gcc/ChangeLog

   1 2014-02-27  Vladimir Makarov  <vmakarov@redhat.com>
   2
   3         PR target/59222
   4         * lra.c (lra_emit_add): Check SUBREG too.
   5
   6 2014-02-27  Andreas Schwab  <schwab@suse.de>
   7
   8         * config/m68k/m68k.c (m68k_option_override): Disable
   9         -flive-range-shrinkage for classic m68k.
  10         (m68k_override_options_after_change): Likewise.
  11
  12 2014-02-27  Marek Polacek  <polacek@redhat.com>
  13
  14         PR middle-end/59223
  15         * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
  16         -Wmaybe-uninitialized.
  17
  18 2014-02-27  Alan Modra  <amodra@gmail.com>
  19
  20         PR target/57936
  21         * reload1.c (emit_input_reload_insns): When reload_override_in,
  22         set old to rl->in_reg when rl->in_reg is a subreg.
  23
  24 2014-02-26  Richard Biener  <rguenther@suse.de>
  25
  26         PR bootstrap/60343
  27         * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
  28
  29 2014-02-25  Ilya Tocar  <ilya.tocar@intel.com>
  30
  31         * common/config/i386/predicates.md (const1256_operand): Remove.
  32         (const2356_operand): New.
  33         (const_1_to_2_operand): Remove.
  34         * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
  35         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
  36         (*avx512pf_gatherpf<mode>sf): Ditto.
  37         (avx512pf_gatherpf<mode>df): Ditto.
  38         (*avx512pf_gatherpf<mode>df_mask): Ditto.
  39         (*avx512pf_gatherpf<mode>df): Ditto.
  40         (avx512pf_scatterpf<mode>sf): Ditto.
  41         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
  42         (*avx512pf_scatterpf<mode>sf): Ditto.
  43         (avx512pf_scatterpf<mode>df): Ditto.
  44         (*avx512pf_scatterpf<mode>df_mask): Ditto.
  45         (*avx512pf_scatterpf<mode>df): Ditto.
  46         * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
  47
  48 2014-02-26  Ilya Tocar  <ilya.tocar@intel.com>
  49
  50         * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
  51         (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
  52         (_mm512_mask_testn_epi64_mask): Move to ...
  53         * config/i386/avx512cdintrin.h: Here.
  54         * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
  55         * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
  56         (avx512f_scalef<mode><mask_name><round_name>): Ditto.
  57         (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
  58         TARGET_AVX512F from TARGET_AVX512CD.
  59
  60 2014-02-26  Richard Biener  <rguenther@suse.de>
  61
  62         PR ipa/60327
  63         * ipa.c (walk_polymorphic_call_targets): Properly guard
  64         call to inline_update_overall_summary.
  65
  66 2014-02-26  Bin Cheng  <bin.cheng@arm.com>
  67
  68         PR target/60280
  69         * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop
  70         preheaders and latches only if requested.  Fix latch if it
  71         is removed.
  72         * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
  73         LOOPS_HAVE_PREHEADERS.
  74
  75 2014-02-25  Andrew Pinski  <apinski@cavium.com>
  76
  77         * builtins.c (expand_builtin_thread_pointer): Create a new target
  78         when the target is NULL.
  79
  80 2014-02-25  Vladimir Makarov  <vmakarov@redhat.com>
  81
  82         PR rtl-optimization/60317
  83         * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
  84         * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
  85         * lra-assigns.c: Include params.h.
  86         (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
  87         other reload pseudos considerations.
  88
  89 2014-02-25  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
  90
  91         * config/rs6000/vector.md (*vector_unordered<mode>): Change split
  92         to use canonical form for nor<mode>3.
  93
  94 2014-02-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
  95
  96         PR target/55426
  97         * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
  98         conversions.
  99
 100 2014-02-25  Ilya Tocar  <ilya.tocar@intel.com>
 101
 102         * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
 103         (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
 104         (ix86_handle_option): Handle OPT_mprefetchwt1.
 105         * config/i386/cpuid.h (bit_PREFETCHWT1): New.
 106         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
 107         PREFETCHWT1 CPUID.
 108         * config/i386/i386-c.c (ix86_target_macros_internal): Handle
 109         OPTION_MASK_ISA_PREFETCHWT1.
 110         * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
 111         (PTA_PREFETCHWT1): New.
 112         (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
 113         (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
 114         * config/i386/i386.h (TARGET_PREFETCHWT1), (TARGET_PREFETCHWT1_P):
 115           New.
 116         * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
 117         (*prefetch_avx512pf_<mode>_: Change into ...
 118          (*prefetch_prefetchwt1_<mode>: This.
 119         * config/i386/i386.opt (mprefetchwt1): New.
 120         * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
 121         (_mm_prefetch): Handle intent to write.
 122         * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
 123
 124 2014-02-25  Richard Biener  <rguenther@suse.de>
 125
 126         PR middle-end/60291
 127         * emit-rtl.c (mem_attrs_htab): Remove.
 128         (mem_attrs_htab_hash): Likewise.
 129         (mem_attrs_htab_eq): Likewise.
 130         (set_mem_attrs): Always allocate new mem-attrs when something
 131         changed.
 132         (init_emit_once): Do not allocate mem_attrs_htab.
 133
 134 2014-02-25  Richard Biener  <rguenther@suse.de>
 135
 136         PR lto/60319
 137         * lto-opts.c (lto_write_options): Output non-explicit conservative
 138         -fwrapv, -fno-trapv and -fno-strict-overflow.
 139         * lto-wrapper.c (merge_and_complain): Handle merging those options.
 140         (run_gcc): And pass them through.
 141
 142 2014-02-25  Andrey Belevantsev  <abel@ispras.ru>
 143
 144         * sel-sched.c (calculate_new_fences): New parameter ptime.
 145         Calculate it as a maximum over all fence cycles.
 146         (sel_sched_region_2): Adjust the call to calculate_new_fences.
 147         Print the final schedule timing when sched_verbose.
 148
 149 2014-02-25  Andrey Belevantsev  <abel@ispras.ru>
 150
 151         PR rtl-optimization/60292
 152         * sel-sched.c (fill_vec_av_set): Do not reset target availability
 153         bit fot the fence instruction.
 154
 155 2014-02-24  Alangi Derick  <alangiderick@gmail.com>
 156
 157         * calls.h: Fix typo in comment.
 158
 159 2014-02-24  John David Anglin  <danglin@gcc.gnu.org>
 160
 161         * config/pa/pa.c (pa_output_move_double): Don't valididate when
 162         adjusting offsetable addresses.
 163
 164 2014-02-24  Guozhi Wei  <carrot@google.com>
 165
 166         * sparseset.h (sparseset_pop): Fix the wrong index.
 167
 168 2014-02-24  Walter Lee  <walt@tilera.com>
 169
 170         * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
 171         (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
 172         triplet.
 173         * common/config/tilegx/tilegx-common.c
 174         (TARGET_DEFAULT_TARGET_FLAGS): Define.
 175         * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
 176         (LINK_SPEC): Ditto.
 177         * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
 178         * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
 179         (tilegx_gimplify_va_arg_expr): Handle big endian.
 180         (tilegx_expand_unaligned_load): Ditto.
 181         (tilegx_expand_unaligned_store): Ditto.
 182         (TARGET_RETURN_IN_MSB): New.
 183         * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
 184         (TARGET_ENDIAN_DEFAULT): New.
 185         (TARGET_BIG_ENDIAN): Handle big endian.
 186         (BYTES_BIG_ENDIAN): Ditto.
 187         (WORDS_BIG_ENDIAN): Ditto.
 188         (FLOAT_WORDS_BIG_ENDIAN): Ditto.
 189         (ENDIAN_SPEC): New.
 190         (EXTRA_SPECS): New.
 191         * config/tilegx/tilegx.md (extv): Handle big endian.
 192         (extzv): Ditto.
 193         (insn_st<n>): Ditto.
 194         (insn_st<n>_add<bitsuffix>): Ditto.
 195         (insn_stnt<n>): Ditto.
 196         (insn_stnt<n>_add<bitsuffix>):Ditto.
 197         (vec_interleave_highv8qi): Handle big endian.
 198         (vec_interleave_highv8qi_be): New.
 199         (vec_interleave_highv8qi_le): New.
 200         (insn_v1int_h): Handle big endian.
 201         (vec_interleave_lowv8qi): Handle big endian.
 202         (vec_interleave_lowv8qi_be): New.
 203         (vec_interleave_lowv8qi_le): New.
 204         (insn_v1int_l): Handle big endian.
 205         (vec_interleave_highv4hi): Handle big endian.
 206         (vec_interleave_highv4hi_be): New.
 207         (vec_interleave_highv4hi_le): New.
 208         (insn_v2int_h): Handle big endian.
 209         (vec_interleave_lowv4hi): Handle big endian.
 210         (vec_interleave_lowv4hi_be): New.
 211         (vec_interleave_lowv4hi_le): New.
 212         (insn_v2int_l): Handle big endian.
 213         (vec_interleave_highv2si): Handle big endian.
 214         (vec_interleave_highv2si_be): New.
 215         (vec_interleave_highv2si_le): New.
 216         (insn_v4int_h): Handle big endian.
 217         (vec_interleave_lowv2si): Handle big endian.
 218         (vec_interleave_lowv2si_be): New.
 219         (vec_interleave_lowv2si_le): New.
 220         (insn_v4int_l): Handle big endian.
 221         * config/tilegx/tilegx.opt (mbig-endian): New option.
 222         (mlittle-endian): New option.
 223         * doc/install.texi: Document tilegxbe-linux.
 224         * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
 225
 226 2014-02-24  Martin Jambor  <mjambor@suse.cz>
 227
 228         PR ipa/60266
 229         * ipa-cp.c (propagate_constants_accross_call): Bail out early if
 230         there are no parameter descriptors.
 231
 232 2014-02-24  Andrey Belevantsev  <abel@ispras.ru>
 233
 234         PR rtl-optimization/60268
 235         * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
 236         initialization to ...
 237         (sched_rgn_init): ... here.
 238         (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
 239
 240 2014-02-23  David Holsgrove <david.holsgrove@xilinx.com>
 241
 242         * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
 243         names.
 244
 245 2014-02-23  Edgar E. Iglesias <edgar.iglesias@xilinx.com>
 246
 247         * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
 248         definition.
 249
 250 2014-02-23  David Holsgrove <david.holsgrove@xilinx.com>
 251
 252         * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk
 253         and define TARGET_ASM_OUTPUT_MI_THUNK and
 254         TARGET_ASM_CAN_OUTPUT_MI_THUNK.
 255
 256 2014-02-23  David Holsgrove <david.holsgrove@xilinx.com>
 257
 258         * config/microblaze/predicates.md: Add cmp_op predicate.
 259         * config/microblaze/microblaze.md: Add branch_compare instruction
 260         which uses cmp_op predicate and emits cmp insn before branch.
 261         * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
 262         to microblaze_expand_conditional_branch and consolidate logic.
 263         (microblaze_expand_conditional_branch): emit branch_compare
 264         insn instead of handling cmp op separate from branch insn.
 265
 266 2014-02-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
 267
 268         * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
 269         to permit subregs.
 270
 271 2014-02-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
 272
 273         * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
 274         define_insn with define_expand and new define_insn
 275         *altivec_lve<VI_char>x_internal.
 276         (altivec_stve<VI_char>x): Replace define_insn with define_expand
 277         and new define_insn *altivec_stve<VI_char>x_internal.
 278         * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
 279         prototype.
 280         * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
 281         lve*x built-ins.
 282         (altivec_expand_stvex_be): New function.
 283
 284 2014-02-22  Joern Rennecke  <joern.rennecke@embecosm.com>
 285
 286         * config/avr/avr.c (avr_can_eliminate): Allow elimination from
 287         ARG_POINTER_REGNUM
 288         to STACK_POINTER_REGNUM if !frame_pointer_needed.
 289         * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
 290         ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
 291
 292 2014-02-21  Vladimir Makarov  <vmakarov@redhat.com>
 293
 294         PR target/60298
 295         * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
 296         instead of emit_move_insn.
 297
 298 2014-02-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
 299
 300         * config/rs6000/altivec.md (altivec_vsumsws): Replace second
 301         vspltw with vsldoi.
 302         (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
 303         gen_altivec_vsumsws.
 304
 305 2014-02-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
 306
 307         * config/rs6000/altivec.md (altivec_lvxl): Rename as
 308         *altivec_lvxl_<mode>_internal and use VM2 iterator instead of
 309         V4SI.
 310         (altivec_lvxl_<mode>): New define_expand incorporating
 311         -maltivec=be semantics where needed.
 312         (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
 313         (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
 314         semantics where needed.
 315         (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
 316         (altivec_stvx_<mode>): New define_expand incorporating
 317         -maltivec=be semantics where needed.
 318         (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
 319         VM2 iterator instead of V4SI.
 320         (altivec_stvxl_<mode>): New define_expand incorporating
 321         -maltivec=be semantics where needed.
 322         * config/rs6000/rs6000-builtin.def: Add new built-in definitions
 323         LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
 324         LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI,
 325         STVX_V2DF, STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI,
 326         STVXL_V2DF, STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI,
 327         STVXL_V16QI.
 328         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
 329         ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
 330         similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
 331         ALTIVEC_BUILTIN_STVXL.
 332         * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New
 333         prototype.
 334         (altivec_expand_stvx_be): Likewise.
 335         * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
 336         (altivec_expand_lvx_be): Likewise.
 337         (altivec_expand_stvx_be): Likewise.
 338         (altivec_expand_builtin): Add cases for
 339         ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
 340         ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
 341         (altivec_init_builtins): Add definitions for
 342         __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
 343         __builtin_altivec_stvx_<mode>, and
 344         __builtin_altivec_stvxl_<mode>.
 345
 346 2014-02-21  Catherine Moore  <clm@codesourcery.com>
 347
 348         * doc/invoke.texi (mvirt, mno-virt): Document.
 349         * config/mips/mips.opt (mvirt): New option.
 350         * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
 351
 352 2014-02-21  Richard Biener  <rguenther@suse.de>
 353
 354         PR tree-optimization/60276
 355         * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
 356         (STMT_VINFO_MIN_NEG_DIST): New macro.
 357         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
 358         STMT_VINFO_MIN_NEG_DIST.
 359         * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
 360         made for negative dependence distances still hold.
 361
 362 2014-02-21  Richard Biener  <rguenther@suse.de>
 363
 364         PR middle-end/60291
 365         * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
 366         DECL_INITIAL for globals not in the current function context.
 367
 368 2014-02-21  Jakub Jelinek  <jakub@redhat.com>
 369
 370         PR tree-optimization/56490
 371         * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
 372         * tree-ssa-uninit.c: Include params.h.
 373         (compute_control_dep_chain): Add num_calls argument, return false
 374         if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
 375         num_calls to recursive call.
 376         (find_predicates): Change dep_chain into normal array,
 377         cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
 378         variable and adjust compute_control_dep_chain caller.
 379         (find_def_preds): Likewise.
 380
 381 2014-02-21  Thomas Schwinge  <thomas@codesourcery.com>
 382
 383         * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
 384         <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
 385
 386 2014-02-21  Nick Clifton  <nickc@redhat.com>
 387
 388         * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
 389         (pushhi1): Likewise.
 390         (popqi1): Add mode to pre_dec.
 391         (pophi1): Likewise.
 392
 393 2014-02-21  Jakub Jelinek  <jakub@redhat.com>
 394
 395         * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
 396         mode for mask of V8SFmode permutation.
 397
 398 2014-02-20  Richard Henderson  <rth@redhat.com>
 399
 400         PR c++/60272
 401         * builtins.c (expand_builtin_atomic_compare_exchange): Always make
 402         a new pseudo for OLDVAL.
 403
 404 2014-02-20  Jakub Jelinek  <jakub@redhat.com>
 405
 406         PR target/57896
 407         * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
 408         gen_reg_rtx if d->testing_p.
 409         (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
 410         if d->testing_p and we will certainly return true.
 411         (expand_vec_perm_even_odd_1): Likewise.  Don't call gen_reg_rtx
 412         if d->testing_p.
 413
 414 2014-02-20  Uros Bizjak  <ubizjak@gmail.com>
 415
 416         * emit-rtl.c (gen_reg_rtx): Assert that
 417         crtl->emit.regno_pointer_align_length is non-zero.
 418
 419 2014-02-20  Richard Henderson  <rth@redhat.com>
 420
 421         PR c++/60272
 422         * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
 423         on failure the store back into EXPECT.
 424
 425 2014-02-20  Chung-Lin Tang  <cltang@codesourcery.com>
 426             Sandra Loosemore  <sandra@codesourcery.com>
 427
 428         * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
 429         * config/nios2/nios2.c (nios2_function_profiler):
 430         Add -fPIC (flag_pic == 2) support.
 431         (nios2_handle_custom_fpu_cfg): Fix warning parameter.
 432         (nios2_large_offset_p): New function.
 433         (nios2_unspec_reloc_p): Move up position, update to use
 434         nios2_large_offset_p.
 435         (nios2_unspec_address): Remove function.
 436         (nios2_unspec_offset): New function.
 437         (nios2_large_got_address): New function.
 438         (nios2_got_address): Add large offset support.
 439         (nios2_legitimize_tls_address): Update usage of removed and new
 440         functions.
 441         (nios2_symbol_binds_local_p): New function.
 442         (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
 443         (nios2_legitimize_address): Update to use nios2_large_offset_p.
 444         (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
 445         (nios2_print_operand): Merge H/L processing, add hiadj/lo
 446         processing for (const (unspec ...)).
 447         (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
 448
 449 2014-02-20  Richard Biener  <rguenther@suse.de>
 450
 451         * tree-cfg.c (replace_uses_by): Mark altered BBs before
 452         doing the substitution.
 453         (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
 454
 455 2014-02-20  Martin Jambor  <mjambor@suse.cz>
 456
 457         PR ipa/55260
 458         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
 459         info when checking whether lattices are bottom.
 460
 461 2014-02-20  Richard Biener  <rguenther@suse.de>
 462
 463         PR middle-end/60221
 464         * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
 465         regions at -O0.
 466
 467 2014-02-20  Jan Hubicka  <hubicka@ucw.cz>
 468
 469         PR ipa/58555
 470         * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
 471         parameter specifying the scaling.
 472         (inline_call): Update.
 473         (want_inline_recursively): Guard division by zero.
 474         (recursive_inlining): Update.
 475         * ipa-inline.h (clone_inlined_nodes): Update.
 476
 477 2014-02-20  Ilya Tocar  <ilya.tocar@intel.com>
 478
 479         PR target/60204
 480         * config/i386/i386.c (classify_argument): Pass structures of size
 481         64 bytes or less in register.
 482
 483 2014-02-20  Ilya Tocar  <ilya.tocar@intel.com>
 484             Kirill Yukhin  <kirill.yukhin@intel.com>
 485
 486         * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
 487         (_mm_rcp28_round_ss): Ditto.
 488         (_mm_rsqrt28_round_sd): Ditto.
 489         (_mm_rsqrt28_round_ss): Ditto.
 490         * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
 491         (_mm_rcp14_round_ss): Ditto.
 492         (_mm_rsqrt14_round_sd): Ditto.
 493         (_mm_rsqrt14_round_ss): Ditto.
 494         * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
 495         the first input operand, get rid of match_dup.
 496         (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
 497         attribute to sse.
 498         (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
 499         Ditto.
 500         (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
 501         operand as the first input operand, set type attribute.
 502         (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
 503         Set type attribute.
 504         (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
 505         operand as the first input operand, set type attribute.
 506
 507 2014-02-19  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
 508
 509         * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
 510         bit of zero.
 511
 512 2014-02-19  H.J. Lu  <hongjiu.lu@intel.com>
 513
 514         PR target/60207
 515         * config/i386/i386.c (construct_container): Remove TFmode check
 516         for X86_64_INTEGER_CLASS.
 517
 518 2014-02-19  Uros Bizjak  <ubizjak@gmail.com>
 519
 520         PR target/59794
 521         * config/i386/i386.c (type_natural_mode): Warn for ABI changes
 522         only when -Wpsabi is enabled.
 523
 524 2014-02-19  Michael Hudson-Doyle  <michael.hudson@linaro.org>
 525
 526          PR target/59799
 527         * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
 528         passing arrays in registers are the same as for structs, so remove the
 529         special case for them.
 530
 531 2014-02-19  Eric Botcazou  <ebotcazou@adacore.com>
 532
 533         * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
 534         destination type, extract only the valid bits if the source type is not
 535         integral and has a different mode.
 536
 537 2014-02-19  Richard Biener  <rguenther@suse.de>
 538
 539         PR ipa/60243
 540         * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
 541         for all calls.
 542
 543 2014-02-19  Richard Biener  <rguenther@suse.de>
 544
 545         PR ipa/60243
 546         * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
 547         (ipa_modify_call_arguments): Emit an argument load explicitely and
 548         preserve virtual SSA form there and for the replacement call.
 549         Do not update SSA form nor free dominance info.
 550
 551 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
 552
 553         * ipa.c (function_and_variable_visibility): Also clear WEAK
 554         flag when disolving COMDAT_GROUP.
 555
 556 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
 557
 558         * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
 559         * ipa-prop.c (ipa_set_jf_known_type): Return early when
 560         not devirtualizing.
 561         (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
 562         do more sanity checks.
 563         (detect_type_change): Return true when giving up early.
 564         (compute_complex_assign_jump_func): Fix type parameter of
 565         ipa_set_ancestor_jf.
 566         (compute_complex_ancestor_jump_func): Likewise.
 567         (update_jump_functions_after_inlining): Fix updating of
 568         ancestor function.
 569         * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
 570
 571 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
 572
 573         * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
 574         inline clones when edge disappears.
 575
 576 2014-02-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
 577
 578         PR target/60203
 579         * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
 580         Split 64-bit moves into 2 patterns.  Do not allow the use of
 581         direct move for TDmode in little endian, since the decimal value
 582         has little endian bytes within a word, but the 64-bit pieces are
 583         ordered in a big endian fashion, and normal subreg's of TDmode are
 584         not allowed.
 585         (mov<mode>_64bit_dm): Likewise.
 586         (movtd_64bit_nodm): Likewise.
 587
 588 2014-02-18  Eric Botcazou  <ebotcazou@adacore.com>
 589
 590         PR tree-optimization/60174
 591         * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
 592         statement of an SSA_NAME that occurs in an abnormal PHI node.
 593
 594 2014-02-18  Jakub Jelinek  <jakub@redhat.com>
 595
 596         PR sanitizer/60142
 597         * final.c (SEEN_BB): Remove.
 598         (SEEN_NOTE, SEEN_EMITTED): Renumber.
 599         (final_scan_insn): Don't force_source_line on second
 600         NOTE_INSN_BASIC_BLOCK.
 601
 602 2014-02-18  Uros Bizjak  <ubizjak@gmail.com>
 603
 604         PR target/60205
 605         * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
 606         * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
 607         (type_natural_mode): Warn ABI change when %zmm register is not
 608         available for AVX512F vector value passing.
 609
 610 2014-02-18  Kai Tietz  <ktietz@redhat.com>
 611
 612         PR target/60193
 613         * config/i386/i386.c (ix86_expand_prologue): Use value in
 614         rax register as displacement when restoring %r10 or %rax.
 615         Fix wrong offset when restoring both registers.
 616
 617 2014-02-18  Eric Botcazou  <ebotcazou@adacore.com>
 618
 619         * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
 620         assertion with conditional return.
 621
 622 2014-02-18  Jakub Jelinek  <jakub@redhat.com>
 623             Uros Bizjak  <ubizjak@gmail.com>
 624
 625         PR driver/60233
 626         * config/i386/driver-i386.c (host_detect_local_cpu): If
 627         YMM state is not saved by the OS, also clear has_f16c.  Move
 628         CPUID 0x80000001 handling before YMM state saving checking.
 629
 630 2014-02-18  Andrey Belevantsev  <abel@ispras.ru>
 631
 632         PR rtl-optimization/58960
 633         * haifa-sched.c (alloc_global_sched_pressure_data): New,
 634         factored out from ...
 635         (sched_init): ... here.
 636         (free_global_sched_pressure_data): New, factored out from ...
 637         (sched_finish): ... here.
 638         * sched-int.h (free_global_sched_pressure_data): Declare.
 639         * sched-rgn.c (nr_regions_initial): New static global.
 640         (haifa_find_rgns): Initialize it.
 641         (schedule_region): Disable sched-pressure for the newly
 642         generated regions.
 643
 644 2014-02-17  Richard Biener  <rguenther@suse.de>
 645
 646         * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
 647         release SSA defs of pattern stmts.
 648
 649 2014-02-17  Richard Biener  <rguenther@suse.de>
 650
 651         * tree-inline.c (expand_call_inline): Release the virtual
 652         operand defined by the call we are about to inline.
 653
 654 2014-02-17  Richard Biener  <rguenther@suse.de>
 655
 656         * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
 657
 658 2014-02-17  Kirill Yukhin  <kirill.yukhin@intel.com>
 659             Ilya Tocar  <ilya.tocar@intel.com>
 660
 661         * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
 662         arguments order in builtin.
 663         (_mm512_permutexvar_epi64): Ditto.
 664         (_mm512_mask_permutexvar_epi64): Ditto
 665         (_mm512_maskz_permutexvar_epi32): Ditto
 666         (_mm512_permutexvar_epi32): Ditto
 667         (_mm512_mask_permutexvar_epi32): Ditto
 668
 669 2014-02-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
 670
 671         * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
 672         (p8_vmrgow): Likewise.
 673
 674 2014-02-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
 675
 676         * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
 677         endian targets.
 678
 679 2014-02-15  Michael Meissner  <meissner@linux.vnet.ibm.com>
 680
 681         PR target/60203
 682         * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
 683         (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
 684         into 64-bit and 32-bit moves.  On 64-bit moves, add support for
 685         using direct move instructions on ISA 2.07.  Also adjust
 686         instruction length for 64-bit.
 687         (mov<mode>_64bit, TFmode/TDmode): Likewise.
 688         (mov<mode>_32bit, TFmode/TDmode): Likewise.
 689
 690 2014-02-15  Alan Modra  <amodra@gmail.com>
 691
 692         PR target/58675
 693         PR target/57935
 694         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
 695         find_replacement on parts of insn rtl that might be reloaded.
 696
 697 2014-02-15  Richard Biener  <rguenther@suse.de>
 698
 699         PR tree-optimization/60183
 700         * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
 701         (tree_ssa_phiprop): Calculate and free post-dominators.
 702
 703 2014-02-14  Jeff Law  <law@redhat.com>
 704
 705         PR rtl-optimization/60131
 706         * ree.c (get_extended_src_reg): New function.
 707         (combine_reaching_defs): Use it rather than assuming location of REG.
 708         (find_and_remove_re): Verify first operand of extension is
 709         a REG before adding the insns to the copy list.
 710
 711 2014-02-14  Roland McGrath  <mcgrathr@google.com>
 712
 713         * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
 714         * configure: Regenerated.
 715         * config.in: Regenerated.
 716         * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
 717         instead of ASM_SHORT.
 718
 719 2014-02-14  Vladimir Makarov  <vmakarov@redhat.com>
 720             Richard Earnshaw  <rearnsha@arm.com>
 721
 722         PR rtl-optimization/59535
 723         * lra-constraints.c (process_alt_operands): Encourage alternative
 724         when unassigned pseudo class is superset of the alternative class.
 725         (inherit_reload_reg): Don't inherit when optimizing for code size.
 726         * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
 727         returning CORE_REGS for anything but Thumb1 and BASE_REGS for
 728         modes not less than 4 for Thumb1.
 729
 730 2014-02-14  Kyle McMartin  <kyle@redhat.com>
 731
 732         PR pch/60010
 733         * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
 734
 735 2014-02-14  Richard Biener  <rguenther@suse.de>
 736
 737         * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
 738         (get_frame_arg): Drop the assert with langhook types_compatible_p.
 739         Do not strip INDIRECT_REFs.
 740
 741 2014-02-14  Richard Biener  <rguenther@suse.de>
 742
 743         PR lto/60179
 744         * lto-streamer-out.c (DFS_write_tree_body): Do not follow
 745         DECL_FUNCTION_SPECIFIC_TARGET.
 746         (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
 747         * tree-streamer-out.c (pack_ts_target_option): Remove.
 748         (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
 749         (write_ts_function_decl_tree_pointers): Do not stream
 750         DECL_FUNCTION_SPECIFIC_TARGET.
 751         * tree-streamer-in.c (unpack_ts_target_option): Remove.
 752         (unpack_value_fields): Do not stream TS_TARGET_OPTION.
 753         (lto_input_ts_function_decl_tree_pointers): Do not stream
 754         DECL_FUNCTION_SPECIFIC_TARGET.
 755
 756 2014-02-14  Jakub Jelinek  <jakub@redhat.com>
 757
 758         * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
 759         (get_initial_def_for_induction, vectorizable_induction): Ignore
 760         debug stmts when looking for exit_phi.
 761         (vectorizable_live_operation): Fix up condition.
 762
 763 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
 764
 765         * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
 766         nreverse() because it changes the content of original tree list.
 767
 768 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
 769
 770         * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
 771         * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
 772
 773 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
 774
 775         * config/nds32/nds32.c (nds32_naked_function_p): Follow the
 776         GNU coding standards.
 777
 778 2014-02-13  Jakub Jelinek  <jakub@redhat.com>
 779
 780         PR debug/60152
 781         * dwarf2out.c (gen_subprogram_die): Don't call
 782         add_calling_convention_attribute if subr_die is old_die.
 783
 784 2014-02-13  Sharad Singhai  <singhai@google.com>
 785
 786         * doc/optinfo.texi: Fix order of nodes.
 787
 788 2014-02-13  Uros Bizjak  <ubizjak@gmail.com>
 789
 790         * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
 791         operands[2], not operands[3].
 792
 793 2014-02-13  Richard Biener  <rguenther@suse.de>
 794
 795         PR bootstrap/59878
 796         * doc/install.texi (ISL): Update recommended version to 0.12.2,
 797         mention the possibility of an in-tree build.
 798         (CLooG): Update recommended version to 0.18.1, mention the
 799         possibility of an in-tree build and clarify that the ISL
 800         bundled with CLooG does not work.
 801
 802 2014-02-13  Jakub Jelinek  <jakub@redhat.com>
 803
 804         PR target/43546
 805         * expr.c (compress_float_constant): If x is a hard register,
 806         extend into a pseudo and then move to x.
 807
 808 2014-02-13  Dominik Vogt  <vogt@linux.vnet.ibm.com>
 809
 810         * config/s390/s390.c (s390_asm_output_function_label): Fix crash
 811         caused by bad second argument to warning_at() with -mhotpatch and
 812         nested functions (e.g. with gfortran).
 813
 814 2014-02-13  Richard Sandiford  <rdsandiford@googlemail.com>
 815
 816         * opts.c (option_name): Remove "enabled by default" rider.
 817
 818 2014-02-12  John David Anglin  <danglin@gcc.gnu.org>
 819
 820         * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
 821
 822 2014-02-12  H.J. Lu  <hongjiu.lu@intel.com>
 823             Uros Bizjak  <ubizjak@gmail.com>
 824
 825         PR target/60151
 826         * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
 827         * configure: Regenerated.
 828
 829 2014-02-12  Richard Biener  <rguenther@suse.de>
 830
 831         * vec.c (vec_prefix::calculate_allocation): Move as
 832         inline variant to vec.h.
 833         (vec_prefix::calculate_allocation_1): New out-of-line version.
 834         * vec.h (vec_prefix::calculate_allocation_1): Declare.
 835         (vec_prefix::m_has_auto_buf): Rename to ...
 836         (vec_prefix::m_using_auto_storage): ... this.
 837         (vec_prefix::calculate_allocation): Inline the easy cases
 838         and dispatch to calculate_allocation_1 which doesn't need the
 839         prefix address.
 840         (va_heap::reserve): Use gcc_checking_assert.
 841         (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
 842         m_using_auto_storage.
 843         (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
 844         member and adjust.
 845         (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
 846         (vec<T, va_heap, vl_ptr>::release): Avoid casting.
 847         (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
 848
 849 2014-02-12  Richard Biener  <rguenther@suse.de>
 850
 851         * gcse.c (compute_transp): break from loop over canon_modify_mem_list
 852         when we found a dependence.
 853
 854 2014-02-12  Thomas Schwinge  <thomas@codesourcery.com>
 855
 856         * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
 857         common code...
 858         (maybe_fold_stmt): ... into this new function.
 859         * omp-low.c (lower_omp): Update comment.
 860
 861         * omp-low.c (lower_omp_target): Add clobber for sizes array, after
 862         last use.
 863
 864         * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
 865         dereference.
 866
 867 2014-02-12  James Greenhalgh  <james.greenhalgh@arm.com>
 868
 869         * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
 870         identifiers in comments.
 871         (cortexa53_extra_costs): Likewise.
 872         * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
 873         (cortexa7_extra_costs): Likewise.
 874         (cortexa12_extra_costs): Likewise.
 875         (cortexa15_extra_costs): Likewise.
 876         (v7m_extra_costs): Likewise.
 877
 878 2014-02-12  Richard Biener  <rguenther@suse.de>
 879
 880         PR middle-end/60092
 881         * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
 882         of posix_memalign being successful.
 883         (lower_stmt): Restrict lowering of posix_memalign to when
 884         -ftree-bit-ccp is enabled.
 885
 886 2014-02-12  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
 887
 888         * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
 889         arg_loc.
 890         * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
 891
 892 2014-02-12  Eric Botcazou  <ebotcazou@adacore.com>
 893
 894         PR rtl-optimization/60116
 895         * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
 896         other_insn once the combination has been validated.
 897
 898 2014-02-11  Jan Hubicka  <hubicka@ucw.cz>
 899
 900         PR lto/59468
 901         * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
 902         and wrapper.
 903         * ipa-devirt.c: Include demangle.h
 904         (odr_violation_reported): New static variable.
 905         (add_type_duplicate): Update odr_violations.
 906         (maybe_record_node): Add completep parameter; update it.
 907         (record_target_from_binfo): Add COMPLETEP parameter;
 908         update it as needed.
 909         (possible_polymorphic_call_targets_1): Likewise.
 910         (struct polymorphic_call_target_d): Add nonconstruction_targets;
 911         rename FINAL to COMPLETE.
 912         (record_targets_from_bases): Sanity check we found the binfo;
 913         fix COMPLETEP updating.
 914         (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
 915         parameter, fix computing of COMPLETEP.
 916         (dump_possible_polymorphic_call_targets): Imrove readability of dump;
 917         at LTO time do demangling.
 918         (ipa_devirt): Use nonconstruction_targets; Improve dumps.
 919         * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
 920         parameter.
 921         (gimple_get_virt_method_for_binfo): Likewise.
 922         * gimple-fold.h (gimple_get_virt_method_for_binfo,
 923         gimple_get_virt_method_for_vtable): Update prototypes.
 924
 925 2014-02-11  Vladimir Makarov  <vmakarov@redhat.com>
 926
 927         PR target/49008
 928         * genautomata.c (add_presence_absence): Fix typo with
 929         {final_}presence_list.
 930
 931 2014-02-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
 932
 933         PR target/60137
 934         * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
 935         for VSX/Altivec vectors that land in GPR registers.
 936
 937 2014-02-11  Richard Henderson  <rth@redhat.com>
 938             Jakub Jelinek  <jakub@redhat.com>
 939
 940         PR debug/59776
 941         * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
 942         around drhs if type conversion to lacc->type is not useless.
 943
 944 2014-02-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
 945
 946         * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
 947         tuning struct.
 948         (cortex-a57.cortex-a53): Likewise.
 949         * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
 950
 951 2014-02-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
 952
 953         * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
 954         arm_restrict_it.
 955
 956 2014-02-11  Renlin Li  <Renlin.Li@arm.com>
 957
 958         * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
 959         add_options_for_arm_vfp3.
 960
 961 2014-02-11  Jeff Law  <law@redhat.com>
 962
 963         PR middle-end/54041
 964         * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
 965         object with an undesirable mode.
 966
 967 2014-02-11  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
 968
 969         PR libgomp/60107
 970         * config/i386/sol2-9.h: New file.
 971         * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
 972         *-*-solaris2.9*): Use it.
 973
 974 2014-02-10  Nagaraju Mekala <nagaraju.mekala@xilinx.com>
 975
 976         * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
 977         * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
 978
 979 2014-02-10  Nagaraju Mekala <nagaraju.mekala@xilinx.com>
 980
 981         * config/microblaze/microblaze.c: Extend mcpu version format
 982
 983 2014-02-10  David Holsgrove <david.holsgrove@xilinx.com>
 984
 985         * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
 986
 987 2014-02-10  Richard Henderson  <rth@redhat.com>
 988
 989         PR target/59927
 990         * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
 991         * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
 992         ms-abi vs -mno-accumulate-outgoing-args.
 993         (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
 994         * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
 995         respect to ms-abi.
 996
 997 2014-02-10  Bernd Edlinger  <bernd.edlinger@hotmail.de>
 998
 999         PR middle-end/60080
1000         * cfgexpand.c (expand_asm_operands): Attach source location to
1001         ASM_INPUT rtx objects.
1002         * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
1003
1004 2014-02-10  Nick Clifton  <nickc@redhat.com>
1005
1006         * config/mn10300/mn10300.c (popcount): New function.
1007         (mn10300_expand_prologue): Include saved registers in stack usage
1008         count.
1009
1010 2014-02-10  Jeff Law  <law@redhat.com>
1011
1012         PR middle-end/52306
1013         * reload1.c (emit_input_reload_insns): Do not create invalid RTL
1014         when changing the SET_DEST of a prior insn to avoid an input reload.
1015
1016 2014-02-10  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
1017
1018         * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
1019         big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
1020         -mcall-openbsd, or -mcall-linux.
1021         (CC1_ENDIAN_BIG_SPEC): Remove.
1022         (CC1_ENDIAN_LITTLE_SPEC): Remove.
1023         (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1024         (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
1025         (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
1026         and %cc1_endian_default.
1027         * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1028
1029 2014-02-10  Richard Biener  <rguenther@suse.de>
1030
1031         PR tree-optimization/60115
1032         * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
1033         MEM_REF handling.  Properly verify that the accesses are not
1034         out of the objects bound.
1035
1036 2014-02-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1037
1038         * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
1039         coretex to cortex.
1040
1041 2014-02-10  Eric Botcazou  <ebotcazou@adacore.com>
1042
1043         * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
1044         proper constants and fix formatting.
1045         (possible_polymorphic_call_targets): Fix formatting.
1046
1047 2014-02-10  Kirill Yukhin  <kirill.yukhin@intel.com>
1048             Ilya Tocar  <ilya.tocar@intel.com>
1049
1050         * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
1051         (_mm512_loadu_epi32): Renamed into...
1052         (_mm512_loadu_si512): This.
1053         (_mm512_storeu_epi32): Renamed into...
1054         (_mm512_storeu_si512): This.
1055         (_mm512_maskz_ceil_ps): Removed.
1056         (_mm512_maskz_ceil_pd): Ditto.
1057         (_mm512_maskz_floor_ps): Ditto.
1058         (_mm512_maskz_floor_pd): Ditto.
1059         (_mm512_floor_round_ps): Ditto.
1060         (_mm512_floor_round_pd): Ditto.
1061         (_mm512_ceil_round_ps): Ditto.
1062         (_mm512_ceil_round_pd): Ditto.
1063         (_mm512_mask_floor_round_ps): Ditto.
1064         (_mm512_mask_floor_round_pd): Ditto.
1065         (_mm512_mask_ceil_round_ps): Ditto.
1066         (_mm512_mask_ceil_round_pd): Ditto.
1067         (_mm512_maskz_floor_round_ps): Ditto.
1068         (_mm512_maskz_floor_round_pd): Ditto.
1069         (_mm512_maskz_ceil_round_ps): Ditto.
1070         (_mm512_maskz_ceil_round_pd): Ditto.
1071         (_mm512_expand_pd): Ditto.
1072         (_mm512_expand_ps): Ditto.
1073         * config/i386/i386.c (ix86_builtins): Remove
1074         IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
1075         (bdesc_args): Ditto.
1076         * config/i386/predicates.md (const1256_operand): New.
1077         (const_1_to_2_operand): Ditto.
1078         * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1079         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1080         (*avx512pf_gatherpf<mode>sf): Ditto.
1081         (avx512pf_gatherpf<mode>df): Ditto.
1082         (*avx512pf_gatherpf<mode>df_mask): Ditto.
1083         (*avx512pf_gatherpf<mode>df): Ditto.
1084         (avx512pf_scatterpf<mode>sf): Ditto.
1085         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1086         (*avx512pf_scatterpf<mode>sf): Ditto.
1087         (avx512pf_scatterpf<mode>df): Ditto.
1088         (*avx512pf_scatterpf<mode>df_mask): Ditto.
1089         (*avx512pf_scatterpf<mode>df): Ditto.
1090         (avx512f_expand<mode>): Removed.
1091         (<shift_insn><mode>3<mask_name>): Change predicate type.
1092
1093 2014-02-08  Jakub Jelinek  <jakub@redhat.com>
1094
1095         * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
1096         not at the end of datarefs vector use ordered_remove to avoid
1097         reordering datarefs vector.
1098
1099         PR c/59984
1100         * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
1101         mark local addressable non-static vars as GOVD_PRIVATE
1102         instead of GOVD_LOCAL.
1103         * omp-low.c (lower_omp_for): Move gimple_bind_vars
1104         and BLOCK_VARS of gimple_bind_block to new_stmt rather
1105         than copying them.
1106
1107         PR middle-end/60092
1108         * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
1109         if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
1110         assume_aligned or alloc_align attributes.
1111         (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
1112         arguments.  Handle also assume_aligned and alloc_align attributes.
1113         (evaluate_stmt): Adjust bit_value_assume_aligned caller.  Handle
1114         calls to functions with assume_aligned or alloc_align attributes.
1115         * doc/extend.texi: Document assume_aligned and alloc_align attributes.
1116
1117 2014-02-08  Terry Guo  <terry.guo@arm.com>
1118
1119         * doc/invoke.texi: Document ARM -march=armv7e-m.
1120
1121 2014-02-08  Jakub Jelinek  <jakub@redhat.com>
1122
1123         * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
1124         flag on __cilkrts_rethrow builtin.
1125
1126         PR ipa/60026
1127         * ipa-cp.c (determine_versionability): Fail at -O0
1128         or __attribute__((optimize (0))) or -fno-ipa-cp functions.
1129         * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
1130
1131         Revert:
1132         2014-02-04  Jakub Jelinek  <jakub@redhat.com>
1133
1134         PR ipa/60026
1135         * tree-inline.c (copy_forbidden): Fail for
1136         __attribute__((optimize (0))) functions.
1137
1138 2014-02-07  Jan Hubicka  <hubicka@ucw.cz>
1139
1140         * varpool.c: Include pointer-set.h.
1141         (varpool_remove_unreferenced_decls): Variables in other partitions
1142         will not be output; be however careful to not lose information
1143         about partitioning.
1144
1145 2014-02-07  Jan Hubicka  <hubicka@ucw.cz>
1146
1147         * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
1148         lookup in the vtable constructor.
1149
1150 2014-02-07  Jeff Law  <law@redhat.com>
1151
1152         PR target/40977
1153         * config/m68k/m68k.md (ashldi_extsi): Turn into a
1154         define_insn_and_split.
1155
1156         * ipa-inline.c (inline_small_functions): Fix typos.
1157
1158 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
1159
1160         * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
1161         (s390_can_use_return_insn): Declare.
1162         * config/s390/s390.h (EPILOGUE_USES): Define.
1163         * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
1164         instructions.
1165         (s390_chunkify_start): Handle return JUMP_LABELs.
1166         (s390_early_mach): Emit a main_pool instruction on the entry edge.
1167         (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
1168         (s390_can_use_return_insn): New functions.
1169         (s390_fix_long_loop_prediction): Handle conditional returns.
1170         (TARGET_SET_UP_BY_PROLOGUE): Define.
1171         * config/s390/s390.md (ANY_RETURN): New code iterator.
1172         (*creturn, *csimple_return, return, simple_return): New patterns.
1173
1174 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
1175
1176         * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
1177         notes to each restore.  Also add REG_CFA_DEF_CFA when restoring %r15.
1178         (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P.  Update the
1179         REG_CFA_RESTORE list when deciding not to restore a register.
1180
1181 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
1182
1183         * config/s390/s390.c: Include tree-pass.h and context.h.
1184         (s390_early_mach): New function, split out from...
1185         (s390_emit_prologue): ...here.
1186         (pass_data_s390_early_mach): New pass structure.
1187         (pass_s390_early_mach): New class.
1188         (s390_option_override): Create and register early_mach pass.
1189         Move to end of file.
1190
1191 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
1192
1193         * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
1194         to match for the exit block.
1195
1196 2014-02-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
1197
1198         * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
1199         ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
1200         Reject misaligned operands.
1201
1202 2014-02-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
1203
1204         * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
1205
1206 2014-02-07  Richard Biener  <rguenther@suse.de>
1207
1208         PR middle-end/60092
1209         * gimple-low.c (lower_builtin_posix_memalign): New function.
1210         (lower_stmt): Call it to lower posix_memalign in a way
1211         to make alignment info accessible.
1212
1213 2014-02-07  Jakub Jelinek  <jakub@redhat.com>
1214
1215         PR c++/60082
1216         * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
1217         __builtin_setjmp_receiver.
1218
1219 2014-02-07  Richard Biener  <rguenther@suse.de>
1220
1221         PR middle-end/60092
1222         * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
1223         * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
1224         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
1225         Handle BUILT_IN_POSIX_MEMALIGN.
1226         (find_func_clobbers): Likewise.
1227         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
1228         (call_may_clobber_ref_p_1): Likewise.
1229
1230 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
1231
1232         PR ipa/59918
1233         * ipa-devirt.c (record_target_from_binfo): Remove overactive
1234         sanity check.
1235
1236 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
1237
1238         PR ipa/59469
1239         * lto-cgraph.c (lto_output_node): Use
1240         symtab_get_symbol_partitioning_class.
1241         (lto_output_varpool_node): likewise.
1242         (symtab_get_symbol_partitioning_class): Move here from
1243         lto/lto-partition.c
1244         * cgraph.h (symbol_partitioning_class): Likewise.
1245         (symtab_get_symbol_partitioning_class): Declare.
1246
1247 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
1248
1249         * ggc.h (ggc_internal_cleared_alloc): New macro.
1250         * vec.h (vec_safe_copy): Handle memory stats.
1251         * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
1252         * target-globals.c (save_target_globals): Likewise.
1253
1254 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
1255
1256         PR target/60077
1257         * expr.c (emit_move_resolve_push): Export; be bit more selective
1258         on when to clear alias set.
1259         * expr.h (emit_move_resolve_push): Declare.
1260         * function.h (struct function): Add tail_call_marked.
1261         * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
1262         * config/i386/i386-protos.h (ix86_expand_push): Remove.
1263         * config/i386/i386.md (TImode move expander): De not call
1264         ix86_expand_push.
1265         (FP push expanders): Preserve memory attributes.
1266         * config/i386/sse.md (push<mode>1): Remove.
1267         * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
1268         (ix86_expand_push): Remove.
1269         * config/i386/mmx.md (push<mode>1): Remove.
1270
1271 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
1272
1273         PR rtl-optimization/60030
1274         * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
1275         lopart with paradoxical subreg before shifting it up by hprec.
1276
1277 2014-02-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1278
1279         * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
1280         Remove extra newline at end of file.
1281         * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
1282         (arm_issue_rate): Handle cortexa57.
1283         * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
1284         (cortex-a57.cortex-a53): Likewise.
1285
1286 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
1287
1288         PR target/59575
1289         * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
1290         don't record in REG_FRAME_RELATED_EXPR registers not set in that
1291         bitmask.
1292         (arm_expand_prologue): Adjust all callers.
1293         (arm_unwind_emit_sequence): Allow saved, but not important for unwind
1294         info, registers also at the lowest numbered registers side.  Use
1295         gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
1296         XEXP.
1297
1298         PR debug/59992
1299         * var-tracking.c (adjust_mems): Before adding a SET to
1300         amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
1301
1302 2014-02-06  Alan Modra  <amodra@gmail.com>
1303
1304         PR target/60032
1305         * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
1306         change SDmode to DDmode when lra_in_progress.
1307
1308 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
1309
1310         PR middle-end/59150
1311         * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
1312         free_data_ref on the dr first, and before goto again also set dr
1313         to the next dr.  For simd_lane_access, free old datarefs[i] before
1314         overwriting it.  For get_vectype_for_scalar_type failure, don't
1315         free_data_ref if simd_lane_access.
1316
1317         * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
1318
1319         PR target/60062
1320         * tree.h (opts_for_fn): New inline function.
1321         (opt_for_fn): Define.
1322         * config/i386/i386.c (ix86_function_regparm): Use
1323         opt_for_fn (decl, optimize) instead of optimize.
1324
1325 2014-02-06  Marcus Shawcroft  <marcus.shawcroft@arm.com>
1326
1327         * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
1328         for SYMBOL_REF in large memory model.
1329
1330 2014-02-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1331
1332         * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
1333         and crypto support.
1334         (cortex-a57): Likewise.
1335         (cortex-a57.cortex-a53): Likewise.
1336
1337 2014-02-06  Yury Gribov  <y.gribov@samsung.com>
1338             Kugan Vivekanandarajah  <kuganv@linaro.org>
1339
1340         * config/arm/arm.c (arm_vector_alignment_reachable): Check
1341         unaligned_access.
1342         * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
1343
1344 2014-02-06  Richard Biener  <rguenther@suse.de>
1345
1346         * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
1347         set_loop_copy and initialize_original_copy_tables.
1348
1349 2014-02-06  Alex Velenko  <Alex.Velenko@arm.com>
1350
1351         * config/aarch64/aarch64-simd.md
1352         (aarch64_ashr_simddi): Change QI to SI.
1353
1354 2014-02-05  Jan Hubicka  <hubicka@ucw.cz>
1355             Jakub Jelinek  <jakub@redhat.com>
1356
1357         PR middle-end/60013
1358         * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
1359         of the dataflow.
1360
1361 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1362
1363         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
1364         CODE_FOR_altivec_vpku[hw]um to
1365         CODE_FOR_altivec_vpku[hw]um_direct.
1366         * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
1367         UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
1368         (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
1369         UNSPEC_VUNPACK_LO_SIGN_DIRECT.
1370
1371 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1372
1373         * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
1374         generation for -maltivec=be.
1375         (altivec_vsumsws): Simplify redundant test.
1376
1377 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1378
1379         * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
1380         (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
1381         (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
1382         (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
1383         gen_altivec_vpkuwum.
1384         (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
1385         BYTES_BIG_ENDIAN.
1386         (altivec_vpks<VI_char>ss): Likewise.
1387         (altivec_vpks<VI_char>us): Likewise.
1388         (altivec_vpku<VI_char>us): Likewise.
1389         (altivec_vpku<VI_char>um): Likewise.
1390         (altivec_vpku<VI_char>um_direct): New (copy of
1391         altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
1392         internal use).
1393         (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
1394         target is little endian and -maltivec=be is not specified.
1395         (*altivec_vupkhs<VU_char>_direct): New (copy of
1396         altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
1397         (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
1398         target is little endian and -maltivec=be is not specified.
1399         (*altivec_vupkls<VU_char>_direct): New (copy of
1400         altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
1401         (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
1402         little endian and -maltivec=be is not specified.
1403         (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
1404         little endian and -maltivec=be is not specified.
1405
1406 2014-02-05  Richard Henderson  <rth@redhat.com>
1407
1408         PR debug/52727
1409         * combine-stack-adj.c: Revert r206943.
1410         * sched-int.h (struct deps_desc): Add last_args_size.
1411         * sched-deps.c (init_deps): Initialize it.
1412         (sched_analyze_insn): Add OUTPUT dependencies between insns that
1413         contain REG_ARGS_SIZE notes.
1414
1415 2014-02-05  Jan Hubicka  <hubicka@ucw.cz>
1416
1417         * lto-cgraph.c (asm_nodes_output): Make global.
1418         * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
1419         * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
1420         (driver_handle_option): Handle OPT_fwpa.
1421
1422 2014-02-05  Jakub Jelinek  <jakub@redhat.com>
1423
1424         PR ipa/59947
1425         * ipa-devirt.c (possible_polymorphic_call_targets): Fix
1426         a comment typo and formatting issue.  If odr_hash hasn't been
1427         created, return vNULL and set *completep to false.
1428
1429         PR middle-end/57499
1430         * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
1431         bb with no successors.
1432
1433 2014-02-05  James Greenhalgh  <james.greenhalgh@arm.com>
1434
1435         PR target/59718
1436         * doc/invoke.texi (-march): Clarify documentation for ARM.
1437         (-mtune): Likewise.
1438         (-mcpu): Likewise.
1439
1440 2014-02-05  Richard Biener  <rguenther@suse.de>
1441
1442         * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
1443         when not vectorizing because of too many alias checks.
1444         * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
1445         Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
1446
1447 2014-02-05  Nick Clifton  <nickc@redhat.com>
1448
1449         * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
1450         accept extended registers in any mode when compiling for the MN10300.
1451
1452 2014-02-05  Yury Gribov  <y.gribov@samsung.com>
1453
1454         * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
1455         * ipa-inline.c (report_inline_failed_reason): Handle mismatched
1456         sanitization attributes.
1457         (can_inline_edge_p): Likewise.
1458         (sanitize_attrs_match_for_inline_p): New function.
1459
1460 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
1461
1462         * ipa-prop.c (detect_type_change): Shor circuit testing of
1463         type changes on THIS pointer.
1464
1465 2014-02-04  John David Anglin  <danglin@gcc.gnu.org>
1466
1467         PR target/59777
1468         * config/pa/pa.c (legitimize_tls_address): Return original address
1469         if not passed a SYMBOL_REF rtx.
1470         (hppa_legitimize_address): Call legitimize_tls_address for all TLS
1471         addresses.
1472         (pa_emit_move_sequence): Simplify TLS source operands.
1473         (pa_legitimate_constant_p): Reject all TLS constants.
1474         * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
1475         (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
1476
1477 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
1478
1479         * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
1480         groups when we know they are controlled by LTO.
1481         * varasm.c (default_binds_local_p_1): If object is in other partition,
1482         it will be resolved locally.
1483
1484 2014-02-04  Bernd Edlinger  <bernd.edlinger@hotmail.de>
1485
1486         * config/host-linux.c (linux_gt_pch_use_address): Don't
1487         use SSIZE_MAX because it is not always defined.
1488
1489 2014-02-04  Vladimir Makarov  <vmakarov@redhat.com>
1490
1491         PR bootstrap/59913
1492         * lra-constraints.c (need_for_split_p): Use more 3 reloads as
1493         threshold for pseudo splitting.
1494         (update_ebb_live_info): Process call argument hard registers and
1495         hard registers from insn definition too.
1496         (max_small_class_regs_num): New constant.
1497         (inherit_in_ebb): Update live hard regs through EBBs.  Update
1498         reloads_num only for small register classes.  Don't split for
1499         outputs of jumps.
1500
1501 2014-02-04  Markus Trippelsdorf  <markus@trippelsdorf.de>
1502
1503         PR ipa/60058
1504         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
1505         is non-null.
1506
1507 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
1508
1509         * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
1510         visibility is safe.
1511
1512 2014-02-04  Marek Polacek  <polacek@redhat.com>
1513
1514         * gdbinit.in (pel): Define.
1515
1516 2014-02-04  Bernd Edlinger  <bernd.edlinger@hotmail.de>
1517
1518         * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
1519         behavior.
1520
1521 2014-02-04  Richard Biener  <rguenther@suse.de>
1522
1523         PR lto/59723
1524         * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
1525         in function context local.
1526         (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
1527         * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
1528         similar to LTO_imported_decl_ref.
1529
1530 2014-02-04  Jakub Jelinek  <jakub@redhat.com>
1531
1532         PR tree-optimization/60002
1533         * cgraphclones.c (build_function_decl_skip_args): Clear
1534         DECL_LANG_SPECIFIC.
1535
1536         PR tree-optimization/60023
1537         * tree-if-conv.c (predicate_mem_writes): Pass true instead of
1538         false to gsi_replace.
1539         * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
1540         has been in some EH region and vec_stmt could throw, add
1541         vec_stmt into the same EH region.
1542         * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
1543         has no lhs, ignore it.
1544         * internal-fn.c (expand_MASK_LOAD): Likewise.
1545
1546         PR ipa/60026
1547         * tree-inline.c (copy_forbidden): Fail for
1548         __attribute__((optimize (0))) functions.
1549
1550         PR other/58712
1551         * omp-low.c (simd_clone_struct_copy): If from->inbranch
1552         is set, copy one less argument.
1553         (expand_simd_clones): Don't subtract clone_info->inbranch
1554         from simd_clone_struct_alloc argument.
1555
1556         PR rtl-optimization/57915
1557         * recog.c (simplify_while_replacing): If all unary/binary/relational
1558         operation arguments are constant, attempt to simplify those.
1559
1560         PR middle-end/59261
1561         * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
1562         if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
1563
1564 2014-02-04  Richard Biener  <rguenther@suse.de>
1565
1566         PR tree-optimization/60012
1567         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
1568         TBAA disambiguation to all DDRs.
1569
1570 2014-02-04  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
1571
1572         PR target/59788
1573         * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
1574         (LINK_SPEC): Use it for -shared, -shared-libgcc.
1575
1576 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
1577
1578         PR ipa/59882
1579         * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
1580
1581 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
1582
1583         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
1584         * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
1585
1586 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
1587
1588         PR ipa/59831
1589         * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
1590         to figure out targets of polymorphic calls with known decl.
1591         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1592         * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
1593         * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
1594         (get_polymorphic_call_info): ... here.
1595         (get_polymorphic_call_info_from_invariant): New function.
1596
1597 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
1598
1599         * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
1600         lookup via vtable pointer; check for type consistency
1601         and turn inconsitent facts into UNREACHABLE.
1602         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1603         * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
1604         type inconsistent querries; return UNREACHABLE instead.
1605
1606 2014-02-03  Richard Henderson  <rth@twiddle.net>
1607
1608         PR tree-opt/59924
1609         * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
1610         already processed this node.
1611         (normalize_one_pred_1): Pass along mark_set.
1612         (normalize_one_pred): Create and destroy a pointer_set_t.
1613         (normalize_one_pred_chain): Likewise.
1614
1615 2014-02-03  Laurent Aflonsi  <laurent.alfonsi@st.com>
1616
1617         PR gcov-profile/58602
1618         * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1619
1620 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
1621
1622         PR ipa/59831
1623         * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1624         -fno-devirtualize; try to devirtualize by the knowledge of
1625         virtual table pointer given by aggregate propagation.
1626         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1627         (ipa_print_node_jump_functions): Dump also offset that
1628         is relevant for polymorphic calls.
1629         (determine_known_aggregate_parts): Add arg_type parameter; use it
1630         instead of determining the type from pointer type.
1631         (ipa_compute_jump_functions_for_edge): Update call of
1632         determine_known_aggregate_parts.
1633         * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1634         (gimple_get_virt_method_for_binfo): ... here; simplify using
1635         vtable_pointer_value_to_vtable.
1636         * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1637         * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1638         to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1639         (vtable_pointer_value_to_vtable): Break out from ...; handle also
1640         POINTER_PLUS_EXPR.
1641         (vtable_pointer_value_to_binfo): ... here.
1642         * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1643
1644 2014-02-03  Teresa Johnson  <tejohnson@google.com>
1645
1646         * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1647         redef of outer loop index variable.
1648
1649 2014-02-03  Marc Glisse  <marc.glisse@inria.fr>
1650
1651         PR c++/53017
1652         PR c++/59211
1653         * doc/extend.texi (Function Attributes): Typo.
1654
1655 2014-02-03  Cong Hou  <congh@google.com>
1656
1657         PR tree-optimization/60000
1658         * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1659         if the vectorized statement is a store.  A store statement can only
1660         appear at the end of pattern statements.
1661
1662 2014-02-03  H.J. Lu  <hongjiu.lu@intel.com>
1663
1664         * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1665         (ix86_option_override_internal): Default long double to 64-bit for
1666         32-bit Bionic and to 128-bit for 64-bit Bionic.
1667
1668         * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1669         TARGET_LONG_DOUBLE_128 is true.
1670         (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1671
1672         * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1673         (mlong-double-64): Negate -mlong-double-128.
1674         (mlong-double-128): New option.
1675
1676         * config/i386/i386-c.c (ix86_target_macros): Define
1677         __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1678
1679         * doc/invoke.texi: Document -mlong-double-128.
1680
1681 2014-02-03  H.J. Lu  <hongjiu.lu@intel.com>
1682
1683         PR rtl-optimization/60024
1684         * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1685
1686 2014-02-03  Markus Trippelsdorf  <markus@trippelsdorf.de>
1687
1688         * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1689
1690 2014-02-03  Andrey Belevantsev  <abel@ispras.ru>
1691
1692         PR rtl-optimization/57662
1693         * sel-sched.c (code_motion_path_driver): Do not mark already not
1694         existing blocks in the visiting bitmap.
1695
1696 2014-02-03  Andrey Belevantsev  <abel@ispras.ru>
1697
1698         * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1699         on the insn being emitted.
1700
1701 2014-02-03  James Greenhalgh  <james.greenhalgh@arm.com>
1702             Will Deacon  <will.deacon@arm.com>
1703
1704         * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1705
1706 2014-02-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1707
1708         * config/arm/arm-tables.opt: Regenerate.
1709
1710 2014-02-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1711
1712         * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1713         for vector types other than V16QImode.
1714         * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1715         define_expand, and call altivec_expand_vec_perm_le when producing
1716         code with little endian element order.
1717         (*altivec_vperm_<mode>_internal): New insn having previous
1718         behavior of altivec_vperm_<mode>.
1719         (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1720         altivec_expand_vec_perm_le when producing code with little endian
1721         element order.
1722         (*altivec_vperm_<mode>_uns_internal): New insn having previous
1723         behavior of altivec_vperm_<mode>_uns.
1724
1725 2014-02-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1726
1727         * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1728         (altivec_vsumsws): Add handling for -maltivec=be with a little
1729         endian target.
1730         (altivec_vsumsws_direct): New.
1731         (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1732         gen_altivec_vsumsws.
1733
1734 2014-02-02  Jan Hubicka  <hubicka@ucw.cz>
1735
1736         * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1737         vtable_pointer_value_to_binfo): New functions.
1738         * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1739         * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1740
1741 2014-02-02  Sandra Loosemore  <sandra@codesourcery.com>
1742
1743         * config/nios2/nios2.md (load_got_register): Initialize GOT
1744         pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1745         * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1746
1747 2014-02-02  Jan Hubicka  <hubicka@ucw.cz>
1748
1749         * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1750         preserverd by passthrough, do not propagate the type.
1751
1752 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
1753
1754         * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1755         (mips_atomic_assign_expand_fenv): New function.
1756         (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1757
1758 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
1759
1760         * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1761         (__builtin_mips_set_fcsr): Likewise.
1762         * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1763         MIPS_USI_FTYPE_VOID.
1764         * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1765         (mips16_expand_set_fcsr): Likewise.
1766         * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1767         (mips16_set_fcsr_stub): Likewise.
1768         (mips16_get_fcsr_one_only_stub): New class.
1769         (mips16_set_fcsr_one_only_stub): Likewise.
1770         (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1771         (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1772         (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1773         (hard_float): New availability predicate.
1774         (mips_builtins): Add get_fcsr and set_fcsr.
1775         (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1776         * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1777         (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1778         (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1779         (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1780         patterns.
1781
1782 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
1783
1784         * config/mips/mips.c (mips_one_only_stub): New class.
1785         (mips_need_mips16_rdhwr_p): Replace with...
1786         (mips16_rdhwr_stub): ...this new variable.
1787         (mips16_stub_call_address): New function.
1788         (mips16_rdhwr_one_only_stub): New class.
1789         (mips_expand_thread_pointer): Use mips16_stub_call_address.
1790         (mips_output_mips16_rdhwr): Delete.
1791         (mips_finish_stub): New function.
1792         (mips_code_end): Use it to handle rdhwr stubs.
1793
1794 2014-02-02  Uros Bizjak  <ubizjak@gmail.com>
1795
1796         PR target/60017
1797         * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1798         when calculating size of integer atomic types.
1799
1800 2014-02-02  H.J. Lu  <hongjiu.lu@intel.com>
1801
1802         * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1803
1804 2014-02-01  Jakub Jelinek  <jakub@redhat.com>
1805
1806         PR tree-optimization/60003
1807         * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1808         * profile.c (branch_prob): Use gimple_call_builtin_p
1809         to check for BUILT_IN_SETJMP_RECEIVER.
1810         * tree-inline.c (copy_bb): Call notice_special_calls.
1811
1812 2014-01-31  Vladimir Makarov  <vmakarov@redhat.com>
1813
1814         PR bootstrap/59985
1815         * lra-constraints.c (process_alt_operands): Update reload_sum only
1816         on the first pass.
1817
1818 2014-01-31  Richard Henderson  <rth@redhat.com>
1819
1820         PR middle-end/60004
1821         * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1822         until after else_eh is processed.
1823
1824 2014-01-31  Ilya Tocar  <ilya.tocar@intel.com>
1825
1826         * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1827         (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1828         (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1829         in smmintrin.h, remove them.
1830         (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1831         * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1832         * config/i386/i386.md (ROUND_SAE): Fix value.
1833         * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1834         (const48_operand): New.
1835         * config/i386/subst.md (round), (round_expand): Use
1836         const_4_or_8_to_11_operand.
1837         (round_saeonly), (round_saeonly_expand): Use const48_operand.
1838
1839 2014-01-31  Ilya Tocar  <ilya.tocar@intel.com>
1840
1841         * config/i386/constraints.md (Yk): Swap meaning with k.
1842         * config/i386/i386.md (movhi_internal): Change Yk to k.
1843         (movqi_internal): Ditto.
1844         (*k<logic><mode>): Ditto.
1845         (*andhi_1): Ditto.
1846         (*andqi_1): Ditto.
1847         (kandn<mode>): Ditto.
1848         (*<code>hi_1): Ditto.
1849         (*<code>qi_1): Ditto.
1850         (kxnor<mode>): Ditto.
1851         (kortestzhi): Ditto.
1852         (kortestchi): Ditto.
1853         (kunpckhi): Ditto.
1854         (*one_cmplhi2_1): Ditto.
1855         (*one_cmplqi2_1): Ditto.
1856         * config/i386/sse.md (): Change k to Yk.
1857         (avx512f_load<mode>_mask): Ditto.
1858         (avx512f_blendm<mode>): Ditto.
1859         (avx512f_store<mode>_mask): Ditto.
1860         (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1861         (avx512f_storedqu<mode>_mask): Ditto.
1862         (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1863         Ditto.
1864         (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1865         (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1866         (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1867         (avx512f_maskcmp<mode>3): Ditto.
1868         (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1869         (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1870         (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1871         (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1872         (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1873         (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1874         (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1875         (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1876         (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1877         (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1878         (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1879         (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1880         (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1881         (vec_extract_lo_<mode>_maskm): Ditto.
1882         (vec_extract_hi_<mode>_maskm): Ditto.
1883         (avx512f_vternlog<mode>_mask): Ditto.
1884         (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1885         (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1886         (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1887         (avx512f_<code>v8div16qi2_mask): Ditto.
1888         (avx512f_<code>v8div16qi2_mask_store): Ditto.
1889         (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1890         (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1891         (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1892         (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1893         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1894         (*avx512pf_gatherpf<mode>df_mask): Ditto.
1895         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1896         (*avx512pf_scatterpf<mode>df_mask): Ditto.
1897         (avx512cd_maskb_vec_dupv8di): Ditto.
1898         (avx512cd_maskw_vec_dupv16si): Ditto.
1899         (avx512f_vpermi2var<mode>3_maskz): Ditto.
1900         (avx512f_vpermi2var<mode>3_mask): Ditto.
1901         (avx512f_vpermi2var<mode>3_mask): Ditto.
1902         (avx512f_vpermt2var<mode>3_maskz): Ditto.
1903         (*avx512f_gathersi<mode>): Ditto.
1904         (*avx512f_gathersi<mode>_2): Ditto.
1905         (*avx512f_gatherdi<mode>): Ditto.
1906         (*avx512f_gatherdi<mode>_2): Ditto.
1907         (*avx512f_scattersi<mode>): Ditto.
1908         (*avx512f_scatterdi<mode>): Ditto.
1909         (avx512f_compress<mode>_mask): Ditto.
1910         (avx512f_compressstore<mode>_mask): Ditto.
1911         (avx512f_expand<mode>_mask): Ditto.
1912         * config/i386/subst.md (mask): Change k to Yk.
1913         (mask_scalar_merge): Ditto.
1914         (sd): Ditto.
1915
1916 2014-01-31  Marc Glisse  <marc.glisse@inria.fr>
1917
1918         * doc/extend.texi (Vector Extensions): Document ?: in C++.
1919
1920 2014-01-31  Richard Biener  <rguenther@suse.de>
1921
1922         PR middle-end/59990
1923         * builtins.c (fold_builtin_memory_op): Make sure to not
1924         use a floating-point mode or a boolean or enumeral type for
1925         the copy operation.
1926
1927 2014-01-30  DJ Delorie  <dj@redhat.com>
1928
1929         * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1930         * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1931         * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1932         whenever main() has an epilogue.
1933
1934 2014-01-30  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1935
1936         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1937         unused variable "field".
1938         * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1939         (vsx_mergeh_<mode>): Likewise.
1940         * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1941         (altivec_vmrghh): Likewise.
1942         (altivec_vmrghw): Likewise.
1943         (altivec_vmrglb): Likewise.
1944         (altivec_vmrglh): Likewise.
1945         (altivec_vmrglw): Likewise.
1946         (altivec_vspltb): Add missing uses.
1947         (altivec_vsplth): Likewise.
1948         (altivec_vspltw): Likewise.
1949         (altivec_vspltsf): Likewise.
1950
1951 2014-01-30  Jakub Jelinek  <jakub@redhat.com>
1952
1953         PR target/59923
1954         * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1955         frame related instructions.
1956
1957 2014-01-30  Vladimir Makarov  <vmakarov@redhat.com>
1958
1959         PR rtl-optimization/59959
1960         * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1961         any reload of register whose subreg is invalid.
1962
1963 2014-01-30  Jakub Jelinek  <jakub@redhat.com>
1964
1965         * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1966         * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1967         Add missing return type - void.
1968
1969 2014-01-30  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1970
1971         * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1972         gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1973         remove element index adjustment for endian (now handled in vsx.md
1974         and altivec.md).
1975         (altivec_expand_vec_perm_const): Use
1976         gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1977         * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1978         (vsx_xxspltw_<mode>): Adjust element index for little endian.
1979         * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1980         define_expand and a new define_insn *altivec_vspltb_internal;
1981         adjust for -maltivec=be on a little endian target.
1982         (altivec_vspltb_direct): New.
1983         (altivec_vsplth): Divide into a define_expand and a new
1984         define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1985         little endian target.
1986         (altivec_vsplth_direct): New.
1987         (altivec_vspltw): Divide into a define_expand and a new
1988         define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1989         little endian target.
1990         (altivec_vspltw_direct): New.
1991         (altivec_vspltsf): Divide into a define_expand and a new
1992         define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1993         a little endian target.
1994
1995 2014-01-30  Richard Biener  <rguenther@suse.de>
1996
1997         PR tree-optimization/59993
1998         * tree-ssa-forwprop.c (associate_pointerplus): Check we
1999         can propagate form the earlier stmt and avoid the transform
2000         when the intermediate result is needed.
2001
2002 2014-01-30  Alangi Derick  <alangiderick@gmail.com>
2003
2004         * README.Portability: Fix typo.
2005
2006 2014-01-30  David Holsgrove <david.holsgrove@xilinx.com>
2007
2008         * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
2009         comparison_operator with ordered_comparison_operator.
2010
2011 2014-01-30  Nick Clifton  <nickc@redhat.com>
2012
2013         * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
2014         Rename to mn10300_store_multiple_regs.
2015         * config/mn10300/mn10300.c: Likewise.
2016         * config/mn10300/mn10300.md (store_movm): Fix typo: call
2017         store_multiple_regs.
2018         * config/mn10300/predicates.md (mn10300_store_multiple_operation):
2019         Call mn10300_store_multiple_regs.
2020
2021 2014-01-30  Nick Clifton  <nickc@redhat.com>
2022             DJ Delorie  <dj@redhat.com>
2023
2024         * config/rl78/rl78.c (register_sizes): Make the "upper half" of
2025         %fp 2 to keep registers after it properly word-aligned.
2026         (rl78_alloc_physical_registers_umul): Handle the case where both
2027         input operands are the same.
2028
2029 2014-01-30  Richard Biener  <rguenther@suse.de>
2030
2031         PR tree-optimization/59903
2032         * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
2033         check properly.
2034
2035 2014-01-30  Jason Merrill  <jason@redhat.com>
2036
2037         PR c++/59633
2038         * tree.c (walk_type_fields): Handle VECTOR_TYPE.
2039
2040         PR c++/59645
2041         * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
2042
2043 2014-01-30  Richard Biener  <rguenther@suse.de>
2044
2045         PR tree-optimization/59951
2046         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
2047
2048 2014-01-30  Savin Zlobec  <savin.zlobec@gmail.com>
2049
2050         PR target/59784
2051         * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
2052         SFmode to DFmode case.
2053
2054 2014-01-29  DJ Delorie  <dj@redhat.com>
2055
2056         * config/msp430/msp430.opt (-minrt): New.
2057         * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
2058         if -minrt given.
2059         (ENDFILE_SPEC): Likewise.
2060
2061 2014-01-29  Jan Hubicka  <hubicka@ucw.cz>
2062
2063         * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
2064         (estimate_function_body_sizes): Use it.
2065
2066 2014-01-29  Paolo Carlini  <paolo.carlini@oracle.com>
2067
2068         PR c++/58561
2069         * dwarf2out.c (is_cxx_auto): New.
2070         (is_base_type): Use it.
2071         (gen_type_die_with_usage): Likewise.
2072
2073 2014-01-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2074
2075         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const):  Use
2076         CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
2077         * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
2078         -maltivec=be with LE targets.
2079         (vsx_mergeh_<mode>): Likewise.
2080         * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
2081         (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
2082         (altivec_vmrghb): Replace with define_expand and new
2083         *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
2084         (altivec_vmrghb_direct): New define_insn.
2085         (altivec_vmrghh): Replace with define_expand and new
2086         *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
2087         (altivec_vmrghh_direct): New define_insn.
2088         (altivec_vmrghw): Replace with define_expand and new
2089         *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
2090         (altivec_vmrghw_direct): New define_insn.
2091         (*altivec_vmrghsf): Adjust for endianness.
2092         (altivec_vmrglb): Replace with define_expand and new
2093         *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
2094         (altivec_vmrglb_direct): New define_insn.
2095         (altivec_vmrglh): Replace with define_expand and new
2096         *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
2097         (altivec_vmrglh_direct): New define_insn.
2098         (altivec_vmrglw): Replace with define_expand and new
2099         *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
2100         (altivec_vmrglw_direct): New define_insn.
2101         (*altivec_vmrglsf): Adjust for endianness.
2102         (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2103         (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2104         (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2105         (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2106         (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
2107         (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
2108         (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
2109         (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
2110
2111 2014-01-29  Marcus Shawcroft  <marcus.shawcroft@arm.com>
2112
2113         * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
2114         (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
2115         whitespace.
2116
2117 2014-01-29  Richard Biener  <rguenther@suse.de>
2118
2119         PR tree-optimization/58742
2120         * tree-ssa-forwprop.c (associate_pointerplus): Rename to
2121         associate_pointerplus_align.
2122         (associate_pointerplus_diff): New function.
2123         (associate_pointerplus): Likewise.  Call associate_pointerplus_align
2124         and associate_pointerplus_diff.
2125
2126 2014-01-29  Richard Biener  <rguenther@suse.de>
2127
2128         * lto-streamer.h (LTO_major_version): Bump to 3.
2129         (LTO_minor_version): Reset to 0.
2130
2131 2014-01-29  Renlin Li  <Renlin.Li@arm.com>
2132
2133         * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
2134         * config/arm/arm.c (FL_FOR_ARCH7VE): New.
2135         (arm_file_start): Generate correct asm header for armv7ve.
2136         * config/arm/bpabi.h: Add multilib support for armv7ve.
2137         * config/arm/driver-arm.c: Change the architectures of cortex-a7
2138         and cortex-a15 to armv7ve.
2139         * config/arm/t-aprofile: Add multilib support for armv7ve.
2140         * doc/invoke.texi: Document -march=armv7ve.
2141
2142 2014-01-29  Richard Biener  <rguenther@suse.de>
2143
2144         PR tree-optimization/58742
2145         * tree-ssa-forwprop.c (associate_plusminus): Return true
2146         if we changed sth, defer EH cleanup to ...
2147         (ssa_forward_propagate_and_combine): ... here.  Call simplify_mult.
2148         (simplify_mult): New function.
2149
2150 2014-01-29  Jakub Jelinek  <jakub@redhat.com>
2151
2152         PR middle-end/59917
2153         PR tree-optimization/59920
2154         * tree.c (build_common_builtin_nodes): Remove
2155         __builtin_setjmp_dispatcher initialization.
2156         * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
2157         * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
2158         instead of gsi_after_labels + manually skipping debug stmts.
2159         Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
2160         ignore bbs with IFN_ABNORMAL_DISPATCHER.
2161         * tree-inline.c (copy_edges_for_bb): Remove
2162         can_make_abnormal_goto argument, instead add abnormal_goto_dest
2163         argument.  Ignore computed_goto_p stmts.  Don't call
2164         make_abnormal_goto_edges.  If a call might need abnormal edges
2165         for non-local gotos, see if it already has an edge to
2166         IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
2167         with true argument, don't do anything then, otherwise add
2168         EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
2169         (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
2170         caller.
2171         * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
2172         (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
2173         (lower_stmt): Don't set data->calls_builtin_setjmp.
2174         (lower_builtin_setjmp): Adjust comment.
2175         * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
2176         * tree-cfg.c (found_computed_goto): Remove.
2177         (factor_computed_gotos): Remove.
2178         (make_goto_expr_edges): Return bool, true for computed gotos.
2179         Don't call make_abnormal_goto_edges.
2180         (build_gimple_cfg): Don't set found_computed_goto, don't call
2181         factor_computed_gotos.
2182         (computed_goto_p): No longer static.
2183         (make_blocks): Don't set found_computed_goto.
2184         (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
2185         (make_edges): If make_goto_expr_edges returns true, push bb
2186         into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
2187         instead of calling make_abnormal_goto_edges push bb into ab_edge_call
2188         vector.  Record mapping between bbs and OpenMP regions if there
2189         are any, adjust make_gimple_omp_edges caller.  Call
2190         handle_abnormal_edges.
2191         (make_abnormal_goto_edges): Remove.
2192         * tree-cfg.h (make_abnormal_goto_edges): Remove.
2193         (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
2194         * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
2195         * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
2196         * internal-fn.def (ABNORMAL_DISPATCHER): New.
2197         * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
2198         filling *region also set *region_idx to (*region)->entry->index.
2199
2200         PR other/58712
2201         * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
2202         For REGs set ORIGINAL_REGNO.
2203
2204 2014-01-29  Bingfeng Mei  <bmei@broadcom.com>
2205
2206         * doc/md.texi: Mention that a target shouldn't implement
2207         vec_widen_(s|u)mul_even/odd pair if it is less efficient
2208         than hi/lo pair.
2209
2210 2014-01-29  Jakub Jelinek  <jakub@redhat.com>
2211
2212         PR tree-optimization/59594
2213         * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
2214         a copy of the datarefs vector rather than the vector itself.
2215
2216 2014-01-28  Jason Merrill  <jason@redhat.com>
2217
2218         PR c++/53756
2219         * dwarf2out.c (auto_die): New static.
2220         (gen_type_die_with_usage): Handle C++1y 'auto'.
2221         (gen_subprogram_die): If in-class DIE had 'auto', emit type again
2222         on definition.
2223
2224 2014-01-28  H.J. Lu  <hongjiu.lu@intel.com>
2225
2226         PR target/59672
2227         * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
2228         (SPEC_X32): Likewise.
2229         (SPEC_64): Likewise.
2230         * config/i386/i386.c (ix86_option_override_internal): Turn off
2231         OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
2232         for TARGET_16BIT.
2233         (x86_file_start): Output .code16gcc for TARGET_16BIT.
2234         * config/i386/i386.h (TARGET_16BIT): New macro.
2235         (TARGET_16BIT_P): Likewise.
2236         * config/i386/i386.opt: Add m16.
2237         * doc/invoke.texi: Document -m16.
2238
2239 2014-01-28  Jakub Jelinek  <jakub@redhat.com>
2240
2241         PR preprocessor/59935
2242         * input.c (location_get_source_line): Bail out on when line number
2243         is zero, and test the return value of lookup_or_add_file_to_cache_tab.
2244
2245 2014-01-28  Richard Biener  <rguenther@suse.de>
2246
2247         PR tree-optimization/58742
2248         * tree-ssa-forwprop.c (associate_plusminus): Handle
2249         pointer subtraction of the form (T)(P + A) - (T)P.
2250
2251 2014-01-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2252
2253         * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
2254         at const_int_cost.
2255
2256 2014-01-28  Richard Biener  <rguenther@suse.de>
2257
2258         Revert
2259         2014-01-28  Richard Biener  <rguenther@suse.de>
2260
2261         PR rtl-optimization/45364
2262         PR rtl-optimization/59890
2263         * var-tracking.c (local_get_addr_clear_given_value): Handle
2264         already cleared slot.
2265         (val_reset): Handle not allocated local_get_addr_cache.
2266         (vt_find_locations): Use post-order on the inverted CFG.
2267
2268 2014-01-28  Richard Biener  <rguenther@suse.de>
2269
2270         * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
2271
2272 2014-01-28  Richard Biener  <rguenther@suse.de>
2273
2274         PR rtl-optimization/45364
2275         PR rtl-optimization/59890
2276         * var-tracking.c (local_get_addr_clear_given_value): Handle
2277         already cleared slot.
2278         (val_reset): Handle not allocated local_get_addr_cache.
2279         (vt_find_locations): Use post-order on the inverted CFG.
2280
2281 2014-01-28  Alan Modra  <amodra@gmail.com>
2282
2283         * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
2284         * configure.ac <recursive call for build != host>: Define
2285         GENERATOR_FILE.  Comment.  Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
2286         and LD_FOR_BUILD too.
2287         * configure: Regenerate.
2288
2289 2014-01-27  Allan Sandfeld Jensen  <sandfeld@kde.org>
2290
2291         * config/i386/i386.c (get_builtin_code_for_version): Separate
2292         Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
2293         Broadwell from Haswell.
2294
2295 2014-01-27  Steve Ellcey  <sellcey@mips.com>
2296
2297         * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
2298         Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
2299         * config/mips/mips.c (mips_option_override): Change setting
2300         of TARGET_DSP.
2301         * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
2302         * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
2303         Change from Mask to Var.
2304
2305 2014-01-27  Jeff Law  <law@redhat.com>
2306
2307         * ipa-inline.c (inline_small_functions): Fix typo.
2308
2309 2014-01-27  Ilya Tocar  <ilya.tocar@intel.com>
2310
2311         * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
2312         (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
2313         (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
2314         (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
2315         (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
2316         (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
2317         (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
2318         (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
2319         (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
2320         (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
2321         (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
2322         (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
2323         (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
2324         (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
2325         (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
2326         (_mm512_storeu_epi64): Ditto.
2327         (_mm512_cmpge_epi32_mask): Ditto.
2328         (_mm512_cmpge_epu32_mask): Ditto.
2329         (_mm512_cmpge_epi64_mask): Ditto.
2330         (_mm512_cmpge_epu64_mask): Ditto.
2331         (_mm512_cmple_epi32_mask): Ditto.
2332         (_mm512_cmple_epu32_mask): Ditto.
2333         (_mm512_cmple_epi64_mask): Ditto.
2334         (_mm512_cmple_epu64_mask): Ditto.
2335         (_mm512_cmplt_epi32_mask): Ditto.
2336         (_mm512_cmplt_epu32_mask): Ditto.
2337         (_mm512_cmplt_epi64_mask): Ditto.
2338         (_mm512_cmplt_epu64_mask): Ditto.
2339         (_mm512_cmpneq_epi32_mask): Ditto.
2340         (_mm512_cmpneq_epu32_mask): Ditto.
2341         (_mm512_cmpneq_epi64_mask): Ditto.
2342         (_mm512_cmpneq_epu64_mask): Ditto.
2343         (_mm512_expand_pd): Ditto.
2344         (_mm512_expand_ps): Ditto.
2345         * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
2346         VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
2347         VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
2348         * config/i386/i386.c (ix86_builtins): Add
2349         IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
2350         IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
2351         IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
2352         IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
2353         IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
2354         IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
2355         IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
2356         IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
2357         IX86_BUILTIN_PMOVUSQW512_MEM.
2358         (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
2359         __builtin_ia32_pmovsqd512mem_mask,
2360         __builtin_ia32_pmovqd512mem_mask,
2361         __builtin_ia32_pmovusqw512mem_mask,
2362         __builtin_ia32_pmovsqw512mem_mask,
2363         __builtin_ia32_pmovqw512mem_mask,
2364         __builtin_ia32_pmovusdw512mem_mask,
2365         __builtin_ia32_pmovsdw512mem_mask,
2366         __builtin_ia32_pmovdw512mem_mask,
2367         __builtin_ia32_pmovqb512mem_mask,
2368         __builtin_ia32_pmovusqb512mem_mask,
2369         __builtin_ia32_pmovsqb512mem_mask,
2370         __builtin_ia32_pmovusdb512mem_mask,
2371         __builtin_ia32_pmovsdb512mem_mask,
2372         __builtin_ia32_pmovdb512mem_mask.
2373         (bdesc_args): Add __builtin_ia32_expanddf512,
2374         __builtin_ia32_expandsf512.
2375         (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
2376         VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
2377         VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
2378         * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
2379         (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
2380         (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
2381         (avx512f_<code>v8div16qi2_mask_store): This.
2382         (avx512f_expand<mode>): New.
2383
2384 2014-01-27  Kirill Yukhin  <kirill.yukhin@intel.com>
2385
2386         * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
2387         New.
2388         (_mm512_mask_prefetch_i64gather_pd): Ditto.
2389         (_mm512_prefetch_i32scatter_pd): Ditto.
2390         (_mm512_mask_prefetch_i32scatter_pd): Ditto.
2391         (_mm512_prefetch_i64scatter_pd): Ditto.
2392         (_mm512_mask_prefetch_i64scatter_pd): Ditto.
2393         (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
2394         (_mm512_mask_prefetch_i64gather_ps): Ditto.
2395         (_mm512_prefetch_i32scatter_ps): Ditto.
2396         (_mm512_mask_prefetch_i32scatter_ps): Ditto.
2397         (_mm512_prefetch_i64scatter_ps): Ditto.
2398         (_mm512_mask_prefetch_i64scatter_ps): Ditto.
2399         * config/i386/i386-builtin-types.def: Define
2400         VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
2401         and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
2402         * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
2403         IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
2404         IX86_BUILTIN_SCATTERPFQPD.
2405         (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
2406         __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
2407         __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
2408         __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
2409         __builtin_ia32_scatterpfqps.
2410         (ix86_expand_builtin): Expand new built-ins.
2411         * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
2412         fix memory access data type.
2413         (*avx512pf_gatherpf<mode>_mask): Ditto.
2414         (*avx512pf_gatherpf<mode>): Ditto.
2415         (avx512pf_scatterpf<mode>): Ditto.
2416         (*avx512pf_scatterpf<mode>_mask): Ditto.
2417         (*avx512pf_scatterpf<mode>): Ditto.
2418         (GATHER_SCATTER_SF_MEM_MODE): New.
2419         (avx512pf_gatherpf<mode>df): Ditto.
2420         (*avx512pf_gatherpf<mode>df_mask): Ditto.
2421         (*avx512pf_scatterpf<mode>df): Ditto.
2422
2423 2014-01-27  Jakub Jelinek  <jakub@redhat.com>
2424
2425         PR bootstrap/59934
2426         * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
2427         know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
2428         reached.
2429
2430 2014-01-27  James Greenhalgh  <james.greenhalgh@arm.com>
2431
2432         * common/config/arm/arm-common.c
2433         (arm_rewrite_mcpu): Handle multiple names.
2434         * config/arm/arm.h
2435         (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2436
2437 2014-01-27  James Greenhalgh  <james.greenhalgh@arm.com>
2438
2439         * gimple-builder.h (create_gimple_tmp): Delete.
2440
2441 2014-01-27  Christian Bruel  <christian.bruel@st.com>
2442
2443         * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
2444         words comparisons.
2445
2446 2014-01-26  John David Anglin  <danglin@gcc.gnu.org>
2447
2448         * config/pa/pa.md (call): Generate indirect long calls to non-local
2449         functions when outputing 32-bit code.
2450         (call_value): Likewise except for special call to buggy powf function.
2451
2452         * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
2453         portable runtime and PIC indirect calls.
2454         (pa_output_indirect_call): Remove unnecessary nop from portable runtime
2455         and PIC call sequences.  Use ldo instead of blr to set return register
2456         in PIC call sequence.
2457
2458 2014-01-25  Walter Lee  <walt@tilera.com>
2459
2460         * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
2461         avoid clobbering a live register.
2462
2463 2014-01-25  Walter Lee  <walt@tilera.com>
2464
2465         * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
2466         Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
2467         * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
2468         Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
2469
2470 2014-01-25  Walter Lee  <walt@tilera.com>
2471
2472         * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
2473         arguments on even registers.
2474         (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
2475         STACK_BOUNDARY.
2476         * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
2477         (BIGGEST_ALIGNMENT): Ditto.
2478         (BIGGEST_FIELD_ALIGNMENT): Ditto.
2479
2480 2014-01-25  Walter Lee  <walt@tilera.com>
2481
2482         * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
2483         insns before bundling.
2484         * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
2485
2486 2014-01-25  Walter Lee  <walt@tilera.com>
2487
2488         * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
2489         PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
2490         * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
2491
2492 2014-01-25  Richard Sandiford  <rdsandiford@googlemail.com>
2493
2494         * config/mips/constraints.md (kl): Delete.
2495         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
2496         define expands, using...
2497         (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
2498         instructions for MIPS16.
2499         (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
2500         non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
2501
2502 2014-01-25  Walter Lee  <walt@tilera.com>
2503
2504         * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
2505         (clzdi2): Ditto.
2506         (ffsdi2): Ditto.
2507
2508 2014-01-25  Walter Lee  <walt@tilera.com>
2509
2510         * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
2511         (TARGET_EXPAND_TO_RTL_HOOK): Define.
2512
2513 2014-01-25  Richard Sandiford  <rdsandiford@googlemail.com>
2514
2515         * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
2516         Handle XOR.
2517
2518 2014-01-25  Jakub Jelinek  <jakub@redhat.com>
2519
2520         * print-rtl.c (in_call_function_usage): New var.
2521         (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
2522         EXPR_LIST mode as mode and not as reg note name.
2523
2524         PR middle-end/59561
2525         * cfgloopmanip.c (copy_loop_info): If
2526         loop->warned_aggressive_loop_optimizations, make sure
2527         the flag is set in target loop too.
2528
2529 2014-01-24  Balaji V. Iyer  <balaji.v.iyer@intel.com>
2530
2531         * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
2532         flag_cilkplus.
2533         * builtins.def: Likewise.
2534         * cilk.h (fn_contains_cilk_spawn_p): Likewise.
2535         * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
2536         * ira.c (ira_setup_eliminable_regset): Likewise.
2537         * omp-low.c (gate_expand_omp): Likewise.
2538         (execute_lower_omp): Likewise.
2539         (diagnose_sb_0): Likewise.
2540         (gate_diagnose_omp_blocks): Likewise.
2541         (simd_clone_clauses_extract): Likewise.
2542         (gate): Likewise.
2543
2544 2014-01-24  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2545
2546         * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
2547         correction for little endian...
2548         * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
2549         here.
2550
2551 2014-01-24  Jeff Law  <law@redhat.com>
2552
2553         PR tree-optimization/59919
2554         * tree-vrp.c (find_assert_locations_1): Do not register asserts
2555         for non-returning calls.
2556
2557 2014-01-24  James Greenhalgh  <james.greenhalgh@arm.com>
2558
2559         * common/config/aarch64/aarch64-common.c
2560         (aarch64_rewrite_mcpu): Handle multiple names.
2561         * config/aarch64/aarch64.h
2562         (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2563
2564 2014-01-24  Dodji Seketeli  <dodji@redhat.com>
2565
2566         * input.c (add_file_to_cache_tab): Handle the case where fopen
2567         returns NULL.
2568
2569 2014-01-23  H.J. Lu  <hongjiu.lu@intel.com>
2570
2571         PR target/59929
2572         * config/i386/i386.md (pushsf splitter): Get stack adjustment
2573         from push operand if code of push isn't PRE_DEC.
2574
2575 2014-01-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
2576
2577         PR target/59909
2578         * doc/invoke.texi (RS/6000 and PowerPC Options): Document
2579         -mquad-memory-atomic.  Update -mquad-memory documentation to say
2580         it is only used for non-atomic loads/stores.
2581
2582         * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
2583         -mquad-memory or -mquad-memory-atomic switches.
2584
2585         * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
2586         -mquad-memory-atomic to ISA 2.07 support.
2587
2588         * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
2589         to separate support of normal quad word memory operations (ldq, stq)
2590         from the atomic quad word memory operations.
2591
2592         * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2593         support to separate non-atomic quad word operations from atomic
2594         quad word operations.  Disable non-atomic quad word operations in
2595         little endian mode so that we don't have to swap words after the
2596         load and before the store.
2597         (quad_load_store_p): Add comment about atomic quad word support.
2598         (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
2599         options printed with -mdebug=reg.
2600
2601         * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
2602         -mquad-memory-atomic as the test for whether we have quad word
2603         atomic instructions.
2604         (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
2605         or -mp8-vector are used, allow byte/half-word atomic operations.
2606
2607         * config/rs6000/sync.md (load_lockedti): Insure that the address
2608         is a proper indexed or indirect address for the lqarx instruction.
2609         On little endian systems, swap the hi/lo registers after the lqarx
2610         instruction.
2611         (load_lockedpti): Use indexed_or_indirect_operand predicate to
2612         insure the address is valid for the lqarx instruction.
2613         (store_conditionalti): Insure that the address is a proper indexed
2614         or indirect address for the stqcrx. instruction.  On little endian
2615         systems, swap the hi/lo registers before doing the stqcrx.
2616         instruction.
2617         (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2618         insure the address is valid for the stqcrx. instruction.
2619
2620         * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2621         Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2622         type of quad memory support is available.
2623
2624 2014-01-23  Vladimir Makarov  <vmakarov@redhat.com>
2625
2626         PR regression/59915
2627         * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2628         there is a danger of looping.
2629
2630 2014-01-23  Pat Haugen  <pthaugen@us.ibm.com>
2631
2632         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2633         force flag_ira_loop_pressure if set via command line.
2634
2635 2014-01-23  Alex Velenko  <Alex.Velenko@arm.com>
2636
2637         * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2638         (ashr_simd): New builtin handling DI mode.
2639         * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2640         (aarch64_sshr_simddi): New match pattern.
2641         * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2642         (vshrd_n_s64): Likewise.
2643         * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2644
2645 2014-01-23  Nick Clifton  <nickc@redhat.com>
2646
2647         * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2648         (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2649         favour of mcu specific scripts.
2650         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2651         430x multilibs.
2652
2653 2014-01-23  James Greenhalgh  <james.greenhalgh@arm.com>
2654             Alex Velenko  <Alex.Velenko@arm.com>
2655
2656         * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2657         (vaddv_s16): Likewise.
2658         (vaddv_s32): Likewise.
2659         (vaddv_u8): Likewise.
2660         (vaddv_u16): Likewise.
2661         (vaddv_u32): Likewise.
2662         (vaddvq_s8): Likewise.
2663         (vaddvq_s16): Likewise.
2664         (vaddvq_s32): Likewise.
2665         (vaddvq_s64): Likewise.
2666         (vaddvq_u8): Likewise.
2667         (vaddvq_u16): Likewise.
2668         (vaddvq_u32): Likewise.
2669         (vaddvq_u64): Likewise.
2670         (vaddv_f32): Likewise.
2671         (vaddvq_f32): Likewise.
2672         (vaddvq_f64): Likewise.
2673         (vmaxv_f32): Likewise.
2674         (vmaxv_s8): Likewise.
2675         (vmaxv_s16): Likewise.
2676         (vmaxv_s32): Likewise.
2677         (vmaxv_u8): Likewise.
2678         (vmaxv_u16): Likewise.
2679         (vmaxv_u32): Likewise.
2680         (vmaxvq_f32): Likewise.
2681         (vmaxvq_f64): Likewise.
2682         (vmaxvq_s8): Likewise.
2683         (vmaxvq_s16): Likewise.
2684         (vmaxvq_s32): Likewise.
2685         (vmaxvq_u8): Likewise.
2686         (vmaxvq_u16): Likewise.
2687         (vmaxvq_u32): Likewise.
2688         (vmaxnmv_f32): Likewise.
2689         (vmaxnmvq_f32): Likewise.
2690         (vmaxnmvq_f64): Likewise.
2691         (vminv_f32): Likewise.
2692         (vminv_s8): Likewise.
2693         (vminv_s16): Likewise.
2694         (vminv_s32): Likewise.
2695         (vminv_u8): Likewise.
2696         (vminv_u16): Likewise.
2697         (vminv_u32): Likewise.
2698         (vminvq_f32): Likewise.
2699         (vminvq_f64): Likewise.
2700         (vminvq_s8): Likewise.
2701         (vminvq_s16): Likewise.
2702         (vminvq_s32): Likewise.
2703         (vminvq_u8): Likewise.
2704         (vminvq_u16): Likewise.
2705         (vminvq_u32): Likewise.
2706         (vminnmv_f32): Likewise.
2707         (vminnmvq_f32): Likewise.
2708         (vminnmvq_f64): Likewise.
2709
2710 2014-01-23  James Greenhalgh  <james.greenhalgh@arm.com>
2711
2712         * config/aarch64/aarch64-simd.md
2713         (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2714         (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2715         (*aarch64_mul3_elt<mode>): Likewise.
2716         (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2717         (*aarch64_mul3_elt_to_64v2df): Likewise.
2718         (*aarch64_mla_elt<mode>): Likewise.
2719         (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2720         (*aarch64_mls_elt<mode>): Likewise.
2721         (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2722         (*aarch64_fma4_elt<mode>): Likewise.
2723         (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2724         (*aarch64_fma4_elt_to_64v2df): Likewise.
2725         (*aarch64_fnma4_elt<mode>): Likewise.
2726         (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2727         (*aarch64_fnma4_elt_to_64v2df): Likewise.
2728         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2729         (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2730         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2731         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2732         (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2733         (aarch64_sqdmull_lane<mode>_internal): Likewise.
2734         (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2735
2736 2013-01-23  Alex Velenko  <Alex.Velenko@arm.com>
2737
2738         * config/aarch64/aarch64-simd.md
2739         (aarch64_be_checked_get_lane<mode>): New define_expand.
2740         * config/aarch64/aarch64-simd-builtins.def
2741         (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2742         New builtin definition.
2743         * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2744         Use new safe be builtin.
2745
2746 2014-01-23  Alex Velenko  <Alex.Velenko@arm.com>
2747
2748         * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2749         New define_insn.
2750         (aarch64_be_st1<mode>): Likewise.
2751         (aarch_ld1<VALL:mode>): Define_expand modified.
2752         (aarch_st1<VALL:mode>): Likewise.
2753         * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2754         (UNSPEC_ST1): Likewise.
2755
2756 2014-01-23  David Holsgrove <david.holsgrove@xilinx.com>
2757
2758         * config/microblaze/microblaze.md: Add trap insn and attribute
2759
2760 2014-01-23  Dodji Seketeli  <dodji@redhat.com>
2761
2762         PR preprocessor/58580
2763         * input.h (location_get_source_line): Take an additional line_size
2764         parameter.
2765         (void diagnostics_file_cache_fini): Declare new function.
2766         * input.c (struct fcache): New type.
2767         (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2768         New static constants.
2769         (diagnostic_file_cache_init, total_lines_num)
2770         (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2771         (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2772         (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2773         (get_next_line, read_next_line, goto_next_line, read_line_num):
2774         New static function definitions.
2775         (diagnostic_file_cache_fini): New function.
2776         (location_get_source_line): Take an additional output line_len
2777         parameter.  Re-write using lookup_or_add_file_to_cache_tab and
2778         read_line_num.
2779         * diagnostic.c (diagnostic_finish): Call
2780         diagnostic_file_cache_fini.
2781         (adjust_line): Take an additional input parameter for the length
2782         of the line, rather than calculating it with strlen.
2783         (diagnostic_show_locus): Adjust the use of
2784         location_get_source_line and adjust_line with respect to their new
2785         signature.  While displaying a line now, do not stop at the first
2786         null byte.  Rather, display the zero byte as a space and keep
2787         going until we reach the size of the line.
2788         * Makefile.in: Add vec.o to OBJS-libcommon
2789
2790 2014-01-23  Kirill Yukhin  <kirill.yukhin@intel.com>
2791             Ilya Tocar     <ilya.tocar@intel.com>
2792
2793         * config/i386/avx512fintrin.h (_mm512_kmov): New.
2794         * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2795         (__builtin_ia32_kmov16): Ditto.
2796         * config/i386/i386.md (UNSPEC_KMOV): New.
2797         (kmovw): Ditto.
2798
2799 2014-01-23  Kirill Yukhin  <kirill.yukhin@intel.com>
2800
2801         * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2802         (_mm512_storeu_si512): Ditto.
2803
2804 2014-01-23  Richard Sandiford  <rdsandiford@googlemail.com>
2805
2806         PR target/52125
2807         * rtl.h (get_referenced_operands): Declare.
2808         * recog.c (get_referenced_operands): New function.
2809         * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2810         operands have been referenced when recording LO_SUM references.
2811
2812 2014-01-22  David Holsgrove <david.holsgrove@xilinx.com>
2813
2814         * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2815
2816 2014-01-22  Jan Hubicka  <hubicka@ucw.cz>
2817
2818         * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2819         Enable for generic and recent AMD targets.
2820
2821 2014-01-22  Jan Hubicka  <hubicka@ucw.cz>
2822
2823         * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2824         ARG_SIZE note when adjustment was eliminated.
2825
2826 2014-01-22  Jeff Law  <law@redhat.com>
2827
2828         PR tree-optimization/59597
2829         * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2830         in file.  Accept new argument REGISTERING and use it to modify
2831         dump output appropriately.
2832         (register_jump_thread): Corresponding changes.
2833         (mark_threaded_blocks): Reinstate code to cancel unprofitable
2834         thread paths involving joiner blocks.  Add code to dump cancelled
2835         jump threading paths.
2836
2837 2014-01-22  Vladimir Makarov  <vmakarov@redhat.com>
2838
2839         PR rtl-optimization/59477
2840         * lra-constraints.c (inherit_in_ebb): Process call for living hard
2841         regs.  Update reloads_num and potential_reload_hard_regs for all insns.
2842
2843 2014-01-22  Tom Tromey  <tromey@redhat.com>
2844
2845         * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2846         PARAMS.
2847         * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2848
2849 2014-01-21  Vladimir Makarov  <vmakarov@redhat.com>
2850
2851         PR rtl-optimization/59896
2852         * lra-constraints.c (process_alt_operands): Check unused note for
2853         matched operands of insn with no output reloads.
2854
2855 2014-01-21  Richard Sandiford  <rdsandiford@googlemail.com>
2856
2857         * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2858         (mips_move_from_gpr_cost): Likewise.
2859
2860 2014-01-21  Vladimir Makarov  <vmakarov@redhat.com>
2861
2862         PR rtl-optimization/59858
2863         * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2864         ira_class_hard_regs_num.
2865         (process_alt_operands): Increase reject for dying matched operand.
2866
2867 2014-01-21  Jakub Jelinek  <jakub@redhat.com>
2868
2869         PR target/59003
2870         * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2871         smaller than size, perform several stores or loads and stores
2872         at dst + count - size to store or copy all of size bytes, rather
2873         than just last modesize bytes.
2874
2875 2014-01-20  DJ Delorie  <dj@redhat.com>
2876
2877         * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2878         that CLOBBERs are REGs before propogating their values.
2879
2880 2014-01-20  H.J. Lu  <hongjiu.lu@intel.com>
2881
2882         PR middle-end/59789
2883         * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2884         (cgraph_inline_failed_type): New function.
2885         * cgraph.h (DEFCIFCODE): Add type.
2886         (cgraph_inline_failed_type_t): New enum.
2887         (cgraph_inline_failed_type): New prototype.
2888         * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2889         FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2890         FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2891         LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2892         MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2893         RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2894         OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2895         INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2896         Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2897         FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2898         EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2899         OPTIMIZATION_MISMATCH.
2900         * tree-inline.c (expand_call_inline): Emit errors during
2901         early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2902
2903 2014-01-20  Uros Bizjak  <ubizjak@gmail.com>
2904
2905         PR target/59685
2906         * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2907         mode attribute in insn output.
2908
2909 2014-01-20  Eric Botcazou  <ebotcazou@adacore.com>
2910
2911         * output.h (output_constant): Delete.
2912         * varasm.c (output_constant): Make private.
2913
2914 2014-01-20  Alex Velenko  <Alex.Velenko@arm.com>
2915
2916         * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2917
2918 2014-01-20  Jakub Jelinek  <jakub@redhat.com>
2919
2920         PR middle-end/59860
2921         * tree.h (fold_builtin_strcat): New prototype.
2922         * builtins.c (fold_builtin_strcat): No longer static.  Add len
2923         argument, if non-NULL, don't call c_strlen.  Optimize
2924         directly into __builtin_memcpy instead of __builtin_strcpy.
2925         (fold_builtin_2): Adjust fold_builtin_strcat caller.
2926         * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2927
2928 2014-01-20  Uros Bizjak  <ubizjak@gmail.com>
2929
2930         * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2931         for SImode_address_operand operands, having only a REG argument.
2932
2933 2014-01-20  Marcus Shawcroft  <marcus.shawcroft@arm.com>
2934
2935         * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2936         loader name using mbig-endian.
2937         (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2938
2939 2014-01-20  James Greenhalgh  <james.greenhalgh@arm.com>
2940
2941         * doc/invoke.texi (-march): Clarify documentation for AArch64.
2942         (-mtune): Likewise.
2943         (-mcpu): Likewise.
2944
2945 2014-01-20  Tejas Belagod  <tejas.belagod@arm.com>
2946
2947         * config/aarch64/aarch64-protos.h
2948         (aarch64_cannot_change_mode_class_ptr): Declare.
2949         * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2950         aarch64_cannot_change_mode_class_ptr): New.
2951         * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2952         backend hook aarch64_cannot_change_mode_class.
2953
2954 2014-01-20  James Greenhalgh  <james.greenhalgh@arm.com>
2955
2956         * common/config/aarch64/aarch64-common.c
2957         (aarch64_handle_option): Don't handle any option order logic here.
2958         * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2959         selected_cpu, warn on architecture version mismatch.
2960         (aarch64_override_options): Fix parsing order for option strings.
2961
2962 2014-01-20  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
2963             Iain Sandoe  <iain@codesourcery.com>
2964
2965         PR bootstrap/59496
2966         * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2967         warning.  Amend comment to reflect current functionality.
2968
2969 2014-01-20  Richard Biener  <rguenther@suse.de>
2970
2971         PR middle-end/59860
2972         * builtins.c (fold_builtin_strcat): Remove case better handled
2973         by tree-ssa-strlen.c.
2974
2975 2014-01-20  Alan Lawrence  <alan.lawrence@arm.com>
2976
2977         * config/aarch64/aarch64.opt
2978         (mcpu, march, mtune): Make case-insensitive.
2979
2980 2014-01-20  Jakub Jelinek  <jakub@redhat.com>
2981
2982         PR target/59880
2983         * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2984         if operands[1] is a REG or ZERO_EXTEND of a REG.
2985
2986 2014-01-19  Jan Hubicka  <hubicka@ucw.cz>
2987
2988         * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2989
2990 2014-01-19  John David Anglin  <danglin@gcc.gnu.org>
2991
2992         * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2993         long non-pic millicode calls.
2994
2995 2014-01-19  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
2996
2997         * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2998
2999 2014-01-19  Kito Cheng  <kito@0xlab.org>
3000
3001         * builtins.c (expand_movstr): Check movstr expand done or fail.
3002
3003 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
3004             H.J. Lu  <hongjiu.lu@intel.com>
3005
3006         PR target/59379
3007         * config/i386/i386.md (*lea<mode>): Zero-extend return register
3008         to DImode for zero-extended addresses.
3009
3010 2014-01-19  Jakub Jelinek  <jakub@redhat.com>
3011
3012         PR rtl-optimization/57763
3013         * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
3014         on the new indirect jump_insn and increment LABEL_NUSES (label).
3015
3016 2014-01-18  H.J. Lu  <hongjiu.lu@intel.com>
3017
3018         PR bootstrap/59580
3019         PR bootstrap/59583
3020         * config.gcc (x86_archs): New variable.
3021         (x86_64_archs): Likewise.
3022         (x86_cpus): Likewise.
3023         Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
3024         --with-arch/--with-cpu= options.
3025         Support --with-arch=/--with-cpu={nehalem,westmere,
3026         sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
3027
3028 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
3029
3030         * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
3031         and PROCESSOR_ATHLON to simplify code.  Move "memory" calculation.
3032
3033 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
3034
3035         * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
3036
3037 2014-01-18  Jakub Jelinek  <jakub@redhat.com>
3038
3039         PR target/58944
3040         * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
3041         clear cpp_get_options (parse_in)->warn_unused_macros for
3042         ix86_target_macros_internal with cpp_define.
3043
3044 2014-01-18  Richard Sandiford  <rdsandiford@googlemail.com>
3045
3046         * jump.c (delete_related_insns): Keep (use (insn))s.
3047         * reorg.c (redundant_insn): Check for barriers too.
3048
3049 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
3050
3051         * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
3052
3053 2014-01-17  John David Anglin  <danglin@gcc.gnu.org>
3054
3055         * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
3056         call to $$dyncall when TARGET_LONG_CALLS is true.
3057
3058 2014-01-17  Jeff Law  <law@redhat.com>
3059
3060         * ree.c (combine_set_extension): Temporarily disable test for
3061         changing number of hard registers.
3062
3063 2014-01-17  Jan Hubicka  <hubicka@ucw.cz>
3064
3065         PR middle-end/58125
3066         * ipa-inline-analysis.c (inline_free_summary):
3067         Do not free summary of aliases.
3068
3069 2014-01-17  Jakub Jelinek  <jakub@redhat.com>
3070
3071         PR middle-end/59706
3072         * gimplify.c (gimplify_expr): Use create_tmp_var
3073         instead of create_tmp_var_raw.  If cond doesn't have
3074         integral type, don't add the IFN_ANNOTATE builtin at all.
3075
3076 2014-01-17  Martin Jambor  <mjambor@suse.cz>
3077
3078         PR ipa/59736
3079         * ipa-cp.c (prev_edge_clone): New variable.
3080         (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
3081         Also resize prev_edge_clone vector.
3082         (ipcp_edge_duplication_hook): Also update prev_edge_clone.
3083         (ipcp_edge_removal_hook): New function.
3084         (ipcp_driver): Register ipcp_edge_removal_hook.
3085
3086 2014-01-17  Andrew Pinski <apinski@cavium.com>
3087             Steve Ellcey  <sellcey@mips.com>
3088
3089         PR target/59462
3090         * config/mips/mips.c (mips_print_operand): Check operand mode instead
3091         of operator mode.
3092
3093 2014-01-17  Jeff Law  <law@redhat.com>
3094
3095         PR middle-end/57904
3096         * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
3097         so that pass_ccp runs first.
3098
3099 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
3100
3101         * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
3102         (ix86_adjust_cost): Use !TARGET_XXX.
3103         (do_reorder_for_imul): Likewise.
3104         (swap_top_of_ready_list): Likewise.
3105         (ix86_sched_reorder): Likewise.
3106
3107 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
3108
3109         * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3110         PROCESSOR_INTEL.  Treat like PROCESSOR_GENERIC.
3111         * config/i386/i386.c (intel_memcpy): New.  Duplicate slm_memcpy.
3112         (intel_memset): New.  Duplicate slm_memset.
3113         (intel_cost): New.  Duplicate slm_cost.
3114         (m_INTEL): New macro.
3115         (processor_target_table): Add "intel".
3116         (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
3117         with PROCESSOR_INTEL for "intel".
3118         (ix86_lea_outperforms): Support PROCESSOR_INTEL.  Duplicate
3119         PROCESSOR_SILVERMONT.
3120         (ix86_issue_rate): Likewise.
3121         (ix86_adjust_cost): Likewise.
3122         (ia32_multipass_dfa_lookahead): Likewise.
3123         (swap_top_of_ready_list): Likewise.
3124         (ix86_sched_reorder): Likewise.
3125         (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
3126         instead of TARGET_OPT_AGU.
3127         * config/i386/i386.h (TARGET_INTEL): New.
3128         (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
3129         (processor_type): Add PROCESSOR_INTEL.
3130         * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
3131         Add X86_TUNE_AVOID_LEA_FOR_ADDR.
3132
3133 2014-01-17  Marek Polacek  <polacek@redhat.com>
3134
3135         PR c/58346
3136         * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
3137         size is zero.
3138
3139 2014-01-17  Richard Biener  <rguenther@suse.de>
3140
3141         PR tree-optimization/46590
3142         * opts.c (default_options_table): Add entries for
3143         OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
3144         all enabled at -O1 but not for -Og.
3145         * common.opt (fbranch-count-reg): Remove Init(1).
3146         (fmove-loop-invariants): Likewise.
3147         (ftree-pta): Likewise.
3148
3149 2014-01-17  Jakub Jelinek  <jakub@redhat.com>
3150
3151         * config/i386/i386.c (ix86_data_alignment): For compatibility with
3152         (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
3153         decls to at least the GCC 4.8 used alignments.
3154
3155         PR fortran/59440
3156         * tree-nested.c (convert_nonlocal_reference_stmt,
3157         convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
3158         of GIMPLE_BIND stmts, adjust associated decls.
3159
3160 2014-01-17  Richard Biener  <rguenther@suse.de>
3161
3162         PR tree-optimization/46590
3163         * vec.h (vec<>::bseach): New member function implementing
3164         binary search according to C89 bsearch.
3165         (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
3166         * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
3167         bitmap pointer again.  Make accesses_in_loop a flat array.
3168         (mem_ref_obstack): New global.
3169         (outermost_indep_loop): Adjust for mem_ref->stored changes.
3170         (mark_ref_stored): Likewise.
3171         (ref_indep_loop_p_2): Likewise.
3172         (set_ref_stored_in_loop): New helper function.
3173         (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
3174         (memref_free): Adjust.
3175         (record_mem_ref_loc): Simplify.
3176         (gather_mem_refs_stmt): Adjust.
3177         (sort_locs_in_loop_postorder_cmp): New function.
3178         (analyze_memory_references): Sort accesses_in_loop after
3179         loop postorder number.
3180         (find_ref_loc_in_loop_cmp): New function.
3181         (for_all_locs_in_loop): Find relevant cluster of locs in
3182         accesses_in_loop and iterate without recursion.
3183         (execute_sm): Avoid uninit warning.
3184         (struct ref_always_accessed): Simplify.
3185         (ref_always_accessed::operator ()): Likewise.
3186         (ref_always_accessed_p): Likewise.
3187         (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
3188         loop postorder numbers here.
3189         (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
3190         numbers.
3191
3192 2014-01-17  Jan Hubicka  <hubicka@ucw.cz>
3193
3194         PR c++/57945
3195         * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
3196         on decls for which assemble_alias has been called.
3197
3198 2014-01-17  Nick Clifton  <nickc@redhat.com>
3199
3200         * config/msp430/msp430.opt: (mcpu): New option.
3201         * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
3202         (msp430_option_override): Parse target_cpu.  If the MCU name
3203         matches a generic string, clear target_mcu.
3204         (msp430_attr): Allow numeric interrupt values up to 63.
3205         (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
3206         * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
3207         option.
3208         * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
3209         Add mcpu matches.
3210         * config/msp430/msp430.md (popm): Use %J rather than %I.
3211         (addsi3): Use msp430_nonimmediate_operand for operand 2.
3212         (addhi_cy_i): Use immediate_operand for operand 2.
3213         * doc/invoke.texi: Document -mcpu option.
3214
3215 2014-01-17  Richard Biener  <rguenther@suse.de>
3216
3217         PR rtl-optimization/38518
3218         * df.h (df_analyze_loop): Declare.
3219         * df-core.c: Include cfgloop.h.
3220         (df_analyze_1): Split out main part of df_analyze.
3221         (df_analyze): Adjust.
3222         (loop_inverted_post_order_compute): New function.
3223         (loop_post_order_compute): Likewise.
3224         (df_analyze_loop): New function avoiding whole-function
3225         postorder computes.
3226         * loop-invariant.c (find_defs): Use df_analyze_loop.
3227         (find_invariants): Adjust.
3228         * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
3229
3230 2014-01-17  Zhenqiang Chen  <zhenqiang.chen@arm.com>
3231
3232         * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
3233         (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
3234
3235 2014-01-16  Ilya Enkovich  <ilya.enkovich@intel.com>
3236
3237         * ipa-ref.c (ipa_remove_stmt_references): Fix references
3238         traversal when removing references.
3239
3240 2014-01-16  Jan Hubicka  <hubicka@ucw.cz>
3241
3242         PR ipa/59775
3243         * tree.c (get_binfo_at_offset): Look harder for virtual bases.
3244
3245 2014-01-16  Bernd Schmidt  <bernds@codesourcery.com>
3246
3247         PR middle-end/56791
3248         * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
3249         pushing a reload for an autoinc when we had previously reloaded an
3250         inner part of the address.
3251
3252 2014-01-16  Jakub Jelinek  <jakub@redhat.com>
3253
3254         * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
3255         field.
3256         (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
3257         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
3258         when not giving up or versioning for alias only because of
3259         loop->safelen.
3260         (vect_analyze_data_ref_dependences): Set to true.
3261         * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
3262         is a GIMPLE_PHI.
3263         (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
3264         LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
3265         to the condition.
3266
3267         PR middle-end/58344
3268         * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
3269
3270         PR target/59839
3271         * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
3272         operand 0 predicate for gathers, use a new pseudo as subtarget.
3273
3274 2014-01-16  Vladimir Makarov  <vmakarov@redhat.com>
3275
3276         PR middle-end/59609
3277         * lra-constraints.c (process_alt_operands): Add printing debug info.
3278         Check absence of input/output reloads for matched operands too.
3279
3280 2014-01-16  Vladimir Makarov  <vmakarov@redhat.com>
3281
3282         PR rtl-optimization/59835
3283         * ira.c (ira_init_register_move_cost): Increase cost for
3284         impossible modes.
3285
3286 2014-01-16  Alan Lawrence  <alan.lawrence@arm.com>
3287
3288         * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
3289
3290 2014-01-16  Richard Earnshaw  <rearnsha@arm.com>
3291
3292         PR target/59780
3293         * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
3294         non-register objects.  Use gen_(high/low)part more consistently.
3295         Fix assertions.
3296
3297 2014-01-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
3298
3299         PR target/59844
3300         * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
3301         endian support, remove tests for WORDS_BIG_ENDIAN.
3302         (p8_mfvsrd_3_<mode>): Likewise.
3303         (reload_gpr_from_vsx<mode>): Likewise.
3304         (reload_gpr_from_vsxsf): Likewise.
3305         (p8_mfvsrd_4_disf): Likewise.
3306
3307 2014-01-16  Richard Biener  <rguenther@suse.de>
3308
3309         PR rtl-optimization/46590
3310         * lcm.c (compute_antinout_edge): Use postorder iteration.
3311         (compute_laterin): Use inverted postorder iteration.
3312
3313 2014-01-16  Nick Clifton  <nickc@redhat.com>
3314
3315         PR middle-end/28865
3316         * varasm.c (output_constant): Return the number of bytes actually
3317         emitted.
3318         (output_constructor_array_range): Update the field size with the
3319         number of bytes emitted by output_constant.
3320         (output_constructor_regular_field): Likewise.  Also do not
3321         complain if the total number of bytes emitted is now greater
3322         than the expected fieldpos.
3323         * output.h (output_constant): Update prototype and descriptive comment.
3324
3325 2014-01-16  Marek Polacek  <polacek@redhat.com>
3326
3327         PR middle-end/59827
3328         * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
3329         it is error_mark_node.
3330
3331 2014-01-15  Uros Bizjak  <ubizjak@gmail.com>
3332
3333         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
3334         VALID_AVX256_REG_OR_OI_MODE.
3335
3336 2014-01-15  Pat Haugen  <pthaugen@us.ibm.com>
3337
3338         * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
3339         current procedure should be profiled.
3340
3341 2014-01-15  Andrew Pinski  <apinski@cavium.com>
3342
3343         * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
3344         of moving from/to the STACK_REG register class.
3345
3346 2014-01-15  Richard Henderson  <rth@redhat.com>
3347
3348         PR debug/54694
3349         * reginfo.c (global_regs_decl): Globalize.
3350         * rtl.h (global_regs_decl): Declare.
3351         * ira.c (do_reload): Diagnose frame_pointer_needed and it
3352         reserved via global_regs.
3353
3354 2014-01-15  Teresa Johnson  <tejohnson@google.com>
3355
3356         * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
3357
3358 2014-01-15  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
3359
3360         * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
3361         and vmulosh rather than call gen_vec_widen_smult_*.
3362         (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
3363         than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
3364         (vec_widen_smult_even_v16qi): Likewise.
3365         (vec_widen_umult_even_v8hi): Likewise.
3366         (vec_widen_smult_even_v8hi): Likewise.
3367         (vec_widen_umult_odd_v16qi): Likewise.
3368         (vec_widen_smult_odd_v16qi): Likewise.
3369         (vec_widen_umult_odd_v8hi): Likewise.
3370         (vec_widen_smult_odd_v8hi): Likewise.
3371         (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
3372         vmuloub rather than call gen_vec_widen_umult_*.
3373         (vec_widen_umult_lo_v16qi): Likewise.
3374         (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
3375         vmulosb rather than call gen_vec_widen_smult_*.
3376         (vec_widen_smult_lo_v16qi): Likewise.
3377         (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
3378         rather than call gen_vec_widen_umult_*.
3379         (vec_widen_umult_lo_v8hi): Likewise.
3380         (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
3381         rather than call gen_vec_widen_smult_*.
3382         (vec_widen_smult_lo_v8hi): Likewise.
3383
3384 2014-01-15  Jeff Law  <law@redhat.com>
3385
3386         PR tree-optimization/59747
3387         * ree.c (find_and_remove_re): Properly handle case where a second
3388         eliminated extension requires widening a copy created for elimination
3389         of a prior extension.
3390         (combine_set_extension): Ensure that the number of hard regs needed
3391         for a destination register does not change when we widen it.
3392
3393 2014-01-15  Sebastian Huber  <sebastian.huber@embedded-brains.de>
3394
3395         * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
3396         (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
3397         (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
3398         (arm*-*-rtems*): Use t-rtems from existing tmake_file.
3399         (avr-*-rtems*): Likewise.
3400         (bfin*-rtems*): Likewise.
3401         (moxie-*-rtems*): Likewise.
3402         (h8300-*-rtems*): Likewise.
3403         (i[34567]86-*-rtems*): Likewise.
3404         (lm32-*-rtems*): Likewise.
3405         (m32r-*-rtems*): Likewise.
3406         (m68k-*-rtems*): Likewise.
3407         (microblaze*-*-rtems*): Likewise.
3408         (mips*-*-rtems*): Likewise.
3409         (powerpc-*-rtems*): Likewise.
3410         (sh-*-rtems*): Likewise.
3411         (sparc-*-rtems*): Likewise.
3412         (sparc64-*-rtems*): Likewise.
3413         (v850-*-rtems*): Likewise.
3414         (m32c-*-rtems*): Likewise.
3415
3416 2014-01-15  Vladimir Makarov  <vmakarov@redhat.com>
3417
3418         PR rtl-optimization/59511
3419         * ira.c (ira_init_register_move_cost): Use memory costs for some
3420         cases of register move cost calculations.
3421         * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
3422         instead of BB frequency.
3423         * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
3424         * lra-assigns.c (find_hard_regno_for): Ditto.
3425
3426 2014-01-15  Richard Biener  <rguenther@suse.de>
3427
3428         PR tree-optimization/59822
3429         * tree-vect-stmts.c (hoist_defs_of_uses): New function.
3430         (vectorizable_load): Use it to hoist defs of uses of invariant
3431         loads out of the loop.
3432
3433 2014-01-15  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
3434             Kugan Vivekanandarajah  <kuganv@linaro.org>
3435
3436         PR target/59695
3437         * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
3438         truncation.
3439
3440 2014-01-15  Richard Biener  <rguenther@suse.de>
3441
3442         PR rtl-optimization/59802
3443         * lcm.c (compute_available): Use inverted postorder to seed
3444         the initial worklist.
3445
3446 2014-01-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3447
3448         PR target/59803
3449         * config/s390/s390.c (s390_preferred_reload_class): Don't return
3450         ADDR_REGS for invalid symrefs in non-PIC code.
3451
3452 2014-01-15  Jakub Jelinek  <jakub@redhat.com>
3453
3454         PR other/58712
3455         * builtins.c (determine_block_size): Initialize *probable_max_size
3456         even if len_rtx is CONST_INT.
3457
3458 2014-01-14  Andrew Pinski  <apinski@cavium.com>
3459
3460         * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
3461         * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
3462         (cortexa53_tunings): Likewise.
3463         (aarch64_sched_issue_rate): New function.
3464         (TARGET_SCHED_ISSUE_RATE): Define.
3465
3466 2014-01-14  Vladimir Makarov  <vmakarov@redhat.com>
3467
3468         * ira-costs.c (find_costs_and_classes): Add missed
3469         ira_init_register_move_cost_if_necessary.
3470
3471 2014-01-14  Vladimir Makarov  <vmakarov@redhat.com>
3472
3473         PR target/59787
3474         * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
3475
3476 2014-01-14  H.J. Lu  <hongjiu.lu@intel.com>
3477
3478         PR target/59794
3479         * config/i386/i386.c (type_natural_mode): Add a bool parameter
3480         to indicate if type is used for function return value.  Warn ABI
3481         change if the vector mode isn't available for function return value.
3482         (ix86_function_arg_advance): Pass false to type_natural_mode.
3483         (ix86_function_arg): Likewise.
3484         (ix86_gimplify_va_arg): Likewise.
3485         (function_arg_32): Don't warn ABI change.
3486         (ix86_function_value): Pass true to type_natural_mode.
3487         (ix86_return_in_memory): Likewise.
3488         (ix86_struct_value_rtx): Removed.
3489         (TARGET_STRUCT_VALUE_RTX): Likewise.
3490
3491 2014-01-14  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
3492
3493         * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
3494         converting a conditional jump into a conditional return.
3495
3496 2014-01-14  Richard Biener  <rguenther@suse.de>
3497
3498         PR tree-optimization/58921
3499         PR tree-optimization/59006
3500         * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
3501         hoisting invariant stmts.
3502         * tree-vect-stmts.c (vectorizable_load): Insert the splat of
3503         invariant loads on the preheader edge if possible.
3504
3505 2014-01-14  Joey Ye  <joey.ye@arm.com>
3506
3507         * doc/plugin.texi (Building GCC plugins): Update to C++.
3508
3509 2014-01-14  Kirill Yukhin  <kirill.yukhin@intel.com>
3510
3511         * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
3512         (_mm_rcp28_round_ss): Ditto.
3513         (_mm_rsqrt28_round_sd): Ditto.
3514         (_mm_rsqrt28_round_ss): Ditto.
3515         (_mm_rcp28_sd): Ditto.
3516         (_mm_rcp28_ss): Ditto.
3517         (_mm_rsqrt28_sd): Ditto.
3518         (_mm_rsqrt28_ss): Ditto.
3519         * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
3520         * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
3521         * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
3522         (IX86_BUILTIN_RCP28SD): Ditto.
3523         (IX86_BUILTIN_RCP28SS): Ditto.
3524         (IX86_BUILTIN_RSQRT28SD): Ditto.
3525         (IX86_BUILTIN_RSQRT28SS): Ditto.
3526         (bdesc_special_args): Define __builtin_ia32_movntdqa512,
3527         __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
3528         __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
3529         (ix86_expand_special_args_builtin): Expand new FTYPE.
3530         * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
3531         (srcp14<mode>): Make insn unary.
3532         (avx512f_vmscalef<mode><round_name>): Use substed predicate.
3533         (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
3534         (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
3535         (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
3536         (avx512er_exp2<mode><mask_name><round_saeonly_name>):
3537         Fix rounding: make it SAE only.
3538         (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3539         Ditto.
3540         (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3541         Ditto.
3542         (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
3543         (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
3544         (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
3545         * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
3546         (round_saeonly_mask_scalar_operand4): Ditto.
3547         (round_saeonly_mask_scalar_op3): Ditto.
3548         (round_saeonly_mask_scalar_op4): Ditto.
3549
3550 2014-01-13  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3551
3552         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3553         Implement -maltivec=be for vec_insert and vec_extract.
3554
3555 2014-01-10  DJ Delorie  <dj@redhat.com>
3556
3557         * config/msp430/msp430.md (call_internal): Don't allow memory
3558         references with SP as the base register.
3559         (call_value_internal): Likewise.
3560         * config/msp430/constraints.md (Yc): New.  For memory references
3561         that don't use SP as a base register.
3562
3563         * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
3564         "an integer without a # prefix"
3565         * config/msp430/msp430.md (epilogue_helper): Use it.
3566
3567 2014-01-13  Jakub Jelinek  <jakub@redhat.com>
3568
3569         PR target/59617
3570         * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
3571         AVX512F gather builtins.
3572         * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
3573         on gather decls with INTEGER_TYPE masktype.
3574         (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
3575         directly into the builtin rather than hoisting it before loop.
3576
3577         PR tree-optimization/59387
3578         * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
3579         (scev_const_prop): If folded_casts and type has undefined overflow,
3580         use force_gimple_operand instead of force_gimple_operand_gsi and
3581         for each added stmt if it is assign with
3582         arith_code_with_undefined_signed_overflow, call
3583         rewrite_to_defined_overflow.
3584         * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
3585         gimple-fold.h instead.
3586         (arith_code_with_undefined_signed_overflow,
3587         rewrite_to_defined_overflow): Moved to ...
3588         * gimple-fold.c (arith_code_with_undefined_signed_overflow,
3589         rewrite_to_defined_overflow): ... here.  No longer static.
3590         Include gimplify-me.h.
3591         * gimple-fold.h (arith_code_with_undefined_signed_overflow,
3592         rewrite_to_defined_overflow): New prototypes.
3593
3594 2014-01-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3595
3596         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
3597
3598 2014-01-13  Eric Botcazou  <ebotcazou@adacore.com>
3599
3600         * builtins.c (get_object_alignment_2): Minor tweak.
3601         * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
3602
3603 2014-01-13  Christian Bruel  <christian.bruel@st.com>
3604
3605         * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
3606         optimized non constant lengths.
3607
3608 2014-01-13  Jakub Jelinek  <jakub@redhat.com>
3609
3610         PR libgomp/59194
3611         * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
3612         load as __atomic_load_N if possible.
3613
3614 2014-01-11  David Edelsohn  <dje.gcc@gmail.com>
3615
3616         * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3617         target parameter.
3618         (rs6000_expand_builtin): Adjust call.
3619
3620 2014-01-11  David Edelsohn  <dje.gcc@gmail.com>
3621
3622         PR target/58115
3623         * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3624         * config/rs6000/rs6000.c: Include target-globals.h.
3625         (rs6000_set_current_function): Instead of doing target_reinit
3626         unconditionally, use save_target_globals_default_opts and
3627         restore_target_globals.
3628
3629         * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3630         FPSCR.
3631         * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3632         (rs6000_expand_builtin): Handle mffs and mtfsf.
3633         (rs6000_init_builtins): Define mffs and mtfsf.
3634         * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3635         (rs6000_mffs): New pattern.
3636         (rs6000_mtfsf): New pattern.
3637
3638 2014-01-11  Bin Cheng  <bin.cheng@arm.com>
3639
3640         * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3641         Start narrowing with START.  Apply candidate-use pair
3642         and check overall cost in narrowing.
3643         (iv_ca_prune): Pass new argument.
3644
3645 2014-01-10  Jeff Law  <law@redhat.com>
3646
3647         PR middle-end/59743
3648         * ree.c (combine_reaching_defs): Ensure the defining statement
3649         occurs before the extension when optimizing extensions with
3650         different source and destination hard registers.
3651
3652 2014-01-10  Jan Hubicka  <hubicka@ucw.cz>
3653
3654         PR ipa/58585
3655         * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3656         vtables into the type inheritance graph.
3657
3658 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
3659
3660         PR rtl-optimization/59754
3661         * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3662         modes in the REGNO != REGNO case.
3663
3664 2014-01-10  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3665
3666         * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3667
3668 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
3669
3670         PR tree-optimization/59745
3671         * tree-predcom.c (tree_predictive_commoning_loop): Call
3672         free_affine_expand_cache if giving up because components is NULL.
3673
3674         * target-globals.c (save_target_globals): Allocate < 4KB structs using
3675         GC in payload of target_globals struct instead of allocating them on
3676         the heap and the larger structs separately using GC.
3677         * target-globals.h (struct target_globals): Make regs, hard_regs,
3678         reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3679         of GTY((skip)) and change type to void *.
3680         (reset_target_globals): Cast loads from those fields to corresponding
3681         types.
3682
3683 2014-01-10  Steve Ellcey  <sellcey@mips.com>
3684
3685         PR plugins/59335
3686         * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3687         gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3688         tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3689
3690 2014-01-10  Richard Earnshaw  <rearnsha@arm.com>
3691
3692         PR target/59744
3693         * aarch64-modes.def (CC_Zmode): New flags mode.
3694         * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3695         represents an equality.
3696         (aarch64_get_condition_code): Handle CC_Zmode.
3697         * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3698
3699 2014-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3700
3701         * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3702         extraction in good case.
3703
3704 2014-01-10  Richard Biener  <rguenther@suse.de>
3705
3706         PR tree-optimization/59374
3707         * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3708         checking after SLP discovery.  Mark stmts not participating
3709         in any SLP instance properly.
3710
3711 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3712
3713         * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3714         when handling a SET rtx.
3715
3716 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3717
3718         * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3719         (cortex-a57): Likewise.
3720         (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3721
3722 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3723
3724         * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3725         non-iwmmxt builtins.
3726
3727 2014-01-10  Jan Hubicka  <hubicka@ucw.cz>
3728
3729         PR ipa/58252
3730         PR ipa/59226
3731         * ipa-devirt.c record_target_from_binfo): Take as argument
3732         stack of binfos and lookup matching one for virtual inheritance.
3733         (possible_polymorphic_call_targets_1): Update.
3734
3735 2014-01-10  Huacai Chen  <chenhc@lemote.com>
3736
3737         * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3738         kernel strings for Loongson-2E/2F/3A.
3739
3740 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
3741
3742         PR middle-end/59670
3743         * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3744         is_gimple_call before calling gimple_call_internal_p.
3745
3746 2014-01-09  Steve Ellcey  <sellcey@mips.com>
3747
3748         * Makefile.in (TREE_FLOW_H): Remove.
3749         (TREE_SSA_H): Add file names from tree-flow.h.
3750         * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3751         * tree.h: Remove tree-flow.h reference.
3752         * hash-table.h: Remove tree-flow.h reference.
3753         * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3754         reference with tree-ssa-loop.h.
3755
3756 2014-01-09  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3757
3758         * doc/invoke.texi: Add -maltivec={be,le} options, and document
3759         default element-order behavior for -maltivec.
3760         * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3761         * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3762         that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3763         when targeting big endian, at least for now.
3764         * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3765
3766 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
3767
3768         PR middle-end/47735
3769         * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3770         var satisfies use_register_for_decl, just take into account type
3771         alignment, rather than decl alignment.
3772
3773         PR tree-optimization/59622
3774         * gimple-fold.c (gimple_fold_call): Fix a typo in message.  For
3775         __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3776         __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3777         Don't devirtualize for inplace at all.  For targets.length () == 1,
3778         if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3779
3780 2014-01-09  H.J. Lu  <hongjiu.lu@intel.com>
3781
3782         * config/i386/i386.md (cpu): Remove the unused btver1.
3783
3784 2014-01-09  H.J. Lu  <hongjiu.lu@intel.com>
3785
3786         * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3787
3788 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
3789
3790         PR target/58115
3791         * tree-core.h (struct target_globals): New forward declaration.
3792         (struct tree_target_option): Add globals field.
3793         * tree.h (TREE_TARGET_GLOBALS): Define.
3794         (prepare_target_option_nodes_for_pch): New prototype.
3795         * target-globals.h (struct target_globals): Define even if
3796         !SWITCHABLE_TARGET.
3797         * tree.c (prepare_target_option_node_for_pch,
3798         prepare_target_option_nodes_for_pch): New functions.
3799         * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3800         * config/i386/i386.c: Include target-globals.h.
3801         (ix86_set_current_function): Instead of doing target_reinit
3802         unconditionally, use save_target_globals_default_opts and
3803         restore_target_globals.
3804
3805 2014-01-09  Richard Biener  <rguenther@suse.de>
3806
3807         PR tree-optimization/59715
3808         * tree-cfg.h (split_critical_edges): Declare.
3809         * tree-cfg.c (split_critical_edges): Export.
3810         * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3811
3812 2014-01-09  Max Ostapenko  <m.ostapenko@partner.samsung.com>
3813
3814         * cfgexpand.c (expand_stack_vars): Optionally disable
3815         asan stack protection.
3816         (expand_used_vars): Likewise.
3817         (partition_stack_vars): Likewise.
3818         * asan.c (asan_emit_stack_protection): Optionally disable
3819         after return stack usage.
3820         (instrument_derefs): Optionally disable memory access instrumentation.
3821         (instrument_builtin_call): Likewise.
3822         (instrument_strlen_call): Likewise.
3823         (asan_protect_global): Optionally disable global variables protection.
3824         * doc/invoke.texi: Added doc for new options.
3825         * params.def: Added new options.
3826         * params.h: Likewise.
3827
3828 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
3829
3830         PR rtl-optimization/59724
3831         * ifcvt.c (cond_exec_process_if_block): Don't call
3832         flow_find_head_matching_sequence with 0 longest_match.
3833         * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3834         non-active insns if !stop_after.
3835         (try_head_merge_bb): Revert 2014-01-07 changes.
3836
3837 2014-01-08  Jeff Law  <law@redhat.com>
3838
3839         * ree.c (get_sub_rtx): New function, extracted from...
3840         (merge_def_and_ext): Here.
3841         (combine_reaching_defs): Use get_sub_rtx.
3842
3843 2014-01-08  Eric Botcazou  <ebotcazou@adacore.com>
3844
3845         * cgraph.h (varpool_variable_node): Do not choke on null node.
3846
3847 2014-01-08  Catherine Moore  <clm@codesourcery.com>
3848
3849         * config/mips/mips.md (simple_return): Attempt to use JRC
3850         for microMIPS.
3851         * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3852
3853 2014-01-08  Richard Sandiford  <rdsandiford@googlemail.com>
3854
3855         PR rtl-optimization/59137
3856         * reorg.c (steal_delay_list_from_target): Call update_block for
3857         elided insns.
3858         (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3859
3860 2014-01-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3861
3862         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3863         two duplicate entries.
3864
3865 2014-01-08  Richard Sandiford  <rdsandiford@googlemail.com>
3866
3867         Revert:
3868         2012-10-07  Richard Sandiford  <rdsandiford@googlemail.com>
3869
3870         * config/mips/mips.c (mips_truncated_op_cost): New function.
3871         (mips_rtx_costs): Adjust test for BADDU.
3872         * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3873
3874         2012-10-02  Richard Sandiford  <rdsandiford@googlemail.com>
3875
3876         * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3877         (*baddu_si): ...this new pattern.
3878
3879 2014-01-08  Jakub Jelinek  <jakub@redhat.com>
3880
3881         PR ipa/59722
3882         * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3883
3884 2014-01-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
3885
3886         PR middle-end/57748
3887         * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3888         inner_reference_p.
3889         (expand_expr, expand_normal): Adjust.
3890         * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3891         inner_reference_p. Use inner_reference_p to expand inner references.
3892         (store_expr): Adjust.
3893         * cfgexpand.c (expand_call_stmt): Adjust.
3894
3895 2014-01-08  Rong Xu  <xur@google.com>
3896
3897         * gcov-io.c (gcov_var): Move from gcov-io.h.
3898         (gcov_position): Ditto.
3899         (gcov_is_error): Ditto.
3900         (gcov_rewrite): Ditto.
3901         * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3902         only part to libgcc/libgcov.h.
3903
3904 2014-01-08  Marek Polacek  <polacek@redhat.com>
3905
3906         PR middle-end/59669
3907         * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3908
3909 2014-01-08  Marek Polacek  <polacek@redhat.com>
3910
3911         PR sanitizer/59667
3912         * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3913
3914 2014-01-08  Jakub Jelinek  <jakub@redhat.com>
3915
3916         PR rtl-optimization/59649
3917         * stor-layout.c (get_mode_bounds): For BImode return
3918         0 and STORE_FLAG_VALUE.
3919
3920 2014-01-08  Richard Biener  <rguenther@suse.de>
3921
3922         PR middle-end/59630
3923         * gimple.h (is_gimple_builtin_call): Remove.
3924         (gimple_builtin_call_types_compatible_p): New.
3925         (gimple_call_builtin_p): New overload.
3926         * gimple.c (is_gimple_builtin_call): Remove.
3927         (validate_call): Rename to ...
3928         (gimple_builtin_call_types_compatible_p): ... this and export.  Also
3929         check return types.
3930         (validate_type): New static function.
3931         (gimple_call_builtin_p): New overload and adjust.
3932         * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3933         (gimple_fold_call): Likewise.  Use gimple_call_builtin_p.
3934         (gimple_fold_stmt_to_constant_1): Likewise.
3935         * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3936
3937 2014-01-08  Richard Biener  <rguenther@suse.de>
3938
3939         PR middle-end/59471
3940         * gimplify.c (gimplify_expr): Gimplify register-register type
3941         VIEW_CONVERT_EXPRs to separate stmts.
3942
3943 2014-01-07  Jeff Law  <law@redhat.com>
3944
3945         PR middle-end/53623
3946         * ree.c (combine_set_extension): Handle case where source
3947         and destination registers in an extension insn are different.
3948         (combine_reaching_defs): Allow source and destination registers
3949         in extension to be different under limited circumstances.
3950         (add_removable_extension): Remove restriction that the
3951         source and destination registers in the extension are the same.
3952         (find_and_remove_re): Emit a copy from the extension's
3953         destination to its source after the defining insn if
3954         the source and destination registers are different.
3955
3956         PR middle-end/59285
3957         * ifcvt.c (merge_if_block): If we are merging a block with more than
3958         one successor with a block with no successors, remove any BARRIER
3959         after the second block.
3960
3961 2014-01-07  Dan Xio Qiang <ziyan01@163.com>
3962
3963         * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3964
3965 2014-01-07  John David Anglin  <danglin@gcc.gnu.org>
3966
3967         PR target/59652
3968         * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3969         for 14-bit register offsets when INT14_OK_STRICT is false.
3970
3971 2014-01-07  Roland Stigge  <stigge@antcom.de>
3972             Michael Meissner  <meissner@linux.vnet.ibm.com>
3973
3974         PR 57386/target
3975         * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3976         Only check TFmode for SPE constants.  Don't check TImode or TDmode.
3977
3978 2014-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
3979
3980         * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3981         -mcpu.
3982
3983 2014-01-07  Yufeng Zhang  <yufeng.zhang@arm.com>
3984
3985         * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3986         with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3987         rtx is const0_rtx or not.
3988
3989 2014-01-07  Richard Sandiford  <rdsandiford@googlemail.com>
3990
3991         PR target/58115
3992         * target-globals.c (save_target_globals): Remove this_fn_optab
3993         handling.
3994         * toplev.c: Include optabs.h.
3995         (target_reinit): Temporarily restore the global options if another
3996         set of options are in force.
3997
3998 2014-01-07  Jakub Jelinek  <jakub@redhat.com>
3999
4000         PR rtl-optimization/58668
4001         * cfgcleanup.c (flow_find_cross_jump): Don't count
4002         any jumps if dir_p is NULL.  Remove p1 variable, use active_insn_p
4003         to determine what is counted.
4004         (flow_find_head_matching_sequence): Use active_insn_p to determine
4005         what is counted.
4006         (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
4007         counting change.
4008         * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
4009         determine what is counted.
4010
4011         PR tree-optimization/59643
4012         * tree-predcom.c (split_data_refs_to_components): If one dr is
4013         read and one write, determine_offset fails and the write isn't
4014         in the bad component, just put the read into the bad component.
4015
4016 2014-01-07  Mike Stump  <mikestump@comcast.net>
4017             Jakub Jelinek  <jakub@redhat.com>
4018
4019         PR pch/59436
4020         * tree-core.h (struct tree_optimization_option): Change optabs
4021         type from unsigned char * to void *.
4022         * optabs.c (init_tree_optimization_optabs): Adjust
4023         TREE_OPTIMIZATION_OPTABS initialization.
4024
4025 2014-01-06  Jakub Jelinek  <jakub@redhat.com>
4026
4027         PR target/59644
4028         * config/i386/i386.h (struct machine_function): Add
4029         no_drap_save_restore field.
4030         * config/i386/i386.c (ix86_save_reg): Use
4031         !cfun->machine->no_drap_save_restore instead of
4032         crtl->stack_realign_needed.
4033         (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
4034         this function clears frame_pointer_needed.  Set
4035         cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
4036         and DRAP reg is needed.
4037
4038 2014-01-06  Marek Polacek  <polacek@redhat.com>
4039
4040         PR c/57773
4041         * doc/implement-c.texi: Mention that other integer types are
4042         permitted as bit-field types in strictly conforming mode.
4043
4044 2014-01-06  Felix Yang  <fei.yang0953@gmail.com>
4045
4046         * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
4047         is newly allocated.
4048
4049 2014-01-06  Richard Earnshaw  <rearnsha@arm.com>
4050
4051         * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
4052
4053 2014-01-06  Martin Jambor  <mjambor@suse.cz>
4054
4055         PR ipa/59008
4056         * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
4057         to int.
4058         * ipa-prop.c (ipa_print_node_params): Fix indentation.
4059
4060 2014-01-06  Eric Botcazou  <ebotcazou@adacore.com>
4061
4062         PR debug/59350
4063         PR debug/59510
4064         * var-tracking.c (add_stores): Preserve the value of the source even if
4065         we don't record the store.
4066
4067 2014-01-06  Terry Guo  <terry.guo@arm.com>
4068
4069         * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
4070
4071 2014-01-05  Iain Sandoe  <iain@codesourcery.com>
4072
4073         PR bootstrap/59541
4074         * config/darwin.c (darwin_function_section): Adjust return values to
4075         correspond to optimisation changes made in r206070.
4076
4077 2014-01-05  Uros Bizjak  <ubizjak@gmail.com>
4078
4079         * config/i386/i386.c (ix86_data_alignment): Calculate max_align
4080         from prefetch_block tune setting.
4081         (nocona_cost): Correct size of prefetch block to 64.
4082
4083 2014-01-04  Eric Botcazou  <ebotcazou@adacore.com>
4084
4085         * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
4086         (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
4087         used to save the static chain register in the computation of the offset
4088         from which the FP registers need to be restored.
4089
4090 2014-01-04  Jakub Jelinek  <jakub@redhat.com>
4091
4092         PR tree-optimization/59519
4093         * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
4094         ICE if get_current_def (current_new_name) is already non-NULL, as long
4095         as it is a phi result of some other phi in *new_exit_bb that has
4096         the same argument.
4097
4098         * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
4099         or vmovdqu* for misaligned_operand.
4100         (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
4101         <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
4102         * config/i386/i386.c (ix86_expand_special_args_builtin): Set
4103         aligned_mem for AVX512F masked aligned load and store builtins and for
4104         non-temporal moves.
4105
4106 2014-01-03  Bingfeng Mei  <bmei@broadcom.com>
4107
4108         PR tree-optimization/59651
4109         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
4110         Address range for negative step should be added by TYPE_SIZE_UNIT.
4111
4112 2014-01-03  Andreas Schwab  <schwab@linux-m68k.org>
4113
4114         * config/m68k/m68k.c (handle_move_double): Handle pushes with
4115         overlapping registers also for registers other than the stack pointer.
4116
4117 2014-01-03  Marek Polacek  <polacek@redhat.com>
4118
4119         PR other/59661
4120         * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
4121         __builtin_FILE.
4122
4123 2014-01-03  Jakub Jelinek  <jakub@redhat.com>
4124
4125         PR target/59625
4126         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
4127         asm goto as jump.
4128
4129         * config/i386/i386.md (MODE_SIZE): New mode attribute.
4130         (push splitter): Use <P:MODE_SIZE> instead of
4131         GET_MODE_SIZE (<P:MODE>mode).
4132         (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
4133         (mov -1, reg peephole2): Likewise.
4134         * config/i386/sse.md (*mov<mode>_internal,
4135         <sse>_storeu<ssemodesuffix><avxsizesuffix>,
4136         <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
4137         *<code><mode>3, *andnot<mode>3<mask_name>,
4138         <mask_codefor><code><mode>3<mask_name>): Likewise.
4139         * config/i386/subst.md (mask_mode512bit_condition,
4140         sd_mask_mode512bit_condition): Likewise.
4141
4142 2014-01-02  Xinliang David Li  <davidxl@google.com>
4143
4144         PR tree-optimization/59303
4145         * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
4146         (dump_predicates): Better output format.
4147         (pred_equal_p): New function.
4148         (is_neq_relop_p): Ditto.
4149         (is_neq_zero_form_p): Ditto.
4150         (pred_expr_equal_p): Ditto.
4151         (pred_neg_p): Ditto.
4152         (simplify_pred): Ditto.
4153         (simplify_preds_2): Ditto.
4154         (simplify_preds_3): Ditto.
4155         (simplify_preds_4): Ditto.
4156         (simplify_preds): Ditto.
4157         (push_pred): Ditto.
4158         (push_to_worklist): Ditto.
4159         (get_pred_info_from_cmp): Ditto.
4160         (is_degenerated_phi): Ditto.
4161         (normalize_one_pred_1): Ditto.
4162         (normalize_one_pred): Ditto.
4163         (normalize_one_pred_chain): Ditto.
4164         (normalize_preds): Ditto.
4165         (normalize_cond_1): Remove function.
4166         (normalize_cond): Ditto.
4167         (is_gcond_subset_of): Ditto.
4168         (is_subset_of_any): Ditto.
4169         (is_or_set_subset_of): Ditto.
4170         (is_and_set_subset_of): Ditto.
4171         (is_norm_cond_subset_of): Ditto.
4172         (pred_chain_length_cmp): Ditto.
4173         (convert_control_dep_chain_into_preds): Type change.
4174         (find_predicates): Ditto.
4175         (find_def_preds): Ditto.
4176         (destroy_predicates_vecs): Ditto.
4177         (find_matching_predicates_in_rest_chains): Ditto.
4178         (use_pred_not_overlap_with_undef_path_pred): Ditto.
4179         (is_pred_expr_subset): Ditto.
4180         (is_pred_chain_subset_of): Ditto.
4181         (is_included_in): Ditto.
4182         (is_superset_of): Ditto.
4183
4184 2014-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
4185
4186         Update copyright years.
4187
4188 2014-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
4189
4190         * common/config/arc/arc-common.c, config/arc/arc-modes.def,
4191         config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
4192         config/arc/arc.md, config/arc/arc.opt,
4193         config/arm/arm_neon_builtins.def, config/arm/crypto.def,
4194         config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
4195         config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
4196         config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
4197         config/linux-protos.h, config/linux.c, config/winnt-c.c,
4198         diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
4199         vtable-verify.c, vtable-verify.h: Use the standard form for the
4200         copyright notice.
4201
4202 2014-01-02  Tobias Burnus  <burnus@net-b.de>
4203
4204         * gcc.c (process_command): Update copyright notice dates.
4205         * gcov-dump.c: Ditto.
4206         * gcov.c: Ditto.
4207         * doc/cpp.texi: Bump @copying's copyright year.
4208         * doc/cppinternals.texi: Ditto.
4209         * doc/gcc.texi: Ditto.
4210         * doc/gccint.texi: Ditto.
4211         * doc/gcov.texi: Ditto.
4212         * doc/install.texi: Ditto.
4213         * doc/invoke.texi: Ditto.
4214
4215 2014-01-01  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
4216
4217         * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
4218
4219 2014-01-01  Jakub Jelinek  <jakub@redhat.com>
4220
4221         * config/i386/sse.md (*mov<mode>_internal): Guard
4222         EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
4223
4224         PR rtl-optimization/59647
4225         * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
4226         new_rtx into UNSIGNED_FLOAT rtxes.
4227 \f
4228 Copyright (C) 2014 Free Software Foundation, Inc.
4229
4230 Copying and distribution of this file, with or without modification,
4231 are permitted in any medium without royalty provided the copyright
4232 notice and this notice are preserved.