gcc/ChangeLog

   1 2019-01-11  Tom de Vries  <tdevries@suse.de>
   2
   3         * config/nvptx/nvptx.c (PTX_CTA_SIZE): Move up.
   4
   5 2019-01-11  Jan Beulich  <jbeulich@suse.com>
   6
   7         * config/i386/i386.md (rex64suffix): Add L suffix for SI.
   8         * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>32<round_name>,
   9         sse2_cvtsi2sd): Add {l}.
  10         (sse2_cvtsi2sdq<round_name>): Make q conditional upon AT&T
  11         syntax.
  12
  13 2019-01-10  Jakub Jelinek  <jakub@redhat.com>
  14
  15         PR target/88785
  16         * config/i386/sse.md (float<floatunssuffix>v2div2sf2): Turn into
  17         define_expand.
  18         (*float<floatunssuffix>v2div2sf2): New define_insn.
  19         (float<floatunssuffix>v2div2sf2_mask): Turn into define_expand.
  20         (*float<floatunssuffix>v2div2sf2_mask): New define_insn.
  21         (*float<floatunssuffix>v2div2sf2_mask_1): Replace
  22         subrtxes (const_vector:V2SF [(const_int 0) (const_int 0)]) with
  23         match_operands with "const0_operand" "C".
  24
  25 2019-01-10  Tamar Christina  <tamar.christina@arm.com>
  26
  27         * config/aarch64/aarch64-builtins.c
  28         (aarch64_init_builtins): Move aarch64_init_fcmla_laneq_builtins...
  29         (aarch64_init_simd_builtins): ...Here
  30
  31 2019-01-10  Vladimir Makarov  <vmakarov@redhat.com>
  32
  33         PR rtl-optimization/87305
  34         * lra-assigns.c
  35         (setup_live_pseudos_and_spill_after_risky_transforms): Check
  36         allocation for big endian pseudos used as paradoxical subregs and
  37         spill them if it is wrong.
  38         * lra-constraints.c (lra_constraints): Add a comment.
  39
  40 2019-01-10  Richard Biener  <rguenther@suse.de>
  41
  42         PR tree-optimization/88792
  43         * tree-ssa-pre.c (get_representative_for): Do not return a
  44         value-number here.
  45
  46 2019-01-10  Jakub Jelinek  <jakub@redhat.com>
  47
  48         PR middle-end/84877
  49         PR bootstrap/88450
  50         * function.c (assign_stack_local_1): Revert the 2018-11-21 changes.
  51         (assign_parm_setup_block): Do the argument slot realignment here
  52         instead.
  53
  54 2019-01-10  Stefan Agner  <stefan@agner.ch>
  55
  56         PR target/88648
  57         * config/arm/arm.c (arm_option_override_internal): Force
  58         opts->x_inline_asm_unified to true only if TARGET_THUMB2_P.
  59
  60 2019-01-10  Jakub Jelinek  <jakub@redhat.com>
  61
  62         PR c/88568
  63         * attribs.c (handle_dll_attribute): Clear TREE_STATIC after setting
  64         DECL_EXTERNAL.
  65
  66 2019-01-10  Tamar Christina  <tamar.christina@arm.com>
  67
  68         * config/arm/arm-builtins.c
  69         (enum arm_type_qualifiers): Add qualifier_lane_pair_index.
  70         (MAC_LANE_PAIR_QUALIFIERS): New.
  71         (arm_expand_builtin_args): Use it.
  72         (arm_expand_builtin_1): Likewise.
  73         * config/arm/arm-protos.h (neon_vcmla_lane_prepare_operands): New.
  74         * config/arm/arm.c (neon_vcmla_lane_prepare_operands): New.
  75         * config/arm/arm-c.c (arm_cpu_builtins): Add __ARM_FEATURE_COMPLEX.
  76         * config/arm/arm_neon.h:
  77         (vcadd_rot90_f16): New.
  78         (vcaddq_rot90_f16): New.
  79         (vcadd_rot270_f16): New.
  80         (vcaddq_rot270_f16): New.
  81         (vcmla_f16): New.
  82         (vcmlaq_f16): New.
  83         (vcmla_lane_f16): New.
  84         (vcmla_laneq_f16): New.
  85         (vcmlaq_lane_f16): New.
  86         (vcmlaq_laneq_f16): New.
  87         (vcmla_rot90_f16): New.
  88         (vcmlaq_rot90_f16): New.
  89         (vcmla_rot90_lane_f16): New.
  90         (vcmla_rot90_laneq_f16): New.
  91         (vcmlaq_rot90_lane_f16): New.
  92         (vcmlaq_rot90_laneq_f16): New.
  93         (vcmla_rot180_f16): New.
  94         (vcmlaq_rot180_f16): New.
  95         (vcmla_rot180_lane_f16): New.
  96         (vcmla_rot180_laneq_f16): New.
  97         (vcmlaq_rot180_lane_f16): New.
  98         (vcmlaq_rot180_laneq_f16): New.
  99         (vcmla_rot270_f16): New.
 100         (vcmlaq_rot270_f16): New.
 101         (vcmla_rot270_lane_f16): New.
 102         (vcmla_rot270_laneq_f16): New.
 103         (vcmlaq_rot270_lane_f16): New.
 104         (vcmlaq_rot270_laneq_f16): New.
 105         (vcadd_rot90_f32): New.
 106         (vcaddq_rot90_f32): New.
 107         (vcadd_rot270_f32): New.
 108         (vcaddq_rot270_f32): New.
 109         (vcmla_f32): New.
 110         (vcmlaq_f32): New.
 111         (vcmla_lane_f32): New.
 112         (vcmla_laneq_f32): New.
 113         (vcmlaq_lane_f32): New.
 114         (vcmlaq_laneq_f32): New.
 115         (vcmla_rot90_f32): New.
 116         (vcmlaq_rot90_f32): New.
 117         (vcmla_rot90_lane_f32): New.
 118         (vcmla_rot90_laneq_f32): New.
 119         (vcmlaq_rot90_lane_f32): New.
 120         (vcmlaq_rot90_laneq_f32): New.
 121         (vcmla_rot180_f32): New.
 122         (vcmlaq_rot180_f32): New.
 123         (vcmla_rot180_lane_f32): New.
 124         (vcmla_rot180_laneq_f32): New.
 125         (vcmlaq_rot180_lane_f32): New.
 126         (vcmlaq_rot180_laneq_f32): New.
 127         (vcmla_rot270_f32): New.
 128         (vcmlaq_rot270_f32): New.
 129         (vcmla_rot270_lane_f32): New.
 130         (vcmla_rot270_laneq_f32): New.
 131         (vcmlaq_rot270_lane_f32): New.
 132         (vcmlaq_rot270_laneq_f32): New.
 133         * config/arm/arm_neon_builtins.def (vcadd90, vcadd270, vcmla0, vcmla90,
 134         vcmla180, vcmla270, vcmla_lane0, vcmla_lane90, vcmla_lane180, vcmla_lane270,
 135         vcmla_laneq0, vcmla_laneq90, vcmla_laneq180, vcmla_laneq270,
 136         vcmlaq_lane0, vcmlaq_lane90, vcmlaq_lane180, vcmlaq_lane270): New.
 137         * config/arm/neon.md (neon_vcmla_lane<rot><mode>,
 138         neon_vcmla_laneq<rot><mode>, neon_vcmlaq_lane<rot><mode>): New.
 139         * config/arm/arm.c (arm_arch8_3, arm_arch8_4): New.
 140         * config/arm/arm.h (TARGET_COMPLEX, arm_arch8_3, arm_arch8_4): New.
 141         (arm_option_reconfigure_globals): Use them.
 142         * config/arm/iterators.md (VDF, VQ_HSF): New.
 143         (VCADD, VCMLA): New.
 144         (VF_constraint, rot, rotsplit1, rotsplit2): Add V4HF and V8HF.
 145         * config/arm/neon.md (neon_vcadd<rot><mode>, neon_vcmla<rot><mode>): New.
 146         * config/arm/unspecs.md (UNSPEC_VCADD90, UNSPEC_VCADD270,
 147         UNSPEC_VCMLA, UNSPEC_VCMLA90, UNSPEC_VCMLA180, UNSPEC_VCMLA270): New.
 148
 149 2019-01-10  Tamar Christina  <tamar.christina@arm.com>
 150
 151         * config/aarch64/aarch64-builtins.c (enum aarch64_type_qualifiers): Add qualifier_lane_pair_index.
 152         (emit-rtl.h): Include.
 153         (TYPES_QUADOP_LANE_PAIR): New.
 154         (aarch64_simd_expand_args): Use it.
 155         (aarch64_simd_expand_builtin): Likewise.
 156         (AARCH64_SIMD_FCMLA_LANEQ_BUILTINS, aarch64_fcmla_laneq_builtin_datum): New.
 157         (FCMLA_LANEQ_BUILTIN, AARCH64_SIMD_FCMLA_LANEQ_BUILTIN_BASE,
 158         AARCH64_SIMD_FCMLA_LANEQ_BUILTINS, aarch64_fcmla_lane_builtin_data,
 159         aarch64_init_fcmla_laneq_builtins, aarch64_expand_fcmla_builtin): New.
 160         (aarch64_init_builtins): Add aarch64_init_fcmla_laneq_builtins.
 161         (aarch64_expand_buildin): Add AARCH64_SIMD_BUILTIN_FCMLA_LANEQ0_V2SF,
 162         AARCH64_SIMD_BUILTIN_FCMLA_LANEQ90_V2SF, AARCH64_SIMD_BUILTIN_FCMLA_LANEQ180_V2SF,
 163         AARCH64_SIMD_BUILTIN_FCMLA_LANEQ2700_V2SF, AARCH64_SIMD_BUILTIN_FCMLA_LANEQ0_V4HF,
 164         AARCH64_SIMD_BUILTIN_FCMLA_LANEQ90_V4HF, AARCH64_SIMD_BUILTIN_FCMLA_LANEQ180_V4HF,
 165         AARCH64_SIMD_BUILTIN_FCMLA_LANEQ270_V4HF.
 166         * config/aarch64/iterators.md (FCMLA_maybe_lane): New.
 167         * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Add __ARM_FEATURE_COMPLEX.
 168         * config/aarch64/aarch64-simd-builtins.def (fcadd90, fcadd270, fcmla0, fcmla90,
 169         fcmla180, fcmla270, fcmla_lane0, fcmla_lane90, fcmla_lane180, fcmla_lane270,
 170         fcmla_laneq0, fcmla_laneq90, fcmla_laneq180, fcmla_laneq270,
 171         fcmlaq_lane0, fcmlaq_lane90, fcmlaq_lane180, fcmlaq_lane270): New.
 172         * config/aarch64/aarch64-simd.md (aarch64_fcmla_lane<rot><mode>,
 173         aarch64_fcmla_laneq<rot>v4hf, aarch64_fcmlaq_lane<rot><mode>,aarch64_fcadd<rot><mode>,
 174         aarch64_fcmla<rot><mode>): New.
 175         * config/aarch64/arm_neon.h:
 176         (vcadd_rot90_f16): New.
 177         (vcaddq_rot90_f16): New.
 178         (vcadd_rot270_f16): New.
 179         (vcaddq_rot270_f16): New.
 180         (vcmla_f16): New.
 181         (vcmlaq_f16): New.
 182         (vcmla_lane_f16): New.
 183         (vcmla_laneq_f16): New.
 184         (vcmlaq_lane_f16): New.
 185         (vcmlaq_rot90_lane_f16): New.
 186         (vcmla_rot90_laneq_f16): New.
 187         (vcmla_rot90_lane_f16): New.
 188         (vcmlaq_rot90_f16): New.
 189         (vcmla_rot90_f16): New.
 190         (vcmlaq_laneq_f16): New.
 191         (vcmla_rot180_laneq_f16): New.
 192         (vcmla_rot180_lane_f16): New.
 193         (vcmlaq_rot180_f16): New.
 194         (vcmla_rot180_f16): New.
 195         (vcmlaq_rot90_laneq_f16): New.
 196         (vcmlaq_rot270_laneq_f16): New.
 197         (vcmlaq_rot270_lane_f16): New.
 198         (vcmla_rot270_laneq_f16): New.
 199         (vcmlaq_rot270_f16): New.
 200         (vcmla_rot270_f16): New.
 201         (vcmlaq_rot180_laneq_f16): New.
 202         (vcmlaq_rot180_lane_f16): New.
 203         (vcmla_rot270_lane_f16): New.
 204         (vcadd_rot90_f32): New.
 205         (vcaddq_rot90_f32): New.
 206         (vcaddq_rot90_f64): New.
 207         (vcadd_rot270_f32): New.
 208         (vcaddq_rot270_f32): New.
 209         (vcaddq_rot270_f64): New.
 210         (vcmla_f32): New.
 211         (vcmlaq_f32): New.
 212         (vcmlaq_f64): New.
 213         (vcmla_lane_f32): New.
 214         (vcmla_laneq_f32): New.
 215         (vcmlaq_lane_f32): New.
 216         (vcmlaq_laneq_f32): New.
 217         (vcmla_rot90_f32): New.
 218         (vcmlaq_rot90_f32): New.
 219         (vcmlaq_rot90_f64): New.
 220         (vcmla_rot90_lane_f32): New.
 221         (vcmla_rot90_laneq_f32): New.
 222         (vcmlaq_rot90_lane_f32): New.
 223         (vcmlaq_rot90_laneq_f32): New.
 224         (vcmla_rot180_f32): New.
 225         (vcmlaq_rot180_f32): New.
 226         (vcmlaq_rot180_f64): New.
 227         (vcmla_rot180_lane_f32): New.
 228         (vcmla_rot180_laneq_f32): New.
 229         (vcmlaq_rot180_lane_f32): New.
 230         (vcmlaq_rot180_laneq_f32): New.
 231         (vcmla_rot270_f32): New.
 232         (vcmlaq_rot270_f32): New.
 233         (vcmlaq_rot270_f64): New.
 234         (vcmla_rot270_lane_f32): New.
 235         (vcmla_rot270_laneq_f32): New.
 236         (vcmlaq_rot270_lane_f32): New.
 237         (vcmlaq_rot270_laneq_f32): New.
 238         * config/aarch64/aarch64.h (TARGET_COMPLEX): New.
 239         * config/aarch64/iterators.md (UNSPEC_FCADD90, UNSPEC_FCADD270,
 240         UNSPEC_FCMLA, UNSPEC_FCMLA90, UNSPEC_FCMLA180, UNSPEC_FCMLA270): New.
 241         (FCADD, FCMLA): New.
 242         (rot): New.
 243         * config/arm/types.md (neon_fcadd, neon_fcmla): New.
 244
 245 2019-01-09  Sandra Loosemore  <sandra@codesourcery.com>
 246
 247         PR other/16615
 248
 249         * config/pa/pa.c: Change "can not" to "cannot".
 250         * gimple-ssa-evrp-analyze.c: Likewise.
 251         * ipa-icf.c: Likewise.
 252         * ipa-polymorphic-call.c: Likewise.
 253         * ipa-pure-const.c: Likewise.
 254         * lra-constraints.c: Likewise.
 255         * lra-remat.c: Likewise.
 256         * reload1.c: Likewise.
 257         * reorg.c: Likewise.
 258         * tree-ssa-uninit.c: Likewise.
 259
 260 2019-01-09  Sandra Loosemore  <sandra@codesourcery.com>
 261
 262         PR other/16615
 263
 264         * Makefile.in: Mechanically replace "can not" with "cannot".
 265         * alias.c: Likewise.
 266         * builtins.c: Likewise.
 267         * calls.c: Likewise.
 268         * cgraph.c: Likewise.
 269         * cgraph.h: Likewise.
 270         * cgraphclones.c: Likewise.
 271         * cgraphunit.c: Likewise.
 272         * combine-stack-adj.c: Likewise.
 273         * combine.c: Likewise.
 274         * common/config/i386/i386-common.c: Likewise.
 275         * config/aarch64/aarch64.c: Likewise.
 276         * config/alpha/sync.md: Likewise.
 277         * config/arc/arc.c: Likewise.
 278         * config/arc/predicates.md: Likewise.
 279         * config/arm/arm-c.c: Likewise.
 280         * config/arm/arm.c: Likewise.
 281         * config/arm/arm.h: Likewise.
 282         * config/arm/arm.md: Likewise.
 283         * config/arm/cortex-r4f.md: Likewise.
 284         * config/csky/csky.c: Likewise.
 285         * config/csky/csky.h: Likewise.
 286         * config/darwin-f.c: Likewise.
 287         * config/epiphany/epiphany.md: Likewise.
 288         * config/i386/i386.c: Likewise.
 289         * config/i386/sol2.h: Likewise.
 290         * config/m68k/m68k.c: Likewise.
 291         * config/mcore/mcore.h: Likewise.
 292         * config/microblaze/microblaze.md: Likewise.
 293         * config/mips/20kc.md: Likewise.
 294         * config/mips/sb1.md: Likewise.
 295         * config/nds32/nds32.c: Likewise.
 296         * config/nds32/predicates.md: Likewise.
 297         * config/pa/pa.c: Likewise.
 298         * config/rs6000/e300c2c3.md: Likewise.
 299         * config/rs6000/rs6000.c: Likewise.
 300         * config/s390/s390.h: Likewise.
 301         * config/sh/sh.c: Likewise.
 302         * config/sh/sh.md: Likewise.
 303         * config/spu/vmx2spu.h: Likewise.
 304         * cprop.c: Likewise.
 305         * dbxout.c: Likewise.
 306         * df-scan.c: Likewise.
 307         * doc/cfg.texi: Likewise.
 308         * doc/extend.texi: Likewise.
 309         * doc/fragments.texi: Likewise.
 310         * doc/gty.texi: Likewise.
 311         * doc/invoke.texi: Likewise.
 312         * doc/lto.texi: Likewise.
 313         * doc/md.texi: Likewise.
 314         * doc/objc.texi: Likewise.
 315         * doc/rtl.texi: Likewise.
 316         * doc/tm.texi: Likewise.
 317         * dse.c: Likewise.
 318         * emit-rtl.c: Likewise.
 319         * emit-rtl.h: Likewise.
 320         * except.c: Likewise.
 321         * expmed.c: Likewise.
 322         * expr.c: Likewise.
 323         * fold-const.c: Likewise.
 324         * genautomata.c: Likewise.
 325         * gimple-fold.c: Likewise.
 326         * hard-reg-set.h: Likewise.
 327         * ifcvt.c: Likewise.
 328         * ipa-comdats.c: Likewise.
 329         * ipa-cp.c: Likewise.
 330         * ipa-devirt.c: Likewise.
 331         * ipa-fnsummary.c: Likewise.
 332         * ipa-icf.c: Likewise.
 333         * ipa-inline-transform.c: Likewise.
 334         * ipa-inline.c: Likewise.
 335         * ipa-polymorphic-call.c: Likewise.
 336         * ipa-profile.c: Likewise.
 337         * ipa-prop.c: Likewise.
 338         * ipa-pure-const.c: Likewise.
 339         * ipa-reference.c: Likewise.
 340         * ipa-split.c: Likewise.
 341         * ipa-visibility.c: Likewise.
 342         * ipa.c: Likewise.
 343         * ira-build.c: Likewise.
 344         * ira-color.c: Likewise.
 345         * ira-conflicts.c: Likewise.
 346         * ira-costs.c: Likewise.
 347         * ira-int.h: Likewise.
 348         * ira-lives.c: Likewise.
 349         * ira.c: Likewise.
 350         * ira.h: Likewise.
 351         * loop-invariant.c: Likewise.
 352         * loop-unroll.c: Likewise.
 353         * lower-subreg.c: Likewise.
 354         * lra-assigns.c: Likewise.
 355         * lra-constraints.c: Likewise.
 356         * lra-eliminations.c: Likewise.
 357         * lra-lives.c: Likewise.
 358         * lra-remat.c: Likewise.
 359         * lra-spills.c: Likewise.
 360         * lra.c: Likewise.
 361         * lto-cgraph.c: Likewise.
 362         * lto-streamer-out.c: Likewise.
 363         * postreload-gcse.c: Likewise.
 364         * predict.c: Likewise.
 365         * profile-count.h: Likewise.
 366         * profile.c: Likewise.
 367         * recog.c: Likewise.
 368         * ree.c: Likewise.
 369         * reload.c: Likewise.
 370         * reload1.c: Likewise.
 371         * reorg.c: Likewise.
 372         * resource.c: Likewise.
 373         * rtl.def: Likewise.
 374         * rtl.h: Likewise.
 375         * rtlanal.c: Likewise.
 376         * sched-deps.c: Likewise.
 377         * sched-ebb.c: Likewise.
 378         * sched-rgn.c: Likewise.
 379         * sel-sched-ir.c: Likewise.
 380         * sel-sched.c: Likewise.
 381         * shrink-wrap.c: Likewise.
 382         * simplify-rtx.c: Likewise.
 383         * symtab.c: Likewise.
 384         * target.def: Likewise.
 385         * toplev.c: Likewise.
 386         * tree-call-cdce.c: Likewise.
 387         * tree-cfg.c: Likewise.
 388         * tree-complex.c: Likewise.
 389         * tree-core.h: Likewise.
 390         * tree-eh.c: Likewise.
 391         * tree-inline.c: Likewise.
 392         * tree-loop-distribution.c: Likewise.
 393         * tree-nrv.c: Likewise.
 394         * tree-profile.c: Likewise.
 395         * tree-sra.c: Likewise.
 396         * tree-ssa-alias.c: Likewise.
 397         * tree-ssa-dce.c: Likewise.
 398         * tree-ssa-dom.c: Likewise.
 399         * tree-ssa-forwprop.c: Likewise.
 400         * tree-ssa-loop-im.c: Likewise.
 401         * tree-ssa-loop-ivcanon.c: Likewise.
 402         * tree-ssa-loop-ivopts.c: Likewise.
 403         * tree-ssa-loop-niter.c: Likewise.
 404         * tree-ssa-phionlycprop.c: Likewise.
 405         * tree-ssa-phiopt.c: Likewise.
 406         * tree-ssa-propagate.c: Likewise.
 407         * tree-ssa-threadedge.c: Likewise.
 408         * tree-ssa-threadupdate.c: Likewise.
 409         * tree-ssa-uninit.c: Likewise.
 410         * tree-ssanames.c: Likewise.
 411         * tree-streamer-out.c: Likewise.
 412         * tree.c: Likewise.
 413         * tree.h: Likewise.
 414         * vr-values.c: Likewise.
 415
 416 2019-01-09  Uroš Bizjak  <ubizjak@gmail.com>
 417
 418         * config/i386/i386-protos.h (ix86_expand_xorsign): New prototype.
 419         (ix86_split_xorsign): Ditto.
 420         * config/i386/i386.c (ix86_expand_xorsign): New function.
 421         (ix86_split_xorsign): Ditto.
 422         * config/i386/i386.md (UNSPEC_XORSIGN): New unspec.
 423         (xorsign<mode>3): New expander.
 424         (xorsign<mode>3_1): New insn_and_split pattern.
 425         * config/i386/sse.md (xorsign<mode>3): New expander.
 426
 427 2019-01-09  Eric Botcazou  <ebotcazou@adacore.com>
 428
 429         * config/sparc/sparc.md (*tablejump_sp32): Merge into...
 430         (*tablejump_sp64): Likewise.
 431         (*tablejump<P:mode>): ...this.
 432         (*call_address_sp32): Merge into...
 433         (*call_address_sp64): Likewise.
 434         (*call_address<P:mode>): ...this.
 435         (*call_symbolic_sp32): Merge into...
 436         (*call_symbolic_sp64): Likewise.
 437         (*call_symbolic<P:mode>): ...this.
 438         (call_value): Remove constraint and add predicate.
 439         (*call_value_address_sp32): Merge into...
 440         (*call_value_address_sp64): Likewise.
 441         (*call_value_address<P:mode>): ...this.
 442         (*call_value_symbolic_sp32): Merge into...
 443         (*call_value_symbolic_sp64): Likewise.
 444         (*call_value_symbolic<P:mode>): ...this.
 445         (*sibcall_symbolic_sp32): Merge into...
 446         (*sibcall_symbolic_sp64): Likewise.
 447         (*sibcall_symbolic<P:mode>): ...this.
 448         (sibcall_value): Remove constraint and add predicate.
 449         (*sibcall_value_symbolic_sp32): Merge into...
 450         (*sibcall_value_symbolic_sp64): Likewise.
 451         (*sibcall_value_symbolic<P:mode>): ...this.
 452         (window_save): Minor tweak.
 453         (*branch_sp32): Merge into...
 454         (*branch_sp64): Likewise.
 455         (*branch<P:mode>): ...this.
 456
 457 2019-01-09  Eric Botcazou  <ebotcazou@adacore.com>
 458             James Clarke  <jrtc27@jrtc27.com>
 459
 460         PR target/84010
 461         * config/sparc/sparc.c (sparc_legitimize_tls_address): Only use Pmode
 462         consistently in TLS address generation and adjust code to the renaming
 463         of patterns.  Mark calls to __tls_get_addr as const.
 464         * config/sparc/sparc.md (tgd_hi22): Turn into...
 465         (tgd_hi22<P:mode>): ...this and use Pmode throughout.
 466         (tgd_lo10): Turn into...
 467         (tgd_lo10<P:mode>): ...this and use Pmode throughout.
 468         (tgd_add32): Merge into...
 469         (tgd_add64): Likewise.
 470         (tgd_add<P:mode>): ...this and use Pmode throughout.
 471         (tldm_hi22): Turn into...
 472         (tldm_hi22<P:mode>): ...this and use Pmode throughout.
 473         (tldm_lo10): Turn into...
 474         (tldm_lo10<P:mode>): ...this and use Pmode throughout.
 475         (tldm_add32): Merge into...
 476         (tldm_add64): Likewise.
 477         (tldm_add<P:mode>): ...this and use Pmode throughout.
 478         (tldm_call32): Merge into...
 479         (tldm_call64): Likewise.
 480         (tldm_call<P:mode>): ...this and use Pmode throughout.
 481         (tldo_hix22): Turn into...
 482         (tldo_hix22<P:mode>): ...this and use Pmode throughout.
 483         (tldo_lox10): Turn into...
 484         (tldo_lox10<P:mode>): ...this and use Pmode throughout.
 485         (tldo_add32): Merge into...
 486         (tldo_add64): Likewise.
 487         (tldo_add<P:mode>): ...this and use Pmode throughout.
 488         (tie_hi22): Turn into...
 489         (tie_hi22<P:mode>): ...this and use Pmode throughout.
 490         (tie_lo10): Turn into...
 491         (tie_lo10<P:mode>): ...this and use Pmode throughout.
 492         (tie_ld64): Use DImode throughout.
 493         (tie_add32): Merge into...
 494         (tie_add64): Likewise.
 495         (tie_add<P:mode>): ...this and use Pmode throughout.
 496         (tle_hix22_sp32): Merge into...
 497         (tle_hix22_sp64): Likewise.
 498         (tle_hix22<P:mode>): ...this and use Pmode throughout.
 499         (tle_lox22_sp32): Merge into...
 500         (tle_lox22_sp64): Likewise.
 501         (tle_lox22<P:mode>): ...this and use Pmode throughout.
 502         (*tldo_ldub_sp32): Merge into...
 503         (*tldo_ldub_sp64): Likewise.
 504         (*tldo_ldub<P:mode>): ...this and use Pmode throughout.
 505         (*tldo_ldub1_sp32): Merge into...
 506         (*tldo_ldub1_sp64): Likewise.
 507         (*tldo_ldub1<P:mode>): ...this and use Pmode throughout.
 508         (*tldo_ldub2_sp32): Merge into...
 509         (*tldo_ldub2_sp64): Likewise.
 510         (*tldo_ldub2<P:mode>): ...this and use Pmode throughout.
 511         (*tldo_ldsb1_sp32): Merge into...
 512         (*tldo_ldsb1_sp64): Likewise.
 513         (*tldo_ldsb1<P:mode>): ...this and use Pmode throughout.
 514         (*tldo_ldsb2_sp32): Merge into...
 515         (*tldo_ldsb2_sp64): Likewise.
 516         (*tldo_ldsb2<P:mode>): ...this and use Pmode throughout.
 517         (*tldo_ldub3_sp64): Use DImode throughout.
 518         (*tldo_ldsb3_sp64): Likewise.
 519         (*tldo_lduh_sp32): Merge into...
 520         (*tldo_lduh_sp64): Likewise.
 521         (*tldo_lduh<P:mode>): ...this and use Pmode throughout.
 522         (*tldo_lduh1_sp32): Merge into...
 523         (*tldo_lduh1_sp64): Likewise.
 524         (*tldo_lduh1<P:mode>): ...this and use Pmode throughout.
 525         (*tldo_ldsh1_sp32): Merge into...
 526         (*tldo_ldsh1_sp64): Likewise.
 527         (*tldo_ldsh1<P:mode>): ...this and use Pmode throughout.
 528         (*tldo_lduh2_sp64): Use DImode throughout.
 529         (*tldo_ldsh2_sp64): Likewise.
 530         (*tldo_lduw_sp32): Merge into...
 531         (*tldo_lduw_sp64): Likewise.
 532         (*tldo_lduw<P:mode>): ...this and use Pmode throughout.
 533         (*tldo_lduw1_sp64): Use DImode throughout.
 534         (*tldo_ldsw1_sp64): Likewise.
 535         (*tldo_ldx_sp64): Likewise.
 536         (*tldo_stb_sp32): Merge into...
 537         (*tldo_stb_sp64): Likewise.
 538         (*tldo_stb<P:mode>): ...this and use Pmode throughout.
 539         (*tldo_sth_sp32): Merge into...
 540         (*tldo_sth_sp64): Likewise.
 541         (*tldo_sth<P:mode>): ...this and use Pmode throughout.
 542         (*tldo_stw_sp32): Merge into...
 543         (*tldo_stw_sp64): Likewise.
 544         (*tldo_stw<P:mode>): ...this and use Pmode throughout.
 545         (*tldo_stx_sp64): Use DImode throughout.
 546
 547 2018-01-09  Sudakshina Das  <sudi.das@arm.com>
 548
 549         * config/aarch64/aarch64.c (aarch64_override_options): Add case to
 550         check configure option to set BTI and Return Address Signing.
 551         * configure.ac: Add --enable-standard-branch-protection and
 552         --disable-standard-branch-protection.
 553         * configure: Regenerated.
 554         * doc/install.texi: Document the same.
 555
 556 2018-01-09  Sudakshina Das  <sudi.das@arm.com>
 557             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
 558
 559         * config.gcc (aarch64*-*-*): Add aarch64-bti-insert.o.
 560         * config/aarch64/aarch64.h: Update comment for TRAMPOLINE_SIZE.
 561         * config/aarch64/aarch64.c (aarch64_asm_trampoline_template): Update
 562         if bti is enabled.
 563         * config/aarch64/aarch64-bti-insert.c: New file.
 564         * config/aarch64/aarch64-passes.def (INSERT_PASS_BEFORE): Insert bti
 565         pass.
 566         * config/aarch64/aarch64-protos.h (make_pass_insert_bti): Declare the
 567         new bti pass.
 568         * config/aarch64/aarch64.md (unspecv): Add UNSPECV_BTI_NOARG,
 569         UNSPECV_BTI_C, UNSPECV_BTI_J and UNSPECV_BTI_JC.
 570         (bti_noarg, bti_j, bti_c, bti_jc): New define_insns.
 571         * config/aarch64/t-aarch64: Add rule for aarch64-bti-insert.o.
 572
 573 2018-01-09  Sudakshina Das  <sudi.das@arm.com>
 574
 575         * config/aarch64/aarch64-protos.h (aarch64_bti_enabled): Declare.
 576         * config/aarch64/aarch64.c (aarch64_handle_no_branch_protection):
 577         Disable bti for -mbranch-protection=none.
 578         (aarch64_handle_standard_branch_protection): Enable bti for
 579         -mbranch-protection=standard.
 580         (aarch64_handle_bti_protection): Enable bti for "bti" in the string to
 581         -mbranch-protection.
 582         (aarch64_bti_enabled): Check if bti is enabled.
 583         * config/aarch64/aarch64.opt: Declare target variable.
 584         * doc/invoke.texi: Add bti to the -mbranch-protection documentation.
 585
 586 2018-01-09  Sudakshina Das  <sudi.das@arm.com>
 587
 588         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use new
 589         epilogue/prologue scratch registers EP0_REGNUM and EP1_REGNUM.
 590         (aarch64_expand_epilogue): Likewise.
 591         (aarch64_output_mi_thunk): Likewise
 592         * config/aarch64/aarch64.h (REG_CLASS_CONTENTS): Change
 593         TAILCALL_ADDR_REGS to x16 and x17.
 594         * config/aarch64/aarch64.md: Define EP0_REGNUM and EP1_REGNUM.
 595
 596 2018-01-09  Sudakshina Das  <sudi.das@arm.com>
 597
 598         * config/aarch64/aarch64-option-extensions.def: Define
 599         AARCH64_OPT_EXTENSION for memtag, rng, sb, ssbs and predres.
 600         * gcc/config/aarch64/aarch64.h (AARCH64_FL_RNG): New.
 601         (AARCH64_FL_MEMTAG, ARCH64_FL_SB, AARCH64_FL_SSBS): New.
 602         (AARCH64_FL_PREDRES): New.
 603         (AARCH64_FL_FOR_ARCH8_5): Add AARCH64_FL_SB, AARCH64_FL_SSBS and
 604         AARCH64_FL_PREDRES by default.
 605         * gcc/doc/invoke.texi: Document rng, memtag, sb, ssbs and predres.
 606
 607 2018-01-09  Sudakshina Das  <sudi.das@arm.com>
 608
 609         * config/aarch64/aarch64-arches.def: Define AARCH64_ARCH for
 610         ARMv8.5-A.
 611         * gcc/config/aarch64/aarch64.h (AARCH64_FL_V8_5): New.
 612         (AARCH64_FL_FOR_ARCH8_5, AARCH64_ISA_V8_5): New.
 613         * gcc/doc/invoke.texi: Document ARMv8.5-A.
 614
 615 2019-01-09  Alejandro Martinez  <alejandro.martinezvicente@arm.com>
 616
 617         * config/aarch64/aarch64-sve.md (copysign<mode>3): New define_expand.
 618         (xorsign<mode>3): Likewise.
 619
 620 2019-01-09  Jelinek  <jakub@redhat.com>
 621
 622         PR middle-end/88758
 623         * tree.c (initializer_each_zero_or_onep) <case VECTOR_CST>: Use
 624         vector_cst_elt instead of VECTOR_CST_ENCODED_ELT.
 625
 626         PR rtl-optimization/88331
 627         * function.c (assign_stack_local_1): Don't set dynamic_align_addr if
 628         not currently_expanding_to_rtl.
 629
 630 2019-01-09  Eric Botcazou  <ebotcazou@adacore.com>
 631
 632         * doc/invoke.texi (-Os): Remove trailing spaces.
 633         (-finline-functions): Remove reference to -O2.
 634
 635 2019-01-08  Jakub Jelinek  <jakub@redhat.com>
 636
 637         PR rtl-optimization/79593
 638         * config/i386/i386.md (reg = mem; mem = reg): New define_peephole2.
 639
 640         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Delegitimize
 641         UNSPEC_FUSION_GPR to its argument.  Formatting fixes.
 642
 643 2019-01-08  Eric Botcazou  <ebotcazou@adacore.com>
 644
 645         PR bootstrap/88721
 646         * config/sparc/sparc.c (function_arg_slotno): Set *PPREGNO & *PPADDING
 647         to -1 on entry.
 648
 649         PR debug/88723
 650         * config/sparc/sparc.c (sparc_delegitimize_address): Deal with naked
 651         UNSPECs and UNSPEC_MOVE_GOTDATA specifically.
 652
 653 2019-01-08  H.J. Lu  <hongjiu.lu@intel.com>
 654
 655         PR target/88717
 656         * config/i386/i386.c (ix86_avx_u128_mode_exit): Call
 657         ix86_avx_u128_mode_entry.
 658
 659 2019-01-08  Martin Liska  <mliska@suse.cz>
 660
 661         PR tree-optimization/88753
 662         * tree-switch-conversion.c (switch_conversion::build_one_array):
 663         Come up with local variable constructor.  Convert first to
 664         type of constructor values.
 665
 666 2019-01-08  Richard Biener  <rguenther@suse.de>
 667
 668         PR tree-optimization/86554
 669         * tree-ssa-sccvn.c (eliminate_dom_walker, rpo_elim,
 670         rpo_avail): Move earlier.
 671         (visit_nary_op): When value-numbering to expressions
 672         with different overflow behavior make sure there's an
 673         available expression on the path.
 674
 675 2019-01-08  Sam Tebbs  <sam.tebbs@arm.com>
 676
 677         * config/aarch64/aarch64.c (BRANCH_PROTECT_STR_MAX,
 678         aarch64_parse_branch_protection,
 679         struct aarch64_branch_protect_type,
 680         aarch64_handle_no_branch_protection,
 681         aarch64_handle_standard_branch_protection,
 682         aarch64_validate_mbranch_protection,
 683         aarch64_handle_pac_ret_protection,
 684         aarch64_handle_attr_branch_protection,
 685         accepted_branch_protection_string,
 686         aarch64_pac_ret_subtypes,
 687         aarch64_branch_protect_types,
 688         aarch64_handle_pac_ret_leaf): Define.
 689         (aarch64_override_options_after_change_1, aarch64_override_options):
 690         Add check for accepted_branch_protection_string.
 691         (aarch64_option_save): Save accepted_branch_protection_string.
 692         (aarch64_option_restore): Save accepted_branch_protection_string.
 693         * config/aarch64/aarch64.c (aarch64_attributes): Add branch-protection.
 694         * config/aarch64/aarch64.opt: Add mbranch-protection. Deprecate
 695         msign-return-address.
 696         * doc/invoke.texi: Add mbranch-protection.
 697
 698 2019-01-08  Alan Modra  <amodra@gmail.com>
 699
 700         PR target/88614
 701         * genattrtab.c (max_attr_value, min_attr_value, or_attr_value):
 702         Delete "unknownp" parameter.  Adjust callers.  Handle
 703         CONST_INT, PLUS, MINUS, and MULT.
 704         (attr_value_aligned): Renamed from or_attr_value.
 705         (min_attr_value): Return INT_MIN for unhandled rtl case..
 706         (min_fn): ..and translate to INT_MAX here.
 707         (write_length_unit_log): Modify to cope without "unknown".
 708         (write_attr_value): Handle IF_THEN_ELSE.
 709
 710 2019-01-07  Richard Sandiford  <richard.sandiford@arm.com>
 711
 712         * tree-vect-stmts.c (vectorizable_store): Don't use the dataref_offset
 713         optimization for masked stores.
 714
 715 2019-01-07  Richard Sandiford  <richard.sandiford@arm.com>
 716
 717         PR middle-end/88567
 718         * tree-vect-loop.c (get_initial_defs_for_reduction): Pass the
 719         output vector directly to duplicate_and_interleave instead of
 720         going through a temporary.  Postpone insertion of ctor_seq to
 721         the end of the loop.
 722
 723 2019-01-07  Richard Earnshaw  <rearnsha@arm.com>
 724
 725         PR target/86891
 726         * config/aarch64/aarch64.c (aarch64_expand_subvti): New parameter
 727         unsigned_p.  Handle signed and unsigned overflow correction as
 728         required.
 729         * config/aarch64/aarch64-protos.h (aarch64_expand_subvti): Update
 730         prototype.
 731         * config/aarch64/aarch64.md (addv<mode>4): Use aarch64_plus_operand
 732         for operand 2.
 733         (add<mode>3_compareV_imm): Make this callable for expanding.
 734         (subv<GPI:mode>4): Use register_operand for operand 1.  Use
 735         aarch64_plus_operand for operand 2.
 736         (subv<GPI:mode>_insn): New insn pattern.
 737         (subv<GPI:mode>_imm): Likewise.
 738         (negv<GPI:mode>3): New expand pattern.
 739         (negv<GPI:mode>_insn): New insn pattern.
 740         (negv<GPI:mode>_cmp_only): Likewise.
 741         (cmpv<GPI:mode>_insn): Likewise.
 742         (subvti4): Use register_operand for operand 1.  Update call to
 743         aarch64_expand_subvti.
 744         (usubvti4): Likewise.
 745         (negvti3): New expand pattern.
 746         (negdi_carryout): New insn pattern.
 747         (negvdi_carryinV): New insn pattern.
 748         (sub<mode3>_compare1_imm): Delete named insn pattern, make anonymous
 749         version the named version.
 750         (peepholes to convert to sub<mode3>_compare1_imm): Adjust order of
 751         operands.
 752         (usub<GPI:mode>3_carryinC, usub<GPI:mode>3_carryinC_z1): New insn
 753         patterns.
 754         (usub<GPI:mode>3_carryinC_z2, usub<GPI:mode>3_carryinC): New insn
 755         patterns.
 756         (sub<mode>3_carryinCV, sub<mode>3_carryinCV_z1_z2): Delete.
 757         (sub<mode>3_carryinCV_z1, sub<mode>3_carryinCV_z2): Delete.
 758         (sub<mode>3_carryinCV): Delete.
 759         (sub<GPI:mode>3_carryinV): New expand pattern.
 760         sub<mode>3_carryinV, sub<mode>3_carryinV_z2): New insn patterns.
 761
 762 2019-01-07  Richard Biener  <rguenther@suse.de>
 763
 764         * tree-ssa-uncprop.c (ssa_equip_hash_traits): Remove in favor
 765         of tree_operand_hash.
 766
 767 2019-01-07  Richard Sandiford  <richard.sandiford@arm.com>
 768
 769         PR tree-optimization/88598
 770         * tree.h (single_nonzero_element): Declare.
 771         * tree.c (single_nonzero_element): New function.
 772         * match.pd: Fold certain reductions of X & CST to X[I] & CST[I]
 773         if I is the only nonzero element of CST.
 774
 775 2019-01-07  Richard Sandiford  <richard.sandiford@arm.com>
 776
 777         PR tree-optimization/88598
 778         * tree.h (initializer_each_zero_or_onep): Declare.
 779         * tree.c (initializer_each_zero_or_onep): New function.
 780         (signed_or_unsigned_type_for): Handle float types too.
 781         (unsigned_type_for, signed_type_for): Update comments accordingly.
 782         * match.pd: Fold x * { 0 or 1, 0 or 1, ...} to
 783         x & { 0 or -1, 0 or -1, ... }.
 784
 785 2019-01-07  Jonathan Wakely  <jwakely@redhat.com>
 786
 787         * doc/install.texi: Replace references to x86_64-unknown-linux-gnu
 788         with x86_64-pc-linux-gnu.
 789
 790 2019-01-07  Tom de Vries  <tdevries@suse.de>
 791
 792         PR target/85486
 793         * config/nvptx/nvptx.c (has_vector_partitionable_routine_calls_p): New
 794         function.
 795         (nvptx_goacc_validate_dims): Force vl32 if calling vector-partitionable
 796         routines.
 797
 798 2019-01-07  Jakub Jelinek  <jakub@redhat.com>
 799
 800         * config/i386/sse.md (vec_extract<mode><ssehalfvecmodelower>): Use
 801         V_256_512 iterator instead of V_512 and TARGET_AVX instead of
 802         TARGET_AVX512F as condition.
 803
 804         PR debug/88723
 805         * dwarf2out.c (const_ok_for_output_1): Remove redundant call to
 806         const_not_ok_for_debug_p target hook.
 807         (mem_loc_descriptor) <case UNSPEC>: Only call const_ok_for_output_1
 808         on UNSPEC and subexpressions thereof if all subexpressions of the
 809         UNSPEC are CONSTANT_P.
 810
 811         PR tree-optimization/88676
 812         * tree-ssa-phiopt.c (two_value_replacement): New function.
 813         (tree_ssa_phiopt_worker): Call it.
 814
 815         PR sanitizer/88619
 816         * cfgexpand.c (expand_stack_vars): Only align prev_offset to
 817         ASAN_MIN_RED_ZONE_SIZE, not to maximum of that and alignb.
 818
 819         PR c++/85052
 820         * tree-vect-generic.c: Include insn-config.h and recog.h.
 821         (expand_vector_piecewise): Add defaulted ret_type argument,
 822         if non-NULL, use that in preference to type for the result type.
 823         (expand_vector_parallel): Formatting fix.
 824         (do_vec_conversion, do_vec_narrowing_conversion,
 825         expand_vector_conversion): New functions.
 826         (expand_vector_operations_1): Call expand_vector_conversion
 827         for VEC_CONVERT ifn calls.
 828         * internal-fn.def (VEC_CONVERT): New internal function.
 829         * internal-fn.c (expand_VEC_CONVERT): New function.
 830         * fold-const-call.c (fold_const_vec_convert): New function.
 831         (fold_const_call): Use it for CFN_VEC_CONVERT.
 832         * doc/extend.texi (__builtin_convertvector): Document.
 833
 834 2019-01-07  Tom de Vries  <tdevries@suse.de>
 835
 836         * config/nvptx/nvptx-protos.h (nvptx_output_red_partition): Declare.
 837         * config/nvptx/nvptx.c (vector_red_size, vector_red_align,
 838         vector_red_partition, vector_red_sym): New global variables.
 839         (nvptx_option_override): Initialize vector_red_sym.
 840         (nvptx_declare_function_name): Restore red_partition register.
 841         (nvptx_file_end): Emit code to declare the vector reduction variables.
 842         (nvptx_output_red_partition): New function.
 843         (nvptx_expand_shared_addr): Add vector argument. Use it to handle
 844         large vector reductions.
 845         (enum nvptx_builtins): Add NVPTX_BUILTIN_VECTOR_ADDR.
 846         (nvptx_init_builtins): Add VECTOR_ADDR.
 847         (nvptx_expand_builtin): Update call to nvptx_expand_shared_addr.
 848         Handle nvptx_expand_shared_addr.
 849         (nvptx_get_shared_red_addr): Add vector argument and handle large
 850         vectors.
 851         (nvptx_goacc_reduction_setup): Add offload_attrs argument and handle
 852         large vectors.
 853         (nvptx_goacc_reduction_init): Likewise.
 854         (nvptx_goacc_reduction_fini): Likewise.
 855         (nvptx_goacc_reduction_teardown): Likewise.
 856         (nvptx_goacc_reduction): Update calls to nvptx_goacc_reduction_{setup,
 857         init,fini,teardown}.
 858         (nvptx_init_axis_predicate): Initialize vector_red_partition.
 859         (nvptx_set_current_function): Init vector_red_partition.
 860         * config/nvptx/nvptx.md (UNSPECV_RED_PART): New unspecv.
 861         (nvptx_red_partition): New insn.
 862         * config/nvptx/nvptx.h (struct machine_function): Add red_partition.
 863
 864 2019-01-07  Tom de Vries  <tdevries@suse.de>
 865
 866         PR target/85381
 867         * config/nvptx/nvptx.c (nvptx_process_pars): Don't emit barriers for
 868         empty loops.
 869
 870 2019-01-07  Tom de Vries  <tdevries@suse.de>
 871
 872         * config/nvptx/nvptx.c (oacc_bcast_partition): Declare.
 873         (nvptx_option_override): Init oacc_bcast_partition.
 874         (nvptx_init_oacc_workers): New function.
 875         (nvptx_declare_function_name): Call nvptx_init_oacc_workers.
 876         (nvptx_needs_shared_bcast): New function.
 877         (nvptx_find_par): Generalize to enable vectors to use shared-memory
 878         to propagate state.
 879         (nvptx_shared_propagate): Initialize vector bcast partition and
 880         synchronization state.
 881         (nvptx_single):  Generalize to enable vectors to use shared-memory
 882         to propagate state.
 883         (nvptx_process_pars): Likewise.
 884         (nvptx_set_current_function): Initialize oacc_broadcast_partition.
 885         * config/nvptx/nvptx.h (struct machine_function): Add
 886         bcast_partition and sync_bar members.
 887
 888 2019-01-07  Tom de Vries  <tdevries@suse.de>
 889
 890         * config/nvptx/nvptx.c (nvptx_welformed_vector_length_p)
 891         (nvptx_apply_dim_limits): New function.
 892         (nvptx_goacc_validate_dims_1): Allow PTX_MAX_VECTOR_LENGTH larger than
 893         PTX_WARP_SIZE.
 894
 895 2019-01-07  Tom de Vries  <tdevries@suse.de>
 896
 897         * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1): Move warnings to
 898         as late as possible.
 899
 900 2019-01-07  Tom de Vries  <tdevries@suse.de>
 901
 902         * config/nvptx/nvptx.c (PTX_VECTOR_LENGTH): Remove.
 903         (PTX_DEFAULT_VECTOR_LENGTH, PTX_MAX_VECTOR_LENGTH): Define.
 904         (nvptx_goacc_validate_dims_1, nvptx_dim_limit)
 905         (nvptx_goacc_reduction_fini): Use PTX_DEFAULT_VECTOR_LENGTH,
 906         PTX_MAX_VECTOR_LENGTH and PTX_WARP_SIZE instead of PTX_VECTOR_LENGTH.
 907
 908 2019-01-07  Tom de Vries  <tdevries@suse.de>
 909
 910         * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Add asserts.
 911
 912 2019-01-07  Tom de Vries  <tdevries@suse.de>
 913
 914         * omp-offload.c (oacc_get_min_dim): New function.
 915         * omp-offload.h (oacc_get_min_dim): Declare.
 916
 917 2018-12-26  Mateusz B  <mateuszb@poczta.onet.pl>
 918
 919         PR target/88521
 920         * config/i386/i386.c (function_value_ms_64): Return small sturct in
 921         AX_REG and float/double in FIRST_SSE_REG for 4 or 8 byte modes.
 922
 923 2019-01-05  Jan Hubicka  <hubicka@ucw.cz>
 924
 925         PR tree-opt/86020
 926         Revert:
 927         2017-05-22  Jan Hubicka  <hubicka@ucw.cz>
 928
 929         * ipa-inline.c (edge_badness): Use inlined_time instead of
 930         inline_summaries->get.
 931
 932 2019-01-05  Jan Hubicka  <hubicka@ucw.cz>
 933
 934         * opts.c (enable_fdo_optimizations): Enable
 935         version-loops-for-strides, loop-interchange, unrol-and-jam
 936         and tree-loop-distribution.
 937         * invoke.texi: Document newly enabled options.
 938
 939 2019-01-05  Jan Hubicka  <hubicka@ucw.cz>
 940
 941         * doc/invoke.texi (max-inline-insns-small): New parameters.
 942         * ipa-inline.c (want_early_inline_function_p): simplify.
 943         (want_inline_small_function_p): Fix pasto from previous patch;
 944         use max-inline-insns-small bound.
 945         * params.def (max-inline-insns-small): New param.
 946         * ipa-fnsummary.c (analyze_function_body): Initialize time/size
 947         variables correctly.
 948
 949 2019-01-05  Jan Hubicka  <hubicka@ucw.cz>
 950
 951         * doc/invoke.texi: Document max-inline-insns-size,
 952         uninlined-function-insns, uninlined-function-time,
 953         uninlined-thunk-insns and uninlined-thunk-time.
 954         * params.def: Add max-inline-insns-size,
 955         uninlined-function-insns, uninlined-function-time,
 956         uninlined-thunk-insns and uninlined-thunk-time.
 957         * ipa-fnsummary.c (compute_fn_summary, analyze_function_body): Use
 958         new parameters.
 959         * ipa-inline.c (can_inline_edge_by_limits_p,
 960         want_inline_small_function_p): Use new parameters.
 961
 962 2019-01-05  Jan Hubicka  <hubicka@ucw.cz>
 963
 964         * ipa-fnsummary.c (analyze_function_body): Fix accounting of time.
 965
 966 2019-01-05  Jakub Jelinek  <jakub@redhat.com>
 967
 968         PR middle-end/82564
 969         PR target/88620
 970         * expr.c (expand_assignment): For calls returning VLA structures
 971         if to_rtx is not a MEM, force it into a stack temporary.
 972
 973         PR debug/88635
 974         * dwarf2out.c (const_ok_for_output_1): Reject MINUS that contains
 975         SYMBOL_REF, CODE_LABEL or UNSPEC in subexpressions of second argument.
 976         Reject PLUS that contains SYMBOL_REF, CODE_LABEL or UNSPEC in
 977         subexpressions of both operands.
 978         (mem_loc_descriptor): Handle UNSPEC if target hook acks it and all the
 979         subrtxes are CONSTANT_P.
 980         * config/i386/i386.c (ix86_const_not_ok_for_debug_p): Revert
 981         2018-11-09 changes.
 982
 983 2019-01-04  Jan Hubicka  <hubicka@ucw.cz>
 984
 985         * params.def (hot-bb-count-ws-permille): Set to 990.
 986
 987 2019-01-04  Martin Sebor  <msebor@redhat.com>
 988
 989         PR c/88546
 990         * attribs.c (decls_mismatched_attributes): Avoid warning for attribute
 991         leaf.
 992
 993 2019-01-04  Martin Sebor  <msebor@redhat.com>
 994
 995         PR c/88363
 996         * doc/extend.texi (attribute alloc_align, alloc_size): Update.
 997
 998 2019-01-04  Jakub Jelinek  <jakub@redhat.com>
 999
1000         * gdbinit.in: Turn off pagination for the skip commands, restore
1001         it to previous state afterwards.
1002
1003 2019-01-04  Sam Tebbs  <sam.tebbs@arm.com>
1004
1005         PR gcc/87763
1006         * gcc.target/aarch64/combine_bfxil.c: Change scan-assembler-times bfxil
1007         count to 18.
1008
1009 2019-01-04  Jakub Jelinek  <jakub@redhat.com>
1010
1011         PR target/88594
1012         * config/i386/i386.c (ix86_expand_divmod_libfunc): Use mode instead
1013         of GET_MODE (opN) as modes of the libcall arguments.
1014
1015 2019-01-04  Jan Beulich  <jbeulich@suse.com>
1016
1017         * config/i386/sse.md
1018         (<avx512>_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>,
1019         <avx512>_cmp<mode>3<mask_scalar_merge_name>,
1020         <avx512>_ucmp<mode>3<mask_scalar_merge_name>,
1021         <avx512>_ucmp<mode>3<mask_scalar_merge_name>,
1022         avx512f_vmcmp<mode>3<round_saeonly_name>,
1023         avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
1024         avx512f_maskcmp<mode>3,
1025         <avx512>_cvt<ssemodesuffix>2mask<mode>,
1026         <avx512>_cvt<ssemodesuffix>2mask<mode>,
1027         *<avx512>_cvtmask2<ssemodesuffix><mode>,
1028         *<avx512>_cvtmask2<ssemodesuffix><mode>,
1029         <avx512>_eq<mode>3<mask_scalar_merge_name>_1,
1030         <avx512>_eq<mode>3<mask_scalar_merge_name>_1,
1031         <avx512>_gt<mode>3<mask_scalar_merge_name>,
1032         <avx512>_gt<mode>3<mask_scalar_merge_name>,
1033         <avx512>_testm<mode>3<mask_scalar_merge_name>,
1034         <avx512>_testnm<mode>3<mask_scalar_merge_name>,
1035         *<avx512>_testm<VI1248_AVX512VLBW:mode>3_zext,
1036         *<avx512>_testm<VI1248_AVX512VLBW:mode>3_zext_mask,
1037         *<avx512>_testnm<VI1248_AVX512VLBW:mode>3_zext,
1038         *<avx512>_testnm<VI1248_AVX512VLBW:mode>3_zext_mask,
1039         avx512cd_maskb_vec_dup<mode>,
1040         avx512cd_maskw_vec_dup<mode>,
1041         avx512dq_fpclass<mode><mask_scalar_merge_name>,
1042         avx512dq_vmfpclass<mode>,
1043         avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): Use =k
1044         instead of =Yk.
1045
1046 2019-01-03  Martin Sebor  <msebor@redhat.com>
1047
1048         PR tree-optimization/88659
1049         * calls.c (maybe_warn_nonstring_arg): Avoid assuming maxlen is set.
1050
1051 2019-01-03  Aaron Sawdey  <acsawdey@linux.ibm.com>
1052
1053         * config/rs6000/rs6000-string.c (expand_block_move): Don't use
1054         unaligned vsx and avoid lxvd2x/stxvd2x.
1055         (gen_lvx_v4si_move): New function.
1056
1057 2019-01-03  Tom de Vries  <tdevries@suse.de>
1058
1059         * config/nvptx/nvptx.c (MACH_VECTOR_LENGTH, MACH_MAX_WORKERS): Define.
1060         (init_axis_dim, nvptx_mach_max_workers, nvptx_mach_vector_length): New
1061         function.
1062         * config/nvptx/nvptx.h (struct machine_function): Add axis_dims.
1063
1064 2019-01-03  Tom de Vries  <tdevries@suse.de>
1065
1066         * config/nvptx/nvptx.c (struct offload_attrs): New.
1067         (populate_offload_attrs): New function.  Factor mask extraction out of
1068         nvptx_reorg.  Add extraction of dimensions.
1069         (nvptx_reorg): Use populate_offload_attrs.
1070
1071 2019-01-03  Tom de Vries  <tdevries@suse.de>
1072
1073         * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1): Add early-out
1074         cases for oacc_min_dims_p and routine_p.  Add asserts for
1075         oacc_default_dims_p and offload_region_p.
1076
1077 2019-01-03  Tom de Vries  <tdevries@suse.de>
1078
1079         * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1): New function,
1080         factored out of ...
1081         (nvptx_goacc_validate_dims): ... here.
1082
1083 2019-01-03  Jan Hubicka  <hubicka@ucw.cz>
1084
1085         PR tree-optimization/85574
1086         * tree-ssa-uncprop.c (struct equiv_hash_elt): Remove unused
1087         structure.
1088         (struct ssa_equip_hash_traits): Declare.
1089         (val_ssa_equiv): Use custom hash traits using operand_equal_p.
1090
1091 2019-01-03  Jakub Jelinek  <jakub@redhat.com>
1092
1093         PR debug/88644
1094         * dwarf2out.c (modified_type_die): If type is equal to sizetype,
1095         change it to qualified_type.
1096
1097 2019-01-03  Jan Hubicka  <hubicka@ucw.cz>
1098
1099         * ipa-utils.c (scale_ipa_profile_for_fn): Break out from ...
1100         (ipa_merge_profiles): ... here; do not ICE on thunks and aliases.
1101
1102 2019-01-02  Martin Sebor  <msebor@redhat.com>
1103             Jeff Law  <law@redhat.com>
1104
1105         * gimple-fold.c (strlen_range_kind): Remove SRK_LENRANGE_2.
1106         (get_range_strlen_tree): Update appropriately.
1107         (get_range_strlen)
1108         * gimple-fold.h (get_range_strlen): Drop unused last argument.
1109
1110         * gimple-fold.c (gimple_fold_builtin_strlen): Use set_strlen_range
1111         rather than set_range_info.
1112         * tree-ssa-strlen.c (set_strlen_range): Extracted from
1113         maybe_set_strlen_range.  Handle potentially boundary crossing
1114         cases more conservatively.
1115         (maybe_set_strlen_range): Parts refactored into set_strlen_range.
1116         Call set_strlen_range.
1117         * tree-ssa-strlen.h (set_strlen_range): Add prototype.
1118
1119         PR middle-end/88663
1120         * gimple-fold.c (get_range_strlen): Update prototype to no longer
1121         need the flexp argument.
1122         (get_range_strlen_tree): Drop flexp argument.  Drop flexp argument
1123         from calls to get_range_strlen.  Update comments.  Just update
1124         VAL for an unterminated const char array and let the reset of the
1125         code handle it normally.  No longer try to set *flexp.  Adjust
1126         return value.
1127         (get_range_strlen): Update for the new get_range_strlen API.
1128         (get_maxval_strlen): Similarly.
1129         (gimple_fold_builtin_strlen): Handle update meaning of return value
1130         from get_range_strlen.
1131         * gimple-ssa-sprintf.c (get_string_length): Update for the new
1132         get_range_strlen API.
1133
1134 2019-01-02  Jan Hubicka  <hubicka@ucw.cz>
1135
1136         PR lto/88130
1137         * varpool.c (varpool_node::ctor_useable_for_folding_p): Also return
1138         false at WPA time when body was removed.
1139
1140 2019-01-02  Martin Liska  <mliska@suse.cz>
1141
1142         PR tree-optimization/88650
1143         * predict.c (set_even_probabilities): Calculate probability
1144         remainer only when really used.
1145
1146 2019-01-02  Richard Biener  <rguenther@suse.de>
1147
1148         PR middle-end/88651
1149         * tree-data-ref.c (analyze_subscript_affine_affine): Use
1150         widest_ints when mangling max_stmt_execution results.
1151
1152 2019-01-02  Richard Biener  <rguenther@suse.de>
1153
1154         PR tree-optimization/88621
1155         * tree-ssa-loop-im.c (gather_mem_refs_stmt): Fix pastos, avoid
1156         bitfields when canoncalizing.
1157
1158 2019-01-02  Richard Biener  <rguenther@suse.de>
1159
1160         PR target/87545
1161         * config/i386/x86-tune-costs.h (intel_cost): Adjust
1162         cost of cheap SSE instruction.
1163
1164 2019-01-02  Richard Biener  <rguenther@suse.de>
1165
1166         PR ipa/85574
1167         * ipa-icf.h (sem_item_optimizer::sort_congruence_split): Declare.
1168         * ipa-icf.c (sem_item_optimizer::sort_congruence_split): New
1169         function.
1170         (sem_item_optimizer::do_congruence_step_f): Sort the congruence
1171         set after UIDs before splitting them.
1172
1173 2019-01-01  Martin Sebor  <msebor@redhat.com>
1174             Jeff Law  <law@redhat.com>
1175
1176         * gimple-fold.c (get_range_strlen_tree): Record if the computed
1177         length is optimistic.  If it is, then arrange to compute the
1178         conservative length as well.
1179
1180         * gimple-fold.h (get_range_strlen): Update prototype.
1181         * builtins.c (check_access): Update call to get_range_strlen to use
1182         c_strlen_data pointer.   Change various variable accesses to instead
1183         pull data from the c_strlen_data structure.
1184         (check_strncat_sizes, expand_builtin_strncat): Likewise.
1185         * calls.c (maybe_warn_nonstring_arg): Likewise.
1186         * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Likewise.  Reset
1187         minimum length if maximum lengh is unknown.
1188         * gimple-ssa-sprintf.c (get_string_length): Likewise.  Drop code
1189         that used c_strlen, it's no longer needed.  Restructure slightly.
1190         (format_string): Set unlikely range appropriately.
1191         * gimple-fold.c (get_range_strlen): Update comments.  Fix minor
1192         formatting issues.
1193         (get_range_strlen):  Accept c_strlen_data pointer for external
1194         call sites as well.  Pass through to call to internal get_range_strlen.
1195         Adjust minlen, maxlen and maxbound as needed.
1196         (get_maxval_strlen): Update comments.
1197         (gimple_fold_builtin_strlen):  Update call to get_range_strlen
1198         to use c_strlen_data pointer.  Change variable accesses to instead
1199         use c_strlen_data data members.
1200
1201         * gimple-fold.c (get_range_strlen): Update prototype.
1202         (get_range_strlen_tree): Update prototype.  Drop minlen/maxlen
1203         local variables.  Use pdata to return information to caller.
1204         Update calls to get_range_strlen.  Update pdata->maxbound.
1205         (get_range_strlen -- static version): Similarly.
1206         (get_range_strlen -- extern version): Update for internal
1207         get_range_strlen API change.  Convert to external data format.
1208         (get_maxval_strlen): Similarly.
1209
1210 2019-01-01  Jan Hubicka  <hubicka@ucw.cz>
1211
1212         * coverage.c (get_coverage_counts): Use current_function_decl.
1213         * profile.c (read_thunk_profile): New function.
1214         (branch_prob): Add THUNK parameter.
1215         * tree-profile.c (tree_profiling): Handle thunks.
1216         * value-prof.c (init_node_map): Handle thunks.
1217         * value-prof.h (branch_prob): Upate prototype.
1218         (read_thunk_profile): Declare.
1219
1220 2019-01-01  Jakub Jelinek  <jakub@redhat.com>
1221
1222         Update copyright years.
1223
1224         * gcc.c (process_command): Update copyright notice dates.
1225         * gcov-dump.c (print_version): Ditto.
1226         * gcov.c (print_version): Ditto.
1227         * gcov-tool.c (print_version): Ditto.
1228         * gengtype.c (create_file): Ditto.
1229         * doc/cpp.texi: Bump @copying's copyright year.
1230         * doc/cppinternals.texi: Ditto.
1231         * doc/gcc.texi: Ditto.
1232         * doc/gccint.texi: Ditto.
1233         * doc/gcov.texi: Ditto.
1234         * doc/install.texi: Ditto.
1235         * doc/invoke.texi: Ditto.
1236 \f
1237 Copyright (C) 2019 Free Software Foundation, Inc.
1238
1239 Copying and distribution of this file, with or without modification,
1240 are permitted in any medium without royalty provided the copyright
1241 notice and this notice are preserved.