gcc/ChangeLog

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