIN to BLOCK_ABSTRACT_ORIGIN.
2356 * tree-inline.c (remap_block): Likewise.
2357 * auto-profile.c (get_function_decl_from_block): Simplify
2358 by eliding the BLOCK_ABSTRACT_ORIGIN chasing.
2359 * langhooks.c (lhd_print_error_function): Likewise.
2360 * optinfo-emit-json.cc (optrecord_json_writer::inlining_chain_to):
2362 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
2363 * tree.c (block_nonartificial_location): Likewise.
2364 (block_ultimate_origin): Likewise.
2365 * tree-pretty-print.c (percent_K_format): Likewise. Remove
2366 no longer needed LTO case.
2368 2018-09-28 Andrew Stubbs <ams@codesourcery.com>
2369 Jan Hubicka <jh@suse.cz>
2370 Martin Jambor <mjambor@suse.cz>
2372 * simplify-rtx.c (simplify_merge_mask): New function.
2373 (simplify_ternary_operation): Use it, also see if VEC_MERGEs with the
2374 same masks are used in op1 or op2.
2375 (test_vec_merge): New function.
2376 (test_vector_ops): Call test_vec_merge.
2378 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
2380 * config/sparc/sparc-protos.h (sparc_branch_cost): Declare.
2381 * config/sparc/sparc.h (BRANCH_COST): Call sparc_branch_cost.
2382 * config/sparc/sparc.c (struct processor_costs): Add branch_cost field.
2383 (cypress_costs): Set it.
2384 (supersparc_costs): Likewise.
2385 (hypersparc_costs): Likewise.
2386 (leon_cost): Likewise.
2387 (leon3_costs): Likewise.
2388 (sparclet_costs): Likewise.
2389 (ultrasparc_costs): Likewise.
2390 (ultrasparc_costs): Likewise.
2391 (niagara_costs): Likewise.
2392 (niagara2_costs): Likewise.
2393 (niagara3_costs): Likewise.
2394 (niagara4_costs): Likewise.
2395 (niagara7_costs): Likewise.
2396 (m8_costs): Likewise.
2397 (TARGET_CAN_FOLLOW_JUMP): Define.
2398 (pass_work_around_errata::gate): Minor tweak.
2399 (sparc_option_override): Remove MASK_FSMULD mask for V7 processors.
2400 Do not set both MASK_VIS4 and MASK_VIS4B for M8 processor.
2401 Automaitcally clear MASK_FSMULD mask for V7 processors.
2402 (sparc_can_follow_jump): New static function.
2403 (output_ubranch): Deal with CROSSING_JUMP_P.
2404 (sparc_use_sched_lookahead): Rewrite using switch statement.
2405 (sparc_issue_rate): Reorder.
2406 (sparc_branch_cost): New function.
2408 2018-09-27 Martin Sebor <msebor@redhat.com>
2410 * tree.h (tree_to_shwi): Add attribute nonnull and pure.
2411 (tree_to_poly_int64, tree_to_uhwi, tree_to_poly_uint64): Same.
2412 (int_fits_type_p): Same.
2414 2018-09-27 Uros Bizjak <ubizjak@gmail.com>
2416 * config/i386/i386.md (FPCR_REG): Remove.
2417 (UNSPEC_FLDCW): Remove.
2418 (x86_fnstcw_1): Use (const_int 0) instead of FPCR_REG.
2419 (x86_fldcw_1): Remove insn pattern.
2420 (fnstenv): Do not clobber FPCR_REG.
2422 * config/i386/i386.h (FIXED_REGISTERS) Remove fpsr register.
2423 (CALL_USED_REGISTERS): Ditto.
2424 (REG_ALLOC_ORDER): Ditto.
2425 (REG_CLASS_CONTENTS): Ditto.
2426 (HI_REGISTER_NAMES): Ditto.
2427 (ADDITIONAL_REGISTER_NAMES): Use defines instead
2428 of numerical constants.
2429 * config/i386/i386.c (regclass_map): Remove fpsr register.
2430 (dbx_register_map): Ditto.
2431 (dbx64_register_map): Ditto.
2432 (svr4_dbx_register_map): Ditto.
2433 (print_reg): Do not handle FPCR_REG.
2435 2018-09-27 Segher Boessenkool <segher@kernel.crashing.org>
2438 * config.in (HAVE_AS_CMPB, HAVE_AS_DFP, HAVE_AS_FPRND, HAVE_AS_MFPGPR,
2439 HAVE_AS_POPCNTB, HAVE_AS_POPCNTD, HAVE_AS_POWER8, HAVE_AS_POWER9):
2440 Delete, always treat as true.
2441 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
2442 Ditto. Simplify remaining code.
2443 * config/powerpcspe/powerpcspe.h: Ditto.
2444 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ditto.
2445 Simplify remaining code.
2446 (rs6000_expand_builtin): Ditto.
2447 * config/rs6000/rs6000.h: Ditto.
2448 * configure.ac: Ditto.
2449 * configure: Regenerate.
2451 2018-09-27 Martin Liska <mliska@suse.cz>
2453 * coverage.c (get_coverage_counts): Revert the formatting
2454 of missing profile opt info.
2456 2018-09-27 Richard Biener <rguenther@suse.de>
2460 * dwarf2out.c (set_block_origin_self): Do not mark outermost
2461 block as we do not output that.
2462 (gen_inlined_subroutine_die): Elide the originally outermost
2463 block, matching what we do for concrete instances.
2464 (decls_for_scope): Add parameter specifying whether to recurse
2467 2018-09-27 Andrew Stubbs <ams@codesourcery.com>
2468 Tom de Vries <tom@codesourcery.com>
2472 * expmed.c (emit_cstore): Fix handling of result_mode == BImode and
2473 STORE_FLAG_VALUE == 1.
2475 2018-09-27 Andreas Krebbel <krebbel@linux.ibm.com>
2477 * config/s390/s390.md (PPA_TX_ABORT, PPA_OOO_BARRIER): New
2478 constant definitions.
2479 ("tx_assist"): Replace magic number with PPA_TX_ABORT.
2480 ("*ppa"): Enable pattern also for -march=zEC12 -mno-htm.
2481 ("speculation_barrier"): New expander definition.
2483 2018-09-26 Indu Bhagat <indu.bhagat@oracle.com>
2485 PR gcov-profile/86957
2486 * common.opt: New warning option -Wmissing-profile.
2487 * coverage.c (get_coverage_counts): Add warning for missing .gcda file.
2488 * doc/invoke.texi: Document -Wmissing-profile.
2490 2018-09-26 Jim Wilson <jimw@sifive.com>
2492 * config/riscv/riscv.md (subsi3_extended2): Add J constraint.
2493 (negdi2, negsi2, negsi2_extended, negsi2_extended2): New.
2495 2018-09-26 Martin Sebor <msebor@redhat.com>
2497 * tree.c (zerop): Change return type to bool.
2498 (integer_zerop, integer_onep, integer_each_onep): Same.
2499 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
2500 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
2501 (real_onep, real_minus_onep, chain_index): Same.
2502 (print_type_hash_statistics, type_list_equal): Same.
2503 * tree.h (zerop): Same.
2504 (zerop, integer_zerop, integer_onep, integer_each_onep): Same.
2505 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
2506 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
2507 (real_onep, real_minus_onep, chain_index): Same.
2508 (print_type_hash_statistics, type_list_equal): Same.
2510 2018-09-26 Jim Wilson <jimw@sifive.com>
2512 * config/riscv/riscv.h (FUNCTION_ARG_REGNO_P): Fix comment.
2514 2018-09-26 Jakub Jelinek <jakub@redhat.com>
2517 * config/i386/i386.c: Include debug.h and dwarf2out.h.
2518 (output_indirect_thunk): Emit DW_CFA_def_cfa_offset after the
2521 2018-09-25 Andrew Stubbs <ams@codesourcery.com>
2523 * builtins.c (get_builtin_sync_mem): Force address mode conversion.
2525 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
2527 * config/i386/i386.h (enum reg_class): Remove FP_TOP_SSE_REGS
2528 and FP_SECOND_SSE_REGS.
2529 (REG_CLASS_NAMES): Ditto.
2530 (REG_CLASS_CONTENTS): Ditto.
2531 * config/i386/i386.c (ix86_preferred_reload_class) Do not handle
2532 FP_TOP_SSE_REGS and FP_SECOND_SSE_REGS classes.
2533 (ix86_preferred_output_reload_class): Ditto.
2534 * config/i386/i386.md (fix_trunc<mode>_i387_fisttp): Change "=&1f"
2535 clobber constraint to "=&f".
2536 (fix_truncdi_i387): Ditto.
2537 (lrintxfdi2): Ditto.
2538 (fistdi2_<rounding>): Ditto.
2539 (fpremxf4_i387): Change "=u" constraint to "=f".
2540 (fprem1xf4_i387): Ditto.
2542 (fptanxf4_i387): Ditto.
2543 (fxtractxf3_i387): Ditto.
2544 (fscalexf4_i387): Ditto.
2545 (atan2xf3): Change "u" constraint to "f".
2546 (fyl2xxf3_i387): Ditto.
2547 (fyl2xp1xf3_i387): Ditto.
2549 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
2552 * config/i386/i386.h (NUM_MODES_FOR_MODE_SWITCHING): Update
2553 for removed I387_MASK_PM entity.
2556 2018-09-26 Jeff Law <law@redhat.com>
2558 2018-09-26 Alexey Neyman <stilor@att.net>
2560 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
2561 headers are no longer pulled in by <isl/val.h>.
2563 2018-09-26 Richard Biener <rguenther@suse.de>
2566 * dwarf2out.c (gen_lexical_block_die): Do not equate inline
2567 or concrete instance DIE to the tree. Create abstract origin
2568 attributes also for concrete instances.
2570 2018-09-26 Alexey Neyman <stilor@att.net>
2572 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
2573 headers are no longer pulled in by <isl/val.h>.
2575 2018-09-26 Matthew Malcomson <matthew.malcomson@arm.com>
2577 * config/arm/arm.c (arm_split_compare_and_swap, arm_split_atomic_op):
2578 Use new helper functions.
2579 * config/arm/sync.md (atomic_load<mode>, atomic_store<mode>):
2580 Use new helper functions.
2581 * config/arm/aarch-common-protos.h (aarch_mm_needs_acquire,
2582 aarch_mm_needs_release): New declarations.
2583 * config/arm/aarch-common.c (aarch_mm_needs_acquire,
2584 aarch_mm_needs_release): New.
2586 2018-09-26 Eric Botcazou <ebotcazou@adacore.com>
2588 * config/arm/arm.c (arm_reorg): Skip Thumb reorg pass for thunks.
2589 (arm32_output_mi_thunk): Deal with long calls.
2591 2018-09-26 Richard Biener <rguenther@suse.de>
2595 * tree-inline.c (expand_call_inline): When the location
2596 of the call is UNKNOWN_LOCATION use DECL_SOURCE_LOCATION
2597 or BUILTINS_LOCATION for the BLOCK_SOURCE_LOCATION of
2598 the inserted BLOCK to make inlined_function_outer_scope_p
2600 * dwarf2out.c (add_call_src_coords_attributes): Do not add
2601 coords for reserved locations.
2603 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
2605 * config/rs6000/rs6000.md (*movcc_internal1): Use set_attr_alternative.
2606 (*call_indirect_nonlocal_sysv<mode>): Ditto.
2607 (*call_value_indirect_nonlocal_sysv<mode>): Ditto.
2608 (*sibcall_nonlocal_sysv<mode>): Ditto.
2609 (*sibcall_value_nonlocal_sysv<mode>): Ditto.
2610 (<bd>_<mode>): Ditto.
2611 (<bd>tf_<mode>): Ditto.
2613 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
2615 * config/rs6000/altivec.md (*altivec_mov<mode>): Write the output
2616 control string as a list of templates instead of as C code.
2617 (*altivec_movti): Ditto.
2618 * config/rs6000/darwin.md (movdf_low_di): Ditto.
2620 2018-09-25 Jim Wilson <jimw@sifive.com>
2622 * config/riscv/riscv.c (riscv_split_symbol): Mark auipc label as weak
2623 when target symbol is weak.
2625 2018-09-25 Bernd Edlinger <bernd.edlinger@hotmail.de>
2628 * builtins.c (unterminated_array): Simplify.
2629 * expr.c (string_constant): Handle SSA_NAME. Add more exceptions
2630 where pointer arithmetic is safe.
2632 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
2635 * config/rs6000/altivec.md (altivec_vspltb): Use
2636 const_0_to_15_operand instead of u5bit_cint_operand.
2637 (*altivec_vspltb_internal): Ditto.
2638 (altivec_vspltb_direct): Ditto.
2639 (altivec_vsplth): Use const_0_to_7_operand instead of
2641 (*altivec_vsplth_internal): Ditto.
2642 (altivec_vsplth_direct): Ditto.
2643 (altivec_vspltw): Use const_0_to_3_operand instead of
2645 (*altivec_vspltw_internal): Ditto.
2646 (altivec_vspltw_direct): Ditto.
2647 (altivec_vspltsf): Ditto.
2648 (*altivec_vspltsf_internal): Ditto.
2649 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Handle the
2650 various splats with the proper size immediate. Reorder the various
2651 cases by ascending size of immediate, and put all such together.
2653 2018-09-25 Richard Biener <rguenther@suse.de>
2656 * dwarf2out.c (add_AT_external_die_ref): Remove now redundant
2658 (maybe_create_die_with_external_ref): Do not create
2659 DW_TAG_imported_unit here.
2660 (add_abstract_origin_attribute): Handle external BLOCK refs.
2661 (dwarf2out_abstract_function): Simplify LTO case.
2662 (dwarf2out_early_finish): Create DW_TAG_imported_unit explicitely
2663 rather than using maybe_create_die_with_external_ref.
2665 2018-09-25 Uros Bizjak <ubizjak@gmail.com>
2668 * config/i386/i386.md (frndintxf2_mask_pm): Remove.
2669 (frndintxf2_mask_pm_i387): Ditto.
2670 (nearbyintxf2): Rewrite expander pattern to match rintxf2.
2671 Enable for !flag_trapping_math.
2672 (nearbyint<mode>2): Enable x87 modes for !flag_trapping_math.
2673 Enable SSE modes for TARGET_SSE4_1 and expand them with round insn.
2674 Change operand 1 predicate to nonimmediate_operand.
2675 (attr "i387_cw"): Remove mask_pm.
2676 * config/i386/i386.h (enum ix86_stack_slot): Remove SLOT_CW_MASK_PM.
2677 (enum ix86_entity): Remove I387_MASK_PM.
2678 * config/i386/i386.c (ix86_i387_mode_needed): Do not
2679 handle I387_MASK_PM.
2680 (ix86_mode_needed): Ditto.
2681 (ix86_mode_after): Ditto.
2682 (ix86_mode_entry): Ditto.
2683 (ix86_mode_exit): Ditto.
2684 (emit_i387_cw_initialization): Do not handle I387_CW_MASK_PM.
2686 2018-09-25 Jakub Jelinek <jakub@redhat.com>
2688 * vr-values.c (vr_values::vr_values): Initialize to_remove_edges and
2689 to_update_switch_stmts to vNULL instead of calling create on them
2692 2018-09-25 Richard Biener <rguenther@suse.de>
2694 PR tree-optimization/87402
2695 * tree-ssa-sccvn.c (SSA_VISITED): Remove unused function.
2696 (visit_phi): Re-instantiate handling of supposed to be VARYING
2697 but non-VARYING backedge value.
2699 2018-09-25 Richard Biener <rguenther@suse.de>
2702 * dwarf2out.c (struct sym_off_pair): New.
2703 (external_die_map): New global.
2704 (lookup_decl_die): When in LTO create DIEs lazily from the
2706 (lookup_block_die): New function, create DIEs lazily in LTO.
2707 (equate_block_to_die): New function.
2708 (dwarf2out_die_ref_for_decl): During WPA get the association
2709 from the external DIE map.
2710 (dwarf2out_register_external_die): Record mapping into the
2712 (maybe_create_die_with_external_ref): New function split out from
2713 DIE generation part of old dwarf2out_register_external_die.
2714 (add_abstract_origin_attribute): Do not return the DIE. When
2715 in LTO reference externals directly.
2716 (dwarf2out_abstract_function): When in LTO ignore calls for
2717 decls with external DIEs (already present abstract instances).
2718 (gen_call_site_die): Adjust.
2719 (add_high_low_attributes): Likewise.
2720 (gen_lexical_block_die): Likewise.
2721 (gen_inlined_subroutine_die): Likewie.
2722 (gen_block_die): Likewise.
2723 (dwarf2out_inline_entry): Likewise.
2724 (dwarf2out_early_finish): In LTRANS phase create DW_TAG_imported_unit
2727 2018-09-25 Martin Liska <mliska@suse.cz>
2729 * ipa-fnsummary.c (estimate_node_size_and_time): Scale by two
2730 integers and not by a float value.
2732 2018-09-25 Martin Liska <mliska@suse.cz>
2735 * dbgcnt.c (dbg_cnt_process_single_pair): Return false
2737 * dumpfile.c (dump_enable_all): Remove extra parenthesis.
2738 * gcov-tool.c: Declare the function with ATTRIBUTE_NORETURN.
2739 * godump.c (go_format_type): Remove extra parenthesis.
2741 2018-09-25 Martin Liska <mliska@suse.cz>
2743 * alias.c (set_dest_equal_p): Remove unused function.
2744 * config/i386/i386.c (def_builtin_pure2): Likewise.
2745 * diagnostic-show-locus.c (class layout): Remove
2747 (layout::layout): Likewise here.
2748 * dump-context.h (class temp_dump_context): Likewise.
2749 * dwarf2out.c (add_AT_fde_ref): Remove unused function.
2750 (add_AT_loclistsptr): Likewise.
2751 (add_AT_offset): Likewise.
2752 (get_AT_hi_pc): Likewise.
2753 (is_comdat_die): Likewise.
2754 (type_is_enum): Likewise.
2755 (ceiling): Likewise.
2756 (add_AT_vms_delta): Likewise.
2757 (is_class_die): Likewise.
2758 * edit-context.c (class line_event): Remove unused field.
2759 * graphite-sese-to-poly.c (tree_int_to_gmp): Remove
2761 * ipa-cp.c (ipa_get_vr_lat): Likewise.
2762 * lra-constraints.c (ok_for_index_p_nonstrict): Likewise.
2763 (ok_for_base_p_nonstrict): Likewise.
2764 * tree-chrec.c (is_not_constant_evolution): Likewise.
2765 (chrec_fold_poly_cst): Likewise.
2766 * tree-if-conv.c (has_pred_critical_p): Likewise.
2767 * tree-ssa-coalesce.c (print_exprs): Likewise.
2768 * tree-ssa-pre.c (bitmap_set_contains_expr): Likewise.
2769 * tree-ssa-uninit.c (is_and_or_or_p): Likewise.
2770 * tree-vrp.c (value_ranges_intersect_p): Likewise.
2771 (value_range_nonnegative_p): Likewise.
2773 2018-09-25 Martin Liska <mliska@suse.cz>
2775 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
2776 Do not handle "GNU Pascal".
2777 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
2779 * config/sparc/sparc.c (sparc_pass_by_reference): Remove Pascal
2780 from documentation. Likewise.
2781 * dbxout.c (dbxout_range_type): Likewise.
2782 * doc/cpp.texi: Likewise.
2783 * doc/extend.texi: Likewise.
2784 * doc/frontends.texi: Likewise.
2785 * doc/invoke.texi: Remove Pascal entry.
2786 * tree.def (CLEANUP_POINT_EXPR): Likewise.
2787 * doc/rtl.texi (MODE_FUNCTION): Remove not used entry.
2789 2018-09-25 Martin Liska <mliska@suse.cz>
2792 * doc/invoke.texi: Document all parameters and remove default
2795 2018-09-25 Ilya Leoshkevich <iii@linux.ibm.com>
2798 * rtl.c (rtx_code_size): Take into account that EQ_ATTR_ALT
2799 contains HOST_WIDE_INTs when computing its size.
2801 2018-09-24 Jim Wilson <jimw@sifive.com>
2804 * config/riscv/riscv.h (STACK_BOUNDARY): Test riscv_abi == ABI_ILP32E
2806 (ABI_STACK_BOUNDARY, MAX_ARGS_IN_REGISTERS): Likewise.
2808 2018-09-24 Andrew Pinski <apinski@marvell.com>
2810 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Don't
2811 access prev before checking it for NULLness in the
2812 AARCH64_FUSE_CMP_BRANCH case.
2814 2018-09-24 H.J. Lu <hongjiu.lu@intel.com>
2817 * config/i386/i386.c (rest_of_insert_endbranch): Set
2818 endbr_queued_at_entrance to true and don't insert ENDBR if
2819 x86_function_profiler will be called.
2820 (x86_function_profiler): Insert ENDBR if endbr_queued_at_entrance
2822 * config/i386/i386.h (machine_function): Add
2823 endbr_queued_at_entrance.
2825 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
2827 * genattrtab.c (mk_attr_alt): Use alternative_mask.
2828 (attr_rtx_1): Adjust caching to match the new EQ_ATTR_ALT field
2830 (check_attr_test): Use alternative_mask.
2831 (get_attr_value): Likewise.
2832 (compute_alternative_mask): Use alternative_mask and XWINT.
2833 (make_alternative_compare): Use alternative_mask.
2834 (attr_alt_subset_p): Use XWINT.
2835 (attr_alt_subset_of_compl_p): Likewise.
2836 (attr_alt_intersection): Use alternative_mask and XWINT.
2837 (attr_alt_union): Likewise.
2838 (attr_alt_complement): Use HOST_WIDE_INT and XWINT.
2839 (mk_attr_alt): Use alternative_mask and HOST_WIDE_INT.
2840 (simplify_test_exp): Use alternative_mask and XWINT.
2841 (write_test_expr): Use alternative_mask and XWINT, adjust bit
2842 number calculation to support 64 bits. Generate code that
2843 checks 64-bit masks.
2844 (main): Use alternative_mask.
2845 * rtl.def (EQ_ATTR_ALT): Change field types from ii to ww.
2847 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
2850 * config/s390/s390.c (s390_emit_epilogue): Do not use PARALLEL
2851 RETURN+USE when returning via %r14.
2853 2018-09-24 Martin Liska <mliska@suse.cz>
2855 * gcov.c (output_lines): Print colorization legend
2856 for both flag_use_colors and flag_use_hotness_colors.
2859 2018-09-24 Martin Liska <mliska@suse.cz>
2861 * coverage.c (get_coverage_counts): Use warning_at
2862 with current_function_decl location. Use %qD in warning
2865 2018-09-24 Martin Liska <mliska@suse.cz>
2867 * memory-block.h (memory_block_pool::release): Annotate with
2868 valgrind that the memory is not accessible.
2870 2018-09-24 Martin Liska <mliska@suse.cz>
2873 * asan.c: Make asan_handled_variables extern.
2875 * cfgexpand.c (expand_stack_vars): Make sure
2876 a representative is unpoison if another
2877 variable in the partition is handled by
2878 use-after-scope sanitization.
2880 2018-09-24 Richard Biener <rguenther@suse.de>
2882 PR tree-optimization/63155
2883 * tree-ssa-propagate.c (add_ssa_edge): Avoid adding PHIs to
2884 the worklist when the edge of the respective argument isn't
2887 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
2889 * config/i386/i386.h (enum reg_class): Rename MASK_REGS to
2890 ALL_MASK_REGS and MASK_EVEX_REGS to MASK_REGS.
2891 (MASK_CLASS_P): Update for rename.
2892 (MAYBE_MASK_CLASS_P): Ditto.
2893 (REG_CLASS_NAMES): Update.
2894 (REG_CLASS_CONTENT): Update.
2895 * config/i386/i386.c (regclass_map): Update for MASK_REG
2896 and ALL_MASK_REGS rename.
2897 * config/i386/constraints.md (Yk): Update for rename.
2900 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
2902 * config/i386/i386.h (enum reg_class): Remove
2903 EVEX_SSE_REGS and MOD4_SSE_REGS.
2904 (REG_CLASS_NAMES): Update.
2905 (REG_CLASS_CONTENT): Update.
2906 * config/i386/i386.c (regclass_map): Declare AVX-512 SSE
2907 registers as ALL_SSE_REGS.
2908 (ix86_additional_allocno_class_p): Remove.
2909 (TARGET_ADDITIONAL_ALLOCNO_CLASS_P): Remove.
2910 (ix86_register_priority): Lower priority of EVEX SSE registers.
2911 Use IN_RANGE macro where appropriate.
2912 (ix86_hard_regno_mode_ok): Merge AVX-5124FMAPS and
2913 AVX-5124VNNIW checks.
2914 (ix86_modes_tieable_p): Tie 512-bit SSE modes.
2915 * config/i386/sse.md (avx5124fmaddps_4fmaddps)
2916 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddps_maskz)
2917 (avx5124fmaddps_4fmaddss, avx5124fmaddps_4fmaddss_mask)
2918 (avx5124fmaddps_4fmaddss_maskz, avx5124fmaddps_4fnmaddps)
2919 (avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddps_maskz)
2920 (avx5124fmaddps_4fnmaddss, avx5124fmaddps_4fnmaddss_mask)
2921 (avx5124fmaddps_4fnmaddss_maskz, avx5124vnniw_vp4dpwssd)
2922 (avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssd_maskz)
2923 (avx5124vnniw_vp4dpwssds, avx5124vnniw_vp4dpwssds_mask)
2924 (avx5124vnniw_vp4dpwssds_maskz): Use "v" instead of "Yh" constraint.
2925 * config/i386/constraints.md (Yh): Remove.
2927 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
2929 * config/i386/i386.c (regclass_map): Declare integer REX registers
2932 2018-09-23 Gerald Pfeifer <gerald@pfeifer.com>
2934 * doc/service.texi (Service): Switch the fsf.org link to https.
2936 2018-09-22 Chung-Ju Wu <jasonwucj@gmail.com>
2939 * config/nds32/nds32.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
2940 Define to speculation_safe_value_not_needed.
2942 2018-09-21 Florian Weimer <fweimer@redhat.com>
2945 * doc/extend.texi (Common Function Attributes): Mention that
2946 noreturn suppresses tail call optimization.
2948 2018-09-21 Jeff Law <law@redhat.com>
2950 * gimple-ssa-evrp.c (evrp_dom_walker::cleanup): Call
2951 vr_values::cleanup_edges_and_switches.
2952 * tree-vrp.c (to_remove_edges, to_update_switch_stmts): Moved into
2954 (identify_jump_threads): Remove EDGE_IGNORE handling.
2955 (execute_vrp): Move handling of to_remove_edges and
2956 to_update_switch_stmts into vr_values class member functions.
2957 * tree-vrp.h (switch_update, to_remove_edges): Remove declarations.
2958 (to_update_switch_stmts): Likewise.
2959 * vr-values.c: Include cfghooks.h.
2960 (vr_values::vr_values): Initialize to_remove_edges and
2961 to_update_switch_stmts.
2962 (vr_values::~vr_values): Verify to_remove_edges and
2963 to_update_switch_stmts are empty.
2964 (vr_values::simplify_switch_using_ranges): Set EDGE_IGNORE as needed.
2965 (vr_values::cleanup_edges_and_switches): New member function.
2966 * vr-values.h (vr_values): Add cleanup_edges_and_switches member
2967 function. Add new data members.
2969 2018-09-21 David Malcolm <dmalcolm@redhat.com>
2971 PR tree-optimization/87309
2972 * dumpfile.c (dump_context::begin_scope): Filter the dump_loc
2973 calls with pflags and alt_flags.
2974 (selftest::test_capture_of_dump_calls): Add test of interaction of
2975 MSG_OPTIMIZED_LOCATIONS with AUTO_DUMP_SCOPE.
2977 2018-09-21 Olivier Hainque <hainque@adacore.com>
2979 * config.gcc: Factorize and comment inclusion of vxworks-dummy.h.
2981 2018-09-21 Olivier Hainque <hainque@adacore.com>
2983 * config/vxworks.h (CLEAR_INSN_CACHE): #define to 1.
2985 2018-09-21 Olivier Hainque <hainque@adacore.com>
2987 * config/vxworks.h (VXWORKS_LIBGCC_SPEC): Remove -lc_internal.
2988 Merge block comment with the one ahead of VXWORKS_LIBS_RTP. Then:
2989 (VXWORKS_LIBS_RTP): Minor reordering.
2991 2018-09-21 Olivier Hainque <hainque@adacore.com>
2993 * config/vxworks.h (STARTFILE_PREFIX_SPEC): Define.
2994 (VXWORKS_LIBS_DIR_RTP): Remove definition and use.
2996 2018-09-21 Olivier Hainque <hainque@adacore.com>
2998 * config/vxworks.h (SIZE_TYPE): Account for TARGET_VXWORKS64.
2999 (PTRDIFF_TYPE): Likewise.
3001 2018-09-21 Olivier Hainque <hainque@adacore.com>
3003 * config.gcc: Enforce def of TARGET_VXWORKS64 to 1 from
3004 triplet, similar to support for VxWorks7.
3005 * config/vxworks-dummy.h: Provide a default definition
3006 of TARGET_VXWORKS64 to 0.
3008 2018-09-21 Olivier Hainque <hainque@adacore.com>
3010 * config/vxworks.h (TARGET_VXWORKS7): Move default definition ...
3011 * config/vxworks-dummy.h: here.
3013 2018-09-21 Olivier Hainque <hainque@adacore.com>
3015 * config.gcc: Prepend vxworks-dummy.h to tm_file for powerpc*
3017 2018-09-21 Shaokun Zhang <zhangshaokun@hisilicon.com>
3018 Bo Zhou <zbo.zhou@hisilicon.com>
3020 * config/aarch64/aarch64-cores.def (tsv110): New CPU.
3021 * config/aarch64/aarch64-tune.md: Regenerated.
3022 * doc/invoke.texi (AArch64 Options/-mtune): Add "tsv110".
3023 * config/aarch64/aarch64.c (tsv110_tunings): New tuning table.
3024 * config/aarch64/aarch64-cost-tables.h: Add "tsv110" extra costs.
3026 2018-09-21 Andrew Stubbs <ams@codesourcery.com>
3027 Julian Brown <julian@codesourcery.com>
3029 * builtins.c (get_builtin_sync_mem): Handle address spaces.
3031 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
3033 * config/rs6000/rs6000.c (rs6000_function_ok_for_sibcall): Return false
3034 if the call takes a static chain.
3036 2018-09-21 Martin Liska <mliska@suse.cz>
3038 * auto-profile.c (autofdo_source_profile::read): Do not
3040 (read_profile): Do not add working sets.
3041 (read_autofdo_file): Remove sum_all.
3042 (afdo_callsite_hot_enough_for_early_inline): Remove const
3044 * coverage.c (struct counts_entry): Remove gcov_summary.
3045 (read_counts_file): Read new GCOV_TAG_OBJECT_SUMMARY,
3046 do not support GCOV_TAG_PROGRAM_SUMMARY.
3047 (get_coverage_counts): Remove summary and expected
3049 * coverage.h (get_coverage_counts): Likewise.
3050 * doc/gcov-dump.texi: Remove -w option.
3051 * gcov-dump.c (dump_working_sets): Remove.
3052 (main): Do not support '-w' option.
3053 (print_usage): Likewise.
3054 (tag_summary): Likewise.
3055 * gcov-io.c (gcov_write_summary): Do not dump
3057 (gcov_read_summary): Likewise.
3058 (gcov_histo_index): Remove.
3059 (gcov_histogram_merge): Likewise.
3060 (compute_working_sets): Likewise.
3061 * gcov-io.h (GCOV_TAG_OBJECT_SUMMARY): Mark
3063 (GCOV_TAG_PROGRAM_SUMMARY): Mark it obsolete.
3064 (GCOV_TAG_SUMMARY_LENGTH): Adjust.
3065 (GCOV_HISTOGRAM_SIZE): Remove.
3066 (GCOV_HISTOGRAM_BITVECTOR_SIZE): Likewise.
3067 (struct gcov_summary): Simplify rapidly just
3068 to runs and sum_max fields.
3069 (gcov_histo_index): Remove.
3070 (NUM_GCOV_WORKING_SETS): Likewise.
3071 (compute_working_sets): Likewise.
3072 * gcov-tool.c (print_overlap_usage_message): Remove
3073 trailing empty line.
3074 * gcov.c (read_count_file): Read GCOV_TAG_OBJECT_SUMMARY.
3075 (output_lines): Remove program related line.
3076 * ipa-profile.c (ipa_profile): Do not consider GCOV histogram.
3077 * lto-cgraph.c (output_profile_summary): Do not stream GCOV
3079 (input_profile_summary): Do not read it.
3080 (merge_profile_summaries): And do not merge it.
3081 (input_symtab): Do not call removed function.
3082 * modulo-sched.c (sms_schedule): Do not print sum_max.
3083 * params.def (HOT_BB_COUNT_FRACTION): Reincarnate param that was
3084 removed when histogram method was invented.
3085 (HOT_BB_COUNT_WS_PERMILLE): Mention that it's used only in LTO
3087 * postreload-gcse.c (eliminate_partially_redundant_load): Fix
3089 * predict.c (get_hot_bb_threshold): Use HOT_BB_COUNT_FRACTION
3090 and dump selected value.
3091 * profile.c (add_working_set): Remove.
3092 (get_working_sets): Likewise.
3093 (find_working_set): Likewise.
3094 (get_exec_counts): Do not work with working sets.
3095 (read_profile_edge_counts): Do not inform as sum_max is removed.
3096 (compute_branch_probabilities): Likewise.
3097 (compute_value_histograms): Remove argument for call of
3098 get_coverage_counts.
3099 * profile.h: Do not make gcov_summary const.
3101 2018-09-21 Monk Chiang <sh.chiang04@gmail.com>
3103 * config.gcc (nds32*-*-*): Set TARGET_DEFAULT_TLSDESC_TRAMPOLINE=0.
3105 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
3107 PR tree-optimization/86990
3108 * gimple-ssa-store-merging.c (imm_store_chain_info:coalesce_immediate):
3109 Check that the entire merged store group is made of constants only for
3112 2018-09-20 Allan Sandfeld Jensen <allan.jensen@qt.io>
3114 * gcc.c (LINK_COMMAND_SPEC): Handle -r like -nostdlib.
3115 (VTABLE_VERIFICATION_SPEC): Likewise.
3116 (SANITIZER_EARLY_SPEC): Likewise.
3117 (SANITIZER_SPEC): Likewise.
3118 * config/darwin.h (LINK_COMMAND_SPEC): Likewise.
3119 * doc/invoke.texi (Link Options): Document -r.
3121 2018-09-20 Richard Biener <rguenther@suse.de>
3124 * gimplify.c (gimplify_expr): Retain alignment of
3125 addressable lvalue in dereference.
3127 2018-09-20 Alexandre Oliva <aoliva@redhat.com>
3130 * configure.ac: Check for .loc is_stmt support.
3131 * configure, config.in: Rebuilt.
3132 * dwarf2out.c (dwarf2out_source_line): Skip is_stmt
3135 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
3137 * config/rs6000/rs6000.opt (misel=no, misel=yes): Delete.
3138 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -misel=yes and
3141 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
3143 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
3145 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete
3148 2018-09-20 Marek Polacek <polacek@redhat.com>
3150 * doc/invoke.texi: Add -Wno-init-list-lifetime to C++ Language Options.
3152 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
3154 PR tree-optimization/87288
3155 * tree-vect-loop.c (vect_analyze_loop_2): Take PEELING_FOR_GAPS
3156 into account when determining PEELING_FOR_NITERS.
3158 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
3160 PR tree-optimization/86877
3161 * tree-vect-loop.c (vect_analyze_loop_2): Call
3162 vect_verify_datarefs_alignment.
3164 2018-09-19 Marek Polacek <polacek@redhat.com>
3166 * doc/invoke.texi: Document -Wclass-conversion.
3168 2018-09-19 John David Anglin <danglin@gcc.gnu.org>
3170 * config/pa/pa.c (pa_adjust_priority): Delete.
3171 (TARGET_SCHED_ADJUST_PRIORITY): Delete define.
3173 * config/pa/pa.md (atomic_storeqi): Restore deleted expander.
3174 (atomic_storehi): Likewise.
3175 (atomic_storesi): Likewise.
3176 (atomic_loaddi): Restore compare and swap exchange loop code.
3178 2018-09-19 Segher Boessenkool <segher@kernel.crashing.org>
3180 PR rtl-optimization/86902
3181 * combine.c (try_combine): When changing the CC mode used, don't change
3182 an unrelated mode in other_insn to that new CC mode.
3184 2018-09-19 David Malcolm <dmalcolm@redhat.com>
3186 * tree-data-ref.c (runtime_alias_check_p): Use formatted printing
3187 with %T in place of calls to dump_generic_expr.
3188 (prune_runtime_alias_test_list): Likewise.
3189 (create_runtime_alias_checks): Likewise.
3190 * tree-vect-data-refs.c (vect_check_nonzero_value): Likewise.
3191 (vect_analyze_data_ref_dependence): Likewise.
3192 (vect_slp_analyze_data_ref_dependence): Likewise.
3193 (vect_record_base_alignment): Likewise. Use %G in place of call
3194 to dump_gimple_stmt.
3195 (vect_compute_data_ref_alignment): Likewise.
3196 (verify_data_ref_alignment): Likewise.
3197 (vect_find_same_alignment_drs): Likewise.
3198 (vect_analyze_group_access_1): Likewise.
3199 (vect_analyze_data_ref_accesses): Likewise.
3200 (dependence_distance_ge_vf): Likewise.
3201 (dump_lower_bound): Likewise.
3202 (vect_prune_runtime_alias_test_list): Likewise.
3203 (vect_find_stmt_data_reference): Likewise.
3204 (vect_analyze_data_refs): Likewise.
3205 (vect_create_addr_base_for_vector_ref): Likewise.
3206 (vect_create_data_ref_ptr): Likewise.
3207 * tree-vect-loop-manip.c (vect_set_loop_condition): Likewise.
3208 (vect_can_advance_ivs_p): Likewise.
3209 (vect_update_ivs_after_vectorizer): Likewise.
3210 (vect_gen_prolog_loop_niters): Likewise.
3211 (vect_prepare_for_masked_peels): Likewise.
3212 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
3213 (vect_determine_vectorization_factor): Likewise.
3214 (vect_is_simple_iv_evolution): Likewise.
3215 (vect_analyze_scalar_cycles_1): Likewise.
3216 (vect_analyze_loop_operations): Likewise.
3217 (report_vect_op): Likewise.
3218 (vect_is_slp_reduction): Likewise.
3219 (check_reduction_path): Likewise.
3220 (vect_is_simple_reduction): Likewise.
3221 (vect_create_epilog_for_reduction): Likewise.
3222 (vect_finalize_reduction:): Likewise.
3223 (vectorizable_induction): Likewise.
3224 (vect_transform_loop_stmt): Likewise.
3225 (vect_transform_loop): Likewise.
3226 (optimize_mask_stores): Likewise.
3227 * tree-vect-patterns.c (vect_pattern_detected): Likewise.
3228 (vect_split_statement): Likewise.
3229 (vect_recog_over_widening_pattern): Likewise.
3230 (vect_recog_average_pattern): Likewise.
3231 (vect_determine_min_output_precision_1): Likewise.
3232 (vect_determine_precisions_from_range): Likewise.
3233 (vect_determine_precisions_from_users): Likewise.
3234 (vect_mark_pattern_stmts): Likewise.
3235 (vect_pattern_recog_1): Likewise.
3236 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
3237 (vect_record_max_nunits): Likewise.
3238 (vect_build_slp_tree_1): Likewise.
3239 (vect_build_slp_tree_2): Likewise.
3240 (vect_print_slp_tree): Likewise.
3241 (vect_analyze_slp_instance): Likewise.
3242 (vect_detect_hybrid_slp_stmts): Likewise.
3243 (vect_detect_hybrid_slp_1): Likewise.
3244 (vect_slp_analyze_operations): Likewise.
3245 (vect_slp_analyze_bb_1): Likewise.
3246 (vect_transform_slp_perm_load): Likewise.
3247 (vect_schedule_slp_instance): Likewise.
3248 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
3249 (vect_mark_stmts_to_be_vectorized): Likewise.
3250 (vect_init_vector_1): Likewise.
3251 (vect_get_vec_def_for_operand): Likewise.
3252 (vect_finish_stmt_generation_1): Likewise.
3253 (vect_check_load_store_mask): Likewise.
3254 (vectorizable_call): Likewise.
3255 (vectorizable_conversion): Likewise.
3256 (vectorizable_operation): Likewise.
3257 (vectorizable_load): Likewise.
3258 (vect_analyze_stmt): Likewise.
3259 (vect_is_simple_use): Likewise.
3260 (vect_get_vector_types_for_stmt): Likewise.
3261 (vect_get_mask_type_for_stmt): Likewise.
3262 * tree-vectorizer.c (increase_alignment): Likewise.
3264 2018-09-19 Andrew Stubbs <ams@codesourcery.com>
3266 * doc/rtl.texi: Adjust vec_select description.
3267 * simplify-rtx.c (simplify_binary_operation_1): Allow VEC_SELECT to use
3268 non-constant selectors.
3270 2018-09-19 Matthew Malcomson <matthew.malcomson@arm.com>
3272 * config/aarch64/aarch64-protos.h
3273 (aarch64_offset_9bit_signed_unscaled_p): New declaration.
3274 * config/aarch64/aarch64.md (arches): New "rcpc8_4" attribute value.
3275 (arch_enabled): Add check for "rcpc8_4" attribute value of "arch".
3276 * config/aarch64/aarch64.h (AARCH64_FL_RCPC8_4): New bitfield.
3277 (AARCH64_FL_FOR_ARCH8_4): Include AARCH64_FL_RCPC8_4.
3278 (AARCH64_FL_PROFILE): Move index so flags are ordered.
3279 (AARCH64_ISA_RCPC8_4): New flag.
3280 * config/aarch64/aarch64.c (offset_9bit_signed_unscaled_p): Renamed
3281 to aarch64_offset_9bit_signed_unscaled_p.
3282 * config/aarch64/atomics.md (atomic_store<mode>): Allow offset
3284 * config/aarch64/constraints.md (Ust): New constraint.
3285 * config/aarch64/predicates.md.
3286 (aarch64_9bit_offset_memory_operand): New predicate.
3287 (aarch64_rcpc_memory_operand): New predicate.
3289 2018-09-19 Eric Botcazou <ebotcazou@adacore.com>
3291 PR rtl-optimization/87361
3292 * rtlanal.c (nonzero_bits1): Revert accidental change.
3294 2018-09-19 Richard Biener <rguenther@suse.de>
3296 PR tree-optimization/87349
3297 PR tree-optimization/87342
3298 * tree-ssa-sccvn.c (do_rpo_vn): Iterate max_rpo computation.
3300 2018-09-18 Marek Polacek <polacek@redhat.com>
3302 P1064R0 - Allowing Virtual Function Calls in Constant Expressions
3303 * gimple-fold.c (gimple_get_virt_method_for_vtable): Adjust assert.
3305 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
3307 * config/rs6000/rs6000.md: Remove old "Cygnus sibcall" comment.
3309 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
3311 PR rtl-optimization/86882
3312 * rtlanal.c (reg_overlap_mentioned_p): Handle CLOBBER.
3314 2018-09-18 Uros Bizjak <ubizjak@gmail.com>
3316 * config/i386/i386.md (*<code>extend<mode>xf2): Macroize insn from
3317 *<code>extendsfxf2 and *<code>extenddfxf2 using MODEF mode iterator.
3319 2018-09-18 Jonathan Wakely <jwakely@redhat.com>
3322 * doc/invoke.texi (Link Options): Fix formatting and grammar.
3324 2018-09-18 Richard Biener <rguenther@suse.de>
3327 * tree-ssa-coalesce.c (tree_int_map_hasher): Remove.
3328 (compute_samebase_partition_bases): Likewise.
3329 (coalesce_ssa_name): Always use compute_optimized_partition_bases.
3330 (gimple_can_coalesce_p): Simplify.
3332 2018-09-18 Hans-Peter Nilsson <hp@bitrange.com>
3334 Handle a library implementation of ffs calling __builtin_ffs.
3335 * config/mmix/mmix.c (TARGET_INIT_LIBFUNCS): Override with...
3336 (mmix_init_libfuncs): New function: make __builtin_ffs expand
3339 2018-09-17 David Malcolm <dmalcolm@redhat.com>
3341 * diagnostic-show-locus.c (class layout_range): Add field
3343 (layout_range::layout_range): Add "original_idx" param and use it
3344 to initialize new field.
3345 (make_range): Use 0 for original_idx.
3346 (layout::layout): Pass in index to calls to
3347 maybe_add_location_range.
3348 (layout::maybe_add_location_range): Add param "original_idx" and
3349 pass it on to layout_range.
3350 (layout::print_any_labels): Pass on range->m_original_idx to
3352 (gcc_rich_location::add_location_if_nearby): Use 0 for
3354 * gcc-rich-location.h (text_range_label::get_text): Update for new
3356 (range_label_for_type_mismatch::get_text): Likewise.
3358 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
3360 * config/i386/i386.c (ix86_emit_i387_log1p): Emit fldln2 earlier.
3362 2018-09-17 David Malcolm <dmalcolm@redhat.com>
3364 * gimple-ssa-sprintf.c (fmtwarn): Update for introduction of
3365 format_string_diagnostic_t.
3366 (fmtwarn_n): Likewise.
3367 * substring-locations.c
3368 (format_string_diagnostic_t::format_string_diagnostic_t) New ctor.
3369 (format_warning_n_va): Convert to...
3370 (format_string_diagnostic_t::emit_warning_n_va): ...this.
3371 (format_warning_va): Convert to...
3372 (format_string_diagnostic_t::emit_warning_va): ...this.
3373 (format_warning_at_substring): Convert to...
3374 (format_string_diagnostic_t::emit_warning): ...this.
3375 (format_warning_at_substring_n): Convert to...
3376 (format_string_diagnostic_t::emit_warning_n): ...this.
3377 * substring-locations.h (class format_string_diagnostic_t): New
3379 (format_warning_va): Convert to
3380 format_string_diagnostic_t::emit_warning_va.
3381 (format_warning_n_va): Convert to
3382 format_string_diagnostic_t::emit_warning_n_va.
3383 (format_warning_at_substring): Convert to
3384 format_string_diagnostic_t::emit_warning.
3385 (format_warning_at_substring_n): Convert to
3386 format_string_diagnostic_t::emit_warning_n.
3388 2018-09-17 Cesar Philippidis <cesar@codesourcery.com>
3389 Bernd Schmidt <bernds_cb1@t-online.de>
3391 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with
3394 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
3396 * config/i386/i386.md (truncxf<mode>2_i387_noop_unspec): Change
3397 operand 0 predicate to nonimmediate operand.
3398 (rint<mode>2_frndint): Remove insn pattern.
3399 (rint<mode>2): Change operand 1 predicate to general_operand.
3400 Extend operand 1 to XFmode and generate rintxf2 insn.
3401 (frndintxf2_<rounding>): Rename from frndint<mode>2_<rounding>.
3402 Do not use X87MODEF mode macro.
3403 (frndintxf2_<rounding>_i387): Rename from
3404 frndint<mode>2_<rounding>_i387. Do not use X87MODEF mode macro.
3405 (<rounding_insn><mode>2): For non-SSE modes, extend operand 1
3406 to XFmode and generate significandxf3 insn.
3408 2018-09-17 Richard Biener <rguenther@suse.de>
3410 PR tree-optimization/87328
3411 * tree-ssa-sccvn.c (process_bb): Remove assertion about not
3412 visiting unexecutable backedges when not iterating.
3413 (do_rpo_vn): Mark all edges not executable even when not
3416 2018-09-17 Martin Jambor <mjambor@suse.cz>
3419 * doc/invoke.texi (Warning Options): Likewise.
3421 2018-09-17 Richard Biener <rguenther@suse.de>
3423 PR tree-optimization/87301
3424 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup): Properly
3425 clean EH info from leftover copy assignments.
3427 2018-09-17 Martin Liska <mliska@suse.cz>
3429 PR gcov-profile/85871
3430 * gcov.c (output_intermediate_file): Fix out of bounds
3433 2018-09-17 Vineet Gupta <vgupta@synopsys.com>
3435 * config/arc/arc.c: Object attributes for core4 not reflected
3437 * config/arc/arc.h: Don't restrict DBNZ to core3 (core4 includes
3440 2018-09-17 Alexey Brodkin <abrodkin@synopsys.com>
3442 * config/arc/linux.h (LINK_EH_SPEC): Add missing space.
3444 2018-09-17 Martin Liska <mliska@suse.cz>
3446 * doc/gcov.texi: Document new option --use-hotness-colors.
3447 * gcov.c (struct source_info): Declare new field.
3448 (source_info::source_info): Set default for maximum_count.
3449 (print_usage): Add new -q option.
3450 (process_args): Process it.
3451 (accumulate_line_info): Save src->maximum_count.
3452 (output_line_beginning): Make color line number if
3453 flag_use_hotness_colors is set.
3454 (output_line_details): Pass default argument value.
3455 (output_lines): Pass src->maximum_count.
3457 2018-09-17 Martin Liska <mliska@suse.cz>
3459 * common/config/i386/i386-common.c (ix86_get_valid_option_values):
3460 Use processor_names table.
3461 * config/i386/i386.c (ix86_default_align): Use
3462 processor_cost_table for alignment values.
3463 (ix86_option_override_internal): Use processor_names.
3464 (ix86_function_specific_print): Likewise.
3465 * config/i386/i386.h (struct processor_costs):
3466 Add alignment values.
3467 (struct ptt): Remove and replace with const char *.
3468 * config/i386/x86-tune-costs.h (struct processor_costs):
3469 Declare default alignments for all costs.
3471 2018-09-17 Aldy Hernandez <aldyh@redhat.com>
3473 * tree-vrp.c (extract_range_from_unary_expr): Do not special case
3474 symbolics or VR_VARYING ranges for ABS_EXPR.
3475 * wide-int-range.cc (wide_int_range_abs): Return positive numbers
3476 when range will wrap.
3478 2018-09-15 Eric Botcazou <ebotcazou@adacore.com>
3481 * cfgexpand.c (expand_gimple_basic_block): Be prepared for a BARRIER
3482 before and after a JUMP_TABLE_DATA.
3484 2018-09-14 John David Anglin <danglin@gcc.gnu.org>
3487 * dojump.c (do_compare_and_jump): Canonicalize function pointers
3488 when one operand is a function pointer. Use POINTER_TYPE_P and
3489 FUNC_OR_METHOD_TYPE_P.
3490 * expr.c (do_store_flag): Use POINTER_TYPE_P and FUNC_OR_METHOD_TYPE_P.
3491 * fold-const.c (build_range_check): Likewise.
3492 * match.pd (simple_comparison): Likewise.
3494 2018-09-14 David Malcolm <dmalcolm@redhat.com>
3497 * spellcheck.c (get_edit_distance_cutoff): New function.
3498 (selftest::test_edit_distance_unit_test_oneway): Rename to...
3499 (selftest::test_get_edit_distance_one_way): ...this.
3500 (selftest::test_get_edit_distance_unit): Rename to...
3501 (selftest::test_get_edit_distance_both_ways): ...this.
3502 (selftest::test_edit_distances): Move tests to this new function,
3503 and test some more pairs of strings. Update for above renaming.
3504 (selftest::get_old_cutoff): New function.
3505 (selftest::test_get_edit_distance_cutoff): New function.
3506 (selftest::assert_suggested_for): New function.
3507 (ASSERT_SUGGESTED_FOR): New macro.
3508 (selftest::assert_not_suggested_for): New function.
3509 (ASSERT_NOT_SUGGESTED_FOR): New macro.
3510 (selftest::test_suggestions): New function.
3511 (selftest::spellcheck_c_tests): Move test_get_edit_distance_unit
3512 tests to selftest::test_edit_distances and call it. Add calls to
3513 selftest::test_get_edit_distance_cutoff and
3514 selftest::test_suggestions.
3515 * spellcheck.h (get_edit_distance_cutoff): New function declaration.
3516 (best_match::consider): Replace hard-coded cutoff calculation with
3518 (best_match::get_cutoff): New declaration.
3519 (best_match::get_best_meaningful_candidate): Likewise.
3521 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
3523 * builtins.c (fold_builtin_strlen): Remove TODO comment.
3525 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
3528 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
3530 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
3531 terminated string literal.
3533 2018-09-14 Martin Sebor <msebor@redhat.com>
3535 * builtins.c (unterminated_array): Handle ARRAY_REF.
3536 (expand_builtin_stpcpy_1): Detect unterminated char arrays.
3537 * builtins.h (unterminated_array): Declare extern.
3538 * gimple-fold.c (gimple_fold_builtin_stpcpy): Detect unterminated
3540 (gimple_fold_builtin_sprintf): Propagate NO_WARNING to transformed
3543 2018-09-14 Martin Sebor <msebor@redhat.com>
3544 Jeff Law <law@redhat.com>
3546 * builtins.c (unterminated_array): New.
3547 (expand_builtin_strcpy): Adjust.
3548 (expand_builtin_strcpy_args): Detect unterminated arrays.
3549 * gimple-fold.c (get_maxval_strlen): Add argument. Detect
3550 unterminated arrays.
3551 * gimple-fold.h (get_maxval_strlen): Add argument.
3552 (gimple_fold_builtin_strcpy): Detec unterminated arrays.
3554 * gimple-fold.c (get_range_strlen): Add argument.
3555 (get_maxval_strlen): Adjust.
3556 * gimple-fold.h (get_range_strlen): Add argument.
3558 2018-09-14 Wei Xiao <wei3.xiao@intel.com>
3560 * config/i386/movdirintrin.h: Fix copyright year.
3562 2018-09-14 Uros Bizjak <ubizjak@gmail.com>
3564 * reg-stack.c: Include regs.h.
3565 (replace_reg): Assert that mode is MODE_FLOAT or MODE_COMPLEX_FLOAT.
3566 (emit_pop_insn): Default pop insn mode to the reg_raw_mode of
3567 FIRST_STACK_REG, not DFmode.
3568 (emit_swap_insn): Default swap insn mode to the reg_raw_mode of
3569 FIRST_STACK_REG, not XFmode. Explicitly construct swap RTX.
3570 (change stack): Default register mode to the reg_raw_mode of
3571 FIRST_STACK_REG, not DFmode.
3572 * config/i386/i386.md (*swap<mode>): Remove insn pattern.
3573 (*swapxf): Rename from swapxf.
3575 2018-09-14 Carl Love <cel@us.ibm.com>
3577 * config/rs6000/emmintrin.h: Add _MM_SHUFFLE2.
3578 * config/rs6000/xmmintrin.h: Add _MM_SHUFFLE.
3580 2018-09-14 Segher Boessenkool <segher@kernel.crashing.org>
3583 * config/rs6000/rs6000.md (*mov<mode>_hardfloat64): Add Z to the Y
3586 2018-09-14 Sam Tebbs <sam.tebbs@arm.com>
3589 * config/aarch64/aarch64.md (*aarch64_bfxilsi_uxtw): Define.
3591 2018-09-14 Jason Merrill <jason@redhat.com>
3593 Fix --enable-gather-detailed-mem-stats.
3594 * hash-table.c (hash_table_usage): Change from variable to function.
3595 * hash-table.h: Adjust.
3596 * Makefile.in: Add missing dependencies on hash-table.h.
3598 2018-09-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3600 PR tree-optimization/87259
3602 (pass_cse_reciprocals::execute): Run optimize_recip_sqrt after
3603 execute_cse_reciprocals_1 has tried transforming.
3605 2018-09-14 Aldy Hernandez <aldyh@redhat.com>
3607 * tree-vrp.c (extract_range_from_binary_expr_1): Normalize
3608 VR_VARYING for PLUS/MINUS_EXPR.
3610 2018-09-14 Ilya Leoshkevich <iii@linux.ibm.com>
3612 * config/s390/s390-passes.def (INSERT_PASS_BEFORE): Improve
3615 2018-09-14 Richard Biener <rguenther@suse.de>
3618 * tree-ssa-coalesce.c (ssa_conflicts_merge): Remove conflict
3619 bits for the merged partition.
3621 2018-09-13 Martin Sebor <msebor@redhat.com>
3622 Bernd Edlinger <bernd.edlinger@hotmail.de>
3624 * builtins.h (c_srlen): Add argument.
3625 * builtins.c (warn_string_no_nul): New function.
3626 (c_strlen): Add argument and use it. Update recursive calls.
3627 Pass DECL argument to string_constant to get info on non
3628 terminated strings. Update *NONSTR as needed.
3629 (fold_builtin_strlen): Add argument to calls to c_strlen.
3630 Warn for unterminated arrays.
3631 (warn_string_no_null): Add prototype.
3632 * expr.c (string_constant): Update arguments. Update recursive
3633 calls appropriately. Detect missing NUL terminator and outermost
3634 declaration its missing in.
3635 Improve checks for arrays with nonzero lower bound or elements
3636 that are not a single byte. Simplify offset computation.
3637 Simplify checks for non-NUL terminated strings.
3638 * gimple-fold.c (get_range_strlen): Add argument to c_strlen call.
3639 * gimple-ssa-sprintf.c (get_string_length): Remove unnecessary code.
3641 2018-09-13 Bernd Edlinger <bernd.edlinger@hotmail.de>
3643 * builtins.c (c_strlen): Handle not zero terminated STRING_CSTs
3645 * fold-const.c (c_getstr): Fix function comment. Remove unused third
3646 argument. Fix range checks.
3647 * fold-const.h (c_getstr): Adjust protoype.
3648 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid folding when
3649 string is constant but contains no NUL byte.
3651 * expr.c (string_constant): Adjust function comment.
3652 Remove bogus check for zero termination.
3654 * fold-const.c (c_getstr): Clamp STRING_LENGTH to STRING_SIZE.
3656 * varasm.c (compare_constant): Compare type size of STRING_CSTs.
3657 (get_constant_size): Don't make STRING_CSTs larger than they are.
3658 (check_string_literal): New check function for STRING_CSTs.
3659 (output_constant): Use it.
3661 2018-09-13 Eric Botcazou <ebotcazou@adacore.com>
3664 * config/visium/visium.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define.
3666 2018-09-13 Richard Biener <rguenther@suse.de>
3668 PR tree-optimization/87263
3669 * tree-ssa-sccvn.c (visit_phi): Revert some earlier changes.
3670 (struct unwind_state): Add max_rpo field.
3671 (do_rpo_vn): Allow up-to-date loop state to be used when not iterating.
3672 Compute max_rpo, the max RPO number a block can be backwards reached
3673 from. Re-write non-iterating mode to a RPO ordered worklist approach,
3674 separating it from the iterating mode.
3676 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
3678 * haifa-sched.c (rank_for_schedule): Schedule by INSN_COST.
3679 (rfs_decision): New scheduling decision.
3681 2018-09-13 Richard Biener <rguenther@suse.de>
3684 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Fix assert.
3685 (vn_nary_op_insert_pieces_predicated): Do not write useless
3686 valid_dominated_by_p entry outside of the allocated storage.
3688 2018-09-13 Omar Sandoval <osandov@osandov.com>
3689 Tom de Vries <tdevries@suse.de>
3692 * dwarf2out.c (is_c): New function.
3693 (add_subscript_info): Add DW_AT_count of 0 for C zero-length arrays.
3695 2018-09-13 Sam Tebbs <sam.tebbs@arm.com>
3698 * config/aarch64/aarch64.md (*aarch64_bfxil):
3700 * config/aarch64/constraints.md (Ulc): Define.
3701 * config/aarch64/aarch64-protos.h (aarch64_high_bits_all_ones_p):
3703 * config/aarch64/aarch64.c (aarch64_high_bits_all_ones_p):
3706 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
3708 * config/aarch64/aarch64.h (TARGET_COMPUTE_FRAME_LAYOUT): Define.
3709 * config/aarch64/aarch64.c (aarch64_expand_prologue): Remove
3710 aarch64_layout_frame call.
3711 (aarch64_expand_epilogue): Likewise.
3712 (aarch64_initial_elimination_offset): Likewise.
3713 (aarch64_get_separate_components): Likewise.
3714 (aarch64_use_return_insn_p): Likewise.
3715 (aarch64_layout_frame): Remove unneeded check.
3717 2018-09-13 Jakub Jelinek <jakub@redhat.com>
3719 * configure.ac: Only append
3720 " : (reconfigured) $TOPLEVEL_CONFIGURE_ARGUMENTS" to
3721 gcc_config_arguments if it was never reconfigured or last reconfigure
3722 was with different arguments.
3723 * configure: Regenerated.
3725 2018-09-13 Jakub Jelinek <jakub@redhat.com>
3726 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3729 * expr.c (maybe_optimize_pow2p_mod_cmp): New function.
3730 (maybe_optimize_mod_cmp): Use it if integer_pow2p treeop1.
3732 2018-09-13 Jakub Jelinek <jakub@redhat.com>
3734 PR tree-optimization/87287
3735 * fold-const.c (fold_binary_loc) <case EQ_EXPR>: Move signed modulo
3736 X % C == 0 to X % (unsigned) C == 0 optimization to ...
3737 * match.pd (X % C == 0): ... here. New optimization.
3739 2018-09-12 Jakub Jelinek <jakub@redhat.com>
3742 * expr.h (maybe_optimize_mod_cmp): Declare.
3743 * expr.c (mod_inv): New function.
3744 (maybe_optimize_mod_cmp): New function.
3745 (do_store_flag): Use it.
3746 * cfgexpand.c (expand_gimple_cond): Likewise.
3748 2018-09-09 Cesar Philippidis <cesar@codesourcery.com>
3749 Julian Brown <julian@codesourcery.com>
3752 * gimplify.c (gimplify_scan_omp_clauses): Set
3753 target_firstprivatize_array_bases in OpenACC parallel and kernels
3754 region contexts. Remove GOMP_MAP_FIRSTPRIVATE_REFERENCE clauses from
3755 OpenACC data regions.
3757 2018-09-12 Uros Bizjak <ubizjak@gmail.com>
3759 * config/i386/i386.md (sqrt_extend<mode>xf3_i387): Remove.
3760 (sqrt<mode>2): Extend operand 1 to XFmode and generate sqrtxf3 insn.
3762 2018-09-12 Richard Biener <rguenther@suse.de>
3764 PR tree-optimization/87280
3765 * tree-ssa-sccvn.c (process_bb): Handle the case of executable
3766 edge but unreachable target.
3767 (do_rpo_vn): For conservatively handling a PHI only mark
3768 the backedge executable but not the block reachable.
3770 2018-09-12 Richard Biener <rguenther@suse.de>
3772 PR tree-optimization/87266
3773 * tree-ssa-sccvn.c (do_rpo_vn): Always iterate to not yet
3776 2018-09-12 Andreas Krebbel <krebbel@linux.ibm.com>
3778 * config/s390/s390.md (PFPO_RND_MODE_DFP, PFPO_RND_MODE_BFP): New
3780 ("trunc<BFP:mode><DFP_ALL:mode>2")
3781 ("trunc<DFP_ALL:mode><BFP:mode>2")
3782 ("extend<BFP:mode><DFP_ALL:mode>2")
3783 ("extend<DFP_ALL:mode><BFP:mode>2"): Set proper rounding mode
3784 according to the target operand type.
3786 2018-09-12 Jakub Jelinek <jakub@redhat.com>
3787 Andreas Krebbel <krebbel@linux.ibm.com>
3789 PR tree-optimization/86844
3790 * gimple-ssa-store-merging.c
3791 (imm_store_chain_info::coalesce_immediate): For overlapping stores, if
3792 there are any overlapping stores in between them, make sure they are
3793 also coalesced or we give up completely.
3795 2018-09-12 Jakub Jelinek <jakub@redhat.com>
3798 * fold-const.c (fold_ternary_loc) <case COND_EXPR>: Verify also that
3799 BIT_AND_EXPR's second operand is a power of two. Formatting fix.
3801 2018-09-12 Tom de Vries <tdevries@suse.de>
3803 * common.opt (gdescribe-dies): Add option.
3804 * dwarf2out.c (add_name_and_src_coords_attributes): Add description
3805 attribute for artifical and nameless decls.
3806 (dwarf2out_register_external_die): Add description attribute to
3807 external reference die.
3808 (add_desc_attribute): New functions.
3809 (gen_subprogram_die): Add description attribute to
3810 DW_TAG_call_site_parameter.
3811 * tree-pretty-print.c (print_generic_expr_to_str): New function.
3812 * tree-pretty-print.h (print_generic_expr_to_str): Declare.
3813 * doc/invoke.texi (@item Debugging Options): Add -gdescribe-dies and
3815 (@item -gdescribe-dies): Add.
3817 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
3819 * tree-vrp.c (vrp_shift_undefined_p): Remove.
3820 (extract_range_from_binary_expr_1: Call
3821 wide_int_range_shift_undefined_p instead of vrp_shift_undefined_p.
3822 * wide-int-range.h (wide_int_range_shift_undefined_p): Do not
3825 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
3827 * gimple-ssa-warn-alloca.c
3828 (alloca_type_and_limit::alloca_type_and_limit): Initialize limit
3829 field for ALLOCA_BOUND_*_LARGE.
3831 2018-09-11 Nathan Sidwell <nathan@acm.org>
3833 * gcc.c (load_specs, execute, run_attempt): Use %qs not '%s'.
3835 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
3837 * reg-stack.c (subst_asm_stack_regs): Call replace_reg also
3838 for clobbers. Remove obsolete comment.
3840 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
3842 * config/i386/i386.md (define_attr "type"): Remove mpxmov, mpxmk,
3843 mpxchk, mpxld and mpxst types.
3844 (define_attr length_immediate): Remove all processing of mpx types.
3845 (define_attr prefix_0f): Ditto.
3846 (define_attr memory): Ditto.
3848 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
3850 * config/i386/i386.md (fyl2x_extend<mode>xf3_i387): Remove.
3851 (log<mode>2): Change operand 1 predicate to general_operand.
3852 Extend operand 1 to XFmode and generate logxf3 insn.
3853 (log10<mode>2): Change operand 1 predicate to general_operand.
3854 Extend operand 1 to XFmode and generate log10xf3 insn.
3855 (log2<mode>2): Change operand 1 predicate to general_operand.
3856 Extend operand 1 to XFmode and generate log2xf3 insn.
3857 (fyl2xp1_extend<mode>xf3_i387): Remove.
3858 (log1p<mode>2): Change operand 1 predicate to general_operand.
3859 Extend operand 1 to XFmode and generate log1pxf3 insn.
3860 (fxtract_extend<mode>xf3_i387): Remove.
3861 (logb<mode>2): Change operand 1 predicate to general_operand.
3862 Extend operand 1 to XFmode and generate logbxf3 insn.
3863 (ilogb<mode>2): Change operand 1 predicate to general_operand.
3864 Extend operand 1 to XFmode and generate fxtractxf3_i387 insn.
3865 (significand<mode>2): Change operand 1 predicate to general_operand.
3866 Extend operand 1 to XFmode and generate significandxf3 insn.
3868 2018-09-11 Nathan Sidwell <nathan@acm.org>
3870 * gcc.c (perror_with_name, pfatal_with_name): Delete.
3871 (load_specs): Use fatal_error.
3872 (DELETE_IF_ORDINARY, process_command): Use error.
3873 (execute, run_attempt): Use fatal_error.
3875 2018-09-11 Andrew Stubbs <ams@codesourcery.com>
3877 * diagnostic-core.h (sorry_at): New prototype.
3878 * diagnostic.c (sorry_at): New function.
3880 2018-09-11 Aldy Hernandez <aldyh@redhat.com>
3882 * tree-vrp.c (extract_range_from_binary_expr_1): Treat all divisions
3883 by zero as VR_UNDEFINED.
3885 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
3887 * config/i386/i386.md (<sincos>xf2): Rename from *<sincos>xf2_i387.
3888 (*<sincos>_extend<mode>xf2_i387): Remove insn pattern.
3889 (<sincos>mode2): New expander.
3890 (sincos_extend<mode>xf3_i387): Remove insn pattern.
3891 (sincos -> sin, cos splitters): Remove splitter patterns.
3892 (sincos<mode>3): Change operand 2 predicate to general_operand.
3893 Extend operand 2 to XFmode and generate sincosxf3 insn.
3894 (fptanxf4_i387): Change mode of operands 0 and 3 to SFmode.
3895 Change operand 3 predicate to const1_operand.
3896 (fptan_extend<mode>xf4_i387): Remove insn pattern.
3897 (tanxf2): Update operands in the call to fptanxf4_i387.
3898 (tan<mode>2): Change operand 1 predicate to general_operand.
3899 Extend operand 1 to XFmode and generate tanxf3 insn.
3900 (atan2xf3): Rename from *fpatanxf3_i387.
3901 (fpatan_extend<mode>xf3_i387): Remove insn pattern.
3902 (atan2xf3): Remove expander.
3903 (atan2<mode<3): Change operand 1 and 2 predicates to general_operand.
3904 Extend operands 1 and 2 to XFmode and generate atan2xf3 insn.
3905 (atan<mode>2): Change operand 1 predicate to general_operand.
3906 Extend operand 1 to XFmode and generate atanxf3 insn.
3908 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
3910 * config/i386/i386.md (x87/SSE constant load splitter): Use
3911 memory_operand instead of nonimmediate_operand for input operand
3914 2018-09-09 Uros Bizjak <ubizjak@gmail.com>
3916 * config/i386/i386.md (float partial SSE register stall splitter): Move
3917 splitter near its instruction pattern.
3918 (float_extend partial SSE register stall splitter): Ditto.
3919 (float_truncate partial SSE register stall splitter): Ditto.
3921 2018-09-09 Hans-Peter Nilsson <hp@bitrange.com>
3924 * config/mmix/mmix.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
3925 to speculation_safe_value_not_needed.
3928 * config/mmix/mmix.c (mmix_assemble_integer): Handle byte-size
3929 non-CONST_INT rtx:es using assemble_integer_with_op ".byte".
3930 (MMIX_CFUN_NEEDS_SAVED_EH_RETURN_ADDRESS): Don't call
3931 leaf_function_p, instead use has_hard_reg_initial_val.
3933 2018-09-09 Nathan Sidwell <nathan@acm.org>
3935 * gcc.h (pfatal_with_name): Don't declare here.
3936 * gcc.c (pfatal_with_name): Make static.
3938 2018-09-09 Xianmiao Qu <xianmiao_qu@c-sky.com>
3940 * config/csky/csky.md (*cskyv2_adddi3, *cskyv2_subdi3): Add
3943 2018-09-08 John David Anglin <danglin@gcc.gnu.org>
3945 PR rtl-optimization/85458
3946 * sel-sched.c (sel_target_adjust_priority): Allow backend adjust
3947 priority hook to reduce the priority of EXPR.
3949 2018-09-07 Uros Bizjak <ubizjak@gmail.com>
3951 * config/i386/i386.md (float<SWI48x:mode><MODEF:mode>2) Enable
3952 DImode for x87 on 32bit targets. Conditionally disable x87 modes
3953 with X87_ENABLE_FLOAT. Remove preparation code.
3954 (*float<SWI48:mode><MODEF:mode>2): Rename from
3955 *float<SWI48:mode><MODEF:mode>2_mixed. Handle x87, SSE and mixed
3956 math using "enabled" attribute.
3957 (*floatdi<MODEF:mode>2_i387): Rename from
3958 *float<SWI48x:mode><MODEF:mode>2_i387. Handle only DImode and
3959 enable for 32bit targets only.
3960 (floatdi<X87MODEF:mode>2_i387_with_xmm pre-reload splitter): New
3962 (floatdi<X87MODEF:mode>2_i387_with_xmm): Use register_operand
3963 as operand 1 predicate. Rewrite as define_insn_and_split.
3964 (floatdi<X87MODEF:mode>2_i387_with_xmm memory input splitter): Remove.
3966 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
3968 * reg-stack.c (get_true_reg) <case FLOAT_TRUNCATE>: Reorder
3969 to fallthru to FLOAT case.
3971 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
3974 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Update logic
3975 around folding of vec_sl to handle out of range shift values.
3977 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
3979 * config/i386/i386.md (fix_trunc<mode>_fisttp_i387_1): Remove.
3980 Update callers to gen_fix_trunc<mode>_i387_fisttp
3981 (fix_trunc<mode>_i387_fisttp): Change operand 0 predicate to
3982 nonimmediate_operand.
3983 (fix_trunc<mode>_i387_fisttp_with_temp): Remove insn pattern
3984 and corresponding splitters.
3985 (*fix_trunc<mode>_i387_1): Always expand with fix_trunc<mode>_i387.
3986 (fix_truncdi_i387): Change operand 0 predicate to nonimmediate_operand.
3987 (fix_truncdi_i387_with_temp): Remove insn pattern
3988 and corresponding splitters.
3989 (fix_trunc<mode>_i387): Change operand 0 predicate to
3990 nonimmediate_operand.
3991 (fix_trunc<mode>_i387_with_temp): Remove insn pattern
3992 and corresponding splitters.
3993 (*fistdi2_1): Remove.
3995 (fistdi2_with_temp): Remove insn pattern and corresponding splitters.
3996 (lrintxfdi2): Remove expander. Reimplement as define_insn.
3997 (*fist<mode>2_1): Remove.
3998 (fist<mode>2): Ditto.
3999 (fist<mode>2_with_temp): Remove insn pattern and corresponding
4001 (lrintxf<mode>2): Remove expander. Reimplement as define_insn.
4002 (*fist<mode>2_<rounding>_1): Always expand with fist<mode>2_<rounding>.
4003 (fistdi2_<rounding>): Change operand 0 predicate to
4004 nonimmediate_operand.
4005 (fistdi2_<rounding>_with_temp): Remove insn pattern
4006 and corresponding splitters.
4007 (fist<mode>2_<rounding>): Change operand 0 predicate to
4008 nonimmediate_operand.
4009 (fist<mode>2_<rounding>_with_temp): Remove insn pattern
4010 and corresponding splitters.
4012 (*fixuns_trunc<mode>si2_avx512f_zext): Depend on TARGET_SSE_MATH.
4014 2018-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
4016 * varasm.c (output_constructor_regular_field): Check TYPE_SIZE_UNIT of
4019 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
4021 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4022 early gimple folding of vec_splat().
4023 * tree-vect-generic.c: Remove static from tree_vec_extract() definition.
4024 * gimple-fold.h: Add an extern define for tree_vec_extract().
4026 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
4028 * config/rs6000/rs6000.c (fold_mergehl_helper): Add types_compatible_p
4029 wrappers around TREE_TYPE comparisons.
4031 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
4034 * config/s390/predicates.md: Add nonsym_memory_operand.
4035 * config/s390/s390.c (s390_legitimize_cs_operand): If operand
4036 contains a SYMBOL_REF, load it into an intermediate pseudo.
4037 (s390_emit_compare_and_swap): Legitimize operand.
4038 * config/s390/s390.md: Use the new nonsym_memory_operand
4039 with UNSPECV_CAS patterns.
4041 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
4044 * config/s390/s390-passes.def: New file.
4045 * config/s390/s390-protos.h (class rtl_opt_pass): Add forward
4047 (make_pass_s390_early_mach): Add declaration.
4048 * config/s390/s390.c (make_pass_s390_early_mach):
4049 (s390_option_override): Remove dynamic registration.
4050 * config/s390/t-s390: Add s390-passes.def.
4052 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
4054 * config/s390/s390.c (s390_decompose_constant_pool_ref):
4055 Remove UNSPEC_LTREL_BASE check.
4056 (annotate_constant_pool_refs): Likewise.
4057 (find_constant_pool_ref): Likewise.
4058 (find_ltrel_base): Removed.
4059 (replace_ltrel_base): Removed.
4060 (s390_mainpool_finish): Remove replace_ltrel_base call.
4061 (s390_chunkify_start): Remove pending LTREL_BASE logic.
4062 (s390_chunkify_finish): Remove replace_ltrel_base call.
4063 * config/s390/s390.md: Remove UNSPEC_LTREL_BASE.
4065 2018-09-06 Hans-Peter Nilsson <hp@axis.com>
4068 * config/cris/cris.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
4069 to speculation_safe_value_not_needed.
4071 2018-09-05 Cesar Philippidis <cesar@codesourcery.com>
4072 Bernd Schmidt <bernds_cb1@t-online.de>
4074 * config/nvptx/nvptx-opts.h: New file.
4075 * config/nvptx/nvptx.c (nvptx_file_start): Print the correct .target.
4076 * config/nvptx/nvptx.h: Include "nvptx-opts.h".
4078 (TARGET_SM35): New macro.
4079 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with the
4081 * config/nvptx/nvptx.opt (ptx_isa, sm_30, sm_35): New enum and its
4083 (misa=): New option.
4084 * doc/invoke.texi (Nvidia PTX Options): Document -misa.
4086 2018-09-05 Uros Bizjak <ubizjak@gmail.com>
4088 * config/i386/i386.md (truncdfsf2): Remove expander.
4089 (truncdfsf2_with_temp): Ditto.
4090 (truncxf<mode>2): Ditto.
4091 (*truncdfsf_fast_mixed): Remove insn pattern.
4092 (*truncdfsf_fast_i387): Ditto.
4093 (*truncdfsf_mixed): Ditto.
4094 (*truncdfsf_i387): Ditto.
4095 (*truncdfsf2_i387_1): Ditto.
4096 (*truncxfsf2_mixed): Ditto.
4097 (*truncxfdf2_mixed): Ditto.
4098 (*truncxf<mode>2_i387_noop): Ditto. Update callers
4099 to call gen_truncxf<mode>2 instead.
4100 (*truncxf<mode>2_i387): Remove.
4101 (reg->reg splitters): Remove splitter pattern.
4102 (reg->mem splitters): Ditto.
4104 (truncdfsf2): New insn pattern.
4105 (truncxf<mode>2): Ditto.
4107 2018-09-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4109 * tree-ssa-math-opts.c (is_mult_by): New function.
4110 (is_square_of): Use the above.
4111 (optimize_recip_sqrt): New function.
4112 (pass_cse_reciprocals::execute): Use the above.
4114 2018-09-05 Richard Biener <rguenther@suse.de>
4117 * tree-ssa-sccvn.c (rpo_elim::eliminate_push_avail): Make sure
4118 to zero-init the emplaced vec.
4120 2018-09-05 Martin Liska <mliska@suse.cz>
4122 PR tree-optimization/87205
4123 * tree-switch-conversion.c (pass_lower_switch::execute):
4124 Group cases for switch statements.
4126 2018-09-05 Richard Biener <rguenther@suse.de>
4128 PR tree-optimization/87217
4129 * tree-ssa-sccvn.c (vuse_valueize): New.
4130 (vn_reference_lookup_pieces): Use it.
4131 (vn_reference_lookup): Likewise.
4133 2018-09-05 Nathan Sidwell <nathan@acm.org>
4136 * stor-layout.c (place_field): Scan forwards to check last
4137 bitfield when ms_bitfield_placement is in effect.
4139 2018-09-05 Richard Biener <rguenther@suse.de>
4142 * tree-vect-stmts.c (vectorizable_simd_clone_call): Fix bogus
4145 2018-09-05 Siddhesh Poyarekar <siddhesh@sourceware.org>
4146 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
4148 * config/aarch64/falkor-tag-collision-avoidance.c: New file.
4149 * config.gcc (extra_objs): Build it.
4150 * config/aarch64/t-aarch64 (falkor-tag-collision-avoidance.o):
4152 * config/aarch64/aarch64-passes.def
4153 (pass_tag_collision_avoidance): New pass.
4154 * config/aarch64/aarch64.c (qdf24xx_tunings): Add
4155 AARCH64_EXTRA_TUNE_RENAME_LOAD_REGS to tuning_flags.
4156 (aarch64_classify_address): Remove static qualifier.
4157 (aarch64_address_info, aarch64_address_type): Move to...
4158 * config/aarch64/aarch64-protos.h: ... here.
4159 (make_pass_tag_collision_avoidance): New function.
4160 * config/aarch64/aarch64-tuning-flags.def (rename_load_regs):
4163 2018-09-05 Martin Liska <mliska@suse.cz>
4165 * doc/gcov.texi: Update documentation of humar
4167 * gcov.c (format_count): Print one decimal place, it provides
4168 more fine number of situations like '1G' vs. '1.4G'.
4170 2018-09-05 Martin Liska <mliska@suse.cz>
4173 * config/rs6000/rs6000.opt: Mark the option as Deprecated.
4174 * optc-gen.awk: Allow 'Var' for Deprecated options in order
4175 to generate a MASK value.
4177 2018-09-04 H.J. Lu <hongjiu.lu@intel.com>
4180 * dwarf2out.c (based_loc_descr): Allow hard frame pointer even
4181 if frame pointer isn't used.
4182 (compute_frame_pointer_to_fb_displacement): Likewise.
4184 2018-09-04 Jakub Jelinek <jakub@redhat.com>
4187 * common/config/i386/i386-common.c (OPTION_MASK_ISA_XSAVEOPT_SET,
4188 OPTION_MASK_ISA_XSAVES_SET, OPTION_MASK_ISA_XSAVEC_SET): Use
4189 OPTION_MASK_ISA_XSAVE_SET instead of OPTION_MASK_ISA_XSAVE.
4190 (OPTION_MASK_ISA_XSAVE_UNSET): Add OPTION_MASK_ISA_XSAVES_UNSET
4191 and OPTION_MASK_ISA_XSAVEC_UNSET.
4193 2018-09-04 Max Filippov <jcmvbkbc@gmail.com>
4195 * config/xtensa/xtensa.c (xtensa_expand_atomic): Reorder AND and
4196 XOR operations in NAND case.
4198 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
4200 * wide-int-range.cc (wide_int_range_convert): New.
4201 * wide-int-range.h (wide_int_range_convert): New.
4202 * tree-vrp.c (extract_range_from_unary_expr): Abstract wide int
4203 code into wide_int_range_convert.
4204 (extract_range_into_wide_ints): Do not munge anti range constants
4205 into the entire domain. Just return the range back.
4207 2018-09-04 Martin Liska <mliska@suse.cz>
4209 * genmatch.c (output_line_directive): Add new argument
4211 (dt_simplify::gen_1): Encapsulate dump within __builtin_expect.
4213 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
4215 * doc/invoke.texi (Option Summary): Add whitespace.
4217 * doc/invoke.texi (Option Summary): Add -Waligned-new.
4219 2018-09-04 Richard Biener <rguenther@suse.de>
4221 PR tree-optimization/87211
4222 * tree-ssa-sccvn.c (visit_phi): When value-numbering to a
4223 backedge value we're supposed to treat as VARYING also number
4224 the PHI to VARYING in case it got a different value-number already.
4226 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
4228 * tree-vrp.c (vrp_can_optimize_bit_op): Remove.
4229 (extract_range_from_binary_expr_1): Do not call
4230 vrp_can_optimize_bit_op.
4231 * wide-int-range.cc (wide_int_range_can_optimize_bit_op): Make
4233 (wide_int_range_get_mask_and_bounds): New.
4234 (wide_int_range_optimize_bit_op): New.
4235 (wide_int_range_bit_ior): Call wide_int_range_optimize_bit_op.
4236 (wide_int_range_bit_and): Same.
4237 * wide-int-range.h (wide_int_range_can_optimize_bit_op): Remove.
4238 (wide_int_range_optimize_bit_op): New.
4239 (wide_int_range_get_mask_and_bounds): New.
4241 2018-09-04 Richard Biener <rguenther@suse.de>
4243 PR tree-optimization/87176
4244 * tree-ssa-sccvn.c (visit_phi): Remove redundant allsame
4245 variable. When value-numbering a virtual PHI node make sure
4246 to not value-number to the backedge value.
4248 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
4250 * doc/extend.texi (Long Long, Hex Floats): Document support for
4251 long long and hex floats in more recent versions of ISO C++.
4253 2018-09-03 Richard Biener <rguenther@suse.de>
4255 PR tree-optimization/87177
4256 * tree-ssa-sccvn.c (vuse_ssa_val): Revert previous change, keep
4259 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
4261 * bb-reorder.c (edge_order): Convert to C-qsort-style
4262 tri-state comparator.
4263 (reorder_basic_blocks_simple): Change std::stable_sort to
4266 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
4268 * tree-loop-distribution.c (offset_cmp): Convert to C-qsort-style
4269 tri-state comparator.
4270 (fuse_memset_builtins): Change std::stable_sort to gcc_stablesort.
4272 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
4274 * sort.cc (struct sort_ctx): New field 'nlim'. Use it...
4275 (mergesort): ... here as maximum count for using netsort.
4276 (gcc_qsort): Set nlim to 3 if stable sort is requested.
4277 (gcc_stablesort): New.
4278 * system.h (gcc_stablesort): Declare.
4280 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
4282 * sort.cc (gcc_qsort) [CHECKING_P]: Call qsort_chk.
4283 * system.h (qsort): Always redirect to gcc_qsort. Update comment.
4284 * vec.c (qsort_chk): Do not call gcc_qsort. Update comment.
4286 2018-09-03 Segher Boessenkool <segher@kernel.crashing.org>
4288 * config/rs6000/rs6000.md (*mov<mode>_hardfloat32): Remove %U from the
4289 lxsdx and stxsdx alternatives.
4290 (*mov<mode>_hardfloat64): Ditto.
4291 * config/rs6000/vsx.md (*vsx_extract_<mode>_store): Ditto.
4293 2018-09-03 Richard Biener <rguenther@suse.de>
4295 PR tree-optimization/87200
4296 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Valueize a
4299 2018-09-03 Martin Liska <mliska@suse.cz>
4301 PR tree-optimization/87201
4302 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
4303 Fix parenthesis in an expression.
4305 2018-09-03 Richard Biener <rguenther@suse.de>
4307 PR tree-optimization/87197
4308 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Mark the new def
4309 visited. CSE the VN_INFO hashtable lookup.
4311 PR tree-optimization/87169
4312 * tree-ssa-sccvn.c (do_rpo_vn): When marking loops for not
4313 iterating make sure there's no extra backedges from irreducible
4314 regions feeding the header. Mark the destination block
4317 2018-09-03 Martin Liska <mliska@suse.cz>
4320 * common/common-target.def: Add TARGET_GET_VALID_OPTION_VALUES.
4321 * common/common-targhooks.c (default_get_valid_option_values):
4323 * common/common-targhooks.h (default_get_valid_option_values):
4325 * common/config/i386/i386-common.c: Move processor_target_table
4327 (ix86_get_valid_option_values): New function.
4328 (TARGET_GET_VALID_OPTION_VALUES): New macro.
4329 * config/i386/i386.c (struct ptt): Move to i386-common.c.
4330 (PTA_*): Move all defined masks into i386-common.c.
4331 (ix86_function_specific_restore): Use new processor_cost_table.
4332 * config/i386/i386.h (struct ptt): Moved from i386.c.
4333 (struct pta): Likewise.
4334 * doc/tm.texi: Document new TARGET_GET_VALID_OPTION_VALUES.
4335 * doc/tm.texi.in: Likewise.
4336 * opt-suggestions.c (option_proposer::suggest_option):
4337 Pass prefix to build_option_suggestions.
4338 (option_proposer::get_completions): Likewise.
4339 (option_proposer::build_option_suggestions): Use the new target
4341 * opts.c (struct option_help_tuple): New struct.
4342 (print_filtered_help): Use the new target hook.
4344 2018-09-03 Martin Liska <mliska@suse.cz>
4347 * predict.c (set_even_probabilities): Add likely_edges
4348 argument and handle cases where we have precisely one
4350 (combine_predictions_for_bb): Catch also likely_edges.
4351 (tree_predict_by_opcode): Handle gswitch statements.
4352 * tree-cfg.h (find_case_label_for_value): New declaration.
4353 (find_taken_edge_switch_expr): Likewise.
4354 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
4355 Find pivot in decision tree based on probabily, not by number of
4358 2018-09-02 Gerald Pfeifer <gerald@pfeifer.com>
4360 * doc/standards.texi (Standards): Update Objective-C reference.
4362 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
4364 * doc/install.texi (Prerequisites): Update link for MPC.
4366 2018-09-01 Michael Matz <matz@suse.de>
4368 PR tree-optimization/87074
4369 * gimple-loop-jam.c (unroll_jam_possible_p): Check loop exit
4370 PHIs for outer-loop uses.
4372 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
4374 * doc/generic.texi (OpenMP): Adjust link to openmp.org.
4375 * doc/invoke.texi (C Dialect Options): Ditto.
4377 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
4379 * doc/install.texi (Prerequisites): Adjust link mpfr.org.
4381 2018-08-31 Richard Biener <rguenther@suse.de>
4383 PR tree-optimization/87168
4384 * tree-ssa-sccvn.c (SSA_VAL): Add visited output parameter.
4385 (rpo_elim::eliminate_avail): When OP was not visited it must
4388 2018-08-31 David Malcolm <dmalcolm@redhat.com>
4390 * tree-vrp.c (copy_value_range): Convert param "from" from
4391 "value_range *" to "const value_range *".
4392 (range_is_null): Likewise for param "vr".
4393 (range_int_cst_p): Likewise.
4394 (range_int_cst_singleton_p): Likewise.
4395 (symbolic_range_p): Likewise.
4396 (value_ranges_intersect_p): Likewise for both params.
4397 (value_range_nonnegative_p): Likewise for param "vr".
4398 (value_range_constant_singleton): Likewise.
4399 (vrp_set_zero_nonzero_bits): Likewise for param "ar".
4400 (extract_range_into_wide_ints): Likewise for param "vr".
4401 (extract_range_from_multiplicative_op): Likewise for params "vr0"
4403 (vrp_can_optimize_bit_op): Likewise.
4404 (extract_range_from_binary_expr_1): Likewise for params "vr0_" and
4406 (extract_range_from_unary_expr): Likewise.
4407 (debug_value_range): Likewise for param "vr".
4408 (value_range::dump): Add "const" qualifier.
4409 (vrp_prop::check_array_ref): Convert local "vr" from
4410 "value_range *" to "const value_range *".
4411 (vrp_prop::check_mem_ref): Likewise.
4412 (vrp_prop::visit_stmt): Likewise for local "old_vr".
4413 (vrp_intersect_ranges_1): Likewise for param "vr_1".
4414 (vrp_intersect_ranges): Likewise.
4415 (simplify_stmt_for_jump_threading): Likewise for local "vr".
4416 (vrp_prop::vrp_finalize): Likewise.
4417 * tree-vrp.h (value_range::dump): Add "const" qualifier.
4418 (vrp_intersect_ranges): Add "const" qualifier to params as above.
4419 (extract_range_from_unary_expr): Likewise.
4420 (value_range_constant_singleton): Likewise.
4421 (symbolic_range_p): Likewise.
4422 (copy_value_range): Likewise.
4423 (extract_range_from_binary_expr_1): Likewise.
4424 (range_int_cst_p): Likewise.
4425 (vrp_set_zero_nonzero_bits): Likewise.
4426 (range_int_cst_singleton_p): Likewise.
4428 2018-08-31 Vlad Lazar <vlad.lazar@arm.com>
4430 * config/aarch64/arm_neon.h (vabsd_s64): New.
4431 (vnegd_s64): Likewise.
4433 2018-08-31 Martin Jambor <mjambor@suse.cz>
4435 * ipa-cp.c (estimate_local_effects): Replace wrong MAX with MIN.
4437 2018-08-31 Martin Liska <mliska@suse.cz>
4439 * ipa-icf.c (sem_item::add_type): Use
4440 sem_item::m_type_hash_cache.
4441 * ipa-icf.h: Move the cache from sem_item_optimizer
4444 2018-08-31 Nathan Sidwell <nathan@acm.org>
4446 * doc/extend.texi (Backwards Compatibility): Remove implicit
4447 extern C leeway of () being (...).
4449 2018-08-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4451 * ipa-inline.c (can_inline_edge_by_limits_p): Fix typos in comment.
4453 2018-08-31 Segher Boessenkool <segher@kernel.crashing.org>
4457 * config/rs6000/rs6000.md (lround<mode>di2): Gate on TARGET_FPRND.
4459 2018-08-31 Jakub Jelinek <jakub@redhat.com>
4462 * expmed.c (expand_mult_const): Use immed_wide_int_const instead of
4463 gen_int_mode. Formatting fixes.
4465 2018-08-30 Sandra Loosemore <sandra@codesourcery.com>
4467 * target.def (custom_function_descriptors): Improve documentation.
4468 * doc/tm.texi.in (Trampolines): Expand discussion of function
4469 descriptors and move TARGET_CUSTOM_FUNCTION_DESCRIPTORS to the
4470 beginning of the section.
4471 * doc/tm.texi: Regenerated.
4473 2018-08-30 Jose E. Marchesi <jose.marchesi@oracle.com>
4475 * cfg.h (class auto_edge_flag): Spell out the template-id of the
4476 base class in the initializer list. This is a workaround for
4477 building with older GCC.
4478 (class auto_bb_flag): Likewise.
4480 2018-08-30 Aaron Sawdey <acsawdey@linux.ibm.com>
4482 * config/rs6000/altivec.md (altivec_eq<mode>): Remove star.
4483 (altivec_vcmpequ<VI_char>_p): Remove star.
4484 * config/rs6000/rs6000-string.c (do_load_for_compare): Support
4486 (expand_strncmp_vec_sequence): New function.
4487 (emit_final_str_compare_vec): New function.
4488 (expand_strn_compare): Add support for vector strncmp.
4489 * config/rs6000/rs6000.opt (-mstring-compare-inline-limit): Change
4490 length specification to bytes.
4491 * config/rs6000/vsx.md (vsx_ld_elemrev_v16qi_internal): Remove star.
4492 (vcmpnezb_p): New pattern.
4493 * doc/invoke.texi (RS/6000 and PowerPC Options): Update documentation
4494 for option -mstring-compare-inline-limit.
4496 2018-08-30 Thiago Macieira <thiago.macieira@intel.com>
4498 * config/i386/i386.c (PTA_WESTMERE): Remove PTA_AES.
4499 (PTA_SKYLAKE): Add PTA_AES.
4500 (PTA_GOLDMONT): Likewise.
4502 2018-08-29 Jan Hubicka <jh@suse.cz>
4505 * lto-opts.c (lto_write_options): Always stream PIC/PIE mode.
4506 * lto-wrapper.c (merge_and_complain): Fix merging of PIC/PIE.
4508 2018-08-29 Jan Hubicka <jh@suse.cz>
4510 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not follow
4512 (hash_tree): Do not visit TYPE_STUB_DECL.
4513 * tree-streamer-out.c (write_ts_type_common_tree_pointers): Do not
4514 stream TYPE_STUB_DECL.
4515 * tree-streamer-in.c (lto_input_ts_type_common_tree_pointers): Likewise.
4516 * ipa-utils.h (type_with_linkage_p): Do not rely on TYPE_STUB_DECL
4517 after free_lang_data.
4518 (type_in_anonymous_namespace_p): Likewise.
4520 2018-08-29 Jan Hubicka <jh@suse.cz>
4522 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
4523 comment that it has to be even number.
4524 (class sreal): Change m_sig type to int32_t.
4525 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
4526 int64_t for temporary calculations.
4527 (sreal_verify_basics): Drop one bit from minimum and maximum.
4529 2018-08-30 Richard Biener <rguenther@suse.de>
4531 PR tree-optimization/87147
4532 * tree-ssa-sccvn.c (SSA_VISITED): New function.
4533 (visit_phi): When the degenerate result is from the backedge and
4534 we didn't visit its definition yet drop to VARYING.
4535 (do_rpo_vn): Properly mark blocks with incoming backedges as executable.
4537 2018-08-29 Jan Hubicka <jh@suse.cz>
4539 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not walk
4541 (hash_tree): Likewise.
4543 2018-08-29 Jan Hubicka <jh@suse.cz>
4545 * tree.c (find_decls_types_r): Walk also TYPE_NEXT_PTR_TO
4546 and TYPE_NEXT_REF_TO.
4548 2018-08-29 Jan Hubicka <jh@suse.cz>
4550 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
4551 comment that it has to be even number.
4552 (class sreal): Change m_sig type to int32_t.
4553 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
4554 int64_t for temporary calculations.
4555 (sreal_verify_basics): Drop one bit from minimum and maximum.
4557 2018-08-30 Tamar Christina <tamar.christina@arm.com>
4559 * config/aarch64/aarch64.c (aarch64_expand_movmem): Set TImode max.
4561 2018-08-30 Vlad Lazar <vlad.lazar@arm.com>
4564 * expmed.c (canonicalize_comparison): Use wi::sub instead of wi::add
4565 if to_add is negative.
4567 2018-08-29 Bernd Edlinger <bernd.edlinger@hotmail.de>
4570 * builtins.c (c_strlen): Improve range checks.
4572 2018-08-29 Martin Sebor <msebor@redhat.com>
4573 Jeff Law <law@redhat.com>
4575 PR tree-optimization/86714
4576 PR tree-optimization/86711
4577 * builtins.c (c_strlen): Add arguments to call to string_constant.
4578 * expr.c (string_constant): Add argument. Detect missing nul
4579 terminator and outermost declaration it's missing in.
4580 * expr.h (string_constant): Add argument.
4581 * fold-const.c (read_from_constant_string): Add arguments to call to
4583 (c_getstr): Likewise.
4584 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
4586 * tree-ssa-strlen.c (get_stridx): Likewise.
4588 2018-08-29 Jan Hubicka <jh@suse.cz>
4590 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers):
4591 Do not stream DECL_VINDEX.
4592 * tree-streamer-out.c (write_ts_function_decl_tree_pointers): Likewise.
4593 * tree.c (free_lang_data_in_decl): Clear DECL_VINDEX.
4594 (decl_function_context): Use DECL_VIRTUAL_P rather than DECL_VINDEX.
4596 2018-08-29 Richard Biener <rguenther@suse.de>
4598 * tree-ssa-sccvn.c (vuse_ssa_val): Return NULL for unvisited
4599 virtual operands that are not default defs to honor region
4601 (rpo_vn_valueize): Remove ineffective code here.
4603 2018-08-29 Richard Biener <rguenther@suse.de>
4605 PR tree-optimization/87132
4606 * tree-ssa-alias.c (get_continuation_for_phi): Do not translate
4607 when skipping defs reachable over backedges.
4609 2018-08-29 Richard Biener <rguenther@suse.de>
4611 * tree-core.h: Document use of deprecated_flag in SSA_NAME.
4612 * tree.h (SSA_NAME_POINTS_TO_READONLY_MEMORY): Define.
4613 * tree-into-ssa.c (pass_build_ssa::execute): Initialize
4614 function parameters SSA_NAME_POINTS_TO_READONLY_MEMORY from fnspec.
4615 * tree-ssa-sccvn.c (const_parms, init_const_parms): Remove.
4616 (vn_reference_lookup_3): Remove use of const_parms.
4617 (free_rpo_vn): Do not free const_parms.
4618 (do_rpo_vn): Do not call init_const_parms.
4619 * tree-ssa-alias.c (refs_may_alias_p_1): Honor
4620 SSA_NAME_POINTS_TO_READONLY_MEMORY.
4621 (call_may_clobber_ref_p_1): Likewise.
4623 2018-08-29 Alexander Monakov <amonakov@ispras.ru>
4626 * invoke.texi (Optimization Options): List -ftree-scev-cprop.
4628 (-ftree-scev-cprop): Document.
4630 2018-08-29 Jan Hubicka <jh@suse.cz>
4632 * sreal.h (normalize, normalize_up, normalize_down): Add new_sig/new_exp
4634 (sreal constructor): Update.
4635 * sreal.c (sreal:operator+, sreal:operator-, sreal:operator*,
4636 sreal:operator/): Update.
4638 2018-08-29 Martin Liska <mliska@suse.cz>
4640 * tree-switch-conversion.c (switch_conversion::expand):
4641 Strenghten assumption about gswitch statements.
4643 2018-08-29 Richard Biener <rguenther@suse.de>
4645 PR tree-optimization/87117
4646 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
4647 re-value-number released SSA VDEFs.
4649 2018-08-29 Richard Biener <rguenther@suse.de>
4651 PR tree-optimization/87126
4652 * tree-ssa-sccvn.c (vn_reference_insert): Remove assert.
4654 2018-08-28 Jim Wilson <jimw@sifive.com>
4656 * config/riscv/pic.md: Rewrite.
4657 * config/riscv/riscv.c (riscv_address_insns): Return cost of 3 for
4659 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Delete.
4660 (SOFTF, default_load, softload, softstore): New.
4662 2018-08-28 Jeff Law <law@redhat.com>
4664 * fold-const.c (fold_binary_loc): Remove recently added assert.
4666 2018-08-28 Joern Rennecke <joern.rennecke@riscy-ip.com>
4668 * genpreds.c (write_predicate_subfunction): Also add ATTRIBUTE_UNUSED
4669 to OP parmeter of generated function.
4671 2018-08-28 MCC CS <deswurstes@users.noreply.github.com>
4673 PR tree-optimization/87009
4674 * match.pd: Add boolean optimizations.
4676 2018-08-28 Martin Sebor <msebor@redhat.com>
4679 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
4680 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
4681 (pass_walloca::gate): Use it.
4682 (alloca_call_type): Same.
4683 (pass_walloca::execute): Same.
4684 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
4686 2018-08-28 David Malcolm <dmalcolm@redhat.com>
4688 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
4689 GCC_VERSION for usage of "__gcc_dump_printf__" format from
4692 2018-08-28 Richard Biener <rguenther@suse.de>
4694 PR tree-optimization/87124
4695 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
4696 constants before looking up avail.
4698 2018-08-28 Jakub Jelinek <jakub@redhat.com>
4701 * calls.c (maybe_warn_nonstring_arg): Punt early if
4702 warn_stringop_overflow is zero. Don't call get_range_strlen
4703 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
4704 Swap comparison operands to have constants on rhs. Only use
4705 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
4706 increment lenrng[0].
4708 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
4710 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
4711 use of tree_to_shwi. Remove duplicated test for the size being
4712 a whole number of bytes.
4714 2018-08-28 Richard Biener <rguenther@suse.de>
4716 PR tree-optimization/87117
4717 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
4718 Handle removed stmt without LHS (GIMPLE_NOP).
4720 2018-08-28 Richard Biener <rguenther@suse.de>
4722 PR tree-optimization/87117
4723 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
4724 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
4726 2018-08-28 Richard Biener <rguenther@suse.de>
4728 PR tree-optimization/87117
4729 * tree-ssa-pre.c (compute_avail): Do not make expressions
4730 with predicated values available.
4731 (get_expr_value_id): Assert we do not run into predicated value
4734 2018-08-28 Richard Biener <rguenther@suse.de>
4736 PR tree-optimization/87117
4737 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
4738 get virtual operands.
4739 (get_expr_operands): Handle STRING_CST like other decls.
4741 2018-08-28 Martin Liska <mliska@suse.cz>
4743 * tree.h: Update documentation of fndecl_built_in_p
4747 2018-08-27 Jeff Law <law@redhat.com>
4748 PR tree-optimization/87110
4749 * tree-ssa-dse.c (compute_trims): Handle non-constant
4752 2018-08-27 Martin Sebor <msebor@redhat.com>
4754 PR tree-optimization/86914
4755 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
4757 2018-08-27 Martin Sebor <msebor@redhat.com>
4759 PR tree-optimization/87112
4760 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
4761 the type of the bound argument.
4763 2018-08-27 Jeff Law <law@redhat.com>
4765 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
4766 type does not have a TYPE_SIZE_UNIT.
4768 2018-08-27 Steve Ellcey <sellcey@cavium.com>
4770 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
4771 with include of backend.h.
4773 2018-08-27 Richard Biener <rguenther@suse.de>
4775 PR tree-optimization/86927
4776 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
4777 use const cond reduction code.
4779 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
4781 PR tree-optimization/85758
4782 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
4784 2018-08-27 David Malcolm <dmalcolm@redhat.com>
4787 * diagnostic-show-locus.c (class layout_range): Update for
4788 conversion of show_caret_p to a tri-state.
4789 (layout_range::layout_range): Likewise.
4790 (make_range): Likewise.
4791 (layout::maybe_add_location_range): Likewise.
4792 (layout::should_print_annotation_line_p): Don't show annotation
4793 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
4794 (layout::get_state_at_point): Update for conversion of
4795 show_caret_p to a tri-state. Bail out early for
4796 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
4797 underlining or source colorization.
4798 (gcc_rich_location::add_location_if_nearby): Update for conversion
4799 of show_caret_p to a tri-state.
4800 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
4801 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
4803 (selftest::test_one_liner_labels): Likewise.
4804 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
4805 conversion of show_caret_p to a tri-state.
4806 * pretty-print.c (text_info::set_location): Likewise.
4807 * pretty-print.h (text_info::set_location): Likewise.
4808 * substring-locations.c (format_warning_n_va): Likewise.
4809 * tree-diagnostic.c (default_tree_printer): Likewise.
4810 * tree-pretty-print.c (newline_and_indent): Likewise.
4812 2018-08-27 David Malcolm <dmalcolm@redhat.com>
4815 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
4816 line above for line-insertion fix-it hints.
4817 (selftest::test_fixit_insert_containing_newline): Update the
4818 expected results, and add a test with line-numbering enabled.
4820 2018-08-27 Martin Liska <mliska@suse.cz>
4823 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
4824 params with DECL_HAS_VALUE_EXPR_P.
4826 2018-08-27 Martin Liska <mliska@suse.cz>
4828 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
4829 selected expansion strategy.
4831 2018-08-27 Martin Liska <mliska@suse.cz>
4833 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
4834 * builtins.c (is_builtin_fn): Likewise.
4835 * attribs.c (diag_attr_exclusions): Use new function
4836 fndecl_built_in_p and remove check for FUNCTION_DECL if
4838 (builtin_mathfn_code): Likewise.
4839 (fold_builtin_expect): Likewise.
4840 (fold_call_expr): Likewise.
4841 (fold_builtin_call_array): Likewise.
4842 (fold_call_stmt): Likewise.
4843 (set_builtin_user_assembler_name): Likewise.
4844 (is_simple_builtin): Likewise.
4845 * calls.c (gimple_alloca_call_p): Likewise.
4846 (maybe_warn_nonstring_arg): Likewise.
4847 * cfgexpand.c (expand_call_stmt): Likewise.
4848 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
4849 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
4850 (cgraph_node::verify_node): Likewise.
4851 * cgraphclones.c (build_function_decl_skip_args): Likewise.
4852 (cgraph_node::create_clone): Likewise.
4853 * config/arm/arm.c (arm_insert_attributes): Likewise.
4854 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
4855 * dse.c (scan_insn): Likewise.
4856 * expr.c (expand_expr_real_1): Likewise.
4857 * fold-const.c (operand_equal_p): Likewise.
4858 (fold_binary_loc): Likewise.
4859 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
4860 * gimple-low.c (lower_stmt): Likewise.
4861 * gimple-pretty-print.c (dump_gimple_call): Likewise.
4862 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
4864 * gimple.c (gimple_build_call_from_tree): Likewise.
4865 (gimple_call_builtin_p): Likewise.
4866 (gimple_call_combined_fn): Likewise.
4867 * gimplify.c (gimplify_call_expr): Likewise.
4868 (gimple_boolify): Likewise.
4869 (gimplify_modify_expr): Likewise.
4870 (gimplify_addr_expr): Likewise.
4871 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
4872 * ipa-cp.c (determine_versionability): Likewise.
4873 * ipa-fnsummary.c (compute_fn_summary): Likewise.
4874 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
4875 * ipa-split.c (visit_bb): Likewise.
4876 (split_function): Likewise.
4877 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
4878 * lto-cgraph.c (input_node): Likewise.
4879 * lto-streamer-out.c (write_symbol): Likewise.
4880 * omp-low.c (setjmp_or_longjmp_p): Likewise.
4881 (lower_omp_1): Likewise.
4882 * predict.c (strip_predict_hints): Likewise.
4883 * print-tree.c (print_node): Likewise.
4884 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
4885 * trans-mem.c (is_tm_irrevocable): Likewise.
4886 (is_tm_load): Likewise.
4887 (is_tm_simple_load): Likewise.
4888 (is_tm_store): Likewise.
4889 (is_tm_simple_store): Likewise.
4890 (is_tm_abort): Likewise.
4891 (tm_region_init_1): Likewise.
4892 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
4893 * tree-cfg.c (verify_gimple_call): Likewise.
4894 (move_stmt_r): Likewise.
4895 (stmt_can_terminate_bb_p): Likewise.
4896 * tree-eh.c (lower_eh_constructs_2): Likewise.
4897 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
4898 * tree-inline.c (remap_gimple_stmt): Likewise.
4899 (copy_bb): Likewise.
4900 (estimate_num_insns): Likewise.
4901 (fold_marked_statements): Likewise.
4902 * tree-sra.c (scan_function): Likewise.
4903 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
4904 (optimize_stack_restore): Likewise.
4905 (pass_fold_builtins::execute): Likewise.
4906 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
4907 (mark_all_reaching_defs_necessary_1): Likewise.
4908 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
4909 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
4910 (pass_forwprop::execute): Likewise.
4911 * tree-ssa-loop-im.c (stmt_cost): Likewise.
4912 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
4913 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
4914 * tree-ssa-strlen.c (get_string_length): Likewise.
4915 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
4916 (find_func_aliases_for_call): Likewise.
4917 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
4918 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
4919 * tree-tailcall.c (find_tail_calls): Likewise.
4920 * tree.c (need_assembler_name_p): Likewise.
4921 (free_lang_data_in_decl): Likewise.
4922 (get_call_combined_fn): Likewise.
4923 * ubsan.c (is_ubsan_builtin_p): Likewise.
4924 * varasm.c (incorporeal_function_p): Likewise.
4925 * tree.h (DECL_BUILT_IN): Remove and replace with
4927 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
4928 (fndecl_built_in_p): New.
4930 2018-08-27 Martin Liska <mliska@suse.cz>
4932 PR tree-optimization/86847
4933 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
4934 Dump also subtree probability.
4935 (switch_decision_tree::do_jump_if_equal): New function.
4936 (switch_decision_tree::emit_case_nodes): Handle special
4937 situations in balanced tree that can be emitted much simpler.
4938 Fix calculation of probabilities that happen in tree expansion.
4939 * tree-switch-conversion.h (struct cluster): Add
4941 (struct simple_cluster): Likewise.
4942 (struct case_tree_node): Add new function has_child.
4943 (do_jump_if_equal): New.
4945 2018-08-27 Martin Liska <mliska@suse.cz>
4947 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
4948 Add new argument to bit_test_cluster constructor.
4949 (bit_test_cluster::emit): Set bits really number of values
4951 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
4952 probability argument.
4953 * tree-switch-conversion.h (struct bit_test_cluster):
4954 Add m_handles_entire_switch.
4956 2018-08-27 Martin Liska <mliska@suse.cz>
4958 PR tree-optimization/86702
4959 * tree-switch-conversion.c (jump_table_cluster::emit):
4960 Make probabilities even for values in jump table
4961 according to number of cases handled.
4962 (switch_decision_tree::compute_cases_per_edge): Pass
4963 argument to reset_out_edges_aux function.
4964 (switch_decision_tree::analyze_switch_statement): Likewise.
4965 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
4968 2018-08-27 Martin Liska <mliska@suse.cz>
4970 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
4971 cfun argument explicitly.
4972 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
4973 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
4974 function gimple_switch_default_bb.
4975 (convert_switch_statements):
4977 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
4978 * stmt.c (label_to_block_fn): Use label_to_block and pass
4979 cfun argument explicitly and use gimple_switch_label_bb.
4980 (expand_case): Likewise.
4981 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
4982 cfun argument explicitly. Likewise.
4983 (make_edges_bb): Likewise.
4984 (make_cond_expr_edges): Likewise.
4985 (get_cases_for_edge): Likewise.
4986 (make_gimple_switch_edges): Likewise.
4987 (label_to_block_fn): Likewise.
4988 (label_to_block): Likewise.
4989 (make_goto_expr_edges): Likewise.
4990 (make_gimple_asm_edges): Likewise.
4991 (main_block_label): Likewise.
4992 (group_case_labels_stmt): Likewise.
4993 (find_taken_edge_computed_goto): Likewise.
4994 (find_taken_edge_switch_expr): Likewise.
4995 (gimple_verify_flow_info): Likewise.
4996 (gimple_redirect_edge_and_branch): Likewise.
4997 (gimple_switch_label_bb): New function.
4998 (gimple_switch_default_bb): Likewise.
4999 (gimple_switch_edge): Likewise.
5000 (gimple_switch_default_edge): Likewise.
5001 * tree-cfg.h (label_to_block_fn): Remove and replace ...
5002 (label_to_block): ... with this.
5003 (gimple_switch_label_bb): New.
5004 (gimple_switch_default_bb): Likewise.
5005 (gimple_switch_edge): Likewise.
5006 (gimple_switch_default_edge): Likewise.
5007 * tree-cfgcleanup.c (convert_single_case_switch): Use
5008 new gimple functions and pass new argument to label_to_block.
5009 (cleanup_control_flow_bb):
5010 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
5011 cfun argument explicitly.
5012 (make_eh_edges): Likewise.
5013 (redirect_eh_dispatch_edge): Likewise.
5014 (lower_resx): Likewise.
5015 (lower_eh_dispatch): Likewise.
5016 (maybe_remove_unreachable_handlers): Likewise.
5017 (unsplit_eh): Likewise.
5018 (cleanup_empty_eh): Likewise.
5019 (verify_eh_edges): Likewise.
5020 (verify_eh_dispatch_edge): Likewise.
5021 * tree-ssa-dom.c (record_edge_info): Likewise.
5022 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
5023 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
5024 (thread_through_normal_block): Likewise.
5025 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
5026 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
5027 * tree-switch-conversion.c (switch_conversion::collect): Use new
5029 (switch_conversion::check_final_bb): Likewise.
5030 (switch_conversion::gather_default_values): Pass new argument
5032 (switch_conversion::build_constructors): Likewise.
5033 (switch_decision_tree::compute_cases_per_edge): Use new
5034 gimple_switch_edge function.
5035 (switch_decision_tree::analyze_switch_statement): Pass new argument
5037 (switch_decision_tree::try_switch_expansion): Use
5038 gimple_switch_default_edge.
5039 * tree-vrp.c (find_switch_asserts): Pass new argument
5041 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
5042 (vr_values::simplify_switch_using_ranges): Likewise.
5044 2018-08-27 Richard Biener <rguenther@suse.de>
5046 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
5047 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
5049 * tree-ssa-sccvn.h (struct vn_pval): New structure.
5050 (struct vn_nary_op_s): Add unwind_to member. Add
5051 predicated_values flag and put result into a union together
5052 with a linked list of vn_pval.
5053 (struct vn_ssa_aux): Add name member to make maintaining
5054 a map of SSA name to vn_ssa_aux possible. Remove no longer
5055 needed info, dfsnum, low, visited, on_sccstack, use_processed
5056 and range_info_anti_range_p members.
5057 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
5058 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
5060 (vn_valueize): New global.
5061 (vn_context_bb): Likewise.
5062 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
5063 VN_INFO_PTR_INFO): Remove.
5064 * tree-ssa-sccvn.c: ... (rewrite)
5065 (pass_fre::execute): For -O2+ initialize loops and run
5066 RPO VN in optimistic mode (iterating). For -O1 and -Og
5067 run RPO VN in non-optimistic mode.
5068 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
5069 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
5070 * doc/invoke.texi (sccvn-max-scc-size): Remove.
5071 (rpo-vn-max-loop-depth): Document.
5072 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
5073 when valuezing the VUSE signals we walked out of the region.
5074 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
5075 (phi_translate): Set VN context block to use for availability
5077 (compute_avail): Likewise.
5078 (pre_valueize): New function.
5079 (pass_pre::execute): Adjust to the RPO VN API.
5081 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
5082 (propagate_constants_for_unrolling): Remove.
5083 (tree_unroll_loops_completely): Perform value-numbering
5084 on the unrolled bodies loop parent.
5086 2018-08-27 Richard Biener <rguenther@suse.de>
5088 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
5089 for partial antic compute.
5091 2018-08-27 Jakub Jelinek <jakub@redhat.com>
5093 PR rtl-optimization/87065
5094 * combine.c (simplify_if_then_else): Formatting fix.
5095 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
5097 (known_cond): Don't return const_true_rtx for vector modes. Use
5098 CONST0_RTX instead of const0_rtx. Formatting fixes.
5100 2018-08-27 Martin Liska <mliska@suse.cz>
5102 PR gcov-profile/87069
5103 * gcov.c (process_file): Record files already processed
5104 and warn about a file being processed multiple times.
5106 2018-08-27 Martin Liska <mliska@suse.cz>
5109 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5110 Set default values for x_aarch64_*_string strings.
5111 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
5112 prefix. For -mabi do not print '=ABI' in help and use
5113 <option_value> format for -msve-vector-bits and -moverride
5116 2018-08-26 Jeff Law <law@redhat.com>
5118 * config/mips/frame-header-opt.c: Include "backend.h" rather than
5121 2018-08-26 Marek Polacek <polacek@redhat.com>
5123 PR c++/87029, Implement -Wredundant-move.
5124 * doc/invoke.texi: Document -Wredundant-move.
5126 2018-08-25 Martin Sebor <msebor@redhat.com>
5128 PR tree-optimization/87059
5129 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
5130 to the same type as the other.
5131 * fold-const.c (fold_binary_loc): Assert expectation.
5133 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
5135 * config/darwin.c (machopic_legitimize_pic_address): Clean up
5136 extraneous parentheses, dead code section and formatting.
5138 2018-08-24 David Malcolm <dmalcolm@redhat.com>
5141 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
5142 wide enough for jumps in the line-numbering to be visible.
5143 (layout::print_gap_in_line_numbering): New member function.
5144 (layout::calculate_line_spans): When using line numbering, merge
5145 line spans that are only 1 line apart.
5146 (diagnostic_show_locus): When printing line numbers, show gaps in
5147 line numbering directly, rather than printing headers.
5148 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
5149 line-numbering with multiple line spans.
5150 (selftest::test_fixit_insert_containing_newline_2): Add test of
5151 line-numbering, in which the spans are close enough to be merged.
5153 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
5155 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
5156 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
5157 * tree-vrp.c (range_is_nonnull): Remove.
5158 (range_includes_zero_p): Accept value_range instead of min/max.
5159 (extract_range_from_binary_expr_1): Do not early bail on
5161 Use range_includes_zero_p instead of range_is_nonnull.
5162 (extract_range_from_unary_expr): Use range_includes_zero_p instead
5163 of range_is_nonnull.
5164 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
5165 special case VR_ANTI_RANGE.
5166 (vrp_finalize): Same.
5167 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
5169 (range_is_nonnull): Remove.
5170 * vr-values.c (vrp_stmt_computes_nonzero): Use
5171 range_includes_zero_p instead of range_is_nonnull.
5172 (extract_range_basic): Pass value_range to range_includes_zero_p
5173 instead of range_is_nonnull.
5175 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
5177 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
5178 * emit-rtl.h (rtl_data): Remove return_bnd.
5179 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
5180 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
5181 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
5182 (POINTER_BOUNDS_MODE): Remove definition.
5183 (make_pointer_bounds_mode): Remove.
5184 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
5185 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
5186 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
5187 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
5188 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
5189 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
5190 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
5192 * config/i386/i386-modes.def (BND32, BND64): Remove.
5193 * config/i386/i386.c (dbx_register_map): Remove bound registers.
5194 (dbx64_register_map): Ditto.
5195 (svr4_dbx_register_map): Ditto.
5196 (indirect_thunk_bnd_needed): Remove.
5197 (indirect_thunks_bnd_used): Ditto.
5198 (indirect_return_bnd_needed): Ditto.
5199 (indirect_return_via_cx_bnd): Ditto.
5200 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
5201 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
5202 (output_indirect_thunk): Ditto. Remove need_prefix argument.
5203 (output_indirect_thunk_function): Remove handling of
5204 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
5205 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
5206 (ix86_save_reg): Remove handling of crtl->return_bnd.
5207 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
5208 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
5209 and UNSPEC_BNDLX_ADDR.
5210 (ix86_output_indirect_branch_via_reg): Remove handling of
5211 indirect_thunk_prefix_bnd.
5212 (ix86_output_indirect_branch_via_push): Ditto.
5213 (ix86_output_function_return): Ditto.
5214 (ix86_output_indirect_function_return): Ditto.
5215 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
5216 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
5217 (CALL_USED_REGISTERS): Ditto.
5218 (REG_ALLOC_ORDER): Update for removal of bound registers.
5219 (HI_REGISTER_NAMES): Ditto.
5220 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
5221 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
5222 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
5223 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
5224 (FIRST_PSEUDO_REG): Update.
5225 (BND): Remove mode iterator.
5226 * config/i386/predicates.md (bnd_mem_operator): Remove.
5228 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
5230 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
5233 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
5235 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
5236 the case in which the permute needs only a single element and
5237 repeats for every vector of the result. Extend that case to
5238 handle variable-length vectors.
5239 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
5241 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
5244 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
5245 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
5247 2018-08-24 Richard Biener <rguenther@suse.de>
5249 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
5250 bb_flags_allocated members.
5251 (auto_flag): New RAII class for allocating flags.
5252 (auto_edge_flag): New RAII class for allocating edge flags.
5253 (auto_bb_flag): New RAII class for allocating bb flags.
5254 * cfgloop.c (verify_loop_structure): Allocate temporary edge
5256 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
5257 in favor of temporarily allocated BB flag.
5258 * hsa-brig.c: Re-order includes.
5259 * hsa-dump.c: Likewise.
5260 * hsa-regalloc.c: Likewise.
5261 * print-rtl.c: Likewise.
5262 * profile-count.c: Likewise.
5264 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
5267 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
5270 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
5273 * wide-int-range.cc (wide_int_range_div): Do not ignore result
5274 from wide_int_range_multiplicative_op.
5276 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5278 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
5281 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
5283 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
5286 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
5288 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
5289 full GENERIC expression used for replacement.
5291 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
5293 * tree-vrp.c (abs_extent_range): Remove.
5294 (extract_range_into_wide_ints): Pass wide ints by reference.
5295 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
5296 Pass wide ints by reference in all calls to
5297 extract_range_into_wide_ints.
5298 * wide-int-range.cc (wide_int_range_div): New.
5299 * wide-int-range.h (wide_int_range_div): New.
5300 (wide_int_range_includes_zero_p): New.
5301 (wide_int_range_zero_p): New.
5303 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
5305 * config/aarch64/aarch64.md (arches): New enum.
5306 (arch): New enum attr.
5307 (arch_enabled): New attr.
5308 (enabled): Now uses arch_enabled only.
5309 (simd, sve, fp16): Removed attribute.
5310 (fp): Attr now defined in terms of 'arch'.
5311 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
5312 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
5313 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
5314 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
5315 attributes into 'arch'.
5316 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
5317 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
5318 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
5319 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
5320 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
5321 'simd' attribute into 'arch'.
5322 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
5323 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
5324 Convert use of 'fp' attribute to 'arch'.
5325 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
5326 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
5328 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
5329 (different modes) Merge 'fp' and 'simd' into 'arch'.
5330 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
5333 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
5335 PR rtl-optimization/87026
5336 * expmed.c (canonicalize_comparison): If we can no longer create
5337 pseudoregisters, don't.
5339 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
5342 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
5344 * config/arm/arm.c (speculation_barrier_libfunc): New static
5346 (arm_init_libfuncs): Initialize it.
5347 (arm_emit_speculation_barrier): New function.
5348 * config/arm/arm.md (speculation_barrier): Call
5349 arm_emit_speculation_barrier for architectures that do not have
5351 (speculation_barrier_insn): Only match on Armv7 or later.
5353 2018-08-23 Richard Biener <rguenther@suse.de>
5356 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
5359 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
5361 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
5362 of single-vector TBLs.
5363 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
5366 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
5369 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
5370 aarch64_evpc_tbl guard.
5372 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
5374 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
5377 2018-08-22 Martin Sebor <msebor@redhat.com>
5380 * tree-pretty-print.c (pretty_print_string): Add argument.
5381 (dump_generic_node): Call to pretty_print_string with string size.
5383 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
5385 PR rtl-optimization/86771
5386 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
5387 of two SETs into those two SETs, one to be placed at i2, if that SETs
5388 destination is modified between i2 and i3.
5390 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
5392 PR tree-optimization/86725
5393 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
5395 (vect_analyze_scalar_cycles_1): Check it.
5397 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
5399 PR tree-optimization/86725
5400 * tree-vect-loop.c (vect_is_simple_reduction): When treating
5401 an outer loop phi as a double reduction, make sure that the
5402 single user of the phi result is an inner loop phi.
5404 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
5406 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
5407 grouped stores with gaps to a strided group.
5409 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
5411 * tree-vect-stmts.c (get_group_load_store_type)
5412 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
5413 first statement in a group.
5415 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
5417 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
5418 the sequence used in gcc/gcc.c.
5420 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
5423 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
5426 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
5428 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
5429 Darwin10-specific unwinder-shim.
5430 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
5431 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
5432 New to cater for Darwin10 Rosetta.
5434 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
5436 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
5437 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
5439 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
5444 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
5445 (dwarf2out_switch_text_section): Generate a local label for the
5446 second function sub-section and apply it as the second FDE start
5448 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
5449 second sub-section start.
5451 2018-08-22 Richard Biener <rguenther@suse.de>
5453 PR tree-optimization/86988
5454 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
5456 2018-08-22 Richard Biener <rguenther@suse.de>
5458 PR tree-optimization/86945
5459 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
5461 2018-08-22 Alexandre Oliva <oliva@adacore.com>
5463 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
5464 a comment about how uses of r2 for .sdata2 come about.
5466 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
5468 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
5470 2018-08-21 Marek Polacek <polacek@redhat.com>
5472 PR c++/86981, Implement -Wpessimizing-move.
5473 * doc/invoke.texi: Document -Wpessimizing-move.
5475 2018-08-21 Jan Hubicka <jh@suse.cz>
5477 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
5478 * tree.h (is_redundant_typedef): Remove.
5479 * dwarf2out.c (is_redundant_typedef): Turn into static function.
5481 2018-08-21 Jan Hubicka <jh@suse.cz>
5483 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
5486 2018-08-21 Tamar Christina <tamar.christina@arm.com>
5488 * expmed.c (extract_low_bits): Reject invalid subregs early.
5490 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
5493 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
5496 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
5498 * config/vxworks.h: Guard vxworks_asm_out_constructor and
5499 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
5500 * config/vxworks.c: Likewise.
5502 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
5504 * config/vxworks.c: Set targetm.have_ctors_dtors
5505 if HAVE_INITFINI_ARRAY_SUPPORT.
5506 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
5507 if HAVE_INITFINI_ARRAY_SUPPORT.
5509 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
5511 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
5512 default search path for VxWorks < 7.
5514 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
5516 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
5517 (get_format_string): Refer to c_getstr.
5519 2018-08-21 Tom de Vries <tdevries@suse.de>
5521 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
5522 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
5523 (debuginfo_early_stop): Declare.
5524 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
5525 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
5526 (debuginfo_early_stop): New function.
5527 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
5528 and debuginfo_early_stop.
5529 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
5530 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
5531 (general_init): Call debuginfo_early_init.
5532 (finalize): Call debuginfo_fini.
5533 (do_compile): Call debuginfo_init.
5534 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
5536 (@item -fdump-debug, @item -fdump-earlydebug): Add.
5538 2018-08-21 Tom de Vries <tdevries@suse.de>
5540 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
5541 flag_dump_noaddr and flag_dump_unnumbered.
5543 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
5545 * wide-int-range.cc (wide_int_range_abs): New.
5546 (wide_int_range_order_set): Rename from wide_int_range_min_max.
5547 * wide-int-range.h (wide_int_range_abs): New.
5548 (wide_int_range_min_max): New.
5549 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
5550 case to call wide_int_range_abs.
5551 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
5552 (extract_range_from_abs_expr): Delete.
5554 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
5557 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
5558 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
5561 2018-08-20 Nathan Sidwell <nathan@acm.org>
5562 Jeff Law <law@redhat.com>
5564 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
5565 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
5567 2018-08-20 David Malcolm <dmalcolm@redhat.com>
5570 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
5571 (decl_attributes): Likewise.
5572 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
5574 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
5575 * diagnostic-core.h (class auto_diagnostic_group): New class.
5576 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
5577 (diagnostic_report_diagnostic): Handle the first diagnostics within
5579 (emit_diagnostic): Add auto_diagnostic_group instance.
5581 (inform_n): Likewise.
5582 (warning): Likewise.
5583 (warning_at): Likewise.
5584 (warning_n): Likewise.
5585 (pedwarn): Likewise.
5586 (permerror): Likewise.
5588 (error_n): Likewise.
5589 (error_at): Likewise.
5591 (fatal_error): Likewise.
5592 (internal_error): Likewise.
5593 (internal_error_no_backtrace): Likewise.
5594 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
5595 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
5596 * diagnostic.h (struct diagnostic_context): Add fields
5597 "diagnostic_group_nesting_depth",
5598 "diagnostic_group_emission_count", "begin_group_cb",
5600 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
5601 Add auto_diagnostic_group instance(s).
5602 (find_explicit_erroneous_behavior): Likewise.
5603 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
5604 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
5605 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
5606 (gimplify_va_arg_expr): Likewise.
5607 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
5608 (HSA_SORRY_AT): Likewise.
5609 * ipa-devirt.c (compare_virtual_tables): Likewise.
5610 (warn_odr): Likewise.
5611 * multiple_target.c (expand_target_clones): Likewise.
5612 * opts-common.c (cmdline_handle_error): Likewise.
5613 * reginfo.c (globalize_reg): Likewise.
5614 * substring-locations.c (format_warning_n_va): Likewise.
5615 * tree-inline.c (expand_call_inline): Likewise.
5616 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
5617 * tree-ssa-loop-niter.c
5618 (do_warn_aggressive_loop_optimizations): Likewise.
5619 * tree-ssa-uninit.c (warn_uninit): Likewise.
5620 * tree.c (warn_deprecated_use): Likewise.
5622 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
5625 * config/i386/i386.md (eh_return): Always update EH return
5626 address in word_mode.
5628 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
5630 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
5631 TARGET_SPLIT_COMPLEX_ARG is defined.
5633 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
5635 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
5637 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
5640 * expr.c (expand_assignment): Assert that bitpos is positive.
5641 (store_field): Likewise
5642 (expand_expr_real_1): Make sure that bitpos is positive.
5643 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
5646 2018-08-20 Nathan Sidwell <nathan@acm.org>
5648 * Makefile.in (CPP_ID_DATA_H): Delete.
5649 (CPP_INTERNAL_H): Don't add it.
5650 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
5651 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
5653 2018-08-20 Richard Biener <rguenther@suse.de>
5655 PR tree-optimization/78655
5656 * tree-vrp.c (extract_range_from_binary_expr_1): Make
5657 pointer + offset nonnull if either operand is nonnull work.
5659 2018-08-20 Tom de Vries <tdevries@suse.de>
5661 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
5662 unless the referenced die describes the added property using
5663 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
5664 Otherwise, add a DW_AT_location to the referenced die.
5666 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
5669 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
5670 register_operand when calling ix86_set_reg_reg_cost.
5671 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
5672 Set *total to 0 for operands that satisfy x86_64_immediate_operand
5673 predicate and to 1 otherwise.
5675 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
5677 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
5678 emit a diagnostic that it is not supported and reset the option.
5679 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
5680 supported and consume the option. (ASM_FINAL_SPEC): New.
5682 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
5684 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
5687 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
5689 C-SKY port: Documentation
5691 * doc/extend.texi (C-SKY Function Attributes): New section.
5692 * doc/invoke.texi (Option Summary): Add C-SKY options.
5693 (C-SKY Options): New section.
5694 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
5696 2018-08-17 Jojo <jijie_rong@c-sky.com>
5697 Huibin Wang <huibin_wang@c-sky.com>
5698 Sandra Loosemore <sandra@codesourcery.com>
5699 Chung-Lin Tang <cltang@codesourcery.com>
5701 C-SKY port: Backend implementation
5703 * config/csky/*: New.
5704 * common/config/csky/*: New.
5706 2018-08-17 Jojo <jijie_rong@c-sky.com>
5707 Huibin Wang <huibin_wang@c-sky.com>
5708 Sandra Loosemore <sandra@codesourcery.com>
5709 Chung-Lin Tang <cltang@codesourcery.com>
5710 Andrew Jenner <andrew@codesourcery.com>
5712 C-SKY port: Configury
5714 * config.gcc (csky-*-*): New.
5715 * configure.ac: Add csky to targets for dwarf2 debug_line support.
5716 * configure: Regenerated.
5718 2018-08-17 David Malcolm <dmalcolm@redhat.com>
5720 * dump-context.h: Include "dumpfile.h".
5721 (dump_context::dump_printf_va): Convert final param from va_list
5722 to va_list *. Convert from ATTRIBUTE_PRINTF to
5723 ATTRIBUTE_GCC_DUMP_PRINTF.
5724 (dump_context::dump_printf_loc_va): Likewise.
5725 * dumpfile.c: Include "stringpool.h".
5726 (make_item_for_dump_printf_va): Delete.
5727 (make_item_for_dump_printf): Delete.
5728 (class dump_pretty_printer): New class.
5729 (dump_pretty_printer::dump_pretty_printer): New ctor.
5730 (dump_pretty_printer::emit_items): New member function.
5731 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
5733 (dump_pretty_printer::emit_item): New member function.
5734 (dump_pretty_printer::stash_item): New member function.
5735 (dump_pretty_printer::format_decoder_cb): New member function.
5736 (dump_pretty_printer::decode_format): New member function.
5737 (dump_context::dump_printf_va): Reimplement in terms of
5738 dump_pretty_printer.
5739 (dump_context::dump_printf_loc_va): Convert final param from va_list
5741 (dump_context::begin_scope): Reimplement call to
5742 make_item_for_dump_printf.
5743 (dump_printf): Update for change to dump_printf_va.
5744 (dump_printf_loc): Likewise.
5745 (selftest::test_capture_of_dump_calls): Convert "stmt" from
5746 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
5747 with %T, %E, and %G.
5748 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
5749 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
5750 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
5751 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
5752 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
5753 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
5754 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
5755 within a dump_printf_loc call to "%wu".
5756 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
5757 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
5758 missing space after "=".
5759 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
5760 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
5761 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
5762 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
5763 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
5764 duplicate "vectorized" from message.
5766 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
5768 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
5769 polyNxK_t element's TYPE_STRING_FLAG.
5771 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
5773 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
5774 (they were unnamed before). Fix comments.
5776 2018-08-17 Nathan Sidwell <nathan@acm.org>
5778 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
5780 2018-08-17 Richard Biener <rguenther@suse.de>
5782 PR tree-optimization/86841
5783 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
5785 2018-08-17 Martin Liska <mliska@suse.cz>
5787 * common.opt: Remove Warn, Init and Report for options with
5788 Ignore/Deprecated flag. Warning is done automatically for
5790 * config/i386/i386.opt: Likewise.
5791 * config/ia64/ia64.opt: Likewise.
5792 * config/rs6000/rs6000.opt: Likewise.
5793 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
5794 Remove usage of flag_check_pointer_bounds.
5795 * lto-wrapper.c (merge_and_complain): Do not handle
5796 OPT_fcheck_pointer_bounds.
5797 (append_compiler_options): Likewise.
5798 * opt-functions.awk: Do not handle Deprecated.
5799 * optc-gen.awk: Check that Var, Report and Init are not
5800 used for an option with Ignore/Deprecated flag.
5801 * opts-common.c (decode_cmdline_option): Do not report
5803 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
5805 * opts.h (struct cl_option): Remove cl_deprecated flag.
5806 (CL_ERR_DEPRECATED): Remove error enum value.
5808 2018-08-17 Richard Biener <rguenther@suse.de>
5811 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
5812 across a va-arg-pack using call adjust its return value accordingly.
5814 2018-08-16 Martin Sebor <msebor@redhat.com>
5816 PR tree-optimization/86853
5817 * gimple-ssa-sprintf.c (struct format_result): Rename member.
5818 (struct fmtresult): Add member and initialize it in ctors.
5819 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
5820 (format_string): Handle %S the same as %ls. Set MAYFAIL.
5821 (format_directive): Set POSUNDER4K when MAYFAIL is set.
5822 (parse_directive): Handle %C same as %c.
5823 (sprintf_dom_walker::compute_format_length): Adjust.
5824 (is_call_safe): Adjust.
5826 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
5828 * builtins.c (c_strlen): Add new parameter eltsize. Use it
5829 for determining how to count the elements.
5830 * builtins.h (c_strlen): Adjust prototype.
5831 * expr.c (string_constant): Add new parameter mem_size.
5832 Set *mem_size appropriately.
5833 * expr.h (string_constant): Adjust protoype.
5834 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
5835 * gimple-fold.h (get_range_strlen): Adjust prototype.
5836 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
5837 (format_string): Call get_string_length with eltsize.
5839 2018-08-16 David Malcolm <dmalcolm@redhat.com>
5841 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
5842 to emit the span, rather than setting it as the prefix.
5844 2018-08-16 David Malcolm <dmalcolm@redhat.com>
5846 * diagnostic-show-locus.c (layout::start_annotation_line): Add
5847 "margin_char" parameter, defaulting to space. Use it in place
5848 of pp_space for the initial part of the margin.
5849 (layout::print_leading_fixits): Use '+' when filling the margin
5850 of line-insertion fix-it hints.
5852 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
5854 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
5857 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
5859 * config/rs6000/altivec.md: Don't set length attribute to the default
5861 * config/rs6000/darwin.md: Ditto.
5862 * config/rs6000/dfp.md: Ditto.
5863 * config/rs6000/htm.md: Ditto.
5864 * config/rs6000/rs6000.md: Ditto.
5865 * config/rs6000/sync.md: Ditto.
5866 * config/rs6000/vsx.md: Ditto.
5868 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
5870 * config/rs6000/altivec.md: Don't set length attribute to the default
5871 value, for branch instructions.
5872 * config/rs6000/darwin.md: Ditto.
5873 * config/rs6000/rs6000.md: Ditto.
5875 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
5877 * config/rs6000/rs6000.md (length): Always define as const_int 4.
5878 (unnamed conditional branch define_insn): Set length to 4 or 8
5879 depending on offset.
5880 (<bd>_<mode>): Similar, for alternative 0.
5881 (<bd>tf_<mode>): Ditto.
5883 2018-08-16 Tamar Christina <tamar.christina@arm.com>
5885 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
5887 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
5889 * doc/rtl.texi: Replace old RTX class names with new names.
5892 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
5894 * expmed.h (canonicalize_comparison): New declaration.
5895 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
5896 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
5897 * optabs.c (prepare_cmp_insn): Likewise.
5898 * rtl.h (unsigned_condition_p): New function which checks if a
5899 comparison operator is unsigned.
5901 2018-08-16 Nathan Sidwell <nathan@acm.org>
5903 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
5904 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
5906 2018-08-16 Tamar Christina <tamar.christina@arm.com>
5909 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
5910 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
5911 (mov<mov>): ..this and enable unconditionally.
5913 2018-08-16 Tamar Christina <tamar.christina@arm.com>
5915 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
5917 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
5919 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
5920 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
5921 "Common" with "Target".
5923 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
5925 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
5926 * doc/invoke.texi (mmitigate-rop): Remove.
5927 * config/i386/i386.c: Do not include "regrename.h".
5928 (ix86_rop_should_change_byte_p, reg_encoded_number)
5929 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
5931 (ix86_reorg): Remove call to ix86_mitigate_rop.
5932 * config/i386/i386.md (attr "modrm_class"): Remove.
5933 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
5934 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
5935 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
5937 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
5939 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
5940 allow folding of mergeh() and mergel() for the float and double types.
5941 (fold_mergehl_helper): Rework to handle building a permute tree
5944 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
5946 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
5949 2018-08-15 David Malcolm <dmalcolm@redhat.com>
5951 * common.opt (fdiagnostics-show-labels): New option.
5952 * diagnostic-show-locus.c (class layout_range): Add field
5954 (class layout): Add field "m_show_labels_p".
5955 (layout_range::layout_range): Add param "label" and use it to
5957 (make_range): Pass in NULL for new "label" param of layout_range's
5959 (layout::layout): Initialize m_show_labels_p.
5960 (layout::maybe_add_location_range): Pass in loc_range->m_label
5961 when constructing layout_range instances.
5962 (struct line_label): New struct.
5963 (layout::print_any_labels): New member function.
5964 (layout::print_line): Call it if label-printing is enabled.
5965 (selftest::test_one_liner_labels): New test.
5966 (selftest::test_diagnostic_show_locus_one_liner): Call it.
5967 * diagnostic.c (diagnostic_initialize): Initialize
5968 context->show_labels_p.
5969 * diagnostic.h (struct diagnostic_context): Add field
5971 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
5972 -fno-diagnostics-show-labels.
5973 * dwarf2out.c (gen_producer_string): Add
5974 OPT_fdiagnostics_show_labels to the ignored options.
5975 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
5977 (gcc_rich_location::maybe_add_expr): Likewise.
5978 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
5979 label" param, defaulting to NULL.
5980 (gcc_rich_location::add_expr): Add "label" param.
5981 (gcc_rich_location::maybe_add_expr): Likewise.
5982 (class text_range_label): New class.
5983 (class range_label_for_type_mismatch): New class.
5984 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
5985 of format_warning_va.
5986 (fmtwarn_n): Likewise for new params of format_warning_n_va.
5987 * lto-wrapper.c (merge_and_complain): Add
5988 OPT_fdiagnostics_show_labels to the "pick one setting" options.
5989 (append_compiler_options): Likewise to the dropped options.
5990 (append_diag_options): Likewise to the passed-on options.
5991 * opts.c (common_handle_option): Handle the new option.
5992 * selftest-diagnostic.c
5993 (test_diagnostic_context::test_diagnostic_context): Enable
5995 * substring-locations.c: Include "gcc-rich-location.h".
5996 (format_warning_n_va): Add "fmt_label" and "param_label" params
5997 and use them as appropriate.
5998 (format_warning_va): Add "fmt_label" and "param_label" params,
5999 passing them on to format_warning_n_va.
6000 (format_warning_at_substring): Likewise.
6001 (format_warning_at_substring_n): Likewise.
6002 * substring-locations.h (format_warning_va): Add "fmt_label" and
6003 "param_label" params.
6004 (format_warning_n_va): Likewise.
6005 (format_warning_at_substring): Likewise.
6006 (format_warning_at_substring_n): Likewise.
6007 * toplev.c (general_init): Initialize global_dc->show_labels_p.
6009 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
6012 * builtins.c (expand_builtin_memcmp): Do not expand the call
6013 when overflow is detected.
6015 2018-08-15 Martin Sebor <msebor@redhat.com>
6017 PR tree-optimization/71625
6018 * config/aarch64/aarch64-builtins.c
6019 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
6021 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
6023 * config/s390/s390.c (s390_reorg): Remove loop.
6025 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
6028 (darwin_function_switched_text_sections): Delete.
6029 * gcc/config/darwin.h
6030 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
6032 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
6035 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
6036 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
6037 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
6039 2018-08-15 Martin Liska <mliska@suse.cz>
6041 PR tree-optimization/86925
6042 * predict.c (expr_expected_value_1): When taking
6043 later predictor, assign also probability.
6044 Use fold_build2_initializer_loc in order to fold
6045 the expression in -frounding-math.
6047 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
6049 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
6051 (expand_vec_perm_1): Try the new method.
6053 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
6056 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
6057 Check whether lra_live_max_point is 0 before dividing.
6059 2018-08-14 Martin Sebor <msebor@redhat.com>
6061 PR tree-optimization/86650
6062 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
6063 (vrp_prop::check_mem_ref): Same.
6065 2018-08-13 Liu Hao <lh_mouse@126.com>
6067 * pretty-print.c (eat_esc_sequence): Swap the foreground and
6068 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
6069 and clear it thereafter, as it only works for DBCS.
6071 2018-08-13 Liu Hao <lh_mouse@126.com>
6073 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
6074 handle returned by _get_osf_handle().
6076 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
6078 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
6079 for folding vec_perm.
6081 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
6083 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
6084 Add support for gimple-folding of vec_pack() and vec_unpack()
6087 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
6089 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
6090 vec_xst variants to the list.
6091 (rs6000_gimple_fold_builtin): Add support for folding unaligned
6092 vector loads and stores.
6094 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
6096 * config.gcc (rs6000-ibm-aix4.x): Delete.
6097 (rs6000-ibm-aix5.1): Delete.
6098 (rs6000-ibm-aix5.2): Delete.
6099 (rs6000-ibm-aix5.3): Delete.
6100 * config/rs6000/aix43.h: Delete.
6101 * config/rs6000/aix51.h: Delete.
6102 * config/rs6000/aix52.h: Delete.
6103 * config/rs6000/t-aix43: Delete.
6105 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
6107 * config/s390/s390.c (s390_decompose_constant_pool_ref):
6109 (s390_decompose_address): Factor out constant pool ref
6112 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
6114 * config/nds32/nds32-predicates.c
6115 (nds32_can_use_bclr_p): Change return type as bool.
6116 (nds32_can_use_bset_p): Ditto.
6117 (nds32_can_use_btgl_p): Ditto.
6118 (nds32_can_use_bitci_p): Ditto.
6119 * config/nds32/nds32-protos.h
6120 (nds32_can_use_bclr_p): Change declaration.
6121 (nds32_can_use_bset_p): Ditto.
6122 (nds32_can_use_btgl_p): Ditto.
6123 (nds32_can_use_bitci_p): Ditto.
6125 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
6127 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
6128 Support -msched-prolog-epilog option.
6129 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
6131 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
6133 * common/config/nds32/nds32-common.c
6134 (nds32_option_optimization_table): Enalbe -malways-align.
6136 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
6138 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
6140 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
6141 OPT_misr_secure_ case.
6142 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
6143 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
6144 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
6146 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
6147 (nds32_isr_info): New field security_level.
6148 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
6149 * config/nds32/nds32.md (return_internal): Consider critical attribute.
6150 * config/nds32/nds32.opt (misr-secure): New option.
6151 * config/nds32/nds32_init.inc: New file.
6152 * config/nds32/nds32_isr.h: New file.
6154 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
6156 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
6157 Update comment for atomic instructions.
6158 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
6159 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
6161 (atomic_loaddi): Revise fence expansion to only emit fence prior to
6162 load for __ATOMIC_SEQ_CST model.
6163 (atomic_loaddi_1): Remove float register target.
6164 (atomic_storedi): Handle CONST_INT values.
6165 (atomic_storedi_1): Remove float register source. Add special case
6167 (memory_barrier): New expander and insn.
6169 2018-08-11 Jakub Jelinek <jakub@redhat.com>
6171 PR tree-optimization/86835
6172 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
6173 new_stmt after def_gsi, make sure to insert new_square_stmt after
6174 that stmt, not 2 stmts before it.
6176 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
6179 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
6182 2018-08-10 Martin Liska <mliska@suse.cz>
6185 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
6187 * builtins.c (expand_builtin_expect_with_probability):
6189 (expand_builtin_expect_with_probability): New function.
6190 (build_builtin_expect_predicate): Add new argumnet probability
6191 for BUILT_IN_EXPECT_WITH_PROBABILITY.
6192 (fold_builtin_expect):
6195 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
6196 * builtins.h (fold_builtin_expect): Set new argument.
6197 * doc/extend.texi: Document __builtin_expect_with_probability.
6198 * doc/invoke.texi: Likewise.
6199 * gimple-fold.c (gimple_fold_call): Pass new argument.
6200 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
6201 also BUILT_IN_EXPECT_WITH_PROBABILITY.
6202 * predict.c (get_predictor_value): New function.
6203 (expr_expected_value): Add new argument probability. Assume
6204 that predictor and probability are always non-null.
6205 (expr_expected_value_1): Likewise. For __builtin_expect and
6206 __builtin_expect_with_probability set probability. Handle
6207 combination in binary expressions.
6208 (tree_predict_by_opcode): Simplify code by simply calling
6209 get_predictor_value.
6210 (pass_strip_predict_hints::execute): Add handling of
6211 BUILT_IN_EXPECT_WITH_PROBABILITY.
6212 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
6214 * tree.h (DECL_BUILT_IN_P): New function.
6216 2018-08-10 Martin Liska <mliska@suse.cz>
6218 PR tree-optimization/85799
6219 * passes.def: Add argument for pass_strip_predict_hints.
6220 * predict.c (class pass_strip_predict_hints): Add new argument
6222 (strip_predictor_early): New function.
6223 (pass_strip_predict_hints::execute): Call the function to
6225 (strip_predict_hints): New function.
6226 * predict.def: Fix comment.
6228 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
6230 * Makefile.in: Clarify which tm.texi to copy over to assert the
6231 right to grant a GFDL license for all.
6233 2018-08-09 Jeff Law <law@redhat.com>
6235 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
6238 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
6240 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
6243 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
6245 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
6246 reductions for variable-length vectors.
6248 2018-08-09 David Malcolm <dmalcolm@redhat.com>
6251 * common.opt (fdiagnostics-show-line-numbers): New option.
6252 * diagnostic-show-locus.c (class layout): Add fields
6253 "m_show_line_numbers_p" and "m_linenum_width";
6254 (num_digits): New function.
6255 (test_num_digits): New function.
6256 (layout::layout): Initialize new fields. Update m_x_offset
6257 logic to handle any left margin.
6258 (layout::print_source_line): Print line number when requested.
6259 (layout::start_annotation_line): New member function.
6260 (layout::print_annotation_line): Call it.
6261 (layout::print_leading_fixits): Likewise.
6262 (layout::print_trailing_fixits): Likewise. Update calls to
6263 move_to_column for new parameter.
6264 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
6265 it to potentially call start_annotation_line.
6266 (layout::show_ruler): Call start_annotation_line.
6267 (selftest::test_line_numbers_multiline_range): New selftest.
6268 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
6269 and selftest::test_line_numbers_multiline_range.
6270 * diagnostic.c (diagnostic_initialize): Initialize
6271 show_line_numbers_p.
6272 * diagnostic.h (struct diagnostic_context): Add field
6273 "show_line_numbers_p".
6274 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
6275 -fno-diagnostics-show-line-numbers.
6276 * dwarf2out.c (gen_producer_string): Add
6277 OPT_fdiagnostics_show_line_numbers to the ignored options.
6278 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
6279 one setting" options.
6280 (append_compiler_options): Likewise to the dropped options.
6281 (append_diag_options): Likewise to the passed-on options.
6282 * opts.c (common_handle_option): Handle the new option.
6283 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
6285 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
6287 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
6288 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
6289 third argument of type "const signed char" to descriptions of
6290 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
6291 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
6292 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
6293 __builtin_bcdsub_ov functions.
6295 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
6297 PR tree-optimization/86858
6298 * tree-vect-loop.c (vect_is_simple_reduction): Restore
6299 flow_bb_inside_loop_p calls.
6301 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
6303 PR tree-optimization/86871
6304 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
6305 instead of gimple_assign_lhs.
6307 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
6310 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
6311 register constraint to operand 0.
6312 (add<mode>3_carryinC): Likewise.
6313 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
6315 2018-08-09 Martin Liska <mliska@suse.cz>
6318 * common.opt: Remove extra line.
6320 2018-08-09 Martin Liska <mliska@suse.cz>
6322 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
6323 at the end of a line, make first letter capital and end up
6324 a sentence with a dot.
6325 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
6326 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
6327 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
6328 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
6329 (PARAM_MAX_ISL_OPERATIONS): Likewise.
6330 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
6331 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
6332 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
6333 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
6334 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
6335 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
6336 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
6337 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
6338 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
6339 (PARAM_TREE_REASSOC_WIDTH): Likewise.
6340 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
6341 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
6342 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
6344 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
6347 * config/s390/s390.c (s390_option_override_internal): Reduce the
6348 stack-clash-protection-probe-interval param if it would be too big
6351 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
6354 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
6355 * config/m68k/m68k.c (handle_move_double): Don't call
6356 m68k_final_prescan_insn.
6357 (m68k_adjust_decorated_operand): Renamed from
6358 m68k_final_prescan_insn, remove first and third operand and
6360 (print_operand): Call it.
6361 (print_operand_address): Call it.
6363 2018-08-08 Nathan Sidwell <nathan@acm.org>
6365 * diagnostic.c (diagnostic_report_current_module): Use
6366 linemap_included_from & linemap_included_from_linemap.
6368 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
6370 * config/aarch64/aarch64-cores.def: Add phecda core.
6371 * config/aarch64/aarch64-tune.md: Regenerate.
6372 * doc/invoke.texi: Add phecda core.
6374 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
6377 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
6379 * config/s390/s390.md ("movti"): Add more alternatives for
6380 constant to GPR copies.
6382 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
6384 * config/s390/s390.c: Fix whitespace damage throughout the file.
6385 * config/s390/s390.h: Likewise.
6386 * config/s390/tpf.h: Likewise.
6388 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
6390 * config/s390/s390.c (s390_loadrelative_operand_p):
6391 Remove TARGET_CPU_ZARCH usages.
6392 (s390_rtx_costs): Likewise.
6393 (s390_legitimate_constant_p): Likewise.
6394 (s390_cannot_force_const_mem): Likewise.
6395 (legitimate_reload_constant_p): Likewise.
6396 (s390_preferred_reload_class): Likewise.
6397 (legitimize_pic_address): Likewise.
6398 (legitimize_tls_address): Likewise.
6399 (s390_split_branches): Removed.
6400 (s390_add_execute): Removed.
6401 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
6402 (s390_mainpool_start): Likewise.
6403 (s390_mainpool_finish): Likewise.
6404 (s390_mainpool_cancel): Removed.
6405 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
6406 (s390_chunkify_cancel): Likewise.
6407 (s390_return_addr_rtx): Likewise.
6408 (s390_register_info): Remove split_branches_pending_p uages.
6409 (s390_optimize_register_info): Likewise.
6410 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
6411 split_branches_pending_p usages.
6412 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
6413 (s390_load_got): Likewise.
6414 (s390_expand_split_stack_prologue): Likewise.
6415 (output_asm_nops): Likewise.
6416 (s390_function_profiler): Likewise.
6417 (s390_emit_call): Likewise.
6418 (s390_conditional_register_usage): Likewise.
6419 (s390_optimize_prologue): Likewise.
6420 (s390_reorg): Remove TARGET_CPU_ZARCH and
6421 split_branches_pending_p usages.
6422 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
6424 (s390_output_indirect_thunk_function): Likewise.
6425 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
6426 (TARGET_CPU_ZARCH_P): Removed.
6427 (struct machine_function): Remove split_branches_pending_p.
6428 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
6430 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
6432 * common/config/s390/s390-common.c (processor_flags_table):
6434 * config.gcc: Remove with_arch/with_tune support.
6435 * config/s390/2064.md: Remove cpu attribute comparisons.
6436 * config/s390/driver-native.c (s390_host_detect_local_cpu):
6438 * config/s390/linux.h (ASM_SPEC):
6439 Remove -march support.
6440 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
6441 Use a table to get an arch level.
6442 * config/s390/s390-opts.h (enum processor_type):
6444 * config/s390/s390.c
6445 (processor_table): Remove entries, add arch_level values.
6446 (s390_issue_rate): Remove cases.
6447 (s390_option_override): Adjust
6448 s390_option_override_internal() call.
6449 (s390_option_override_internal): Remove deprecation warning.
6450 (s390_valid_target_attribute_tree): Adjust
6451 s390_option_override_internal() call.
6452 * config/s390/s390.h (struct s390_processor):
6453 Share with s390-c.c, add arch_level field.
6454 * config/s390/s390.md:
6455 Remove occurrences in cpu attribute.
6456 * config/s390/s390.opt: Remove -march/-mtune support.
6457 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
6458 * doc/invoke.texi: Remove deprecation warning.
6460 2018-08-08 Luis Machado <luis.machado@linaro.org>
6462 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
6464 (qdf24xx_tunings): Set vector cost structure to
6465 qdf24xx_vector_cost.
6467 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
6468 <register_sextend>: Set to 3.
6470 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
6473 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
6474 * config/aarch64/aarch64-simd.md
6475 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
6476 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
6477 (aarch64_frecpx<mode>): ...this new pattern.
6478 * config/aarch64/aarch64-simd-builtins.def: Remove comment
6479 about aarch64_frecp<FRECP:frecp_suffix><mode>.
6481 2018-08-07 Martin Liska <mliska@suse.cz>
6484 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
6485 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
6486 * predict.def (PRED_MALLOC_NONNULL): New predictor.
6487 * doc/extend.texi: Document that malloc attribute adds
6490 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
6493 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
6494 Define to speculation_safe_value_not_needed.
6496 2018-08-06 Jeff Law <law@redhat.com>
6498 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
6499 the vr_values instance to cprop_into_stmt.
6500 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
6501 (cprop_operand): Also query EVRP to determine if OP is a constant.
6503 2018-08-06 Nathan Sidwell <nathan@acm.org>
6505 * diagnostic.c (diagnostic_report_current_module): Reroll
6506 included-at loop. Translate text.
6508 2018-08-06 David Malcolm <dmalcolm@redhat.com>
6510 * function-tests.c (selftest::test_expansion_to_rtl): Call
6511 free_after_compilation.
6513 2018-08-06 Alan Hayward <alan.hayward@arm.com>
6515 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
6517 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
6519 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
6520 loops with memory block operations from getting unrolled.
6522 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
6525 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
6526 Define to speculation_safe_value_not_needed.
6528 2018-08-06 Jeff Law <law@redhat.com>
6530 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
6533 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
6536 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
6537 with all enabled __intN types.
6539 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
6541 2018-08-06 Alan Hayward <alan.hayward@arm.com>
6543 * alias.c (record_set): Check for clobber high.
6544 * cfgexpand.c (expand_gimple_stmt): Likewise.
6545 * combine-stack-adj.c (single_set_for_csa): Likewise.
6546 * combine.c (find_single_use_1): Likewise.
6547 (set_nonzero_bits_and_sign_copies): Likewise.
6548 (get_combine_src_dest): Likewise.
6549 (is_parallel_of_n_reg_sets): Likewise.
6550 (try_combine): Likewise.
6551 (record_dead_and_set_regs_1): Likewise.
6552 (reg_dead_at_p_1): Likewise.
6553 (reg_dead_at_p): Likewise.
6554 * dce.c (deletable_insn_p): Likewise.
6555 (mark_nonreg_stores_1): Likewise.
6556 (mark_nonreg_stores_2): Likewise.
6557 * df-scan.c (df_find_hard_reg_defs): Likewise.
6558 (df_uses_record): Likewise.
6559 (df_get_call_refs): Likewise.
6560 * dwarf2out.c (mem_loc_descriptor): Likewise.
6561 * haifa-sched.c (haifa_classify_rtx): Likewise.
6562 * ira-build.c (create_insn_allocnos): Likewise.
6563 * ira-costs.c (scan_one_insn): Likewise.
6564 * ira.c (equiv_init_movable_p): Likewise.
6565 (rtx_moveable_p): Likewise.
6566 (interesting_dest_for_shprep): Likewise.
6567 * jump.c (mark_jump_label_1): Likewise.
6568 * postreload-gcse.c (record_opr_changes): Likewise.
6569 * postreload.c (reload_cse_simplify): Likewise.
6570 (struct reg_use): Add source expr.
6571 (reload_combine): Check for clobber high.
6572 (reload_combine_note_use): Likewise.
6573 (reload_cse_move2add): Likewise.
6574 (move2add_note_store): Likewise.
6575 * print-rtl.c (print_pattern): Likewise.
6576 * recog.c (decode_asm_operands): Likewise.
6577 (store_data_bypass_p): Likewise.
6578 (if_test_bypass_p): Likewise.
6579 * regcprop.c (kill_clobbered_value): Likewise.
6580 (kill_set_value): Likewise.
6581 * reginfo.c (reg_scan_mark_refs): Likewise.
6582 * reload1.c (maybe_fix_stack_asms): Likewise.
6583 (eliminate_regs_1): Likewise.
6584 (elimination_effects): Likewise.
6585 (mark_not_eliminable): Likewise.
6586 (scan_paradoxical_subregs): Likewise.
6587 (forget_old_reloads_1): Likewise.
6588 * reorg.c (find_end_label): Likewise.
6589 (try_merge_delay_insns): Likewise.
6590 (redundant_insn): Likewise.
6591 (own_thread_p): Likewise.
6592 (fill_simple_delay_slots): Likewise.
6593 (fill_slots_from_thread): Likewise.
6594 (dbr_schedule): Likewise.
6595 * resource.c (update_live_status): Likewise.
6596 (mark_referenced_resources): Likewise.
6597 (mark_set_resources): Likewise.
6598 * rtl.c (copy_rtx): Likewise.
6599 * rtlanal.c (reg_referenced_p): Likewise.
6600 (single_set_2): Likewise.
6601 (noop_move_p): Likewise.
6602 (note_stores): Likewise.
6603 * sched-deps.c (sched_analyze_reg): Likewise.
6604 (sched_analyze_insn): Likewise.
6606 2018-08-06 Alan Hayward <alan.hayward@arm.com>
6608 * cse.c (invalidate_reg): New function extracted from...
6609 (invalidate): ...here.
6610 (canonicalize_insn): Check for clobber high.
6611 (invalidate_from_clobbers): invalidate clobber highs.
6612 (invalidate_from_sets_and_clobbers): Likewise.
6613 (count_reg_usage): Check for clobber high.
6614 (insn_live_p): Likewise.
6615 * cselib.c (cselib_expand_value_rtx_1):Likewise.
6616 (cselib_invalidate_regno): Check for clobber in setter.
6617 (cselib_invalidate_rtx): Pass through setter.
6618 (cselib_invalidate_rtx_note_stores):
6619 (cselib_process_insn): Check for clobber high.
6620 * cselib.h (cselib_invalidate_rtx): Add operand.
6622 2018-08-06 Alan Hayward <alan.hayward@arm.com>
6624 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
6625 (mark_not_eliminable): Likewise.
6626 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
6627 * lra-lives.c (process_bb_lives): Check for clobber high.
6628 * lra.c (new_insn_reg): Remember clobber highs.
6629 (collect_non_operand_hard_regs): Check for clobber high.
6630 (lra_set_insn_recog_data): Likewise.
6631 (add_regs_to_insn_regno_info): Likewise.
6632 (lra_update_insn_regno_info): Likewise.
6634 2018-08-06 Alan Hayward <alan.hayward@arm.com>
6636 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
6637 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
6639 2018-08-06 Alan Hayward <alan.hayward@arm.com>
6641 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
6642 (copy_insn_1): Likewise.
6643 (gen_hard_reg_clobber_high): New gen function.
6644 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
6645 * genemit.c (gen_exp): Likewise.
6646 (gen_emit_seq): Pass through info.
6647 (gen_insn): Check for CLOBBER_HIGH.
6648 (gen_expand): Pass through info.
6649 (gen_split): Likewise.
6650 (output_add_clobbers): Likewise.
6651 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
6652 (remove_clobbers): Likewise.
6653 * rtl.h (gen_hard_reg_clobber_high): New declaration.
6655 2018-08-06 Alan Hayward <alan.hayward@arm.com>
6657 * doc/rtl.texi (clobber_high): Add.
6658 (parallel): Add in clobber high
6659 * rtl.c (rtl_check_failed_code3): Add function.
6660 * rtl.def (CLOBBER_HIGH): Add expression.
6661 * rtl.h (RTL_CHECKC3): Add macro.
6662 (rtl_check_failed_code3): Add declaration.
6663 (XC3EXP): Add macro.
6665 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
6668 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
6669 cfun->machine->max_used_stack_alignment if needed.
6671 2018-08-04 Martin Sebor <msebor@redhat.com>
6673 PR tree-optimization/86571
6674 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
6677 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
6680 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
6683 2018-08-03 Jeff Law <law@redhat.com>
6686 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
6687 Define to speculation_safe_value_not_needed.
6689 2018-08-03 David Malcolm <dmalcolm@redhat.com>
6691 * doc/gcov.texi (-x): Remove duplicate "to".
6692 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
6693 (-Wif-not-aligned): Remove duplicate "is".
6694 (-flto): Remove duplicate "the".
6695 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
6696 duplicate "v5.00.b".
6697 (MSP430 Options): Remove duplicate "and" from the description
6698 of "-mgprel-sec=regexp".
6699 (x86 Options): Remove duplicate copies of "vmldLog102" and
6700 vmlsLog104 from description of "-mveclibabi=type".
6702 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
6704 * internal-fn.h (first_commutative_argument): Declare.
6705 * internal-fn.c (first_commutative_argument): New function.
6706 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
6707 restrictions for pattern statements. Use first_commutative_argument
6708 to look for commutative operands in calls to internal functions.
6710 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
6712 * Makefile.in (wide-int-range.o): New.
6713 * tree-vrp.c: Move all the wide_int_* functions to...
6714 * wide-int-range.cc: ...here.
6715 * tree-vrp.h: Move all the wide_int_* prototypes to...
6716 * wide-int-range.h: ...here.
6718 2018-08-03 Tom de Vries <tdevries@suse.de>
6720 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
6722 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
6723 * except.c (output_function_exception_table): Do early exit if
6724 targetm_common.except_unwind_info (&global_options) == UI_NONE.
6726 2018-08-03 Martin Liska <mliska@suse.cz>
6728 * predict.c (dump_prediction): Change to 2 digits
6731 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
6733 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
6735 (falkor_am_1_gtov_gtov): ... a new insn reservation.
6737 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
6739 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
6740 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
6741 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
6742 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
6743 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
6745 2018-08-02 David Malcolm <dmalcolm@redhat.com>
6747 * diagnostic-show-locus.c (diagnostic_show_locus): Use
6748 pp_take_prefix when saving the existing prefix.
6749 * diagnostic.c (diagnostic_append_note): Likewise.
6750 * langhooks.c (lhd_print_error_function): Likewise.
6751 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
6752 param's type. Free the existing prefix.
6753 (pp_take_prefix): New function.
6754 (pretty_printer::pretty_printer): Drop the prefix parameter.
6755 Rename the length parameter to match the comment.
6756 (pretty_printer::~pretty_printer): Free the prefix.
6757 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
6759 (struct pretty_printer): Drop the "const" from "prefix" field's
6760 type and clarify memory management.
6761 (pp_set_prefix): Drop the "const" from the 2nd param.
6762 (pp_take_prefix): New decl.
6764 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
6766 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
6767 for word_mode_ok here instead of passing as argument.
6768 (expand_block_compare): Change select_block_compare_mode() call.
6769 (expand_strncmp_gpr_sequence): New function.
6770 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
6772 2018-08-02 Jeff Law <law@redhat.com>
6775 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
6776 Define to speculation_safe_value_not_needed.
6779 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
6780 Define to speculation_safe_value_not_needed.
6782 2018-08-02 Tom de Vries <tdevries@suse.de>
6785 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
6786 function. Return UI_TARGET unconditionally.
6787 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
6788 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
6790 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
6792 * genemit.c (print_overload_test): Fix typo.
6794 2018-08-02 Richard Biener <rguenther@suse.de>
6796 PR tree-optimization/86816
6797 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
6798 which checks for value availability before querying it.
6799 (gvn_uses_equal): Use it.
6800 (same_succ_hash): Likewise.
6801 (gimple_equal_p): Likewise.
6803 2018-08-02 Nick Clifton <nickc@redhat.com>
6806 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
6807 Define to speculation_safe_value_not_needed.
6810 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
6811 Define to speculation_safe_value_not_needed.
6814 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
6815 Define to speculation_safe_value_not_needed.
6818 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
6819 Define to speculation_safe_value_not_needed.
6822 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
6823 Define to speculation_safe_value_not_needed.
6826 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
6827 Define to speculation_safe_value_not_needed.
6830 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
6831 Define to speculation_safe_value_not_needed.
6834 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
6835 Define to speculation_safe_value_not_needed.
6838 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
6839 speculation_safe_value_not_needed.
6842 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
6843 to speculation_safe_value_not_needed.
6845 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
6847 * doc/md.texi: Expand the documentation of instruction names
6848 to mention port-local uses. Document '@' in pattern names.
6849 * read-md.h (overloaded_instance, overloaded_name): New structs.
6851 (md_reader::handle_overloaded_name): New member function.
6852 (md_reader::get_overloads): Likewise.
6853 (md_reader::m_first_overload): New member variable.
6854 (md_reader::m_next_overload_ptr): Likewise.
6855 (md_reader::m_overloads_htab): Likewise.
6856 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
6857 m_next_overload_ptr and m_overloads_htab.
6858 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
6859 (get_mode_token, get_code_token, get_int_token): New functions.
6860 (map_attr_string): Add an optional argument that passes back
6861 the associated iterator.
6862 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
6863 (md_reader::handle_overloaded_name, add_overload_instance): New
6865 (apply_iterators): Handle '@' names. Report an error if '@'
6866 is used without iterators.
6867 (initialize_iterators): Initialize the new iterator_group fields.
6868 * genopinit.c (handle_overloaded_code_for)
6869 (handle_overloaded_gen): New functions.
6870 (main): Use them to print declarations of maybe_code_for_* and
6871 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
6872 * genemit.c (print_overload_arguments, print_overload_test)
6873 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
6874 (main): Use it to print definitions of maybe_code_for_* and
6875 maybe_gen_* functions.
6876 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
6877 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
6878 instead of explicit mode checks.
6879 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
6880 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
6881 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
6882 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
6883 (aarch64_expand_compare_and_swap): Likewise
6884 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
6885 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
6886 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
6887 (aarch64_constant_pool_reload_icode): Delete.
6888 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
6889 instead of aarch64_constant_pool_reload_icode. Use
6890 code_for_aarch64_reload_mov instead of explicit mode checks.
6891 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
6892 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
6893 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
6894 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
6895 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
6896 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
6897 (aarch64_atomic_load_op_code): Delete.
6898 (aarch64_emit_atomic_load_op): Likewise.
6899 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
6900 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
6901 instead of aarch64_emit_atomic_load_op.
6902 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
6903 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
6904 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
6905 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
6906 character before the pattern name.
6907 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
6908 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
6909 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
6910 (aarch64_frecps<mode>): Likewise.
6911 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
6912 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
6913 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
6914 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
6915 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
6917 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
6919 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
6920 Allow HFmode constants if TARGET_FP_F16INST.
6922 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
6925 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
6926 No longer check last store for clobber of address register.
6928 2018-08-02 Martin Liska <mliska@suse.cz>
6930 PR gcov-profile/86817
6931 * gcov.c (process_all_functions): New function.
6933 (process_file): Move functions processing to
6934 process_all_functions.
6936 2018-08-02 David Malcolm <dmalcolm@redhat.com>
6938 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
6939 "const" to the "gimple *" and "rtx_insn *" parameters.
6940 * dumpfile.h (dump_user_location_t::dump_user_location_t):
6942 (dump_location_t::dump_location_t): Likewise.
6944 2018-08-01 Martin Sebor <msebor@redhat.com>
6946 PR tree-optimization/86650
6947 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
6948 rather than a "gcall *". Directly pass the data of interest
6949 to percent_K_format, rather than building a temporary CALL_EXPR
6951 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
6952 (gimple_fold_builtin_strncat): Adjust.
6953 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
6954 gcall* argument with gimple*.
6955 * gimple-ssa-warn-restrict.c (check_call): Same.
6956 (wrestrict_dom_walker::before_dom_children): Same.
6957 (builtin_access::builtin_access): Same.
6958 (check_bounds_or_overlap): Same
6959 (maybe_diag_overlap): Same.
6960 (maybe_diag_offset_bounds): Same.
6961 * tree-diagnostic.c (default_tree_printer): Move usage of
6962 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
6964 * tree-pretty-print.c (percent_K_format): Add argument.
6965 * tree-pretty-print.h: Add argument.
6966 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
6967 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
6968 (maybe_diag_stxncpy_trunc): Same.
6969 (handle_builtin_stxncpy): Same.
6970 (handle_builtin_strcat): Same.
6972 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
6974 * match.pd: Optimise pointer range checks.
6976 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
6978 PR tree-optimization/86758
6979 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
6980 to remove pattern statements.
6982 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
6984 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
6985 result of dfs_enumerate_from when constructing stmt_vec_infos,
6986 instead of additionally calling get_loop_body.
6988 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
6990 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
6992 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
6993 When creating an iv, assert that the step is not known to be zero.
6994 (vect_setup_realignment): Update call accordingly.
6995 * tree-vect-stmts.c (vectorizable_store): Likewise.
6996 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
6998 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7000 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
7001 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
7002 (vectorizable_reduction): Likewise.
7003 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
7004 (vect_detect_hybrid_slp_stmts): Likewise.
7005 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
7007 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
7009 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
7010 (wide_int_set_zero_nonzero_bits): ...this.
7011 (zero_nonzero_bits_from_vr): Rename to...
7012 (vrp_set_zero_nonzero_bits): ...this.
7013 (extract_range_from_multiplicative_op_1): Abstract wide int
7015 (wide_int_range_multiplicative_op): ...here.
7016 (extract_range_from_binary_expr_1): Extract wide int binary
7017 operations into their own functions.
7018 (wide_int_range_lshift): New.
7019 (wide_int_range_can_optimize_bit_op): New.
7020 (wide_int_range_shift_undefined_p): New.
7021 (wide_int_range_bit_xor): New.
7022 (wide_int_range_bit_ior): New.
7023 (wide_int_range_bit_and): New.
7024 (wide_int_range_trunc_mod): New.
7025 (extract_range_into_wide_ints): New.
7026 (vrp_shift_undefined_p): New.
7027 (extract_range_from_multiplicative_op): New.
7028 (vrp_can_optimize_bit_op): New.
7029 * tree-vrp.h (value_range::dump): New.
7030 (wide_int_range_multiplicative_op): New.
7031 (wide_int_range_lshift):New.
7032 (wide_int_range_shift_undefined_p): New.
7033 (wide_int_range_bit_xor): New.
7034 (wide_int_range_bit_ior): New.
7035 (wide_int_range_bit_and): New.
7036 (wide_int_range_trunc_mod): New.
7037 (zero_nonzero_bits_from_bounds): Rename to...
7038 (wide_int_set_zero_nonzero_bits): ...this.
7039 (zero_nonzero_bits_from_vr): Rename to...
7040 (vrp_set_zero_nonzero_bits): ...this.
7041 (range_easy_mask_min_max): Rename to...
7042 (wide_int_range_can_optimize_bit_op): this.
7043 * vr-values.c (simplify_bit_ops_using_ranges): Rename
7044 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
7046 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7048 * tree-vectorizer.h (vect_orig_stmt): New function.
7049 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
7050 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
7051 (vect_create_epilog_for_reduction): Likewise.
7052 (vectorizable_live_operation): Likewise.
7053 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
7054 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
7055 * tree-vect-stmts.c (vectorizable_call): Likewise.
7056 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
7058 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7060 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
7062 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
7063 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
7064 (vect_transform_loop): Likewise.
7065 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
7067 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7069 * tree-vectorizer.h (vect_schedule_slp): Return void.
7070 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
7071 (vect_schedule_slp): Likewise.
7073 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7075 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
7077 (vect_transform_loop): Update calls accordingly. Schedule SLP
7078 instances before the main loop, if any exist.
7080 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7082 PR tree-optimization/86749
7083 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
7084 If the lhs is used in a COND_EXPR, check that it is being used
7085 as the "then" or "else" value.
7087 2018-08-01 Tom de Vries <tdevries@suse.de>
7090 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
7091 speculation_safe_value_not_needed.
7093 2018-08-01 Richard Biener <rguenther@suse.de>
7095 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
7098 2018-08-01 Martin Liska <mliska@suse.cz>
7100 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
7101 format how successful transformation is dumped.
7102 (gimple_mod_pow2_value_transform): Likewise.
7103 (gimple_mod_subtract_transform): Likewise.
7104 (gimple_stringops_transform): Likewise.
7106 2018-08-01 Martin Liska <mliska@suse.cz>
7109 * value-prof.c (interesting_stringop_to_profile_p):
7110 Simplify the code and add BUILT_IN_MEMMOVE.
7111 (gimple_stringops_transform): Likewise.
7113 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
7115 * config/aarch64/aarch64-simd.md
7116 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
7117 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
7118 use GPI iterator instead of SI mode.
7120 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
7122 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
7123 rs6000_speculation_barrier.
7124 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
7125 new barrier pattern name.
7127 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
7129 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
7130 (speculation_barrier): New insn.
7132 2018-08-01 Richard Biener <rguenther@suse.de>
7135 * graphite.h: Include isl/id.h and isl/space.h to allow build
7138 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
7141 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
7142 mode for COFF targets.
7143 * defaults.h (TARGET_COFF): Define.
7144 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
7145 TARGET_COFF): Define.
7146 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
7147 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
7148 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
7150 2018-07-31 Alexandre Oliva <oliva@adacore.com>
7152 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
7153 * gimple-streamer-out.c (output_bb): Save it.
7154 * lto-streamer-in.c (input_struct_function_base): Restore
7155 instance discriminator if available. Create map on demand.
7156 * lto-streamer-out.c (output_struct_function_base): Save it if
7158 * final.c (decl_to_instance_map): Document LTO strategy.
7160 2018-07-31 Alexandre Oliva <oliva@adacore.com>
7161 Olivier Hainque <hainque@adacore.com>
7163 * debug.h (decl_to_instance_map_t): New type.
7164 (decl_to_instance_map): Declare.
7165 (maybe_create_decl_to_instance_map): New inline function.
7166 * final.c (bb_discriminator, last_bb_discriminator): New statics,
7167 to track basic block discriminators.
7168 (final_start_function_1): Initialize them.
7169 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
7171 (decl_to_instance_map): New variable.
7172 (map_decl_to_instance, maybe_set_discriminator): New functions.
7173 (notice_source_line): Set discriminator.
7175 2018-07-31 Ian Lance Taylor <iant@golang.org>
7177 * targhooks.c (default_have_speculation_safe_value): Add
7180 2018-07-31 David Malcolm <dmalcolm@redhat.com>
7182 * dump-context.h: Include "pretty-print.h".
7183 (dump_context::refresh_dumps_are_enabled): New decl.
7184 (dump_context::emit_item): New decl.
7185 (class dump_context): Add fields "m_test_pp" and
7187 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
7188 (temp_dump_context::get_dumped_text): New decl.
7189 (class temp_dump_context): Add field "m_pp".
7190 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
7191 (dump_context::refresh_dumps_are_enabled): ...and add a test for
7193 (set_dump_file): Update for above change.
7194 (set_alt_dump_file): Likewise.
7195 (dump_loc): New overload, taking a pretty_printer *.
7196 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
7197 to any test pretty-printer.
7198 (make_item_for_dump_gimple_stmt): New function, adapted from
7199 optinfo::add_gimple_stmt.
7200 (dump_context::dump_gimple_stmt): Call it, and use the result,
7201 eliminating the direct usage of dump_file and alt_dump_file in
7202 favor of indirectly using them via emit_item.
7203 (make_item_for_dump_gimple_expr): New function, adapted from
7204 optinfo::add_gimple_expr.
7205 (dump_context::dump_gimple_expr): Call it, and use the result,
7206 eliminating the direct usage of dump_file and alt_dump_file in
7207 favor of indirectly using them via emit_item.
7208 (make_item_for_dump_generic_expr): New function, adapted from
7210 (dump_context::dump_generic_expr): Call it, and use the result,
7211 eliminating the direct usage of dump_file and alt_dump_file in
7212 favor of indirectly using them via emit_item.
7213 (make_item_for_dump_printf_va): New function, adapted from
7214 optinfo::add_printf_va.
7215 (make_item_for_dump_printf): New function.
7216 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
7217 and use the result, eliminating the direct usage of dump_file and
7218 alt_dump_file in favor of indirectly using them via emit_item.
7219 (make_item_for_dump_dec): New function.
7220 (dump_context::dump_dec): Call it, and use the result,
7221 eliminating the direct usage of dump_file and alt_dump_file in
7222 favor of indirectly using them via emit_item.
7223 (make_item_for_dump_symtab_node): New function, adapted from
7224 optinfo::add_symtab_node.
7225 (dump_context::dump_symtab_node): Call it, and use the result,
7226 eliminating the direct usage of dump_file and alt_dump_file in
7227 favor of indirectly using them via emit_item.
7228 (dump_context::begin_scope): Reimplement, avoiding direct usage
7229 of dump_file and alt_dump_file in favor of indirectly using them
7231 (dump_context::emit_item): New member function.
7232 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
7233 Set up test pretty-printer on the underlying context. Call
7234 refresh_dumps_are_enabled.
7235 (temp_dump_context::~temp_dump_context): Call
7236 refresh_dumps_are_enabled.
7237 (temp_dump_context::get_dumped_text): New member function.
7238 (selftest::verify_dumped_text): New function.
7239 (ASSERT_DUMPED_TEXT_EQ): New macro.
7240 (selftest::test_capture_of_dump_calls): Run all tests twice, with
7241 and then without optinfo enabled. Add uses of
7242 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
7243 * dumpfile.h: Update comment for the dump_* API.
7244 * optinfo-emit-json.cc
7245 (selftest::test_building_json_from_dump_calls): Update for new
7246 param for temp_dump_context ctor.
7247 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
7248 and "m_owned" field.
7249 (optinfo_item::~optinfo_item): Likewise.
7250 (optinfo::add_item): New member function.
7251 (optinfo::emit): Update comment.
7252 (optinfo::add_string): Delete.
7253 (optinfo::add_printf): Delete.
7254 (optinfo::add_printf_va): Delete.
7255 (optinfo::add_gimple_stmt): Delete.
7256 (optinfo::add_gimple_expr): Delete.
7257 (optinfo::add_tree): Delete.
7258 (optinfo::add_symtab_node): Delete.
7259 (optinfo::add_dec): Delete.
7260 * optinfo.h (class dump_context): New forward decl.
7261 (optinfo::add_item): New decl.
7262 (optinfo::add_string): Delete.
7263 (optinfo::add_printf): Delete.
7264 (optinfo::add_printf_va): Delete.
7265 (optinfo::add_gimple_stmt): Delete.
7266 (optinfo::add_gimple_expr): Delete.
7267 (optinfo::add_tree): Delete.
7268 (optinfo::add_symtab_node): Delete.
7269 (optinfo::add_dec): Delete.
7270 (optinfo::add_poly_int): Delete.
7271 (optinfo_item::optinfo_item): Remove "owned" param.
7272 (class optinfo_item): Remove field "m_owned".
7274 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
7277 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
7278 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
7279 requested variable alignment.
7280 (expand_one_ssa_partition): Likewise.
7281 (expand_one_var): Likewise.
7283 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7285 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
7286 to speculation_safe_value_not_needed.
7288 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7290 * targhooks.h (speculation_safe_value_not_needed): New prototype.
7291 * targhooks.c (speculation_safe_value_not_needed): New function.
7292 * target.def (have_speculation_safe_value): Update documentation.
7293 * doc/tm.texi: Regenerated.
7295 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7297 * config/aarch64/iterators.md (ALLI_TI): New iterator.
7298 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
7300 (despeculate_copy<ALLI:mode>_insn): New insn.
7301 (despeculate_copyti_insn): New insn.
7302 (despeculate_simple<ALLI:mode>): New insn
7303 (despeculate_simpleti): New insn.
7304 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
7306 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
7307 aarch64_speculation_safe_value.
7308 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
7310 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7312 * config/aarch64/aarch64-speculation.cc: New file.
7313 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
7314 before pass_reorder_blocks.
7315 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
7317 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
7318 X14 and X15 when tracking speculation.
7319 * config/aarch64/aarch64.md (register name constants): Add
7320 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
7321 (unspec): Add UNSPEC_SPECULATION_TRACKER.
7322 (speculation_barrier): New insn attribute.
7323 (cmp<mode>): Allow SP in comparisons.
7324 (speculation_tracker): New insn.
7325 (speculation_barrier): Add speculation_barrier attribute.
7326 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
7327 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
7328 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
7330 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7332 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
7333 aarch64_track_speculation is true.
7334 (tb<optab><mode>1): Likewise.
7335 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
7336 generate CB[N]Z when tracking speculation.
7337 (aarch64_split_compare_and_swap): Likewise.
7338 (aarch64_split_atomic_op): Likewise.
7340 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7342 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
7344 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7346 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
7347 (speculation_barrier): New insn.
7349 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7351 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
7352 * config/arm/arm.md (speculation_barrier): New expand.
7353 (speculation_barrier_insn): New pattern.
7355 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7357 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
7358 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
7359 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
7360 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
7362 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
7363 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
7364 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
7365 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
7366 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
7367 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
7368 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
7369 * builtins.c (expand_speculation_safe_value): New function.
7370 (expand_builtin): Call it.
7371 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
7372 * doc/extend.texi: Document __builtin_speculation_safe_value.
7373 * doc/md.texi: Document "speculation_barrier" pattern.
7374 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
7375 TARGET_HAVE_SPECULATION_SAFE_VALUE.
7376 * doc/tm.texi: Regenerated.
7377 * target.def (have_speculation_safe_value, speculation_safe_value): New
7379 * targhooks.c (default_have_speculation_safe_value): New function.
7380 (default_speculation_safe_value): New function.
7381 * targhooks.h (default_have_speculation_safe_value): Add prototype.
7382 (default_speculation_safe_value): Add prototype.
7384 2018-07-31 David Malcolm <dmalcolm@redhat.com>
7386 * dump-context.h (dump_context::dump_loc): New decl.
7387 * dumpfile.c (dump_context::dump_loc): New member function.
7388 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
7389 and dump_gimple_stmt.
7390 (dump_context::dump_gimple_expr_loc): Likewise, using
7392 (dump_context::dump_generic_expr_loc): Likewise, using
7394 (dump_context::dump_printf_loc_va): Likewise, using
7396 (dump_context::begin_scope): Explicitly using the global function
7397 "dump_loc", rather than the member function.
7399 2018-07-31 Martin Sebor <msebor@redhat.com>
7401 PR tree-optimization/86741
7402 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
7404 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
7406 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
7407 depend on whether prefetch instructions will be emitted or not.
7408 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
7409 will be emitted or not.
7410 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
7411 (TARGET_SETMEM_PFD): New macros.
7413 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7415 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
7416 (NULL_STMT_VEC_INFO): Delete.
7417 (stmt_vec_info::operator*): Likewise.
7418 (stmt_vec_info::operator gimple *): Likewise.
7419 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
7420 of NULL_STMT_VEC_INFO.
7421 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
7422 (vect_reassociating_reduction_p): Likewise.
7423 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
7424 (vectorizable_store): Likewise.
7425 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
7426 (vec_info::free_stmt_vec_infos): Likewise.
7428 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7430 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
7431 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
7433 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7435 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
7436 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
7437 (vec_info::free_stmt_vec_info): New private member functions.
7438 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
7439 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
7440 * tree-parloops.c (gather_scalar_reductions): Remove calls to
7441 set_stmt_vec_info_vec and free_stmt_vec_infos.
7442 * tree-vect-loop.c (_loop_vec_info): Remove call to
7443 set_stmt_vec_info_vec.
7444 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
7445 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
7446 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
7447 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
7448 (vec_info::free_stmt_vec_info): ...these new functions. Remove
7449 assignments in {vec_info::,}new_stmt_vec_info that are redundant
7450 with the clearing in the xcalloc.
7451 (stmt_vec_info_vec): Delete.
7452 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
7453 (vectorize_loops): Likewise.
7454 (vec_info::~vec_info): Remove argument from call to
7455 free_stmt_vec_infos.
7456 (vec_info::add_stmt): Remove vinfo argument from call to
7459 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7461 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
7462 rather than a gimple stmt.
7463 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
7464 information for pattern statements when passed the original
7465 statement; instead wait to be passed the pattern statement itself.
7466 Don't call set_vinfo_for_stmt here.
7467 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
7468 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
7469 stmt_vec_infos here.
7470 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
7471 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
7472 stmt_vec_infos entry.
7474 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7476 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
7477 * tree-vectorizer.c (vec_info::replace_stmt): New function.
7478 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
7479 * tree-vect-stmts.c (vectorizable_call): Likewise.
7480 (vectorizable_simd_clone_call): Likewise.
7482 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7484 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
7485 * tree-vectorizer.c (vec_info::remove_stmt): New function.
7486 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
7487 * tree-vect-loop.c (vect_transform_loop): Likewise.
7488 * tree-vect-slp.c (vect_schedule_slp): Likewise.
7489 * tree-vect-stmts.c (vect_remove_stores): Likewise.
7491 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7493 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
7494 (vect_dr_stmt): Delete.
7495 * tree-vectorizer.c (vec_info::lookup_dr): New function.
7496 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
7498 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
7499 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
7500 (vect_verify_datarefs_alignment, vect_peeling_supportable)
7501 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
7502 (vect_analyze_data_refs): Likewise.
7503 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
7505 (vect_find_same_alignment_drs): Likewise.
7506 (vect_slp_analyze_node_dependences): Update calls accordingly.
7507 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
7508 instead of DR_VECT_AUX.
7509 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
7510 of a vector data references. Use vec_info::lookup_dr instead of
7512 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
7513 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
7514 instead of DR_VECT_AUX.
7516 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7518 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
7520 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
7522 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
7523 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
7524 (vect_gen_prolog_loop_niters): Likewise.
7526 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7528 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
7529 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
7530 (vect_known_alignment_in_bytes, vect_dr_behavior)
7531 (vect_get_scalar_dr_size): Take references as dr_vec_infos
7532 instead of data_references. Update calls to other routines for
7533 which the same change has been made.
7534 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
7535 dr_vec_infos instead of stmt_vec_infos.
7536 (vect_analyze_data_ref_dependence): Update call accordingly.
7537 (vect_slp_analyze_data_ref_dependence)
7538 (vect_record_base_alignments): Use DR_VECT_AUX.
7539 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
7540 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
7541 (vector_alignment_reachable_p, vect_get_data_access_cost)
7542 (vect_peeling_supportable, vect_analyze_group_access_1)
7543 (vect_analyze_group_access, vect_analyze_data_ref_access)
7544 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
7545 (vect_compile_time_alias, vect_small_gap_p)
7546 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
7547 (vect_supportable_dr_alignment): Take references as dr_vec_infos
7548 instead of data_references. Update calls to other routines for
7549 which the same change has been made.
7550 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
7551 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
7552 (vect_slp_analyze_and_verify_node_alignment)
7553 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
7554 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
7555 (vect_setup_realignment): Use dr_vec_infos. Update calls after
7557 (_vect_peel_info::dr): Replace with...
7558 (_vect_peel_info::dr_info): ...this new field.
7559 (vect_peeling_hash_get_most_frequent)
7560 (vect_peeling_hash_choose_best_peeling): Update accordingly.
7561 (vect_peeling_hash_get_lowest_cost):
7562 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
7563 routines for which the same change has been made.
7564 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
7566 * tree-vect-loop-manip.c (get_misalign_in_elems)
7567 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
7569 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
7570 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
7571 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
7572 (get_group_load_store_type, get_negative_load_store_type)
7573 (vect_get_data_ptr_increment, vectorizable_store)
7574 (vectorizable_load): Likewise.
7575 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
7576 Update calls to other routines for which the same change has been made.
7578 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7580 * tree-vectorizer.h (vec_info::move_dr): New member function.
7581 (dataref_aux): Rename to...
7582 (dr_vec_info): ...this and add "dr" and "stmt" fields.
7583 (_stmt_vec_info::dr_aux): Update accordingly.
7584 (_stmt_vec_info::data_ref_info): Delete.
7585 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
7586 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
7587 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
7588 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
7590 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
7591 (STMT_VINFO_DR_INFO): New macro.
7592 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
7593 (set_dr_misalignment): Update after rename of dataref_aux.
7594 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
7595 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
7596 initialization of STMT_VINFO_DATA_REF.
7597 * tree-vectorizer.c (vec_info::move_dr): New function.
7598 * tree-vect-patterns.c (vect_recog_bool_pattern)
7599 (vect_recog_mask_conversion_pattern)
7600 (vect_recog_gather_scatter_pattern): Use it.
7601 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
7602 the "dr" and "stmt" fields of dr_vec_info instead of
7603 STMT_VINFO_DATA_REF.
7605 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7607 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
7608 (is_pattern_stmt_p): Use it.
7609 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
7610 on pattern statements.
7612 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7614 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
7615 original stmt as a stmt_vec_info rather than a gimple stmt.
7616 (vect_pattern_recog_1): Take the statement directly as a
7617 stmt_vec_info, rather than via a gimple_stmt_iterator.
7618 Update call to vect_mark_pattern_stmts.
7619 (vect_pattern_recog): Update calls accordingly.
7621 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7623 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
7624 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
7625 a vect_def_type for the first argument.
7626 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
7627 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
7628 operand if it isn't defined by a vectorized statement.
7629 (vect_build_gather_load_calls): Remove the mask_dt argument and
7630 update calls to vect_get_vec_def_for_stmt_copy.
7631 (vectorizable_bswap): Likewise the dt argument.
7632 (vectorizable_call): Update calls to vectorizable_bswap and
7633 vect_get_vec_def_for_stmt_copy.
7634 (vectorizable_simd_clone_call, vectorizable_assignment)
7635 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
7636 (vectorizable_comparison): Update calls to
7637 vect_get_vec_def_for_stmt_copy.
7638 (vectorizable_store): Likewise. Remove now-unnecessary calls to
7640 (vect_get_loop_based_defs): Remove dt argument and update call
7641 to vect_get_vec_def_for_stmt_copy.
7642 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
7643 and vect_get_vec_def_for_stmt_copy.
7644 (vectorizable_load): Update calls to vect_build_gather_load_calls
7645 and vect_get_vec_def_for_stmt_copy.
7646 * tree-vect-loop.c (vect_create_epilog_for_reduction)
7647 (vectorizable_reduction, vectorizable_live_operation): Update calls
7648 to vect_get_vec_def_for_stmt_copy.
7650 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7652 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
7653 and gimple stmt arguments with a stmt_vec_info.
7654 (vect_record_base_alignments): Update calls accordingly.
7655 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
7656 and gimple stmt arguments with a stmt_vec_info.
7657 (vect_build_slp_tree_1): Remove vinfo argument and update call
7658 to vect_record_max_nunits.
7659 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
7660 and vect_record_max_nunits.
7662 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7664 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
7665 file and take a stmt_vec_info instead of a gimple stmt.
7666 (supportable_widening_operation, vect_finish_replace_stmt)
7667 (vect_finish_stmt_generation, vect_get_store_rhs)
7668 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
7669 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
7670 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
7671 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
7672 (vect_create_data_ref_ptr, bump_vector_ptr)
7673 (vect_permute_store_chain, vect_setup_realignment)
7674 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
7675 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
7676 (vectorizable_reduction, vectorizable_induction)
7677 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
7678 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
7679 than gimple stmts as arguments.
7680 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
7681 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
7682 (can_group_stmts_p, vect_check_gather_scatter)
7683 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
7684 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
7685 (vect_permute_load_chain, vect_shift_permute_load_chain)
7686 (vect_transform_grouped_load)
7687 (vect_record_grouped_load_vectors): Likewise.
7688 * tree-vect-loop.c (vect_fixup_reduc_chain)
7689 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
7690 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
7691 (vectorizable_reduction, vectorizable_induction)
7692 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
7693 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
7694 (vect_get_load_store_mask): Likewise.
7695 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
7696 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
7697 * tree-vect-stmts.c (vect_mark_relevant)
7698 (is_simple_and_all_uses_invariant)
7699 (exist_non_indexing_operands_for_use_p, process_use)
7700 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
7701 (vect_get_vec_def_for_operand, vect_get_vec_defs)
7702 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
7703 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
7704 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
7705 (get_negative_load_store_type, get_load_store_type)
7706 (vect_check_load_store_mask, vect_check_store_rhs)
7707 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
7708 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
7709 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
7710 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
7711 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
7712 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
7713 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
7714 (supportable_widening_operation): Likewise.
7716 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7718 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
7719 a stmt_vec_info instead of a gcall.
7720 (vect_check_gather_scatter): Update call accordingly.
7721 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
7723 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
7724 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
7725 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
7726 instead of a gimple stmt.
7727 (vect_transform_loop): Update calls accordingly.
7728 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
7729 stmt_vec_infos instead of gimple stmts.
7730 (vect_analyze_slp_instance): Update use accordingly.
7731 * tree-vect-stmts.c (read_vector_array, write_vector_array)
7732 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
7733 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
7734 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
7735 (vect_gen_widened_results_half, vect_get_loop_based_defs)
7736 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
7737 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
7738 down to subroutines.
7740 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7742 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
7743 of the worklist from a vector of gimple stmts to a vector of
7745 * tree-vect-stmts.c (vect_mark_relevant, process_use)
7746 (vect_mark_stmts_to_be_vectorized): Likewise
7748 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7750 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
7751 statement before passing it to vect_analyze_stmt.
7752 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
7753 the chain of phi vector definitions. Track the exit phi via its
7755 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
7756 STMT_VINFO_REDUC_DEF.
7757 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
7758 stmt_vec_infos to handle the statement chains.
7759 (vect_get_slp_defs): Record the first statement in the node
7760 using a stmt_vec_info.
7761 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
7762 statements here and pass their stmt_vec_info down to subroutines.
7763 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
7764 down to vect_finish_stmt_generation.
7765 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
7766 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
7767 stmt_vec_infos to subroutines.
7768 (vect_remove_stores): Use stmt_vec_infos to handle the statement
7771 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7773 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
7774 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
7775 (vect_permute_store_chain, vect_setup_realignment)
7776 (vect_permute_load_chain, vect_shift_permute_load_chain)
7777 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
7778 stmts internally, and when passing values to other vectorizer routines.
7779 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
7780 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
7781 (vect_analyze_loop_operations, get_initial_def_for_reduction)
7782 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
7783 (vectorizable_reduction, vectorizable_induction)
7784 (vectorizable_live_operation, vect_transform_loop_stmt)
7785 (vect_transform_loop): Likewise.
7786 * tree-vect-patterns.c (vect_reassociating_reduction_p)
7787 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
7788 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
7789 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
7790 (vect_slp_analyze_node_operations_1): Likewise.
7791 * tree-vect-stmts.c (vect_mark_relevant, process_use)
7792 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
7793 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
7794 (vect_finish_stmt_generation_1, get_group_load_store_type)
7795 (get_load_store_type, vect_build_gather_load_calls)
7796 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
7797 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
7798 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
7799 (vectorizable_store, vectorizable_load, vectorizable_condition)
7800 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
7801 (supportable_widening_operation): Likewise.
7802 (vect_get_vector_types_for_stmt): Likewise.
7803 * tree-vectorizer.h (vect_dr_behavior): Likewise.
7805 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7807 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
7808 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
7809 (vect_permute_store_chain, vect_permute_load_chain)
7810 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
7811 repeated stmt_vec_info lookups.
7812 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
7813 (vect_update_ivs_after_vectorizer): Likewise.
7814 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
7815 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
7816 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
7817 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
7818 (vect_bb_slp_scalar_cost): Likewise.
7819 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
7821 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7823 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
7824 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
7825 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
7826 to get gassigns and gcalls, rather than operating on generc gimple
7828 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
7829 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
7830 (vectorizable_load, vect_analyze_stmt): Likewise.
7831 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
7833 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7835 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
7836 return stmt_vec_infos rather than gimple stmts. Do not accept
7838 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
7840 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
7841 Update use of get_later_stmt.
7842 (vect_get_constant_vectors): Update call accordingly.
7843 (vect_schedule_slp_instance): Likewise
7844 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
7845 (vect_slp_analyze_instance_dependence): Likewise.
7846 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
7848 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7850 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
7851 (stmt_info_for_cost::stmt_info): ...this new field.
7852 (add_stmt_costs): Update accordingly.
7853 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
7854 (vect_get_known_peeling_cost): Likewise.
7855 (vect_estimate_min_profitable_iters): Likewise.
7856 * tree-vect-stmts.c (record_stmt_cost): Likewise.
7858 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7860 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
7861 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
7862 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
7864 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
7866 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7868 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
7869 a gimple stmt to a stmt_vec_info.
7870 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
7872 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7874 * tree-vectorizer.h (vec_info::grouped_stores): Change from
7875 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
7876 (_loop_vec_info::reduction_chains): Likewise.
7877 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
7879 * tree-vect-slp.c (vect_analyze_slp): Likewise.
7881 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7883 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
7884 a gimple stmt to a stmt_vec_info.
7885 (_stmt_vec_info::next_element): Likewise.
7886 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
7887 (vect_slp_analyze_and_verify_node_alignment)
7888 (vect_analyze_group_access_1, vect_analyze_group_access)
7889 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
7890 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
7891 (vect_supportable_dr_alignment): Update accordingly.
7892 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
7893 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
7894 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
7895 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
7896 * tree-vect-slp.c (vect_build_slp_tree_1)
7897 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
7898 (vect_split_slp_store_group, vect_analyze_slp_instance)
7899 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
7900 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
7901 (get_group_load_store_type, get_load_store_type)
7902 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
7903 (vect_transform_stmt, vect_remove_stores): Likewise.
7905 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7907 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
7909 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
7910 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
7911 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
7912 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
7913 (vector_alignment_reachable_p, vect_get_data_access_cost)
7914 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
7915 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
7916 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
7917 (vect_analyze_group_access_1, vect_analyze_group_access)
7918 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
7919 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
7920 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
7921 result of vect_dr_stmt and use the stmt_vec_info instead of
7922 the associated gimple stmt.
7923 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
7924 (vect_gen_prolog_loop_niters): Likewise.
7925 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
7927 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7929 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
7930 to a vec<stmt_vec_info>.
7931 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
7932 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
7934 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
7935 to a vec<stmt_vec_info>.
7936 (bst_traits::value_type, bst_traits::value_type): Likewise.
7937 (bst_traits::hash): Update accordingly.
7938 (vect_get_and_check_slp_defs): Change the stmts parameter from
7939 a vec<gimple *> to a vec<stmt_vec_info>.
7940 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
7941 (vect_build_slp_tree): Likewise.
7942 (vect_build_slp_tree_2): Likewise. Update uses of
7943 SLP_TREE_SCALAR_STMTS.
7944 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
7945 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
7946 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
7947 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
7948 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
7949 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
7950 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
7951 (vect_get_constant_vectors, vect_get_slp_defs)
7952 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
7953 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
7954 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
7955 instead of gimple stmts.
7956 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
7957 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
7958 (vect_slp_analyze_instance_dependence): Update uses of
7959 SLP_TREE_SCALAR_STMTS.
7960 (vect_slp_analyze_and_verify_node_alignment): Likewise.
7961 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
7962 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
7963 (get_initial_defs_for_reduction): Likewise.
7964 (vect_create_epilog_for_reduction): Likewise.
7965 (vectorize_fold_left_reduction): Likewise.
7966 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
7967 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
7968 (can_vectorize_live_stmts): Likewise.
7970 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7972 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
7973 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
7974 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
7976 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
7978 (gather_scalar_reductions): Update after above interface changes.
7979 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
7980 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
7982 (vect_force_simple_reduction): Likewise.
7983 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
7984 LOOP_VINFO_REDUCTIONS.
7985 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
7987 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7989 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
7990 a gimple stmt to a stmt_vec_info.
7991 * tree-vect-loop.c (vect_active_double_reduction_p)
7992 (vect_force_simple_reduction, vectorizable_reduction): Update
7995 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7997 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
7998 vec<gimple *> to a vec<stmt_vec_info>.
7999 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
8000 the reduction_phis argument from a vec<gimple *> to a
8002 (vectorizable_reduction): Likewise the phis local variable that
8003 is passed to vect_create_epilog_for_reduction. Update for new type
8004 of SLP_TREE_VEC_STMTS.
8005 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
8006 (vectorizable_live_operation): Likewise.
8007 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
8008 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
8010 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8012 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
8013 a gimple stmt to a stmt_vec_info.
8014 (vectorizable_condition, vectorizable_live_operation)
8015 (vectorizable_reduction, vectorizable_induction): Pass back the
8016 vectorized statement as a stmt_vec_info.
8017 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
8018 use of STMT_VINFO_VEC_STMT.
8019 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
8020 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
8021 as stmt_vec_infos rather than gimple stmts.
8022 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
8024 (vectorizable_live_operation): Likewise.
8025 (vectorizable_reduction, vectorizable_induction): Likewise,
8026 updating use of STMT_VINFO_VEC_STMT.
8027 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
8028 of STMT_VINFO_VEC_STMT.
8029 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
8030 (vectorizable_simd_clone_call, vectorizable_conversion)
8031 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
8032 (vectorizable_store, vectorizable_load, vectorizable_condition)
8033 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
8034 from a gimple stmt to a stmt_vec_info.
8035 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
8036 pointer to a stmt_vec_info to the vectorizable_* routines.
8038 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8040 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
8041 a gimple stmt to a stmt_vec_info.
8042 (is_pattern_stmt_p): Update accordingly.
8043 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
8044 (vect_record_grouped_load_vectors): Likewise.
8045 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
8046 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
8047 (vect_model_reduction_cost): Likewise.
8048 (vect_create_epilog_for_reduction): Likewise.
8049 (vectorizable_reduction, vectorizable_induction): Likewise.
8050 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
8051 Return the stmt_vec_info for the pattern statement.
8052 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
8053 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
8054 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
8055 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
8056 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
8057 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
8058 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
8059 (free_stmt_vec_info, vect_is_simple_use): Likewise.
8061 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8063 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
8064 (vect_finish_stmt_generation): Likewise.
8065 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
8066 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
8067 (vect_build_gather_load_calls): Use the return value of the above
8068 functions instead of a separate call to vinfo_for_stmt. Use narrow
8069 scopes for the input gimple stmt and wider scopes for the associated
8070 stmt_vec_info. Use vec_info::lookup_def when setting these
8071 stmt_vec_infos from an SSA_NAME definition.
8072 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
8073 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
8074 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
8075 (vectorizable_store, vectorizable_load, vectorizable_condition)
8076 (vectorizable_comparison): Likewise.
8077 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
8078 (vectorizable_reduction): Likewise.
8080 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8082 * tree-vectorizer.h (vect_is_simple_use): Add an optional
8083 stmt_vec_info * parameter before the optional gimple **.
8084 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
8085 (process_use, vect_get_vec_def_for_operand_1): Update callers.
8086 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
8087 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8088 (vectorizable_live_operation): Likewise.
8089 * tree-vect-patterns.c (type_conversion_p): Likewise.
8090 (vect_look_through_possible_promotion): Likewise.
8091 (vect_recog_rotate_pattern): Likewise.
8092 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
8094 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8096 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
8097 a typedef to a wrapper class.
8098 (NULL_STMT_VEC_INFO): New macro.
8099 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
8100 (stmt_vec_info::operator*): New function.
8101 (stmt_vec_info::operator gimple *): Likewise.
8102 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
8103 (add_stmt_costs): Likewise.
8104 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
8105 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
8106 (vect_get_known_peeling_cost): Likewise.
8107 (vect_estimate_min_profitable_iters): Likewise.
8108 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
8109 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
8110 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
8111 (vectorizable_store, free_stmt_vec_infos): Likewise.
8112 (new_stmt_vec_info): Change return type of xcalloc to
8115 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8117 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
8118 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
8119 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
8120 a single_imm_use-based sequence.
8121 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
8123 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8125 * tree-vectorizer.h (vec_info::lookup_def): Declare.
8126 * tree-vectorizer.c (vec_info::lookup_def): New function.
8127 * tree-vect-patterns.c (vect_get_internal_def): Use it.
8128 (vect_widened_op_tree): Likewise.
8129 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
8130 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
8131 (vectorizable_reduction): Likewise.
8132 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
8134 (vect_is_slp_reduction): Update calls accordingly. Use
8135 vec_info::lookup_def.
8136 (vect_is_simple_reduction): Likewise
8137 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
8139 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8141 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
8142 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
8143 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
8145 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
8146 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
8147 (vect_update_vf_for_slp, vect_analyze_loop_operations)
8148 (vect_is_slp_reduction, vectorizable_induction)
8149 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
8150 * tree-vect-patterns.c (vect_init_pattern_stmt):
8151 (vect_determine_min_output_precision_1, vect_determine_precisions)
8152 (vect_pattern_recog): Likewise.
8153 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
8154 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
8155 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
8156 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
8157 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
8158 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
8159 info field from a loop to a loop_vec_info.
8161 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8163 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
8164 (vec_info::add_stmt): Declare.
8165 * tree-vectorizer.c (vec_info::add_stmt): New function.
8166 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
8167 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
8168 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
8169 (vectorizable_induction): Likewise.
8170 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
8171 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
8172 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
8173 (vectorizable_load): Likewise.
8174 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
8175 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
8176 (vect_recog_gather_scatter_pattern): Likewise.
8177 (append_pattern_def_seq): Likewise. Remove a check that is
8178 performed by add_stmt itself.
8180 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8182 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
8183 which make_ssa_name was called with new_stmt before new_stmt
8186 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8188 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
8190 (vect_is_slp_reduction): ...here...
8191 (vect_is_simple_reduction): ...and here. Remove repetition of tests
8192 that are already known to be false.
8194 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8196 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
8197 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
8198 STMT_VINFO_NUM_SLP_USES when it's true.
8199 (vect_free_slp_instance): Add a final_p parameter and pass it to
8201 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
8202 (vect_analyze_slp_instance): Likewise.
8203 (vect_slp_analyze_operations): Likewise.
8204 (vect_slp_analyze_bb_1): Likewise.
8205 * tree-vectorizer.c (vec_info): Likewise.
8206 * tree-vect-loop.c (vect_transform_loop): Likewise.
8208 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8210 * tree-vect-loop.c (vectorizable_reduction): Assert that the
8211 function is not called for second and subsequent members of
8214 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8216 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
8217 cases for nested loops from here to ...
8218 (vect_create_epilog_for_reduction): ...here. Only call
8219 vect_is_simple_use for inner-loop reductions.
8221 2018-07-31 Martin Liska <mliska@suse.cz>
8223 PR gcov-profile/85338
8224 PR gcov-profile/85350
8225 PR gcov-profile/85372
8226 * profile.c (struct location_triplet): New.
8227 (struct location_triplet_hash): Likewise.
8228 (output_location): Do not output a BB that
8229 is already recorded for a line.
8230 (branch_prob): Use streamed_locations.
8232 2018-07-31 Martin Liska <mliska@suse.cz>
8234 PR gcov-profile/85370
8235 * coverage.c (coverage_begin_function): Do not mark target
8236 clones as artificial functions.
8238 2018-07-31 Martin Liska <mliska@suse.cz>
8240 PR gcov-profile/83813
8241 PR gcov-profile/84758
8242 PR gcov-profile/85217
8243 PR gcov-profile/85332
8244 * profile.c (branch_prob): Do not record GOTO expressions
8245 for GIMPLE statements which locations are already streamed.
8247 2018-07-31 Olivier Hainque <hainque@adacore.com>
8249 * gcc.c (handle_spec_function): Accept a soft_matched_part
8250 argument, as do_spec_1. Pass it down to ...
8251 (eval_spec_function): Accept a soft_matched_part argument,
8252 and pass it down to ...
8253 (do_spec_2): Accept a soft_matched_part argument, and pass
8254 it down to do_spec_1.
8255 (do_spec_1): Pass soft_matched_part to handle_spec_function.
8256 (handle_braces): Update call to handle_spec_function.
8257 (driver::set_up_specs): Update calls to do_spec_2.
8258 (compare_debug_dump_opt_spec_function): Likewise.
8259 (compare_debug_self_opt_spec_function): Likewise.
8261 2018-07-31 Olivier Hainque <hainque@adacore.com>
8263 * common.opt (nolibc): New option.
8264 * doc/invoke.texi (Link Options): Document it.
8265 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
8266 * config/alpha/linux.h: Likewise.
8267 * config/arc/elf.h: Likewise.
8268 * config/arm/uclinux-elf.h: Likewise.
8269 * config/arm/unknown-elf.h: Likewise.
8270 * config/avr/avrlibc.h: Likewise.
8271 * config/bfin/bfin.h: Likewise.
8272 * config/bfin/linux.h: Likewise.
8273 * config/bfin/uclinux.h: Likewise.
8274 * config/darwin.h: Likewise.
8275 * config/darwin10.h: Likewise.
8276 * config/darwin12.h: Likewise.
8277 * config/gnu-user.h: Likewise.
8278 * config/lm32/uclinux-elf.h: Likewise.
8279 * config/pa/pa-hpux11.h: Likewise.
8280 * config/pa/pa64-hpux.h: Likewise.
8281 * config/sparc/sparc.h: Likewise.
8283 2018-07-31 Olivier Hainque <hainque@adacore.com>
8285 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
8286 undefined variables.
8288 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
8291 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
8294 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
8296 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
8297 terminated string literal.
8299 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
8301 PR rtl-optimization/85160
8302 * combine.c (is_just_move): New function.
8303 (try_combine): Allow combining two instructions into two if neither of
8304 the original instructions was a move.
8306 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
8309 * doc/extend.texi (Global Register Variables): Discourage use of type
8311 (Local Register Variables): Likewise.
8313 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
8315 PR tree-optimization/86506
8316 * hwint.h (ceil_log2): Resync with hwint.c implementation.
8318 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
8321 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
8322 hard_regno, make sure no insn between `from` and `to` clobbers it.
8324 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
8325 Tom de Vries <tdevries@suse.de>
8327 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
8328 (PTX_DEFAULT_RUNTIME_DIM): ... this.
8329 (nvptx_goacc_validate_dims): Set default worker and gang dims to
8330 PTX_DEFAULT_RUNTIME_DIM.
8331 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
8333 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
8335 * config/pa/pa.c (pa_output_addr_vec): Align address table.
8336 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
8337 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
8339 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
8341 * config/rs6000/constraints.md (wG constraint): Delete, no longer
8343 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
8344 predicate to reflect toc fusion has been deleted.
8345 (toc_fusion_mem_raw): Delete, no longer used.
8346 (toc_fusion_mem_wrapped): Likewise.
8347 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
8349 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
8350 Delete, no longer used.
8351 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
8352 meant to be used for toc fusion.
8353 (rs6000_debug_print_mode): Delete toc fusion debugging.
8354 (rs6000_debug_reg_global): Likewise.
8355 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
8356 fusion and secondary reload support that were never used.
8357 (rs6000_option_override_internal): Delete TOC fusion, that was only
8358 partially defined, and it did not work unless you also used the
8360 (rs6000_legitimate_address_p): Delete TOC fusion support.
8361 (rs6000_opt_masks): Likewise.
8362 (fusion_wrap_memory_address): Delete function, no longer used.
8363 (fusion_split_address); Delete TOC fusion support.
8364 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
8365 longer used with toc fusion being deleted.
8366 (TARGET_TOC_FUSION_FP): Likewise.
8367 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
8369 (toc fusion spliter): Delete TOC fusion support.
8370 (toc_fusionload_<mode>): Likewise.
8371 (toc_fusionload_di): Likewise.
8372 (fusion_gpr_load_<mode>): Delete generator function, this insn no
8373 longer needs to be named. Rename predicate to delete TOC fusion.
8374 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
8375 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
8376 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
8377 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
8378 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
8380 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
8382 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
8383 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
8384 __int128 in built-in function prototypes.
8385 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
8386 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
8388 2018-07-27 Martin Sebor <msebor@redhat.com>
8390 PR tree-optimization/86696
8391 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
8392 types, including enums.
8393 (handle_char_store): Be prepared for the above function to fail.
8395 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
8397 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
8398 when optimization level is lower than 2 or optimize for size.
8400 2018-07-26 Martin Sebor <msebor@redhat.com>
8402 PR tree-optimization/86043
8403 PR tree-optimization/86042
8404 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
8405 (get_string_cst_length): Rename...
8406 (get_min_string_length): ...to this. Add argument.
8407 (handle_char_store): Extend to handle multi-character stores by
8409 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
8410 * tree.h (initializer_zerop): Add argument.
8412 2018-07-26 Jakub Jelinek <jakub@redhat.com>
8415 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
8416 declare target to variables if they have always,{to,from,tofrom} map
8419 2018-07-26 Martin Liska <mliska@suse.cz>
8422 * lto-wrapper.c: Add linker_output as prefix
8423 for ltrans_output_file.
8425 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
8427 PR rtl-optimization/85805
8428 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
8429 value for hard registers if that was written in the same mode.
8431 2018-07-26 Martin Liska <mliska@suse.cz>
8433 PR gcov-profile/86536
8434 * gcov.c (format_gcov): Use printf format %.*f directly
8435 and do not handle special values.
8437 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
8439 * common/config/arc/arc-common.c (arc_option_optimization_table):
8440 Update default optimizations for size.
8442 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
8444 * config/arc/arc.md (movsf_insn): Add short instruction selection.
8445 * config/arc/constraints.md (CfZ): New constraint.
8446 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
8447 (subsf3_fpu): Likewise.
8448 (cmpsf_fpu): Likewise.
8449 (cmpsf_fpu_uneq): Likewise.
8451 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
8453 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
8455 (arc_is_uncached_mem_p): uncached applies to both the variable and
8458 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
8460 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
8463 2018-07-25 David Malcolm <dmalcolm@redhat.com>
8465 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
8466 field "m_scopes" from vec to auto_vec.
8468 2018-07-25 Martin Liska <mliska@suse.cz>
8470 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
8473 2018-07-25 Richard Biener <rguenther@suse.de>
8476 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
8477 special wrt context_die late.
8478 (gen_subprogram_die): Re-use DIEs in local scope.
8480 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
8482 PR tree-optimization/86644
8483 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
8485 2018-07-25 Martin Liska <mliska@suse.cz>
8488 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
8489 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
8491 2018-07-25 Martin Liska <mliska@suse.cz>
8494 * params.def: Explain ASan abbreviation and provide
8495 a documentation link.
8497 2018-07-24 Martin Sebor <msebor@redhat.com>
8499 PR tree-optimization/86622
8500 PR tree-optimization/86532
8501 * builtins.h (string_length): Declare.
8502 * builtins.c (c_strlen): Correct handling of non-constant offsets.
8503 (check_access): Be prepared for non-constant length ranges.
8504 (string_length): Make extern.
8505 * expr.c (string_constant): Only handle the minor non-constant
8506 array index. Use string_constant to compute the length of
8507 a generic string constant.
8509 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
8511 PR tree-optimization/86618
8512 * tree-vect-stmts.c (vectorizable_call): Don't take the address
8513 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
8515 2018-07-24 David Malcolm <dmalcolm@redhat.com>
8517 PR tree-optimization/86636
8518 * json.cc (json::object::set): Fix comment. Add assertions.
8519 (json::array::append): Move here from json.h. Add comment and an
8521 (json::string::string): Likewise.
8522 * json.h (json::array::append): Move to json.cc.
8523 (json::string::string): Likewise.
8524 * optinfo-emit-json.cc
8525 (optrecord_json_writer::impl_location_to_json): Assert that we
8526 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
8527 wrapper around it. Expand the location once, rather than three
8529 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
8530 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
8532 (optrecord_json_writer::optinfo_to_json): Likewise, in four
8533 places. Fix some overlong lines.
8535 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
8537 * config/aarch64/aarch64-simd.md
8538 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
8539 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
8540 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
8541 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
8542 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
8543 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
8544 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
8545 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
8546 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
8548 2018-07-24 Jakub Jelinek <jakub@redhat.com>
8551 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
8552 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
8553 and abs_d == d, do the power of two handling if profitable.
8555 2018-07-24 Richard Biener <rguenther@suse.de>
8557 * match.pd: Add BIT_FIELD_REF canonicalizations.
8559 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
8562 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
8564 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
8566 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
8567 terminated STRING_CST object.
8569 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
8571 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
8573 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
8575 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
8576 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
8577 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
8578 the elements into a register.
8579 (rs6000_split_v4si_init_di_reg): Delete.
8580 (rs6000_split_v4si_init): Delete.
8581 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
8582 (vsx_init_v4si): Rewrite as a define_expand.
8584 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
8586 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
8587 zero_extend argument from memory): New.
8589 2018-07-22 Martin Sebor <msebor@redhat.com>
8592 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
8593 diagnosing calls with unknown arguments unless -Walloca-larger-than
8594 is restricted to less than PTRDIFF_MAX bytes.
8596 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
8598 * doc/gcov.texi (Invoking Gcov): Editorial changes.
8600 2018-07-20 David Malcolm <dmalcolm@redhat.com>
8602 * pretty-print.c (text_info::set_location): Remove redundant
8603 "line_table" parameter from call to rich_location::set_range.
8605 2018-07-20 Martin Sebor <msebor@redhat.com>
8608 * builtins.c (expand_builtin_alloca): Adjust.
8609 * calls.c (alloc_max_size): Simplify.
8610 * cgraphunit.c (cgraph_node::expand): Adjust.
8611 * common.opt (larger_than_size, warn_frame_larger_than): Remove
8613 (frame_larger_than_size): Same.
8614 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
8615 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
8617 * doc/invoke.texi (GCC Command Options): Document option arguments.
8618 Explain byte-size arguments and suffixes.
8619 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
8620 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
8621 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
8622 * doc/options.texi (UInteger): Expand.
8623 (Host_Wide_Int, ByteSize): Document new properties.
8624 * final.c (final_start_function_1): Include sizes in an error message.
8625 * function.c (frame_offset_overflow): Same.
8626 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
8627 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
8628 Diagnose unbounded alloca calls only for limits of less than
8630 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
8631 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
8633 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
8634 only for limits of less than PTRDIFF_MAX.
8635 * langhooks-def.h (lhd_handle_option): Change function argument
8637 * langhooks.c (lhd_handle_option): Same.
8638 * langhooks.h (handle_option): Same.
8639 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
8641 (var_type, var_type_struct): Same.
8642 (var_set): Handle ByteSize flag.
8643 * optc-gen.awk: Add comments to output to ease debugging. Make
8644 use of HOST_WIDE_INT where appropriate.
8645 * opts-gen-save.awk: Use %lx to format unsigned long.
8646 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
8647 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
8648 arguments. Parse bytes-size suffixes.
8649 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
8650 (enum_value_to_arg): Same.
8651 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
8652 (handle_option): Adjust.
8653 (generate_option): Change function argument to HOST_WIDE_INT.
8654 (cmdline_handle_error): Adjust.
8655 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
8656 (set_option): Change function argument to HOST_WIDE_INT.
8657 (option_enabled): Handle cl_host_wide_int.
8658 (get_option_state): Handle CLVC_SIZE.
8659 (control_warning_option): Same.
8660 * opts.c (common_handle_option): Change function argument to
8661 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
8662 OPT_Wvla_larger_than_.
8663 * opts.h (enum cl_var_type): Add an enumerator.
8664 * stor-layout.c (layout_decl): Print a more meaningful warning.
8665 * toplev.c (output_stack_usage): Adjust.
8667 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
8669 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
8670 call to inline_expand_builtin_string_cmp.
8671 (expand_builtin_strcmp): Likewise.
8672 (expand_builtin_strncmp): Likewise.
8673 (inline_string_cmp): Delete the last parameter, change char_type_node
8674 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
8676 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
8677 the inlining expansion on target where the type of the call has same or
8678 narrower precision than unsigned char.
8680 2018-07-20 David Malcolm <dmalcolm@redhat.com>
8682 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
8683 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
8684 * common.opt (fsave-optimization-record): New option.
8685 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
8686 * doc/invoke.texi (-fsave-optimization-record): New option.
8687 * dumpfile.c: Include "optinfo-emit-json.h".
8688 (struct kv_pair): Move to coretypes.h.
8689 (optgroup_options): Make non-static.
8690 (dump_context::end_scope): Call
8691 optimization_records_maybe_pop_dump_scope.
8692 * dumpfile.h (optgroup_options): New decl.
8693 * json.cc: New file.
8695 * optinfo-emit-json.cc: New file.
8696 * optinfo-emit-json.h: New file.
8697 * optinfo.cc: Include "optinfo-emit-json.h".
8698 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
8699 (optinfo_enabled_p): Check optimization_records_enabled_p.
8700 (optinfo_wants_inlining_info_p): Likewise.
8701 * optinfo.h: Update comment.
8702 * profile-count.c (profile_quality_as_string): New function.
8703 * profile-count.h (profile_quality_as_string): New decl.
8704 (profile_count::quality): New accessor.
8705 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
8706 and optinfo_emit_json_cc_tests.
8707 * selftest.h (selftest::json_cc_tests): New decl.
8708 (selftest::optinfo_emit_json_cc_tests): New decl.
8709 * toplev.c: Include "optinfo-emit-json.h".
8710 (compile_file): Call optimization_records_finish.
8711 (do_compile): Call optimization_records_start.
8712 * tree-ssa-live.c: Include optinfo.h.
8713 (remove_unused_scope_block_p): Retain inlining information if
8714 optinfo_wants_inlining_info_p returns true.
8716 2018-07-20 Richard Biener <rguenther@suse.de>
8719 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
8720 to cover -flto-partition=none.
8722 2018-07-20 Martin Liska <mliska@suse.cz>
8724 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
8725 (get_decl_source_range): Remove unused function.
8727 2018-07-20 Richard Biener <rguenther@suse.de>
8729 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
8730 (struct vn_phi_s): Likewise.
8731 (struct vn_reference_s): Likewise.
8732 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
8733 for searching the slot of an entry known to be in the hash itself.
8734 (vn_phi_hasher::equal): Likewise.
8735 (vn_reference_hasher::equal): Likewise.
8736 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
8738 (optimistic_info, current_info): Remove, keeping only valid_info.
8739 (vn_reference_lookup_1): Remove fallback lookup.
8740 (vn_reference_lookup_2): Likewise.
8741 (vn_nary_op_lookup_1): Likewise.
8742 (vn_phi_lookup): Likewise.
8743 (vn_nary_build_or_lookup_1): Make sure to not chain the built
8745 (vn_reference_insert): Adjust, chain the inserted hash element
8746 at last_inserted_ref.
8747 (vn_reference_insert_pieces): Likewise.
8748 (visit_reference_op_call): Likewise.
8749 (vn_nary_op_insert_into): Chain the inserted hash element at
8751 (vn_nary_op_insert_pieces): Adjust.
8752 (vn_nary_op_insert): Likewise.
8753 (vn_nary_op_insert_stmt): Likewise.
8754 (vn_phi_insert): Adjust, chain the inserted hash element at
8756 (process_scc): Remove clearing and copying the optimistic
8757 table. Instead remove elements inserted during an optimistic
8758 iteration from the single table we maintain.
8759 (init_scc_vn): Adjust.
8760 (free_scc_vn): Likewise.
8761 (sccvn_dom_walker::record_cond): Likewise.
8762 (sccvn_dom_walker::after_dom_children): Likewise.
8764 2018-07-19 Martin Sebor <msebor@redhat.com>
8766 PR tree-optimization/84047
8767 PR tree-optimization/83776
8768 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
8769 (check_array_bounds): Call it.
8771 2018-07-19 Martin Sebor <msebor@redhat.com>
8773 * align.h (align_flags): Use member initialization.
8775 2018-07-19 David Malcolm <dmalcolm@redhat.com>
8777 * Makefile.in (OBJS): Add optinfo.o.
8778 * coretypes.h (class symtab_node): New forward decl.
8779 (struct cgraph_node): New forward decl.
8780 (class varpool_node): New forward decl.
8781 * dump-context.h: New file.
8782 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
8784 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
8785 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
8786 (set_alt_dump_file): Likewise.
8787 (dump_context::~dump_context): New dtor.
8788 (dump_gimple_stmt): Move implementation to...
8789 (dump_context::dump_gimple_stmt): ...this new member function.
8790 Add the stmt to any pending optinfo, creating one if need be.
8791 (dump_gimple_stmt_loc): Move implementation to...
8792 (dump_context::dump_gimple_stmt_loc): ...this new member function.
8793 Start a new optinfo and add the stmt to it.
8794 (dump_gimple_expr): Move implementation to...
8795 (dump_context::dump_gimple_expr): ...this new member function.
8796 Add the stmt to any pending optinfo, creating one if need be.
8797 (dump_gimple_expr_loc): Move implementation to...
8798 (dump_context::dump_gimple_expr_loc): ...this new member function.
8799 Start a new optinfo and add the stmt to it.
8800 (dump_generic_expr): Move implementation to...
8801 (dump_context::dump_generic_expr): ...this new member function.
8802 Add the tree to any pending optinfo, creating one if need be.
8803 (dump_generic_expr_loc): Move implementation to...
8804 (dump_context::dump_generic_expr_loc): ...this new member
8805 function. Add the tree to any pending optinfo, creating one if
8807 (dump_printf): Move implementation to...
8808 (dump_context::dump_printf_va): ...this new member function. Add
8809 the text to any pending optinfo, creating one if need be.
8810 (dump_printf_loc): Move implementation to...
8811 (dump_context::dump_printf_loc_va): ...this new member function.
8812 Start a new optinfo and add the stmt to it.
8813 (dump_dec): Move implementation to...
8814 (dump_context::dump_dec): ...this new member function. Add the
8815 value to any pending optinfo, creating one if need be.
8816 (dump_context::dump_symtab_node): New member function.
8817 (dump_context::get_scope_depth): New member function.
8818 (dump_context::begin_scope): New member function.
8819 (dump_context::end_scope): New member function.
8820 (dump_context::ensure_pending_optinfo): New member function.
8821 (dump_context::begin_next_optinfo): New member function.
8822 (dump_context::end_any_optinfo): New member function.
8823 (dump_context::s_current): New global.
8824 (dump_context::s_default): New global.
8825 (dump_scope_depth): Delete global.
8826 (dumpfile_ensure_any_optinfo_are_flushed): New function.
8827 (dump_symtab_node): New function.
8828 (get_dump_scope_depth): Reimplement in terms of dump_context.
8829 (dump_begin_scope): Likewise.
8830 (dump_end_scope): Likewise.
8831 (selftest::temp_dump_context::temp_dump_context): New ctor.
8832 (selftest::temp_dump_context::~temp_dump_context): New dtor.
8833 (selftest::verify_item): New function.
8834 (ASSERT_IS_TEXT): New macro.
8835 (ASSERT_IS_TREE): New macro.
8836 (ASSERT_IS_GIMPLE): New macro.
8837 (selftest::test_capture_of_dump_calls): New test.
8838 (selftest::dumpfile_c_tests): Call it.
8839 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
8840 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
8841 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
8842 descriptive comment.
8843 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
8844 (dump_node, dump_bb): Move these unrelated decls.
8845 (class dump_manager): Add leading comment.
8846 * optinfo.cc: New file.
8847 * optinfo.h: New file.
8849 2018-07-19 Michael Collison <michael.collison@arm.com>
8850 Richard Henderson <rth@redhat.com>
8852 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
8853 (subti): Handle op1 zero.
8854 (subvti4, usub4ti4): New.
8855 (*sub<GPI>3_compare1_imm): New.
8856 (sub<GPI>3_carryinCV): New.
8857 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
8858 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
8860 2018-07-19 Michael Collison <michael.collison@arm.com>
8861 Richard Henderson <rth@redhat.com>
8863 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
8864 (addti3): Create simpler code if low part is already known to be 0.
8865 (addvti4, uaddvti4): New.
8866 (*add<GPI>3_compareC_cconly_imm): New.
8867 (*add<GPI>3_compareC_cconly): New.
8868 (*add<GPI>3_compareC_imm): New.
8869 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
8870 handle constants within this pattern..
8871 (*add<GPI>3_compareV_cconly_imm): New.
8872 (*add<GPI>3_compareV_cconly): New.
8873 (*add<GPI>3_compareV_imm): New.
8874 (add<GPI>3_compareV): New.
8875 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
8876 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
8877 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
8878 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
8879 with 'comparison' operator.
8880 (*add<GPI>3_compareV_cconly_imm): Ditto.
8881 (*add<GPI>3_compareV_cconly): Ditto.
8882 (*add<GPI>3_compareV_imm): Ditto.
8883 (add<GPI>3_compareV): Ditto.
8884 (add<mode>3_carryinC): Ditto.
8885 (*add<mode>3_carryinC_zero): Ditto.
8886 (*add<mode>3_carryinC): Ditto.
8887 (add<mode>3_carryinV): Ditto.
8888 (*add<mode>3_carryinV_zero): Ditto.
8889 (*add<mode>3_carryinV): Ditto.
8891 2018-07-19 Michael Collison <michael.collison@arm.com>
8892 Richard Henderson <rth@redhat.com>
8894 * config/aarch64/aarch64-modes.def (CC_V): New.
8895 * config/aarch64/aarch64-protos.h
8896 (aarch64_addti_scratch_regs): Declare
8897 (aarch64_subvti_scratch_regs): Declare.
8898 (aarch64_expand_subvti): Declare.
8899 (aarch64_gen_unlikely_cbranch): Declare
8900 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
8901 for signed overflow using CC_Vmode.
8902 (aarch64_get_condition_code_1): Handle CC_Vmode.
8903 (aarch64_gen_unlikely_cbranch): New function.
8904 (aarch64_addti_scratch_regs): New function.
8905 (aarch64_subvti_scratch_regs): New function.
8906 (aarch64_expand_subvti): New function.
8908 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
8910 * config/aarch64/aarch64-option-extensions.def: New entry for profile
8912 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
8913 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
8916 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
8919 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
8920 address check not strict.
8922 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
8924 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
8926 (store_pair_lanes<mode>): Likewise.
8927 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
8928 enum value 'ADDR_QUERY_LDP_STP_N'.
8929 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
8930 (aarch64_print_address_internal): Add declaration.
8931 (aarch64_print_ldpstp_address): Remove.
8932 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
8933 (aarch64_print_operand): Change printing of 'y'.
8934 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
8935 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
8936 'true' rather than '1'.
8937 * config/aarch64/constraints.md (Uml): Likewise.
8938 (Uml): Rename to Umn.
8941 2018-07-19 Richard Biener <rguenther@suse.de>
8943 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
8945 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
8946 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
8947 (vn_reference_hasher): Likewise.
8948 (struct vn_tables_s): Remove obstack and alloc-pool members.
8949 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
8950 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
8951 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
8952 (vn_reference_insert_pieces): Likewise.
8953 (alloc_vn_nary_op_noinit): Adjust.
8954 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
8955 (vn_phi_eq): Adjust.
8956 (shared_lookup_phiargs): Remove.
8957 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
8958 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
8959 (visit_reference_op_call): Likewise.
8960 (copy_nary, copy_phi, copy_reference): Remove.
8961 (process_scc): Rewind the obstack when iterating. Do not
8962 copy the elements to valid_info but just move them from one
8963 hashtable to the other.
8964 (allocate_vn_table): Adjust.
8965 (free_vn_table): Likewise.
8966 (init_scc_vn): Likewise.
8967 (free_scc_vn): Likewise.
8969 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
8972 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
8973 indirect_return as function type attribute.
8974 (ix86_attribute_table): Change indirect_return to function
8976 * doc/extend.texi: Update indirect_return attribute.
8978 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
8980 * wide-int.h (widest2_int): New.
8981 * gimple-fold.c (arith_overflowed_p): Use it.
8982 * tree.h (widest2_int_cst): New.
8983 * tree-vrp.c (wide_int_binop_overflow): Rename from
8984 vrp_int_const_binop.
8985 Rewrite to work on trees.
8986 (extract_range_from_multiplicative_op_1): Abstract code to...
8987 (wide_int_range_min_max): ...here.
8988 (wide_int_range_cross_product): ...and here.
8989 (extract_range_from_binary_expr_1): Abstract overflow code to...
8990 (wide_int_range_mult_wrapping): ...here.
8991 * tree-vrp.h (wide_int_range_cross_product): New.
8992 (wide_int_range_mult_wrapping): New.
8994 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
8995 Julia Koval <julia.koval@intel.com>
8997 * config/i386/x86-tune-costs.h (skylake_memcpy,
8998 skylake_memset): Replace rep_prefix with unrolling for size 512.
9000 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
9003 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
9004 comparision with EQ_EXPR in last stmt.
9006 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
9008 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
9009 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
9010 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
9011 previously known as "PowerPC AltiVec Built-in Functions". Move
9012 some material to new subsubsections "PowerPC AltiVec Built-in
9013 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
9015 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
9016 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
9017 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
9018 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
9020 2018-07-18 Richard Biener <rguenther@suse.de>
9022 PR tree-optimization/86557
9023 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
9026 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
9028 * config/s390/s390.c (s390_function_profiler): Generate CFI.
9030 2018-07-17 Jeff Law <law@redhat.com>
9032 * config/arm/arm.c (get_label_padding): Update for recent
9033 changes to label_to_alignment.
9035 PR tree-optimization/86010
9036 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
9038 * config/mips/mips.c (vr4130_align_insns): Update for recent
9039 changes to label_to_alignment.
9041 * config/frv/frv.c (frv_label_align): Update for recent changes
9042 to label_to_alignment.
9044 * config/nios2/nios2.c (nios2_label_align): Update for recent
9045 changes which dropped ALIGN_LABELS_LOG.
9047 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
9049 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
9050 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
9052 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9054 * config/arc/arc.c (arc_label_align): Use align_labels instead of
9055 deprecated align_labels_log.
9057 2018-07-17 Richard Biener <rguenther@suse.de>
9060 * dwarf2out.c (init_sections_and_labels): Always generate
9061 a debug_line_str_section for early LTO debug.
9062 (dwarf2out_finish): Reset debug_line_str_hash output early.
9063 Bump counter for extra dwarf5 .debug_loc labels to not conflict
9065 (dwarf2out_early_finish): Output debug_line_str.
9067 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
9069 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
9070 index register on z196 or later.
9072 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
9074 * config/s390/s390.c (s390_default_align): Set default function
9076 (s390_override_options_after_change): Call s390_default align.
9077 (s390_option_override_internal): Call s390_default_align.
9078 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
9080 2018-07-17 Jakub Jelinek <jakub@redhat.com>
9083 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
9084 to _looptemp_ clauses, other than the first two.
9086 2018-07-17 Martin Liska <mliska@suse.cz>
9088 * opts.c: Do not enable OPT_falign_* for -Os.
9090 2018-07-17 Martin Liska <mliska@suse.cz>
9092 * align.h (MAX_CODE_ALIGN): New.
9093 (MAX_CODE_ALIGN_VALUE): New.
9094 * common/config/i386/i386-common.c (ix86_handle_option):
9095 (MAX_CODE_ALIGN): Moved to align.h.
9096 * final.c (MAX_CODE_ALIGN): Likewise.
9097 * opts.c (parse_and_check_align_values):
9098 (MAX_CODE_ALIGN): Likewise.
9099 (MAX_CODE_ALIGN_VALUE): Likewise.
9101 2018-07-17 Martin Liska <mliska@suse.cz>
9103 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
9104 in order to fulfil coding style.
9105 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
9106 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
9107 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
9108 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
9109 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
9110 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
9111 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
9112 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
9113 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
9115 2018-07-17 Martin Liska <mliska@suse.cz>
9117 * align.h: New file.
9118 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
9120 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
9121 align_flags of label_to_alignment.
9122 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
9124 * config/m68k/m68k.c: Do not use removed align_labels_value and
9126 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
9127 (LOOP_ALIGN): Likewise.
9128 (LABEL_ALIGN): Likewise.
9129 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
9130 Remove not used macro.
9131 (rs6000_loop_align): Change return type to align_flags.
9132 (rs6000_loop_align_max_skip): Remove.
9133 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
9134 Change return type to align_flags.
9135 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
9136 Remove not used macro.
9137 (rs6000_loop_align): Change return type to align_flags.
9138 (rs6000_loop_align_max_skip): Remove.
9139 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
9140 * config/rx/rx-protos.h (rx_align_for_label): Make it
9142 * config/rx/rx.c (rx_align_for_label): Change return type
9144 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
9146 into align_flags class.
9147 (LABEL_ALIGN): Likewise.
9148 (LOOP_ALIGN): Likewise.
9149 * config/s390/s390.c (s390_label_align): Use align_flags
9151 (s390_asm_output_function_label): Likewise.
9152 * config/sh/sh.c (sh_override_options_after_change):
9153 Use align_flags class directly without macros.
9154 (find_barrier): Likewise.
9155 (barrier_align): Likewise.
9156 (sh_loop_align): Likewise.
9157 * config/spu/spu.c (spu_option_override):
9158 Use align_flags_tuple::get_value instead of removed macros.
9159 (spu_sched_init): Likewise.
9160 * config/spu/spu.h (GTY): Likewise.
9161 * config/visium/visium.c (visium_option_override):
9162 Set "8" as default secondary alignment.
9163 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
9164 in order to guarantee secondary alignment of 8.
9165 * coretypes.h: Include align.h header file.
9166 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
9167 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
9168 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
9169 * doc/tm.texi.in: Likewise.
9170 * final.c (struct label_alignment): Remove not used structure.
9171 (LABEL_ALIGN): Change type to align_flags.
9172 (LOOP_ALIGN): Likewise.
9173 (JUMP_ALIGN): Likewise.
9174 (default_loop_align_max_skip): Remove.
9175 (default_label_align_max_skip): Likewise.
9176 (default_jump_align_max_skip): Likewise.
9177 (default_label_align_after_barrier_max_skip):
9178 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
9179 (LABEL_TO_MAX_SKIP): Remove.
9180 (label_to_alignment): Return align_flags type instead of integer.
9181 (label_to_max_skip): Remove.
9182 (align_fuzz): Use align_flags type.
9183 (compute_alignments): Use align_flags type and use align_flags::max
9184 to combine multiple alignments.
9185 (grow_label_align): Grow vec instead of C array.
9186 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
9187 (shorten_branches): Use align_flags type and use align_flags::max
9188 to combine multiple alignments.
9189 (final_scan_insn_1): Remove usage of secondary alignment that comes
9190 from label alignment, but instead use proper secondary alignment
9191 which is computed in grow_label_align.
9192 * flags.h (struct align_flags_tuple): Move to align.h.
9193 (struct align_flags): Likewise.
9194 (state_align_loops): Rename to align_loops.
9195 (state_align_jumps): Rename to align_jumps.
9196 (state_align_labels): Rename to align_labels.
9197 (state_align_functions): Rename to align_functions.
9198 (align_loops_log): Remove.
9199 (align_jumps_log): Remove.
9200 (align_labels_log): Remove.
9201 (align_functions_log): Remove.
9202 (align_loops_max_skip): Remove.
9203 (align_jumps_max_skip): Remove.
9204 (align_labels_max_skip): Remove.
9205 (align_functions_max_skip): Remove.
9206 (align_loops_value): Remove.
9207 (align_jumps_value): Remove.
9208 (align_labels_value): Remove.
9209 (align_functions_value): Remove.
9210 * output.h (label_to_alignment): Change return type to align_flags.
9211 (label_to_max_skip): Remove.
9212 * target.def: Remove loop_align_max_skip, label_align_max_skip,
9213 jump_align_max_skip macros.
9214 * targhooks.h (default_loop_align_max_skip): Remove.
9215 (default_label_align_max_skip): Likewise.
9216 (default_jump_align_max_skip): Likewise.
9217 (default_label_align_after_barrier_max_skip): Remove.
9218 * toplev.c (read_log_maxskip): Use ::normalize function.
9219 (parse_N_M): Remove not used argument and also call ::normalize.
9220 (parse_alignment_opts): Do not pass unused arguments.
9221 * varasm.c (assemble_start_function): Use directly align_functions
9222 instead of removed macros.
9223 * system.h: Do not poison removed macros.
9225 2018-07-17 Jakub Jelinek <jakub@redhat.com>
9228 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
9229 and cond temporaries don't have reference type if iterator has
9230 pointer type. For init use &for_pre_body instead of pre_p if
9231 for_pre_body is non-empty.
9233 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
9235 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
9236 double-double modes to SFmode directly directly.
9237 (trunc<mode>sf2_fprs): Delete.
9239 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
9241 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
9242 for conversions between IFmode and the decimal floating point modes.
9243 (init_float128_ieee): Use the correct names for conversions between
9244 KFmode and the decimal floating point modes.
9246 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
9248 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
9249 for the conversions between TDmode and IFmode.
9250 (init_float128_ieee): Use more correct names for the conversions
9251 between TDmode and KFmode.
9253 2018-07-16 Jakub Jelinek <jakub@redhat.com>
9255 PR tree-optimization/86526
9256 * builtins.c (expand_builtin_memcmp): Formatting fixes.
9257 (inline_expand_builtin_string_cmp): Likewise.
9258 (inline_string_cmp): Likewise. Use c_readstr instead of
9259 builtin_memcpy_read_str. Add unit_mode temporary.
9261 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
9264 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
9265 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
9267 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
9269 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
9270 Alphabetize prototypes of built-in functions, separating out
9271 built-in functions that are listed in this section but should be
9272 described elsewhere.
9274 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
9277 * expmed.c (emit_store_flag): Do not emit setcc followed by a
9278 conditional move when trapping comparison was split to a
9279 non-trapping one (and vice versa).
9281 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
9283 * config/s390/s390.c (s390_function_profiler): Generate nops
9284 instead of profiler call sequences.
9285 * config/s390/s390.opt: Add the new option.
9287 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
9289 * config/s390/s390.c (s390_function_profiler): Generate
9290 __mcount_loc section.
9291 * config/s390/s390.opt: Add the new option.
9293 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
9295 * common.opt: Add the new warning.
9296 * config/s390/s390.c (s390_function_profiler): Emit "brasl
9297 %r0,__fentry__" when -mfentry is specified.
9298 (s390_option_override_internal): Disallow -mfentry for 31-bit
9300 * config/s390/s390.opt: Add the new option.
9302 2018-07-16 Richard Biener <rguenther@suse.de>
9305 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
9306 for function-local FUNCTION_DECL and RESULT_DECL immediately.
9308 2018-07-16 Martin Liska <mliska@suse.cz>
9311 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
9314 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
9316 * config/arc/arcHS.md: Update ARCHS scheduling rules.
9318 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
9320 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
9322 * config/arc/arc-cpus.def (hs4x): New cpu.
9324 * config/arc/arc-tables.opt: Regenerate.
9325 * config/arc/arc.c (arc_sched_issue_rate): New function.
9326 (TARGET_SCHED_ISSUE_RATE): Define.
9327 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
9328 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
9330 (attr tune): Add ARCHS4x tune values.
9331 (attr tune_dspmpy): Define.
9332 (*tst): Correct instruction type.
9333 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
9334 * config/arc/arcHS4x.md: New file.
9335 * config/arc/fpu.md: Update instruction type attributes.
9336 * config/arc/t-multilib: Regenerate.
9338 2018-07-16 Tom de Vries <tdevries@suse.de>
9341 * var-tracking.c (vt_initialize): Fix pre_dec handling.
9343 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9345 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
9348 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
9350 PR tree-optimization/86514
9351 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
9352 conversion to a boolean type from a type with greater precision.
9354 2018-07-16 Tom de Vries <tdevries@suse.de>
9356 * var-tracking.c (vt_initialize): Print adjusted insn slim if
9357 dump_flags request TDF_SLIM.
9359 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
9361 * fold-const.c (int_const_binop_1): Abstract...
9362 (wide_int_binop): ...wide int code here.
9363 (poly_int_binop): ...poly int code here.
9364 Abstract the rest of int_const_binop_1 into int_const_binop.
9365 * fold-const.h (wide_int_binop): New.
9366 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
9367 Remove useless PLUS/MINUS_EXPR case.
9368 (zero_nonzero_bits_from_vr): Move wide int code...
9369 (zero_nonzero_bits_from_bounds): ...here.
9370 (extract_range_from_binary_expr_1): Move mask optimization code...
9371 (range_easy_mask_min_max): ...here.
9372 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
9373 (range_easy_mask_min_max): New.
9375 2018-07-15 Jeff Law <law@redhat.com>
9378 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
9381 2018-07-14 Jim Wilson <jimw@sifive.com>
9383 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
9385 2018-07-14 Paul Koning <ni1d@arrl.net>
9387 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
9389 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
9391 * lto-streamer-out.c (copy_function_or_variable): Dump info about
9394 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
9395 Steve Munroe <munroesj52@gmail.com>
9397 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
9398 (_mm_andnot_si128): Likewise.
9399 (_mm_or_si128): Likewise.
9400 (_mm_xor_si128): Likewise.
9402 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
9405 * builtins.c (expand_builtin_memcmp): Inline the calls first
9406 when result_eq is false.
9407 (expand_builtin_strcmp): Inline the calls first.
9408 (expand_builtin_strncmp): Likewise.
9409 (inline_string_cmp): New routine. Expand a string compare
9410 call by using a sequence of char comparison.
9411 (inline_expand_builtin_string_cmp): New routine. Inline expansion
9412 a call to str(n)cmp/memcmp.
9413 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
9415 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
9417 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
9419 * config/arm/driver-arm.c: Include arm-native.h.
9420 (host_detect_local_cpu): Use auto-generated data tables.
9421 (vendors, arm_cpu_table): Delete. Move part information to ...
9422 * config/arm/arm-cpus.in: ... here.
9423 * config/arm/parsecpu.awk (gen_native): New function.
9424 (vendor, part): New CPU fields.
9425 (END): Add support for building the native CPU detection tables.
9426 * config/arm/t-arm (arm-native.h): Add build rule.
9427 (driver-arm.o): Add dependency on arm-native.h.
9429 2018-07-13 Richard Biener <rguenther@suse.de>
9432 * match.pd (addr1 - addr2): Allow either of the operand to
9435 2018-07-13 Tom de Vries <tdevries@suse.de>
9437 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
9440 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
9442 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
9443 arrays instead of numbered variables.
9445 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
9447 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
9448 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
9450 2018-07-13 Richard Biener <rguenther@suse.de>
9453 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
9454 instead of get_context_die.
9456 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
9457 Richard Biener <rguenther@suse.de>
9460 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
9461 that the loop latch destination where phi is defined.
9463 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
9465 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
9466 (riscv_expand_epilogue): Add assertion to check interrupt mode.
9467 (riscv_set_current_function): Extract getting interrupt type to new
9469 (riscv_get_interrupt_type): New function.
9470 (riscv_merge_decl_attributes): New function, checking interrupt type is
9472 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
9474 2018-07-12 Paul Koning <ni1d@arrl.net>
9476 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
9479 2018-07-12 Paul Koning <ni1d@arrl.net>
9481 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
9482 zero reference, add doloop_end instead.
9483 * doc/md.texi (decrement_and_branch_until_zero): Remove.
9484 (Looping patterns): Remove decrement_and_branch_until_zero. Add
9485 detail for doloop_end.
9487 2018-07-12 Martin Sebor <msebor@redhat.com>
9490 * attribs.c (decl_attributes): Reject conflicting attributes before
9491 calling attribute handlers.
9493 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
9495 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
9497 (gcc::dump_manager::get_dump_file_name): likewise.
9498 (dump_begin): Likewise.
9499 * dumpfile.h (dump_begin): Update prototype.
9500 (gcc::dump_manager::get_dump_file_name,
9501 gcc::dump_manager::get_dump_file_name): Update prototype.
9503 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
9505 * internal-fn.h (vectorizable_internal_fn_p): New function.
9506 * tree-vect-slp.c (compatible_calls_p): Likewise.
9507 (vect_build_slp_tree_1): Remove nops argument. Handle calls
9508 to internal functions.
9509 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
9511 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
9513 * fold-const.h (inverse_conditions_p): Declare.
9514 * fold-const.c (inverse_conditions_p): New function.
9515 * match.pd: Use inverse_conditions_p. Add folds of view_converts
9516 that test the inverse condition of a conditional internal function.
9517 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
9518 * internal-fn.c (internal_fn_mask_index): Handle conditional
9520 (vectorized_internal_fn_supported_p): New function.
9521 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
9522 (any_pred_load_store): Replace with...
9523 (need_to_predicate): ...this new variable.
9524 (redundant_ssa_names): New variable.
9525 (ifcvt_can_use_mask_load_store): Move initial checks to...
9526 (ifcvt_can_predicate): ...this new function. Handle tree codes
9527 for which a conditional internal function exists.
9528 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
9529 instead of ifcvt_can_use_mask_load_store. Update after variable
9531 (predicate_load_or_store): New function, split out from
9532 predicate_mem_writes.
9533 (check_redundant_cond_expr): New function.
9534 (value_available_p): Likewise.
9535 (predicate_rhs_code): Likewise.
9536 (predicate_mem_writes): Rename to...
9537 (predicate_statements): ...this. Use predicate_load_or_store
9538 and predicate_rhs_code.
9539 (combine_blocks, tree_if_conversion): Update after above name changes.
9540 (ifcvt_local_dce): Handle redundant_ssa_names.
9541 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
9542 general conditional functions.
9543 * tree-vect-stmts.c (vectorizable_call): Likewise.
9545 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
9546 Alan Hayward <alan.hayward@arm.com>
9547 David Sherwood <david.sherwood@arm.com>
9549 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
9550 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
9551 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
9552 plus and minus and convert them into IFN_COND_FMA-based sequences.
9553 (convert_mult_to_fma): Handle conditional plus and minus.
9555 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
9557 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
9558 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
9559 (cond_fnms_optab): New optabs.
9560 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
9562 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
9563 * internal-fn.h (get_conditional_internal_fn): Declare.
9564 (get_unconditional_internal_fn): Likewise.
9565 * internal-fn.c (cond_ternary_direct): New macro.
9566 (expand_cond_ternary_optab_fn): Likewise.
9567 (direct_cond_ternary_optab_supported_p): Likewise.
9568 (FOR_EACH_COND_FN_PAIR): Likewise.
9569 (get_conditional_internal_fn): New function.
9570 (get_unconditional_internal_fn): Likewise.
9571 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
9572 (gimple_match_op::gimple_match_op): Add a new overload for 5
9574 (gimple_match_op::set_op): Likewise.
9575 (gimple_resimplify5): Declare.
9576 * genmatch.c (decision_tree::gen): Generate simplifications for
9578 * gimple-match-head.c (gimple_simplify): Define an overload for
9579 5 operands. Handle calls with 5 arguments in the top-level overload.
9580 (convert_conditional_op): Handle conversions from unconditional
9581 internal functions to conditional ones.
9582 (gimple_resimplify5): New function.
9583 (build_call_internal): Pass a fifth operand.
9584 (maybe_push_res_to_seq): Likewise.
9585 (try_conditional_simplification): Try converting conditional
9586 internal functions to unconditional internal functions.
9587 Handle 3-operand unconditional forms.
9588 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
9589 Define ternary equivalents of the current rules for binary conditional
9591 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
9593 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
9594 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
9595 (optab): Handle them.
9596 (SVE_COND_FP_TERNARY): New int iterator.
9597 (sve_fmla_op, sve_fmad_op): New int attributes.
9598 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
9599 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
9600 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
9602 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
9604 * target.def (preferred_else_value): New target hook.
9605 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
9606 * doc/tm.texi: Regenerate.
9607 * targhooks.h (default_preferred_else_value): Declare.
9608 * targhooks.c (default_preferred_else_value): New function.
9609 * internal-fn.h (conditional_internal_fn_code): Declare.
9610 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
9611 (get_conditional_internal_fn): Use it.
9612 (conditional_internal_fn_code): New function.
9613 * gimple-match.h (gimple_match_cond): New struct.
9614 (gimple_match_op): Add a cond member function.
9615 (gimple_match_op::gimple_match_op): Update all forms to take a
9617 * genmatch.c (expr::gen_transform): Use the same condition as res_op
9618 for the suboperation, but don't specify a particular else_value.
9619 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
9620 (visit_nary_op, visit_reference_op_load): Pass
9621 gimple_match_cond::UNCOND to the gimple_match_op constructor.
9622 * gimple-match-head.c: Include tree-eh.h
9623 (convert_conditional_op): New function.
9624 (maybe_resimplify_conditional_op): Likewise.
9625 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
9626 (gimple_resimplify2): Likewise.
9627 (gimple_resimplify3): Likewise.
9628 (gimple_resimplify4): Likewise.
9629 (maybe_push_res_to_seq): Return null for conditional operations.
9630 (try_conditional_simplification): New function.
9631 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
9633 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
9635 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
9637 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
9639 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
9641 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
9643 (hash_tree): Do not hash DECL_FCONTEXT
9644 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
9645 Do not stream DECL_FCONTEXT.
9646 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
9647 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
9649 2018-07-12 Richard Biener <rguenther@suse.de>
9652 * dwarf2out.c (gen_block_die): Only output blocks when they have
9653 at least one !DECL_IGNORED_P variable.
9655 2018-07-12 Richard Biener <rguenther@suse.de>
9658 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
9659 Remove -mieee-fp handling.
9661 2018-07-12 Richard Biener <rguenther@suse.de>
9663 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
9664 left-over from last patch.
9666 2018-07-12 Jakub Jelinek <jakub@redhat.com>
9668 PR tree-optimization/86492
9669 * gimple-ssa-store-merging.c
9670 (imm_store_chain_info::coalesce_immediate_stores): Call
9671 check_no_overlap even for the merge_overlapping case. Formatting fix.
9673 2018-07-12 Richard Biener <rguenther@suse.de>
9676 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
9677 move possibly trapping operations into the conditional.
9679 2018-07-12 Richard Biener <rguenther@suse.de>
9681 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
9682 (vn_lookup_simplify_result): Remove recursion limit applied
9684 (vn_nary_build_or_lookup_1): Adjust.
9685 (try_to_simplify): Likewise.
9686 * gimple-match-head.c (gimple_resimplify1): Instead apply one
9688 (gimple_resimplify2): Likewise.
9689 (gimple_resimplify3): Likewise.
9690 (gimple_resimplify4): Likewise.
9692 2018-07-11 Jakub Jelinek <jakub@redhat.com>
9694 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
9695 Use __mmask64 type instead of __mmask8 for __M argument.
9696 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
9697 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
9699 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
9700 __mmask16 for __M argument.
9701 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
9702 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
9703 to __mmask16 instead of __mmask8.
9704 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
9705 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
9706 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
9707 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
9708 instead of __mmask16 for __U argument.
9709 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
9710 __mmask16 instead of __mmask8 for __U argument.
9711 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
9713 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
9715 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
9717 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
9719 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
9721 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
9723 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
9725 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
9727 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
9729 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
9731 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
9732 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
9733 return type as well as __M argument type and all casts from __mmask8
9735 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
9736 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
9737 return type as well as __M argument type and all casts from __mmask8
9739 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
9740 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
9741 return type as well as __M argument type and all casts from __mmask8
9743 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
9744 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
9745 return type as well as __M argument type and all casts from __mmask8
9747 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
9748 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
9751 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
9753 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
9754 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
9757 2018-07-11 Paul Koning <ni1d@arrl.net>
9759 * doc/md.texi (define_subst): Document how multiple occurrences of
9760 the same argument in the replacement pattern are handled.
9762 2018-07-11 Paul Koning <ni1d@arrl.net>
9764 * doc/extend.texi (Common Variable Attributes): Move "mode" into
9766 (Common Type Attributes): Add "mode" attribute.
9768 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
9770 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
9771 stream DECL_ORIGINAL_TYPE.
9772 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
9773 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
9774 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
9775 Do not walk original type.
9776 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
9778 (write_ts_decl_non_common_tree_pointers): Do not stream
9780 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
9781 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
9783 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
9785 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
9786 thread twice from the same starting edge.
9788 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
9790 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
9791 * gimple.c (gimple_call_nonnull_result_p): ...here...
9792 (gimple_call_nonnull_arg): ...and here.
9793 * gimple.h (gimple_call_nonnull_result_p): New.
9794 (gimple_call_nonnull_arg): New.
9796 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
9798 * config/arm/arm-cpus.in: Move information from fpu field of each
9799 cpu definition to the isa field.
9800 * config/arm/parsecpu.awk (fpu): Delete match rule.
9801 (gen_comm_data): Don't add bits from the CPU's FPU entry.
9803 2018-07-11 Richard Biener <rguenther@suse.de>
9806 * dwarf2out.c (init_sections_and_labels): Use
9807 output_asm_line_debug_info consistently.
9808 (dwarf2out_early_finish): Likewise.
9809 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
9812 2018-07-11 Richard Biener <rguenther@suse.de>
9814 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
9815 Rework father_bb setting in a way to avoid propagating constants
9816 multiple times on a loop body.
9818 2018-07-10 Mark Wielaard <mark@klomp.org>
9821 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
9823 2018-07-10 Richard Biener <rguenther@suse.de>
9825 * hash-map.h (hash_map::iterator::operator*): Return
9826 references to key and value.
9828 2018-07-10 Jakub Jelinek <jakub@redhat.com>
9831 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
9832 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
9833 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
9834 (gimplify_omp_for): For composite loops, move outer
9835 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
9836 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
9837 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
9838 TREE_LIST for both the original class iterator and the "last" helper
9839 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
9840 loop, remember has_decl_expr from outer composite loops for the
9841 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
9843 2018-07-09 Martin Sebor <msebor@redhat.com>
9847 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
9848 accessing implicitly initialized array elements.
9849 * expr.c (string_constant): Handle string initializers of
9850 character arrays within aggregates.
9851 * gimple-fold.c (fold_array_ctor_reference): Add argument.
9852 Store element offset. As a special case, handle zero size.
9853 (fold_nonarray_ctor_reference): Same.
9854 (fold_ctor_reference): Add argument. Store subobject offset.
9855 * gimple-fold.h (fold_ctor_reference): Add argument.
9857 2018-07-09 Paul Koning <ni1d@arrl.net>
9859 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
9860 (pdp11_insn_cost): New function.
9861 (pdp11_md_asm_adjust): New function.
9862 (TARGET_INVALID_WITHIN_DOLOOP): Define.
9863 (pdp11_rtx_costs): Update to match machine better.
9864 (output_addr_const_pdp11): Correct format mismatch warnings.
9865 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
9866 * config/pdp11/pdp11.md: General change to add base_cost and/or
9867 length attributes for use by new pdp11_insn_cost function.
9868 (MIN_BRANCH): Correct definition.
9870 (doloop_end): Use standard pattern name for looping pattern.
9871 (doloop_end_nocc): New.
9872 (movsf): Add another constraint alternative.
9873 (zero_extendqihi2): Add constraint alternatives for not in place
9875 (zero_extendhisi2): Remove.
9876 (shift patterns): Add CC handling variants.
9880 (define_peephole2): New peephole to recognize mov that sets CC for
9883 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9885 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
9886 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
9889 2018-07-09 Jakub Jelinek <jakub@redhat.com>
9892 * real.c (real_nextafter): Return true if result is denormal.
9894 2018-07-09 Martin Liska <mliska@suse.cz>
9896 * common.opt: Add back wrongly removed attribute.
9898 2018-07-09 Richard Biener <rguenther@suse.de>
9901 * dwarf2out.c (gen_block_die): For an early generated DIE
9902 always output high/low PC attributes.
9904 2018-07-09 Tom de Vries <tdevries@suse.de>
9906 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
9907 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
9908 onto VAR_DECL with abstract origin.
9910 2018-07-07 Jim Wilson <jimw@sifive.com>
9912 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
9914 2018-07-07 Tom de Vries <tdevries@suse.de>
9916 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
9918 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
9920 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
9922 (combine_bound): Use wide-int overflow calculation instead of
9924 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
9926 * fold-const.c (int_const_binop_2): Same.
9927 (extract_muldiv_1): Same.
9928 (fold_div_compare): Same.
9929 (fold_abs_const): Same.
9931 * poly-int.h (add): Same.
9935 * predict.c (predict_iv_comparison): Same.
9936 * profile-count.c (slow_safe_scale_64bit): Same.
9937 * simplify-rtx.c (simplify_const_binary_operation): Same.
9938 * tree-chrec.c (tree_fold_binomial): Same.
9939 * tree-data-ref.c (split_constant_offset_1): Same.
9940 * tree-if-conv.c (idx_within_array_bound): Same.
9941 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
9942 * tree-ssa-phiopt.c (minmax_replacement): Same.
9943 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
9944 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
9945 * vr-values.c (vr_values::adjust_range_with_scev): Same.
9946 * wide-int.cc (wi::add_large): Same.
9947 (wi::mul_internal): Same.
9948 (wi::sub_large): Same.
9949 (wi::divmod_internal): Same.
9950 * wide-int.h: Change overflow type to overflow_type for neg, add,
9951 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
9952 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
9953 mul_internal, divmod_internal.
9954 (overflow_type): New enum.
9955 (accumulate_overflow): New.
9957 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
9959 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
9960 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
9962 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
9964 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
9965 argument is checked for zero before entering loop, avoid checking again.
9967 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
9969 * gimplify.h (generic_expr_could_trap_p): Set as global function.
9970 * gimplify.h (generic_expr_could_trap_p): Likwise.
9971 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
9973 2018-07-06 Jakub Jelinek <jakub@redhat.com>
9975 PR tree-optimization/86401
9976 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
9977 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
9978 (fold_bit_and_mask): ... here. New helper function for match.pd.
9979 * fold-const.h (fold_bit_and_mask): Declare.
9980 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
9982 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
9985 * target.def (translate_mode_attribute): New hook.
9986 * targhooks.h (default_translate_mode_attribute): Declare.
9987 * targhooks.c (default_translate_mode_attribute): New function.
9988 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
9989 * doc/tm.texi: Regenerate.
9990 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
9991 (rs6000_translate_mode_attribute): New function.
9993 2018-07-06 Paul Koning <ni1d@arrl.net>
9995 * doc/md.texi (define_split): Document DONE and FAIL.
9996 (define_peephole2): Ditto.
9998 2018-07-05 Jeff Law <law@redhat.com>
10000 PR tree-optimization/86010
10001 * tree-ssa-dse.c (compute_trims): More aggressively trim at
10002 both the head and tail of mem* and str* calls.
10004 2018-07-05 Jim Wilson <jimw@sifive.com>
10006 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
10008 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
10010 * config/aarch64/aarch64-simd.md: correct flags text for
10011 MIN_EXPR replacement.
10013 2018-07-05 James Clarke <jrtc27@jrtc27.com>
10015 * configure: Regenerated.
10017 2018-07-05 Carl Love <cel@us.ibm.com>
10019 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
10020 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
10021 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
10022 VSX_BUILTIN_DOUBLEL_V4SF.
10024 2018-07-05 Martin Sebor <msebor@redhat.com>
10027 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
10028 than its domain to compute its the upper bound of a char array.
10030 2018-07-05 Nathan Sidwell <nathan@acm.org>
10032 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
10033 * doc/cpp.texi: Update comment.
10034 * doc/tm.texi: Rebuilt.
10035 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
10036 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
10037 * doc/extend.texi (Backwards Compatibility): Clarify it is system
10038 headers affected by extern "C".
10039 * system.h: Poison NO_IMPLICIT_EXTERN_C.
10040 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
10041 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
10042 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
10043 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
10044 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
10045 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
10046 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
10047 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
10048 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
10049 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
10050 config/sparc/sp64-elf.h, config/spu/spu.h,
10051 config/stormy16/stormy16.h, config/v850/v850.h,
10052 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
10053 define NO_IMPLICIT_EXTERN_C.
10054 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
10056 2018-07-05 Tamar Christina <tamar.christina@arm.com>
10059 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
10060 instead of GET_MODE_SIZE when comparing Units.
10062 2018-07-05 Tamar Christina <tamar.christina@arm.com>
10065 * rtlanal.c (set_noop_p): Constrain on mode change,
10066 include hard-reg-set.h
10068 2018-07-05 Tamar Christina <tamar.christina@arm.com>
10070 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
10072 2018-07-05 Jakub Jelinek <jakub@redhat.com>
10075 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
10078 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
10080 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
10082 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
10085 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
10087 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
10089 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
10091 PR middle-end/86380
10092 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
10094 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
10096 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
10097 neg_*_op* variables.
10099 2018-07-04 Martin Liska <mliska@suse.cz>
10101 * tree-switch-conversion.c: Define
10102 max_ratio_for_speed and max_ratio_for_size constants.
10104 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
10105 Martin Liska <mliska@suse.cz>
10107 PR middle-end/66240
10110 * common.opt: Rename align options with 'str_' prefix.
10111 * common/config/i386/i386-common.c (set_malign_value): New
10113 (ix86_handle_option): Use it to set -falign-* options/
10114 * config/aarch64/aarch64-protos.h (struct tune_params): Change
10115 type from int to string.
10116 * config/aarch64/aarch64.c: Update default values from int
10118 * config/alpha/alpha.c (alpha_override_options_after_change):
10120 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
10121 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
10122 max skip conditionally.
10123 * config/i386/freebsd.h (SUBALIGN_LOG): New.
10124 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
10125 max skip conditionally.
10126 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
10127 max skip conditionally.
10128 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
10129 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
10130 max skip conditionally.
10131 * config/i386/i386.c (struct ptt): Change type from int to
10133 (ix86_default_align): Set default values.
10134 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
10135 max skip conditionally.
10136 * config/i386/iamcu.h (SUBALIGN_LOG): New.
10137 (ASM_OUTPUT_MAX_SKIP_ALIGN):
10138 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
10139 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
10140 max skip conditionally.
10141 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
10142 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
10143 * config/i386/x86-64.h (SUBALIGN_LOG): New.
10144 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
10145 max skip conditionally.
10146 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
10147 * config/ia64/ia64.c (ia64_option_override): Set default values
10148 for alignment options.
10149 * config/m68k/m68k.c: Handle new str_align_* options.
10150 * config/mips/mips.c (mips_set_compression_mode): Change
10152 (mips_option_override): Set default values for options.
10153 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
10155 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10157 * config/rx/rx.c (rx_option_override): Likewise.
10158 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
10159 (LABEL_ALIGN): Use align_labels_log.
10160 (LOOP_ALIGN): Use align_loops_align.
10161 * config/s390/s390.c (s390_asm_output_function_label): Use new
10163 * config/sh/sh.c (sh_override_options_after_change):
10164 Change type of constants.
10165 * config/spu/spu.c (spu_sched_init): Likewise.
10166 * config/sparc/sparc.c (sparc_option_override): Set default
10167 values for options.
10168 * config/visium/visium.c (visium_option_override): Likewise.
10169 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
10170 emit p2align format with last argument if it's not needed.
10171 * doc/invoke.texi: Document extended format of -falign-*.
10172 * final.c: Use align_labels alignment.
10173 * flags.h (struct target_flag_state): Change type to use
10175 (struct align_flags_tuple): New.
10176 (struct align_flags): Likewise.
10177 (align_loops_log): Redefine macro to use new types.
10178 (align_loops_max_skip): Redefine macro to use new types.
10179 (align_jumps_log): Redefine macro to use new types.
10180 (align_jumps_max_skip): Redefine macro to use new types.
10181 (align_labels_log): Redefine macro to use new types.
10182 (align_labels_max_skip): Redefine macro to use new types.
10183 (align_functions_log): Redefine macro to use new types.
10184 (align_loops): Redefine macro to use new types.
10185 (align_jumps): Redefine macro to use new types.
10186 (align_labels): Redefine macro to use new types.
10187 (align_functions): Redefine macro to use new types.
10188 (align_functions_max_skip): Redefine macro to use new types.
10189 (align_loops_value): New macro.
10190 (align_jumps_value): New macro.
10191 (align_labels_value): New macro.
10192 (align_functions_value): New macro.
10193 * function.c (invoke_set_current_function_hook): Propagate
10194 alignment values from flags to global variables default in
10196 * ipa-icf.c (sem_function::equals_wpa): Use
10197 cl_optimization_option_eq instead of memcmp.
10198 * lto-streamer.h (cl_optimization_stream_out): Support streaming
10200 (cl_optimization_stream_in): Likewise.
10201 * optc-save-gen.awk: Support strings in cl_optimization.
10202 * opth-gen.awk: Likewise.
10203 * opts.c (finish_options): Remove error checking of invalid
10205 (MAX_CODE_ALIGN): Remove.
10206 (MAX_CODE_ALIGN_VALUE): Likewise.
10207 (parse_and_check_align_values): New function.
10208 (check_alignment_argument): Likewise.
10209 (common_handle_option): Use check_alignment_argument.
10210 * opts.h (parse_and_check_align_values): Declare.
10211 * toplev.c (init_alignments): Remove.
10212 (read_log_maxskip): New.
10213 (parse_N_M): Likewise.
10214 (parse_alignment_opts): Likewise.
10215 (backend_init_target): Remove usage of init_alignments.
10216 * toplev.h (parse_alignment_opts): Declare.
10217 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
10219 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
10220 * tree.c (cl_option_hasher::equal): New.
10221 * varasm.c: Use new global macros.
10223 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
10225 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
10226 Use a simpler align directive also if MAXSKIP = ALIGN-1.
10227 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10228 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10229 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10230 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
10231 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
10232 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
10233 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
10234 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10235 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10236 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10237 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10239 2018-07-04 Martin Liska <mliska@suse.cz>
10240 Jonathan Wakely <jwakely@redhat.com>
10242 * coverage.c: Use correct type.
10243 * doc/invoke.texi: Language correction.
10245 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
10248 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
10249 ENDBRANCH for non-tail call which may return via indirect branch.
10250 * doc/extend.texi: Document indirect_return attribute.
10252 2018-07-03 Martin Sebor <msebor@redhat.com>
10254 PR tree-optimization/86274
10255 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
10257 (format_floating): Correct handling of infinities and NaNs.
10259 2018-07-03 Martin Sebor <msebor@redhat.com>
10261 * print-tree.c (print_real_cst): New function.
10262 (print_node_brief): Call it.
10263 (print_node): Ditto.
10265 2018-07-03 Jeff Law <law@redhat.com>
10267 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
10268 into a single pattern.
10270 * config/h8300/h8300.md (ors code_iterator): New.
10271 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
10272 a single pattern and single splitter.
10273 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
10274 (iorqi3_1, xorqi3_1): Likewise.
10275 (iorqi3, xorqi3 expanders): Similarly.
10277 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
10278 (movmd_internal) into a single pattern using the P mode iterator.
10279 (movmd splitters): Similarly.
10280 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
10281 (movsd splitters): Similarly.
10283 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
10284 ADDB, ADDW and ADDL into a single ADD attribute which selects the
10285 right table based on the size of the operand.
10286 * config/h8300/h8300.md (length_table): Corresponding changes. All
10287 references to "addb", "addw" and "addl" changed to "add".
10288 (btst patterns): Merge two variants into a single pattern.
10289 (tstqi, tsthi): Likewise.
10290 (addhi3_incdec, addsi3_incdec): Likewise.
10291 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
10292 (mulhi3, mulsi3): Likewise.
10293 (udivhi3, udivsi3): Likewise.
10294 (divhi3, divsi3): Likewise.
10295 (andorqi3, andorhi3, andorsi3): Likewise.
10297 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
10300 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
10301 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
10303 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10305 PR tree-optimization/85694
10306 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
10307 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
10309 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
10310 (<u>avg<mode>3_ceil): New patterns.
10312 2018-07-03 David Malcolm <dmalcolm@redhat.com>
10314 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
10315 scan-tree-dump directive.
10316 * gcc.dg/vect/slp-perm-2.c: Likewise.
10317 * gcc.dg/vect/slp-perm-3.c: Likewise.
10318 * gcc.dg/vect/slp-perm-5.c: Likewise.
10319 * gcc.dg/vect/slp-perm-6.c: Likewise.
10320 * gcc.dg/vect/slp-perm-7.c: Likewise.
10321 * gcc.dg/vect/slp-perm-8.c: Likewise.
10323 2018-07-03 Marek Polacek <polacek@redhat.com>
10325 PR middle-end/86202
10326 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
10328 2018-07-03 Richard Biener <rguenther@suse.de>
10331 * tree-ssa-structalias.c (find_func_clobbers): Properly
10332 handle indirect calls.
10334 2018-07-03 Jeff Law <law@redhat.com>
10336 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
10337 (shifts): New code iterator.
10338 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
10339 expander. Fix HImode handling on H8/SX.
10340 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
10341 (subqi3, subhi3, subsi3 expanders): Likewise.
10342 (andqi3, andhi3, andsi3 expanders): Likewise.
10343 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
10344 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
10345 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
10346 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
10347 (zero_extendqihi2, zero_extendqisi2): Likewise.
10348 (extendqihi2, extendqisi2): Likewise.
10349 (rotlqi3, rotlhi3, rotlsi3): Likewise.
10350 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
10351 (rotlqi3_1, rotlhi3_1): Likewise.
10352 (logicalhi3_sn, logicalsi3_sn): Likewise.
10353 (logicalhi3, logicalsi3): Likewise.
10355 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10357 * tree-vect-patterns.c (vect_recog_rotate_pattern)
10358 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
10359 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
10360 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
10361 type to append_pattern_def_seq instead of creating a stmt_vec_info
10363 (build_mask_conversion): Likewise. Remove vinfo argument.
10364 (vect_add_conversion_to_patterm): Likewise, renaming to...
10365 (vect_add_conversion_to_pattern): ...this.
10366 (vect_recog_mask_conversion_pattern): Update call to
10367 build_mask_conversion. Pass the vector type to
10368 append_pattern_def_seq here too.
10369 (vect_recog_gather_scatter_pattern): Update call to
10370 vect_add_conversion_to_pattern.
10372 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10374 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
10375 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
10376 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
10377 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
10378 STMT_VINFO_PATTERN_DEF_SEQ to null here.
10379 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
10380 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
10381 append_pattern_def_seq instead of new_pattern_def_seq.
10382 (vect_recog_divmod_pattern): Do both of the above.
10383 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
10386 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10388 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
10389 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
10390 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
10391 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
10392 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
10393 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
10394 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
10395 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
10396 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
10397 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
10398 parameter with a single stmt_vec_info.
10399 (vect_recog_func_ptr): Likewise.
10400 (vect_recog_gather_scatter_pattern): Likewise, folding in...
10401 (vect_try_gather_scatter_pattern): ...this.
10402 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
10403 the stmt_vec_info of the statement to be matched. Don't clear
10404 STMT_VINFO_RELATED_STMT.
10405 (vect_pattern_recog): Update call accordingly.
10407 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10409 PR tree-optimization/85694
10410 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
10411 (uavgM3_ceil): Document new optabs.
10412 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
10413 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
10415 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
10416 (savg_ceil_optab): New optabs.
10417 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
10418 (vect_vect_recog_func_ptrs): Add it.
10419 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
10420 constant directly from the associated lhs.
10422 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10424 * tree-vect-patterns.c (vect_split_statement): New function.
10425 (vect_convert_input): Use it to try to split an existing cast.
10427 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10429 * poly-int.h (print_hex): New function.
10430 * dumpfile.h (dump_dec, dump_hex): Declare.
10431 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
10432 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
10433 min_input_precision, operation_precision and operation_sign.
10434 * tree-vect-patterns.c (vect_get_range_info): New function.
10435 (vect_same_loop_or_bb_p, vect_single_imm_use)
10436 (vect_operation_fits_smaller_type): Delete.
10437 (vect_look_through_possible_promotion): Add an optional
10438 single_use_p parameter.
10439 (vect_recog_over_widening_pattern): Rewrite to use new
10440 stmt_vec_info infomration. Handle one operation at a time.
10441 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
10442 (vect_truncatable_operation_p, vect_set_operation_type)
10443 (vect_set_min_input_precision): New functions.
10444 (vect_determine_min_output_precision_1): Likewise.
10445 (vect_determine_min_output_precision): Likewise.
10446 (vect_determine_precisions_from_range): Likewise.
10447 (vect_determine_precisions_from_users): Likewise.
10448 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
10449 (vect_vect_recog_func_ptrs): Put over_widening first.
10451 (vect_pattern_recog): Call vect_determine_precisions.
10453 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10455 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
10456 statements that have been replaced by further pattern statements.
10457 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
10459 2018-07-03 Richard Biener <rguenther@suse.de>
10461 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
10462 always set *dt. Dump vectype in vectype overload.
10463 * dumpfile.h (dump_gimple_expr): New function.
10464 (dump_gimple_expr_loc): Likewise.
10465 * dumpfile.c (dump_gimple_expr): New function.
10466 (dump_gimple_expr_loc): Likewise.
10468 2018-07-02 Jeff Law <law@redhat.com>
10470 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
10471 the H8/300, H8/300H and H8/S variants into a single pattern.
10472 (movhi_h8300, movqi_h8300hs): Similarly.
10473 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
10474 (QHI mode iterator): New.
10476 * config/h8300/h8300.md: Remove trailing whitespace.
10478 2018-07-02 Jim Wilson <jimw@sifive.com>
10480 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
10481 instead of emit_insn for interrupt returns.
10482 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
10483 (riscv_sret, riscv_uret): Likewise.
10485 2018-07-02 David Malcolm <dmalcolm@redhat.com>
10487 * pretty-print.c (selftest::test_pp_format): Move save and restore
10488 of quotes to class auto_fix_quotes, and add an instance.
10489 * selftest.c: Include "intl.h".
10490 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
10491 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
10492 * selftest.h (selftest::auto_fix_quotes): New class.
10494 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
10496 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
10497 (aarch64_sve_prepare_conditional_op): Remove.
10498 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
10499 Allow aarch64_simd_reg_or_zero as select operand; remove
10500 the aarch64_sve_prepare_conditional_op call.
10501 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
10502 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
10503 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
10504 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
10505 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
10506 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
10507 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
10508 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
10509 and a splitters to match all of the *_any patterns.
10510 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
10512 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
10513 (SVE_COND_FP_BINARY_REV): Remove.
10514 (sve_int_op_rev, sve_fp_op_rev): New.
10515 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
10516 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
10517 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
10518 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
10519 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
10520 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
10521 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
10522 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
10523 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
10525 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
10526 Remove match_dup 1 from the inner unspec.
10527 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
10529 * config/aarch64/aarch64.md (movprfx): New attr.
10530 (length): Default movprfx to 8.
10531 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
10532 (*madd<SVE_I>, *msub<SVE_I): Likewise.
10533 (*<su>mul<SVE_I>3_highpart): Likewise.
10534 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
10535 (*v<ASHIFT><SVE_I>3): Likewise.
10536 (*<su><MAXMIN><SVE_I>3): Likewise.
10537 (*<su><MAXMIN><SVE_F>3): Likewise.
10538 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
10539 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
10540 (*div<SVE_F>4): Likewise.
10542 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
10544 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
10547 2018-07-02 Richard Biener <rguenther@suse.de>
10549 PR tree-optimization/86363
10550 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
10551 memset argument refers to a non-variable address.
10553 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
10555 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
10556 {PLUS,MINUS}_EXPR code to...
10557 (adjust_symbolic_bound): ...here,
10558 (combine_bound): ...here,
10559 (set_value_range_with_overflow): ...and here.
10561 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
10563 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
10565 (extract_range_from_abs_expr): ...here.
10567 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
10569 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
10570 -fno-omit-frame-pointer when not optimizing.
10572 2018-07-02 Martin Liska <mliska@suse.cz>
10575 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
10576 (propagate_nothrow): Likewise.
10578 2018-07-02 Martin Liska <mliska@suse.cz>
10581 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
10583 2018-07-02 David Malcolm <dmalcolm@redhat.com>
10585 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
10586 function in r262149, changing "loc" param from source_location to
10587 const dump_location_t &.
10588 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
10589 declaration, as above.
10591 2018-07-01 Paul Koning <ni1d@arrl.net>
10593 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
10594 -munit-asm, -mgnu-asm, -mdec-asm.
10595 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
10596 (pdp11_output_labelref): New.
10597 (pdp11_output_def): New.
10598 (pdp11_output_addr_vec_elt): New.
10599 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
10600 %# and %@ format codes.
10601 (pdp11_option_override): New.
10602 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
10603 (pdp11_output_ident): New.
10604 (pdp11_asm_named_section): New.
10605 (pdp11_asm_init_sections): New.
10606 (pdp11_file_start): New.
10607 (pdp11_file_end): New.
10608 (output_ascii): Use .ascii/.asciz for -mdec-asm.
10609 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
10610 %o, like %c but octal.
10611 (pdp11_option_override): New.
10612 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
10614 (DATA_SECTION_ASM_OP): Ditto.
10615 (READONLY_DATA_SECTION_ASM_OP): New.
10616 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
10617 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
10618 (ASM_OUTPUT_LABELREF): Ditto.
10619 (ASM_OUTPUT_DEF): Ditto.
10620 (ASM_OUTPUT_EXTERNAL): New.
10621 (ASM_OUTPUT_SOURCE_FILENAME): New.
10622 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
10623 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
10624 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
10625 %# and %@ format codes.
10626 * config/pdp11/pdp11.opt (mgnu-asm): New.
10627 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
10628 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
10629 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
10631 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
10633 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
10634 dereferencing path[] beyond its length.
10636 (debug_all_paths): New.
10637 (rewire_first_differing_edge): New.
10638 (adjust_paths_after_duplication): New.
10639 (duplicate_thread_path): Call adjust_paths_after_duplication.
10641 (thread_through_all_blocks): Add new argument to
10642 duplicate_thread_path.
10644 2018-06-30 Jim Wilson <jimw@sifive.com>
10646 * config/riscv/predicates.md (p2m1_shift_operand): New.
10647 (high_mask_shift_operand): New.
10648 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
10649 pattern using p2m1_shift_operand.
10650 (lshsi3_zero_extend_3+2): New combiner pattern using
10651 high_mask_shift_operand.
10653 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
10655 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
10657 (vect_recog_rotate_pattern): ...here.
10658 (vect_convert_input): Try to insert casts of invariants in the
10660 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
10661 preheader to be empty.
10663 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
10665 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
10666 vector type. If given, install it in the new statement's
10667 STMT_VINFO_VECTYPE.
10668 (vect_element_precision): New function.
10669 (vect_unpromoted_value): New struct.
10670 (vect_unpromoted_value::vect_unpromoted_value): New function.
10671 (vect_unpromoted_value::set_op): Likewise.
10672 (vect_look_through_possible_promotion): Likewise.
10673 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
10674 (vect_widened_op_tree, vect_convert_input): Likewise.
10675 (vect_convert_inputs, vect_convert_output): Likewise.
10676 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
10677 to handle the optional cast of the multiplication result and
10678 vect_widened_op_tree to detect the widened multiplication itself.
10679 Do not require the input and output of promotion casts to have
10680 the same sign, but base the signedness of the operation on the
10681 input rather than the result. If the pattern includes two
10682 promotions, check that those promotions have the same sign.
10683 Do not restrict the MULT_EXPR handling to a double-width result;
10684 handle quadruple-width results and wider. Use vect_convert_inputs
10685 to convert the inputs to the common type.
10686 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
10687 to handle the optional cast of the ABS result. Also allow a sign
10688 change or a sign extension between the ABS and MINUS.
10689 Use vect_widened_op_tree to detect the widened subtraction and use
10690 vect_convert_inputs to convert the inputs to the common type.
10691 (vect_handle_widen_op_by_const): Delete.
10692 (vect_recog_widen_op_pattern): New function.
10693 (vect_recog_widen_mult_pattern): Use it.
10694 (vect_recog_widen_shift_pattern): Likewise.
10695 (vect_recog_widen_sum_pattern): Use
10696 vect_look_through_possible_promotion to handle the promoted
10699 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
10701 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
10702 the containing gimple_seq *.
10703 * gimple-iterator.h (gsi_for_stmt): Declare it.
10704 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
10705 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
10706 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
10707 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
10708 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
10710 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
10712 (vect_mark_pattern_stmts): ...here. Handle cases in which the
10713 statement being replaced is part of an existing pattern
10714 definition sequence, inserting the new pattern statements before
10716 (vect_pattern_recog_1): Don't return a bool. If the statement
10717 is already part of a pattern, instead apply pattern matching
10718 to the pattern definition statements. Don't clear the
10719 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
10720 (vect_pattern_recog): Don't break after the first match;
10721 continue processing the pattern definition statements instead.
10722 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
10724 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
10726 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
10727 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
10728 (vect_recog_widen_sum_pattern): Use it.
10730 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
10732 * tree-vect-loop.c (vectorizable_reduction): Assert that the
10733 phi is not a pattern statement and has not been replaced by
10734 a pattern statement.
10735 * tree-vect-patterns.c (type_conversion_p): Don't check
10736 STMT_VINFO_IN_PATTERN_P.
10737 (vect_recog_vector_vector_shift_pattern): Likewise.
10738 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
10739 the pattern statement rather than the original statement; check
10740 directly for a WIDEN_MULT_EXPR here.
10741 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
10742 vect_is_simple_use to return the pattern statement rather
10743 than the original statement; use is_pattern_stmt_p to check
10744 for such a pattern statement.
10745 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
10746 to return the pattern statement rather than the original statement;
10747 don't do the same transformation here.
10748 (vect_is_simple_use): If the defining statement has been replaced
10749 by a pattern statement, return the pattern statement instead.
10750 Remove the corresponding (local) transformation from the vectype
10753 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
10755 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
10756 end and default to null.
10757 * tree-vect-loop.c (vect_create_epilog_for_reduction)
10758 (vectorizable_reduction): Update calls accordingly, dropping the
10759 gimple ** argument if the passed-back statement isn't needed.
10760 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
10761 (vect_recog_rotate_pattern): Likewise.
10762 (vect_recog_mask_conversion_pattern): Likewise.
10763 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
10764 (vect_mask_constant_operand_p): Likewise.
10765 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
10766 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
10767 (get_group_load_store_type, get_load_store_type): Likewise.
10768 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
10769 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
10770 (vectorizable_conversion, vectorizable_assignment): Likewise.
10771 (vectorizable_shift, vectorizable_operation): Likewise.
10772 (vectorizable_store, vect_is_simple_cond): Likewise.
10773 (vectorizable_condition, vectorizable_comparison): Likewise.
10774 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
10775 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
10776 and move it to the end. Cope with null def_stmt_outs.
10778 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
10780 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
10782 2018-06-29 Jeff Law <law@redhat.com>
10784 * config/v850/v850.c (v850_legitimate_address_p): Handle large
10785 displacements for TARGET_V850E2V3 and newer.
10786 (TARGET_LRA_P): Remove. Defaults to LRA now.
10787 * config/v850/v850.md (sign23byte_load): Remove.
10788 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
10789 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
10791 2018-06-29 Martin Liska <mliska@suse.cz>
10794 * coverage.c (coverage_init): Mangle full path name.
10795 * doc/invoke.texi: Document the change.
10796 * gcov-io.c (mangle_path): New.
10797 * gcov-io.h (mangle_path): Likewise.
10798 * gcov.c (mangle_name): Use mangle_path for path mangling.
10800 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10802 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
10803 if starting source register is not even.
10805 2018-06-29 Martin Liska <mliska@suse.cz>
10807 PR tree-optimization/86263
10808 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
10809 Make edge redirection.
10811 2018-06-29 David Malcolm <dmalcolm@redhat.com>
10813 * dumpfile.c (dump_loc): Add indentation based on scope depth.
10814 (dump_scope_depth): New variable.
10815 (get_dump_scope_depth): New function.
10816 (dump_begin_scope): New function.
10817 (dump_end_scope): New function.
10818 * dumpfile.h (get_dump_scope_depth): New declaration.
10819 (dump_begin_scope): New declaration.
10820 (dump_end_scope): New declaration.
10821 (class auto_dump_scope): New class.
10822 (AUTO_DUMP_SCOPE): New macro.
10823 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
10826 2018-06-29 Richard Biener <rguenther@suse.de>
10828 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
10829 compute_all_dependences succeeds.
10830 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
10831 exceed --param loop-max-datarefs-for-datadeps.
10833 2018-06-29 Jakub Jelinek <jakub@redhat.com>
10835 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
10837 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
10840 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
10841 alternative 0 in preferred_for_speed attribute.
10843 2018-06-28 Paul Koning <ni1d@arrl.net>
10845 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
10846 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
10847 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
10848 * config/pdp11/pdp11.md: Correct "length" attribute calculation
10849 for shift insn patterns.
10851 2018-06-28 David Malcolm <dmalcolm@redhat.com>
10853 * cgraph.c (cgraph_node::get_body): Replace assignments to
10854 "dump_file" with calls to set_dump_file.
10855 * dumpfile.c (alt_dump_file): Make static, and group with...
10856 (alt_flags): ...this definition.
10857 (dumps_are_enabled): New variable.
10858 (refresh_dumps_are_enabled): New function.
10859 (set_dump_file): New function.
10860 (set_alt_dump_file): New function.
10861 (gcc::dump_manager::dump_start): Replace assignments to
10862 "dump_file" and "alt_dump_file" with calls to set_dump_file and
10864 (gcc::dump_manager::dump_finish): Likewise.
10865 * dumpfile.h (alt_dump_file): Delete decl.
10866 (dumps_are_enabled): New variable decl.
10867 (set_dump_file): New function decl.
10868 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
10870 * tree-nested.c (lower_nested_functions): Replace assignments to
10871 "dump_file" with calls to set_dump_file.
10873 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
10875 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
10876 goto_locus of each outgoing edge of each basic block.
10878 2018-06-28 Richard Biener <rguenther@suse.de>
10880 * dwarf2out.c (decl_scope_table): Remove.
10881 (push_decl_scope): Likewise.
10882 (pop_decl_scope): Likewise.
10883 (gen_type_die_for_member): Do not call push/pop_decl_scope.
10884 (gen_struct_or_union_type_die): Likewise.
10885 (gen_tagged_type_die): Likewise.
10886 (dwarf2out_init): Do not initialize decl_scope_table.
10887 (dwarf2out_c_finalize): Do not free it.
10889 2018-06-28 Richard Biener <rguenther@suse.de>
10891 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
10892 deciding whether to not re-use a DIE.
10894 2018-06-28 Richard Biener <rguenther@suse.de>
10896 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
10897 DW_AT_abstract_origin attribute.
10899 2018-06-28 Martin Liska <mliska@suse.cz>
10901 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
10902 Use newly introduced constants.
10903 * tree-switch-conversion.h (struct jump_table_cluster):
10904 Define max_ratio_for_size and max_ratio_for_speed.
10906 2018-06-28 Martin Liska <mliska@suse.cz>
10908 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
10909 Add new checking assert to catch invalid state.
10910 (jump_table_cluster::can_be_handled): Handle single case
10912 (jump_table_cluster::is_beneficial): Bail out for such case.
10913 (bit_test_cluster::find_bit_tests):
10914 Add new checking assert to catch invalid state.
10915 (bit_test_cluster::can_be_handled): Handle single case
10917 (bit_test_cluster::is_beneficial): Bail out for such case.
10918 (switch_decision_tree::analyze_switch_statement):
10921 2018-06-28 Martin Liska <mliska@suse.cz>
10923 * common.opt: Introduce -completion option.
10924 * gcc.c (driver_handle_option): Handle it.
10925 (driver::main): Print completions if completion
10927 * opt-suggestions.c (option_proposer::get_completions):
10929 (option_proposer::suggest_completion): Likewise.
10930 (option_proposer::find_param_completions): Likewise.
10931 (verify_autocompletions): Likewise.
10932 (test_completion_valid_options): Likewise.
10933 (test_completion_valid_params): Likewise.
10934 (in_completion_p): Likewise.
10935 (empty_completion_p): Likewise.
10936 (test_completion_partial_match): Likewise.
10937 (test_completion_garbage): Likewise.
10938 (opt_proposer_c_tests): Likewise.
10939 * opt-suggestions.h: Declare new functions.
10940 * opts.c (common_handle_option): Handle OPT__completion_.
10941 * selftest-run-tests.c (selftest::run_tests): Add
10942 opt_proposer_c_tests.
10943 * selftest.c (assert_str_startswith): New.
10944 * selftest.h (assert_str_startswith): Likewise.
10945 (opt_proposer_c_tests): New.
10946 (ASSERT_STR_STARTSWITH): Likewise.
10948 2018-06-28 Martin Liska <mliska@suse.cz>
10950 * Makefile.in: Add opt-suggestions.o.
10951 * gcc-main.c: Include opt-suggestions.h.
10952 * gcc.c (driver::driver): Likewise.
10953 (driver::~driver): Remove m_option_suggestions.
10954 (driver::build_option_suggestions): Moved to option_proposer.
10955 (driver::suggest_option): Likewise.
10956 (driver::handle_unrecognized_options): Use option_proposer.
10957 * gcc.h (class driver): Add new memver m_option_proposer.
10958 * opt-suggestions.c: New file.
10959 * opt-suggestions.h: New file.
10961 2018-06-28 Martin Liska <mliska@suse.cz>
10963 * vec.h (class auto_string_vec): New (moved from auto_argvec).
10964 (auto_string_vec::~auto_string_vec): Likewise.
10966 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
10968 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
10969 prevent_decl_creation_for_types fields up and add reset_location field.
10970 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
10971 statement if id->reset_location is true.
10972 (copy_edges_for_bb): Do not set goto_locus on the new edges if
10973 id->reset_location is true.
10974 (copy_phis_for_bb): Force input_location on the arguments if
10975 id->reset_location is true.
10976 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
10977 is set on the function to be inlined.
10979 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
10981 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
10983 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
10985 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
10986 registers for Pmode.
10987 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
10988 hard registers for the clobbered pseudo.
10990 2018-06-27 Paul Koning <ni1d@arrl.net>
10992 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
10993 mutually exclusive options.
10994 * config/pdp11/constraints.md (h): New constraint.
10995 (O): Update definition to match shift code generation.
10996 (D): New constraint.
10997 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
10999 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
11001 (output_jump): Change arguments.
11002 (pdp11_fixed_cc_regs): New function.
11003 (pdp11_cc_mode): Ditto.
11004 (pdp11_expand_shift): Ditto.
11005 (pdp11_assemble_shift): Ditto.
11006 (pdp11_small_shift): Ditto.
11007 (pdp11_branch_cost): Remove.
11008 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
11010 (pdp11_register_move_cost): Update for CC registers.
11011 (pdp11_rtx_costs): Add case for LSHIFTRT.
11012 (pdp11_output_jump): Add CCNZ mode conditional branches.
11013 (notice_update_cc_on_set): Remove.
11014 (pdp11_cc_mode): New function.
11015 (simple_memory_operand): Correct pre/post decrement case.
11016 (no_side_effect_operand): New function.
11017 (pdp11_regno_reg_class): Add CC_REGS class.
11018 (pdp11_fixed_cc_regs): New function.
11019 (pdp11_small_shift): New function.
11020 (pdp11_expand_shift): New function to expand shift insns.
11021 (pdp11_assemble_shift): New function to output shifts.
11022 (pdp11_branch_cost): Remove.
11023 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
11024 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
11025 (WCHAR_TYPE): Ditto.
11026 (PTRDIFF_TYPE): Ditto.
11027 (ADJUST_INSN_LENGTH): New macro.
11028 (FIXED_REGISTERS): Add CC registers.
11029 (CALL_USED_REGISTERS): Ditto.
11030 (reg_class): Ditto.
11031 (REG_CLASS_NAMES): Ditto.
11032 (REG_CLASS_CONTENTS): Ditto.
11033 (SELECT_CC_MODE): Use new function.
11034 (TARGET_FLAGS_REGNUM): New macro.
11035 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
11036 (cc0_reg_rtx): Remove.
11037 (CC_STATUS_MDEP): Remove.
11038 (CC_STATUS_MDEFP_INIT): Remove.
11039 (CC_IN_FPU): Remove.
11040 (NOTICE_UPDATE_CC): Remove.
11041 (REGISTER_NAMES): Add CC registers.
11042 (BRANCH_COST): Change to constant 1.
11043 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
11045 * config/pdp11/pdp11.opt (mbcopy): Remove.
11046 (mbcopy-builtin): Remove.
11047 (mbranch-cheap): Remove.
11048 (mbranch-expensive): Remove.
11049 * config/pdp11/predicates.md (expand_shift_operand): Update to
11050 match shift code generation.
11051 (ccnz_operator): New predicate.
11052 * doc/invoke.texi (PDP-11 Options): Remove deleted options
11053 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
11054 Remove non-existent option -mabshi, -mno-abshi. Document mutually
11056 * doc/md.texi (PDP-11): Document new D and h constraints. Update
11057 description of O constraint.
11059 2018-06-27 Jeff Law <law@redhat.com>
11060 Austin Law <austinklaw@gmail.com>
11062 * config/v850/v850.md (addsi3_set_flags): New pattern.
11063 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
11064 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
11065 (zero_extendhisi2_v850_set_flags): Likewise.
11066 (zero_extendqisi2_v850_set_flags): Likewise.
11067 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
11068 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
11069 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
11071 * config/v850/v850-protos.h (notice_update_cc): Remove.
11072 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
11073 (v850_print_operand): Handle 'D' and "d".
11074 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
11075 Add handling of arithmetic/logical operations compared against zero.
11076 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
11077 Do not look at v850_compare_op, instead get mode from last argument.
11078 (v850_gen_compare): Remove
11079 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
11080 after reload for prologue insns.
11081 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
11083 (construct_save_jarl): Likewise.
11084 (TARGET_FLAGS_REGNUM): Define.
11085 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
11086 (NOTICE_UPDATE_CC): Remove.
11087 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
11088 than cc0. Conditionalize on reload_completed.
11089 (cmpsi_insn, setfcc_insn): Likewise.
11090 (tst1 splitter): Turn into define_and_split which sets the flags
11092 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
11093 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
11094 (cstoresf4, cstoredf4): Clobber the flags.
11095 (cmpsi, cmpsf, cmpdf): Remove expanders.
11096 (setf_insn): Remove pattern.
11097 (addsi3): Turn into define_and_split which clobbers the flags after
11098 reload and a suitable pattern (addsi3_clobber_flags) for use after
11100 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
11101 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
11102 (ashrsi3, ashrsi3_v850e2): Likewise.
11103 (bins): Clobber the flags.
11104 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
11105 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
11106 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
11107 (call_value_internal_short, call_value_internal_long): Likewise.
11108 (callt_save_interrupt, callt_return_interrupt): Likewise.
11109 (save_interrupt, return_interrupt): Likewise.
11110 (callt_save_all_interrupt, save_all_interrupt): Likewise.
11111 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
11112 (restore_all_interrupt, _restore_all_interrupt): Likewise.
11113 (All FP comparisons): Only allow after reload has completed.
11115 (divh, divhu): Tweak output template.
11116 (branch_z_normal, branch_z_invert): Remove
11117 (branch_nz_normal, branch_nz_invert): Likewise.
11118 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
11120 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
11121 * config/v850/v850.c (notice_update_cc): Remove.
11122 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
11123 (CC_NO_CARRY): Likewise.
11124 (NOTICE_UPDATE_CC): Define to nothing.
11125 * config/v850/v850.md: Remove block comment on cc0 handling
11126 Remove "cc" attribute from all patterns. Remove cc_status handling
11127 from all patterns. Minor formatting fixes.
11129 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11131 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
11132 (cortex-a76.cortex-a55): Likewise.
11133 * config/aarch64/aarch64-tune.md: Regenerate.
11134 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
11135 cortex-a76.cortex-a55.
11137 2018-06-27 Jeff Law <law@redhat.com>
11139 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
11140 (MULTILIB_DIRNAMES): Similarly.
11142 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
11144 * gimple.h (gimple_return_retbnd): Delete.
11145 (gimple_return_set_retbnd): Likewise.
11146 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
11147 gimple_return_set_retbnd.
11148 * gimple-pretty-print.c (dump_gimple_return): Remove call to
11149 gimple_return_retbnd and adjust.
11150 * tree-inline.h (struct copy_body_data): Remove retbnd field.
11151 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
11152 Explicitly return NULL in a couple more cases. Move assertion
11153 on debug statements and remove unreachable code.
11154 (reset_debug_binding): Do not test id->retbnd.
11155 (expand_call_inline): Do not set it.
11157 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
11159 * configure.ac: Add --disable-gcov option.
11160 * configure: Regenerate.
11161 * Makefile.in: Honour @enable_gcov@.
11162 * doc/install.texi: Document --disable-gcov.
11164 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11166 * config/arm/arm-cpus.in (cortex-a76): New entry.
11167 (cortex-a76.cortex-a55): Likewise.
11168 * config/arm/arm-tables.opt: Regenerate.
11169 * config/arm/arm-tune.md: Likewise.
11170 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
11171 * doc/invoke.texi (ARM Options): Document cortex-a76 and
11172 cortex-a76.cortex-a55.
11174 2018-06-27 Tamar Christina <tamar.christina@arm.com>
11177 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
11179 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
11181 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
11183 (EPILOGUE_USES): Likewise.
11185 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
11187 * tree-inline.c (remap_location): New function extracted from...
11188 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
11189 (copy_phis_for_bb): ...here. Call remap_location.
11190 (copy_cfg_body): Adjust call to copy_edges_for_bb.
11192 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
11194 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
11195 unaligned vsx for 16B memset.
11197 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
11200 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
11201 ieee128_float_type_node to long_double_type_node unless
11202 TARGET_LONG_DOUBLE_128 is set.
11204 2018-06-26 David Malcolm <dmalcolm@redhat.com>
11206 * cfgloop.c (get_loop_location): Convert return type from
11207 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
11208 by implicit construction from rtx_insn *, and using
11209 dump_user_location_t::from_function_decl for the fallback case.
11210 * cfgloop.h (get_loop_location): Convert return type from
11211 location_t to dump_user_location_t.
11212 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
11213 dump_printf_loc to pass in a dump_location_t rather than a
11214 location_t, via the gimple stmt.
11215 * coverage.c (get_coverage_counts): Update calls to
11216 dump_printf_loc to pass in dump_location_t rather than a
11218 * doc/optinfo.texi (Dump types): Convert example of
11219 dump_printf_loc from taking "locus" to taking "insn". Update
11220 description of the "_loc" calls to cover dump_location_t.
11221 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
11223 (dump_user_location_t::dump_user_location_t): New constructors,
11224 from gimple *stmt and rtx_insn *.
11225 (dump_user_location_t::from_function_decl): New function.
11226 (dump_loc): Make static.
11227 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
11228 const dump_location_t &.
11229 (dump_generic_expr_loc): Delete.
11230 (dump_printf_loc): Convert param "loc" from location_t to
11231 const dump_location_t &.
11232 (selftest::test_impl_location): New function.
11233 (selftest::dumpfile_c_tests): New function.
11234 * dumpfile.h: Include "profile-count.h".
11235 (class dump_user_location_t): New class.
11236 (struct dump_impl_location_t): New struct.
11237 (class dump_location_t): New class.
11238 (dump_printf_loc): Convert 2nd param from source_location to
11239 const dump_location_t &.
11240 (dump_generic_expr_loc): Delete.
11241 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
11242 const dump_location_t &.
11243 * gimple-fold.c (fold_gimple_assign): Update call to
11244 dump_printf_loc to pass in a dump_location_t rather than a
11245 location_t, via the gimple stmt.
11246 (gimple_fold_call): Likewise.
11247 * gimple-loop-interchange.cc
11248 (loop_cand::analyze_iloop_reduction_var): Update for change to
11249 check_reduction_path.
11250 (tree_loop_interchange::interchange): Update for change to
11251 find_loop_location.
11252 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
11253 change in return-type of find_loop_location.
11254 (graphite_regenerate_ast_isl): Likewise.
11255 * graphite-optimize-isl.c (optimize_isl): Likewise.
11256 * graphite.c (graphite_transform_loops): Likewise.
11257 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
11258 pass in a dump_location_t rather than a location_t, via the
11260 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
11261 * ipa.c (walk_polymorphic_call_targets): Likewise.
11262 * loop-unroll.c (report_unroll): Convert "locus" param from
11263 location_t to dump_location_t.
11264 (decide_unrolling): Update for change to get_loop_location's
11266 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
11267 location_t to dump_user_location_t.
11268 (grid_find_single_omp_among_assignments_1): Updates calls to
11269 dump_printf_loc to pass in a dump_location_t rather than a
11270 location_t, via the gimple stmt.
11271 (grid_parallel_clauses_gridifiable): Convert "tloc" from
11272 location_t to dump_location_t. Updates calls to dump_printf_loc
11273 to pass in a dump_location_t rather than a location_t, via the
11275 (grid_inner_loop_gridifiable_p): Likewise.
11276 (grid_dist_follows_simple_pattern): Likewise.
11277 (grid_gfor_follows_tiling_pattern): Likewise.
11278 (grid_target_follows_gridifiable_pattern): Likewise.
11279 (grid_attempt_target_gridification): Convert initialization
11280 of local "grid" from memset to zero-initialization; FIXME: does
11281 this require C++11? Update call to dump_printf_loc to pass in a
11282 optinfo_location rather than a location_t, via the gimple stmt.
11283 * profile.c (read_profile_edge_counts): Updates call to
11284 dump_printf_loc to pass in a dump_location_t rather than a
11286 (compute_branch_probabilities): Likewise.
11287 * selftest-run-tests.c (selftest::run_tests): Call
11289 * selftest.h (dumpfile_c_tests): New decl.
11290 * tree-loop-distribution.c (pass_loop_distribution::execute):
11291 Update for change in return type of find_loop_location.
11292 * tree-parloops.c (parallelize_loops): Likewise.
11293 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
11294 "locus" from location_t to dump_user_location_t.
11295 (canonicalize_loop_induction_variables): Likewise.
11296 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
11297 for change in return type of find_loop_location.
11298 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
11299 to dump_printf_loc to pass in a dump_location_t rather than a
11300 location_t, via the stmt.
11301 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
11303 * tree-vect-loop-manip.c (find_loop_location): Convert return
11304 type from source_location to dump_user_location_t.
11305 (vect_do_peeling): Update for above change.
11306 (vect_loop_versioning): Update for change in type of
11308 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
11309 from location_t to dump_user_location_t.
11310 (vect_estimate_min_profitable_iters): Update for change in type
11312 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
11313 location_t to dump_location_t.
11314 (vect_slp_bb): Update for change in type of vect_location.
11315 * tree-vectorizer.c (vect_location): Convert from source_location
11316 to dump_user_location_t.
11317 (try_vectorize_loop_1): Update for change in vect_location's type.
11318 (vectorize_loops): Likewise.
11319 (increase_alignment): Likewise.
11320 * tree-vectorizer.h (vect_location): Convert from source_location
11321 to dump_user_location_t.
11322 (find_loop_location): Convert return type from source_location to
11323 dump_user_location_t.
11324 (check_reduction_path): Convert 1st param from location_t to
11325 dump_user_location_t.
11326 * value-prof.c (check_counter): Update call to dump_printf_loc to
11327 pass in a dump_user_location_t rather than a location_t; update
11328 call to error_at for change in type of "locus".
11329 (check_ic_target): Update call to dump_printf_loc to
11330 pass in a dump_user_location_t rather than a location_t, via the
11333 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
11335 * config/s390/s390.h (enum processor_flags): Do not use
11336 default tune parameter when -march was specified.
11338 2018-06-26 Jakub Jelinek <jakub@redhat.com>
11341 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
11342 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
11345 2018-06-26 Richard Biener <rguenther@suse.de>
11347 PR tree-optimization/86287
11349 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
11350 (vect_analyze_loop): Initialize n_stmts.
11352 2018-06-26 Richard Biener <rguenther@suse.de>
11354 PR middle-end/86271
11355 * fold-const.c (fold_convertible_p): Pointer extension
11358 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
11361 * dwarf2out.c (loc_list_has_views): Adjust comments.
11362 (dw_loc_list): Split single cross-partition range with
11365 2018-06-25 Jeff Law <law@redhat.com>
11367 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
11368 on -mbig-switch by default.
11370 * config/v850/predicates.md (const_float_1_operand): Fix match_code
11372 (const_float_0_operand): Remove unused predicate.
11373 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
11374 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
11375 (recipsf2): New expander. Original pattern now called
11377 (recipdf2, recipdf2_insn): Similarly.
11378 (rsqrtsf2, rsqrtsf2_insn): Similarly
11379 (rsqrtdf2, rsqrtdf2_insn): Similarly
11381 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
11383 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
11384 Simplify logic for FreeBSD (twice).
11386 2018-06-25 Martin Sebor <msebor@redhat.com>
11388 PR tree-optimization/86204
11389 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
11390 a strnlen result if it's less than the length of the string.
11392 2018-06-25 Martin Sebor <msebor@redhat.com>
11394 PR tree-optimization/85700
11395 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
11396 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
11397 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
11399 2018-06-25 Martin Sebor <msebor@redhat.com>
11401 * doc/extend.texi (Zero-length arrays): Update and clarify.
11403 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
11405 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
11406 added IEEE/IBM long double multilib support on PowerPC little
11407 endian Linux systems.
11408 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
11409 (MULTILIB_DEFAULTS): Likewise.
11410 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11412 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
11413 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
11414 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
11416 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
11418 PR middle-end/86311
11419 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
11420 (REORDER_45): Likewise.
11422 2018-06-25 Jeff Law <law@redhat.com>
11424 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
11425 dividend to 32 bits. Adjust length.
11426 (udivmodhi4): Cleanup output template. Fix length.
11428 2018-06-25 Carl Love <cel@us.ibm.com>
11430 * config/rs6000/vsx.md: Change word selector to prefered location.
11432 2018-06-25 Richard Biener <rguenther@suse.de>
11434 PR tree-optimization/86304
11435 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
11436 epilogue-if-converted loops as well.
11438 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
11440 * lto-section-out.c (lto_begin_section): Do not print section
11441 name for noaddr and unnumbered dumps.
11443 2018-06-25 Richard Biener <rguenther@suse.de>
11445 * tree-vectorizer.h (struct vec_info_shared): New structure
11446 with parts split out from struct vec_info and loop_nest from
11447 struct _loop_vec_info.
11448 (struct vec_info): Adjust accordingly.
11449 (struct _loop_vec_info): Likewise.
11450 (LOOP_VINFO_LOOP_NEST): Adjust.
11451 (LOOP_VINFO_DATAREFS): Likewise.
11452 (LOOP_VINFO_DDRS): Likewise.
11453 (struct _bb_vec_info): Likewise.
11454 (BB_VINFO_DATAREFS): Likewise.
11455 (BB_VINFO_DDRS): Likewise.
11456 (struct _stmt_vec_info): Add dr_aux member.
11457 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
11458 (DR_MISALIGNMENT_UNINITIALIZED): New.
11459 (set_dr_misalignment): Adjust.
11460 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
11461 (vect_analyze_loop): Adjust prototype.
11462 (vect_analyze_loop_form): Likewise.
11463 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
11464 Compute dependences lazily.
11465 (vect_record_base_alignments): Use shared datarefs/ddrs.
11466 (vect_verify_datarefs_alignment): Likewise.
11467 (vect_analyze_data_refs_alignment): Likewise.
11468 (vect_analyze_data_ref_accesses): Likewise.
11469 (vect_analyze_data_refs): Likewise.
11470 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
11471 constructor parameter for shared part.
11472 (vect_analyze_loop_form): Pass in shared part and adjust.
11473 (vect_analyze_loop_2): Pass in storage for the number of
11474 stmts. Move loop nest finding to the caller. Compute
11476 (vect_analyze_loop): Pass in shared part.
11477 (vect_transform_loop): Verify shared datarefs are unchanged.
11478 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
11479 constructor parameter for shared part.
11480 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
11481 (vect_slp_bb): Verify shared datarefs are unchanged before
11483 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
11485 (new_stmt_vec_info): Initialize DR_AUX misalignment to
11486 DR_MISALIGNMENT_UNINITIALIZED.
11487 * tree-vectorizer.c (vec_info::vec_info): Add constructor
11488 parameter for shared part.
11489 (vec_info::~vec_info): Adjust.
11490 (vec_info_shared::vec_info_shared): New.
11491 (vec_info_shared::~vec_info_shared): Likewise.
11492 (vec_info_shared::save_datarefs): Likewise.
11493 (vec_info_shared::check_datarefs): Likewise.
11494 (try_vectorize_loop_1): Construct shared part live for analyses
11495 of a single loop for multiple vector sizes.
11496 * tree-parloops.c (gather_scalar_reductions): Adjust.
11498 2018-06-25 Richard Biener <rguenther@suse.de>
11500 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
11501 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
11502 (vect_analyze_data_refs): Remove similar code from here and
11503 simplify accordingly.
11505 2018-06-25 Richard Biener <rguenther@suse.de>
11507 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
11508 for reverse storage order accesses rather than asserting
11509 they cannot happen here.
11511 2018-06-25 Tom de Vries <tdevries@suse.de>
11514 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
11515 Use data16 instead of .byte for insn prefix.
11517 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
11520 * parser.c (make_char_string_pack): Pass this literal chars
11521 through cpp_interpret_string.
11522 (cp_parser_userdef_numeric_literal): Check the result of
11523 make_char_string_pack.
11525 2018-06-24 Maya Rashish <coypu@sdf.org>
11527 * ginclude/stddef.h: Simplify conditions around avoiding
11528 re-definition of __size_t.
11530 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
11532 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
11533 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
11535 2018-06-22 Maya Rashish <coypu@sdf.org>
11537 * doc/invoke.texi (mno-fancy-math-387): Update for changes
11538 made to OpenBSD and NetBSD through the years.
11540 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
11542 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
11543 behavior of vec_pack (vector double, vector double) to match
11544 behavior of vec_float2 (vector double, vector double).
11546 2018-06-22 Olivier Hainque <hainque@adacore.com>
11548 * gimplify.c (gimplify_function_tree): Prevent creation
11549 of a trampoline for the address of the current function
11550 passed to entry/exit instrumentation hooks.
11552 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
11555 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
11558 2018-06-22 Martin Liska <mliska@suse.cz>
11560 PR tree-optimization/86263
11561 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
11562 Bail out if is_enabled is false.
11563 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
11565 (jump_table_cluster::is_enabled): New function.
11567 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
11569 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
11570 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
11571 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
11572 (lto_input_ts_binfo_tree_pointers): Likewise.
11573 * tree-streamer-out.c (streamer_write_tree_bitfields,
11574 write_ts_binfo_tree_pointers): Likewise.
11575 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
11577 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
11579 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
11581 2018-06-22 Martin Liska <mliska@suse.cz>
11583 * symbol-summary.h (get): Make it pure and inline move
11584 functionality from ::get function.
11585 (get): Remove and inline into ::get and ::get_create.
11586 (get_create): Move code from ::get function.
11588 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11591 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
11592 -x assembler-with-cpp.
11594 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11596 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
11597 _FILE_OFFSET_BITS=64 for C++.
11599 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
11601 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
11602 conversion insn that shows up when pr85657-3.c is compiled using
11603 IEEE 128-bit long double.
11604 (neg<mode>2_internal): Use the correct mode to check whether the
11605 mode is IBM extended.
11606 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
11607 multiply and divide external functions from being created more
11610 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
11612 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
11614 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
11615 the edge can be forwarded.
11616 (cfg_layout_merge_blocks): Likewise.
11618 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
11620 * except.c (finish_eh_generation): Commit edge insertions only after
11621 the EH edges have been redirected from post-landing to landing pads.
11623 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
11625 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
11626 create_tmp_var_for to create the FRAME decl.
11627 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
11629 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
11631 * tree-inline.c (copy_edges_for_bb): Minor tweak.
11632 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
11633 debug statement when resetting its value.
11634 (expand_call_inline): Copy the locus of the call onto the assignment
11635 of the return value, if any. Use local variable in more cases.
11637 2018-06-21 Martin Liska <mliska@suse.cz>
11639 * ipa-pure-const.c (propagate_nothrow): Use
11640 funct_state_summaries->get.
11641 (dump_malloc_lattice): Likewise.
11642 (propagate_malloc): Likewise.
11644 2018-06-21 Richard Biener <rguenther@suse.de>
11646 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
11647 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
11648 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
11650 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
11651 BLOCK_ABSTRACT_ORIGIN unconditionally.
11653 2018-06-21 David Malcolm <dmalcolm@redhat.com>
11655 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
11657 * ipa-reference.c (ipa_reference_c_finalize): Delete
11658 ipa_ref_opt_sum_summaries and set it to NULL.
11660 2018-06-21 Tom de Vries <tdevries@suse.de>
11662 PR tree-optimization/85859
11663 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
11664 test with comment from bb_no_side_effects_p.
11666 2018-06-21 Richard Biener <rguenther@suse.de>
11668 PR tree-optimization/86232
11669 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
11670 max for constant niter.
11672 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
11674 * config/aarch64/aarch64-simd.md
11675 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
11677 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
11679 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
11680 Make opernads of the unspec commutative.
11682 2018-06-21 Richard Biener <rguenther@suse.de>
11684 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
11685 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
11686 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
11687 (vect_analyze_data_ref_dependence): Re-order checks to deal with
11689 (vect_record_base_alignments): Do not record base alignment
11690 for gathers or scatters.
11691 (vect_compute_data_ref_alignment): Drop return value that is always
11692 true. Bail out early for gathers or scatters.
11693 (vect_enhance_data_refs_alignment): Bail out early for gathers
11695 (vect_find_same_alignment_drs): Likewise.
11696 (vect_analyze_data_refs_alignment): Remove dead code.
11697 (vect_slp_analyze_and_verify_node_alignment): Likewise.
11698 (vect_analyze_data_refs): For possible gathers or scatters do
11699 not create an alternate DR, just check their possible validity
11700 and mark them. Adjust DECL_NONALIASED handling to not rely
11701 on DR_BASE_ADDRESS.
11702 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
11703 update inits of gathers or scatters.
11704 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
11705 Also copy gather/scatter flag to pattern vinfo.
11707 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
11709 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
11710 behavior of vec_packsu (vector unsigned long long, vector unsigned
11711 long long) to match behavior of vec_packs with same signature.
11713 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
11714 Thomas Schwinge <thomas@codesourcery.com>
11715 Cesar Philippidis <cesar@codesourcery.com>
11717 * gimplify.c (gimplify_scan_omp_clauses): Add support for
11718 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
11719 (gimplify_adjust_omp_clauses): Likewise.
11720 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
11721 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
11722 (gimplify_omp_target_update): Update handling of acc update and
11724 * omp-low.c (install_var_field): Remove unused parameter
11725 base_pointers_restrict.
11726 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
11727 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
11729 (omp_target_base_pointers_restrict_p): Delete.
11730 (scan_omp_target): Update call to scan_sharing_clauses.
11731 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
11733 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
11734 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
11735 (convert_local_omp_clauses): Likewise.
11736 * tree-pretty-print.c (dump_omp_clause): Likewise.
11737 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
11739 (omp_clause_code_name): Likewise.
11741 2018-06-20 Jakub Jelinek <jakub@redhat.com>
11744 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
11747 PR tree-optimization/86231
11748 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
11749 anti-range don't overwrite *vr0min before using it to compute *vr0max.
11751 2018-06-20 Tom de Vries <tdevries@suse.de>
11753 PR tree-optimization/86097
11754 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
11755 iv type if signedness of iv type is not the same as that of *nit.
11757 2018-06-20 Jakub Jelinek <jakub@redhat.com>
11759 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
11760 EDGE_EH edges, verify they are all EDGE_EH.
11762 2018-06-20 Maya Rashish <coypu@sdf.org>
11764 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
11766 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11768 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
11769 * config/aarch64/aarch64.c (xgene1_tunings): Add
11770 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
11771 (aarch64_mode_valid_for_sched_fusion_p):
11772 Allow 16-byte modes.
11773 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
11774 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
11776 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
11778 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
11779 * config/aarch64/iterators.md (VQ2): New mode iterator.
11781 2018-06-20 Martin Liska <mliska@suse.cz>
11783 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
11784 Change default ratio from 10 to 8.
11786 2018-06-20 Martin Liska <mliska@suse.cz>
11788 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
11790 (bit_test_cluster::find_bit_tests): Likewise.
11791 (switch_decision_tree::analyze_switch_statement): Find clusters.
11792 * tree-switch-conversion.h (struct jump_table_cluster): Document
11795 2018-06-20 Martin Liska <mliska@suse.cz>
11797 * tree-switch-conversion.c (switch_conversion::collect):
11798 Record m_uniq property.
11799 (switch_conversion::expand): Bail out for special conditions.
11800 (group_cluster::~group_cluster): New.
11801 (group_cluster::group_cluster): Likewise.
11802 (group_cluster::dump): Likewise.
11803 (jump_table_cluster::emit): New.
11804 (switch_decision_tree::fix_phi_operands_for_edges): New.
11805 (struct case_node): Remove struct.
11806 (jump_table_cluster::can_be_handled): New.
11807 (case_values_threshold): Moved to header.
11808 (reset_out_edges_aux): Likewise.
11809 (jump_table_cluster::is_beneficial): New.
11810 (bit_test_cluster::can_be_handled): Likewise.
11811 (add_case_node): Remove.
11812 (bit_test_cluster::is_beneficial): New.
11813 (case_bit_test::cmp): New.
11814 (bit_test_cluster::emit): New.
11815 (expand_switch_as_decision_tree_p): Remove.
11816 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
11817 (fix_phi_operands_for_edge): Likewise.
11818 (switch_decision_tree::analyze_switch_statement): New.
11819 (compute_cases_per_edge): Move ...
11820 (switch_decision_tree::compute_cases_per_edge): ... here.
11821 (try_switch_expansion): Likewise.
11822 (switch_decision_tree::try_switch_expansion): Likewise.
11823 (record_phi_operand_mapping): Likewise.
11824 (switch_decision_tree::record_phi_operand_mapping): Likewise.
11825 (emit_case_decision_tree): Likewise.
11826 (switch_decision_tree::emit): Likewise.
11827 (balance_case_nodes): Likewise.
11828 (switch_decision_tree::balance_case_nodes): Likewise.
11829 (dump_case_nodes): Likewise.
11830 (switch_decision_tree::dump_case_nodes): Likewise.
11831 (emit_jump): Likewise.
11832 (switch_decision_tree::emit_jump): Likewise.
11833 (emit_cmp_and_jump_insns): Likewise.
11834 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
11835 (emit_case_nodes): Likewise.
11836 (switch_decision_tree::emit_case_nodes): Likewise.
11837 (conditional_probability): Remove.
11838 * tree-switch-conversion.h (enum cluster_type): New.
11840 (struct cluster): Likewise.
11841 (cluster::cluster): Likewise.
11842 (struct simple_cluster): Likewise.
11843 (simple_cluster::simple_cluster): Likewise.
11844 (struct group_cluster): Likewise.
11845 (struct jump_table_cluster): Likewise.
11846 (struct bit_test_cluster): Likewise.
11847 (struct min_cluster_item): Likewise.
11848 (struct case_tree_node): Likewise.
11849 (case_tree_node::case_tree_node): Likewise.
11850 (jump_table_cluster::case_values_threshold): Likewise.
11851 (struct case_bit_test): Likewise.
11852 (struct switch_decision_tree): Likewise.
11853 (struct switch_conversion): Likewise.
11854 (switch_decision_tree::reset_out_edges_aux): Likewise.
11856 2018-06-20 Martin Liska <mliska@suse.cz>
11858 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
11859 (hoist_edge_and_branch_if_true): Likewise.
11860 (expand_switch_using_bit_tests_p): Likewise.
11861 (struct case_bit_test): Likewise.
11862 (case_bit_test_cmp): Likewise.
11863 (emit_case_bit_tests): Likewise.
11864 (switch_conversion::switch_conversion): New class.
11865 (struct switch_conv_info): Remove old struct.
11866 (collect_switch_conv_info): More to ...
11867 (switch_conversion::collect): ... this.
11868 (check_range): Likewise.
11869 (switch_conversion::check_range): Likewise.
11870 (check_all_empty_except_final): Likewise.
11871 (switch_conversion::check_all_empty_except_final): Likewise.
11872 (check_final_bb): Likewise.
11873 (switch_conversion::check_final_bb): Likewise.
11874 (create_temp_arrays): Likewise.
11875 (switch_conversion::create_temp_arrays): Likewise.
11876 (free_temp_arrays): Likewise.
11877 (gather_default_values): Likewise.
11878 (switch_conversion::gather_default_values): Likewise.
11879 (build_constructors): Likewise.
11880 (switch_conversion::build_constructors): Likewise.
11881 (constructor_contains_same_values_p): Likewise.
11882 (switch_conversion::contains_same_values_p): Likewise.
11883 (array_value_type): Likewise.
11884 (switch_conversion::array_value_type): Likewise.
11885 (build_one_array): Likewise.
11886 (switch_conversion::build_one_array): Likewise.
11887 (build_arrays): Likewise.
11888 (switch_conversion::build_arrays): Likewise.
11889 (gen_def_assigns): Likewise.
11890 (switch_conversion::gen_def_assigns): Likewise.
11891 (prune_bbs): Likewise.
11892 (switch_conversion::prune_bbs): Likewise.
11893 (fix_phi_nodes): Likewise.
11894 (switch_conversion::fix_phi_nodes): Likewise.
11895 (gen_inbound_check): Likewise.
11896 (switch_conversion::gen_inbound_check): Likewise.
11897 (process_switch): Use the newly created class.
11898 (switch_conversion::expand): New.
11899 (switch_conversion::~switch_conversion): New.
11900 * tree-switch-conversion.h: New file.
11902 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
11904 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
11905 tree-vect-patterns.c.
11906 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
11907 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
11908 (vect_recog_sad_pattern): Likewise.
11909 (vect_recog_widen_sum_pattern): Likewise.
11910 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
11911 (vect_recog_widen_shift_pattern): Remove the type_in argument.
11912 (vect_recog_rotate_pattern): Likewise.
11913 (vect_recog_mult_pattern): Likewise.
11914 (vect_recog_vector_vector_shift_pattern): Likewise.
11915 (vect_recog_divmod_pattern): Likewise.
11916 (vect_recog_mixed_size_cond_pattern): Likewise.
11917 (vect_recog_bool_pattern): Likewise.
11918 (vect_recog_mask_conversion_pattern): Likewise.
11919 (vect_try_gather_scatter_pattern): Likewise.
11920 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
11921 (vect_recog_over_widening_pattern): Likewise.
11922 (vect_recog_gather_scatter_pattern): Likewise.
11923 (vect_recog_func_ptr): Move from tree-vectorizer.h
11924 (vect_vect_recog_func_ptrs): Move further down the file.
11925 (vect_recog_func): Likewise. Remove the third argument.
11926 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
11927 (vect_pattern_recog_1): Expect the pattern function to do any
11928 necessary target tests. Also expect it to provide a vector type.
11929 Remove the type_in handling.
11931 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
11933 * tree-vect-patterns.c (vect_pattern_detected): New function.
11934 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
11935 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
11936 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
11937 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
11938 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
11939 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
11940 (vect_recog_mask_conversion_pattern)
11941 (vect_try_gather_scatter_pattern): Likewise.
11943 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
11945 * tree-vect-patterns.c (vect_get_internal_def): New function.
11946 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
11947 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
11948 (search_type_for_mask_1): Use it.
11950 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
11952 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
11953 redundant WIDEN_SUM_EXPR handling.
11954 (vect_recog_sad_pattern): Likewise.
11956 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
11958 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
11959 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
11960 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
11961 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
11962 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
11964 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
11966 * tree-vect-stmts.c (vectorizable_call): Make sure that we
11967 use the stmt_vec_info of the original bb statement for the
11968 new zero assignment, even if the call is part of a pattern.
11970 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
11972 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
11973 that the sequence is attached to the original statement rather
11974 than the pattern statement.
11975 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
11976 PATTERN_DEF_SEQ from the original statement rather than
11977 the main pattern statement.
11978 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
11979 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
11980 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
11982 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
11984 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
11985 definition statements before the early exit for statements that aren't
11987 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
11989 (vect_transform_loop): ...here. Process pattern definition
11990 statements without first checking whether the main pattern
11991 statement is live or relevant.
11993 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
11995 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
11996 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
11998 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
12000 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
12001 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
12002 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
12003 (expand_block_compare): Change select_block_compare_mode call.
12004 (expand_strncmp_align_check): Use new functions, fix comment.
12005 (emit_final_str_compare_gpr): New function.
12006 (expand_strn_compare): Refactor and clean up code.
12007 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
12009 2018-06-19 Tony Reix <tony.reix@atos.com>
12010 Damien Bergamini <damien.bergamini@atos.com>
12011 David Edelsohn <dje.gcc@gmail.com>
12013 * collect2.c (static_obj): New variable.
12014 (static_libs): New variable.
12015 (is_in_list): Uncomment declaration.
12016 (main): Track AIX libraries linked statically.
12017 (is_in_list): Uncomment definition.
12018 (scan_prog_file): Don't add AIX shared libraries initializer
12019 to constructor list if linking statically.
12021 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
12023 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
12025 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
12027 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
12029 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
12032 2018-06-19 Martin Liska <mliska@suse.cz>
12034 * config/i386/i386.c (ix86_can_inline_p): Do not use
12035 ipa_fn_summaries::get_create.
12036 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
12038 (devirtualization_time_bonus): Likewise.
12039 (ipcp_propagate_stage): Likewise.
12040 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
12041 (edge_set_predicate): Likewise.
12042 (evaluate_conditions_for_known_args): Likewise.
12043 (evaluate_properties_for_edge): Likewise.
12044 (ipa_call_summary::reset): Tranform to ...
12045 (ipa_call_summary::~ipa_call_summary): ... this.
12046 (ipa_fn_summary::reset): Transform to ...
12047 (ipa_fn_summary::~ipa_fn_summary): ... this.
12048 (ipa_fn_summary_t::remove): Rename to ...
12049 (ipa_fn_summary_t::remove_callees): ... this.
12050 (ipa_fn_summary_t::duplicate): Use placement new
12051 instead of memory copy.
12052 (ipa_call_summary_t::duplicate): Likewise.
12053 (ipa_call_summary_t::remove): Remove.
12054 (dump_ipa_call_summary): Change get_create to get.
12055 (ipa_dump_fn_summary): Dump only when summary exists.
12056 (analyze_function_body): Use symbol_summary::get instead
12058 (compute_fn_summary): Likewise.
12059 (estimate_edge_devirt_benefit): Likewise.
12060 (estimate_edge_size_and_time): Likewise.
12061 (inline_update_callee_summaries): Likewise.
12062 (remap_edge_change_prob): Likewise.
12063 (remap_edge_summaries): Likewise.
12064 (ipa_merge_fn_summary_after_inlining): Likewise.
12065 (write_ipa_call_summary): Likewise.
12066 (ipa_fn_summary_write): Likewise.
12067 (ipa_free_fn_summary): Likewise.
12068 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
12069 (struct ipa_call_summary): Likewise.
12070 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
12072 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
12073 (estimate_size_after_inlining): Likewise.
12074 (estimate_growth): Likewise.
12075 (growth_likely_positive): Likewise.
12076 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
12077 (inline_call): Likewise.
12078 * ipa-inline.c (caller_growth_limits): Likewise.
12079 (can_inline_edge_p): Likewise.
12080 (can_inline_edge_by_limits_p): Likewise.
12081 (compute_uninlined_call_time): Likewise.
12082 (compute_inlined_call_time): Likewise.
12083 (want_inline_small_function_p): Likewise.
12084 (edge_badness): Likewise.
12085 (update_caller_keys): Likewise.
12086 (update_callee_keys): Likewise.
12087 (inline_small_functions): Likewise.
12088 (inline_to_all_callers_1): Likewise.
12089 (dump_overall_stats): Likewise.
12090 (early_inline_small_functions): Likewise.
12091 (early_inliner): Likewise.
12092 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
12093 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
12094 * ipa-pure-const.c (malloc_candidate_p): Likewise.
12095 * ipa-split.c (execute_split_functions): Likewise.
12096 * symbol-summary.h: Likewise.
12097 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
12099 2018-06-19 Richard Biener <rguenther@suse.de>
12101 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
12102 (vectorize_loops): ... here. Fix dbgcnt handling.
12103 (try_vectorize_loop): Wrap try_vectorize_loop_1.
12105 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
12108 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
12109 ieee128 argument takes up only one (vector) register, not two (floating
12112 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
12114 * gimplify.c (gimplify_init_constructor): Really never clear for an
12115 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
12117 2018-06-19 Richard Biener <rguenther@suse.de>
12119 PR tree-optimization/86179
12120 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
12121 after failed recognition.
12123 2018-06-18 Martin Sebor <msebor@redhat.com>
12125 PR middle-end/85602
12126 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
12127 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
12128 Handle integer subtraction.
12129 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
12130 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
12132 2018-06-18 David Malcolm <dmalcolm@redhat.com>
12134 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
12135 param from rtx to rtx_insn *.
12136 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
12138 (frv_ifcvt_modify_insn): Likwise.
12139 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
12140 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
12141 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
12142 as_a <rtx_insn *> cast to local "unprotected_region" once
12143 it's been established that it's not NULL or pc_rtx.
12144 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
12145 param "sethi" from rtx to rtx_insn *.
12146 (nds32_group_float_insns): Likewise for param "insn".
12147 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
12149 (vax_output_int_subtract): Likewise.
12150 * config/vax/vax.c (vax_output_int_add): Likewise for param
12152 (vax_output_int_subtract): Likewise.
12153 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
12154 (emit_pattern_after): Likewise for param "after".
12155 (emit_insn_after): Likewise.
12156 (emit_jump_insn_after): Likewise.
12157 (emit_call_insn_after): Likewise.
12158 (emit_debug_insn_after): Likewise.
12159 (emit_pattern_before): Likewise for param "before".
12160 (emit_insn_before): Likewise.
12161 (emit_jump_insn_before): Likewise.
12162 * final.c (get_insn_template): Likewise for param "insn", removing
12164 * output.h (get_insn_template): Likewise for 2nd param.
12165 * rtl.h (emit_insn_before): Likewise.
12166 (emit_jump_insn_before): Likewise.
12167 (emit_debug_insn_before_noloc): Likewise.
12168 (emit_insn_after): Likewise.
12169 (emit_jump_insn_after): Likewise.
12170 (emit_call_insn_after): Likewise.
12171 (emit_debug_insn_after): Likewise.
12172 (set_insn_deleted): Likewise for param.
12174 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
12177 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
12178 floating point modes, so that IFmode is numerically greater than
12179 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
12180 to declare the ordering. This prevents IFmode from being
12181 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
12182 machine. Include rs6000-modes.h to share the fractional values
12183 between genmodes* and the rest of the compiler.
12184 (IFmode): Likewise.
12185 (KFmode): Likewise.
12186 (TFmode): Likewise.
12187 * config/rs6000/rs6000-modes.h: New file.
12188 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
12189 meaning of rs6000_long_double_size so that 126..128 selects an
12190 appropriate 128-bit floating point type.
12191 (rs6000_option_override_internal): Likewise.
12192 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
12193 (TARGET_LONG_DOUBLE_128): Change the meaning of
12194 rs6000_long_double_size so that 126..128 selects an appropriate
12195 128-bit floating point type.
12196 (LONG_DOUBLE_TYPE_SIZE): Update comment.
12197 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
12198 source and destination to match the standard usage.
12199 (truncifkf2): Likewise.
12200 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
12201 ISA 2.07 to use an explicit clobber, instead of passing in a
12203 (copysign<mode>3_soft): Likewise.
12205 2018-06-18 David Malcolm <dmalcolm@redhat.com>
12207 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
12208 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
12209 (vect_slp_analyze_instance_dependence): Likewise.
12210 (vect_enhance_data_refs_alignment): Likewise.
12211 (vect_analyze_data_refs_alignment): Likewise.
12212 (vect_slp_analyze_and_verify_instance_alignment
12213 (vect_analyze_data_ref_accesses): Likewise.
12214 (vect_prune_runtime_alias_test_list): Likewise.
12215 (vect_analyze_data_refs): Likewise.
12216 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
12217 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
12218 (vect_analyze_scalar_cycles_1): Likewise.
12219 (vect_get_loop_niters): Likewise.
12220 (vect_analyze_loop_form_1): Likewise.
12221 (vect_update_vf_for_slp): Likewise.
12222 (vect_analyze_loop_operations): Likewise.
12223 (vect_analyze_loop): Likewise.
12224 (vectorizable_induction): Likewise.
12225 (vect_transform_loop): Likewise.
12226 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
12227 * tree-vect-slp.c (vect_analyze_slp): Likewise.
12228 (vect_make_slp_decision): Likewise.
12229 (vect_detect_hybrid_slp): Likewise.
12230 (vect_slp_analyze_operations): Likewise.
12231 (vect_slp_bb): Likewise.
12232 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
12233 (vectorizable_bswap): Likewise.
12234 (vectorizable_call): Likewise.
12235 (vectorizable_simd_clone_call): Likewise.
12236 (vectorizable_conversion): Likewise.
12237 (vectorizable_assignment): Likewise.
12238 (vectorizable_shift): Likewise.
12239 (vectorizable_operation): Likewise.
12240 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
12242 2018-06-18 Martin Sebor <msebor@redhat.com>
12244 PR tree-optimization/81384
12245 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
12246 * builtins.c (expand_builtin_strnlen): New function.
12247 (expand_builtin): Call it.
12248 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
12249 * builtins.def (BUILT_IN_STRNLEN): New.
12250 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
12251 Warn for bounds in excess of maximum object size.
12252 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
12253 single-value ranges. Handle strnlen.
12254 (handle_builtin_strlen): Handle strnlen.
12255 (strlen_check_and_optimize_stmt): Same.
12256 * doc/extend.texi (Other Builtins): Document strnlen.
12258 2018-06-18 Maya Rashish <coypu@sdf.org>
12260 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
12261 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
12262 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
12264 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
12266 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
12268 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12270 * tree.c (escaped_string::escape): Replace cast to char * by
12271 const_cast<char *> (unescaped).
12273 2018-06-18 Nick Clifton <nickc@redhat.com>
12276 * tree.c (escaped_string): New class. Converts an unescaped
12277 string into its escaped equivalent.
12278 (warn_deprecated_use): Use the new class to convert the
12279 deprecation message, if present.
12280 (test_escaped_strings): New self test.
12281 (test_c_tests): Add test_escaped_strings.
12282 * doc/extend.texi (deprecated): Add a note that the
12283 deprecation message is affected by the -fmessage-length
12284 option, and that control characters will be escaped.
12285 (#pragma GCC error): Document this pragma.
12286 (#pragma GCC warning): Likewise.
12287 * doc/invoke.texi (-fmessage-length): Document this option's
12288 effect on the #warning and #error preprocessor directives and
12289 the deprecated attribute.
12291 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
12293 * tree.c (decl_value_expr_lookup): Revert latest change.
12294 (decl_value_expr_insert): Likewise.
12296 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
12298 * gimplify.c (nonlocal_vlas): Delete.
12299 (nonlocal_vla_vars): Likewise.
12300 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
12302 (gimplify_body): Do not create and destroy nonlocal_vlas.
12303 * tree-nested.c: Include diagnostic.h.
12304 (use_pointer_in_frame): Tweak.
12305 (lookup_field_for_decl): Add assertion and declare the transformation.
12306 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
12307 internal error when the reference is in a wrong context. Do not
12308 create a debug decl by default.
12309 (note_nonlocal_block_vlas): Delete.
12310 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
12311 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
12312 create a debug decl by default.
12313 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
12314 call is in a wrong context.
12315 (fixup_vla_decls): New function.
12316 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
12317 debug variables were created.
12318 * tree.c (decl_value_expr_lookup): Add checking assertion.
12319 (decl_value_expr_insert): Likewise.
12321 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
12323 PR middle-end/82479
12324 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
12325 * tree-scalar-evolution.c (interpret_expr): Likewise.
12326 (expression_expensive_p): Likewise.
12327 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
12328 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
12329 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
12330 (ssa_defined_by_minus_one_stmt_p): New.
12332 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
12334 PR middle-end/64946
12335 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
12336 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
12337 * dojump.c (do_jump): Likewise.
12338 * expr.c (expand_expr_real_2): Check operand type's sign.
12339 * fold-const.c (const_unop): Handle ABSU_EXPR.
12340 (fold_abs_const): Likewise.
12341 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
12342 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
12343 (strip_sign_op_1): Likesise.
12344 * match.pd: Add new pattern to generate ABSU_EXPR.
12345 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
12346 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
12347 * tree-eh.c (operation_could_trap_helper_p): Likewise.
12348 * tree-inline.c (estimate_operator_cost): Likewise.
12349 * tree-pretty-print.c (dump_generic_node): Likewise.
12350 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
12351 * tree.def (ABSU_EXPR): New.
12353 2018-06-16 Jakub Jelinek <jakub@redhat.com>
12355 PR middle-end/86095
12356 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
12357 documented as preserved for backward compatibility only.
12358 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
12360 PR rtl-optimization/86108
12361 * bb-reorder.c (create_forwarder_block): Renamed to ...
12362 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
12363 jump from new landing pad to the second part.
12364 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
12367 2018-06-15 Jakub Jelinek <jakub@redhat.com>
12369 PR middle-end/85878
12370 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
12371 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
12372 Only call store_expr for halves if the mode is the same.
12374 PR middle-end/86123
12375 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
12376 Fix up comment formatting.
12378 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
12380 * typed-splay-tree.h (typed_splay_tree::remove): New function.
12381 (typed_splay_tree::closure,
12382 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
12383 (typed_splay_tree::typed_splay_tree,
12384 typed_splay_tree::operator =): Declared private.
12385 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
12386 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
12387 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
12388 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
12389 typed_splay_tree::splay_tree_splay,
12390 typed_splay_tree::splay_tree_foreach_helper,
12391 typed_splay_tree::splay_tree_insert,
12392 typed_splay_tree::splay_tree_remove,
12393 typed_splay_tree::splay_tree_lookup,
12394 typed_splay_tree::splay_tree_predecessor,
12395 typed_splay_tree::splay_tree_successor,
12396 typed_splay_tree::splay_tree_min,
12397 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
12398 (typed_splay_tree::root, typed_splay_tree::comp,
12399 typed_splay_tree::delete_key,
12400 typed_splay_tree::delete_value): New data members.
12401 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
12402 typed_splay_tree::remove.
12404 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
12406 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
12407 -mginv and -mno-ginv to the assembler.
12408 * config/mips/mips.opt (-mcrc): New option.
12409 (-mginv): Likewise.
12410 * doc/invoke.text (-mcrc): Document.
12411 (-mginv): Likewise.
12413 2018-06-15 Nick Clifton <nickc@redhat.com>
12416 * tree.c (escaped_string): New class. Converts an unescaped
12417 string into its escaped equivalent.
12418 (warn_deprecated_use): Use the new class to convert the
12419 deprecation message, if present.
12420 (test_escaped_strings): New self test.
12421 (test_c_tests): Add test_escaped_strings.
12422 * doc/extend.texi (deprecated): Add a note that the
12423 deprecation message is affected by the -fmessage-length
12424 option, and that control characters will be escaped.
12425 (#pragma GCC error): Document this pragma.
12426 (#pragma GCC warning): Likewise.
12427 * doc/invoke.texi (-fmessage-length): Document this option's
12428 effect on the #warning and #error preprocessor directives and
12429 the deprecated attribute.
12431 2018-06-15 Richard Biener <rguenther@suse.de>
12433 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
12434 here, also noting vector size used.
12435 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
12436 size used in MSG_OPTIMIZED_LOCATIONS dump.
12437 (pass_slp_vectorize::execute): Adjust.
12439 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
12442 * config/arc/arc.c (arc_return_address_register): Fix
12445 2018-06-15 Richard Biener <rguenther@suse.de>
12447 PR middle-end/86159
12448 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
12449 leave useless conversion stripping to force_gimple_operand_gsi.
12450 (gimplify_build2): Likewise.
12451 (gimplify_build1): Likewise.
12453 2018-06-15 Richard Biener <rguenther@suse.de>
12455 PR middle-end/86076
12456 * tree-cfg.c (move_stmt_op): unshare invariant addresses
12457 before adjusting their block.
12459 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
12461 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
12462 multilibs for *-*-rtems*.
12463 * config/riscv/t-rtems: New file.
12465 2018-06-14 Jakub Jelinek <jakub@redhat.com>
12467 PR middle-end/86122
12468 * match.pd ((A +- CST1) +- CST2): Punt if last resort
12469 unsigned_type_for returns NULL.
12472 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
12473 subregs of multi-word pseudos unless the float mode has word size.
12475 2018-06-14 Richard Biener <rguenther@suse.de>
12477 PR middle-end/86139
12478 * tree-vect-generic.c (build_word_mode_vector_type): Remove
12479 duplicate and harmful type_hash_canon.
12480 * tree.c (type_hash_canon): Assert we didn't find ourselves.
12482 2018-06-14 Richard Biener <rguenther@suse.de>
12485 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
12488 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
12490 * config/rtems.h (STDINT_LONG32): Define.
12492 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
12493 Prachi Godbole <prachi.godbole@imgtec.com>
12495 * config/mips/mips-cpus.def: Define P6600.
12496 * config/mips/mips-tables.opt: Regenerate.
12497 * config/mips/mips.c (mips_ucbranch_type): New enum.
12498 (mips_rtx_cost_data): Add support for P6600.
12499 (mips_issue_rate): Likewise.
12500 (mips_multipass_dfa_lookahead): Likewise.
12501 (mips_avoid_hazard): Likewise.
12502 (mips_reorg_process_insns): Likewise.
12503 (mips_classify_branch_p6600): New function.
12504 * config/mips/mips.h (TUNE_P6600): New define.
12505 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
12506 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
12507 * config/mips/mips.md: Include p6600.md.
12508 (processor): Add p6600.
12509 * config/mips/p6600.md: New file.
12510 * doc/invoke.texi: Add p6600 to supported architectures.
12512 2018-06-13 Martin Sebor <msebor@redhat.com>
12514 PR tree-optimization/86114
12515 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
12517 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
12519 2018-06-13 Richard Biener <rguenther@suse.de>
12521 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
12522 Properly set vector type of the intermediate stmt.
12523 * tree-vect-stmts.c (vectorizable_operation): The destination
12524 var always has vectype_out type.
12526 2018-06-13 Jeff Law <law@redhat.com>
12528 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
12529 integer 0 for argument to print_rtl_with_bb.
12530 (rl78_reorg): Likewise.
12532 2018-06-13 David Malcolm <dmalcolm@redhat.com>
12534 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
12535 from rtx to rtx_insn *.
12536 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
12538 (add_sched_insns_for_speculation): Likewise for local "target",
12539 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
12540 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
12541 from rtx_insn ** to rtx_code_label **.
12542 (reorg_emit_nops): Likewise.
12543 (c6x_reorg): Likewise for local "call_labels".
12544 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
12546 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
12547 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
12548 the loops over LABEL_REFS.
12549 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
12551 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
12552 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
12553 (split_branches): Strengthen local "olabel" from rtx to
12554 rtx_insn *, adding a safe_as_a cast.
12555 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
12557 (add_insn_after): Likewise for first two params.
12558 (add_insn_before): Likewise.
12559 (remove_insn): Likewise for param.
12560 (emit_pattern_before_noloc): Likewise for second and third params.
12561 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
12562 (emit_call_insn_before_noloc): Likewise.
12563 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
12565 (emit_barrier_before): Likewise.
12566 (emit_label_before): Strengthen "label" param from "rtx" to
12567 "rtx_code_label *". Strengthen "before" param from "rtx" to
12569 (emit_insn_after_1): Strengthen "after" param from "rtx" to
12571 (emit_pattern_after_noloc): Likewise.
12572 (emit_insn_after_noloc): Likewise.
12573 (emit_jump_insn_after_noloc): Likewise.
12574 (emit_call_insn_after_noloc): Likewise.
12575 (emit_debug_insn_after_noloc): Likewise.
12576 (emit_barrier_after): Likewise.
12577 (emit_label_after): Likewise for both params.
12578 (emit_pattern_after_setloc): Likewise for "after" param. Convert
12579 "loc" param from "int" to "location_t".
12580 (emit_insn_after_setloc): Likewise.
12581 (emit_jump_insn_after_setloc): Likewise.
12582 (emit_call_insn_after_setloc): Likewise.
12583 (emit_debug_insn_after_setloc): Likewise.
12584 (emit_pattern_before_setloc): Likewise for "before" param. Convert
12585 "loc" param from "int" to "location_t".
12586 (emit_pattern_before): Convert NULL_RTX to NULL.
12587 (emit_insn_before_setloc): Convert "loc" param from "int" to
12589 (emit_jump_insn_before_setloc): Likewise.
12590 (emit_call_insn_before_setloc): Likewise.
12591 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
12592 rtx_insn *. Convert "loc" param from "int" to "location_t".
12593 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
12594 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
12595 Convert 3rd param from "int" to "location_t".
12596 (emit_barrier_before, emit_barrier_after, next_real_insn):
12597 Strengthen param from rtx to rtx_insn *.
12598 (emit_label_before): Strengthen 1st param from "rtx" to
12599 "rtx_code_label *". Strengthen 2nd param from "rtx" to
12601 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
12602 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
12603 Strengthen 2nd param from "rtx" to "rtx_insn *".
12604 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
12605 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
12606 Likewise. Convert 3rd param from "int" to "location_t".
12607 (emit_label_after): Strengthen 1st param from "rtx" to
12608 "rtx_code_label *".
12609 (next_real_insn, remove_insn): Strengthen param from "rtx" to
12611 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
12612 from "rtx" to "rtx_insn *".
12614 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
12616 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
12617 bodies streamed in with -Q.
12618 * dumpfile.c (dump_files): Add lto-stream-out dump file.
12619 * dumpfile.h (tree_dump_index): Add lto_stream_out.
12620 * gimple-streamer-out.c: Include gimple-pretty-print.h
12621 (output_bb): Dump stmts streamed.
12622 * lto-section-out.c: Include print-tree.h
12623 (lto_begin_section): Dump sections created.
12624 (lto_output_decl_index): Dump decl encoded.
12625 * lto-streamer-out.c: Include print-tree.h
12626 (create_output_block): Dump output block created.
12627 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
12628 (output_function): Dump function output.
12629 (output_constructor): Dump constructor streamed.
12630 (write_global_stream): Output indexes encoded.
12631 (produce_asm_for_decls): Dump streams encoded.
12632 * lto-streamer.c (streamer_dump_file): New global var.
12633 * lto-streamer.h (streamer_dump_file): Declare.
12634 * passes.c (ipa_write_summaries): Initialize streamer dump.
12635 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
12638 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
12641 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
12642 offsets for register save directives. Emit a second batch of save
12643 directives, if need be, when the function accesses prior frames.
12645 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
12647 * config/arc/fpu.md (fmasf4): Force operand to register.
12648 (fnmasf4): Likewise.
12650 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
12652 * config/arc/arc-protos.h (arc_pad_return): Remove.
12653 * config/arc/arc.c (machine_function): Remove force_short_suffix
12655 (arc_print_operand): Adjust printing of '&'.
12656 (arc_verify_short): Remove conditional printing of short suffix.
12657 (arc_final_prescan_insn): Remove reference to size_reason.
12658 (pad_return): New function.
12659 (arc_reorg): Call pad_return.
12660 (arc_pad_return): Remove.
12661 (arc_init_machine_status): Remove reference to force_short_suffix.
12662 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
12663 (attr length): When attribute iscompact is true force to 2
12664 regardless; in the case of maybe check if we want to force the
12665 instruction to have 4 bytes length.
12666 (nopv): Change it to generate 4 byte long nop as well.
12667 (blockage): New pattern.
12668 (simple_return): Remove call to arc_pad_return.
12669 (p_return_i): Likewise.
12671 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
12673 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
12675 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
12677 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
12680 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
12682 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
12683 for ARC700 and ARCv2.
12685 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
12688 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
12689 operands[2] instead of operands[1].
12692 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
12694 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
12695 case, check whether the outer register overlaps an unallocatable
12696 register, not just whether it fits the required class.
12698 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
12700 * poly-int.h (can_div_trunc_p): Add new overload in which all values
12702 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
12703 (memrefs_conflict_p): Likewise.
12704 (init_alias_analysis): Likewise.
12705 * cfgexpand.c (expand_debug_expr): Likewise.
12706 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
12707 * cse.c (fold_rtx): Likewise.
12708 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
12709 * expr.c (emit_block_move_hints): Likewise.
12710 (clear_storage_hints, push_block, emit_push_insn): Likewise.
12711 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
12712 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
12713 (emit_group_store): Likewise.
12714 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
12715 to read the PRE/POST_MODIFY increment.
12716 * calls.c (store_one_arg): Use strip_offset.
12717 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
12719 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
12721 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
12722 (simplify_binary_operation_1): Extend CONST_INT handling to
12724 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
12725 than a HOST_WIDE_INT.
12726 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
12728 (adjust_mems, add_stores): Update accodingly.
12729 (vt_canonicalize_addr): Track polynomial offsets.
12730 (emit_note_insn_var_location): Likewise.
12731 (vt_add_function_parameter): Likewise.
12732 (vt_initialize): Likewise.
12734 2018-06-12 Jeff Law <law@redhat.com>
12736 * config.gcc (alpha*-*-freebsd*): Remove.
12737 * config/alpha/freebsd.h: Remove.
12739 2018-06-12 David Malcolm <dmalcolm@redhat.com>
12742 * spellcheck-tree.c (levenshtein_distance): Rename to...
12743 (get_edit_distance): ...this, and update for underlying renaming.
12744 * spellcheck-tree.h (levenshtein_distance): Rename to...
12745 (get_edit_distance): ...this.
12746 * spellcheck.c (levenshtein_distance): Rename to...
12747 (get_edit_distance): ...this. Convert from Levenshtein distance
12748 to Damerau-Levenshtein distance by supporting transpositions of
12749 adjacent characters. Rename "v1" to "v_next" and "v0" to
12751 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
12752 (selftest::test_edit_distance_unit_test_oneway): ...this, and
12753 update for underlying renaming.
12754 (selftest::levenshtein_distance_unit_test): Rename to...
12755 (selftest::test_get_edit_distance_unit): ...this, and update for
12756 underlying renaming.
12757 (selftest::test_find_closest_string): Add example from PR 69968
12758 where transposition helps
12759 (selftest::test_metric_conditions): Update for renaming.
12760 (selftest::test_metric_conditions): Likewise.
12761 (selftest::spellcheck_c_tests): Likewise.
12762 * spellcheck.h (levenshtein_distance): Rename both overloads to...
12763 (get_edit_distance): ...this.
12764 (best_match::consider): Update for renaming.
12766 2018-06-12 Martin Sebor <msebor@redhat.com>
12768 PR tree-optimization/85259
12769 * builtins.c (compute_objsize): Handle constant offsets.
12770 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
12771 true iff a warning has been issued.
12772 * gimple.h (gimple_nonartificial_location): New function.
12773 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
12774 gimple_nonartificial_location and handle -Wno-system-headers.
12775 (handle_builtin_stxncpy): Same.
12777 2018-06-12 Martin Sebor <msebor@redhat.com>
12780 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
12782 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
12784 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12785 BUILTIN_VEC_XST entries for pointer to double and long long.
12787 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
12790 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
12792 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
12795 2018-06-12 Martin Liska <mliska@suse.cz>
12797 * doc/options.texi: Document IntegerRange.
12799 2018-06-12 Martin Liska <mliska@suse.cz>
12801 * config/i386/i386.opt: Make MPX-related options as Deprecated.
12802 * opt-functions.awk: Handle Deprecated flag.
12803 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
12805 (read_cmdline_option): Report warning for a deprecated option.
12806 * opts.h (struct cl_option): Add new field cl_deprecated.
12807 (CL_ERR_DEPRECATED): New.
12809 2018-06-12 Martin Liska <mliska@suse.cz>
12811 * doc/options.texi: Document Deprecated option flag.
12813 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
12815 * config/arc/arc-arch.h (arc_extras): New enum.
12816 (arc_cpu_t):Add field extra.
12817 (arc_cpu_types): Consider the extras.
12818 * config/arc/arc-cpus.def: Add extras info.
12819 * config/arc/arc-opts.h (processor_type): Consider extra field.
12820 * config/arc/arc.c (arc_override_options): Handle extra field.
12822 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
12824 * config/arc/arc-arch.h: Update ARC_OPTX macro.
12825 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
12827 * config/arc/arc.c (arc_init): Update pic warning.
12828 (irq_range): Update irq range parsing warnings.
12829 (arc_override_options): Update various warning messages.
12830 (arc_handle_aux_attribute): Likewise.
12832 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
12834 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
12836 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
12838 * doc/sourcebuild.texi: Document usage of line number 0 in verify
12839 compiler messages directives.
12841 2018-06-12 Matthew Fortune <mfortune@gmail.com>
12843 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
12844 * config/mips/mips-tables.opt: Regenerate.
12845 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
12847 * doc/invoke.texi: Document -march=i6500.
12849 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
12851 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
12852 (i6400_gpmul): Add cpu_unit.
12853 (i6400_gpdiv): Likewise.
12854 (i6400_msa_add_d): Update reservations.
12855 (i6400_msa_int_add) Likewise.
12856 (i6400_msa_short_logic3) Likewise.
12857 (i6400_msa_short_logic2) Likewise.
12858 (i6400_msa_short_logic) Likewise.
12859 (i6400_msa_move) Likewise.
12860 (i6400_msa_cmp) Likewise.
12861 (i6400_msa_short_float2) Likewise.
12862 (i6400_msa_div_d) Likewise.
12863 (i6400_msa_long_logic1) Likewise.
12864 (i6400_msa_long_logic2) Likewise.
12865 (i6400_msa_mult) Likewise.
12866 (i6400_msa_long_float2) Likewise.
12867 (i6400_msa_long_float4) Likewise.
12868 (i6400_msa_long_float5) Likewise.
12869 (i6400_msa_long_float8) Likewise.
12870 (i6400_fpu_fadd): Include frint type.
12871 (i6400_fpu_store): New define_insn_reservation.
12872 (i6400_fpu_load): Likewise.
12873 (i6400_fpu_move): Likewise.
12874 (i6400_fpu_fcmp): Likewise.
12875 (i6400_fpu_fmadd): Likewise.
12876 (i6400_int_mult): Include imul3nc type and update reservation.
12877 (i6400_int_div): Include idiv3 type and update reservation.
12878 (i6400_int_load): Update to check type not move_type.
12879 (i6400_int_store): Likewise.
12880 (i6400_int_prefetch): Set zero latency.
12882 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
12884 * gcc.c: Document new %@{...} sequence.
12885 (LINK_COMMAND_SPEC): Use it for the -L switches.
12886 (cpp_unique_options): Use it for the -I switches.
12887 (at_file_argbuf): New global variable.
12888 (in_at_file): Likewise.
12889 (alloc_args): Create at_file_argbuf.
12890 (clear_args): Truncate at_file_argbuf.
12891 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
12892 (open_at_file): New function.
12893 (close_at_file): Likewise.
12894 (create_at_file): Delete.
12895 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
12898 (validate_switches_from_spec): Deal with %@{...} sequence.
12899 (validate_switches): Likewise.
12900 (driver::finalize): Call clear_args.
12902 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
12904 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
12906 2018-06-11 Martin Sebor <msebor@redhat.com>
12908 * doc/invoke.texi (-Wall): List -Wc++17-compat.
12909 (Wno-class-memaccess): Add @opindex.
12910 (Wno-templates, Wno-multiple-inheritance): Same.
12911 (Wno-virtual-inheritance, Wno-namespaces): Same.
12912 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
12913 (Wno-format-overflow, Wno-format-truncation): Same.
12914 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
12915 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
12916 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
12917 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
12918 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
12919 (Wno-misspelled-isr): Same.
12921 2018-06-11 Martin Sebor <msebor@redhat.com>
12923 * PR tree-optimization/86083
12924 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
12926 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
12928 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
12930 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
12933 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
12934 on the correct operand.
12935 (*movdi_internal64): Ditto.
12937 2018-06-11 Martin Liska <mliska@suse.cz>
12939 PR tree-optimization/86089
12940 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
12942 2018-06-11 Julia Koval <julia.koval@intel.com>
12944 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
12945 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
12946 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
12948 2018-06-11 Olivier Hainque <hainque@adacore.com>
12950 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
12951 for Ada with strict dwarf2.
12953 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
12956 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
12959 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
12961 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
12963 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
12965 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
12968 2018-06-08 Martin Liska <mliska@suse.cz>
12970 * tree-cfg.h (debug_function): Fix argument type to match
12973 2018-06-08 Martin Liska <mliska@suse.cz>
12975 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
12976 Remove usage of MPX-related (and removed) fields.
12977 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
12979 2018-06-08 David Malcolm <dmalcolm@redhat.com>
12981 * cfg.c (debug): Use TDF_NONE rather than 0.
12982 * cfghooks.c (debug): Likewise.
12983 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
12984 (struct dump_option_value_info): Convert to...
12985 (struct kv_pair): ...this template type.
12986 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
12988 (optinfo_verbosity_options): Likewise.
12989 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
12991 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
12992 than int for "optgroup_flags" param.
12993 (dump_generic_expr_loc): Use dump_flags_t rather than int for
12995 (dump_dec): Likewise.
12996 (dump_finish): Use TDF_NONE rather than 0.
12997 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
12998 rather than int for "optgroup_flags" param. Use TDF_NONE rather
12999 than 0. Update for change to option_ptr.
13000 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
13001 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
13002 0. Update for changes to optinfo_verbosity_options and
13004 (opt_info_switch_p): Convert optgroup_flags from int to
13006 (dump_basic_block): Use dump_flags_t rather than int
13007 for "dump_kind" param.
13008 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
13009 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
13010 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
13011 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
13012 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
13013 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
13014 TDF_NONE): Convert from macros to...
13015 (enum dump_flag): ...this new enum.
13016 (dump_flags_t): Update to use enum.
13017 (operator|, operator&, operator~, operator|=, operator&=):
13018 Implement for dump_flags_t.
13019 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
13020 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
13021 Convert from macros to...
13022 (enum optgroup_flag): ...this new enum.
13023 (optgroup_flags_t): New typedef.
13024 (operator|, operator|=): Implement for optgroup_flags_t.
13025 (struct dump_file_info): Convert field "alt_flags" to
13026 dump_flags_t. Convert field "optgroup_flags" to
13028 (dump_basic_block): Use dump_flags_t rather than int for param.
13029 (dump_generic_expr_loc): Likewise.
13030 (dump_dec): Likewise.
13031 (dump_register): Convert param "optgroup_flags" to
13033 (opt_info_enable_passes): Likewise.
13034 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
13036 * gimple-pretty-print.c (debug): Likewise.
13037 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
13038 (merged_store_group::apply_stores): Likewise.
13039 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
13040 * gimple.c (verify_gimple_pp): Likewise.
13041 * graphite-poly.c (print_pbb_body): Likewise.
13042 * passes.c (pass_manager::register_one_dump_file): Convert
13043 local "optgroup_flags" to optgroup_flags_t.
13044 * print-tree.c (print_node): Use TDF_NONE rather than 0.
13046 (debug_body): Likewise.
13047 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
13048 to optgroup_flags_t.
13049 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
13051 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
13052 (convert_mult_to_fma): Likewise.
13053 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
13054 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
13055 * tree-vect-data-refs.c (dump_lower_bound): Convert param
13056 "dump_kind" to dump_flags_t.
13058 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
13060 * config/rs6000/rs6000.c (min, max): Delete.
13062 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
13064 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
13065 -mabi=spe and -mabi=no-spe.
13067 2018-06-08 Martin Liska <mliska@suse.cz>
13069 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
13070 where we expect an existing summary.
13072 2018-06-08 Martin Liska <mliska@suse.cz>
13074 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
13075 * ipa-inline.h (estimate_edge_growth): Likewise.
13077 2018-06-08 Martin Liska <mliska@suse.cz>
13079 * cgraph.c (function_version_hasher::hash): Use
13080 cgraph_node::get_uid ().
13081 (function_version_hasher::equal):
13082 * cgraph.h (cgraph_node::get_uid): New method.
13083 * ipa-inline.c (update_caller_keys): Use
13084 cgraph_node::get_uid ().
13085 (update_callee_keys): Likewise.
13086 * ipa-utils.c (searchc): Likewise.
13087 (ipa_reduced_postorder): Likewise.
13088 * lto-cgraph.c (input_node): Likewise.
13089 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
13090 * symbol-summary.h (symtab_insertion): Likewise.
13091 (symtab_removal): Likewise.
13092 (symtab_duplication): Likewise.
13093 * tree-pretty-print.c (dump_function_header): Likewise.
13094 * tree-sra.c (convert_callers_for_node): Likewise.
13096 2018-06-08 Martin Liska <mliska@suse.cz>
13098 * cgraph.c (symbol_table::create_edge): Always assign a new
13100 (symbol_table::free_edge): Do not recycle numbers.
13101 * cgraph.h (cgraph_edge::get): New method.
13102 * symbol-summary.h (symtab_removal): Use it.
13103 (symtab_duplication): Likewise.
13104 (call_summary::hashable_uid): Remove.
13106 2018-06-08 Martin Liska <mliska@suse.cz>
13108 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
13109 (initialize_growth_caches): Remove.
13110 (free_growth_caches): Likewise.
13111 (do_estimate_edge_time): Use edge_growth_cache.
13112 (do_estimate_edge_size): Likewise.
13113 (do_estimate_edge_hints): Likewise.
13114 * ipa-inline.c (reset_edge_caches): Likewise.
13115 (recursive_inlining): Likewise.
13116 (inline_small_functions): Likewise.
13117 * ipa-inline.h (initialize_growth_caches): Remove.
13118 (estimate_edge_size): Likewise.
13119 (estimate_edge_time): Likewise.
13120 (estimate_edge_hints): Likewise.
13121 (reset_edge_growth_cache): Likewise.
13122 * symbol-summary.h (call_summary::remove): New method.
13124 2018-06-08 Martin Liska <mliska@suse.cz>
13126 * ipa-cp.c (class edge_clone_summary): New summary.
13127 (grow_edge_clone_vectors): Remove.
13128 (ipcp_edge_duplication_hook): Remove.
13129 (class edge_clone_summary_t): New call_summary class.
13130 (ipcp_edge_removal_hook): Remove.
13131 (edge_clone_summary_t::duplicate): New function.
13132 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
13133 (create_specialized_node): Likewise.
13134 (ipcp_driver): Initialize edge_clone_summaries and do not
13137 2018-06-08 Martin Liska <mliska@suse.cz>
13139 * symbol-summary.h (get): New function.
13140 (call_summary::m_initialize_when_cloning): New class member.
13142 2018-06-08 Martin Liska <mliska@suse.cz>
13144 * cgraph.c (cgraph_node::remove): Do not recycle uid.
13145 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
13146 (symbol_table::allocate_cgraph_symbol): Do not set uid.
13147 * passes.c (uid_hash_t): Record removed_nodes by their uids.
13148 (remove_cgraph_node_from_order): Use the removed_nodes set.
13149 (do_per_function_toporder): Likwise.
13150 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
13151 instead of summary_uid.
13152 (symtab_removal): Likewise.
13153 (symtab_duplication): Likewise.
13155 2018-06-08 Martin Liska <mliska@suse.cz>
13157 * ipa-cp.c (ipcp_store_bits_results): Use
13158 ipcp_transformation_sum.
13159 (ipcp_store_vr_results): Likewise.
13160 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
13162 (ipcp_transformation_initialize): ... this.
13163 (ipa_set_node_agg_value_chain):
13164 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
13165 (write_ipcp_transformation_info): Likewise.
13166 (read_ipcp_transformation_info): Likewise.
13167 (ipcp_update_bits): Likewise.
13168 (ipcp_update_vr): Likewise.
13169 (ipcp_transform_function): Likewise.
13170 * ipa-prop.h: Rename ipcp_transformation_summary to
13171 ipcp_transformation.
13172 (class ipcp_transformation_t): New function summary.
13173 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
13174 (ipa_get_agg_replacements_for_node): Likewise.
13176 2018-06-08 Martin Liska <mliska@suse.cz>
13178 * ipa-pure-const.c (struct funct_state_d): Do it class instead
13180 (class funct_state_summary_t): New function_summary class.
13181 (has_function_state): Remove.
13182 (get_function_state): Likewise.
13183 (set_function_state): Likewise.
13184 (add_new_function): Likewise.
13185 (funct_state_summary_t::insert): New function.
13186 (duplicate_node_data): Remove.
13187 (remove_node_data): Remove.
13188 (funct_state_summary_t::duplicate): New function.
13189 (register_hooks): Create new funct_state_summaries.
13190 (pure_const_generate_summary): Use it.
13191 (pure_const_write_summary): Likewise.
13192 (pure_const_read_summary): Likewise.
13193 (propagate_pure_const): Likewise.
13194 (propagate_nothrow): Likewise.
13195 (dump_malloc_lattice): Likewise.
13196 (propagate_malloc): Likewise.
13197 (execute): Do not register hooks, just remove summary
13199 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
13202 2018-06-08 Martin Liska <mliska@suse.cz>
13204 * ipa-reference.c (remove_node_data): Remove.
13205 (duplicate_node_data): Likewise.
13206 (class ipa_ref_var_info_summary_t): New class.
13207 (class ipa_ref_opt_summary_t): Likewise.
13208 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
13209 (get_reference_optimization_summary): Use
13210 ipa_ref_opt_sum_summaries.
13211 (set_reference_vars_info): Remove.
13212 (set_reference_optimization_summary): Likewise.
13213 (ipa_init): Create summaries.
13214 (init_function_info): Use function summary.
13215 (ipa_ref_opt_summary_t::duplicate): New function.
13216 (ipa_ref_opt_summary_t::remove): New function.
13217 (get_read_write_all_from_node): Fix GNU coding style.
13218 (propagate): Use function summary.
13219 (write_node_summary_p): Fix GNU coding style.
13220 (stream_out_bitmap): Likewise.
13221 (ipa_reference_read_optimization_summary): Use function summary.
13222 (ipa_reference_c_finalize): Do not release hooks.
13224 2018-06-08 Martin Liska <mliska@suse.cz>
13226 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
13227 (analyze_function_body): Extract multiple calls of get_create.
13228 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
13229 * ipa-inline.c (recursive_inlining): Use ::get method.
13230 * ipa-inline.h (estimate_edge_growth): Likewise.
13232 2018-06-08 Martin Liska <mliska@suse.cz>
13234 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
13236 (hsa_function_summary::hsa_function_summary): Use the new enum
13238 (hsa_gpu_implementation_p): Use hsa_summaries::get.
13239 * hsa-gen.c (hsa_get_host_function): Likewise.
13240 (get_brig_function_name): Likewise.
13241 * ipa-hsa.c (process_hsa_functions): Likewise.
13242 (ipa_hsa_write_summary): Likewise.
13243 * symbol-summary.h (symtab_duplication): Use ::get function/
13244 (get): New function.
13246 2018-06-08 Martin Liska <mliska@suse.cz>
13248 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
13250 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
13251 (hsa_register_kernel): Likewise.
13252 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
13253 * hsa-gen.c (hsa_get_host_function): Likewise.
13254 (get_brig_function_name): Likewise.
13255 (generate_hsa): Likewise.
13256 (pass_gen_hsail::execute): Likewise.
13257 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
13258 (devirtualization_time_bonus): Likewise.
13259 (ipcp_propagate_stage): Likewise.
13260 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
13261 (edge_set_predicate): Likewise.
13262 (evaluate_conditions_for_known_args): Likewise.
13263 (evaluate_properties_for_edge): Likewise.
13264 (ipa_fn_summary::reset): Likewise.
13265 (ipa_fn_summary_t::duplicate): Likewise.
13266 (dump_ipa_call_summary): Likewise.
13267 (ipa_dump_fn_summary): Likewise.
13268 (analyze_function_body): Likewise.
13269 (compute_fn_summary): Likewise.
13270 (estimate_edge_devirt_benefit): Likewise.
13271 (estimate_edge_size_and_time): Likewise.
13272 (estimate_calls_size_and_time): Likewise.
13273 (estimate_node_size_and_time): Likewise.
13274 (inline_update_callee_summaries): Likewise.
13275 (remap_edge_change_prob): Likewise.
13276 (remap_edge_summaries): Likewise.
13277 (ipa_merge_fn_summary_after_inlining): Likewise.
13278 (ipa_update_overall_fn_summary): Likewise.
13279 (read_ipa_call_summary): Likewise.
13280 (inline_read_section): Likewise.
13281 (write_ipa_call_summary): Likewise.
13282 (ipa_fn_summary_write): Likewise.
13283 (ipa_free_fn_summary): Likewise.
13284 * ipa-hsa.c (process_hsa_functions): Likewise.
13285 (ipa_hsa_write_summary): Likewise.
13286 (ipa_hsa_read_section): Likewise.
13287 * ipa-icf.c (sem_function::merge): Likewise.
13288 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
13289 (do_estimate_edge_time): Likewise.
13290 (estimate_size_after_inlining): Likewise.
13291 (estimate_growth): Likewise.
13292 (growth_likely_positive): Likewise.
13293 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
13294 (inline_call): Likewise.
13295 * ipa-inline.c (caller_growth_limits): Likewise.
13296 (can_inline_edge_p): Likewise.
13297 (can_inline_edge_by_limits_p): Likewise.
13298 (compute_uninlined_call_time): Likewise.
13299 (compute_inlined_call_time): Likewise.
13300 (want_inline_small_function_p): Likewise.
13301 (edge_badness): Likewise.
13302 (update_caller_keys): Likewise.
13303 (update_callee_keys): Likewise.
13304 (recursive_inlining): Likewise.
13305 (inline_small_functions): Likewise.
13306 (inline_to_all_callers_1): Likewise.
13307 (dump_overall_stats): Likewise.
13308 (early_inline_small_functions): Likewise.
13309 (early_inliner): Likewise.
13310 * ipa-inline.h (estimate_edge_growth): Likewise.
13311 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
13312 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
13313 * ipa-prop.h (IPA_NODE_REF): Likewise.
13314 (IPA_EDGE_REF): Likewise.
13315 * ipa-pure-const.c (malloc_candidate_p): Likewise.
13316 (propagate_malloc): Likewise.
13317 * ipa-split.c (execute_split_functions): Likewise.
13318 * symbol-summary.h: Rename get to get_create.
13320 (get_create): Likewise.
13321 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
13323 2018-06-08 Martin Liska <mliska@suse.cz>
13325 * symbol-summary.h (release): Move definition out of class
13327 (symtab_removal): Likewise.
13328 (symtab_duplication): Likewise.
13330 2018-06-08 Martin Liska <mliska@suse.cz>
13332 * symbol-summary.h (function_summary): Move constructor
13333 implementation out of class declaration.
13334 (release): Likewise.
13335 (symtab_insertion): Likewise.
13336 (symtab_removal): Likewise.
13337 (symtab_duplication): Likewise.
13340 2018-06-08 Martin Liska <mliska@suse.cz>
13342 * Makefile.in: Remove support for MPX (macros, related functions,
13343 fields in cgraph_node, ...).
13344 * builtin-types.def (BT_BND): Likewise.
13345 (BT_FN_BND_CONST_PTR): Likewise.
13346 (BT_FN_CONST_PTR_BND): Likewise.
13347 (BT_FN_VOID_PTR_BND): Likewise.
13348 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
13349 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
13350 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
13351 (expand_builtin_mempcpy_with_bounds): Likewise.
13352 (expand_builtin_memset_with_bounds): Likewise.
13353 (expand_builtin_memset_args): Likewise.
13354 (std_expand_builtin_va_start): Likewise.
13355 (expand_builtin): Likewise.
13356 (expand_builtin_with_bounds): Likewise.
13357 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
13358 (DEF_LIB_BUILTIN_CHKP): Likewise.
13359 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
13360 (DEF_CHKP_BUILTIN): Likewise.
13361 (BUILT_IN_MEMCPY): Likewise.
13362 (BUILT_IN_MEMMOVE): Likewise.
13363 (BUILT_IN_MEMPCPY): Likewise.
13364 (BUILT_IN_MEMSET): Likewise.
13365 (BUILT_IN_STPCPY): Likewise.
13366 (BUILT_IN_STRCAT): Likewise.
13367 (BUILT_IN_STRCHR): Likewise.
13368 (BUILT_IN_STRCPY): Likewise.
13369 (BUILT_IN_STRLEN): Likewise.
13370 (BUILT_IN_MEMCPY_CHK): Likewise.
13371 (BUILT_IN_MEMMOVE_CHK): Likewise.
13372 (BUILT_IN_MEMPCPY_CHK): Likewise.
13373 (BUILT_IN_MEMSET_CHK): Likewise.
13374 (BUILT_IN_STPCPY_CHK): Likewise.
13375 (BUILT_IN_STRCAT_CHK): Likewise.
13376 (BUILT_IN_STRCPY_CHK): Likewise.
13377 * calls.c (store_bounds): Likewise.
13378 (emit_call_1): Likewise.
13379 (special_function_p): Likewise.
13380 (maybe_warn_nonstring_arg): Likewise.
13381 (initialize_argument_information): Likewise.
13382 (finalize_must_preallocate): Likewise.
13383 (compute_argument_addresses): Likewise.
13384 (expand_call): Likewise.
13385 * cfgexpand.c (expand_call_stmt): Likewise.
13386 (expand_return): Likewise.
13387 (expand_gimple_stmt_1): Likewise.
13388 (pass_expand::execute): Likewise.
13389 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
13390 (cgraph_node::remove): Likewise.
13391 (cgraph_node::dump): Likewise.
13392 (cgraph_node::verify_node): Likewise.
13393 * cgraph.h (chkp_function_instrumented_p): Likewise.
13394 (symtab_node::get_alias_target): Likewise.
13395 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
13396 (cgraph_local_p): Likewise.
13397 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
13398 (cgraph_edge::rebuild_references): Likewise.
13399 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
13400 (walk_polymorphic_call_targets): Likewise.
13401 (cgraph_node::expand_thunk): Likewise.
13402 (symbol_table::output_weakrefs): Likewise.
13403 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
13404 (ix86_handle_option): Likewise.
13405 * config/i386/constraints.md: Likewise.
13406 * config/i386/i386-builtin-types.def (BND): Likewise.
13410 * config/i386/i386-builtin.def (BDESC_END): Likewise.
13411 (BDESC_FIRST): Likewise.
13413 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
13414 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
13415 * config/i386/i386.c (enum reg_class): Likewise.
13416 (ix86_target_string): Likewise.
13417 (ix86_option_override_internal): Likewise.
13418 (ix86_conditional_register_usage): Likewise.
13419 (ix86_valid_target_attribute_inner_p): Likewise.
13420 (ix86_set_indirect_branch_type): Likewise.
13421 (ix86_set_current_function): Likewise.
13422 (ix86_function_arg_regno_p): Likewise.
13423 (init_cumulative_args): Likewise.
13424 (ix86_function_arg_advance): Likewise.
13425 (ix86_function_arg): Likewise.
13426 (ix86_pass_by_reference): Likewise.
13427 (ix86_function_value_regno_p): Likewise.
13428 (ix86_function_value_1): Likewise.
13429 (ix86_function_value_bounds): Likewise.
13430 (ix86_return_in_memory): Likewise.
13431 (ix86_setup_incoming_vararg_bounds): Likewise.
13432 (ix86_va_start): Likewise.
13433 (indirect_thunk_need_prefix): Likewise.
13434 (print_reg): Likewise.
13435 (ix86_print_operand): Likewise.
13436 (ix86_expand_call): Likewise.
13437 (ix86_output_function_return): Likewise.
13438 (reg_encoded_number): Likewise.
13439 (BDESC_VERIFYS): Likewise.
13440 (ix86_init_mpx_builtins): Likewise.
13441 (ix86_init_builtins): Likewise.
13442 (ix86_emit_cmove): Likewise.
13443 (ix86_emit_move_max): Likewise.
13444 (ix86_expand_builtin): Likewise.
13445 (ix86_builtin_mpx_function): Likewise.
13446 (ix86_get_arg_address_for_bt): Likewise.
13447 (ix86_load_bounds): Likewise.
13448 (ix86_store_bounds): Likewise.
13449 (ix86_load_returned_bounds): Likewise.
13450 (ix86_store_returned_bounds): Likewise.
13451 (ix86_class_likely_spilled_p): Likewise.
13452 (ix86_hard_regno_mode_ok): Likewise.
13453 (x86_order_regs_for_local_alloc): Likewise.
13454 (ix86_mitigate_rop): Likewise.
13455 (ix86_bnd_prefixed_insn_p): Likewise.
13456 (ix86_mpx_bound_mode): Likewise.
13457 (ix86_make_bounds_constant): Likewise.
13458 (ix86_initialize_bounds): Likewise.
13459 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
13460 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
13461 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
13462 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
13463 (TARGET_CHKP_BOUND_MODE): Likewise.
13464 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
13465 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
13466 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
13467 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
13468 * config/i386/i386.h (TARGET_MPX): Likewise.
13469 (TARGET_MPX_P): Likewise.
13470 (VALID_BND_REG_MODE): Likewise.
13471 (FIRST_BND_REG): Likewise.
13472 (LAST_BND_REG): Likewise.
13473 (enum reg_class): Likewise.
13474 (BND_REG_P): Likewise.
13475 (BND_REGNO_P): Likewise.
13476 (BNDmode): Likewise.
13477 (ADJUST_INSN_LENGTH): Likewise.
13478 * config/i386/i386.md: Likewise.
13479 * config/i386/i386.opt: Likewise.
13480 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
13481 (defined): Likewise.
13482 (LINK_MPX): Likewise.
13483 (MPX_SPEC): Likewise.
13484 (LIBMPX_SPEC): Likewise.
13485 (LIBMPXWRAPPERS_SPEC): Likewise.
13486 (CHKP_SPEC): Likewise.
13487 * config/i386/predicates.md: Likewise.
13488 * dbxout.c (dbxout_type): Likewise.
13489 * doc/extend.texi: Likewise.
13490 * doc/invoke.texi: Likewise.
13491 * doc/md.texi: Likewise.
13492 * doc/tm.texi: Likewise.
13493 * doc/tm.texi.in: Likewise.
13494 * dwarf2out.c (is_base_type): Likewise.
13495 (gen_formal_types_die): Likewise.
13496 (gen_subprogram_die): Likewise.
13497 (gen_type_die_with_usage): Likewise.
13498 (gen_decl_die): Likewise.
13499 (dwarf2out_late_global_decl): Likewise.
13500 * expr.c (expand_assignment): Likewise.
13501 (emit_storent_insn): Likewise.
13502 (store_expr_with_bounds): Likewise.
13503 (store_expr): Likewise.
13504 (expand_expr_real_1): Likewise.
13505 * expr.h (store_expr_with_bounds): Likewise.
13506 * function.c (use_register_for_decl): Likewise.
13507 (struct bounds_parm_data): Likewise.
13508 (assign_parms_augmented_arg_list): Likewise.
13509 (assign_parm_find_entry_rtl): Likewise.
13510 (assign_parm_is_stack_parm): Likewise.
13511 (assign_parm_load_bounds): Likewise.
13512 (assign_bounds): Likewise.
13513 (assign_parms): Likewise.
13514 (expand_function_start): Likewise.
13515 * gcc.c (CHKP_SPEC): Likewise.
13516 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
13517 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
13518 (wrestrict_dom_walker::check_call): Likewise.
13519 * gimple.c (gimple_build_call_from_tree): Likewise.
13520 * gimple.h (enum gf_mask): Likewise.
13521 (gimple_call_with_bounds_p): Likewise.
13522 (gimple_call_set_with_bounds): Likewise.
13523 * gimplify.c (gimplify_init_constructor): Likewise.
13524 * ipa-cp.c (initialize_node_lattices): Likewise.
13525 (propagate_constants_across_call): Likewise.
13526 (find_more_scalar_values_for_callers_subset): Likewise.
13527 * ipa-hsa.c (process_hsa_functions): Likewise.
13528 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
13529 * ipa-icf.c (sem_function::merge): Likewise.
13530 * ipa-inline.c (early_inliner): Likewise.
13531 * ipa-pure-const.c (warn_function_noreturn): Likewise.
13532 (warn_function_cold): Likewise.
13533 (propagate_pure_const): Likewise.
13534 * ipa-ref.h (enum GTY): Likewise.
13535 * ipa-split.c (find_retbnd): Likewise.
13536 (consider_split): Likewise.
13537 (split_function): Likewise.
13538 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
13539 * ipa.c (walk_polymorphic_call_targets): Likewise.
13540 (symbol_table::remove_unreachable_nodes): Likewise.
13541 (process_references): Likewise.
13542 (cgraph_build_static_cdtor_1): Likewise.
13543 * lto-cgraph.c (lto_output_node): Likewise.
13544 (output_refs): Likewise.
13545 (compute_ltrans_boundary): Likewise.
13546 (input_overwrite_node): Likewise.
13547 (input_node): Likewise.
13548 (input_cgraph_1): Likewise.
13549 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
13550 * passes.c (pass_manager::execute_early_local_passes): Likewise.
13551 (class pass_chkp_instrumentation_passes): Likewise.
13552 (make_pass_chkp_instrumentation_passes): Likewise.
13553 * passes.def: Likewise.
13554 * rtl.h (struct GTY): Likewise.
13555 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
13556 * stor-layout.c (layout_type): Likewise.
13557 * symtab.c: Likewise.
13558 * target.def: Likewise.
13559 * targhooks.c (default_chkp_bound_type): Likewise.
13560 (default_chkp_bound_mode): Likewise.
13561 (default_builtin_chkp_function): Likewise.
13562 (default_chkp_function_value_bounds): Likewise.
13563 (default_chkp_make_bounds_constant): Likewise.
13564 (default_chkp_initialize_bounds): Likewise.
13565 * targhooks.h (default_chkp_bound_type): Likewise.
13566 (default_chkp_bound_mode): Likewise.
13567 (default_builtin_chkp_function): Likewise.
13568 (default_chkp_function_value_bounds): Likewise.
13569 (default_chkp_make_bounds_constant): Likewise.
13570 (default_chkp_initialize_bounds): Likewise.
13571 * toplev.c (compile_file): Likewise.
13572 (process_options): Likewise.
13573 * tree-core.h (DEF_BUILTIN): Likewise.
13574 (DEF_BUILTIN_CHKP): Likewise.
13575 * tree-inline.c (declare_return_variable): Likewise.
13576 (remap_gimple_stmt): Likewise.
13577 (copy_bb): Likewise.
13578 (initialize_inlined_parameters): Likewise.
13579 (expand_call_inline): Likewise.
13580 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
13581 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
13582 (make_pass_ipa_chkp_produce_thunks): Likewise.
13583 (make_pass_chkp): Likewise.
13584 (make_pass_chkp_opt): Likewise.
13585 (make_pass_chkp_instrumentation_passes): Likewise.
13586 * tree-pretty-print.c (dump_generic_node): Likewise.
13587 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
13588 * tree-ssa-dce.c (propagate_necessity): Likewise.
13589 (eliminate_unnecessary_stmts): Likewise.
13590 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
13591 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
13592 * tree-ssa-sccvn.h: Likewise.
13593 * tree-ssa-strlen.c (get_string_length): Likewise.
13594 (valid_builtin_call): Likewise.
13595 (adjust_last_stmt): Likewise.
13596 (handle_builtin_strchr): Likewise.
13597 (handle_builtin_strcpy): Likewise.
13598 (handle_builtin_stxncpy): Likewise.
13599 (handle_builtin_memcpy): Likewise.
13600 (handle_builtin_strcat): Likewise.
13601 (strlen_check_and_optimize_stmt): Likewise.
13602 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
13603 * tree-streamer-in.c: Likewise.
13604 * tree-streamer.c (record_common_node): Likewise.
13605 * tree.c (tree_code_size): Likewise.
13606 (wide_int_to_tree_1): Likewise.
13607 (type_contains_placeholder_1): Likewise.
13608 (build_common_tree_nodes): Likewise.
13609 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
13610 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
13611 (POINTER_BOUNDS_P): Likewise.
13612 (BOUNDED_TYPE_P): Likewise.
13613 (BOUNDED_P): Likewise.
13614 (CALL_WITH_BOUNDS_P): Likewise.
13615 (pointer_bounds_type_node): Likewise.
13616 * value-prof.c (gimple_ic): Likewise.
13617 * var-tracking.c (vt_add_function_parameters): Likewise.
13618 * varasm.c (make_decl_rtl): Likewise.
13619 (assemble_start_function): Likewise.
13620 (output_constant): Likewise.
13621 (maybe_assemble_visibility): Likewise.
13622 * varpool.c (ctor_for_folding): Likewise.
13623 * chkp-builtins.def: Remove.
13624 * ipa-chkp.c: Remove.
13625 * ipa-chkp.h: Remove.
13626 * rtl-chkp.c: Remove.
13627 * rtl-chkp.h: Remove.
13628 * tree-chkp-opt.c: Remove.
13629 * tree-chkp.c: Remove.
13630 * tree-chkp.h: Remove.
13632 2018-06-07 Carl Love <cel@us.ibm.com>
13634 * config/rs6000/vsx.md (vextract_fp_from_shorth,
13635 vextract_fp_from_shortl): Add BE support.
13637 2018-06-07 Paul Koning <ni1d@arrl.net>
13639 * compare-elim.c (try_merge_compare): Don't merge compare if
13640 address contains a side effect.
13641 (try_eliminate_compare): Likewise.
13643 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
13645 * config.gcc: Support "tremont".
13646 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
13647 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13649 * config/i386/i386.c (m_TREMONT): Define.
13650 (processor_target_table): Add "tremont".
13651 (PTA_TREMONT): Define.
13652 (ix86_lea_outperforms): Add TARGET_TREMONT.
13653 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
13654 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
13655 and M_INTEL_GOLDMONT_PLUS.
13656 (fold_builtin_cpu): Add "tremont".
13657 (ix86_add_stmt_cost): Add TARGET_TREMONT.
13658 (ix86_option_override_internal): Add "tremont".
13659 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
13660 (processor_type): Add PROCESSOR_TREMONT.
13661 * config/i386/x86-tune.def: Add m_TREMONT.
13662 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
13664 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
13666 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
13667 symbol defined for msp430i* devices to be lower case.
13669 2018-06-07 Richard Biener <rguenther@suse.de>
13671 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
13672 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
13673 Properly wrap signed arithmetic if overflow wraps.
13675 2018-06-07 Jakub Jelinek <jakub@redhat.com>
13677 PR tree-optimization/69615
13678 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
13679 of a cast from a same precision integral SSA_NAME in a bb dominated
13680 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
13681 cast to utype if rhs2 has already a compatible type.
13683 2018-06-07 Richard Biener <rguenther@suse.de>
13685 PR tree-optimization/85935
13686 * graphite-scop-detection.c (find_params_in_bb): Analyze
13687 condition operands with respect to the correct loop. Assert
13688 the analysis doesn't fail.
13690 2018-06-04 Carl Love <cel@us.ibm.com>
13692 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
13693 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
13694 as it is slightly cheaper.
13695 (first_match_or_eos_index_<mode>):
13696 Calculate index using natural element order.
13697 (first_match_index_<mode>):
13698 Calculate index using natural element order.
13699 (first_match_or_eos_index_<mode>):
13700 Calculate index using natural order.
13701 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
13702 for BE and LE modes.
13703 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
13704 P9V_BUILTIN_VCLZLSBB_V16QI.
13705 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
13708 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
13710 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
13711 indentation and line wrap for many prototypes. Add missing
13712 @smallexample directives around block of prototypes for vec_xl and
13715 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
13717 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
13718 track if we pass or return IEEE 128-bit floating point.
13719 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
13720 C++ mangling that is compatible with GCC 8.1.
13721 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
13722 (init_cumulative_args): Note if we pass or return IEEE 128-bit
13723 floating point types.
13724 (rs6000_function_arg_advance_1): Likewise.
13725 (rs6000_mangle_type): Optionally generate mangled names that match
13726 what GCC 8.1 generated for IEEE 128-bit floating point types.
13727 (rs6000_globalize_decl_name): If we have an external function that
13728 passes or returns IEEE 128-bit types, generate a weak reference
13729 from the mangled name used in GCC 8.1 to the current mangled
13731 (rs6000_init_builtins): Make __ibm128 use the long double type if
13732 long double is IBM extended double. Make __float128 use the long
13733 double type if long double is IEEE 128-bit.
13736 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
13737 macro for __ibm128 built-in functions.
13738 (PACK_IF): Add __ibm128 pack/unpack functions.
13739 (UNPACK_IF): Likewise.
13740 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
13741 enable long double built-in functions if long double is IEEE
13742 128-bit floating point.
13743 (rs6000_invalid_builtin): Update long double built-in function
13745 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
13746 functions, adjust the built-in function to use the long double
13747 built-in function if __ibm128 and long double are the same type.
13748 * doc/extend.texi (PowerPC builtins): Update documention for
13749 __builtin_{,un}pack_longdouble. Add documentation for
13750 __builtin_{,un}pack_ibm128.
13752 2018-06-06 Jim Wilson <jimw@sifive.com>
13754 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
13755 (struct machine_function): New field interrupt_mode.
13756 (riscv_handle_type_attribute): New function. Add forward declaration.
13757 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
13758 (riscv_expand_epilogue): Check interrupt_mode field.
13759 (riscv_set_current_function): Check interrupt attribute args and
13760 set interrupt_mode field.
13761 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
13762 (riscv_sret, riscv_uret): New.
13763 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
13764 new arguments to interrupt attribute.
13766 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
13769 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
13770 Don't handle -mcpu=power8 if -mpower9-vector is also used.
13772 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
13774 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
13775 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
13776 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
13777 several redundant entries.
13779 2018-06-06 David Malcolm <dmalcolm@redhat.com>
13781 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
13782 type from "rtx" to "rtx_insn *".
13783 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
13784 for local "call_insn", removing cast.
13785 (ix86_expand_call): Likewise, introducing a "call_insn" local.
13787 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
13789 PR tree-optimization/86066
13790 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
13791 for BIT_INSERT_EXPR stores.
13793 2018-06-06 Richard Biener <rguenther@suse.de>
13795 PR tree-optimization/86062
13796 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
13797 component refs ontop
13798 of to be offsetted base.
13800 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
13802 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
13803 to be static and remove check on interrupt attribute name.
13805 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
13807 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
13808 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
13810 2018-06-05 Steve Ellcey <sellcey@cavium.com>
13813 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
13815 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
13816 Remove second argument, change how error is called.
13817 (aarch64_layout_arg): Remove second argument from
13818 aarch64_err_no_fpadvsimd call.
13819 (aarch64_init_cumulative_args): Ditto.
13820 (aarch64_gimplify_va_arg_expr): Ditto.
13821 * config/aarch64/aarch64.md (mov<mode>): Ditto.
13823 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
13825 * config/i386/i386.md (simple_return_indirect_internal): New expander.
13826 (*simple_return_indirect_internal<mode>): Rename from
13827 simple_return_indirect_internal. Use W mode iterator.
13828 (rstorssp): New expander.
13829 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
13830 (clrssbsy): New expander.
13831 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
13833 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
13835 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
13837 (cmse_check_pointed_object): Likewise.
13839 2018-06-05 Martin Liska <mliska@suse.cz>
13841 PR gcov-profile/47618
13842 * doc/invoke.texi: Document how -fprofile-dir format
13845 2018-06-05 Richard Biener <rguenther@suse.de>
13847 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
13848 removal pretend DOM info isn't available so we do not update
13849 it and only remove edges, not dominated blocks. Actually free
13850 DOM info in case we removed something. Remove unreachable blocks.
13851 (mfb_keep_latches): Work with either DOM info or marked backedges.
13852 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
13853 first. Mark backedges if DOM info isn't available.
13854 (Re-)compute DOM info after cleanup_control_flow_pre.
13856 2018-06-05 Richard Biener <rguenther@suse.de>
13858 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
13859 (locus_discrim_hasher::hash): Adjust.
13860 (locus_discrim_hasher::equal): Likewise.
13861 (next_discriminator_for_locus): Work on line directly.
13862 (same_line_p): Pass in expanded locus1 as well.
13863 (assign_discriminators): Avoid redundant location expansions.
13865 2018-06-05 Richard Biener <rguenther@suse.de>
13867 PR tree-optimization/86046
13868 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
13869 if required after clearing TREE_ADDRESSABLE.
13871 2018-06-05 Richard Biener <rguenther@suse.de>
13873 PR tree-optimization/86047
13874 * tree-ssa-loop.c (for_each_index): Glob handling of all
13875 decls and constants and really handle all of them.
13877 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13880 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
13881 qualifier_void_pointer and qualifier_const_void_pointer.
13882 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
13883 (arm_init_builtins): Handle the above.
13884 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
13885 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
13886 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
13889 2018-06-05 Martin Liska <mliska@suse.cz>
13891 * auto-profile.c (read_autofdo_file): Do not use
13892 gcov_ctr_summary struct.
13893 (afdo_callsite_hot_enough_for_early_inline): Likewise.
13894 * coverage.c (struct counts_entry): Likewise.
13895 (read_counts_file): Read just single summary entry.
13896 (get_coverage_counts): Use gcov_summary struct.
13897 * coverage.h (get_coverage_counts): Likewise.
13898 * gcov-dump.c (dump_working_sets): Likewise.
13899 (tag_summary): Dump just single summary.
13900 * gcov-io.c (gcov_write_summary): Write just histogram
13902 (gcov_read_summary): Read just single summary.
13903 (compute_working_sets): Use gcov_summary struct.
13904 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
13905 of GCOV_COUNTERS_SUMMABLE.
13906 (GCOV_COUNTERS_SUMMABLE): Remove.
13907 (GCOV_FIRST_VALUE_COUNTER): Replace with
13908 GCOV_COUNTER_V_INTERVAL.
13909 (struct gcov_ctr_summary): Remove.
13910 (struct gcov_summary): Directly use fields of former
13912 (compute_working_sets): Use gcov_summary struct.
13913 * gcov.c (read_count_file): Do not use ctrs fields.
13914 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
13916 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
13918 * profile.c: Likewise.
13919 * profile.h: Likewise.
13921 2018-06-05 Martin Liska <mliska@suse.cz>
13923 PR gcov-profile/84846
13924 * gcov.c (output_lines): Print working directory only
13925 in intermediate format.
13927 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
13929 * config/s390/s390-builtin-types.def: Add void function type.
13930 * config/s390/s390-builtins.def: Use the function type for the
13933 2018-06-04 Jim Wilson <jimw@sifive.com>
13935 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
13937 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
13938 and maybe_eh_return. Change regno to unsigned int. Use new args to
13939 handle EH_RETURN_DATA_REGNO registers properly.
13940 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
13941 (riscv_expand_epilogue): Update comment. Change argument name and
13942 type. Update code to use new name and type. Pass new args to
13943 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
13945 * config/riscv/riscv.md (NORMAL_RETURN): New.
13946 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
13947 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
13948 (eh_return): Call gen_eh_return_internal and emit barrier.
13949 (eh_return_internal): Call riscv_expand_epilogue.
13951 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
13953 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
13954 bit_insertion field and declare can_be_merged_into method.
13955 (merged_store_group::can_be_merged_into): New method.
13956 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
13957 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
13958 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
13960 2018-06-04 Richard Biener <rguenther@suse.de>
13962 PR tree-optimization/85955
13963 * builtins.c (fold_builtin_sincos): Convert pointers to
13964 destination to appropriate type before dereferencing.
13966 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
13968 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
13970 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
13972 * expr.c (expand_expr_real_1): Force the operand into memory if
13973 its TYPE_MODE is BLKmode and if there is no integer mode for
13974 the number of bits being extracted.
13976 2018-06-04 Jakub Jelinek <jakub@redhat.com>
13980 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
13981 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
13983 2018-06-04 Richard Biener <rguenther@suse.de>
13985 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
13986 (cleanup_tree_cfg_noloop): ... single caller. Do
13987 start_recording_case_labels later.
13989 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
13991 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
13992 to _IMMINTRIN_H_INCLUDED.
13993 * config/i386/pconfigintrin.h: Ditto.
13994 * config/i386/waitpkgintrin.h: Ditto.
13995 * config/i386/immintrin.h: Add includes for sgxintrin.h,
13996 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
13997 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
13998 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
13999 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
14000 waitpkgintrin.h and cldemoteintrin.h.
14002 2018-06-04 Richard Biener <rguenther@suse.de>
14004 PR tree-optimization/86038
14005 * tracer.c (find_best_successor): Check probability for
14006 being initialized, bail out if not.
14008 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
14011 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
14012 of bits to ignore when comparing architectures.
14014 2018-06-04 Jakub Jelinek <jakub@redhat.com>
14016 PR tree-optimization/69615
14017 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
14018 maximum or minimum of the type, try to merge it also as if
14019 range1 is + [-, x - 1] or + [x + 1, -].
14022 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
14024 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
14026 PR tree-optimization/86034
14027 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
14028 the unsigned bitfield type in a bit insertion sequence if it does not
14029 have a larger precision than the bitfield size.
14030 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
14032 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
14034 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
14036 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
14038 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
14039 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
14040 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
14041 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
14043 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
14045 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
14046 Disable -fdelete-null-pointer-checks for ELF toolchain.
14048 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
14049 Kito Cheng <kito.cheng@gmail.com>
14051 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
14052 (nds32le-*-*, nds32be-*-*): Integrate checking process.
14053 (nds32*-*-*): Add glibc and uclibc conditions.
14054 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
14055 (TARGET_EXCEPT_UNWIND_INFO): Define.
14056 * config/nds32/elf.h: New file.
14057 * config/nds32/linux.h: New file.
14058 * config/nds32/nds32-elf.opt: New file.
14059 * config/nds32/nds32-linux.opt: New file.
14060 * config/nds32/nds32-fp-as-gp.c
14061 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
14062 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
14064 (nds32_asm_file_end): Ditto.
14065 (nds32_print_operand): Ditto.
14066 (nds32_insert_attributes): Ditto.
14067 (nds32_init_libfuncs): New function.
14068 (TARGET_HAVE_TLS): Define.
14069 (TARGET_INIT_LIBFUNCS): Define.
14070 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
14072 (TARGET_ELF): Apply different mcmodel setting.
14073 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
14074 been migrated into elf.h and linux.h files.
14075 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
14076 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
14077 (mcmodel): The content has been migrated into nds32-elf.opt and
14078 nds32-linux.opt files.
14079 * config/nds32/t-elf: New file.
14080 * config/nds32/t-linux: New file.
14082 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
14083 Shiva Chen <shiva0217@gmail.com>
14085 * config/nds32/constants.md (unspec_volatile_element): Add
14086 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
14087 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
14089 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
14090 (make_pass_nds32_fp_as_gp): Declare.
14091 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
14093 (nds32_asm_function_end_prologue): Remove unused asm output.
14094 (nds32_asm_function_begin_epilogue): Remove unused asm output.
14095 (nds32_asm_file_start): Output necessary fp_as_gp information.
14096 (nds32_option_override): Adjust register usage.
14097 (nds32_expand_prologue): Consider fp_as_gp situation.
14098 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
14099 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
14102 (simple_return): Ditto.
14103 (omit_fp_begin): Output special directive for fp_as_gp.
14104 (omit_fp_end): Output special directive for fp_as_gp.
14105 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
14106 mforbid-fp-as-gp): New options.
14108 2018-06-01 Mark Wielaard <mark@klomp.org>
14110 * dwarf2out.c (dwarf2out_finish): Remove generation of
14111 DW_AT_loclists_base.
14113 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
14115 * gimple-ssa-store-merging.c: Include gimple-fold.h.
14116 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
14117 (struct merged_store_group): Add bit_insertion field.
14118 (dump_char_array): Use standard hexadecimal format.
14119 (merged_store_group::merged_store_group): Set bit_insertion to false.
14120 (merged_store_group::apply_stores): Use optimal buffer size. Deal
14121 with BIT_INSERT_EXPR stores. Move up code updating the mask and
14122 also print the mask in the dump file.
14123 (pass_store_merging::gate): Minor tweak.
14124 (imm_store_chain_info::coalesce_immediate): Fix wrong association
14125 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
14126 stores with INTEGER_CST stores.
14127 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
14128 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
14129 and use it throughout. Generate bit insertion sequences if need be.
14130 (pass_store_merging::process_store): Remove redundant condition.
14131 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
14133 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
14135 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
14136 the 128-bit floating point types. Fix function comment.
14138 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14140 * config/aarch64/aarch64-simd.md
14141 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
14143 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
14146 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
14148 PR tree-optimization/85989
14149 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
14151 (backprop::intersect_uses): Check it when deciding whether this
14152 is a backedge reference.
14153 (backprop::process_block): Add each phi to m_visited_phis
14154 after visiting it, then clear it at the end.
14156 2018-06-01 Richard Biener <rguenther@suse.de>
14158 * tree-vectorizer.h (vect_dr_stmt): New function.
14159 (vect_get_load_cost): Adjust.
14160 (vect_get_store_cost): Likewise.
14161 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
14162 Use vect_dr_stmt instead of DR_SMTT.
14163 (vect_record_base_alignments): Likewise.
14164 (vect_calculate_target_alignment): Likewise.
14165 (vect_compute_data_ref_alignment): Likewise and make static.
14166 (vect_update_misalignment_for_peel): Likewise.
14167 (vect_verify_datarefs_alignment): Likewise.
14168 (vector_alignment_reachable_p): Likewise.
14169 (vect_get_data_access_cost): Likewise. Pass down
14170 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
14171 (vect_get_peeling_costs_all_drs): Likewise.
14172 (vect_peeling_hash_get_lowest_cost): Likewise.
14173 (vect_enhance_data_refs_alignment): Likewise.
14174 (vect_find_same_alignment_drs): Likewise.
14175 (vect_analyze_data_refs_alignment): Likewise.
14176 (vect_analyze_group_access_1): Likewise.
14177 (vect_analyze_group_access): Likewise.
14178 (vect_analyze_data_ref_access): Likewise.
14179 (vect_analyze_data_ref_accesses): Likewise.
14180 (vect_vfa_segment_size): Likewise.
14181 (vect_small_gap_p): Likewise.
14182 (vectorizable_with_step_bound_p): Likewise.
14183 (vect_prune_runtime_alias_test_list): Likewise.
14184 (vect_analyze_data_refs): Likewise.
14185 (vect_supportable_dr_alignment): Likewise.
14186 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
14187 (vect_gen_prolog_loop_niters): Likewise.
14188 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
14189 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
14191 (vect_recog_mask_conversion_pattern): Likewise.
14192 (vect_try_gather_scatter_pattern): Likewise.
14193 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
14194 to vect_get_store_cost.
14195 (vect_get_store_cost): Get stmt_info instead of DR.
14196 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
14197 (vect_get_load_cost): Get stmt_info instead of DR.
14199 2018-06-01 Richard Biener <rguenther@suse.de>
14201 PR middle-end/86017
14202 * gimple-fold.c (var_decl_component_p): Also allow offsetted
14203 vars wrapped in MEM_REFs.
14205 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
14207 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
14208 Fix subreg tests so that we only return a choice between
14209 GENERAL_REGS and FP_REGS if the original classes included both.
14211 2018-06-01 Richard Biener <rguenther@suse.de>
14214 * tree-ssa-structalias.c (get_function_part_constraint):
14215 Handle NULL fi->decl.
14216 (find_func_aliases_for_call): Properly handle indirect
14217 fi from direct call.
14218 (find_func_clobbers): Likewise.
14219 (ipa_pta_execute): Likewise.
14220 (create_variable_info_for): For functions that are ifunc_resolver
14221 resolve to a varinfo that contains the result of the resolver call.
14222 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
14225 2018-05-31 Michael Collison <michael.collison@arm.com>
14227 * config/aarch64/aarch64.md:
14228 (*fix_to_zero_extenddfdi2): New pattern.
14229 * gcc.target/aarch64/fix_extend1.c: New testcase.
14231 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
14233 PR middle-end/78809
14234 PR middle-end/83026
14235 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
14236 and BUILT_IN_STRNCMP_EQ.
14237 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
14238 BUILT_IN_STRNCMP_EQ.
14239 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
14240 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
14241 (gimple_fold_builtin): Likewise.
14242 * tree-ssa-strlen.c (compute_string_length): New function.
14243 (determine_min_obsize): New function.
14244 (handle_builtin_string_cmp): New function to handle calls to
14245 string compare functions.
14246 (strlen_optimize_stmt): Add handling to builtin string compare
14248 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
14249 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
14250 * tree.c (build_common_builtin_nodes): Add new defines of
14251 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
14253 2018-05-31 Jakub Jelinek <jakub@redhat.com>
14256 * bb-reorder.c (pass_partition_blocks::gate): Return false for
14257 functions with naked attribute.
14259 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
14261 * config/i386/sse.md (avx_vec_concat<mode>):
14262 Substitute concat_tg_mode mode attribute with xtg_mode.
14263 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
14264 (concat_tg_mode): Remove mode attribute.
14266 2018-05-31 Martin Sebor <msebor@redhat.com>
14269 * calls.c (alloc_max_size): Correct a logic error/typo.
14270 Treat excessive arguments as infinite. Warn for invalid arguments.
14271 * doc/invoke.texi (-Walloc-size-larger-than): Update.
14273 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
14276 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
14277 and movx for Haswell.
14279 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
14280 Cesar Philippidis <cesar@codesourcery.com>
14282 PR middle-end/85879
14283 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
14284 when emitting error on private/firstprivate reductions.
14285 * omp-low.c (lower_omp_target): Avoid reference-type processing
14286 on pointers for firstprivate clause.
14288 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
14290 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
14293 * config/aarch64/aarch64-simd.md
14294 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
14295 (aarch64_ld1_x3_<mode>): Likewise
14296 (aarch64_st1x2<VALLDIF:mode>): Likewise
14297 (aarch64_st1_x2_<mode>): Likewise
14298 (aarch64_st1x3<VALLDIF:mode>): Likewise
14299 (aarch64_st1_x3_<mode>): Likewise
14300 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
14301 (vld1_s8_x3): Likewise.
14302 (vld1_u16_x3): Likewise.
14303 (vld1_s16_x3): Likewise.
14304 (vld1_u32_x3): Likewise.
14305 (vld1_s32_x3): Likewise.
14306 (vld1_u64_x3): Likewise.
14307 (vld1_s64_x3): Likewise.
14308 (vld1_f16_x3): Likewise.
14309 (vld1_f32_x3): Likewise.
14310 (vld1_f64_x3): Likewise.
14311 (vld1_p8_x3): Likewise.
14312 (vld1_p16_x3): Likewise.
14313 (vld1_p64_x3): Likewise.
14314 (vld1q_u8_x3): Likewise.
14315 (vld1q_s8_x3): Likewise.
14316 (vld1q_u16_x3): Likewise.
14317 (vld1q_s16_x3): Likewise.
14318 (vld1q_u32_x3): Likewise.
14319 (vld1q_s32_x3): Likewise.
14320 (vld1q_u64_x3): Likewise.
14321 (vld1q_s64_x3): Likewise.
14322 (vld1q_f16_x3): Likewise.
14323 (vld1q_f32_x3): Likewise.
14324 (vld1q_f64_x3): Likewise.
14325 (vld1q_p8_x3): Likewise.
14326 (vld1q_p16_x3): Likewise.
14327 (vld1q_p64_x3): Likewise.
14328 (vst1_s64_x2): Likewise.
14329 (vst1_u64_x2): Likewise.
14330 (vst1_f64_x2): Likewise.
14331 (vst1_s8_x2): Likewise.
14332 (vst1_p8_x2): Likewise.
14333 (vst1_s16_x2): Likewise.
14334 (vst1_p16_x2): Likewise.
14335 (vst1_s32_x2): Likewise.
14336 (vst1_u8_x2): Likewise.
14337 (vst1_u16_x2): Likewise.
14338 (vst1_u32_x2): Likewise.
14339 (vst1_f16_x2): Likewise.
14340 (vst1_f32_x2): Likewise.
14341 (vst1_p64_x2): Likewise.
14342 (vst1q_s8_x2): Likewise.
14343 (vst1q_p8_x2): Likewise.
14344 (vst1q_s16_x2): Likewise.
14345 (vst1q_p16_x2): Likewise.
14346 (vst1q_s32_x2): Likewise.
14347 (vst1q_s64_x2): Likewise.
14348 (vst1q_u8_x2): Likewise.
14349 (vst1q_u16_x2): Likewise.
14350 (vst1q_u32_x2): Likewise.
14351 (vst1q_u64_x2): Likewise.
14352 (vst1q_f16_x2): Likewise.
14353 (vst1q_f32_x2): Likewise.
14354 (vst1q_f64_x2): Likewise.
14355 (vst1q_p64_x2): Likewise.
14356 (vst1_s64_x3): Likewise.
14357 (vst1_u64_x3): Likewise.
14358 (vst1_f64_x3): Likewise.
14359 (vst1_s8_x3): Likewise.
14360 (vst1_p8_x3): Likewise.
14361 (vst1_s16_x3): Likewise.
14362 (vst1_p16_x3): Likewise.
14363 (vst1_s32_x3): Likewise.
14364 (vst1_u8_x3): Likewise.
14365 (vst1_u16_x3): Likewise.
14366 (vst1_u32_x3): Likewise.
14367 (vst1_f16_x3): Likewise.
14368 (vst1_f32_x3): Likewise.
14369 (vst1_p64_x3): Likewise.
14370 (vst1q_s8_x3): Likewise.
14371 (vst1q_p8_x3): Likewise.
14372 (vst1q_s16_x3): Likewise.
14373 (vst1q_p16_x3): Likewise.
14374 (vst1q_s32_x3): Likewise.
14375 (vst1q_s64_x3): Likewise.
14376 (vst1q_u8_x3): Likewise.
14377 (vst1q_u16_x3): Likewise.
14378 (vst1q_u32_x3): Likewise.
14379 (vst1q_u64_x3): Likewise.
14380 (vst1q_f16_x3): Likewise.
14381 (vst1q_f32_x3): Likewise.
14382 (vst1q_f64_x3): Likewise.
14383 (vst1q_p64_x3): Likewise.
14385 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
14387 * config/msp430/msp430.c (msp430_output_labelref): Prepend
14388 user_label_prefix to name.
14390 * tree-core.h: Update comment about the format of NAME string
14391 passed to handler in attribute_spec.
14393 * config/msp430/msp430.md: Remove erroneous subreg expression from
14394 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
14395 zero_extend{q,h}isi2.
14397 2018-05-30 Borislav Petkov <bp@suse.de>
14399 * doc/extend.texi: Document some architecture specific
14400 constraints and sort entries.
14402 2018-05-30 Martin Sebor <msebor@redhat.com>
14404 PR middle-end/85369
14405 * builtins.c (expand_builtin_stpcpy_1): New function.
14406 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
14407 only if the former succeeds.
14409 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
14411 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
14414 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
14416 * doc/invoke.texi (-flinker-output): Document
14418 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
14420 * passes.c (ipa_write_summaries): Only modify statements if body
14422 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
14423 incrementally linking.
14424 (ipa_passes): Likewise.
14425 * lto-cgraph.c (lto_output_node): When incrementally linking do not
14426 pass down resolution info.
14427 * common.opt (flag_incremental_link): Update info.
14428 * gcc.c (plugin specs): Turn flinker-output=* to
14429 -plugin-opt=-linker-output-known
14430 * toplev.c (compile_file): Also cut compilation when doing incremental
14432 * flag-types. (enum lto_partition_model): Add
14433 LTO_LINKER_OUTPUT_NOLTOREL.
14434 (invoke.texi): Add -flinker-output docs.
14435 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
14436 link same way as WPA; do not stream in dead initializers.
14438 * dwarf2out.c (dwarf2out_die_ref_for_decl,
14439 darf2out_register_external_decl): Support incremental link.
14441 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
14443 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
14445 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
14447 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
14448 it down to simple_object_copy_lto_debug_sections.
14449 (run_gcc): Determine incremental LTO link time and configure
14450 lto1 into non-wpa mode, disable renaming of debug sections.
14452 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
14454 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
14455 descriptions of various incorrectly documented functions.
14457 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
14460 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
14461 address check not strict.
14463 2018-05-30 Richard Biener <rguenther@suse.de>
14465 PR tree-optimization/85964
14466 * tracer.c (better_p): Drop initialized count check, we only
14467 call the function with initialized counts now.
14468 (find_best_successor): Do find a best edge if one
14469 has uninitialized count.
14470 (find_best_predecessor): Likewise. Do BB frequency check only
14471 if count is initialized.
14473 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
14475 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
14476 (aarch64_ldrstr_offset_compare): New.
14477 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
14478 load/store orderings.
14479 (aarch64_gen_adjusted_ldpstp): Likewise.
14481 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
14483 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
14484 Check for subset of GENERAL_REGS and FP_REGS.
14485 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
14488 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
14490 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
14491 and wi::to_poly_offset. Add the current offset and then check
14492 whether the sum fits, rather than using an unchecked addition of
14493 a checked term. Check for a shwi rather than a uhwi.
14494 * expr.c (get_bit_range): Use tree_to_poly_uint64.
14495 (store_constructor): Use poly_int_tree_p.
14496 (expand_expr_real_1): Likewise.
14497 * function.c (assign_temp): Likewise.
14498 * fold-const.c (const_binop): Use poly_int_tree_p and
14499 wi::to_poly_offset.
14500 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
14502 * ipa-icf-gimple.c (func_checker::compare_operand): Use
14503 to_poly_offset for MEM offsets.
14504 * ipa-icf.c (sem_variable::equals): Likewise.
14505 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
14506 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
14507 wi::to_poly_offset for BIT_FIELD_REF offsets.
14508 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
14509 wi::to_poly_offset.
14510 * var-tracking.c (emit_note_insn_var_location): Use
14511 tree_to_poly_uint64.
14513 2018-05-29 Jim Wilson <jimw@sifive.com>
14515 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
14517 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
14520 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
14521 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
14523 (sse4_1_round<mode>2): Use nonimmediate_operand
14524 for operand 1 predicate.
14526 2018-05-29 Martin Sebor <msebor@redhat.com>
14527 Richard Biener <rguenther@suse.de>
14530 * calls.c (get_size_range): Call determine_value_range instead
14531 of get_value_range..
14532 * tree-vrp.h (determine_value_range): Declared new function.
14533 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
14535 2018-05-29 Richard Biener <rguenther@suse.de>
14537 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
14538 sure to use non-pattern stmts for get_earlier_stmt arguments.
14539 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
14540 called on pattern stmts.
14541 (get_later_stmt): Likewise.
14543 2018-05-29 Martin Liska <mliska@suse.cz>
14545 PR gcov-profile/85759
14546 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
14549 2018-05-29 Jakub Jelinek <jakub@redhat.com>
14551 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
14553 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
14557 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
14558 VEC_PACK_FLOAT_EXPR): New tree codes.
14559 * tree-pretty-print.c (op_code_prio): Handle
14560 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
14561 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
14562 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
14563 * tree-inline.c (estimate_operator_cost): Likewise.
14564 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
14565 * fold-const.c (const_binop): Likewise.
14566 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
14567 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
14568 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
14569 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
14570 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
14571 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
14572 * expr.c (expand_expr_real_2): Likewise.
14573 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
14574 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
14575 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
14577 * optabs.c (expand_widen_pattern_expr): For
14578 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
14579 sign from result type rather than operand's type.
14580 (expand_binop_directly): For vec_packu_float_optab and
14581 vec_packs_float_optab allow result type to be different from operand's
14583 * optabs-tree.c (optab_for_tree_code): Handle
14584 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
14585 VEC_PACK_FLOAT_EXPR. Formatting fixes.
14586 * tree-vect-generic.c (expand_vector_operations_1): Handle
14587 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
14588 VEC_PACK_FLOAT_EXPR.
14589 * tree-vect-stmts.c (supportable_widening_operation): Handle
14591 (supportable_narrowing_operation): Handle FLOAT_EXPR.
14592 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
14593 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
14594 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
14595 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
14597 (vec_pack<floatprefix>_float_<mode>): New expander.
14598 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
14600 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
14601 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
14602 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
14603 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
14604 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
14606 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
14607 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
14608 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
14609 VEC_PACK_FLOAT_EXPR): Document.
14611 2018-05-29 Richard Biener <rguenther@suse.de>
14613 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
14615 (stmt_vec_info_vec): Make pointer.
14616 (init_stmt_vec_info_vec): Remove.
14617 (free_stmt_vec_info_vec): Likewise.
14618 (set_stmt_vec_info_vec): New function.
14619 (free_stmt_vec_infos): Likewise.
14620 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
14621 (set_vinfo_for_stmt): Likewise.
14622 (get_earlier_stmt): Likewise.
14623 (get_later_stmt): Likewise.
14624 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
14625 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
14626 (vec_info::~vec_info): Free stmt_vec_infos.
14627 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
14628 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
14629 (pass_slp_vectorize::execute): Likewise.
14630 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
14631 (free_stmt_vec_info_vec): Likewise.
14632 (set_stmt_vec_info_vec): New function.
14633 (free_stmt_vec_infos): Likewise.
14634 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
14635 the global stmt_vec_info_vec.
14636 * tree-parloops.c (gather_scalar_reductions): Use
14637 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
14640 2018-05-29 Richard Biener <rguenther@suse.de>
14642 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
14644 2018-05-29 Martin Liska <mliska@suse.cz>
14645 David Malcolm <dmalcolm@redhat.com>
14647 * vec.c (test_reverse): New.
14648 (vec_c_tests): Add new test.
14649 * vec.h (vl_ptr>::reverse): New function.
14651 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
14653 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
14655 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
14658 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
14660 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
14662 2018-05-28 Richard Biener <rguenther@suse.de>
14664 PR tree-optimization/85933
14665 * tree-vect-data-refs.c (vect_record_base_alignments): Only
14666 look at stmts marked as vectorizable.
14668 2018-05-28 Richard Biener <rguenther@suse.de>
14670 PR tree-optimization/85934
14671 * tree-vect-generic.c (expand_vector_operations_1): Hoist
14672 vector boolean check before scalar optimization.
14674 2018-05-28 Jakub Jelinek <jakub@redhat.com>
14676 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
14679 2018-05-28 Mark Wielaard <mark@klomp.org>
14681 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
14682 if it is an expression containing a minus sign.
14684 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
14686 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
14688 2018-05-27 Paul Koning <ni1d@arrl.net>
14690 * config/pdp11/pdp11.md (truncsihi2): Remove.
14692 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
14693 Chung-Ju Wu <jasonwucj@gmail.com>
14695 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
14697 (unaligned_store_dw): Ditto.
14698 * config/nds32/nds32-memory-manipulation.c
14699 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
14700 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
14701 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
14702 (emit_setmem_word_loop): Rename to ...
14703 (emit_setmem_doubleword_loop): ... this.
14704 (nds32_gen_dup_4_byte_to_word_value): New function.
14705 (nds32_gen_dup_8_byte_to_double_word_value): New function.
14706 (nds32_expand_setmem_loop): Refine implementation.
14707 (nds32_expand_setmem_loop_v3m): Ditto.
14708 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
14711 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
14713 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
14715 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
14717 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
14718 (nds32_init_machine_status): Initialize machine->attr_naked_p and
14719 machine->attr_no_prologue_p.
14720 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
14721 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
14722 (nds32_expand_epilogue): Consider attr_naked_p.
14723 (nds32_expand_epilogue_v3pop): Likewise.
14724 (nds32_can_use_return_insn): Likewise.
14725 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
14726 attr_no_prologue_p fields.
14727 * config/nds32/nds32.opt (mret-in-naked-func): New option.
14729 2018-05-27 Jakub Jelinek <jakub@redhat.com>
14732 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
14734 * config/i386/sse.md
14735 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
14737 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
14739 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
14741 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
14743 (*<floatsuffix>floatv2div2sf2): Rename to ...
14744 (*float<floatunssuffix>v2div2sf2): ... this.
14745 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
14746 (float<floatunssuffix>v2div2sf2_mask): ... this.
14747 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
14748 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
14749 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
14751 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
14753 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
14755 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
14757 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
14759 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
14761 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
14762 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
14763 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
14764 gen_ufix_truncv8dfv8si2.
14765 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
14766 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
14767 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
14768 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
14769 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
14770 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
14771 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
14772 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
14774 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
14778 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
14780 2018-05-25 Jim Wilson <jimw@sifive.com>
14782 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
14783 * config/riscv/riscv.c (struct machine_function): Add
14784 interrupt_handler_p and attribute_checked_p fields.
14785 (riscv_attribute_table): Add interrupt.
14786 (riscv_interrupt_type_p): New.
14787 (riscv_save_reg_p): Save extra regs for interrupt handler.
14788 (riscv_use_save_libcall): Return false for interrupt handler.
14789 (riscv_first_stack_step): Add forward declaration.
14790 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
14791 for interrupt handler with large frame. Use it for saved reg list.
14792 (riscv_expand_prologue): Move flag_stack_usage_info support to
14793 eliminate duplication.
14794 (riscv_expand_epilogue): Generate mret for interrupt handler.
14795 (riscv_epilogue_uses): New.
14796 (riscv_can_use_return_insn): Return false for interrupt handler.
14797 (riscv_function_ok_for_sibcall): Likewise.
14798 (riscv_set_current_function): Add interrupt handler support.
14799 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
14800 * config/riscv/riscv.md (UNSPECV_MRET): New.
14802 (riscv_frflags, riscv_fsflags): Use tab after opcode.
14804 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
14806 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
14808 PR tree-optimization/85712
14809 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
14810 this candidate has already been replaced in-situ by a copy.
14812 2018-05-25 Jason Merrill <jason@redhat.com>
14814 PR c++/80485 - inline function non-zero address.
14815 * symtab.c (nonzero_address): Check DECL_COMDAT.
14817 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
14820 * config/alpha/alpha.md (ashlsi3): New insn pattern.
14821 (*ashlsi_se): Rename from *ashldi_se. Define as sign
14822 extension of SImode operation. Use const123_operand predicate.
14823 (*saddsi_1): Remove.
14824 (*saddl_se_1): Ditto.
14825 (*ssubsi_1): Ditto.
14826 (*ssubl_se_1): Ditto.
14827 * config/alpha/predicates.md (const123_operand): New predicate.
14828 * config/alpha/constraints.md (P): Use IN_RANGE.
14830 2018-05-25 Richard Biener <rguenther@suse.de>
14832 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
14834 (ref_maybe_used_by_stmt_p): Likewise.
14835 (stmt_may_clobber_ref_p): Likewise.
14836 (stmt_may_clobber_ref_p_1): Likewise.
14837 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
14839 (ref_maybe_used_by_stmt_p): Likewise.
14840 (stmt_may_clobber_ref_p): Likewise.
14841 (stmt_may_clobber_ref_p_1): Likewise.
14842 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
14843 the alias oracle to disambiguate DRs with stmts DR analysis
14845 (vect_analyze_data_refs): Do not give up on not analyzable
14846 DRs for BB vectorization. Remove code truncating the dataref
14849 2018-05-25 Jakub Jelinek <jakub@redhat.com>
14852 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
14853 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
14854 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
14856 2018-05-25 Richard Biener <rguenther@suse.de>
14858 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
14859 function, combining stmt data ref gathering and fatal analysis
14861 (vect_analyze_data_refs): Remove now redudnant code and simplify.
14862 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
14863 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
14864 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
14865 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
14867 2018-05-25 Bin Cheng <bin.cheng@arm.com>
14869 PR tree-optimization/85720
14870 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
14871 SCC if all partitions are builtins.
14872 (version_loop_by_alias_check): New parameter. Generate cancelable
14873 runtime alias check if all partitions are builtins.
14874 (distribute_loop): Update call to above function.
14876 2018-05-25 Bin Cheng <bin.cheng@arm.com>
14878 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
14879 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
14880 (parm_default_def_partition_arg): Ditto.
14881 (set_parm_default_def_partition): Ditto.
14882 (get_parm_default_def_partitions): Ditto and make it static.
14883 (get_undefined_value_partitions): Ditto and make it static.
14884 (remove_ssa_form): Refactor call to init_var_map here.
14885 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
14886 computation for loop region.
14887 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
14888 (register_default_def): Delete.
14889 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
14890 (parm_default_def_partition_arg): Ditto.
14891 (set_parm_default_def_partition): Ditto.
14892 (get_parm_default_def_partitions): Ditto and make it static.
14893 (get_undefined_value_partitions): Ditto and make it static.
14894 (coalesce_with_default, coalesce_with_default): Update comment.
14895 (create_coalesce_list_for_region): New func factored out from
14896 create_outofssa_var_map.
14897 (populate_coalesce_list_for_outofssa): New func factored out from
14898 create_outofssa_var_map and coalesce_ssa_name.
14899 (create_outofssa_var_map): Delete.
14900 (coalesce_ssa_name): Refactor to support live range computation.
14901 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
14902 (get_parm_default_def_partitions): Delete.
14903 (get_undefined_value_partitions): Ditto.
14904 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
14905 computation for loop region.
14906 (new_tree_live_info, loe_visit_block): Ditto.
14907 (live_worklist, set_var_live_on_entry): Ditto.
14908 (calculate_live_on_exit, verify_live_on_entry): Ditto.
14909 * tree-ssa-live.h (struct _var_map): New fields.
14910 (init_var_map): Change decl.
14911 (region_contains_p): New.
14913 2018-05-25 Bin Cheng <bin.cheng@arm.com>
14915 * tree-ssa-live.h (live_merge_and_clear): Delete.
14917 2018-05-25 Richard Biener <rguenther@suse.de>
14920 * tree-dump.c (dequeue_and_dump): Remove access to removed
14921 operand 2 of a SWITCH_EXPR.
14923 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
14925 * doc/sourcebuild.texi (vect_double_cond_arith): Include
14926 multiplication and division.
14927 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
14928 (cond_udiv@var{m}, cond_umod@var{m}): Document.
14929 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
14930 (cond_udiv_optab, cond_umod_optab): New optabs.
14931 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
14932 (IFN_COND_RDIV): New internal functions.
14933 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
14934 TRUNC_MOD_EXPR and RDIV_EXPR.
14935 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
14936 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
14938 (SVE_INT_BINARY): Include mult.
14939 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
14940 (optab, sve_int_op): Handle mult.
14941 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
14943 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
14944 for SVE_INT_BINARY_SD.
14946 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
14948 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
14949 (optab, sve_int_op): Handle div and udiv.
14950 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
14951 for SVE_INT_BINARY_SD.
14952 (*<optab><mode>3): New insn for the same.
14954 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
14956 * tree-vect-patterns.c: Include predict.h.
14957 (vect_recog_divmod_pattern): Restrict check for division support
14958 to when optimizing for size.
14960 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
14962 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
14963 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
14964 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
14965 (gimple_match_op::set_op): Likewise.
14966 (gimple_resimplify4): Declare.
14967 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
14968 (expr::gen_transform): Likewise.
14969 (decision_tree::gen): Generate a simplification routine for 4 operands.
14970 * gimple-match-head.c (gimple_simplify): Add an overload for
14971 4 operands. In the top-level function, handle up to 4 call
14972 arguments and call gimple_resimplify4.
14973 (gimple_resimplify4): New function.
14974 (build_call_internal): Pass a fourth operand.
14975 (maybe_push_to_seq): Likewise.
14976 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
14977 Fold VEC_COND_EXPRs of an operation and a default value into
14978 an IFN_COND_* function if possible.
14979 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
14981 (SVE_COND_FP_BINARY): Include them.
14982 (optab, sve_fp_op): Handle them.
14983 (SVE_INT_BINARY_REV): New code iterator.
14984 (SVE_COND_FP_BINARY_REV): New int iterator.
14985 (commutative): New int attribute.
14986 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
14988 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
14990 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
14991 (*cond_<optab><mode>): New patterns for reversed operands.
14993 2018-05-25 Richard Biener <rguenther@suse.de>
14995 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
14996 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
14997 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
14998 (STMT_VINFO_GROUPED_ACCESS): Adjust.
14999 * tree-vect-data-refs.c (everywhere): Adjust users.
15000 * tree-vect-loop.c (everywhere): Likewise.
15001 * tree-vect-slp.c (everywhere): Likewise.
15002 * tree-vect-stmts.c (everywhere): Likewise.
15003 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
15005 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15007 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
15009 (gcc_cv_as_section_exclude): ... this.
15010 Try Solaris as #exclude syntax.
15011 * configure: Regenerate.
15012 * config.in: Regenerate.
15013 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
15015 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
15016 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
15018 * varasm.c (default_elf_asm_named_section): Don't check if
15019 HAVE_GAS_SECTION_EXCLUDE is defined.
15021 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15023 * doc/md.texi: Update the documentation of the cond_* optabs
15024 to mention the new final operand. Fix GET_MODE_NUNITS call.
15025 Describe the scalar case too.
15026 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
15027 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
15029 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
15030 (get_conditional_internal_fn): Update comment.
15031 * tree-vect-loop.c (vectorizable_reduction): Pass the original
15032 accumulator value as a final argument to conditional functions.
15033 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
15034 a define_expand and add an "else" operand. Assert for now that
15035 the else operand is equal to operand 2. Use SVE_INT_BINARY and
15036 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
15037 (*cond_<optab><mode>): New patterns.
15038 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
15039 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
15040 (UNSPEC_COND_EOR): Delete.
15041 (optab): Remove associated mappings.
15042 (SVE_INT_BINARY): New code iterator.
15043 (sve_int_op): Remove int attribute and add "minus" to the code
15045 (SVE_COND_INT_OP): Delete.
15046 (SVE_COND_FP_OP): Rename to...
15047 (SVE_COND_FP_BINARY): ...this.
15049 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15051 * optabs.c (can_reuse_operands_p): New function.
15052 (maybe_legitimize_operands): Try to reuse the results for
15055 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
15057 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
15058 Add {q} suffix to insn mnemonic.
15060 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
15062 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
15063 (msp430_warn_func_return): New.
15065 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
15067 * fold-const.c (tree_nonzero_bits): New function.
15068 * fold-const.h (tree_nonzero_bits): Likewise.
15069 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
15070 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
15072 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
15076 * varasm.c (assemble_alias): Check ifunc_resolver only on
15079 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
15082 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
15083 when memory input operand is handled.
15085 2018-05-24 Luis Machado <luis.machado@linaro.org>
15087 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
15089 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
15091 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
15093 * match.pd: Delay FMA folds until after vectorization.
15095 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
15098 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
15099 address check not strict.
15101 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
15103 * gimple-match.h (gimple_match_op): New class.
15104 (mprts_hook): Replace parameters with a gimple_match_op *.
15105 (maybe_build_generic_op): Likewise.
15106 (gimple_simplified_result_is_gimple_val): Replace parameters with
15107 a const gimple_match_op *.
15108 (gimple_simplify): Replace code_helper * and tree * parameters with
15109 a gimple_match_op * parameter.
15110 (gimple_resimplify1): Replace code_helper *, tree and tree *
15111 parameters with a gimple_match_op * parameter.
15112 (gimple_resimplify2): Likewise.
15113 (gimple_resimplify3): Likewise.
15114 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
15115 parameters with a gimple_match_op * parameter.
15116 * gimple-match-head.c (gimple_simplify): Change prototypes of
15117 auto-generated functions to take a gimple_match_op * instead of
15118 separate code_helper * and tree * parameters. Make the same
15119 change in the top-level overload and update calls to the
15120 gimple_resimplify routines. Update calls to the auto-generated
15121 functions and to maybe_push_res_to_seq in the publicly-facing
15122 operation-specific gimple_simplify overloads.
15123 (gimple_match_op::MAX_NUM_OPS): Define.
15124 (gimple_resimplify1): Replace rcode and ops with a single res_op
15125 parameter. Update call to gimple_simplify.
15126 (gimple_resimplify2): Likewise.
15127 (gimple_resimplify3): Likewise.
15128 (mprts_hook): Replace parameters with a gimple_match_op *.
15129 (maybe_build_generic_op): Likewise.
15130 (build_call_internal): Replace type, nargs and ops with
15131 a gimple_match_op *.
15132 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
15133 with a single gimple_match_op *. Update calls to mprts_hook,
15134 build_call_internal and gimple_simplified_result_is_gimple_val.
15135 Factor out code that is common to the tree_code and combined_fn cases.
15136 * genmatch.c (expr::gen_transform): Replace tem_code and
15137 tem_ops with a gimple_match_op called tem_op. Update calls
15138 to the gimple_resimplify functions and maybe_push_res_to_seq.
15139 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
15140 res_ops. Update call to the gimple_resimplify functions.
15141 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
15142 (decision_tree::gen): Make the functions take a gimple_match_op *
15143 called res_op instead of separate res_code and res_ops parameters.
15144 Update call accordingly.
15145 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
15146 and ops with a single res_op parameter. Update calls to
15147 maybe_build_generic_op and maybe_push_res_to_seq.
15148 (fold_stmt_1): Update calls to gimple_simplify and
15149 replace_stmt_with_simplification.
15150 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
15151 and gimple_simplified_result_is_gimple_val.
15152 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
15154 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
15155 with a gimple_match_op *.
15156 (vn_nary_build_or_lookup): Likewise. Update call to
15157 vn_nary_build_or_lookup_1.
15158 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
15159 gimple_match_op *. Update calls to the gimple_resimplify routines
15160 and to gimple_simplified_result_is_gimple_val.
15161 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
15162 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
15163 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
15164 (visit_nary_op): Likewise.
15165 (visit_reference_op_load): Likewise.
15167 2018-05-23 Luis Machado <luis.machado@linaro.org>
15169 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
15170 modifier for printing the step amount.
15172 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15175 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
15178 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
15180 * doc/sourcebuild.texi (Endianness): New subsubsection.
15182 2018-05-23 Luis Machado <luis.machado@linaro.org>
15184 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
15185 <prefetch_dynamic_strides>: New const bool field.
15186 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
15187 prefetch_dynamic_strides.
15188 (exynosm1_prefetch_tune): Likewise.
15189 (thunderxt88_prefetch_tune): Likewise.
15190 (thunderx_prefetch_tune): Likewise.
15191 (thunderx2t99_prefetch_tune): Likewise.
15192 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
15194 (aarch64_override_options_internal): Update to set
15195 PARAM_PREFETCH_DYNAMIC_STRIDES.
15196 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
15197 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
15198 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
15199 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
15200 prefetch-dynamic-strides setting.
15202 2018-05-23 Luis Machado <luis.machado@linaro.org>
15204 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
15205 <minimum_stride>: New const int field.
15206 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
15207 minimum_stride field defaulting to -1.
15208 (exynosm1_prefetch_tune): Likewise.
15209 (thunderxt88_prefetch_tune): Likewise.
15210 (thunderx_prefetch_tune): Likewise.
15211 (thunderx2t99_prefetch_tune): Likewise.
15212 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
15213 <default_opt_level>: Set to 3.
15214 (aarch64_override_options_internal): Update to set
15215 PARAM_PREFETCH_MINIMUM_STRIDE.
15216 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
15217 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
15218 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
15219 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
15220 stride is constant and is below the minimum stride threshold.
15222 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15224 * config/arm/arm-cpus.in (mode26): Delete.
15225 (armv4): Delete mode26 reference.
15226 * config/arm/arm.c (arm_configure_build_target): Delete use of
15229 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
15231 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
15233 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
15234 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
15236 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
15237 (floatunsdidf2): Ditto.
15239 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
15241 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
15242 (fixuns_trunc<mode>si2_avx512f): Ditto.
15243 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
15244 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
15245 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
15247 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
15249 PR rtl-optimization/79985
15250 * df-scan.c (df_insn_refs_collect): Remove special case for
15251 global registers and asm statements.
15253 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
15255 * extend.texi (Global Register Variables): Rewrite the bullet list.
15256 Note that the register is available for allocation. Note that access
15257 via inline asm must use constraints. Add note about async-signal
15258 handlers. Remove paragraph about automagic register selection.
15260 2018-05-23 Richard Biener <rguenther@suse.de>
15262 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
15263 of fixed offset from memset VN.
15265 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
15267 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
15268 first_interp field.
15269 (alloc_cand_and_find_basis): Initialize first_interp field.
15270 (slsr_process_mul): Modify first_interp field.
15271 (slsr_process_add): Likewise.
15272 (slsr_process_cast): Modify first_interp field for each new
15274 (slsr_process_copy): Likewise.
15275 (dump_candidate): Dump first_interp field.
15276 (replace_mult_candidate): Process all interpretations, not just
15278 (replace_rhs_if_not_dup): Likewise.
15279 (replace_one_candidate): Likewise.
15281 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
15283 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
15285 (aarch64_needs_frame_chain): New function.
15286 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
15288 2018-05-23 Sudakshina Das <sudi.das@arm.com>
15291 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
15292 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
15293 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
15294 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
15295 as true for strict-align.
15296 (aarch64_can_inline_p): Perform checks even when callee has no
15297 attributes to check for strict alignment.
15298 * doc/extend.texi (AArch64 Function Attributes): Document
15300 * doc/invoke.texi: (AArch64 Options): Likewise.
15302 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
15304 PR tree-optimization/85853
15305 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
15306 the handling of the root of the node to...
15307 (vect_slp_analyze_node_operations_1): ...this new function,
15308 and run the whole thing with the child nodes' def types
15309 set according to their SLP node's def type.
15311 2018-05-23 Richard Biener <rguenther@suse.de>
15313 PR middle-end/85874
15314 * tree-data-ref.c (create_runtime_alias_checks): Defer
15315 and ignore overflow warnings.
15317 2018-05-23 Yury Gribov <tetra2005@gmail.com>
15319 PR tree-optimization/85822
15320 * tree-vrp.c (is_masked_range_test): Fix handling of negative
15323 2018-05-23 Richard Biener <rguenther@suse.de>
15325 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
15326 memset constants via native_interpret_expr.
15328 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
15331 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
15333 (cgraph_node::create_alias): Likewise.
15334 (cgraph_node::get_availability): Check ifunc_resolver instead
15335 of looking up ifunc attribute.
15336 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
15337 * varasm.c (do_assemble_alias): Likewise.
15338 (assemble_alias): Likewise.
15339 (default_binds_local_p_3): Likewise.
15340 * cgraph.h (cgraph_node): Add ifunc_resolver.
15341 (cgraph_node::only_called_directly_or_aliased_p): Return false
15342 for IFUNC resolver.
15343 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
15345 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
15346 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
15347 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
15348 instead of looking up ifunc attribute.
15350 2018-05-22 Luis Machado <luis.machado@linaro.org>
15352 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
15354 2018-05-22 Martin Sebor <msebor@redhat.com>
15356 PR middle-end/85359
15357 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
15358 only when expasion succeeds.
15359 (expand_builtin_strcmp): Same.
15360 (expand_builtin_strncmp): Same.
15362 2018-05-22 Martin Sebor <msebor@redhat.com>
15364 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
15366 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
15367 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15369 * config/aarch64/aarch64-ldpstp.md: Replace uses of
15370 aarch64_mem_pair_operand with memory_operand and delete operand swapping
15372 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
15373 Add check for legitimate_address.
15374 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
15375 (aarch64_swap_ldrstr_operands): New.
15376 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
15379 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
15380 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15382 * config/aarch64/aarch64.md: New patterns to generate stp
15384 (store_pair_sw, store_pair_dw): New patterns to generate stp for
15385 single words and double words.
15386 (load_pair_sw, load_pair_dw): Likewise.
15387 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
15389 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
15391 * config/aarch64/aarch64-ldpstp.md: Modify peephole
15392 for different mode ldpstp and add peephole for merged zero stores.
15393 Likewise for loads.
15394 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
15396 (aarch64_gen_store_pair): Rename calls to match new patterns.
15397 (aarch64_gen_load_pair): Rename calls to match new patterns.
15398 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
15399 (load_pair<DREG:mode><DREG2:mode>): ... This.
15400 (store_pair<mode>): Rename to...
15401 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
15402 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
15403 New mode iterators.
15404 (V_INT_EQUIV): Handle SImode.
15405 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
15408 2018-05-22 Martin Sebor <msebor@redhat.com>
15411 * calls.c (maybe_warn_nonstring_arg): Use string length to set
15412 or ajust the presumed bound on an operation to avoid unnecessary
15415 2018-05-22 Martin Sebor <msebor@redhat.com>
15417 PR tree-optimization/85826
15418 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
15419 assuming that a DECL necesarily has a constant size.
15421 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
15423 PR middle-end/85862
15424 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
15426 2018-05-22 Richard Biener <rguenther@suse.de>
15428 PR tree-optimization/85834
15429 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
15430 non-constant and non-zero memset arguments.
15432 2018-05-22 Martin Liska <mliska@suse.cz>
15435 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
15437 2018-05-22 Richard Biener <rguenther@suse.de>
15439 PR tree-optimization/85863
15440 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
15441 comparisons when vectype is specified.
15442 (vectorizable_condition): Do not specify vectype for
15443 vect_is_simple_cond when SLP vectorizing.
15445 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
15448 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
15449 define __ibm128 as long double.
15450 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
15451 as a distinct type when IEEE 128-bit support is enabled.
15452 (init_float128_ieee): Fix up conversions between IFmode and IEEE
15453 128-bit types to use the correct functions.
15454 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
15455 convert between 128-bit floating point types that have different
15456 modes but the same representation, instead of using gen_lowpart to
15458 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
15460 (IFKF_reg): New attributes to give the register constraints for
15462 (extend<mode>tf2_internal): New insns to mark an explicit
15463 conversion between 128-bit floating point types that have a
15464 different mode but share the same representation.
15466 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
15468 PR tree-optimization/85814
15469 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
15470 a null return from get_strinfo when unsharing the next
15471 strinfo in the chain.
15473 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
15476 * varasm.c (weak_finish): Clean up weak_decls.
15478 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15480 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
15481 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
15482 UNSPEC_UADALP values.
15483 * config/aarch64/iterators.md (ABAL): New int iterator.
15486 (sur): Add mappings for the above.
15487 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
15489 (aarch64_<sur>abal<mode>_4): Likewise.
15490 (aarch64_<sur>adalp<mode>_3): Likewise.
15491 (<sur>sadv16qi): New define_expand.
15493 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
15495 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
15496 (*movdf_internal): Ditto.
15497 (*rcpsf2_sse): Ditto.
15498 (*rsqrtsf2_sse): Ditto.
15499 (*sqrt<mode>2_sse): Ditto.
15501 2018-05-21 Tamar Christina <tamar.christina@arm.com>
15503 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
15505 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
15506 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
15507 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
15508 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
15510 * config/aarch64/arm_neon.h: Likewise.
15511 * config/aarch64/iterators.md (VQ_I): New.
15513 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
15515 * config.gcc: Add arc/t-multilib-linux to tmake_file for
15517 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
15520 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
15522 * config/nds32/constraints.md (S): New constraint.
15523 * config/nds32/nds32.md (call_internal): Use constraint S.
15524 (call_value_internal): Likewise.
15525 (sibcall_internal): Likewise.
15526 (sibcall_value_internal): Likewise.
15528 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
15529 Chung-Ju Wu <jasonwucj@gmail.com>
15531 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
15532 into consideration.
15534 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
15535 Chung-Ju Wu <jasonwucj@gmail.com>
15537 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
15538 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
15539 (nds32_rtx_costs_impl): Simplify.
15540 (nds32_address_cost_impl): Simplify.
15541 (nds32_init_rtx_costs): New function.
15542 (nds32_rtx_costs_speed_prefer): Likewise.
15543 (nds32_rtx_costs_size_prefer): Likewise.
15544 (nds32_address_cost_speed_prefer): Likewise.
15545 (nds32_address_cost_speed_fwprop): Likewise.
15546 (nds32_address_cost_size_prefer): Likewise.
15547 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
15548 * config/nds32/nds32.c (nds32_option_override): Use
15549 nds32_init_rtx_costs function.
15551 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
15553 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
15554 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
15555 (TARGET_PIPELINE_N8): Likewise.
15556 (TARGET_PIPELINE_N10): Likewise.
15557 (TARGET_PIPELINE_N13): Likewise.
15558 (TARGET_PIPELINE_GRAYWOLF): Likewise.
15560 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
15562 * config/nds32/nds32-fpu.md: Update copyright year.
15564 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
15566 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
15568 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
15570 * config/nds32/nds32.c
15571 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
15572 * config/nds32/nds32.opt (minline-asm-r15): New option.
15574 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
15576 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
15578 * config/nds32/nds32.md (abssi2): New pattern.
15580 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
15582 * config/i386/i386.md (rex64namesuffix): New mode attribute.
15583 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
15584 Merge insn pattern from sse_cvtsi2ss<round_name> and
15585 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
15586 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
15587 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
15588 using SWI48 mode iterator.
15589 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
15590 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
15591 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
15592 pattern from sse_cvttss2si<round_saeonly_name>
15593 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
15594 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
15595 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
15596 using SWI48 mode iterator.
15597 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
15598 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
15599 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
15600 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
15601 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
15602 using SWI48 mode iterator.
15603 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
15604 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
15605 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
15606 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
15607 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
15608 SWI48 mode iterator.
15609 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
15610 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
15611 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
15612 pattern from sse_cvttsd2si<round_saeonly_name>
15613 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
15615 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
15617 * config/nds32/nds32-md-auxiliary.c
15618 (nds32_valid_smw_lwm_base_p): Refine.
15619 (nds32_output_smw_single_word): Refine.
15620 (nds32_output_smw_double_word): New.
15621 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
15623 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
15625 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
15626 (nds32_output_stack_pop): Refine.
15627 (nds32_expand_unaligned_load): Refine.
15628 (nds32_expand_unaligned_store): Refine.
15630 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
15631 Chung-Ju Wu <jasonwucj@gmail.com>
15633 * config/nds32/constants.md: Add TP_REGNUM constant.
15634 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
15635 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
15637 * config/nds32/nds32-doubleword.md: Consider flag_pic.
15638 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
15639 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
15640 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
15641 and PIC code generation.
15642 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
15644 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
15646 * config/nds32/nds32.md: Support TLS and PIC.
15647 * config/nds32/nds32.c: Support TLS and PIC.
15648 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
15649 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
15652 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
15654 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
15655 mode with E_ prefix.
15657 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
15658 Chung-Ju Wu <jasonwucj@gmail.com>
15660 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
15661 * config/nds32/nds32-md-auxiliary.c
15662 (symbolic_reference_mentioned_p): New.
15663 (nds32_legitimize_ict_address): New.
15664 (nds32_expand_ict_move): New.
15665 (nds32_indirect_call_referenced_p): New.
15666 (nds32_symbol_binds_local_p): Delete.
15667 (nds32_long_call_p): Modify.
15668 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
15669 * config/nds32/nds32-protos.h
15670 (symbolic_reference_mentioned_p): Declare.
15671 (nds32_legitimize_ict_address): Declare.
15672 (nds32_expand_ict_move): Declare.
15673 (nds32_indirect_call_referenced_p): Declare.
15674 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
15675 (nds32_relax_group): Use nds32_ict_const_p as condition.
15676 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
15677 (nds32_asm_file_start): Output ict_model directive in asm code.
15678 (nds32_legitimate_address_p): Consider indirect call.
15679 (nds32_print_operand): Consider indirect call.
15680 (nds32_print_operand_address): Consider indirect call.
15681 (nds32_insert_attributes): Handle "indirect_call" attribute.
15682 (TARGET_LEGITIMATE_ADDRESS_P): Define.
15683 (TARGET_LEGITIMATE_CONSTANT_P): Define.
15684 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
15685 (TARGET_DELEGITIMIZE_ADDRESS): Define.
15686 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
15687 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
15688 (TARGET_ICT_MODEL_SMALL): Define.
15689 (TARGET_ICT_MODEL_LARGE): Define.
15690 * config/nds32/nds32.md (movsi): Consider ict model.
15691 (call, call_value): Consider ict model.
15692 (sibcall, sibcall_value): Consider ict model.
15693 * config/nds32/nds32.opt (mict-model): New option.
15694 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
15697 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
15698 Monk Chiang <sh.chiang04@gmail.com>
15699 Jim Wilson <jimw@sifive.com>
15701 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
15702 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
15703 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
15704 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
15705 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
15706 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
15707 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
15708 compute save_libcall_adjustment properly.
15709 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
15710 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
15711 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
15712 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
15713 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
15714 (ABI_SPEC): Handle mabi=ilp32e.
15715 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
15716 (RVE): Add RVE mask.
15717 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
15718 <-march>: Add rv32e as an example.
15720 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
15723 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
15724 (intra_create_variable_infos): Handle C++ constructors.
15726 2018-05-18 Martin Liska <mliska@suse.cz>
15728 * passes.def: Remove a redundant pass.
15730 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
15733 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
15735 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15737 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
15739 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
15741 (armv2, armv2a, armv3, armv3m): Delete architectures.
15742 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
15743 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
15744 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
15746 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
15747 (*mulsidi3adddi): Likewise.
15748 (mulsidi3): Likewise.
15749 (*mulsidi3_nov6): Likewise.
15750 (umulsidi3): Likewise.
15751 (umulsidi3_nov6): Likewise.
15752 (umaddsidi4): Likewise.
15753 (*umulsidi3adddi): Likewise.
15754 (smulsi3_highpart): Likewise.
15755 (*smulsi3_highpart_nov6): Likewise.
15756 (umulsi3_highpart): Likewise.
15757 (*umulsi3_highpart_nov6): Likewise.
15758 * config/arm/arm.h (arm_arch3m): Delete.
15759 * config/arm/arm.c (arm_arch3m): Delete.
15760 (arm_option_override_internal): Update armv3-related comment.
15761 (arm_configure_build_target): Delete use of isa_bit_mode32.
15762 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
15763 (arm_rtx_costs_internal): Delete check of arm_arch3m.
15764 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
15765 (mulsa3): Likewise.
15766 (mulusa3): Likewise.
15767 * config/arm/arm-protos.h (arm_arch3m): Delete.
15768 * config/arm/arm-tables.opt: Regenerate.
15769 * config/arm/arm-tune.md: Likewise.
15770 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
15771 deleted architectures.
15773 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15775 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
15776 (armv5t, armv5te): New features.
15777 (ARMv5, ARMv5e): Delete fgroups.
15778 (ARMv5t, ARMv5te): Adjust for above changes.
15779 (ARMv6m): Likewise.
15780 (armv5, armv5e): Delete arches.
15781 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
15783 (*call_reg_arm): Likewise.
15784 (*call_value_reg_armv5): Likewise.
15785 (*call_value_reg_arm): Likewise.
15786 (*call_symbol): Likewise.
15787 (*call_value_symbol): Likewise.
15788 (*sibcall_insn): Likewise.
15789 (*sibcall_value_insn): Likewise.
15790 (clzsi2): Likewise.
15791 (prefetch): Likewise.
15792 (define_split and define_peephole2 dependent on arm_arch5):
15794 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
15796 (TARGET_ARM_QBIT): Likewise.
15797 (TARGET_DSP_MULTIPLY): Likewise.
15798 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
15799 (arm_arch5, arm_arch5e): Delete.
15800 (arm_arch5t, arm_arch5te): Declare.
15801 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
15802 (arm_arch5t): Declare.
15803 (arm_option_reconfigure_globals): Update for the above.
15804 (arm_options_perform_arch_sanity_checks): Update comment, replace
15805 use of arm_arch5 with arm_arch5t.
15806 (use_return_insn): Likewise.
15807 (arm_emit_call_insn): Likewise.
15808 (output_return_instruction): Likewise.
15809 (arm_final_prescan_insn): Likewise.
15810 (arm_coproc_builtin_available): Likewise.
15811 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
15812 arm_arch5e with arm_arch5t and arm_arch5te.
15813 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
15814 (arm_arch5t, arm_arch5te): Declare.
15815 * config/arm/arm-tables.opt: Regenerate.
15816 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
15817 * config/arm/t-multilib: Likewise.
15818 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
15819 instead of arm_arch5.
15820 (*call_reg_thumb1): Likewise.
15821 (*call_value_reg_thumb1_v5): Likewise.
15822 (*call_value_reg_thumb1): Likewise.
15823 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
15825 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
15827 2018-05-18 Martin Liska <mliska@suse.cz>
15829 PR gcov-profile/84846
15830 * doc/gcov.texi: Document -t option of gcov tool.
15832 2018-05-18 Martin Liska <mliska@suse.cz>
15834 PR gcov-profile/84846
15835 * gcov.c (print_usage): Add new -t option.
15836 (process_args): Handle the option.
15837 (generate_results): Use stdout as output when requested by
15840 2018-05-18 Martin Liska <mliska@suse.cz>
15842 PR gcov-profile/84846
15843 * coverage.c (coverage_init): Write PWD to .gcno file.
15844 * doc/gcov.texi: Document how working directory is printed.
15845 * gcov-dump.c (dump_gcov_file): Print PWD.
15846 * gcov.c (output_intermediate_file): Likewise.
15847 (read_graph_file): Read PWD string.
15848 (output_lines): Print PWD.
15850 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15852 PR middle-end/85817
15853 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
15854 for retval and return false if all args to phi are zero.
15856 2018-05-18 Richard Biener <rguenther@suse.de>
15858 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
15860 (evrp_dom_walker::before_dom_children): Call it.
15862 2018-05-18 Richard Biener <rguenther@suse.de>
15864 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
15865 results when processing array refs with variable index.
15867 2018-05-18 Toon Moene <toon@moene.org>
15869 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
15870 directly after that of -floop-interchange. Indicate that both
15871 options are enabled by default when specifying -O3.
15873 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15875 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
15876 iterator. Delete separate integer-mode vec_set<mode> expander.
15877 (aarch64_simd_vec_setv2di): Delete.
15878 (vec_setv2di): Delete.
15879 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
15880 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
15881 the "w, r" alternative.
15883 2018-05-18 Martin Liska <mliska@suse.cz>
15885 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
15886 * tree-pass.h (make_pass_lower_switch_O0): New function.
15887 * tree-switch-conversion.c (node_has_low_bound): Remove.
15888 (node_has_high_bound): Likewise.
15889 (node_is_bounded): Likewise.
15890 (class pass_lower_switch): Make it a template type and create
15892 (pass_lower_switch::execute): Add template argument.
15893 (make_pass_lower_switch): New function.
15894 (make_pass_lower_switch_O0): New function.
15895 (do_jump_if_equal): Remove.
15896 (emit_case_nodes): Simplify to just handle all 3 cases and leave
15897 all the hard work to tree optimization passes.
15899 2018-05-18 Martin Liska <mliska@suse.cz>
15901 * dbgcnt.c (limit_low): Renamed from limit.
15902 (limit_high): New variable.
15903 (dbg_cnt_is_enabled): Check for upper limit.
15904 (dbg_cnt): Adjust dumping.
15905 (dbg_cnt_set_limit_by_index): Add new argument for high
15907 (dbg_cnt_set_limit_by_name): Likewise.
15908 (dbg_cnt_process_single_pair): Parse new format.
15909 (dbg_cnt_process_opt): Use strtok.
15910 (dbg_cnt_list_all_counters): Remove 'value' and add
15912 * doc/invoke.texi: Document changes.
15914 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
15916 * doc/sourcebuild.texi (scalar_all_fma): Document.
15917 * tree.def (FMA_EXPR): Delete.
15918 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
15919 * internal-fn.c (ternary_direct): New macro.
15920 (expand_ternary_optab_fn): Likewise.
15921 (direct_ternary_optab_supported_p): Likewise.
15922 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
15923 * builtins.c (fold_builtin_fma): Delete.
15924 (fold_builtin_3): Don't call it.
15925 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
15926 * expr.c (expand_expr_real_2): Likewise.
15927 * fold-const.c (operand_equal_p): Likewise.
15928 (fold_ternary_loc): Likewise.
15929 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
15930 * gimple.c (DEFTREECODE): Likewise.
15931 * gimplify.c (gimplify_expr): Likewise.
15932 * optabs-tree.c (optab_for_tree_code): Likewise.
15933 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
15934 * tree-eh.c (operation_could_trap_p): Likewise.
15935 (stmt_could_throw_1_p): Likewise.
15936 * tree-inline.c (estimate_operator_cost): Likewise.
15937 * tree-pretty-print.c (dump_generic_node): Likewise.
15938 (op_code_prio): Likewise.
15939 * tree-ssa-loop-im.c (stmt_cost): Likewise.
15940 * tree-ssa-operands.c (get_expr_operands): Likewise.
15941 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
15942 * fold-const-call.h (fold_fma): Delete.
15943 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
15944 CFN_FNMA and CFN_FNMS.
15945 (fold_fma): Delete.
15946 * genmatch.c (combined_fn): New enum.
15947 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
15948 (commutative_op): New function.
15949 (commutate): Use it. Handle more than 2 operands.
15950 (dt_operand::gen_gimple_expr): Use commutative_op.
15951 (parser::parse_expr): Allow :c to be used with non-binary
15952 operators if the commutative operand is known.
15953 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
15954 CFN_FMS, CFN_FNMA and CFN_FNMS.
15955 (backprop::process_assign_use): Remove FMA_EXPR handling.
15956 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
15957 (gen_hsa_fma): New function.
15958 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
15959 IFN_FNMA and IFN_FNMS.
15960 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
15961 * gimple-fold.h (follow_all_ssa_edges): Declare.
15962 * gimple-fold.c (follow_all_ssa_edges): New function.
15963 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
15964 gimple_build interface and use follow_all_ssa_edges to fold the result.
15965 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
15966 instead of checking for optabs directly.
15967 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
15968 rather than FMA_EXPRs.
15969 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
15970 call to IFN_FMA instead of an FMA_EXPR.
15972 2018-05-17 Jim Wilson <jimw@sifive.com>
15974 * expr.c (do_tablejump): When converting index to Pmode, if we have a
15975 sign extended promoted subreg, and the range does not have the sign bit
15976 set, then do a sign extend.
15978 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
15979 test, check for sign extended subreg and/or constant operands, and
15980 do a sign extend in that case.
15982 2018-05-17 Steve Ellcey <sellcey@cavium.com>
15984 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
15985 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
15987 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
15988 Change logics_shift_reg to logics_shift_imm.
15989 (thunderx2t99_fp_loadpair_basic): Delete.
15990 (thunderx2t99_fp_storepair_basic): Delete.
15991 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
15992 (thunderx2t99_asimd_polynomial): Delete.
15993 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
15994 and neon_fp_mul_d_scalar_q.
15995 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
15996 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
15997 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
15998 (thunderx2t99_asimd_lut): Add missing tbl types.
15999 (thunderx2t99_asimd_ext): Delete.
16000 (thunderx2t99_asimd_load1_1_mult): Delete.
16001 (thunderx2t99_asimd_load1_2_mult): Delete.
16002 (thunderx2t99_asimd_load1_ldp): New.
16003 (thunderx2t99_asimd_load1): New.
16004 (thunderx2t99_asimd_load2): Add missing *load2* types.
16005 (thunderx2t99_asimd_load3): New.
16006 (thunderx2t99_asimd_load4): New.
16007 (thunderx2t99_asimd_store1_1_mult): Delete.
16008 (thunderx2t99_asimd_store1_2_mult): Delete.
16009 (thunderx2t99_asimd_store2_mult): Delete.
16010 (thunderx2t99_asimd_store2_onelane): Delete.
16011 (thunderx2t99_asimd_store_stp): New.
16012 (thunderx2t99_asimd_store1): New.
16013 (thunderx2t99_asimd_store2): New.
16014 (thunderx2t99_asimd_store3): New.
16015 (thunderx2t99_asimd_store4): New.
16017 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
16019 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
16020 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
16022 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
16023 Segher Boessenkool <segher@kernel.crashing.org>
16026 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
16029 2018-05-17 Richard Biener <rguenther@suse.de>
16031 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
16032 for pruning loop and prune defs feeding only already visited PHIs.
16034 2018-05-17 Richard Biener <rguenther@suse.de>
16036 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
16038 2018-05-17 Bin Cheng <bin.cheng@arm.com>
16039 Richard Biener <rguenther@suse.de>
16041 PR tree-optimization/85793
16042 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
16043 for VMAT_ELEMENTWISE.
16045 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
16047 * internal-fn.h (lookup_internal_fn): Declare
16048 * internal-fn.c (lookup_internal_fn): New function.
16049 * gimple.c (gimple_build_call_from_tree): Handle calls to
16050 internal functions.
16051 * gimple-pretty-print.c (dump_gimple_call): Print "." before
16052 internal function names.
16053 * tree-pretty-print.c (dump_generic_node): Likewise.
16054 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
16056 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
16058 * gimple-fold.h (gimple_build): Make the function forms take
16059 combined_fn rather than built_in_function.
16060 (gimple_simplify): Likewise.
16061 * gimple-match-head.c (gimple_simplify): Likewise.
16062 * gimple-fold.c (gimple_build): Likewise.
16063 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
16064 rather than gimple_build_call_internal.
16065 (get_initial_defs_for_reduction): Likewise.
16066 (vect_create_epilog_for_reduction): Likewise.
16067 (vectorizable_live_operation): Likewise.
16069 2018-05-17 Martin Liska <mliska@suse.cz>
16071 * gimple-ssa-sprintf.c (format_directive): Do not use
16072 space in between 'G_' and '('.
16074 2018-05-17 Jakub Jelinek <jakub@redhat.com>
16077 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
16078 even if the mask is not all ones.
16081 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
16083 (ix86_gimple_fold_builtin): Likewise.
16086 * config/i386/i386.c: Include tree-vector-builder.h.
16087 (ix86_vector_shift_count): New function.
16088 (ix86_fold_builtin): Fold shift builtins by scalar count.
16089 (ix86_gimple_fold_builtin): Likewise.
16091 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
16092 _mm512_setzero): New intrinsics.
16094 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
16095 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16097 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
16098 code generation for cases where splatting a value is not useful.
16099 * simplify-rtx.c (simplify_ternary_operation): Simplify
16100 vec_merge across a vec_duplicate and a paradoxical subreg forming
16101 a vector mode to a vec_concat.
16103 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
16105 * config.gcc: Support "goldmont-plus".
16106 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
16108 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
16109 PROCESSOR_GOLDMONT_PLUS.
16110 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
16111 (processor_target_table): Add "goldmont-plus".
16112 (PTA_GOLDMONT_PLUS): Define.
16113 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
16114 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
16115 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
16116 (fold_builtin_cpu): Add "goldmont-plus".
16117 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
16118 (ix86_option_override_internal): Add "goldmont-plus".
16119 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
16120 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
16121 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
16122 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
16124 2018-05-17 Richard Biener <rguenther@suse.de>
16126 PR tree-optimization/85757
16127 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
16128 remove defs that only feed that PHI from further processing.
16130 2018-05-16 Jim Wilson <jimw@sifive.com>
16132 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
16134 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
16136 2018-05-16 Mark Wielaard <mark@klomp.org>
16138 * dwarf2out.c (count_index_strings): New function.
16139 (output_indirect_strings): Call count_index_strings and generate
16140 header for dwarf_version >= 5.
16142 2018-05-16 Mark Wielaard <mark@klomp.org>
16144 * dwarf2out.c (dwarf_FORM): New function.
16145 (set_indirect_string): Use dwarf_FORM.
16146 (reset_indirect_string): Likewise.
16147 (size_of_die): Likewise.
16148 (value_format): Likewise.
16149 (output_die): Likewise.
16150 (add_skeleton_AT_string): Likewise.
16151 (output_macinfo_op): Likewise.
16152 (index_string): Likewise.
16153 (output_index_string_offset): Likewise.
16154 (output_index_string): Likewise.
16155 (count_index_strings): Likewise.
16157 2018-05-16 Carl Love <cel@us.ibm.com>
16159 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
16160 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
16162 2018-05-16 Martin Jambor <mjambor@suse.cz>
16164 * ipa-prop.c (ipa_free_all_edge_args): Remove.
16165 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
16167 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
16169 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
16170 (fnma<mode>4): Likewise.
16171 (fms<mode>4): Likewise.
16172 (fnms<mode>4): Likewise.
16173 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
16174 (aarch64_fnma<mode>4): Likewise.
16175 (aarch64_fms<mode>4): Likewise.
16176 (aarch64_fnms<mode>4): Likewise.
16177 (aarch64_fnmadd<mode>4): Likewise.
16179 2018-05-16 Jason Merrill <jason@redhat.com>
16181 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
16183 2018-05-16 Richard Biener <rguenther@suse.de>
16185 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
16186 (dump_stmt_cost): Declare.
16187 (add_stmt_cost): Dump cost we add.
16188 (add_stmt_costs): New function.
16189 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
16190 No longer exported.
16191 (vect_analyze_stmt): Adjust prototype.
16192 (vectorizable_condition): Likewise.
16193 (vectorizable_live_operation): Likewise.
16194 (vectorizable_reduction): Likewise.
16195 (vectorizable_induction): Likewise.
16196 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
16197 cost vector to pass to vectorizable_ and record afterwards.
16198 (vect_model_reduction_cost): Take cost vector argument and adjust.
16199 (vect_model_induction_cost): Likewise.
16200 (vectorizable_reduction): Likewise.
16201 (vectorizable_induction): Likewise.
16202 (vectorizable_live_operation): Likewise.
16203 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
16204 SLP_TREE_NUMBER_OF_VEC_STMTS.
16205 (vect_analyze_slp_cost_1): Remove.
16206 (vect_analyze_slp_cost): Likewise.
16207 (vect_slp_analyze_node_operations): Take visited args and
16208 a target cost vector. Avoid processing already visited stmt sets.
16209 (vect_slp_analyze_operations): Use a local cost vector to gather
16210 costs and register those of non-discarded instances.
16211 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
16212 (vect_schedule_slp_instance): Remove copying of
16213 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
16215 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
16216 adding cost. Record cost entry location.
16217 (vect_prologue_cost_for_slp_op): Function to compute cost of
16218 a constant or invariant generated for SLP vect in the prologue,
16219 split out from vect_analyze_slp_cost_1.
16220 (vect_model_simple_cost): Make static. Adjust for SLP costing.
16221 (vect_model_promotion_demotion_cost): Likewise.
16222 (vect_model_store_cost): Likewise, make static.
16223 (vect_model_load_cost): Likewise.
16224 (vectorizable_bswap): Add cost vector arg and adjust.
16225 (vectorizable_call): Likewise.
16226 (vectorizable_simd_clone_call): Likewise.
16227 (vectorizable_conversion): Likewise.
16228 (vectorizable_assignment): Likewise.
16229 (vectorizable_shift): Likewise.
16230 (vectorizable_operation): Likewise.
16231 (vectorizable_store): Likewise.
16232 (vectorizable_load): Likewise.
16233 (vectorizable_condition): Likewise.
16234 (vectorizable_comparison): Likewise.
16235 (can_vectorize_live_stmts): Likewise.
16236 (vect_analyze_stmt): Likewise.
16237 (vect_transform_stmt): Adjust calls to vectorizable_*.
16238 * tree-vectorizer.c: Include gimple-pretty-print.h.
16239 (dump_stmt_cost): New function.
16241 2018-05-16 Richard Biener <rguenther@suse.de>
16243 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
16244 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
16245 * tree-ssa-dse.c: Include tree-ssa-loop.h.
16246 (check_name): New callback.
16247 (dse_classify_store): Track cycles via a visited bitmap of PHI
16248 defs and simplify handling of in-loop and across loop dead stores
16249 and properly fail for loop-variant refs. Handle byte-tracking with
16250 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
16253 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
16255 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
16256 (vect_get_mask_type_for_stmt): Likewise.
16257 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
16259 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
16260 to determine the statement's vector type and the vector type that
16261 should be used for calculating nunits. Deal with cases in which
16262 the type has to be deferred.
16263 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
16264 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
16265 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
16266 (vect_determine_vf_for_stmt): New functions, split out from...
16267 (vect_determine_vectorization_factor): ...here.
16268 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
16269 (vect_get_mask_type_for_stmt): New functions, split out from
16270 vect_determine_vectorization_factor.
16272 2018-05-16 Richard Biener <rguenther@suse.de>
16274 * tree-cfg.c (verify_gimple_assign_ternary): Properly
16275 verify the [VEC_]COND_EXPR embedded comparison.
16277 2018-05-15 Martin Sebor <msebor@redhat.com>
16279 PR tree-optimization/85753
16280 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
16281 RECORD_TYPE in addition to ARRAY_TYPE.
16283 2018-05-15 Martin Sebor <msebor@redhat.com>
16285 PR middle-end/85643
16286 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
16288 2018-05-15 Richard Biener <rguenther@suse.de>
16290 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
16291 add by_clobber_p one. Change algorithm to collect all defs
16292 representing uses we need to walk and try reducing them to
16293 a single one before failing.
16294 (dse_dom_walker::dse_optimize_stmt): Adjust.
16296 2018-05-13 Mark Wielaard <mark@klomp.org>
16298 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
16299 (size_of_loc_descr): Likewise.
16300 (output_loc_operands): Likewise.
16301 (output_loc_operands_raw): Likewise.
16302 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
16303 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
16304 (hash_loc_operands): Likewise.
16305 (compare_loc_operands): Likewise.
16307 2018-05-14 Mark Wielaard <mark@klomp.org>
16309 * dwarf2out.c (count_index_addrs): New function.
16310 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
16312 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16314 PR tree-optimization/83648
16315 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
16316 return value as malloc candidate.
16318 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16321 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
16322 param as true in call to suggest_attribute.
16324 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
16326 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
16327 -mreadonly-in-sdata.
16329 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16331 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
16333 (aarch64_crypto_aesd_fused): Likewise.
16335 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
16337 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
16338 (movsi_aarch64): Likewise.
16339 (load_pairsi): Likewise.
16340 (load_pairdi): Likewise.
16341 (store_pairsi): Likewise.
16342 (store_pairdi): Likewise.
16343 (load_pairsf): Likewise.
16344 (load_pairdf): Likewise.
16345 (store_pairsf): Likewise.
16346 (store_pairdf): Likewise.
16347 (zero_extend): Likewise.
16348 (trunc): Swap alternatives.
16349 (fcvt_target): Add '?' to prefer w over r.
16351 2018-05-14 Jakub Jelinek <jakub@redhat.com>
16354 * config/i386/i386.md: Disallow non-commutative arithmetics in
16355 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
16356 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
16357 in the peephole2 before it.
16359 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
16361 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
16362 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
16363 (ix86_handle_option): Handle -mcldemote.
16364 * config.gcc: New header.
16365 * config/i386/cldemoteintrin.h: New file.
16366 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
16367 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
16369 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
16370 OPTION_MASK_ISA_CLDEMOTE.
16371 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
16372 (ix86_valid_target_attribute_inner_p): Ditto.
16373 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
16374 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
16375 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
16376 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
16377 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
16379 * config/i386/i386.opt: Add -mcldemote.
16380 * config/i386/x86intrin.h: New header.
16381 * doc/invoke.texi: Add -mcldemote.
16383 2018-05-14 Richard Biener <rguenther@suse.de>
16385 * doc/match-and-simplify.texi: Adjust :s documentation.
16387 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
16389 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
16390 intended memcpy size.
16391 (REORDER_45): Likewise.
16393 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
16395 * sort.cc: New file.
16396 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
16397 * vec.c (qsort_chk): Use gcc_qsort.
16398 * Makefile.in (OBJS-libcommon): Add sort.o.
16399 (build/sort.o): New target. Use it...
16400 (BUILD_RTL): ... here, and...
16401 (build/gencfn-macros): ... here, and...
16402 (build/genmatch): ... here.
16404 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
16405 Chung-Ju Wu <jasonwucj@gmail.com>
16407 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
16408 * config/nds32/nds32-graywolf.md: New file.
16409 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
16410 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
16412 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
16413 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
16414 * config/nds32/nds32.md (pipeline_model): Add graywolf.
16415 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
16416 * config/nds32/pipelines.md: Include n15 settings.
16418 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
16419 Chung-Ju Wu <jasonwucj@gmail.com>
16421 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
16422 * config/nds32/nds32-n13.md: New file.
16423 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
16424 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
16426 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
16427 * config/nds32/nds32.md (pipeline_model): Add n13.
16428 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
16429 * config/nds32/pipelines.md: Include n13 settings.
16431 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
16432 Chung-Ju Wu <jasonwucj@gmail.com>
16434 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
16435 * config/nds32/nds32-n10.md: New file.
16436 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
16437 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
16439 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
16440 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
16441 * config/nds32/nds32.md (pipeline_model): Add n10.
16442 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
16443 * config/nds32/pipelines.md: Include n10 settings.
16445 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
16446 Kito Cheng <kito.cheng@gmail.com>
16447 Chung-Ju Wu <jasonwucj@gmail.com>
16449 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
16450 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
16451 Add enum values for DSP extension instructions.
16452 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
16454 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
16455 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
16456 New code iterators.
16457 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
16458 * config/nds32/nds32-dspext.md: New file for DSP implementation.
16459 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
16460 * config/nds32/nds32-intrinsic.md: Likewise.
16461 * config/nds32/nds32_intrinsic.h: Likewise.
16462 * config/nds32/nds32-md-auxiliary.c: Likewise.
16463 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
16464 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
16465 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
16466 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
16467 * config/nds32/nds32-protos.h: New declarations for DSP extension.
16468 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
16469 TYPE_DMAC in switch statement.
16470 * config/nds32/nds32.c: New checking and implementation for DSP
16471 extension instructions.
16472 * config/nds32/nds32.h: Likewise.
16473 * config/nds32/nds32.md: Likewise.
16474 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
16475 * config/nds32/predicates.md: Implement new predicates for DSP
16478 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
16480 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
16481 Reformat alternatives and attributes so it is easier to identify
16482 which constraints/attributes go with which instruction.
16483 (mov<mode>_hardfloat32, FMOVE64): Likewise.
16484 (mov<mode>_softfloat32, FMOVE64): Likewise.
16485 (mov<mode>_hardfloat64, FMOVE64): Likewise.
16486 (mov<mode>_softfloat64, FMOVE64): Likewise.
16488 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
16490 * doc/extend.texi (PowerPC Built-in Functions): Rename this
16492 (Basic PowerPC Built-in Functions): The new name of the
16493 subsection previously known as "PowerPC Built-in Functions".
16494 (Basic PowerPC Built-in Functions Available on all Configurations):
16496 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
16497 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
16498 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
16499 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
16501 2018-05-11 Martin Jambor <mjambor@suse.cz>
16504 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
16507 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
16510 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
16512 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
16514 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
16515 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
16516 (ix86_handle_option): Handle -mwaitpkg.
16517 * config.gcc: New header.
16518 * config/i386/cpuid.h (bit_WAITPKG): New bit.
16519 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
16520 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
16522 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
16523 OPTION_MASK_ISA_WAITPKG.
16524 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
16525 (ix86_option_override_internal): Add PTA_WAITPKG.
16526 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
16527 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
16528 IX86_BUILTIN_TPAUSE.
16529 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
16530 __builtin_ia32_umwait and __builtin_ia32_tpause.
16531 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
16532 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
16533 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
16534 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
16535 UNSPECV_TPAUSE): New.
16536 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
16537 * config/i386/i386.opt: Add -mwaitpkg.
16538 * config/i386/waitpkgintrin.h: New file.
16539 * config/i386/x86intrin.h: New header.
16540 * doc/invoke.texi: Add -mwaitpkg.
16542 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
16545 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
16547 (cortex-m0): Use armv6s-m isa.
16548 (cortex-m0plus): Likewise.
16549 (cortex-m1): Likewise.
16550 (cortex-m0.small-multiply): Likewise.
16551 (cortex-m0plus.small-multiply): Likewise.
16552 (cortex-m1.small-multiply): Likewise.
16554 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
16555 Jakub Jelinek <jakub@redhat.com>
16557 PR tree-optimization/85692
16558 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
16559 source permute as well.
16561 2018-05-11 Martin Liska <mliska@suse.cz>
16564 * doc/extend.texi: Document LLVM style format for no_sanitize
16567 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
16569 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
16570 mode_supports_vsx_dform_quad to mode_supports_dq_form.
16571 (mode_supports_vsx_dform_quad): Likewise.
16572 (mode_supports_vmx_dform): Move these functions to be next to the
16573 other mode_supports functions.
16574 (mode_supports_dq_form): Likewise.
16575 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
16576 mode_supports_dq_form.
16577 (reg_offset_addressing_ok_p): Likewise.
16578 (offsettable_ok_by_alignment): Likewise.
16579 (rs6000_legitimate_offset_address_p): Likewise.
16580 (legitimate_lo_sum_address_p): Likewise.
16581 (rs6000_legitimize_address): Likewise.
16582 (rs6000_legitimize_reload_address): Likewise.
16583 (rs6000_secondary_reload_inner): Likewise.
16584 (rs6000_preferred_reload_class): Likewise.
16585 (rs6000_output_move_128bit): Likewise.
16587 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
16589 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
16590 Generate SImode target register for null target.
16591 <case IX86_BUILTIN_XGETBV>: Ditto.
16592 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
16593 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
16595 2018-05-10 Carl Love <cel@us.ibm.com>
16597 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
16598 dcbtt and dcbtstt if operands[2] is 0.
16600 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
16603 * config/i386/sse.md (usadv64qi): New expander.
16605 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
16607 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
16608 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
16609 -maltivec=be support.
16610 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
16611 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
16612 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
16613 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
16614 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
16615 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
16616 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
16617 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
16618 altivec_vsumsws): Adjust.
16619 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
16620 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
16621 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
16623 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
16624 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
16625 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
16626 (altivec_lve<VI_char>x): Delete expand.
16627 (*altivec_lve<VI_char>x_internal): Rename to...
16628 (altivec_lve<VI_char>x): ... this.
16629 (altivec_lvxl_<mode>): Delete expand.
16630 (*altivec_lvxl_<mode>_internal): Rename to ...
16631 (altivec_lvxl_<mode>): ... this.
16632 (altivec_stvxl_<mode>): Delete expand.
16633 (*altivec_stvxl_<mode>_internal): Rename to ...
16634 (altivec_stvxl_<mode>): ... this.
16635 (altivec_stve<VI_char>x): Delete expand.
16636 (*altivec_stve<VI_char>x_internal): Rename to ...
16637 (altivec_stve<VI_char>x): ... this.
16638 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
16639 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
16640 reduc_plus_scal_<mode>): Adjust.
16641 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
16643 (rs6000_cpu_cpp_builtins): Adjust.
16644 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
16645 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
16646 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
16647 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
16648 -maltivec=be support.
16649 (rs6000_split_vec_extract_var): Adjust.
16650 (rs6000_split_v4si_init): Adjust.
16651 (swap_selector_for_mode): Delete.
16652 (altivec_expand_lvx_be, altivec_expand_stvx_be,
16653 altivec_expand_stvex_be): Delete.
16654 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
16655 -maltivec=be support.
16656 (rs6000_gimple_fold_builtin): Ditto.
16657 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
16659 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
16660 (TARGET_DIRECT_MOVE_64BIT): Adjust.
16661 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
16662 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
16663 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
16664 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
16665 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
16666 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
16667 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
16668 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
16669 anonymous split): Adjust.
16670 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
16671 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
16673 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
16675 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
16676 when --with-gxx-include-dir is also specified.
16677 * configure: Regenerate.
16679 2018-05-09 Jim Wilson <jimw@sifive.com>
16682 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
16683 * config/riscv/t-withmultilib: New.
16684 * config/riscv/withmultilib.h: New.
16685 * doc/install.texi: Document RISC-V --with-multilib-list support.
16687 2018-05-09 Richard Biener <rguenther@suse.de>
16689 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
16691 (vect_bb_vectorization_profitable_p): Adjust. Compute
16692 actual scalar cost using the cost vector and the add_stmt_cost
16695 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
16697 PR rtl-optimization/85645
16698 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
16699 in the REG_CFA_REGISTER note for LR, don't leave it empty.
16701 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
16703 PR rtl-optimization/85645
16704 * shrink-wrap.c (spread_components): Return a boolean saying if
16705 anything was changed.
16706 (try_shrink_wrapping_separate): Iterate spread_components until
16707 nothing changes anymore.
16709 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
16711 PR rtl-optimization/85645
16712 * regrename.c (build_def_use): Also kill the chains that include the
16713 destination of a REG_CFA_REGISTER note.
16715 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
16717 PR rtl-optimization/85645
16718 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
16719 insn that has a REG_CFA_REGISTER note.
16721 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
16723 * cfgexpand.c (expand_clobber): New function.
16724 (expand_gimple_stmt_1): Use it.
16725 * tree-vect-stmts.c (vect_clobber_variable): New function,
16727 (vectorizable_simd_clone_call): ...here.
16728 (vectorizable_store): Emit a clobber either side of an
16729 IFN_STORE_LANES sequence.
16730 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
16732 2018-05-09 Tom de Vries <tom@codesourcery.com>
16735 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
16736 (define_insn "trap_if_false"): Add exit after trap.
16738 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
16740 PR rtl-optimization/85638
16741 * bb-reorder.c: Include common/common-target.h.
16742 (create_forwarder_block): New function extracted from...
16743 (fix_up_crossing_landing_pad): ...here. Rename into...
16744 (dw2_fix_up_crossing_landing_pad): ...this.
16745 (sjlj_fix_up_crossing_landing_pad): New function.
16746 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
16747 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
16748 from both partitions and exit the loop after one iteration.
16750 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
16753 * doc/extend.texi (PowerPC Built-in Functions): Rename this
16755 (Basic PowerPC Built-in Functions): The new name of the
16756 subsection previously known as "PowerPC Built-in Functions".
16757 (Basic PowerPC Built-in Functions Available on all Configurations):
16759 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
16761 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
16763 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
16765 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
16768 2018-05-08 Jim Wilson <jimw@sifive.com>
16770 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
16771 (LD_EMUL_SUFFIX): New.
16772 (LINK_SPEC): Use it.
16774 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
16776 * doc/extend.texi (PowerPC Built-in Functions): Rename this
16778 (Basic PowerPC Built-in Functions): The new name of the
16779 subsection previously known as "PowerPC Built-in Functions".
16780 (Basic PowerPC Built-in Functions Available on all Configurations):
16782 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
16784 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
16786 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
16788 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
16791 2018-05-08 Jakub Jelinek <jakub@redhat.com>
16794 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
16795 after cmpelim optimization.
16797 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
16799 * config.gcc: Support "goldmont".
16800 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
16801 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
16802 PROCESSOR_GOLDMONT.
16803 * config/i386/i386.c (m_GOLDMONT): Define.
16804 (processor_target_table): Add "goldmont".
16805 (PTA_GOLDMONT): Define.
16806 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
16807 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
16808 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
16809 (fold_builtin_cpu): Add "goldmont".
16810 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
16811 (ix86_option_override_internal): Add "goldmont".
16812 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
16813 (processor_type): Add PROCESSOR_GOLDMONT.
16814 * config/i386/i386.md: Add CPU "glm".
16815 * config/i386/glm.md: New file.
16816 * config/i386/x86-tune.def: Add m_GOLDMONT.
16817 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
16819 2018-05-08 Jakub Jelinek <jakub@redhat.com>
16822 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
16824 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
16825 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
16826 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
16829 * config/i386/i386.c (ix86_fold_builtin): Handle
16830 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
16833 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
16834 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
16836 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
16839 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
16840 (check_arch): Likewise.
16841 (check_fpu): Return the result rather than printing it.
16842 (end arch): Fix operator precedence.
16843 (end cpu): Likewise.
16844 (END): Print the result from check_fpu.
16846 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
16847 Alan Hayward <alan.hayward@arm.com>
16848 David Sherwood <david.sherwood@arm.com>
16850 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
16851 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
16852 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
16853 (*fcmuo<mode>_and): New patterns.
16855 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
16857 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
16858 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
16859 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
16860 (cmp_op, sve_imm_con): New code attributes.
16861 (SVE_COND_INT_CMP, imm_con): Delete.
16862 (cmp_op): Remove above unspecs from int attribute.
16863 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
16865 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
16866 comparison-specific unspecs.
16867 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
16868 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
16869 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
16870 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
16871 (*vec_fcm<cmp_op><mode>): Rename to...
16872 (*fcm<cmp_op><mode>): ...this and adjust likewise.
16873 (*vec_fcmuo<mode>): Rename to...
16874 (*fcmuo<mode>): ...this and adjust likewise.
16875 (*pred_fcm<cmp_op><mode>): New pattern.
16876 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
16877 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
16879 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
16881 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
16882 (aarch64_emit_sve_predicated_cond): New function.
16883 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
16884 (aarch64_emit_unspec_cond_or): Replace with...
16885 (aarch64_emit_sve_or_conds): ...this new function. Use
16886 aarch64_emit_sve_ptrue_op for the individual comparisons and
16887 aarch64_emit_binop to OR them together.
16888 (aarch64_emit_inverted_unspec_cond): Replace with...
16889 (aarch64_emit_sve_inverted_cond): ...this new function. Use
16890 aarch64_emit_sve_ptrue_op for the comparison and
16891 aarch64_emit_unop to invert the result.
16892 (aarch64_expand_sve_vec_cmp_float): Update after the above
16893 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
16895 2018-05-07 Nathan Sidwell <nathan@acm.org>
16897 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
16898 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
16899 (Backwards Compatibility): Likewise.
16901 2018-05-07 Luis Machado <luis.machado@linaro.org>
16905 2018-05-07 Luis Machado <luis.machado@linaro.org>
16907 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
16908 <prefetch_dynamic_strides>: New const bool field.
16909 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
16910 prefetch_dynamic_strides.
16911 (exynosm1_prefetch_tune): Likewise.
16912 (thunderxt88_prefetch_tune): Likewise.
16913 (thunderx_prefetch_tune): Likewise.
16914 (thunderx2t99_prefetch_tune): Likewise.
16915 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
16917 (aarch64_override_options_internal): Update to set
16918 PARAM_PREFETCH_DYNAMIC_STRIDES.
16919 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
16920 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
16921 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
16922 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
16923 prefetch-dynamic-strides setting.
16925 2018-05-07 Luis Machado <luis.machado@linaro.org>
16927 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
16928 <minimum_stride>: New const int field.
16929 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
16930 minimum_stride field.
16931 (exynosm1_prefetch_tune): Likewise.
16932 (thunderxt88_prefetch_tune): Likewise.
16933 (thunderx_prefetch_tune): Likewise.
16934 (thunderx2t99_prefetch_tune): Likewise.
16935 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
16936 (aarch64_override_options_internal): Update to set
16937 PARAM_PREFETCH_MINIMUM_STRIDE.
16938 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
16939 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
16940 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
16941 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
16942 stride is constant and is below the minimum stride threshold.
16944 2018-05-07 Luis Machado <luis.machado@linaro.org>
16946 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
16949 2018-05-07 Luis Machado <luis.machado@linaro.org>
16951 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
16952 <prefetch_dynamic_strides>: New const bool field.
16953 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
16954 prefetch_dynamic_strides.
16955 (exynosm1_prefetch_tune): Likewise.
16956 (thunderxt88_prefetch_tune): Likewise.
16957 (thunderx_prefetch_tune): Likewise.
16958 (thunderx2t99_prefetch_tune): Likewise.
16959 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
16961 (aarch64_override_options_internal): Update to set
16962 PARAM_PREFETCH_DYNAMIC_STRIDES.
16963 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
16964 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
16965 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
16966 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
16967 prefetch-dynamic-strides setting.
16969 2018-05-07 Luis Machado <luis.machado@linaro.org>
16971 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
16972 <minimum_stride>: New const int field.
16973 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
16974 minimum_stride field.
16975 (exynosm1_prefetch_tune): Likewise.
16976 (thunderxt88_prefetch_tune): Likewise.
16977 (thunderx_prefetch_tune): Likewise.
16978 (thunderx2t99_prefetch_tune): Likewise.
16979 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
16980 (aarch64_override_options_internal): Update to set
16981 PARAM_PREFETCH_MINIMUM_STRIDE.
16982 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
16983 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
16984 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
16985 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
16986 stride is constant and is below the minimum stride threshold.
16988 2018-05-06 Jakub Jelinek <jakub@redhat.com>
16991 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
16992 the type is addressable. Don't force op into register if it has
16995 2018-05-05 Roland McGrath <mcgrathr@google.com>
16998 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
16999 any section for which we don't know a specific type it should have,
17000 regardless of name. Previously this was done only for the exact
17001 names ".init_array", ".fini_array", and ".preinit_array".
17002 (default_elf_asm_named_section): Add comment about
17003 relationship with default_section_type_flags and SECTION_NOTYPE.
17004 (get_section): Don't consider it a type conflict if one side has
17005 SECTION_NOTYPE and the other doesn't, as long as neither has the
17006 SECTION_BSS et al used in the default_section_type_flags logic.
17008 2018-05-05 Tom de Vries <tom@codesourcery.com>
17011 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
17012 (workaround_barsyncs): New function.
17013 (nvptx_reorg): Use workaround_barsyncs.
17014 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
17015 (define_expand "nvptx_membar_cta"): New define_expand.
17016 (define_insn "*nvptx_membar_cta"): New insn.
17018 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
17020 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
17021 To improve optimization opportunities.
17022 * builtin-types.def: The new needed builtin types for the above.
17024 2018-05-04 Richard Biener <rguenther@suse.de>
17026 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
17027 * gimple-ssa-store-merging.c
17028 (imm_store_chain_info::output_merged_store): Remove redundant create,
17029 release split_store vector contents on failure.
17030 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
17031 scalar stmt vector on cache hit.
17033 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
17035 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
17037 * config.gcc (powerpc-xilinx-eabi*): Remove.
17038 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
17040 (fusion_addis_mem_combo_load): Ditto.
17041 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
17043 (rs6000_cpu_cpp_builtins): Ditto.
17044 * config/rs6000/rs6000-linux.c
17045 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
17046 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
17047 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
17049 (rs6000_setup_reg_addr_masks): Ditto.
17050 (rs6000_init_hard_regno_mode_ok): Ditto.
17051 (rs6000_option_override_internal): Ditto.
17052 (legitimate_lo_sum_address_p): Ditto.
17053 (rs6000_legitimize_address): Ditto.
17054 (rs6000_legitimize_reload_address): Ditto.
17055 (rs6000_legitimate_address_p): Ditto.
17056 (abi_v4_pass_in_fpr): Ditto.
17057 (setup_incoming_varargs): Ditto.
17058 (rs6000_gimplify_va_arg): Ditto.
17059 (rs6000_split_multireg_move): Ditto.
17060 (rs6000_savres_strategy): Ditto.
17061 (rs6000_emit_prologue_components): Ditto.
17062 (rs6000_emit_epilogue_components): Ditto.
17063 (rs6000_emit_prologue): Ditto.
17064 (rs6000_emit_epilogue): Ditto.
17065 (rs6000_elf_file_end): Ditto.
17066 (rs6000_function_value): Ditto.
17067 (rs6000_libcall_value): Ditto.
17068 * config/rs6000/rs6000.h: Ditto.
17069 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
17070 (TARGET_MINMAX): ... this. New.
17071 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
17072 * config/rs6000/rs6000.md: Remove Xilinx FP support.
17073 (*movsi_internal1_single): Delete.
17074 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
17075 mfpu=, mxilinx-fpu): Delete.
17076 * config/rs6000/singlefp.h: Delete.
17077 * config/rs6000/sysv4.h: Remove Xilinx FP support.
17078 * config/rs6000/t-rs6000: Ditto.
17079 * config/rs6000/t-xilinx: Delete.
17080 * config/rs6000/titan.md: Adjust for fp_type removal.
17081 * config/rs6000/vsx.md: Remove Xilinx FP support.
17082 (VStype_simple): Delete.
17083 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
17084 * config/rs6000/xfpu.h: Delete.
17085 * config/rs6000/xfpu.md: Delete.
17086 * config/rs6000/xilinx.h: Delete.
17087 * config/rs6000/xilinx.opt: Delete.
17088 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
17089 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
17091 2018-05-04 Tom de Vries <tom@codesourcery.com>
17094 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
17097 2018-05-04 Richard Biener <rguenther@suse.de>
17099 PR middle-end/85627
17100 * tree-complex.c (update_complex_assignment): We are always in SSA form.
17101 (expand_complex_div_wide): Likewise.
17102 (expand_complex_operations_1): Likewise.
17103 (expand_complex_libcall): Preserve EH info of the original stmt.
17104 (tree_lower_complex): Handle removed blocks.
17105 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
17106 on complex multiplication and division libcall builtins.
17108 2018-05-04 Richard Biener <rguenther@suse.de>
17110 PR middle-end/85574
17111 * fold-const.c (negate_expr_p): Restrict negation of operand
17112 zero of a division to when we know that can happen without
17114 (fold_negate_expr_1): Likewise.
17116 2018-05-04 Jakub Jelinek <jakub@redhat.com>
17119 * real.h (real_nextafter): Declare.
17120 * real.c (real_nextafter): New function.
17121 * fold-const-call.c (fold_const_nextafter): New function.
17122 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
17123 CASE_CFN_NEXTTOWARD.
17124 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
17125 even when arg1_mode is different from arg0_mode.
17127 2018-05-03 Nathan Sidwell <nathan@acm.org>
17129 * doc/extend.texi (Deprecated Features): Remove
17130 -ffriend-injection.
17131 (Backwards Compatibility): Likewise.
17132 * doc/invoke.texi (C++ Language Options): Likewise.
17133 (C++ Dialect Options): Likewise.
17135 2018-05-03 Jakub Jelinek <jakub@redhat.com>
17138 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
17139 _mm512_mask_mullox_epi64): New intrinsics.
17141 2018-05-03 Tom de Vries <tom@codesourcery.com>
17144 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
17145 dump files): Add offload-tree.
17147 2018-05-03 Richard Biener <rguenther@suse.de>
17149 PR tree-optimization/85615
17150 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
17151 to loops not nested in BBs loop father to avoid creating multi-entry
17154 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17156 PR tree-optimization/70291
17157 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
17158 arguments. Change return type to tree. Emit libcall as a new
17159 statement rather than replacing existing one when inplace_p is true.
17160 (expand_complex_multiplication_components): New function.
17161 (expand_complex_multiplication): Expand floating-point complex
17162 multiplication using the above.
17163 (expand_complex_division): Rename inner_type parameter to type.
17164 Update expand_complex_libcall call-site.
17165 (expand_complex_operations_1): Update expand_complex_multiplication
17166 and expand_complex_division call-sites.
17168 2018-05-02 Jakub Jelinek <jakub@redhat.com>
17171 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
17172 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
17173 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
17174 the highest significant bit of the shift count mask is clear. In
17175 check whether and[sq]i3 is needed verify that all significant bits
17176 of the shift count other than the highest are set.
17178 2018-05-02 Tom de Vries <tom@codesourcery.com>
17181 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
17182 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
17183 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
17184 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
17185 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
17186 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
17187 __builtin_goacc_parlevel_size.
17189 2018-05-02 Richard Biener <rguenther@suse.de>
17191 PR tree-optimization/85597
17192 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
17193 do not use split vect_get_vec_defs call but call vect_get_slp_defs
17196 2018-05-02 Tom de Vries <tom@codesourcery.com>
17199 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
17200 dump files): Add ltrans-tree.
17202 2018-05-02 Tom de Vries <tom@codesourcery.com>
17205 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
17206 dump files): Add wpa-ipa.
17208 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
17210 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
17211 powerpc*-*-linux*paired* target.
17212 * config/rs6000/750cl.h: Delete.
17213 * config/rs6000/paired.h: Delete.
17214 * config/rs6000/paired.md: Delete.
17215 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
17217 * config/rs6000/rs6000-builtin.def: Remove paired float support.
17218 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
17219 comment. Remove paired float support.
17220 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
17221 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
17223 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
17224 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
17226 * config/rs6000/rs6000.c: Remove paired float support.
17227 (paired_expand_vector_init, paired_expand_vector_move,
17228 paired_emit_vector_compare, paired_emit_vector_cond_expr,
17229 (paired_expand_lv_builtin, paired_expand_stv_builtin,
17230 paired_expand_builtin, paired_expand_predicate_builtin,
17231 paired_init_builtins): Delete.
17232 * config/rs6000/rs6000.h: Remove paired float support.
17233 * config/rs6000/rs6000.md: Remove paired float support.
17234 (move_from_CR_ov_bit): Delete.
17235 * config/rs6000/rs6000.opt (mpaired): Delete.
17236 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
17237 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
17239 2018-05-02 Richard Biener <rguenther@suse.de>
17241 PR middle-end/85567
17242 * gimplify.c (gimplify_save_expr): When in SSA form allow
17243 SAVE_EXPRs to compute to SSA vars.
17245 2018-05-02 Jakub Jelinek <jakub@redhat.com>
17248 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
17249 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
17250 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
17251 clobber operands[2], instead use a new pseudo. Formatting fixes.
17253 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
17255 PR tree-optimization/85586
17256 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
17257 exit early for statements in the same group if the accesses are
17260 2018-05-02 Tom de Vries <tom@codesourcery.com>
17263 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
17266 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
17268 PR tree-optimization/85143
17269 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
17271 2018-05-01 Tom de Vries <tom@codesourcery.com>
17274 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
17275 not found" error message.
17277 2018-05-01 Tom de Vries <tom@codesourcery.com>
17280 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
17281 * vec.c (test_ordered_remove_if): New function.
17282 (vec_c_tests): Call test_ordered_remove_if.
17283 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
17284 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
17285 * tree-vect-patterns.c (vect_pattern_recog_1): Use
17286 VEC_ORDERED_REMOVE_IF.
17288 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17290 PR tree-optimization/82665
17291 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
17292 pointer subtraction where arguments come from a memchr call.
17294 2018-05-01 Jakub Jelinek <jakub@redhat.com>
17296 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
17297 --push-state --as-needed and --pop-state instead of --as-needed and
17298 --no-as-needed if ld supports it.
17299 * configure: Regenerated.
17302 * doc/install.texi2html: Replace _002d with - and _002a with * in
17303 generated html files using sed.
17305 2018-04-30 David Malcolm <dmalcolm@redhat.com>
17308 * gcc-rich-location.c (blank_line_before_p): New function.
17309 (use_new_line): New function.
17310 (gcc_rich_location::add_fixit_insert_formatted): New function.
17311 * gcc-rich-location.h
17312 (gcc_rich_location::add_fixit_insert_formatted): New function.
17314 2018-04-30 David Malcolm <dmalcolm@redhat.com>
17316 * selftest.c (assert_streq): Rename "expected" and "actual" to
17317 "val1" and "val2". Extend NULL-handling to cover both inputs
17318 symmetrically, while still requiring both to be non-NULL for a pass.
17319 * selftest.h (assert_streq): Rename "expected" and "actual" to
17321 (ASSERT_EQ): Likewise.
17322 (ASSERT_EQ_AT): Likewise.
17323 (ASSERT_KNOWN_EQ): Likewise.
17324 (ASSERT_KNOWN_EQ_AT): Likewise.
17325 (ASSERT_NE): Likewise.
17326 (ASSERT_MAYBE_NE): Likewise.
17327 (ASSERT_MAYBE_NE_AT): Likewise.
17328 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
17329 the assertion to pass.
17330 (ASSERT_STREQ_AT): Likewise.
17332 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
17334 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
17335 interaction with -pie.
17337 2018-04-30 David Malcolm <dmalcolm@redhat.com>
17339 * selftest.h: Fix alphabetization of per-source-file selftest
17342 2018-04-30 Jason Merrill <jason@redhat.com>
17344 PR c++/61982 - dead stores to destroyed objects.
17345 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
17348 2018-04-30 Jason Merrill <jason@redhat.com>
17350 * tree.c (build_clobber): New.
17351 * tree.h: Declare it.
17352 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
17354 2018-04-30 David Malcolm <dmalcolm@redhat.com>
17356 * diagnostic-show-locus.c (layout::layout): Update for
17357 location_get_source_line returning a char_span.
17358 (struct char_span): Move to input.h.
17359 (struct correction): Update for fields in char_span becoming
17361 (struct source_line): Update for location_get_source_line
17362 returning a char_span.
17363 (layout::print_line): Likewise.
17364 * edit-context.c (edited_file::print_content): Likewise.
17365 (edited_file::print_diff_hunk): Likewise.
17366 (edited_file::print_run_of_changed_lines): Likewise.
17367 (edited_file::get_num_lines): Likewise.
17368 (edited_line::edited_line): Likewise.
17369 * final.c (asm_show_source): Likewise.
17370 * input.c (location_get_source_line): Convert return type
17371 from const char * to char_span, losing the final "line_len"
17373 (dump_location_info): Update for the above.
17374 (get_substring_ranges_for_loc): Likewise. Use a char_span
17375 when handling the literal within the line.
17376 (test_reading_source_line): Update for location_get_source_line
17377 returning a char_span.
17378 * input.h (class char_span): Move here from
17379 diagnostic-show-locus.c, converting from a struct to a class.
17380 Make data members private.
17381 (char_span::operator bool): New.
17382 (char_span::length): New.
17383 (char_span::get_buffer): New.
17384 (char_span::operator[]): New.
17385 (char_span::subspan): Make const.
17386 (char_span::xstrdup): New.
17387 (location_get_source_line): Convert return type from const char *
17388 to char_span, losing the final "line_size" param.
17390 2018-04-30 Jan Hubicka <jh@suse.cz>
17392 * lto-wrapper.c (ltrans_priorities): New static var.
17393 (cmp_priority): New.
17394 (run_gcc): Read priorities and if doing parallel build order
17395 the Makefile by them.
17397 2018-04-30 David Malcolm <dmalcolm@redhat.com>
17399 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
17401 2018-04-30 Richard Biener <rguenther@suse.de>
17403 * tree-cfg.c (verify_address): Remove base argument, add
17404 flag whether to check TREE_ADDRESSABLE and do that.
17405 (verify_expr): Remove.
17406 (verify_types_in_gimple_reference): Add pieces from verify_expr.
17407 (verify_gimple_assign_single): Likewise.
17408 (verify_gimple_switch): Likewise.
17409 (verify_expr_location_1): Dereference tp once. Add (disabled)
17410 piece from verify_expr.
17411 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
17413 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
17415 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
17417 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
17419 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
17420 (small_data_pattern): Likewise.
17421 (arc_rewrite_small_data): Likewise.
17422 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
17423 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
17424 (get_symbol_alignment): New function.
17425 (legitimate_small_data_address_p): Likewise.
17426 (legitimate_scaled_address): Update, call
17427 legitimate_small_data_address_p.
17428 (output_sdata): New static variable.
17429 (arc_print_operand): Update how we handle small data operands.
17430 (arc_print_operand_address): Likewise.
17431 (arc_legitimate_address_p): Update, use
17432 legitimate_small_data_address_p.
17433 (arc_rewrite_small_data_p): Remove.
17434 (arc_rewrite_small_data_1): Likewise.
17435 (arc_rewrite_small_data): Likewise.
17436 (small_data_pattern): Likewise.
17437 (compact_sda_memory_operand): Update to use
17438 legitimate_small_data_address_p and get_symbol_alignment.
17439 (prepare_move_operands): Don't rewite sdata pattern.
17440 (prepare_extend_operands): Remove.
17441 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
17443 (zero_extendqisi2): Likewise.
17444 (zero_extendhisi2): Likewise.
17445 (extendqihi2): Likewise.
17446 (extendqisi2): Likewise.
17447 (extendhisi2): Likewise.
17448 (addsi3): Likewise.
17449 (subsi3): Likewise.
17450 (andsi3): Likewise.
17451 * config/arc/constraints.md (Usd): Change it to memory constraint.
17453 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
17455 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
17456 as source of std instructions.
17457 * config/arc/arc.md (movsi_insn): Update pattern predicate to
17458 allow 6-bit constants as source for store instructions.
17459 (movdi_insn): Update instruction pattern to allow 6-bit constants
17460 as source for store instructions.
17462 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
17464 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
17466 2018-04-30 Nathan Sidwell <nathan@acm.org>
17467 Sandra Loosemore <sandra@codesourcery.com>
17469 * dumpfile.c (dump_open): Allow '-' for stdout.
17470 * doc/invoke.texi (Developer Options): Document dump filename
17471 determination early. Document stdin/stdout selection.
17473 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
17475 Microblaze Target: PIC data text relative
17477 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
17478 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
17480 * config/microblaze/microblaze.h (microblaze_constant_address_p):
17481 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
17482 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
17483 New addressing mode for data-text relative position indepenedent code.
17484 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
17485 'ADDRESS_SYMBOLIC_TXT_REL'.
17486 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
17487 (microblaze_legitimate_pic_operand): Exclude function calls from
17488 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
17489 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
17491 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
17492 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
17493 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
17494 for 'address + offset'.
17495 (microblaze_expand_prologue): Add new function prologue call for
17497 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
17498 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
17499 table in case of TARGET_PIC_DATA_TEXT_REL.
17500 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
17501 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
17502 Add new macros 'UNSPEC_TEXT',
17503 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
17504 + exclude function calls from 'UNSPEC_PLT' in case of data text
17506 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
17507 new target hook for generating address diff vector tables in case of
17509 * doc/tm.texi : Regenerate.
17510 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
17511 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
17512 of addr diff vector generation.
17513 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
17514 target hook definition.
17515 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
17516 Add default function for generate_pic_addr_diff_vec -> flag_pic.
17517 * doc/invoke.texi (Add new pic option): Add new microblaze pic
17518 option for data text relative.
17520 2018-04-30 Richard Biener <rguenther@suse.de>
17522 * tree-chrec.h (evolution_function_is_constant_p): Remove
17524 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
17526 2018-04-30 Richard Biener <rguenther@suse.de>
17529 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
17531 2018-04-30 Richard Biener <rguenther@suse.de>
17533 PR tree-optimization/28364
17534 PR tree-optimization/85275
17535 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
17536 copying first exit test.
17538 2018-04-28 Mark Wielaard <mark@klomp.org>
17540 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
17541 dwarf_version >= 5.
17542 (dwarf_AT): Handle DW_AT_addr_base.
17543 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
17545 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
17548 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
17549 (*ashl<dwi>3_doubleword_mask_1): Ditto.
17550 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
17551 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
17553 2018-04-28 Richard Biener <rguenther@suse.de>
17555 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
17556 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
17557 to reflect use. Only add interesting stmts.
17559 2018-04-27 Martin Jambor <mjambor@suse.cz>
17562 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
17563 the jump function allows for passing through aggregate values.
17565 2018-04-27 David Malcolm <dmalcolm@redhat.com>
17567 * input.h (in_system_header_at): Convert from macro to inline
17569 (from_macro_expansion_at): Likewise.
17570 (from_macro_definition_at): Likewise.
17572 2018-04-27 Jeff Law <law@redhat.com>
17574 * config.gcc: Mark tile* targets as deprecated/obsolete.
17576 2018-04-27 Richard Biener <rguenther@suse.de>
17578 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
17581 2018-04-27 Richard Biener <rguenther@suse.de>
17583 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
17585 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
17587 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
17588 with Yd constraint. Set "preferred_for_speed" attribute from
17589 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
17590 with Yd constraint.
17591 (*movdi_internal): Ditto.
17592 (movti_interunit splitters): Remove
17593 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
17594 (movdi_interunit splitters): Ditto.
17595 * config/i386/constraints.md (Ye): Remove.
17596 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
17598 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17601 * config/aarch64/constraints.md (Usg): Limit to 31.
17602 (Usj): Limit to 63.
17604 2018-04-27 Jakub Jelinek <jakub@redhat.com>
17606 PR tree-optimization/85529
17607 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
17608 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
17609 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
17610 zero extension or masking of the MSB bit.
17611 (optimize_range_tests): Add FIRST_BB argument, pass it through
17612 to optimize_range_tests_var_bound.
17613 (maybe_optimize_range_tests, reassociate_bb): Adjust
17614 optimize_range_tests callers.
17616 2018-04-26 Richard Biener <rguenther@suse.de>
17617 Jakub Jelinek <jakub@redhat.com>
17619 * cgraph.h (symbol_table): Just declare debug method here.
17620 * symtab.c (symbol_table::debug): Define.
17622 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
17624 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
17626 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
17628 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
17629 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
17630 (*movdi_internal): Substitute Yi and Yj constraint with x
17631 and Ym and Yn constraint with y constraint. Update "isa"
17632 attribute and set "preferred_for_speed" attribute from
17633 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
17634 (*movsi_internal): Ditto.
17635 (*movdf_internal): Ditto.
17636 (*movsf_internal): Ditto.
17637 (*zero_extendsidi2): Ditto.
17638 * config/i386/sse.md (vec_set<mode>_0): Ditto.
17639 (sse2_loadld): Ditto.
17640 (*vec_extract<ssevecmodelower>_0): Ditto.
17641 (*vec_extractv4si_0_zext_sse4): Ditto.
17642 (vec_concatv2di): Ditto.
17643 (*vec_dup<mode>): Ditto.
17644 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
17645 * config/i386/constraints.md (Yi): Remove.
17650 2018-04-26 Nathan Sidwell <nathan@acm.org>
17652 * dumpfile.c (dump_open): New.
17653 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
17654 (dump_finish): Detect stdio/stderr by value not name.
17656 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
17658 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
17660 2018-04-26 Tom de Vries <tom@codesourcery.com>
17663 * config/nvptx/nvptx.c (verify_neutering_jumps)
17664 (verify_neutering_labels): New function
17665 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
17667 2018-04-26 Tom de Vries <tom@codesourcery.com>
17670 * config/nvptx/nvptx.c (needs_neutering_p): New function.
17671 (nvptx_single): Use needs_neutering_p to skip over insns that do not
17674 2018-04-26 Richard Biener <rguenther@suse.de>
17675 Tom de Vries <tom@codesourcery.com>
17678 * lto-streamer-out.c (output_function): Fixup loops if required to match
17679 discovery done in the reader.
17681 2018-04-26 Richard Biener <rguenther@suse.de>
17683 PR tree-optimization/85116
17684 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
17685 have a loop exit from the single latch predecessor. Remove
17686 case of header with just condition.
17687 (ch_base::copy_headers): Exclude infinite loops from any
17689 (pass_ch::execute): Record exits.
17691 2018-04-26 Richard Biener <rguenther@suse.de>
17693 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
17694 prologue cost vector and pass it to vect_get_load_cost.
17695 (vect_get_peeling_costs_all_drs): Likewise.
17696 (vect_peeling_hash_get_lowest_cost): Likewise.
17697 (vect_enhance_data_refs_alignment): Likewise.
17699 2018-04-26 Richard Biener <rguenther@suse.de>
17701 PR middle-end/85450
17702 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
17703 checking of integer<->pointer conversions.
17704 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
17705 sign-/zero-extending pointer types.
17706 (expand_omp_for_static_chunk): Likewise.
17708 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
17709 Jean Lee <xiaoyur347@gmail.com>
17711 * config/mips/mips.c (mips_asan_shadow_offset): New function.
17712 (TARGET_ASAN_SHADOW_OFFSET): Define.
17713 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
17714 true for -fsanitize=address.
17716 2018-04-25 Mark Wielaard <mark@klomp.org>
17718 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
17721 2018-04-25 Jakub Jelinek <jakub@redhat.com>
17723 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
17724 than "alu", remove explicit "memory" and "imm_disp" attributes.
17725 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
17727 PR middle-end/85414
17728 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
17729 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
17730 gen_lowpart_no_emit.
17732 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
17735 * config/i386/i386.c (ix86_expand_builtin): Change memory
17736 operand to XI, extend p0 to Pmode.
17737 * config/i386/i386.md: Change unspec volatile and operand
17738 1 mode to XI, change operand 0 mode to P.
17740 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
17742 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
17743 GET_MODE_MASK before any checking.
17744 (nds32_can_use_bset_p): Likewise.
17745 (nds32_can_use_btgl_p): Likewise.
17747 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
17749 * config/nds32/nds32-doubleword.md: New define_split pattern for
17750 illegal register number.
17752 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
17754 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
17756 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
17758 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
17760 2018-04-25 Richard Biener <rguenther@suse.de>
17762 * lto-streamer.h (LTO_major_version): Bump to 8.
17764 2018-04-25 Jakub Jelinek <jakub@redhat.com>
17766 * BASE-VER: Set to 9.0.0.
17768 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
17770 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
17771 in __abskf2 and __powikf2.
17773 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17776 * config/aarch64/constraints.md (Usg, Usj): New constraints.
17777 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
17778 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
17779 Use the above on operand 2. Reindent.
17780 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
17782 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
17785 * common/config/i386/i386-common.c (ix86_handle_option): Don't
17787 * config/i386/i386.opt (mcet): Removed.
17788 * doc/install.texi: Remove -mcet documentation.
17789 * doc/invoke.texi: Likewise.
17791 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
17794 * doc/install.texi: Remove -mcet from bootstrap-cet.
17796 2018-04-24 Jakub Jelinek <jakub@redhat.com>
17799 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
17800 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
17804 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
17805 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
17806 containing a CONST_VECTOR.
17808 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
17810 * doc/install.texi: Update newlib dependency for nvptx.
17812 2018-04-24 Jakub Jelinek <jakub@redhat.com>
17815 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
17816 instead of INTVAL when shifting x left.
17818 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
17820 PR tree-optimization/85478
17821 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
17822 vect_grouped_store_supported for single element vectors.
17824 2018-04-24 Richard Biener <rguenther@suse.de>
17827 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
17828 load cost increase to the case of non-constant step.
17830 2018-04-24 Jakub Jelinek <jakub@redhat.com>
17833 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
17834 destination if any_malformed_asm.
17836 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
17838 PR middle-end/85496
17839 * expr.c (store_field): In the bitfield case, if the value comes from
17840 a function call and is returned in registers by means of a PARALLEL,
17841 do not change the mode of the temporary unless BLKmode and VOIDmode.
17843 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
17845 PR rtl-optimization/85423
17846 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
17847 dependencies to debug insns when the previous insn is non-debug.
17849 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
17851 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
17852 enums into a single definition.
17853 (fls): Fix predicates and printing.
17856 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
17858 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
17859 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
17860 and short u6 immediate.
17861 (check_if_valid_sleep_operand): Remove.
17862 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
17864 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
17866 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
17867 flag_always_save_lp condition.
17868 * config/nds32/nds32.opt (malways-save-lp): New option.
17870 2018-04-22 Shiva Chen <shiva0217@gmail.com>
17872 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
17873 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
17874 * config/nds32/nds32.h
17875 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
17876 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
17878 2018-04-22 Shiva Chen <shiva0217@gmail.com>
17880 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
17881 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
17883 2018-04-22 Shiva Chen <shiva0217@gmail.com>
17884 Chung-Ju Wu <jasonwucj@gmail.com>
17886 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
17888 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
17889 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
17891 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
17893 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
17895 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
17897 * config/nds32/nds32-protos.h (nds32_data_alignment,
17898 nds32_local_alignment): Declare.
17899 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
17900 nds32_local_alignment): New functions.
17901 (TARGET_CONSTANT_ALIGNMENT): Define.
17902 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
17904 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
17906 * config/nds32/nds32.c
17907 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
17908 (TARGET_MODES_TIEABLE_P): Likewise.
17910 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
17912 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
17913 level Ofast and Og.
17915 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
17916 Chung-Ju Wu <jasonwucj@gmail.com>
17918 * config/nds32/constants.md (unspec_volatile_element): Add enum values
17919 for unaligned access.
17920 * config/nds32/nds32-intrinsic.c: Implementation of expanding
17922 * config/nds32/nds32-intrinsic.md: Likewise.
17923 * config/nds32/nds32_intrinsic.h: Likewise.
17924 * config/nds32/nds32.h (nds32_builtins): Likewise.
17925 * config/nds32/nds32.opt (munaligned-access): New option.
17926 * config/nds32/nds32.c (nds32_asm_file_start): Display
17927 flag_unaligned_access status.
17929 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
17931 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
17932 -mno-relax is present.
17933 * config/riscv/linux.h (LINK_SPEC): Ditto.
17935 2018-04-20 Martin Sebor <msebor@redhat.com>
17938 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
17940 (gimple_fold_builtin_stxcpy_chk): Same.
17941 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
17943 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
17946 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
17947 __powikf2 when long double is IEEE 128-bit.
17949 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
17951 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
17952 step to make sure stack always aligned.
17954 2018-04-20 Carl Love <cel@us.ibm.com>
17957 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
17958 size check for arg0.
17960 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
17961 Tom de Vries <tom@codesourcery.com>
17964 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
17965 Emit insns for calls too.
17966 (nvptx_find_par): Always look for worker-level predecessor insn.
17967 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
17969 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
17970 (nvptx_process_pars): Propagate frames for calls.
17972 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
17975 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
17977 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
17978 (ix86_handle_option): Don't handle OPT_mibt.
17979 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
17981 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
17983 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
17984 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
17985 (ix86_target_macros): Define __CET__ with flag_cf_protection
17986 for -fcf-protection.
17987 * config/i386/i386.c (isa2_opts): Remove -mibt.
17988 * config/i386/i386.h (TARGET_IBT): Removed.
17989 (TARGET_IBT_P): Likewise.
17990 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
17991 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
17992 * config/i386/i386.opt (mcet): Update help message.
17993 (mshstk): Likewise.
17995 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
17996 -mcet as an alias for -mshstk.
17998 2018-04-20 Richard Biener <rguenther@suse.de>
18000 PR middle-end/85475
18001 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
18002 complexity by forcing a single use of the multiply operand.
18004 2018-04-20 Martin Jambor <mjambor@suse.cz>
18007 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
18008 recursion dependency to only apply to non-clones.
18010 2018-04-20 Martin Jambor <mjambor@suse.cz>
18013 * ipa-cp.c (create_specialized_node): Check that clones of
18014 self-recursive edges exist during IPA-CP.
18016 2018-04-19 Toon Moene <toon@moene.org>
18018 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
18021 2018-04-19 Jakub Jelinek <jakub@redhat.com>
18023 PR tree-optimization/85467
18024 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
18025 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
18026 VECTOR_CST element to type.
18028 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
18031 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
18032 * config/i386/i386.md (builtin_setjmp_setup): Removed.
18033 (builtin_longjmp): Likewise.
18034 (save_stack_nonlocal): New pattern.
18035 (restore_stack_nonlocal): Likewise.
18037 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
18040 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
18041 Replace ASM_OUTPUT_LABEL with fprintf.
18043 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
18046 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
18047 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
18048 * config/i386/i386-c.c (ix86_target_macros_internal): Also
18049 define __IBT__ and __SHSTK__ for -fcf-protection.
18050 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
18052 (ix86_trampoline_init): Likewise.
18053 (x86_output_mi_thunk): Likewise.
18054 (ix86_notrack_prefixed_insn_p): Likewise.
18055 (ix86_option_override_internal): Don't disallow -fcf-protection.
18056 * config/i386/i386.md (rdssp<mode>): Also enable for
18058 (incssp<mode>): Likewise.
18059 (nop_endbr): Likewise.
18060 * config/i386/i386.opt (mcet): Change help message to built-in
18063 (mshstk): Likewise.
18064 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
18065 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
18066 enable CET built-in functions.
18068 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
18070 * common/config/i386/i386-common.c
18071 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
18072 OPTION_MASK_ISA_MOVDIRI_UNSET,
18073 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
18074 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
18075 * config.gcc (movdirintrin.h): New header.
18076 * config/i386/cpuid.h (bit_MOVDIRI,
18077 bit_MOVDIR64B): New bits.
18078 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
18080 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
18081 (VOID, PVOID, PCVOID)): New function types.
18082 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
18083 __builtin_ia32_directstoreu_u64,
18084 __builtin_ia32_movdir64b): New builtins.
18085 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
18086 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
18088 (ix86_valid_target_attribute_inner_p): Ditto.
18089 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
18090 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
18091 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
18092 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
18093 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
18094 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
18095 (movdiri<mode>, movdir64b_<mode>): New.
18096 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
18097 * config/i386/immintrin.h: Include movdirintrin.h.
18098 * config/i386/movdirintrin.h: New file.
18099 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
18101 2018-04-19 Richard Biener <rguenther@suse.de>
18103 PR middle-end/85455
18104 * cfg.c (clear_bb_flags): When loop state says we have
18105 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
18107 2018-04-19 Richard Biener <rguenther@suse.de>
18109 PR tree-optimization/84737
18110 * tree-vect-data-refs.c (vect_copy_ref_info): New function
18111 copying restrict info.
18112 (vect_setup_realignment): Use it.
18113 * tree-vectorizer.h (vect_copy_ref_info): Declare.
18114 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
18115 the first DR to all generated stores.
18116 (vectorizable_load): Likewise for loads.
18118 2018-04-19 Jakub Jelinek <jakub@redhat.com>
18120 PR tree-optimization/85446
18121 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
18122 the integral and pointer types to have the same precision.
18124 * doc/install.texi: Document --disable-cet being the default and
18127 2018-04-18 Martin Liska <mliska@suse.cz>
18129 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
18132 2018-04-18 Martin Liska <mliska@suse.cz>
18135 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
18138 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
18139 arguments if they are comparable.
18141 2018-04-18 Martin Liska <mliska@suse.cz>
18144 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
18147 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
18150 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
18153 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
18154 ENDBR after calling __morestack.
18156 2018-04-18 David Malcolm <dmalcolm@redhat.com>
18159 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
18160 by using gcc_base_ver to generate a gcc_driver_version, and use
18161 it when generating GCC_DRIVER_NAME.
18162 * configure: Regenerate.
18164 2018-04-18 Jakub Jelinek <jakub@redhat.com>
18167 * config.gcc: Obsolete powerpc*-*-*spe*.
18169 2018-04-17 Jakub Jelinek <jakub@redhat.com>
18172 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
18173 (stabstr_D): Change type of unum from unsigned int to
18174 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
18177 2018-04-17 Jim Wilson <jimw@sifive.com>
18180 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
18181 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
18182 Set arg_pointer_offset after using pretend_args_size.
18184 2018-04-17 Jakub Jelinek <jakub@redhat.com>
18186 PR rtl-optimization/85431
18187 * dse.c (record_store): Ignore zero width stores.
18190 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
18191 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
18192 __builtin_stack_restore rather than after it.
18193 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
18194 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
18195 argument instead of virtual_dynamic_stack_rtx.
18197 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
18199 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
18201 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18202 Add note to error message to explain internal mapping of overloaded
18203 built-in function name to non-overloaded built-in function name.
18204 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
18207 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
18210 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
18211 where the inputs overlap with the output.
18213 2018-04-17 Jakub Jelinek <jakub@redhat.com>
18215 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
18216 (=v, v) alternative and explicit "memory" attribute.
18217 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
18218 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
18220 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
18221 "sselog1" type instead of "sselog".
18222 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
18223 "sselog". Remove explicit "memory" attribute.
18224 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
18225 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
18227 (vec_extract_hi_v32hi): Merge all alternatives into one, use
18228 "sselog1" type instead of "sselog". Remove explicit "memory"
18230 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
18231 use "sselog1" type instead of "sselog". Remove explicit "memory"
18233 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
18234 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
18236 (vec_extract_hi_v64qi): Merge all alternatives into one, use
18237 "sselog1" type instead of "sselog". Remove explicit "memory"
18239 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
18240 use "sselog1" type instead of "sselog". Remove explicit "memory"
18244 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
18246 PR middle-end/85414
18247 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
18250 2018-04-17 Martin Jambor <mjambor@suse.cz>
18253 * ipa-cp.c (create_specialized_node): Call
18254 expand_all_artificial_thunks if necessary.
18256 2018-04-17 Martin Liska <mliska@suse.cz>
18259 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
18260 in message, remote space in between '_G' and '('.
18262 2018-04-17 Jakub Jelinek <jakub@redhat.com>
18265 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
18266 avx512f_vmcmp<mode>3<round_saeonly_name>,
18267 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
18268 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
18269 avx512f_rndscale<mode><round_saeonly_name>,
18270 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
18271 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
18272 Use %<iptr>2 instead of %2 for -masm=intel.
18273 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
18274 avx512f_vcvttss2usi<round_saeonly_name>,
18275 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
18277 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
18278 avx512f_vcvttsd2usi<round_saeonly_name>,
18279 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
18280 Use %q1 instead of %1 for -masm=intel.
18281 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
18282 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
18283 of %3 for -masm=intel.
18284 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
18286 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
18288 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
18290 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
18291 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
18293 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
18295 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
18296 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
18297 %g1 and one with %0 and %1.
18298 (avx512er_vmrcp28<mode><round_saeonly_name>,
18299 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
18300 %1 for -masm=intel.
18301 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
18302 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
18303 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
18304 of %0 and %{%4%} for -masm=intel.
18305 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
18306 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
18307 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
18308 order of %0 and %{%5%}%{z%} for -masm=intel.
18310 2018-04-17 Jan Hubicka <jh@suse.cz>
18313 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
18315 2018-04-17 Martin Liska <mliska@suse.cz>
18318 * multiple_target.c (create_dispatcher_calls): Set apostrophes
18319 for target_clone error message. Make default implementation
18320 clone to be a local declaration.
18321 (separate_attrs): Add new argument and check for an empty
18323 (expand_target_clones): Handle it.
18324 (ipa_target_clone): Make redirection just for target_clones
18327 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
18328 Tom de Vries <tom@codesourcery.com>
18330 PR middle-end/84955
18331 * omp-expand.c (expand_oacc_for): Add dummy false branch for
18332 tiled basic blocks without omp continue statements.
18334 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
18337 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
18338 vec_extract expression as having side effects to make sure it gets
18341 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
18344 * config/i386/i386.c (get_builtin_code_for_version): Check
18347 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
18350 * config.gcc: Support "skylake".
18351 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
18353 * config/i386/i386.c (m_SKYLAKE): Define.
18354 (processor_target_table): Add "skylake".
18355 (ix86_option_override_internal): Add "skylake".
18356 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
18357 PROCESSOR_CANNONLAKE.
18358 (get_builtin_code_for_version): Fix priority for
18359 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
18360 PROCESSOR_SKYLAKE-AVX512.
18361 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
18362 (processor_type): Add PROCESSOR_SKYLAKE.
18364 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
18365 Jason Merrill <jason@redhat.com>
18368 * convert.c (convert_to_integer_1): Use direct recursion for
18369 enumeral types and types with a precision less than the number
18370 of bits in their mode.
18372 2018-04-16 Julia Koval <julia.koval@intel.com>
18375 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
18376 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
18378 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
18381 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
18382 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
18383 and -mno-direct-move.
18385 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
18388 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
18389 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
18390 Ensure negative shifts result in {0}.
18392 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
18394 PR rtl-optimization/79916
18395 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
18396 regs (if any) to define how to gnerate SD moves when LRA is in
18399 2018-04-13 Jakub Jelinek <jakub@redhat.com>
18401 PR rtl-optimization/85393
18402 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
18403 * except.c (expand_dw2_landing_pad_for_region): Make static.
18404 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
18405 a label and unconditional jump to old_bb, rather than
18406 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
18409 PR rtl-optimization/85376
18410 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
18411 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
18412 instead of a specific value.
18414 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
18415 Bin Cheng <bin.cheng@arm.com>
18417 PR tree-optimization/82965
18418 PR tree-optimization/83991
18419 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
18420 by_profile_only parameter.
18421 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
18422 information if the loop was predicted to iterate too many times.
18423 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
18425 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
18428 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
18431 2018-04-13 Martin Liska <mliska@suse.cz>
18432 Jakub Jelinek <jakub@redhat.com>
18434 PR middle-end/81657
18435 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
18436 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
18437 * builtins.c (expand_builtin_memory_copy_args): Use
18438 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
18439 handle dest_addr == pc_rtx.
18441 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
18444 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
18445 asked to not generate direct moves.
18446 (fix_trunc<mode>si2_stfiwx): Similar.
18447 (fix_trunc<mode>si2_internal): Similar.
18449 2018-04-12 Jakub Jelinek <jakub@redhat.com>
18452 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
18453 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
18454 lookup if dest in some wider mode is known to be const0_rtx and
18455 if so, record permanent equivalence for it to be ZERO_EXTEND of
18456 the narrower mode destination.
18458 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
18460 * lto-streamer-out.c (output_function): Revert 259346.
18461 * omp-expand.c (expand_oacc_for): Likewise.
18463 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
18465 PR rtl-optimization/85354
18466 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
18467 * sel-sched.c (sel_global_init): ... here.
18469 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
18472 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
18473 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
18474 mode for PE-COFF targets.
18475 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
18476 (i386_pe_asm_lto_end): Likewise.
18477 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
18478 (TARGET_ASM_LTO_END): Likewise.
18479 * config/i386/winnt.c (saved_debug_info_level): New static variable.
18480 (i386_pe_asm_lto_start): New function.
18481 (i386_pe_asm_lto_end): Likewise.
18483 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
18484 Richard Biener <rguenther@suse.de>
18486 PR middle-end/84955
18487 * lto-streamer-out.c (output_function): Fix CFG loop state before
18489 * omp-expand.c (expand_oacc_for): Handle calls to internal
18490 functions like regular functions.
18492 2018-04-12 Richard Biener <rguenther@suse.de>
18495 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
18496 for the early LTO debug to properly generate references to it
18497 during DIE emission. Do not re-use that for the skeleton for
18499 (dwarf2out_early_finish): Likewise.
18501 2018-04-12 Jakub Jelinek <jakub@redhat.com>
18504 * config/i386/sse.md
18505 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
18506 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
18507 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
18508 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
18509 and output is a reg, avoid creating invalid lowpart subreg, but
18510 instead split into a 512-bit move. Don't split if not AVX512VL,
18511 input is xmm16+ reg and output is a mem.
18512 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
18513 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
18514 xmm16+ reg and output is a mem.
18516 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18518 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
18519 also for flag_dwarf2_cfi_asm.
18521 2018-04-12 Jakub Jelinek <jakub@redhat.com>
18523 PR rtl-optimization/85342
18524 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
18525 a bool scalar var inside of the loop instead. Don't try to update
18526 recog_data.operand after failed apply_change_group.
18528 2018-04-12 Tom de Vries <tom@codesourcery.com>
18531 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
18532 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
18533 array with flexible array member as array without given dimension.
18534 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
18535 argument for undefined param to true.
18537 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
18540 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
18541 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
18542 from PowerPC section.
18543 * config/rs6000/sysv4.opt (mcall-): Improve help text.
18544 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
18545 help text that is too long.
18546 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
18547 help text that is too long.
18548 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
18549 help text that is too long.
18551 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
18553 * config/alpha/alpha.md (stack_probe_internal): Rename
18554 from "probe_stack". Update all callers.
18556 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
18558 PR rtl-optimization/84566
18559 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
18560 sched_macro_fuse_insns.
18562 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
18565 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
18566 (compute_block_dependences): ... from here.
18568 2018-04-11 Jakub Jelinek <jakub@redhat.com>
18570 PR tree-optimization/85331
18571 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
18572 from int to HOST_WIDE_INT.
18574 2018-04-11 Martin Jambor <mjambor@suse.cz>
18577 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
18578 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
18579 not the same as the source val.
18580 (cgraph_edge_brings_value_p): New parameter.
18581 (gather_edges_for_value): Pass destination value to
18582 cgraph_edge_brings_value_p.
18583 (perhaps_add_new_callers): Likewise.
18584 (get_info_about_necessary_edges): Likewise and exclude values brought
18585 only by self-recursive edges.
18586 (create_specialized_node): Redirect only clones of self-calling edges.
18587 (+self_recursive_pass_through_p): New function.
18588 (find_more_scalar_values_for_callers_subset): Use it.
18589 (find_aggregate_values_for_callers_subset): Likewise.
18590 (known_aggs_to_agg_replacement_list): Removed.
18591 (decide_whether_version_node): Re-calculate known constants for all
18592 remaining context clones.
18594 2018-04-11 Richard Biener <rguenther@suse.de>
18597 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
18598 from early DWARF output.
18599 (dwarf2out_early_finish): Output line info unconditionally into
18600 early DWARF and add reference to it.
18602 2018-04-11 Jakub Jelinek <jakub@redhat.com>
18605 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
18606 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
18607 other than V2DFmode using iptr mode attribute.
18608 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
18610 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
18612 PR rtl-optimization/84659
18613 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
18615 2018-04-11 Jakub Jelinek <jakub@redhat.com>
18618 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
18620 (output_loc_list): Pass address of a dummy size variable even in the
18621 locview handling loop.
18622 (index_location_lists): Add comment on why skip_loc_list_entry can't
18625 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
18628 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
18631 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
18634 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
18635 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
18636 and -mstring-compare-inline-limit.
18638 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
18641 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
18642 for stack clash protection in a register whenever we need it to be in
18645 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
18647 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
18648 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
18650 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
18653 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
18655 (mlong-double-): Ditto.
18656 * config/rs6000/sysv4.opt (msdata=): Ditto.
18657 (mtls-size=): Ditto.
18659 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
18661 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
18662 erroneous entries for
18663 "vector int vec_ldl (int, long int *)", and
18664 "vector unsigned int vec_ldl (int, unsigned long int *)".
18665 Add comments and entries for
18666 "vector bool char vec_ldl (int, bool char *)",
18667 "vector bool short vec_ldl (int, bool short *)",
18668 "vector bool int vec_ldl (int, bool int *)",
18669 "vector bool long long vec_ldl (int, bool long long *)",
18670 "vector pixel vec_ldl (int, pixel *)",
18671 "vector long long vec_ldl (int, long long *)",
18672 "vector unsigned long long vec_ldl (int, unsigned long long *)".
18673 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
18674 type tree bool_long_long_type_node and correct definition of
18675 bool_V2DI_type_node to make reference to this new type tree.
18676 (rs6000_mangle_type): Replace erroneous reference to
18677 bool_long_type_node with bool_long_long_type_node.
18678 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
18679 comments to emphasize sign distinctions for char and int types and
18680 replace RS6000_BTI_bool_long constant with
18681 RS6000_BTI_bool_long_long constant. Also add comment to restrict
18682 use of RS6000_BTI_pixel.
18683 (bool_long_type_node): Remove this macro definition.
18684 (bool_long_long_type_node): New macro definition
18686 2018-04-10 Jakub Jelinek <jakub@redhat.com>
18688 PR rtl-optimization/85300
18689 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
18690 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
18691 simplify_unary_operation fails.
18693 2018-04-10 Martin Liska <mliska@suse.cz>
18695 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
18696 cgraph_edge and ipa_ref.
18698 2018-04-10 Jakub Jelinek <jakub@redhat.com>
18702 * config/i386/sse.md
18703 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
18704 computation of the VEC_MERGE selector from mask.
18705 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
18706 Fix decoding of the VEC_MERGE selector into mask.
18708 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
18710 PR tree-optimization/85286
18711 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
18713 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
18715 * final.c (final_1): Set insn_last_address as well as
18716 insn_current_address.
18718 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18721 * explow.c (emit_stack_probe): Call validize_mem on memory location
18722 before passing it to gen_probe_stack. Create address operand and
18723 legitimize it for the probe_stack_address case.
18725 2018-04-09 Jan Hubicka <jh@suse.cz>
18728 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
18729 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
18730 * tree.c (free_lang_data_in_type): Fix handling of binfos;
18732 (free_lang_data): Rebuild type inheritance graph.
18734 2018-04-09 Martin Sebor <msebor@redhat.com>
18736 * invoke.texi (-finline-small-functions): Mention other optimization
18738 (-findirect-inlining, -fpartial-inlining): Same.
18739 (-finline-functions-called-once): Same.
18740 (-freorder-blocks-and-partition): Same.
18742 2018-04-09 Jan Hubicka <jh@suse.cz>
18745 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
18746 jumps; choose last target that matches the criteria (i.e.
18747 no partition changes for non-crossing jumps).
18748 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
18749 support for redirecting crossing jumps to non-crossing.
18751 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
18753 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
18754 also for naked functions.
18756 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
18758 * config/arc/arc.md (add_shift): New pattern.
18759 (add_shift2): Likewise.
18760 (sub_shift): Likewise.
18761 (sub_shift_cmp0_noout): Likewise.
18762 (compare_si_ashiftsi): Likewise.
18763 (xbfu_cmp0_noout): New combine pattern.
18764 (xbfu_cmp0"): Likewise.
18765 (movsi_set_cc_insn): Place the predicable variant first.
18766 (commutative_binary_cmp0_noout): Remove clobber.
18767 (commutative_binary_cmp0): New pattern.
18768 (noncommutative_binary_cmp0): Likewise.
18769 (noncommutative_binary_cmp0_noout): Likewise.
18770 (noncommutative_binary_comparison_result_used): Removed.
18771 (rsub_cmp0): New pattern.
18772 (rsub_cmp0_noout): Likewise.
18773 (extzvsi): Changed, keep only meaningful variants.
18774 (SQH, SEZ): New iterators.
18775 (SQH_postfix): New mode attribute.
18776 (SEZ_prefix): New code attribute.
18777 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
18778 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
18779 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
18780 of numerical value.
18781 (noncommutative_operator): Check the availability of barrel
18784 2018-04-09 Richard Biener <rguenther@suse.de>
18786 PR tree-optimization/85284
18787 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
18788 Only use the niter constraining form of simple_iv when the exit
18789 is always executed.
18791 2018-04-09 Tom de Vries <tom@codesourcery.com>
18794 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
18795 (define_expand "*memory_barrier"): New define_expand.
18796 (define_insn "memory_barrier"): New insn.
18798 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
18800 PR rtl-optimization/80463
18801 PR rtl-optimization/83972
18802 PR rtl-optimization/83480
18804 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
18805 correct producer for the insn.
18806 (tidy_control_flow): Fixup seqnos in case of debug insns.
18808 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
18810 PR rtl-optimization/83913
18812 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
18813 different sched-times when merging exprs.
18815 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
18817 PR rtl-optimization/83962
18819 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
18820 tidy_fallthru_edge and tidy_control_flow.
18822 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
18824 PR rtl-optimization/83530
18826 * sel-sched.c (force_next_insn): New global variable.
18827 (remove_insn_for_debug): When force_next_insn is true, also leave only
18828 next insn in the ready list.
18829 (sel_sched_region): When the region wasn't scheduled, make another pass
18830 over it with force_next_insn set to 1.
18832 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
18834 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
18836 * config/nds32/constants.md (unspec_volatile_element): Add enum values
18837 for interrupt control.
18838 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
18839 functions for interrupt control.
18840 * config/nds32/nds32-intrinsic.md: Likewise.
18841 * config/nds32/nds32_intrinsic.h: Likewise.
18842 * config/nds32/nds32.h (nds32_builtins): Likewise.
18844 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
18846 * config/nds32/nds32.c (nds32_init_machine_status,
18847 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
18848 strict_aligned_p field.
18849 (nds32_expand_to_rtl_hook): New function.
18850 (TARGET_EXPAND_TO_RTL_HOOK): Define.
18851 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
18853 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
18854 Chung-Ju Wu <jasonwucj@gmail.com>
18856 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
18857 * config/nds32/nds32-n7.md: New file.
18858 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
18859 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
18861 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
18862 * config/nds32/nds32.md (pipeline_model): Add n7.
18863 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
18864 * config/nds32/pipelines.md: Include n7 settings.
18866 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
18867 Chung-Ju Wu <jasonwucj@gmail.com>
18869 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
18870 * config/nds32/nds32-e8.md: New file.
18871 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
18872 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
18874 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
18875 * config/nds32/nds32.md (pipeline_model): Add e8.
18876 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
18877 * config/nds32/pipelines.md: Include e8 settings.
18879 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
18880 Chung-Ju Wu <jasonwucj@gmail.com>
18882 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
18883 * config/nds32/nds32-n8.md: New file.
18884 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
18885 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
18887 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
18888 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
18889 * config/nds32/nds32.md (pipeline_model): Add n8.
18890 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
18891 * config/nds32/pipelines.md: Include n8 settings.
18893 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
18894 Chung-Ju Wu <jasonwucj@gmail.com>
18896 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
18897 * config/nds32/nds32-n9-2r1w.md: New file.
18898 * config/nds32/nds32-n9-3r2w.md: New file.
18899 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
18900 nds32_register_ports): New or modify for cpu n9.
18901 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
18903 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
18904 * config/nds32/nds32-utils.c: New file.
18905 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
18906 TARGET_MUL_SLOW): Define.
18907 * config/nds32/nds32.md (pipeline_model): New attribute.
18908 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
18909 New options that support cpu n9.
18910 * config/nds32/pipelines.md: Include n9 settings.
18911 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
18913 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
18915 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
18916 information if necessary.
18917 (output_cond_branch_compare_zero): Likewise.
18918 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
18919 (nds32_target_alignment): Refine for alignment.
18920 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
18921 (FUNCTION_BOUNDARY): Modify.
18922 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
18924 * config/nds32/nds32.opt (malways-align, malign-functions): New.
18926 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
18928 * config/nds32/constants.md (unspec_volatile_element): Add values for
18929 TLB operation and data prefetch.
18930 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
18931 functions for TLB operation and data prefetch.
18932 * config/nds32/nds32-intrinsic.md: Likewise.
18933 * config/nds32/nds32_intrinsic.h: Likewise.
18934 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
18935 (nds32_print_operand): Likewise.
18936 * config/nds32/nds32.h (nds32_builtins): Likewise.
18938 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
18939 Andrew Pinski <pinsika@gcc.gnu.org>
18941 PR middle-end/82976
18942 * match.pd: Use constant_boolean_node of correct type instead of
18943 boolean_true_node or boolean_false_node for simplifying
18944 pointer comparisons to zero.
18946 2018-04-07 Jakub Jelinek <jakub@redhat.com>
18948 PR tree-optimization/80021
18949 * tree.c (verify_type_variant): Make error call in verify_variant_match
18950 translatable and remove final full stop.
18952 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
18954 * config/nds32/constants.md (unspec_volatile_element): Add
18955 UNSPEC_VOLATILE_EH_RETURN.
18956 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
18957 nds32_output_stack_pop): Support dwarf exception handling process.
18958 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
18959 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
18960 exception handling process.
18961 (nds32_compute_stack_frame): Likewise.
18962 (nds32_return_addr_rtx): Likewise.
18963 (nds32_initial_elimination_offset): Likewise.
18964 (nds32_expand_prologue): Likewise.
18965 (nds32_expand_epilogue): Likewise.
18966 (nds32_dynamic_chain_address): New function.
18967 * config/nds32/nds32.h (machine_function): Add fields for dwarf
18968 exception handling.
18969 (DYNAMIC_CHAIN_ADDRESS): Define.
18970 (EH_RETURN_DATA_REGNO): Define.
18971 (EH_RETURN_STACKADJ_RTX): Define.
18972 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
18973 patterns for dwarf exception handling.
18975 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
18977 * config/nds32/nds32.h: Clean up obsolete macros.
18979 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
18981 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
18982 Add enum values for particular instructions.
18983 * config/nds32/nds32-intrinsic.c: Implementation of expanding
18984 particular intrinsic functions.
18985 * config/nds32/nds32-intrinsic.md: Likewise.
18986 * config/nds32/nds32_intrinsic.h: Likewise.
18987 * config/nds32/nds32.h (nds32_builtins): Likewise.
18988 * config/nds32/nds32.md (type): Add pbsad and pbsada.
18989 (btst, ave): New patterns for particular instructions.
18991 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
18993 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
18994 Add enum values for atomic load/store and memory sync.
18995 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
18997 * config/nds32/nds32-intrinsic.md: Likewise.
18998 * config/nds32/nds32_intrinsic.h: Likewise.
18999 * config/nds32/nds32.h (nds32_builtins): Likewise.
19001 2018-04-07 Jakub Jelinek <jakub@redhat.com>
19003 PR tree-optimization/85257
19004 * fold-const.c (native_encode_vector): If not all elts could fit
19005 and off is -1, return 0 rather than offset.
19006 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
19007 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
19008 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
19009 adjust buffer in native_interpret_expr call.
19011 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
19013 * config/nds32/constants.md (unspec_volatile_element): Add cache
19014 control enum values.
19015 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
19016 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
19017 * config/nds32/nds32.c (nds32_cctl_names): New.
19018 (nds32_print_operand): Handle cache control register names.
19019 * config/nds32/nds32.h (nds32_builtins): New enum values.
19020 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
19022 * config/nds32/nds32.md (type): Add mmu.
19023 * config/nds32/pipelines.md (simple_insn): Add mmu.
19025 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
19027 * config/nds32/nds32.md (type): Remove call.
19028 * config/nds32/pipelines.md (simple_insn): Likewise.
19030 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
19032 * config/nds32/constants.md (unspec_volatile_element): Add
19033 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
19034 UNSPEC_VOLATILE_FMFCFG.
19035 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
19036 description for fmfcfg and fmfcsr.
19037 (bdesc_1arg): Add fmtcsr.
19038 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
19039 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
19040 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
19041 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
19042 unspec_fmfcfg): New patterns.
19043 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
19044 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
19045 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
19046 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
19047 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
19048 __nds32__fmfcfg): Define.
19050 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
19052 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
19053 intrinsic register names.
19054 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
19055 intrinsic register enum values and macros.
19057 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
19059 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
19060 for load/store addressing form.
19061 (nds32_print_operand_address): Likewise.
19063 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
19066 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
19067 based on LABEL_REF. Remove useless assertion.
19068 (pic_address_needs_scratch): Fix formatting.
19069 (sparc_legitimize_pic_address): Minor tweaks.
19070 (sparc_delegitimize_address): Adjust assertion accordingly.
19071 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
19072 into symbolic_operand.
19073 (movsi_high_pic_label_ref): Likewise.
19074 (movsi_lo_sum_pic_label_ref): Likewise.
19075 (movdi_pic_label_ref): Likewise.
19076 (movdi_high_pic_label_ref): Likewise.
19077 (movdi_lo_sum_pic_label_ref): Likewise.
19079 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
19081 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
19082 custom LIB_SPEC setup.
19084 2018-04-06 Ruslan Bukin <br@bsdpad.com>
19085 Kito Cheng <kito.cheng@gmail.com>
19087 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
19088 * config/riscv/freebsd.h: New.
19090 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
19092 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
19093 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
19096 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
19097 Kito Cheng <kito.cheng@gmail.com>
19099 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
19100 nds32_output_call, nds32_symbol_binds_local_p): New functions.
19101 * config/nds32/nds32-protos.h (nds32_output_call,
19102 nds32_output_return): Declare.
19103 * config/nds32/nds32.md: Refine all the call and return patterns.
19105 2018-04-06 Jakub Jelinek <jakub@redhat.com>
19108 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
19109 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
19111 PR rtl-optimization/84872
19112 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
19113 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
19114 EDGE_CROSSING edge.
19116 2018-04-06 Tamar Christina <tamar.christina@arm.com>
19118 * expr.c (copy_blkmode_to_reg): Revert 254862.
19119 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
19121 2018-04-06 Richard Biener <rguenther@suse.de>
19123 PR middle-end/85244
19124 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
19125 after seeing a component reference with an adjacent field. Treat
19126 refs to arrays at struct end of external decls similar to
19127 refs to unconstrained commons.
19129 2018-04-06 Jakub Jelinek <jakub@redhat.com>
19132 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
19133 look through SAVE_EXPRs with non-side-effects argument. Adjust
19135 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
19138 2018-04-06 Richard Biener <rguenther@suse.de>
19140 PR middle-end/85180
19141 * alias.c (find_base_term): New wrapper around find_base_term
19142 unwinding CSELIB_VAL_PTR changes.
19143 (find_base_term): Do not restore CSELIB_VAL_PTR during the
19146 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19148 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
19150 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
19151 constant definitions.
19152 ("nop"): lr 0,0 -> nopr r0
19153 ("nop_lr0", "nop_lr1"): New insn definitions.
19155 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
19157 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
19158 NDS32_V3PUSH_AVAILABLE_P macro.
19160 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
19161 Chung-Ju Wu <jasonwucj@gmail.com>
19163 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
19164 (nds32*-*-*): Add float and fpu_config into supported_defaults.
19165 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
19166 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
19167 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
19168 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
19169 * config/nds32/constraints.md: New constraints and checking for hard
19170 float configuration.
19171 * config/nds32/iterators.md: New mode iterator and attribute for hard
19172 float configuration.
19173 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
19175 * config/nds32/nds32-fpu.md: New file.
19176 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
19177 deal with hard float code generation.
19178 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
19180 (abi_type, float_reg_number): New enum type.
19181 * config/nds32/nds32-predicates.c: New predicates for hard float.
19182 * config/nds32/nds32-protos.h: Declare functions for hard float.
19183 * config/nds32/nds32.c: Implementation for hard float configuration.
19184 * config/nds32/nds32.h: Definitions for hard float configuration.
19185 * config/nds32/nds32.md: Include hard float machine description and
19186 modify patterns for hard float configuration.
19187 * config/nds32/nds32.opt: New options for hard float configuration.
19188 * config/nds32/predicates.md: New predicates for hard float
19191 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
19193 * common/config/nds32/nds32-common.c
19194 (nds32_option_optimization_table): Enable -mreleax-hint by default.
19196 2018-04-05 Jakub Jelinek <jakub@redhat.com>
19198 PR middle-end/85195
19199 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
19200 CONSTRUCTOR_ELT (ctor, ...)->value.
19202 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
19205 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
19207 2018-04-05 Tom de Vries <tom@codesourcery.com>
19210 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
19213 2018-04-05 Shiva Chen <shiva0217@gmail.com>
19214 Kito Cheng <kito.cheng@gmail.com>
19216 * config/nds32/constraints.md (U33): Fine-tune checking condition.
19217 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
19218 * config/nds32/nds32.h (nds32_16bit_address_type): Add
19219 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
19221 2018-04-05 Shiva Chen <shiva0217@gmail.com>
19222 Kito Cheng <kito.cheng@gmail.com>
19224 * config/nds32/constraints.md (Ufe): New memory constraint.
19225 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
19226 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
19227 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
19229 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
19230 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
19232 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19234 * config/nds32/nds32.md: Use optimize_size in the condition for
19235 alu-shift instructions.
19237 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19239 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
19241 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19243 * config/nds32/nds32.md (negsi2): Refine pattern.
19245 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
19246 Chung-Ju Wu <jasonwucj@gmail.com>
19248 * config/nds32/iterators.md (shift_rotate): New code iterator.
19249 (shift): New code attribute.
19250 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
19251 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
19252 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
19253 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
19254 bit-wise operations.
19255 (andsi3, *andsi3): Ditto.
19256 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
19257 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
19258 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
19259 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
19260 nds32_ior_operand, nds32_xor_operand): New predicates.
19262 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19264 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
19265 (addsi3, subsi3): ... this.
19267 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19269 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
19271 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19273 * config/nds32/nds32.md: Adjust indention.
19275 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
19277 * config/nds32/nds32.md (feature): New attribute.
19279 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19281 * config/nds32/nds32.md (subtype): New attribute.
19283 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19286 * config/arm/arm-builtins.c (arm_expand_builtin): Change
19287 expansion to perform a bitwise AND of the argument followed by a
19288 boolean negation of the result.
19290 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
19292 PR rtl-optimization/84878
19293 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
19294 the basic block. Assert the use reference is not artificial and that
19295 it has an associated insn.
19297 2018-04-04 Michael Matz <matz@suse.de>
19299 * builtins.c (compute_objsize): Pass correct operand
19300 to array_at_struct_end_p.
19302 2018-04-04 Richard Biener <rguenther@suse.de>
19305 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
19306 from contexts for DINFO_LEVEL_TERSE and below.
19308 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
19310 * config/nds32/nds32-doubleword.md (move_<mode>): Require
19311 resiter_operand condition.
19312 * config/nds32/nds32.md (*move<mode>): Ditto.
19314 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
19315 Monk Chiang <sh.chiang04@gmail.com>
19317 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
19319 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
19321 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
19323 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
19324 Kito Cheng <kito.cheng@gmail.com>
19326 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
19327 nds32_cond_code_str, output_cond_branch,
19328 output_cond_branch_compare_zero, nds32_expand_cbranch,
19329 nds32_expand_cstore, nds32_expand_movcc,
19330 nds32_output_cbranchsi4_equality_zero,
19331 nds32_output_cbranchsi4_equality_reg,
19332 nds32_output_cbranchsi4_equality_reg_or_const_int,
19333 nds32_output_cbranchsi4_greater_less_zero: New functions.
19334 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
19335 nds32_expand_cstore, nds32_expand_movcc,
19336 nds32_output_cbranchsi4_equality_zero,
19337 nds32_output_cbranchsi4_equality_reg,
19338 nds32_output_cbranchsi4_equality_reg_or_const_int,
19339 nds32_output_cbranchsi4_greater_less_zero): Declare.
19340 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
19341 nds32_rimm11s_operand): New predicates.
19342 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
19343 * config/nds32/nds32.md: Rewrite all the branch and conditional move
19346 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
19348 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
19349 * config/nds32/nds32.md: Ditto.
19350 * config/nds32/pipelines.md: Ditto.
19352 2018-04-04 Richard Biener <rguenther@suse.de>
19354 PR tree-optimization/85168
19355 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
19356 propagating abnormals.
19358 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
19360 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
19362 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
19363 Kito Cheng <kito.cheng@gmail.com>
19365 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
19366 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
19367 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
19368 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
19369 * config/nds32/nds32.md (sibcall_internal): New.
19370 (sibcall_register): Remove.
19371 (sibcall_immediate): Remove.
19372 (sibcall_value_internal): New.
19373 (sibcall_value_register): Remove.
19374 (sibcall_value_immediate): Remove.
19375 * config/nds32/predicates.md (nds32_general_register_operand): New.
19376 (nds32_call_address_operand): New.
19378 2018-04-03 Jakub Jelinek <jakub@redhat.com>
19380 PR rtl-optimization/85167
19381 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
19382 bb_defs if *split_p, instead preinitialize it to NULL.
19384 PR tree-optimization/85156
19385 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
19386 evaluating the argument multiple times.
19388 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
19390 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
19392 (_mm_cvtpd_ps): Likewise.
19393 (_mm_cvttpd_epi32): Likewise.
19394 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
19395 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
19396 vector, pixel, and bool following altivec.h include.
19398 2018-04-03 Martin Sebor <msebor@redhat.com>
19400 * doc/extend.texi (Common Function Attributes): Clarify.
19401 (const attribute): Likewise.
19402 (pure attribute): Likewise.
19404 2018-04-03 Jakub Jelinek <jakub@redhat.com>
19407 * config/i386/i386.c (ix86_expand_vector_set): Use
19408 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
19410 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
19412 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
19413 instructions when changing rounding bits to preserve precision bits
19414 in the x87 control word.
19416 2018-04-03 Martin Liska <mliska@suse.cz>
19418 PR tree-optimization/82491
19419 * rtl.h (strip_offset_and_add): Replace += suboffset with
19420 poly_uint64 () + suboffset.
19422 2018-03-29 Martin Liska <mliska@suse.cz>
19423 Martin Jambor <mjambor@suse.cz>
19426 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
19427 param_type is not an integral or pointer type.
19429 2018-04-03 Richard Biener <rguenther@suse.de>
19431 * sese.h (recompute_all_dominators): Remove.
19433 2018-04-02 Martin Sebor <msebor@redhat.com>
19435 * doc/invoke.texi (-Wrestrict): Fix typos.
19437 2018-04-02 Jim Wilson <jimw@sifive.com>
19439 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
19440 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
19441 (<optab>di3, <optab>si3_extend): Likewise.
19442 (<optab>si3_mask, <optab>si3_mask_1): New.
19443 (<optab>di3_mask, <optab>di3_mask_1): New.
19444 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
19445 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
19446 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
19448 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
19450 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
19453 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
19455 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
19456 (nds32_canonicalize_comparison): New function.
19458 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
19459 Kito Cheng <kito.cheng@gmail.com>
19460 Kuan-Lin Chen <kuanlinchentw@gmail.com>
19462 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
19463 * config/nds32/constants.md (unspec_volatile_element): Add
19464 UNSPEC_VOLATILE_RELAX_GROUP.
19465 * config/nds32/nds32-relax-opt.c: New file.
19466 * config/nds32/nds32-predicates.c
19467 (nds32_symbol_load_store_p): New function.
19468 * config/nds32/nds32-protos.h
19469 (nds32_symbol_load_store_p): Declare function.
19470 (make_pass_nds32_relax_opt): Declare new rtl pass function.
19471 * config/nds32/nds32.c
19472 (nds32_register_pass): New function to register pass.
19473 (nds32_register_passes): New function to register passes.
19474 * config/nds32/nds32.md (relax_group): New pattern.
19475 * config/nds32/nds32.opt (mrelax-hint): New option.
19476 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
19478 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
19480 * config/nds32/t-nds32: Modify files dependency.
19482 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
19484 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
19485 (PROFILE_HOOK): Define its implementation.
19487 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
19489 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
19490 type and 32-bit size.
19492 2018-04-01 Jakub Jelinek <jakub@redhat.com>
19494 PR middle-end/85090
19495 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
19496 (V_128_256): New mode iterator.
19497 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
19498 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
19499 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
19501 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
19502 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
19504 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
19507 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
19508 NaN inputs correctly.
19510 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
19513 * config/rs6000/vsx.md (??r): New mode attribute.
19514 (*vsx_mov<mode>_64bit): Use it.
19515 (*vsx_mov<mode>_32bit): Likewise.
19517 2018-03-30 Martin Sebor <msebor@redhat.com>
19519 PR tree-optimization/84818
19520 * builtins.c (check_access): Use warning_n.
19522 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19525 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
19527 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
19530 2018-03-30 Julia Koval <julia.koval@intel.com>
19533 * x86-tune.def (movx, partial_reg_dependency): Enable for
19536 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
19538 PR inline-asm/84985
19539 * lra-constraints.c (process_alt_operands): Move setting
19540 this_alternative_matches below.
19542 2018-03-29 Martin Liska <mliska@suse.cz>
19545 * doc/invoke.texi: Document how LTO works with debug info.
19546 Describe auto-load support of binutils. Mention 'x86-64'
19547 as valid option value of -march option.
19549 2018-03-29 Jakub Jelinek <jakub@redhat.com>
19551 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
19554 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
19555 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
19556 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
19559 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
19562 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
19563 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
19564 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
19565 for RS6000_BTM_POWERPC64.
19566 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
19567 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
19568 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
19571 (DIVDEU): Likewise.
19573 2018-03-28 Carl Love <cel@us.ibm.com>
19576 2017-09-27 Carl Love <cel@us.ibm.com>
19578 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
19579 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
19580 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
19583 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19585 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
19586 instead of __vector bool.
19587 (_mm_max_pu8): Likewise.
19588 (_mm_min_pi16): Likewise.
19590 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
19593 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
19594 (DIVWEUO): Likewise.
19595 (DIVDEO): Likewise.
19596 (DIVDEUO): Likewise.
19597 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
19598 DIVWEUO and DIVDEUO.
19599 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
19600 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
19601 (div_extend): Likewise.
19602 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
19604 (__builtin_divweuo): Likewise.
19605 (__builtin_divdeo): Likewise.
19606 (__builtin_divdeuo): Likewise.
19608 2018-03-28 Jakub Jelinek <jakub@redhat.com>
19611 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
19612 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
19614 PR tree-optimization/82004
19615 * gimple-match-head.c (optimize_pow_to_exp): New function.
19616 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
19617 Don't fold to exp if optimize_pow_to_exp is false.
19619 2018-03-28 Martin Liska <mliska@suse.cz>
19622 * calls.c (initialize_argument_information): Fix trailing space.
19623 * common.opt: Fix typo and provide better explanation for
19624 -fsanitize-coverage option.
19625 * config/i386/i386.opt: Fix typo.
19627 2018-03-28 Jakub Jelinek <jakub@redhat.com>
19628 Martin Liska <mliska@suse.cz>
19631 * gimplify.c (asan_poison_variable): Don't do the check for
19632 gimplify_omp_ctxp here.
19633 (gimplify_decl_expr): Do it here.
19634 (gimplify_target_expr): Likewise.
19636 2018-03-28 Martin Liska <mliska@suse.cz>
19639 * config/i386/i386.c (ix86_function_arg_advance): Do not call
19640 chkp_type_bounds_count if MPX is not enabled.
19642 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
19644 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
19646 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
19649 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
19650 function to create the function decl for complex long double
19651 multiply and divide for -mabi=ieeelongdouble.
19652 (init_float128_ieee): Call it.
19654 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
19657 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
19658 -fcf-protection=branch -mibt.
19659 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
19661 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19664 * config/arm/arm.c (arm_valid_symbolic_address): Handle
19665 arm_word_relocations.
19667 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
19670 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
19671 extern array declarations.
19673 2018-03-27 Richard Biener <rguenther@suse.de>
19675 PR middle-end/84067
19676 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
19677 explicit single_use checks.
19679 2018-03-27 Richard Biener <rguenther@suse.de>
19681 PR tree-optimization/85082
19682 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
19685 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19687 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
19688 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
19689 Turn on fasynchronous-unwind-tables and funwind-tables.
19691 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
19694 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
19695 (*bmi_blsr_<mode>_ccz): Ditto.
19697 2018-03-26 Tom de Vries <tom@codesourcery.com>
19699 PR tree-optimization/85063
19700 * omp-general.c (offloading_function_p): New function. Factor out
19702 * omp-offload.c (pass_omp_target_link::gate): ... here.
19703 * omp-general.h (offloading_function_p): Declare.
19704 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
19705 with attribute omp declare target for offloading functions.
19707 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
19709 PR tree-optimization/84005
19710 * tree-data-ref.h (get_base_for_alignment): Declare.
19711 * tree-data-ref.c (get_base_for_alignment_1): New function.
19712 (get_base_for_alignment): Likewise.
19713 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
19714 get_base_for_alignment to find a suitable base object, instead
19715 of always using drb->base_address.
19717 2018-03-23 Jakub Jelinek <jakub@redhat.com>
19719 PR inline-asm/85022
19720 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
19721 known size by default.
19723 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
19725 PR inline-asm/85030
19726 * lra-constraints.c (process_alt_operands): Don't match BLKmode
19727 and non BLKmode operands.
19729 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19732 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
19733 Clean up attributes.
19735 2018-03-23 Richard Biener <rguenther@suse.de>
19738 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
19739 we are going to emit early debug for LTO.
19741 2018-03-23 Jakub Jelinek <jakub@redhat.com>
19743 PR inline-asm/85034
19744 * function.c (match_asm_constraints_1): Don't optimize if input
19745 doesn't satisfy general_operand predicate for output's mode.
19747 PR inline-asm/85022
19748 * alias.c (write_dependence_p): Don't require for x_canonicalized
19749 non-VOIDmode if x has VOIDmode.
19752 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
19753 just don't try to optimize it rather than assert it never happens.
19755 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
19757 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
19758 macro expansions for definition of ST_INTERNAL_<mode> and
19759 LD_INTERNAL_<mode> builtins.
19760 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
19762 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
19764 (altivec_expand_st_builtin): Likewise.
19765 (altivec_expand_builtin): Remove calls to deleted functions.
19766 (rs6000_address_for_altivec): Delete this function.
19767 * config/rs6000/vector.md: Remove expands for
19768 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
19770 2018-03-22 Sudakshina Das <sudi.das@arm.com>
19773 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
19774 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
19775 re-computing once computed.
19776 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
19777 (arm_init_machine_status): Initialize
19778 machine->static_chain_stack_bytes.
19780 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
19783 * doc/extend.texi: Add four new prototypes for vec_ld.
19784 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
19785 definitions for more logical presentation.
19786 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
19787 entries for V1TI variants of __builtin_altivec_ld builtin.
19788 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
19789 handling of V1TI variant of LVX icode pattern.
19790 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
19791 (rs6000_gimple_fold_builtin): Likewise.
19792 (altivec_init_builtins): Add code to define
19793 __builtin_altivec_lvx_v1ti function.
19795 2018-03-22 Jakub Jelinek <jakub@redhat.com>
19797 PR inline-asm/84941
19798 * function.c (match_asm_constraints_1): Don't do the optimization
19799 if input isn't a REG, SUBREG, MEM or constant.
19801 2018-03-22 Tom de Vries <tom@codesourcery.com>
19803 PR tree-optimization/84956
19804 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
19805 bb_has_abnormal_pred.
19807 2018-03-22 Jakub Jelinek <jakub@redhat.com>
19810 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
19811 DECL_INITIAL (decl) to decl at the end.
19812 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
19813 adjust the comment.
19815 2018-03-21 Joseph Myers <joseph@codesourcery.com>
19817 * doc/extend.texi (__builtin_tgmath): Document when complex
19818 integer types are treated as _Complex _Float64.
19820 2018-03-21 Tom de Vries <tom@codesourcery.com>
19822 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
19824 2018-03-21 Jakub Jelinek <jakub@redhat.com>
19826 PR tree-optimization/84960
19827 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
19828 if it is ENTRY block, move them into single succ of ENTRY in that case.
19830 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
19832 PR tree-optimization/84811
19833 * poly-int.h (poly_span_traits): Remove the T3 parameter and
19834 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
19835 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
19836 (known_subrange_p): Update accordingly. Cast each value involved
19837 in the size comparison, rather than casting the result of the
19840 2018-03-21 Jakub Jelinek <jakub@redhat.com>
19842 PR tree-optimization/84982
19843 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
19844 by flipping the least significant bit rather than all bits from
19845 bitpos to bitpos + bitsize - 1.
19847 2018-03-21 Nathan Sidwell <nathan@acm.org>
19849 * doc/extend.texi (Deprecated Features): Remove mention of
19850 long-deleted deprecations.
19852 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19855 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
19856 * configure: Regenerate.
19858 2018-03-21 Tom de Vries <tom@codesourcery.com>
19860 PR tree-optimization/83126
19861 * tree-parloops.c (num_phis): New function.
19862 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
19864 2018-03-21 Nathan Sidwell <nathan@acm.org>
19866 * doc/extend.texi (Deprecated Features): Update deprecated flags,
19867 mention anon-struct/union members and trailing attributes.
19869 2018-03-21 Bin Cheng <bin.cheng@arm.com>
19871 PR tree-optimization/84969
19872 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
19873 builtin memset partitions if they set different rhs values.
19875 2018-03-21 Jakub Jelinek <jakub@redhat.com>
19877 PR rtl-optimization/84989
19878 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
19879 VEC_DUPLICATE with scalar result mode.
19881 2018-03-21 Martin Liska <mliska@suse.cz>
19884 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
19885 not intended return statement.
19887 2018-03-21 Martin Liska <mliska@suse.cz>
19890 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
19891 (chkp_find_bound_slots_1): Limit number of iterations.
19893 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
19896 * Minor grammar fixes for x86 options.
19898 2018-03-20 Jakub Jelinek <jakub@redhat.com>
19901 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
19902 holding REG_CFA_RESTORE notes, instead turn them into a USE.
19904 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
19907 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
19908 (altivec_lvx_<mode>_1op): Likewise.
19909 (altivec_stvx_<mode>_2op): Likewise.
19910 (altivec_stvx_<mode>_1op): Likewise.
19911 (altivec_lvx_<VM2:mode>): New define_expand.
19912 (altivec_stvx_<VM2:mode>): Likewise.
19913 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
19914 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
19915 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
19916 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
19917 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
19918 (rs6000_gen_lvx): Likewise.
19919 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
19920 (altivec_expand_stv_builtin): Likewise.
19921 (altivec_expand_builtin): Likewise.
19922 * config/rs6000/vector.md: Likewise.
19924 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19927 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
19930 2018-03-20 Richard Biener <rguenther@suse.de>
19933 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
19934 sign-conversions as zero, fall back to standard scalar_stmt
19937 2018-03-20 Martin Liska <mliska@suse.cz>
19940 * predict.c (rebuild_frequencies): Handle case when we have
19941 PROFILE_ABSENT, but flag_guess_branch_prob is false.
19943 2018-03-20 Jakub Jelinek <jakub@redhat.com>
19946 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
19947 flag_section_anchors.
19948 * varasm.c (use_blocks_for_decl_p): Remove hack for
19949 dw2_force_const_mem.
19952 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
19954 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
19955 be created, use lowpart_subreg of operands[0] rather than operands[0]
19957 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
19958 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
19959 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
19960 and n constraint instead of aarch64_shift_imm_di and Usd.
19961 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
19962 (*aarch64_<optab>_reg_minus<mode>3): ... this.
19964 2018-03-20 Sudakshina Das <sudi.das@arm.com>
19967 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
19968 to favor GPR over NEON registers.
19969 (<shift>di3_neon): Likewise.
19971 2018-03-20 Tom de Vries <tom@codesourcery.com>
19974 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
19975 (nvptx_process_pars): Emit bar.sync asap and alap.
19977 2018-03-20 Tom de Vries <tom@codesourcery.com>
19980 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
19981 seen_label if seen_label is already set.
19983 2018-03-20 Jakub Jelinek <jakub@redhat.com>
19986 * config/i386/i386.c (fold_builtin_cpu): For features above 31
19987 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
19988 Use 1U instead of 1. Formatting fixes.
19991 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
19992 instead of TREE_TYPE (s1) for the return value.
19994 2018-03-19 Jakub Jelinek <jakub@redhat.com>
19996 PR tree-optimization/84946
19997 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
19998 bitsize + bitsize in poly_uint64 rather than poly_int64.
20001 * dwarf2asm.c: Include fold-const.c.
20002 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
20003 of decl rather than decl itself.
20005 PR rtl-optimization/84643
20006 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
20008 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
20011 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
20012 calling assemble_variable.
20014 2018-03-19 Sudakshina Das <sudi.das@arm.com>
20017 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
20018 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
20020 2018-03-19 Jim Wilson <jimw@sifive.com>
20023 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
20024 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
20025 (riscv_first_stack_step): Likewise.
20026 (riscv_option_override): Use STACK_BOUNDARY instead of
20027 MIN_STACK_BOUNDARY.
20028 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
20029 MIN_STACK_BOUNDARY.
20030 (BIGGEST_ALIGNMENT): Set to 128.
20031 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
20032 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
20035 2018-03-19 Richard Biener <rguenther@suse.de>
20037 PR tree-optimization/84933
20038 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
20039 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
20041 2018-03-19 Richard Biener <rguenther@suse.de>
20043 PR tree-optimization/84859
20044 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
20045 (cond_if_else_store_replacement): Perform sinking operation on
20046 single-store BBs regardless of MAX_STORES_TO_SINK setting.
20047 Generalize what a BB with a single eligible store is.
20049 2018-03-19 Richard Biener <rguenther@suse.de>
20051 PR tree-optimization/84929
20052 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
20053 chrec_is_positive against non-chrec arg.
20055 2018-03-19 Tamar Christina <tamar.christina@arm.com>
20058 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
20060 2018-03-18 Martin Liska <mliska@suse.cz>
20062 PR rtl-optimization/84635
20063 * regrename.c (build_def_use): Use matches_mode only when
20066 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
20068 PR tree-optimization/84913
20069 * tree-vect-loop.c (vectorizable_reduction): Don't try to
20070 vectorize chains of COND_EXPRs.
20072 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
20074 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
20076 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
20078 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
20080 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
20082 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
20084 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
20085 Kito Cheng <kito.cheng@gmail.com>
20087 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
20088 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
20089 (nds32_adjust_reg_alloc_order): New function.
20090 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
20092 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
20094 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
20095 nds32_print_operand, nds32_print_operand_address): Use
20096 HOST_WIDE_INT_PRINT_DEC instead.
20098 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
20100 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
20102 2018-03-17 Jakub Jelinek <jakub@redhat.com>
20105 * config/i386/i386.c (initial_ix86_tune_features,
20106 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
20107 unsigned long long.
20108 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
20109 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
20110 rather than 1u << ix86_tune. Formatting fix.
20111 (ix86_option_override_internal): Change ix86_arch_mask from
20112 unsigned int to unsigned HOST_WIDE_INT, initialize to
20113 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
20114 (ix86_function_specific_restore): Likewise.
20116 2018-03-16 Jakub Jelinek <jakub@redhat.com>
20119 * postreload.c (reload_combine_recognize_pattern): Perform
20120 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
20121 truncate_int_for_mode the result for the destination's mode.
20124 * hsa-gen.c (mem_type_for_type): Fix comment typo.
20125 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
20127 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
20130 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
20133 * lra-assigns.c (lra_split_hard_reg_for): Don't use
20134 regno_allocno_class_array and sorted_pseudos.
20135 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
20136 insns where regno is used.
20138 2018-03-16 Martin Liska <mliska@suse.cz>
20141 * multiple_target.c (create_dispatcher_calls): Redirect
20142 reference in the symbol table.
20144 2018-03-16 Martin Liska <mliska@suse.cz>
20147 * multiple_target.c (create_dispatcher_calls): Redirect also
20150 2018-03-16 Jakub Jelinek <jakub@redhat.com>
20154 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
20155 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
20156 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
20158 2018-03-16 Julia Koval <julia.koval@intel.com>
20160 * doc/invoke.texi (Skylake Server): Add CLWB.
20161 Cannonlake): Remove CLWB.
20163 2018-03-16 Jakub Jelinek <jakub@redhat.com>
20165 PR tree-optimization/84841
20166 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
20168 (FLOAT_ONE_CONST_TYPE): Define.
20169 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
20170 (sort_by_operand_rank): Put entries with higher constant_type last
20171 rather than first to match comments.
20173 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
20175 * config/nios2/nios2.md (movsi_internal): Fix thinko in
20178 2018-03-15 Jakub Jelinek <jakub@redhat.com>
20181 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
20182 check and use address of target always.
20184 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
20187 * config/i386/i386.c (indirect_thunk_needed): Update comments.
20188 (indirect_thunk_bnd_needed): Likewise.
20189 (indirect_thunks_used): Likewise.
20190 (indirect_thunks_bnd_used): Likewise.
20191 (indirect_return_needed): New.
20192 (indirect_return_bnd_needed): Likewise.
20193 (output_indirect_thunk_function): Add a bool argument for
20195 (output_indirect_thunk_function): Don't generate alias for
20196 function return thunk.
20197 (ix86_code_end): Call output_indirect_thunk_function to generate
20198 function return thunks.
20199 (ix86_output_function_return): Set indirect_return_bnd_needed
20200 and indirect_return_needed instead of indirect_thunk_bnd_needed
20201 and indirect_thunk_needed.
20203 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
20205 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
20206 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
20207 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
20209 2018-03-15 David Malcolm <dmalcolm@redhat.com>
20210 Paul Hua <paul.hua.gm@gmail.com>
20213 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
20215 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
20217 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
20218 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
20219 resp. SFmode cases.
20221 2018-03-15 Tamar Christina <tamar.christina@arm.com>
20224 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
20225 instead of GET_MODE_SIZE when comparing Units.
20227 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
20230 * varasm.c (hash_section): Return an unchangeble hash value
20231 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
20232 Return !aarch64_can_use_per_function_literal_pools_p ().
20234 2018-03-15 Jakub Jelinek <jakub@redhat.com>
20237 * optabs.c (emit_conditional_move): Pass address of cmode's copy
20238 rather than address of cmode as last argument to prepare_cmp_insn.
20240 2018-03-15 Julia Koval <julia.koval@intel.com>
20242 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
20243 F_AVX512VNNI, F_AVX512BITALG): New.
20245 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
20248 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
20249 insn for floating-point loads and stores.
20251 2018-03-14 Carl Love <cel@us.ibm.com>
20253 * config/rs6000/rs6000-c.c: Add macro definitions for
20254 ALTIVEC_BUILTIN_VEC_PERMXOR.
20255 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
20256 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
20257 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
20258 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
20260 * config/doc/extend.texi: Add prototypes for vec_permxor.
20262 2018-03-14 David Malcolm <dmalcolm@redhat.com>
20265 * diagnostic-show-locus.c (class layout_point): Convert m_line
20266 from int to linenum_type.
20267 (line_span::comparator): Use linenum "compare" function when
20268 comparing line numbers.
20269 (test_line_span): New function.
20270 (layout_range::contains_point): Convert param "row" from int to
20272 (layout_range::intersects_line_p): Likewise.
20273 (layout::will_show_line_p): Likewise.
20274 (layout::print_source_line): Likewise.
20275 (layout::should_print_annotation_line_p): Likewise.
20276 (layout::print_annotation_line): Likewise.
20277 (layout::print_leading_fixits): Likewise.
20278 (layout::annotation_line_showed_range_p): Likewise.
20279 (struct line_corrections): Likewise for field m_row.
20280 (line_corrections::line_corrections): Likewise for param "row".
20281 (layout::print_trailing_fixits): Likewise.
20282 (layout::get_state_at_point): Likewise.
20283 (layout::get_x_bound_for_row): Likewise.
20284 (layout::print_line): Likewise.
20285 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
20286 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
20287 * input.c (selftest::test_linenum_comparisons): New function.
20288 (selftest::input_c_tests): Call it.
20289 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
20290 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
20291 * selftest.h (ASSERT_GT): New macro.
20292 (ASSERT_GT_AT): New macro.
20293 (ASSERT_LT): New macro.
20294 (ASSERT_LT_AT): New macro.
20296 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
20298 PR rtl-optimization/84780
20299 * combine.c (distribute_links): Don't make a link based on pc_rtx.
20301 2018-03-14 Martin Liska <mliska@suse.cz>
20303 * tree.c (record_node_allocation_statistics): Use
20304 get_stats_node_kind.
20305 (get_stats_node_kind): New function extracted from
20306 record_node_allocation_statistics.
20307 (free_node): Use get_stats_node_kind.
20309 2018-03-14 Richard Biener <rguenther@suse.de>
20311 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
20312 that the value-set of ANTIC_IN doesn't grow.
20315 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
20317 (BB_VISITED_WITH_VISITED_SUCCS): New define.
20318 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
20320 2018-03-14 Julia Koval <julia.koval@intel.com>
20322 * config.gcc (icelake-client, icelake-server): New.
20324 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
20325 (initial_ix86_arch_features): Ditto.
20326 (PTA_SKYLAKE): Add SGX.
20327 (PTA_ICELAKE): Remove.
20328 (PTA_ICELAKE_CLIENT): New.
20329 (PTA_ICELAKE_SERVER): New.
20330 (ix86_option_override_internal): Split up icelake on icelake client and
20332 (get_builtin_code_for_version): Ditto.
20333 (fold_builtin_cpu): Ditto.
20334 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
20335 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
20336 * config/i386/i386.h (processor_type): Ditto.
20337 * doc/invoke.texi: Ditto.
20339 2018-03-14 Jakub Jelinek <jakub@redhat.com>
20342 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
20343 INTEGER_CST offset, add it together with bitpos / 8 and
20344 sign extend based on POINTER_SIZE.
20348 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
20351 * config/i386/constraints.md (Yc): New register constraint.
20352 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
20353 Use Yc constraint for alternative 2 of operand 0. Remove
20354 preferred_for_speed attribute.
20356 2018-03-14 Richard Biener <rguenther@suse.de>
20358 PR tree-optimization/84830
20359 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
20360 with the old one to avoid oscillations.
20362 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
20365 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
20367 (assign_by_spills): Return a flag of reload assignment failure.
20368 Do not process the reload assignment failures. Do not spill other
20369 reload pseudos if they has the same reg class. Update n if
20371 (lra_assign): Add a return arg. Set up from the result of
20372 assign_by_spills call.
20373 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
20374 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
20375 usage_insns if it is not NULL.
20376 (spill_hard_reg_in_range): New function.
20377 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
20378 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
20379 function prototypes.
20380 (lra_assign): Change prototype.
20381 * lra.c (lra): Add code to deal with fails by splitting hard reg
20384 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
20386 * config/riscv/riscv.opt (mrelax): New option.
20387 * config/riscv/riscv.c (riscv_file_start): Emit ".option
20388 "norelax" when riscv_mrelax is disabled.
20389 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
20391 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20394 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
20395 reassociation for int modes.
20397 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
20399 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
20400 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
20402 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
20403 * config/aarch64/aarch64-sve.md
20404 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
20405 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
20406 (*extend<mode><Vwide>2): Rename to...
20407 (aarch64_sve_extend<mode><Vwide>2): ...this.
20408 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
20409 renaming the old pattern to...
20410 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
20412 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
20413 define_expand, renaming the old pattern to...
20414 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
20415 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
20416 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
20417 account when deciding which SVE instruction the optab should use.
20418 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
20420 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
20422 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
20423 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
20424 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
20425 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
20426 (tlsdesc_small_<mode>): Turn a define_expand and use
20427 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
20428 (tlsdesc_small_advsimd_<mode>): ...this.
20429 (tlsdesc_small_sve_<mode>): New pattern.
20431 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
20433 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
20434 (UNSPEC_UMUL_HIGHPART): New constants.
20435 (MUL_HIGHPART): New int iteraor.
20436 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
20437 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
20439 (*<su>mul<mode>3_highpart): New define_insn.
20441 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
20444 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
20447 2018-03-13 Martin Liska <mliska@suse.cz>
20450 * (sem_item_optimizer::sem_item_optimizer): Initialize new
20452 (sem_item_optimizer::~sem_item_optimizer): Release it.
20453 (sem_item_optimizer::merge_classes): Register variable aliases.
20454 (sem_item_optimizer::fixup_pt_set): New function.
20455 (sem_item_optimizer::fixup_points_to_sets): Likewise.
20456 * ipa-icf.h: Declare new variables and functions.
20458 2018-03-13 Jakub Jelinek <jakub@redhat.com>
20460 PR middle-end/84834
20461 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
20462 integer_pow2p@2 and test integer_pow2p in condition.
20463 (A < 0 ? C : 0): Similarly for @1.
20465 PR middle-end/84831
20466 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
20467 characters starting at p contain '\0' character, don't look beyond
20471 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
20472 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
20475 * reg-stack.c (change_stack): Change update_end var from int to
20476 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
20477 also call set_block_for_insn on the newly added insns and rescan.
20480 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
20481 on the last operand.
20484 * tree.c (stabilize_reference_1): Return save_expr (e) for
20485 STATEMENT_LIST even if it doesn't have side-effects.
20487 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
20489 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
20491 2018-03-12 Renlin Li <renlin.li@arm.com>
20493 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
20494 aarch64_output_scalar_simd_mov_immediate.
20496 2018-03-12 Martin Sebor <msebor@redhat.com>
20498 PR tree-optimization/83456
20499 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
20500 for perfectly overlapping calls to memcpy.
20501 (gimple_fold_builtin_memory_chk): Same.
20502 (gimple_fold_builtin_strcpy): Handle no-warning.
20503 (gimple_fold_builtin_stxcpy_chk): Same.
20504 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
20506 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
20508 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
20509 parameter. Use it for SFmode.
20510 (rs6000_function_arg_advance_1): Adjust.
20511 (rs6000_function_arg): Adjust.
20512 (rs6000_gimplify_va_arg): Pass false for that new parameter.
20514 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
20516 PR rtl-optimization/84169
20517 PR rtl-optimization/84780
20518 * combine.c (can_combine_p): Check for a 2-insn combination whether
20519 the destination register is used between the two insns, too.
20521 2018-03-12 Richard Biener <rguenther@suse.de>
20523 PR tree-optimization/84803
20524 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
20525 for refs DR analysis didn't process.
20527 2018-03-12 Richard Biener <rguenther@suse.de>
20529 PR tree-optimization/84777
20530 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
20531 force-vectorize loops ignore whether we are optimizing for size.
20533 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
20535 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
20536 (TARGET_MD_ASM_ADJUST): Define.
20538 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
20539 Kito Cheng <kito.cheng@gmail.com>
20540 Chung-Ju Wu <jasonwucj@gmail.com>
20542 * config/nds32/nds32.c (nds32_compute_stack_frame,
20543 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
20544 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
20545 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
20546 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
20547 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
20548 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
20549 * config/nds32/nds32.md (prologue, epilogue): Use macro
20550 NDS32_V3PUSH_AVAILABLE_P to do checking.
20552 2018-03-11 Jakub Jelinek <jakub@redhat.com>
20555 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
20556 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
20557 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
20558 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
20559 addition of most attributes on !orig_type_die or the attribute not
20560 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
20562 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
20563 Chung-Ju Wu <jasonwucj@gmail.com>
20565 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
20566 __NDS32_VH__ macro.
20567 * config/nds32/nds32.opt (mvh): New option.
20569 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
20570 Chung-Ju Wu <jasonwucj@gmail.com>
20572 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
20574 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
20575 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
20578 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
20579 Chung-Ju Wu <jasonwucj@gmail.com>
20581 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
20583 * config/nds32/nds32-multiple.md (strlensi): New pattern.
20584 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
20586 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
20587 Kito Cheng <kito.cheng@gmail.com>
20588 Chung-Ju Wu <jasonwucj@gmail.com>
20590 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
20591 UNSPEC_FFMISM and UNSPEC_FLMISM.
20592 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
20593 for ffb, ffmism and flmism.
20594 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
20595 (unspec_ffmism): Ditto.
20596 (unspec_flmism): Ditto.
20597 (nds32_expand_builtin_impl): Check if string extension is available.
20598 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
20599 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
20601 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
20604 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
20607 * lra-assigns.c (assign_by_spills): Return a flag of reload
20608 assignment failure. Do not process the reload assignment
20609 failures. Do not spill other reload pseudos if they has the same
20611 (lra_assign): Add a return arg. Set up from the result of
20612 assign_by_spills call.
20613 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
20614 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
20615 usage_insns if it is not NULL.
20616 (spill_hard_reg_in_range): New function.
20617 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
20618 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
20619 function prototypes.
20620 (lra_assign): Change prototype.
20621 * lra.c (lra): Add code to deal with fails by splitting hard reg
20624 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
20627 * config/i386/i386.opt: Replace Enforcment with Enforcement.
20629 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
20632 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
20633 (dw_val_node): Add val_symbolic_view.
20634 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
20635 (symview_upper_bound): New.
20636 (new_line_info_table): Initialize symviews_since_reset.
20637 (dwarf2out_source_line): Count symviews_since_reset and set
20638 symview_upper_bound.
20639 (dw_val_equal_p): Handle symview.
20640 (add_AT_symview): New.
20641 (print_dw_val): Handle symview.
20642 (attr_checksum, attr_checksum_ordered): Likewise.
20643 (same_dw_val_p, size_of_die): Likewise.
20644 (value_format, output_die): Likewise.
20645 (add_high_low_attributes): Use add_AT_symview for entry_view.
20646 (dwarf2out_finish): Reset symview_upper_bound, clear
20649 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
20652 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
20653 Add strict argument and use it.
20654 (rs6000_split_multireg_move): Update for new strict argument.
20655 (mem_operand_gpr): Disallow all non-offsettable addresses.
20656 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
20658 2018-03-09 Jakub Jelinek <jakub@redhat.com>
20661 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
20662 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
20663 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
20666 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
20667 decl, use remap_type if we want to use the type.
20669 2018-03-09 Martin Sebor <msebor@redhat.com>
20671 PR tree-optimization/84526
20672 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
20674 (builtin_access::generic_overlap): Be prepared to handle non-array
20677 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
20679 PR rtl-optimization/84682
20680 * lra-constraints.c (process_address_1): Check is_address flag
20681 for address constraints.
20682 (process_alt_operands): Likewise.
20683 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
20684 preprocess_constraints.
20685 * recog.h (preprocess_constraints): Add oploc parameter.
20687 * recog.c (preprocess_constraints): Test address_operand for
20688 CT_ADDRESS constraints.
20690 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
20693 * lra-assigns.c (assign_by_spills): Return a flag of reload
20694 assignment failure. Do not process the reload assignment
20695 failures. Do not spill other reload pseudos if they has the same
20697 (lra_assign): Add a return arg. Set up from the result of
20698 assign_by_spills call.
20699 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
20700 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
20701 usage_insns if it is not NULL.
20702 (spill_hard_reg_in_range): New function.
20703 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
20704 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
20705 function prototypes.
20706 (lra_assign): Change prototype.
20707 * lra.c (lra): Add code to deal with fails by splitting hard reg
20710 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20713 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
20714 Accept complain bool parameter. Only emit errors if it is true.
20715 (arm_parse_cpu_option_name): Likewise.
20716 (arm_target_thumb_only): Adjust callers of the above.
20717 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
20718 prototype to take a default true bool parameter.
20719 (arm_parse_arch_option_name): Likewise.
20721 2018-03-09 David Malcolm <dmalcolm@redhat.com>
20722 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
20726 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
20727 * configure: Regenerate.
20728 * configure.ac ("linker --version-script option"): New.
20729 ("linker soname option"): New.
20731 2018-03-09 Richard Biener <rguenther@suse.de>
20733 PR tree-optimization/84775
20734 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
20735 immediate uses of predicate stmts and mark them modified.
20738 PR tree-optimization/84178
20739 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
20741 (version_loop_for_if_conversion): Delay update_ssa call.
20742 (tree_if_conversion): Delay update_ssa until after predicate
20745 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
20748 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
20749 when the function accesses prior frames.
20751 2018-03-08 Jakub Jelinek <jakub@redhat.com>
20754 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
20755 gen_llsym, otherwise call maybe_gen_llsym.
20757 PR inline-asm/84742
20758 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
20759 has ',' character inside of it.
20761 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20764 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
20765 as clobbering CC_REGNUM.
20767 2018-03-08 Richard Biener <rguenther@suse.de>
20769 PR middle-end/84552
20770 * tree-scalar-evolution.c: Include tree-into-ssa.h.
20771 (follow_copies_to_constant): Do not follow SSA names registered
20774 2018-03-08 Richard Biener <rguenther@suse.de>
20776 PR tree-optimization/84178
20777 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
20779 (version_loop_for_if_conversion): Delay update_ssa call.
20780 (tree_if_conversion): Delay update_ssa until after predicate
20783 2018-03-08 David Malcolm <dmalcolm@redhat.com>
20785 PR tree-optimization/84178
20786 * tree-if-conv.c (release_bb_predicate): Remove the
20787 the assertion that the stmts have NULL use_ops.
20788 Discard the statements, asserting that they haven't
20789 yet been added to a BB.
20791 2018-03-08 Richard Biener <rguenther@suse.de>
20793 PR tree-optimization/84746
20794 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
20795 (phi_translate): Pass in destination ANTIC_OUT set.
20796 (phi_translate_1): Likewise. For a simplified result lookup
20797 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
20798 (phi_translate_set): Adjust.
20799 (do_pre_regular_insertion): Likewise.
20800 (do_pre_partial_partial_insertion): Likewise.
20802 2018-03-08 Martin Liska <mliska@suse.cz>
20804 PR gcov-profile/84735
20805 * doc/gcov.texi: Document usage of profile files.
20806 * gcov-io.h: Document changes in the format.
20808 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
20812 * dwarf2out.c (struct dw_line_info_table): Update comments for
20814 (FORCE_RESET_NEXT_VIEW): New.
20815 (FORCE_RESETTING_VIEW_P): New.
20816 (RESETTING_VIEW_P): Check for -1 too.
20817 (ZERO_VIEW_P): Likewise.
20818 (new_line_info_table): Force-reset next view.
20819 (dwarf2out_begin_function): Likewise.
20820 (dwarf2out_source_line): Simplify zero_view_p initialization.
20821 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
20822 view directly. Omit view when omitting .loc at line 0.
20824 2018-03-08 Jakub Jelinek <jakub@redhat.com>
20826 PR tree-optimization/84740
20827 * tree-switch-conversion.c (process_switch): Call build_constructors
20828 only if info.phi_count is non-zero.
20830 PR tree-optimization/84739
20831 * tree-tailcall.c (find_tail_calls): Check call arguments against
20832 DECL_ARGUMENTS (current_function_decl) rather than
20833 DECL_ARGUMENTS (func) when checking for tail recursion.
20835 2018-03-07 Jakub Jelinek <jakub@redhat.com>
20837 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
20838 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
20839 Volker Reichelt's entry and add entries for people that perform
20840 GCC fuzzy testing and report numerous bugs.
20842 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
20845 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
20846 readonly data in sdata, if that is disabled.
20847 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
20848 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
20849 -mreadonly-in-sdata option.
20851 2018-03-07 Martin Sebor <msebor@redhat.com>
20853 PR tree-optimization/84468
20854 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
20855 basic block when looking for nul assignment.
20857 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
20860 * except.h (output_function_exception_table): Adjust prototype.
20861 * except.c (output_function_exception_table): Remove FNNAME parameter
20862 and add SECTION parameter. Ouput one part of the table at a time.
20863 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
20864 the first part of the exception table and emit unwind directives.
20865 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
20866 (i386_pe_seh_cold_init): Likewise.
20867 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
20868 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
20869 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
20870 (ix86_output_call_insn): Emit a nop in one more case for SEH.
20871 * config/i386/winnt.c: Include except.h.
20872 (struct seh_frame_state): Add reg_offset, after_prologue and
20873 in_cold_section fields.
20874 (i386_pe_seh_end_prologue): Set seh->after_prologue.
20875 (i386_pe_seh_cold_init): New function.
20876 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
20877 to seh->in_cold_section.
20878 (seh_emit_push): Record the offset of the push.
20879 (seh_emit_save): Record the offet of the save.
20880 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
20881 Test seh->after_prologue to disregard the epilogue.
20882 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
20883 (i386_pe_end_cold_function): New function.
20885 2018-03-07 Jakub Jelinek <jakub@redhat.com>
20888 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
20889 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
20892 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
20894 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
20895 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
20897 PR middle-end/84723
20898 * multiple_target.c: Include tree-inline.h and intl.h.
20899 (expand_target_clones): Diagnose and fail if node->definition and
20900 !tree_versionable_function_p (node->decl).
20902 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
20904 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
20906 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
20907 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20908 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
20910 2018-03-06 Jakub Jelinek <jakub@redhat.com>
20913 * combine.c (try_combine): Use reg_or_subregno instead of handling
20914 just paradoxical SUBREGs and REGs.
20916 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
20918 * config/arc/arc.c (arc_finalize_pic): Remove function.
20919 (arc_must_save_register): We use single base PIC register, remove
20920 checks to save/restore the PIC register.
20921 (arc_expand_prologue): Likewise.
20922 * config/arc/arc-protos.h (arc_set_default_type_attributes):
20924 (arc_verify_short): Likewise.
20925 (arc_attr_type): Likewise.
20926 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
20927 (walk_stores): Likewise.
20928 (arc_address_cost): Make it static.
20929 (arc_verify_short): Likewise.
20930 (branch_dest): Likewise.
20931 (arc_attr_type): Likewise.
20932 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
20933 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
20934 (arc_final_prescan_insn): Remove inserting the nops due to
20935 hardware hazards. It is done in reorg step.
20936 (insn_length_variant_t): Remove.
20937 (insn_length_parameters_t): Likewise.
20938 (arc_insn_length_parameters): Likewise.
20939 (arc_get_insn_variants): Likewise.
20940 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
20942 2018-03-06 Jakub Jelinek <jakub@redhat.com>
20944 PR inline-asm/84683
20945 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
20948 PR tree-optimization/84687
20949 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
20951 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
20953 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20955 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
20956 Rename to ppc_speculation_barrier.
20957 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
20958 __builtin_ppc_speculation_barrier.
20960 2018-03-05 Jakub Jelinek <jakub@redhat.com>
20963 * combine.c (combine_simplify_rtx): Don't try to simplify if
20964 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
20967 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
20969 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
20970 to 32 bytes when compiling for POWER9.
20972 2018-03-05 Jakub Jelinek <jakub@redhat.com>
20975 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
20976 regparm >= 3 with no arg reg available also for calls with
20977 flag_force_indirect_call. Pass decl to ix86_function_regparm.
20980 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
20982 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
20984 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
20987 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
20989 2018-03-05 Richard Biener <rguenther@suse.de>
20991 PR tree-optimization/84486
20992 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
20993 When inserting a __builtin_assume_aligned call set the LHS
20994 SSA name alignment info accordingly.
20996 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
20998 PR tree-optimization/84114
20999 * config/aarch64/aarch64.c (aarch64_reassociation_width)
21000 Avoid reassociation of FLOAT_MODE addition.
21002 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
21004 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
21005 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
21006 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
21007 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
21008 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
21009 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
21010 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
21012 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
21013 __builtin_ia32_wbinvd): New builtins.
21014 (SPECIAL_ARGS2): New.
21015 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
21016 (SPECIAL_ARGS2): New.
21017 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
21018 (ix86_valid_target_attribute_inner_p): Ditto.
21019 (ix86_init_mmx_sse_builtins): Add special_args2.
21020 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
21021 TARGET_WBNOINVD_P): New.
21022 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
21023 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
21024 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
21025 * config/i386/immintrin.h (_wbinvd): New intrinsic.
21026 * config/i386/pconfigintrin.h: New file.
21027 * config/i386/wbnoinvdintrin.h: Ditto.
21028 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
21030 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
21032 2018-03-05 Richard Biener <rguenther@suse.de>
21034 PR tree-optimization/84670
21035 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
21037 (BB_VISITED_WITH_VISITED_SUCCS): New define.
21038 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
21039 (compute_antic_aux): Only assert the number of values in ANTIC_IN
21040 doesn't grow if all successors (recursively) were visited at least
21043 2018-03-05 Richard Biener <rguenther@suse.de>
21045 PR tree-optimization/84650
21046 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
21047 if executed in the loop pipeline.
21049 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
21051 * doc/configfiles.texi (Configuration Files): Move info about
21052 conditionalizing $target-protos.h to...
21053 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
21054 differs from $target-protos.h.
21056 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
21057 Chung-Ju Wu <jasonwucj@gmail.com>
21059 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
21060 * config/nds32/nds32-multiple.md (setmemsi): Define.
21061 * config/nds32/nds32-memory-manipulation.c
21062 (nds32_gen_dup_4_byte_to_word_value): New.
21063 (emit_setmem_word_loop): New.
21064 (emit_setmem_byte_loop): New.
21065 (nds32_expand_setmem_loop): New.
21066 (nds32_expand_setmem_loop_v3m): New.
21067 (nds32_expand_setmem_unroll): New.
21068 (nds32_expand_setmem): New.
21070 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
21071 Chung-Ju Wu <jasonwucj@gmail.com>
21073 * config/nds32/nds32-memory-manipulation.c
21074 (nds32_emit_load_store): New.
21075 (nds32_emit_post_inc_load_store): New.
21076 (nds32_emit_mem_move): New.
21077 (nds32_emit_mem_move_block): New.
21078 (nds32_expand_movmemsi_loop_unknown_size): New.
21079 (nds32_expand_movmemsi_loop_known_size): New.
21080 (nds32_expand_movmemsi_loop): New.
21081 (nds32_expand_movmemsi_unroll): New.
21082 (nds32_expand_movmemqi): Rename ...
21083 (nds32_expand_movmemsi): ... to this.
21084 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
21085 (movmemsi): ... to this.
21086 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
21087 (nds32_expand_movmemsi): ... to this.
21089 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
21090 Monk Chiang <sh.chiang04@gmail.com>
21091 Chung-Ju Wu <jasonwucj@gmail.com>
21093 * config/nds32/nds32-protos.h
21094 (nds32_expand_load_multiple): New arguments.
21095 (nds32_expand_store_multiple): Ditto.
21096 (nds32_valid_multiple_load_store): Rename ...
21097 (nds32_valid_multiple_load_store_p): ... to this.
21098 * config/nds32/nds32-memory-manipulation.c
21099 (nds32_expand_load_multiple): Refine implementation.
21100 (nds32_expand_store_multiple): Ditto.
21101 * config/nds32/nds32-multiple.md
21102 (load_multiple): Update nds32_expand_load_multiple interface.
21103 (store_multiple): Update nds32_expand_store_multiple interface.
21104 * config/nds32/nds32-predicates.c
21105 (nds32_valid_multiple_load_store): Rename ...
21106 (nds32_valid_multiple_load_store_p): ... to this and refine
21108 * config/nds32/predicates.md
21109 (nds32_load_multiple_and_update_address_operation): New predicate.
21110 (nds32_store_multiple_and_update_address_operation): New predicate.
21112 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
21113 Chung-Ju Wu <jasonwucj@gmail.com>
21115 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
21116 (combo): New attribute.
21117 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
21119 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
21121 * config/nds32/nds32.opt: Change -mcmodel= default value.
21123 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
21124 Monk Chiang <sh.chiang04@gmail.com>
21125 Chung-Ju Wu <jasonwucj@gmail.com>
21127 * config/nds32/constants.md (unspec_element): New enum.
21128 * config/nds32/constraints.md (Umw): New constraint.
21129 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
21130 * config/nds32/nds32-intrinsic.md: Likewise.
21131 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
21132 (nds32_valid_smw_lwm_base_p): New.
21133 (nds32_output_smw_single_word): New.
21134 (nds32_output_lmw_single_word): New.
21135 (nds32_expand_unaligned_load): New.
21136 (nds32_expand_unaligned_store): New.
21137 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
21138 (nds32_output_smw_single_word): Declare.
21139 (nds32_output_lmw_single_word): Declare.
21140 (nds32_expand_unaligned_load): Declare.
21141 (nds32_expand_unaligned_store): Declare.
21142 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
21143 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
21144 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
21145 NDS32_BUILTIN_UASTORE_DW.
21146 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
21149 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
21150 Kito Cheng <kito.cheng@gmail.com>
21151 Chung-Ju Wu <jasonwucj@gmail.com>
21153 * config/nds32/nds32-intrinsic.c
21154 (nds32_expand_builtin_null_ftype_reg): Delete.
21155 (nds32_expand_builtin_reg_ftype_imm): Ditto.
21156 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
21157 (nds32_read_argument): New.
21158 (nds32_legitimize_target): Ditto.
21159 (nds32_legitimize_argument): Ditto.
21160 (nds32_check_constant_argument): Ditto.
21161 (nds32_expand_unop_builtin): Ditto.
21162 (nds32_expand_unopimm_builtin): Ditto.
21163 (nds32_expand_binop_builtin): Ditto.
21164 (nds32_builtin_decl_impl): Ditto.
21165 (builtin_description): Ditto.
21166 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
21167 (nds32_init_builtins_impl): Ditto.
21168 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
21169 (nds32_builtin_decl): New.
21170 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
21171 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
21173 2018-03-02 Jeff Law <law@redhat.com>
21175 * reorg.c (stop_search_p): Handle DEBUG_INSN.
21176 (redundant_insn, fill_simple_delay_slots): Likewise.
21177 (fill_slots_from_thread): Likewise.
21178 * resource.c (mark_referenced_resources): Likewise.
21179 (mark_set_resources, find_dead_or_set_registers): Likewise.
21181 2018-03-02 Jakub Jelinek <jakub@redhat.com>
21183 * substring-locations.h (format_warning_va): Formatting fix for
21184 ATTRIBUTE_GCC_DIAG.
21185 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
21187 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
21188 * substring-locations.c: Include intl.h.
21189 (format_warning_va): Turned into small wrapper around
21190 format_warning_n_va, renamed to ...
21191 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
21192 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
21194 (format_warning_at_substring_n): New function.
21195 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
21196 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
21197 format_warning_at_substring with just a shorter name instead of
21198 const function pointer.
21199 (fmtwarn_n): New function.
21200 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
21201 appropriate, get rid of all the fmtstr temporaries, move conditionals
21202 with G_() wrapped string literals directly into fmtwarn arguments,
21203 cast dir.len to (int), formatting fixes.
21205 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
21207 * doc/invoke.texi: Remove "Cilk Plus" references.
21209 2018-03-02 Jakub Jelinek <jakub@redhat.com>
21210 Richard Biener <rguenther@suse.de>
21213 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
21214 for error or warning attributes if CALL_FROM_THUNK_P is set.
21217 2018-03-02 Jakub Jelinek <jakub@redhat.com>
21220 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
21221 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
21224 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
21225 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
21227 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
21228 instead of -1U in last predictors element's probability member.
21230 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
21233 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
21234 arguments if they are comparable.
21236 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
21238 PR tree-optimization/84634
21239 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
21240 masks and masked_loop_p with a single loop_masks, making sure it's
21241 null for bb vectorization.
21243 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
21245 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
21246 (vect_analyze_data_ref_access): Use loop->safe_len rather than
21247 loop->force_vectorize to check whether there is no alias.
21249 2018-03-02 Jakub Jelinek <jakub@redhat.com>
21252 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
21254 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
21256 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
21257 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
21258 instead of a loop around prev_real_insn.
21259 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
21262 PR inline-asm/84625
21263 * config/i386/i386.c (ix86_print_operand): Use conditional
21264 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
21267 2018-03-02 Richard Biener <rguenther@suse.de>
21269 PR tree-optimization/84427
21270 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
21271 (bitmap_set_subtract_values): Rewrite to handle multiple
21274 (prune_clobbered_mems): Likewise.
21275 (phi_translate): Take edge instead of pred/phiblock.
21276 (phi_translate_1): Likewise.
21277 (phi_translate_set): Likewise. Insert all translated
21278 exprs for a value into the set, keeping possibly multiple
21279 expressions per value.
21280 (compute_antic_aux): Adjust for phi_translate changes.
21281 When intersecting union the expressions and prune those
21282 not in the final value set, keeping possibly multiple
21283 expressions per value. Do not use value-insertion
21284 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
21285 all expressions. Add verification that the value-sets
21286 only shrink during iteration.
21287 (compute_partial_antic_aux): Adjust for the phi_translate changes.
21288 (do_pre_regular_insertion): Likewise.
21289 (do_pre_partial_partial_insertion): Likewise.
21291 2018-03-02 Richard Biener <rguenther@suse.de>
21294 * config/darwin.c (saved_debug_info_level): New static global.
21295 (darwin_asm_lto_start): Disable debug info generation for LTO out.
21296 (darwin_asm_lto_end): Restore debug info generation settings.
21298 2018-03-01 Martin Liska <mliska@suse.cz>
21301 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
21302 volatile arguments.
21304 2018-03-01 Richard Biener <rguenther@suse.de>
21307 * dwarf2out.c (gen_variable_die): Properly handle late VLA
21308 type annotation with LTO when debug was disabled at compile-time.
21310 2018-03-01 Matthew Fortune <mfortune@gmail.com>
21312 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
21314 (mips_final_postscan_insn): Likewise.
21316 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
21318 PR rtl-optimization/84528
21319 * alias.c (init_alias_target): Add commentary.
21320 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
21321 a unique base value if the frame pointer is not eliminated
21322 to the stack pointer.
21324 2018-03-01 Tom de Vries <tom@codesourcery.com>
21326 PR rtl-optimization/83327
21327 * lra-int.h (hard_regs_spilled_into): Declare.
21328 * lra.c (hard_regs_spilled_into): Define.
21329 (init_reg_info): Init hard_regs_spilled_into.
21330 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
21331 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
21332 (process_bb_lives): Handle hard_regs_spilled_into.
21333 (lra_create_live_ranges_1): Before doing liveness propagation, clear
21334 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
21336 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
21338 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
21339 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
21340 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
21341 * config/rs6000/aix72.h: New file.
21343 2018-02-28 Jakub Jelinek <jakub@redhat.com>
21345 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
21346 instead of warning_at with conditional singular and plural messages
21350 * stor-layout.c (update_alignment_for_field): For
21351 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
21352 && !DECL_PACKED (field), do the alignment update, just use
21353 only desired_align instead of MAX (type_align, desired_align)
21355 (place_field): Don't do known_align < desired_align handling
21356 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
21357 is non-NULL, instead do it after rli->prev_field handling and
21358 only if not within a bitfield word. For DECL_PACKED (field)
21359 use type_align of BITS_PER_UNIT.
21361 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
21363 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
21364 superfluous parentheses and trailing spaces.
21366 2018-02-28 Richard Biener <rguenther@suse.de>
21368 PR tree-optimization/84584
21369 * graphite-scop-detection.c (scop_detection::add_scop): Discard
21370 SCoPs with fake exit edge.
21372 2018-02-28 Martin Liska <mliska@suse.cz>
21375 * timevar.c (timer::print): Fix format to properly print 100%
21378 2018-02-28 Richard Biener <rguenther@suse.de>
21380 PR middle-end/84607
21381 * genmatch.c (capture_info::walk_match): Do not mark
21382 captured expressions without operands as expr_p given
21383 they act more like predicates and should be subject to
21384 "lost tail" side-effect preserving.
21386 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
21388 PR rtl-optimization/81611
21389 * auto-inc-dec.c (attempt_change): Move dead note from
21390 mem_insn if it's the next use of regno
21391 (find_address): Take address use of reg holding
21392 non-incremented value. Add parm to limit search to the named
21394 (merge_in_block): Attempt to use a mem insn that is the next
21395 use of the original regno.
21397 2018-02-27 Martin Sebor <msebor@redhat.com>
21400 * doc/invoke.texi (-Wmissing-attributes): New option.
21401 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
21403 2018-02-27 Martin Sebor <msebor@redhat.com>
21405 PR translation/84207
21406 * diagnostic-core.h (warning_n, error_n, inform_n): Change
21407 n argument to unsigned HOST_WIDE_INT.
21408 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
21409 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
21410 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
21411 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
21413 2018-02-27 Richard Biener <rguenther@suse.de>
21415 PR tree-optimization/84512
21416 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
21417 Do not use the estimate returned from record_stmt_cost for
21418 the scalar iteration cost but sum properly using add_stmt_cost.
21420 2018-02-27 Richard Biener <rguenther@suse.de>
21422 PR tree-optimization/84466
21423 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
21424 Adjust last change to less strictly validate use operands.
21426 2018-02-27 Martin Liska <mliska@suse.cz>
21428 PR gcov-profile/84548
21429 * gcov.c (process_file): Allow partial overlap and consider it
21430 also as group functions.
21431 (output_lines): Properly calculate range of lines for a group.
21433 2018-02-27 Martin Liska <mliska@suse.cz>
21435 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
21436 'ggc' suffixes. Change first column width.
21437 (timer::print): Fix formatting of the column.
21439 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
21441 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
21442 preserve inline entry blocks for the sake of debug inline
21443 entry point markers alone.
21444 (remove_unused_locals): Suggest in comments a better place to
21445 force the preservation of inline entry blocks that are
21446 otherwise unused, but do not preserve them.
21448 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
21450 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
21452 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
21455 * config/i386/constraints.md (Bs): Replace
21456 ix86_indirect_branch_register with
21457 TARGET_INDIRECT_BRANCH_REGISTER.
21459 * config/i386/i386.md (indirect_jump): Likewise.
21460 (tablejump): Likewise.
21461 (*sibcall_memory): Likewise.
21462 (*sibcall_value_memory): Likewise.
21463 Peepholes of indirect call and jump via memory: Likewise.
21464 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
21465 (*sibcall_value_GOT_32): Likewise.
21466 * config/i386/predicates.md (indirect_branch_operand): Likewise.
21467 (GOT_memory_operand): Likewise.
21468 (call_insn_operand): Likewise.
21469 (sibcall_insn_operand): Likewise.
21470 (GOT32_symbol_operand): Likewise.
21471 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
21473 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
21475 PR rtl-optimization/83496
21476 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
21477 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
21478 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
21479 redundant insn, if any.
21480 (relax_delay_slots): Likewise.
21481 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
21483 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
21485 PR tree-optimization/83965
21486 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
21487 that grouped statements are part of a reduction chain. Return
21488 true if the statement is not marked as a reduction itself but
21489 is part of a group.
21490 (vect_recog_dot_prod_pattern): Don't check whether the statement
21491 is part of a group here.
21492 (vect_recog_sad_pattern): Likewise.
21493 (vect_recog_widen_sum_pattern): Likewise.
21495 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
21498 * final.c (rest_of_clean_state): Also look for calls inside sequences.
21500 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
21503 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
21505 (ix86_output_indirect_function_return): New prototype.
21506 (ix86_split_simple_return_pop_internal): Likewise.
21507 * config/i386/i386.c (indirect_return_via_cx): New.
21508 (indirect_return_via_cx_bnd): Likewise.
21509 (indirect_thunk_name): Handle return va CX_REG.
21510 (output_indirect_thunk_function): Create alias for
21511 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
21512 (ix86_output_indirect_jmp): Remove the bool argument.
21513 (ix86_output_indirect_function_return): New function.
21514 (ix86_split_simple_return_pop_internal): Likewise.
21515 * config/i386/i386.md (*indirect_jump): Don't pass false
21516 to ix86_output_indirect_jmp.
21517 (*tablejump_1): Likewise.
21518 (simple_return_pop_internal): Change it to define_insn_and_split.
21519 Call ix86_split_simple_return_pop_internal to split it for
21520 -mfunction-return=.
21521 (simple_return_indirect_internal): Call
21522 ix86_output_indirect_function_return instead of
21523 ix86_output_indirect_jmp.
21525 2018-02-26 Jakub Jelinek <jakub@redhat.com>
21528 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
21529 memset and value initialization afterwards.
21531 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
21533 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
21535 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21538 * common/config/aarch64/aarch64-common.c
21539 (aarch_option_optimization_table[]): Switch
21540 off fomit-frame-pointer
21542 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
21543 Chung-Ju Wu <jasonwucj@gmail.com>
21545 * config/nds32/nds32-multiple.md (load_multiple): Disallow
21547 (store_multiple): Ditto.
21549 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
21551 * config.gcc: Add --with-cpu support for nds32 target.
21552 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
21553 * config/nds32/nds32.opt: Add -mcpu= option.
21555 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
21557 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
21558 isel=yes): Warn for these deprecated options.
21560 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
21562 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
21563 ISA_2_5_MASKS_EMBEDDED.
21565 2018-02-23 Jakub Jelinek <jakub@redhat.com>
21567 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
21568 p->max as pointers rather than using iterative_hash_expr.
21570 2018-02-23 Carl Love <cel@us.ibm.com>
21572 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
21573 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
21575 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
21576 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
21577 P8V_BUILTIN_VEC_VUNSIGNED2.
21579 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
21582 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
21583 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
21585 * lra-constraints.c (curr_insn_transform): Set up
21586 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
21588 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
21589 * lra-eliminations.c (spill_pseudos): Ditto.
21590 (process_insn_for_elimination): Ditto.
21591 * lra-lives.c (reg_early_clobber_p): Use the new macros.
21592 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
21593 LRA_NON_CLOBBERED_ALT.
21595 2018-02-22 Martin Sebor <msebor@redhat.com>
21597 PR tree-optimization/84480
21598 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
21599 to maybe_diag_stxncpy_trunc. Call it.
21600 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
21601 from gimple_fold_builtin_strcpy. Print inlining stack.
21602 (handle_builtin_stxncpy): Print inlining stack.
21603 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
21605 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
21608 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
21609 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
21610 and -fcheck-pointer-bounds are used together.
21611 (indirect_thunk_prefix): New enum.
21612 (indirect_thunk_need_prefix): New function.
21613 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
21614 "_nt" instead of "_bnd" for NOTRACK prefix.
21615 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
21616 (output_indirect_thunk_function): Likewise.
21618 (ix86_code_end): Update output_indirect_thunk_function calls.
21619 (ix86_output_indirect_branch_via_reg): Replace
21620 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
21621 (ix86_output_indirect_branch_via_push): Likewise.
21622 (ix86_output_function_return): Likewise.
21623 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
21624 incompatible with -fcf-protection=branch and
21625 -fcheck-pointer-bounds.
21627 2018-02-22 Steve Ellcey <sellcey@cavium.com>
21630 * config/aarch64/aarch64.c (aarch64_print_address_internal):
21631 Change gcc_assert call to output_operand_lossage.
21633 2018-02-22 Steve Ellcey <sellcey@cavium.com>
21635 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
21637 2018-02-22 DJ Delorie <dj@redhat.com>
21638 Sebastian Perta <sebastian.perta@renesas.com>
21639 Oleg Endo <olegendo@gcc.gnu.org>
21641 * config/rx/rx.c (rx_rtx_costs): New function.
21642 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
21644 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
21646 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
21648 2018-02-22 Martin Liska <mliska@suse.cz>
21651 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
21652 Add "native" as a possible value.
21654 2018-02-22 Martin Liska <mliska@suse.cz>
21657 * config/i386/i386.c (ix86_option_override_internal):
21658 Add "native" as a possible value for -march and -mtune.
21660 2018-02-22 Jakub Jelinek <jakub@redhat.com>
21663 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
21664 to all type variants.
21666 PR tree-optimization/84503
21667 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
21668 width as info->bitpos + info->bitsize - start.
21669 (merged_store_group::merge_overlapping): Simplify width computation.
21670 (check_no_overlap): New function.
21671 (imm_store_chain_info::try_coalesce_bswap): Compute expected
21672 start + width and last_order of the group, fail if check_no_overlap
21674 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
21675 to group if check_no_overlap fails.
21677 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
21679 * config/rs6000/altivec.md: Delete contraint arguments to
21680 define_expand, define_split, and define_peephole2, and in
21681 define_insn_and_split if always unused.
21682 * config/rs6000/darwin.md: Ditto.
21683 * config/rs6000/dfp.md: Ditto.
21684 * config/rs6000/rs6000.md: Ditto.
21685 * config/rs6000/sync.md: Ditto.
21686 * config/rs6000/vector.md: Ditto.
21687 * config/rs6000/vsx.md: Ditto.
21689 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
21691 * config/rs6000/altivec.md: Write output control strings as braced
21692 blocks instead of double-quoted strings.
21693 * config/rs6000/darwin.md: Ditto.
21694 * config/rs6000/rs6000.md: Ditto.
21695 * config/rs6000/vector.md: Ditto.
21696 * config/rs6000/vsx.md: Ditto.
21698 2018-02-21 Jason Merrill <jason@redhat.com>
21700 PR c++/84314 - ICE with templates and fastcall attribute.
21701 * attribs.c (build_type_attribute_qual_variant): Remove assert.
21703 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
21705 * ipa-cp.c (determine_versionability): Fix comment typos.
21707 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
21710 * ipa-cp.c (determine_versionability): Do not version functions caling
21713 2018-02-21 Martin Liska <mliska@suse.cz>
21716 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
21717 Add "native" as a possible value.
21718 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
21719 the macro when native cpu detection is available.
21721 2018-02-21 Martin Liska <mliska@suse.cz>
21724 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
21725 Add "native" as a possible value.
21726 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
21727 when native cpu detection is available.
21729 2018-02-21 Jakub Jelinek <jakub@redhat.com>
21730 Martin Sebor <msebor@redhat.com>
21732 PR tree-optimization/84478
21733 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
21735 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
21736 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
21737 support which is conservatively correct, for 2 only stay conservative
21738 for maxlen. Formatting and comment capitalization fixes. Add STRICT
21739 argument to the 2 argument get_range_strlen, adjust 6 arg
21740 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
21742 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
21743 (gimple_fold_builtin_strlen): Pass true as last argument to
21746 2018-02-20 Martin Sebor <msebor@redhat.com>
21748 PR middle-end/84095
21749 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
21750 (builtin_memref::set_base_and_offset): Same. Handle inner references.
21751 (builtin_memref::builtin_memref): Factor out parts into
21752 set_base_and_offset and call it.
21754 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
21756 PR middle-end/84406
21757 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
21758 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
21759 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
21760 search at the associated MODE_INT.
21762 2018-02-20 Jeff Law <law@redhat.com>
21764 PR middle-end/82123
21765 PR tree-optimization/81592
21766 PR middle-end/79257
21767 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
21768 for range data rather than using global data.
21769 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
21770 range data rather than using global data.
21771 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
21772 pass it to children as needed.
21773 (struct directive::fmtresult): Similarly.
21774 (struct directive::set_width): Similarly.
21775 (struct directive::set_precision): Similarly.
21776 (format_integer, format_directive, parse_directive): Similarly.
21777 (format_none): Accept unnamed vr_values parameter.
21778 (format_percent, format_floating, format_character): Similarly.
21779 (format_string, format_plain): Similarly.
21780 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
21781 the EVRP range analyzer for range data rather than using global data.
21782 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
21783 gimple-ssa-evrp-analyze.h
21784 (class sprintf_dom_walker): Add after_dom_children member function.
21785 Add evrp_range_analyzer member.
21786 (sprintf_dom_walker::before_dom_children): Call into the EVRP
21787 range analyzer as needed.
21788 (sprintf_dom_walker::after_dom_children): New member function.
21789 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
21791 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
21792 (evrp_range_analyzer::pop_to_marker): Likewise.
21794 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
21796 PR tree-optimization/84419
21797 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
21798 with the required type if its current type is compatible but
21801 2018-02-20 Jakub Jelinek <jakub@redhat.com>
21803 PR middle-end/82004
21804 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
21805 after vectorization.
21807 2018-02-20 Martin Liska <mliska@suse.cz>
21810 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
21811 possible values if we don't have a hint.
21813 2018-02-20 Martin Liska <mliska@suse.cz>
21817 * final.c (shorten_branches): Build align_tab array with one
21819 * opts.c (finish_options): Add alignment option limit check.
21820 (MAX_CODE_ALIGN): Likewise.
21821 (MAX_CODE_ALIGN_VALUE): Likewise.
21822 * doc/invoke.texi: Document maximum allowed option value for
21823 all -falign-* options.
21825 2018-02-19 Jakub Jelinek <jakub@redhat.com>
21828 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
21829 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
21830 * var-tracking.c (emit_note_insn_var_location): Remove all references
21831 to NOTE_INSN_CALL_ARG_LOCATION.
21832 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
21833 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
21834 Use copy_rtx_if_shared.
21835 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
21836 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
21837 (dwarf2out_var_location): Remove handling of
21838 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
21840 * final.c (final_scan_insn): Remove all references to
21841 NOTE_INSN_CALL_ARG_LOCATION.
21842 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
21843 before dumping final insns.
21844 * except.c (emit_note_eh_region_end): Remove all references to
21845 NOTE_INSN_CALL_ARG_LOCATION.
21846 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
21847 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
21848 * config/arc/arc.c (hwloop_optimize): Likewise.
21849 * config/arm/arm.c (create_fix_barrier): Likewise.
21850 * config/s390/s390.c (s390_chunkify_start): Likewise.
21851 * config/sh/sh.c (find_barrier): Likewise.
21852 * config/i386/i386.c (rest_of_insert_endbranch,
21853 ix86_seh_fixup_eh_fallthru): Likewise.
21854 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
21855 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
21856 * config/frv/frv.c (frv_function_prologue): Likewise.
21857 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
21859 (note_outside_basic_block_p): Remove all references to
21860 NOTE_INSN_CALL_ARG_LOCATION.
21861 * gengtype.c (adjust_field_rtx_def): Likewise.
21862 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
21864 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
21865 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
21868 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
21871 PR tree-optimization/84452
21872 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
21873 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
21876 2018-02-19 Martin Liska <mliska@suse.cz>
21879 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
21881 2018-02-19 Martin Liska <mliska@suse.cz>
21882 Richard Sandiford <richard.sandiford@linaro.org>
21884 PR tree-optimization/82491
21885 * gimple-fold.c (get_base_constructor): Make earlier bail out
21888 2018-02-19 Carl Love <cel@us.ibm.com>
21890 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
21891 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
21893 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
21894 P8V_BUILTIN_VEC_NEG.
21896 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
21898 * config/rl78/rl78.md (movdf): New define expand.
21900 2018-02-19 Martin Liska <mliska@suse.cz>
21903 * doc/invoke.texi: Fix typo.
21904 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
21906 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
21908 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
21909 handle rs6000_single_float and rs6000_double_float specially for
21912 2018-02-16 Jeff Law <law@redhat.com>
21914 * config/rx/rx.c (add_pop_cfi_notes): New function.;
21915 (pop_regs): Use it.
21917 2018-02-16 Jakub Jelinek <jakub@redhat.com>
21920 * ipa-inline.c (inline_small_functions): Fix a typo.
21922 2018-02-16 Nathan Sidwell <nathan@acm.org>
21924 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
21926 2018-02-16 Carl Love <cel@us.ibm.com>
21928 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
21929 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
21930 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
21931 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
21932 expansion to P8V_BUILTIN_VEC_FLOAT2.
21934 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
21936 PR rtl-optimization/70023
21937 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
21938 src_regno into account.
21940 2018-02-16 Carl Love <cel@us.ibm.com>
21942 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
21943 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
21944 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
21945 * config/rs6000/rs6000.c: Remove case statements for
21946 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
21947 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
21948 and P9V_BUILTIN_VEC_VINSERT4B.
21949 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
21950 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
21951 * config/rs6000/vsx.md:
21952 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
21955 2018-02-16 Carl Love <cel@us.ibm.com>
21957 * config/rs6000/altivec.h: Add builtin names vec_extract4b
21959 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
21961 * config/rs6000/rs6000-c.c: Add the definitions for
21962 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
21963 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
21964 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
21965 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
21966 definition for insert4b and define insn *insert3b_internal.
21967 * doc/extend.texi: Add documentation for vec_extract4b.
21969 2018-02-16 Nathan Sidwell <nathan@acm.org>
21971 * doc/extend.texi (Backwards Compatibility): Mention friend
21972 injection. Note for-scope is deprecated.
21973 * doc/invoke.texi (-ffriend-injection): Deprecate.
21975 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
21977 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
21978 that moved to I2, also allow destinations that are a paradoxical
21979 subreg (instead of a normal reg).
21981 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
21984 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
21987 2018-02-16 Richard Biener <rguenther@suse.de>
21989 PR tree-optimization/84037
21990 PR tree-optimization/84016
21992 * config/i386/i386.c (ix86_builtin_vectorization_cost):
21993 Adjust vec_construct for the fact we need additional higher latency
21994 128bit inserts for AVX256 and AVX512 vector builds.
21995 (ix86_add_stmt_cost): Scale vector construction cost for
21998 2018-02-16 Richard Biener <rguenther@suse.de>
22000 PR tree-optimization/84417
22001 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
22002 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
22003 (non_rewritable_lvalue_p): Likewise, use poly-ints.
22005 2018-02-16 Martin Liska <mliska@suse.cz>
22008 * internal-fn.def (ASAN_CHECK): Set proper flags.
22009 (ASAN_MARK): Likewise.
22011 2018-02-16 Julia Koval <julia.koval@intel.com>
22013 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
22014 from PTA_CANNONLAKE.
22016 2018-02-16 Jakub Jelinek <jakub@redhat.com>
22019 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
22020 Use ++iter rather than iter++ for std::list iterators.
22021 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
22022 defer deleting them until all nodes in the forest are processed. Do
22023 free even leaf nodes. Change to_process into auto_vec.
22026 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
22027 * vec.h (vec_default_construct): Use memset instead of placement new
22028 if BROKEN_VALUE_INITIALIZATION is defined.
22029 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
22030 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
22033 PR rtl-optimization/83723
22034 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
22035 * lra.c (lra_substitute_pseudo): Likewise. If true, use
22036 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
22038 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
22040 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
22042 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
22044 PR rtl-optimization/81443
22045 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
22046 from inner REGs to paradoxical SUBREGs.
22048 2018-02-16 Richard Biener <rguenther@suse.de>
22050 PR tree-optimization/84399
22051 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
22052 For operands we can analyze at their definition make sure we can
22053 analyze them at each use as well.
22055 2018-02-16 Richard Biener <rguenther@suse.de>
22057 PR tree-optimization/84190
22058 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
22059 volatile accesses if the decl isn't volatile.
22061 2018-02-15 Jason Merrill <jason@redhat.com>
22063 PR c++/84314 - ICE with templates and fastcall attribute.
22064 * attribs.c (build_type_attribute_qual_variant): Don't clobber
22065 TYPE_CANONICAL on an existing type.
22067 2018-02-15 Jakub Jelinek <jakub@redhat.com>
22069 PR tree-optimization/84383
22070 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
22071 dstoff nor call operand_equal_p if dstbase is NULL.
22073 PR tree-optimization/84334
22074 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
22075 also a CONSTANT_CLASS_P, punt.
22077 2018-02-14 Jim Wilson <jimw@sifive.com>
22079 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
22080 first SMALL_OPERAND check. New local min_second_step. Move assert
22081 to where locals are set. Add TARGET_RVC support.
22082 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
22084 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
22086 * doc/invoke.texi: Correct -Wformat-overflow code sample.
22088 2018-02-14 Martin Sebor <msebor@redhat.com>
22090 PR tree-optimization/83698
22091 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
22092 arrays constrain the offset range to their bounds.
22093 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
22094 (builtin_access::overlap): Avoid setting the size of overlap if it's
22096 (maybe_diag_overlap): Also consider arrays when deciding what values
22097 of offsets to include in diagnostics.
22099 2018-02-14 Martin Sebor <msebor@redhat.com>
22102 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
22103 that correspond to the kind of a declaration.
22105 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
22108 * config/pa/pa.md: Load address of PIC label using the linkage table
22109 if the label is nonlocal.
22111 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
22113 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
22114 warning message if user requests -maltivec=be.
22115 * doc/invoke.texi: Document deprecation of -maltivec=be.
22117 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
22120 * config/rs6000/rs6000-c.c: Update definitions for
22121 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
22122 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
22124 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
22127 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
22128 add _get_ssp intrinsics. Remove argument from
22129 __builtin_ia32_rdssp[d|q].
22130 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
22131 * config/i386/i386-builtin.def: Remove argument from
22132 __builtin_ia32_rdssp[d|q].
22133 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
22134 ix86_expand_special_args_builtin for _rdssp[d|q].
22135 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
22136 Clear register before usage.
22137 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
22138 Add documentation for new _get_ssp and _inc_ssp intrinsics.
22140 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
22142 PR tree-optimization/84357
22143 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
22144 operand 1 of an ARRAY_REF too.
22146 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
22149 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
22150 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
22152 (set_of_reg): New struct.
22153 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
22154 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
22155 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
22157 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
22158 Split into bitclr, bitset, bitinvert patterns if appropriate.
22159 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
22160 use rx_fuse_in_memory_bitop.
22161 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
22162 to named insn, correct maximum insn length.
22164 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
22167 * machmode.def: Define a complex mode for PARTIAL_INT.
22168 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
22170 * doc/rtl.texi: Document CSPImode.
22171 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
22173 (msp430_hard_regno_nregs_with_padding): Likewise.
22175 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
22178 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
22180 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
22182 PR rtl-optimization/84169
22183 * combine.c (try_combine): New variable split_i2i3. Set it to true if
22184 we generated a parallel as new i3 and we split that to new i2 and i3
22185 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
22186 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
22187 those to i2, not i1. Partially rewrite this scan code.
22189 2018-02-13 Jakub Jelinek <jakub@redhat.com>
22192 * stor-layout.c (place_field): For variable length fields, adjust
22193 offset_align afterwards not just based on the field's alignment,
22194 but also on the size.
22196 PR middle-end/84309
22197 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
22198 of exps and logs in the use_exp2 case.
22200 2018-02-13 Jeff Law <law@redhat.com>
22202 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
22203 entry for "vector".
22205 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
22208 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
22212 * common.opt (gas-loc-support, gas-locview-support): New.
22213 (ginline-points, ginternal-reset-location-views): New.
22214 * doc/invoke.texi: Document them. Use @itemx where intended.
22215 (gvariable-location-views): Adjust.
22216 * target.def (reset_location_view): New.
22217 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
22218 (TARGET_RESET_LOCATION_VIEW): New.
22219 * doc/tm.texi: Rebuilt.
22220 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
22221 (dwarf2out_default_as_locview_support): New.
22222 (output_asm_line_debug_info): Use option variables.
22223 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
22224 (output_loc_list): Likewise.
22225 (add_high_low_attributes): Check option variables.
22226 Don't output entry view attribute in strict mode.
22227 (gen_inlined_subroutine_die): Check option variables.
22228 (dwarf2out_inline_entry): Likewise.
22229 (init_sections_and_labels): Likewise.
22230 (dwarf2out_early_finish): Likewise.
22231 (maybe_reset_location_view): New, from...
22232 (dwarf2out_var_location): ... here. Call it.
22233 * debug.h (dwarf2out_default_as_loc_support): Declare.
22234 (dwarf2out_default_as_locview_support): Declare.
22235 * hooks.c (hook_int_rtx_insn_0): New.
22236 * hooks.h (hook_int_rtx_insn_0): Declare.
22237 * toplev.c (process_options): Take -gas-loc-support and
22238 -gas-locview-support from dwarf2out. Enable
22239 -gvariable-location-views by default only with locview
22240 assembler support. Enable -ginternal-reset-location-views by
22241 default only if the target defines the corresponding hook.
22242 Enable -ginline-points by default if location views are
22243 enabled; force it disabled if statement frontiers are
22245 * tree-inline.c (expand_call_inline): Check option variables.
22246 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
22248 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
22250 PR tree-optimization/84321
22251 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
22252 handling. Also check whether the anti-range contains any values
22253 that satisfy the mask; switch to a VR_RANGE if not.
22255 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
22258 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
22260 2018-02-13 Martin Jambor <mjambor@suse.cz>
22263 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
22264 of call statements, also set location of a load to a temporary.
22266 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
22268 * config/rl78/rl78.c (add_vector_labels): New function.
22269 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
22270 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
22271 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
22272 which checks that no arguments are passed.
22273 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
22274 * doc/extend.texi: Documentation for the new attribute.
22276 2018-02-13 Andreas Schwab <schwab@suse.de>
22278 * config/riscv/linux.h (CPP_SPEC): Define.
22280 2018-02-13 Jakub Jelinek <jakub@redhat.com>
22283 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
22284 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
22285 OPTION_MASK_ISA_AES as first argument to def_builtin_const
22286 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
22287 instead of OPTION_MASK_ISA_PCLMUL as first argument to
22288 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
22289 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
22290 temporarily for AES and PCLMUL builtins.
22292 PR tree-optimization/84339
22293 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
22294 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
22297 PR middle-end/84309
22298 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
22299 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
22300 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
22302 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
22304 * omp-simd-clone.h: New file.
22305 * omp-simd-clone.c: Include omp-simd-clone.h.
22306 (expand_simd_clones): No longer static.
22307 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
22308 cgraph.h and omp-simd-clone.h.
22309 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
22310 (vect_recog_widen_shift_pattern): Formatting fix.
22311 (vect_pattern_recog_1): Don't check optab for calls.
22314 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
22315 operands[2] into a REG before using gen_lowpart on it.
22317 2018-02-12 Jeff Law <law@redhat.com>
22320 * config/sh/sh.c (find_barrier): Consider a sibling call
22323 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
22324 successfully back substituting a reg.
22326 2018-02-12 Richard Biener <rguenther@suse.de>
22328 PR tree-optimization/84037
22329 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
22330 parameter, move visited init to caller.
22331 (vect_slp_analyze_operations): Separate cost from validity
22332 check, initialize visited once for all instances.
22333 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
22335 * tree-vect-stmts.c (vect_model_simple_cost): Make early
22337 (vect_model_promotion_demotion_cost): Likewise.
22338 (vectorizable_bswap): Guard cost modeling with !slp_node
22339 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
22341 (vectorizable_call): Likewise.
22342 (vectorizable_conversion): Likewise.
22343 (vectorizable_assignment): Likewise.
22344 (vectorizable_shift): Likewise.
22345 (vectorizable_operation): Likewise.
22346 (vectorizable_store): Likewise.
22347 (vectorizable_load): Likewise.
22348 (vectorizable_condition): Likewise.
22349 (vectorizable_comparison): Likewise.
22351 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
22354 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
22355 (ASAN_MARK): Fix fnspec to account for return value, change pointer
22356 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
22358 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
22360 PR middle-end/83665
22361 * params.def (inline-min-speedup): Increase from 8 to 15.
22362 (max-inline-insns-auto): Decrease from 40 to 30.
22363 * ipa-split.c (consider_split): Add some buffer for function to
22364 be considered inlining candidate.
22365 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
22368 2018-02-12 Richard Biener <rguenther@suse.de>
22370 PR tree-optimization/84037
22371 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
22372 matched stmts if we cannot swap the non-matched ones.
22374 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
22376 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
22377 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
22378 _mm_maskz_scalef_round_ss): New intrinsics.
22379 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
22380 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
22381 __builtin_ia32_scalefss_round): Remove.
22382 (__builtin_ia32_scalefsd_mask_round,
22383 __builtin_ia32_scalefss_mask_round): New intrinsics.
22384 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
22385 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
22386 ((match_operand:VF_128 2 "<round_nimm_predicate>"
22387 "<round_constraint>")): Changed to ...
22388 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
22389 "<round_scalar_constraint>")): ... this.
22390 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
22391 %0, %1, %2<round_op3>}"): Changed to ...
22392 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
22393 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
22394 %2<round_scalar_mask_op3>}"): ... this.
22395 * config/i386/subst.md (round_scalar_nimm_predicate): New.
22397 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
22399 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
22400 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
22401 (_mm_maskz_sqrt_round_ss): New intrinsics.
22402 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
22403 (__builtin_ia32_sqrtsd_mask_round)
22404 (__builtin_ia32_sqrtss_mask_round): New builtins.
22405 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
22406 (__builtin_ia32_sqrtss_round): Remove.
22407 (__builtin_ia32_sqrtsd_mask_round)
22408 (__builtin_ia32_sqrtss_mask_round): New builtins.
22409 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
22410 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
22411 ((match_operand:VF_128 1 "vector_operand"
22412 "xBm,<round_constraint>")): Changed to ...
22413 ((match_operand:VF_128 1 "vector_operand"
22414 "xBm,<round_scalar_constraint>")): ... this.
22415 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
22416 %0, %2, %<iptr>1<round_op3>}): Changed to ...
22417 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
22418 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
22419 %<iptr>1<round_scalar_mask_op3>}): ... this.
22420 ((set_attr "prefix" "<round_prefix>")): Changed to ...
22421 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
22423 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
22426 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
22427 Cast vec_cmpeq result to correct type.
22428 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
22429 Cast vec_cmpgt result to correct type.
22431 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
22433 * final.c (final_scan_insn_1): Renamed from...
22434 (final_scan_insn): ... this. New wrapper, to recover
22435 seen from the outermost call in recursive ones.
22436 * config/sparc/sparc.c (output_return): Drop seen from call.
22437 (output_sibcall): Likewise.
22438 * config/visium/visium.c (output_branch): Likewise.
22440 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
22442 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
22445 2018-02-10 Alan Modra <amodra@gmail.com>
22448 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
22449 Specify LR as an input.
22451 2018-02-10 Jakub Jelinek <jakub@redhat.com>
22454 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
22455 remove_member_access_dummy_vars): New functions.
22456 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
22457 lower_omp_1, execute_lower_omp): Use them.
22459 PR rtl-optimization/84308
22460 * shrink-wrap.c (spread_components): Release todo vector.
22462 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
22464 PR rtl-optimization/57193
22465 * ira-color.c (struct allocno_color_data): Add member
22466 conflict_allocno_hard_prefs.
22467 (update_conflict_allocno_hard_prefs): New.
22468 (bucket_allocno_compare_func): Add a preference based on
22469 conflict_allocno_hard_prefs.
22470 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
22471 (color_allocnos): Remove a dead code. Initiate
22472 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
22474 2018-02-09 Jakub Jelinek <jakub@redhat.com>
22477 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
22478 constraint from =wa to wa. Avoid a subreg on the output operand,
22479 instead use a pseudo and subreg it in a move.
22480 (p9_xxbrd_<mode>): Changed to ...
22481 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
22482 (p9_xxbrd_v2df): New expander.
22483 (p9_xxbrw_<mode>): Changed to ...
22484 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
22485 (p9_xxbrw_v4sf): New expander.
22487 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
22489 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
22491 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
22494 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
22495 multiply in 32-bit mode.
22496 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
22497 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
22500 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
22502 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
22503 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
22504 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
22505 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
22507 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
22510 * dwarf2out.c (is_trivial_indirect_ref): New function.
22511 (dwarf2out_late_global_decl): Do not generate a location
22512 attribute for variables that have a non-trivial DECL_VALUE_EXPR
22513 and that are not defined in the current unit.
22515 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
22517 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
22518 instead of a libcall for UNORDERED.
22520 2018-02-09 Tamar Christina <tamar.christina@arm.com>
22523 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
22524 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
22526 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22529 * config/s390/s390.c (s390_set_current_function): Invoke
22530 s390_indirect_branch_settings also if fndecl didn't change.
22532 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
22534 * config/rs6000/rs6000.md (blockage): Set length to zero.
22536 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
22538 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
22540 2018-02-09 Jakub Jelinek <jakub@redhat.com>
22543 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
22544 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
22548 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
22549 PARALLEL incoming that failed vt_get_decl_and_offset check.
22551 PR middle-end/84237
22552 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
22553 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
22555 (get_variable_section): For decls in named .bss* sections pass true as
22556 second argument to bss_initializer_p.
22558 2018-02-09 Marek Polacek <polacek@redhat.com>
22559 Jakub Jelinek <jakub@redhat.com>
22562 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
22563 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
22564 Sync some changes from cxx_fold_indirect_ref.
22566 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
22568 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
22570 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
22571 (BLOCK_INLINE_ENTRY_LABEL): New.
22572 (dwarf2out_var_location): Disregard inline entry markers.
22573 (inline_entry_data): New struct.
22574 (inline_entry_data_hasher): New hashtable type.
22575 (inline_entry_data_hasher::hash): New.
22576 (inline_entry_data_hasher::equal): New.
22577 (inline_entry_data_table): New variable.
22578 (add_high_low_attributes): Add DW_AT_entry_pc and
22579 DW_AT_GNU_entry_view attributes if a pending entry is found
22580 in inline_entry_data_table. Add old entry_pc attribute only
22581 if debug nonbinding markers are disabled.
22582 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
22583 markers are enabled.
22584 (block_within_block_p, dwarf2out_inline_entry): New.
22585 (dwarf2out_finish): Check that no entries remained in
22586 inline_entry_data_table.
22587 * final.c (reemit_insn_block_notes): Handle inline entry notes.
22588 (final_scan_insn, notice_source_line): Likewise.
22589 (rest_of_clean_state): Skip inline entry markers.
22590 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
22592 * gimple.c (gimple_build_debug_inline_entry): New.
22593 * gimple.h (enum gimple_debug_subcode): Add
22594 GIMPLE_DEBUG_INLINE_ENTRY.
22595 (gimple_build_debug_inline_entry): Declare.
22596 (gimple_debug_inline_entry_p): New.
22597 (gimple_debug_nonbind_marker_p): Adjust.
22598 * insn-notes.def (INLINE_ENTRY): New.
22599 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
22600 inline entry marker notes.
22601 (print_insn): Likewise.
22602 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
22603 (INSN_DEBUG_MARKER_KIND): Likewise.
22604 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
22605 * tree-inline.c (expand_call_inline): Build and insert
22606 debug_inline_entry stmt.
22607 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
22608 inline entry blocks early, if nonbind markers are enabled.
22609 (dump_scope_block): Dump fragment info.
22610 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
22611 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
22612 (gimple_build_debug_inline_entry): New.
22613 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
22614 Enable/disable inline entry points too.
22615 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
22616 (DEBUG_INSN): Describe inline entry markers.
22618 * common.opt (gvariable-location-views): New.
22619 (gvariable-location-views=incompat5): New.
22620 * config.in: Rebuilt.
22621 * configure: Rebuilt.
22622 * configure.ac: Test assembler for view support.
22623 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
22624 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
22625 * dwarf2out.c (var_loc_view): New typedef.
22626 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
22627 (dwarf2out_locviews_in_attribute): New.
22628 (dwarf2out_locviews_in_loclist): New.
22629 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
22630 (enum dw_line_info_opcode): Add LI_adv_address.
22631 (struct dw_line_info_table): Add view.
22632 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
22633 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
22634 (zero_view_p): New variable.
22635 (ZERO_VIEW_P): New macro.
22636 (output_asm_line_debug_info): New.
22637 (struct var_loc_node): Add view.
22638 (add_AT_view_list, AT_loc_list): New.
22639 (add_var_loc_to_decl): Add view param. Test it against last.
22640 (new_loc_list): Add view params. Record them.
22641 (AT_loc_list_ptr): Handle loc and view lists.
22642 (view_list_to_loc_list_val_node): New.
22643 (print_dw_val): Handle dw_val_class_view_list.
22644 (size_of_die): Likewise.
22645 (value_format): Likewise.
22646 (loc_list_has_views): New.
22647 (gen_llsym): Set vl_symbol too.
22648 (maybe_gen_llsym, skip_loc_list_entry): New.
22649 (dwarf2out_maybe_output_loclist_view_pair): New.
22650 (output_loc_list): Output view list or entries too.
22651 (output_view_list_offset): New.
22652 (output_die): Handle dw_val_class_view_list.
22653 (output_dwarf_version): New.
22654 (output_compilation_unit_header): Use it.
22655 (output_skeleton_debug_sections): Likewise.
22656 (output_rnglists, output_line_info): Likewise.
22657 (output_pubnames, output_aranges): Update version comments.
22658 (output_one_line_info_table): Output view numbers in asm comments.
22659 (dw_loc_list): Determine current endview, pass it to new_loc_list.
22660 Call maybe_gen_llsym.
22661 (loc_list_from_tree_1): Adjust.
22662 (add_AT_location_description): Create view list attribute if
22663 needed, check it's absent otherwise.
22664 (convert_cfa_to_fb_loc_list): Adjust.
22665 (maybe_emit_file): Call output_asm_line_debug_info for test.
22666 (dwarf2out_var_location): Reset views as needed. Precompute
22667 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
22668 attribute. Set view.
22669 (new_line_info_table): Reset next view.
22670 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
22671 (dwarf2out_source_line): Likewise. Output view resets and labels to
22672 the assembler, or select appropriate line info opcodes.
22673 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
22674 (optimize_string_length): Catch it. Adjust.
22675 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
22676 dw_val_class_view_list, and remove it if no longer needed.
22677 (hash_loc_list): Hash view numbers.
22678 (loc_list_hasher::equal): Compare them.
22679 (optimize_location_lists): Check whether a view list symbol is
22680 needed, and whether the locview attribute is present, and
22681 whether they match. Remove the locview attribute if no longer
22683 (index_location_lists): Call skip_loc_list_entry for test.
22684 (dwarf2out_finish): Call output_asm_line_debug_info for test.
22685 Use output_dwarf_version.
22686 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
22687 (struct dw_val_node): Add val_view_list.
22688 * final.c (SEEN_NEXT_VIEW): New.
22689 (set_next_view_needed): New.
22690 (clear_next_view_needed): New.
22691 (maybe_output_next_view): New.
22692 (final_start_function): Rename to...
22693 (final_start_function_1): ... this. Take pointer to FIRST,
22694 add SEEN parameter. Emit param bindings in the initial view.
22695 (final_start_function): Reintroduce SEEN-less interface.
22696 (final): Rename to...
22697 (final_1): ... this. Take SEEN parameter. Output final pending
22698 next view at the end.
22699 (final): Reintroduce seen-less interface.
22700 (final_scan_insn): Output pending next view before switching
22701 sections or ending a block. Mark the next view as needed when
22702 outputting variable locations. Notify debug backend of section
22703 changes, and of location view changes.
22704 (rest_of_handle_final): Adjust.
22705 * toplev.c (process_options): Autodetect value for debug variable
22706 location views option. Warn on incompat5 without -gdwarf-5.
22707 * doc/invoke.texi (gvariable-location-views): New.
22708 (gvariable-location-views=incompat5): New.
22709 (gno-variable-location-views): New.
22711 2018-02-08 David Malcolm <dmalcolm@redhat.com>
22713 PR tree-optimization/84136
22714 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
22715 that the result of find_edge is non-NULL.
22717 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
22720 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
22721 storing integer register in SImode. Fix cost of 256 and 512
22722 byte aligned SSE register store.
22724 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
22726 * config/i386/i386.c (ix86_multiplication_cost): Fix
22727 multiplication cost for TARGET_AVX512DQ.
22729 2018-02-08 Marek Polacek <polacek@redhat.com>
22731 PR tree-optimization/84238
22732 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
22735 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
22737 PR tree-optimization/84265
22738 * tree-vect-stmts.c (vectorizable_store): Don't treat
22739 VMAT_CONTIGUOUS accesses as grouped.
22740 (vectorizable_load): Likewise.
22742 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
22744 PR tree-optimization/81635
22745 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
22746 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
22747 (test_round_for_mask): New functions.
22748 (wide_int_cc_tests): Call test_round_for_mask.
22749 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
22750 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
22751 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
22752 range returned by get_range_info.
22754 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
22757 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
22758 * symtab.c: Include builtins.h
22759 (symtab_node::output_to_lto_symbol_table_p): Move here
22760 from lto-streamer-out.c:output_symbol_p.
22761 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
22762 (output_symbol_p): Move all logic to symtab.c
22763 (produce_symtab): Update.
22765 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22767 * config/s390/s390-opts.h (enum indirect_branch): Define.
22768 * config/s390/s390-protos.h (s390_return_addr_from_memory)
22769 (s390_indirect_branch_via_thunk)
22770 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
22771 (enum s390_indirect_branch_type): Define.
22772 * config/s390/s390.c (struct s390_frame_layout, struct
22773 machine_function): Remove.
22774 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
22775 (indirect_branch_table_label_no, indirect_branch_table_name):
22777 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
22778 (enum s390_indirect_branch_option): Define.
22779 (s390_return_addr_from_memory): New function.
22780 (s390_handle_string_attribute): New function.
22781 (s390_attribute_table): Add new attribute handler.
22782 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
22783 (s390_indirect_branch_via_thunk): New function.
22784 (s390_indirect_branch_via_inline_thunk): New function.
22785 (s390_function_ok_for_sibcall): When jumping via thunk disallow
22786 sibling call optimization for non z10 compiles.
22787 (s390_emit_call): Force indirect branch target to be a single
22788 register. Add r1 clobber for non-z10 compiles.
22789 (s390_emit_epilogue): Emit return jump via return_use expander.
22790 (s390_reorg): Handle JUMP_INSNs as execute targets.
22791 (s390_option_override_internal): Perform validity checks for the
22792 new command line options.
22793 (s390_indirect_branch_attrvalue): New function.
22794 (s390_indirect_branch_settings): New function.
22795 (s390_set_current_function): Invoke s390_indirect_branch_settings.
22796 (s390_output_indirect_thunk_function): New function.
22797 (s390_code_end): Implement target hook.
22798 (s390_case_values_threshold): Implement target hook.
22799 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
22801 * config/s390/s390.h (struct s390_frame_layout)
22802 (struct machine_function): Move here from s390.c.
22803 (TARGET_INDIRECT_BRANCH_NOBP_RET)
22804 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
22805 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
22806 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
22807 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
22808 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
22809 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
22810 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
22811 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
22812 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
22813 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
22814 (mnemonic attribute): Add values which aren't recognized
22816 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
22817 pattern for branch conversion. Fix mnemonic attribute.
22818 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
22819 indirect branch via thunk if requested.
22820 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
22821 ("*indirect_jump"): Disable for branch conversion using out of
22823 ("indirect_jump_via_thunk<mode>_z10")
22824 ("indirect_jump_via_thunk<mode>")
22825 ("indirect_jump_via_inlinethunk<mode>_z10")
22826 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
22827 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
22828 ("casesi_jump_via_inlinethunk<mode>_z10")
22829 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
22830 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
22831 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
22832 ("*indirect2_jump"): Disable for branch conversion.
22833 ("casesi_jump"): Turn into expander and expand patterns for branch
22835 ("return_use"): New expander.
22836 ("*return"): Emit return via thunk and rename it to ...
22837 ("*return<mode>"): ... this one.
22838 * config/s390/s390.opt: Add new options and and enum for the
22841 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
22843 * lra-constraints.c (match_reload): Unconditionally use
22844 gen_lowpart_SUBREG, rather than selecting between that
22845 and equivalent gen_rtx_SUBREG code.
22847 2018-02-08 Richard Biener <rguenther@suse.de>
22849 PR tree-optimization/84233
22850 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
22851 changed flag instead of boguously re-using phi_inserted.
22853 2018-02-08 Martin Jambor <mjambor@suse.cz>
22855 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
22856 static local variables.
22858 2018-02-08 Richard Biener <rguenther@suse.de>
22860 PR tree-optimization/84278
22861 * tree-vect-stmts.c (vectorizable_store): When looking for
22862 smaller vector types to perform grouped strided loads/stores
22863 make sure the mode is supported by the target.
22864 (vectorizable_load): Likewise.
22866 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
22868 * config/aarch64/aarch64.c (aarch64_components_for_bb):
22869 Increase LDP/STP opportunities by adding adjacent callee-saves.
22871 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
22873 PR rtl-optimization/84068
22874 PR rtl-optimization/83459
22875 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
22877 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
22879 PR tree-optimization/84224
22880 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
22881 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
22882 non-zero arguments.
22884 2018-02-07 Iain Sandoe <iain@codesourcery.com>
22887 * config/rs6000/altivec.md (*restore_world): Remove LR use.
22888 * config/rs6000/predicates.md (restore_world_operation): Adjust op
22889 count, remove one USE.
22891 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
22893 * doc/install.texi (Configuration): Document the
22894 --with-long-double-format={ibm,ieee} PowerPC configuration
22898 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
22899 Convert from define_expand to be define_insn_and_split. Rework
22900 float/double/_Float128 conversions to QI/HI/SImode to work with
22901 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
22902 conversions to QI/HImode types did a store and then a load to
22903 truncate the value. For conversions to VSX registers, don't split
22904 the insn, instead emit the code directly. Use the code iterator
22905 any_fix to combine signed and unsigned conversions.
22906 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
22907 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
22908 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
22909 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
22910 (fix_<mode>di2_hw): Likewise.
22911 (fixuns_<mode>di2_hw): Likewise.
22912 (fix_<mode>si2_hw): Likewise.
22913 (fixuns_<mode>si2_hw): Likewise.
22914 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
22915 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
22916 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
22917 fix<uns>_trunc<SFDF:mode>si2_p8.
22918 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
22919 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22920 (fix<uns>_<mode>_mem): Likewise.
22921 (fctiw<u>z_<mode>_mem): Likewise.
22922 (fix<uns>_<mode>_mem): Likewise.
22923 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
22924 the register allocator from doing a direct move to the GPRs to do
22925 a store, and instead use the ISA 3.0 store byte/half-word from
22926 vector register instruction. For IEEE 128-bit floating point,
22927 also optimize stores of 32-bit ints.
22928 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
22930 2018-02-07 Alan Hayward <alan.hayward@arm.com>
22932 * genextract.c (push_pathstr_operand): New function to support
22934 (walk_rtx): Call push_pathstr_operand.
22935 (print_path): Support [a-zA-Z].
22937 2018-02-07 Richard Biener <rguenther@suse.de>
22939 PR tree-optimization/84037
22940 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
22941 (cse_and_gimplify_to_preheader): Declare.
22942 (vect_get_place_in_interleaving_chain): Likewise.
22943 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
22945 (_loop_vec_info::~_loop_vec_info): Delete it.
22946 (cse_and_gimplify_to_preheader): New function.
22947 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
22948 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
22949 (vectorizable_load): Likewise. For grouped stores always base
22950 the IV on the first element.
22951 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
22952 condition before gimplifying.
22954 2018-02-07 Jakub Jelinek <jakub@redhat.com>
22956 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
22957 *DIV_EXPR and *MOD_EXPR.
22959 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
22962 * config/i386/i386.c (ix86_option_override_internal): Mask out
22963 the CF_SET bit when checking -fcf-protection.
22965 2018-02-07 Tom de Vries <tom@codesourcery.com>
22968 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
22971 2018-02-07 Richard Biener <rguenther@suse.de>
22973 PR tree-optimization/84204
22974 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
22977 PR tree-optimization/84205
22978 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
22979 special-case isl_ast_op_zdiv_r.
22981 PR tree-optimization/84223
22982 * graphite-scop-detection.c (gather_bbs::before_dom_children):
22983 Only add conditions from within the region.
22984 (gather_bbs::after_dom_children): Adjust.
22986 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
22989 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
22990 * config/avr/avr.md: Only post-reload split REG-REG moves if
22991 either register is GENERAL_REG_P.
22993 2018-02-07 Jakub Jelinek <jakub@redhat.com>
22995 PR tree-optimization/84235
22996 * tree-ssa-scopedtables.c
22997 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
22998 if the subtraction is performed in floating point type where NaNs are
22999 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
23000 build 1. Formatting fix.
23002 2018-02-06 Jakub Jelinek <jakub@redhat.com>
23005 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
23006 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
23007 and skip it regardless of bb boundaries. Use CALL_P macro,
23008 don't test INSN_P (insn) together with CALL_P or JUMP_P check
23009 unnecessarily, formatting fix.
23011 2018-02-06 Michael Collison <michael.collison@arm.com>
23013 * config/arm/thumb2.md:
23014 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
23015 (*thumb_mov_notscc): Ditto.
23017 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
23020 * config/rs6000/rs6000.md (su code attribute): Use "u" for
23021 unsigned_fix, not "s".
23023 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23025 * configure.ac (gcc_fn_eh_frame_ro): New function.
23026 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
23027 correct .eh_frame permissions.
23028 * configure: Regenerate.
23030 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
23032 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
23033 irrelevant options.
23035 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23037 * config/rs6000/rs6000.c (rs6000_option_override_internal):
23038 Display warning message for -mno-speculate-indirect-jumps.
23040 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
23042 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
23044 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
23046 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
23048 PR tree-optimization/84225
23049 * tree-eh.c (find_trapping_overflow): Only call
23050 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
23052 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
23055 * config/i386/i386.c: Reimplement the check of possible options
23056 -mibt/-mshstk conbination. Change error messages.
23057 * doc/invoke.texi: Fix a typo: remove extra '='.
23059 2018-02-06 Marek Polacek <polacek@redhat.com>
23061 PR tree-optimization/84228
23062 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
23064 2018-02-06 Tamar Christina <tamar.christina@arm.com>
23067 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
23068 emitted arch directives.
23069 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
23070 __ARM_FEATURE_COPROC before changing architectures.
23072 2018-02-06 Richard Biener <rguenther@suse.de>
23074 * config/i386/i386.c (print_reg): Fix typo.
23075 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
23077 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
23079 * configure: Regenerate.
23081 2018-02-05 Martin Sebor <msebor@redhat.com>
23083 PR tree-optimization/83369
23084 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
23087 2018-02-05 Martin Liska <mliska@suse.cz>
23089 * doc/invoke.texi: Cherry-pick upstream r323995.
23091 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
23093 * ira.c (ira_init_register_move_cost): Adjust comment.
23095 2018-02-05 Martin Liska <mliska@suse.cz>
23097 PR gcov-profile/84137
23098 * doc/gcov.texi: Fix typo in documentation.
23100 2018-02-05 Martin Liska <mliska@suse.cz>
23102 PR gcov-profile/83879
23103 * doc/gcov.texi: Document necessity of --dynamic-list-data when
23104 using dlopen functionality.
23106 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
23108 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
23109 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
23110 _mm_maskz_range_ss, _mm_mask_range_round_ss,
23111 _mm_maskz_range_round_ss): New intrinsics.
23112 (__builtin_ia32_rangesd128_round)
23113 (__builtin_ia32_rangess128_round): Remove.
23114 (__builtin_ia32_rangesd128_mask_round,
23115 __builtin_ia32_rangess128_mask_round): New builtins.
23116 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
23117 __builtin_ia32_rangess128_round): Remove.
23118 (__builtin_ia32_rangesd128_mask_round,
23119 __builtin_ia32_rangess128_mask_round): New builtins.
23120 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
23121 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
23122 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
23123 "<round_saeonly_constraint>")): Changed to ...
23124 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
23125 "<round_saeonly_scalar_constraint>")): ... this.
23126 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
23127 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
23128 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
23129 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
23130 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
23132 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
23134 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
23136 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
23137 Remove all values except native, 8540 and 8548.
23139 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
23141 * config/i386/i386.c (ix86_output_function_return): Pass
23142 INVALID_REGNUM, instead of -1, as invalid register number to
23143 indirect_thunk_name and output_indirect_thunk.
23145 2018-02-02 Julia Koval <julia.koval@intel.com>
23147 * config.gcc: Add -march=icelake.
23148 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
23149 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
23150 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
23151 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
23152 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
23153 (processor_target_table): Add icelake.
23154 (ix86_option_override_internal): Handle new PTAs.
23155 (get_builtin_code_for_version): Handle icelake.
23156 (M_INTEL_COREI7_ICELAKE): New.
23157 (fold_builtin_cpu): Handle icelake.
23158 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
23159 * doc/invoke.texi: Add -march=icelake.
23161 2018-02-02 Julia Koval <julia.koval@intel.com>
23163 * config/i386/i386.c (ix86_option_override_internal): Change flags type
23164 to wide_int_bitmask.
23165 * wide-int-bitmask.h: New.
23167 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
23170 * config/i386/i386.md: Replace Pmode with word_mode in
23171 builtin_setjmp_setup and builtin_longjmp to support x32.
23173 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
23177 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
23179 (rs6000_supported_cpu_names): New static variable.
23180 (linux_cpu_translation_table): Likewise.
23181 (elf_platform) <cpu>: Define new static variable and use it.
23182 Translate kernel AT_PLATFORM name to canonical name if needed.
23183 Error if platform name is unknown.
23185 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
23188 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
23190 2018-02-01 Jeff Law <law@redhat.com>
23193 * config/i386/i386.c (release_scratch_register_on_entry): Add new
23194 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
23195 the scratch if RELEASE_VIA_POP is false.
23196 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
23197 If we have to save a temporary register, decrement SIZE appropriately.
23198 Pass new arguments to release_scratch_register_on_entry.
23199 (ix86_adjust_stack_and_probe): Likewise.
23200 (ix86_emit_probe_stack_range): Pass new arguments to
23201 release_scratch_register_on_entry.
23203 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
23205 PR rtl-optimization/84157
23206 * combine.c (change_zero_ext): Use REG_P predicate in
23207 front of HARD_REGISTER_P predicate.
23209 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
23211 * config/avr/avr.c (avr_option_override): Move disabling of
23212 -fdelete-null-pointer-checks to...
23213 * common/config/avr/avr-common.c (avr_option_optimization_table):
23216 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
23218 PR tree-optimization/81635
23219 * tree-data-ref.c (split_constant_offset_1): For types that
23220 wrap on overflow, try to use range info to prove that wrapping
23223 2018-02-01 Renlin Li <renlin.li@arm.com>
23226 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
23227 TAILCALL_ADDR_REGS.
23228 (aarch64_register_move_cost): Likewise.
23229 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
23230 TAILCALL_ADDR_REGS.
23231 (REG_CLASS_NAMES): Likewise.
23232 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
23233 TAILCALL_ADDR_REGS. Remove IP registers.
23234 * config/aarch64/aarch64.md (Ucs): Update register constraint.
23236 2018-02-01 Richard Biener <rguenther@suse.de>
23238 * domwalk.h (dom_walker::dom_walker): Add additional constructor
23239 for specifying RPO order and allow NULL for that.
23240 * domwalk.c (dom_walker::dom_walker): Likewise.
23241 (dom_walker::walk): Handle NULL RPO order.
23242 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
23244 (rewrite_update_dom_walker): Likewise.
23245 (mark_def_dom_walker): Likewise.
23247 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
23249 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
23250 (aarch64_maybe_expand_sve_subreg_move): Declare.
23251 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
23252 * config/aarch64/predicates.md (aarch64_any_register_operand): New
23254 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
23255 that are semantically a reverse operation.
23256 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
23257 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
23258 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
23260 (aarch64_can_change_mode_class): For big-endian, forbid changes
23261 between two SVE modes if they have different element sizes.
23263 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
23265 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
23266 the TImode handling for big-endian targets.
23268 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
23270 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
23271 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
23273 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
23274 Remove BSWAP handing for big-endian targets and use the form of
23275 LD1RQ appropariate for the mode.
23277 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
23279 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
23280 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
23281 duplicated element.
23283 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
23286 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
23287 check for operands that need to go through aarch64_sve_reload_be.
23289 2018-02-01 Jakub Jelinek <jakub@redhat.com>
23291 PR tree-optimization/81661
23292 PR tree-optimization/84117
23293 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
23294 * tree-eh.c: Include gimplify.h.
23295 (find_trapping_overflow, replace_trapping_overflow,
23296 rewrite_to_non_trapping_overflow): New functions.
23297 * tree-vect-loop.c: Include tree-eh.h.
23298 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
23299 * tree-data-ref.c: Include tree-eh.h.
23300 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
23302 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
23304 PR rtl-optimization/84123
23305 * combine.c (change_zero_ext): Check if hard register satisfies
23306 can_change_dest_mode before calling gen_lowpart_SUBREG.
23308 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
23311 * ira.c (ira_init_register_move_cost): Remove assert.
23313 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
23315 PR rtl-optimization/84071
23316 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
23317 * doc/tm.texi: Regenerate.
23319 2018-01-31 Richard Biener <rguenther@suse.de>
23321 PR tree-optimization/84132
23322 * tree-data-ref.c (analyze_miv_subscript): Properly
23323 check whether evolution_function_is_affine_multivariate_p
23324 before calling gcd_of_steps_may_divide_p.
23326 2018-01-31 Julia Koval <julia.koval@intel.com>
23329 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
23330 * config/i386/i386.md (rdpid_rex64) New.
23331 (rdpid): Make 32bit only.
23333 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
23336 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
23337 an IDENTIFIER_NODE for FUNCTION_TYPE's.
23339 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
23342 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
23344 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
23346 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
23348 PR rtl-optimization/84071
23349 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
23350 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
23352 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
23354 * config/arc/arc.c (arc_handle_aux_attribute): New function.
23355 (arc_attribute_table): Add 'aux' attribute.
23356 (arc_in_small_data_p): Consider aux like variables.
23357 (arc_is_aux_reg_p): New function.
23358 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
23359 (arc_get_aux_arg): New function.
23360 (prepare_move_operands): Handle aux-register access.
23361 (arc_handle_aux_attribute): New function.
23362 * doc/extend.texi (ARC Variable attributes): Add subsection.
23364 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
23366 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
23367 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
23368 (arc_attribute_table): Add 'uncached' attribute.
23369 (arc_print_operand): Print '.di' flag for uncached memory
23371 (arc_in_small_data_p): Do not consider for small data the uncached
23373 (arc_is_uncached_mem_p): New function.
23374 * config/arc/predicates.md (compact_store_memory_operand): Check
23375 for uncached memory accesses.
23376 (nonvol_nonimm_operand): Likewise.
23377 * doc/extend.texi (ARC Type Attribute): New subsection.
23379 2018-01-31 Jakub Jelinek <jakub@redhat.com>
23382 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
23383 falign-loops=): Add Optimization flag.
23385 2018-01-30 Jeff Law <law@redhat.com>
23388 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
23389 INT_REGISTERS_SAVED. Check it prior to calling
23390 get_scratch_register_on_entry.
23391 (ix86_adjust_stack_and_probe): Similarly.
23392 (ix86_emit_probe_stack_range): Similarly.
23393 (ix86_expand_prologue): Corresponding changes.
23395 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23398 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
23399 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
23401 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
23404 * lra-constraints.c (curr_insn_transform): Process AND in the
23407 2018-01-30 Jakub Jelinek <jakub@redhat.com>
23409 PR rtl-optimization/83986
23410 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
23411 dependence against last_pending_memory_flush in addition to
23412 pending_jump_insns.
23414 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
23416 PR tree-optimization/81611
23417 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
23420 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23423 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
23426 2018-01-30 Richard Biener <rguenther@suse.de>
23427 Jakub Jelinek <jakub@redhat.com>
23429 PR tree-optimization/84111
23430 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
23431 inner loops added during recursion, as they don't have up-to-date
23434 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
23437 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
23438 (can_inline_edge_by_limits_p): ... here.
23439 (can_early_inline_edge_p, check_callers,
23440 update_caller_keys, update_callee_keys, recursive_inlining,
23441 add_new_edges_to_heap, speculation_useful_p,
23442 inline_small_functions,
23443 inline_small_functions, flatten_function,
23444 inline_to_all_callers_1): Update.
23446 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
23448 * profile-count.c (profile_count::combine_with_ipa_count): Handle
23451 2018-01-30 Richard Biener <rguenther@suse.de>
23453 PR tree-optimization/83008
23454 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
23455 invariant and constant vector uses in stmts when they need
23456 more than one stmt.
23458 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23461 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
23462 * configure: Regenerate.
23464 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
23466 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
23468 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
23469 Use gen_rtx_REG rather than gen_lowpart.
23471 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
23473 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
23474 rather than 0 when creating partial subregs.
23476 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
23478 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
23481 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
23484 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
23485 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
23486 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
23487 flags. This restores the settings used before the 2017-07-24.
23488 Turning off pre increment/decrement/modify allows IVOPTS to
23489 optimize DF/SF loops where the index is an int.
23491 2018-01-29 Richard Biener <rguenther@suse.de>
23492 Kelvin Nilsen <kelvin@gcc.gnu.org>
23495 * tree-vect-stmts.c (vectorizable_call): Don't call
23496 targetm.vectorize_builtin_md_vectorized_function if callee is
23499 2018-01-22 Carl Love <cel@us.ibm.com>
23501 * doc/extend.tex: Fix typo in second arg in
23502 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
23504 2018-01-29 Richard Biener <rguenther@suse.de>
23506 PR tree-optimization/84086
23507 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
23508 (flush_ssaname_freelist): When SSA names were released reset
23509 the SCEV hash table.
23511 2018-01-29 Richard Biener <rguenther@suse.de>
23513 PR tree-optimization/84057
23514 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
23515 removed paths when removing edges.
23517 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
23519 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
23520 -mfunction-return=@var{choice}.
23522 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
23524 PR diagnostic/84034
23525 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
23526 Handle CR like TAB.
23527 (layout::print_source_line): Likewise.
23528 (test_get_line_width_without_trailing_whitespace): Add test cases.
23530 2018-01-27 Jakub Jelinek <jakub@redhat.com>
23532 PR middle-end/84040
23533 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
23536 2018-01-26 Jim Wilson <jimw@sifive.com>
23538 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
23540 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
23543 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23545 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
23546 and CMP + SUB-immediate -> SUBS.
23548 2018-01-26 Martin Sebor <msebor@redhat.com>
23550 PR tree-optimization/83896
23551 * tree-ssa-strlen.c (get_string_len): Rename...
23552 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
23553 Avoid assuming length is constant.
23554 (handle_char_store): Use HOST_WIDE_INT for string length.
23556 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
23559 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
23560 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
23562 2018-01-26 Richard Biener <rguenther@suse.de>
23564 PR rtl-optimization/84003
23565 * dse.c (record_store): Only record redundant stores when
23566 the earlier store aliases at least all accesses the later one does.
23568 2018-01-26 Jakub Jelinek <jakub@redhat.com>
23570 PR rtl-optimization/83985
23571 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
23572 REG_CFA_RESTORE insns.
23573 (delete_unmarked_insns): Don't ignore separate shrink wrapping
23574 REG_CFA_RESTORE insns here.
23577 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
23578 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
23580 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
23582 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
23583 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
23584 (arc_init): Likewise.
23585 (arc_override_options): Likewise.
23586 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
23588 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
23590 * config/arc/arc.h (TARGET_DBNZ): Define.
23591 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
23592 properly set the tune attribute.
23593 (dbnz): Use TARGET_DBNZ guard.
23594 * config/arc/arc.opt (mtune): Add core3 option.
23596 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
23598 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
23599 recognize new pic like addresses.
23600 (arc_delegitimize_address): Clean up.
23602 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
23604 * config/arc/arc-arches.def: Option mrf16 valid for all
23606 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
23607 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
23608 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
23609 * config/arc/arc-tables.opt: Regenerate.
23610 * config/arc/arc.c (arc_conditional_register_usage): Handle
23611 reduced register file case.
23612 (arc_file_start): Set must have build attributes.
23613 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
23614 mrf16 option value.
23615 * config/arc/arc.opt (mrf16): Add new option.
23616 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
23617 * config/arc/genmultilib.awk: Handle new mrf16 option.
23618 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
23619 * config/arc/t-multilib: Regenerate.
23620 * doc/invoke.texi (ARC Options): Document mrf16 option.
23622 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
23624 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
23625 * config/arc/arc.c (arc_handle_secure_attribute): New function.
23626 (arc_attribute_table): Add 'secure_call' attribute.
23627 (arc_print_operand): Print secure call operand.
23628 (arc_function_ok_for_sibcall): Don't optimize tail calls when
23630 (arc_is_secure_call_p): New function. * config/arc/arc.md
23631 (call_i): Add support for sjli instruction.
23632 (call_value_i): Likewise.
23633 * config/arc/constraints.md (Csc): New constraint.
23635 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
23636 John Eric Martin <John.Martin@emmicro-us.com>
23638 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
23639 * config/arc/arc.c (_arc_jli_section): New struct.
23640 (arc_jli_section): New type.
23641 (rc_jli_sections): New static variable.
23642 (arc_handle_jli_attribute): New function.
23643 (arc_attribute_table): Add jli_always and jli_fixed attribute.
23644 (arc_file_end): New function.
23645 (TARGET_ASM_FILE_END): Define.
23646 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
23647 (arc_add_jli_section): New function.
23648 (jli_call_scan): Likewise.
23649 (arc_reorg): Call jli_call_scan.
23650 (arc_output_addsi): Remove 'S' from printing asm operand.
23651 (arc_is_jli_call_p): New function.
23652 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
23654 (movhi_insn): Likewise.
23655 (movsi_insn): Likewise.
23656 (movsi_set_cc_insn): Likewise.
23657 (loadqi_update): Likewise.
23658 (load_zeroextendqisi_update): Likewise.
23659 (load_signextendqisi_update): Likewise.
23660 (loadhi_update): Likewise.
23661 (load_zeroextendhisi_update): Likewise.
23662 (load_signextendhisi_update): Likewise.
23663 (loadsi_update): Likewise.
23664 (loadsf_update): Likewise.
23665 (movsicc_insn): Likewise.
23666 (bset_insn): Likewise.
23667 (bxor_insn): Likewise.
23668 (bclr_insn): Likewise.
23669 (bmsk_insn): Likewise.
23670 (bicsi3_insn): Likewise.
23671 (cmpsi_cc_c_insn): Likewise.
23672 (movsi_ne): Likewise.
23673 (movsi_cond_exec): Likewise.
23674 (clrsbsi2): Likewise.
23675 (norm_f): Likewise.
23685 (call_i): Remove 'S' asm letter, add jli instruction.
23686 (call_value_i): Likewise.
23687 * config/arc/arc.op (mjli-always): New option.
23688 * config/arc/constraints.md (Cji): New constraint.
23689 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
23691 (subsf3_fpx): Likewise.
23692 (mulsf3_fpx): Likewise.
23693 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
23695 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
23696 function attrbutes.
23697 * doc/invoke.texi (ARC): Document mjli-always option.
23699 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
23701 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
23702 avoid addition with 0 and use incw and decw where possible.
23704 2018-01-26 Richard Biener <rguenther@suse.de>
23706 PR tree-optimization/81082
23707 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
23708 association if it requires casting to unsigned.
23709 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
23710 from fold_plusminus_mult_expr to catch important cases late when
23711 range info is available.
23713 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23715 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
23716 * configure.ac (hidden_linkonce): New test.
23717 * configure: Regenerate.
23718 * config.in: Regenerate.
23720 2018-01-26 Julia Koval <julia.koval@intel.com>
23722 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
23723 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
23724 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
23725 _mm_mask_bitshuffle_epi64_mask): Fix type.
23726 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
23727 USI_FTYPE_V4DI_V4DI_USI): Remove.
23728 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
23729 __builtin_ia32_vpshufbitqmb256_mask,
23730 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
23731 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
23732 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
23734 2018-01-26 Alan Modra <amodra@gmail.com>
23737 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
23738 UNSPEC_VBPERMQ. Sort other unspecs.
23740 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
23742 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
23744 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
23746 PR middle-end/83055
23747 * predict.c (drop_profile): Do not push/pop cfun; update also
23749 (handle_missing_profiles): Fix logic looking for zero profiles.
23751 2018-01-25 Jakub Jelinek <jakub@redhat.com>
23753 PR middle-end/83977
23754 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
23755 on functions with #pragma omp declare simd or functions with simd
23757 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
23758 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
23759 Remove trailing \n from warning_at calls.
23761 2018-01-25 Tom de Vries <tom@codesourcery.com>
23764 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
23765 for neutered workers.
23767 2018-01-24 Joseph Myers <joseph@codesourcery.com>
23770 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
23771 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
23773 2018-01-24 Jeff Law <law@redhat.com>
23776 * i386.c (get_probe_interval): Move to earlier point.
23777 (ix86_compute_frame_layout): If -fstack-clash-protection and
23778 the frame is larger than the probe interval, then use pushes
23779 to save registers rather than reg->mem moves.
23780 (ix86_expand_prologue): Remove conditional for int_registers_saved
23783 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
23786 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
23787 min/max for never referenced object.
23789 2018-01-24 Jakub Jelinek <jakub@redhat.com>
23791 PR middle-end/83977
23792 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
23794 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
23795 attributes from DECL_ATTRIBUTES (decl) without affecting
23796 DECL_ATTRIBUTES (current_function_decl).
23797 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
23798 functions with non-NULL DECL_ABSTRACT_ORIGIN.
23800 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
23802 PR tree-optimization/83979
23803 * fold-const.c (fold_comparison): Use constant_boolean_node
23804 instead of boolean_{true,false}_node.
23806 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
23808 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
23811 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23813 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
23814 Simplify the clause that sets the length attribute.
23815 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
23816 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
23817 clause that sets the length attribute.
23818 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
23820 2018-01-24 Tom de Vries <tom@codesourcery.com>
23823 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
23824 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
23825 Add strict parameter.
23826 (prevent_branch_around_nothing): Insert dummy insn between branch to
23827 label and label with no ptx insn inbetween.
23828 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
23830 2018-01-24 Tom de Vries <tom@codesourcery.com>
23833 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
23834 for neutered threads in warp.
23835 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
23837 2018-01-24 Richard Biener <rguenther@suse.de>
23839 PR tree-optimization/83176
23840 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
23843 2018-01-24 Richard Biener <rguenther@suse.de>
23845 PR tree-optimization/82819
23846 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
23847 code generating pluses that are no-ops in the target precision.
23849 2018-01-24 Richard Biener <rguenther@suse.de>
23851 PR middle-end/84000
23852 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
23854 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
23856 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
23857 to merge probabilities.
23858 * predict.c (probably_never_executed): Also mark as cold functions
23859 with global 0 profile and guessed local profile.
23860 * profile-count.c (profile_probability::combine_with_count): New
23862 * profile-count.h (profile_probability::operator*,
23863 profile_probability::operator*=, profile_probability::operator/,
23864 profile_probability::operator/=): Reduce precision to adjusted
23865 and set value to guessed on contradictory divisions.
23866 (profile_probability::combine_with_freq): Remove.
23867 (profile_probability::combine_wiht_count): Declare.
23868 (profile_count::force_nonzero):: Set to adjusted.
23869 (profile_count::probability_in):: Set quality to adjusted.
23870 * tree-ssa-tail-merge.c (replace_block_by): Use
23871 combine_with_count.
23873 2018-01-23 Andrew Waterman <andrew@sifive.com>
23874 Jim Wilson <jimw@sifive.com>
23876 * config/riscv/riscv.c (riscv_stack_boundary): New.
23877 (riscv_option_override): Set riscv_stack_boundary. Handle
23878 riscv_preferred_stack_boundary_arg.
23879 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
23880 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
23881 (STACK_BOUNDARY): Set to riscv_stack_boundary.
23882 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
23883 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
23884 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
23886 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
23889 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
23890 of struct ix86_frame.
23891 (ix86_expand_epilogue): Likewise. Add a local variable for
23892 the reg_save_offset field in struct ix86_frame.
23894 2018-01-23 Bin Cheng <bin.cheng@arm.com>
23896 PR tree-optimization/82604
23897 * tree-loop-distribution.c (enum partition_kind): New enum item
23898 PKIND_PARTIAL_MEMSET.
23899 (partition_builtin_p): Support above new enum item.
23900 (generate_code_for_partition): Ditto.
23901 (compute_access_range): Differentiate cases that equality can be
23902 proven at all loops, the innermost loops or no loops.
23903 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
23904 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
23905 (finalize_partitions, distribute_loop): Don't fuse partition of
23906 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
23907 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
23908 parloop is enabled.
23910 2018-01-23 Martin Liska <mliska@suse.cz>
23912 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
23913 order to ignore the predictor.
23914 (PRED_POLYMORPHIC_CALL): Likewise.
23915 (PRED_RECURSIVE_CALL): Likewise.
23917 2018-01-23 Martin Liska <mliska@suse.cz>
23919 * tree-profile.c (tree_profiling): Print function header to
23920 aware reader which function we are working on.
23921 * value-prof.c (gimple_find_values_to_profile): Do not print
23922 not interesting value histograms.
23924 2018-01-23 Martin Liska <mliska@suse.cz>
23926 * profile-count.h (enum profile_quality): Add
23927 profile_uninitialized as the first value. Do not number values
23928 as they are zero based.
23929 (profile_count::verify): Update sanity check.
23930 (profile_probability::verify): Likewise.
23932 2018-01-23 Nathan Sidwell <nathan@acm.org>
23934 * doc/invoke.texi (ffor-scope): Deprecate.
23936 2018-01-23 David Malcolm <dmalcolm@redhat.com>
23938 PR tree-optimization/83510
23939 * domwalk.c (set_all_edges_as_executable): New function.
23940 (dom_walker::dom_walker): Convert bool param
23941 "skip_unreachable_blocks" to enum reachability. Move setup of
23942 edge flags to set_all_edges_as_executable and only do it when
23943 reachability is REACHABLE_BLOCKS.
23944 * domwalk.h (enum dom_walker::reachability): New enum.
23945 (dom_walker::dom_walker): Convert bool param
23946 "skip_unreachable_blocks" to enum reachability.
23947 (set_all_edges_as_executable): New decl.
23948 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
23949 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
23951 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
23952 but converting true to REACHABLE_BLOCKS.
23953 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
23955 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
23956 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
23957 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
23959 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
23960 if check_all_array_refs will be called.
23962 2018-01-23 David Malcolm <dmalcolm@redhat.com>
23964 * tree.c (selftest::test_location_wrappers): Add more test
23967 2018-01-23 David Malcolm <dmalcolm@redhat.com>
23969 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
23970 (selftest::test_bit_in_range): Likewise.
23972 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
23975 * doc/sourcebuild.texi (vect_float): Say that the selector
23976 only describes the situation when -funsafe-math-optimizations is on.
23977 (vect_float_strict): Document.
23979 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
23981 PR tree-optimization/83965
23982 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
23983 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
23984 instead of checking only for a reduction.
23985 (vect_recog_widen_sum_pattern): Likewise.
23987 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
23989 * predict.c (probably_never_executed): Only use precise profile info.
23990 (compute_function_frequency): Skip after inlining hack since we now
23991 have quality checking.
23993 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
23995 * profile-count.h (profile_probability::very_unlikely,
23996 profile_probability::unlikely, profile_probability::even): Set
23997 precision to guessed.
23999 2018-01-23 Richard Biener <rguenther@suse.de>
24001 PR tree-optimization/83963
24002 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
24003 Properly terminate dominator walk when crossing the exit edge not
24004 when visiting its source block.
24006 2018-01-23 Jakub Jelinek <jakub@redhat.com>
24009 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
24010 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
24012 2018-01-22 Jakub Jelinek <jakub@redhat.com>
24014 PR tree-optimization/83957
24015 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
24016 semicolon after for body surrounded by braces.
24018 PR tree-optimization/83081
24019 * profile-count.h (profile_probability::split): New method.
24020 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
24021 Use profile_probability::split.
24022 (do_compare_rtx_and_jump): Fix adjustment of probabilities
24023 when splitting a single conditional jump into 2.
24025 2018-01-22 David Malcolm <dmalcolm@redhat.com>
24027 PR tree-optimization/69452
24028 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
24031 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24033 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
24034 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
24035 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
24037 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24039 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
24041 * config/rl78/rl78.md (movdi): New define_expand.
24042 * config/rl78/rl78.c (rl78_split_movdi): New function.
24044 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
24047 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
24049 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
24050 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
24051 128-bit to produce an UNSPEC move to get the double word with the
24052 signbit and then a shift directly to do signbit.
24053 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
24054 implementation with a new version that just does either a direct
24055 move or a regular move. Move memory interface to separate insns.
24056 Move insns so they are next to the expander.
24057 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
24058 with signbit move. Split big and little endian case.
24059 (signbit<mode>2_dm_mem_le): Likewise.
24060 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
24061 (signbit<mode>2_dm2): Likewise.
24063 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24065 * config/rl78/rl78.md (anddi3): New define_expand.
24067 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24069 * config/rl78/rl78.md (umindi3): New define_expand.
24071 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24073 * config/rl78/rl78.md (smindi3): New define_expand.
24075 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24077 * config/rl78/rl78.md (smaxdi3): New define_expand.
24079 2018-01-22 Carl Love <cel@us.ibm.com>
24081 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
24082 LVX_V1TI): Add macro expansion.
24083 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
24084 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
24085 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
24086 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
24087 Change check to determine if the instruction is a byte reversing
24088 entry. Fix typo in comment.
24089 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
24090 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
24091 Add def_builtin calls for new builtins.
24092 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
24093 Add define_insn expansion.
24095 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24097 * config/rl78/rl78.md (umaxdi3): New define_expand.
24099 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24101 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
24102 for non-QImode registers.
24104 2018-01-22 Richard Biener <rguenther@suse.de>
24106 PR tree-optimization/83963
24107 * graphite-scop-detection.c (scop_detection::get_sese): Delay
24108 including the loop exit block.
24109 (scop_detection::merge_sese): Likewise.
24110 (scop_detection::add_scop): Do it here instead.
24112 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24114 * doc/sourcebuild.texi (arm_softfloat): Document.
24116 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
24119 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
24120 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
24121 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
24123 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24124 David Edelsohn <dje.gcc@gmail.com>
24127 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
24128 Change "crset eq" to "crset 2".
24129 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
24130 (*call_indirect_aix<mode>_nospec): Likewise.
24131 (*call_value_indirect_aix<mode>_nospec): Likewise.
24132 (*call_indirect_elfv2<mode>_nospec): Likewise.
24133 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
24134 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
24135 change assembly output from . to $.
24136 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
24137 (indirect_jump<mode>_nospec): Change assembly output from . to $.
24138 (*tablejump<mode>_internal1_nospec): Likewise.
24140 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
24143 * config/sh/sh_optimize_sett_clrt.cc:
24144 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
24146 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
24148 PR tree-optimization/83940
24149 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
24150 offset_dt to vect_constant_def rather than vect_unknown_def_type.
24151 (vect_check_load_store_mask): Add a mask_dt_out parameter and
24152 use it to pass back the definition type.
24153 (vect_check_store_rhs): Likewise rhs_dt_out.
24154 (vect_build_gather_load_calls): Add a mask_dt argument and use
24155 it instead of a call to vect_is_simple_use.
24156 (vectorizable_store): Update calls to vect_check_load_store_mask
24157 and vect_check_store_rhs. Use the dt returned by the latter instead
24158 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
24159 instead of calls to vect_is_simple_use. Pass the scalar rather
24160 than the vector operand to vect_is_simple_use when handling
24161 second and subsequent copies of an rhs value.
24162 (vectorizable_load): Update calls to vect_check_load_store_mask
24163 and vect_build_gather_load_calls. Use the cached mask_dt and
24164 gs_info.offset_dt instead of calls to vect_is_simple_use.
24166 2018-01-20 Jakub Jelinek <jakub@redhat.com>
24168 PR middle-end/83945
24169 * tree-emutls.c: Include gimplify.h.
24170 (lower_emutls_2): New function.
24171 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
24172 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
24173 it before further processing.
24176 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
24177 UINTVAL (trueop1) instead of INTVAL (op1).
24179 2018-01-19 Jakub Jelinek <jakub@redhat.com>
24183 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
24184 INCOMING_FRAME_SP_OFFSET if not defined.
24185 (scan_trace): Add ENTRY argument. If true and
24186 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
24187 emit a note to adjust the CFA offset.
24188 (create_cfi_notes): Adjust scan_trace callers.
24189 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
24190 INCOMING_FRAME_SP_OFFSET in the CIE.
24191 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
24192 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
24194 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
24195 * doc/tm.texi: Regenerated.
24197 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24199 PR rtl-optimization/83147
24200 * lra-constraints.c (remove_inheritance_pseudos): Use
24201 lra_substitute_pseudo_within_insn.
24203 2018-01-19 Tom de Vries <tom@codesourcery.com>
24204 Cesar Philippidis <cesar@codesourcery.com>
24207 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
24209 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
24212 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
24213 spaces for function labels.
24215 2018-01-19 Martin Liska <mliska@suse.cz>
24217 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
24218 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
24219 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
24220 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
24221 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
24222 (PRED_CONST_RETURN): Change from 69 to 65.
24223 (PRED_NULL_RETURN): Change from 91 to 71.
24224 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
24225 (PRED_LOOP_GUARD): Change from 66 to 73.
24227 2018-01-19 Martin Liska <mliska@suse.cz>
24229 * predict.c (predict_insn_def): Add new assert.
24230 (struct branch_predictor): Change type to signed integer.
24231 (test_prediction_value_range): Amend test to cover
24232 PROB_UNINITIALIZED.
24233 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
24234 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
24235 (PRED_LOOP_ITERATIONS_MAX): Likewise.
24236 (PRED_LOOP_IV_COMPARE): Likewise.
24237 * predict.h (PROB_UNINITIALIZED): Define new constant.
24239 2018-01-19 Martin Liska <mliska@suse.cz>
24241 * predict.c (dump_prediction): Add new format for
24242 analyze_brprob.py script which is enabled with -details
24244 * profile-count.h (precise_p): New function.
24246 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
24248 PR tree-optimization/83922
24249 * tree-vect-loop.c (vect_verify_full_masking): Return false if
24250 there are no statements that need masking.
24251 (vect_active_double_reduction_p): New function.
24252 (vect_analyze_loop_operations): Use it when handling phis that
24253 are not in the loop header.
24255 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
24257 PR tree-optimization/83914
24258 * tree-vect-loop.c (vectorizable_induction): Don't convert
24259 init_expr or apply the peeling adjustment for inductions
24260 that are nested within the vectorized loop.
24262 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24264 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
24267 2018-01-18 Jakub Jelinek <jakub@redhat.com>
24271 * function.h (gimplify_parameters): Add gimple_seq * argument.
24272 * function.c: Include gimple.h and options.h.
24273 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
24274 for the added local temporaries if needed.
24275 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
24276 if there are any parameter cleanups, wrap whole body into a
24277 try/finally with the cleanups.
24279 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
24282 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
24283 Use GET_MODE_CLASS for scalar floating point.
24285 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
24289 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
24290 Fix call of call_cgraph_insertion_hooks.
24292 2018-01-18 Martin Sebor <msebor@redhat.com>
24294 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
24296 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
24299 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
24302 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
24305 * common.opt: (-ffile-prefix-map): New option.
24306 * opts.c (common_handle_option): Defer it.
24307 * opts-global.c (handle_common_deferred_options): Handle it.
24308 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
24309 * file-prefix-map.h: New file.
24310 (remap_debug_filename, add_debug_prefix_map): ...here.
24311 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
24312 * final.c (debug_prefix_map, add_debug_prefix_map
24313 remap_debug_filename): Move to...
24314 * file-prefix-map.c: New file.
24315 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
24316 generalize, get rid of alloca(), use strrchr() instead of strchr().
24317 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
24318 Implement in terms of add_prefix_map().
24319 (remap_macro_filename, remap_debug_filename): Implement in term of
24321 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
24322 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
24323 * dbxout.c: Include file-prefix-map.h.
24324 * varasm.c: Likewise.
24325 * vmsdbgout.c: Likewise.
24326 * xcoffout.c: Likewise.
24327 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
24328 * doc/cppopts.texi (-fmacro-prefix-map): Document.
24329 * doc/invoke.texi (-ffile-prefix-map): Document.
24330 (-fdebug-prefix-map): Update description.
24332 2018-01-18 Martin Liska <mliska@suse.cz>
24334 * config/i386/i386.c (indirect_thunk_name): Document that also
24336 (output_indirect_thunk): Document why both instructions
24337 (pause and lfence) are generated.
24339 2018-01-18 Richard Biener <rguenther@suse.de>
24341 PR tree-optimization/83887
24342 * graphite-scop-detection.c
24343 (scop_detection::get_nearest_dom_with_single_entry): Remove.
24344 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
24345 (scop_detection::merge_sese): Re-implement with a flood-fill
24346 algorithm that properly finds a SESE region if it exists.
24348 2018-01-18 Jakub Jelinek <jakub@redhat.com>
24351 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
24352 pointer_diff optimizations use view_convert instead of convert.
24354 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24356 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
24357 Generate different code for -mno-speculate-indirect-jumps.
24358 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
24359 (*call_indirect_aix<mode>): Disable for
24360 -mno-speculate-indirect-jumps.
24361 (*call_indirect_aix<mode>_nospec): New define_insn.
24362 (*call_value_indirect_aix<mode>): Disable for
24363 -mno-speculate-indirect-jumps.
24364 (*call_value_indirect_aix<mode>_nospec): New define_insn.
24365 (*sibcall_nonlocal_sysv<mode>): Generate different code for
24366 -mno-speculate-indirect-jumps.
24367 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
24369 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
24371 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
24372 long double type, set the flags for noting the default long double
24373 type, even if we don't pass or return a long double type.
24375 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
24378 * ipa-inline.c (flatten_function): Do not overwrite final inlining
24381 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
24383 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
24384 support for merge[hl].
24385 (fold_mergehl_helper): New helper function.
24386 (tree-vector-builder.h): New #include for tree_vector_builder usage.
24387 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
24388 (altivec_vmrglw_direct): Add xxmrglw insn.
24390 2018-01-17 Andrew Waterman <andrew@sifive.com>
24392 * config/riscv/riscv.c (riscv_conditional_register_usage): If
24393 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
24395 2018-01-17 David Malcolm <dmalcolm@redhat.com>
24398 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
24399 call the lto_location_cache before reading the
24400 DECL_SOURCE_LOCATION of the types.
24402 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
24403 Richard Sandiford <richard.sandiford@linaro.org>
24405 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
24406 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
24407 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
24408 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
24409 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
24411 * config/aarch64/constraints.md (aarch64_movti_operand):
24413 * config/aarch64/predicates.md (Uti): Add new constraint.
24415 2018-01-17 Carl Love <cel@us.ibm.com>
24417 * config/rs6000/vsx.md (define_expand xl_len_r,
24418 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
24419 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
24421 (define_expand, define_insn): Move the shift left from the
24422 define_insn to the define_expand for lxvl and stxvl instructions.
24423 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
24424 and XL_LEN_R definitions to PURE.
24426 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
24428 * config/i386/i386.c (indirect_thunk_name): Declare regno
24429 as unsigned int. Compare regno with INVALID_REGNUM.
24430 (output_indirect_thunk): Ditto.
24431 (output_indirect_thunk_function): Ditto.
24432 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
24433 in the call to output_indirect_thunk_function.
24435 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
24437 PR middle-end/83884
24438 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
24439 rather than the size of inner_type to determine the stack slot size
24440 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
24442 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
24445 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
24448 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
24450 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
24451 endian Linux systems to optionally enable multilibs for selecting
24452 the long double type if the user configured an explicit type.
24453 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
24454 have no long double multilibs if not defined.
24455 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
24456 warn if the user used -mabi={ieee,ibm}longdouble and we built
24457 multilibs for long double.
24458 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
24459 appropriate multilib option.
24460 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
24462 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
24463 for building long double multilibs.
24464 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
24466 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
24468 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
24471 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
24473 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
24476 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
24479 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
24482 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
24484 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
24485 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
24487 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
24489 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
24490 different rtl trees depending on TARGET_64BIT.
24491 (rs6000_gen_lvx): Likewise.
24493 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
24495 * config/visium/visium.md (nop): Tweak comment.
24496 (hazard_nop): Likewise.
24498 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24500 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
24501 -mspeculate-indirect-jumps.
24502 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
24503 for -mno-speculate-indirect-jumps.
24504 (*call_indirect_elfv2<mode>_nospec): New define_insn.
24505 (*call_value_indirect_elfv2<mode>): Disable for
24506 -mno-speculate-indirect-jumps.
24507 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
24508 (indirect_jump): Emit different RTL for
24509 -mno-speculate-indirect-jumps.
24510 (*indirect_jump<mode>): Disable for
24511 -mno-speculate-indirect-jumps.
24512 (*indirect_jump<mode>_nospec): New define_insn.
24513 (tablejump): Emit different RTL for
24514 -mno-speculate-indirect-jumps.
24515 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
24516 (tablejumpsi_nospec): New define_expand.
24517 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
24518 (tablejumpdi_nospec): New define_expand.
24519 (*tablejump<mode>_internal1): Disable for
24520 -mno-speculate-indirect-jumps.
24521 (*tablejump<mode>_internal1_nospec): New define_insn.
24522 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
24525 2018-01-16 Artyom Skrobov tyomitch@gmail.com
24527 * caller-save.c (insert_save): Drop unnecessary parameter. All
24530 2018-01-16 Jakub Jelinek <jakub@redhat.com>
24531 Richard Biener <rguenth@suse.de>
24534 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
24535 return early, inline manually is_gimple_sizepos. Make sure if we
24536 call gimplify_expr we don't end up with a gimple constant.
24537 * tree.c (variably_modified_type_p): Don't return true for
24538 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
24539 * gimplify.h (is_gimple_sizepos): Remove.
24541 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
24543 PR tree-optimization/83857
24544 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
24545 vectorizable_live_operation for pure SLP statements.
24546 (vectorizable_live_operation): Handle PHIs.
24548 2018-01-16 Richard Biener <rguenther@suse.de>
24550 PR tree-optimization/83867
24551 * tree-vect-stmts.c (vect_transform_stmt): Precompute
24552 nested_in_vect_loop_p since the scalar stmt may get invalidated.
24554 2018-01-16 Jakub Jelinek <jakub@redhat.com>
24557 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
24558 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
24559 If off is not INTEGER_CST, issue a may not be aligned warning
24560 rather than isn't aligned. Use isn%'t rather than isn't.
24561 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
24563 <case MULT_EXPR>: Improve the case when bottom and one of the
24564 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
24565 operand, in that case check if the other operand is multiple of
24566 bottom divided by the INTEGER_CST operand.
24568 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
24571 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
24572 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
24573 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
24574 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
24575 * config/pa/pa.c (pa_function_arg_advance): Likewise.
24576 (pa_function_arg, pa_arg_partial_bytes): Likewise.
24577 (pa_function_arg_size): New function.
24579 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
24581 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
24582 in a separate statement.
24584 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
24586 PR tree-optimization/83847
24587 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
24588 group gathers and scatters.
24590 2018-01-16 Jakub Jelinek <jakub@redhat.com>
24592 PR rtl-optimization/86620
24593 * params.def (max-sched-ready-insns): Bump minimum value to 1.
24595 PR rtl-optimization/83213
24596 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
24597 to last if both are JUMP_INSNs.
24599 PR tree-optimization/83843
24600 * gimple-ssa-store-merging.c
24601 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
24602 store_immediate_info for bswap/nop orig_stores.
24604 2018-01-15 Andrew Waterman <andrew@sifive.com>
24606 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
24608 <UDIV>: Increase cost if !TARGET_DIV.
24610 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
24612 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
24613 (define_attr "cr_logical_3op"): New.
24614 (cceq_ior_compare): Adjust.
24615 (cceq_ior_compare_complement): Adjust.
24616 (*cceq_rev_compare): Adjust.
24617 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
24618 (is_cracked_insn): Adjust.
24619 (insn_must_be_first_in_group): Adjust.
24620 * config/rs6000/40x.md: Adjust.
24621 * config/rs6000/440.md: Adjust.
24622 * config/rs6000/476.md: Adjust.
24623 * config/rs6000/601.md: Adjust.
24624 * config/rs6000/603.md: Adjust.
24625 * config/rs6000/6xx.md: Adjust.
24626 * config/rs6000/7450.md: Adjust.
24627 * config/rs6000/7xx.md: Adjust.
24628 * config/rs6000/8540.md: Adjust.
24629 * config/rs6000/cell.md: Adjust.
24630 * config/rs6000/e300c2c3.md: Adjust.
24631 * config/rs6000/e500mc.md: Adjust.
24632 * config/rs6000/e500mc64.md: Adjust.
24633 * config/rs6000/e5500.md: Adjust.
24634 * config/rs6000/e6500.md: Adjust.
24635 * config/rs6000/mpc.md: Adjust.
24636 * config/rs6000/power4.md: Adjust.
24637 * config/rs6000/power5.md: Adjust.
24638 * config/rs6000/power6.md: Adjust.
24639 * config/rs6000/power7.md: Adjust.
24640 * config/rs6000/power8.md: Adjust.
24641 * config/rs6000/power9.md: Adjust.
24642 * config/rs6000/rs64.md: Adjust.
24643 * config/rs6000/titan.md: Adjust.
24645 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
24647 * config/i386/predicates.md (indirect_branch_operand): Rewrite
24648 ix86_indirect_branch_register logic.
24650 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
24652 * config/i386/constraints.md (Bs): Update
24653 ix86_indirect_branch_register check. Don't check
24654 ix86_indirect_branch_register with GOT_memory_operand.
24656 * config/i386/predicates.md (GOT_memory_operand): Don't check
24657 ix86_indirect_branch_register here.
24658 (GOT32_symbol_operand): Likewise.
24660 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
24662 * config/i386/predicates.md (constant_call_address_operand):
24663 Rewrite ix86_indirect_branch_register logic.
24664 (sibcall_insn_operand): Likewise.
24666 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
24668 * config/i386/constraints.md (Bs): Replace
24669 ix86_indirect_branch_thunk_register with
24670 ix86_indirect_branch_register.
24672 * config/i386/i386.md (indirect_jump): Likewise.
24673 (tablejump): Likewise.
24674 (*sibcall_memory): Likewise.
24675 (*sibcall_value_memory): Likewise.
24676 Peepholes of indirect call and jump via memory: Likewise.
24677 * config/i386/i386.opt: Likewise.
24678 * config/i386/predicates.md (indirect_branch_operand): Likewise.
24679 (GOT_memory_operand): Likewise.
24680 (call_insn_operand): Likewise.
24681 (sibcall_insn_operand): Likewise.
24682 (GOT32_symbol_operand): Likewise.
24684 2018-01-15 Jakub Jelinek <jakub@redhat.com>
24686 PR middle-end/83837
24687 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
24688 type rather than type addr's type points to.
24689 (expand_omp_atomic_mutex): Likewise.
24690 (expand_omp_atomic): Likewise.
24692 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
24695 * config/i386/i386.c (output_indirect_thunk_function): Use
24696 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
24697 for __x86_return_thunk.
24699 2018-01-15 Richard Biener <rguenther@suse.de>
24701 PR middle-end/83850
24702 * expmed.c (extract_bit_field_1): Fix typo.
24704 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24707 * config/arm/iterators.md (VF): New mode iterator.
24708 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
24709 Remove integer-related logic from pattern.
24710 (neon_vabd<mode>_3): Likewise.
24712 2018-01-15 Jakub Jelinek <jakub@redhat.com>
24714 PR middle-end/82694
24715 * common.opt (fstrict-overflow): No longer an alias.
24716 (fwrapv-pointer): New option.
24717 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
24718 also for pointer types based on flag_wrapv_pointer.
24719 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
24720 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
24721 opts->x_flag_wrapv got set.
24722 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
24723 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
24724 POINTER_TYPE_OVERFLOW_UNDEFINED.
24725 * match.pd: Likewise in address comparison pattern.
24726 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
24728 2018-01-15 Richard Biener <rguenther@suse.de>
24731 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
24732 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
24733 Reset type names to their identifier if their TYPE_DECL doesn't
24734 have linkage (and thus is used for ODR and devirt).
24735 (save_debug_info_for_decl): Remove.
24736 (save_debug_info_for_type): Likewise.
24737 (add_tree_to_fld_list): Adjust.
24738 * tree-pretty-print.c (dump_generic_node): Make dumping of
24739 type names more robust.
24741 2018-01-15 Richard Biener <rguenther@suse.de>
24743 * BASE-VER: Bump to 8.0.1.
24745 2018-01-14 Martin Sebor <msebor@redhat.com>
24748 * builtins.c (check_access): Avoid warning when the no-warning bit
24751 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
24753 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
24754 * ira-color (allocno_hard_regs_compare): Likewise.
24756 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
24759 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
24760 Use .pushsection/.popsection.
24762 2018-01-14 Martin Sebor <msebor@redhat.com>
24765 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
24767 2018-01-14 Jakub Jelinek <jakub@redhat.com>
24769 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
24770 entry from extra_headers.
24771 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
24772 extra_headers, make the list bitwise identical to the i?86-*-* one.
24774 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
24776 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
24777 -mcmodel=large with -mindirect-branch=thunk,
24778 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
24779 -mfunction-return=thunk-extern.
24780 * doc/invoke.texi: Document -mcmodel=large is incompatible with
24781 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
24782 -mfunction-return=thunk and -mfunction-return=thunk-extern.
24784 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
24786 * config/i386/i386.c (print_reg): Print the name of the full
24787 integer register without '%'.
24788 (ix86_print_operand): Handle 'V'.
24789 * doc/extend.texi: Document 'V' modifier.
24791 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
24793 * config/i386/constraints.md (Bs): Disallow memory operand for
24794 -mindirect-branch-register.
24796 * config/i386/predicates.md (indirect_branch_operand): Likewise.
24797 (GOT_memory_operand): Likewise.
24798 (call_insn_operand): Likewise.
24799 (sibcall_insn_operand): Likewise.
24800 (GOT32_symbol_operand): Likewise.
24801 * config/i386/i386.md (indirect_jump): Call convert_memory_address
24802 for -mindirect-branch-register.
24803 (tablejump): Likewise.
24804 (*sibcall_memory): Likewise.
24805 (*sibcall_value_memory): Likewise.
24806 Disallow peepholes of indirect call and jump via memory for
24807 -mindirect-branch-register.
24808 (*call_pop): Replace m with Bw.
24809 (*call_value_pop): Likewise.
24810 (*sibcall_pop_memory): Replace m with Bs.
24811 * config/i386/i386.opt (mindirect-branch-register): New option.
24812 * doc/invoke.texi: Document -mindirect-branch-register option.
24814 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
24816 * config/i386/i386-protos.h (ix86_output_function_return): New.
24817 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
24818 set function_return_type.
24819 (indirect_thunk_name): Add ret_p to indicate thunk for function
24821 (output_indirect_thunk_function): Pass false to
24822 indirect_thunk_name.
24823 (ix86_output_indirect_branch_via_reg): Likewise.
24824 (ix86_output_indirect_branch_via_push): Likewise.
24825 (output_indirect_thunk_function): Create alias for function
24826 return thunk if regno < 0.
24827 (ix86_output_function_return): New function.
24828 (ix86_handle_fndecl_attribute): Handle function_return.
24829 (ix86_attribute_table): Add function_return.
24830 * config/i386/i386.h (machine_function): Add
24831 function_return_type.
24832 * config/i386/i386.md (simple_return_internal): Use
24833 ix86_output_function_return.
24834 (simple_return_internal_long): Likewise.
24835 * config/i386/i386.opt (mfunction-return=): New option.
24836 (indirect_branch): Mention -mfunction-return=.
24837 * doc/extend.texi: Document function_return function attribute.
24838 * doc/invoke.texi: Document -mfunction-return= option.
24840 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
24842 * config/i386/i386-opts.h (indirect_branch): New.
24843 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
24844 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
24845 with local indirect jump when converting indirect call and jump.
24846 (ix86_set_indirect_branch_type): New.
24847 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
24848 (indirectlabelno): New.
24849 (indirect_thunk_needed): Likewise.
24850 (indirect_thunk_bnd_needed): Likewise.
24851 (indirect_thunks_used): Likewise.
24852 (indirect_thunks_bnd_used): Likewise.
24853 (INDIRECT_LABEL): Likewise.
24854 (indirect_thunk_name): Likewise.
24855 (output_indirect_thunk): Likewise.
24856 (output_indirect_thunk_function): Likewise.
24857 (ix86_output_indirect_branch_via_reg): Likewise.
24858 (ix86_output_indirect_branch_via_push): Likewise.
24859 (ix86_output_indirect_branch): Likewise.
24860 (ix86_output_indirect_jmp): Likewise.
24861 (ix86_code_end): Call output_indirect_thunk_function if needed.
24862 (ix86_output_call_insn): Call ix86_output_indirect_branch if
24864 (ix86_handle_fndecl_attribute): Handle indirect_branch.
24865 (ix86_attribute_table): Add indirect_branch.
24866 * config/i386/i386.h (machine_function): Add indirect_branch_type
24867 and has_local_indirect_jump.
24868 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
24870 (tablejump): Likewise.
24871 (*indirect_jump): Use ix86_output_indirect_jmp.
24872 (*tablejump_1): Likewise.
24873 (simple_return_indirect_internal): Likewise.
24874 * config/i386/i386.opt (mindirect-branch=): New option.
24875 (indirect_branch): New.
24878 (thunk-inline): Likewise.
24879 (thunk-extern): Likewise.
24880 * doc/extend.texi: Document indirect_branch function attribute.
24881 * doc/invoke.texi: Document -mindirect-branch= option.
24883 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
24886 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
24888 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
24890 * ipa-inline.c (want_inline_small_function_p): Return false if
24891 inlining has already failed with CIF_FINAL_ERROR.
24892 (update_caller_keys): Call want_inline_small_function_p before
24894 (update_callee_keys): Likewise.
24896 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
24898 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
24900 (rs6000_quadword_masked_address_p): Likewise.
24901 (quad_aligned_load_p): Likewise.
24902 (quad_aligned_store_p): Likewise.
24903 (const_load_sequence_p): Add comment to describe the outer-most loop.
24904 (mimic_memory_attributes_and_flags): New function.
24905 (rs6000_gen_stvx): Likewise.
24906 (replace_swapped_aligned_store): Likewise.
24907 (rs6000_gen_lvx): Likewise.
24908 (replace_swapped_aligned_load): Likewise.
24909 (replace_swapped_load_constant): Capitalize argument name in
24910 comment describing this function.
24911 (rs6000_analyze_swaps): Add a third pass to search for vector loads
24912 and stores that access quad-word aligned addresses and replace
24913 with stvx or lvx instructions when appropriate.
24914 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
24915 New function prototype.
24916 (rs6000_quadword_masked_address_p): Likewise.
24917 (rs6000_gen_lvx): Likewise.
24918 (rs6000_gen_stvx): Likewise.
24919 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
24920 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
24921 when memory address is aligned.
24922 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
24923 this split to select lvx instruction when memory address is aligned.
24924 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
24925 instruction when memory address is aligned.
24926 (*vsx_le_perm_load_v16qi): Likewise.
24927 (four unnamed splitters): Modify to select the stvx instruction
24928 when memory is aligned.
24930 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
24932 * predict.c (determine_unlikely_bbs): Handle correctly BBs
24933 which appears in the queue multiple times.
24935 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24936 Alan Hayward <alan.hayward@arm.com>
24937 David Sherwood <david.sherwood@arm.com>
24939 * tree-vectorizer.h (vec_lower_bound): New structure.
24940 (_loop_vec_info): Add check_nonzero and lower_bounds.
24941 (LOOP_VINFO_CHECK_NONZERO): New macro.
24942 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
24943 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
24944 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
24945 fields. Make seg_len the distance travelled, not including the
24947 (dr_direction_indicator): Declare.
24948 (dr_zero_step_indicator): Likewise.
24949 (dr_known_forward_stride_p): Likewise.
24950 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
24952 (runtime_alias_check_p): Allow runtime alias checks with
24954 (operator ==): Compare access_size and align.
24955 (prune_runtime_alias_test_list): Rework for new distinction between
24956 the access_size and seg_len.
24957 (create_intersect_range_checks_index): Likewise. Cope with polynomial
24959 (get_segment_min_max): New function.
24960 (create_intersect_range_checks): Use it.
24961 (dr_step_indicator): New function.
24962 (dr_direction_indicator): Likewise.
24963 (dr_zero_step_indicator): Likewise.
24964 (dr_known_forward_stride_p): Likewise.
24965 * tree-loop-distribution.c (data_ref_segment_size): Return
24966 DR_STEP * (niters - 1).
24967 (compute_alias_check_pairs): Update call to the dr_with_seg_len
24969 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
24970 (vect_preserves_scalar_order_p): New function, split out from...
24971 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
24972 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
24973 (vect_vfa_access_size): New function.
24974 (vect_vfa_align): Likewise.
24975 (vect_compile_time_alias): Take access_size_a and access_b arguments.
24976 (dump_lower_bound): New function.
24977 (vect_check_lower_bound): Likewise.
24978 (vect_small_gap_p): Likewise.
24979 (vectorizable_with_step_bound_p): Likewise.
24980 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
24981 depencies if the vectorization factor is 1. Convert the checks
24982 for nonzero steps into checks on the bounds of DR_STEP. Try using
24983 a bunds check for variable steps if the minimum required step is
24984 relatively small. Update calls to the dr_with_seg_len
24985 constructor and to vect_compile_time_alias.
24986 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
24988 (vect_loop_versioning): Call it.
24989 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
24991 (vect_estimate_min_profitable_iters): Account for any bounds checks.
24993 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24994 Alan Hayward <alan.hayward@arm.com>
24995 David Sherwood <david.sherwood@arm.com>
24997 * doc/sourcebuild.texi (vect_scatter_store): Document.
24998 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
25000 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
25002 * genopinit.c (main): Add supports_vec_scatter_store and
25003 supports_vec_scatter_store_cached to target_optabs.
25004 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
25005 IFN_MASK_SCATTER_STORE.
25006 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
25008 * internal-fn.h (internal_store_fn_p): Declare.
25009 (internal_fn_stored_value_index): Likewise.
25010 * internal-fn.c (scatter_store_direct): New macro.
25011 (expand_scatter_store_optab_fn): New function.
25012 (direct_scatter_store_optab_supported_p): New macro.
25013 (internal_store_fn_p): New function.
25014 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
25015 IFN_MASK_SCATTER_STORE.
25016 (internal_fn_mask_index): Likewise.
25017 (internal_fn_stored_value_index): New function.
25018 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
25019 for scatter stores.
25020 * optabs-query.h (supports_vec_scatter_store_p): Declare.
25021 * optabs-query.c (supports_vec_scatter_store_p): New function.
25022 * tree-vectorizer.h (vect_get_store_rhs): Declare.
25023 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
25024 true for scatter stores.
25025 (vect_gather_scatter_fn_p): Handle scatter stores too.
25026 (vect_check_gather_scatter): Consider using scatter stores if
25027 supports_vec_scatter_store_p.
25028 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
25029 scatter stores too.
25030 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
25031 internal_fn_stored_value_index.
25032 (check_load_store_masking): Handle scatter stores too.
25033 (vect_get_store_rhs): Make public.
25034 (vectorizable_call): Use internal_store_fn_p.
25035 (vectorizable_store): Handle scatter store internal functions.
25036 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
25037 when deciding whether the end of the group has been reached.
25038 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
25039 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
25040 (mask_scatter_store<mode>): New insns.
25042 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25043 Alan Hayward <alan.hayward@arm.com>
25044 David Sherwood <david.sherwood@arm.com>
25046 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
25047 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
25048 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
25050 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
25051 Use vect_truncate_gather_scatter_offset if we can't treat the
25052 operation as a normal gather load or scatter store.
25053 (get_group_load_store_type): Take the gather_scatter_info
25054 as argument. Try using a gather load or scatter store for
25055 single-element groups.
25056 (get_load_store_type): Update calls to get_group_load_store_type
25057 and vect_use_strided_gather_scatters_p.
25059 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25060 Alan Hayward <alan.hayward@arm.com>
25061 David Sherwood <david.sherwood@arm.com>
25063 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
25064 optional tree argument.
25065 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
25067 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
25068 but continue to use the current value as a fallback.
25069 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
25070 to compare the updates.
25071 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
25072 (get_load_store_type): Use it when handling a strided access.
25073 (vect_get_strided_load_store_ops): New function.
25074 (vect_get_data_ptr_increment): Likewise.
25075 (vectorizable_load): Handle strided gather loads. Always pass
25076 a step to vect_create_data_ref_ptr and bump_vector_ptr.
25078 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25079 Alan Hayward <alan.hayward@arm.com>
25080 David Sherwood <david.sherwood@arm.com>
25082 * doc/md.texi (gather_load@var{m}): Document.
25083 (mask_gather_load@var{m}): Likewise.
25084 * genopinit.c (main): Add supports_vec_gather_load and
25085 supports_vec_gather_load_cached to target_optabs.
25086 * optabs-tree.c (init_tree_optimization_optabs): Use
25087 ggc_cleared_alloc to allocate target_optabs.
25088 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
25089 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
25091 * internal-fn.h (internal_load_fn_p): Declare.
25092 (internal_gather_scatter_fn_p): Likewise.
25093 (internal_fn_mask_index): Likewise.
25094 (internal_gather_scatter_fn_supported_p): Likewise.
25095 * internal-fn.c (gather_load_direct): New macro.
25096 (expand_gather_load_optab_fn): New function.
25097 (direct_gather_load_optab_supported_p): New macro.
25098 (direct_internal_fn_optab): New function.
25099 (internal_load_fn_p): Likewise.
25100 (internal_gather_scatter_fn_p): Likewise.
25101 (internal_fn_mask_index): Likewise.
25102 (internal_gather_scatter_fn_supported_p): Likewise.
25103 * optabs-query.c (supports_at_least_one_mode_p): New function.
25104 (supports_vec_gather_load_p): Likewise.
25105 * optabs-query.h (supports_vec_gather_load_p): Declare.
25106 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
25107 and memory_type field.
25108 (NUM_PATTERNS): Bump to 15.
25109 * tree-vect-data-refs.c: Include internal-fn.h.
25110 (vect_gather_scatter_fn_p): New function.
25111 (vect_describe_gather_scatter_call): Likewise.
25112 (vect_check_gather_scatter): Try using internal functions for
25113 gather loads. Recognize existing calls to a gather load function.
25114 (vect_analyze_data_refs): Consider using gather loads if
25115 supports_vec_gather_load_p.
25116 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
25117 (vect_get_gather_scatter_offset_type): Likewise.
25118 (vect_convert_mask_for_vectype): Likewise.
25119 (vect_add_conversion_to_patterm): Likewise.
25120 (vect_try_gather_scatter_pattern): Likewise.
25121 (vect_recog_gather_scatter_pattern): New pattern recognizer.
25122 (vect_vect_recog_func_ptrs): Add it.
25123 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
25124 internal_fn_mask_index and internal_gather_scatter_fn_p.
25125 (check_load_store_masking): Take the gather_scatter_info as an
25126 argument and handle gather loads.
25127 (vect_get_gather_scatter_ops): New function.
25128 (vectorizable_call): Check internal_load_fn_p.
25129 (vectorizable_load): Likewise. Handle gather load internal
25131 (vectorizable_store): Update call to check_load_store_masking.
25132 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
25133 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
25134 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
25135 (aarch64_gather_scale_operand_d): New predicates.
25136 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
25137 (mask_gather_load<mode>): New insns.
25139 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25140 Alan Hayward <alan.hayward@arm.com>
25141 David Sherwood <david.sherwood@arm.com>
25143 * optabs.def (fold_left_plus_optab): New optab.
25144 * doc/md.texi (fold_left_plus_@var{m}): Document.
25145 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
25146 * internal-fn.c (fold_left_direct): Define.
25147 (expand_fold_left_optab_fn): Likewise.
25148 (direct_fold_left_optab_supported_p): Likewise.
25149 * fold-const-call.c (fold_const_fold_left): New function.
25150 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
25151 * tree-parloops.c (valid_reduction_p): New function.
25152 (gather_scalar_reductions): Use it.
25153 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
25154 (vect_finish_replace_stmt): Declare.
25155 * tree-vect-loop.c (fold_left_reduction_fn): New function.
25156 (needs_fold_left_reduction_p): New function, split out from...
25157 (vect_is_simple_reduction): ...here. Accept reductions that
25158 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
25159 (vect_force_simple_reduction): Also store the reduction type in
25160 the assignment's STMT_VINFO_REDUC_TYPE.
25161 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
25162 (merge_with_identity): New function.
25163 (vect_expand_fold_left): Likewise.
25164 (vectorize_fold_left_reduction): Likewise.
25165 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
25166 scalar phi in place for it. Check for target support and reject
25167 cases that would reassociate the operation. Defer the transform
25168 phase to vectorize_fold_left_reduction.
25169 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
25170 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
25171 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
25173 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25175 * tree-if-conv.c (predicate_mem_writes): Remove redundant
25176 call to ifc_temp_var.
25178 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25179 Alan Hayward <alan.hayward@arm.com>
25180 David Sherwood <david.sherwood@arm.com>
25182 * target.def (legitimize_address_displacement): Take the original
25183 offset as a poly_int.
25184 * targhooks.h (default_legitimize_address_displacement): Update
25186 * targhooks.c (default_legitimize_address_displacement): Likewise.
25187 * doc/tm.texi: Regenerate.
25188 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
25189 as an argument, moving assert of ad->disp == ad->disp_term to...
25190 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
25191 Try calling targetm.legitimize_address_displacement before expanding
25192 the address rather than afterwards, and adjust for the new interface.
25193 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
25194 Match the new hook interface. Handle SVE addresses.
25195 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
25196 new hook interface.
25198 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25200 * Makefile.in (OBJS): Add early-remat.o.
25201 * target.def (select_early_remat_modes): New hook.
25202 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
25203 * doc/tm.texi: Regenerate.
25204 * targhooks.h (default_select_early_remat_modes): Declare.
25205 * targhooks.c (default_select_early_remat_modes): New function.
25206 * timevar.def (TV_EARLY_REMAT): New timevar.
25207 * passes.def (pass_early_remat): New pass.
25208 * tree-pass.h (make_pass_early_remat): Declare.
25209 * early-remat.c: New file.
25210 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
25212 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
25214 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25215 Alan Hayward <alan.hayward@arm.com>
25216 David Sherwood <david.sherwood@arm.com>
25218 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
25219 vfm1 with a bound_epilog parameter.
25220 (vect_do_peeling): Update calls accordingly, and move the prologue
25221 call earlier in the function. Treat the base bound_epilog as 0 for
25222 fully-masked loops and retain vf - 1 for other loops. Add 1 to
25223 this base when peeling for gaps.
25224 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
25225 with fully-masked loops.
25226 (vect_estimate_min_profitable_iters): Handle the single peeled
25227 iteration in that case.
25229 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25230 Alan Hayward <alan.hayward@arm.com>
25231 David Sherwood <david.sherwood@arm.com>
25233 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
25234 single-element interleaving even if the size is not a power of 2.
25235 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
25236 accesses for single-element interleaving if the group size is
25239 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25240 Alan Hayward <alan.hayward@arm.com>
25241 David Sherwood <david.sherwood@arm.com>
25243 * doc/md.texi (fold_extract_last_@var{m}): Document.
25244 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
25245 * optabs.def (fold_extract_last_optab): New optab.
25246 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
25247 * internal-fn.c (fold_extract_direct): New macro.
25248 (expand_fold_extract_optab_fn): Likewise.
25249 (direct_fold_extract_optab_supported_p): Likewise.
25250 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
25251 * tree-vect-loop.c (vect_model_reduction_cost): Handle
25252 EXTRACT_LAST_REDUCTION.
25253 (get_initial_def_for_reduction): Do not create an initial vector
25254 for EXTRACT_LAST_REDUCTION reductions.
25255 (vectorizable_reduction): Leave the scalar phi in place for
25256 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
25257 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
25258 epilogue code for EXTRACT_LAST_REDUCTION and defer the
25259 transform phase to vectorizable_condition.
25260 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
25262 (vect_finish_stmt_generation): ...here.
25263 (vect_finish_replace_stmt): New function.
25264 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
25265 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
25267 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
25269 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25270 Alan Hayward <alan.hayward@arm.com>
25271 David Sherwood <david.sherwood@arm.com>
25273 * doc/md.texi (extract_last_@var{m}): Document.
25274 * optabs.def (extract_last_optab): New optab.
25275 * internal-fn.def (EXTRACT_LAST): New internal function.
25276 * internal-fn.c (cond_unary_direct): New macro.
25277 (expand_cond_unary_optab_fn): Likewise.
25278 (direct_cond_unary_optab_supported_p): Likewise.
25279 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
25280 loops using EXTRACT_LAST.
25281 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
25282 (extract_last_<mode>): ...this optab.
25283 (vec_extract<mode><Vel>): Update accordingly.
25285 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25286 Alan Hayward <alan.hayward@arm.com>
25287 David Sherwood <david.sherwood@arm.com>
25289 * target.def (empty_mask_is_expensive): New hook.
25290 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
25291 * doc/tm.texi: Regenerate.
25292 * targhooks.h (default_empty_mask_is_expensive): Declare.
25293 * targhooks.c (default_empty_mask_is_expensive): New function.
25294 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
25295 if the target says that empty masks are expensive.
25296 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
25298 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
25300 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25301 Alan Hayward <alan.hayward@arm.com>
25302 David Sherwood <david.sherwood@arm.com>
25304 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
25305 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
25306 (vect_use_loop_mask_for_alignment_p): New function.
25307 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
25308 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
25309 niters_skip argument. Make sure that the first niters_skip elements
25310 of the first iteration are inactive.
25311 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
25312 Update call to vect_set_loop_masks_directly.
25313 (get_misalign_in_elems): New function, split out from...
25314 (vect_gen_prolog_loop_niters): ...here.
25315 (vect_update_init_of_dr): Take a code argument that specifies whether
25316 the adjustment should be added or subtracted.
25317 (vect_update_init_of_drs): Likewise.
25318 (vect_prepare_for_masked_peels): New function.
25319 (vect_do_peeling): Skip prologue peeling if we're using a mask
25320 instead. Update call to vect_update_inits_of_drs.
25321 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
25323 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
25324 alignment. Do not include the number of peeled iterations in
25325 the minimum threshold in that case.
25326 (vectorizable_induction): Adjust the start value down by
25327 LOOP_VINFO_MASK_SKIP_NITERS iterations.
25328 (vect_transform_loop): Call vect_prepare_for_masked_peels.
25329 Take the number of skipped iterations into account when calculating
25331 * tree-vect-stmts.c (vect_gen_while_not): New function.
25333 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25334 Alan Hayward <alan.hayward@arm.com>
25335 David Sherwood <david.sherwood@arm.com>
25337 * doc/sourcebuild.texi (vect_fully_masked): Document.
25338 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
25339 default value to 0.
25340 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
25342 (vect_analyze_loop_2): ...here. Don't check the vectorization
25343 factor against the number of loop iterations if the loop is
25346 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25347 Alan Hayward <alan.hayward@arm.com>
25348 David Sherwood <david.sherwood@arm.com>
25350 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
25351 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
25352 (dump_groups): Update accordingly.
25353 (iv_use::mem_type): New member variable.
25354 (address_p): New function.
25355 (record_use): Add a mem_type argument and initialize the new
25357 (record_group_use): Add a mem_type argument. Use address_p.
25358 Remove obsolete null checks of base_object. Update call to record_use.
25359 (find_interesting_uses_op): Update call to record_group_use.
25360 (find_interesting_uses_cond): Likewise.
25361 (find_interesting_uses_address): Likewise.
25362 (get_mem_type_for_internal_fn): New function.
25363 (find_address_like_use): Likewise.
25364 (find_interesting_uses_stmt): Try find_address_like_use before
25365 calling find_interesting_uses_op.
25366 (addr_offset_valid_p): Use the iv mem_type field as the type
25367 of the addressed memory.
25368 (add_autoinc_candidates): Likewise.
25369 (get_address_cost): Likewise.
25370 (split_small_address_groups_p): Use address_p.
25371 (split_address_groups): Likewise.
25372 (add_iv_candidate_for_use): Likewise.
25373 (autoinc_possible_for_pair): Likewise.
25374 (rewrite_groups): Likewise.
25375 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
25376 (determine_group_iv_cost): Update after split of USE_ADDRESS.
25377 (get_alias_ptr_type_for_ptr_address): New function.
25378 (rewrite_use_address): Rewrite address uses in calls that were
25379 identified by find_address_like_use.
25381 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25382 Alan Hayward <alan.hayward@arm.com>
25383 David Sherwood <david.sherwood@arm.com>
25385 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
25387 * gimple-expr.h (is_gimple_addressable: Likewise.
25388 * gimple-expr.c (is_gimple_address): Likewise.
25389 * internal-fn.c (expand_call_mem_ref): New function.
25390 (expand_mask_load_optab_fn): Use it.
25391 (expand_mask_store_optab_fn): Likewise.
25393 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25394 Alan Hayward <alan.hayward@arm.com>
25395 David Sherwood <david.sherwood@arm.com>
25397 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
25398 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
25399 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
25400 (cond_umax@var{mode}): Document.
25401 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
25402 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
25403 (cond_umin_optab, cond_umax_optab): New optabs.
25404 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
25405 (COND_IOR, COND_XOR): New internal functions.
25406 * internal-fn.h (get_conditional_internal_fn): Declare.
25407 * internal-fn.c (cond_binary_direct): New macro.
25408 (expand_cond_binary_optab_fn): Likewise.
25409 (direct_cond_binary_optab_supported_p): Likewise.
25410 (get_conditional_internal_fn): New function.
25411 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
25412 Cope with reduction statements that are vectorized as calls rather
25414 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
25415 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
25416 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
25417 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
25418 (UNSPEC_COND_EOR): New unspecs.
25419 (optab): Add mappings for them.
25420 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
25421 (sve_int_op, sve_fp_op): New int attributes.
25423 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25424 Alan Hayward <alan.hayward@arm.com>
25425 David Sherwood <david.sherwood@arm.com>
25427 * optabs.def (while_ult_optab): New optab.
25428 * doc/md.texi (while_ult@var{m}@var{n}): Document.
25429 * internal-fn.def (WHILE_ULT): New internal function.
25430 * internal-fn.h (direct_internal_fn_supported_p): New override
25431 that takes two types as argument.
25432 * internal-fn.c (while_direct): New macro.
25433 (expand_while_optab_fn): New function.
25434 (convert_optab_supported_p): Likewise.
25435 (direct_while_optab_supported_p): New macro.
25436 * wide-int.h (wi::udiv_ceil): New function.
25437 * tree-vectorizer.h (rgroup_masks): New structure.
25438 (vec_loop_masks): New typedef.
25439 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
25440 and fully_masked_p.
25441 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
25442 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
25443 (vect_max_vf): New function.
25444 (slpeel_make_loop_iterate_ntimes): Delete.
25445 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
25446 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
25447 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
25448 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
25449 internal-fn.h, stor-layout.h and optabs-query.h.
25450 (vect_set_loop_mask): New function.
25451 (add_preheader_seq): Likewise.
25452 (add_header_seq): Likewise.
25453 (interleave_supported_p): Likewise.
25454 (vect_maybe_permute_loop_masks): Likewise.
25455 (vect_set_loop_masks_directly): Likewise.
25456 (vect_set_loop_condition_masked): Likewise.
25457 (vect_set_loop_condition_unmasked): New function, split out from
25458 slpeel_make_loop_iterate_ntimes.
25459 (slpeel_make_loop_iterate_ntimes): Rename to..
25460 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
25461 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
25462 (vect_do_peeling): Update call accordingly.
25463 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
25465 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
25466 mask_compare_type, can_fully_mask_p and fully_masked_p.
25467 (release_vec_loop_masks): New function.
25468 (_loop_vec_info): Use it to free the loop masks.
25469 (can_produce_all_loop_masks_p): New function.
25470 (vect_get_max_nscalars_per_iter): Likewise.
25471 (vect_verify_full_masking): Likewise.
25472 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
25473 retries, and free the mask rgroups before retrying. Check loop-wide
25474 reasons for disallowing fully-masked loops. Make the final decision
25475 about whether use a fully-masked loop or not.
25476 (vect_estimate_min_profitable_iters): Do not assume that peeling
25477 for the number of iterations will be needed for fully-masked loops.
25478 (vectorizable_reduction): Disable fully-masked loops.
25479 (vectorizable_live_operation): Likewise.
25480 (vect_halve_mask_nunits): New function.
25481 (vect_double_mask_nunits): Likewise.
25482 (vect_record_loop_mask): Likewise.
25483 (vect_get_loop_mask): Likewise.
25484 (vect_transform_loop): Handle the case in which the final loop
25485 iteration might handle a partial vector. Call vect_set_loop_condition
25486 instead of slpeel_make_loop_iterate_ntimes.
25487 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
25488 (check_load_store_masking): New function.
25489 (prepare_load_store_mask): Likewise.
25490 (vectorizable_store): Handle fully-masked loops.
25491 (vectorizable_load): Likewise.
25492 (supportable_widening_operation): Use vect_halve_mask_nunits for
25494 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
25495 (vect_gen_while): New function.
25496 * config/aarch64/aarch64.md (umax<mode>3): New expander.
25497 (aarch64_uqdec<mode>): New insn.
25499 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25500 Alan Hayward <alan.hayward@arm.com>
25501 David Sherwood <david.sherwood@arm.com>
25503 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
25504 (reduc_xor_scal_optab): New optabs.
25505 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
25506 (reduc_xor_scal_@var{m}): Document.
25507 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
25508 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
25509 internal functions.
25510 * fold-const-call.c (fold_const_call): Handle them.
25511 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
25512 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
25513 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
25514 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
25515 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
25516 (UNSPEC_XORV): New unspecs.
25517 (optab): Add entries for them.
25518 (BITWISEV): New int iterator.
25519 (bit_reduc_op): New int attributes.
25521 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25522 Alan Hayward <alan.hayward@arm.com>
25523 David Sherwood <david.sherwood@arm.com>
25525 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
25526 * internal-fn.def (VEC_SHL_INSERT): New internal function.
25527 * optabs.def (vec_shl_insert_optab): New optab.
25528 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
25529 (duplicate_and_interleave): Likewise.
25530 * tree-vect-loop.c: Include internal-fn.h.
25531 (neutral_op_for_slp_reduction): New function, split out from
25532 get_initial_defs_for_reduction.
25533 (get_initial_def_for_reduction): Handle option 2 for variable-length
25534 vectors by loading the neutral value into a vector and then shifting
25535 the initial value into element 0.
25536 (get_initial_defs_for_reduction): Replace the code argument with
25537 the neutral value calculated by neutral_op_for_slp_reduction.
25538 Use gimple_build_vector for constant-length vectors.
25539 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
25540 but the first group_size elements have a neutral value.
25541 Use duplicate_and_interleave otherwise.
25542 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
25543 Update call to get_initial_defs_for_reduction. Handle SLP
25544 reductions for variable-length vectors by creating one vector
25545 result for each scalar result, with the elements associated
25546 with other scalar results stubbed out with the neutral value.
25547 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
25548 Require IFN_VEC_SHL_INSERT for double reductions on
25549 variable-length vectors, or SLP reductions that have
25550 a neutral value. Require can_duplicate_and_interleave_p
25551 support for variable-length unchained SLP reductions if there
25552 is no neutral value, such as for MIN/MAX reductions. Also require
25553 the number of vector elements to be a multiple of the number of
25554 SLP statements when doing variable-length unchained SLP reductions.
25555 Update call to vect_create_epilog_for_reduction.
25556 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
25557 and remove initial values.
25558 (duplicate_and_interleave): Make public.
25559 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
25560 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
25562 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25563 Alan Hayward <alan.hayward@arm.com>
25564 David Sherwood <david.sherwood@arm.com>
25566 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
25567 (can_duplicate_and_interleave_p): New function.
25568 (vect_get_and_check_slp_defs): Take the vector of statements
25569 rather than just the current one. Remove excess parentheses.
25570 Restriction rejectinon of vect_constant_def and vect_external_def
25571 for variable-length vectors to boolean types, or types for which
25572 can_duplicate_and_interleave_p is false.
25573 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
25574 (duplicate_and_interleave): New function.
25575 (vect_get_constant_vectors): Use gimple_build_vector for
25576 constant-length vectors and suitable variable-length constant
25577 vectors. Use duplicate_and_interleave for other variable-length
25578 vectors. Don't defer the update when inserting new statements.
25580 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25581 Alan Hayward <alan.hayward@arm.com>
25582 David Sherwood <david.sherwood@arm.com>
25584 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
25585 min_profitable_iters doesn't go negative.
25587 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25588 Alan Hayward <alan.hayward@arm.com>
25589 David Sherwood <david.sherwood@arm.com>
25591 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
25592 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
25593 * optabs.def (vec_mask_load_lanes_optab): New optab.
25594 (vec_mask_store_lanes_optab): Likewise.
25595 * internal-fn.def (MASK_LOAD_LANES): New internal function.
25596 (MASK_STORE_LANES): Likewise.
25597 * internal-fn.c (mask_load_lanes_direct): New macro.
25598 (mask_store_lanes_direct): Likewise.
25599 (expand_mask_load_optab_fn): Handle masked operations.
25600 (expand_mask_load_lanes_optab_fn): New macro.
25601 (expand_mask_store_optab_fn): Handle masked operations.
25602 (expand_mask_store_lanes_optab_fn): New macro.
25603 (direct_mask_load_lanes_optab_supported_p): Likewise.
25604 (direct_mask_store_lanes_optab_supported_p): Likewise.
25605 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
25607 (vect_load_lanes_supported): Likewise.
25608 * tree-vect-data-refs.c (strip_conversion): New function.
25609 (can_group_stmts_p): Likewise.
25610 (vect_analyze_data_ref_accesses): Use it instead of checking
25611 for a pair of assignments.
25612 (vect_store_lanes_supported): Take a masked_p parameter.
25613 (vect_load_lanes_supported): Likewise.
25614 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
25615 vect_store_lanes_supported and vect_load_lanes_supported.
25616 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
25617 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
25618 parameter. Don't allow gaps for masked accesses.
25619 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
25620 and vect_load_lanes_supported.
25621 (get_load_store_type): Take a masked_p parameter and update
25622 call to get_group_load_store_type.
25623 (vectorizable_store): Update call to get_load_store_type.
25624 Handle IFN_MASK_STORE_LANES.
25625 (vectorizable_load): Update call to get_load_store_type.
25626 Handle IFN_MASK_LOAD_LANES.
25628 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25629 Alan Hayward <alan.hayward@arm.com>
25630 David Sherwood <david.sherwood@arm.com>
25632 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
25634 * config/aarch64/aarch64-protos.h
25635 (aarch64_sve_struct_memory_operand_p): Declare.
25636 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
25637 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
25638 (VPRED, vpred): Handle SVE structure modes.
25639 * config/aarch64/constraints.md (Utx): New constraint.
25640 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
25641 (aarch64_sve_struct_nonimmediate_operand): New predicates.
25642 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
25643 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
25644 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
25645 structure modes. Split into pieces after RA.
25646 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
25647 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
25649 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
25650 SVE structure modes.
25651 (aarch64_classify_address): Likewise.
25652 (sizetochar): Move earlier in file.
25653 (aarch64_print_operand): Handle SVE register lists.
25654 (aarch64_array_mode): New function.
25655 (aarch64_sve_struct_memory_operand_p): Likewise.
25656 (TARGET_ARRAY_MODE): Redefine.
25658 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25659 Alan Hayward <alan.hayward@arm.com>
25660 David Sherwood <david.sherwood@arm.com>
25662 * target.def (array_mode): New target hook.
25663 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
25664 * doc/tm.texi: Regenerate.
25665 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
25666 * hooks.c (hook_optmode_mode_uhwi_none): New function.
25667 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
25668 targetm.array_mode.
25669 * stor-layout.c (mode_for_array): Likewise. Support polynomial
25672 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25673 Alan Hayward <alan.hayward@arm.com>
25674 David Sherwood <david.sherwood@arm.com>
25676 * fold-const.c (fold_binary_loc): Check the argument types
25677 rather than the result type when testing for a vector operation.
25679 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25681 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
25682 * doc/tm.texi: Regenerate.
25684 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25685 Alan Hayward <alan.hayward@arm.com>
25686 David Sherwood <david.sherwood@arm.com>
25688 * doc/invoke.texi (-msve-vector-bits=): Document new option.
25689 (sve): Document new AArch64 extension.
25690 * doc/md.texi (w): Extend the description of the AArch64
25691 constraint to include SVE vectors.
25692 (Upl, Upa): Document new AArch64 predicate constraints.
25693 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
25695 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
25696 (msve-vector-bits=): New option.
25697 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
25698 SVE when these are disabled.
25699 (sve): New extension.
25700 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
25701 modes. Adjust their number of units based on aarch64_sve_vg.
25702 (MAX_BITSIZE_MODE_ANY_MODE): Define.
25703 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
25704 aarch64_addr_query_type.
25705 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
25706 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
25707 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
25708 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
25709 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
25710 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
25711 (aarch64_simd_imm_zero_p): Delete.
25712 (aarch64_check_zero_based_sve_index_immediate): Declare.
25713 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
25714 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
25715 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
25716 (aarch64_sve_float_mul_immediate_p): Likewise.
25717 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
25718 rather than an rtx.
25719 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
25720 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
25721 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
25722 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
25723 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
25724 (aarch64_regmode_natural_size): Likewise.
25725 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
25726 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
25728 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
25729 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
25730 for VG and the SVE predicate registers.
25731 (V_ALIASES): Add a "z"-prefixed alias.
25732 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
25733 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
25734 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
25735 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
25736 (REG_CLASS_NAMES): Add entries for them.
25737 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
25738 and the predicate registers.
25739 (aarch64_sve_vg): Declare.
25740 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
25741 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
25742 (REGMODE_NATURAL_SIZE): Define.
25743 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
25745 * config/aarch64/aarch64.c: Include cfgrtl.h.
25746 (simd_immediate_info): Add a constructor for series vectors,
25747 and an associated step field.
25748 (aarch64_sve_vg): New variable.
25749 (aarch64_dbx_register_number): Handle VG and the predicate registers.
25750 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
25751 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
25752 (VEC_ANY_DATA, VEC_STRUCT): New constants.
25753 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
25754 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
25755 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
25756 (aarch64_get_mask_mode): New functions.
25757 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
25758 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
25759 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
25760 predicate modes and predicate registers. Explicitly restrict
25761 GPRs to modes of 16 bytes or smaller. Only allow FP registers
25762 to store a vector mode if it is recognized by
25763 aarch64_classify_vector_mode.
25764 (aarch64_regmode_natural_size): New function.
25765 (aarch64_hard_regno_caller_save_mode): Return the original mode
25767 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
25768 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
25769 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
25770 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
25772 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
25773 does not overlap dest if the function is frame-related. Handle
25775 (aarch64_split_add_offset): New function.
25776 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
25777 them aarch64_add_offset.
25778 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
25779 and update call to aarch64_sub_sp.
25780 (aarch64_add_cfa_expression): New function.
25781 (aarch64_expand_prologue): Pass extra temporary registers to the
25782 functions above. Handle the case in which we need to emit new
25783 DW_CFA_expressions for registers that were originally saved
25784 relative to the stack pointer, but now have to be expressed
25785 relative to the frame pointer.
25786 (aarch64_output_mi_thunk): Pass extra temporary registers to the
25788 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
25789 IP0 and IP1 values for SVE frames.
25790 (aarch64_expand_vec_series): New function.
25791 (aarch64_expand_sve_widened_duplicate): Likewise.
25792 (aarch64_expand_sve_const_vector): Likewise.
25793 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
25794 Handle SVE constants. Use emit_move_insn to move a force_const_mem
25795 into the register, rather than emitting a SET directly.
25796 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
25797 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
25798 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
25799 (offset_9bit_signed_scaled_p): New functions.
25800 (aarch64_replicate_bitmask_imm): New function.
25801 (aarch64_bitmask_imm): Use it.
25802 (aarch64_cannot_force_const_mem): Reject expressions involving
25803 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
25804 (aarch64_classify_index): Handle SVE indices, by requiring
25805 a plain register index with a scale that matches the element size.
25806 (aarch64_classify_address): Handle SVE addresses. Assert that
25807 the mode of the address is VOIDmode or an integer mode.
25808 Update call to aarch64_classify_symbol.
25809 (aarch64_classify_symbolic_expression): Update call to
25810 aarch64_classify_symbol.
25811 (aarch64_const_vec_all_in_range_p): New function.
25812 (aarch64_print_vector_float_operand): Likewise.
25813 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
25814 "vN" for FP registers with SVE modes. Handle (const ...) vectors
25815 and the FP immediates 1.0 and 0.5.
25816 (aarch64_print_address_internal): Handle SVE addresses.
25817 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
25818 (aarch64_regno_regclass): Handle predicate registers.
25819 (aarch64_secondary_reload): Handle big-endian reloads of SVE
25821 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
25822 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
25823 (aarch64_convert_sve_vector_bits): New function.
25824 (aarch64_override_options): Use it to handle -msve-vector-bits=.
25825 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
25826 rather than an rtx.
25827 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
25828 Handle SVE vector and predicate modes. Accept VL-based constants
25829 that need only one temporary register, and VL offsets that require
25830 no temporary registers.
25831 (aarch64_conditional_register_usage): Mark the predicate registers
25832 as fixed if SVE isn't available.
25833 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
25834 Return true for SVE vector and predicate modes.
25835 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
25836 rather than an unsigned int. Handle SVE modes.
25837 (aarch64_preferred_simd_mode): Update call accordingly. Handle
25839 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
25841 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
25842 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
25843 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
25844 (aarch64_sve_float_mul_immediate_p): New functions.
25845 (aarch64_sve_valid_immediate): New function.
25846 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
25847 Explicitly reject structure modes. Check for INDEX constants.
25848 Handle PTRUE and PFALSE constants.
25849 (aarch64_check_zero_based_sve_index_immediate): New function.
25850 (aarch64_simd_imm_zero_p): Delete.
25851 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
25852 vector modes. Accept constants in the range of CNT[BHWD].
25853 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
25854 ask for an Advanced SIMD mode.
25855 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
25856 (aarch64_simd_vector_alignment): Handle SVE predicates.
25857 (aarch64_vectorize_preferred_vector_alignment): New function.
25858 (aarch64_simd_vector_alignment_reachable): Use it instead of
25860 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
25861 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
25863 (MAX_VECT_LEN): Delete.
25864 (expand_vec_perm_d): Add a vec_flags field.
25865 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
25866 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
25867 (aarch64_evpc_ext): Don't apply a big-endian lane correction
25869 (aarch64_evpc_rev): Rename to...
25870 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
25871 (aarch64_evpc_rev_global): New function.
25872 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
25873 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
25875 (aarch64_evpc_sve_tbl): New function.
25876 (aarch64_expand_vec_perm_const_1): Update after rename of
25877 aarch64_evpc_rev. Handle SVE permutes too, trying
25878 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
25879 than aarch64_evpc_tbl.
25880 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
25881 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
25882 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
25883 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
25884 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
25885 (aarch64_expand_sve_vcond): New functions.
25886 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
25887 of aarch64_vector_mode_p.
25888 (aarch64_dwarf_poly_indeterminate_value): New function.
25889 (aarch64_compute_pressure_classes): Likewise.
25890 (aarch64_can_change_mode_class): Likewise.
25891 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
25892 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
25893 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
25894 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
25895 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
25896 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
25897 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
25898 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
25900 (Dn, Dl, Dr): Accept const as well as const_vector.
25901 (Dz): Likewise. Compare against CONST0_RTX.
25902 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
25903 of "vector" where appropriate.
25904 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
25905 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
25906 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
25907 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
25908 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
25909 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
25910 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
25911 (v_int_equiv): Extend to SVE modes.
25912 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
25914 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
25915 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
25916 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
25917 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
25918 (SVE_COND_FP_CMP): New int iterators.
25919 (perm_hilo): Handle the new unpack unspecs.
25920 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
25922 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
25923 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
25924 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
25925 (aarch64_equality_operator, aarch64_constant_vector_operand)
25926 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
25927 (aarch64_sve_nonimmediate_operand): Likewise.
25928 (aarch64_sve_general_operand): Likewise.
25929 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
25930 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
25931 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
25932 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
25933 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
25934 (aarch64_sve_float_arith_immediate): Likewise.
25935 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
25936 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
25937 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
25938 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
25939 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
25940 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
25941 (aarch64_sve_float_arith_operand): Likewise.
25942 (aarch64_sve_float_arith_with_sub_operand): Likewise.
25943 (aarch64_sve_float_mul_operand): Likewise.
25944 (aarch64_sve_vec_perm_operand): Likewise.
25945 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
25946 (aarch64_mov_operand): Accept const_poly_int and const_vector.
25947 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
25948 as well as const_vector.
25949 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
25950 in file. Use CONST0_RTX and CONSTM1_RTX.
25951 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
25952 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
25953 Use aarch64_simd_imm_zero.
25954 * config/aarch64/aarch64-sve.md: New file.
25955 * config/aarch64/aarch64.md: Include it.
25956 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
25957 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
25958 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
25959 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
25960 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
25961 (sve): New attribute.
25962 (enabled): Disable instructions with the sve attribute unless
25964 (movqi, movhi): Pass CONST_POLY_INT operaneds through
25965 aarch64_expand_mov_immediate.
25966 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
25967 CNT[BHSD] immediates.
25968 (movti): Split CONST_POLY_INT moves into two halves.
25969 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
25970 Split additions that need a temporary here if the destination
25971 is the stack pointer.
25972 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
25973 (*add<mode>3_poly_1): New instruction.
25974 (set_clobber_cc): New expander.
25976 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25978 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
25979 parameter and use it instead of GET_MODE_SIZE (innermode). Use
25980 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
25981 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
25982 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
25983 Change innermode from fixed_mode_size to machine_mode.
25984 (simplify_subreg): Update call accordingly. Handle a constant-sized
25985 subreg of a variable-length CONST_VECTOR.
25987 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25988 Alan Hayward <alan.hayward@arm.com>
25989 David Sherwood <david.sherwood@arm.com>
25991 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
25992 (add_offset_to_base): New function, split out from...
25993 (create_mem_ref): ...here. When handling a scale other than 1,
25994 check first whether the address is valid without the offset.
25995 Add it into the base if so, leaving the index and scale as-is.
25997 2018-01-12 Jakub Jelinek <jakub@redhat.com>
26000 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
26001 fold_for_warn before checking if arg2 is INTEGER_CST.
26003 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
26005 * config/rs6000/predicates.md (load_multiple_operation): Delete.
26006 (store_multiple_operation): Delete.
26007 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
26008 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
26009 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
26010 guarded by TARGET_STRING.
26011 (rs6000_output_load_multiple): Delete.
26012 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
26013 OPTION_MASK_STRING / TARGET_STRING handling.
26014 (print_operand) <'N', 'O'>: Add comment that these are unused now.
26015 (const rs6000_opt_masks) <"string">: Change mask to 0.
26016 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
26017 (MASK_STRING): Delete.
26018 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
26020 (load_multiple): Delete.
26027 (store_multiple): Delete.
26034 (movmemsi_8reg): Delete.
26035 (corresponding unnamed define_insn): Delete.
26036 (movmemsi_6reg): Delete.
26037 (corresponding unnamed define_insn): Delete.
26038 (movmemsi_4reg): Delete.
26039 (corresponding unnamed define_insn): Delete.
26040 (movmemsi_2reg): Delete.
26041 (corresponding unnamed define_insn): Delete.
26042 (movmemsi_1reg): Delete.
26043 (corresponding unnamed define_insn): Delete.
26044 * config/rs6000/rs6000.opt (mno-string): New.
26045 (mstring): Replace by deprecation warning stub.
26046 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
26048 2018-01-12 Jakub Jelinek <jakub@redhat.com>
26050 * regrename.c (regrename_do_replace): If replacing the same
26051 reg multiple times, try to reuse last created gen_raw_REG.
26054 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
26055 main to workaround a bug in GDB.
26057 2018-01-12 Tom de Vries <tom@codesourcery.com>
26060 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
26062 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
26064 PR rtl-optimization/80481
26065 * ira-color.c (get_cap_member): New function.
26066 (allocnos_conflict_by_live_ranges_p): Use it.
26067 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
26068 (setup_slot_coalesced_allocno_live_ranges): Ditto.
26070 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
26073 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
26074 (*saddl_se_1): Ditto.
26075 (*ssubsi_1): Ditto.
26076 (*ssubl_se_1): Ditto.
26078 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
26080 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
26081 rather than wi::to_widest for DR_INITs.
26082 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
26083 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
26084 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
26086 (vect_analyze_group_access_1): Note that here.
26088 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
26090 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
26091 polynomial type sizes.
26093 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
26095 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
26096 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
26097 (gimple_add_tmp_var): Likewise.
26099 2018-01-12 Martin Liska <mliska@suse.cz>
26101 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
26102 (gimple_alloc_sizes): Likewise.
26103 (dump_gimple_statistics): Use PRIu64 in printf format.
26104 * gimple.h: Change uint64_t to int.
26106 2018-01-12 Martin Liska <mliska@suse.cz>
26108 * tree-core.h: Use uint64_t instead of int.
26109 * tree.c (tree_node_counts): Likewise.
26110 (tree_node_sizes): Likewise.
26111 (dump_tree_statistics): Use PRIu64 in printf format.
26113 2018-01-12 Martin Liska <mliska@suse.cz>
26115 * Makefile.in: As qsort_chk is implemented in vec.c, add
26116 vec.o to linkage of gencfn-macros.
26117 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
26118 passing the info to record_node_allocation_statistics.
26119 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
26121 * ggc-common.c (struct ggc_usage): Add operator== and use
26122 it in operator< and compare function.
26123 * mem-stats.h (struct mem_usage): Likewise.
26124 * vec.c (struct vec_usage): Remove operator< and compare
26125 function. Can be simply inherited.
26127 2018-01-12 Martin Jambor <mjambor@suse.cz>
26130 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
26131 * tree-ssa-math-opts.c: Include domwalk.h.
26132 (convert_mult_to_fma_1): New function.
26133 (fma_transformation_info): New type.
26134 (fma_deferring_state): Likewise.
26135 (cancel_fma_deferring): New function.
26136 (result_of_phi): Likewise.
26137 (last_fma_candidate_feeds_initial_phi): Likewise.
26138 (convert_mult_to_fma): Added deferring logic, split actual
26139 transformation to convert_mult_to_fma_1.
26140 (math_opts_dom_walker): New type.
26141 (math_opts_dom_walker::after_dom_children): New method, body moved
26142 here from pass_optimize_widening_mul::execute, added deferring logic
26144 (pass_optimize_widening_mul::execute): Moved most of code to
26145 math_opts_dom_walker::after_dom_children.
26146 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
26147 * config/i386/i386.c (ix86_option_override_internal): Added
26148 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
26150 2018-01-12 Richard Biener <rguenther@suse.de>
26153 * dwarf2out.c (gen_variable_die): Do not reset old_die for
26154 inline instance vars.
26156 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
26159 * config/rx/rx.c (rx_is_restricted_memory_address):
26160 Handle SUBREG case.
26162 2018-01-12 Richard Biener <rguenther@suse.de>
26164 PR tree-optimization/80846
26165 * target.def (split_reduction): New target hook.
26166 * targhooks.c (default_split_reduction): New function.
26167 * targhooks.h (default_split_reduction): Declare.
26168 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
26169 target requests first reduce vectors by combining low and high
26171 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
26172 (get_vectype_for_scalar_type_and_size): Export.
26173 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
26174 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
26175 * doc/tm.texi: Regenerate.
26176 * config/i386/i386.c (ix86_split_reduction): Implement
26177 TARGET_VECTORIZE_SPLIT_REDUCTION.
26179 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
26182 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
26183 in PIC mode except for TARGET_VXWORKS_RTP.
26184 * config/sparc/sparc.c: Include cfgrtl.h.
26185 (TARGET_INIT_PIC_REG): Define.
26186 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
26187 (sparc_pic_register_p): New predicate.
26188 (sparc_legitimate_address_p): Use it.
26189 (sparc_legitimize_pic_address): Likewise.
26190 (sparc_delegitimize_address): Likewise.
26191 (sparc_mode_dependent_address_p): Likewise.
26192 (gen_load_pcrel_sym): Remove 4th parameter.
26193 (load_got_register): Adjust call to above. Remove obsolete stuff.
26194 (sparc_expand_prologue): Do not call load_got_register here.
26195 (sparc_flat_expand_prologue): Likewise.
26196 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
26197 (sparc_use_pseudo_pic_reg): New function.
26198 (sparc_init_pic_reg): Likewise.
26199 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
26200 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
26202 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
26204 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
26205 Add item for branch_cost.
26207 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
26209 PR rtl-optimization/83565
26210 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
26211 not extend the result to a larger mode for rotate operations.
26212 (num_sign_bit_copies1): Likewise.
26214 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26217 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
26219 Use values-Xc.o for -pedantic.
26220 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
26222 2018-01-12 Martin Liska <mliska@suse.cz>
26225 * ipa-devirt.c (final_warning_record::grow_type_warnings):
26227 (possible_polymorphic_call_targets): Use it.
26228 (ipa_devirt): Likewise.
26230 2018-01-12 Martin Liska <mliska@suse.cz>
26232 * profile-count.h (enum profile_quality): Use 0 as invalid
26233 enum value of profile_quality.
26235 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
26237 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
26238 -mext-string options.
26240 2018-01-12 Richard Biener <rguenther@suse.de>
26242 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
26243 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
26244 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
26246 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
26248 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
26250 * configure.ac (--with-long-double-format): Add support for the
26251 configuration option to change the default long double format on
26253 * config.gcc (powerpc*-linux*-*): Likewise.
26254 * configure: Regenerate.
26255 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
26256 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
26257 used without modification.
26259 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26261 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
26262 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
26263 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
26264 MISC_BUILTIN_SPEC_BARRIER.
26265 (rs6000_init_builtins): Likewise.
26266 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
26268 (speculation_barrier): New define_insn.
26269 * doc/extend.texi: Document __builtin_speculation_barrier.
26271 2018-01-11 Jakub Jelinek <jakub@redhat.com>
26274 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
26275 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
26276 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
26278 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
26279 integral modes instead of "ss" and "sd".
26280 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
26281 vectors with 32-bit and 64-bit elements.
26282 (vecdupssescalarmodesuffix): New mode attribute.
26283 (vec_dup<mode>): Use it.
26285 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
26288 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
26289 frame if argument is passed on stack.
26291 2018-01-11 Jakub Jelinek <jakub@redhat.com>
26294 * ree.c (combine_reaching_defs): Optimize also
26295 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
26296 reg2=any_extend(exp); reg1=reg2;, formatting fix.
26298 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
26300 PR middle-end/83189
26301 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
26303 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
26305 PR middle-end/83718
26306 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
26307 after they are computed.
26309 2018-01-11 Bin Cheng <bin.cheng@arm.com>
26311 PR tree-optimization/83695
26312 * gimple-loop-linterchange.cc
26313 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
26314 reset cached scev information after interchange.
26315 (pass_linterchange::execute): Remove call to scev_reset_htab.
26317 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26319 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
26320 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
26321 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
26322 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
26323 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
26324 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
26325 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
26326 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
26327 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
26328 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
26329 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
26330 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
26331 (V_lane_reg): Likewise.
26332 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
26334 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
26335 (vfmal_lane_low<mode>_intrinsic,
26336 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
26337 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
26338 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
26339 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
26340 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
26341 vfmsl_lane_high<mode>_intrinsic): New define_insns.
26343 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26345 * config/arm/arm-cpus.in (fp16fml): New feature.
26346 (ALL_SIMD): Add fp16fml.
26347 (armv8.2-a): Add fp16fml as an option.
26348 (armv8.3-a): Likewise.
26349 (armv8.4-a): Add fp16fml as part of fp16.
26350 * config/arm/arm.h (TARGET_FP16FML): Define.
26351 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
26353 * config/arm/arm-modes.def (V2HF): Define.
26354 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
26355 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
26356 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
26357 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
26358 vfmsl_low, vfmsl_high): New set of builtins.
26359 * config/arm/iterators.md (PLUSMINUS): New code iterator.
26360 (vfml_op): New code attribute.
26361 (VFMLHALVES): New int iterator.
26362 (VFML, VFMLSEL): New mode attributes.
26363 (V_reg): Define mapping for V2HF.
26364 (V_hi, V_lo): New mode attributes.
26365 (VF_constraint): Likewise.
26366 (vfml_half, vfml_half_selector): New int attributes.
26367 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
26369 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
26370 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
26372 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
26373 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
26374 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
26375 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
26377 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
26378 Document new effective target and option set.
26380 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26382 * config/arm/arm-cpus.in (armv8_4): New feature.
26383 (ARMv8_4a): New fgroup.
26384 (armv8.4-a): New arch.
26385 * config/arm/arm-tables.opt: Regenerate.
26386 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
26387 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
26388 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
26389 Add matching rules for -march=armv8.4-a and extensions.
26390 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
26392 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
26395 * config/rx/rx.md (BW): New mode attribute.
26396 (sync_lock_test_and_setsi): Add mode suffix to insn output.
26398 2018-01-11 Richard Biener <rguenther@suse.de>
26400 PR tree-optimization/83435
26401 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
26402 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
26403 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
26405 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
26406 Alan Hayward <alan.hayward@arm.com>
26407 David Sherwood <david.sherwood@arm.com>
26409 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
26411 (aarch64_classify_address): Initialize it. Track polynomial offsets.
26412 (aarch64_print_address_internal): Use it to check for a zero offset.
26414 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
26415 Alan Hayward <alan.hayward@arm.com>
26416 David Sherwood <david.sherwood@arm.com>
26418 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
26419 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
26420 Return a poly_int64 rather than a HOST_WIDE_INT.
26421 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
26422 rather than a HOST_WIDE_INT.
26423 * config/aarch64/aarch64.h (aarch64_frame): Protect with
26424 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
26425 hard_fp_offset, frame_size, initial_adjust, callee_offset and
26426 final_offset from HOST_WIDE_INT to poly_int64.
26427 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
26428 to_constant when getting the number of units in an Advanced SIMD
26430 (aarch64_builtin_vectorized_function): Check for a constant number
26432 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
26434 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
26435 attribute instead of GET_MODE_NUNITS.
26436 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
26437 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
26438 GET_MODE_SIZE for fixed-size registers.
26439 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
26440 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
26441 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
26442 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
26443 (aarch64_print_operand, aarch64_print_address_internal)
26444 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
26445 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
26446 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
26447 Handle polynomial GET_MODE_SIZE.
26448 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
26449 wider than SImode without modification.
26450 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
26451 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
26452 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
26453 passing and returning SVE modes.
26454 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
26455 rather than GEN_INT.
26456 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
26457 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
26458 (aarch64_allocate_and_probe_stack_space): Likewise.
26459 (aarch64_layout_frame): Cope with polynomial offsets.
26460 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
26461 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
26462 polynomial offsets.
26463 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
26464 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
26465 poly_int64 rather than a HOST_WIDE_INT.
26466 (aarch64_get_separate_components, aarch64_process_components)
26467 (aarch64_expand_prologue, aarch64_expand_epilogue)
26468 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
26469 (aarch64_anchor_offset): New function, split out from...
26470 (aarch64_legitimize_address): ...here.
26471 (aarch64_builtin_vectorization_cost): Handle polynomial
26472 TYPE_VECTOR_SUBPARTS.
26473 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
26475 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
26476 number of elements from the PARALLEL rather than the mode.
26477 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
26478 rather than GET_MODE_BITSIZE.
26479 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
26480 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
26481 (aarch64_expand_vec_perm_const_1): Handle polynomial
26482 d->perm.length () and d->perm elements.
26483 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
26484 Apply to_constant to d->perm elements.
26485 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
26486 polynomial CONST_VECTOR_NUNITS.
26487 (aarch64_move_pointer): Take amount as a poly_int64 rather
26489 (aarch64_progress_pointer): Avoid temporary variable.
26490 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
26491 the mode attribute instead of GET_MODE.
26493 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
26494 Alan Hayward <alan.hayward@arm.com>
26495 David Sherwood <david.sherwood@arm.com>
26497 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
26498 x exists before using it.
26499 (aarch64_add_constant_internal): Rename to...
26500 (aarch64_add_offset_1): ...this. Replace regnum with separate
26501 src and dest rtxes. Handle the case in which they're different,
26502 including when the offset is zero. Replace scratchreg with an rtx.
26503 Use 2 additions if there is no spare register into which we can
26504 move a 16-bit constant.
26505 (aarch64_add_constant): Delete.
26506 (aarch64_add_offset): Replace reg with separate src and dest
26507 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
26508 Use aarch64_add_offset_1.
26509 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
26510 an rtx rather than an int. Take the delta as a poly_int64
26511 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
26512 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
26513 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
26514 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
26515 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
26516 and aarch64_add_sp.
26517 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
26518 aarch64_add_constant.
26520 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
26522 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
26523 Use scalar_float_mode.
26525 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
26527 * config/aarch64/aarch64-simd.md
26528 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
26529 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
26530 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
26531 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
26532 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
26533 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
26534 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
26535 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
26536 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
26537 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
26539 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26542 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
26543 targ_options->x_arm_arch_string is non NULL.
26545 2018-01-11 Tamar Christina <tamar.christina@arm.com>
26547 * config/aarch64/aarch64.h
26548 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
26550 2018-01-11 Sudakshina Das <sudi.das@arm.com>
26553 * expmed.c (emit_store_flag_force): Swap if const op0
26554 and change VOIDmode to mode of op0.
26556 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
26558 PR rtl-optimization/83761
26559 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
26560 than bytes to mode_for_size.
26562 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
26564 PR middle-end/83189
26565 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
26566 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
26569 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
26571 PR middle-end/83575
26572 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
26573 when in layout mode.
26574 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
26575 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
26578 2018-01-10 Michael Collison <michael.collison@arm.com>
26580 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
26581 * config/aarch64/aarch64-option-extension.def: Add
26582 AARCH64_OPT_EXTENSION of 'fp16fml'.
26583 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
26584 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
26585 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
26586 * config/aarch64/constraints.md (Ui7): New constraint.
26587 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
26588 (VFMLA_SEL_W): Ditto.
26591 (VFMLA16_LOW): New int iterator.
26592 (VFMLA16_HIGH): Ditto.
26593 (UNSPEC_FMLAL): New unspec.
26594 (UNSPEC_FMLSL): Ditto.
26595 (UNSPEC_FMLAL2): Ditto.
26596 (UNSPEC_FMLSL2): Ditto.
26597 (f16mac): New code attribute.
26598 * config/aarch64/aarch64-simd-builtins.def
26599 (aarch64_fmlal_lowv2sf): Ditto.
26600 (aarch64_fmlsl_lowv2sf): Ditto.
26601 (aarch64_fmlalq_lowv4sf): Ditto.
26602 (aarch64_fmlslq_lowv4sf): Ditto.
26603 (aarch64_fmlal_highv2sf): Ditto.
26604 (aarch64_fmlsl_highv2sf): Ditto.
26605 (aarch64_fmlalq_highv4sf): Ditto.
26606 (aarch64_fmlslq_highv4sf): Ditto.
26607 (aarch64_fmlal_lane_lowv2sf): Ditto.
26608 (aarch64_fmlsl_lane_lowv2sf): Ditto.
26609 (aarch64_fmlal_laneq_lowv2sf): Ditto.
26610 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
26611 (aarch64_fmlalq_lane_lowv4sf): Ditto.
26612 (aarch64_fmlsl_lane_lowv4sf): Ditto.
26613 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
26614 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
26615 (aarch64_fmlal_lane_highv2sf): Ditto.
26616 (aarch64_fmlsl_lane_highv2sf): Ditto.
26617 (aarch64_fmlal_laneq_highv2sf): Ditto.
26618 (aarch64_fmlsl_laneq_highv2sf): Ditto.
26619 (aarch64_fmlalq_lane_highv4sf): Ditto.
26620 (aarch64_fmlsl_lane_highv4sf): Ditto.
26621 (aarch64_fmlalq_laneq_highv4sf): Ditto.
26622 (aarch64_fmlsl_laneq_highv4sf): Ditto.
26623 * config/aarch64/aarch64-simd.md:
26624 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
26625 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
26626 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
26627 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
26628 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
26629 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
26630 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
26631 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
26632 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
26633 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
26634 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
26635 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
26636 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
26637 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
26638 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
26639 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
26640 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
26641 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
26642 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
26643 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
26644 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
26645 (vfmlsl_low_u32): Ditto.
26646 (vfmlalq_low_u32): Ditto.
26647 (vfmlslq_low_u32): Ditto.
26648 (vfmlal_high_u32): Ditto.
26649 (vfmlsl_high_u32): Ditto.
26650 (vfmlalq_high_u32): Ditto.
26651 (vfmlslq_high_u32): Ditto.
26652 (vfmlal_lane_low_u32): Ditto.
26653 (vfmlsl_lane_low_u32): Ditto.
26654 (vfmlal_laneq_low_u32): Ditto.
26655 (vfmlsl_laneq_low_u32): Ditto.
26656 (vfmlalq_lane_low_u32): Ditto.
26657 (vfmlslq_lane_low_u32): Ditto.
26658 (vfmlalq_laneq_low_u32): Ditto.
26659 (vfmlslq_laneq_low_u32): Ditto.
26660 (vfmlal_lane_high_u32): Ditto.
26661 (vfmlsl_lane_high_u32): Ditto.
26662 (vfmlal_laneq_high_u32): Ditto.
26663 (vfmlsl_laneq_high_u32): Ditto.
26664 (vfmlalq_lane_high_u32): Ditto.
26665 (vfmlslq_lane_high_u32): Ditto.
26666 (vfmlalq_laneq_high_u32): Ditto.
26667 (vfmlslq_laneq_high_u32): Ditto.
26668 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
26669 (AARCH64_FL_FOR_ARCH8_4): New.
26670 (AARCH64_ISA_F16FML): New ISA flag.
26671 (TARGET_F16FML): New feature flag for fp16fml.
26672 (doc/invoke.texi): Document new fp16fml option.
26674 2018-01-10 Michael Collison <michael.collison@arm.com>
26676 * config/aarch64/aarch64-builtins.c:
26677 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
26678 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
26679 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
26680 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
26681 (AARCH64_ISA_SHA3): New ISA flag.
26682 (TARGET_SHA3): New feature flag for sha3.
26683 * config/aarch64/iterators.md (sha512_op): New int attribute.
26684 (CRYPTO_SHA512): New int iterator.
26685 (UNSPEC_SHA512H): New unspec.
26686 (UNSPEC_SHA512H2): Ditto.
26687 (UNSPEC_SHA512SU0): Ditto.
26688 (UNSPEC_SHA512SU1): Ditto.
26689 * config/aarch64/aarch64-simd-builtins.def
26690 (aarch64_crypto_sha512hqv2di): New builtin.
26691 (aarch64_crypto_sha512h2qv2di): Ditto.
26692 (aarch64_crypto_sha512su0qv2di): Ditto.
26693 (aarch64_crypto_sha512su1qv2di): Ditto.
26694 (aarch64_eor3qv8hi): Ditto.
26695 (aarch64_rax1qv2di): Ditto.
26696 (aarch64_xarqv2di): Ditto.
26697 (aarch64_bcaxqv8hi): Ditto.
26698 * config/aarch64/aarch64-simd.md:
26699 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
26700 (aarch64_crypto_sha512su0qv2di): Ditto.
26701 (aarch64_crypto_sha512su1qv2di): Ditto.
26702 (aarch64_eor3qv8hi): Ditto.
26703 (aarch64_rax1qv2di): Ditto.
26704 (aarch64_xarqv2di): Ditto.
26705 (aarch64_bcaxqv8hi): Ditto.
26706 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
26707 (vsha512h2q_u64): Ditto.
26708 (vsha512su0q_u64): Ditto.
26709 (vsha512su1q_u64): Ditto.
26710 (veor3q_u16): Ditto.
26711 (vrax1q_u64): Ditto.
26712 (vxarq_u64): Ditto.
26713 (vbcaxq_u16): Ditto.
26714 * config/arm/types.md (crypto_sha512): New type attribute.
26715 (crypto_sha3): Ditto.
26716 (doc/invoke.texi): Document new sha3 option.
26718 2018-01-10 Michael Collison <michael.collison@arm.com>
26720 * config/aarch64/aarch64-builtins.c:
26721 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
26722 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
26723 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
26724 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
26725 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
26726 (AARCH64_ISA_SM4): New ISA flag.
26727 (TARGET_SM4): New feature flag for sm4.
26728 * config/aarch64/aarch64-simd-builtins.def
26729 (aarch64_sm3ss1qv4si): Ditto.
26730 (aarch64_sm3tt1aq4si): Ditto.
26731 (aarch64_sm3tt1bq4si): Ditto.
26732 (aarch64_sm3tt2aq4si): Ditto.
26733 (aarch64_sm3tt2bq4si): Ditto.
26734 (aarch64_sm3partw1qv4si): Ditto.
26735 (aarch64_sm3partw2qv4si): Ditto.
26736 (aarch64_sm4eqv4si): Ditto.
26737 (aarch64_sm4ekeyqv4si): Ditto.
26738 * config/aarch64/aarch64-simd.md:
26739 (aarch64_sm3ss1qv4si): Ditto.
26740 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
26741 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
26742 (aarch64_sm4eqv4si): Ditto.
26743 (aarch64_sm4ekeyqv4si): Ditto.
26744 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
26745 (sm3part_op): Ditto.
26746 (CRYPTO_SM3TT): Ditto.
26747 (CRYPTO_SM3PART): Ditto.
26748 (UNSPEC_SM3SS1): New unspec.
26749 (UNSPEC_SM3TT1A): Ditto.
26750 (UNSPEC_SM3TT1B): Ditto.
26751 (UNSPEC_SM3TT2A): Ditto.
26752 (UNSPEC_SM3TT2B): Ditto.
26753 (UNSPEC_SM3PARTW1): Ditto.
26754 (UNSPEC_SM3PARTW2): Ditto.
26755 (UNSPEC_SM4E): Ditto.
26756 (UNSPEC_SM4EKEY): Ditto.
26757 * config/aarch64/constraints.md (Ui2): New constraint.
26758 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
26759 * config/arm/types.md (crypto_sm3): New type attribute.
26760 (crypto_sm4): Ditto.
26761 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
26762 (vsm3tt1aq_u32): Ditto.
26763 (vsm3tt1bq_u32): Ditto.
26764 (vsm3tt2aq_u32): Ditto.
26765 (vsm3tt2bq_u32): Ditto.
26766 (vsm3partw1q_u32): Ditto.
26767 (vsm3partw2q_u32): Ditto.
26768 (vsm4eq_u32): Ditto.
26769 (vsm4ekeyq_u32): Ditto.
26770 (doc/invoke.texi): Document new sm4 option.
26772 2018-01-10 Michael Collison <michael.collison@arm.com>
26774 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
26775 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
26776 (AARCH64_FL_FOR_ARCH8_4): New.
26777 (AARCH64_FL_V8_4): New flag.
26778 (doc/invoke.texi): Document new armv8.4-a option.
26780 2018-01-10 Michael Collison <michael.collison@arm.com>
26782 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
26783 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
26784 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
26785 * config/aarch64/aarch64-option-extension.def: Add
26786 AARCH64_OPT_EXTENSION of 'sha2'.
26787 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
26788 (crypto): Disable sha2 and aes if crypto disabled.
26789 (crypto): Enable aes and sha2 if enabled.
26790 (simd): Disable sha2 and aes if simd disabled.
26791 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
26793 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
26794 (TARGET_SHA2): New feature flag for sha2.
26795 (TARGET_AES): New feature flag for aes.
26796 * config/aarch64/aarch64-simd.md:
26797 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
26798 conditional on TARGET_AES.
26799 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
26800 (aarch64_crypto_sha1hsi): Make pattern conditional
26802 (aarch64_crypto_sha1hv4si): Ditto.
26803 (aarch64_be_crypto_sha1hv4si): Ditto.
26804 (aarch64_crypto_sha1su1v4si): Ditto.
26805 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
26806 (aarch64_crypto_sha1su0v4si): Ditto.
26807 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
26808 (aarch64_crypto_sha256su0v4si): Ditto.
26809 (aarch64_crypto_sha256su1v4si): Ditto.
26810 (doc/invoke.texi): Document new aes and sha2 options.
26812 2018-01-10 Martin Sebor <msebor@redhat.com>
26814 PR tree-optimization/83781
26815 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
26818 2018-01-11 Martin Sebor <msebor@gmail.com>
26819 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26821 PR tree-optimization/83501
26822 PR tree-optimization/81703
26824 * tree-ssa-strlen.c (get_string_cst): Rename...
26825 (get_string_len): ...to this. Handle global constants.
26826 (handle_char_store): Adjust.
26828 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
26829 Jim Wilson <jimw@sifive.com>
26831 * config/riscv/riscv-protos.h (riscv_output_return): New.
26832 * config/riscv/riscv.c (struct machine_function): New naked_p field.
26833 (riscv_attribute_table, riscv_output_return),
26834 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
26835 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
26836 (riscv_compute_frame_info): Only compute frame->mask if not a naked
26838 (riscv_expand_prologue): Add early return for naked function.
26839 (riscv_expand_epilogue): Likewise.
26840 (riscv_function_ok_for_sibcall): Return false for naked function.
26841 (riscv_set_current_function): New.
26842 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
26843 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
26844 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
26845 * doc/extend.texi (RISC-V Function Attributes): New.
26847 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
26849 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
26850 check for 128-bit long double before checking TCmode.
26851 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
26852 128-bit long doubles before checking TFmode or TCmode.
26853 (FLOAT128_IBM_P): Likewise.
26855 2018-01-10 Martin Sebor <msebor@redhat.com>
26857 PR tree-optimization/83671
26858 * builtins.c (c_strlen): Unconditionally return zero for the empty
26860 Use -Warray-bounds for warnings.
26861 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
26862 for non-constant array indices with COMPONENT_REF, arrays of
26863 arrays, and pointers to arrays.
26864 (gimple_fold_builtin_strlen): Determine and set length range for
26865 non-constant character arrays.
26867 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
26869 PR middle-end/81897
26870 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
26873 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
26875 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
26877 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
26880 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
26881 VECTOR_MEM_ALTIVEC_OR_VSX_P.
26882 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
26883 indexed_or_indirect_operand predicate.
26884 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
26885 (*vsx_le_perm_load_v8hi): Likewise.
26886 (*vsx_le_perm_load_v16qi): Likewise.
26887 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
26888 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
26889 (*vsx_le_perm_store_v8hi): Likewise.
26890 (*vsx_le_perm_store_v16qi): Likewise.
26891 (eight unnamed splitters): Likewise.
26893 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
26895 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
26896 * config/rs6000/emmintrin.h: Likewise.
26897 * config/rs6000/mmintrin.h: Likewise.
26898 * config/rs6000/xmmintrin.h: Likewise.
26900 2018-01-10 David Malcolm <dmalcolm@redhat.com>
26903 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
26905 * tree.c (tree_nop_conversion): Return true for location wrapper
26907 (maybe_wrap_with_location): New function.
26908 (selftest::check_strip_nops): New function.
26909 (selftest::test_location_wrappers): New function.
26910 (selftest::tree_c_tests): Call it.
26911 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
26912 (maybe_wrap_with_location): New decl.
26913 (EXPR_LOCATION_WRAPPER_P): New macro.
26914 (location_wrapper_p): New inline function.
26915 (tree_strip_any_location_wrapper): New inline function.
26917 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
26920 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
26921 stack_realign_offset for the largest alignment of stack slot
26923 (ix86_find_max_used_stack_alignment): New function.
26924 (ix86_finalize_stack_frame_flags): Use it. Set
26925 max_used_stack_alignment if we don't realign stack.
26926 * config/i386/i386.h (machine_function): Add
26927 max_used_stack_alignment.
26929 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
26931 * config/arm/arm.opt (-mbranch-cost): New option.
26932 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
26935 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
26938 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
26939 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
26941 2018-01-10 Richard Biener <rguenther@suse.de>
26944 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
26945 early out so it also covers the case where we have a non-NULL
26948 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
26950 PR tree-optimization/83753
26951 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
26952 for non-strided grouped accesses if the number of elements is 1.
26954 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
26957 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
26958 * i386.h (TARGET_USE_GATHER): Define.
26959 * x86-tune.def (X86_TUNE_USE_GATHER): New.
26961 2018-01-10 Martin Liska <mliska@suse.cz>
26964 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
26965 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
26967 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
26968 CLEANUP_NO_PARTITIONING is not set.
26970 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
26972 * doc/rtl.texi: Remove documentation of (const ...) wrappers
26973 for vectors, as a partial revert of r254296.
26974 * rtl.h (const_vec_p): Delete.
26975 (const_vec_duplicate_p): Don't test for vector CONSTs.
26976 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
26977 * expmed.c (make_tree): Likewise.
26980 * common.md (E, F): Use CONSTANT_P instead of checking for
26982 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
26983 checking for CONST_VECTOR.
26985 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
26987 PR middle-end/83575
26988 * predict.c (force_edge_cold): Handle in more sane way edges
26989 with no prediction.
26991 2018-01-09 Carl Love <cel@us.ibm.com>
26993 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
26995 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
26996 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
26997 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
26998 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
26999 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
27000 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
27001 * config/rs6000/rs6000-protos.h: Add extern defition for
27002 rs6000_generate_float2_double_code.
27003 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
27005 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
27006 (float2_v2df): Add define_expand.
27008 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
27011 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
27012 op_mode in the force_to_mode call.
27014 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
27016 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
27017 instead of checking each element individually.
27018 (aarch64_evpc_uzp): Likewise.
27019 (aarch64_evpc_zip): Likewise.
27020 (aarch64_evpc_ext): Likewise.
27021 (aarch64_evpc_rev): Likewise.
27022 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
27023 instead of checking each element individually. Return true without
27025 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
27026 whether all selected elements come from the same input, instead of
27027 checking each element individually. Remove calls to gen_rtx_REG,
27028 start_sequence and end_sequence and instead assert that no rtl is
27031 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
27033 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
27034 order of HIGH and CONST checks.
27036 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
27038 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
27039 if the destination isn't an SSA_NAME.
27041 2018-01-09 Richard Biener <rguenther@suse.de>
27043 PR tree-optimization/83668
27044 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
27046 (canonicalize_loop_form): ... here, renamed from ...
27047 (canonicalize_loop_closed_ssa_form): ... this and amended to
27048 swap successor edges for loop exit blocks to make us use
27049 the RPO order we need for initial schedule generation.
27051 2018-01-09 Joseph Myers <joseph@codesourcery.com>
27053 PR tree-optimization/64811
27054 * match.pd: When optimizing comparisons with Inf, avoid
27055 introducing or losing exceptions from comparisons with NaN.
27057 2018-01-09 Martin Liska <mliska@suse.cz>
27060 * asan.c (shadow_mem_size): Add gcc_assert.
27062 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
27064 Don't save registers in main().
27067 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
27068 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
27069 * config/avr/avr.c (avr_set_current_function): Don't error if
27070 naked, OS_task or OS_main are specified at the same time.
27071 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
27073 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
27075 * common/config/avr/avr-common.c (avr_option_optimization_table):
27076 Switch on -mmain-is-OS_task for optimizing compilations.
27078 2018-01-09 Richard Biener <rguenther@suse.de>
27080 PR tree-optimization/83572
27081 * graphite.c: Include cfganal.h.
27082 (graphite_transform_loops): Connect infinite loops to exit
27083 and remove fake edges at the end.
27085 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
27087 * ipa-inline.c (edge_badness): Revert accidental checkin.
27089 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
27092 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
27093 symbols; not inline clones.
27095 2018-01-09 Jakub Jelinek <jakub@redhat.com>
27098 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
27099 hard registers. Formatting fixes.
27101 PR preprocessor/83722
27102 * gcc.c (try_generate_repro): Pass
27103 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
27104 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
27107 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
27108 Kito Cheng <kito.cheng@gmail.com>
27110 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
27111 (riscv_leaf_function_p): Delete.
27112 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
27114 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27116 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
27118 (do_ifelse): New function.
27119 (do_isel): New function.
27120 (do_sub3): New function.
27121 (do_add3): New function.
27122 (do_load_mask_compare): New function.
27123 (do_overlap_load_compare): New function.
27124 (expand_compare_loop): New function.
27125 (expand_block_compare): Call expand_compare_loop() when appropriate.
27126 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
27127 option description.
27128 (-mblock-compare-inline-loop-limit): New option.
27130 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27133 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
27134 Reverse order of second and third operands in first alternative.
27135 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
27136 of first and second elements in UNSPEC_VPERMR vector.
27137 (altivec_expand_vec_perm_le): Likewise.
27139 2018-01-08 Jeff Law <law@redhat.com>
27141 PR rtl-optimizatin/81308
27142 * tree-switch-conversion.c (cfg_altered): New file scoped static.
27143 (process_switch): If group_case_labels makes a change, then set
27145 (pass_convert_switch::execute): If a switch is converted, then
27146 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
27148 PR rtl-optimization/81308
27149 * recog.c (split_all_insns): Conditionally cleanup the CFG after
27152 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
27154 PR target/83663 - Revert r255946
27155 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
27156 generation for cases where splatting a value is not useful.
27157 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
27158 across a vec_duplicate and a paradoxical subreg forming a vector
27159 mode to a vec_concat.
27161 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27163 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
27164 -march=armv8.3-a variants.
27165 * config/arm/t-multilib: Likewise.
27166 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
27168 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27170 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
27172 (cceq_ior_compare_complement): Give it a name so I can use it, and
27173 change boolean_or_operator predicate to boolean_operator so it can
27174 be used to generate a crand.
27175 (eqne): New code iterator.
27176 (bd/bd_neg): New code_attrs.
27177 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
27178 a single define_insn.
27179 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
27180 decrement (bdnzt/bdnzf/bdzt/bdzf).
27181 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
27182 with the new names of the branch decrement patterns, and added the
27183 names of the branch decrement conditional patterns.
27185 2018-01-08 Richard Biener <rguenther@suse.de>
27187 PR tree-optimization/83563
27188 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
27191 2018-01-08 Richard Biener <rguenther@suse.de>
27193 PR middle-end/83713
27194 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
27196 2018-01-08 Richard Biener <rguenther@suse.de>
27198 PR tree-optimization/83685
27199 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
27200 references to abnormals.
27202 2018-01-08 Richard Biener <rguenther@suse.de>
27205 * dwarf2out.c (output_indirect_strings): Handle empty
27206 skeleton_debug_str_hash.
27207 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
27209 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
27211 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
27212 (emit_store_direct): Likewise.
27213 (arc_trampoline_adjust_address): Likewise.
27214 (arc_asm_trampoline_template): New function.
27215 (arc_initialize_trampoline): Use asm_trampoline_template.
27216 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
27217 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
27218 * config/arc/arc.md (flush_icache): Delete pattern.
27220 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
27222 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
27223 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
27226 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
27229 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
27230 by not USED_FOR_TARGET.
27231 (make_pass_resolve_sw_modes): Likewise.
27233 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
27235 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
27238 2018-01-08 Richard Biener <rguenther@suse.de>
27240 PR middle-end/83580
27241 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
27243 2018-01-08 Richard Biener <rguenther@suse.de>
27245 PR middle-end/83517
27246 * match.pd ((t * 2) / 2) -> t): Add missing :c.
27248 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
27250 PR middle-end/81897
27251 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
27252 basic blocks with a small number of successors.
27253 (convert_control_dep_chain_into_preds): Improve handling of
27255 (dump_predicates): Split apart into...
27256 (dump_pred_chain): ...here...
27257 (dump_pred_info): ...and here.
27258 (can_one_predicate_be_invalidated_p): Add debugging printfs.
27259 (can_chain_union_be_invalidated_p): Improve check for invalidation
27261 (uninit_uses_cannot_happen): Avoid unnecessary if
27262 convert_control_dep_chain_into_preds yielded nothing.
27264 2018-01-06 Martin Sebor <msebor@redhat.com>
27266 PR tree-optimization/83640
27267 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
27268 subtracting negative offset from size.
27269 (builtin_access::overlap): Adjust offset bounds of the access to fall
27270 within the size of the object if possible.
27272 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
27274 PR rtl-optimization/83699
27275 * expmed.c (extract_bit_field_1): Restrict the vector usage of
27276 extract_bit_field_as_subreg to cases in which the extracted
27277 value is also a vector.
27279 * lra-constraints.c (process_alt_operands): Test for the equivalence
27280 substitutions when detecting a possible reload cycle.
27282 2018-01-06 Jakub Jelinek <jakub@redhat.com>
27285 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
27286 by default if flag_selective_schedling{,2}. Formatting fixes.
27288 PR rtl-optimization/83682
27289 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
27290 if it has non-VECTOR_MODE element mode.
27291 (vec_duplicate_p): Likewise.
27293 PR middle-end/83694
27294 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
27295 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
27297 2018-01-05 Jakub Jelinek <jakub@redhat.com>
27300 * config/i386/i386-builtin.def
27301 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
27302 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
27303 Require also OPTION_MASK_ISA_AVX512F in addition to
27304 OPTION_MASK_ISA_GFNI.
27305 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
27306 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
27307 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
27308 to OPTION_MASK_ISA_GFNI.
27309 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
27310 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
27311 OPTION_MASK_ISA_AVX512BW.
27312 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
27313 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
27314 addition to OPTION_MASK_ISA_GFNI.
27315 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
27316 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
27317 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
27318 to OPTION_MASK_ISA_GFNI.
27319 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
27320 a requirement for all ISAs rather than any of them with a few
27322 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
27324 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
27325 bitmasks to be enabled with 3 exceptions, instead of requiring any
27326 enabled ISA with lots of exceptions.
27327 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
27328 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
27329 Change avx512bw in isa attribute to avx512f.
27330 * config/i386/sgxintrin.h: Add license boilerplate.
27331 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
27332 to __AVX512F__ and __AVX512VL to __AVX512VL__.
27333 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
27334 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
27336 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
27337 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
27338 temporarily sse2 rather than sse if not enabled already.
27341 * config/i386/sse.md (VI248_VLBW): Rename to ...
27342 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
27343 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
27344 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
27345 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
27346 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
27347 mode iterator instead of VI248_VLBW.
27349 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
27351 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
27352 (record_modified): Skip clobbers; add debug output.
27353 (param_change_prob): Use sreal frequencies.
27355 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
27357 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
27358 punt for user-aligned variables.
27360 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
27362 * tree-chrec.c (chrec_contains_symbols): Return true for
27365 2018-01-05 Sudakshina Das <sudi.das@arm.com>
27368 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
27369 of (x|y) == x for BICS pattern.
27371 2018-01-05 Jakub Jelinek <jakub@redhat.com>
27373 PR tree-optimization/83605
27374 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
27375 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
27378 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
27380 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
27381 * config/epiphany/rtems.h: New file.
27383 2018-01-04 Jakub Jelinek <jakub@redhat.com>
27384 Uros Bizjak <ubizjak@gmail.com>
27387 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
27388 QIreg_operand instead of register_operand predicate.
27389 * config/i386/i386.c (ix86_rop_should_change_byte_p,
27390 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
27391 comments instead of -fmitigate[-_]rop.
27393 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27396 * cgraphunit.c (symbol_table::compile): Switch to text_section
27397 before calling assembly_start debug hook.
27398 * run-rtl-passes.c (run_rtl_passes): Likewise.
27401 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
27403 * tree-vrp.c (extract_range_from_binary_expr_1): Check
27404 range_int_cst_p rather than !symbolic_range_p before calling
27405 extract_range_from_multiplicative_op_1.
27407 2018-01-04 Jeff Law <law@redhat.com>
27409 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
27410 redundant test in assertion.
27412 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
27414 * doc/rtl.texi: Document machine_mode wrapper classes.
27416 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
27418 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
27419 using tree_to_uhwi.
27421 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
27423 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
27424 the VEC_PERM_EXPR fold to fail.
27426 2018-01-04 Jakub Jelinek <jakub@redhat.com>
27429 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
27430 to switched_sections.
27432 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
27435 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
27436 test for d.testing.
27438 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
27441 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
27442 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
27444 2018-01-04 Jakub Jelinek <jakub@redhat.com>
27447 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
27448 is BLKmode and bitpos not zero or mode change is needed.
27450 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
27453 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
27456 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
27459 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
27460 instead of MULT rtx. Update all corresponding splitters.
27461 (*saddl_se): Ditto.
27462 (*ssub<modesuffix>): Ditto.
27463 (*ssubl_se): Ditto.
27464 (*cmp_sadd_di): Update split patterns.
27465 (*cmp_sadd_si): Ditto.
27466 (*cmp_sadd_sidi): Ditto.
27467 (*cmp_ssub_di): Ditto.
27468 (*cmp_ssub_si): Ditto.
27469 (*cmp_ssub_sidi): Ditto.
27470 * config/alpha/predicates.md (const23_operand): New predicate.
27471 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
27472 Look for ASHIFT, not MULT inner operand.
27473 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
27475 2018-01-04 Martin Liska <mliska@suse.cz>
27477 PR gcov-profile/83669
27478 * gcov.c (output_intermediate_file): Add version to intermediate
27480 * doc/gcov.texi: Document new field 'version' in intermediate
27481 file format. Fix location of '-k' option of gcov command.
27483 2018-01-04 Martin Liska <mliska@suse.cz>
27486 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
27488 2018-01-04 Jakub Jelinek <jakub@redhat.com>
27490 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
27492 2018-01-03 Martin Sebor <msebor@redhat.com>
27494 PR tree-optimization/83655
27495 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
27496 checking calls with invalid arguments.
27498 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27500 * tree-vect-stmts.c (vect_get_store_rhs): New function.
27501 (vectorizable_mask_load_store): Delete.
27502 (vectorizable_call): Return false for masked loads and stores.
27503 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
27504 instead of gimple_assign_rhs1.
27505 (vectorizable_load): Handle IFN_MASK_LOAD.
27506 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
27508 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27510 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
27512 (vectorizable_mask_load_store): ...here.
27513 (vectorizable_load): ...and here.
27515 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27517 * tree-vect-stmts.c (vect_build_all_ones_mask)
27518 (vect_build_zero_merge_argument): New functions, split out from...
27519 (vectorizable_load): ...here.
27521 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27523 * tree-vect-stmts.c (vect_check_store_rhs): New function,
27525 (vectorizable_mask_load_store): ...here.
27526 (vectorizable_store): ...and here.
27528 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27530 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
27532 (vectorizable_mask_load_store): ...here.
27534 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27536 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
27537 (vect_model_store_cost): Take a vec_load_store_type instead of a
27539 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
27540 (vect_model_store_cost): Take a vec_load_store_type instead of a
27542 (vectorizable_mask_load_store): Update accordingly.
27543 (vectorizable_store): Likewise.
27544 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
27546 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27548 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
27549 IFN_MASK_LOAD calls here rather than...
27550 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
27552 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27553 Alan Hayward <alan.hayward@arm.com>
27554 David Sherwood <david.sherwood@arm.com>
27556 * expmed.c (extract_bit_field_1): For vector extracts,
27557 fall back to extract_bit_field_as_subreg if vec_extract
27560 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27561 Alan Hayward <alan.hayward@arm.com>
27562 David Sherwood <david.sherwood@arm.com>
27564 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
27565 they are variable or constant sized.
27566 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
27567 slots for constant-sized data.
27569 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27570 Alan Hayward <alan.hayward@arm.com>
27571 David Sherwood <david.sherwood@arm.com>
27573 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
27574 handling COND_EXPRs with boolean comparisons, try to find a better
27575 basis for the mask type than the boolean itself.
27577 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27579 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
27580 is calculated and how it can be overridden.
27581 * genmodes.c (max_bitsize_mode_any_mode): New variable.
27582 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
27584 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
27587 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27588 Alan Hayward <alan.hayward@arm.com>
27589 David Sherwood <david.sherwood@arm.com>
27591 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
27592 Remove the mode argument.
27593 (aarch64_simd_valid_immediate): Remove the mode and inverse
27595 * config/aarch64/iterators.md (bitsize): New iterator.
27596 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
27597 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
27598 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
27599 aarch64_simd_valid_immediate.
27600 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
27601 (aarch64_reg_or_bic_imm): Likewise.
27602 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
27603 with an insn_type enum and msl with a modifier_type enum.
27604 Replace element_width with a scalar_mode. Change the shift
27605 to unsigned int. Add constructors for scalar_float_mode and
27606 scalar_int_mode elements.
27607 (aarch64_vect_float_const_representable_p): Delete.
27608 (aarch64_can_const_movi_rtx_p)
27609 (aarch64_simd_scalar_immediate_valid_for_move)
27610 (aarch64_simd_make_constant): Update call to
27611 aarch64_simd_valid_immediate.
27612 (aarch64_advsimd_valid_immediate_hs): New function.
27613 (aarch64_advsimd_valid_immediate): Likewise.
27614 (aarch64_simd_valid_immediate): Remove mode and inverse
27615 arguments. Rewrite to use the above. Use const_vec_duplicate_p
27616 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
27617 and aarch64_float_const_representable_p on the result.
27618 (aarch64_output_simd_mov_immediate): Remove mode argument.
27619 Update call to aarch64_simd_valid_immediate and use of
27620 simd_immediate_info.
27621 (aarch64_output_scalar_simd_mov_immediate): Update call
27624 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27625 Alan Hayward <alan.hayward@arm.com>
27626 David Sherwood <david.sherwood@arm.com>
27628 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
27629 (mode_nunits): Likewise CONST_MODE_NUNITS.
27630 * machmode.def (ADJUST_NUNITS): Document.
27631 * genmodes.c (mode_data::need_nunits_adj): New field.
27632 (blank_mode): Update accordingly.
27633 (adj_nunits): New variable.
27634 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
27636 (emit_mode_size_inline): Set need_bytesize_adj for all modes
27637 listed in adj_nunits.
27638 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
27639 listed in adj_nunits. Don't emit case statements for such modes.
27640 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
27641 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
27642 nothing if adj_nunits is nonnull.
27643 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
27644 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
27645 (emit_mode_fbit): Update use of print_maybe_const_decl.
27646 (emit_move_size): Likewise. Treat the array as non-const
27648 (emit_mode_adjustments): Handle adj_nunits.
27650 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27652 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
27653 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
27654 (VECTOR_MODES): Use it.
27655 (make_vector_modes): Take the prefix as an argument.
27657 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27658 Alan Hayward <alan.hayward@arm.com>
27659 David Sherwood <david.sherwood@arm.com>
27661 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
27662 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
27663 for MODE_VECTOR_BOOL.
27664 * machmode.def (VECTOR_BOOL_MODE): Document.
27665 * genmodes.c (VECTOR_BOOL_MODE): New macro.
27666 (make_vector_bool_mode): New function.
27667 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
27669 * lto-streamer-in.c (lto_input_mode_table): Likewise.
27670 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
27672 * stor-layout.c (int_mode_for_mode): Likewise.
27673 * tree.c (build_vector_type_for_mode): Likewise.
27674 * varasm.c (output_constant_pool_2): Likewise.
27675 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
27676 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
27677 for MODE_VECTOR_BOOL.
27678 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
27679 of mode class checks.
27680 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
27681 instead of a list of mode class checks.
27682 (expand_vector_scalar_condition): Likewise.
27683 (type_for_widest_vector_mode): Handle BImode as an inner mode.
27685 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27686 Alan Hayward <alan.hayward@arm.com>
27687 David Sherwood <david.sherwood@arm.com>
27689 * machmode.h (mode_size): Change from unsigned short to
27691 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
27692 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
27693 or if measurement_type is not polynomial.
27694 (fixed_size_mode::includes_p): Check for constant-sized modes.
27695 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
27696 return a poly_uint16 rather than an unsigned short.
27697 (emit_mode_size): Change the type of mode_size from unsigned short
27698 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
27699 (emit_mode_adjustments): Cope with polynomial vector sizes.
27700 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
27702 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
27704 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
27705 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
27706 * caller-save.c (setup_save_areas): Likewise.
27707 (replace_reg_with_saved_mem): Likewise.
27708 * calls.c (emit_library_call_value_1): Likewise.
27709 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
27710 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
27711 (gen_lowpart_for_combine): Likewise.
27712 * convert.c (convert_to_integer_1): Likewise.
27713 * cse.c (equiv_constant, cse_insn): Likewise.
27714 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
27715 (cselib_subst_to_values): Likewise.
27716 * dce.c (word_dce_process_block): Likewise.
27717 * df-problems.c (df_word_lr_mark_ref): Likewise.
27718 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
27719 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
27720 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
27721 (rtl_for_decl_location): Likewise.
27722 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
27723 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
27724 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
27725 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
27726 (expand_expr_real_1): Likewise.
27727 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
27728 (pad_below): Likewise.
27729 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
27730 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
27731 * ira.c (get_subreg_tracking_sizes): Likewise.
27732 * ira-build.c (ira_create_allocno_objects): Likewise.
27733 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
27734 (ira_sort_regnos_for_alter_reg): Likewise.
27735 * ira-costs.c (record_operand_costs): Likewise.
27736 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
27737 (resolve_simple_move): Likewise.
27738 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
27739 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
27740 (lra_constraints): Likewise.
27741 (CONST_POOL_OK_P): Reject variable-sized modes.
27742 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
27743 (add_pseudo_to_slot, lra_spill): Likewise.
27744 * omp-low.c (omp_clause_aligned_alignment): Likewise.
27745 * optabs-query.c (get_best_extraction_insn): Likewise.
27746 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
27747 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
27748 (expand_mult_highpart, valid_multiword_target_p): Likewise.
27749 * recog.c (offsettable_address_addr_space_p): Likewise.
27750 * regcprop.c (maybe_mode_change): Likewise.
27751 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
27752 * regrename.c (build_def_use): Likewise.
27753 * regstat.c (dump_reg_info): Likewise.
27754 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
27755 (find_reloads, find_reloads_subreg_address): Likewise.
27756 * reload1.c (eliminate_regs_1): Likewise.
27757 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
27758 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
27759 (simplify_binary_operation_1, simplify_subreg): Likewise.
27760 * targhooks.c (default_function_arg_padding): Likewise.
27761 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
27762 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
27763 (verify_gimple_assign_ternary): Likewise.
27764 * tree-inline.c (estimate_move_cost): Likewise.
27765 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
27766 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
27767 (get_address_cost_ainc): Likewise.
27768 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
27769 (vect_supportable_dr_alignment): Likewise.
27770 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
27771 (vectorizable_reduction): Likewise.
27772 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
27773 (vectorizable_operation, vectorizable_load): Likewise.
27774 * tree.c (build_same_sized_truth_vector_type): Likewise.
27775 * valtrack.c (cleanup_auto_inc_dec): Likewise.
27776 * var-tracking.c (emit_note_insn_var_location): Likewise.
27777 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
27778 (ADDR_VEC_ALIGN): Likewise.
27780 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27781 Alan Hayward <alan.hayward@arm.com>
27782 David Sherwood <david.sherwood@arm.com>
27784 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
27786 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
27787 or if measurement_type is polynomial.
27788 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
27789 * combine.c (make_extraction): Likewise.
27790 * dse.c (find_shift_sequence): Likewise.
27791 * dwarf2out.c (mem_loc_descriptor): Likewise.
27792 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
27793 (extract_bit_field, extract_low_bits): Likewise.
27794 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
27795 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
27796 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
27797 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
27798 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
27799 * reload.c (find_reloads): Likewise.
27800 * reload1.c (alter_reg): Likewise.
27801 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
27802 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
27803 * tree-if-conv.c (predicate_mem_writes): Likewise.
27804 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
27805 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
27806 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
27807 * valtrack.c (dead_debug_insert_temp): Likewise.
27808 * varasm.c (mergeable_constant_section): Likewise.
27809 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
27811 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27812 Alan Hayward <alan.hayward@arm.com>
27813 David Sherwood <david.sherwood@arm.com>
27815 * expr.c (expand_assignment): Cope with polynomial mode sizes
27816 when assigning to a CONCAT.
27818 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27819 Alan Hayward <alan.hayward@arm.com>
27820 David Sherwood <david.sherwood@arm.com>
27822 * machmode.h (mode_precision): Change from unsigned short to
27824 (mode_to_precision): Return a poly_uint16 rather than an unsigned
27826 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
27827 or if measurement_type is not polynomial.
27828 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
27829 in which the mode is already known to be a scalar_int_mode.
27830 * genmodes.c (emit_mode_precision): Change the type of mode_precision
27831 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
27833 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
27834 for GET_MODE_PRECISION.
27835 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
27836 for GET_MODE_PRECISION.
27837 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
27839 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
27840 (expand_field_assignment, make_extraction): Likewise.
27841 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
27842 (get_last_value): Likewise.
27843 * convert.c (convert_to_integer_1): Likewise.
27844 * cse.c (cse_insn): Likewise.
27845 * expr.c (expand_expr_real_1): Likewise.
27846 * lra-constraints.c (simplify_operand_subreg): Likewise.
27847 * optabs-query.c (can_atomic_load_p): Likewise.
27848 * optabs.c (expand_atomic_load): Likewise.
27849 (expand_atomic_store): Likewise.
27850 * ree.c (combine_reaching_defs): Likewise.
27851 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
27852 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
27853 * tree.h (type_has_mode_precision_p): Likewise.
27854 * ubsan.c (instrument_si_overflow): Likewise.
27856 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27857 Alan Hayward <alan.hayward@arm.com>
27858 David Sherwood <david.sherwood@arm.com>
27860 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
27861 polynomial numbers of units.
27862 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
27863 (valid_vector_subparts_p): New function.
27864 (build_vector_type): Remove temporary shim and take the number
27865 of units as a poly_uint64 rather than an int.
27866 (build_opaque_vector_type): Take the number of units as a
27867 poly_uint64 rather than an int.
27868 * tree.c (build_vector_from_ctor): Handle polynomial
27869 TYPE_VECTOR_SUBPARTS.
27870 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
27871 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
27872 (build_vector_from_val): If the number of units is variable,
27873 use build_vec_duplicate_cst for constant operands and
27874 VEC_DUPLICATE_EXPR otherwise.
27875 (make_vector_type): Remove temporary is_constant ().
27876 (build_vector_type, build_opaque_vector_type): Take the number of
27877 units as a poly_uint64 rather than an int.
27878 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
27880 * cfgexpand.c (expand_debug_expr): Likewise.
27881 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
27882 (store_constructor, expand_expr_real_1): Likewise.
27883 (const_scalar_mask_from_tree): Likewise.
27884 * fold-const-call.c (fold_const_reduction): Likewise.
27885 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
27886 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
27887 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
27888 (fold_relational_const): Likewise.
27889 (native_interpret_vector): Likewise. Change the size from an
27890 int to an unsigned int.
27891 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
27892 TYPE_VECTOR_SUBPARTS.
27893 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
27894 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
27895 duplicating a non-constant operand into a variable-length vector.
27896 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
27897 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
27898 * ipa-icf.c (sem_variable::equals): Likewise.
27899 * match.pd: Likewise.
27900 * omp-simd-clone.c (simd_clone_subparts): Likewise.
27901 * print-tree.c (print_node): Likewise.
27902 * stor-layout.c (layout_type): Likewise.
27903 * targhooks.c (default_builtin_vectorization_cost): Likewise.
27904 * tree-cfg.c (verify_gimple_comparison): Likewise.
27905 (verify_gimple_assign_binary): Likewise.
27906 (verify_gimple_assign_ternary): Likewise.
27907 (verify_gimple_assign_single): Likewise.
27908 * tree-pretty-print.c (dump_generic_node): Likewise.
27909 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
27910 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
27911 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
27912 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
27913 (vect_shift_permute_load_chain): Likewise.
27914 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
27915 (expand_vector_condition, optimize_vector_constructor): Likewise.
27916 (lower_vec_perm, get_compute_type): Likewise.
27917 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
27918 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
27919 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
27920 (vect_recog_mask_conversion_pattern): Likewise.
27921 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
27922 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
27923 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
27924 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
27925 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
27926 (vectorizable_shift, vectorizable_operation, vectorizable_store)
27927 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
27928 (supportable_widening_operation): Likewise.
27929 (supportable_narrowing_operation): Likewise.
27930 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
27932 * varasm.c (output_constant): Likewise.
27934 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27935 Alan Hayward <alan.hayward@arm.com>
27936 David Sherwood <david.sherwood@arm.com>
27938 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
27939 so that both the length == 3 and length != 3 cases set up their
27940 own permute vectors. Add comments explaining why we know the
27941 number of elements is constant.
27942 (vect_permute_load_chain): Likewise.
27944 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27945 Alan Hayward <alan.hayward@arm.com>
27946 David Sherwood <david.sherwood@arm.com>
27948 * machmode.h (mode_nunits): Change from unsigned char to
27950 (ONLY_FIXED_SIZE_MODES): New macro.
27951 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
27952 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
27953 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
27955 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
27956 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
27957 or if measurement_type is not polynomial.
27958 * genmodes.c (ZERO_COEFFS): New macro.
27959 (emit_mode_nunits_inline): Make mode_nunits_inline return a
27961 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
27962 Use ZERO_COEFFS when emitting initializers.
27963 * data-streamer.h (bp_pack_poly_value): New function.
27964 (bp_unpack_poly_value): Likewise.
27965 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
27966 for GET_MODE_NUNITS.
27967 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
27968 for GET_MODE_NUNITS.
27969 * tree.c (make_vector_type): Remove temporary shim and make
27970 the real function take the number of units as a poly_uint64
27971 rather than an int.
27972 (build_vector_type_for_mode): Handle polynomial nunits.
27973 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
27974 * emit-rtl.c (const_vec_series_p_1): Likewise.
27975 (gen_rtx_CONST_VECTOR): Likewise.
27976 * fold-const.c (test_vec_duplicate_folding): Likewise.
27977 * genrecog.c (validate_pattern): Likewise.
27978 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
27979 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
27980 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
27981 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
27982 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
27983 * rtlanal.c (subreg_get_info): Likewise.
27984 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
27985 (vect_grouped_load_supported): Likewise.
27986 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
27987 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
27988 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
27989 (simplify_const_unary_operation, simplify_binary_operation_1)
27990 (simplify_const_binary_operation, simplify_ternary_operation)
27991 (test_vector_ops_duplicate, test_vector_ops): Likewise.
27992 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
27993 instead of CONST_VECTOR_NUNITS.
27994 * varasm.c (output_constant_pool_2): Likewise.
27995 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
27996 explicit-encoded elements in the XVEC for variable-length vectors.
27998 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28000 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
28002 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28003 Alan Hayward <alan.hayward@arm.com>
28004 David Sherwood <david.sherwood@arm.com>
28006 * coretypes.h (fixed_size_mode): Declare.
28007 (fixed_size_mode_pod): New typedef.
28008 * builtins.h (target_builtins::x_apply_args_mode)
28009 (target_builtins::x_apply_result_mode): Change type to
28010 fixed_size_mode_pod.
28011 * builtins.c (apply_args_size, apply_result_size, result_vector)
28012 (expand_builtin_apply_args_1, expand_builtin_apply)
28013 (expand_builtin_return): Update accordingly.
28015 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28017 * cse.c (hash_rtx_cb): Hash only the encoded elements.
28018 * cselib.c (cselib_hash_rtx): Likewise.
28019 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
28020 CONST_VECTOR encoding.
28022 2018-01-03 Jakub Jelinek <jakub@redhat.com>
28023 Jeff Law <law@redhat.com>
28026 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
28027 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
28028 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
28029 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
28032 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
28033 explicitly probe *sp in a noreturn function if there were any callee
28034 register saves or frame pointer is needed.
28036 2018-01-03 Jakub Jelinek <jakub@redhat.com>
28039 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
28040 BLKmode for ternary, binary or unary expressions.
28043 * var-tracking.c (delete_vta_debug_insn): New inline function.
28044 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
28045 insns from get_insns () to NULL instead of each bb separately.
28046 Use delete_vta_debug_insn. No longer static.
28047 (vt_debug_insns_local, variable_tracking_main_1): Adjust
28048 delete_vta_debug_insns callers.
28049 * rtl.h (delete_vta_debug_insns): Declare.
28050 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
28051 instead of variable_tracking_main.
28053 2018-01-03 Martin Sebor <msebor@redhat.com>
28055 PR tree-optimization/83603
28056 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
28057 arguments past the endof the argument list in functions declared
28058 without a prototype.
28059 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
28060 Avoid checking when arguments are null.
28062 2018-01-03 Martin Sebor <msebor@redhat.com>
28065 * doc/extend.texi (attribute const): Fix a typo.
28066 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
28067 issuing -Wsuggest-attribute for void functions.
28069 2018-01-03 Martin Sebor <msebor@redhat.com>
28071 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
28072 offset_int::from instead of wide_int::to_shwi.
28073 (maybe_diag_overlap): Remove assertion.
28074 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
28075 * gimple-ssa-sprintf.c (format_directive): Same.
28076 (parse_directive): Same.
28077 (sprintf_dom_walker::compute_format_length): Same.
28078 (try_substitute_return_value): Same.
28080 2018-01-03 Jeff Law <law@redhat.com>
28082 PR middle-end/83654
28083 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
28084 non-constant residual for zero at runtime and avoid probing in
28085 that case. Reorganize code for trailing problem to mirror handling
28088 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28090 PR tree-optimization/83501
28091 * tree-ssa-strlen.c (get_string_cst): New.
28092 (handle_char_store): Call get_string_cst.
28094 2018-01-03 Martin Liska <mliska@suse.cz>
28096 PR tree-optimization/83593
28097 * tree-ssa-strlen.c: Include tree-cfg.h.
28098 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
28099 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
28100 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
28102 (strlen_dom_walker::before_dom_children): Call
28103 gimple_purge_dead_eh_edges. Dump tranformation with details
28105 (strlen_dom_walker::before_dom_children): Update call by adding
28106 new argument cleanup_eh.
28107 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
28109 2018-01-03 Martin Liska <mliska@suse.cz>
28112 * cif-code.def (VARIADIC_THUNK): New enum value.
28113 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
28116 2018-01-03 Jan Beulich <jbeulich@suse.com>
28118 * sse.md (mov<mode>_internal): Tighten condition for when to use
28119 vmovdqu<ssescalarsize> for TI and OI modes.
28121 2018-01-03 Jakub Jelinek <jakub@redhat.com>
28123 Update copyright years.
28125 2018-01-03 Martin Liska <mliska@suse.cz>
28128 * ipa-visibility.c (function_and_variable_visibility): Skip
28129 functions with noipa attribure.
28131 2018-01-03 Jakub Jelinek <jakub@redhat.com>
28133 * gcc.c (process_command): Update copyright notice dates.
28134 * gcov-dump.c (print_version): Ditto.
28135 * gcov.c (print_version): Ditto.
28136 * gcov-tool.c (print_version): Ditto.
28137 * gengtype.c (create_file): Ditto.
28138 * doc/cpp.texi: Bump @copying's copyright year.
28139 * doc/cppinternals.texi: Ditto.
28140 * doc/gcc.texi: Ditto.
28141 * doc/gccint.texi: Ditto.
28142 * doc/gcov.texi: Ditto.
28143 * doc/install.texi: Ditto.
28144 * doc/invoke.texi: Ditto.
28146 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28148 * vector-builder.h (vector_builder::m_full_nelts): Change from
28149 unsigned int to poly_uint64.
28150 (vector_builder::full_nelts): Update prototype accordingly.
28151 (vector_builder::new_vector): Likewise.
28152 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
28153 (vector_builder::operator ==): Likewise.
28154 (vector_builder::finalize): Likewise.
28155 * int-vector-builder.h (int_vector_builder::int_vector_builder):
28156 Take the number of elements as a poly_uint64 rather than an
28158 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
28159 from unsigned int to poly_uint64.
28160 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
28161 (vec_perm_indices::new_vector): Likewise.
28162 (vec_perm_indices::length): Likewise.
28163 (vec_perm_indices::nelts_per_input): Likewise.
28164 (vec_perm_indices::input_nelts): Likewise.
28165 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
28166 number of elements per input as a poly_uint64 rather than an
28167 unsigned int. Use the original encoding for variable-length
28168 vectors, rather than clamping each individual element.
28169 For the second and subsequent elements in each pattern,
28170 clamp the step and base before clamping their sum.
28171 (vec_perm_indices::series_p): Handle polynomial element counts.
28172 (vec_perm_indices::all_in_range_p): Likewise.
28173 (vec_perm_indices_to_tree): Likewise.
28174 (vec_perm_indices_to_rtx): Likewise.
28175 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
28176 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
28177 (tree_vector_builder::new_binary_operation): Handle polynomial
28178 element counts. Return false if we need to know the number
28179 of elements at compile time.
28180 * fold-const.c (fold_vec_perm): Punt if the number of elements
28181 isn't known at compile time.
28183 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28185 * vec-perm-indices.h (vec_perm_builder): Change element type
28186 from HOST_WIDE_INT to poly_int64.
28187 (vec_perm_indices::element_type): Update accordingly.
28188 (vec_perm_indices::clamp): Handle polynomial element_types.
28189 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
28190 (vec_perm_indices::all_in_range_p): Likewise.
28191 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
28193 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
28194 polynomial vec_perm_indices element types.
28195 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
28196 * fold-const.c (fold_vec_perm): Likewise.
28197 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
28198 * tree-vect-generic.c (lower_vec_perm): Likewise.
28199 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
28200 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
28201 element type to HOST_WIDE_INT.
28203 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28204 Alan Hayward <alan.hayward@arm.com>
28205 David Sherwood <david.sherwood@arm.com>
28207 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
28208 rather than an int. Use plus_constant.
28209 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
28210 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
28212 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28213 Alan Hayward <alan.hayward@arm.com>
28214 David Sherwood <david.sherwood@arm.com>
28216 * calls.c (emit_call_1, expand_call): Change struct_value_size from
28217 a HOST_WIDE_INT to a poly_int64.
28219 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28220 Alan Hayward <alan.hayward@arm.com>
28221 David Sherwood <david.sherwood@arm.com>
28223 * calls.c (load_register_parameters): Cope with polynomial
28224 mode sizes. Require a constant size for BLKmode parameters
28225 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
28226 forces a parameter to be padded at the lsb end in order to
28227 fill a complete number of words, require the parameter size
28228 to be ordered wrt UNITS_PER_WORD.
28230 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28231 Alan Hayward <alan.hayward@arm.com>
28232 David Sherwood <david.sherwood@arm.com>
28234 * reload1.c (spill_stack_slot_width): Change element type
28235 from unsigned int to poly_uint64_pod.
28236 (alter_reg): Treat mode sizes as polynomial.
28238 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28239 Alan Hayward <alan.hayward@arm.com>
28240 David Sherwood <david.sherwood@arm.com>
28242 * reload.c (complex_word_subreg_p): New function.
28243 (reload_inner_reg_of_subreg, push_reload): Use it.
28245 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28246 Alan Hayward <alan.hayward@arm.com>
28247 David Sherwood <david.sherwood@arm.com>
28249 * lra-constraints.c (process_alt_operands): Reject matched
28250 operands whose sizes aren't ordered.
28251 (match_reload): Refer to this check here.
28253 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28254 Alan Hayward <alan.hayward@arm.com>
28255 David Sherwood <david.sherwood@arm.com>
28257 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
28258 that the mode size is in the set {1, 2, 4, 8, 16}.
28260 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28261 Alan Hayward <alan.hayward@arm.com>
28262 David Sherwood <david.sherwood@arm.com>
28264 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
28265 Use plus_constant instead of gen_rtx_PLUS.
28267 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28268 Alan Hayward <alan.hayward@arm.com>
28269 David Sherwood <david.sherwood@arm.com>
28271 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
28272 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
28273 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
28274 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
28275 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
28276 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
28277 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
28278 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
28279 * config/i386/i386.c (ix86_push_rounding): ...this new function.
28280 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
28282 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
28283 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
28284 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
28285 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
28286 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
28287 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
28288 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
28289 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
28290 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
28291 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
28293 * expr.c (emit_move_resolve_push): Treat the input and result
28294 of PUSH_ROUNDING as a poly_int64.
28295 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
28296 (emit_push_insn): Likewise.
28297 * lra-eliminations.c (mark_not_eliminable): Likewise.
28298 * recog.c (push_operand): Likewise.
28299 * reload1.c (elimination_effects): Likewise.
28300 * rtlanal.c (nonzero_bits1): Likewise.
28301 * calls.c (store_one_arg): Likewise. Require the padding to be
28302 known at compile time.
28304 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28305 Alan Hayward <alan.hayward@arm.com>
28306 David Sherwood <david.sherwood@arm.com>
28308 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
28309 Use plus_constant instead of gen_rtx_PLUS.
28311 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28312 Alan Hayward <alan.hayward@arm.com>
28313 David Sherwood <david.sherwood@arm.com>
28315 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
28316 rather than an int.
28318 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28319 Alan Hayward <alan.hayward@arm.com>
28320 David Sherwood <david.sherwood@arm.com>
28322 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
28323 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
28324 via stack temporaries. Treat the mode size as polynomial too.
28326 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28327 Alan Hayward <alan.hayward@arm.com>
28328 David Sherwood <david.sherwood@arm.com>
28330 * expr.c (expand_expr_real_2): When handling conversions involving
28331 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
28332 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
28333 as a poly_uint64 too.
28335 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28336 Alan Hayward <alan.hayward@arm.com>
28337 David Sherwood <david.sherwood@arm.com>
28339 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
28341 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28342 Alan Hayward <alan.hayward@arm.com>
28343 David Sherwood <david.sherwood@arm.com>
28345 * combine.c (can_change_dest_mode): Handle polynomial
28346 REGMODE_NATURAL_SIZE.
28347 * expmed.c (store_bit_field_1): Likewise.
28348 * expr.c (store_constructor): Likewise.
28349 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
28350 and polynomial REGMODE_NATURAL_SIZE.
28351 (gen_lowpart_common): Likewise.
28352 * reginfo.c (record_subregs_of_mode): Likewise.
28353 * rtlanal.c (read_modify_subreg_p): Likewise.
28355 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28356 Alan Hayward <alan.hayward@arm.com>
28357 David Sherwood <david.sherwood@arm.com>
28359 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
28360 numbers of elements.
28362 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28363 Alan Hayward <alan.hayward@arm.com>
28364 David Sherwood <david.sherwood@arm.com>
28366 * match.pd: Cope with polynomial numbers of vector elements.
28368 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28369 Alan Hayward <alan.hayward@arm.com>
28370 David Sherwood <david.sherwood@arm.com>
28372 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
28373 in a POINTER_PLUS_EXPR.
28375 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28376 Alan Hayward <alan.hayward@arm.com>
28377 David Sherwood <david.sherwood@arm.com>
28379 * omp-simd-clone.c (simd_clone_subparts): New function.
28380 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
28381 (ipa_simd_modify_function_body): Likewise.
28383 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28384 Alan Hayward <alan.hayward@arm.com>
28385 David Sherwood <david.sherwood@arm.com>
28387 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
28388 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
28389 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
28390 (expand_vector_condition, vector_element): Likewise.
28391 (subparts_gt): New function.
28392 (get_compute_type): Use subparts_gt.
28393 (count_type_subparts): Delete.
28394 (expand_vector_operations_1): Use subparts_gt instead of
28395 count_type_subparts.
28397 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28398 Alan Hayward <alan.hayward@arm.com>
28399 David Sherwood <david.sherwood@arm.com>
28401 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
28402 (vect_compile_time_alias): ...this new function. Do the calculation
28403 on poly_ints rather than trees.
28404 (vect_prune_runtime_alias_test_list): Update call accordingly.
28406 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28407 Alan Hayward <alan.hayward@arm.com>
28408 David Sherwood <david.sherwood@arm.com>
28410 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
28412 (vect_schedule_slp_instance): Likewise.
28414 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28415 Alan Hayward <alan.hayward@arm.com>
28416 David Sherwood <david.sherwood@arm.com>
28418 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
28419 constant and extern definitions for variable-length vectors.
28420 (vect_get_constant_vectors): Note that the number of units
28421 is known to be constant.
28423 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28424 Alan Hayward <alan.hayward@arm.com>
28425 David Sherwood <david.sherwood@arm.com>
28427 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
28428 of units as polynomial. Choose between WIDE and NARROW based
28431 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28432 Alan Hayward <alan.hayward@arm.com>
28433 David Sherwood <david.sherwood@arm.com>
28435 * tree-vect-stmts.c (simd_clone_subparts): New function.
28436 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
28438 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28439 Alan Hayward <alan.hayward@arm.com>
28440 David Sherwood <david.sherwood@arm.com>
28442 * tree-vect-stmts.c (vectorizable_call): Treat the number of
28443 vectors as polynomial. Use build_index_vector for
28444 IFN_GOMP_SIMD_LANE.
28446 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28447 Alan Hayward <alan.hayward@arm.com>
28448 David Sherwood <david.sherwood@arm.com>
28450 * tree-vect-stmts.c (get_load_store_type): Treat the number of
28451 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
28452 for variable-length vectors.
28453 (vectorizable_mask_load_store): Treat the number of units as
28454 polynomial, asserting that it is constant if the condition has
28455 already been enforced.
28456 (vectorizable_store, vectorizable_load): Likewise.
28458 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28459 Alan Hayward <alan.hayward@arm.com>
28460 David Sherwood <david.sherwood@arm.com>
28462 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
28463 of units as polynomial. Punt if we can't tell at compile time
28464 which vector contains the final result.
28466 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28467 Alan Hayward <alan.hayward@arm.com>
28468 David Sherwood <david.sherwood@arm.com>
28470 * tree-vect-loop.c (vectorizable_induction): Treat the number
28471 of units as polynomial. Punt on SLP inductions. Use an integer
28472 VEC_SERIES_EXPR for variable-length integer reductions. Use a
28473 cast of such a series for variable-length floating-point
28476 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28477 Alan Hayward <alan.hayward@arm.com>
28478 David Sherwood <david.sherwood@arm.com>
28480 * tree.h (build_index_vector): Declare.
28481 * tree.c (build_index_vector): New function.
28482 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
28483 of units as polynomial, forcibly converting it to a constant if
28484 vectorizable_reduction has already enforced the condition.
28485 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
28486 to create a {1,2,3,...} vector.
28487 (vectorizable_reduction): Treat the number of units as polynomial.
28488 Choose vectype_in based on the largest scalar element size rather
28489 than the smallest number of units. Enforce the restrictions
28492 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28493 Alan Hayward <alan.hayward@arm.com>
28494 David Sherwood <david.sherwood@arm.com>
28496 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
28497 number of units as polynomial.
28499 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28500 Alan Hayward <alan.hayward@arm.com>
28501 David Sherwood <david.sherwood@arm.com>
28503 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
28504 * target.def (autovectorize_vector_sizes): Return the vector sizes
28505 by pointer, using vector_sizes rather than a bitmask.
28506 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
28507 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
28508 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
28510 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
28511 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
28512 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
28513 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
28514 * omp-general.c (omp_max_vf): Likewise.
28515 * omp-low.c (omp_clause_aligned_alignment): Likewise.
28516 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
28517 * tree-vect-loop.c (vect_analyze_loop): Likewise.
28518 * tree-vect-slp.c (vect_slp_bb): Likewise.
28519 * doc/tm.texi: Regenerate.
28520 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
28522 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
28523 the vector size as a poly_uint64 rather than an unsigned int.
28524 (current_vector_size): Change from an unsigned int to a poly_uint64.
28525 (get_vectype_for_scalar_type): Update accordingly.
28526 * tree.h (build_truth_vector_type): Take the size and number of
28527 units as a poly_uint64 rather than an unsigned int.
28528 (build_vector_type): Add a temporary overload that takes
28529 the number of units as a poly_uint64 rather than an unsigned int.
28530 * tree.c (make_vector_type): Likewise.
28531 (build_truth_vector_type): Take the number of units as a poly_uint64
28532 rather than an unsigned int.
28534 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28535 Alan Hayward <alan.hayward@arm.com>
28536 David Sherwood <david.sherwood@arm.com>
28538 * target.def (get_mask_mode): Take the number of units and length
28539 as poly_uint64s rather than unsigned ints.
28540 * targhooks.h (default_get_mask_mode): Update accordingly.
28541 * targhooks.c (default_get_mask_mode): Likewise.
28542 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
28543 * doc/tm.texi: Regenerate.
28545 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28546 Alan Hayward <alan.hayward@arm.com>
28547 David Sherwood <david.sherwood@arm.com>
28549 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
28550 * omp-general.c (omp_max_vf): Likewise.
28551 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
28552 (expand_omp_simd): Handle polynomial safelen.
28553 * omp-low.c (omplow_simd_context): Add a default constructor.
28554 (omplow_simd_context::max_vf): Change from int to poly_uint64.
28555 (lower_rec_simd_input_clauses): Update accordingly.
28556 (lower_rec_input_clauses): Likewise.
28558 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28559 Alan Hayward <alan.hayward@arm.com>
28560 David Sherwood <david.sherwood@arm.com>
28562 * tree-vectorizer.h (vect_nunits_for_cost): New function.
28563 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
28564 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
28565 (vect_analyze_slp_cost): Likewise.
28566 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
28567 (vect_model_load_cost): Likewise.
28569 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28570 Alan Hayward <alan.hayward@arm.com>
28571 David Sherwood <david.sherwood@arm.com>
28573 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
28574 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
28575 from an unsigned int * to a poly_uint64_pod *.
28576 (calculate_unrolling_factor): New function.
28577 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
28579 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28580 Alan Hayward <alan.hayward@arm.com>
28581 David Sherwood <david.sherwood@arm.com>
28583 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
28584 from an unsigned int to a poly_uint64.
28585 (_loop_vec_info::slp_unrolling_factor): Likewise.
28586 (_loop_vec_info::vectorization_factor): Change from an int
28588 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
28589 (vect_get_num_vectors): New function.
28590 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
28591 (vect_get_num_copies): Use vect_get_num_vectors.
28592 (vect_analyze_data_ref_dependences): Change max_vf from an int *
28593 to an unsigned int *.
28594 (vect_analyze_data_refs): Change min_vf from an int * to a
28596 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
28597 than an unsigned HOST_WIDE_INT.
28598 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
28599 (vect_analyze_data_ref_dependence): Change max_vf from an int *
28600 to an unsigned int *.
28601 (vect_analyze_data_ref_dependences): Likewise.
28602 (vect_compute_data_ref_alignment): Handle polynomial vf.
28603 (vect_enhance_data_refs_alignment): Likewise.
28604 (vect_prune_runtime_alias_test_list): Likewise.
28605 (vect_shift_permute_load_chain): Likewise.
28606 (vect_supportable_dr_alignment): Likewise.
28607 (dependence_distance_ge_vf): Take the vectorization factor as a
28608 poly_uint64 rather than an unsigned HOST_WIDE_INT.
28609 (vect_analyze_data_refs): Change min_vf from an int * to a
28611 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
28612 vfm1 as a poly_uint64 rather than an int. Make the same change
28613 for the returned bound_scalar.
28614 (vect_gen_vector_loop_niters): Handle polynomial vf.
28615 (vect_do_peeling): Likewise. Update call to
28616 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
28617 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
28619 * tree-vect-loop.c (vect_determine_vectorization_factor)
28620 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
28621 (vect_get_known_peeling_cost): Likewise.
28622 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
28623 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
28624 (vect_transform_loop): Likewise. Use the lowest possible VF when
28625 updating the upper bounds of the loop.
28626 (vect_min_worthwhile_factor): Make static. Return an unsigned int
28627 rather than an int.
28628 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
28629 polynomial unroll factors.
28630 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
28631 (vect_make_slp_decision): Likewise.
28632 (vect_supported_load_permutation_p): Likewise, and polynomial
28634 (vect_analyze_slp_cost): Handle polynomial vf.
28635 (vect_slp_analyze_node_operations): Likewise.
28636 (vect_slp_analyze_bb_1): Likewise.
28637 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
28638 than an unsigned HOST_WIDE_INT.
28639 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
28640 (vectorizable_load): Handle polynomial vf.
28641 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
28643 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
28645 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28646 Alan Hayward <alan.hayward@arm.com>
28647 David Sherwood <david.sherwood@arm.com>
28649 * match.pd: Handle bit operations involving three constants
28650 and try to fold one pair.
28652 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28654 * tree-vect-loop-manip.c: Include gimple-fold.h.
28655 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
28656 niters_maybe_zero parameters. Handle other cases besides a step of 1.
28657 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
28658 Add a path that uses a step of VF instead of 1, but disable it
28660 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
28661 and niters_no_overflow parameters. Update calls to
28662 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
28663 Create a new SSA name if the latter choses to use a ste other
28664 than zero, and return it via niters_vector_mult_vf_var.
28665 * tree-vect-loop.c (vect_transform_loop): Update calls to
28666 vect_do_peeling, vect_gen_vector_loop_niters and
28667 slpeel_make_loop_iterate_ntimes.
28668 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
28669 (vect_gen_vector_loop_niters): Update declarations after above changes.
28671 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
28673 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
28674 128-bit round to integer instructions.
28675 (ceil<mode>2): Likewise.
28676 (btrunc<mode>2): Likewise.
28677 (round<mode>2): Likewise.
28679 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28681 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
28682 unaligned VSX load/store on P8/P9.
28683 (expand_block_clear): Allow the use of unaligned VSX
28684 load/store on P8/P9.
28686 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28688 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
28690 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
28691 swap associated with both a load and a store.
28693 2018-01-02 Andrew Waterman <andrew@sifive.com>
28695 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
28696 * config/riscv/riscv.md (clear_cache): Use it.
28698 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
28700 * web.c: Remove out-of-date comment.
28702 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
28704 * expr.c (fixup_args_size_notes): Check that any existing
28705 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
28706 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
28707 (emit_single_push_insn): ...here.
28709 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
28711 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
28712 (const_vector_encoded_nelts): New function.
28713 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
28714 (const_vector_int_elt, const_vector_elt): Declare.
28715 * emit-rtl.c (const_vector_int_elt_1): New function.
28716 (const_vector_elt): Likewise.
28717 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
28718 of CONST_VECTOR_ELT.
28720 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
28722 * expr.c: Include rtx-vector-builder.h.
28723 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
28724 directly on the tree encoding.
28725 (const_vector_from_tree): Likewise.
28726 * optabs.c: Include rtx-vector-builder.h.
28727 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
28728 sequence of "u" values.
28729 * vec-perm-indices.c: Include rtx-vector-builder.h.
28730 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
28731 directly on the vec_perm_indices encoding.
28733 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
28735 * doc/rtl.texi (const_vector): Describe new encoding scheme.
28736 * Makefile.in (OBJS): Add rtx-vector-builder.o.
28737 * rtx-vector-builder.h: New file.
28738 * rtx-vector-builder.c: Likewise.
28739 * rtl.h (rtx_def::u2): Add a const_vector field.
28740 (CONST_VECTOR_NPATTERNS): New macro.
28741 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
28742 (CONST_VECTOR_DUPLICATE_P): Likewise.
28743 (CONST_VECTOR_STEPPED_P): Likewise.
28744 (CONST_VECTOR_ENCODED_ELT): Likewise.
28745 (const_vec_duplicate_p): Check for a duplicated vector encoding.
28746 (unwrap_const_vec_duplicate): Likewise.
28747 (const_vec_series_p): Check for a non-duplicated vector encoding.
28748 Say that the function only returns true for integer vectors.
28749 * emit-rtl.c: Include rtx-vector-builder.h.
28750 (gen_const_vec_duplicate_1): Delete.
28751 (gen_const_vector): Call gen_const_vec_duplicate instead of
28752 gen_const_vec_duplicate_1.
28753 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
28754 (gen_const_vec_duplicate): Use rtx_vector_builder.
28755 (gen_const_vec_series): Likewise.
28756 (gen_rtx_CONST_VECTOR): Likewise.
28757 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
28758 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
28759 Build a new vector rather than modifying a CONST_VECTOR in-place.
28760 (handle_special_swappables): Update call accordingly.
28761 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
28762 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
28763 Build a new vector rather than modifying a CONST_VECTOR in-place.
28764 (handle_special_swappables): Update call accordingly.
28766 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
28768 * simplify-rtx.c (simplify_const_binary_operation): Use
28769 CONST_VECTOR_ELT instead of XVECEXP.
28771 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
28773 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
28774 the selector elements to be different from the data elements
28775 if the selector is a VECTOR_CST.
28776 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
28777 ssizetype for the selector.
28779 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
28781 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
28782 before testing each element individually.
28783 * tree-vect-generic.c (lower_vec_perm): Likewise.
28785 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
28787 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
28788 * selftest-run-tests.c (selftest::run_tests): Call it.
28789 * vector-builder.h (vector_builder::operator ==): New function.
28790 (vector_builder::operator !=): Likewise.
28791 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
28792 (vec_perm_indices::all_from_input_p): New function.
28793 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
28794 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
28795 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
28796 instead of reading the VECTOR_CST directly. Detect whether both
28797 vector inputs are the same before constructing the vec_perm_indices,
28798 and update the number of inputs argument accordingly. Use the
28799 utility functions added above. Only construct sel2 if we need to.
28801 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
28803 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
28804 the broadcast of the low byte.
28805 (expand_mult_highpart): Use an explicit encoding for the permutes.
28806 * optabs-query.c (can_mult_highpart_p): Likewise.
28807 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
28808 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
28809 (vectorizable_bswap): Likewise.
28810 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
28811 explicit encoding for the power-of-2 permutes.
28812 (vect_permute_store_chain): Likewise.
28813 (vect_grouped_load_supported): Likewise.
28814 (vect_permute_load_chain): Likewise.
28816 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
28818 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
28819 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
28820 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
28821 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
28822 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
28823 (vect_gen_perm_mask_any): Likewise.
28825 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
28827 * int-vector-builder.h: New file.
28828 * vec-perm-indices.h: Include int-vector-builder.h.
28829 (vec_perm_indices): Redefine as an int_vector_builder.
28830 (auto_vec_perm_indices): Delete.
28831 (vec_perm_builder): Redefine as a stand-alone class.
28832 (vec_perm_indices::vec_perm_indices): New function.
28833 (vec_perm_indices::clamp): Likewise.
28834 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
28835 (vec_perm_indices::new_vector): New function.
28836 (vec_perm_indices::new_expanded_vector): Update for new
28837 vec_perm_indices class.
28838 (vec_perm_indices::rotate_inputs): New function.
28839 (vec_perm_indices::all_in_range_p): Operate directly on the
28840 encoded form, without computing elided elements.
28841 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
28842 encoding. Update for new vec_perm_indices class.
28843 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
28844 the given vec_perm_builder.
28845 (expand_vec_perm_var): Update vec_perm_builder constructor.
28846 (expand_mult_highpart): Use vec_perm_builder instead of
28847 auto_vec_perm_indices.
28848 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
28849 vec_perm_indices instead of auto_vec_perm_indices. Use a single
28850 or double series encoding as appropriate.
28851 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
28852 vec_perm_indices instead of auto_vec_perm_indices.
28853 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
28854 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
28855 (vect_permute_store_chain): Likewise.
28856 (vect_grouped_load_supported): Likewise.
28857 (vect_permute_load_chain): Likewise.
28858 (vect_shift_permute_load_chain): Likewise.
28859 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
28860 (vect_transform_slp_perm_load): Likewise.
28861 (vect_schedule_slp_instance): Likewise.
28862 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
28863 (vectorizable_mask_load_store): Likewise.
28864 (vectorizable_bswap): Likewise.
28865 (vectorizable_store): Likewise.
28866 (vectorizable_load): Likewise.
28867 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
28868 vec_perm_indices instead of auto_vec_perm_indices. Use
28869 tree_to_vec_perm_builder to read the vector from a tree.
28870 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
28871 vec_perm_builder instead of a vec_perm_indices.
28872 (have_whole_vector_shift): Use vec_perm_builder and
28873 vec_perm_indices instead of auto_vec_perm_indices. Leave the
28874 truncation to calc_vec_perm_mask_for_shift.
28875 (vect_create_epilog_for_reduction): Likewise.
28876 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
28877 from auto_vec_perm_indices to vec_perm_indices.
28878 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
28879 instead of changing individual elements.
28880 (aarch64_vectorize_vec_perm_const): Use new_vector to install
28881 the vector in d.perm.
28882 * config/arm/arm.c (expand_vec_perm_d::perm): Change
28883 from auto_vec_perm_indices to vec_perm_indices.
28884 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
28885 instead of changing individual elements.
28886 (arm_vectorize_vec_perm_const): Use new_vector to install
28887 the vector in d.perm.
28888 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
28889 Update vec_perm_builder constructor.
28890 (rs6000_expand_interleave): Likewise.
28891 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
28892 (rs6000_expand_interleave): Likewise.
28894 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
28896 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
28897 to qimode could truncate the indices.
28898 * optabs.c (expand_vec_perm_var): Likewise.
28900 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
28902 * Makefile.in (OBJS): Add vec-perm-indices.o.
28903 * vec-perm-indices.h: New file.
28904 * vec-perm-indices.c: Likewise.
28905 * target.h (vec_perm_indices): Replace with a forward class
28907 (auto_vec_perm_indices): Move to vec-perm-indices.h.
28908 * optabs.h: Include vec-perm-indices.h.
28909 (expand_vec_perm): Delete.
28910 (selector_fits_mode_p, expand_vec_perm_var): Declare.
28911 (expand_vec_perm_const): Declare.
28912 * target.def (vec_perm_const_ok): Replace with...
28913 (vec_perm_const): ...this new hook.
28914 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
28915 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
28916 * doc/tm.texi: Regenerate.
28917 * optabs.def (vec_perm_const): Delete.
28918 * doc/md.texi (vec_perm_const): Likewise.
28919 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
28920 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
28921 expand_vec_perm for constant permutation vectors. Assert that
28922 the mode of variable permutation vectors is the integer equivalent
28923 of the mode that is being permuted.
28924 * optabs-query.h (selector_fits_mode_p): Declare.
28925 * optabs-query.c: Include vec-perm-indices.h.
28926 (selector_fits_mode_p): New function.
28927 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
28928 is defined, instead of checking whether the vec_perm_const_optab
28929 exists. Use targetm.vectorize.vec_perm_const instead of
28930 targetm.vectorize.vec_perm_const_ok. Check whether the indices
28931 fit in the vector mode before using a variable permute.
28932 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
28933 vec_perm_indices instead of an rtx.
28934 (expand_vec_perm): Replace with...
28935 (expand_vec_perm_const): ...this new function. Take the selector
28936 as a vec_perm_indices rather than an rtx. Also take the mode of
28937 the selector. Update call to shift_amt_for_vec_perm_mask.
28938 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
28939 Use vec_perm_indices::new_expanded_vector to expand the original
28940 selector into bytes. Check whether the indices fit in the vector
28941 mode before using a variable permute.
28942 (expand_vec_perm_var): Make global.
28943 (expand_mult_highpart): Use expand_vec_perm_const.
28944 * fold-const.c: Includes vec-perm-indices.h.
28945 * tree-ssa-forwprop.c: Likewise.
28946 * tree-vect-data-refs.c: Likewise.
28947 * tree-vect-generic.c: Likewise.
28948 * tree-vect-loop.c: Likewise.
28949 * tree-vect-slp.c: Likewise.
28950 * tree-vect-stmts.c: Likewise.
28951 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
28953 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
28954 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
28955 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
28956 (aarch64_vectorize_vec_perm_const): ...this new function.
28957 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
28958 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
28959 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
28960 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
28961 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
28962 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
28963 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
28965 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
28966 check for NEON modes.
28967 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
28968 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
28969 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
28970 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
28972 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
28973 the old VEC_PERM_CONST conditions.
28974 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
28975 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
28976 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
28977 (ia64_vectorize_vec_perm_const_ok): Merge into...
28978 (ia64_vectorize_vec_perm_const): ...this new function.
28979 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
28980 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
28981 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
28982 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
28983 * config/mips/mips.c (mips_expand_vec_perm_const)
28984 (mips_vectorize_vec_perm_const_ok): Merge into...
28985 (mips_vectorize_vec_perm_const): ...this new function.
28986 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
28987 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
28988 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
28989 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
28990 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
28991 (rs6000_expand_vec_perm_const): Delete.
28992 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
28994 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
28995 (altivec_expand_vec_perm_const_le): Take each operand individually.
28996 Operate on constant selectors rather than rtxes.
28997 (altivec_expand_vec_perm_const): Likewise. Update call to
28998 altivec_expand_vec_perm_const_le.
28999 (rs6000_expand_vec_perm_const): Delete.
29000 (rs6000_vectorize_vec_perm_const_ok): Delete.
29001 (rs6000_vectorize_vec_perm_const): New function.
29002 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
29003 an element count and rtx array.
29004 (rs6000_expand_extract_even): Update call accordingly.
29005 (rs6000_expand_interleave): Likewise.
29006 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
29007 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
29008 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
29009 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
29010 (rs6000_expand_vec_perm_const): Delete.
29011 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
29012 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
29013 (altivec_expand_vec_perm_const_le): Take each operand individually.
29014 Operate on constant selectors rather than rtxes.
29015 (altivec_expand_vec_perm_const): Likewise. Update call to
29016 altivec_expand_vec_perm_const_le.
29017 (rs6000_expand_vec_perm_const): Delete.
29018 (rs6000_vectorize_vec_perm_const_ok): Delete.
29019 (rs6000_vectorize_vec_perm_const): New function. Remove stray
29020 reference to the SPE evmerge intructions.
29021 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
29022 an element count and rtx array.
29023 (rs6000_expand_extract_even): Update call accordingly.
29024 (rs6000_expand_interleave): Likewise.
29025 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
29026 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
29028 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
29030 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29032 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
29033 vector mode and that that mode matches the mode of the data
29035 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
29036 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
29037 directly using expand_vec_perm_1 when forcing selectors into
29039 (expand_vec_perm_var): New function, split out from expand_vec_perm.
29041 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29043 * optabs-query.h (can_vec_perm_p): Delete.
29044 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
29045 * optabs-query.c (can_vec_perm_p): Split into...
29046 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
29047 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
29048 particular selector is valid.
29049 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
29050 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
29051 (vect_grouped_load_supported): Likewise.
29052 (vect_shift_permute_load_chain): Likewise.
29053 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
29054 (vect_transform_slp_perm_load): Likewise.
29055 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
29056 (vectorizable_bswap): Likewise.
29057 (vect_gen_perm_mask_checked): Likewise.
29058 * fold-const.c (fold_ternary_loc): Likewise. Don't take
29059 implementations of variable permutation vectors into account
29060 when deciding which selector to use.
29061 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
29062 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
29063 with a false third argument.
29064 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
29065 to test whether the constant selector is valid and can_vec_perm_var_p
29066 to test whether a variable selector is valid.
29068 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29070 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
29071 * optabs-query.c (can_vec_perm_p): Likewise.
29072 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
29073 instead of vec_perm_indices.
29074 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
29075 (vect_gen_perm_mask_checked): Likewise,
29076 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
29077 (vect_gen_perm_mask_checked): Likewise,
29079 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29081 * optabs-query.h (qimode_for_vec_perm): Declare.
29082 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
29083 (qimode_for_vec_perm): ...this new function.
29084 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
29086 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29088 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
29089 does not have a conditional at the top.
29091 2018-01-02 Richard Biener <rguenther@suse.de>
29093 * ipa-inline.c (big_speedup_p): Fix expression.
29095 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
29098 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
29101 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
29105 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
29106 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
29107 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
29108 cond_taken_branch_cost 3->4.
29110 2018-01-01 Jakub Jelinek <jakub@redhat.com>
29112 PR tree-optimization/83581
29113 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
29114 TODO_cleanup_cfg if any changes have been made.
29116 PR middle-end/83608
29117 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
29118 convert_modes if target mode has the right side, but different mode
29121 PR middle-end/83609
29122 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
29123 last argument when extracting from CONCAT. If either from_real or
29124 from_imag is NULL, use expansion through memory. If result is not
29125 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
29126 the parts directly to inner mode, if even that fails, use expansion
29129 PR middle-end/83623
29130 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
29131 check for bswap in mode rather than HImode and use that in expand_unop
29134 Copyright (C) 2018 Free Software Foundation, Inc.
29136 Copying and distribution of this file, with or without modification,
29137 are permitted in any medium without royalty provided the copyright
29138 notice and this notice are preserved.