gcc/ChangeLog

   1 2019-02-12  Eric Botcazou  <ebotcazou@adacore.com>
   2
   3         * asan.c (asan_expand_mark_ifn): Take into account the alignment of
   4         the object to pick the size of stores on strict-alignment platforms.
   5
   6         * config/sparc/sparc.md (*movsi_insn): Minor tweak.
   7         (*movdi_insn_sp32): Likewise.
   8         (*movdi_insn_sp64): Likewise.
   9
  10 2019-02-12  Jan Hubicka  <hubicka@ucw.cz>
  11
  12         PR lto/88777
  13         * cgraphunit.c (analyze_functions): Clear READONLY flag for external
  14         types that needs constructiong.
  15         * tree.h (may_be_aliased): Do not check TYPE_NEEDS_CONSTRUCTING.
  16
  17 2019-02-12  Richard Biener  <rguenther@suse.de>
  18
  19         PR tree-optimization/89253
  20         * tree-ssa-loop-split.c (tree_ssa_split_loops): Check we can
  21         duplicate the loop.
  22
  23 2019-02-11  David Malcolm  <dmalcolm@redhat.com>
  24
  25         PR lto/88147
  26         * input.c (selftest::test_line_offset_overflow): New selftest.
  27         (selftest::input_c_tests): Call it.
  28
  29 2019-02-11  Martin Sebor  <msebor@redhat.com>
  30
  31         PR tree-optimization/88771
  32         * gimple-ssa-warn-restrict.c (pass_wrestrict::gate): Also enable
  33         when -Wstringop-overflow is set.
  34         (builtin_memref::builtin_memref): Adjust excessive upper bound
  35         only when lower bound is not excessive.
  36         (maybe_diag_overlap): Detect and diagnose excessive bounds via
  37         -Wstringop-ovefflow.
  38         (maybe_diag_offset_bounds): Rename...
  39         (maybe_diag_access_bounds): ...to this.
  40         (check_bounds_or_overlap): Adjust for name change above.
  41
  42 2019-02-11  Martin Sebor  <msebor@redhat.com>
  43
  44         PR c++/87996
  45         * builtins.c (max_object_size): Move from here...
  46         * builtins.h (max_object_size): ...and here...
  47         * tree.c (max_object_size): ...to here...
  48         * tree.h (max_object_size): ...and here.
  49
  50 2019-02-11  Bill Schmidt  <wschmidt@linux.ibm.com>
  51
  52         * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Shift-right
  53         and shift-left vector built-ins need to include a TRUNC_MOD_EXPR
  54         for correct semantics.
  55
  56 2019-02-11  Alan Modra  <amodra@gmail.com>
  57
  58         * doc/invoke.texi (man page RS/6000 and PowerPC Options): Mention
  59         -mlongcall and -mpltseq.
  60         (RS/6000 and PowerPC Options <-mlongcall>): Mention inline PLT calls.
  61         (RS/6000 and PowerPC Options <-mpltseq>): Document.
  62         * config/rs6000/rs6000.h (TARGET_PLTSEQ): Define.
  63         * config/rs6000/sysv4.opt (mpltseq): New option.
  64         * config/rs6000/sysv4.h (TARGET_PLTSEQ): Redefine.
  65         (SUBTARGET_OVERRIDE_OPTIONS): Error if given -mpltseq when assembler
  66         support is lacking.  Don't allow -mpltseq with -mbss-plt.
  67         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Warn if
  68         -mpltseq given for ELFv1.
  69         * config/rs6000/rs6000.c (rs6000_call_aix): Comment on UNSPEC_PLTSEQ.
  70         Only use UNSPEC_PLTSEQ for inline PLT calls.
  71         (rs6000_call_sysv, rs6000_sibcall_sysv): Expand comments.  Only
  72         use UNSPEC_PLTSEQ for inline PLT calls.
  73         (rs6000_indirect_call_template_1, rs6000_longcall_ref),
  74         (rs6000_call_aix, rs6000_call_sysv, rs6000_sibcall_sysv): Replace
  75         uses of HAVE_AS_PLTSEQ with TARGET_PLTSEQ, simplifying.
  76         * config/rs6000/rs6000.md (pltseq_tocsave_<mode>),
  77         (pltseq_plt16_ha_<mode>, pltseq_plt16_lo_<mode>),
  78         (pltseq_mtctr_<mode>): Likewise.
  79
  80 2019-02-11  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
  81
  82         * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Use 8-byte strings with
  83         Solaris ld.
  84         * configure: Regenerate.
  85
  86 2019-02-11  Jakub Jelinek  <jakub@redhat.com>
  87
  88         PR bootstrap/88714
  89         * config/arm/ldrdstrd.md (*arm_ldrd, *arm_strd): Use q constraint
  90         instead of r.
  91
  92 2019-02-11  Hans-Peter Nilsson  <hp@axis.com>
  93
  94         * function.c (assign_parm_setup_block): Use the stored
  95         size, not the passed size, when allocating stack-space,
  96         also for a parameter with alignment larger than
  97         MAX_SUPPORTED_STACK_ALIGNMENT.
  98
  99 2019-02-11  Martin Liska  <mliska@suse.cz>
 100
 101         PR ipa/89009
 102         * ipa-cp.c (build_toporder_info): Remove usage of a param.
 103         * ipa-inline.c (inline_small_functions): Likewise.
 104         * ipa-pure-const.c (propagate_pure_const): Likewise.
 105         (propagate_nothrow): Likewise.
 106         * ipa-reference.c (propagate): Likewise.
 107         * ipa-utils.c (struct searchc_env): Remove unused field.
 108         (searchc): Always search across AVAIL_INTERPOSABLE.
 109         (ipa_reduced_postorder): Always allow AVAIL_INTERPOSABLE as
 110         the only called IPA pure const can properly not propagate
 111         across interposable boundary.
 112         * ipa-utils.h (ipa_reduced_postorder): Remove param.
 113
 114 2019-02-11  Chung-Ju Wu  <jasonwucj@gmail.com>
 115
 116         * config/nds32/nds32.md (call_internal, call_value_internal,
 117         sibcall_internal, sibcall_value_internal): Use SImode for mem rtx.
 118
 119 2019-02-11  Hans-Peter Nilsson  <hp@axis.com>
 120
 121         * config/cris/cris.c (cris_preferred_minimum_alignment): Fix name
 122         typo.
 123
 124 2019-02-10  H.J. Lu  <hongjiu.lu@intel.com>
 125
 126         * config/i386/constraints.md (Yd): Replace AVX512BW with AVX512DQ
 127         in comments
 128
 129 2019-02-10  Chung-Ju Wu  <jasonwucj@gmail.com>
 130
 131         * config.gcc (with_nds32_lib): Set default --with-nds32-lib correctly.
 132
 133 2019-02-10  Jakub Jelinek  <jakub@redhat.com>
 134
 135         PR tree-optimization/89268
 136         * tree-if-conv.c (version_loop_for_if_conversion): Push to preds only
 137         if preds is non-NULL.
 138
 139 2019-02-09  Jan Hubicka  <hubicka@ucw.cz>
 140
 141         PR lto/89272
 142         * tree.c (fld_simplified_type_name): Also keep TYPE_DECL for
 143         polymorphic types.
 144
 145 2019-02-10  Monk Chiang  <sh.chiang04@gmail.com>
 146
 147         * config/nds32/nds32.md (trap): New pattern.
 148
 149 2019-02-10  Monk Chiang  <sh.chiang04@gmail.com>
 150
 151         * config/nds32/nds32.c (nds32_dwarf_register_span): Refine register
 152         dwarf span.
 153
 154 2019-02-10  Chung-Ju Wu  <jasonwucj@gmail.com>
 155
 156         * config/nds32/nds32-md-auxiliary.c (nds32_spilt_doubleword): Support
 157         to split POST_INC.
 158
 159 2019-02-09  Jan Hubicka  <hubicka@ucw.cz>
 160
 161         * ipa-visibility.c (localize_node): Also do not localize
 162         LDPR_PREVAILING_DEF_IRONLY_EXP.
 163
 164 2019-02-09  Jan Hubicka  <hubicka@ucw.cz>
 165
 166         PR lto/87957
 167         * tree.c (fld_simplified_type_name): Use DECL_ASSEMBLER_NAME_SET_P
 168         instead of type_with_linkage.
 169
 170 2019-02-09  Jan Hubicka  <hubicka@ucw.cz>
 171
 172         PR ipa/88755
 173         * params.def (uninlined-function-insns, uninlined-function-time,
 174         uninlined-thunk-insns, uninlined-thunk-time): Add artificial upper
 175         bound so we don't get overflows.
 176
 177 2019-02-09  Aaron Sawdey  <acsawdey@linux.ibm.com>
 178
 179         * config/rs6000/rs6000-string.c (expand_compare_loop,
 180         expand_block_compare): Insert REG_BR_PROB notes in inline expansion of
 181         memcmp/strncmp.
 182
 183 2019-02-09  Jakub Jelinek  <jakub@redhat.com>
 184
 185         PR middle-end/89246
 186         * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
 187         If !node->definition and TYPE_ARG_TYPES is non-NULL, use
 188         TYPE_ARG_TYPES instead of DECL_ARGUMENTS.
 189
 190 2019-02-09  Alan Modra  <amodra@gmail.com>
 191
 192         PR target/88343
 193         * config/rs6000/rs6000.c (save_reg_p): Correct calls_eh_return
 194         case.  Match logic in rs6000_emit_prologue emitting pic_offset_table
 195         setup.
 196
 197 2019-02-08  Vladimir Makarov  <vmakarov@redhat.com>
 198
 199         PR middle-end/88560
 200         * lra-constraints.c (process_alt_operands): Don't increase reject
 201         for memory when offset memory is required.
 202
 203 2019-02-08  Robin Dapp  <rdapp@linux.ibm.com>
 204
 205         * config/s390/vector.md: Implement vector copysign.
 206
 207 2019-02-08  H.J. Lu  <hongjiu.lu@intel.com>
 208
 209         * expr.c (expand_constructor): Correct indentations.
 210
 211 2019-02-08  Richard Biener  <rguenther@suse.de>
 212
 213         PR tree-optimization/89247
 214         * tree-if-conv.c: Include tree-cfgcleanup.h.
 215         (version_loop_for_if_conversion): Record LOOP_VECTORIZED call.
 216         (tree_if_conversion): Pass through predicate vector.
 217         (pass_if_conversion::execute): Do CFG cleanup and SSA update
 218         inline, see if any if-converted loops we refrece in
 219         LOOP_VECTORIZED calls vanished and fixup.
 220         * tree-if-conv.h (tree_if_conversion): Adjust prototype.
 221
 222 2019-02-08  Ilya Leoshkevich  <iii@linux.ibm.com>
 223
 224         * config/s390/constraints.md (jdd): New constraint.
 225
 226 2019-02-08  H.J. Lu  <hongjiu.lu@intel.com>
 227
 228         PR target/89229
 229         * config/i386/i386.md (*movoi_internal_avx): Set mode to XI for
 230         upper 16 vector registers without TARGET_AVX512VL.
 231         (*movti_internal): Likewise.
 232
 233 2019-02-08  Jakub Jelinek  <jakub@redhat.com>
 234
 235         PR rtl-optimization/89234
 236         * except.c (copy_reg_eh_region_note_forward): Return if note_or_insn
 237         is a NOTE, CODE_LABEL etc. - rtx_insn * other than INSN_P.
 238         (copy_reg_eh_region_note_backward): Likewise.
 239
 240 2019-02-08  Richard Biener  <rguenther@suse.de>
 241
 242         PR middle-end/89223
 243         * tree-data-ref.c (initialize_matrix_A): Fail if constant
 244         doesn't fit in HWI.
 245         (analyze_subscript_affine_affine): Handle failure from
 246         initialize_matrix_A.
 247
 248 2019-02-08  Jakub Jelinek  <jakub@redhat.com>
 249
 250         * cfganal.c (pre_and_rev_post_order_compute_fn): Use fn instead of
 251         cfun everywhere.
 252
 253 2019-02-07  David Malcolm  <dmalcolm@redhat.com>
 254
 255         PR tree-optimization/86637
 256         PR tree-optimization/89235
 257         * tree-vect-loop.c (optimize_mask_stores): Add an
 258         auto_purge_vect_location sentinel to ensure that vect_location is
 259         purged on exit.
 260         * tree-vectorizer.c
 261         (auto_purge_vect_location::~auto_purge_vect_location): New dtor.
 262         (try_vectorize_loop_1): Add an auto_purge_vect_location sentinel
 263         to ensure that vect_location is purged on exit.
 264         (pass_slp_vectorize::execute): Likewise, replacing the manual
 265         reset.
 266         * tree-vectorizer.h (class auto_purge_vect_location): New class.
 267
 268 2019-02-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
 269
 270         * config/aarch64/iterators.md (max_opp): New code_attr.
 271         (USMAX): New code iterator.
 272         * config/aarch64/predicates.md (aarch64_smin): New predicate.
 273         (aarch64_smax): Likewise.
 274         * config/aarch64/aarch64-simd.md (abd<mode>_3): Rename to...
 275         (*aarch64_<su>abd<mode>_3): ... Change RTL representation to
 276         MINUS (MAX MIN).
 277
 278 2019-02-07  H.J. Lu  <hongjiu.lu@intel.com>
 279
 280         PR target/89229
 281         * config/i386/i386.md (*movoi_internal_avx): Set mode to OI
 282         for TARGET_AVX512VL.
 283         (*movti_internal): Set mode to TI for TARGET_AVX512VL.
 284
 285 2019-02-07  Andreas Krebbel  <krebbel@linux.ibm.com>
 286
 287         * config/s390/s390-builtin-types.def: Add new types.
 288         * config/s390/s390-builtins.def: (s390_vec_xl, s390_vec_xld2)
 289         (s390_vec_xlw4): Make the memory operand into a const pointer.
 290         (s390_vec_xld2, s390_vec_xlw4): Add a variant for single precision
 291         float.
 292         * config/s390/s390-c.c (s390_expand_overloaded_builtin): Generate
 293         a new vector type with the alignment of the scalar memory operand.
 294
 295 2019-02-07  Matthew Malcomson  <matthew.malcomson@arm.com>
 296             Jakub Jelinek  <jakub@redhat.com>
 297
 298         PR bootstrap/88714
 299         * config/arm/arm-protos.h (valid_operands_ldrd_strd,
 300         arm_count_ldrdstrd_insns): New declarations.
 301         * config/arm/arm.c (mem_ok_for_ldrd_strd): Remove broken handling of
 302         MINUS.
 303         (valid_operands_ldrd_strd): New function.
 304         (arm_count_ldrdstrd_insns): New function.
 305         * config/arm/ldrdstrd.md: Change peepholes to generate PARALLEL SImode
 306         sets instead of single DImode set and define new insns to match this.
 307
 308 2019-02-07  Tamar Christina  <tamar.christina@arm.com>
 309
 310         * config/aarch64/aarch64-builtins.c (aarch64_fcmla_lane_builtin_data):
 311         Make it a C initializer.
 312
 313 2019-02-07  Tamar Christina  <tamar.christina@arm.com>
 314
 315         PR/target 88850
 316         * config/arm/neon.md (*neon_mov<mode>): Add r -> r case.
 317
 318 2019-02-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
 319
 320         * config/arm/neon.md (neon_<sup>dot<vsi2qi>):
 321         Use neon_dot<q> for type.
 322         (neon_<sup>dot_lane<vsi2qi>): Likewise.
 323
 324 2019-02-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
 325
 326         * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>):
 327         Use neon_dot<q> for type.
 328         (aarch64_<sur>dot_lane<vsi2qi>): Likewise.
 329         (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
 330
 331 2019-02-06  Vladimir Makarov  <vmakarov@redhat.com>
 332
 333         PR rtl-optimization/89225
 334         * lra-constaints.c (simplify_operand_subreg): Add subreg mode
 335         sizes check.
 336
 337 2019-02-06  Eric Botcazou  <ebotcazou@adacore.com>
 338
 339         * config/i386/i386.c (ix86_expand_prologue): Emit a memory blockage
 340         after restoring registers saved to allocate the frame on Windows.
 341
 342 2019-02-06  Richard Biener  <rguenther@suse.de>
 343
 344         PR tree-optimization/89182
 345         * graphite.h (cached_scalar_evolution_in_region): Declare.
 346         * graphite.c (struct seir_cache_key): New.
 347         (struct sese_scev_hash): Likewise.
 348         (seir_cache): New global.
 349         (cached_scalar_evolution_in_region): New function.
 350         (graphite_transform_loops): Allocate and release seir_cache.
 351         * graphite-isl-ast-to-gimple.c (get_rename_from_scev): Use
 352         cached_scalar_evolution_in_region.
 353         * graphite-scop-detection.c (scop_detection::can_represent_loop):
 354         Simplify.
 355         (scop_detection::graphite_can_represent_expr: Use
 356         cached_scalar_evolution_in_region.
 357         (scop_detection::stmt_simple_for_scop_p): Likewise.
 358         (find_params_in_bb): Likewise.
 359         (gather_bbs::before_dom_children): Likewise.
 360         * graphite-sese-to-poly.c (create_pw_aff_from_tree): Likewise.
 361         (add_loop_constraints): Likewise.
 362
 363 2019-02-06  Jakub Jelinek  <jakub@redhat.com>
 364
 365         PR middle-end/89210
 366         * fold-const-call.c (fold_const_vec_convert): Pass true as last
 367         operand to new_unary_operation only if both element types are integral
 368         and it isn't a widening conversion.  Return NULL_TREE if
 369         new_unary_operation failed.
 370
 371 2019-02-05  Andreas Krebbel  <krebbel@linux.ibm.com>
 372
 373         PR target/88856
 374         * config/s390/s390.md: Remove load and test FP splitter.
 375
 376 2019-02-05  Aaron Sawdey  <acsawdey@linux.ibm.com>
 377
 378         PR target/89112
 379         * config/rs6000/rs6000-string.c (do_ifelse, expand_cmp_vec_sequence,
 380         expand_compare_loop, expand_block_compare_gpr,
 381         expand_strncmp_align_check, expand_strncmp_gpr_sequence): Insert
 382         REG_BR_PROB notes in inline expansion of memcmp/strncmp. Add
 383         #include "profile-count.h" and "predict.h" for types and functions
 384         needed to work with REG_BR_PROB notes.
 385
 386 2019-02-05  Aaron Sawdey  <acsawdey@linux.ibm.com>
 387
 388         PR target/89112
 389         * config/rs6000/rs6000.md (<bd>tf_<mode>): Generate a local label
 390         for the long branch case.
 391
 392 2019-02-05  Jakub Jelinek  <jakub@redhat.com>
 393
 394         PR target/89188
 395         * dce.c (delete_unmarked_insns): Don't remove no-op moves if they
 396         can throw, non-call exceptions are enabled and we can't delete
 397         dead exceptions or alter cfg.  Set must_clean if
 398         delete_insn_and_edges returns true, don't set it blindly for calls.
 399         Assert that delete_unreachable_blocks is called only if can_alter_cfg.
 400
 401         PR rtl-optimization/89195
 402         * combine.c (make_extraction): For MEMs, don't extract bytes outside
 403         of the original MEM.
 404
 405 2019-02-05  Martin Liska  <mliska@suse.cz>
 406
 407         PR gcov-profile/89000
 408         * gcov.c (function_summary): Remove argument.
 409         (file_summary): New function.
 410         (print_usage): Replace tabs with spaces.
 411         (generate_results): Use new function file_summary.
 412
 413 2019-02-05  Jakub Jelinek  <jakub@redhat.com>
 414
 415         PR target/89186
 416         * optabs.c (prepare_cmp_insn): Pass x and y to
 417         emit_block_comp_via_libcall rather than XEXP (x, 0) and XEXP (y, 0).
 418
 419 2019-02-05  Richard Biener  <rguenther@suse.de>
 420
 421         PR middle-end/89150
 422         * bitmap.h (struct bitmap_obstack): Do not mark GTY.
 423         (struct bitmap_element): Drop chain_prev so we properly recurse on
 424         the prev member, supporting tree views.
 425         (struct bitmap_head): GTY skip the obstack member.
 426
 427 2019-02-04  Alexander Monakov  <amonakov@ispras.ru>
 428
 429         PR c/88698
 430         * doc/extend.texi (Vector Extensions): Add an example of using vector
 431         types together with x86 intrinsics.
 432
 433 2019-02-04  Alan Modra  <amodra@gmail.com>
 434
 435         * config/rs6000/rs6000.c (rs6000_indirect_call_template_1): Increase
 436         str[] size to 160, and comment.
 437
 438 2019-02-04  Alan Modra  <amodra@gmail.com>
 439
 440         * config/rs6000/rs6000.c (rs6000_indirect_call_template_1),
 441         (rs6000_pltseq_template): Guard output of TLS markers with
 442         TARGET_TLS_MARKERS.
 443         (rs6000_longcall_ref, rs6000_call_aix, rs6000_call_sysv),
 444         (rs6000_sibcall_sysv): Ignore TARGET_TLS_MARKERS when deciding
 445         to use inline PLT sequences.
 446         * config/rs6000/rs6000.md (pltseq_tocsave_<mode>),
 447         (pltseq_plt16_ha_<mode>, pltseq_plt16_lo_<mode>),
 448         (pltseq_mtctr_<mode>): Don't test TARGET_TLS_MARKERS in predicate.
 449
 450 2019-02-04  Martin Liska  <mliska@suse.cz>
 451
 452         PR ipa/88985
 453         * ipa-fnsummary.c (estimate_edge_devirt_benefit): Bail
 454         out when ipa_fn_summaries does not contain entry for callee.
 455
 456 2019-02-04  Eric Botcazou  <ebotcazou@adacore.com>
 457
 458         * config/sparc/sparc.h: Remove superfluous blank lines.
 459         * config/sparc/sparc.c (global_offset_table_rtx): Rename into...
 460         (got_register_rtx): ...this.
 461         (sparc_got): Adjust to above renaming.
 462         (sparc_tls_got): Likewise.
 463         (sparc_delegitimize_address): Likewise.
 464         (sparc_output_mi_thunk): Likewise.
 465         (sparc_init_pic_reg): Likewise.
 466         (save_local_or_in_reg_p): Fix test on the GOT register.
 467         (USE_HIDDEN_LINKONCE): Move around.
 468         (get_pc_thunk_name): Likewise.
 469         (gen_load_pcrel_sym): Likewise.
 470         (load_got_register): Likewise.
 471
 472 2019-02-04  Kito Cheng  <kito.cheng@gmail.com>
 473
 474         * config/nds32/linux.h (GLIBC_DYNAMIC_LINKER): Define the naming rule
 475         of the dynamic linker: "ld-linux-nds32[le|be][f].so.1".
 476
 477 2019-02-04  Chung-Ju Wu  <jasonwucj@gmail.com>
 478
 479         * config/nds32/nds32.c (nds32_legitimate_address_p): Add TLS model
 480         into consideration.
 481
 482 2019-02-04  Chung-Ju Wu  <jasonwucj@gmail.com>
 483
 484         * config.gcc (with_nds32_lib, glibc):
 485         Remove TARGET_DEFAULT_TLSDESC_TRAMPOLINE=0 setting.
 486         * config/nds32/linux.h (TARGET_DEFAULT_TLSDESC_TRAMPOLINE): Delete.
 487         (NDS32_TLSDESC_TRAMPOLINE_SPEC): Delete.
 488
 489 2019-02-03  Uroš Bizjak  <ubizjak@gmail.com>
 490
 491         PR target/89071
 492         * config/i386/i386.md (*sqrt<mode>2_sse): Add (v,0) alternative.
 493         Do not prefer (v,v) alternative for non-AVX targets and (m,v)
 494         alternative for speed when TARGET_SSE_PARTIAL_REG_DEPENDENCY is set.
 495         (*rcpsf2_sse): Ditto.
 496         (*rsqrtsf2_sse): Ditto.
 497         (sse4_1_round<mode<2): Ditto.
 498
 499 2019-02-03  Richard Biener  <rguenther@suse.de>
 500
 501         PR debug/87295
 502         * dwarf2out.c (copy_ancestor_tree): Register non-stubs as
 503         orig.
 504
 505 2019-02-02  Jakub Jelinek  <jakub@redhat.com>
 506
 507         PR middle-end/87887
 508         * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
 509         Punt with warning on aggregate return or argument types.  Ignore
 510         type/mode checking for uniform arguments.
 511
 512 2019-02-01  Segher Boessenkool  <segher@kernel.crashing.org>
 513
 514         * combine.c (try_combine): Do not print "Can't combine" messages unless
 515         printing failed combination attempts.
 516
 517 2019-02-01  Martin Jambor  <mjambor@suse.cz>
 518
 519         PR hsa/87863
 520         * omp-grid.c (grid_mark_variable_segment): Set assembler name of group
 521         segment and global segment variables before making them static.
 522
 523 2019-02-01  Martin Jambor  <mjambor@suse.cz>
 524
 525         * omp-grid.c (grid_target_follows_gridifiable_pattern): Guard two
 526         missed optimization dump with dump_enabled_p.
 527
 528 2019-02-01  Richard Biener  <rguenther@suse.de>
 529
 530         PR middle-end/88597
 531         * tree-scalar-evolution.c (analyze_scalar_evolution): Set up
 532         the instantiate cache.
 533         (instantiate_scev_binary): Elide second operand procesing
 534         if equal to the first.
 535         * tree-chrec.c (chrec_contains_symbols): Add visited set.
 536         (chrec_contains_undetermined): Likewise.
 537         (tree_contains_chrecs): Likewise.
 538
 539 2019-02-01  Jan Hubicka  <hubicka@ucw.cz>
 540
 541         * parms.def (MAX_INLINE_INSNS_SINGLE): Reduce from 400 to 200.
 542
 543 2019-02-01  Jakub Jelinek  <jakub@redhat.com>
 544
 545         PR tree-optimization/89143
 546         * wide-int-range.h (wide_int_range_absu): Declare.
 547         * wide-int-range.cc (wide_int_range_absu): New function.
 548         * tree-vrp.c (extract_range_from_unary_expr): Handle ABSU_EXPR.
 549
 550         PR tree-optimization/88107
 551         * tree-cfg.c (find_outermost_region_in_block): Add ALL argument,
 552         instead of assertion that eh_region_outermost is non-NULL, if it
 553         is NULL, set *ALL to true and return NULL.
 554         (move_sese_region_to_fn): Adjust caller, if all is set, call
 555         duplicate_eh_regions with NULL region.
 556
 557 2019-02-01  Richard Biener  <rguenth@suse.de>
 558
 559         PR rtl-optimization/88593
 560         * mode-switching.c (optimize_mode_switching): Free dominators before
 561         calling cleanup_cfg.
 562
 563 2019-02-01  Bin Cheng  <bin.cheng@linux.alibaba.com>
 564
 565         PR tree-optimization/88932
 566         * tree-predcom.c (try_combine_chains): Get loop bbs in dom order.
 567
 568 2019-01-31  Jakub Jelinek  <jakub@redhat.com>
 569
 570         PR middle-end/89137
 571         * omp-low.c (lower_omp_task_reductions): Drop redundant test to avoid
 572         bogus clang warning.
 573
 574 2019-01-31  Uroš Bizjak  <ubizjak@gmail.com>
 575
 576         PR target/89071
 577         * config/i386/i386.md (*extendsfdf2): Split out reg->reg
 578         alternative to avoid partial SSE register stall for TARGET_AVX.
 579         (truncdfsf2): Ditto.
 580         (sse4_1_round<mode>2): Ditto.
 581
 582 2018-01-31  Bill Schmidt  <wschmidt@linux.ibm.com>
 583
 584         PR tree-optimization/89008
 585         * gimple-ssa-strength-reduction.c (slsr_process_mul): Don't
 586         process anything of the form X * 0.
 587
 588 2019-01-31  Richard Biener  <rguenther@suse.de>
 589
 590         PR tree-optimization/89135
 591         * tree-ssa-phiprop.c (pass_phiprop::execute): Skip blocks
 592         with abnormal preds.
 593
 594 2019-01-31  Jakub Jelinek  <jakub@redhat.com>
 595
 596         PR sanitizer/89124
 597         * ipa-inline.c (sanitize_attrs_match_for_inline_p): Allow inlining
 598         always_inline callees into no_sanitize_address callers.
 599
 600 2019-01-31  Richard Biener  <rguenther@suse.de>
 601
 602         PR rtl-optimization/89115
 603         * lra.c (lra_rtx_hash): Properly hash CONST_INT values.
 604
 605 2019-01-30  Martin Sebor  <msebor@redhat.com>
 606
 607         PR other/89106
 608         * doc/extend.texi (cast to a union): Correct and expand.
 609
 610 2019-01-30  Vladimir Makarov  <vmakarov@redhat.com>
 611
 612         PR rtl-optimization/87246
 613         * lra-constraints.c (simplify_operand_subreg): Reload memory
 614         in subreg if the address became invalid.
 615
 616 2019-01-30  Bill Schmidt  <wschmidt@linux.ibm.com>
 617
 618         PR target/87064
 619         * config/rs6000/vsx.md (*vsx_reduc_<VEC_reduc_name>_v4sf_scalar):
 620         Disable for little-endian.
 621
 622 2019-01-30  Richard Biener  <rguenther@suse.de>
 623
 624         PR rtl-optimization/89115
 625         * opts.c (default_options_optimization): Reduce
 626         PARAM_MAX_DSE_ACTIVE_LOCAL_STORES by a factor of 10 at -O1.
 627         Make PARAM_LOOP_INVARIANT_MAX_BBS_IN_LOOP reduction relative
 628         to the default.
 629
 630 2019-01-30  Kelvin Nilsen  <kelvin@gcc.gnu.org>
 631
 632         * config/rs6000/rs6000-c.c (altivec-resolve_overloaded_builtin):
 633         Change handling of ALTIVEC_BUILTIN_VEC_EXTRACT.  Coerce result to
 634         type of vector element when vec_extract is implemented by direct
 635         move.
 636
 637 2019-01-30  Thomas Schwinge  <thomas@codesourcery.com>
 638
 639         * doc/invoke.texi (C Language Options): List "-fopenacc-dim".
 640
 641 2019-01-30  Richard Biener  <rguenther@suse.de>
 642
 643         PR tree-optimization/89111
 644         * tree-ssa-loop-im.c (gather_mem_refs_stmt): Restrict
 645         canonicalization to appropriately sized access types.
 646
 647 2019-01-30  Jakub Jelinek  <jakub@redhat.com>
 648
 649         PR c++/89105
 650         * config/i386/i386.c (ix86_warn_parameter_passing_abi): Don't warn
 651         for arguments to functions that are TU-local and shouldn't be
 652         referenced by assembly.
 653
 654 2019-01-30  Ulrich Drepper  <drepper@redhat.com>
 655
 656         * dumpfile.c (opt_info_switch_p_1): Ignore '-' if it appears
 657         after '='.
 658
 659 2019-01-29  Martin Sebor  <msebor@redhat.com>
 660
 661         PR c/88956
 662         * gimple-fold.c (fold_array_ctor_reference): Avoid zero-length arrays.
 663
 664 2019-01-29  Jakub Jelinek  <jakub@redhat.com>
 665
 666         PR c++/66676
 667         PR ipa/89104
 668         * omp-simd-clone.c (simd_clone_clauses_extract)
 669         <case OMP_CLAUSE_ALIGNED>: Ignore clauses with NULL
 670         OMP_CLAUSE_ALIGNED_ALIGNMENT.
 671
 672 2019-01-29  Vineet Gupta  <vgupta@synopsys.com>
 673
 674         * config.gcc: Force .init_array for ARC.
 675
 676 2019-01-29  Richard Biener  <rguenther@suse.de>
 677
 678         PR debug/87295
 679         * dwarf2out.c (collect_skeleton_dies): New helper.
 680         (copy_decls_for_unworthy_types): Call it.
 681         (build_abbrev_table): Assert we do not try to replace
 682         DW_AT_signature refs with local refs.
 683
 684 2019-01-28  Jakub Jelinek  <jakub@redhat.com>
 685
 686         PR middle-end/89002
 687         * gimplify.c (gimplify_omp_for): When adding OMP_CLAUSE_*_GIMPLE_SEQ
 688         for lastprivate/linear IV, push gimplify context around gimplify_assign
 689         and, if it needed any temporaries, pop it into a gimple bind around the
 690         sequence.
 691
 692 2019-01-28  Bernd Edlinger  <bernd.edlinger@hotmail.de>
 693
 694         * common.opt (-Wattribute-alias): Remove "no-" from name.
 695         Make -Wattribute-alias command line option and
 696         #pragma GCC diagnostic ignored "-Wattribute-alias" work again.
 697
 698 2019-01-28  Jakub Jelinek  <jakub@redhat.com>
 699
 700         PR target/89073
 701         * doc/invoke.texi (-mclwb, -mprfchw, -mrdpid, -mrdseed, -msgx,
 702         -madx, -mhle, -mavx5124fmaps, -mavx512vnni, -mavx5124vnniw): Document
 703         x86 ISA options.
 704         (bmi2): Add missing @opindex.
 705         * doc/extend.texi (x86 target attribute): Move fma4, lwp, ssse3
 706         options alphabetically.  Add missing 3dnow, 3dnowa, adx, avx, avx2,
 707         avx5124fmaps, avx5124vnniw, avx512bitalg, avx512bw, avx512cd,
 708         avx512dq, avx512er, avx512f, avx512ifma, avx512pf, avx512vbmi,
 709         avx512vbmi2, avx512vl, avx512vnni, avx512vpopcntdq, bmi, bmi2,
 710         cldemote, clflushopt, clwb, clzero, crc32, cx16, f16c, fma, fsgsbase,
 711         fxsr, gfni, hle, lzcnt, movbe, movdir64b, movdiri, mwaitx, pconfig,
 712         pku, prefetchwt1, prfchw, ptwrite, rdpid, rdrnd, rdseed, rtm, sahf,
 713         sgx, sha, shstk, tbm, vaes, vpclmulqdq, waitpkg, wbnoinvd, xsave,
 714         xsavec, xsaveopt and xsaves options.
 715
 716 2019-01-28  Richard Biener  <rguenther@suse.de>
 717
 718         PR debug/89076
 719         * dwarf2out.c (gen_subprogram_die): Remove leftover from MPX
 720         support removal.
 721
 722 2019-01-28  Richard Biener  <rguenther@suse.de>
 723
 724         PR tree-optimization/88739
 725         * tree-cfg.c (verify_types_in_gimple_reference): Verify
 726         BIT_FIELD_REFs only are applied to mode-precision operands
 727         when they are integral.
 728         (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR.
 729         * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid generating
 730         BIT_FIELD_REFs of non-mode-precision integral operands.
 731
 732 2019-01-27  Jakub Jelinek  <jakub@redhat.com>
 733
 734         PR target/87214
 735         * config/i386/sse.md
 736         (<mask_codefor>avx512dq_shuf_<shuffletype>64x2_1<mask_name>,
 737         avx512f_shuf_<shuffletype>64x2_1<mask_name>): Ensure the
 738         first constants in pairs are multiples of 2.  Formatting fixes.
 739         (avx512vl_shuf_<shuffletype>32x4_1<mask_name>,
 740         avx512vl_shuf_<shuffletype>32x4_1<mask_name>): Ensure the
 741         first constants in each quadruple are multiples of 4.  Formatting fixes.
 742
 743 2019-01-26  Martin Jambor  <mjambor@suse.cz>
 744
 745         PR ipa/88933
 746         * tree-inline.c: Include tree-cfgcleanup.h.
 747         (delete_unreachable_blocks_update_callgraph): Move...
 748         * tree-cfgcleanup.c (delete_unreachable_blocks_update_callgraph):
 749         ...here, make externally visible, make second argument bool, adjust
 750         all callers.
 751         * tree-cfgcleanup.c: Include cgraph.h.
 752         * tree-cfgcleanup.h (delete_unreachable_blocks_update_callgraph):
 753         Declare.
 754         * ipa-prop.c: Include tree-cfgcleanup.h.
 755         (ipcp_transform_function): Call
 756         delete_unreachable_blocks_update_callgraph instead of cleaning uo CFG.
 757
 758 2019-01-25  Vladimir Makarov  <vmakarov@redhat.com>
 759
 760         PR rtl-optimization/88846
 761         * ira.c (process_set_for_memref_referenced_p): New.
 762         (memref_referenced_p): Add new param.  Use
 763         process_set_for_memref_referenced_p.  Add new switch cases.
 764         (memref_used_between_p): Pass new arg to memref_referenced_p.
 765
 766 2019-01-25  Richard Earnshaw  <rearnsha@arm.com>
 767
 768         PR target/88469
 769         * config/aarch64/aarch64.c (aarch64_function_arg_alignment): Add new
 770         argument ABI_BREAK.  Set to true if the calculated alignment has
 771         changed in gcc-9.  Check bit-fields for their base type alignment.
 772         (aarch64_layout_arg): Warn if argument passing has changed in gcc-9.
 773         (aarch64_function_arg_boundary): Likewise.
 774         (aarch64_gimplify_va_arg_expr): Likewise.
 775
 776 2019-01-25  Richard Sandiford  <richard.sandiford@arm.com>
 777
 778         PR middle-end/89037
 779         * varasm.c (output_constructor_bitfield): Use wi::extract_uhwi
 780         instead of accessing TREE_INT_CST_ELT directly.
 781
 782 2019-01-25  Christophe Lyon  <christophe.lyon@linaro.org>
 783
 784         * doc/sourcebuild.texi (Environment attributes): Add fenv and
 785         fenv_exceptions description.
 786
 787 2019-01-25  Wilco Dijkstra  <wdijkstr@arm.com>
 788
 789         PR rtl-optimization/87763
 790         * config/aarch64/aarch64.c (aarch64_select_cc_mode):
 791         Allow SUBREG when matching CC_NZmode compare.
 792
 793 2019-01-25  Richard Biener  <rguenther@suse.de>
 794
 795         PR tree-optimization/89049
 796         * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
 797         Look at the pattern stmt to determine if the stmt is vectorized.
 798
 799 2019-01-25  Richard Sandiford  <richard.sandiford@arm.com>
 800
 801         * config/aarch64/aarch64-sve.md (*pred_mov<mode>)
 802         (pred_mov<mode>): Handle all-register forms using both a new
 803         alternative and a split.
 804
 805 2019-01-25  Richard Biener  <rguenther@suse.de>
 806
 807         PR tree-optimization/86865
 808         * graphite-scop-detection.c (scop_detection::can_represent_loop):
 809         Reject non-do-while loops.
 810
 811 2019-01-24  Peter Bergner  <bergner@linux.ibm.com>
 812
 813         * config/rs6000/altivec.md (build_vector_mask_for_load): Use MEM_P.
 814         * config/rs6000/constraints.md (Q constraint): Use REG_P.
 815         * config/rs6000/darwin.h (PREFERRED_RELOAD_CLASS): Use SYMBOL_REF_P.
 816         * config/rs6000/freebsd64.h (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Use
 817         SYMBOL_REF_P, CONST_INT_P and CONST_DOUBLE_P.
 818         * config/rs6000/linux64.h (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
 819         * config/rs6000/predicates.md (altivec_register_operand, vint_operand,
 820         vsx_register_operand, vsx_reg_sfsubreg_ok, vfloat_operand,
 821         vlogical_operand, gpc_reg_operand, int_reg_operand,
 822         int_reg_operand_not_pseudo): Use SUBREG_P and HARD_REGISTER_P.
 823         (ca_operand, base_reg_operand, htm_spr_reg_operand, cc_reg_operand,
 824         cc_reg_not_cr0_operand, input_operand): Use SUBREG_P.
 825         (save_world_operation, restore_world_operation, lmw_operation,
 826         stmw_operation): Use MEM_P and REG_P.
 827         (tie_operand): Use MEM_P.
 828         (vrsave_operation, crsave_operation): Use REG_P.
 829         (mfcr_operation, mtcrf_operation): Use REG_P and CONST_INT_P.
 830         (fpr_reg_operand): Use SUBREG_P and HARD_REGISTER_NUM_P.
 831         (quad_int_reg_operand): Use HARD_REGISTER_NUM_P.
 832         (call_operand): Use HARD_REGISTER_P.
 833         (indexed_or_indirect_operand, altivec_indexed_or_indirect_operand):
 834         Use CONST_INT_P.
 835         (lwa_operand): Use SUBREG_P, REG_P and CONST_INT_P.
 836         * config/rs6000/rs6000-p8swap.c (insn_is_load_p, insn_is_store_p,
 837         quad_aligned_load_p, replace_swapped_aligned_store,
 838         recombine_lvx_pattern, replace_swapped_aligned_load,
 839         recombine_stvx_pattern): Use MEM_P.
 840         (const_load_sequence_p, adjust_vperm, replace_swapped_load_constant):
 841         Use MEM_P and SYMBOL_REF_P.
 842         (rtx_is_swappable_p): Use REG_P and CONST_INT_P.
 843         (insn_is_swappable_p): Use REG_P and MEM_P.
 844         (insn_is_swap_p, (alignment_mask): Use CONST_INT_P.
 845         * config/rs6000/rs6000-string.c (expand_block_clear, expand_block_move):
 846         Use CONST_INT_P.
 847         * config/rs6000/rs6000.c (rs6000_secondary_reload, rs6000_emit_cmove):
 848         Use CONST_DOUBLE_P.
 849         (rs6000_output_move_128bit): Use CONST_DOUBLE_P, CONST_INT_P and
 850         CONST_WIDE_INT_P.
 851         (rs6000_legitimize_address): Use CONST_DOUBLE_P, CONST_INT_P,
 852         CONST_WIDE_INT_P, REG_P and SYMBOL_REF_P.
 853         (rs6000_emit_move): Use CONST_DOUBLE_P, CONST_INT_P, HARD_REGISTER_P,
 854         HARD_REGISTER_NUM_P, MEM_P, REG_P, SUBREG_P, SYMBOL_REF_P and
 855         reg_or_subregno:
 856         (output_toc): Use CONST_DOUBLE_P, CONST_INT_P and SYMBOL_REF_P.
 857         (easy_altivec_constant, rs6000_legitimate_offset_address_p,
 858         rs6000_mode_dependent_address, rs6000_expand_mtfsf_builtin,
 859         rs6000_expand_set_fpscr_rn_builtin, rs6000_expand_set_fpscr_drn_builtin,
 860         rs6000_expand_unop_builtin, INT_P, rs6000_generate_compare,
 861         rs6000_machopic_legitimize_pic_address, rs6000_split_logical_inner,
 862         rs6000_split_logical_di): Use CONST_INT_P.
 863         (rs6000_legitimize_reload_address): Use CONST_INT_P, HARD_REGISTER_P,
 864         REG_P and SYMBOL_REF_P.
 865         (setup_incoming_varargs, rs6000_rtx_costs): Use CONST_INT_P and MEM_P.
 866         (print_operand): Use CONST_INT_P, MEM_P and REG_P.
 867         (virtual_stack_registers_memory_p, rs6000_legitimate_address_p,
 868         mems_ok_for_quad_peep): Use CONST_INT_P and REG_P.
 869         (rs6000_secondary_reload_memory): Use CONST_INT_P and SUBREG_P.
 870         (small_data_operand, print_operand_address): Use CONST_INT_P and
 871         SYMBOL_REF_P.
 872         (split_stack_arg_pointer_used_p): Use HARD_REGISTER_P.
 873         (rs6000_init_hard_regno_mode_ok, direct_move_p):
 874         Use HARD_REGISTER_NUM_P.
 875         (rs6000_secondary_reload_gpr): Use HARD_REGISTER_NUM_P and MEM_P.
 876         (rs6000_secondary_reload_class): Use HARD_REGISTER_NUM_P, REG_P,
 877         SUBREG_P and SYMBOL_REF_P.
 878         (register_to_reg_type, rs6000_secondary_reload_inner): Use SUBREG_P
 879         and HARD_REGISTER_NUM_P.
 880         (rs6000_adjust_vec_address): Use HARD_REGISTER_NUM_P and
 881         reg_or_subregno.
 882         (rs6000_adjust_cost, find_mem_ref): Use MEM_P.
 883         (macho_lo_sum_memory_operand, rs6000_eliminate_indexed_memrefs): Use
 884         MEM_P and REG_P.
 885         (legitimate_indirect_address_p, legitimate_lo_sum_address_p,
 886         registers_ok_for_quad_peep, rs6000_output_function_epilogue,
 887         find_addr_reg): Use REG_P.
 888         (altivec_expand_vec_perm_const): Use REG_P and SUBREG_P.
 889         (rs6000_emit_le_vsx_move): Use SUBREG_P.
 890         (offsettable_ok_by_alignment, constant_pool_expr_p,
 891         legitimate_small_data_p, rs6000_output_dwarf_dtprel,
 892         rs6000_delegitimize_address, rs6000_const_not_ok_for_debug_p,
 893         rs6000_cannot_force_const_mem, rs6000_output_addr_const_extra,
 894         rs6000_assemble_integer, create_TOC_reference,
 895         rs6000_emit_allocate_stack, rs6000_xcoff_encode_section_info,
 896         rs6000_call_aix, rs6000_call_aix): Use SYMBOL_REF_P.
 897         (rs6000_split_vec_extract_var): Use reg_or_subregno.
 898         * config/rs6000/rtems.h (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Use
 899         CONST_DOUBLE_P, CONST_INT_P and SYMBOL_REF_P.
 900         * config/rs6000/sysv4.h (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
 901         * config/rs6000/xcoff.h (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
 902         * config/rs6000/rs6000.h (RS6000_SYMBOL_REF_TLS_P): Use SYMBOL_REF_P.
 903         (REGNO_OK_FOR_INDEX_P, REGNO_OK_FOR_BASE_P): Use HARD_REGISTER_NUM_P.
 904         (INT_REG_OK_FOR_INDEX_P, INT_REG_OK_FOR_BASE_P): Use HARD_REGISTER_P.
 905         (CONSTANT_ADDRESS_P): Use CONST_INT_P and SYMBOL_REF_P.
 906         * config/rs6000/rs6000.md (define_expands strlensi, mod<mode>3
 907         and cbranch<mode>4): Use CONST_INT_P.
 908         (multiple define_splits): Use REG_P and SUBREG_P.
 909         (define_expands call, call_value): Use MEM_P.
 910         (define_expands sibcall, sibcall_value): Use CONST_INT_P and  MEM_P.
 911         (define insn *mtcrfsi): Use CONST_INT_P and REG_P.
 912         * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>,
 913         *vsx_le_perm_load_v8hi, *vsx_le_perm_load_v16qi): Use HARD_REGISTER_P
 914         and HARD_REGISTER_NUM_P.
 915         (multiple define_splits): Use HARD_REGISTER_NUM_P.
 916
 917 2019-01-24  Uroš Bizjak  <ubizjak@gmail.com>
 918
 919         PR rtl-optimization/88948
 920         * rtl.h (prepare_copy_insn): New prototype.
 921         * gcse.c (prepare_copy_insn): New function, split out from
 922         process_insert_insn.
 923         (process_insert_insn): Use prepare_copy_insn.
 924         * store-motion.c (replace_store_insn): Use prepare_copy_insn
 925         instead of gen_move_insn.
 926
 927 2019-01-24  Jakub Jelinek  <jakub@redhat.com>
 928
 929         PR debug/89006
 930         * config/i386/i386.c (ix86_pic_register_p): Return true for
 931         UNSPEC_SET_GOT too.
 932
 933         PR tree-optimization/88964
 934         * gimple-loop-interchange.cc (loop_cand::analyze_induction_var): Also
 935         punt if HONOR_SNANS (chrec).
 936
 937         PR middle-end/89015
 938         * tree-nested.c (convert_nonlocal_reference_stmt,
 939         convert_local_reference_stmt, convert_tramp_reference_stmt,
 940         convert_gimple_call) <case GIMPLE_OMP_TEAMS>: Treat
 941         gimple_omp_teams_host teams stmts like GIMPLE_OMP_PARALLEL
 942         or GIMPLE_OMP_TASK.
 943
 944         PR tree-optimization/89027
 945         * tree-inline.c (add_clobbers_to_eh_landing_pad): Don't add clobbers
 946         for "omp simd array" variables.
 947
 948 2019-01-24  Richard Earnshaw  <rearnsha@arm.com>
 949
 950         PR target/88469
 951         * profile-count.h (profile_count): On ARM systems using GCC 6/7/8
 952         force the alignment of m_val.
 953
 954 2019-01-24  Richard Biener  <rguenther@suse.de>
 955
 956         PR lto/87187
 957         * tree-streamer-out.c (write_ts_decl_common_tree_pointers):
 958         When in "legacy" debug mode make sure to reset self-origins.
 959
 960 2019-01-24  Martin Liska  <mliska@suse.cz>
 961
 962         PR gcov-profile/88994
 963         * gcov-io.c (mangle_path): Do not allocate a bigger buffer,
 964         result will be always smaller or equal to the original.
 965         * gcov.c (mangle_name): Fix else branch where we should
 966         also copy to PTR and shift the pointer.
 967
 968 2019-01-24  Xiong Hu Luo  <luoxhu@linux.vnet.ibm.com>
 969
 970         * tree-ssa-dom.c (test_for_singularity): Fix a comment typo.
 971         * vr-values.c (find_case_label_ranges): Fix a comment typo.
 972
 973 2019-01-23  Bin Cheng  <bin.cheng@arm.com>
 974             Steve Ellcey <sellcey@marvell.com>
 975
 976         PR target/85711
 977         * recog.c (address_operand): Return false on wrong mode for address.
 978         (constrain_operands): Check for mode with 'p' constraint.
 979
 980 2019-01-23  Uroš Bizjak  <ubizjak@gmail.com>
 981
 982         PR target/88998
 983         * config/i386/sse.md (sse2_cvtpi2pd): Add SSE alternatives.
 984         Disparage MMX alternative.
 985         (sse2_cvtpd2pi): Ditto.
 986         (sse2_cvttpd2pi): Ditto.
 987
 988 2019-01-23  David Malcolm  <dmalcolm@redhat.com>
 989
 990         PR driver/89014
 991         * config/aarch64/driver-aarch64.c (host_detect_local_cpu): Fix
 992         use-after-free of the result of
 993         aarch64_get_extension_string_for_isa_flags.
 994
 995 2019-01-23  Jakub Jelinek  <jakub@redhat.com>
 996
 997         PR c/44715
 998         * doc/extend.texi: Document break and continue behavior in
 999         statement expressions.
1000
1001 2019-01-23  Richard Biener  <rguenther@suse.de>
1002
1003         PR tree-optimization/89008
1004         * tree-ssa-reassoc.c (eliminate_using_constants): For * 0 do
1005         not leave another stray operand.
1006
1007 2019-01-23  Jakub Jelinek  <jakub@redhat.com>
1008
1009         * BASE-VER: Bump to 9.0.1.
1010
1011 2019-01-23  Eric Botcazou  <ebotcazou@adacore.com>
1012
1013         * cgraphunit.c (cgraph_node::expand_thunk): When expanding a GIMPLE
1014         thunk that returns by reference, use the type of the return object
1015         of the thunk instead of that of the alias to build the dereference.
1016
1017 2019-01-23  Vineet Gupta  <vgupta@synopsys.com>
1018
1019         * config/arc/atomic.md: Add operand to DMB instruction.
1020
1021 2019-01-23  Jakub Jelinek  <jakub@redhat.com>
1022
1023         PR tree-optimization/88964
1024         * gimple-loop-interchange.cc (loop_cand::analyze_induction_var): Use
1025         build_zero_cst instead of build_int_cst.  Return false for loop
1026         invariants which honor signed zeros.
1027
1028 2019-01-22  Segher Boessenkool  <segher@kernel.crashing.org>
1029
1030         * doc/invoke.texi (-fsplit-paths): This is enabled by default at -O3.
1031
1032 2019-01-22  Jakub Jelinek  <jakub@redhat.com>
1033
1034         PR target/88965
1035         * config/rs6000/rs6000.c: Include tree-vrp.h and tree-ssanames.h.
1036         (rs6000_gimple_fold_builtin): If MEM_REF address doesn't satisfy
1037         is_gimple_mem_ref_addr predicate, force it into a SSA_NAME first.
1038
1039         PR middle-end/88968
1040         * gimplify.c (gimplify_omp_atomic): Handle bitfield atomics with
1041         non-integral DECL_BIT_FIELD_REPRESENTATIVEs.
1042
1043         PR target/87064
1044         * config/rs6000/vsx.md (*vsx_reduc_<VEC_reduc_name>_v2df_scalar):
1045         Disable for little endian.
1046
1047 2019-01-22  Richard Earnshaw  <rearnsha@arm.com>
1048
1049         PR target/88469
1050         * config/arm/arm.c (arm_needs_double_word_align): Check
1051         DECL_BIT_FIELD_TYPE.
1052
1053 2019-01-22  Hongtao Liu  <hongtao.liu@intel.com>
1054             H.J. Lu  <hongjiu.lu@intel.com>
1055
1056         PR target/88909
1057         * config/i386/i386-builtin.def: Add mask2 to all builtin
1058         initializations.  Merge ARGS2 and SPECIAL_ARGS2 into ARGS and
1059         SPECIAL_ARGS.
1060         * config/i386/i386.c (BDESC): Add mask2 to the definition.
1061         (BDESC_FIRST): Likewise.
1062         (define_builtin): Add an argument for mask2.  Updated to handle
1063         both ix86_isa_flags and ix86_isa_flags2.
1064         (define_builtin_const): Likewise.
1065         (define_builtin_pure): Likewise.
1066         (define_builtin2): Deleted.
1067         (define_builtin_const2): Likewise.
1068         (builtin_description): Add a member, mask2.
1069         (bdesc_*): Add mask2 to builtin initializations.
1070         (ix86_init_mmx_sse_builtins): Update calls to def_builtin,
1071         def_builtin_const and def_builtin_pure.  Remove SPECIAL_ARGS2
1072         support.
1073         (ix86_get_builtin_func_type): Remove SPECIAL_ARGS2 support.
1074
1075 2019-01-22  H.J. Lu  <hongjiu.lu@intel.com>
1076
1077         PR target/88954
1078         * config/i386/i386.c (ix86_force_load_from_GOT_p): Also check
1079         noplt attribute.
1080
1081 2019-01-22  Richard Earnshaw  <rearnsha@arm.com>
1082
1083         PR target/88469
1084         * config/arm/arm.c (arm_needs_doubleword_align): Return 2 if a record's
1085         alignment is dominated by a bitfield with 64-bit aligned base type.
1086         (arm_function_arg): Emit a warning if the alignment has changed since
1087         earlier GCC releases.
1088         (arm_function_arg_boundary): Likewise.
1089         (arm_setup_incoming_varargs): Likewise.
1090
1091 2019-01-22  Richard Biener  <rguenther@suse.de>
1092
1093         PR tree-optimization/88862
1094         * graphite-scop-detection.c
1095         (scop_detection::graphite_can_represent_scev): Reject ADDR_EXPR.
1096
1097 2019-01-22  Andrew Stubbs  <ams@codesourcery.com>
1098
1099         * doc/extend.tex (AMD GCN Function Attributes): New section.
1100         * doc/install.texi (amdgcn-unknown-amdhsa): New instructions.
1101         * doc/invoke.texi (AMD GCN Options): New section.
1102         * doc/md.texi (Constraints for Particular Machines): Add AMD GCN.
1103
1104 2019-01-22  Eric Botcazou  <ebotcazou@adacore.com>
1105
1106         * config/sparc/sparc.c (parc_delegitimize_address): Recognize the GOT
1107         register and decoded HIGH/LO_SUM combinations for labels in PIC mode.
1108
1109 2019-01-22  Jakub Jelinek  <jakub@redhat.com>
1110
1111         PR tree-optimization/88044
1112         * tree-ssa-loop-niter.c (number_of_iterations_cond): If condition
1113         is false in the first iteration, but !every_iteration, return false
1114         instead of true with niter->niter zero.
1115
1116         PR rtl-optimization/88904
1117         * cfgcleanup.c (thread_jump): Verify cond2 doesn't mention
1118         any nonequal registers before processing BB_END (b).
1119
1120         PR target/88905
1121         * optabs.c (add_equal_note): Add op0_mode argument, use it instead of
1122         GET_MODE (op0).
1123         (expand_binop_directly, expand_doubleword_clz,
1124         expand_doubleword_popcount, expand_ctz, expand_ffs,
1125         expand_unop_direct, maybe_emit_unop_insn): Adjust callers.
1126
1127         PR rtl-optimization/49429
1128         PR target/49454
1129         PR rtl-optimization/86334
1130         PR target/88906
1131         * expr.c (emit_block_move_hints): Move marking of MEM_EXPRs
1132         addressable from here...
1133         (emit_block_op_via_libcall): ... to here.
1134
1135 2019-01-22  Richard Biener  <rguenther@suse.de>
1136
1137         * tree-vect-loop.c (vect_analyze_loop_operations): Use
1138         auto_vec for cost vector to fix memleak.
1139         (vectorize_fold_left_reduction): Properly gather SLP defs.
1140         (vectorizable_comparison): Do not swap operands to properly
1141         gather SLP defs.
1142
1143 2019-01-22  Alan Modra  <amodra@gmail.com>
1144
1145         PR target/88614
1146         * config/rs6000/predicates.md (unspec_tls): Ensure GOT reg
1147         stays a reg.  Allow a const_int.
1148         * config/rs6000/rs6000-protos.h (rs6000_output_tlsargs): Declare.
1149         * config/rs6000/rs6000.h (IS_V4_FP_ARGS): Define.
1150         (IS_NOMARK_TLSGETADDR): Define.
1151         * config/rs6000/rs6000.c (edit_tls_call_insn): Delete.
1152         (rs6000_output_tlsargs): New function.
1153         (rs6000_legitimize_tls_address): Don't say a !TARGET_TLS_MARKERS
1154         __tls_get_addr call takes an arg.
1155         (rs6000_call_sysv): Generate sysv4 secure plt call pattern here..
1156         * config/rs6000/rs6000.md (call_nonlocal_sysv): ..rather than here,
1157         delete split..
1158         (call_value_nonlocal_sysv): ..or here, delete split.
1159         (tls_gdld_nomark): Delete.
1160         (call_value_indirect_nonlocal_sysv): Use unspec_tls as operand2
1161         predicate.  Call rs6000_output_tlsargs.  Adjust length to suit.
1162         (call_value_nonlocal_sysv): Likewise.
1163         (call_value_nonlocal_sysv_secure): Likewise.
1164         (call_value_nonlocal_aix): Likewise.
1165         (call_value_indirect_aix): Likewise.
1166         (call_value_indirect_elfv2): Likewise.
1167         (call_value_local32, call_value_local64): Disable for no-mark tls.
1168         (call_value_local_aix): Likewise.
1169
1170 2019-01-21  Uroš Bizjak  <ubizjak@gmail.com>
1171
1172         PR target/88938
1173         * config/i386/i386.c (ix86_expand_builtin) [case IX86_BUILTIN_BEXTRI32,
1174         case IX86_BUILTIN_BEXTRI64]: Sanitize operands.
1175
1176 2019-01-21  Michael Ploujnikov  <michael.ploujnikov@oracle.com>
1177
1178         * hash-map-tests.c (test_map_of_strings_to_int): Show how to use
1179         string contents as hash_map keys.
1180
1181 2019-01-21  Bernd Edlinger  <bernd.edlinger@hotmail.de>
1182
1183         PR c/88928
1184         * c-warn.c (check_alignment_of_packed_member): Add a boolean parameter
1185         for rvalue context.  Handle rvalues correctly.  Use min_align_of_type
1186         instead of TYPE_ALIGN.
1187         (check_address_or_pointer_of_packed_member): Handle rvalues coorrectly.
1188         Use min_align_of_type instead of TYPE_ALIGN_UNIT.  Check for NULL
1189         pointer from TYPE_STUB_DECL.
1190
1191 2019-01-21  Richard Biener  <rguenther@suse.de>
1192
1193         PR tree-optimization/88934
1194         * tree-vect-slp.c (vect_mask_constant_operand_p): Always look
1195         at the possibly non-constant operand.
1196         (vect_get_constant_vectors): Adjust.
1197
1198 2019-01-21  H.J. Lu  <hongjiu.lu@intel.com>
1199
1200         PR target/71659
1201         * config/i386/adxintrin.h: Just check _IMMINTRIN_H_INCLUDED.
1202         * config/i386/clflushoptintrin.h: Check _IMMINTRIN_H_INCLUDED
1203         instead of _X86INTRIN_H_INCLUDED.
1204         * onfig/i386/clwbintrin.h: Likewise.
1205         * config/i386/pkuintrin.h: Likewise.
1206         * config/i386/prfchwintrin.h: Likewise.
1207         * config/i386/rdseedintrin.h: Likewise.
1208         * config/i386/wbnoinvdintrin.h: Likewise.
1209         * config/i386/xsavecintrin.h: Likewise.
1210         * config/i386/xsavesintrin.h: Likewise.
1211         * config/i386/fxsrintrin.h: Enable _IMMINTRIN_H_INCLUDED check.
1212         * config/i386/xsaveintrin.h: Likewise.
1213         * config/i386/xsaveoptintrin.h: Likewise.
1214         * config/i386/x86intrin.h: Move "#include" <rdseedintrin.h>,
1215         <prfchwintrin.h>, <fxsrintrin.h>, <xsaveintrin.h>,
1216         <xsaveoptintrin.h>, <adxintrin.h>, <clwbintrin.h>,
1217         <clflushoptintrin.h>, <xsavesintrin.h>, <xsavecintrin.h>,
1218         <wbnoinvdintrin.h> and <pkuintrin.h> to ...
1219         * config/i386/immintrin.h: Here.
1220
1221 2019-01-20  Martin Jambor  <mjambor@suse.cz>
1222
1223         PR ipa/87615
1224         * ipa-prop.h (struct ipa_func_body_info): Replaced field aa_walked
1225         with aa_walk_budget.
1226         * cgraph.h (ipa_polymorphic_call_context::get_dynamic_type): Add
1227         aa_walk_budget_p parameter.
1228         * ipa-fnsummary.c (unmodified_parm_1): New parameter fbi.  Limit AA
1229         walk.  Updated all callers.
1230         (unmodified_parm): New parameter fbi, pass it to unmodified_parm_1.
1231         (eliminated_by_inlining_prob): New parameter fbi, pass it on to
1232         unmodified_parm.
1233         (will_be_nonconstant_expr_predicate): New parameter fbi, removed
1234         parameter info.  Extract info from fbi.  Pass fbi to recursive calls
1235         and to unmodified_parm.
1236         (phi_result_unknown_predicate): New parameter fbi, removed parameter
1237         info, updated call to will_be_nonconstant_expr_predicate.
1238         (param_change_prob): New parameter fbi, limit AA walking.
1239         (analyze_function_body): Initialize aa_walk_budget in fbi.  Update
1240         calls to various above functions.
1241         * ipa-polymorphic-call.c (get_dynamic_type): Add aa_walk_budget_p
1242         parameter.  Use it to limit AA walking.
1243         * ipa-prop.c (detect_type_change_from_memory_writes): New parameter
1244         fbi, limit AA walk.
1245         (detect_type_change): New parameter fbi, pass it on to
1246         detect_type_change_from_memory_writes.
1247         (detect_type_change_ssa): Likewise.
1248         (aa_overwalked): Removed.
1249         (parm_preserved_before_stmt_p): Assume fbi is never NULL, stream line
1250         accordingly, adjust to the neew AA limiting scheme.
1251         (parm_ref_data_preserved_p): Likewise.
1252         (ipa_compute_jump_functions_for_edge): Adjust call to
1253         get_dynamic_type.
1254         (ipa_analyze_call_uses): Likewise.
1255         (ipa_analyze_virtual_call_uses): Pass fbi to detect_type_change_ssa.
1256         (ipa_analyze_node): Initialize aa_walk_budget.
1257         (ipcp_transform_function): Likewise.
1258         * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Update call
1259         to get_dynamic_type.
1260
1261 2019-01-19  Jakub Jelinek  <jakub@redhat.com>
1262
1263         * config/aarch64/aarch64.c (aarch64_stack_protect_guard): Move
1264         outside of #if CHECKING_P code.
1265
1266 2019-01-19  Richard Sandiford  <richard.sandiford@arm.com>
1267
1268         * gimple-loop-versioning.cc (loop_versioning::dump_inner_likelihood):
1269         New function, split out from...
1270         (loop_versioning::analyze_stride): ...here.
1271         (loop_versioning::find_per_loop_multiplication): Use gassign.
1272         (loop_versioning::analyze_term_using_scevs): Return a success code.
1273         (loop_versioning::analyze_arbitrary_term): New function.
1274         (loop_versioning::analyze_address_fragment): Use
1275         analyze_arbitrary_term if all else fails.
1276
1277 2019-01-18  Segher Boessenkool  <segher@kernel.crashing.org>
1278
1279         PR target/88892
1280         * config/rs6000/rs6000.md (*movsi_from_df): Allow only register
1281         operands.
1282
1283 2019-01-18  Richard Biener  <rguenther@suse.de>
1284
1285         PR tree-optimization/88903
1286         * tree-vect-stmts.c (vectorizable_shift): Verify we see all
1287         scalar stmts a SLP shift amount is composed of when detecting
1288         shifts by scalars.
1289
1290 2019-01-18  Richard Earnshaw  <rearnsha@arm.com>
1291
1292         PR target/88799
1293         * config/arm/arm-cpus.in (mp): New feature.
1294         (sec): New feature.
1295         (fgroup ARMv7ve): Add mp and sec features.
1296         (arch armv7-a): Add options to allow mp and sec extensions.
1297         (cpu generic-armv7-a): Add options to allow mp and sec extensions.
1298         (cpu cortex-a5, cpu cortex-7, cpu cortex-a9): Add mp and sec
1299         extenstions to the base architecture.
1300         (cpu cortex-a8): Add sec extension to the base architecture.
1301         (cpu marvell-pj4): Add mp and sec extensions to the base architecture.
1302         * config/arm/t-aprofile (MULTILIB_MATCHES): Map all armv7-a arch
1303         variants down to the base v7-a varaint.
1304         * config/arm/t-multilib (v7_a_arch_variants): New variable.
1305         * doc/invoke.texi (ARM Options): Add +mp and +sec to the list
1306         of permitted extensions for -march=armv7-a and for
1307         -mcpu=generic-armv7-a.
1308
1309 2019-01-18  Martin Liska  <mliska@suse.cz>
1310
1311         * params.def: Fix comment.
1312         * tree-profile.c (gimple_init_gcov_profiler): Bump function
1313         name.
1314         (gimple_gen_ic_func_profiler): Likewise.
1315
1316 2019-01-18  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
1317
1318         * config/aarch64/aarch64-opts.h (enum stack_protector_guard): New
1319         * config/aarch64/aarch64.c (aarch64_override_options_internal): Handle
1320         and put in error checks for stack protector guard options.
1321         (aarch64_stack_protect_guard): New.
1322         (TARGET_STACK_PROTECT_GUARD): Define.
1323         * config/aarch64/aarch64.md (UNSPEC_SSP_SYSREG): New.
1324         (reg_stack_protect_address<mode>): New.
1325         (stack_protect_set): Adjust for SSP_GLOBAL.
1326         (stack_protect_test): Likewise.
1327         * config/aarch64/aarch64.opt (-mstack-protector-guard-reg): New.
1328         (-mstack-protector-guard): Likewise.
1329         (-mstack-protector-guard-offset): Likewise.
1330
1331 2019-01-18  Jakub Jelinek  <jakub@redhat.com>
1332
1333         PR tree-optimization/86214
1334         * tree-inline.h (struct copy_body_data): Add
1335         add_clobbers_to_eh_landing_pads member.
1336         * tree-inline.c (add_clobbers_to_eh_landing_pad): New function.
1337         (copy_edges_for_bb): Call it if EH edge destination is <
1338         id->add_clobbers_to_eh_landing_pads.  Fix a comment typo.
1339         (expand_call_inline): Set id->add_clobbers_to_eh_landing_pads
1340         if flag_stack_reuse != SR_NONE and clear it afterwards.
1341
1342 2019-01-18  Christophe Lyon  <christophe.lyon@linaro.org>
1343
1344         PR target/85596
1345         * doc/install.texi (with-multilib-list): Document for aarch64.
1346
1347 2019-01-18  Jakub Jelinek  <jakub@redhat.com>
1348
1349         PR target/88734
1350         * config/arm/arm_neon.h: Fix #pragma GCC target syntax - replace
1351         (("..."))) with ("...").
1352
1353 2019-01-18  Sebastian Huber  <sebastian.huber@embedded-brains.de>
1354
1355         * doc/extend.texi (Built-in Functions for Memory Model Aware
1356         Atomic Operations): Document atomic fetch and nand.
1357
1358 2019-01-18  Martin Liska  <mliska@suse.cz>
1359             Richard Biener  <rguenther@suse.de>
1360
1361         PR middle-end/88587
1362         * cgraph.h (create_version_clone_with_body): Add new argument
1363         with attributes.
1364         * cgraphclones.c (cgraph_node::create_version_clone): Add
1365         DECL_ATTRIBUTES to a newly created decl.  And call
1366         valid_attribute_p so that proper cl_target_optimization_node
1367         is set for the newly created declaration.
1368         * multiple_target.c (create_target_clone): Set DECL_ATTRIBUTES
1369         for declaration.
1370         (expand_target_clones): Do not call valid_attribute_p, it must
1371         be already done.
1372         * tree-inline.c (copy_decl_for_dup_finish): Reset mode for
1373         vector types.
1374
1375 2019-01-17  Jakub Jelinek  <jakub@redhat.com>
1376
1377         PR target/88734
1378         * config/aarch64/arm_neon.h: Fix #pragma GCC target syntax - replace
1379         (("..."))) with ("...").  Use arch=armv8.2-a+sha3 instead of
1380         arch=armv8.2-a+crypto for vsha512hq_u64 etc. intrinsics.
1381
1382 2019-01-17  Martin Sebor  <msebor@redhat.com>
1383
1384         PR middle-end/88273
1385         * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range):
1386         Handle anti-ranges the same as no range at all.
1387
1388 2018-01-17  Steve Ellcey  <sellcey@cavium.com>
1389
1390         * config/aarch64/aarch64.c (cgraph.h): New include.
1391         (intl.h): New include.
1392         (supported_simd_type): New function.
1393         (currently_supported_simd_type): Ditto.
1394         (aarch64_simd_clone_compute_vecsize_and_simdlen): Ditto.
1395         (aarch64_simd_clone_adjust): Ditto.
1396         (aarch64_simd_clone_usable): Ditto.
1397         (TARGET_SIMD_CLONE_COMPUTE_VECSIZE_AND_SIMDLEN): New macro.
1398         (TARGET_SIMD_CLONE_ADJUST): Ditto.
1399         (TARGET_SIMD_CLONE_USABLE): Ditto.
1400         * config/i386/i386.c (ix86_simd_clone_adjust): Add definition check.
1401         * omp-simd-clone.c (expand_simd_clones): Add targetm.simd_clone.adjust
1402         call.
1403
1404 2019-01-17  Martin Sebor  <msebor@redhat.com>
1405
1406         PR tree-optimization/88800
1407         * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid checking
1408         NO_WARNING bit here.  Avoid folding out-of-bounds calls.
1409         * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Remove
1410         redundant argument.  Add new argument and issue diagnostics under
1411         its control.  Detect out-of-bounds access even with warnings
1412         disabled.
1413         (check_bounds_or_overlap): Change return type.  Add argument.
1414         (wrestrict_dom_walker::check_call): Adjust.
1415         * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Add argument.
1416         * tree-ssa-strlen.c (handle_builtin_strcpy): Adjust to change in
1417         check_bounds_or_overlap's return value.
1418         (handle_builtin_stxncpy): Same.
1419         (handle_builtin_strcat): Same.
1420
1421 2019-01-17  Andrew Stubbs  <ams@codesourcery.com>
1422             Kwok Cheung Yeung  <kcy@codesourcery.com>
1423             Julian Brown  <julian@codesourcery.com>
1424             Tom de Vries  <tom@codesourcery.com>
1425
1426         * doc/sourcebuild.texi: Document dg-add-options sqrt_insn.
1427
1428 2019-01-17  Andrew Stubbs  <ams@codesourcery.com>
1429
1430         * doc/sourcebuild.texi: Document dg-require-effective-target
1431         llvm_binutils and offload_gcn.
1432
1433 2019-01-17  Andrew Stubbs  <ams@codesourcery.com>
1434             Kwok Cheung Yeung  <kcy@codesourcery.com>
1435             Julian Brown  <julian@codesourcery.com>
1436             Tom de Vries  <tom@codesourcery.com>
1437
1438         * doc/sourcebuild.texi: Document dg-required-effective-target
1439         exceptions.
1440
1441 2019-01-17  Andrew Stubbs  <ams@codesourcery.com>
1442             Kwok Cheung Yeung  <kcy@codesourcery.com>
1443             Julian Brown  <julian@codesourcery.com>
1444             Tom de Vries  <tom@codesourcery.com>
1445             Jan Hubicka  <hubicka@ucw.cz>
1446             Martin Jambor  <mjambor@suse.cz>
1447
1448         * config.gcc: Add amdgcn*-*-amdhsa configuration.
1449         * configure.ac: Check for dlopen.
1450         * configure: Regenerate.
1451
1452 2019-01-17  Andrew Stubbs  <ams@codesourcery.com>
1453             Kwok Cheung Yeung  <kcy@codesourcery.com>
1454             Julian Brown  <julian@codesourcery.com>
1455             Tom de Vries  <tom@codesourcery.com>
1456             Jan Hubicka  <hubicka@ucw.cz>
1457             Martin Jambor  <mjambor@suse.cz>
1458
1459         * common/config/gcn/gcn-common.c: New file.
1460         * config/gcn/driver-gcn.c: New file.
1461         * config/gcn/gcn-builtins.def: New file.
1462         * config/gcn/gcn-hsa.h: New file.
1463         * config/gcn/gcn-modes.def: New file.
1464         * config/gcn/gcn-opts.h: New file.
1465         * config/gcn/gcn-passes.def: New file.
1466         * config/gcn/gcn-protos.h: New file.
1467         * config/gcn/gcn-run.c: New file.
1468         * config/gcn/gcn-tree.c: New file.
1469         * config/gcn/gcn.c: New file.
1470         * config/gcn/gcn.h: New file.
1471         * config/gcn/gcn.opt: New file.
1472         * config/gcn/t-gcn-hsa: New file.
1473
1474 2019-01-17  Andrew Stubbs  <ams@codesourcery.com>
1475             Kwok Cheung Yeung  <kcy@codesourcery.com>
1476             Julian Brown  <julian@codesourcery.com>
1477             Tom de Vries  <tom@codesourcery.com>
1478             Jan Hubicka  <hubicka@ucw.cz>
1479             Martin Jambor  <mjambor@suse.cz>
1480
1481         * config/gcn/constraints.md: New file.
1482         * config/gcn/gcn-valu.md: New file.
1483         * config/gcn/gcn.md: New file.
1484         * config/gcn/predicates.md: New file.
1485
1486 2019-01-17  Eric Botcazou  <ebotcazou@adacore.com>
1487
1488         * gimple-ssa-isolate-paths.c (stmt_uses_name_in_undefined_way): Replace
1489         flag_non_call_exceptions with cfun->can_throw_non_call_exceptions.
1490         (stmt_uses_0_or_null_in_undefined_way): Likewise.
1491         * tree-ssa-alias.c (same_addr_size_stores_p): Likewise.
1492
1493 2019-01-17  Tamar Christina  <tamar.christina@arm.com>
1494
1495         PR target/88851
1496         * config/aarch64/aarch64.md (STACK_CLASH_SVE_CFA_REGNUM): New.
1497         * config/aarch64/aarch64.c (aarch64_allocate_and_probe_stack_space): Use
1498         it and document registers.
1499
1500 2019-01-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1501
1502         * config/aarch64/aarch64.c (ares_tunings): Define.
1503         * config/aarch64/aarch64-cores.def (ares): Use the above.
1504
1505 2019-01-17  Wei Xiao  <wei3.xiao@intel.com>
1506
1507         PR target/88794
1508         Revert:
1509         2018-11-06  Wei Xiao  <wei3.xiao@intel.com>
1510
1511         * config/i386/avx512fintrin.h: Update VFIXUPIMM* intrinsics.
1512         (_mm512_fixupimm_round_pd): Update parameters and builtin.
1513         (_mm512_maskz_fixupimm_round_pd): Ditto.
1514         (_mm512_fixupimm_round_ps): Ditto.
1515         (_mm512_maskz_fixupimm_round_ps): Ditto.
1516         (_mm_fixupimm_round_sd): Ditto.
1517         (_mm_maskz_fixupimm_round_sd): Ditto.
1518         (_mm_fixupimm_round_ss): Ditto.
1519         (_mm_maskz_fixupimm_round_ss): Ditto.
1520         (_mm512_fixupimm_pd): Ditto.
1521         (_mm512_maskz_fixupimm_pd): Ditto.
1522         (_mm512_fixupimm_ps): Ditto.
1523         (_mm512_maskz_fixupimm_ps): Ditto.
1524         (_mm_fixupimm_sd): Ditto.
1525         (_mm_maskz_fixupimm_sd): Ditto.
1526         (_mm_fixupimm_ss): Ditto.
1527         (_mm_maskz_fixupimm_ss): Ditto.
1528         (_mm512_mask_fixupimm_round_pd): Update builtin.
1529         (_mm512_mask_fixupimm_round_ps): Ditto.
1530         (_mm_mask_fixupimm_round_sd): Ditto.
1531         (_mm_mask_fixupimm_round_ss): Ditto.
1532         (_mm512_mask_fixupimm_pd): Ditto.
1533         (_mm512_mask_fixupimm_ps): Ditto.
1534         (_mm_mask_fixupimm_sd): Ditto.
1535         (_mm_mask_fixupimm_ss): Ditto.
1536         * config/i386/avx512vlintrin.h:
1537         (_mm256_fixupimm_pd): Update parameters and builtin.
1538         (_mm256_maskz_fixupimm_pd): Ditto.
1539         (_mm256_fixupimm_ps): Ditto.
1540         (_mm256_maskz_fixupimm_ps): Ditto.
1541         (_mm_fixupimm_pd): Ditto.
1542         (_mm_maskz_fixupimm_pd): Ditto.
1543         (_mm_fixupimm_ps): Ditto.
1544         (_mm_maskz_fixupimm_ps): Ditto.
1545         (_mm256_mask_fixupimm_pd): Update builtin.
1546         (_mm256_mask_fixupimm_ps): Ditto.
1547         (_mm_mask_fixupimm_pd): Ditto.
1548         (_mm_mask_fixupimm_ps): Ditto.
1549         * config/i386/i386-builtin-types.def: Add new types and remove useless ones.
1550         * config/i386/i386-builtin.def: Update builtin definitions.
1551         * config/i386/i386.c: Handle new builtin types and remove useless ones.
1552         * config/i386/sse.md: Update VFIXUPIMM* patterns.
1553         (<avx512>_fixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
1554         (<avx512>_fixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
1555         (<avx512>_fixupimm<mode>_mask<round_saeonly_name>): Update.
1556         (avx512f_sfixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
1557         (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
1558         (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Update.
1559         * config/i386/subst.md:
1560         (round_saeonly_sd_mask_operand4): Add new subst_attr.
1561         (round_saeonly_sd_mask_op4): Ditto.
1562         (round_saeonly_expand_operand5): Ditto.
1563         (round_saeonly_expand): Update.
1564
1565 2019-01-17  Wei Xiao  <wei3.xiao@intel.com>
1566
1567         PR target/88794
1568         Revert:
1569         2018-11-12  Wei Xiao  <wei3.xiao@intel.com>
1570
1571         * config/i386/sse.md: Combine VFIXUPIMM* patterns
1572         (<avx512>_fixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
1573         (<avx512>_fixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
1574         (<avx512>_fixupimm<mode>_mask<round_saeonly_name>): Remove.
1575         (avx512f_sfixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
1576         (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
1577         (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Remove.
1578
1579 2019-01-17  Wei Xiao  <wei3.xiao@intel.com>
1580
1581         PR target/88794
1582         Revert:
1583         2018-12-15  Jakub Jelinek  <jakub@redhat.com>
1584
1585         PR target/88489
1586         * config/i386/sse.md (UNSPEC_SFIXUPIMM): New unspec enumerator.
1587         (avx512f_sfixupimm<mode><mask_name><round_saeonly_name>): Use it
1588         instead of UNSPEC_FIXUPIMM.
1589
1590 2019-01-17  Richard Biener  <rguenther@suse.de>
1591
1592         PR lto/86736
1593         * dwarf2out.c (want_pubnames): Never generate pubnames sections
1594         and friends for the LTO part of debug info.
1595
1596 2019-01-17  Jakub Jelinek  <jakub@redhat.com>
1597
1598         PR tree-optimization/86214
1599         * cfgexpand.c (add_stack_var_conflict): Don't add any conflicts
1600         if x == y.
1601
1602         PR rtl-optimization/88870
1603         * dce.c (deletable_insn_p): Never delete const/pure calls that can
1604         throw if we can't alter the cfg or delete dead exceptions.
1605         (mark_insn): Don't call find_call_stack_args for such calls.
1606
1607 2019-01-17  Kewen Lin  <linkw@gcc.gnu.org>
1608
1609         * doc/extend.texi: Add four new prototypes for vec_ld and seven new
1610         prototypes for vec_st.
1611         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add entries
1612         for scalar address type variants of altivec_vec_ld/altivec_vec_st,
1613         mainly on signed/unsigned long long and double.
1614
1615 2019-01-16  David Malcolm  <dmalcolm@redhat.com>
1616
1617         PR target/88861
1618         * combine.c (delete_noop_moves): Convert to "bool" return,
1619         returning true if any edges are eliminated.
1620         (combine_instructions): Also return true if delete_noop_moves
1621         returns true.
1622
1623 2019-01-16  Tamar Christina  <tamar.christina@arm.com>
1624
1625         * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
1626         correct max nunits for endian swap.
1627         (aarch64_expand_fcmla_builtin): Correct subreg code.
1628         * config/aarch64/aarch64-simd.md (aarch64_fcmla_lane<rot><mode>,
1629         aarch64_fcmla_laneq<rot>v4hf, aarch64_fcmlaq_lane<rot><mode>): Correct
1630         lane endianness.
1631
1632 2019-01-16  Uroš Bizjak  <ubizjak@gmail.com>
1633
1634         * config/alpha/alpha.c (alpha_gimplify_va_arg):
1635         Handle split indirect COMPLEX_TYPE arguments.
1636
1637 2019-01-16  Richard Earnshaw  <rearnsha@arm.com>
1638
1639         PR target/86891
1640         * config/aarch64/aarch64-modes.def: Add comment about how the carry
1641         bit is set by add and compare.
1642         (CC_ADC): New CC_MODE.
1643         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Use variables
1644         to cache the code and mode of X.  Adjust the shape of a CC_Cmode
1645         comparison.  Add detection for CC_ADCmode.
1646         (aarch64_get_condition_code_1): Update code support for CC_Cmode.  Add
1647         CC_ADCmode.
1648         * config/aarch64/aarch64.md (uaddv<mode>4): Use LTU with CCmode.
1649         (uaddvti4): Comparison result is in CC_ADCmode and the condition is GEU.
1650         (add<mode>3_compareC_cconly_imm): Delete.  Merge into...
1651         (add<mode>3_compareC_cconly): ... this.  Restructure the comparison
1652         to eliminate the need for zero-extending the operands.
1653         (add<mode>3_compareC_imm): Delete.  Merge into ...
1654         (add<mode>3_compareC): ... this.  Restructure the comparison to
1655         eliminate the need for zero-extending the operands.
1656         (add<mode>3_carryin): Use LTU for the overflow detection.
1657         (add<mode>3_carryinC): Use CC_ADCmode for the result of the carry out.
1658         Reexpress comparison for overflow.
1659         (add<mode>3_carryinC_zero): Update for change to add<mode>3_carryinC.
1660         (add<mode>3_carryinC): Likewise.
1661         (add<mode>3_carryinV): Use LTU for carry between partials.
1662         * config/aarch64/predicates.md (aarch64_carry_operation): Update
1663         handling of CC_Cmode and add CC_ADCmode.
1664         (aarch64_borrow_operation): Likewise.
1665
1666 2019-01-16  Tamar Christina  <tamar.christina@arm.com>
1667
1668         * config/arm/arm-protos.h (neon_vcmla_lane_prepare_operands): Remove patternmode.
1669         * config/arm/arm.c (neon_vcmla_lane_prepare_operands): Likewise.
1670         * config/arm/neon.md (neon_vcmla_lane<rot><mode>, neon_vcmla_laneq<rot><mode>,
1671         neon_vcmlaq_lane<rot><mode>): Remove endianness conversion.
1672
1673 2019-01-16  Martin Liska  <mliska@suse.cz>
1674
1675         * Makefile.in: Set TOOL_INCLUDE_DIR and NATIVE_SYSTEM_HEADER_DIR
1676         for GCC driver.
1677         * config/gnu-user.h (TARGET_F951_OPTIONS): Add 'finclude%s/' as
1678         a new argument.
1679         * gcc.c (add_sysrooted_hdrs_prefix): New function.
1680         (path_prefix_reset): Move up in the source file.
1681         (find_fortran_preinclude_file): Make complex search for the
1682         fortran header files.
1683
1684 2019-01-15  Nikhil Benesch  <nikhil.benesch@gmail.com>
1685
1686         * godump.c (go_output_typedef): When outputting a typedef, refer
1687         to the underlying type by its name and not its structure.
1688
1689 2019-01-15  David Malcolm  <dmalcolm@redhat.com>
1690
1691         PR c++/88795
1692         * tree.c (build_function_type): Assert that arg_types is not
1693         error_mark_node.
1694
1695 2019-01-15  Richard Sandiford  <richard.sandiford@arm.com>
1696
1697         PR inline-asm/52813
1698         * doc/extend.texi: Document that listing the stack pointer in the
1699         clobber list of an asm is a deprecated feature.
1700         * common.opt (Wdeprecated): Moved from c-family/c.opt.
1701         * cfgexpand.c (asm_clobber_reg_is_valid): Issue a -Wdeprecated
1702         warning instead of an error for clobbers of the stack pointer.
1703         Add a note explaining why.
1704
1705 2019-01-15  Richard Biener  <rguenther@suse.de>
1706
1707         PR debug/88046
1708         * dwarf2out.c (gen_member_die): Do not generate inheritance
1709         DIEs late.
1710
1711 2019-01-15  Richard Biener  <rguenther@suse.de>
1712
1713         PR tree-optimization/88855
1714         * tree-if-conv.c (combine_blocks): Collect
1715         SSA_NAME_OCCURS_IN_ABNORMAL_PHI from propagated out virtuals.
1716
1717 2019-01-15  Tom de Vries  <tdevries@suse.de>
1718
1719         PR target/80547
1720         * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Handle
1721         lhs == NULL_TREE for gang-level reduction.
1722
1723 2019-01-15  Richard Biener  <rguenther@suse.de>
1724             Prathamesh Kulkarni  <prathamesh.kulkarni@linaro.org>
1725
1726         PR ipa/88788
1727         * ipa-pure-const.c (malloc_candidate_p_1): Add parameter visited and
1728         return true if SSA_NAME is already marked in visited bitmap.
1729         (malloc_candidate_p): Pass visited to malloc_candidate_p_1.
1730
1731 2019-01-15  Jakub Jelinek  <jakub@redhat.com>
1732
1733         PR tree-optimization/88775
1734         * match.pd (cmp (convert1?@2 addr@0) (convert2? addr@1)): Optimize
1735         equal == 0 equality pointer comparisons some more if compared in
1736         integral types and either one points to an automatic var and the
1737         other to a global, or we can prove at least one points to the middle
1738         or both point to start or both point to end.
1739
1740 2019-01-14  Andi Kleen  <ak@linux.intel.com>
1741
1742         * Makefile.in: Lower autofdo sampling rate by 10x.
1743         * Makefile.tpl: Dito.
1744
1745 2019-01-14  Tom Honermann  <tom@honermann.net>
1746
1747         * defaults.h: Define CHAR8_TYPE.
1748
1749 2019-01-14  Martin Sebor  <msebor@redhat.com>
1750
1751         PR target/88638
1752         * doc/extend.texi (Darwin Format Checks): Clarify.
1753
1754 2019-01-14  Richard Biener  <rguenther@suse.de>
1755
1756         * genmatch.c (dt_simplify::gen_1): Change dumping dependent on
1757         whether we are in (simplify ...) or (match ...) context.
1758
1759 2019-01-14  Jakub Jelinek  <jakub@redhat.com>
1760
1761         PR rtl-optimization/88796
1762         * emit-rtl.h (struct rtl_data): Add stack_protect_guard_decl field.
1763         * cfgexpand.c (stack_protect_prologue): Initialize
1764         crtl->stack_protect_guard_decl.
1765         * function.c (stack_protect_epilogue): Use it instead of calling
1766         targetm.stack_protect_guard again.
1767         * dse.c (check_mem_read_rtx): Ignore MEM_VOLATILE_P reads from
1768         MEMs with MEM_EXPR equal to crtl->stack_protect_guard or
1769         crtl->stack_protect_guard_decl.
1770         * config/i386/i386.c (ix86_stack_protect_guard): Set TREE_THIS_VOLATILE
1771         on the returned MEM_EXPR.
1772
1773 2019-01-12  Tom de Vries  <tdevries@suse.de>
1774
1775         * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1): Alow setting
1776         vector length using -fopenacc-dim.
1777
1778 2019-01-12  Tom de Vries  <tdevries@suse.de>
1779
1780         * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Take larger vector
1781         lengths into account.
1782
1783 2019-01-12  Svante Signell  <svante.signell@gmail.com>
1784
1785         * config/i386/gnu.h (TARGET_THREAD_SSP_OFFSET): Define.
1786         (TARGET_CAN_SPLIT_STACK): Define.
1787         (TARGET_THREAD_SPLIT_STACK_OFFSET): Define.
1788
1789 2019-01-05  Jan Hubicka  <hubicka@ucw.cz>
1790
1791         * params.def (inline-unit-growth): Set to 40.
1792
1793 2019-01-12  Jakub Jelinek  <jakub@redhat.com>
1794
1795         * tree-ssa-loop-ivopts.c (find_inv_vars): Fix a comment typo.
1796
1797 2019-01-12  Tom de Vries  <tdevries@suse.de>
1798
1799         * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1): In offloading
1800         region calling vector-partitionable routine, set default_vector_length
1801         to WARP_SIZE.
1802
1803 2019-01-12  Tom de Vries  <tdevries@suse.de>
1804
1805         * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1): Add an use new
1806         variable default_vector_length.
1807
1808 2019-01-12  Tom de Vries  <tdevries@suse.de>
1809
1810         PR middle-end/88703
1811         * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1): Apply defaults
1812         from oacc_default_dims, as oacc_validate_dims would do it, and apply
1813         dimensions limits.
1814
1815 2019-01-12  Tom de Vries  <tdevries@suse.de>
1816
1817         * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1)
1818         (nvptx_goacc_validate_dims): Add used parameter.
1819         * doc/tm.texi: Regenerate.
1820         * omp-offload.c (oacc_parse_default_dims, oacc_validate_dims): Add
1821         argument to call to targetm.goacc.validate_dims.
1822         (default_goacc_validate_dims): Add used
1823         parameter.
1824         * target.def (validate_dims): Add used parameter in DEFHOOK.
1825         * targhooks.h (default_goacc_validate_dims): Add used parameter.
1826
1827 2019-01-11  Jakub Jelinek  <jakub@redhat.com>
1828
1829         PR middle-end/85956
1830         PR lto/88733
1831         * tree-inline.h (struct copy_body_data): Add adjust_array_error_bounds
1832         field.
1833         * tree-inline.c (remap_type_1): Formatting fix.  If TYPE_MAX_VALUE of
1834         ARRAY_TYPE's TYPE_DOMAIN is newly error_mark_node, replace it with
1835         a dummy "omp dummy var" variable if id->adjust_array_error_bounds.
1836         * omp-low.c (new_omp_context): Set cb.adjust_array_error_bounds.
1837
1838 2019-01-11  Vladimir Makarov  <vmakarov@redhat.com>
1839
1840         PR rtl-optimization/87305
1841         * lra-assigns.c
1842         (setup_live_pseudos_and_spill_after_risky_transforms): Add code
1843         for little endian pseudos used as paradoxical subreg.
1844
1845 2019-01-11  Jakub Jelinek  <jakub@redhat.com>
1846
1847         PR tree-optimization/88693
1848         * tree-ssa-strlen.c (get_min_string_length): Don't set *full_string_p
1849         for STRING_CSTs that don't contain any NUL characters in the first
1850         TREE_STRING_LENGTH bytes.
1851
1852 2019-01-11  Alan Modra  <amodra@gmail.com>
1853
1854         PR 88777
1855         PR 88614
1856         * genattrtab.c (min_fn): Don't translate values.
1857         (min_attr_value): Return INT_MAX when the value can't be calculated.
1858         Return minimum among any values that can be calculated.
1859         (max_attr_value): Adjust.
1860
1861 2019-01-11  Jakub Jelinek  <jakub@redhat.com>
1862
1863         * Makefile.in (PLUGIN_HEADERS): Add $(INSN_ATTR_H).
1864
1865 2019-01-11  Steve Ellcey  <sellcey@marvell.com>
1866
1867         * config/aarch64/aarch64.c (aarch64_simd_call_p): New function.
1868         (aarch64_hard_regno_call_part_clobbered): Add insn argument.
1869         (aarch64_return_call_with_max_clobbers): New function.
1870         (TARGET_RETURN_CALL_WITH_MAX_CLOBBERS): New macro.
1871         * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Add insn
1872         argument.
1873         * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): Ditto.
1874         * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Ditto.
1875         * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered): Ditto.
1876         * config/s390/s390.c (s390_hard_regno_call_part_clobbered): Ditto.
1877         * cselib.c (cselib_process_insn): Add argument to
1878         targetm.hard_regno_call_part_clobbered call.
1879         * ira-conflicts.c (ira_build_conflicts): Ditto.
1880         * ira-costs.c (ira_tune_allocno_costs): Ditto.
1881         * lra-constraints.c (inherit_reload_reg): Ditto.
1882         * lra-int.h (struct lra_reg): Add call_insn field, remove call_p field.
1883         * lra-lives.c (check_pseudos_live_through_calls): Add call_insn
1884         argument.  Call targetm.return_call_with_max_clobbers.
1885         Add argument to targetm.hard_regno_call_part_clobbered call.
1886         (calls_have_same_clobbers_p): New function.
1887         (process_bb_lives): Add call_insn and last_call_insn variables.
1888         Pass call_insn to check_pseudos_live_through_calls.
1889         Modify if stmt to check targetm.return_call_with_max_clobbers.
1890         Update setting of flush variable.
1891         (lra_create_live_ranges_1): Set call_insn to NULL instead of call_p
1892         to false.
1893         * lra.c (initialize_lra_reg_info_element): Set call_insn to NULL.
1894         * regcprop.c (copyprop_hardreg_forward_1): Add argument to
1895         targetm.hard_regno_call_part_clobbered call.
1896         * reginfo.c (choose_hard_reg_mode): Ditto.
1897         * regrename.c (check_new_reg_p): Ditto.
1898         * reload.c (find_equiv_reg): Ditto.
1899         * reload1.c (emit_reload_insns): Ditto.
1900         * sched-deps.c (deps_analyze_insn): Ditto.
1901         * sel-sched.c (init_regs_for_mode): Ditto.
1902         (mark_unavailable_hard_regs): Ditto.
1903         * targhooks.c (default_dwarf_frame_reg_mode): Ditto.
1904         * target.def (hard_regno_call_part_clobbered): Add insn argument.
1905         (return_call_with_max_clobbers): New target function.
1906         * doc/tm.texi: Regenerate.
1907         * doc/tm.texi.in (TARGET_RETURN_CALL_WITH_MAX_CLOBBERS): New hook.
1908         * hooks.c (hook_bool_uint_mode_false): Change to
1909         hook_bool_insn_uint_mode_false.
1910         * hooks.h (hook_bool_uint_mode_false): Ditto.
1911
1912 2019-01-11  Steve Ellcey  <sellcey@marvell.com>
1913
1914         * config/aarch64/aarch64.c (aarch64_simd_call_p): New function.
1915         (aarch64_remove_extra_call_preserved_regs): New function.
1916         (TARGET_REMOVE_EXTRA_CALL_PRESERVED_REGS): New macro.
1917         * doc/tm.texi.in (TARGET_REMOVE_EXTRA_CALL_PRESERVED_REGS): New hook.
1918         * doc/tm.texi: Regenerate.
1919         * final.c (get_call_reg_set_usage): Call new hook.
1920         * target.def (remove_extra_call_preserved_regs): New hook.
1921         * targhooks.c (default_remove_extra_call_preserved_regs): New function.
1922         * targhooks.h (default_remove_extra_call_preserved_regs): New function.
1923
1924 2019-01-11  Jakub Jelinek  <jakub@redhat.com>
1925
1926         PR bootstrap/88714
1927         * passes.c (finish_optimization_passes): Call print_combine_total_stats
1928         inside of pass_combine_1 dump rather than pass_profile_1.
1929
1930 2019-01-11  Tom de Vries  <tdevries@suse.de>
1931
1932         * config/nvptx/nvptx.c (PTX_CTA_NUM_BARRIERS, PTX_PER_CTA_BARRIER)
1933         (PTX_NUM_PER_CTA_BARRIER, PTX_FIRST_PER_WORKER_BARRIER)
1934         (PTX_NUM_PER_WORKER_BARRIERS): Define.
1935         (nvptx_apply_dim_limits): Prevent vector_length 64 and
1936         num_workers 16.
1937
1938 2019-01-11  Tom de Vries  <tdevries@suse.de>
1939
1940         * config/nvptx/nvptx.c (PTX_CTA_SIZE): Move up.
1941
1942 2019-01-11  Jan Beulich  <jbeulich@suse.com>
1943
1944         * config/i386/i386.md (rex64suffix): Add L suffix for SI.
1945         * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>32<round_name>,
1946         sse2_cvtsi2sd): Add {l}.
1947         (sse2_cvtsi2sdq<round_name>): Make q conditional upon AT&T
1948         syntax.
1949
1950 2019-01-10  Jakub Jelinek  <jakub@redhat.com>
1951
1952         PR target/88785
1953         * config/i386/sse.md (float<floatunssuffix>v2div2sf2): Turn into
1954         define_expand.
1955         (*float<floatunssuffix>v2div2sf2): New define_insn.
1956         (float<floatunssuffix>v2div2sf2_mask): Turn into define_expand.
1957         (*float<floatunssuffix>v2div2sf2_mask): New define_insn.
1958         (*float<floatunssuffix>v2div2sf2_mask_1): Replace
1959         subrtxes (const_vector:V2SF [(const_int 0) (const_int 0)]) with
1960         match_operands with "const0_operand" "C".
1961
1962 2019-01-10  Tamar Christina  <tamar.christina@arm.com>
1963
1964         * config/aarch64/aarch64-builtins.c
1965         (aarch64_init_builtins): Move aarch64_init_fcmla_laneq_builtins...
1966         (aarch64_init_simd_builtins): ...Here
1967
1968 2019-01-10  Vladimir Makarov  <vmakarov@redhat.com>
1969
1970         PR rtl-optimization/87305
1971         * lra-assigns.c
1972         (setup_live_pseudos_and_spill_after_risky_transforms): Check
1973         allocation for big endian pseudos used as paradoxical subregs and
1974         spill them if it is wrong.
1975         * lra-constraints.c (lra_constraints): Add a comment.
1976
1977 2019-01-10  Richard Biener  <rguenther@suse.de>
1978
1979         PR tree-optimization/88792
1980         * tree-ssa-pre.c (get_representative_for): Do not return a
1981         value-number here.
1982
1983 2019-01-10  Jakub Jelinek  <jakub@redhat.com>
1984
1985         PR middle-end/84877
1986         PR bootstrap/88450
1987         * function.c (assign_stack_local_1): Revert the 2018-11-21 changes.
1988         (assign_parm_setup_block): Do the argument slot realignment here
1989         instead.
1990
1991 2019-01-10  Stefan Agner  <stefan@agner.ch>
1992
1993         PR target/88648
1994         * config/arm/arm.c (arm_option_override_internal): Force
1995         opts->x_inline_asm_unified to true only if TARGET_THUMB2_P.
1996
1997 2019-01-10  Jakub Jelinek  <jakub@redhat.com>
1998
1999         PR c/88568
2000         * attribs.c (handle_dll_attribute): Clear TREE_STATIC after setting
2001         DECL_EXTERNAL.
2002
2003 2019-01-10  Tamar Christina  <tamar.christina@arm.com>
2004
2005         * config/arm/arm-builtins.c
2006         (enum arm_type_qualifiers): Add qualifier_lane_pair_index.
2007         (MAC_LANE_PAIR_QUALIFIERS): New.
2008         (arm_expand_builtin_args): Use it.
2009         (arm_expand_builtin_1): Likewise.
2010         * config/arm/arm-protos.h (neon_vcmla_lane_prepare_operands): New.
2011         * config/arm/arm.c (neon_vcmla_lane_prepare_operands): New.
2012         * config/arm/arm-c.c (arm_cpu_builtins): Add __ARM_FEATURE_COMPLEX.
2013         * config/arm/arm_neon.h:
2014         (vcadd_rot90_f16): New.
2015         (vcaddq_rot90_f16): New.
2016         (vcadd_rot270_f16): New.
2017         (vcaddq_rot270_f16): New.
2018         (vcmla_f16): New.
2019         (vcmlaq_f16): New.
2020         (vcmla_lane_f16): New.
2021         (vcmla_laneq_f16): New.
2022         (vcmlaq_lane_f16): New.
2023         (vcmlaq_laneq_f16): New.
2024         (vcmla_rot90_f16): New.
2025         (vcmlaq_rot90_f16): New.
2026         (vcmla_rot90_lane_f16): New.
2027         (vcmla_rot90_laneq_f16): New.
2028         (vcmlaq_rot90_lane_f16): New.
2029         (vcmlaq_rot90_laneq_f16): New.
2030         (vcmla_rot180_f16): New.
2031         (vcmlaq_rot180_f16): New.
2032         (vcmla_rot180_lane_f16): New.
2033         (vcmla_rot180_laneq_f16): New.
2034         (vcmlaq_rot180_lane_f16): New.
2035         (vcmlaq_rot180_laneq_f16): New.
2036         (vcmla_rot270_f16): New.
2037         (vcmlaq_rot270_f16): New.
2038         (vcmla_rot270_lane_f16): New.
2039         (vcmla_rot270_laneq_f16): New.
2040         (vcmlaq_rot270_lane_f16): New.
2041         (vcmlaq_rot270_laneq_f16): New.
2042         (vcadd_rot90_f32): New.
2043         (vcaddq_rot90_f32): New.
2044         (vcadd_rot270_f32): New.
2045         (vcaddq_rot270_f32): New.
2046         (vcmla_f32): New.
2047         (vcmlaq_f32): New.
2048         (vcmla_lane_f32): New.
2049         (vcmla_laneq_f32): New.
2050         (vcmlaq_lane_f32): New.
2051         (vcmlaq_laneq_f32): New.
2052         (vcmla_rot90_f32): New.
2053         (vcmlaq_rot90_f32): New.
2054         (vcmla_rot90_lane_f32): New.
2055         (vcmla_rot90_laneq_f32): New.
2056         (vcmlaq_rot90_lane_f32): New.
2057         (vcmlaq_rot90_laneq_f32): New.
2058         (vcmla_rot180_f32): New.
2059         (vcmlaq_rot180_f32): New.
2060         (vcmla_rot180_lane_f32): New.
2061         (vcmla_rot180_laneq_f32): New.
2062         (vcmlaq_rot180_lane_f32): New.
2063         (vcmlaq_rot180_laneq_f32): New.
2064         (vcmla_rot270_f32): New.
2065         (vcmlaq_rot270_f32): New.
2066         (vcmla_rot270_lane_f32): New.
2067         (vcmla_rot270_laneq_f32): New.
2068         (vcmlaq_rot270_lane_f32): New.
2069         (vcmlaq_rot270_laneq_f32): New.
2070         * config/arm/arm_neon_builtins.def (vcadd90, vcadd270, vcmla0, vcmla90,
2071         vcmla180, vcmla270, vcmla_lane0, vcmla_lane90, vcmla_lane180, vcmla_lane270,
2072         vcmla_laneq0, vcmla_laneq90, vcmla_laneq180, vcmla_laneq270,
2073         vcmlaq_lane0, vcmlaq_lane90, vcmlaq_lane180, vcmlaq_lane270): New.
2074         * config/arm/neon.md (neon_vcmla_lane<rot><mode>,
2075         neon_vcmla_laneq<rot><mode>, neon_vcmlaq_lane<rot><mode>): New.
2076         * config/arm/arm.c (arm_arch8_3, arm_arch8_4): New.
2077         * config/arm/arm.h (TARGET_COMPLEX, arm_arch8_3, arm_arch8_4): New.
2078         (arm_option_reconfigure_globals): Use them.
2079         * config/arm/iterators.md (VDF, VQ_HSF): New.
2080         (VCADD, VCMLA): New.
2081         (VF_constraint, rot, rotsplit1, rotsplit2): Add V4HF and V8HF.
2082         * config/arm/neon.md (neon_vcadd<rot><mode>, neon_vcmla<rot><mode>): New.
2083         * config/arm/unspecs.md (UNSPEC_VCADD90, UNSPEC_VCADD270,
2084         UNSPEC_VCMLA, UNSPEC_VCMLA90, UNSPEC_VCMLA180, UNSPEC_VCMLA270): New.
2085
2086 2019-01-10  Tamar Christina  <tamar.christina@arm.com>
2087
2088         * config/aarch64/aarch64-builtins.c (enum aarch64_type_qualifiers): Add qualifier_lane_pair_index.
2089         (emit-rtl.h): Include.
2090         (TYPES_QUADOP_LANE_PAIR): New.
2091         (aarch64_simd_expand_args): Use it.
2092         (aarch64_simd_expand_builtin): Likewise.
2093         (AARCH64_SIMD_FCMLA_LANEQ_BUILTINS, aarch64_fcmla_laneq_builtin_datum): New.
2094         (FCMLA_LANEQ_BUILTIN, AARCH64_SIMD_FCMLA_LANEQ_BUILTIN_BASE,
2095         AARCH64_SIMD_FCMLA_LANEQ_BUILTINS, aarch64_fcmla_lane_builtin_data,
2096         aarch64_init_fcmla_laneq_builtins, aarch64_expand_fcmla_builtin): New.
2097         (aarch64_init_builtins): Add aarch64_init_fcmla_laneq_builtins.
2098         (aarch64_expand_buildin): Add AARCH64_SIMD_BUILTIN_FCMLA_LANEQ0_V2SF,
2099         AARCH64_SIMD_BUILTIN_FCMLA_LANEQ90_V2SF, AARCH64_SIMD_BUILTIN_FCMLA_LANEQ180_V2SF,
2100         AARCH64_SIMD_BUILTIN_FCMLA_LANEQ2700_V2SF, AARCH64_SIMD_BUILTIN_FCMLA_LANEQ0_V4HF,
2101         AARCH64_SIMD_BUILTIN_FCMLA_LANEQ90_V4HF, AARCH64_SIMD_BUILTIN_FCMLA_LANEQ180_V4HF,
2102         AARCH64_SIMD_BUILTIN_FCMLA_LANEQ270_V4HF.
2103         * config/aarch64/iterators.md (FCMLA_maybe_lane): New.
2104         * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Add __ARM_FEATURE_COMPLEX.
2105         * config/aarch64/aarch64-simd-builtins.def (fcadd90, fcadd270, fcmla0, fcmla90,
2106         fcmla180, fcmla270, fcmla_lane0, fcmla_lane90, fcmla_lane180, fcmla_lane270,
2107         fcmla_laneq0, fcmla_laneq90, fcmla_laneq180, fcmla_laneq270,
2108         fcmlaq_lane0, fcmlaq_lane90, fcmlaq_lane180, fcmlaq_lane270): New.
2109         * config/aarch64/aarch64-simd.md (aarch64_fcmla_lane<rot><mode>,
2110         aarch64_fcmla_laneq<rot>v4hf, aarch64_fcmlaq_lane<rot><mode>,aarch64_fcadd<rot><mode>,
2111         aarch64_fcmla<rot><mode>): New.
2112         * config/aarch64/arm_neon.h:
2113         (vcadd_rot90_f16): New.
2114         (vcaddq_rot90_f16): New.
2115         (vcadd_rot270_f16): New.
2116         (vcaddq_rot270_f16): New.
2117         (vcmla_f16): New.
2118         (vcmlaq_f16): New.
2119         (vcmla_lane_f16): New.
2120         (vcmla_laneq_f16): New.
2121         (vcmlaq_lane_f16): New.
2122         (vcmlaq_rot90_lane_f16): New.
2123         (vcmla_rot90_laneq_f16): New.
2124         (vcmla_rot90_lane_f16): New.
2125         (vcmlaq_rot90_f16): New.
2126         (vcmla_rot90_f16): New.
2127         (vcmlaq_laneq_f16): New.
2128         (vcmla_rot180_laneq_f16): New.
2129         (vcmla_rot180_lane_f16): New.
2130         (vcmlaq_rot180_f16): New.
2131         (vcmla_rot180_f16): New.
2132         (vcmlaq_rot90_laneq_f16): New.
2133         (vcmlaq_rot270_laneq_f16): New.
2134         (vcmlaq_rot270_lane_f16): New.
2135         (vcmla_rot270_laneq_f16): New.
2136         (vcmlaq_rot270_f16): New.
2137         (vcmla_rot270_f16): New.
2138         (vcmlaq_rot180_laneq_f16): New.
2139         (vcmlaq_rot180_lane_f16): New.
2140         (vcmla_rot270_lane_f16): New.
2141         (vcadd_rot90_f32): New.
2142         (vcaddq_rot90_f32): New.
2143         (vcaddq_rot90_f64): New.
2144         (vcadd_rot270_f32): New.
2145         (vcaddq_rot270_f32): New.
2146         (vcaddq_rot270_f64): New.
2147         (vcmla_f32): New.
2148         (vcmlaq_f32): New.
2149         (vcmlaq_f64): New.
2150         (vcmla_lane_f32): New.
2151         (vcmla_laneq_f32): New.
2152         (vcmlaq_lane_f32): New.
2153         (vcmlaq_laneq_f32): New.
2154         (vcmla_rot90_f32): New.
2155         (vcmlaq_rot90_f32): New.
2156         (vcmlaq_rot90_f64): New.
2157         (vcmla_rot90_lane_f32): New.
2158         (vcmla_rot90_laneq_f32): New.
2159         (vcmlaq_rot90_lane_f32): New.
2160         (vcmlaq_rot90_laneq_f32): New.
2161         (vcmla_rot180_f32): New.
2162         (vcmlaq_rot180_f32): New.
2163         (vcmlaq_rot180_f64): New.
2164         (vcmla_rot180_lane_f32): New.
2165         (vcmla_rot180_laneq_f32): New.
2166         (vcmlaq_rot180_lane_f32): New.
2167         (vcmlaq_rot180_laneq_f32): New.
2168         (vcmla_rot270_f32): New.
2169         (vcmlaq_rot270_f32): New.
2170         (vcmlaq_rot270_f64): New.
2171         (vcmla_rot270_lane_f32): New.
2172         (vcmla_rot270_laneq_f32): New.
2173         (vcmlaq_rot270_lane_f32): New.
2174         (vcmlaq_rot270_laneq_f32): New.
2175         * config/aarch64/aarch64.h (TARGET_COMPLEX): New.
2176         * config/aarch64/iterators.md (UNSPEC_FCADD90, UNSPEC_FCADD270,
2177         UNSPEC_FCMLA, UNSPEC_FCMLA90, UNSPEC_FCMLA180, UNSPEC_FCMLA270): New.
2178         (FCADD, FCMLA): New.
2179         (rot): New.
2180         * config/arm/types.md (neon_fcadd, neon_fcmla): New.
2181
2182 2019-01-09  Sandra Loosemore  <sandra@codesourcery.com>
2183
2184         PR other/16615
2185
2186         * config/pa/pa.c: Change "can not" to "cannot".
2187         * gimple-ssa-evrp-analyze.c: Likewise.
2188         * ipa-icf.c: Likewise.
2189         * ipa-polymorphic-call.c: Likewise.
2190         * ipa-pure-const.c: Likewise.
2191         * lra-constraints.c: Likewise.
2192         * lra-remat.c: Likewise.
2193         * reload1.c: Likewise.
2194         * reorg.c: Likewise.
2195         * tree-ssa-uninit.c: Likewise.
2196
2197 2019-01-09  Sandra Loosemore  <sandra@codesourcery.com>
2198
2199         PR other/16615
2200
2201         * Makefile.in: Mechanically replace "can not" with "cannot".
2202         * alias.c: Likewise.
2203         * builtins.c: Likewise.
2204         * calls.c: Likewise.
2205         * cgraph.c: Likewise.
2206         * cgraph.h: Likewise.
2207         * cgraphclones.c: Likewise.
2208         * cgraphunit.c: Likewise.
2209         * combine-stack-adj.c: Likewise.
2210         * combine.c: Likewise.
2211         * common/config/i386/i386-common.c: Likewise.
2212         * config/aarch64/aarch64.c: Likewise.
2213         * config/alpha/sync.md: Likewise.
2214         * config/arc/arc.c: Likewise.
2215         * config/arc/predicates.md: Likewise.
2216         * config/arm/arm-c.c: Likewise.
2217         * config/arm/arm.c: Likewise.
2218         * config/arm/arm.h: Likewise.
2219         * config/arm/arm.md: Likewise.
2220         * config/arm/cortex-r4f.md: Likewise.
2221         * config/csky/csky.c: Likewise.
2222         * config/csky/csky.h: Likewise.
2223         * config/darwin-f.c: Likewise.
2224         * config/epiphany/epiphany.md: Likewise.
2225         * config/i386/i386.c: Likewise.
2226         * config/i386/sol2.h: Likewise.
2227         * config/m68k/m68k.c: Likewise.
2228         * config/mcore/mcore.h: Likewise.
2229         * config/microblaze/microblaze.md: Likewise.
2230         * config/mips/20kc.md: Likewise.
2231         * config/mips/sb1.md: Likewise.
2232         * config/nds32/nds32.c: Likewise.
2233         * config/nds32/predicates.md: Likewise.
2234         * config/pa/pa.c: Likewise.
2235         * config/rs6000/e300c2c3.md: Likewise.
2236         * config/rs6000/rs6000.c: Likewise.
2237         * config/s390/s390.h: Likewise.
2238         * config/sh/sh.c: Likewise.
2239         * config/sh/sh.md: Likewise.
2240         * config/spu/vmx2spu.h: Likewise.
2241         * cprop.c: Likewise.
2242         * dbxout.c: Likewise.
2243         * df-scan.c: Likewise.
2244         * doc/cfg.texi: Likewise.
2245         * doc/extend.texi: Likewise.
2246         * doc/fragments.texi: Likewise.
2247         * doc/gty.texi: Likewise.
2248         * doc/invoke.texi: Likewise.
2249         * doc/lto.texi: Likewise.
2250         * doc/md.texi: Likewise.
2251         * doc/objc.texi: Likewise.
2252         * doc/rtl.texi: Likewise.
2253         * doc/tm.texi: Likewise.
2254         * dse.c: Likewise.
2255         * emit-rtl.c: Likewise.
2256         * emit-rtl.h: Likewise.
2257         * except.c: Likewise.
2258         * expmed.c: Likewise.
2259         * expr.c: Likewise.
2260         * fold-const.c: Likewise.
2261         * genautomata.c: Likewise.
2262         * gimple-fold.c: Likewise.
2263         * hard-reg-set.h: Likewise.
2264         * ifcvt.c: Likewise.
2265         * ipa-comdats.c: Likewise.
2266         * ipa-cp.c: Likewise.
2267         * ipa-devirt.c: Likewise.
2268         * ipa-fnsummary.c: Likewise.
2269         * ipa-icf.c: Likewise.
2270         * ipa-inline-transform.c: Likewise.
2271         * ipa-inline.c: Likewise.
2272         * ipa-polymorphic-call.c: Likewise.
2273         * ipa-profile.c: Likewise.
2274         * ipa-prop.c: Likewise.
2275         * ipa-pure-const.c: Likewise.
2276         * ipa-reference.c: Likewise.
2277         * ipa-split.c: Likewise.
2278         * ipa-visibility.c: Likewise.
2279         * ipa.c: Likewise.
2280         * ira-build.c: Likewise.
2281         * ira-color.c: Likewise.
2282         * ira-conflicts.c: Likewise.
2283         * ira-costs.c: Likewise.
2284         * ira-int.h: Likewise.
2285         * ira-lives.c: Likewise.
2286         * ira.c: Likewise.
2287         * ira.h: Likewise.
2288         * loop-invariant.c: Likewise.
2289         * loop-unroll.c: Likewise.
2290         * lower-subreg.c: Likewise.
2291         * lra-assigns.c: Likewise.
2292         * lra-constraints.c: Likewise.
2293         * lra-eliminations.c: Likewise.
2294         * lra-lives.c: Likewise.
2295         * lra-remat.c: Likewise.
2296         * lra-spills.c: Likewise.
2297         * lra.c: Likewise.
2298         * lto-cgraph.c: Likewise.
2299         * lto-streamer-out.c: Likewise.
2300         * postreload-gcse.c: Likewise.
2301         * predict.c: Likewise.
2302         * profile-count.h: Likewise.
2303         * profile.c: Likewise.
2304         * recog.c: Likewise.
2305         * ree.c: Likewise.
2306         * reload.c: Likewise.
2307         * reload1.c: Likewise.
2308         * reorg.c: Likewise.
2309         * resource.c: Likewise.
2310         * rtl.def: Likewise.
2311         * rtl.h: Likewise.
2312         * rtlanal.c: Likewise.
2313         * sched-deps.c: Likewise.
2314         * sched-ebb.c: Likewise.
2315         * sched-rgn.c: Likewise.
2316         * sel-sched-ir.c: Likewise.
2317         * sel-sched.c: Likewise.
2318         * shrink-wrap.c: Likewise.
2319         * simplify-rtx.c: Likewise.
2320         * symtab.c: Likewise.
2321         * target.def: Likewise.
2322         * toplev.c: Likewise.
2323         * tree-call-cdce.c: Likewise.
2324         * tree-cfg.c: Likewise.
2325         * tree-complex.c: Likewise.
2326         * tree-core.h: Likewise.
2327         * tree-eh.c: Likewise.
2328         * tree-inline.c: Likewise.
2329         * tree-loop-distribution.c: Likewise.
2330         * tree-nrv.c: Likewise.
2331         * tree-profile.c: Likewise.
2332         * tree-sra.c: Likewise.
2333         * tree-ssa-alias.c: Likewise.
2334         * tree-ssa-dce.c: Likewise.
2335         * tree-ssa-dom.c: Likewise.
2336         * tree-ssa-forwprop.c: Likewise.
2337         * tree-ssa-loop-im.c: Likewise.
2338         * tree-ssa-loop-ivcanon.c: Likewise.
2339         * tree-ssa-loop-ivopts.c: Likewise.
2340         * tree-ssa-loop-niter.c: Likewise.
2341         * tree-ssa-phionlycprop.c: Likewise.
2342         * tree-ssa-phiopt.c: Likewise.
2343         * tree-ssa-propagate.c: Likewise.
2344         * tree-ssa-threadedge.c: Likewise.
2345         * tree-ssa-threadupdate.c: Likewise.
2346         * tree-ssa-uninit.c: Likewise.
2347         * tree-ssanames.c: Likewise.
2348         * tree-streamer-out.c: Likewise.
2349         * tree.c: Likewise.
2350         * tree.h: Likewise.
2351         * vr-values.c: Likewise.
2352
2353 2019-01-09  Uroš Bizjak  <ubizjak@gmail.com>
2354
2355         * config/i386/i386-protos.h (ix86_expand_xorsign): New prototype.
2356         (ix86_split_xorsign): Ditto.
2357         * config/i386/i386.c (ix86_expand_xorsign): New function.
2358         (ix86_split_xorsign): Ditto.
2359         * config/i386/i386.md (UNSPEC_XORSIGN): New unspec.
2360         (xorsign<mode>3): New expander.
2361         (xorsign<mode>3_1): New insn_and_split pattern.
2362         * config/i386/sse.md (xorsign<mode>3): New expander.
2363
2364 2019-01-09  Eric Botcazou  <ebotcazou@adacore.com>
2365
2366         * config/sparc/sparc.md (*tablejump_sp32): Merge into...
2367         (*tablejump_sp64): Likewise.
2368         (*tablejump<P:mode>): ...this.
2369         (*call_address_sp32): Merge into...
2370         (*call_address_sp64): Likewise.
2371         (*call_address<P:mode>): ...this.
2372         (*call_symbolic_sp32): Merge into...
2373         (*call_symbolic_sp64): Likewise.
2374         (*call_symbolic<P:mode>): ...this.
2375         (call_value): Remove constraint and add predicate.
2376         (*call_value_address_sp32): Merge into...
2377         (*call_value_address_sp64): Likewise.
2378         (*call_value_address<P:mode>): ...this.
2379         (*call_value_symbolic_sp32): Merge into...
2380         (*call_value_symbolic_sp64): Likewise.
2381         (*call_value_symbolic<P:mode>): ...this.
2382         (*sibcall_symbolic_sp32): Merge into...
2383         (*sibcall_symbolic_sp64): Likewise.
2384         (*sibcall_symbolic<P:mode>): ...this.
2385         (sibcall_value): Remove constraint and add predicate.
2386         (*sibcall_value_symbolic_sp32): Merge into...
2387         (*sibcall_value_symbolic_sp64): Likewise.
2388         (*sibcall_value_symbolic<P:mode>): ...this.
2389         (window_save): Minor tweak.
2390         (*branch_sp32): Merge into...
2391         (*branch_sp64): Likewise.
2392         (*branch<P:mode>): ...this.
2393
2394 2019-01-09  Eric Botcazou  <ebotcazou@adacore.com>
2395             James Clarke  <jrtc27@jrtc27.com>
2396
2397         PR target/84010
2398         * config/sparc/sparc.c (sparc_legitimize_tls_address): Only use Pmode
2399         consistently in TLS address generation and adjust code to the renaming
2400         of patterns.  Mark calls to __tls_get_addr as const.
2401         * config/sparc/sparc.md (tgd_hi22): Turn into...
2402         (tgd_hi22<P:mode>): ...this and use Pmode throughout.
2403         (tgd_lo10): Turn into...
2404         (tgd_lo10<P:mode>): ...this and use Pmode throughout.
2405         (tgd_add32): Merge into...
2406         (tgd_add64): Likewise.
2407         (tgd_add<P:mode>): ...this and use Pmode throughout.
2408         (tldm_hi22): Turn into...
2409         (tldm_hi22<P:mode>): ...this and use Pmode throughout.
2410         (tldm_lo10): Turn into...
2411         (tldm_lo10<P:mode>): ...this and use Pmode throughout.
2412         (tldm_add32): Merge into...
2413         (tldm_add64): Likewise.
2414         (tldm_add<P:mode>): ...this and use Pmode throughout.
2415         (tldm_call32): Merge into...
2416         (tldm_call64): Likewise.
2417         (tldm_call<P:mode>): ...this and use Pmode throughout.
2418         (tldo_hix22): Turn into...
2419         (tldo_hix22<P:mode>): ...this and use Pmode throughout.
2420         (tldo_lox10): Turn into...
2421         (tldo_lox10<P:mode>): ...this and use Pmode throughout.
2422         (tldo_add32): Merge into...
2423         (tldo_add64): Likewise.
2424         (tldo_add<P:mode>): ...this and use Pmode throughout.
2425         (tie_hi22): Turn into...
2426         (tie_hi22<P:mode>): ...this and use Pmode throughout.
2427         (tie_lo10): Turn into...
2428         (tie_lo10<P:mode>): ...this and use Pmode throughout.
2429         (tie_ld64): Use DImode throughout.
2430         (tie_add32): Merge into...
2431         (tie_add64): Likewise.
2432         (tie_add<P:mode>): ...this and use Pmode throughout.
2433         (tle_hix22_sp32): Merge into...
2434         (tle_hix22_sp64): Likewise.
2435         (tle_hix22<P:mode>): ...this and use Pmode throughout.
2436         (tle_lox22_sp32): Merge into...
2437         (tle_lox22_sp64): Likewise.
2438         (tle_lox22<P:mode>): ...this and use Pmode throughout.
2439         (*tldo_ldub_sp32): Merge into...
2440         (*tldo_ldub_sp64): Likewise.
2441         (*tldo_ldub<P:mode>): ...this and use Pmode throughout.
2442         (*tldo_ldub1_sp32): Merge into...
2443         (*tldo_ldub1_sp64): Likewise.
2444         (*tldo_ldub1<P:mode>): ...this and use Pmode throughout.
2445         (*tldo_ldub2_sp32): Merge into...
2446         (*tldo_ldub2_sp64): Likewise.
2447         (*tldo_ldub2<P:mode>): ...this and use Pmode throughout.
2448         (*tldo_ldsb1_sp32): Merge into...
2449         (*tldo_ldsb1_sp64): Likewise.
2450         (*tldo_ldsb1<P:mode>): ...this and use Pmode throughout.
2451         (*tldo_ldsb2_sp32): Merge into...
2452         (*tldo_ldsb2_sp64): Likewise.
2453         (*tldo_ldsb2<P:mode>): ...this and use Pmode throughout.
2454         (*tldo_ldub3_sp64): Use DImode throughout.
2455         (*tldo_ldsb3_sp64): Likewise.
2456         (*tldo_lduh_sp32): Merge into...
2457         (*tldo_lduh_sp64): Likewise.
2458         (*tldo_lduh<P:mode>): ...this and use Pmode throughout.
2459         (*tldo_lduh1_sp32): Merge into...
2460         (*tldo_lduh1_sp64): Likewise.
2461         (*tldo_lduh1<P:mode>): ...this and use Pmode throughout.
2462         (*tldo_ldsh1_sp32): Merge into...
2463         (*tldo_ldsh1_sp64): Likewise.
2464         (*tldo_ldsh1<P:mode>): ...this and use Pmode throughout.
2465         (*tldo_lduh2_sp64): Use DImode throughout.
2466         (*tldo_ldsh2_sp64): Likewise.
2467         (*tldo_lduw_sp32): Merge into...
2468         (*tldo_lduw_sp64): Likewise.
2469         (*tldo_lduw<P:mode>): ...this and use Pmode throughout.
2470         (*tldo_lduw1_sp64): Use DImode throughout.
2471         (*tldo_ldsw1_sp64): Likewise.
2472         (*tldo_ldx_sp64): Likewise.
2473         (*tldo_stb_sp32): Merge into...
2474         (*tldo_stb_sp64): Likewise.
2475         (*tldo_stb<P:mode>): ...this and use Pmode throughout.
2476         (*tldo_sth_sp32): Merge into...
2477         (*tldo_sth_sp64): Likewise.
2478         (*tldo_sth<P:mode>): ...this and use Pmode throughout.
2479         (*tldo_stw_sp32): Merge into...
2480         (*tldo_stw_sp64): Likewise.
2481         (*tldo_stw<P:mode>): ...this and use Pmode throughout.
2482         (*tldo_stx_sp64): Use DImode throughout.
2483
2484 2018-01-09  Sudakshina Das  <sudi.das@arm.com>
2485
2486         * config/aarch64/aarch64.c (aarch64_override_options): Add case to
2487         check configure option to set BTI and Return Address Signing.
2488         * configure.ac: Add --enable-standard-branch-protection and
2489         --disable-standard-branch-protection.
2490         * configure: Regenerated.
2491         * doc/install.texi: Document the same.
2492
2493 2018-01-09  Sudakshina Das  <sudi.das@arm.com>
2494             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2495
2496         * config.gcc (aarch64*-*-*): Add aarch64-bti-insert.o.
2497         * config/aarch64/aarch64.h: Update comment for TRAMPOLINE_SIZE.
2498         * config/aarch64/aarch64.c (aarch64_asm_trampoline_template): Update
2499         if bti is enabled.
2500         * config/aarch64/aarch64-bti-insert.c: New file.
2501         * config/aarch64/aarch64-passes.def (INSERT_PASS_BEFORE): Insert bti
2502         pass.
2503         * config/aarch64/aarch64-protos.h (make_pass_insert_bti): Declare the
2504         new bti pass.
2505         * config/aarch64/aarch64.md (unspecv): Add UNSPECV_BTI_NOARG,
2506         UNSPECV_BTI_C, UNSPECV_BTI_J and UNSPECV_BTI_JC.
2507         (bti_noarg, bti_j, bti_c, bti_jc): New define_insns.
2508         * config/aarch64/t-aarch64: Add rule for aarch64-bti-insert.o.
2509
2510 2018-01-09  Sudakshina Das  <sudi.das@arm.com>
2511
2512         * config/aarch64/aarch64-protos.h (aarch64_bti_enabled): Declare.
2513         * config/aarch64/aarch64.c (aarch64_handle_no_branch_protection):
2514         Disable bti for -mbranch-protection=none.
2515         (aarch64_handle_standard_branch_protection): Enable bti for
2516         -mbranch-protection=standard.
2517         (aarch64_handle_bti_protection): Enable bti for "bti" in the string to
2518         -mbranch-protection.
2519         (aarch64_bti_enabled): Check if bti is enabled.
2520         * config/aarch64/aarch64.opt: Declare target variable.
2521         * doc/invoke.texi: Add bti to the -mbranch-protection documentation.
2522
2523 2018-01-09  Sudakshina Das  <sudi.das@arm.com>
2524
2525         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use new
2526         epilogue/prologue scratch registers EP0_REGNUM and EP1_REGNUM.
2527         (aarch64_expand_epilogue): Likewise.
2528         (aarch64_output_mi_thunk): Likewise
2529         * config/aarch64/aarch64.h (REG_CLASS_CONTENTS): Change
2530         TAILCALL_ADDR_REGS to x16 and x17.
2531         * config/aarch64/aarch64.md: Define EP0_REGNUM and EP1_REGNUM.
2532
2533 2018-01-09  Sudakshina Das  <sudi.das@arm.com>
2534
2535         * config/aarch64/aarch64-option-extensions.def: Define
2536         AARCH64_OPT_EXTENSION for memtag, rng, sb, ssbs and predres.
2537         * config/aarch64/aarch64.h (AARCH64_FL_RNG): New.
2538         (AARCH64_FL_MEMTAG, ARCH64_FL_SB, AARCH64_FL_SSBS): New.
2539         (AARCH64_FL_PREDRES): New.
2540         (AARCH64_FL_FOR_ARCH8_5): Add AARCH64_FL_SB, AARCH64_FL_SSBS and
2541         AARCH64_FL_PREDRES by default.
2542         * doc/invoke.texi: Document rng, memtag, sb, ssbs and predres.
2543
2544 2018-01-09  Sudakshina Das  <sudi.das@arm.com>
2545
2546         * config/aarch64/aarch64-arches.def: Define AARCH64_ARCH for
2547         ARMv8.5-A.
2548         * config/aarch64/aarch64.h (AARCH64_FL_V8_5): New.
2549         (AARCH64_FL_FOR_ARCH8_5, AARCH64_ISA_V8_5): New.
2550         * doc/invoke.texi: Document ARMv8.5-A.
2551
2552 2019-01-09  Alejandro Martinez  <alejandro.martinezvicente@arm.com>
2553
2554         * config/aarch64/aarch64-sve.md (copysign<mode>3): New define_expand.
2555         (xorsign<mode>3): Likewise.
2556
2557 2019-01-09  Jelinek  <jakub@redhat.com>
2558
2559         PR middle-end/88758
2560         * tree.c (initializer_each_zero_or_onep) <case VECTOR_CST>: Use
2561         vector_cst_elt instead of VECTOR_CST_ENCODED_ELT.
2562
2563         PR rtl-optimization/88331
2564         * function.c (assign_stack_local_1): Don't set dynamic_align_addr if
2565         not currently_expanding_to_rtl.
2566
2567 2019-01-09  Eric Botcazou  <ebotcazou@adacore.com>
2568
2569         * doc/invoke.texi (-Os): Remove trailing spaces.
2570         (-finline-functions): Remove reference to -O2.
2571
2572 2019-01-08  Jakub Jelinek  <jakub@redhat.com>
2573
2574         PR rtl-optimization/79593
2575         * config/i386/i386.md (reg = mem; mem = reg): New define_peephole2.
2576
2577         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Delegitimize
2578         UNSPEC_FUSION_GPR to its argument.  Formatting fixes.
2579
2580 2019-01-08  Eric Botcazou  <ebotcazou@adacore.com>
2581
2582         PR bootstrap/88721
2583         * config/sparc/sparc.c (function_arg_slotno): Set *PPREGNO & *PPADDING
2584         to -1 on entry.
2585
2586         PR debug/88723
2587         * config/sparc/sparc.c (sparc_delegitimize_address): Deal with naked
2588         UNSPECs and UNSPEC_MOVE_GOTDATA specifically.
2589
2590 2019-01-08  H.J. Lu  <hongjiu.lu@intel.com>
2591
2592         PR target/88717
2593         * config/i386/i386.c (ix86_avx_u128_mode_exit): Call
2594         ix86_avx_u128_mode_entry.
2595
2596 2019-01-08  Martin Liska  <mliska@suse.cz>
2597
2598         PR tree-optimization/88753
2599         * tree-switch-conversion.c (switch_conversion::build_one_array):
2600         Come up with local variable constructor.  Convert first to
2601         type of constructor values.
2602
2603 2019-01-08  Richard Biener  <rguenther@suse.de>
2604
2605         PR tree-optimization/86554
2606         * tree-ssa-sccvn.c (eliminate_dom_walker, rpo_elim,
2607         rpo_avail): Move earlier.
2608         (visit_nary_op): When value-numbering to expressions
2609         with different overflow behavior make sure there's an
2610         available expression on the path.
2611
2612 2019-01-08  Sam Tebbs  <sam.tebbs@arm.com>
2613
2614         * config/aarch64/aarch64.c (BRANCH_PROTECT_STR_MAX,
2615         aarch64_parse_branch_protection,
2616         struct aarch64_branch_protect_type,
2617         aarch64_handle_no_branch_protection,
2618         aarch64_handle_standard_branch_protection,
2619         aarch64_validate_mbranch_protection,
2620         aarch64_handle_pac_ret_protection,
2621         aarch64_handle_attr_branch_protection,
2622         accepted_branch_protection_string,
2623         aarch64_pac_ret_subtypes,
2624         aarch64_branch_protect_types,
2625         aarch64_handle_pac_ret_leaf): Define.
2626         (aarch64_override_options_after_change_1, aarch64_override_options):
2627         Add check for accepted_branch_protection_string.
2628         (aarch64_option_save): Save accepted_branch_protection_string.
2629         (aarch64_option_restore): Save accepted_branch_protection_string.
2630         * config/aarch64/aarch64.c (aarch64_attributes): Add branch-protection.
2631         * config/aarch64/aarch64.opt: Add mbranch-protection. Deprecate
2632         msign-return-address.
2633         * doc/invoke.texi: Add mbranch-protection.
2634
2635 2019-01-08  Alan Modra  <amodra@gmail.com>
2636
2637         PR target/88614
2638         * genattrtab.c (max_attr_value, min_attr_value, or_attr_value):
2639         Delete "unknownp" parameter.  Adjust callers.  Handle
2640         CONST_INT, PLUS, MINUS, and MULT.
2641         (attr_value_aligned): Renamed from or_attr_value.
2642         (min_attr_value): Return INT_MIN for unhandled rtl case..
2643         (min_fn): ..and translate to INT_MAX here.
2644         (write_length_unit_log): Modify to cope without "unknown".
2645         (write_attr_value): Handle IF_THEN_ELSE.
2646
2647 2019-01-07  Richard Sandiford  <richard.sandiford@arm.com>
2648
2649         * tree-vect-stmts.c (vectorizable_store): Don't use the dataref_offset
2650         optimization for masked stores.
2651
2652 2019-01-07  Richard Sandiford  <richard.sandiford@arm.com>
2653
2654         PR middle-end/88567
2655         * tree-vect-loop.c (get_initial_defs_for_reduction): Pass the
2656         output vector directly to duplicate_and_interleave instead of
2657         going through a temporary.  Postpone insertion of ctor_seq to
2658         the end of the loop.
2659
2660 2019-01-07  Richard Earnshaw  <rearnsha@arm.com>
2661
2662         PR target/86891
2663         * config/aarch64/aarch64.c (aarch64_expand_subvti): New parameter
2664         unsigned_p.  Handle signed and unsigned overflow correction as
2665         required.
2666         * config/aarch64/aarch64-protos.h (aarch64_expand_subvti): Update
2667         prototype.
2668         * config/aarch64/aarch64.md (addv<mode>4): Use aarch64_plus_operand
2669         for operand 2.
2670         (add<mode>3_compareV_imm): Make this callable for expanding.
2671         (subv<GPI:mode>4): Use register_operand for operand 1.  Use
2672         aarch64_plus_operand for operand 2.
2673         (subv<GPI:mode>_insn): New insn pattern.
2674         (subv<GPI:mode>_imm): Likewise.
2675         (negv<GPI:mode>3): New expand pattern.
2676         (negv<GPI:mode>_insn): New insn pattern.
2677         (negv<GPI:mode>_cmp_only): Likewise.
2678         (cmpv<GPI:mode>_insn): Likewise.
2679         (subvti4): Use register_operand for operand 1.  Update call to
2680         aarch64_expand_subvti.
2681         (usubvti4): Likewise.
2682         (negvti3): New expand pattern.
2683         (negdi_carryout): New insn pattern.
2684         (negvdi_carryinV): New insn pattern.
2685         (sub<mode3>_compare1_imm): Delete named insn pattern, make anonymous
2686         version the named version.
2687         (peepholes to convert to sub<mode3>_compare1_imm): Adjust order of
2688         operands.
2689         (usub<GPI:mode>3_carryinC, usub<GPI:mode>3_carryinC_z1): New insn
2690         patterns.
2691         (usub<GPI:mode>3_carryinC_z2, usub<GPI:mode>3_carryinC): New insn
2692         patterns.
2693         (sub<mode>3_carryinCV, sub<mode>3_carryinCV_z1_z2): Delete.
2694         (sub<mode>3_carryinCV_z1, sub<mode>3_carryinCV_z2): Delete.
2695         (sub<mode>3_carryinCV): Delete.
2696         (sub<GPI:mode>3_carryinV): New expand pattern.
2697         sub<mode>3_carryinV, sub<mode>3_carryinV_z2): New insn patterns.
2698
2699 2019-01-07  Richard Biener  <rguenther@suse.de>
2700
2701         * tree-ssa-uncprop.c (ssa_equip_hash_traits): Remove in favor
2702         of tree_operand_hash.
2703
2704 2019-01-07  Richard Sandiford  <richard.sandiford@arm.com>
2705
2706         PR tree-optimization/88598
2707         * tree.h (single_nonzero_element): Declare.
2708         * tree.c (single_nonzero_element): New function.
2709         * match.pd: Fold certain reductions of X & CST to X[I] & CST[I]
2710         if I is the only nonzero element of CST.
2711
2712 2019-01-07  Richard Sandiford  <richard.sandiford@arm.com>
2713
2714         PR tree-optimization/88598
2715         * tree.h (initializer_each_zero_or_onep): Declare.
2716         * tree.c (initializer_each_zero_or_onep): New function.
2717         (signed_or_unsigned_type_for): Handle float types too.
2718         (unsigned_type_for, signed_type_for): Update comments accordingly.
2719         * match.pd: Fold x * { 0 or 1, 0 or 1, ...} to
2720         x & { 0 or -1, 0 or -1, ... }.
2721
2722 2019-01-07  Jonathan Wakely  <jwakely@redhat.com>
2723
2724         * doc/install.texi: Replace references to x86_64-unknown-linux-gnu
2725         with x86_64-pc-linux-gnu.
2726
2727 2019-01-07  Tom de Vries  <tdevries@suse.de>
2728
2729         PR target/85486
2730         * config/nvptx/nvptx.c (has_vector_partitionable_routine_calls_p): New
2731         function.
2732         (nvptx_goacc_validate_dims): Force vl32 if calling vector-partitionable
2733         routines.
2734
2735 2019-01-07  Jakub Jelinek  <jakub@redhat.com>
2736
2737         * config/i386/sse.md (vec_extract<mode><ssehalfvecmodelower>): Use
2738         V_256_512 iterator instead of V_512 and TARGET_AVX instead of
2739         TARGET_AVX512F as condition.
2740
2741         PR debug/88723
2742         * dwarf2out.c (const_ok_for_output_1): Remove redundant call to
2743         const_not_ok_for_debug_p target hook.
2744         (mem_loc_descriptor) <case UNSPEC>: Only call const_ok_for_output_1
2745         on UNSPEC and subexpressions thereof if all subexpressions of the
2746         UNSPEC are CONSTANT_P.
2747
2748         PR tree-optimization/88676
2749         * tree-ssa-phiopt.c (two_value_replacement): New function.
2750         (tree_ssa_phiopt_worker): Call it.
2751
2752         PR sanitizer/88619
2753         * cfgexpand.c (expand_stack_vars): Only align prev_offset to
2754         ASAN_MIN_RED_ZONE_SIZE, not to maximum of that and alignb.
2755
2756         PR c++/85052
2757         * tree-vect-generic.c: Include insn-config.h and recog.h.
2758         (expand_vector_piecewise): Add defaulted ret_type argument,
2759         if non-NULL, use that in preference to type for the result type.
2760         (expand_vector_parallel): Formatting fix.
2761         (do_vec_conversion, do_vec_narrowing_conversion,
2762         expand_vector_conversion): New functions.
2763         (expand_vector_operations_1): Call expand_vector_conversion
2764         for VEC_CONVERT ifn calls.
2765         * internal-fn.def (VEC_CONVERT): New internal function.
2766         * internal-fn.c (expand_VEC_CONVERT): New function.
2767         * fold-const-call.c (fold_const_vec_convert): New function.
2768         (fold_const_call): Use it for CFN_VEC_CONVERT.
2769         * doc/extend.texi (__builtin_convertvector): Document.
2770
2771 2019-01-07  Tom de Vries  <tdevries@suse.de>
2772
2773         * config/nvptx/nvptx-protos.h (nvptx_output_red_partition): Declare.
2774         * config/nvptx/nvptx.c (vector_red_size, vector_red_align,
2775         vector_red_partition, vector_red_sym): New global variables.
2776         (nvptx_option_override): Initialize vector_red_sym.
2777         (nvptx_declare_function_name): Restore red_partition register.
2778         (nvptx_file_end): Emit code to declare the vector reduction variables.
2779         (nvptx_output_red_partition): New function.
2780         (nvptx_expand_shared_addr): Add vector argument. Use it to handle
2781         large vector reductions.
2782         (enum nvptx_builtins): Add NVPTX_BUILTIN_VECTOR_ADDR.
2783         (nvptx_init_builtins): Add VECTOR_ADDR.
2784         (nvptx_expand_builtin): Update call to nvptx_expand_shared_addr.
2785         Handle nvptx_expand_shared_addr.
2786         (nvptx_get_shared_red_addr): Add vector argument and handle large
2787         vectors.
2788         (nvptx_goacc_reduction_setup): Add offload_attrs argument and handle
2789         large vectors.
2790         (nvptx_goacc_reduction_init): Likewise.
2791         (nvptx_goacc_reduction_fini): Likewise.
2792         (nvptx_goacc_reduction_teardown): Likewise.
2793         (nvptx_goacc_reduction): Update calls to nvptx_goacc_reduction_{setup,
2794         init,fini,teardown}.
2795         (nvptx_init_axis_predicate): Initialize vector_red_partition.
2796         (nvptx_set_current_function): Init vector_red_partition.
2797         * config/nvptx/nvptx.md (UNSPECV_RED_PART): New unspecv.
2798         (nvptx_red_partition): New insn.
2799         * config/nvptx/nvptx.h (struct machine_function): Add red_partition.
2800
2801 2019-01-07  Tom de Vries  <tdevries@suse.de>
2802
2803         PR target/85381
2804         * config/nvptx/nvptx.c (nvptx_process_pars): Don't emit barriers for
2805         empty loops.
2806
2807 2019-01-07  Tom de Vries  <tdevries@suse.de>
2808
2809         * config/nvptx/nvptx.c (oacc_bcast_partition): Declare.
2810         (nvptx_option_override): Init oacc_bcast_partition.
2811         (nvptx_init_oacc_workers): New function.
2812         (nvptx_declare_function_name): Call nvptx_init_oacc_workers.
2813         (nvptx_needs_shared_bcast): New function.
2814         (nvptx_find_par): Generalize to enable vectors to use shared-memory
2815         to propagate state.
2816         (nvptx_shared_propagate): Initialize vector bcast partition and
2817         synchronization state.
2818         (nvptx_single):  Generalize to enable vectors to use shared-memory
2819         to propagate state.
2820         (nvptx_process_pars): Likewise.
2821         (nvptx_set_current_function): Initialize oacc_broadcast_partition.
2822         * config/nvptx/nvptx.h (struct machine_function): Add
2823         bcast_partition and sync_bar members.
2824
2825 2019-01-07  Tom de Vries  <tdevries@suse.de>
2826
2827         * config/nvptx/nvptx.c (nvptx_welformed_vector_length_p)
2828         (nvptx_apply_dim_limits): New function.
2829         (nvptx_goacc_validate_dims_1): Allow PTX_MAX_VECTOR_LENGTH larger than
2830         PTX_WARP_SIZE.
2831
2832 2019-01-07  Tom de Vries  <tdevries@suse.de>
2833
2834         * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1): Move warnings to
2835         as late as possible.
2836
2837 2019-01-07  Tom de Vries  <tdevries@suse.de>
2838
2839         * config/nvptx/nvptx.c (PTX_VECTOR_LENGTH): Remove.
2840         (PTX_DEFAULT_VECTOR_LENGTH, PTX_MAX_VECTOR_LENGTH): Define.
2841         (nvptx_goacc_validate_dims_1, nvptx_dim_limit)
2842         (nvptx_goacc_reduction_fini): Use PTX_DEFAULT_VECTOR_LENGTH,
2843         PTX_MAX_VECTOR_LENGTH and PTX_WARP_SIZE instead of PTX_VECTOR_LENGTH.
2844
2845 2019-01-07  Tom de Vries  <tdevries@suse.de>
2846
2847         * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Add asserts.
2848
2849 2019-01-07  Tom de Vries  <tdevries@suse.de>
2850
2851         * omp-offload.c (oacc_get_min_dim): New function.
2852         * omp-offload.h (oacc_get_min_dim): Declare.
2853
2854 2018-12-26  Mateusz B  <mateuszb@poczta.onet.pl>
2855
2856         PR target/88521
2857         * config/i386/i386.c (function_value_ms_64): Return small sturct in
2858         AX_REG and float/double in FIRST_SSE_REG for 4 or 8 byte modes.
2859
2860 2019-01-05  Jan Hubicka  <hubicka@ucw.cz>
2861
2862         PR tree-opt/86020
2863         Revert:
2864         2017-05-22  Jan Hubicka  <hubicka@ucw.cz>
2865
2866         * ipa-inline.c (edge_badness): Use inlined_time instead of
2867         inline_summaries->get.
2868
2869 2019-01-05  Jan Hubicka  <hubicka@ucw.cz>
2870
2871         * opts.c (enable_fdo_optimizations): Enable
2872         version-loops-for-strides, loop-interchange, unrol-and-jam
2873         and tree-loop-distribution.
2874         * invoke.texi: Document newly enabled options.
2875
2876 2019-01-05  Jan Hubicka  <hubicka@ucw.cz>
2877
2878         * doc/invoke.texi (max-inline-insns-small): New parameters.
2879         * ipa-inline.c (want_early_inline_function_p): simplify.
2880         (want_inline_small_function_p): Fix pasto from previous patch;
2881         use max-inline-insns-small bound.
2882         * params.def (max-inline-insns-small): New param.
2883         * ipa-fnsummary.c (analyze_function_body): Initialize time/size
2884         variables correctly.
2885
2886 2019-01-05  Jan Hubicka  <hubicka@ucw.cz>
2887
2888         * doc/invoke.texi: Document max-inline-insns-size,
2889         uninlined-function-insns, uninlined-function-time,
2890         uninlined-thunk-insns and uninlined-thunk-time.
2891         * params.def: Add max-inline-insns-size,
2892         uninlined-function-insns, uninlined-function-time,
2893         uninlined-thunk-insns and uninlined-thunk-time.
2894         * ipa-fnsummary.c (compute_fn_summary, analyze_function_body): Use
2895         new parameters.
2896         * ipa-inline.c (can_inline_edge_by_limits_p,
2897         want_inline_small_function_p): Use new parameters.
2898
2899 2019-01-05  Jan Hubicka  <hubicka@ucw.cz>
2900
2901         * ipa-fnsummary.c (analyze_function_body): Fix accounting of time.
2902
2903 2019-01-05  Jakub Jelinek  <jakub@redhat.com>
2904
2905         PR middle-end/82564
2906         PR target/88620
2907         * expr.c (expand_assignment): For calls returning VLA structures
2908         if to_rtx is not a MEM, force it into a stack temporary.
2909
2910         PR debug/88635
2911         * dwarf2out.c (const_ok_for_output_1): Reject MINUS that contains
2912         SYMBOL_REF, CODE_LABEL or UNSPEC in subexpressions of second argument.
2913         Reject PLUS that contains SYMBOL_REF, CODE_LABEL or UNSPEC in
2914         subexpressions of both operands.
2915         (mem_loc_descriptor): Handle UNSPEC if target hook acks it and all the
2916         subrtxes are CONSTANT_P.
2917         * config/i386/i386.c (ix86_const_not_ok_for_debug_p): Revert
2918         2018-11-09 changes.
2919
2920 2019-01-04  Jan Hubicka  <hubicka@ucw.cz>
2921
2922         * params.def (hot-bb-count-ws-permille): Set to 990.
2923
2924 2019-01-04  Martin Sebor  <msebor@redhat.com>
2925
2926         PR c/88546
2927         * attribs.c (decls_mismatched_attributes): Avoid warning for attribute
2928         leaf.
2929
2930 2019-01-04  Martin Sebor  <msebor@redhat.com>
2931
2932         PR c/88363
2933         * doc/extend.texi (attribute alloc_align, alloc_size): Update.
2934
2935 2019-01-04  Jakub Jelinek  <jakub@redhat.com>
2936
2937         * gdbinit.in: Turn off pagination for the skip commands, restore
2938         it to previous state afterwards.
2939
2940 2019-01-04  Jakub Jelinek  <jakub@redhat.com>
2941
2942         PR target/88594
2943         * config/i386/i386.c (ix86_expand_divmod_libfunc): Use mode instead
2944         of GET_MODE (opN) as modes of the libcall arguments.
2945
2946 2019-01-04  Jan Beulich  <jbeulich@suse.com>
2947
2948         * config/i386/sse.md
2949         (<avx512>_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>,
2950         <avx512>_cmp<mode>3<mask_scalar_merge_name>,
2951         <avx512>_ucmp<mode>3<mask_scalar_merge_name>,
2952         <avx512>_ucmp<mode>3<mask_scalar_merge_name>,
2953         avx512f_vmcmp<mode>3<round_saeonly_name>,
2954         avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
2955         avx512f_maskcmp<mode>3,
2956         <avx512>_cvt<ssemodesuffix>2mask<mode>,
2957         <avx512>_cvt<ssemodesuffix>2mask<mode>,
2958         *<avx512>_cvtmask2<ssemodesuffix><mode>,
2959         *<avx512>_cvtmask2<ssemodesuffix><mode>,
2960         <avx512>_eq<mode>3<mask_scalar_merge_name>_1,
2961         <avx512>_eq<mode>3<mask_scalar_merge_name>_1,
2962         <avx512>_gt<mode>3<mask_scalar_merge_name>,
2963         <avx512>_gt<mode>3<mask_scalar_merge_name>,
2964         <avx512>_testm<mode>3<mask_scalar_merge_name>,
2965         <avx512>_testnm<mode>3<mask_scalar_merge_name>,
2966         *<avx512>_testm<VI1248_AVX512VLBW:mode>3_zext,
2967         *<avx512>_testm<VI1248_AVX512VLBW:mode>3_zext_mask,
2968         *<avx512>_testnm<VI1248_AVX512VLBW:mode>3_zext,
2969         *<avx512>_testnm<VI1248_AVX512VLBW:mode>3_zext_mask,
2970         avx512cd_maskb_vec_dup<mode>,
2971         avx512cd_maskw_vec_dup<mode>,
2972         avx512dq_fpclass<mode><mask_scalar_merge_name>,
2973         avx512dq_vmfpclass<mode>,
2974         avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): Use =k
2975         instead of =Yk.
2976
2977 2019-01-03  Martin Sebor  <msebor@redhat.com>
2978
2979         PR tree-optimization/88659
2980         * calls.c (maybe_warn_nonstring_arg): Avoid assuming maxlen is set.
2981
2982 2019-01-03  Aaron Sawdey  <acsawdey@linux.ibm.com>
2983
2984         * config/rs6000/rs6000-string.c (expand_block_move): Don't use
2985         unaligned vsx and avoid lxvd2x/stxvd2x.
2986         (gen_lvx_v4si_move): New function.
2987
2988 2019-01-03  Tom de Vries  <tdevries@suse.de>
2989
2990         * config/nvptx/nvptx.c (MACH_VECTOR_LENGTH, MACH_MAX_WORKERS): Define.
2991         (init_axis_dim, nvptx_mach_max_workers, nvptx_mach_vector_length): New
2992         function.
2993         * config/nvptx/nvptx.h (struct machine_function): Add axis_dims.
2994
2995 2019-01-03  Tom de Vries  <tdevries@suse.de>
2996
2997         * config/nvptx/nvptx.c (struct offload_attrs): New.
2998         (populate_offload_attrs): New function.  Factor mask extraction out of
2999         nvptx_reorg.  Add extraction of dimensions.
3000         (nvptx_reorg): Use populate_offload_attrs.
3001
3002 2019-01-03  Tom de Vries  <tdevries@suse.de>
3003
3004         * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1): Add early-out
3005         cases for oacc_min_dims_p and routine_p.  Add asserts for
3006         oacc_default_dims_p and offload_region_p.
3007
3008 2019-01-03  Tom de Vries  <tdevries@suse.de>
3009
3010         * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1): New function,
3011         factored out of ...
3012         (nvptx_goacc_validate_dims): ... here.
3013
3014 2019-01-03  Jan Hubicka  <hubicka@ucw.cz>
3015
3016         PR tree-optimization/85574
3017         * tree-ssa-uncprop.c (struct equiv_hash_elt): Remove unused
3018         structure.
3019         (struct ssa_equip_hash_traits): Declare.
3020         (val_ssa_equiv): Use custom hash traits using operand_equal_p.
3021
3022 2019-01-03  Jakub Jelinek  <jakub@redhat.com>
3023
3024         PR debug/88644
3025         * dwarf2out.c (modified_type_die): If type is equal to sizetype,
3026         change it to qualified_type.
3027
3028 2019-01-03  Jan Hubicka  <hubicka@ucw.cz>
3029
3030         * ipa-utils.c (scale_ipa_profile_for_fn): Break out from ...
3031         (ipa_merge_profiles): ... here; do not ICE on thunks and aliases.
3032
3033 2019-01-02  Martin Sebor  <msebor@redhat.com>
3034             Jeff Law  <law@redhat.com>
3035
3036         * gimple-fold.c (strlen_range_kind): Remove SRK_LENRANGE_2.
3037         (get_range_strlen_tree): Update appropriately.
3038         (get_range_strlen)
3039         * gimple-fold.h (get_range_strlen): Drop unused last argument.
3040
3041         * gimple-fold.c (gimple_fold_builtin_strlen): Use set_strlen_range
3042         rather than set_range_info.
3043         * tree-ssa-strlen.c (set_strlen_range): Extracted from
3044         maybe_set_strlen_range.  Handle potentially boundary crossing
3045         cases more conservatively.
3046         (maybe_set_strlen_range): Parts refactored into set_strlen_range.
3047         Call set_strlen_range.
3048         * tree-ssa-strlen.h (set_strlen_range): Add prototype.
3049
3050         PR middle-end/88663
3051         * gimple-fold.c (get_range_strlen): Update prototype to no longer
3052         need the flexp argument.
3053         (get_range_strlen_tree): Drop flexp argument.  Drop flexp argument
3054         from calls to get_range_strlen.  Update comments.  Just update
3055         VAL for an unterminated const char array and let the reset of the
3056         code handle it normally.  No longer try to set *flexp.  Adjust
3057         return value.
3058         (get_range_strlen): Update for the new get_range_strlen API.
3059         (get_maxval_strlen): Similarly.
3060         (gimple_fold_builtin_strlen): Handle update meaning of return value
3061         from get_range_strlen.
3062         * gimple-ssa-sprintf.c (get_string_length): Update for the new
3063         get_range_strlen API.
3064
3065 2019-01-02  Jan Hubicka  <hubicka@ucw.cz>
3066
3067         PR lto/88130
3068         * varpool.c (varpool_node::ctor_useable_for_folding_p): Also return
3069         false at WPA time when body was removed.
3070
3071 2019-01-02  Martin Liska  <mliska@suse.cz>
3072
3073         PR tree-optimization/88650
3074         * predict.c (set_even_probabilities): Calculate probability
3075         remainer only when really used.
3076
3077 2019-01-02  Richard Biener  <rguenther@suse.de>
3078
3079         PR middle-end/88651
3080         * tree-data-ref.c (analyze_subscript_affine_affine): Use
3081         widest_ints when mangling max_stmt_execution results.
3082
3083 2019-01-02  Richard Biener  <rguenther@suse.de>
3084
3085         PR tree-optimization/88621
3086         * tree-ssa-loop-im.c (gather_mem_refs_stmt): Fix pastos, avoid
3087         bitfields when canoncalizing.
3088
3089 2019-01-02  Richard Biener  <rguenther@suse.de>
3090
3091         PR target/87545
3092         * config/i386/x86-tune-costs.h (intel_cost): Adjust
3093         cost of cheap SSE instruction.
3094
3095 2019-01-02  Richard Biener  <rguenther@suse.de>
3096
3097         PR ipa/85574
3098         * ipa-icf.h (sem_item_optimizer::sort_congruence_split): Declare.
3099         * ipa-icf.c (sem_item_optimizer::sort_congruence_split): New
3100         function.
3101         (sem_item_optimizer::do_congruence_step_f): Sort the congruence
3102         set after UIDs before splitting them.
3103
3104 2019-01-01  Martin Sebor  <msebor@redhat.com>
3105             Jeff Law  <law@redhat.com>
3106
3107         * gimple-fold.c (get_range_strlen_tree): Record if the computed
3108         length is optimistic.  If it is, then arrange to compute the
3109         conservative length as well.
3110
3111         * gimple-fold.h (get_range_strlen): Update prototype.
3112         * builtins.c (check_access): Update call to get_range_strlen to use
3113         c_strlen_data pointer.   Change various variable accesses to instead
3114         pull data from the c_strlen_data structure.
3115         (check_strncat_sizes, expand_builtin_strncat): Likewise.
3116         * calls.c (maybe_warn_nonstring_arg): Likewise.
3117         * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Likewise.  Reset
3118         minimum length if maximum lengh is unknown.
3119         * gimple-ssa-sprintf.c (get_string_length): Likewise.  Drop code
3120         that used c_strlen, it's no longer needed.  Restructure slightly.
3121         (format_string): Set unlikely range appropriately.
3122         * gimple-fold.c (get_range_strlen): Update comments.  Fix minor
3123         formatting issues.
3124         (get_range_strlen):  Accept c_strlen_data pointer for external
3125         call sites as well.  Pass through to call to internal get_range_strlen.
3126         Adjust minlen, maxlen and maxbound as needed.
3127         (get_maxval_strlen): Update comments.
3128         (gimple_fold_builtin_strlen):  Update call to get_range_strlen
3129         to use c_strlen_data pointer.  Change variable accesses to instead
3130         use c_strlen_data data members.
3131
3132         * gimple-fold.c (get_range_strlen): Update prototype.
3133         (get_range_strlen_tree): Update prototype.  Drop minlen/maxlen
3134         local variables.  Use pdata to return information to caller.
3135         Update calls to get_range_strlen.  Update pdata->maxbound.
3136         (get_range_strlen -- static version): Similarly.
3137         (get_range_strlen -- extern version): Update for internal
3138         get_range_strlen API change.  Convert to external data format.
3139         (get_maxval_strlen): Similarly.
3140
3141 2019-01-01  Jan Hubicka  <hubicka@ucw.cz>
3142
3143         * coverage.c (get_coverage_counts): Use current_function_decl.
3144         * profile.c (read_thunk_profile): New function.
3145         (branch_prob): Add THUNK parameter.
3146         * tree-profile.c (tree_profiling): Handle thunks.
3147         * value-prof.c (init_node_map): Handle thunks.
3148         * value-prof.h (branch_prob): Upate prototype.
3149         (read_thunk_profile): Declare.
3150
3151 2019-01-01  Jakub Jelinek  <jakub@redhat.com>
3152
3153         Update copyright years.
3154
3155         * gcc.c (process_command): Update copyright notice dates.
3156         * gcov-dump.c (print_version): Ditto.
3157         * gcov.c (print_version): Ditto.
3158         * gcov-tool.c (print_version): Ditto.
3159         * gengtype.c (create_file): Ditto.
3160         * doc/cpp.texi: Bump @copying's copyright year.
3161         * doc/cppinternals.texi: Ditto.
3162         * doc/gcc.texi: Ditto.
3163         * doc/gccint.texi: Ditto.
3164         * doc/gcov.texi: Ditto.
3165         * doc/install.texi: Ditto.
3166         * doc/invoke.texi: Ditto.
3167 \f
3168 Copyright (C) 2019 Free Software Foundation, Inc.
3169
3170 Copying and distribution of this file, with or without modification,
3171 are permitted in any medium without royalty provided the copyright
3172 notice and this notice are preserved.