gcc/ChangeLog

   1 2014-05-23  Segher Boessenkool  <segher@kernel.crashing.org>
   2
   3         * config/rs6000/440.md (ppc440-integer): Include shift without
   4         dot.
   5         (ppc440-compare): Include shift with dot.
   6         * config/rs6000/e300c2c3.md (ppce300c3_iu): Include shift without
   7         dot.
   8         * config/rs6000/e5500.md (e5500_sfx2): Include constant shift
   9         without dot.
  10         * config/rs6000/e6500.md (e6500_sfx): Exclude constant shift
  11         without dot.
  12         (e6500_sfx2): Include it.
  13         * config/rs6000/rs6000.md ( *zero_extend<mode>di2_internal1,
  14         *zero_extend<mode>di2_internal2, *zero_extend<mode>di2_internal3,
  15         *zero_extendsidi2_lfiwzx, andsi3_mc, andsi3_nomc,
  16         andsi3_internal0_nomc, extzvsi_internal, extzvdi_internal,
  17         *extzvdi_internal1, *extzvdi_internal2, rotlsi3, *rotlsi3_64,
  18         *rotlsi3_internal4, *rotlsi3_internal7le, *rotlsi3_internal7be,
  19         *rotlsi3_internal10le, *rotlsi3_internal10be, rlwinm,
  20         *lshiftrt_internal1le, *lshiftrt_internal1be,
  21         *lshiftrt_internal4le, *lshiftrt_internal4be, rotldi3,
  22         *rotldi3_internal4, *rotldi3_internal7le, *rotldi3_internal7be,
  23         *rotldi3_internal10le, *rotldi3_internal10be,
  24         *rotldi3_internal13le, *rotldi3_internal13be, *ashldi3_internal4,
  25         ashldi3_internal5, *ashldi3_internal6, *ashldi3_internal7,
  26         ashldi3_internal8, *ashldi3_internal9, anddi3_mc, anddi3_nomc,
  27         *anddi3_internal2_mc, *anddi3_internal3_mc, and 4 anonymous
  28         define_insns): Use type "shift" in the appropriate alternatives.
  29
  30 2014-05-23  Segher Boessenkool  <segher@kernel.crashing.org>
  31
  32         * config/rs6000/rs6000.md (type): Add "logical".  Delete
  33         "fast_compare".
  34         (dot): Adjust comment.
  35         (andsi3_mc, *andsi3_internal2_mc, *andsi3_internal3_mc,
  36         *andsi3_internal4, *andsi3_internal5_mc, *boolsi3_internal2,
  37         *boolsi3_internal3, *boolccsi3_internal2, *boolccsi3_internal3,
  38         anddi3_mc, *anddi3_internal2_mc, *anddi3_internal3_mc,
  39         *booldi3_internal2, *booldi3_internal3, *boolcdi3_internal2,
  40         *boolcdi3_internal3, *boolccdi3_internal2, *boolccdi3_internal3,
  41         *mov<mode>_internal2, and 10 anonymous define_insns): Use
  42         "logical".
  43         * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
  44
  45         * config/rs6000/40x.md: (ppc403-integer, ppc403-compare): Adjust.
  46         * config/rs6000/440.md: (ppc440-integer, ppc440-compare): Adjust.
  47         * config/rs6000/476.md: (ppc476-simple-integer, ppc476-compare):
  48         Adjust.
  49         * config/rs6000/603.md: (ppc603-integer, ppc603-compare): Adjust.
  50         * config/rs6000/6xx.md: (ppc604-integer, ppc604-compare): Adjust.
  51         * config/rs6000/7450.md: (ppc7450-integer, ppc7450-compare):
  52         Adjust.
  53         * config/rs6000/7xx.md: (ppc750-integer, ppc750-compare): Adjust.
  54         * config/rs6000/8540.md: (ppc8540_su): Adjust.
  55         * config/rs6000/cell.md: (cell-integer, cell-fast-cmp,
  56         cell-cmp-microcoded): Adjust.
  57         * config/rs6000/e300c2c3.md: (ppce300c3_cmp, ppce300c3_iu):
  58         Adjust.
  59         * config/rs6000/e500mc.md: (e500mc_su): Adjust.
  60         * config/rs6000/e500mc64.md: (e500mc64_su, e500mc64_su2): Adjust.
  61         * config/rs6000/e5500.md: (e5500_sfx, e5500_sfx2): Adjust.
  62         * config/rs6000/e6500.md: (e6500_sfx, e6500_sfx2): Adjust.
  63         * config/rs6000/mpc.md: (mpccore-integer, mpccore-compare):
  64         Adjust.
  65         * config/rs6000/power4.md: (power4-integer, power4-cmp): Adjust.
  66         * config/rs6000/power5.md: (power5-integer, power5-cmp): Adjust.
  67         * config/rs6000/power6.md: (power6-integer, power6-fast-compare):
  68         Adjust.
  69         * config/rs6000/power7.md: (power7-integer, power7-cmp): Adjust.
  70         * config/rs6000/power8.md: (power8-1cyc, power8-fast-compare):
  71         Adjust.  Adjust comment.
  72         * config/rs6000/rs64.md: (rs64a-integer, rs64a-compare): Adjust.
  73         * config/rs6000/titan.md: (titan_fxu_adder, titan_fxu_alu):
  74         Adjust.
  75
  76 2014-05-23  Segher Boessenkool  <segher@kernel.crashing.org>
  77
  78         * config/rs6000/rs6000.md (type): Add "add".
  79         (*add<mode>3_internal1, addsi3_high, *add<mode>3_internal2,
  80         *add<mode>3_internal3, *neg<mode>2_internal, and 5 anonymous
  81         define_insns): Use it.
  82         * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
  83
  84         * config/rs6000/40x.md (ppc403-integer, ppc403-compare): Adjust.
  85         * config/rs6000/440.md (ppc440-integer, ppc440-compare): Adjust.
  86         * config/rs6000/476.md (ppc476-simple-integer, ppc476-compare):
  87         Adjust.
  88         * config/rs6000/601.md (ppc601-integer): Adjust.
  89         * config/rs6000/603.md (ppc603-integer, ppc603-compare): Adjust.
  90         * config/rs6000/6xx.md (ppc604-integer, ppc604-compare): Adjust.
  91         * config/rs6000/7450.md (ppc7450-integer, ppc7450-compare): Adjust.
  92         * config/rs6000/7xx.md (ppc750-integer, ppc750-compare): Adjust.
  93         * config/rs6000/8540.md (ppc8540_su): Adjust.
  94         * config/rs6000/cell.md (cell-integer, cell-fast-cmp,
  95         cell-cmp-microcoded): Adjust.
  96         * config/rs6000/e300c2c3.md (ppce300c3_cmp, ppce300c3_iu): Adjust.
  97         * config/rs6000/e500mc.md (e500mc_su): Adjust.
  98         * config/rs6000/e500mc64.md (e500mc64_su, e500mc64_su2): Adjust.
  99         * config/rs6000/e5500.md (e5500_sfx, e5500_sfx2): Adjust.
 100         * config/rs6000/e6500.md (e6500_sfx, e6500_sfx2): Adjust.
 101         * config/rs6000/mpc.md (mpccore-integer, mpccore-compare): Adjust.
 102         * config/rs6000/power4.md (power4-integer, power4-cmp): Adjust.
 103         * config/rs6000/power5.md (power5-integer, power5-cmp): Adjust.
 104         * config/rs6000/power6.md (power6-integer, power6-fast-compare):
 105         Adjust.
 106         * config/rs6000/power7.md (power7-integer, power7-cmp): Adjust.
 107         * config/rs6000/power8.md (power8-1cyc, power8-fast-compare):
 108         Adjust.
 109         * config/rs6000/rs64.md (rs64a-integer, rs64a-compare): Adjust.
 110         * config/rs6000/titan.md (titan_fxu_adder, titan_fxu_alu): Adjust.
 111
 112 2014-05-23  Segher Boessenkool  <segher@kernel.crashing.org>
 113
 114         * config/rs6000/rs6000.md (type): Delete "var_shift_rotate",
 115         "delayed_compare", "var_delayed_compare".
 116         (var_shift): New attribute.
 117         (cell_micro): Adjust.
 118         (*andsi3_internal2_mc, *andsi3_internal3_mc, *andsi3_internal4,
 119         *andsi3_internal5_mc, *extzvsi_internal1, *extzvsi_internal2,
 120         rotlsi3, *rotlsi3_64, *rotlsi3_internal2, *rotlsi3_internal3,
 121         *rotlsi3_internal4, *rotlsi3_internal5, *rotlsi3_internal6,
 122         *rotlsi3_internal8le, *rotlsi3_internal8be, *rotlsi3_internal9le,
 123         *rotlsi3_internal9be, *rotlsi3_internal10le, *rotlsi3_internal10be,
 124         *rotlsi3_internal11le, *rotlsi3_internal11be, *rotlsi3_internal12le,
 125         *rotlsi3_internal12be, ashlsi3, *ashlsi3_64, lshrsi3, *lshrsi3_64,
 126         *lshiftrt_internal2le, *lshiftrt_internal2be, *lshiftrt_internal3le,
 127         *lshiftrt_internal3be, *lshiftrt_internal5le, *lshiftrt_internal5be,
 128         *lshiftrt_internal5le, *lshiftrt_internal5be, ashrsi3, *ashrsi3_64,
 129         rotldi3, *rotldi3_internal2, *rotldi3_internal3, *rotldi3_internal4,
 130         *rotldi3_internal5, *rotldi3_internal6, *rotldi3_internal7le,
 131         *rotldi3_internal7be, *rotldi3_internal8le, *rotldi3_internal8be,
 132         *rotldi3_internal9le, *rotldi3_internal9be, *rotldi3_internal10le,
 133         *rotldi3_internal10be, *rotldi3_internal11le, *rotldi3_internal11be,
 134         *rotldi3_internal12le, *rotldi3_internal12be, *rotldi3_internal13le,
 135         *rotldi3_internal13be, *rotldi3_internal14le, *rotldi3_internal14be,
 136         *rotldi3_internal15le, *rotldi3_internal15be, *ashldi3_internal1,
 137         *ashldi3_internal2, *ashldi3_internal3, *lshrdi3_internal1,
 138         *lshrdi3_internal2, *lshrdi3_internal3, *ashrdi3_internal1,
 139         *ashrdi3_internal2, *ashrdi3_internal3, *anddi3_internal2_mc,
 140         *anddi3_internal3_mc, as well as 11 anonymous define_insns): Adjust.
 141         * config/rs6000/rs6000.c (rs6000_adjust_cost, is_cracked_insn,
 142         insn_must_be_first_in_group, insn_must_be_last_in_group): Adjust.
 143
 144         * config/rs6000/40x.md (ppc403-integer, ppc403-compare): Adjust.
 145         * config/rs6000/440.md (ppc440-integer): Adjust.
 146         * config/rs6000/476.md (ppc476-simple-integer, ppc476-compare):
 147         Adjust.
 148         * config/rs6000/601.md (ppc601-integer, ppc601-compare): Adjust.
 149         * config/rs6000/603.md (ppc603-integer, ppc603-compare): Adjust.
 150         * config/rs6000/6xx.md (ppc604-integer, ppc604-compare): Adjust.
 151         * config/rs6000/7450.md (ppc7450-integer, ppc7450-compare):
 152         Adjust.
 153         * config/rs6000/7xx.md (ppc750-integer, ppc750-compare): Adjust.
 154         * config/rs6000/8540.md (ppc8540_su): Adjust.
 155         * config/rs6000/cell.md (cell-integer, cell-fast-cmp,
 156         cell-cmp-microcoded): Adjust.
 157         * config/rs6000/e300c2c3.md (ppce300c3_cmp): Adjust.
 158         * config/rs6000/e500mc.md (e500mc_su): Adjust.
 159         * config/rs6000/e500mc64.md (e500mc64_su, e500mc64_su2,
 160         e500mc64_delayed): Adjust.
 161         * config/rs6000/e5500.md (e5500_sfx, e5500_delayed): Adjust.
 162         * config/rs6000/e6500.md (e6500_sfx, e6500_delayed): Adjust.
 163         * config/rs6000/mpc.md (mpccore-integer, mpccore-compare): Adjust.
 164         * config/rs6000/power4.md (power4-integer, power4-compare):
 165         Adjust.
 166         * config/rs6000/power5.md (power5-integer, power5-compare):
 167         Adjust.
 168         * config/rs6000/power6.md (power6-shift, power6-var-rotate,
 169         power6-delayed-compare, power6-var-delayed-compare): Adjust.
 170         * config/rs6000/power7.md (power7-integer, power7-compare):
 171         Adjust.
 172         * config/rs6000/power8.md (power8-1cyc, power8-compare): Adjust.
 173         Adjust comment.
 174         * config/rs6000/rs64.md (rs64a-integer, rs64a-compare): Adjust.
 175         * config/rs6000/titan.md (titan_fxu_shift_and_rotate): Adjust.
 176
 177 2014-05-23  Segher Boessenkool  <segher@kernel.crashing.org>
 178
 179         * config/rs6000/rs6000.md (type): Delete "idiv", "ldiv".  Add
 180         "div".
 181         (bits): New mode_attr.
 182         (idiv_ldiv): Delete mode_attr.
 183         (udiv<mode>3, *div<mode>3, div<div_extend>_<mode>): Adjust.
 184         * config/rs6000/rs6000.c (rs6000_adjust_cost, is_cracked_insn,
 185         rs6000_adjust_priority, is_nonpipeline_insn,
 186         insn_must_be_first_in_group, insn_must_be_last_in_group): Adjust.
 187
 188         * config/rs6000/40x.md (ppc403-idiv): Adjust.
 189         * config/rs6000/440.md (ppc440-idiv): Adjust.
 190         * config/rs6000/476.md (ppc476-idiv): Adjust.
 191         * config/rs6000/601.md (ppc601-idiv): Adjust.
 192         * config/rs6000/603.md (ppc603-idiv): Adjust.
 193         * config/rs6000/6xx.md (ppc604-idiv, ppc620-idiv, ppc630-idiv,
 194         ppc620-ldiv): Adjust.
 195         * config/rs6000/7450.md (ppc7450-idiv): Adjust.
 196         * config/rs6000/7xx.md (ppc750-idiv): Adjust.
 197         * config/rs6000/8540.md (ppc8540_divide): Adjust.
 198         * config/rs6000/a2.md (ppca2-idiv, ppca2-ldiv): Adjust.
 199         * config/rs6000/cell.md (cell-idiv, cell-ldiv): Adjust.
 200         * config/rs6000/e300c2c3.md (ppce300c3_divide): Adjust.
 201         * config/rs6000/e500mc.md (e500mc_divide): Adjust.
 202         * config/rs6000/e500mc64.md (e500mc64_divide): Adjust.
 203         * config/rs6000/e5500.md (e5500_divide, e5500_divide_d): Adjust.
 204         * config/rs6000/e6500.md (e6500_divide, e6500_divide_d): Adjust.
 205         * config/rs6000/mpc.md (mpccore-idiv): Adjust.
 206         * config/rs6000/power4.md (power4-idiv, power4-ldiv): Adjust.
 207         * config/rs6000/power5.md (power5-idiv, power5-ldiv): Adjust.
 208         * config/rs6000/power6.md (power6-idiv, power6-ldiv): Adjust.
 209         * config/rs6000/power7.md (power7-idiv, power7-ldiv): Adjust.
 210         * config/rs6000/power8.md (power8-idiv, power8-ldiv): Adjust.
 211         * config/rs6000/rs64.md (rs64a-idiv, rs64a-ldiv): Adjust.
 212         * config/rs6000/titan.md (titan_fxu_div): Adjust.
 213
 214 2014-05-23  Segher Boessenkool  <segher@kernel.crashing.org>
 215
 216         * config/rs6000/rs6000.md (type): Delete "insert_word",
 217         "insert_dword".  Add "insert".
 218         (size): Update comment.
 219         * config/rs6000/rs6000.c (rs6000_adjust_cost, is_cracked_insn,
 220         insn_must_be_first_in_group): Adjust.
 221         (insvsi_internal, *insvsi_internal1, *insvsi_internal2,
 222         *insvsi_internal3, *insvsi_internal4, *insvsi_internal5,
 223         *insvsi_internal6, insvdi_internal): Adjust.
 224
 225         * config/rs6000/40x.md (ppc403-integer): Adjust.
 226         * config/rs6000/440.md (ppc440-integer): Adjust.
 227         * config/rs6000/476.md (ppc476-simple-integer): Adjust.
 228         * config/rs6000/601.md (ppc601-integer): Adjust.
 229         * config/rs6000/603.md (ppc603-integer): Adjust.
 230         * config/rs6000/6xx.md (ppc604-integer): Adjust.
 231         * config/rs6000/7450.md (ppc7450-integer): Adjust.
 232         * config/rs6000/7xx.md (ppc750-integer): Adjust.
 233         * config/rs6000/8540.md (ppc8540_su): Adjust.
 234         * config/rs6000/cell.md (cell-integer, cell-insert): Adjust.
 235         * config/rs6000/e300c2c3.md (ppce300c3_iu): Adjust.
 236         * config/rs6000/e500mc.md (e500mc_su): Adjust.
 237         * config/rs6000/e500mc64.md (e500mc64_su): Adjust.
 238         * config/rs6000/e5500.md (e5500_sfx): Adjust.
 239         * config/rs6000/e6500.md (e6500_sfx): Adjust.
 240         * config/rs6000/mpc.md (mpccore-integer): Adjust.
 241         * config/rs6000/power4.md (power4-integer, power4-insert): Adjust.
 242         * config/rs6000/power5.md (power5-integer, power5-insert): Adjust.
 243         * config/rs6000/power6.md (power6-insert, power6-insert-dword):
 244         Adjust.
 245         * config/rs6000/power7.md (power7-integer): Adjust.
 246         * config/rs6000/power8.md (power8-1cyc): Adjust.
 247         * config/rs6000/rs64.md (rs64a-integer): Adjust.
 248         * config/rs6000/titan.md (titan_fxu_shift_and_rotate): Adjust.
 249
 250 2014-05-23  Segher Boessenkool  <segher@kernel.crashing.org>
 251
 252         * config/rs6000/rs6000.md (type): Add "mul".  Delete "imul",
 253         "imul2", "imul3", "lmul", "imul_compare", "lmul_compare".
 254         (size): New attribute.
 255         (dot): New attribute.
 256         (cell_micro): Adjust.
 257         (mulsi3, *mulsi3_internal1, *mulsi3_internal2, mulsidi3,
 258         umulsidi3, smulsi3_highpart, umulsi3_highpart, muldi3,
 259         *muldi3_internal1, *muldi3_internal2, smuldi3_highpart,
 260         umuldi3_highpart): Adjust.
 261         * config/rs6000/rs6000.c (rs6000_adjust_cost, is_cracked_insn,
 262         rs6000_adjust_priority, is_nonpipeline_insn,
 263         insn_must_be_first_in_group, insn_must_be_last_in_group): Adjust.
 264
 265         * config/rs6000/40x.md (ppc403-imul, ppc405-imul, ppc405-imul2,
 266         ppc405-imul3): Adjust.
 267         * config/rs6000/440.md (ppc440-imul, ppc440-imul2): Adjust.
 268         * config/rs6000/476.md (ppc476-imul): Adjust.
 269         * config/rs6000/601.md (ppc601-imul): Adjust.
 270         * config/rs6000/603.md (ppc603-imul, ppc603-imul2): Adjust.
 271         * config/rs6000/6xx.md (ppc604-imul, ppc604e-imul, ppc620-imul,
 272         ppc620-imul2, ppc620-imul3, ppc620-lmul): Adjust.
 273         * config/rs6000/7450.md (ppc7450-imul, ppc7450-imul2): Adjust.
 274         * config/rs6000/7xx.md (ppc750-imul, ppc750-imul2, ppc750-imul3):
 275         Adjust.
 276         * config/rs6000/8540.md (ppc8540_multiply): Adjust.
 277         * config/rs6000/a2.md (ppca2-imul, ppca2-lmul): Adjust.
 278         * config/rs6000/cell.md (cell-lmul, cell-lmul-cmp, cell-imul23,
 279         cell-imul): Adjust.
 280         * config/rs6000/e300c2c3.md (ppce300c3_multiply): Adjust.
 281         * config/rs6000/e500mc.md (e500mc_multiply): Adjust.
 282         * config/rs6000/e500mc64.md (e500mc64_multiply): Adjust.
 283         * config/rs6000/e5500.md (e5500_multiply, e5500_multiply_i): Adjust.
 284         * config/rs6000/e6500.md (e6500_multiply, e6500_multiply_i): Adjust.
 285         * config/rs6000/mpc.md (mpccore-imul): Adjust.
 286         * config/rs6000/power4.md (power4-lmul-cmp, power4-imul-cmp,
 287         power4-lmul, power4-imul, power4-imul3): Adjust.
 288         * config/rs6000/power5.md (power5-lmul-cmp, power5-imul-cmp,
 289         power5-lmul, power5-imul, power5-imul3): Adjust.
 290         * config/rs6000/power6.md (power6-lmul-cmp, power6-imul-cmp,
 291         power6-lmul, power6-imul, power6-imul3): Adjust.
 292         * config/rs6000/power7.md (power7-mul, power7-mul-compare): Adjust.
 293         * config/rs6000/power8.md (power8-mul, power8-mul-compare): Adjust.
 294
 295         * config/rs6000/rs64.md (rs64a-imul, rs64a-imul2, rs64a-imul3,
 296         rs64a-lmul): Adjust.
 297         * config/rs6000/titan.md (titan_imul): Adjust.
 298
 299 2014-05-23  Segher Boessenkool  <segher@kernel.crashing.org>
 300
 301         * config/rs6000/rs6000.md (type): Add new value "halfmul".
 302         (*macchwc, *macchw, *macchwuc, *macchwu, *machhwc, *machhw,
 303         *machhwuc, *machhwu, *maclhwc, *maclhw, *maclhwuc, *maclhwu,
 304         *nmacchwc, *nmacchw, *nmachhwc, *nmachhw, *nmaclhwc, *nmaclhw,
 305         *mulchwc, *mulchw, *mulchwuc, *mulchwu, *mulhhwc, *mulhhw,
 306         *mulhhwuc, *mulhhwu, *mullhwc, *mullhw, *mullhwuc, *mullhwu):
 307         Use it.
 308         * config/rs6000/40x.md (ppc405-imul3): Add type halfmul.
 309         * config/rs6000/440.md (ppc440-imul2): Add type halfmul.
 310         * config/rs6000/476.md (ppc476-imul): Add type halfmul.
 311         * config/rs6000/titan.md: Delete nonsensical comment.
 312         (titan_imul): Add type imul3.
 313         (titan_mulhw): Remove type imul3; add type halfmul.
 314
 315 2014-05-23  Segher Boessenkool  <segher@kernel.crashing.org>
 316
 317         * config/rs6000/rs6000.md (type): Reorder, reformat.
 318
 319 2014-05-23  Martin Jambor  <mjambor@suse.cz>
 320
 321         PR tree-optimization/53787
 322         * params.def (PARAM_IPA_MAX_AA_STEPS): New param.
 323         * ipa-prop.h (ipa_node_params): Rename uses_analysis_done to
 324         analysis_done, update all uses.
 325         * ipa-prop.c: Include domwalk.h
 326         (param_analysis_info): Removed.
 327         (param_aa_status): New type.
 328         (ipa_bb_info): Likewise.
 329         (func_body_info): Likewise.
 330         (ipa_get_bb_info): New function.
 331         (aa_overwalked): Likewise.
 332         (find_dominating_aa_status): Likewise.
 333         (parm_bb_aa_status_for_bb): Likewise.
 334         (parm_preserved_before_stmt_p): Changed to use new param AA info.
 335         (load_from_unmodified_param): Accept func_body_info as a parameter
 336         instead of parms_ainfo.
 337         (parm_ref_data_preserved_p): Changed to use new param AA info.
 338         (parm_ref_data_pass_through_p): Likewise.
 339         (ipa_load_from_parm_agg_1): Likewise.  Update callers.
 340         (compute_complex_assign_jump_func): Changed to use new param AA info.
 341         (compute_complex_ancestor_jump_func): Likewise.
 342         (ipa_compute_jump_functions_for_edge): Likewise.
 343         (ipa_compute_jump_functions): Removed.
 344         (ipa_compute_jump_functions_for_bb): New function.
 345         (ipa_analyze_indirect_call_uses): Likewise, moved variable
 346         declarations down.
 347         (ipa_analyze_virtual_call_uses): Accept func_body_info instead of node
 348         and info, moved variable declarations down.
 349         (ipa_analyze_call_uses): Accept and pass on func_body_info instead of
 350         node and info.
 351         (ipa_analyze_stmt_uses): Likewise.
 352         (ipa_analyze_params_uses): Removed.
 353         (ipa_analyze_params_uses_in_bb): New function.
 354         (ipa_analyze_controlled_uses): Likewise.
 355         (free_ipa_bb_info): Likewise.
 356         (analysis_dom_walker): New class.
 357         (ipa_analyze_node): Handle node-specific forbidden analysis,
 358         initialize and free func_body_info, use dominator walker.
 359         (ipcp_modif_dom_walker): New class.
 360         (ipcp_transform_function): Create and free func_body_info, use
 361         ipcp_modif_dom_walker, moved a lot of functionality there.
 362
 363 2014-05-23  Marek Polacek  <polacek@redhat.com>
 364             Jakub Jelinek  <jakub@redhat.com>
 365
 366         * builtins.def: Change SANITIZE_FLOAT_DIVIDE to SANITIZE_NONDEFAULT.
 367         * gcc.c (sanitize_spec_function): Likewise.
 368         * convert.c (convert_to_integer): Include "ubsan.h".  Add
 369         floating-point to integer instrumentation.
 370         * doc/invoke.texi: Document -fsanitize=float-cast-overflow.
 371         * flag-types.h (enum sanitize_code): Add SANITIZE_FLOAT_CAST and
 372         SANITIZE_NONDEFAULT.
 373         * opts.c (common_handle_option): Handle -fsanitize=float-cast-overflow.
 374         * sanitizer.def (BUILT_IN_UBSAN_HANDLE_FLOAT_CAST_OVERFLOW,
 375         BUILT_IN_UBSAN_HANDLE_FLOAT_CAST_OVERFLOW_ABORT): Add.
 376         * ubsan.c: Include "realmpfr.h" and "dfp.h".
 377         (get_ubsan_type_info_for_type): Handle REAL_TYPEs.
 378         (ubsan_type_descriptor): Set tkind to 0xffff for types other than
 379         float/double/long double.
 380         (ubsan_instrument_float_cast): New function.
 381         * ubsan.h (ubsan_instrument_float_cast): Declare.
 382
 383 2014-05-23  Jiong Wang   <jiong.wang@arm.com>
 384
 385         * config/aarch64/predicates.md (aarch64_call_insn_operand): New
 386         predicate.
 387         * config/aarch64/constraints.md ("Ucs", "Usf"):  New constraints.
 388         * config/aarch64/aarch64.md (*sibcall_insn, *sibcall_value_insn):
 389         Adjust for tailcalling through registers.
 390         * config/aarch64/aarch64.h (enum reg_class): New caller save
 391         register class.
 392         (REG_CLASS_NAMES): Likewise.
 393         (REG_CLASS_CONTENTS): Likewise.
 394         * config/aarch64/aarch64.c (aarch64_function_ok_for_sibcall):
 395         Allow tailcalling without decls.
 396
 397 2014-05-23  Thomas Schwinge  <thomas@codesourcery.com>
 398
 399         * gimplify.c (omp_notice_variable) <case OMP_CLAUSE_DEFAULT_NONE>:
 400         Rewrite check for ORT_PARALLEL and ORT_COMBINED_PARALLEL.
 401
 402         * omp-low.c (expand_omp_for_static_chunk): Rename variable si to
 403         gsi, and variables v_* to v*.
 404
 405 2014-05-23  Eric Botcazou  <ebotcazou@adacore.com>
 406
 407         * varasm.c (output_constructor_bitfield): Fix thinkos in latest change.
 408
 409 2014-05-23  Thomas Schwinge  <thomas@codesourcery.com>
 410
 411         * gimple.h (enum gf_mask): Add and use GF_OMP_FOR_SIMD.
 412         * omp-low.c: Update accordingly.
 413
 414         * gimple.h (enum gf_mask): Rewrite "<< 0" shift expressions used
 415         for GF_OMP_FOR_KIND_MASK, GF_OMP_FOR_KIND_FOR,
 416         GF_OMP_FOR_KIND_DISTRIBUTE, GF_OMP_FOR_KIND_SIMD,
 417         GF_OMP_FOR_KIND_CILKSIMD, GF_OMP_TARGET_KIND_MASK,
 418         GF_OMP_TARGET_KIND_REGION, GF_OMP_TARGET_KIND_DATA,
 419         GF_OMP_TARGET_KIND_UPDATE.
 420
 421         * gimplify.c (omp_notice_variable) <case OMP_CLAUSE_DEFAULT_NONE>:
 422         Explicitly enumerate the expected region types.
 423
 424 2014-05-23  Paul Eggert  <eggert@cs.ucla.edu>
 425
 426         PR other/56955
 427         * doc/extend.texi (Function Attributes): Fix  __attribute__ ((malloc))
 428         documentation; the old documentation didn't clearly state the
 429         constraints on the contents of the pointed-to storage.
 430
 431 2014-05-23  Maxim Kuvyrkov  <maxim.kuvyrkov@linaro.org>
 432
 433         Fix bootstrap error on ia64
 434         * config/ia64/ia64.c (ia64_first_cycle_multipass_dfa_lookahead_guard):
 435         Return default value.
 436
 437 2014-05-23  Thomas Preud'homme  <thomas.preudhomme@arm.com>
 438
 439         PR tree-optimization/54733
 440         * tree-ssa-math-opts.c (nop_stats): New "bswap_stats" structure.
 441         (CMPNOP): Define.
 442         (find_bswap_or_nop_load): New.
 443         (find_bswap_1): Renamed to ...
 444         (find_bswap_or_nop_1): This. Also add support for memory source.
 445         (find_bswap): Renamed to ...
 446         (find_bswap_or_nop): This. Also add support for memory source and
 447         detection of bitwise operations equivalent to load in host endianness.
 448         (execute_optimize_bswap): Likewise. Also move its leading comment back
 449         in place and split statement transformation into ...
 450         (bswap_replace): This.
 451
 452 2014-05-22  Vladimir Makarov  <vmakarov@redhat.com>
 453
 454         PR rtl-optimization/61215
 455         * lra-elelimination.c (lra_eliminate_regs_1): Don't use
 456         simplify_gen_subreg until final substitution.
 457
 458 2014-05-23  Alan Modra  <amodra@gmail.com>
 459
 460         PR target/61231
 461         * config/rs6000/rs6000.c (mem_operand_gpr): Handle SImode.
 462         * config/rs6000/rs6000.md (extendsidi2_lfiwax, extendsidi2_nocell):
 463         Use "Y" constraint rather than "m".
 464
 465 2014-05-23  Kugan Vivekanandarajah  <kuganv@linaro.org>
 466
 467         * config/aarch64/aarch64.c (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): New
 468         define.
 469         * config/aarch64/aarch64-protos.h (aarch64_atomic_assign_expand_fenv):
 470         New function declaration.
 471         * config/aarch64/aarch64-builtins.c (aarch64_builtins) : Add
 472         AARCH64_BUILTIN_GET_FPCR, AARCH64_BUILTIN_SET_FPCR.
 473         AARCH64_BUILTIN_GET_FPSR and AARCH64_BUILTIN_SET_FPSR.
 474         (aarch64_init_builtins) : Initialize builtins
 475         __builtins_aarch64_set_fpcr, __builtins_aarch64_get_fpcr.
 476         __builtins_aarch64_set_fpsr and __builtins_aarch64_get_fpsr.
 477         (aarch64_expand_builtin) : Expand builtins __builtins_aarch64_set_fpcr
 478         __builtins_aarch64_get_fpcr, __builtins_aarch64_get_fpsr,
 479         and __builtins_aarch64_set_fpsr.
 480         (aarch64_atomic_assign_expand_fenv): New function.
 481         * config/aarch64/aarch64.md (set_fpcr): New pattern.
 482         (get_fpcr) : Likewise.
 483         (set_fpsr) : Likewise.
 484         (get_fpsr) : Likewise.
 485         (unspecv): Add UNSPECV_GET_FPCR and UNSPECV_SET_FPCR, UNSPECV_GET_FPSR
 486          and UNSPECV_SET_FPSR.
 487         * doc/extend.texi (AARCH64 Built-in Functions) : Document
 488         __builtins_aarch64_set_fpcr, __builtins_aarch64_get_fpcr.
 489         __builtins_aarch64_set_fpsr and __builtins_aarch64_get_fpsr.
 490
 491 2014-05-22  Vladimir Makarov  <vmakarov@redhat.com>
 492
 493         PR rtl-optimization/60969
 494         * ira-costs.c (record_reg_classes): Process NO_REGS for matching
 495         constraints.  Set up mem cost for NO_REGS case.
 496
 497 2014-05-22  Thomas Schwinge  <thomas@codesourcery.com>
 498
 499         * builtin-types.def: Simplify examples for DEF_FUNCTION_TYPE_*.
 500
 501 2012-05-22  Bernd Schmidt  <bernds@codesourcery.com>
 502
 503         * config/darwin.c: Include "lto-section-names.h".
 504         (LTO_SEGMENT_NAME): Don't define.
 505         * config/i386/winnt.c: Include "lto-section-names.h".
 506         * lto-streamer.c: Include "lto-section-names.h".
 507         * lto-streamer.h (LTO_SECTION_NAME_PREFIX): Don't define.
 508         * lto-wrapper.c: Include "lto-section-names.h".
 509         (LTO_SECTION_NAME_PREFIX): Don't define.
 510         * lto-section-names.h: New file.
 511         * cgraphunit.c: Include "lto-section-names.h".
 512
 513 2014-05-22  Peter Bergner  <bergner@vnet.ibm.com>
 514
 515         * config/rs6000/htm.md (ttest): Use correct shift value to get CR0.
 516
 517 2014-05-22  Richard Earnshaw  <rearnsha@arm.com>
 518
 519         PR target/61208
 520         * arm.md (arm_cmpdi_unsigned): Fix length calculation for Thumb2.
 521
 522 2014-05-22  Nick Clifton  <nickc@redhat.com>
 523
 524         * config/msp430/msp430.h (ASM_SPEC): Add spaces after inserted
 525         options.
 526
 527 2014-05-22  Eric Botcazou  <ebotcazou@adacore.com>
 528
 529         * tree-ssa-forwprop.c (associate_plusminus): Extend (T)(P + A) - (T)P
 530         -> (T)A transformation to integer types.
 531
 532 2014-05-22  Teresa Johnson  <tejohnson@google.com>
 533
 534         * gcov-io.c (gcov_position): Use gcov_nonruntime_assert.
 535         (gcov_is_error): Remove gcc_assert from IN_LIBGCOV code.
 536         (gcov_rewrite): Use gcov_nonruntime_assert.
 537         (gcov_open): Ditto.
 538         (gcov_write_words): Ditto.
 539         (gcov_write_length): Ditto.
 540         (gcov_read_words): Use gcov_nonruntime_assert, and remove
 541         gcc_assert from IN_LIBGCOV code.
 542         (gcov_read_summary): Use gcov_error to flag profile corruption.
 543         (gcov_sync): Use gcov_nonruntime_assert.
 544         (gcov_seek): Remove gcc_assert from IN_LIBGCOV code.
 545         (gcov_histo_index): Use gcov_nonruntime_assert.
 546         (static void gcov_histogram_merge): Ditto.
 547         (compute_working_sets): Ditto.
 548         * gcov-io.h (gcov_nonruntime_assert): Define.
 549         (gcov_error): Define for !IN_LIBGCOV
 550
 551 2014-05-22  Richard Biener  <rguenther@suse.de>
 552
 553         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Handle
 554         BUILT_IN_REALLOC like BUILT_IN_STRDUP.
 555         (call_may_clobber_ref_p_1): Handle BUILT_IN_REALLOC as allocation
 556         and deallocation site.
 557         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
 558         Handle BUILT_IN_REALLOC similar to BUILT_IN_STRDUP with also
 559         passing through the incoming points-to set.
 560         (handle_lhs_call): Use flags argument instead of recomputing it.
 561         (find_func_aliases_for_call): Call handle_lhs_call with proper
 562         call return flags.
 563
 564 2014-05-22  Jakub Jelinek  <jakub@redhat.com>
 565
 566         * tree-streamer-in.c (unpack_ts_real_cst_value_fields): Make sure
 567         all padding bits in REAL_VALUE_TYPE are cleared.
 568
 569 2014-05-22  Maxim Kuvyrkov  <maxim.kuvyrkov@linaro.org>
 570
 571         Cleanup and improve multipass_dfa_lookahead_guard
 572         * config/i386/i386.c (core2i7_first_cycle_multipass_filter_ready_try,)
 573         (core2i7_first_cycle_multipass_begin,)
 574         (core2i7_first_cycle_multipass_issue,)
 575         (core2i7_first_cycle_multipass_backtrack): Update signature.
 576         * config/ia64/ia64.c
 577         (ia64_first_cycle_multipass_dfa_lookahead_guard_spec): Remove.
 578         (ia64_first_cycle_multipass_dfa_lookahead_guard): Update signature.
 579         (TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD_SPEC): Remove
 580         hook definition.
 581         (ia64_first_cycle_multipass_dfa_lookahead_guard): Merge logic from
 582         ia64_first_cycle_multipass_dfa_lookahead_guard_spec.  Update return
 583         values.
 584         * config/rs6000/rs6000.c (rs6000_use_sched_lookahead_guard): Update
 585         return values.
 586         * doc/tm.texi: Regenerate.
 587         * doc/tm.texi.in
 588         (TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD_SPEC): Remove.
 589         * haifa-sched.c (ready_try): Make signed to allow negative values.
 590         (rebug_ready_list_1): Update.
 591         (choose_ready): Simplify.
 592         (sched_extend_ready_list): Update.
 593
 594 2014-05-22  Maxim Kuvyrkov  <maxim.kuvyrkov@linaro.org>
 595
 596         Remove IA64 speculation tweaking flags
 597         * config/ia64/ia64.c (ia64_set_sched_flags): Delete handling of
 598         speculation tuning flags.
 599         (msched-prefer-non-data-spec-insns,)
 600         (msched-prefer-non-control-spec-insns): Obsolete options.
 601         * haifa-sched.c (choose_ready): Remove handling of
 602         PREFER_NON_CONTROL_SPEC and PREFER_NON_DATA_SPEC.
 603         * sched-int.h (enum SPEC_SCHED_FLAGS): Remove PREFER_NON_CONTROL_SPEC
 604         and PREFER_NON_DATA_SPEC.
 605         * sel-sched.c (process_spec_exprs): Remove handling of
 606         PREFER_NON_CONTROL_SPEC and PREFER_NON_DATA_SPEC.
 607
 608 2014-05-22  Maxim Kuvyrkov  <maxim.kuvyrkov@linaro.org>
 609
 610         Improve scheduling debug output
 611         * haifa-sched.c (debug_ready_list): Remove unnecessary prototype.
 612         (advance_one_cycle): Update.
 613         (schedule_insn, queue_to_ready): Add debug printouts.
 614         (debug_ready_list_1): New static function.
 615         (debug_ready_list): Update.
 616         (max_issue): Add debug printouts.
 617         (dump_insn_stream): New static function.
 618         (schedule_block): Use it.  Also better indent printouts.
 619
 620 2014-05-22  Maxim Kuvyrkov  <maxim.kuvyrkov@linaro.org>
 621
 622         Fix sched_insn debug counter
 623         * haifa-sched.c (schedule_insn): Update.
 624         (struct haifa_saved_data): Add nonscheduled_insns_begin.
 625         (save_backtrack_point, restore_backtrack_point): Update.
 626         (first_nonscheduled_insn): New static function.
 627         (queue_to_ready, choose_ready): Use it.
 628         (schedule_block): Init nonscheduled_insns_begin.
 629         (sched_emit_insn): Update.
 630
 631
 632 2014-05-22  Kugan Vivekanandarajah  <kuganv@linaro.org>
 633
 634         * config/aarch64/aarch64.c (aarch64_regno_regclass) : Change CORE_REGS
 635         to GENERAL_REGS.
 636         (aarch64_secondary_reload) : LikeWise.
 637         (aarch64_class_max_nregs) : Remove CORE_REGS.
 638         * config/aarch64/aarch64.h (enum reg_class) : Remove CORE_REGS.
 639         (REG_CLASS_NAMES) : Likewise.
 640         (REG_CLASS_CONTENTS) : LikeWise.
 641         (INDEX_REG_CLASS) : Change CORE_REGS to GENERAL_REGS.
 642
 643 2014-05-21  Guozhi Wei  <carrot@google.com>
 644
 645         PR target/61202
 646         * config/aarch64/arm_neon.h (vqdmulh_n_s16): Change the last operand's
 647         constraint.
 648         (vqdmulhq_n_s16): Likewise.
 649
 650 2014-05-21  Segher Boessenkool  <segher@kernel.crashing.org>
 651
 652         * config/rs6000/predicates.md (update_indexed_address_mem): Delete.
 653
 654 2014-05-21  Marek Polacek  <polacek@redhat.com>
 655
 656         PR sanitizer/61272
 657         * ubsan.c (is_ubsan_builtin_p): Turn assert into a condition.
 658
 659 2014-05-21  Martin Jambor  <mjambor@suse.cz>
 660
 661         * doc/invoke.texi (Optimize Options): Document parameters
 662         ipa-cp-eval-threshold, ipa-max-agg-items, ipa-cp-loop-hint-bonus and
 663         ipa-cp-array-index-hint-bonus.
 664
 665 2014-05-21  Mark Wielaard  <mjw@redhat.com>
 666
 667         PR debug/16063
 668         * dwarf2out.c (gen_enumeration_type_die): Add DW_AT_type if DWARF
 669         version >= 3 or not strict DWARF.
 670         * langhooks.h (struct lang_hooks_for_types): Add
 671         enum_underlying_base_type.
 672         * langhooks.c (lhd_enum_underlying_base_type): New function.
 673         * gcc/langhooks.h (struct lang_hooks_for_types): Add
 674         enum_underlying_base_type.
 675         * langhooks-def.h (lhd_enum_underlying_base_type): New declaration.
 676         (LANG_HOOKS_ENUM_UNDERLYING_BASE_TYPE): New define.
 677         (LANG_HOOKS_FOR_TYPES_INITIALIZER): Add new lang hook.
 678
 679 2014-05-21  Richard Biener  <rguenther@suse.de>
 680
 681         * doc/invoke.texi (-flto-partition=): Document one and
 682         none algorithms.
 683
 684 2014-05-21  John Marino  <gnugcc@marino.st>
 685
 686         * config.gcc (*-*-dragonfly*): New target.
 687         * configure.ac: Detect dl_iterate_phdr (*freebsd*, *dragonfly*).
 688         * configure: Regenerate.
 689         * config/dragonfly-stdint.h: New.
 690         * config/dragonfly.h: New.
 691         * config/dragonfly.opt: New.
 692         * config/i386/dragonfly.h: New.
 693         * ginclude/stddef.h: Detect _PTRDIFF_T_DECLARED for DragonFly.
 694
 695 2014-05-21  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
 696
 697         * tree.def (VOID_CST): New.
 698         * tree-core.h (TI_VOID): New.
 699         * tree.h (void_node): New.
 700         * tree.c (tree_node_structure_for_code, tree_code_size)
 701         (iterative_hash_expr): Handle VOID_CST.
 702         (build_common_tree_nodes): Initialize void_node.
 703
 704 2014-05-21  Bernd Schmidt  <bernds@codesourcery.com>
 705
 706         * reload1.c (remove_init_insns, will_delete_init_insn_p): New static
 707         functions.
 708         (reload, calculate_needs_all_insns, reload_as_needed): Use them.
 709
 710         * config/bfin/bfin.c (split_load_immediate): Use gen_int_mode in a few
 711         more places.
 712
 713         * cfgrtl.c (cfg_layout_initialize): Weaken assert to only trigger if
 714         flag_reorder_blocks_and_partition.
 715         * hw-doloop.c (reorg_loops): Avoid reordering if that flag is set.
 716
 717 2014-05-21  Oleg Endo  <olegendo@gcc.gnu.org>
 718
 719         PR target/54236
 720         * config/sh/sh.md (*addc_r_1): Rename to addc_t_r.  Remove empty
 721         constraints.
 722         (*addc_r_t): Add new insn_and_split.
 723
 724 2014-05-21  Jakub Jelinek  <jakub@redhat.com>
 725
 726         PR middle-end/61252
 727         * omp-low.c (handle_simd_reference): New function.
 728         (lower_rec_input_clauses): Use it.  Defer adding reference
 729         initialization even for reduction without placeholder if in simd,
 730         handle it properly later on.
 731
 732 2014-05-20  Jan Hubicka  <hubicka@ucw.cz>
 733
 734         PR tree-optimization/60899
 735         * gimple-fold.c (can_refer_decl_in_current_unit_p): Cleanup;
 736         assume all static symbols will have definition wile parsing and
 737         check the do have definition later in compilation; check that
 738         variable referring symbol will be output before concluding that
 739         reference is safe; be conservative for referring local statics;
 740         be more precise about when comdat is output in other partition.
 741
 742 2014-05-20  Jan Hubicka  <hubicka@ucw.cz>
 743
 744         PR bootstrap/60984
 745         * ipa-inline-transform.c (inline_call): Use add CALLEE_REMOVED parameter.
 746         * ipa-inline.c (inline_to_all_callers): If callee was removed; return.
 747         (ipa_inline): Loop inline_to_all_callers until no more aliases are removed.
 748
 749 2014-05-20  Jan Hubicka  <hubicka@ucw.cz>
 750
 751         * ipa.c (ipa_discover_readonly_nonaddressable_var): Fix dumping;
 752         set writeonly flag only for vars actually written to.
 753
 754 2014-05-20  Dehao Chen  <dehao@google.com>
 755
 756         * ipa-inline-transform.c (clone_inlined_nodes): Use min of edge count
 757         and callee count to get clone count.
 758         * tree-inline.c (expand_call_inline): Use callee count instead of bb
 759         count in copy_body.
 760
 761 2014-05-20  Richard Sandiford  <rdsandiford@googlemail.com>
 762
 763         PR rtl-optimization/61243
 764         * emit-rtl.c (emit_copy_of_insn_after): Copy CROSSING_JUMP_P.
 765
 766 2014-05-20  Xinliang David Li  <davidxl@google.com>
 767
 768         * cgraphunit.c (walk_polymorphic_call_targets): Add
 769         dbgcnt and fopt-info support.
 770         * ipa-prop.c (ipa_make_edge_direct_to_target): Ditto.
 771         * ipa-devirt.c (ipa_devirt): Ditto.
 772         * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
 773         Ditto.
 774         * ipa.c (walk_polymorphic_call_targets): Ditto.
 775         * gimple-fold.c (fold_gimple_assign): Ditto.
 776         (gimple_fold_call): Ditto.
 777         * dbgcnt.def: New counter.
 778
 779 2014-05-20  DJ Delorie  <dj@redhat.com>
 780
 781         * config/msp430/msp430.md (split): Don't allow subregs when
 782         splitting SImode adds.
 783         (andneghi): Fix subtraction logic.
 784         * config/msp430/predicates.md (msp430_nonsubreg_or_imm_operand): New.
 785
 786 2014-05-20  Jan Hubicka  <hubicka@ucw.cz>
 787
 788         * tree.h (DECL_ONE_ONLY): Return true only for externally visible
 789         symbols.
 790         * except.c (switch_to_exception_section, resolve_unique_section,
 791         get_named_text_section, default_function_rodata_section,
 792         align_variable, get_block_for_decl, default_section_type_flags):
 793         Use DECL_COMDAT_GROUP instead of DECL_ONE_ONLY.
 794         * symtab.c (symtab_add_to_same_comdat_group,
 795         symtab_make_decl_local, fixup_same_cpp_alias_visibility,
 796         symtab_nonoverwritable_alias, symtab_get_symbol_partitioning_class):
 797         Likewise.
 798         * cgraphclones.c (cgraph_create_virtual_clone): Likewise.
 799         * bb-reorder.c (pass_partition_blocks::gate): Likewise.
 800         * config/c6x/c6x.c (c6x_elf_unique_section): Likewise.
 801         (c6x_function_in_section_p): Likewise.
 802         * config/darwin.c (machopic_select_section): Likewise.
 803         * config/arm/arm.c (arm_function_in_section_p): Likewise.
 804         * config/mips/mips.c (mips_function_rodata_section): Likewise.
 805         * config/mep/mep.c (mep_select_section): LIkewise.
 806         * config/i386/i386.c (x86_64_elf_unique_section): Likewise.
 807
 808 2014-05-20  Eric Botcazou  <ebotcazou@adacore.com>
 809
 810         * tree-ssa-dom.c (hashable_expr_equal_p) <EXPR_CALL>: Also compare the
 811         EH region of calls to pure functions that can throw an exception.
 812         * tree-ssa-sccvn.c (vn_reference_eq): Remove duplicated test.
 813         (copy_reference_ops_from_call): Also copy the EH region of the call if
 814         it can throw an exception.
 815
 816 2014-05-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
 817
 818         * simplify-rtx.c (simplify_binary_operation_1): Optimize case of
 819         nested VEC_SELECTs that are inverses of each other.
 820
 821 2014-05-20  Richard Biener  <rguenther@suse.de>
 822
 823         * tree-ssa-sccvn.c (process_scc): Dump SCC here, when
 824         iterating,
 825         (extract_and_process_scc_for_name): not here.
 826         (cond_dom_walker::before_dom_children): Only process
 827         stmts that end the BB in interesting ways.
 828         (run_scc_vn): Mark param uses as visited.
 829
 830 2014-05-20  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
 831
 832         * config/arm/arm.md (arith_shiftsi): Do not predicate for
 833         arm_restrict_it.
 834
 835 2014-05-20  Nick Clifton  <nickc@redhat.com>
 836
 837         * config/msp430/msp430.c (TARGET_GIMPLIFY_VA_ARG_EXPR): Define.
 838         (msp430_gimplify_va_arg_expr): New function.
 839         (msp430_print_operand): Handle (CONST (ZERO_EXTRACT)).
 840
 841         * config/msp430/msp430.md (zero_extendpsisi2): Use + constraint on
 842         operand 0 in order to prevent confusion about the number of
 843         registers involved.
 844
 845 2014-05-20  Richard Biener  <rguenther@suse.de>
 846
 847         PR tree-optimization/61221
 848         * tree-ssa-pre.c (el_to_update): Remove.
 849         (eliminate_dom_walker::before_dom_children): Handle released
 850         VDEFs by value-numbering them to the associated VUSE.  Update
 851         stmt immediately for substituted call address.
 852         (eliminate): Remove delayed stmt updating code.
 853         * tree-ssa-sccvn.c (vuse_ssa_val): New function valueizing
 854         possibly late re-numbered vuses.
 855         (vn_reference_lookup_2): Adjust.
 856         (vn_reference_lookup_pieces): Likewise.
 857         (vn_reference_lookup): Likewise.
 858
 859 2014-05-20  Richard Biener  <rguenther@suse.de>
 860
 861         * config.gcc: Remove need_64bit_hwint.
 862         * configure.ac: Do not define NEED_64BIT_HOST_WIDE_INT.
 863         * hwint.h: Do not check NEED_64BIT_HOST_WIDE_INT but assume
 864         it to be true.
 865         * config.in: Regenerate.
 866         * configure: Likewise.
 867
 868 2014-05-19  David Wohlferd <dw@LimeGreenSocks.com>
 869
 870         * doc/extend.texi: Create Label Attributes section,
 871         move all label attributes into it and reference it.
 872
 873 2014-05-19  Richard Earnshaw  <rearnsha@arm.com>
 874
 875         * arm.c (thumb1_reorg): When scanning backwards skip anything
 876         that's not a proper insn.
 877
 878 2014-05-19  Richard Biener  <rguenther@suse.de>
 879
 880         PR tree-optimization/61221
 881         * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
 882         Do nothing for unreachable blocks.
 883         * tree-ssa-sccvn.c (cond_dom_walker::before_dom_children):
 884         Improve unreachability detection.
 885
 886 2014-05-19  Richard Biener  <rguenther@suse.de>
 887
 888         PR tree-optimization/61209
 889         * tree-ssa-sccvn.c (visit_phi): Avoid setting expr to VN_TOP.
 890
 891 2014-05-19  Nick Clifton  <nickc@redhat.com>
 892
 893         * except.c (init_eh): Fix computation of builtin setjmp buffer
 894         size to allow for targets where POINTER_SIZE > BITS_PER_WORD.
 895
 896 2014-05-19  Richard Biener  <rguenther@suse.de>
 897
 898         PR tree-optimization/61184
 899         * tree-vrp.c (is_negative_overflow_infinity): Use
 900         TREE_OVERFLOW_P and do that check first.
 901         (is_positive_overflow_infinity): Likewise.
 902         (is_overflow_infinity): Likewise.
 903         (vrp_operand_equal_p): Properly treat operands with
 904         differing overflow as not equal.
 905
 906 2014-05-19  Bernd Schmidt  <bernds@codesourcery.com>
 907
 908         * simplify-rtx.c (simplify_unary_operation_1): Use CONST_INT_P in
 909         shift simplification where it was intended.
 910
 911 2014-05-19  Christian Bruel  <christian.bruel@st.com>
 912
 913         PR target/61195
 914         * config/sh/sh.md (movsf_ie): Unset fp_mode for fmov.
 915
 916 2014-05-19  Richard Sandiford  <r.sandiford@uk.ibm.com>
 917
 918         PR target/61084
 919         * config/sparc/sparc.c (sparc_fold_builtin): Use widest_int rather
 920         than wide_int.
 921
 922 2014-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
 923
 924         * reg-notes.def (CROSSING_JUMP): Likewise.
 925         * rtl.h (rtx_def): Update comment for jump flag.
 926         (CROSSING_JUMP_P): Define.
 927         * cfgcleanup.c (try_forward_edges, try_optimize_cfg): Use it instead
 928         of a REG_CROSSING_JUMP note.
 929         * cfghooks.c (tidy_fallthru_edges): Likewise.
 930         * cfgrtl.c (fixup_partition_crossing, rtl_verify_edges): Likewise.
 931         * emit-rtl.c (try_split): Likewise.
 932         * haifa-sched.c (sched_create_recovery_edges): Likewise.
 933         * ifcvt.c (find_if_case_1, find_if_case_2): Likewise.
 934         * jump.c (redirect_jump_2): Likewise.
 935         * reorg.c (follow_jumps, fill_slots_from_thread): Likewise.
 936         (relax_delay_slots): Likewise.
 937         * config/arc/arc.md (jump_i, cbranchsi4_scratch, *bbit): Likewise.
 938         (bbit_di): Likewise.
 939         * config/arc/arc.c (arc_reorg, arc_can_follow_jump): Likewise.
 940         * config/sh/sh.md (jump_compact): Likewise.
 941         * bb-reorder.c (rotate_loop): Likewise.
 942         (pass_duplicate_computed_gotos::execute): Likewise.
 943         (add_reg_crossing_jump_notes): Rename to...
 944         (update_crossing_jump_flags): ...this.
 945         (pass_partition_blocks::execute): Update accordingly.
 946
 947 2014-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
 948
 949         * tree.h: Remove extraneous template <>.
 950
 951 2014-05-17  Jan Hubicka  <hubicka@ucw.cz>
 952
 953         * ipa.c (symtab_remove_unreachable_nodes): Remove
 954         symbol from comdat group if its body was eliminated.
 955         (comdat_can_be_unshared_p_1): Static symbols can always
 956         be privatized.
 957         * symtab.c (symtab_remove_from_same_comdat_group): Break out
 958         from ...
 959         (symtab_unregister_node): ... this one.
 960         (verify_symtab_base): More strict checking of comdats.
 961         * cgraph.h (symtab_remove_from_same_comdat_group): Declare.
 962
 963 2014-05-17  Jan Hubicka  <hubicka@ucw.cz>
 964
 965         * tree-pass.h (make_pass_ipa_comdats): New pass.
 966         * timevar.def (TV_IPA_COMDATS): New timevar.
 967         * passes.def (pass_ipa_comdats): Add.
 968         * Makefile.in (OBJS): Add ipa-comdats.o
 969         * ipa-comdats.c: New file.
 970
 971 2014-05-17  Jan Hubicka  <hubicka@ucw.cz>
 972
 973         * ipa.c (update_visibility_by_resolution_info): New function.
 974         (function_and_variable_visibility): Use it.
 975
 976 2014-05-17  Jan Hubicka  <hubicka@ucw.cz>
 977
 978         * cgraph.h (symtab_first_defined_symbol, symtab_next_defined_symbol):
 979         New functions.
 980         (FOR_EACH_DEFINED_SYMBOL): New macro.
 981         (varpool_first_static_initializer, varpool_next_static_initializer,
 982         varpool_first_defined_variable, varpool_next_defined_variable): Fix comments.
 983         (symtab_in_same_comdat_p): Correctly deal with inline functions.
 984
 985 2014-05-17  Trevor Saunders  <tsaunders@mozilla.com>
 986
 987         * ggc-page.c (ggc_handle_finalizers): Add comment.
 988
 989 2014-05-17  Trevor Saunders  <tsaunders@mozilla.com>
 990
 991         * ggc-common.c (ggc_internal_cleared_alloc): Adjust.
 992         * ggc-none.c (ggc_internal_alloc): Assert if a finalizer is passed.
 993         (ggc_internal_cleared_alloc): Likewise.
 994         * ggc-page.c (finalizer): New class.
 995         (vec_finalizer): Likewise.
 996         (globals::finalizers): New member.
 997         (globals::vec_finalizers): Likewise.
 998         (ggc_internal_alloc): Record the finalizer if any for the block being
 999         allocated.
1000         (ggc_handle_finalizers): New function.
1001         (ggc_collect): Call ggc_handle_finalizers.
1002         * ggc.h (ggc_internal_alloc): Add arguments to allow installing a
1003         finalizer.
1004         (ggc_internal_cleared_alloc): Likewise.
1005         (finalize): New function.
1006         (need_finalization_p): Likewise.
1007         (ggc_alloc): Install the type's destructor as the finalizer if it
1008         might do something.
1009         (ggc_cleared_alloc): Likewise.
1010         (ggc_vec_alloc): Likewise.
1011         (ggc_cleared_vec_alloc): Likewise.
1012
1013 2014-05-17  Trevor Saunders  <tsaunders@mozilla.com>
1014
1015         * ggc.h (ggc_alloc_cleared_simd_clone_stat): Remove function.
1016
1017 2014-05-17  Trevor Saunders  <tsaunders@mozilla.com>
1018
1019         * alias.c (record_alias_subset): Adjust.
1020         * bitmap.c (bitmap_element_allocate): Likewise.
1021         (bitmap_gc_alloc_stat): Likewise.
1022         * cfg.c (init_flow): Likewise.
1023         (alloc_block): Likewise.
1024         (unchecked_make_edge): Likewise.
1025         * cfgloop.c (alloc_loop): Likewise.
1026         (flow_loops_find): Likewise.
1027         (rescan_loop_exit): Likewise.
1028         * cfgrtl.c (init_rtl_bb_info): Likewise.
1029         * cgraph.c (insert_new_cgraph_node_version): Likewise.
1030         (cgraph_allocate_node): Likewise.
1031         (cgraph_create_edge_1): Likewise.
1032         (cgraph_allocate_init_indirect_info): Likewise.
1033         * cgraphclones.c (cgraph_clone_edge): Likewise.
1034         * cgraphunit.c (add_asm_node): Likewise.
1035         (init_lowered_empty_function): Likewise.
1036         * config/aarch64/aarch64.c (aarch64_init_machine_status): Likewise.
1037         * config/alpha/alpha.c (alpha_init_machine_status): Likewise.
1038         (alpha_use_linkage): Likewise.
1039         * config/arc/arc.c (arc_init_machine_status): Likewise.
1040         * config/arm/arm.c (arm_init_machine_status): Likewise.
1041         * config/avr/avr.c (avr_init_machine_status): Likewise.
1042         * config/bfin/bfin.c (bfin_init_machine_status): Likewise.
1043         * config/c6x/c6x.c (c6x_init_machine_status): Likewise.
1044         * config/cris/cris.c (cris_init_machine_status): Likewise.
1045         * config/darwin.c (machopic_indirection_name): Likewise.
1046         (darwin_build_constant_cfstring): Likewise.
1047         (darwin_enter_string_into_cfstring_table): Likewise.
1048         * config/epiphany/epiphany.c (epiphany_init_machine_status): Likewise.
1049         * config/frv/frv.c (frv_init_machine_status): Likewise.
1050         * config/i386/i386.c (get_dllimport_decl): Likewise.
1051         (ix86_init_machine_status): Likewise.
1052         (assign_386_stack_local): Likewise.
1053         * config/i386/winnt.c (i386_pe_record_external_function): Likewise.
1054         (i386_pe_maybe_record_exported_symbol): Likewise.
1055         (i386_pe_record_stub): Likewise.
1056         * config/ia64/ia64.c (ia64_init_machine_status): Likewise.
1057         * config/iq2000/iq2000.c (iq2000_init_machine_status): Likewise.
1058         * config/m32c/m32c.c (m32c_init_machine_status): Likewise.
1059         (m32c_note_pragma_address): Likewise.
1060         * config/mep/mep.c (mep_init_machine_status): Likewise.
1061         (mep_note_pragma_flag): Likewise.
1062         * config/mips/mips.c (mflip_mips16_use_mips16_p): Likewise.
1063         (mips16_local_alias): Likewise.
1064         (mips_init_machine_status): Likewise.
1065         * config/mmix/mmix.c (mmix_init_machine_status): Likewise.
1066         * config/moxie/moxie.c (moxie_init_machine_status): Likewise.
1067         * config/msp430/msp430.c (msp430_init_machine_status): Likewise.
1068         * config/nds32/nds32.c (nds32_init_machine_status): Likewise.
1069         * config/nios2/nios2.c (nios2_init_machine_status): Likewise.
1070         * config/pa/pa.c (pa_init_machine_status): Likewise.
1071         (pa_get_deferred_plabel): Likewise.
1072         * config/rl78/rl78.c (rl78_init_machine_status): Likewise.
1073         * config/rs6000/rs6000.c (builtin_function_type): Likewise.
1074         (rs6000_init_machine_status): Likewise.
1075         (output_toc): Likewise.
1076         * config/s390/s390.c (s390_init_machine_status): Likewise.
1077         * config/score/score.c (score_output_external): Likewise.
1078         * config/sparc/sparc.c (sparc_init_machine_status): Likewise.
1079         * config/spu/spu.c (spu_init_machine_status): Likewise.
1080         * config/tilegx/tilegx.c (tilegx_init_machine_status): Likewise.
1081         * config/tilepro/tilepro.c (tilepro_init_machine_status): Likewise.
1082         * config/xtensa/xtensa.c (xtensa_init_machine_status): Likewise.
1083         * coverage.c (coverage_end_function): Likewise.
1084         * dbxout.c (dbxout_init): Likewise.
1085         * doc/gty.texi: Don't mention variable_size attribute.
1086         * dwarf2cfi.c (new_cfi): Adjust.
1087         (new_cfi_row): Likewise.
1088         (copy_cfi_row): Likewise.
1089         (create_cie_data): Likewise.
1090         * dwarf2out.c (dwarf2out_alloc_current_fde): Likewise.
1091         (new_loc_descr): Likewise.
1092         (find_AT_string_in_table): Likewise.
1093         (add_addr_table_entry): Likewise.
1094         (new_die): Likewise.
1095         (add_var_loc_to_decl): Likewise.
1096         (clone_die): Likewise.
1097         (clone_as_declaration): Likewise.
1098         (break_out_comdat_types): Likewise.
1099         (new_loc_list): Likewise.
1100         (add_loc_descr_to_each): Likewise.
1101         (add_location_or_const_value_attribute): Likewise.
1102         (add_linkage_name): Likewise.
1103         (lookup_filename): Likewise.
1104         (dwarf2out_var_location): Likewise.
1105         (new_line_info_table): Likewise.
1106         (dwarf2out_init): Likewise.
1107         (mem_loc_descriptor): Likewise.
1108         (loc_descriptor): Likewise.
1109         (add_const_value_attribute): Likewise.
1110         (tree_add_const_value_attribute): Likewise.
1111         (comp_dir_string): Likewise.
1112         (dwarf2out_vms_debug_main_pointer): Likewise.
1113         (string_cst_pool_decl): Likewise.
1114         * emit-rtl.c (set_mem_attrs): Likewise.
1115         (get_reg_attrs): Likewise.
1116         (start_sequence): Likewise.
1117         (init_emit): Likewise.
1118         (init_emit_regs): Likewise.
1119         * except.c (init_eh_for_function): Likewise.
1120         (gen_eh_region): Likewise.
1121         (gen_eh_region_catch): Likewise.
1122         (gen_eh_landing_pad): Likewise.
1123         (add_call_site): Likewise.
1124         * function.c (add_frame_space): Likewise.
1125         (insert_temp_slot_address): Likewise.
1126         (assign_stack_temp_for_type): Likewise.
1127         (get_hard_reg_initial_val): Likewise.
1128         (allocate_struct_function): Likewise.
1129         (prepare_function_start): Likewise.
1130         (types_used_by_var_decl_insert): Likewise.
1131         * gengtype.c (variable_size_p): Remove function.
1132         (enum alloc_quantity): Remove enum.
1133         (write_typed_alloc_def): Remove function.
1134         (write_typed_struct_alloc_def): Likewise.
1135         (write_typed_typedef_alloc_def): Likewise.
1136         (write_typed_alloc_defns): Likewise.
1137         (main): Adjust.
1138         * ggc-common.c (ggc_cleared_alloc_htab_ignore_args): Adjust.
1139         (ggc_cleared_alloc_ptr_array_two_args): Likewise.
1140         * ggc.h (ggc_alloc): new function.
1141         (ggc_cleared_alloc): Likewise.
1142         (ggc_vec_alloc): Template on type of vector element, and remove
1143         element size argument.
1144         (ggc_cleared_vec_alloc): Likewise.
1145         * gimple.c (gimple_build_omp_for): Adjust.
1146         (gimple_copy): Likewise.
1147         * ipa-cp.c (get_replacement_map): Likewise.
1148         (find_aggregate_values_for_callers_subset): Likewise.
1149         (known_aggs_to_agg_replacement_list): Likewise.
1150         * ipa-devirt.c (get_odr_type): Likewise.
1151         * ipa-prop.c (ipa_node_duplication_hook): Likewise.
1152         (read_agg_replacement_chain): Likewise.
1153         * loop-iv.c (get_simple_loop_desc): Likewise.
1154         * lto-cgraph.c (input_node_opt_summary): Likewise.
1155         * lto-section-in.c (lto_new_in_decl_state): Likewise.
1156         * lto-streamer-in.c (lto_input_eh_catch_list): Likewise.
1157         (input_eh_region): Likewise.
1158         (input_eh_lp): Likewise.
1159         (input_cfg): Likewise.
1160         * optabs.c (set_optab_libfunc): Likewise.
1161         (init_tree_optimization_optabs): Likewise.
1162         (set_conv_libfunc): Likewise.
1163         * passes.c (do_per_function_toporder): Likewise.
1164         * rtl.h: Don't use variable_size gty attribute.
1165         * sese.c (if_region_set_false_region): Adjust.
1166         * stringpool.c (gt_pch_save_stringpool): Likewise.
1167         * target-globals.c (save_target_globals): Likewise.
1168         * toplev.c (general_init): Likewise.
1169         * trans-mem.c (record_tm_replacement): Likewise.
1170         (split_bb_make_tm_edge): Likewise.
1171         * tree-cfg.c (move_sese_region_to_fn): Likewise.
1172         * tree-data-ref.h (lambda_vector_new): Likewise.
1173         * tree-eh.c (add_stmt_to_eh_lp_fn): Likewise.
1174         * tree-iterator.c (tsi_link_before): Likewise.
1175         (tsi_link_after): Likewise.
1176         * tree-scalar-evolution.c (new_scev_info_str): Likewise.
1177         * tree-ssa-loop-niter.c (record_estimate): Likewise.
1178         * tree-ssa-operands.c (ssa_operand_alloc): Likewise.
1179         * tree-ssa-operands.h: Don't use variable_size gty attribute.
1180         * tree-ssa.c (init_tree_ssa): Adjust.
1181         * tree-ssanames.c (set_range_info): Likewise.
1182         (get_ptr_info): Likewise.
1183         (duplicate_ssa_name_ptr_info): Likewise.
1184         (duplicate_ssa_name_range_info): Likewise.
1185         * tree-streamer-in.c (unpack_ts_real_cst_value_fields): Likewise.
1186         (unpack_ts_fixed_cst_value_fields): Likewise.
1187         * tree.c (build_fixed): Likewise.
1188         (build_real): Likewise.
1189         (build_string): Likewise.
1190         (decl_priority_info): Likewise.
1191         (decl_debug_expr_insert): Likewise.
1192         (decl_value_expr_insert): Likewise.
1193         (decl_debug_args_insert): Likewise.
1194         (type_hash_add): Likewise.
1195         (build_omp_clause): Likewise.
1196         * ubsan.c (decl_for_type_insert): Likewise.
1197         * varasm.c (get_unnamed_section): Likewise.
1198         (get_noswitch_section): Likewise.
1199         (get_section): Likewise.
1200         (get_block_for_section): Likewise.
1201         (create_block_symbol): Likewise.
1202         (build_constant_desc): Likewise.
1203         (create_constant_pool): Likewise.
1204         (force_const_mem): Likewise.
1205         (record_tm_clone_pair): Likewise.
1206         * varpool.c (varpool_create_empty_node): Likewise.
1207
1208 2014-05-17  Trevor Saunders  <tsaunders@mozilla.com>
1209
1210         * dwarf2out.c (tree_add_const_value_attribute): Call
1211         ggc_internal_cleared_alloc instead of ggc_alloc_cleared_atomic.
1212         * gengtype.c (write_typed_alloc_def): Call ggc_internal_<x>alloc
1213         instead of ggc_internal_<x>alloc_stat.
1214         * ggc-common.c (ggc_internal_cleared_alloc): Drop _stat suffix.
1215         (ggc_realloc): Likewise.
1216         * ggc-none.c (ggc_internal_alloc): Likewise.
1217         (ggc_internal_cleared_alloc): Likewise.
1218         * ggc-page.c: Likewise.
1219         * ggc.h (ggc_internal_alloc_stat): Likewise.
1220         (ggc_internal_alloc): Remove macro.
1221         (ggc_internal_cleared_alloc_stat): Drop _stat suffix.
1222         (ggc_internal_cleared_alloc): Remove macro.
1223         (GGC_RESIZEVEC): Adjust.
1224         (ggc_resizevar): Remove macro.
1225         (ggc_internal_vec_alloc_stat): Drop _stat suffix.
1226         (ggc_internal_cleared_vec_alloc_stat): Likewise.
1227         (ggc_internal_vec_cleared_alloc): Remove macro.
1228         (ggc_alloc_atomic_stat): Drop _stat suffix.
1229         (ggc_alloc_atomic): Remove macro.
1230         (ggc_alloc_cleared_atomic): Remove macro.
1231         (ggc_alloc_string_stat): Drop _stat suffix.
1232         (ggc_alloc_string): Remove macro.
1233         (ggc_alloc_rtx_def_stat): Adjust.
1234         (ggc_alloc_tree_node_stat): Likewise.
1235         (ggc_alloc_cleared_tree_node_stat): Likewise.
1236         (ggc_alloc_cleared_gimple_statement_stat): Likewise.
1237         (ggc_alloc_cleared_simd_clone_stat): Likewise.
1238         * gimple.c (gimple_build_omp_for): Likewise.
1239         (gimple_copy): Likewise.
1240         * stringpool.c (ggc_alloc_string_stat): Drop _stat suffix.
1241         * toplev.c (realloc_for_line_map): Adjust.
1242         * tree-data-ref.h (lambda_vector_new): Likewise.
1243         * tree-phinodes.c (allocate_phi_node): Likewise.
1244         * tree.c (grow_tree_vec_stat): Likewise.
1245         * vec.h (va_gc::reserve): Adjust.
1246
1247 2014-05-17  Ajit Agarwal  <ajitkum@xilinx.com>
1248
1249         * config/microblaze/microblaze.c (break_handler): New Declaration.
1250         (microblaze_break_function_p,microblaze_is_break_handler): New.
1251         (compute_frame_size): Use microblaze_break_function_p.
1252         Add the test of break_handler.
1253         (microblaze_function_prologue) : Add the test of variable
1254         break_handler.  Check the fnname by BREAK_HANDLER_NAME.
1255         (microblaze_function_epilogue) : Add the test of break_handler.
1256         (microblaze_globalize_label) : Add the test of break_handler.
1257         Check the name by BREAK_HANDLER_NAME.
1258
1259         * config/microblaze/microblaze.h (BREAK_HANDLER_NAME): New macro
1260
1261         * config/microblaze/microblaze.md (*<optab>,<optab>_internal): Add
1262         microblaze_is_break_handler test.
1263         (call_internal1,call_value_intern): Use microblaze_break_function_p.
1264         Use SYMBOL_REF_DECL.
1265
1266         * config/microblaze/microblaze-protos.h
1267         (microblaze_break_function_p,microblaze_is_break_handler):
1268         New Declaration.
1269
1270         * doc/extend.texi (MicroBlaze break_handler Functions): Document
1271         new MicroBlaze break_handler functions.
1272
1273 2014-05-17  Uros Bizjak  <ubizjak@gmail.com>
1274
1275         * doc/extend.texi (Size of an asm): Move node text according
1276         to its @menu entry position.
1277
1278 2014-05-17  Marc Glisse  <marc.glisse@inria.fr>
1279
1280         PR tree-optimization/61140
1281         PR tree-optimization/61150
1282         PR tree-optimization/61197
1283         * tree-ssa-phiopt.c (value_replacement): Punt on multiple phis.
1284
1285 2014-05-17  Uros Bizjak  <ubizjak@gmail.com>
1286
1287         * doc/invoke.texi (free): Mention Alpha.  Also enabled at -Os.
1288
1289 2014-05-17  Richard Sandiford  <r.sandiford@uk.ibm.com>
1290
1291         * wide-int.cc: Only include longlong.h if W_TYPE_SIZE==32 or
1292         __SIZEOF_INT128__ is defined.
1293
1294 2014-05-17  Richard Sandiford  <rdsandiford@googlemail.com>
1295
1296         * config/rs6000/rs6000.c (rs6000_real_tls_symbol_ref_p): New function.
1297         (rs6000_delegitimize_address): Use it.
1298
1299 2014-05-17  Richard Sandiford  <rdsandiford@googlemail.com>
1300
1301         * emit-rtl.h (replace_equiv_address, replace_equiv_address_nv): Add an
1302         inplace argument.  Store the new address in the original MEM when true.
1303         * emit-rtl.c (change_address_1): Likewise.
1304         (adjust_address_1, adjust_automodify_address_1, offset_address):
1305         Update accordingly.
1306         * rtl.h (plus_constant): Add an inplace argument.
1307         * explow.c (plus_constant): Likewise.  Try to reuse the original PLUS
1308         when true.  Avoid generating (plus X (const_int 0)).
1309         * function.c (instantiate_virtual_regs_in_rtx): Adjust the PLUS
1310         in-place.  Pass true to plus_constant.
1311         (instantiate_virtual_regs_in_insn): Pass true to replace_equiv_address.
1312
1313 2014-05-16  Dehao Chen  <dehao@google.com>
1314
1315         * tree-cfg.c (gimple_merge_blocks): Updates bb count with max count.
1316
1317 2014-05-16  Oleg Endo  <olegendo@gcc.gnu.org>
1318
1319         PR target/54089
1320         * config/sh/predicates.md (negt_reg_shl31_operand): Match additional
1321         patterns.
1322         * config/sh/sh.md (*negt_msb): Merge SH2A and non-SH2A variants.
1323
1324 2014-05-16  Dehao Chen  <dehao@google.com>
1325
1326         * ira-int.h (REG_FREQ_FROM_EDGE_FREQ): Use
1327         optimize_function_for_size_p.
1328         * regs.h (REG_FREQ_FROM_BB): Likewise.
1329
1330 2014-05-16  Oleg Endo  <olegendo@gcc.gnu.org>
1331
1332         PR target/51244
1333         * config/sh/sh.c (sh_eval_treg_value): Handle t_reg_operand and
1334         negt_reg_operand cases.
1335         * config/sh/sh.md (*cset_zero): Likewise by using cbranch_treg_value
1336         predicate.
1337         * config/sh/predicates.md (cbranch_treg_value): Simplify.
1338
1339 2014-05-16  Oleg Endo  <olegendo@gcc.gnu.org>
1340
1341         * config/sh/sh.c (sh_option_override): Set branch cost to 2 for all
1342         target variants.
1343
1344 2014-05-16  David Malcolm  <dmalcolm@redhat.com>
1345
1346         Revert:
1347         2014-04-29  David Malcolm  <dmalcolm@redhat.com>
1348
1349         * tree-cfg.c (dump_function_to_file): Dump the return type of
1350         functions, in a line to itself before the function body, mimicking
1351         the layout of a C function.
1352
1353 2014-05-16  Dehao Chen  <dehao@google.com>
1354
1355         * cfghooks.c (make_forwarder_block): Use direct computation to
1356         get fall-through edge's count and frequency.
1357
1358 2014-05-16  Benno Schulenberg  <bensberg@justemail.net>
1359
1360         * config/arc/arc.c (arc_init): Fix typo in error message.
1361         * config/i386/i386.c (ix86_expand_builtin): Likewise.
1362         (split_stack_prologue_scratch_regno): Likewise.
1363         * fortran/check.c (gfc_check_fn_rc2008): Remove duplicate
1364         word from error message.
1365
1366 2014-05-16  Zhouyi Zhou <yizhouzhou@ict.ac.cn>
1367
1368         * ira-costs.c: Fix typo in comment.
1369
1370 2014-05-16  David Wohlferd <dw@LimeGreenSocks.com>
1371
1372         * doc/extend.texi: (Visibility Pragmas) Fix misplaced @xref
1373
1374 2014-05-16  Jan Hubicka  <hubicka@ucw.cz>
1375
1376         * varpool.c (dump_varpool_node): Dump write-only flag.
1377         * lto-cgraph.c (lto_output_varpool_node, input_varpool_node): Stream
1378         write-only flag.
1379         * tree-cfg.c (execute_fixup_cfg): Remove statements setting
1380         write-only variables.
1381         * ipa.c (process_references): New function.
1382         (set_readonly_bit): New function.
1383         (set_writeonly_bit): New function.
1384         (clear_addressable_bit): New function.
1385         (ipa_discover_readonly_nonaddressable_var): Mark write only variables;
1386         fix handling of aliases.
1387         * cgraph.h (struct varpool_node): Add writeonly flag.
1388
1389 2014-05-16  Vladimir Makarov  <vmakarov@redhat.com>
1390
1391         PR rtl-optimization/60969
1392         * ira-costs.c (record_reg_classes): Allow only memory for pseudo.
1393         Calculate costs for this case.
1394
1395 2014-05-16  Eric Botcazou  <ebotcazou@adacore.com>
1396
1397         * fold-const (fold_unary_loc) <NON_LVALUE_EXPR>: New case.
1398         <CASE_CONVERT>: Pass arg0 instead of op0 to fold_convert_const.
1399
1400 2014-05-16  Richard Biener  <rguenther@suse.de>
1401
1402         PR tree-optimization/61194
1403         * tree-vect-patterns.c (adjust_bool_pattern): Also handle
1404         bool patterns ending in a COND_EXPR.
1405
1406 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
1407
1408         * config/aarch64/aarch64.c (aarch64_rtx_mult_cost): Fix FNMUL case.
1409
1410 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
1411
1412         * config/aarch64/aarch64.c (aarch64_rtx_costs): Handle the case
1413         where we were unable to cost an RTX.
1414
1415 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
1416
1417         * config/aarch64/aarch64.c (aarch64_rtx_costs): Cost SYMBOL_REF,
1418         HIGH, LO_SUM.
1419
1420 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
1421             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
1422
1423         * config/aarch64/aarch64.c (aarch64_rtx_costs): Cost TRUNCATE.
1424
1425 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
1426             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
1427
1428         * config/aarch64/aarch64.c (aarch64_rtx_costs): Cost FMA,
1429         FLOAT_EXTEND, FLOAT_TRUNCATE, ABS, SMAX, and SMIN.
1430
1431 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
1432             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
1433
1434         * config/aarch64/aarch64.c (aarch64_rtx_costs): Cost comparison
1435         operators.
1436
1437 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
1438             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
1439
1440         * config/aarch64/aarch64.c (aarch64_rtx_costs): Improve costs for
1441         DIV/MOD.
1442
1443 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
1444             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
1445
1446         * config/aarch64/aarch64.c (aarch64_rtx_arith_op_extract_p): New.
1447         (aarch64_rtx_costs): Improve costs for SIGN/ZERO_EXTRACT.
1448
1449 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
1450             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
1451
1452         * config/aarch64/aarch64.c (aarch64_rtx_costs): Improve costs for
1453         rotates and shifts.
1454
1455 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
1456             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
1457
1458         * config/aarch64/aarch64.c (aarch64_rtx_costs): Cost
1459         ZERO_EXTEND and SIGN_EXTEND better.
1460
1461 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
1462             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
1463
1464         * config/aarch64/aarch64.c (aarch64_rtx_costs): Improve cost for
1465         logical operations.
1466
1467 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
1468             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
1469
1470         * config/aarch64/aarch64.c (aarch64_rtx_costs): Use address
1471         costs when costing loads and stores to memory.
1472
1473 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
1474             Philip Tomsich  <philipp.tomsich@theobroma-systems.com>
1475
1476         * config/aarch64/aarch64.c (aarch64_rtx_costs): Improve costing
1477         for SET RTX.
1478
1479 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
1480
1481         * config/aarch64/aarch64.c (aarch64_rtx_costs): Set default costs.
1482
1483 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
1484             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
1485
1486         * config/aarch64/aarch64.c (aarch64_strip_shift_or_extend): Rename
1487         to...
1488         (aarch64_strip_extend): ...this, don't strip shifts, check RTX is
1489         well formed.
1490         (aarch64_rtx_mult_cost): New.
1491         (aarch64_rtx_costs): Use it, refactor as appropriate.
1492
1493 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
1494             Philipp Tomsich  <philipp.tomsich@theobroma-systems.com>
1495
1496         * config/aarch64/aarch64.c (aarch64_build_constant): Conditionally
1497         emit instructions, return number of instructions which would
1498         be emitted.
1499         (aarch64_add_constant): Update call to aarch64_build_constant.
1500         (aarch64_output_mi_thunk): Likewise.
1501         (aarch64_rtx_costs): Estimate cost of a CONST_INT, cost of
1502         a CONST_DOUBLE.
1503
1504 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
1505
1506         * config/aarch64/aarch64.c (aarch64_rtx_costs_wrapper): New.
1507         (TARGET_RTX_COSTS): Call it.
1508
1509 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
1510
1511         * config/aarch64/aarch64.c (cortexa57_addrcost_table): New.
1512         (cortexa57_vector_cost): Likewise.
1513         (cortexa57_tunings): Use them.
1514
1515 2014-05-16  James Greenhalgh  <james.greenhalgh@arm.com>
1516
1517         * config/aarch64/aarch64-protos.h (scale_addr_mode_cost): New.
1518         (cpu_addrcost_table): Use it.
1519         * config/aarch64/aarch64.c (generic_addrcost_table): Initialize it.
1520         (aarch64_address_cost): Rewrite using aarch64_classify_address,
1521         move it.
1522
1523 2014-05-16  Richard Biener  <rguenther@suse.de>
1524
1525         * tree-ssa-sccvn.c: Include tree-cfg.h and domwalk.h.
1526         (set_ssa_val_to): Handle unexpected sets to VN_TOP.
1527         (visit_phi): Ignore edges marked as not executable.
1528         (class cond_dom_walker): New.
1529         (cond_dom_walker::before_dom_children): Value-number
1530         control statements and mark successor edges as not
1531         executable if possible.
1532         (run_scc_vn): First walk all control statements in
1533         dominator order, marking edges as not executable.
1534         * tree-inline.c (copy_edges_for_bb): Be not confused
1535         about random edge flags.
1536
1537 2014-05-16  Richard Biener  <rguenther@suse.de>
1538
1539         * tree-ssa-sccvn.c (visit_use): Also constant-fold calls.
1540
1541 2014-05-15  Peter Bergner  <bergner@vnet.ibm.com>
1542
1543         PR target/61193
1544         * config/rs6000/htmxlintrin.h (_HTM_TBEGIN_STARTED): New define.
1545         (__TM_simple_begin): Use it.
1546         (__TM_begin): Likewise.
1547
1548 2014-05-15  Martin Jambor  <mjambor@suse.cz>
1549
1550         PR ipa/61085
1551         * ipa-prop.c (update_indirect_edges_after_inlining): Check
1552         type_preserved flag when the indirect edge is polymorphic.
1553
1554 2014-05-15  Martin Jambor  <mjambor@suse.cz>
1555
1556         PR tree-optimization/61090
1557         * tree-sra.c (sra_modify_expr): Pass the current gsi to
1558         build_ref_for_model.
1559
1560 2014-05-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1561
1562         * config/arm/arm.c (arm_option_override): Use the SCHED_PRESSURE_MODEL
1563         enum name for PARAM_SCHED_PRESSURE_ALGORITHM.
1564
1565 2014-05-15  Jakub Jelinek  <jakub@redhat.com>
1566
1567         PR tree-optimization/61158
1568         * fold-const.c (fold_binary_loc): If X is zero-extended and
1569         shiftc >= prec, make sure zerobits is all ones instead of
1570         invoking undefined behavior.
1571
1572 2014-05-15  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
1573
1574         * regcprop.h: New file.
1575         * regcprop.c (skip_debug_insn_p): New decl.
1576         (replace_oldest_value_reg): Check skip_debug_insn_p.
1577         (copyprop_hardreg_forward_bb_without_debug_insn): New function.
1578         * shrink-wrap.c: Include regcprop.h.
1579         (prepare_shrink_wrap): Call
1580         copyprop_hardreg_forward_bb_without_debug_insn.
1581
1582 2014-05-15  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
1583
1584         * shrink-wrap.h: Update comment.
1585         * shrink-wrap.c: Update comment.
1586         (next_block_for_reg): Rename to live_edge_for_reg.
1587         (live_edge_for_reg): Allow live_edge->dest has two predecessors.
1588         (move_insn_for_shrink_wrap): Split live_edge.
1589         (prepre_shrink_wrap): One more parameter for move_insn_for_shrink_wrap.
1590
1591 2014-05-14  Eric Botcazou  <ebotcazou@adacore.com>
1592
1593         * config/sparc/sparc-protos.h (sparc_absnegfloat_split_legitimate):
1594         Delete.
1595         * config/sparc/sparc.c (sparc_absnegfloat_split_legitimate): Likewise.
1596         * config/sparc/sparc.md (fptype_ut699): New attribute.
1597         (in_branch_delay): Return false if -mfix-ut699 is specified and
1598         fptype_ut699 is set to single.
1599         (truncdfsf2): Add fptype_ut699 attribute.
1600         (fix_truncdfsi2): Likewise.
1601         (floatsisf2): Change fptype attribute.
1602         (fix_truncsfsi2): Likewise.
1603         (negtf2_notv9): Delete.
1604         (negtf2_v9): Likewise.
1605         (negtf2_hq): New instruction.
1606         (negtf2): New instruction and splitter.
1607         (negdf2_notv9): Rewrite.
1608         (abstf2_notv9): Delete.
1609         (abstf2_hq_v9): Likewise.
1610         (abstf2_v9): Likewise.
1611         (abstf2_hq): New instruction.
1612         (abstf2): New instruction and splitter.
1613         (absdf2_notv9): Rewrite.
1614
1615 2014-05-14  Cary Coutant  <ccoutant@google.com>
1616
1617         PR debug/61013
1618         * opts.c (common_handle_option): Don't special-case "-g".
1619         (set_debug_level): Default to at least level 2 with "-g".
1620
1621 2014-05-14  DJ Delorie  <dj@redhat.com>
1622
1623         * config/msp430/msp430.c (msp430_builtin): Add
1624         MSP430_BUILTIN_DELAY_CYCLES.
1625         (msp430_init_builtins): Register void __delay_cycles(long long).
1626         (msp430_builtin_decl): Add it.
1627         (cg_magic_constant): New.
1628         (msp430_expand_delay_cycles): New.
1629         (msp430_expand_builtin): Call it.
1630         (msp430_print_operand_raw): Change integer printing from "int" to
1631         HOST_WIDE_INT.
1632         * config/msp430/msp430.md (define_constants): Add delay_cycles tags.
1633         (delay_cycles_start): New.
1634         (delay_cycles_end): New.
1635         (delay_cycles_32): New.
1636         (delay_cycles_32x): New.
1637         (delay_cycles_16): New.
1638         (delay_cycles_16x): New.
1639         (delay_cycles_2): New.
1640         (delay_cycles_1): New.
1641         * doc/extend.texi: Document __delay_cycles().
1642
1643 2014-05-14  Sandra Loosemore  <sandra@codesourcery.com>
1644
1645         * config/nios2/nios2.md (nios2_cbranch): Fix paste-o in
1646         length attribute computation.
1647
1648 2014-05-14  Richard Sandiford  <rdsandiford@googlemail.com>
1649
1650         PR debug/61188
1651         * print-rtl.c (print_rtx): Suppress uids if flag_dump_unnumbered.
1652
1653 2014-05-14  Richard Sandiford  <r.sandiford@uk.ibm.com>
1654
1655         PR target/61084
1656         * config/sparc/sparc.md: Fix types of low and high in DI constant
1657         splitter.  Use gen_int_mode in some other splitters.
1658
1659 2014-05-14  Martin Jambor  <mjambor@suse.cz>
1660
1661         PR ipa/60897
1662         * ipa-prop.c (ipa_modify_formal_parameters): Reset DECL_LANG_SPECIFIC.
1663
1664 2014-05-14  James Norris  <jnorris@codesourcery.com>
1665
1666         * omp-low.c (expand_parallel_call): Remove shadow variable.
1667         (expand_omp_taskreg): Likewise.
1668
1669 2014-05-14  Ilya Tocar  <ilya.tocar@intel.com>
1670
1671         * common/config/i386/i386-common.c
1672         (OPTION_MASK_ISA_CLFLUSHOPT_SET): Define.
1673         (OPTION_MASK_ISA_XSAVES_SET): Ditto.
1674         (OPTION_MASK_ISA_XSAVEC_SET): Ditto.
1675         (OPTION_MASK_ISA_CLFLUSHOPT_UNSET): Ditto.
1676         (OPTION_MASK_ISA_XSAVES_UNSET): Ditto.
1677         (OPTION_MASK_ISA_XSAVEC_UNSET): Ditto.
1678         (ix86_handle_option): Handle OPT_mxsavec, OPT_mxsaves, OPT_mclflushopt.
1679         * config.gcc (i[34567]86-*-*): Add clflushoptintrin.h,
1680         xsavecintrin.h, xsavesintrin.h.
1681         (x86_64-*-*): Ditto.
1682         * config/i386/clflushoptintrin.h: New.
1683         * config/i386/xsavecintrin.h: Ditto.
1684         * config/i386/xsavesintrin.h: Ditto.
1685         * config/i386/cpuid.h (bit_CLFLUSHOPT): Define.
1686         (bit_XSAVES): Ditto.
1687         (bit_XSAVES): Ditto.
1688         * config/i386/driver-i386.c (host_detect_local_cpu): Handle
1689         -mclflushopt, -mxsavec, -mxsaves, -mno-xsaves, -mno-xsavec,
1690         -mno-clflushopt.
1691         * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1692         OPTION_MASK_ISA_CLFLUSHOPT, OPTION_MASK_ISA_XSAVEC,
1693         OPTION_MASK_ISA_XSAVES.
1694         * config/i386/i386.c (ix86_target_string): Handle -mclflushopt,
1695         -mxsavec, -mxsaves.
1696         (PTA_CLFLUSHOPT) Define.
1697         (PTA_XSAVEC): Ditto.
1698         (PTA_XSAVES): Ditto.
1699         (ix86_option_override_internal): Handle new options.
1700         (ix86_valid_target_attribute_inner_p): Ditto.
1701         (ix86_builtins): Add IX86_BUILTIN_XSAVEC, IX86_BUILTIN_XSAVEC64,
1702         IX86_BUILTIN_XSAVES, IX86_BUILTIN_XRSTORS, IX86_BUILTIN_XSAVES64,
1703         IX86_BUILTIN_XRSTORS64, IX86_BUILTIN_CLFLUSHOPT.
1704         (bdesc_special_args): Add __builtin_ia32_xsaves,
1705         __builtin_ia32_xrstors, __builtin_ia32_xsavec, __builtin_ia32_xsaves64,
1706         __builtin_ia32_xrstors64, __builtin_ia32_xsavec64.
1707         (ix86_init_mmx_sse_builtins): Add __builtin_ia32_clflushopt.
1708         (ix86_expand_builtin): Handle new builtins.
1709         * config/i386/i386.h (TARGET_CLFLUSHOPT) Define.
1710         (TARGET_CLFLUSHOPT_P): Ditto.
1711         (TARGET_XSAVEC): Ditto.
1712         (TARGET_XSAVEC_P): Ditto.
1713         (TARGET_XSAVES): Ditto.
1714         (TARGET_XSAVES_P): Ditto.
1715         * config/i386/i386.md (ANY_XSAVE): Add UNSPECV_XSAVEC, UNSPECV_XSAVES.
1716         (ANY_XSAVE64)" Add UNSPECV_XSAVEC64, UNSPECV_XSAVES64.
1717         (attr xsave): Add xsavec, xsavec64, xsaves, xsaves64.
1718         (ANY_XRSTOR): New.
1719         (ANY_XRSTOR64): Ditto.
1720         (xrstor): Ditto.
1721         (xrstor): Change into <xrstor>.
1722         (xrstor_rex64): Change into <xrstor>_rex64.
1723         (xrstor64): Change into <xrstor>64
1724         (clflushopt): New.
1725         * config/i386/i386.opt (mclflushopt): New.
1726         (mxsavec): Ditto.
1727         (mxsaves): Ditto.
1728         * config/i386/x86intrin.h: Add clflushoptintrin.h, xsavesintrin.h,
1729         xsavecintrin.h.
1730         * doc/invoke.texi: Document new options.
1731
1732 2014-05-14  Andrey Belevantsev  <abel@ispras.ru>
1733
1734         PR rtl-optimization/60866
1735         * sel-sched-ir (sel_init_new_insn): New parameter old_seqno.
1736         Default it to -1.  Pass it down to init_simplejump_data.
1737         (init_simplejump_data): New parameter old_seqno.  Pass it down
1738         to get_seqno_for_a_jump.
1739         (get_seqno_for_a_jump): New parameter old_seqno.  Use it for
1740         initializing new jump seqno as a last resort.  Add comment.
1741         (sel_redirect_edge_and_branch): Save old seqno of the conditional
1742         jump and pass it down to sel_init_new_insn.
1743         (sel_redirect_edge_and_branch_force): Likewise.
1744
1745 2014-05-14  Georg-Johann Lay  <avr@gjlay.de>
1746
1747         * config/avr/avr.h (REG_CLASS_CONTENTS): Use unsigned suffix for
1748         shifted values to avoid build warning.
1749
1750 2014-05-14  Eric Botcazou  <ebotcazou@adacore.com>
1751
1752         * cfgcleanup.c (try_forward_edges): Use location_t for locations.
1753         * cfgrtl.c (rtl_merge_blocks): Fix comment.
1754         (cfg_layout_merge_blocks): Likewise.
1755         * except.c (emit_to_new_bb_before): Remove prev_bb local variable.
1756
1757 2014-05-14  Andrey Belevantsev  <abel@ispras.ru>
1758
1759         PR rtl-optimization/60901
1760         * config/i386/i386.c (ix86_dependencies_evaluation_hook): Check that
1761         bb predecessor belongs to the same scheduling region.  Adjust comment.
1762
1763 2014-05-13  Peter Bergner  <bergner@vnet.ibm.com>
1764
1765         * doc/sourcebuild.texi: (dfp_hw): Document.
1766         (p8vector_hw): Likewise.
1767         (powerpc_eabi_ok): Likewise.
1768         (powerpc_elfv2): Likewise.
1769         (powerpc_htm_ok): Likewise.
1770         (ppc_recip_hw): Likewise.
1771         (vsx_hw): Likewise.
1772
1773 2014-05-13  Cary Coutant  <ccoutant@google.com>
1774
1775         * opts.c (finish_options): Use -ggnu-pubnames with -gsplit-dwarf.
1776
1777 2014-05-13  David Malcolm  <dmalcolm@redhat.com>
1778
1779         * gengtype-parse.c (require3): Eliminate in favor of...
1780         (require4): New.
1781         (require_template_declaration): Update to support optional single *
1782         on a type.
1783
1784         * gengtype.c (get_ultimate_base_class): Add a non-const overload.
1785         (create_user_defined_type): Handle a single level of explicit
1786         pointerness within template arguments.
1787         (struct write_types_data): Add field "kind".
1788         (filter_type_name): Handle "*" character.
1789         (write_user_func_for_structure_ptr): Require a write_types_data
1790         rather than just a prefix string, so that we can look up the kind
1791         of the wtd and use it as an index into wrote_user_func_for_ptr,
1792         ensuring that such functions are written at most once.  Support
1793         subclasses by invoking the marking function of the ultimate base class.
1794         (write_user_func_for_structure_body): Require a write_types_data
1795         rather than just a prefix string, so that we can pass this to
1796         write_user_func_for_structure_ptr.
1797         (write_func_for_structure): Likewise.
1798         (ggc_wtd): Add initializer of new "kind" field.
1799         (pch_wtd): Likewise.
1800
1801         * gengtype.h (enum write_types_kinds): New.
1802         (struct type): Add field wrote_user_func_for_ptr to the "s"
1803         union member.
1804
1805 2014-05-13  Richard Sandiford  <r.sandiford@uk.ibm.com>
1806
1807         * fold-const.c (optimize_bit_field_compare): Use wi:: operations
1808         instead of const_binop.
1809         (fold_binary_loc): Likewise.
1810
1811 2014-05-13  Richard Sandiford  <r.sandiford@uk.ibm.com>
1812
1813         * tree-dfa.h (get_addr_base_and_unit_offset_1): Update array index
1814         calculation to match get_ref_base_and_extent.
1815
1816 2014-05-13  Catherine Moore  <clm@codesourcery.com>
1817             Sandra Loosemore  <sandra@codesourcery.com>
1818
1819         * configure.ac: Fix assembly for explicit JALR relocation check.
1820         * configure: Regenerate.
1821
1822 2014-05-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1823
1824         * config/arm/arm.c (neon_itype): Remove NEON_RESULTPAIR.
1825         (arm_init_neon_builtins): Remove handling of NEON_RESULTPAIR.
1826         Remove associated type declarations and initialisations.
1827         (arm_expand_neon_builtin): Likewise.
1828         (neon_emit_pair_result_insn): Delete.
1829         * config/arm/arm_neon_builtins (vtrn, vzip, vuzp): Delete.
1830         * config/arm/neon.md (neon_vtrn<mode>): Delete.
1831         (neon_vzip<mode>): Likewise.
1832         (neon_vuzp<mode>): Likewise.
1833
1834 2014-05-13  Richard Biener  <rguenther@suse.de>
1835
1836         PR ipa/60973
1837         * tree-inline.c (remap_gimple_stmt): Clear tail call flag,
1838         it needs revisiting whether the call still may be tail-called.
1839
1840 2014-05-13  Richard Sandiford  <rdsandiford@googlemail.com>
1841
1842         * rtl.def (SYMBOL_REF): Remove middle "0" field.
1843         * rtl.h (block_symbol): Reduce number of fields to 2.
1844         (rtx_def): Add u2.symbol_ref_flags.
1845         (SYMBOL_REF_FLAGS): Use it.
1846         (SYMBOL_REF_DATA, SET_SYMBOL_REF_DECL, SYMBOL_REF_DECL)
1847         (SET_SYMBOL_REF_CONSTANT, SYMBOL_REF_CONSTANT): Lower index.
1848         * gengtype.c (adjust_field_rtx_def): Remove SYMBOL_REF_FLAGS handling.
1849         Lower index of SYMBOL_REF_DATA.
1850         * print-rtl.c (print_rtx): Lower index for SYMBOL_REF_DATA.
1851         Print SYMBOL_REF_FLAGS at the same time.
1852         * genattrtab.c (attr_rtx_1): Only initialize 1 "0" SYMBOL_REF field.
1853
1854 2014-05-13  Richard Sandiford  <rdsandiford@googlemail.com>
1855
1856         * rtl.def (VAR_LOCATION): Remove "i" field.
1857         * rtl.h (rtx_def): Add u2.var_location_status.
1858         (PAT_VAR_LOCATION_STATUS): Use it.
1859         (gen_rtx_VAR_LOCATION): Declare.
1860         * gengenrtl.c (excluded_rtx): Add VAR_LOCATION.
1861         * emit-rtl.c (gen_rtx_VAR_LOCATION): New function.
1862         * var-tracking.c (emit_note_insn_var_location): Remove casts.
1863
1864 2014-05-13  Richard Sandiford  <rdsandiford@googlemail.com>
1865
1866         * rtl.def (scratch): Fix outdated comment and remove "0" field.
1867         * gengtype.c (adjust_field_rtx_def): Update accordingly.
1868
1869 2014-05-13  Richard Sandiford  <rdsandiford@googlemail.com>
1870
1871         * rtl.def (DEBUG_INSN, INSN, JUMP_INSN, CALL_INSN, JUMP_TABLE_DATA)
1872         (BARRIER, CODE_LABEL, NOTE): Remove first "i" field.
1873         * rtl.h (rtx_def): Add insn_uid to u2 field.
1874         (RTX_FLAG_CHECK8): Delete in favor of...
1875         (RTL_INSN_CHAIN_FLAG_CHECK): ...this new macro.
1876         (INSN_DELETED_P): Update accordingly.
1877         (INSN_UID): Use u2.insn_uid.
1878         (INSN_CHAIN_CODE_P): Define.
1879         (PREV_INSN, NEXT_INSN, BLOCK_FOR_INSN, PATTERN, INSN_LOCATION)
1880         (INSN_CODE, REG_NOTES, CALL_INSN_FUNCTION_USAGE, CODE_LABEL_NUMBER)
1881         (NOTE_DATA, NOTE_DELETED_LABEL_NAME, NOTE_BLOCK, NOTE_EH_HANDLER)
1882         (NOTE_BASIC_BLOCK, NOTE_VAR_LOCATION, NOTE_CFI, NOTE_LABEL_NUMBER)
1883         (NOTE_KIND, LABEL_NAME, LABEL_NUSES, JUMP_LABEL, LABEL_REFS): Lower
1884         indices accordingly.
1885         * print-rtl.c (print_rtx): Print INSN_UIDs before the main loop.
1886         Update indices for insn-chain rtxes.
1887         * gengtype.c (gen_rtx_next): Adjust test for insn-chain rtxes.
1888         (adjust_field_rtx_def): Lower '0' indices for all insn-chain rtxes.
1889         * emit-rtl.c (gen_label_rtx): Update gen_rtx_LABEL call.
1890         * caller-save.c (init_caller_save): Update gen_rtx_INSN calls.
1891         * combine.c (try_combine): Likewise.
1892         * ira.c (setup_prohibited_mode_move_regs): Likewise.
1893
1894 2014-05-13  Richard Sandiford  <rdsandiford@googlemail.com>
1895
1896         * rtl.def (REG): Remove middle field.
1897         * rtl.h (rtx_def): Add orignal_regno to u2.
1898         (ORIGINAL_REGNO): Use it instead of field 1.
1899         (REG_ATTRS): Lower field index accordingly.
1900         * gengtype.c (adjust_field_rtx_def): Remove handling of
1901         ORIGINAL_REGNO.  Move REG_ATTRS index down.
1902         * print-rtl.c (print_rtx): Move ORIGINAL_REGNO handling to the
1903         code that prints the REGNO.
1904
1905 2014-05-13  Richard Sandiford  <rdsandiford@googlemail.com>
1906
1907         * print-rtl.c (print_rtx): Guard whole '0' block with ifndef
1908         GENERATOR_FILE.
1909
1910 2014-05-13  Richard Sandiford  <rdsandiford@googlemail.com>
1911
1912         * rtl.h (rtx_def): Mark u2 as GTY ((skip)).
1913
1914 2014-05-13  Bin Cheng  <bin.cheng@arm.com>
1915
1916         * tree-ssa-loop-ivopts.c (contain_complex_addr_expr): New.
1917         (alloc_iv): Lower base expressions containing ADDR_EXPR.
1918
1919 2014-05-13  Ian Bolton  <ian.bolton@arm.com>
1920
1921         * config/aarch64/aarch64-protos.h
1922         (aarch64_hard_regno_caller_save_mode): New prototype.
1923         * config/aarch64/aarch64.c (aarch64_hard_regno_caller_save_mode):
1924         New function.
1925         * config/aarch64/aarch64.h (HARD_REGNO_CALLER_SAVE_MODE): New macro.
1926
1927 2014-05-13  Christian Bruel  <christian.bruel@st.com>
1928
1929         * target.def (mode_switching): New hook vector.
1930         (mode_emit, mode_needed, mode_after, mode_entry): New hooks.
1931         (mode_exit, modepriority_to_mode): Likewise.
1932         * mode-switching.c (MODE_NEEDED, MODE_AFTER, MODE_ENTRY): Hookify.
1933         (MODE_EXIT, MODE_PRIORITY_TO_MODE, EMIT_MODE_SET): Likewise.
1934         * target.h: Include tm.h and hard-reg-set.h.
1935         * doc/tm.texi.in (EMIT_MODE_SET, MODE_NEEDED, MODE_AFTER, MODE_ENTRY)
1936         (MODE_EXIT, MODE_PRIORITY_TO_MODE): Delete and hookify.
1937         * doc/tm.texi Regenerate.
1938         * config/sh/sh.h (MODE_NEEDED, MODE_AFTER, MODE_ENTRY): Delete
1939         (MODE_EXIT, MODE_PRIORITY_TO_MODE, EMIT_MODE_SET): Likewise.
1940         * config/sh/sh.c (sh_emit_mode_set, sh_mode_priority): Hookify.
1941         (sh_mode_needed, sh_mode_after, sh_mode_entry, sh_mode_exit): Likewise.
1942         * config/i386/i386.h (MODE_NEEDED, MODE_AFTER, MODE_ENTRY): Delete
1943         (MODE_EXIT, MODE_PRIORITY_TO_MODE, EMIT_MODE_SET): Likewise.
1944         * config/i386/i386-protos.h (ix86_mode_needed, ix86_mode_after)
1945         (ix86_mode_entrym, ix86_emit_mode_set): Remove external declaration.
1946         * config/i386/i386.c (ix86_mode_needed, ix86_mode_after,
1947         (ix86_mode_exit, ix86_mode_entry, ix86_mode_priority)
1948         (ix86_emit_mode_set): Hookify.
1949         * config/epiphany/epiphany.h (MODE_NEEDED, MODE_AFTER, MODE_ENTRY):
1950         Delete.
1951         (MODE_EXIT, MODE_PRIORITY_TO_MODE, EMIT_MODE_SET): Likewise.
1952         * config/epiphany/epiphany-protos.h (epiphany_mode_needed)
1953         (emit_set_fp_mode, epiphany_mode_entry_exit, epiphany_mode_after)
1954         (epiphany_mode_priority_to_mode): Remove declaration.
1955         * config/epiphany/epiphany.c (emit_set_fp_mode): Hookify.
1956         (epiphany_mode_needed, epiphany_mode_priority_to_mode): Likewise.
1957         (epiphany_mode_entry, epiphany_mode_exit, epiphany_mode_after):
1958         Likewise.
1959         (epiphany_mode_priority_to_mode): Change priority type.  Hookify.
1960         (epiphany_mode_needed, epiphany_mode_entry_exit): Hookify.
1961         (epiphany_mode_after, epiphany_mode_entry, emit_set_fp_mode): Hookify.
1962
1963 2014-05-13  Jakub Jelinek  <jakub@redhat.com>
1964
1965         PR target/61060
1966         * config/i386/i386.c (ix86_expand_set_or_movmem): If count_exp
1967         is const0_rtx, return immediately.  Don't test count == 0 when
1968         it is always true.
1969
1970 2014-05-13  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
1971
1972         * Makefile.in: add shrink-wrap.o.
1973         * config/i386/i386.c: include "shrink-wrap.h"
1974         * function.c: Likewise.
1975         (requires_stack_frame_p, next_block_for_reg,
1976         move_insn_for_shrink_wrap, prepare_shrink_wrap,
1977         dup_block_and_redirect): Move to shrink-wrap.c
1978         (thread_prologue_and_epilogue_insns): Extract three code segments
1979         as functions in shrink-wrap.c
1980         * function.h: Move #ifdef HAVE_simple_return ... #endif block to
1981         shrink-wrap.h
1982         * shrink-wrap.c: New file.
1983         * shrink-wrap.h: New file.
1984
1985 2014-05-12  David Wohlferd  <dw@LimeGreenSocks.com>
1986
1987         * doc/extend.texi: Reflect current numbers of pragmas.  Remove
1988         reference to Solaris.
1989
1990 2014-05-12  Mike Stump  <mikestump@comcast.net>
1991
1992         PR other/31778
1993         * genattrtab.c (filename): Add.
1994         (convert_set_attr_alternative): Improve error message.
1995         (check_defs): Restore read_md_filename for error messages.
1996         (gen_insn): Save filename.
1997
1998 2014-05-12  Dimitris Papavasiliou  <dpapavas@gmail.com>
1999
2000         * doc/invoke.texi: Document new switches -Wno-shadow-ivar,
2001         -fno-local-ivars and -fivar-visibility.
2002         * c-family/c.opt: Make -Wshadow also implicitly enable
2003         -Wshadow-ivar.
2004
2005 2014-05-12  David Wohlferd  <dw@LimeGreenSocks.com>
2006
2007         * doc/tm.texi: Remove reference to deleted macro.
2008         * doc/tm.texi.in: Likewise.
2009
2010 2014-05-12  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
2011
2012         PR target/60991
2013         * config/avr/avr.c (avr_out_store_psi): Use correct constant
2014         to restore Y.
2015
2016 2014-05-12  Georg-Johann Lay  <avr@gjlay.de>
2017
2018         * config/arm/arm.h (License): Add GCC Runtime Library Exception.
2019         * config/arm/aout.h (License): Same.
2020         * config/arm/bpabi.h (License): Same.
2021         * config/arm/elf.h (License): Same.
2022         * config/arm/linux-elf.h (License): Same.
2023         * config/arm/linux-gas.h (License): Same.
2024         * config/arm/netbsd-elf.h (License): Same.
2025         * config/arm/uclinux-eabi.h (License): Same.
2026         * config/arm/uclinux-elf.h (License): Same.
2027         * config/arm/vxworks.h (License): Same.
2028
2029 2014-05-11  Jakub Jelinek  <jakub@redhat.com>
2030
2031         * tree.h (OMP_CLAUSE_LINEAR_STMT): Define.
2032         * tree.c (omp_clause_num_ops): Increase OMP_CLAUSE_LINEAR
2033         number of operands to 3.
2034         (walk_tree_1): Walk all operands of OMP_CLAUSE_LINEAR.
2035         * tree-nested.c (convert_nonlocal_omp_clauses,
2036         convert_local_omp_clauses): Handle OMP_CLAUSE_DEPEND.
2037         * gimplify.c (gimplify_scan_omp_clauses): Handle
2038         OMP_CLAUSE_LINEAR_STMT.
2039         * omp-low.c (lower_rec_input_clauses): Fix typo.
2040         (maybe_add_implicit_barrier_cancel, lower_omp_1): Add
2041         cast between Fortran boolean_type_node and C _Bool if
2042         needed.
2043
2044 2014-05-11  Richard Sandiford  <rdsandiford@googlemail.com>
2045
2046         PR tree-optimization/61136
2047         * wide-int.h (multiple_of_p): Define a version that doesn't return
2048         the quotient.
2049         * fold-const.c (extract_muldiv_1): Use wi::multiple_of_p instead of an
2050         integer_zerop/const_binop pair.
2051         (multiple_of_p): Likewise, converting both operands to widest_int
2052         precision.
2053
2054 2014-05-09  Teresa Johnson  <tejohnson@google.com>
2055
2056         * cgraphunit.c (analyze_functions): Use correct dump file.
2057
2058 2014-05-09  Florian Weimer  <fweimer@redhat.com>
2059
2060         * cfgexpand.c (stack_protect_decl_p): New function, extracted from
2061         expand_used_vars.
2062         (stack_protect_return_slot_p): New function.
2063         (expand_used_vars): Call stack_protect_decl_p and
2064         stack_protect_return_slot_p for -fstack-protector-strong.
2065
2066 2014-05-09  Kenneth Zadeck  <zadeck@naturalbridge.com>
2067
2068         PR middle-end/61111
2069         * fold-const.c (fold_binary_loc): Changed width of mask.
2070
2071 2014-05-09  Georg-Johann Lay  <avr@gjlay.de>
2072
2073         * config/avr/avr-fixed.md (round<mode>3): Use -1U instead of -1 in
2074         unsigned int initializers for regno_in, regno_out.
2075
2076 2014-05-09  Georg-Johann Lay  <avr@gjlay.de>
2077
2078         PR target/61055
2079         * config/avr/avr.md (cc): Add new attribute set_vzn.
2080         (addqi3, addqq3, adduqq3, subqi3, subqq3, subuqq3, negqi2) [cc]:
2081         Set cc insn attribute to set_vzn instead of set_zn for alternatives
2082         with INC, DEC or NEG.
2083         * config/avr/avr.c (avr_notice_update_cc): Handle SET_VZN.
2084         (avr_out_plus_1): ADIW sets cc0 to CC_SET_CZN.
2085         INC, DEC and ADD+ADC set cc0 to CC_CLOBBER.
2086
2087 2014-05-09  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2088
2089         Revert:
2090         2014-05-08  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2091
2092         * wide-int.cc (UTItype): Define.
2093         (UDWtype): Define for appropriate W_TYPE_SIZE.
2094
2095 2014-05-09  Richard Biener  <rguenther@suse.de>
2096
2097         * Makefile.in (GTFILES): Remove tree-ssa-propagate.c.
2098         * tree-ssa-propagate.c: Do not include gt-tree-ssa-propagate.h.
2099         (interesting_ssa_edges, varying_ssa_edges): Move out of GC space.
2100         (add_ssa_edge, process_ssa_edge_worklist, ssa_prop_init,
2101         ssa_propagate): Adjust.
2102
2103 2014-05-08  Jeff Law  <law@redhat.com>
2104
2105         PR tree-optimization/61009
2106         * tree-ssa-threadedge.c (thread_through_normal_block): Return a
2107         tri-state rather than a boolean.  When a block is too big to
2108         thread through, inform caller via negative return value.
2109         (thread_across_edge): If a block was too big for normal threading,
2110         then it's too big for a joiner too, so remove temporary equivalences
2111         and return immediately.
2112
2113 2014-05-08  Manuel López-Ibáñez  <manu@gcc.gnu.org>
2114             Matthias Klose  <doko@ubuntu.com>
2115
2116         PR driver/61106
2117         * optc-gen.awk: Fix option handling for -Wunused-parameter.
2118
2119 2014-05-08  Uros Bizjak  <ubizjak@gmail.com>
2120
2121         PR target/59952
2122         * config/i386/i386.c (PTA_HASWELL): Remove PTA_RTM.
2123
2124 2014-05-08  Uros Bizjak  <ubizjak@gmail.com>
2125
2126         PR target/61092
2127         * config/alpha/alpha.c: Include gimple-iterator.h.
2128         (alpha_gimple_fold_builtin): New function.  Move
2129         ALPHA_BUILTIN_UMULH folding from ...
2130         (alpha_fold_builtin): ... here.
2131         (TARGET_GIMPLE_FOLD_BUILTIN): New define.
2132
2133 2014-05-08  Wei Mi  <wmi@google.com>
2134
2135         PR target/58066
2136         * config/i386/i386.c (ix86_compute_frame_layout): Update
2137         preferred_stack_boundary for call, expanded from tls descriptor.
2138         * config/i386/i386.md (*tls_global_dynamic_32_gnu): Update RTX
2139         to depend on SP register.
2140         (*tls_local_dynamic_base_32_gnu): Ditto.
2141         (*tls_local_dynamic_32_once): Ditto.
2142         (tls_global_dynamic_64_<mode>): Set
2143         ix86_tls_descriptor_calls_expanded_in_cfun.
2144         (tls_local_dynamic_base_64_<mode>): Ditto.
2145         (tls_global_dynamic_32): Set
2146         ix86_tls_descriptor_calls_expanded_in_cfun. Update RTX
2147         to depend on SP register.
2148         (tls_local_dynamic_base_32): Ditto.
2149
2150 2014-05-08  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2151
2152         * config/arm/arm_neon.h: Update comment.
2153         * config/arm/neon-docgen.ml: Delete.
2154         * config/arm/neon-gen.ml: Delete.
2155         * doc/arm-neon-intrinsics.texi: Update comment.
2156
2157 2014-05-08  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2158
2159         * config/arm/arm_neon_builtins.def (vadd, vsub): Only define the v2sf
2160         and v4sf versions.
2161         (vand, vorr, veor, vorn, vbic): Remove.
2162         * config/arm/neon.md (neon_vadd, neon_vsub, neon_vadd_unspec): Adjust
2163         iterator.
2164         (neon_vsub_unspec): Likewise.
2165         (neon_vorr, neon_vand, neon_vbic, neon_veor, neon_vorn): Remove.
2166
2167 2014-05-08  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2168
2169         * config/arm/arm_neon.h (vadd_s8): GNU C implementation
2170         (vadd_s16): Likewise.
2171         (vadd_s32): Likewise.
2172         (vadd_f32): Likewise.
2173         (vadd_u8): Likewise.
2174         (vadd_u16): Likewise.
2175         (vadd_u32): Likewise.
2176         (vadd_s64): Likewise.
2177         (vadd_u64): Likewise.
2178         (vaddq_s8): Likewise.
2179         (vaddq_s16): Likewise.
2180         (vaddq_s32): Likewise.
2181         (vaddq_s64): Likewise.
2182         (vaddq_f32): Likewise.
2183         (vaddq_u8): Likewise.
2184         (vaddq_u16): Likewise.
2185         (vaddq_u32): Likewise.
2186         (vaddq_u64): Likewise.
2187         (vmul_s8): Likewise.
2188         (vmul_s16): Likewise.
2189         (vmul_s32): Likewise.
2190         (vmul_f32): Likewise.
2191         (vmul_u8): Likewise.
2192         (vmul_u16): Likewise.
2193         (vmul_u32): Likewise.
2194         (vmul_p8): Likewise.
2195         (vmulq_s8): Likewise.
2196         (vmulq_s16): Likewise.
2197         (vmulq_s32): Likewise.
2198         (vmulq_f32): Likewise.
2199         (vmulq_u8): Likewise.
2200         (vmulq_u16): Likewise.
2201         (vmulq_u32): Likewise.
2202         (vsub_s8): Likewise.
2203         (vsub_s16): Likewise.
2204         (vsub_s32): Likewise.
2205         (vsub_f32): Likewise.
2206         (vsub_u8): Likewise.
2207         (vsub_u16): Likewise.
2208         (vsub_u32): Likewise.
2209         (vsub_s64): Likewise.
2210         (vsub_u64): Likewise.
2211         (vsubq_s8): Likewise.
2212         (vsubq_s16): Likewise.
2213         (vsubq_s32): Likewise.
2214         (vsubq_s64): Likewise.
2215         (vsubq_f32): Likewise.
2216         (vsubq_u8): Likewise.
2217         (vsubq_u16): Likewise.
2218         (vsubq_u32): Likewise.
2219         (vsubq_u64): Likewise.
2220         (vand_s8): Likewise.
2221         (vand_s16): Likewise.
2222         (vand_s32): Likewise.
2223         (vand_u8): Likewise.
2224         (vand_u16): Likewise.
2225         (vand_u32): Likewise.
2226         (vand_s64): Likewise.
2227         (vand_u64): Likewise.
2228         (vandq_s8): Likewise.
2229         (vandq_s16): Likewise.
2230         (vandq_s32): Likewise.
2231         (vandq_s64): Likewise.
2232         (vandq_u8): Likewise.
2233         (vandq_u16): Likewise.
2234         (vandq_u32): Likewise.
2235         (vandq_u64): Likewise.
2236         (vorr_s8): Likewise.
2237         (vorr_s16): Likewise.
2238         (vorr_s32): Likewise.
2239         (vorr_u8): Likewise.
2240         (vorr_u16): Likewise.
2241         (vorr_u32): Likewise.
2242         (vorr_s64): Likewise.
2243         (vorr_u64): Likewise.
2244         (vorrq_s8): Likewise.
2245         (vorrq_s16): Likewise.
2246         (vorrq_s32): Likewise.
2247         (vorrq_s64): Likewise.
2248         (vorrq_u8): Likewise.
2249         (vorrq_u16): Likewise.
2250         (vorrq_u32): Likewise.
2251         (vorrq_u64): Likewise.
2252         (veor_s8): Likewise.
2253         (veor_s16): Likewise.
2254         (veor_s32): Likewise.
2255         (veor_u8): Likewise.
2256         (veor_u16): Likewise.
2257         (veor_u32): Likewise.
2258         (veor_s64): Likewise.
2259         (veor_u64): Likewise.
2260         (veorq_s8): Likewise.
2261         (veorq_s16): Likewise.
2262         (veorq_s32): Likewise.
2263         (veorq_s64): Likewise.
2264         (veorq_u8): Likewise.
2265         (veorq_u16): Likewise.
2266         (veorq_u32): Likewise.
2267         (veorq_u64): Likewise.
2268         (vbic_s8): Likewise.
2269         (vbic_s16): Likewise.
2270         (vbic_s32): Likewise.
2271         (vbic_u8): Likewise.
2272         (vbic_u16): Likewise.
2273         (vbic_u32): Likewise.
2274         (vbic_s64): Likewise.
2275         (vbic_u64): Likewise.
2276         (vbicq_s8): Likewise.
2277         (vbicq_s16): Likewise.
2278         (vbicq_s32): Likewise.
2279         (vbicq_s64): Likewise.
2280         (vbicq_u8): Likewise.
2281         (vbicq_u16): Likewise.
2282         (vbicq_u32): Likewise.
2283         (vbicq_u64): Likewise.
2284         (vorn_s8): Likewise.
2285         (vorn_s16): Likewise.
2286         (vorn_s32): Likewise.
2287         (vorn_u8): Likewise.
2288         (vorn_u16): Likewise.
2289         (vorn_u32): Likewise.
2290         (vorn_s64): Likewise.
2291         (vorn_u64): Likewise.
2292         (vornq_s8): Likewise.
2293         (vornq_s16): Likewise.
2294         (vornq_s32): Likewise.
2295         (vornq_s64): Likewise.
2296         (vornq_u8): Likewise.
2297         (vornq_u16): Likewise.
2298         (vornq_u32): Likewise.
2299         (vornq_u64): Likewise.
2300
2301 2014-05-08  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2302
2303         * wide-int.cc (UTItype): Define.
2304         (UDWtype): Define for appropriate W_TYPE_SIZE.
2305
2306 2014-05-08  Marc Glisse  <marc.glisse@inria.fr>
2307
2308         PR tree-optimization/59100
2309         * tree-ssa-phiopt.c: Include tree-inline.h.
2310         (neutral_element_p, absorbing_element_p): New functions.
2311         (value_replacement): Handle conditional binary operations with a
2312         neutral or absorbing element.
2313
2314 2014-05-08  Richard Biener  <rguenther@suse.de>
2315
2316         * tree-ssa-sccvn.c (vn_get_expr_for): Valueize operands before
2317         folding the expression.
2318         (valueize_expr): Remove.
2319         (visit_reference_op_load): Do not valueize the result of
2320         vn_get_expr_for.
2321         (simplify_binary_expression): Likewise.
2322         (simplify_unary_expression): Likewise.
2323
2324 2014-05-08  Richard Biener  <rguenther@suse.de>
2325
2326         * gimplify.c (gimplify_call_expr): Use saved fnptrtype for
2327         looking at TYPE_ARG_TYPES.
2328
2329 2014-05-08  Richard Biener  <rguenther@suse.de>
2330
2331         * gimple-fold.c (gimple_fold_stmt_to_constant_1): Remove
2332         pointer propagation special-case.
2333
2334 2014-05-08  Bin Cheng  <bin.cheng@arm.com>
2335
2336         * tree-affine.c (tree_to_aff_combination): Handle MEM_REF for
2337         core part of address expressions.
2338
2339 2014-05-08  Alan Modra  <amodra@gmail.com>
2340
2341         PR target/60737
2342         * config/rs6000/rs6000.c (expand_block_move): Allow 64-bit
2343         loads and stores when -mno-strict-align at any alignment.
2344         (expand_block_clear): Similarly.  Also correct calculation of
2345         instruction count.
2346
2347 2014-05-07  Thomas Preud'homme  <thomas.preudhomme@arm.com>
2348
2349         PR middle-end/39246
2350         * tree-complex.c (expand_complex_move): Keep line info when expanding
2351         complex move.
2352         * tree-ssa-uninit.c (warn_uninit): New argument. Ignore assignment
2353         of complex expression. Use new argument to display correct location
2354         for values coming from phi statement.
2355         (warn_uninitialized_vars): Adapt to new signature of warn_uninit.
2356         (warn_uninitialized_phi): Pass location of phi argument to
2357         warn_uninit.
2358         * tree-ssa.c (ssa_undefined_value_p): For SSA_NAME initialized by a
2359         COMPLEX_EXPR, recurse on each part of the COMPLEX_EXPR.
2360
2361 2014-05-07  Segher Boessenkool  <segher@kernel.crashing.org>
2362
2363         * config/rs6000/predicates.md (indexed_address_mem): New.
2364         * config/rs6000/rs6000.md (type): Remove load_ext, load_ext_u,
2365         load_ext_ux, load_ux, load_u, store_ux, store_u, fpload_ux, fpload_u,
2366         fpstore_ux, fpstore_u.
2367         (sign_extend, indexed, update): New.
2368         (cell_micro): Adjust.
2369         (*zero_extend<mode>di2_internal1, *zero_extendsidi2_lfiwzx,
2370         *extendsidi2_lfiwax, *extendsidi2_nocell, *extendsfdf2_fpr,
2371         *movsi_internal1, *movsi_internal1_single, *movhi_internal,
2372         *movqi_internal, *movcc_internal1, mov<mode>_hardfloat,
2373         *mov<mode>_softfloat, *mov<mode>_hardfloat32, *mov<mode>_hardfloat64,
2374         *mov<mode>_softfloat64, *movdi_internal32, *movdi_internal64,
2375         *mov<mode>_string, *ldmsi8, *ldmsi7, *ldmsi6, *ldmsi5, *ldmsi4,
2376         *ldmsi3, *stmsi8, *stmsi7, *stmsi6, *stmsi5, *stmsi4, *stmsi3,
2377         *movdi_update1, movdi_<mode>_update, movdi_<mode>_update_stack,
2378         *movsi_update1, *movsi_update2, movsi_update, movsi_update_stack,
2379         *movhi_update1, *movhi_update2, *movhi_update3, *movhi_update4,
2380         *movqi_update1, *movqi_update2, *movqi_update3, *movsf_update1,
2381         *movsf_update2, *movsf_update3, *movsf_update4, *movdf_update1,
2382         *movdf_update2, load_toc_aix_si, load_toc_aix_di, probe_stack_<mode>,
2383         *stmw, *lmw, as well as 10 anonymous patterns): Adjust.
2384
2385         * config/rs6000/dfp.md (movsd_store, movsd_load): Adjust.
2386         * config/rs6000/vsx.md (*vsx_movti_32bit, *vsx_extract_<mode>_load,
2387         *vsx_extract_<mode>_store): Adjust.
2388         * config/rs6000/rs6000.c (rs6000_adjust_cost, is_microcoded_insn,
2389         is_cracked_insn, insn_must_be_first_in_group,
2390         insn_must_be_last_in_group): Adjust.
2391
2392         * config/rs6000/40x.md (ppc403-load, ppc403-store, ppc405-float):
2393         Adjust.
2394         * config/rs6000/440.md (ppc440-load, ppc440-store, ppc440-fpload,
2395         ppc440-fpstore): Adjust.
2396         * config/rs6000/476.md (ppc476-load, ppc476-store, ppc476-fpload,
2397         ppc476-fpstore): Adjust.
2398         * config/rs6000/601.md (ppc601-load, ppc601-store, ppc601-fpload,
2399         ppc601-fpstore): Adjust.
2400         * config/rs6000/603.md (ppc603-load, ppc603-store, ppc603-fpload):
2401         Adjust.
2402         * config/rs6000/6xx.md (ppc604-load, ppc604-store, ppc604-fpload):
2403         Adjust.
2404         * config/rs6000/7450.md (ppc7450-load, ppc7450-store, ppc7450-fpload,
2405         ppc7450-fpstore): Adjust.
2406         * config/rs6000/7xx.md (ppc750-load, ppc750-store): Adjust.
2407         * config/rs6000/8540.md (ppc8540_load, ppc8540_store): Adjust.
2408         * config/rs6000/a2.md (ppca2-load, ppca2-fp-load, ppca2-fp-store):
2409         Adjust.
2410         * config/rs6000/cell.md (cell-load, cell-load-ux, cell-load-ext,
2411         cell-fpload, cell-fpload-update, cell-store, cell-store-update,
2412         cell-fpstore, cell-fpstore-update): Adjust.
2413         * config/rs6000/e300c2c3.md (ppce300c3_load, ppce300c3_fpload,
2414         ppce300c3_store, ppce300c3_fpstore): Adjust.
2415         * config/rs6000/e500mc.md (e500mc_load, e500mc_fpload, e500mc_store,
2416         e500mc_fpstore): Adjust.
2417         * config/rs6000/e500mc64.md (e500mc64_load, e500mc64_fpload,
2418         e500mc64_store, e500mc64_fpstore): Adjust.
2419         * config/rs6000/e5500.md (e5500_load, e5500_fpload, e5500_store,
2420         e5500_fpstore): Adjust.
2421         * config/rs6000/e6500.md (e6500_load, e6500_fpload, e6500_store,
2422         e6500_fpstore): Adjust.
2423         * config/rs6000/mpc.md (mpccore-load, mpccore-store, mpccore-fpload):
2424         Adjust.
2425         * config/rs6000/power4.md (power4-load, power4-load-ext,
2426         power4-load-ext-update, power4-load-ext-update-indexed,
2427         power4-load-update-indexed, power4-load-update, power4-fpload,
2428         power4-fpload-update, power4-store, power4-store-update,
2429         power4-store-update-indexed, power4-fpstore, power4-fpstore-update):
2430         Adjust.
2431         * config/rs6000/power5.md (power5-load, power5-load-ext,
2432         power5-load-ext-update, power5-load-ext-update-indexed,
2433         power5-load-update-indexed, power5-load-update, power5-fpload,
2434         power5-fpload-update, power5-store, power5-store-update,
2435         power5-store-update-indexed, power5-fpstore, power5-fpstore-update):
2436         Adjust.
2437         * config/rs6000/power6.md (power6-load, power6-load-ext,
2438         power6-load-update, power6-load-update-indexed,
2439         power6-load-ext-update, power6-load-ext-update-indexed, power6-fpload,
2440         power6-fpload-update, power6-store, power6-store-update,
2441         power6-store-update-indexed, power6-fpstore, power6-fpstore-update):
2442         Adjust.
2443         * config/rs6000/power7.md (power7-load, power7-load-ext,
2444         power7-load-update, power7-load-update-indexed,
2445         power7-load-ext-update, power7-load-ext-update-indexed, power7-fpload,
2446         power7-fpload-update, power7-store, power7-store-update,
2447         power7-store-update-indexed, power7-fpstore, power7-fpstore-update):
2448         Adjust.
2449         * config/rs6000/power8.md (power8-load, power8-load-update,
2450         power8-load-ext, power8-load-ext-update, power8-fpload,
2451         power8-fpload-update, power8-store, power8-store-update-indexed,
2452         power8-fpstore, power8-fpstore-update): Adjust.
2453         * config/rs6000/rs64.md (rs64a-load, rs64a-store, rs64a-fpload):
2454         Adjust.
2455         * config/rs6000/titan.md (titan_lsu_load, titan_lsu_fpload,
2456         titan_lsu_store, titan_lsu_fpstore): Adjust.
2457         * config/rs6000/xfpu.md (fp-load, fp-store): Adjust.
2458
2459 2014-05-07  Oleg Endo  <olegendo@gcc.gnu.org>
2460
2461         PR target/60884
2462         * config/sh/sh-mem.cc (sh_expand_strlen): Use loop when emitting
2463         unrolled byte insns.  Emit address increments after move insns.
2464
2465 2014-05-07  David Malcolm  <dmalcolm@redhat.com>
2466
2467         * gimple.h (gimple_builtin_call_types_compatible_p): Accept a
2468         const_gimple, rather than a gimple.
2469         (gimple_call_builtin_p): Likewise, for the three variants.
2470
2471         * gimple.c (gimple_builtin_call_types_compatible_p): Likewise.
2472         (gimple_call_builtin_p): Likewise, for the three variants.
2473
2474 2014-05-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
2475
2476         PR tree-optimization/61095
2477         * tree-ssanames.c (get_nonzero_bits): Fix type extension in wi::shwi.
2478
2479 2014-05-07  Richard Biener  <rguenther@suse.de>
2480
2481         PR tree-optimization/61034
2482         * tree-ssa-alias.c (call_may_clobber_ref_p_1): Export.
2483         (maybe_skip_until): Use translate to take into account
2484         lattices when trying to do disambiguations.
2485         (get_continuation_for_phi_1): Likewise.
2486         (get_continuation_for_phi): Adjust for added translate arguments.
2487         (walk_non_aliased_vuses): Likewise.
2488         * tree-ssa-alias.h (get_continuation_for_phi): Adjust prototype.
2489         (walk_non_aliased_vuses): Likewise.
2490         (call_may_clobber_ref_p_1): Declare.
2491         * tree-ssa-sccvn.c (vn_reference_lookup_3): Also disambiguate against
2492         calls.  Stop early if we are only supposed to disambiguate.
2493         * tree-ssa-pre.c (translate_vuse_through_block): Adjust.
2494
2495 2014-05-07  Joern Rennecke  <joern.rennecke@embecosm.com>
2496
2497         * config/epiphany/epiphany.c (epiphany_handle_interrupt_attribute):
2498         Emit an error when the function has arguments.
2499
2500 2014-05-07  Thomas Schwinge  <thomas@codesourcery.com>
2501
2502         * cfgloop.h (unswitch_loops): Remove.
2503         * doc/passes.texi: Remove references to loop-unswitch.c
2504         * timevar.def (TV_LOOP_UNSWITCH): Remove.
2505
2506 2014-05-07  Evgeny Stupachenko  <evstupac@gmail.com>
2507
2508         * tree-vect-data-refs.c (vect_grouped_load_supported): New
2509         check for loads group of length 3.
2510         (vect_permute_load_chain): New permutations for loads group of
2511         length 3.
2512         * tree-vect-stmts.c (vect_model_load_cost): Change cost
2513         of vec_perm_shuffle for the new permutations.
2514
2515 2014-05-07  Alan Lawrence  <alan.lawrence@arm.com>
2516
2517         * config/aarch64/arm_neon.h (vtrn1_f32, vtrn1_p8, vtrn1_p16, vtrn1_s8,
2518         vtrn1_s16, vtrn1_s32, vtrn1_u8, vtrn1_u16, vtrn1_u32, vtrn1q_f32,
2519         vtrn1q_f64, vtrn1q_p8, vtrn1q_p16, vtrn1q_s8, vtrn1q_s16, vtrn1q_s32,
2520         vtrn1q_s64, vtrn1q_u8, vtrn1q_u16, vtrn1q_u32, vtrn1q_u64, vtrn2_f32,
2521         vtrn2_p8, vtrn2_p16, vtrn2_s8, vtrn2_s16, vtrn2_s32, vtrn2_u8,
2522         vtrn2_u16, vtrn2_u32, vtrn2q_f32, vtrn2q_f64, vtrn2q_p8, vtrn2q_p16,
2523         vtrn2q_s8, vtrn2q_s16, vtrn2q_s32, vtrn2q_s64, vtrn2q_u8, vtrn2q_u16,
2524         vtrn2q_u32, vtrn2q_u64): Replace temporary asm with __builtin_shuffle.
2525
2526 2014-05-07  Thomas Schwinge  <thomas@codesourcery.com>
2527
2528         * loop-unswitch.c: Delete.
2529
2530 2014-05-07  Richard Biener  <rguenther@suse.de>
2531
2532         * config.gcc: Always set need_64bit_hwint to yes.
2533
2534 2014-05-07  Chung-Ju Wu  <jasonwucj@gmail.com>
2535
2536         * config/nds32/nds32.h (HONOR_REG_ALLOC_ORDER): Have it in favor
2537         of using optimize_size.
2538
2539 2014-05-06  Mike Stump  <mikestump@comcast.net>
2540
2541         * wide-int.h (wi::int_traits <HOST_WIDE_INT>): Always define.
2542
2543 2014-05-06  Joseph Myers  <joseph@codesourcery.com>
2544
2545         * config/i386/sse.md (*mov<mode>_internal)
2546         (*<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>)
2547         (*<sse2_avx_avx512f>_loaddqu<mode><mask_name>)
2548         (<sse>_andnot<mode>3, <code><mode>3, *andnot<mode>3)
2549         (*<code><mode>3, *andnot<mode>3<mask_name>)
2550         (<mask_codefor><code><mode>3<mask_name>): Only consider
2551         TARGET_SSE_PACKED_SINGLE_INSN_OPTIMAL for modes of size 16.
2552
2553 2014-05-06  Richard Sandiford  <rdsandiford@googlemail.com>
2554
2555         Revert:
2556         2014-05-03  Richard Sandiford  <rdsandiford@googlemail.com>
2557
2558         * lra-constraints.c (valid_address_p): Move earlier in file.
2559         Add a constraint argument to the address_info version.
2560         (satisfies_memory_constraint_p): New function.
2561         (satisfies_address_constraint_p): Likewise.
2562         (process_alt_operands, curr_insn_transform): Use them.
2563         (process_address): Pass the constraint to valid_address_p when
2564         checking address operands.
2565
2566 2014-05-06  Richard Sandiford  <r.sandiford@uk.ibm.com>
2567
2568         * lto-cgraph.c (compute_ltrans_boundary): Make node variables local
2569         to their respective blocks.  Fix inadvertent use of "node".
2570
2571 2014-05-06  Richard Sandiford  <rdsandiford@googlemail.com>
2572
2573         * emit-rtl.c (init_derived_machine_modes): New functionm, split
2574         out from...
2575         (init_emit_once): ...here.
2576         * rtl.h (init_derived_machine_modes): Declare.
2577         * toplev.c (do_compile): Call it even if no_backend.
2578
2579 2014-05-06  Kenneth Zadeck  <zadeck@naturalbridge.com>
2580             Mike Stump  <mikestump@comcast.net>
2581             Richard Sandiford  <rdsandiford@googlemail.com>
2582             Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2583
2584         * alias.c (ao_ref_from_mem): Use wide-int interfaces.
2585         (rtx_equal_for_memref_p): Update comment.
2586         (adjust_offset_for_component_ref): Use wide-int interfaces.
2587         * builtins.c (get_object_alignment_2): Likewise.
2588         (c_readstr): Likewise.
2589         (target_char_cast): Add comment.
2590         (determine_block_size): Use wide-int interfaces.
2591         (expand_builtin_signbit): Likewise.
2592         (fold_builtin_int_roundingfn): Likewise.
2593         (fold_builtin_bitop): Likewise.
2594         (fold_builtin_bswap): Likewise.
2595         (fold_builtin_logarithm): Use signop.
2596         (fold_builtin_pow): Likewise.
2597         (fold_builtin_memory_op): Use wide-int interfaces.
2598         (fold_builtin_object_size): Likewise.
2599         * cfgloop.c (alloc_loop): Initialize nb_iterations_upper_bound and
2600         nb_iterations_estimate.
2601         (record_niter_bound): Use wide-int interfaces.
2602         (get_estimated_loop_iterations_int): Likewise.
2603         (get_estimated_loop_iterations): Likewise.
2604         (get_max_loop_iterations): Likewise.
2605         * cfgloop.h: Include wide-int.h.
2606         (struct nb_iter_bound): Change bound to widest_int.
2607         (struct loop): Change nb_iterations_upper_bound and
2608         nb_iterations_estimate to widest_int.
2609         (record_niter_bound): Switch to use widest_int.
2610         (get_estimated_loop_iterations): Likewise.
2611         (get_max_loop_iterations): Likewise.
2612         (gcov_type_to_double_int): Rename to gcov_type_to_wide_int and
2613         update for wide-int.
2614         * cgraph.c (cgraph_add_thunk): Use wide-int interfaces.
2615         * combine.c (try_combine): Likewise.
2616         (subst): Use CONST_SCALAR_INT_P rather than CONST_INT_P.
2617         * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Use wide-int
2618         interfaces.
2619         (aarch64_float_const_representable_p): Likewise.
2620         * config/arc/arc.c: Include wide-int.h.
2621         (arc_can_use_doloop_p): Use wide-int interfaces.
2622         * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
2623         (vfp3_const_double_index): Likewise.
2624         * config/avr/avr.c (avr_out_round): Likewise.
2625         (avr_fold_builtin): Likewise.
2626         * config/bfin/bfin.c (bfin_local_alignment): Likewise.
2627         (bfin_can_use_doloop_p): Likewise.
2628         * config/darwin.c (darwin_mergeable_constant_section): Likewise.
2629         (machopic_select_rtx_section): Update to handle CONST_WIDE_INT.
2630         * config/i386/i386.c: Include wide-int.h.
2631         (ix86_data_alignment): Use wide-int interfaces.
2632         (ix86_local_alignment): Likewise.
2633         (ix86_emit_swsqrtsf): Update real_from_integer.
2634         * config/msp430/msp430.c (msp430_attr): Use wide-int interfaces.
2635         * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
2636         * config/rs6000/predicates.md (any_operand): Add const_wide_int.
2637         (zero_constant): Likewise.
2638         (input_operand): Likewise.
2639         (splat_input_operand): Likewise.
2640         (non_logical_cint_operand): Change const_double to const_wide_int.
2641         * config/rs6000/rs6000.c (num_insns_constant): Handle CONST_WIDE_INT.
2642         (easy_altivec_constant): Remove comment.
2643         (paired_expand_vector_init): Use CONSTANT_P.
2644         (rs6000_legitimize_address): Handle CONST_WIDE_INT.
2645         (rs6000_emit_move): Update checks.
2646         (rs6000_aggregate_candidate): Use wide-int interfaces.
2647         (rs6000_expand_ternop_builtin): Likewise.
2648         (rs6000_output_move_128bit): Handle CONST_WIDE_INT.
2649         (rs6000_assemble_integer): Likewise.
2650         (rs6000_hash_constant): Likewise.
2651         (output_toc): Likewise.
2652         (rs6000_rtx_costs): Likewise.
2653         (rs6000_emit_swrsqrt); Update call to real_from_integer.
2654         * config/rs6000/rs6000-c.c: Include wide-int.h.
2655         (altivec_resolve_overloaded_builtin): Use wide-int interfaces.
2656         * config/rs6000/rs6000.h (TARGET_SUPPORTS_WIDE_INT): New.
2657         * config/rs6000/rs6000.md: Use const_scalar_int_operand.
2658         Handle CONST_WIDE_INT.
2659         * config/sol2-c.c (solaris_pragma_align): Change low to unsigned HWI.
2660         Use tree_fits_uhwi_p.
2661         * config/sparc/sparc.c: Include wide-int.h.
2662         (sparc_fold_builtin): Use wide-int interfaces.
2663         * config/vax/vax.c: Include wide-int.h.
2664         (vax_float_literal): Use real_from_integer.
2665         * coretypes.h (struct hwivec_def): New.
2666         (hwivec): New.
2667         (const_hwivec): New.
2668         * cse.c (hash_rtx_cb): Handle CONST_WIDE_INT.
2669         (equiv_constant): Handle CONST_WIDE_INT.
2670         * cselib.c (rtx_equal_for_cselib_1): Use CASE_CONST_UNIQUE.
2671         (cselib_hash_rtx): Handle CONST_WIDE_INT.
2672         * dbxout.c (stabstr_U): Use wide-int interfaces.
2673         (dbxout_type): Update to use cst_fits_shwi_p.
2674         * defaults.h (LOG2_BITS_PER_UNIT): Define.
2675         (TARGET_SUPPORTS_WIDE_INT): Add default.
2676         * dfp.c: Include wide-int.h.
2677         (decimal_real_to_integer2): Use wide-int interfaces and rename to
2678         decimal_real_to_integer.
2679         * dfp.h (decimal_real_to_integer2): Return a wide_int and rename to
2680         decimal_real_to_integer.
2681         * doc/generic.texi (Constant expressions): Update for wide_int.
2682         * doc/rtl.texi (const_double): Likewise.
2683         (const_wide_int, CONST_WIDE_INT, CONST_WIDE_INT_VEC): New.
2684         (CONST_WIDE_INT_NUNITS, CONST_WIDE_INT_ELT): New.
2685         * doc/tm.texi.in (REAL_VALUE_TO_INT): Remove.
2686         (REAL_VALUE_FROM_INT): Remove.
2687         (TARGET_SUPPORTS_WIDE_INT): New.
2688         * doc/tm.texi: Regenerate.
2689         * dojump.c (prefer_and_bit_test): Use wide-int interfaces.
2690         * double-int.h: Include wide-int.h.
2691         (struct wi::int_traits): New.
2692         * dwarf2out.c (get_full_len): New.
2693         (dw_val_equal_p): Add case dw_val_class_wide_int.
2694         (size_of_loc_descr): Likewise.
2695         (output_loc_operands): Likewise.
2696         (insert_double): Remove.
2697         (insert_wide_int): New.
2698         (add_AT_wide): New.
2699         (print_die): Add case dw_val_class_wide_int.
2700         (attr_checksum): Likewise.
2701         (attr_checksum_ordered): Likewise.
2702         (same_dw_val_p): Likewise.
2703         (size_of_die): Likewise.
2704         (value_format): Likewise.
2705         (output_die): Likewise.
2706         (double_int_type_size_in_bits): Rename to offset_int_type_size_in_bits.
2707         Use wide-int.
2708         (clz_loc_descriptor): Use wide-int interfaces.
2709         (mem_loc_descriptor): Likewise.  Handle CONST_WIDE_INT.
2710         (loc_descriptor): Use wide-int interfaces.  Handle CONST_WIDE_INT.
2711         (round_up_to_align): Use wide-int interfaces.
2712         (field_byte_offset): Likewise.
2713         (insert_double): Rename to insert_wide_int.  Use wide-int interfaces.
2714         (add_const_value_attribute): Handle CONST_WIDE_INT.  Update
2715         CONST_DOUBLE handling.  Use wide-int interfaces.
2716         (add_bound_info): Use tree_fits_uhwi_p.  Use wide-int interfaces.
2717         (gen_enumeration_type_die): Use add_AT_wide.
2718         (hash_loc_operands): Add case dw_val_class_wide_int.
2719         (compare_loc_operands): Likewise.
2720         * dwarf2out.h: Include wide-int.h.
2721         (wide_int_ptr): New.
2722         (enum dw_val_class): Add dw_val_class_wide_int.
2723         (struct dw_val_struct): Add val_wide.
2724         * emit-rtl.c (const_wide_int_htab): New.
2725         (const_wide_int_htab_hash): New.
2726         (const_wide_int_htab_eq): New.
2727         (lookup_const_wide_int): New.
2728         (const_double_htab_hash): Use wide-int interfaces.
2729         (const_double_htab_eq): Likewise.
2730         (rtx_to_double_int): Conditionally compile for wide-int.
2731         (immed_double_int_const): Rename to immed_wide_int_const and
2732         update for wide-int.
2733         (immed_double_const): Conditionally compile for wide-int.
2734         (init_emit_once): Use wide-int interfaces.
2735         * explow.c (plus_constant): Likewise.
2736         * expmed.c (mask_rtx): Move further up file.  Use wide-int interfaces.
2737         (lshift_value): Use wide-int interfaces.
2738         (expand_mult): Likewise.
2739         (choose_multiplier): Likewise.
2740         (expand_smod_pow2): Likewise.
2741         (make_tree): Likewise.
2742         * expr.c (convert_modes): Consolidate handling of constants.
2743         Use wide-int interfaces.
2744         (emit_group_load_1): Add note.
2745         (store_expr): Update comment.
2746         (get_inner_reference): Use wide-int interfaces.
2747         (expand_constructor): Update comment.
2748         (expand_expr_real_2): Use wide-int interfaces.
2749         (expand_expr_real_1): Likewise.
2750         (reduce_to_bit_field_precision): Likewise.
2751         (const_vector_from_tree): Likewise.
2752         * final.c: Include wide-int-print.h.
2753         (output_addr_const): Handle CONST_WIDE_INT.  Use CONST_DOUBLE_AS_INT_P.
2754         * fixed-value.c: Include wide-int.h.
2755         (fixed_from_string): Use wide-int interfaces.
2756         (fixed_to_decimal): Likewise.
2757         (fixed_convert_from_real): Likewise.
2758         (real_convert_from_fixed): Likewise.
2759         * fold-const.h (mem_ref_offset): Return an offset_int.
2760         (div_if_zero_remainder): Remove code parameter.
2761         * fold-const.c (div_if_zero_remainder): Remove code parameter.
2762         Use wide-int interfaces.
2763         (may_negate_without_overflow_p): Use wide-int interfaces.
2764         (negate_expr_p): Likewise.
2765         (fold_negate_expr): Likewise.
2766         (int_const_binop_1): Likewise.
2767         (const_binop): Likewise.
2768         (fold_convert_const_int_from_int): Likewise.
2769         (fold_convert_const_int_from_real): Likewise.
2770         (fold_convert_const_int_from_fixed): Likewise.
2771         (fold_convert_const_fixed_from_int): Likewise.
2772         (all_ones_mask_p): Take an unsigned size.  Use wide-int interfaces.
2773         (sign_bit_p): Use wide-int interfaces.
2774         (make_range_step): Likewise.
2775         (build_range_check): Likewise.  Pass an integer of the correct type
2776         instead of using integer_one_node.
2777         (range_predecessor): Pass an integer of the correct type instead
2778         of using integer_one_node.
2779         (range_successor): Likewise.
2780         (merge_ranges): Likewise.
2781         (unextend): Use wide-int interfaces.
2782         (extract_muldiv_1): Likewise.
2783         (fold_div_compare): Likewise.
2784         (fold_single_bit_test): Likewise.
2785         (fold_sign_changed_comparison): Likewise.
2786         (try_move_mult_to_index): Update calls to div_if_zero_remainder.
2787         (fold_plusminus_mult_expr): Use wide-int interfaces.
2788         (native_encode_int): Likewise.
2789         (native_interpret_int): Likewise.
2790         (fold_unary_loc): Likewise.
2791         (pointer_may_wrap_p): Likewise.
2792         (size_low_cst): Likewise.
2793         (mask_with_tz): Likewise.
2794         (fold_binary_loc): Likewise.
2795         (fold_ternary_loc): Likewise.
2796         (multiple_of_p): Likewise.
2797         (tree_call_nonnegative_warnv_p): Update calls to
2798         tree_int_cst_min_precision and real_from_integer.
2799         (fold_negate_const): Use wide-int interfaces.
2800         (fold_abs_const): Likewise.
2801         (fold_relational_const): Use tree_int_cst_lt.
2802         (round_up_loc): Use wide-int interfaces.
2803         * genemit.c (gen_exp): Add CONST_WIDE_INT case.
2804         * gengenrtl.c (excluded_rtx): Add CONST_WIDE_INT case.
2805         * gengtype.c: Remove include of double-int.h.
2806         (do_typedef): Use wide-int interfaces.
2807         (open_base_files): Add wide-int.h.
2808         (main): Add offset_int and widest_int typedefs.
2809         * gengtype-lex.l: Handle "^".
2810         (CXX_KEYWORD): Add "static".
2811         * gengtype-parse.c (require3): New.
2812         (require_template_declaration): Handle constant template arguments
2813         and nested templates.
2814         * gengtype-state.c: Don't include "double-int.h".
2815         * genpreds.c (write_one_predicate_function): Update comment.
2816         (write_tm_constrs_h): Add check for hval and lval use in
2817         CONST_WIDE_INT.
2818         * genrecog.c (validate_pattern): Add CONST_WIDE_INT case.
2819         (add_to_sequence): Likewise.
2820         * gensupport.c (struct std_pred_table): Add const_scalar_int_operand
2821         and const_double_operand.
2822         * gimple.c (preprocess_case_label_vec_for_gimple): Use wide-int
2823         interfaces.
2824         * gimple-fold.c (get_base_constructor): Likewise.
2825         (fold_array_ctor_reference): Likewise.
2826         (fold_nonarray_ctor_reference): Likewise.
2827         (fold_const_aggregate_ref_1): Likewise.
2828         (gimple_val_nonnegative_real_p): Likewise.
2829         (gimple_fold_indirect_ref): Likewise.
2830         * gimple-pretty-print.c (dump_ssaname_info): Likewise.
2831         * gimple-ssa-strength-reduction.c: Include wide-int-print.h.
2832         (struct slsr_cand_d): Change index to be widest_int.
2833         (struct incr_info_d): Change incr to be widest_int.
2834         (alloc_cand_and_find_basis): Use wide-int interfaces.
2835         (slsr_process_phi): Likewise.
2836         (backtrace_base_for_ref): Likewise.  Return a widest_int.
2837         (restructure_reference): Take a widest_int instead of a double_int.
2838         (slsr_process_ref): Use wide-int interfaces.
2839         (create_mul_ssa_cand): Likewise.
2840         (create_mul_imm_cand): Likewise.
2841         (create_add_ssa_cand): Likewise.
2842         (create_add_imm_cand): Take a widest_int instead of a double_int.
2843         (slsr_process_add): Use wide-int interfaces.
2844         (slsr_process_cast): Likewise.
2845         (slsr_process_copy): Likewise.
2846         (dump_candidate): Likewise.
2847         (dump_incr_vec): Likewise.
2848         (replace_ref): Likewise.
2849         (cand_increment): Likewise.  Return a widest_int.
2850         (cand_abs_increment): Likewise.
2851         (replace_mult_candidate): Take a widest_int instead of a double_int.
2852         (replace_unconditional_candidate): Use wide-int interfaces.
2853         (incr_vec_index): Take a widest_int instead of a double_int.
2854         (create_add_on_incoming_edge): Likewise.
2855         (create_phi_basis): Use wide-int interfaces.
2856         (replace_conditional_candidate): Likewise.
2857         (record_increment): Take a widest_int instead of a double_int.
2858         (record_phi_increments): Use wide-int interfaces.
2859         (phi_incr_cost): Take a widest_int instead of a double_int.
2860         (lowest_cost_path): Likewise.
2861         (total_savings): Likewise.
2862         (analyze_increments): Use wide-int interfaces.
2863         (ncd_with_phi): Take a widest_int instead of a double_int.
2864         (ncd_of_cand_and_phis): Likewise.
2865         (nearest_common_dominator_for_cands): Likewise.
2866         (insert_initializers): Use wide-int interfaces.
2867         (all_phi_incrs_profitable): Likewise.
2868         (replace_one_candidate): Likewise.
2869         (replace_profitable_candidates): Likewise.
2870         * godump.c: Include wide-int-print.h.
2871         (go_output_typedef): Use wide-int interfaces.
2872         * graphite-clast-to-gimple.c (gmp_cst_to_tree): Likewise.
2873         * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
2874         (build_loop_iteration_domains): Likewise.
2875         * hooks.h: Include wide-int.h rather than double-int.h.
2876         (hook_bool_dint_dint_uint_bool_true): Delete.
2877         (hook_bool_wint_wint_uint_bool_true): Declare.
2878         * hooks.c (hook_bool_dint_dint_uint_bool_true): Removed.
2879         (hook_bool_wint_wint_uint_bool_true): New.
2880         * internal-fn.c (ubsan_expand_si_overflow_addsub_check): Use wide-int
2881         interfaces.
2882         (ubsan_expand_si_overflow_mul_check): Likewise.
2883         * ipa-devirt.c (get_polymorphic_call_info): Likewise.
2884         * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
2885         (get_ancestor_addr_info): Likewise.
2886         (ipa_modify_call_arguments): Likewise.
2887         * loop-doloop.c (doloop_modify): Likewise.
2888         (doloop_optimize): Likewise.
2889         * loop-iv.c (iv_number_of_iterations): Likewise.
2890         * loop-unroll.c (decide_unroll_constant_iterations): Likewise.
2891         (unroll_loop_constant_iterations): Likewise.
2892         (decide_unroll_runtime_iterations): Likewise.
2893         (unroll_loop_runtime_iterations): Likewise.
2894         (decide_peel_simple): Likewise.
2895         (decide_unroll_stupid): Likewise.
2896         * lto-streamer-in.c (streamer_read_wi): Add.
2897         (input_cfg): Use wide-int interfaces.
2898         (lto_input_tree_1): Likewise.
2899         * lto-streamer-out.c (streamer_write_wi): Add.
2900         (hash_tree): Use wide-int interfaces.
2901         (output_cfg): Likewise.
2902         * Makefile.in (OBJS): Add wide-int.o and wide-int-print.o.
2903         (GTFILES): Add wide-int.h and signop.h.
2904         (TAGS): Look for .cc files too.
2905         * omp-low.c (scan_omp_1_op): Use wide-int interfaces.
2906         * optabs.c (expand_subword_shift): Likewise.
2907         (expand_doubleword_shift): Likewise.
2908         (expand_absneg_bit): Likewise.
2909         (expand_copysign_absneg): Likewise.
2910         (expand_copysign_bit): Likewise.
2911         * postreload.c (reload_cse_simplify_set): Likewise.
2912         * predict.c (predict_iv_comparison): Likewise.
2913         * pretty-print.h: Include wide-int-print.h.
2914         (pp_wide_int) New.
2915         * print-rtl.c (print_rtx): Add CONST_WIDE_INT case.
2916         * print-tree.c: Include wide-int-print.h.
2917         (print_node_brief): Use wide-int interfaces.
2918         (print_node): Likewise.
2919         * read-rtl.c (validate_const_wide_int): New.
2920         (read_rtx_code): Add CONST_WIDE_INT case.
2921         * real.c: Include wide-int.h.
2922         (real_to_integer2): Delete.
2923         (real_to_integer): New function, returning a wide_int.
2924         (real_from_integer): Take a wide_int rather than two HOST_WIDE_INTs.
2925         (ten_to_ptwo): Update call to real_from_integer.
2926         (real_digit): Likewise.
2927         * real.h: Include signop.h, wide-int.h and insn-modes.h.
2928         (real_to_integer2, REAL_VALUE_FROM_INT, REAL_VALUE_FROM_UNSIGNED_INT)
2929         (REAL_VALUE_TO_INT): Delete.
2930         (real_to_integer): Declare a wide-int form.
2931         (real_from_integer): Take a wide_int rather than two HOST_WIDE_INTs.
2932         * recog.c (const_int_operand): Improve comment.
2933         (const_scalar_int_operand): New.
2934         (const_double_operand): Add a separate definition for CONST_WIDE_INT.
2935         * rtlanal.c (commutative_operand_precedence): Handle CONST_WIDE_INT.
2936         (split_double): Likewise.
2937         * rtl.c (DEF_RTL_EXPR): Handle CONST_WIDE_INT.
2938         (rtx_size): Likewise.
2939         (rtx_alloc_stat_v): New.
2940         (rtx_alloc_stat): Now calls rtx_alloc_stat_v.
2941         (cwi_output_hex): New.
2942         (iterative_hash_rtx): Handle CONST_WIDE_INT.
2943         (cwi_check_failed_bounds): New.
2944         * rtl.def (CONST_WIDE_INT): New.
2945         * rtl.h: Include <utility> and wide-int.h.
2946         (struct hwivec_def): New.
2947         (CWI_GET_NUM_ELEM): New.
2948         (CWI_PUT_NUM_ELEM): New.
2949         (struct rtx_def): Add num_elem and hwiv.
2950         (CASE_CONST_SCALAR_INT): Modify for TARGET_SUPPORTS_WIDE_INT.
2951         (CASE_CONST_UNIQUE): Likewise.
2952         (CASE_CONST_ANY): Likewise.
2953         (CONST_SCALAR_INT_P): Likewise.
2954         (CONST_WIDE_INT_P): New.
2955         (CWI_ELT): New.
2956         (HWIVEC_CHECK): New.
2957         (cwi_check_failed_bounds): New.
2958         (CWI_ELT): New.
2959         (HWIVEC_CHECK): New.
2960         (CONST_WIDE_INT_VEC) New.
2961         (CONST_WIDE_INT_NUNITS) New.
2962         (CONST_WIDE_INT_ELT) New.
2963         (rtx_mode_t): New type.
2964         (wi::int_traits <rtx_mode_t>): New.
2965         (wi::shwi): New.
2966         (wi::min_value): New.
2967         (wi::max_value): New.
2968         (rtx_alloc_v) New.
2969         (const_wide_int_alloc): New.
2970         (immed_wide_int_const): New.
2971         * sched-vis.c (print_value): Handle CONST_WIDE_INT.
2972         * sel-sched-ir.c (lhs_and_rhs_separable_p): Update comment.
2973         * signop.h: New file.
2974         * simplify-rtx.c (mode_signbit_p): Handle CONST_WIDE_INT.
2975         (simplify_const_unary_operation): Use wide-int interfaces.
2976         (simplify_binary_operation_1): Likewise.
2977         (simplify_const_binary_operation): Likewise.
2978         (simplify_const_relational_operation): Likewise.
2979         (simplify_immed_subreg): Likewise.
2980         * stmt.c (expand_case): Likewise.
2981         * stor-layout.h (set_min_and_max_values_for_integral_type): Take a
2982         signop rather than a bool.
2983         * stor-layout.c (layout_type): Use wide-int interfaces.
2984         (initialize_sizetypes): Update calls to
2985         set_min_and_max_values_for_integral_type.
2986         (set_min_and_max_values_for_integral_type): Take a signop rather
2987         than a bool.  Use wide-int interfaces.
2988         (fixup_signed_type): Update accordingly.  Remove
2989         HOST_BITS_PER_DOUBLE_INT limit.
2990         (fixup_unsigned_type): Likewise.
2991         * system.h (STATIC_CONSTANT_P): New.
2992         (STATIC_ASSERT): New.
2993         * target.def (can_use_doloop_p): Take widest_ints rather than
2994         double_ints.
2995         * target.h: Include wide-int.h rather than double-int.h.
2996         * targhooks.h (can_use_doloop_if_innermost): Take widest_ints rather
2997         than double_ints.
2998         * targhooks.c (default_cxx_get_cookie_size): Use tree_int_cst_lt
2999         rather than INT_CST_LT_UNSIGNED.
3000         (can_use_doloop_if_innermost): Take widest_ints rather than
3001         double_ints.
3002         * tree-affine.c: Include wide-int-print.h.
3003         (double_int_ext_for_comb): Delete.
3004         (wide_int_ext_for_comb): New.
3005         (aff_combination_zero): Use wide-int interfaces.
3006         (aff_combination_const): Take a widest_int instead of a double_int.
3007         (aff_combination_elt): Use wide-int interfaces.
3008         (aff_combination_scale): Take a widest_int instead of a double_int.
3009         (aff_combination_add_elt): Likewise.
3010         (aff_combination_add_cst): Likewise.
3011         (aff_combination_add): Use wide-int interfaces.
3012         (aff_combination_convert): Likewise.
3013         (tree_to_aff_combination): Likewise.
3014         (add_elt_to_tree): Take a widest_int instead of a double_int.
3015         (aff_combination_to_tree): Use wide-int interfaces.
3016         (aff_combination_remove_elt): Likewise.
3017         (aff_combination_add_product): Take a widest_int instead of
3018         a double_int.
3019         (aff_combination_mult): Use wide-int interfaces.
3020         (aff_combination_expand): Likewise.
3021         (double_int_constant_multiple_p): Delete.
3022         (wide_int_constant_multiple_p): New.
3023         (aff_combination_constant_multiple_p): Take a widest_int pointer
3024         instead of a double_int pointer.
3025         (print_aff): Use wide-int interfaces.
3026         (get_inner_reference_aff): Take a widest_int pointer
3027         instead of a double_int pointer.
3028         (aff_comb_cannot_overlap_p): Take widest_ints instead of double_ints.
3029         * tree-affine.h: Include wide-int.h.
3030         (struct aff_comb_elt): Change type of coef to widest_int.
3031         (struct affine_tree_combination): Change type of offset to widest_int.
3032         (double_int_ext_for_comb): Delete.
3033         (wide_int_ext_for_comb): New.
3034         (aff_combination_const): Use widest_int instead of double_int.
3035         (aff_combination_scale): Likewise.
3036         (aff_combination_add_elt): Likewise.
3037         (aff_combination_constant_multiple_p): Likewise.
3038         (get_inner_reference_aff): Likewise.
3039         (aff_comb_cannot_overlap_p): Likewise.
3040         (aff_combination_zero_p): Use wide-int interfaces.
3041         * tree.c: Include tree.h.
3042         (init_ttree): Use make_int_cst.
3043         (tree_code_size): Removed code for INTEGER_CST case.
3044         (tree_size): Add INTEGER_CST case.
3045         (make_node_stat): Update comment.
3046         (get_int_cst_ext_nunits, build_new_int_cst, build_int_cstu): New.
3047         (build_int_cst_type): Use wide-int interfaces.
3048         (double_int_to_tree): Likewise.
3049         (double_int_fits_to_tree_p): Delete.
3050         (force_fit_type_double): Delete.
3051         (force_fit_type): New.
3052         (int_cst_hash_hash): Use wide-int interfaces.
3053         (int_cst_hash_eq): Likewise.
3054         (build_int_cst_wide): Delete.
3055         (wide_int_to_tree): New.
3056         (cache_integer_cst): Use wide-int interfaces.
3057         (build_low_bits_mask): Likewise.
3058         (cst_and_fits_in_hwi): Likewise.
3059         (real_value_from_int_cst): Likewise.
3060         (make_int_cst_stat): New.
3061         (integer_zerop): Use wide_int interfaces.
3062         (integer_onep): Likewise.
3063         (integer_all_onesp): Likewise.
3064         (integer_pow2p): Likewise.
3065         (integer_nonzerop): Likewise.
3066         (tree_log2): Likewise.
3067         (tree_floor_log2): Likewise.
3068         (tree_ctz): Likewise.
3069         (int_size_in_bytes): Likewise.
3070         (mem_ref_offset): Return an offset_int rather than a double_int.
3071         (build_type_attribute_qual_variant): Use wide_int interfaces.
3072         (type_hash_eq): Likewise
3073         (tree_int_cst_equal): Likewise.
3074         (tree_int_cst_lt): Delete.
3075         (tree_int_cst_compare): Likewise.
3076         (tree_fits_shwi_p): Use wide_int interfaces.
3077         (tree_fits_uhwi_p): Likewise.
3078         (tree_int_cst_sign_bit): Likewise.
3079         (tree_int_cst_sgn): Likewise.
3080         (tree_int_cst_min_precision): Take a signop rather than a bool.
3081         (simple_cst_equal): Use wide_int interfaces.
3082         (compare_tree_int): Likewise.
3083         (iterative_hash_expr): Likewise.
3084         (int_fits_type_p): Likewise.  Use tree_int_cst_lt rather than
3085         INT_CST_LT.
3086         (get_type_static_bounds): Use wide_int interfaces.
3087         (tree_int_cst_elt_check_failed): New.
3088         (build_common_tree_nodes): Reordered to set prec before filling in
3089         value.
3090         (int_cst_value): Check cst_and_fits_in_hwi.
3091         (widest_int_cst_value): Use wide_int interfaces.
3092         (upper_bound_in_type): Likewise.
3093         (lower_bound_in_type): Likewise.
3094         (num_ending_zeros): Likewise.
3095         (drop_tree_overflow): Likewise.
3096         * tree-call-cdce.c (check_pow): Update call to real_from_integer.
3097         (gen_conditions_for_pow_cst_base): Likewise.
3098         * tree-cfg.c: Include wide-int.h and wide-int-print.h.
3099         (group_case_labels_stmt): Use wide-int interfaces.
3100         (verify_gimple_assign_binary): Likewise.
3101         (print_loop): Likewise.
3102         * tree-chrec.c (tree_fold_binomial): Likewise.
3103         * tree-core.h (struct tree_base): Add int_length.
3104         (struct tree_int_cst): Change rep of value.
3105         * tree-data-ref.c (dr_analyze_innermost): Use wide-int interfaces.
3106         (dr_may_alias_p): Likewise.
3107         (max_stmt_executions_tree): Likewise.
3108         * tree.def (INTEGER_CST): Update comment.
3109         * tree-dfa.c (get_ref_base_and_extent): Use wide-int interfaces.
3110         * tree-dfa.h (get_addr_base_and_unit_offset_1): Likewise.
3111         * tree-dump.c: Include wide-int.h and wide-int-print.h.
3112         (dequeue_and_dump): Use wide-int interfaces.
3113         * tree.h: Include wide-int.h.
3114         (NULL_TREE): Moved to earlier loc in file.
3115         (TREE_INT_CST_ELT_CHECK): New.
3116         (tree_int_cst_elt_check_failed): New.
3117         (TYPE_SIGN): New.
3118         (TREE_INT_CST): Delete.
3119         (TREE_INT_CST_LOW): Use wide-int interfaces.
3120         (TREE_INT_CST_HIGH): Delete.
3121         (TREE_INT_CST_NUNITS): New.
3122         (TREE_INT_CST_EXT_NUNITS): Likewise.
3123         (TREE_INT_CST_OFFSET_NUNITS): Likewise.
3124         (TREE_INT_CST_ELT): Likewise.
3125         (INT_CST_LT): Delete.
3126         (tree_int_cst_elt_check): New (two forms).
3127         (type_code_size): Update comment.
3128         (make_int_cst_stat, make_int_cst): New.
3129         (tree_to_double_int): Delete.
3130         (double_int_fits_to_tree_p): Delete.
3131         (force_fit_type_double): Delete.
3132         (build_int_cstu): Replace with out-of-line function.
3133         (build_int_cst_wide): Delete.
3134         (tree_int_cst_lt): Define inline.
3135         (tree_int_cst_le): New.
3136         (tree_int_cst_compare): Define inline.
3137         (tree_int_cst_min_precision): Take a signop rather than a bool.
3138         (wi::int_traits <const_tree>): New.
3139         (wi::int_traits <tree>): New.
3140         (wi::extended_tree): New.
3141         (wi::int_traits <wi::extended_tree>): New.
3142         (wi::to_widest): New.
3143         (wi::to_offset): New.
3144         (wi::fits_to_tree_p): New.
3145         (wi::min_value): New.
3146         (wi::max_value): New.
3147         * tree-inline.c (remap_gimple_op_r): Use wide-int interfaces.
3148         (copy_tree_body_r): Likewise.
3149         * tree-object-size.c (compute_object_offset): Likewise.
3150         (addr_object_size): Likewise.
3151         * tree-predcom.c: Include wide-int-print.h.
3152         (struct dref_d): Change type of offset to widest_int.
3153         (dump_dref): Call wide-int printer.
3154         (aff_combination_dr_offset): Use wide-int interfaces.
3155         (determine_offset): Take a widest_int pointer rather than a
3156         double_int pointer.
3157         (split_data_refs_to_components): Use wide-int interfaces.
3158         (suitable_component_p): Likewise.
3159         (order_drefs): Likewise.
3160         (add_ref_to_chain): Likewise.
3161         (valid_initializer_p): Likewise.
3162         (determine_roots_comp): Likewise.
3163         * tree-pretty-print.c: Include wide-int-print.h.
3164         (dump_generic_node): Use wide-int interfaces.
3165         * tree-sra.c (sra_ipa_modify_expr): Likewise.
3166         * tree-ssa-address.c (addr_for_mem_ref): Likewise.
3167         (move_fixed_address_to_symbol): Likewise.
3168         (move_hint_to_base): Likewise.
3169         (move_pointer_to_base): Likewise.
3170         (move_variant_to_index): Likewise.
3171         (most_expensive_mult_to_index): Likewise.
3172         (addr_to_parts): Likewise.
3173         (copy_ref_info): Likewise.
3174         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
3175         (indirect_refs_may_alias_p): Likewise.
3176         (stmt_kills_ref_p_1): Likewise.
3177         * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
3178         * tree-ssa-ccp.c: Update comment at top of file.  Include
3179         wide-int-print.h.
3180         (struct prop_value_d): Change type of mask to widest_int.
3181         (extend_mask): New function.
3182         (dump_lattice_value): Use wide-int interfaces.
3183         (get_default_value): Likewise.
3184         (set_constant_value): Likewise.
3185         (set_value_varying): Likewise.
3186         (valid_lattice_transition): Likewise.
3187         (set_lattice_value): Likewise.
3188         (value_to_double_int): Delete.
3189         (value_to_wide_int): New.
3190         (get_value_from_alignment): Use wide-int interfaces.
3191         (get_value_for_expr): Likewise.
3192         (do_dbg_cnt): Likewise.
3193         (ccp_finalize): Likewise.
3194         (ccp_lattice_meet): Likewise.
3195         (bit_value_unop_1): Use widest_ints rather than double_ints.
3196         (bit_value_binop_1): Likewise.
3197         (bit_value_unop): Use wide-int interfaces.
3198         (bit_value_binop): Likewise.
3199         (bit_value_assume_aligned): Likewise.
3200         (evaluate_stmt): Likewise.
3201         (ccp_fold_stmt): Likewise.
3202         (visit_cond_stmt): Likewise.
3203         (ccp_visit_stmt): Likewise.
3204         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
3205         (constant_pointer_difference): Likewise.
3206         (associate_pointerplus): Likewise.
3207         (combine_conversions): Likewise.
3208         * tree-ssa-loop.h: Include wide-int.h.
3209         (struct tree_niter_desc): Change type of max to widest_int.
3210         * tree-ssa-loop-im.c (mem_refs_may_alias_p): Use wide-int interfaces.
3211         * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Likewise.
3212         (remove_redundant_iv_tests): Likewise.
3213         (canonicalize_loop_induction_variables): Likewise.
3214         * tree-ssa-loop-ivopts.c (alloc_iv): Likewise.
3215         (constant_multiple_of): Take a widest_int pointer instead of
3216         a double_int pointer.
3217         (get_computation_aff): Use wide-int interfaces.
3218         (ptr_difference_cost): Likewise.
3219         (difference_cost): Likewise.
3220         (get_loop_invariant_expr_id): Likewise.
3221         (get_computation_cost_at): Likewise.
3222         (iv_elimination_compare_lt): Likewise.
3223         (may_eliminate_iv): Likewise.
3224         * tree-ssa-loop-niter.h (estimated_loop_iterations): Use widest_int
3225         instead of double_int.
3226         (max_loop_iterations): Likewise.
3227         (max_stmt_executions): Likewise.
3228         (estimated_stmt_executions): Likewise.
3229         * tree-ssa-loop-niter.c: Include wide-int-print.h.
3230         (split_to_var_and_offset): Use wide-int interfaces.
3231         (determine_value_range): Likewise.
3232         (bound_difference_of_offsetted_base): Likewise.
3233         (bounds_add): Take a widest_int instead of a double_int.
3234         (number_of_iterations_ne_max): Use wide-int interfaces.
3235         (number_of_iterations_ne): Likewise.
3236         (number_of_iterations_lt_to_ne): Likewise.
3237         (assert_loop_rolls_lt): Likewise.
3238         (number_of_iterations_lt): Likewise.
3239         (number_of_iterations_le): Likewise.
3240         (number_of_iterations_cond): Likewise.
3241         (number_of_iterations_exit): Likewise.
3242         (finite_loop_p): Likewise.
3243         (derive_constant_upper_bound_assign): Likewise.
3244         (derive_constant_upper_bound): Return a widest_int.
3245         (derive_constant_upper_bound_ops): Likewise.
3246         (do_warn_aggressive_loop_optimizations): Use wide-int interfaces.
3247         (record_estimate): Take a widest_int rather than a double_int.
3248         (record_nonwrapping_iv): Use wide-int interfaces.
3249         (double_int_cmp): Delete.
3250         (wide_int_cmp): New.
3251         (bound_index): Take a widest_int rather than a double_int.
3252         (discover_iteration_bound_by_body_walk): Use wide-int interfaces.
3253         (maybe_lower_iteration_bound): Likewise.
3254         (estimate_numbers_of_iterations_loop): Likewise.
3255         (estimated_loop_iterations): Take a widest_int pointer than than
3256         a double_int pointer.
3257         (estimated_loop_iterations_int): Use wide-int interfaces.
3258         (max_loop_iterations): Take a widest_int pointer than than
3259         a double_int pointer.
3260         (max_loop_iterations_int): Use wide-int interfaces.
3261         (max_stmt_executions): Take a widest_int pointer than than
3262         a double_int pointer.
3263         (estimated_stmt_executions): Likewise.
3264         (n_of_executions_at_most): Use wide-int interfaces.
3265         (scev_probably_wraps_p): Likewise.
3266         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Update calls
3267         to real_to_integer.
3268         * tree-scalar-evolution.c (simplify_peeled_chrec): Use wide-int
3269         interfaces.
3270         * tree-ssanames.c (set_range_info): Use wide_int_refs rather than
3271         double_ints.  Adjust for trailing_wide_ints <3> representation.
3272         (set_nonzero_bits): Likewise.
3273         (get_range_info): Return wide_ints rather than double_ints.
3274         Adjust for trailing_wide_ints <3> representation.
3275         (get_nonzero_bits): Likewise.
3276         (duplicate_ssa_name_range_info): Adjust for trailing_wide_ints <3>
3277         representation.
3278         * tree-ssanames.h (struct range_info_def): Replace min, max and
3279         nonzero_bits with a trailing_wide_ints <3>.
3280         (set_range_info): Use wide_int_refs rather than double_ints.
3281         (set_nonzero_bits): Likewise.
3282         (get_range_info): Return wide_ints rather than double_ints.
3283         (get_nonzero_bits): Likewise.
3284         * tree-ssa-phiopt.c (jump_function_from_stmt): Use wide-int interfaces.
3285         * tree-ssa-pre.c (phi_translate_1): Likewise.
3286         * tree-ssa-reassoc.c (decrement_power): Use calls to real_from_integer.
3287         (acceptable_pow_call): Likewise.
3288         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Use wide-int
3289         interfaces.
3290         (vn_reference_fold_indirect): Likewise.
3291         (vn_reference_maybe_forwprop_address): Likewise.
3292         (valueize_refs_1): Likewise.
3293         * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
3294         * tree-ssa-uninit.c (is_value_included_in): Use wide-int interfaces,
3295         tree_int_cst_lt and tree_int_cst_le.
3296         * tree-streamer-in.c (unpack_ts_base_value_fields): Use wide-int
3297         interfaces.
3298         (streamer_alloc_tree): Likewise.
3299         * tree-streamer-out.c (pack_ts_int_cst_value_fields): Likewise.
3300         (streamer_write_tree_header): Likewise.
3301         (streamer_write_integer_cst): Likewise.
3302         * tree-switch-conversion.c (emit_case_bit_tests): Likewise.
3303         (build_constructors): Likewise.
3304         (array_value_type): Likewise.
3305         * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Likewise.
3306         (vect_check_gather): Likewise.
3307         * tree-vect-generic.c (build_replicated_const): Likewise.
3308         (expand_vector_divmod): Likewise.
3309         * tree-vect-loop.c (vect_transform_loop): Likewise.
3310         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
3311         (vect_do_peeling_for_alignment): Likewise.
3312         * tree-vect-patterns.c (vect_recog_divmod_pattern): Likewise.
3313         * tree-vrp.c: Include wide-int.h.
3314         (operand_less_p): Use wide-int interfaces and tree_int_cst_lt.
3315         (extract_range_from_assert): Use wide-int interfaces.
3316         (vrp_int_const_binop): Likewise.
3317         (zero_nonzero_bits_from_vr): Take wide_int pointers rather than
3318         double_int pointers.
3319         (ranges_from_anti_range): Use wide-int interfaces.
3320         (quad_int_cmp): Delete.
3321         (quad_int_pair_sort): Likewise.
3322         (extract_range_from_binary_expr_1): Use wide-int interfaces.
3323         (extract_range_from_unary_expr_1): Likewise.
3324         (adjust_range_with_scev): Likewise.
3325         (masked_increment): Take and return wide_ints rather than double_ints.
3326         (register_edge_assert_for_2): Use wide-int interfaces.
3327         (check_array_ref): Likewise.
3328         (search_for_addr_array): Likewise.
3329         (maybe_set_nonzero_bits): Likewise.
3330         (union_ranges): Pass an integer of the correct type instead of
3331         using integer_one_node.
3332         (intersect_ranges): Likewise.
3333         (simplify_truth_ops_using_ranges): Likewise.
3334         (simplify_bit_ops_using_ranges): Use wide-int interfaces.
3335         (range_fits_type_p): Likewise.
3336         (simplify_cond_using_ranges): Likewise.  Take a signop rather than
3337         a bool.
3338         (simplify_conversion_using_ranges): Use wide-int interfaces.
3339         (simplify_float_conversion_using_ranges): Likewise.
3340         (vrp_finalize): Likewise.
3341         * value-prof.c (gimple_divmod_fixed_value_transform): Likewise.
3342         (gimple_stringops_transform): Likewise.
3343         * varasm.c (decode_addr_const): Likewise.
3344         (const_hash_1): Likewise.
3345         (const_rtx_hash_1): Likewise
3346         (output_constant): Likewise.
3347         (array_size_for_constructor): Likewise.
3348         (output_constructor_regular_field): Likewise.
3349         (output_constructor_bitfield): Likewise.
3350         * var-tracking.c (loc_cmp): Handle CONST_WIDE_INT.
3351         * mkconfig.sh: Include machmode.h to pick up BITS_PER_UNIT for
3352         GENERATOR_FILEs.
3353         * gencheck.c: Define BITS_PER_UNIT.
3354         * wide-int.cc: New.
3355         * wide-int.h: New.
3356         * wide-int-print.cc: New.
3357         * wide-int-print.h: New.
3358
3359 2014-05-06  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
3360
3361         * config/avr/avr.c (avr_can_eliminate): Mark unused argument.
3362
3363 2014-05-06  Richard Biener  <rguenther@suse.de>
3364
3365         * tree-pass.h (TODO_verify_ssa, TODO_verify_flow,
3366         TODO_verify_stmts, TODO_verify_rtl_sharing): Remove.
3367         (TODO_verify_all): Adjust.
3368         * asan.c: Remove references to TODO_verify_ssa, TODO_verify_flow,
3369         TODO_verify_stmts and TODO_verify_rtl_sharing.
3370         * bb-reorder.c: Likewise.
3371         * cfgexpand.c: Likewise.
3372         * cprop.c: Likewise.
3373         * cse.c: Likewise.
3374         * function.c: Likewise.
3375         * fwprop.c: Likewise.
3376         * gcse.c: Likewise.
3377         * gimple-ssa-isolate-paths.c: Likewise.
3378         * gimple-ssa-strength-reduction.c: Likewise.
3379         * ipa-split.c: Likewise.
3380         * loop-init.c: Likewise.
3381         * loop-unroll.c: Likewise.
3382         * lower-subreg.c: Likewise.
3383         * modulo-sched.c: Likewise.
3384         * postreload-gcse.c: Likewise.
3385         * predict.c: Likewise.
3386         * recog.c: Likewise.
3387         * sched-rgn.c: Likewise.
3388         * store-motion.c: Likewise.
3389         * tracer.c: Likewise.
3390         * trans-mem.c: Likewise.
3391         * tree-call-cdce.c: Likewise.
3392         * tree-cfg.c: Likewise.
3393         * tree-cfgcleanup.c: Likewise.
3394         * tree-complex.c: Likewise.
3395         * tree-eh.c: Likewise.
3396         * tree-emutls.c: Likewise.
3397         * tree-if-conv.c: Likewise.
3398         * tree-into-ssa.c: Likewise.
3399         * tree-loop-distribution.c: Likewise.
3400         * tree-object-size.c: Likewise.
3401         * tree-parloops.c: Likewise.
3402         * tree-pass.h: Likewise.
3403         * tree-sra.c: Likewise.
3404         * tree-ssa-ccp.c: Likewise.
3405         * tree-ssa-copy.c: Likewise.
3406         * tree-ssa-copyrename.c: Likewise.
3407         * tree-ssa-dce.c: Likewise.
3408         * tree-ssa-dom.c: Likewise.
3409         * tree-ssa-dse.c: Likewise.
3410         * tree-ssa-forwprop.c: Likewise.
3411         * tree-ssa-ifcombine.c: Likewise.
3412         * tree-ssa-loop-ch.c: Likewise.
3413         * tree-ssa-loop-ivcanon.c: Likewise.
3414         * tree-ssa-loop.c: Likewise.
3415         * tree-ssa-math-opts.c: Likewise.
3416         * tree-ssa-phiopt.c: Likewise.
3417         * tree-ssa-phiprop.c: Likewise.
3418         * tree-ssa-pre.c: Likewise.
3419         * tree-ssa-reassoc.c: Likewise.
3420         * tree-ssa-sink.c: Likewise.
3421         * tree-ssa-strlen.c: Likewise.
3422         * tree-ssa-tail-merge.c: Likewise.
3423         * tree-ssa-uncprop.c: Likewise.
3424         * tree-switch-conversion.c: Likewise.
3425         * tree-tailcall.c: Likewise.
3426         * tree-vect-generic.c: Likewise.
3427         * tree-vectorizer.c: Likewise.
3428         * tree-vrp.c: Likewise.
3429         * tsan.c: Likewise.
3430         * var-tracking.c: Likewise.
3431         * bt-load.c: Likewise.
3432         * cfgcleanup.c: Likewise.
3433         * combine-stack-adj.c: Likewise.
3434         * combine.c: Likewise.
3435         * compare-elim.c: Likewise.
3436         * config/epiphany/resolve-sw-modes.c: Likewise.
3437         * config/i386/i386.c: Likewise.
3438         * config/mips/mips.c: Likewise.
3439         * config/s390/s390.c: Likewise.
3440         * config/sh/sh_treg_combine.cc: Likewise.
3441         * config/sparc/sparc.c: Likewise.
3442         * dce.c: Likewise.
3443         * dse.c: Likewise.
3444         * final.c: Likewise.
3445         * ifcvt.c: Likewise.
3446         * mode-switching.c: Likewise.
3447         * passes.c: Likewise.
3448         * postreload.c: Likewise.
3449         * ree.c: Likewise.
3450         * reg-stack.c: Likewise.
3451         * regcprop.c: Likewise.
3452         * regrename.c: Likewise.
3453         * web.c: Likewise.
3454
3455 2014-05-06  Richard Biener  <rguenther@suse.de>
3456
3457         PR middle-end/61070
3458         * bitmap.c (debug_bitmap): Dump to stderr, not stdout.
3459         * tree-ssa-structalias.c (dump_solution_for_var): Likewise.
3460
3461 2014-05-05  Jan Hubicka  <hubicka@ucw.cz>
3462
3463         PR ipa/60965
3464         * ipa-devirt.c (get_class_context): Allow POD to change to non-POD.
3465
3466 2014-05-05  Radovan Obradovic  <robradovic@mips.com>
3467             Tom de Vries  <tom@codesourcery.com>
3468
3469         * target.def (call_fusage_contains_non_callee_clobbers): New
3470         DEFHOOKPOD.
3471         * doc/tm.texi.in (@node Stack and Calling): Add Miscellaneous Register
3472         Hooks to @menu.
3473         (@node Miscellaneous Register Hooks): New node.
3474         (@hook TARGET_CALL_FUSAGE_CONTAINS_NON_CALLEE_CLOBBERS): New hook.
3475         * doc/tm.texi: Regenerate.
3476
3477 2014-05-05  Marek Polacek  <polacek@redhat.com>
3478
3479         PR driver/61065
3480         * opts.c (common_handle_option): Call error_at instead of warning_at.
3481
3482 2014-05-05  Richard Biener  <rguenther@suse.de>
3483
3484         * passes.c (execute_function_todo): Don't reset TODO_verify_ssa
3485         from last_verified if update_ssa ran.  Move TODO_verify_rtl_sharing
3486         under the TODO_verify_il umbrella.
3487
3488 2014-05-05  Richard Biener  <rguenther@suse.de>
3489
3490         * passes.c (execute_function_todo): Move TODO_verify_flow under
3491         the TODO_verify_ul umbrella.
3492
3493 2014-05-05  Richard Biener  <rguenther@suse.de>
3494
3495         PR middle-end/61010
3496         * fold-const.c (fold_binary_loc): Consistently avoid canonicalizing
3497         X & CST away from a CST that is the mask of a mode.
3498
3499 2014-05-05  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
3500
3501         * config/picochip/picochip-protos.h (picochip_regno_nregs): Change
3502         int argument to enum machine_mode.
3503         (picochip_class_max_nregs): Ditto.
3504         * config/picochip/picochip.c (picochip_regno_nregs): Ditto.
3505         (picochip_class_max_nregs): Ditto.
3506
3507 2014-05-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3508
3509         * target.def: Add new target hook.
3510         * doc/tm.texi: Regenerate.
3511         * targhooks.h (default_keep_leaf_when_profiled): Add prototype.
3512         * targhooks.c (default_keep_leaf_when_profiled): New function.
3513
3514         * config/s390/s390.c (s390_keep_leaf_when_profiled): New function.
3515         (TARGET_KEEP_LEAF_WHEN_PROFILED): Define.
3516
3517 2014-05-05  Bin Cheng  <bin.cheng@arm.com>
3518
3519         PR tree-optimization/60363
3520         * gcc/tree-ssa-threadupdate.c (get_value_locus_in_path): New.
3521         (copy_phi_args): New parameters.  Call get_value_locus_in_path.
3522         (update_destination_phis): New parameter.
3523         (create_edge_and_update_destination_phis): Ditto.
3524         (ssa_fix_duplicate_block_edges): Pass new arguments.
3525         (thread_single_edge): Ditto.
3526
3527 2014-05-04  Peter Bergner  <bergner@vnet.ibm.com>
3528
3529         * config/rs6000/rs6000.h (RS6000_BTM_HARD_FLOAT): New define.
3530         (RS6000_BTM_COMMON): Add RS6000_BTM_HARD_FLOAT.
3531         (TARGET_EXTRA_BUILTINS): Add TARGET_HARD_FLOAT.
3532         * config/rs6000/rs6000-builtin.def (BU_MISC_1):
3533         Use RS6000_BTM_HARD_FLOAT.
3534         (BU_MISC_2): Likewise.
3535         * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Handle
3536         RS6000_BTM_HARD_FLOAT.
3537         (rs6000_option_override_internal): Enforce -mhard-float if -mhard-dfp
3538         is explicitly used.
3539         (rs6000_invalid_builtin): Add hard floating builtin support.
3540         (rs6000_expand_builtin): Relax the gcc_assert to allow the new
3541         hard float builtins.
3542         (rs6000_builtin_mask_names): Add RS6000_BTM_HARD_FLOAT.
3543
3544 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
3545
3546         * config/sh/sh_optimize_sett_clrt.cc (sh_optimize_sett_clrt::execute):
3547         Add missing function* argument.
3548
3549 2014-05-03  Richard Sandiford  <rdsandiford@googlemail.com>
3550
3551         * lra-constraints.c (valid_address_p): Move earlier in file.
3552         Add a constraint argument to the address_info version.
3553         (satisfies_memory_constraint_p): New function.
3554         (satisfies_address_constraint_p): Likewise.
3555         (process_alt_operands, curr_insn_transform): Use them.
3556         (process_address): Pass the constraint to valid_address_p when
3557         checking address operands.
3558
3559 2014-05-03  Richard Sandiford  <rdsandiford@googlemail.com>
3560
3561         * config/mips/mips.c (mips_isa_rev): New variable.
3562         (mips_set_architecture): Set it.
3563         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Set __mips_isa_rev
3564         from mips_isa_rev.
3565         (ISA_HAS_MUL3, ISA_HAS_FP_CONDMOVE, ISA_HAS_8CC, ISA_HAS_FP4)
3566         (ISA_HAS_PAIRED_SINGLE, ISA_HAS_MADD_MSUB, ISA_HAS_FP_RECIP_RSQRT)
3567         (ISA_HAS_CLZ_CLO, ISA_HAS_ROR, ISA_HAS_WSBH, ISA_HAS_PREFETCH)
3568         (ISA_HAS_SEB_SEH, ISA_HAS_EXT_INS, ISA_HAS_MXHC1)
3569         (ISA_HAS_HILO_INTERLOCKS, ISA_HAS_SYNCI, MIN_FPRS_PER_FMT): Reexpress
3570         conditions in terms of mips_isa_rev.
3571         (mips_isa_rev): Declare.
3572
3573 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
3574
3575         * config/sh/sh-mem.cc: Use tabs instead of spaces.
3576         (prob_unlikely, prob_likely): Make variables const.
3577
3578 2014-05-03  Denis Chertykov  <chertykov@gmail.com>
3579
3580         * config/avr/avr.c (avr_adjust_insn_length): Handle JUMP_TABLE_DATA.
3581
3582 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
3583
3584         * config/sh/sh.h (SH_ASM_SPEC): Handle m1, m2*, m3* and m4* cases.
3585
3586 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
3587
3588         * config/sh/sh.h (ROUND_ADVANCE): Delete macro.
3589         (ROUND_REG, PASS_IN_REG_P): Move and rename macros to ...
3590         * config/sh/sh.c (sh_round_reg, sh_pass_in_reg_p): ... these new
3591         functions.
3592         (sh_arg_partial_bytes, sh_function_arg, sh_function_arg_advance,
3593         sh_setup_incoming_varargs): Replace usage of PASS_IN_REG_P with
3594         sh_pass_in_reg_p.
3595         Replace usage of ROUND_REG with sh_round_reg.
3596         Use CEIL instead of ROUND_ADVANCE.
3597
3598 2014-05-03  Oleg Endo  <olegendo@gcc.gnu.org>
3599
3600         PR target/61026
3601         * config/sh/sh.c: Include stdlib headers before everything else.
3602
3603 2014-05-02  Jakub Jelinek  <jakub@redhat.com>
3604
3605         * gimplify.c (gimplify_adjust_omp_clauses_1): Handle
3606         GOVD_FIRSTPRIVATE | GOVD_LASTPRIVATE.
3607         (gimplify_adjust_omp_clauses): Simd region is never
3608         directly nested in combined parallel.  Instead, for linear
3609         with copyin/copyout, if in combined for simd loop, make decl
3610         firstprivate/lastprivate on OMP_FOR.
3611         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
3612         expand_omp_for_static_chunk): When setting endvar, also set
3613         fd->loop.v to the same value.
3614
3615 2014-05-02  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
3616
3617         * hwint.h (zext_hwi): Fix signed overflow for prec == 63.
3618
3619 2014-05-02  Alan Lawrence  <alan.lawrence@arm.com>
3620
3621         * config/aarch64/aarch64.c (aarch64_expand_vec_perm_1): Tidy bit-flip
3622         expression.
3623
3624 2014-05-02  Marek Polacek  <polacek@redhat.com>
3625
3626         * doc/invoke.texi: Describe -fsanitize=float-divide-by-zero.
3627
3628 2014-05-02  Kito Cheng  <kito@0xlab.org>
3629
3630         * defaults.h (HONOR_REG_ALLOC_ORDER): Change HONOR_REG_ALLOC_ORDER
3631         to a C expression marco.
3632         * ira-color.c (HONOR_REG_ALLOC_ORDER) : Ditto.
3633         * config/arm/arm.h (HONOR_REG_ALLOC_ORDER): Ditto.
3634         * config/nds32/nds32.h (HONOR_REG_ALLOC_ORDER): Ditto.
3635         * doc/tm.texi (HONOR_REG_ALLOC_ORDER): Update document for
3636         HONOR_REG_ALLOC_ORDER.
3637         * doc/tm.texi.in (HONOR_REG_ALLOC_ORDER): Ditto.
3638
3639 2014-05-01  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
3640
3641         * config/arc/arc.c (TARGET_LRA_P): Undef before redefine.
3642
3643 2014-05-01  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
3644
3645         * config/arc/arc.c (arc_select_cc_mode): Fix typo.
3646
3647 2014-05-01  Yuri Rumyantsev  <ysrumyan@gmail.com>
3648
3649         * tree-if-conv.c (is_cond_scalar_reduction): New function.
3650         (convert_scalar_cond_reduction): Likewise.
3651         (predicate_scalar_phi): Add recognition and transformation
3652         of simple conditioanl reduction to be vectorizable.
3653
3654 2014-05-01  Marek Polacek  <polacek@redhat.com>
3655
3656         PR c/43245
3657         * doc/invoke.texi: Document -Wdiscarded-qualifiers.
3658
3659 2014-04-30  Alan Lawrence  <alan.lawrence@arm.com>
3660
3661         * config/aarch64/arm_neon.h (vuzp1_f32, vuzp1_p8, vuzp1_p16, vuzp1_s8,
3662         vuzp1_s16, vuzp1_s32, vuzp1_u8, vuzp1_u16, vuzp1_u32, vuzp1q_f32,
3663         vuzp1q_f64, vuzp1q_p8, vuzp1q_p16, vuzp1q_s8, vuzp1q_s16, vuzp1q_s32,
3664         vuzp1q_s64, vuzp1q_u8, vuzp1q_u16, vuzp1q_u32, vuzp1q_u64, vuzp2_f32,
3665         vuzp2_p8, vuzp2_p16, vuzp2_s8, vuzp2_s16, vuzp2_s32, vuzp2_u8,
3666         vuzp2_u16, vuzp2_u32, vuzp2q_f32, vuzp2q_f64, vuzp2q_p8, vuzp2q_p16,
3667         vuzp2q_s8, vuzp2q_s16, vuzp2q_s32, vuzp2q_s64, vuzp2q_u8, vuzp2q_u16,
3668         vuzp2q_u32, vuzp2q_u64): Replace temporary asm with __builtin_shuffle.
3669
3670 2014-04-30  Joern Rennecke  <joern.rennecke@embecosm.com>
3671
3672         * config/arc/arc.opt (mlra): Move comment above option name
3673         to avoid mis-parsing as language options.
3674
3675 2014-04-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3676
3677         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Move ...
3678         * config/sol2.h: ... here.
3679         * config/sol2-10.h: Remove.
3680
3681         * config/sol2-bi.h (WCHAR_TYPE, WCHAR_TYPE_SIZE, WINT_TYPE)
3682         (WINT_TYPE_SIZE, MULTILIB_DEFAULTS, DEF_ARCH32_SPEC)
3683         (DEF_ARCH64_SPEC, ASM_CPU_DEFAULT_SPEC, LINK_ARCH64_SPEC_BASE)
3684         (LINK_ARCH64_SPEC, ARCH_DEFAULT_EMULATION, TARGET_LD_EMULATION)
3685         (LINK_ARCH_SPEC, SUBTARGET_EXTRA_SPECS): Move ...
3686         * config/sol2.h: ... here.
3687         (SECTION_NAME_FORMAT): Don't redefine.
3688         (STARTFILE_ARCH32_SPEC): Rename to ...
3689         (STARTFILE_ARCH_SPEC): ... this.
3690         (ASM_OUTPUT_ALIGNED_COMMON): Move ...
3691         * config/sparc/sol2.h: ... here.
3692         (SECTION_NAME_FORMAT): Don't undef.
3693         * config/i386/sol2.h (ASM_CPU_DEFAULT_SPEC)
3694         (SUBTARGET_EXTRA_SPECS): Remove.
3695         * config/sparc/sol2.h (ASM_CPU_DEFAULT_SPEC): Remove.
3696
3697         * config/i386/sol2-bi.h (TARGET_SUBTARGET_DEFAULT)
3698         (MD_STARTFILE_PREFIX): Remove.
3699         (SUBTARGET_OPTIMIZATION_OPTIONS, ASM_CPU32_DEFAULT_SPEC)
3700         (ASM_CPU64_DEFAULT_SPEC, ASM_CPU_SPEC, ASM_SPEC, DEFAULT_ARCH32_P)
3701         (ARCH64_SUBDIR, ARCH32_EMULATION, ARCH64_EMULATION)
3702         (ASM_COMMENT_START, JUMP_TABLES_IN_TEXT_SECTION)
3703         (ASM_OUTPUT_DWARF_PCREL, ASM_OUTPUT_ALIGNED_COMMON)
3704         (USE_IX86_FRAME_POINTER, USE_X86_64_FRAME_POINTER): Move ...
3705         * config/i386/sol2.h: ... here.
3706         (TARGET_SUBTARGET_DEFAULT, SIZE_TYPE, PTRDIFF_TYPE): Remove.
3707         * config/i386/sol2-bi.h: Remove.
3708         * config/sol2.h (MD_STARTFILE_PREFIX): Remove.
3709         (LINK_ARCH32_SPEC_BASE): Remove /usr/ccs/lib/libp, /usr/ccs/lib.
3710
3711         * config/i386/t-sol2-64: Rename to ...
3712         * config/i386/t-sol2: ... this.
3713         * config/sparc/t-sol2-64: Rename to ...
3714         * config/sparc/t-sol2: ... this.
3715
3716         * config.gcc (*-*-solaris2*): Split sol2_tm_file into
3717         sol2_tm_file_head, sol2_tm_file_tail.
3718         Include ${cpu_type}/sol2.h before sol2.h.
3719         Remove sol2-10.h.
3720         (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Include
3721         i386/x86-64.h between sol2_tm_file_head and sol2_tm_file_tail.
3722         Remove i386/sol2-bi.h, sol2-bi.h from tm_file.
3723         Reflect i386/t-sol2-64 renaming.
3724         (sparc*-*-solaris2*): Remove sol2-bi.h from tm_file.
3725         Reflect sparc/t-sol2-64 renaming.
3726
3727 2014-04-30  Richard Biener  <rguenther@suse.de>
3728
3729         * passes.c (execute_function_todo): Move TODO_verify_stmts
3730         and TODO_verify_ssa under the TODO_verify_il umbrella.
3731         * tree-ssa.h (verify_ssa): Adjust prototype.
3732         * tree-ssa.c (verify_ssa): Add parameter to tell whether
3733         we should verify SSA operands.
3734         * tree-cfg.h (verify_gimple_in_cfg): Adjust prototype.
3735         * tree-cfg.c (verify_gimple_in_cfg): Add parameter to tell
3736         whether we should verify whether not throwing stmts have EH info.
3737         * graphite-scop-detection.c (create_sese_edges): Adjust.
3738         * tree-ssa-loop-manip.c (verify_loop_closed_ssa): Likewise.
3739         * tree-eh.c (lower_try_finally_switch): Do not add the
3740         default case label twice.
3741
3742 2014-04-30  Marek Polacek  <polacek@redhat.com>
3743
3744         * gcc.c (sanitize_spec_function): Handle SANITIZE_FLOAT_DIVIDE.
3745         * builtins.def: Initialize builtins even for SANITIZE_FLOAT_DIVIDE.
3746         * flag-types.h (enum sanitize_code): Add SANITIZE_FLOAT_DIVIDE.
3747         * opts.c (common_handle_option): Add -fsanitize=float-divide-by-zero.
3748
3749 2014-04-29  Alan Lawrence  <alan.lawrence@arm.com>
3750
3751         * config/aarch64/arm_neon.h (vzip1_f32, vzip1_p8, vzip1_p16, vzip1_s8,
3752         vzip1_s16, vzip1_s32, vzip1_u8, vzip1_u16, vzip1_u32, vzip1q_f32,
3753         vzip1q_f64, vzip1q_p8, vzip1q_p16, vzip1q_s8, vzip1q_s16, vzip1q_s32,
3754         vzip1q_s64, vzip1q_u8, vzip1q_u16, vzip1q_u32, vzip1q_u64, vzip2_f32,
3755         vzip2_p8, vzip2_p16, vzip2_s8, vzip2_s16, vzip2_s32, vzip2_u8,
3756         vzip2_u16, vzip2_u32, vzip2q_f32, vzip2q_f64, vzip2q_p8, vzip2q_p16,
3757         vzip2q_s8, vzip2q_s16, vzip2q_s32, vzip2q_s64, vzip2q_u8, vzip2q_u16,
3758         vzip2q_u32, vzip2q_u64): Replace inline __asm__ with __builtin_shuffle.
3759
3760 2014-04-29  David Malcolm  <dmalcolm@redhat.com>
3761
3762         * tree-cfg.c (dump_function_to_file): Dump the return type of
3763         functions, in a line to itself before the function body, mimicking
3764         the layout of a C function.
3765
3766 2014-04-29  Jakub Jelinek  <jakub@redhat.com>
3767
3768         PR tree-optimization/60971
3769         * tree-tailcall.c (process_assignment): Reject conversions which
3770         reduce precision.
3771
3772 2014-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
3773
3774         * calls.c (initialize_argument_information): Always treat
3775         PUSH_ARGS_REVERSED as 1, simplify code accordingly.
3776         (expand_call): Likewise.
3777         (emit_library_call_calue_1): Likewise.
3778         * expr.c (PUSH_ARGS_REVERSED): Do not define.
3779         (emit_push_insn): Always treat PUSH_ARGS_REVERSED as 1, simplify
3780         code accordingly.
3781
3782 2014-04-29  Nick Clifton  <nickc@redhat.com>
3783
3784         * config/msp430/msp430.md (umulsidi): Fix typo.
3785         (mulhisi3): Enable even inside interrupt handlers.
3786         * config/msp430/msp430.c (msp430_print_operand): %O: Allow for the
3787         bigger return address pushed in large mode.
3788
3789 2014-04-29  Nick Clifton  <nickc@redhat.com>
3790
3791         * config/arc/arc.c (arc_select_cc_mode): Fix parentheses.
3792         (arc_init_reg_tables): Use a machine_mode enum to iterate over
3793         available modes.
3794         * config/m32r/m32r.c (init_reg_tables): Likewise.
3795         * config/m32c/m32c.c (m32c_illegal_subreg_p): Use a machine_mode
3796         enum to hold the modes.
3797
3798 2014-04-29  Richard Biener  <rguenther@suse.de>
3799
3800         * dominance.c (free_dominance_info): Add overload with
3801         function parameter.
3802         (dom_info_state): Likewise.
3803         (dom_info_available_p): Likewise.
3804         * basic-block.h (free_dominance_info, dom_info_state,
3805         dom_info_available_p): Declare overloads.
3806         * passes.c (execute_function_todo): Verify that verifiers
3807         don't change dominator info state.  Drop dominator info
3808         for IPA pass invocations.
3809         * cgraph.c (release_function_body): Restore asserts that
3810         dominator information is released.
3811
3812 2014-04-29  Patrick Palka  <patrick@parcs.ath.cx>
3813
3814         * doc/invoke.texi: Fix typo.
3815         * tree-vrp.c: Fix typos.
3816         * gimple.c (infer_nonnull_range): Reorder operands of an && condition.
3817
3818 2014-04-29  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
3819
3820         * config/aarch64/aarch64.md (mov<mode>cc): New for GPF.
3821
3822 2014-04-28  James Greenhalgh  <james.greenhalgh@arm.com>
3823
3824         * config/aarch64/aarch64-builtins.c
3825         (aarch64_types_storestruct_lane_qualifiers): New.
3826         (TYPES_STORESTRUCT_LANE): Likewise.
3827         * config/aarch64/aarch64-simd-builtins.def (st2_lane): New.
3828         (st3_lane): Likewise.
3829         (st4_lane): Likewise.
3830         * config/aarch64/aarch64-simd.md (vec_store_lanesoi_lane<mode>): New.
3831         (vec_store_lanesci_lane<mode>): Likewise.
3832         (vec_store_lanesxi_lane<mode>): Likewise.
3833         (aarch64_st2_lane<VQ:mode>): Likewise.
3834         (aarch64_st3_lane<VQ:mode>): Likewise.
3835         (aarch64_st4_lane<VQ:mode>): Likewise.
3836         * config/aarch64/aarch64.md (unspec): Add UNSPEC_ST{2,3,4}_LANE.
3837         * config/aarch64/arm_neon.h
3838         (__ST2_LANE_FUNC): Rewrite using builtins, update use points to
3839         use new macro arguments.
3840         (__ST3_LANE_FUNC): Likewise.
3841         (__ST4_LANE_FUNC): Likewise.
3842         * config/aarch64/iterators.md (V_TWO_ELEM): New.
3843         (V_THREE_ELEM): Likewise.
3844         (V_FOUR_ELEM): Likewise.
3845
3846 2014-04-28  David Malcolm  <dmalcolm@redhat.com>
3847
3848         * doc/gimple.texi: Replace the description of the now-defunct
3849         union gimple_statement_d with a diagram showing the
3850         gimple_statement_base class hierarchy and its relationships to
3851         the GSS_ and GIMPLE_ enums.
3852
3853 2014-04-28  James Greenhalgh  <james.greenhalgh@arm.com>
3854
3855         * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): New.
3856         * config/aarch64/aarch64.c
3857         (aarch64_cannot_change_mode_class): Weaken conditions.
3858         (aarch64_modes_tieable_p): New.
3859         * config/aarch64/aarch64.h (MODES_TIEABLE_P): Use it.
3860
3861 2014-04-28  Pat Haugen  <pthaugen@us.ibm.com>
3862
3863         * config/rs6000/sync.md (AINT mode_iterator): Move definition.
3864         (loadsync_<mode>): Change mode.
3865         (load_quadpti, store_quadpti): New.
3866         (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
3867         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
3868
3869 2014-04-28  Martin Jambor  <mjambor@suse.cz>
3870
3871         * tree-sra.c (sra_modify_expr): Generate new memory accesses with
3872         same alias type as the original statement.
3873         (subreplacement_assignment_data): New type.
3874         (handle_unscalarized_data_in_subtree): New type of parameter,
3875         generate new memory accesses with same alias type as the original
3876         statement.
3877         (load_assign_lhs_subreplacements): Likewise.
3878         (sra_modify_constructor_assign): Generate new memory accesses with
3879         same alias type as the original statement.
3880
3881 2014-04-28  Richard Biener  <rguenther@suse.de>
3882
3883         * tree-pass.h (TODO_verify_il): Define.
3884         (TODO_verify_all): Complete properly.
3885         * passes.c (execute_function_todo): Move existing loop-closed
3886         SSA verification under TODO_verify_il.
3887         (execute_one_pass): Trigger TODO_verify_il at todo-after time.
3888         * graphite-sese-to-poly.c (rewrite_cross_bb_scalar_deps):
3889         Fix tree sharing issue.
3890
3891 2014-04-28  Richard Biener  <rguenther@suse.de>
3892
3893         PR middle-end/60092
3894         * builtins.def (DEF_C11_BUILTIN): Add.
3895         (BUILT_IN_ALIGNED_ALLOC): Likewise.
3896         * coretypes.h (enum function_class): Add function_c11_misc.
3897         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Handle
3898         BUILT_IN_ALIGNED_ALLOC like BUILT_IN_MALLOC.
3899         (call_may_clobber_ref_p_1): Likewise.
3900         * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
3901         (mark_all_reaching_defs_necessary_1): Likewise.
3902         (propagate_necessity): Likewise.
3903         (eliminate_unnecessary_stmts): Likewise.
3904         * tree-ssa-ccp.c (evaluate_stmt): Handle BUILT_IN_ALIGNED_ALLOC.
3905
3906 2014-04-28  Richard Biener  <rguenther@suse.de>
3907
3908         * tree-vrp.c (vrp_var_may_overflow): Remove.
3909         (vrp_visit_phi_node): Rather than bumping to +-INF possibly
3910         with overflow immediately bump to one before that value and
3911         let iteration figure out overflow status.
3912
3913 2014-04-28  Richard Biener  <rguenther@suse.de>
3914
3915         * configure.ac: Do valgrind header checks unconditionally.
3916         Add --enable-valgrind-annotations.
3917         * system.h: Guard valgrind header inclusion with
3918         ENABLE_VALGRIND_ANNOTATIONS instead of ENABLE_VALGRIND_CHECKING.
3919         * alloc-pool.c (pool_alloc, pool_free): Use
3920         ENABLE_VALGRIND_ANNOTATIONS instead of ENABLE_VALGRIND_CHECKING
3921         to guard possibly dead code.
3922         * config.in: Regenerated.
3923         * configure: Likewise.
3924
3925 2014-04-28  Jeff Law  <law@redhat.com>
3926
3927         PR tree-optimization/60902
3928         * tree-ssa-threadedge.c
3929         (record_temporary_equivalences_from_stmts_at_dest): Only iterate
3930         over real defs when invalidating outputs from statements that do not
3931         produce useful outputs for threading.
3932
3933 2014-04-28  Richard Biener  <rguenther@suse.de>
3934
3935         PR tree-optimization/60979
3936         * graphite-scop-detection.c (scopdet_basic_block_info): Reject
3937         SCOPs that end in a block with a successor with abnormal
3938         predecessors.
3939
3940 2014-04-28  Richard Biener  <rguenther@suse.de>
3941
3942         * tree-pass.h (execute_pass_list): Adjust prototype.
3943         * passes.c (pass_manager::execute_early_local_passes): Adjust.
3944         (do_per_function): Change callback signature, push all actual
3945         work to the callbals.
3946         (do_per_function_toporder): Likewise.
3947         (execute_function_dump): Adjust.
3948         (execute_function_todo): Likewise.
3949         (clear_last_verified): Likewise.
3950         (verify_curr_properties): Likewise.
3951         (update_properties_after_pass): Likewise.
3952         (execute_pass_list_1): Split out from ...
3953         (execute_pass_list): ... here.  Adjust.
3954         (execute_ipa_pass_list): Likewise.
3955         * cgraphunit.c (cgraph_add_new_function): Adjust.
3956         (analyze_function): Likewise.
3957         (expand_function): Likewise.
3958         * cgraph.c (release_function_body): Free dominance info
3959         here instead of asserting it was magically freed elsewhere.
3960
3961 2014-04-28  Eric Botcazou  <ebotcazou@adacore.com>
3962
3963         * configure.ac: Tweak GAS check for LEON instructions on SPARC.
3964         * configure: Regenerate.
3965         * config/sparc/sparc.opt (muser-mode): New option.
3966         * config/sparc/sync.md (atomic_compare_and_swap<mode>_1): Do not enable
3967         for LEON3.
3968         (atomic_compare_and_swap_leon3_1): New instruction for LEON3.
3969         * doc/invoke.texi (SPARC options): Document -muser-mode.
3970
3971 2014-04-27  Richard Sandiford  <rdsandiford@googlemail.com>
3972
3973         * cselib.c (find_slot_memmode): Delete.
3974         (cselib_hasher): Change compare_type to a struct.
3975         (cselib_hasher::equal): Update accordingly.  Don't expect wrapped
3976         constants.
3977         (preserve_constants_and_equivs): Adjust for new compare_type.
3978         (cselib_find_slot): Likewise.  Take the mode of the rtx as argument.
3979         (wrap_constant): Delete.
3980         (cselib_lookup_mem, cselib_lookup_1): Update calls to cselib_find_slot.
3981
3982 2014-04-26  Markus Trippelsdorf  <markus@trippelsdorf.de>
3983
3984         * doc/install.texi (Building with profile feedback): Remove
3985         outdated sentence.
3986
3987 2014-04-26  Tom de Vries  <tom@codesourcery.com>
3988
3989         * config/i386/i386.md (define_expand "ldexpxf3"): Fix out-of-bounds
3990         array accesses.
3991
3992 2014-04-25  Cary Coutant  <ccoutant@google.com>
3993
3994         PR debug/60929
3995         * dwarf2out.c (should_move_die_to_comdat): A type definition
3996         can contain a subprogram definition, but don't move it to a
3997         comdat unit.
3998         (clone_as_declaration): Copy DW_AT_abstract_origin attribute.
3999         (generate_skeleton_bottom_up): Remove DW_AT_object_pointer attribute
4000         from original DIE.
4001         (clone_tree_hash): Rename to...
4002         (clone_tree_partial): ...this; change callers.  Copy
4003         DW_TAG_subprogram DIEs as declarations.
4004         (copy_decls_walk): Don't copy children of a declaration into a
4005         type unit.
4006
4007 2014-04-25  H.J. Lu  <hongjiu.lu@intel.com>
4008
4009         PR target/60969
4010         * config/i386/i386.md (*movsf_internal): Set MODE to SI for
4011         alternative 12.
4012
4013 2014-04-25  Jiong Wang  <jiong.wang@arm.com>
4014
4015         * config/arm/predicates.md (call_insn_operand): Add long_call check.
4016         * config/arm/arm.md (sibcall, sibcall_value): Force the address to
4017         reg for long_call.
4018         * config/arm/arm.c (arm_function_ok_for_sibcall): Remove long_call
4019         restriction.
4020
4021 2014-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4022
4023         * config/arm/arm.c (arm_cortex_a8_tune): Initialise T16-related fields.
4024
4025 2014-04-25  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4026
4027         PR tree-optimization/60930
4028         * gimple-ssa-strength-reduction.c (create_mul_imm_cand):  Reject
4029         creating a multiply candidate by folding two constant
4030         multiplicands when the result overflows.
4031
4032 2014-04-25  Jakub Jelinek  <jakub@redhat.com>
4033
4034         PR tree-optimization/60960
4035         * tree-vect-generic.c (expand_vector_operation): Only call
4036         expand_vector_divmod if type's mode satisfies VECTOR_MODE_P.
4037
4038 2014-04-25  Tom de Vries  <tom@codesourcery.com>
4039
4040         * expr.c (clobber_reg_mode): New function.
4041         * expr.h (clobber_reg): New function.
4042
4043 2014-04-25  Tom de Vries  <tom@codesourcery.com>
4044
4045         * rtlanal.c (find_all_hard_reg_sets): Note INSN_CALL_FUNCTION_USAGE
4046         clobbers.
4047
4048 2014-04-25  Radovan Obradovic  <robradovic@mips.com>
4049             Tom de Vries  <tom@codesourcery.com>
4050
4051         * rtlanal.c (find_all_hard_reg_sets): Add bool implicit parameter and
4052         handle.
4053         * rtl.h (find_all_hard_reg_sets): Add bool parameter.
4054         * haifa-sched.c (recompute_todo_spec, check_clobbered_conditions): Add
4055         new argument to find_all_hard_reg_sets call.
4056
4057 2014-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4058
4059         * config/arm/aarch-common.c (aarch_rev16_shright_mask_imm_p):
4060         Use HOST_WIDE_INT_C for mask literal.
4061         (aarch_rev16_shleft_mask_imm_p): Likewise.
4062
4063 2014-04-25  Eric Botcazou  <ebotcazou@adacore.com>
4064
4065         PR target/60941
4066         * config/sparc/sparc.md (ashlsi3_extend): Delete.
4067
4068 2014-04-25  Marc Glisse  <marc.glisse@inria.fr>
4069
4070         PR preprocessor/56540
4071         * config/i386/i386-c.c (ix86_target_macros): Define
4072         __SIZEOF_FLOAT80__ and __SIZEOF_FLOAT128__.
4073
4074 2014-04-25  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
4075
4076         * configure.ac (tga_func): Remove.
4077         (LIB_TLS_SPEC): Remove.
4078         * configure: Regenerate.
4079         * config.in: Regenerate.
4080         * config/sol2.h (LIB_SPEC): Don't use LIB_TLS_SPEC.
4081
4082 2014-04-25  Richard Biener  <rguenther@suse.de>
4083
4084         PR ipa/60912
4085         * tree-ssa-structalias.c (ipa_pta_execute): Compute direct
4086         call stmt use/clobber sets during stmt walk instead of
4087         walking the possibly incomplete set of caller edges.
4088
4089 2014-04-25  Richard Biener  <rguenther@suse.de>
4090
4091         PR ipa/60911
4092         * passes.c (apply_ipa_transforms): Inline into only caller ...
4093         (execute_one_pass): ... here.  Properly bring in function
4094         bodies for nodes we want to apply IPA transforms to.
4095
4096 2014-04-24  Cong Hou  <congh@google.com>
4097
4098         PR tree-optimization/60896
4099         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Pick up
4100         all statements in PATTERN_DEF_SEQ in recognized widen-mult pattern.
4101         (vect_mark_pattern_stmts): Set the def type of all statements in
4102         PATTERN_DEF_SEQ as vect_internal_def.
4103
4104 2014-04-24  Michael Meissner  <meissner@linux.vnet.ibm.com>
4105
4106         * doc/extend.texi (PowerPC Built-in Functions): Document new
4107         powerpc extended divide, bcd, pack/unpack 128-bit, builtin functions.
4108         (PowerPC AltiVec/VSX Built-in Functions): Likewise.
4109
4110         * config/rs6000/predicates.md (const_0_to_3_operand): New
4111         predicate to match 0..3 integer constants.
4112
4113         * config/rs6000/rs6000-builtin.def (BU_DFP_MISC_1): Add new macros
4114         to support adding miscellaneous builtin functions.
4115         (BU_DFP_MISC_2): Likewise.
4116         (BU_P7_MISC_1): Likewise.
4117         (BU_P7_MISC_2): Likewise.
4118         (BU_P8V_MISC_3): Likewise.
4119         (BU_MISC_1): Likewise.
4120         (BU_MISC_2): Likewise.
4121         (DIVWE): Add extended divide builtin functions.
4122         (DIVWEO): Likewise.
4123         (DIVWEU): Likewise.
4124         (DIVWEUO): Likewise.
4125         (DIVDE): Likewise.
4126         (DIVDEO): Likewise.
4127         (DIVDEU): Likewise.
4128         (DIVDEUO): Likewise.
4129         (DXEX): Add decimal floating-point builtin functions.
4130         (DXEXQ): Likewise.
4131         (DDEDPD): Likewise.
4132         (DDEDPDQ): Likewise.
4133         (DENBCD): Likewise.
4134         (DENBCDQ): Likewise.
4135         (DIEX): Likewise.
4136         (DIEXQ): Likewise.
4137         (DSCLI): Likewise.
4138         (DSCLIQ): Likewise.
4139         (DSCRI): Likewise.
4140         (DSCRIQ): Likewise.
4141         (CDTBCD): Add new BCD builtin functions.
4142         (CBCDTD): Likewise.
4143         (ADDG6S): Likewise.
4144         (BCDADD): Likewise.
4145         (BCDADD_LT): Likewise.
4146         (BCDADD_EQ): Likewise.
4147         (BCDADD_GT): Likewise.
4148         (BCDADD_OV): Likewise.
4149         (BCDSUB): Likewise.
4150         (BCDSUB_LT): Likewise.
4151         (BCDSUB_EQ): Likewise.
4152         (BCDSUB_GT): Likewise.
4153         (BCDSUB_OV): Likewise.
4154         (PACK_TD): Add new pack/unpack 128-bit type builtin functions.
4155         (UNPACK_TD): Likewise.
4156         (PACK_TF): Likewise.
4157         (UNPACK_TF): Likewise.
4158         (UNPACK_TF_0): Likewise.
4159         (UNPACK_TF_1): Likewise.
4160         (PACK_V1TI): Likewise.
4161         (UNPACK_V1TI): Likewise.
4162
4163         * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
4164         support for decimal floating point builtin functions.
4165         (rs6000_expand_ternop_builtin): Add checks for the new builtin
4166         functions that take constant arguments.
4167         (rs6000_invalid_builtin): Add decimal floating point builtin support.
4168         (rs6000_init_builtins): Setup long double, _Decimal64, and
4169         _Decimal128 types for new builtin functions.
4170         (builtin_function_type): Set the unsigned flags appropriately for
4171         the new builtin functions.
4172         (rs6000_opt_masks): Add support for decimal floating point builtin
4173         functions.
4174
4175         * config/rs6000/rs6000.h (RS6000_BTM_DFP): Add support for decimal
4176         floating point builtin functions.
4177         (RS6000_BTM_COMMON): Likewise.
4178         (RS6000_BTI_long_double): Likewise.
4179         (RS6000_BTI_dfloat64): Likewise.
4180         (RS6000_BTI_dfloat128): Likewise.
4181         (long_double_type_internal_node): Likewise.
4182         (dfloat64_type_internal_node): Likewise.
4183         (dfloat128_type_internal_node): Likewise.
4184
4185         * config/rs6000/altivec.h (UNSPEC_BCDADD): Add support for ISA
4186         2.07 bcd arithmetic instructions.
4187         (UNSPEC_BCDSUB): Likewise.
4188         (UNSPEC_BCD_OVERFLOW): Likewise.
4189         (UNSPEC_BCD_ADD_SUB): Likewise.
4190         (bcd_add_sub): Likewise.
4191         (BCD_TEST): Likewise.
4192         (bcd<bcd_add_sub>): Likewise.
4193         (bcd<bcd_add_sub>_test): Likewise.
4194         (bcd<bcd_add_sub>_test2): Likewise.
4195         (bcd<bcd_add_sub>_<code>): Likewise.
4196         (peephole2 for combined bcd ops): Likewise.
4197
4198         * config/rs6000/dfp.md (UNSPEC_DDEDPD): Add support for new
4199         decimal floating point builtin functions.
4200         (UNSPEC_DENBCD): Likewise.
4201         (UNSPEC_DXEX): Likewise.
4202         (UNSPEC_DIEX): Likewise.
4203         (UNSPEC_DSCLI): Likewise.
4204         (UNSPEC_DSCRI): Likewise.
4205         (D64_D128): Likewise.
4206         (dfp_suffix): Likewise.
4207         (dfp_ddedpd_<mode>): Likewise.
4208         (dfp_denbcd_<mode>): Likewise.
4209         (dfp_dxex_<mode>): Likewise.
4210         (dfp_diex_<mode>): Likewise.
4211         (dfp_dscli_<mode>): Likewise.
4212         (dfp_dscri_<mode>): Likewise.
4213
4214         * config/rs6000/rs6000.md (UNSPEC_ADDG6S): Add support for new BCD
4215         builtin functions.
4216         (UNSPEC_CDTBCD): Likewise.
4217         (UNSPEC_CBCDTD): Likewise.
4218         (UNSPEC_DIVE): Add support for new extended divide builtin functions.
4219         (UNSPEC_DIVEO): Likewise.
4220         (UNSPEC_DIVEU): Likewise.
4221         (UNSPEC_DIVEUO): Likewise.
4222         (UNSPEC_UNPACK_128BIT): Add support for new builtin functions to
4223         pack/unpack 128-bit types.
4224         (UNSPEC_PACK_128BIT): Likewise.
4225         (idiv_ldiv): New mode attribute to set the 32/64-bit divide type.
4226         (udiv<mode>3): Use idiv_ldiv mode attribute.
4227         (div<mode>3): Likewise.
4228         (addg6s): Add new BCD builtin functions.
4229         (cdtbcd): Likewise.
4230         (cbcdtd): Likewise.
4231         (UNSPEC_DIV_EXTEND): Add support for new extended divide instructions.
4232         (div_extend): Likewise.
4233         (div<div_extend>_<mode>"): Likewise.
4234         (FP128_64): Add support for new builtin functions to pack/unpack
4235         128-bit types.
4236         (unpack<mode>): Likewise.
4237         (unpacktf_0): Likewise.
4238         (unpacktf_1): Likewise.
4239         (unpack<mode>_dm): Likewise.
4240         (unpack<mode>_nodm): Likewise.
4241         (pack<mode>): Likewise.
4242         (unpackv1ti): Likewise.
4243         (packv1ti): Likewise.
4244
4245 2014-04-24  Vishnu K S  <Vishnu.k_s@atmel.com>
4246
4247         * gcc/config/avr/avr.c: Add comment on why -fdelete-null-pointer-checks
4248         is disabled.
4249
4250 2014-04-24  Jakub Jelinek  <jakub@redhat.com>
4251
4252         * tree.h (OMP_CLAUSE_LINEAR_GIMPLE_SEQ): Define.
4253         * gimplify.c (omp_is_private): Change last argument's type to int.
4254         Only diagnose lastprivate if the simd argument is 1, only diagnose
4255         linear if the simd argument is 2.
4256         (gimplify_omp_for): Adjust omp_is_private callers.  When adding
4257         lastprivate or private, add the clause to OMP_FOR_CLAUSES.  Pass
4258         GOVD_EXPLICIT to omp_add_variable.  For simd with collapse == 1
4259         create OMP_CLAUSE_LINEAR rather than OMP_CLAUSE_PRIVATE for var.
4260         If var != decl and decl is in OMP_CLAUSE_LINEAR, gimplify decl
4261         increment to OMP_CLAUSE_LINEAR_GIMPLE_SEQ.
4262         * omp-low.c (scan_sharing_clauses, lower_lastprivate_clauses): Handle
4263         OMP_CLAUSE_LINEAR_GIMPLE_SEQ.
4264         * tree-nested.c (convert_nonlocal_omp_clauses,
4265         convert_local_omp_clauses): Handle OMP_CLAUSE_LINEAR.
4266
4267 2014-04-24  Segher Boessenkool  <segher@kernel.crashing.org>
4268
4269         PR target/60822
4270         * config/m68k/m68k.md (extendplussidi): Don't allow memory for
4271         operand 1.
4272
4273 2014-04-24  Dimitris Papavasiliou  <dpapavas@gmail.com>
4274
4275         * flag-types.h (enum ivar_visibility): Add.
4276
4277 2014-04-24  Trevor Saunders  <tsaunders@mozilla.com>
4278
4279         * config/sh/sh_treg_combine.c (sh_treg_combine::execute): Take
4280         function * argument.
4281
4282 2014-04-24  Alan Lawrence  <alan.lawrence@arm.com>
4283
4284         * config/aarch64/aarch64.c (aarch64_evpc_tbl): Enable for bigendian.
4285
4286 2014-04-24  Radovan Obradovic  <robradovic@mips.com>
4287             Tom de Vries  <tom@codesourcery.com>
4288
4289         * reg-notes.def (REG_NOTE (CALL_DECL)): New reg-note REG_CALL_DECL.
4290         * calls.c (expand_call, emit_library_call_value_1): Add REG_CALL_DECL
4291         reg-note.
4292         * combine.c (distribute_notes): Handle REG_CALL_DECL reg-note.
4293         * emit-rtl.c (try_split): Same.
4294
4295 2014-04-24  Radovan Obradovic  <robradovic@mips.com>
4296             Tom de Vries  <tom@codesourcery.com>
4297
4298         * common.opt (fuse-caller-save): New option.
4299
4300 2014-04-24  Tejas Belagod  <tejas.belagod@arm.com>
4301
4302         * config/aarch64/aarch64.c (aarch64_evpc_tbl): Reverse order of
4303         elements for big-endian.
4304
4305 2014-04-24  Richard Biener  <rguenther@suse.de>
4306
4307         * expr.c (expand_expr_real_1): Avoid gimple_assign_rhs_to_tree
4308         during TER and instead use the sepops interface for expanding
4309         non-GIMPLE_SINGLE_RHS.
4310
4311 2014-04-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
4312
4313         * config/i386/sol2.h (ASM_PREFERRED_EH_DATA_FORMAT): Only redefine
4314         if not HAVE_AS_IX86_DIFF_SECT_DELTA.
4315
4316 2014-04-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
4317
4318         * configure.ac (gcc_cv_as_cfi_directive): Support Solaris/x86
4319         assembler 64-bit option.
4320         * configure: Regenerate.
4321
4322 2014-04-24  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4323
4324         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Check
4325         TARGET_SIMD rather than TARGET_GENERAL_REGS_ONLY.
4326         (TARGET_SIMD): Take AARCH64_ISA_SIMD into account.
4327         (TARGET_FLOAT): Take AARCH64_ISA_FP into account.
4328         (TARGET_CRYPTO): Take TARGET_SIMD into account.
4329
4330 2014-04-24  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4331
4332         * config/aarch64/aarch64-builtins.c
4333         (aarch64_builtin_vectorized_function): Handle BUILT_IN_BSWAP16,
4334         BUILT_IN_BSWAP32, BUILT_IN_BSWAP64.
4335         * config/aarch64/aarch64-simd.md (bswap<mode>): New pattern.
4336         * config/aarch64/aarch64-simd-builtins.def: Define vector bswap
4337         builtins.
4338         * config/aarch64/iterator.md (VDQHSD): New mode iterator.
4339         (Vrevsuff): New mode attribute.
4340
4341 2014-04-24  Terry Guo  <terry.guo@arm.com>
4342
4343         * config/arm/arm.h (machine_function): Define variable
4344         after_arm_reorg here.
4345         * config/arm/arm.c (after_arm_reorg): Remove the definition.
4346         (arm_split_constant): Update the way to access variable
4347         after_arm_reorg.
4348         (arm_reorg): Ditto.
4349         (arm_output_function_epilogue): Remove the reset of after_arm_reorg.
4350
4351 2014-04-23  Tom de Vries  <tom@codesourcery.com>
4352
4353         * target-hooks-macros.h: Fix DEFHOOKPOD argument order in comment.
4354
4355 2014-04-23  David Malcolm  <dmalcolm@redhat.com>
4356
4357         * is-a.h: Update comments to reflect the following changes to the
4358         "pointerness" of the API, making the template parameter match the
4359         return type, allowing use of is-a.h with typedefs of pointers.
4360         (is_a_helper::cast): Return a T rather then a pointer to a T, so
4361         that the return type matches the parameter to the is_a_helper.
4362         (as_a): Likewise.
4363         (dyn_cast): Likewise.
4364
4365         * cgraph.c (cgraph_node_for_asm): Update for removal of implicit
4366         pointer from the is-a.h API.
4367
4368         * cgraph.h (is_a_helper <cgraph_node>::test): Convert to...
4369         (is_a_helper <cgraph_node *>::test): ...this, matching change to
4370         is-a.h API.
4371         (is_a_helper <varpool_node>::test): Likewise, convert to...
4372         (is_a_helper <varpool_node *>::test): ...this.
4373
4374         (varpool_first_variable): Update for removal of implicit pointer
4375         from the is-a.h API.
4376         (varpool_next_variable): Likewise.
4377         (varpool_first_static_initializer): Likewise.
4378         (varpool_next_static_initializer): Likewise.
4379         (varpool_first_defined_variable): Likewise.
4380         (varpool_next_defined_variable): Likewise.
4381         (cgraph_first_defined_function): Likewise.
4382         (cgraph_next_defined_function): Likewise.
4383         (cgraph_first_function): Likewise.
4384         (cgraph_next_function): Likewise.
4385         (cgraph_first_function_with_gimple_body): Likewise.
4386         (cgraph_next_function_with_gimple_body): Likewise.
4387         (cgraph_alias_target): Likewise.
4388         (varpool_alias_target): Likewise.
4389         (cgraph_function_or_thunk_node): Likewise.
4390         (varpool_variable_node): Likewise.
4391         (symtab_real_symbol_p): Likewise.
4392         * cgraphunit.c (referred_to_p): Likewise.
4393         (analyze_functions): Likewise.
4394         (handle_alias_pairs): Likewise.
4395         * gimple-fold.c (can_refer_decl_in_current_unit_p): Likewise.
4396         * gimple-ssa.h (gimple_vuse_op): Likewise.
4397         (gimple_vdef_op): Likewise.
4398         * gimple-streamer-in.c (input_gimple_stmt): Likewise.
4399         * gimple.c (gimple_build_asm_1): Likewise.
4400         (gimple_build_try): Likewise.
4401         (gimple_build_resx): Likewise.
4402         (gimple_build_eh_dispatch): Likewise.
4403         (gimple_build_omp_for): Likewise.
4404         (gimple_omp_for_set_clauses): Likewise.
4405
4406         * gimple.h (is_a_helper <gimple_statement_asm>::test): Convert to...
4407         (is_a_helper <gimple_statement_asm *>::test): ...this.
4408         (is_a_helper <gimple_statement_bind>::test): Convert to...
4409         (is_a_helper <gimple_statement_bind *>::test): ...this.
4410         (is_a_helper <gimple_statement_call>::test): Convert to...
4411         (is_a_helper <gimple_statement_call *>::test): ...this.
4412         (is_a_helper <gimple_statement_catch>::test): Convert to...
4413         (is_a_helper <gimple_statement_catch *>::test): ...this.
4414         (is_a_helper <gimple_statement_resx>::test): Convert to...
4415         (is_a_helper <gimple_statement_resx *>::test): ...this.
4416         (is_a_helper <gimple_statement_eh_dispatch>::test): Convert to...
4417         (is_a_helper <gimple_statement_eh_dispatch *>::test): ...this.
4418         (is_a_helper <gimple_statement_eh_else>::test): Convert to...
4419         (is_a_helper <gimple_statement_eh_else *>::test): ...this.
4420         (is_a_helper <gimple_statement_eh_filter>::test): Convert to...
4421         (is_a_helper <gimple_statement_eh_filter *>::test): ...this.
4422         (is_a_helper <gimple_statement_eh_mnt>::test): Convert to...
4423         (is_a_helper <gimple_statement_eh_mnt *>::test): ...this.
4424         (is_a_helper <gimple_statement_omp_atomic_load>::test): Convert to...
4425         (is_a_helper <gimple_statement_omp_atomic_load *>::test): ...this.
4426         (is_a_helper <gimple_statement_omp_atomic_store>::test): Convert to...
4427         (is_a_helper <gimple_statement_omp_atomic_store *>::test): ...this.
4428         (is_a_helper <gimple_statement_omp_return>::test): Convert to...
4429         (is_a_helper <gimple_statement_omp_return *>::test): ...this.
4430         (is_a_helper <gimple_statement_omp_continue>::test): Convert to...
4431         (is_a_helper <gimple_statement_omp_continue *>::test): ...this.
4432         (is_a_helper <gimple_statement_omp_critical>::test): Convert to...
4433         (is_a_helper <gimple_statement_omp_critical *>::test): ...this.
4434         (is_a_helper <gimple_statement_omp_for>::test): Convert to...
4435         (is_a_helper <gimple_statement_omp_for *>::test): ...this.
4436         (is_a_helper <gimple_statement_omp_taskreg>::test): Convert to...
4437         (is_a_helper <gimple_statement_omp_taskreg *>::test): ...this.
4438         (is_a_helper <gimple_statement_omp_parallel>::test): Convert to...
4439         (is_a_helper <gimple_statement_omp_parallel *>::test): ...this.
4440         (is_a_helper <gimple_statement_omp_target>::test): Convert to...
4441         (is_a_helper <gimple_statement_omp_target *>::test): ...this.
4442         (is_a_helper <gimple_statement_omp_sections>::test): Convert to...
4443         (is_a_helper <gimple_statement_omp_sections *>::test): ...this.
4444         (is_a_helper <gimple_statement_omp_single>::test): Convert to...
4445         (is_a_helper <gimple_statement_omp_single *>::test): ...this.
4446         (is_a_helper <gimple_statement_omp_teams>::test): Convert to...
4447         (is_a_helper <gimple_statement_omp_teams *>::test): ...this.
4448         (is_a_helper <gimple_statement_omp_task>::test): Convert to...
4449         (is_a_helper <gimple_statement_omp_task *>::test): ...this.
4450         (is_a_helper <gimple_statement_phi>::test): Convert to...
4451         (is_a_helper <gimple_statement_phi *>::test): ...this.
4452         (is_a_helper <gimple_statement_transaction>::test): Convert to...
4453         (is_a_helper <gimple_statement_transaction *>::test): ...this.
4454         (is_a_helper <gimple_statement_try>::test): Convert to...
4455         (is_a_helper <gimple_statement_try *>::test): ...this.
4456         (is_a_helper <gimple_statement_wce>::test): Convert to...
4457         (is_a_helper <gimple_statement_wce *>::test): ...this.
4458         (is_a_helper <const gimple_statement_asm>::test): Convert to...
4459         (is_a_helper <const gimple_statement_asm *>::test): ...this.
4460         (is_a_helper <const gimple_statement_bind>::test): Convert to...
4461         (is_a_helper <const gimple_statement_bind *>::test): ...this.
4462         (is_a_helper <const gimple_statement_call>::test): Convert to...
4463         (is_a_helper <const gimple_statement_call *>::test): ...this.
4464         (is_a_helper <const gimple_statement_catch>::test): Convert to...
4465         (is_a_helper <const gimple_statement_catch *>::test): ...this.
4466         (is_a_helper <const gimple_statement_resx>::test): Convert to...
4467         (is_a_helper <const gimple_statement_resx *>::test): ...this.
4468         (is_a_helper <const gimple_statement_eh_dispatch>::test): Convert to...
4469         (is_a_helper <const gimple_statement_eh_dispatch *>::test): ...this.
4470         (is_a_helper <const gimple_statement_eh_filter>::test): Convert to...
4471         (is_a_helper <const gimple_statement_eh_filter *>::test): ...this.
4472         (is_a_helper <const gimple_statement_omp_atomic_load>::test):
4473         Convert to...
4474         (is_a_helper <const gimple_statement_omp_atomic_load *>::test):
4475         ...this.
4476         (is_a_helper <const gimple_statement_omp_atomic_store>::test):
4477         Convert to...
4478         (is_a_helper <const gimple_statement_omp_atomic_store *>::test):
4479         ...this.
4480         (is_a_helper <const gimple_statement_omp_return>::test): Convert to...
4481         (is_a_helper <const gimple_statement_omp_return *>::test): ...this.
4482         (is_a_helper <const gimple_statement_omp_continue>::test): Convert
4483         to...
4484         (is_a_helper <const gimple_statement_omp_continue *>::test): ...this.
4485         (is_a_helper <const gimple_statement_omp_critical>::test): Convert
4486         to...
4487         (is_a_helper <const gimple_statement_omp_critical *>::test): ...this.
4488         (is_a_helper <const gimple_statement_omp_for>::test): Convert to...
4489         (is_a_helper <const gimple_statement_omp_for *>::test): ...this.
4490         (is_a_helper <const gimple_statement_omp_taskreg>::test): Convert to...
4491         (is_a_helper <const gimple_statement_omp_taskreg *>::test): ...this.
4492         (is_a_helper <const gimple_statement_omp_parallel>::test): Convert
4493         to...
4494         (is_a_helper <const gimple_statement_omp_parallel *>::test): ...this.
4495         (is_a_helper <const gimple_statement_omp_target>::test): Convert to...
4496         (is_a_helper <const gimple_statement_omp_target *>::test): ...this.
4497         (is_a_helper <const gimple_statement_omp_sections>::test): Convert
4498         to...
4499         (is_a_helper <const gimple_statement_omp_sections *>::test): ...this.
4500         (is_a_helper <const gimple_statement_omp_single>::test): Convert to...
4501         (is_a_helper <const gimple_statement_omp_single *>::test): ...this.
4502         (is_a_helper <const gimple_statement_omp_teams>::test): Convert to...
4503         (is_a_helper <const gimple_statement_omp_teams *>::test): ...this.
4504         (is_a_helper <const gimple_statement_omp_task>::test): Convert to...
4505         (is_a_helper <const gimple_statement_omp_task *>::test): ...this.
4506         (is_a_helper <const gimple_statement_phi>::test): Convert to...
4507         (is_a_helper <const gimple_statement_phi *>::test): ...this.
4508         (is_a_helper <const gimple_statement_transaction>::test): Convert to...
4509         (is_a_helper <const gimple_statement_transaction *>::test): ...this.
4510         (is_a_helper <const gimple_statement_with_ops>::test): Convert to...
4511         (is_a_helper <const gimple_statement_with_ops *>::test): ...this.
4512         (is_a_helper <gimple_statement_with_ops>::test): Convert to...
4513         (is_a_helper <gimple_statement_with_ops *>::test): ...this.
4514         (is_a_helper <const gimple_statement_with_memory_ops>::test): Convert
4515         to...
4516         (is_a_helper <const gimple_statement_with_memory_ops *>::test):
4517         ...this.
4518         (is_a_helper <gimple_statement_with_memory_ops>::test): Convert to...
4519         (is_a_helper <gimple_statement_with_memory_ops *>::test): ...this.
4520
4521         (gimple_use_ops): Update for removal of implicit pointer from the
4522         is-a.h API.
4523         (gimple_set_use_ops): Likewise.
4524         (gimple_vuse): Likewise.
4525         (gimple_vdef): Likewise.
4526         (gimple_vuse_ptr): Likewise.
4527         (gimple_vdef_ptr): Likewise.
4528         (gimple_set_vuse): Likewise.
4529         (gimple_set_vdef): Likewise.
4530         (gimple_omp_return_set_lhs): Likewise.
4531         (gimple_omp_return_lhs): Likewise.
4532         (gimple_omp_return_lhs_ptr): Likewise.
4533         (gimple_call_fntype): Likewise.
4534         (gimple_call_set_fntype): Likewise.
4535         (gimple_call_set_internal_fn): Likewise.
4536         (gimple_call_use_set): Likewise.
4537         (gimple_call_clobber_set): Likewise.
4538         (gimple_bind_vars): Likewise.
4539         (gimple_bind_set_vars): Likewise.
4540         (gimple_bind_body_ptr): Likewise.
4541         (gimple_bind_set_body): Likewise.
4542         (gimple_bind_add_stmt): Likewise.
4543         (gimple_bind_block): Likewise.
4544         (gimple_bind_set_block): Likewise.
4545         (gimple_asm_ninputs): Likewise.
4546         (gimple_asm_noutputs): Likewise.
4547         (gimple_asm_nclobbers): Likewise.
4548         (gimple_asm_nlabels): Likewise.
4549         (gimple_asm_input_op): Likewise.
4550         (gimple_asm_input_op_ptr): Likewise.
4551         (gimple_asm_output_op): Likewise.
4552         (gimple_asm_output_op_ptr): Likewise.
4553         (gimple_asm_set_output_op): Likewise.
4554         (gimple_asm_clobber_op): Likewise.
4555         (gimple_asm_set_clobber_op): Likewise.
4556         (gimple_asm_label_op): Likewise.
4557         (gimple_asm_set_label_op): Likewise.
4558         (gimple_asm_string): Likewise.
4559         (gimple_catch_types): Likewise.
4560         (gimple_catch_types_ptr): Likewise.
4561         (gimple_catch_handler_ptr): Likewise.
4562         (gimple_catch_set_types): Likewise.
4563         (gimple_catch_set_handler): Likewise.
4564         (gimple_eh_filter_types): Likewise.
4565         (gimple_eh_filter_types_ptr): Likewise.
4566         (gimple_eh_filter_failure_ptr): Likewise.
4567         (gimple_eh_filter_set_types): Likewise.
4568         (gimple_eh_filter_set_failure): Likewise.
4569         (gimple_eh_must_not_throw_fndecl): Likewise.
4570         (gimple_eh_must_not_throw_set_fndecl): Likewise.
4571         (gimple_eh_else_n_body_ptr): Likewise.
4572         (gimple_eh_else_e_body_ptr): Likewise.
4573         (gimple_eh_else_set_n_body): Likewise.
4574         (gimple_eh_else_set_e_body): Likewise.
4575         (gimple_try_eval_ptr): Likewise.
4576         (gimple_try_cleanup_ptr): Likewise.
4577         (gimple_try_set_eval): Likewise.
4578         (gimple_try_set_cleanup): Likewise.
4579         (gimple_wce_cleanup_ptr): Likewise.
4580         (gimple_wce_set_cleanup): Likewise.
4581         (gimple_phi_capacity): Likewise.
4582         (gimple_phi_num_args): Likewise.
4583         (gimple_phi_result): Likewise.
4584         (gimple_phi_result_ptr): Likewise.
4585         (gimple_phi_set_result): Likewise.
4586         (gimple_phi_arg): Likewise.
4587         (gimple_phi_set_arg): Likewise.
4588         (gimple_resx_region): Likewise.
4589         (gimple_resx_set_region): Likewise.
4590         (gimple_eh_dispatch_region): Likewise.
4591         (gimple_eh_dispatch_set_region): Likewise.
4592         (gimple_omp_critical_name): Likewise.
4593         (gimple_omp_critical_name_ptr): Likewise.
4594         (gimple_omp_critical_set_name): Likewise.
4595         (gimple_omp_for_clauses): Likewise.
4596         (gimple_omp_for_clauses_ptr): Likewise.
4597         (gimple_omp_for_set_clauses): Likewise.
4598         (gimple_omp_for_collapse): Likewise.
4599         (gimple_omp_for_index): Likewise.
4600         (gimple_omp_for_index_ptr): Likewise.
4601         (gimple_omp_for_set_index): Likewise.
4602         (gimple_omp_for_initial): Likewise.
4603         (gimple_omp_for_initial_ptr): Likewise.
4604         (gimple_omp_for_set_initial): Likewise.
4605         (gimple_omp_for_final): Likewise.
4606         (gimple_omp_for_final_ptr): Likewise.
4607         (gimple_omp_for_set_final): Likewise.
4608         (gimple_omp_for_incr): Likewise.
4609         (gimple_omp_for_incr_ptr): Likewise.
4610         (gimple_omp_for_set_incr): Likewise.
4611         (gimple_omp_for_pre_body_ptr): Likewise.
4612         (gimple_omp_for_set_pre_body): Likewise.
4613         (gimple_omp_parallel_clauses): Likewise.
4614         (gimple_omp_parallel_clauses_ptr): Likewise.
4615         (gimple_omp_parallel_set_clauses): Likewise.
4616         (gimple_omp_parallel_child_fn): Likewise.
4617         (gimple_omp_parallel_child_fn_ptr): Likewise.
4618         (gimple_omp_parallel_set_child_fn): Likewise.
4619         (gimple_omp_parallel_data_arg): Likewise.
4620         (gimple_omp_parallel_data_arg_ptr): Likewise.
4621         (gimple_omp_parallel_set_data_arg): Likewise.
4622         (gimple_omp_task_clauses): Likewise.
4623         (gimple_omp_task_clauses_ptr): Likewise.
4624         (gimple_omp_task_set_clauses): Likewise.
4625         (gimple_omp_task_child_fn): Likewise.
4626         (gimple_omp_task_child_fn_ptr): Likewise.
4627         (gimple_omp_task_set_child_fn): Likewise.
4628         (gimple_omp_task_data_arg): Likewise.
4629         (gimple_omp_task_data_arg_ptr): Likewise.
4630         (gimple_omp_task_set_data_arg): Likewise.
4631         (gimple_omp_taskreg_clauses): Likewise.
4632         (gimple_omp_taskreg_clauses_ptr): Likewise.
4633         (gimple_omp_taskreg_set_clauses): Likewise.
4634         (gimple_omp_taskreg_child_fn): Likewise.
4635         (gimple_omp_taskreg_child_fn_ptr): Likewise.
4636         (gimple_omp_taskreg_set_child_fn): Likewise.
4637         (gimple_omp_taskreg_data_arg): Likewise.
4638         (gimple_omp_taskreg_data_arg_ptr): Likewise.
4639         (gimple_omp_taskreg_set_data_arg): Likewise.
4640         (gimple_omp_task_copy_fn): Likewise.
4641         (gimple_omp_task_copy_fn_ptr): Likewise.
4642         (gimple_omp_task_set_copy_fn): Likewise.
4643         (gimple_omp_task_arg_size): Likewise.
4644         (gimple_omp_task_arg_size_ptr): Likewise.
4645         (gimple_omp_task_set_arg_size): Likewise.
4646         (gimple_omp_task_arg_align): Likewise.
4647         (gimple_omp_task_arg_align_ptr): Likewise.
4648         (gimple_omp_task_set_arg_align): Likewise.
4649         (gimple_omp_single_clauses): Likewise.
4650         (gimple_omp_single_clauses_ptr): Likewise.
4651         (gimple_omp_single_set_clauses): Likewise.
4652         (gimple_omp_target_clauses): Likewise.
4653         (gimple_omp_target_clauses_ptr): Likewise.
4654         (gimple_omp_target_set_clauses): Likewise.
4655         (gimple_omp_target_child_fn): Likewise.
4656         (gimple_omp_target_child_fn_ptr): Likewise.
4657         (gimple_omp_target_set_child_fn): Likewise.
4658         (gimple_omp_target_data_arg): Likewise.
4659         (gimple_omp_target_data_arg_ptr): Likewise.
4660         (gimple_omp_target_set_data_arg): Likewise.
4661         (gimple_omp_teams_clauses): Likewise.
4662         (gimple_omp_teams_clauses_ptr): Likewise.
4663         (gimple_omp_teams_set_clauses): Likewise.
4664         (gimple_omp_sections_clauses): Likewise.
4665         (gimple_omp_sections_clauses_ptr): Likewise.
4666         (gimple_omp_sections_set_clauses): Likewise.
4667         (gimple_omp_sections_control): Likewise.
4668         (gimple_omp_sections_control_ptr): Likewise.
4669         (gimple_omp_sections_set_control): Likewise.
4670         (gimple_omp_for_set_cond): Likewise.
4671         (gimple_omp_for_cond): Likewise.
4672         (gimple_omp_atomic_store_set_val): Likewise.
4673         (gimple_omp_atomic_store_val): Likewise.
4674         (gimple_omp_atomic_store_val_ptr): Likewise.
4675         (gimple_omp_atomic_load_set_lhs): Likewise.
4676         (gimple_omp_atomic_load_lhs): Likewise.
4677         (gimple_omp_atomic_load_lhs_ptr): Likewise.
4678         (gimple_omp_atomic_load_set_rhs): Likewise.
4679         (gimple_omp_atomic_load_rhs): Likewise.
4680         (gimple_omp_atomic_load_rhs_ptr): Likewise.
4681         (gimple_omp_continue_control_def): Likewise.
4682         (gimple_omp_continue_control_def_ptr): Likewise.
4683         (gimple_omp_continue_set_control_def): Likewise.
4684         (gimple_omp_continue_control_use): Likewise.
4685         (gimple_omp_continue_control_use_ptr): Likewise.
4686         (gimple_omp_continue_set_control_use): Likewise.
4687         (gimple_transaction_body_ptr): Likewise.
4688         (gimple_transaction_label): Likewise.
4689         (gimple_transaction_label_ptr): Likewise.
4690         (gimple_transaction_set_body): Likewise.
4691         (gimple_transaction_set_label): Likewise.
4692
4693         * ipa-devirt.c (build_type_inheritance_graph): Likewise.
4694         * ipa-inline-analysis.c (inline_write_summary): Likewise.
4695         * ipa-ref.c (ipa_record_reference): Likewise.
4696         * ipa-reference.c (analyze_function): Likewise.
4697         (ipa_reference_write_optimization_summary): Likewise.
4698         * ipa.c (symtab_remove_unreachable_nodes): Likewise.
4699         (address_taken_from_non_vtable_p): Likewise.
4700         (comdat_can_be_unshared_p_1): Likewise.
4701         * lto-cgraph.c (lto_output_ref): Likewise.
4702         (add_references): Likewise.
4703         (compute_ltrans_boundary): Likewise.
4704         (output_symtab): Likewise.
4705         (input_ref): Likewise.
4706         (input_cgraph_1): Likewise.
4707         (output_cgraph_opt_summary): Likewise.
4708         * lto-streamer-out.c (lto_output): Likewise.
4709         (output_symbol_p): Likewise.
4710         * lto-streamer.h (lsei_next_function_in_partition): Likewise.
4711         (lsei_start_function_in_partition): Likewise.
4712         (lsei_next_variable_in_partition): Likewise.
4713         (lsei_start_variable_in_partition): Likewise.
4714         * symtab.c (insert_to_assembler_name_hash): Likewise.
4715         (unlink_from_assembler_name_hash): Likewise.
4716         (symtab_unregister_node): Likewise.
4717         (symtab_remove_node): Likewise.
4718         (dump_symtab_node): Likewise.
4719         (verify_symtab_base): Likewise.
4720         (verify_symtab_node): Likewise.
4721         (symtab_make_decl_local): Likewise.
4722         (symtab_alias_ultimate_target): Likewise.
4723         (symtab_resolve_alias): Likewise.
4724         (symtab_get_symbol_partitioning_class): Likewise.
4725         * tree-phinodes.c (allocate_phi_node): Likewise.
4726         (reserve_phi_args_for_new_edge): Likewise.
4727         (remove_phi_args): Likewise.
4728         * varpool.c (varpool_node_for_asm): Likewise.
4729         (varpool_remove_unreferenced_decls): Likewise.
4730
4731 2014-04-23  Jeff Law  <law@redhat.com>
4732
4733         PR tree-optimization/60902
4734         * tree-ssa-threadedge.c
4735         (record_temporary_equivalences_from_stmts_at_dest): Make sure to
4736         invalidate outputs from statements that do not produce useful
4737         outputs for threading.
4738
4739 2014-04-23  Venkataramanan Kumar  <venkataramanan.kumar@linaro.org>
4740
4741         * config/aarch64/aarch64.md (stack_protect_set, stack_protect_test)
4742         (stack_protect_set_<mode>, stack_protect_test_<mode>): Add
4743         machine descriptions for Stack Smashing Protector.
4744
4745 2014-04-23  Richard Earnshaw  <rearnsha@arm.com>
4746
4747         * aarch64.md (<optab>_rol<mode>3): New pattern.
4748         (<optab>_rolsi3_uxtw): Likewise.
4749         * aarch64.c (aarch64_strip_shift): Handle ROTATE and ROTATERT.
4750
4751 2014-04-23  James Greenhalgh  <james.greenhalgh@arm.com>
4752
4753         * config/arm/arm.c (arm_cortex_a57_tune): Initialize all fields.
4754         (arm_cortex_a12_tune): Likewise.
4755
4756 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4757
4758         * config/aarch64/aarch64.c (aarch64_rtx_costs): Handle BSWAP.
4759
4760 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4761
4762         * config/arm/arm.md (arm_rev16si2): New pattern.
4763         (arm_rev16si2_alt): Likewise.
4764         * config/arm/arm.c (arm_new_rtx_costs): Handle rev16 case.
4765
4766 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4767
4768         * config/aarch64/aarch64.md (rev16<mode>2): New pattern.
4769         (rev16<mode>2_alt): Likewise.
4770         * config/aarch64/aarch64.c (aarch64_rtx_costs): Handle rev16 case.
4771         * config/arm/aarch-common.c (aarch_rev16_shright_mask_imm_p): New.
4772         (aarch_rev16_shleft_mask_imm_p): Likewise.
4773         (aarch_rev16_p_1): Likewise.
4774         (aarch_rev16_p): Likewise.
4775         * config/arm/aarch-common-protos.h (aarch_rev16_p): Declare extern.
4776         (aarch_rev16_shright_mask_imm_p): Likewise.
4777         (aarch_rev16_shleft_mask_imm_p): Likewise.
4778
4779 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4780
4781         * config/arm/aarch-common-protos.h (alu_cost_table): Add rev field.
4782         * config/arm/aarch-cost-tables.h (generic_extra_costs): Specify
4783         rev cost.
4784         (cortex_a53_extra_costs): Likewise.
4785         (cortex_a57_extra_costs): Likewise.
4786         * config/arm/arm.c (cortexa9_extra_costs): Likewise.
4787         (cortexa7_extra_costs): Likewise.
4788         (cortexa8_extra_costs): Likewise.
4789         (cortexa12_extra_costs): Likewise.
4790         (cortexa15_extra_costs): Likewise.
4791         (v7m_extra_costs): Likewise.
4792         (arm_new_rtx_costs): Handle BSWAP.
4793
4794 2013-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4795
4796         * config/arm/arm.c (cortexa8_extra_costs): New table.
4797         (arm_cortex_a8_tune): New tuning struct.
4798         * config/arm/arm-cores.def (cortex-a8): Use cortex_a8 tuning struct.
4799
4800 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4801
4802         * config/arm/arm.c (arm_new_rtx_costs): Handle FMA.
4803
4804 2014-04-23  Richard Biener  <rguenther@suse.de>
4805
4806         * Makefile.in (OBJS): Remove loop-unswitch.o.
4807         * tree-pass.h (make_pass_rtl_unswitch): Remove.
4808         * passes.def (pass_rtl_unswitch): Likewise.
4809         * loop-init.c (gate_rtl_unswitch): Likewise.
4810         (rtl_unswitch): Likewise.
4811         (pass_data_rtl_unswitch): Likewise.
4812         (pass_rtl_unswitch): Likewise.
4813         (make_pass_rtl_unswitch): Likewise.
4814         * rtl.h (reversed_condition): Likewise.
4815         (compare_and_jump_seq): Likewise.
4816         * loop-iv.c (reversed_condition): Move here from loop-unswitch.c
4817         and make static.
4818         * loop-unroll.c (compare_and_jump_seq): Likewise.
4819
4820 2014-04-23  Richard Biener  <rguenther@suse.de>
4821
4822         PR tree-optimization/60903
4823         * tree-ssa-loop-im.c (analyze_memory_references): Remove
4824         commented code block.
4825         (execute_sm_if_changed): Properly apply IRREDUCIBLE_LOOP
4826         loop flags to newly created BBs and edges.
4827
4828 2014-04-23  Nick Clifton  <nickc@redhat.com>
4829
4830         * config/msp430/msp430.c (msp430_handle_option): Move function
4831         to msp430-common.c
4832         (msp430_option_override): Simplify mcu and mcpu option handling.
4833         (msp430_is_f5_mcu): Rename to msp430_use_f5_series_hwmult.  Add
4834         support for -mhwmult command line option.
4835         (has_32bit_hwmult): Rename to use_32bit_hwmult.  Add support for
4836         -mhwmult command line option.
4837         (msp430_hwmult_enabled): Delete.
4838         (msp43o_output_labelref): Add support for -mhwmult command line option.
4839         * config/msp430/msp430.md (mulhisi3, umulhisi3, mulsidi3)
4840         (umulsidi3): Likewise.
4841         * config/msp430/msp430.opt (mmcu): Add Report attribute.
4842         (mcpu, mlarge, msmall): Likewise.
4843         (mhwmult): New option.
4844         * config/msp430/msp430-protos.h (msp430_hwmult_enabled): Remove
4845         prototype.
4846         (msp430_is_f5_mcu): Remove prototype.
4847         (msp430_use_f5_series_hwmult): Add prototype.
4848         * config/msp430/msp430-opts.h: New file.
4849         * common/config/msp430: New directory.
4850         * common/config/msp430/msp430-common.c: New file.
4851         * config.gcc (msp430): Remove target_has_targetm_common.
4852         * doc/invoke.texi: Document -mhwmult command line option.
4853
4854 2014-04-23  Nick Clifton  <nickc@redhat.com>
4855
4856         * config/i386/cygwin.h (ENDFILE_SPEC): Include
4857         default-manifest.o if it can be found in the search path.
4858         * config/i386/mingw32.h (ENDFILE_SPEC): Likewise.
4859
4860 2014-04-23  Terry Guo  <terry.guo@arm.com>
4861
4862         * config/arm/arm.h (ASM_APP_OFF): Re-define it in a cleaner way.
4863
4864 2014-04-23  Richard Biener  <rguenther@suse.de>
4865
4866         PR middle-end/60895
4867         * tree-inline.c (declare_return_variable): Use mark_addressable.
4868
4869 2014-04-23  Richard Biener  <rguenther@suse.de>
4870
4871         PR middle-end/60891
4872         * loop-init.c (loop_optimizer_init): Make sure to apply
4873         LOOPS_MAY_HAVE_MULTIPLE_LATCHES before fixing up loops.
4874
4875 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
4876
4877         PR sanitizer/60275
4878         * common.opt (fsanitize-recover, fsanitize-undefined-trap-on-error):
4879         New options.
4880         * gcc.c (sanitize_spec_function): Don't return "" for "undefined"
4881         if flag_sanitize_undefined_trap_on_error.
4882         * sanitizer.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW_ABORT,
4883         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS_ABORT,
4884         BUILT_IN_UBSAN_HANDLE_VLA_BOUND_NOT_POSITIVE_ABORT,
4885         BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
4886         BUILT_IN_UBSAN_HANDLE_ADD_OVERFLOW_ABORT,
4887         BUILT_IN_UBSAN_HANDLE_SUB_OVERFLOW_ABORT,
4888         BUILT_IN_UBSAN_HANDLE_MUL_OVERFLOW_ABORT,
4889         BUILT_IN_UBSAN_HANDLE_NEGATE_OVERFLOW_ABORT,
4890         BUILT_IN_UBSAN_HANDLE_LOAD_INVALID_VALUE_ABORT): New builtins.
4891         * ubsan.c (ubsan_instrument_unreachable): Return
4892         __builtin_trap () if flag_sanitize_undefined_trap_on_error.
4893         (ubsan_expand_null_ifn): Emit __builtin_trap ()
4894         if flag_sanitize_undefined_trap_on_error and
4895         __ubsan_handle_type_mismatch_abort if !flag_sanitize_recover.
4896         (ubsan_expand_null_ifn, ubsan_build_overflow_builtin,
4897         instrument_bool_enum_load): Emit __builtin_trap () if
4898         flag_sanitize_undefined_trap_on_error and
4899         __builtin_handle_*_abort () if !flag_sanitize_recover.
4900         * doc/invoke.texi (-fsanitize-recover,
4901         -fsanitize-undefined-trap-on-error): Document.
4902
4903 2014-04-22  Christian Bruel  <christian.bruel@st.com>
4904
4905         * config/sh/sh.md (mov<mode>): Replace movQIHI.
4906          Force immediates to SImode.
4907
4908 2014-04-22  Sandra Loosemore  <sandra@codesourcery.com>
4909
4910         * config/nios2/nios2.md (UNSPEC_ROUND): New.
4911         (lroundsfsi2): New.
4912         * config/nios2/nios2.opt (mno-custom-round, mcustom-round=): New.
4913         * config/nios2/nios2-opts.h (N2FPU_ALL_CODES): Add round.
4914         * config/nios2/nios2.c (N2F_NO_ERRNO): Define.
4915         (nios2_fpu_insn): Add entry for round.
4916         (N2FPU_NO_ERRNO_P): Define.
4917         (nios2_custom_check_insns): Add check for N2F_NO_ERRNO and
4918         flag_errno_math.
4919         * doc/invoke.texi (Nios II Options): Document -mcustom-round.
4920
4921 2014-04-22  Richard Henderson  <rth@redhat.com>
4922
4923         * config/aarch64/aarch64 (addti3, subti3): New expanders.
4924         (add<GPI>3_compare0): Remove leading * from name.
4925         (add<GPI>3_carryin): Likewise.
4926         (sub<GPI>3_compare0): Likewise.
4927         (sub<GPI>3_carryin): Likewise.
4928         (<su_optab>mulditi3): New expander.
4929         (multi3): New expander.
4930         (madd<GPI>): Remove leading * from name.
4931
4932 2014-04-22  Martin Jambor  <mjambor@suse.cz>
4933
4934         * cgraphclones.c (cgraph_function_versioning): Copy
4935         ipa_transforms_to_apply instead of asserting it is empty.
4936
4937 2014-04-22  H.J. Lu  <hongjiu.lu@intel.com>
4938
4939         PR target/60868
4940         * config/i386/i386.c (ix86_expand_set_or_movmem): Call counter_mode
4941         on count_exp to get mode.
4942
4943 2014-04-22  Andrew Pinski  <apinski@cavium.com>
4944
4945         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
4946         Handle TLS for ILP32.
4947         * config/aarch64/aarch64.md (tlsie_small): Rename to ...
4948         (tlsie_small_<mode>): this and handle PTR.
4949         (tlsie_small_sidi): New pattern.
4950         (tlsle_small): Change to an expand to handle ILP32.
4951         (tlsle_small_<mode>): New pattern.
4952         (tlsdesc_small): Rename to ...
4953         (tlsdesc_small_<mode>): this and handle PTR.
4954
4955 2014-04-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
4956
4957         * config/aarch64/aarch64.c (TARGET_FLAGS_REGNUM): Define.
4958
4959 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
4960
4961         * config/aarch64/aarch64-builtins.c (TYPES_REINTERP): Removed.
4962         (aarch64_types_signed_unsigned_qualifiers): Qualifier added.
4963         (aarch64_types_signed_poly_qualifiers): Likewise.
4964         (aarch64_types_unsigned_signed_qualifiers): Likewise.
4965         (aarch64_types_poly_signed_qualifiers): Likewise.
4966         (TYPES_REINTERP_SS): Type macro added.
4967         (TYPES_REINTERP_SU): Likewise.
4968         (TYPES_REINTERP_SP): Likewise.
4969         (TYPES_REINTERP_US): Likewise.
4970         (TYPES_REINTERP_PS): Likewise.
4971         (aarch64_fold_builtin): New expression folding added.
4972         * config/aarch64/aarch64-simd-builtins.def (REINTERP):
4973         Declarations removed.
4974         (REINTERP_SS): Declarations added.
4975         (REINTERP_US): Likewise.
4976         (REINTERP_PS): Likewise.
4977         (REINTERP_SU): Likewise.
4978         (REINTERP_SP): Likewise.
4979         * config/aarch64/arm_neon.h (vreinterpret_p8_f64): Implemented.
4980         (vreinterpretq_p8_f64): Likewise.
4981         (vreinterpret_p16_f64): Likewise.
4982         (vreinterpretq_p16_f64): Likewise.
4983         (vreinterpret_f32_f64): Likewise.
4984         (vreinterpretq_f32_f64): Likewise.
4985         (vreinterpret_f64_f32): Likewise.
4986         (vreinterpret_f64_p8): Likewise.
4987         (vreinterpret_f64_p16): Likewise.
4988         (vreinterpret_f64_s8): Likewise.
4989         (vreinterpret_f64_s16): Likewise.
4990         (vreinterpret_f64_s32): Likewise.
4991         (vreinterpret_f64_s64): Likewise.
4992         (vreinterpret_f64_u8): Likewise.
4993         (vreinterpret_f64_u16): Likewise.
4994         (vreinterpret_f64_u32): Likewise.
4995         (vreinterpret_f64_u64): Likewise.
4996         (vreinterpretq_f64_f32): Likewise.
4997         (vreinterpretq_f64_p8): Likewise.
4998         (vreinterpretq_f64_p16): Likewise.
4999         (vreinterpretq_f64_s8): Likewise.
5000         (vreinterpretq_f64_s16): Likewise.
5001         (vreinterpretq_f64_s32): Likewise.
5002         (vreinterpretq_f64_s64): Likewise.
5003         (vreinterpretq_f64_u8): Likewise.
5004         (vreinterpretq_f64_u16): Likewise.
5005         (vreinterpretq_f64_u32): Likewise.
5006         (vreinterpretq_f64_u64): Likewise.
5007         (vreinterpret_s64_f64): Likewise.
5008         (vreinterpretq_s64_f64): Likewise.
5009         (vreinterpret_u64_f64): Likewise.
5010         (vreinterpretq_u64_f64): Likewise.
5011         (vreinterpret_s8_f64): Likewise.
5012         (vreinterpretq_s8_f64): Likewise.
5013         (vreinterpret_s16_f64): Likewise.
5014         (vreinterpretq_s16_f64): Likewise.
5015         (vreinterpret_s32_f64): Likewise.
5016         (vreinterpretq_s32_f64): Likewise.
5017         (vreinterpret_u8_f64): Likewise.
5018         (vreinterpretq_u8_f64): Likewise.
5019         (vreinterpret_u16_f64): Likewise.
5020         (vreinterpretq_u16_f64): Likewise.
5021         (vreinterpret_u32_f64): Likewise.
5022         (vreinterpretq_u32_f64): Likewise.
5023
5024 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
5025
5026         * config/aarch64/aarch64/aarch64-builtins.c (TYPES_REINTERP): Removed.
5027         * config/aarch64/aarch64/aarch64-simd-builtins.def (REINTERP): Removed.
5028         (vreinterpret_p8_s8): Likewise.
5029         * config/aarch64/aarch64/arm_neon.h (vreinterpret_p8_s8): Uses cast.
5030         (vreinterpret_p8_s16): Likewise.
5031         (vreinterpret_p8_s32): Likewise.
5032         (vreinterpret_p8_s64): Likewise.
5033         (vreinterpret_p8_f32): Likewise.
5034         (vreinterpret_p8_u8): Likewise.
5035         (vreinterpret_p8_u16): Likewise.
5036         (vreinterpret_p8_u32): Likewise.
5037         (vreinterpret_p8_u64): Likewise.
5038         (vreinterpret_p8_p16): Likewise.
5039         (vreinterpretq_p8_s8): Likewise.
5040         (vreinterpretq_p8_s16): Likewise.
5041         (vreinterpretq_p8_s32): Likewise.
5042         (vreinterpretq_p8_s64): Likewise.
5043         (vreinterpretq_p8_f32): Likewise.
5044         (vreinterpretq_p8_u8): Likewise.
5045         (vreinterpretq_p8_u16): Likewise.
5046         (vreinterpretq_p8_u32): Likewise.
5047         (vreinterpretq_p8_u64): Likewise.
5048         (vreinterpretq_p8_p16): Likewise.
5049         (vreinterpret_p16_s8): Likewise.
5050         (vreinterpret_p16_s16): Likewise.
5051         (vreinterpret_p16_s32): Likewise.
5052         (vreinterpret_p16_s64): Likewise.
5053         (vreinterpret_p16_f32): Likewise.
5054         (vreinterpret_p16_u8): Likewise.
5055         (vreinterpret_p16_u16): Likewise.
5056         (vreinterpret_p16_u32): Likewise.
5057         (vreinterpret_p16_u64): Likewise.
5058         (vreinterpret_p16_p8): Likewise.
5059         (vreinterpretq_p16_s8): Likewise.
5060         (vreinterpretq_p16_s16): Likewise.
5061         (vreinterpretq_p16_s32): Likewise.
5062         (vreinterpretq_p16_s64): Likewise.
5063         (vreinterpretq_p16_f32): Likewise.
5064         (vreinterpretq_p16_u8): Likewise.
5065         (vreinterpretq_p16_u16): Likewise.
5066         (vreinterpretq_p16_u32): Likewise.
5067         (vreinterpretq_p16_u64): Likewise.
5068         (vreinterpretq_p16_p8): Likewise.
5069         (vreinterpret_f32_s8): Likewise.
5070         (vreinterpret_f32_s16): Likewise.
5071         (vreinterpret_f32_s32): Likewise.
5072         (vreinterpret_f32_s64): Likewise.
5073         (vreinterpret_f32_u8): Likewise.
5074         (vreinterpret_f32_u16): Likewise.
5075         (vreinterpret_f32_u32): Likewise.
5076         (vreinterpret_f32_u64): Likewise.
5077         (vreinterpret_f32_p8): Likewise.
5078         (vreinterpret_f32_p16): Likewise.
5079         (vreinterpretq_f32_s8): Likewise.
5080         (vreinterpretq_f32_s16): Likewise.
5081         (vreinterpretq_f32_s32): Likewise.
5082         (vreinterpretq_f32_s64): Likewise.
5083         (vreinterpretq_f32_u8): Likewise.
5084         (vreinterpretq_f32_u16): Likewise.
5085         (vreinterpretq_f32_u32): Likewise.
5086         (vreinterpretq_f32_u64): Likewise.
5087         (vreinterpretq_f32_p8): Likewise.
5088         (vreinterpretq_f32_p16): Likewise.
5089         (vreinterpret_s64_s8): Likewise.
5090         (vreinterpret_s64_s16): Likewise.
5091         (vreinterpret_s64_s32): Likewise.
5092         (vreinterpret_s64_f32): Likewise.
5093         (vreinterpret_s64_u8): Likewise.
5094         (vreinterpret_s64_u16): Likewise.
5095         (vreinterpret_s64_u32): Likewise.
5096         (vreinterpret_s64_u64): Likewise.
5097         (vreinterpret_s64_p8): Likewise.
5098         (vreinterpret_s64_p16): Likewise.
5099         (vreinterpretq_s64_s8): Likewise.
5100         (vreinterpretq_s64_s16): Likewise.
5101         (vreinterpretq_s64_s32): Likewise.
5102         (vreinterpretq_s64_f32): Likewise.
5103         (vreinterpretq_s64_u8): Likewise.
5104         (vreinterpretq_s64_u16): Likewise.
5105         (vreinterpretq_s64_u32): Likewise.
5106         (vreinterpretq_s64_u64): Likewise.
5107         (vreinterpretq_s64_p8): Likewise.
5108         (vreinterpretq_s64_p16): Likewise.
5109         (vreinterpret_u64_s8): Likewise.
5110         (vreinterpret_u64_s16): Likewise.
5111         (vreinterpret_u64_s32): Likewise.
5112         (vreinterpret_u64_s64): Likewise.
5113         (vreinterpret_u64_f32): Likewise.
5114         (vreinterpret_u64_u8): Likewise.
5115         (vreinterpret_u64_u16): Likewise.
5116         (vreinterpret_u64_u32): Likewise.
5117         (vreinterpret_u64_p8): Likewise.
5118         (vreinterpret_u64_p16): Likewise.
5119         (vreinterpretq_u64_s8): Likewise.
5120         (vreinterpretq_u64_s16): Likewise.
5121         (vreinterpretq_u64_s32): Likewise.
5122         (vreinterpretq_u64_s64): Likewise.
5123         (vreinterpretq_u64_f32): Likewise.
5124         (vreinterpretq_u64_u8): Likewise.
5125         (vreinterpretq_u64_u16): Likewise.
5126         (vreinterpretq_u64_u32): Likewise.
5127         (vreinterpretq_u64_p8): Likewise.
5128         (vreinterpretq_u64_p16): Likewise.
5129         (vreinterpret_s8_s16): Likewise.
5130         (vreinterpret_s8_s32): Likewise.
5131         (vreinterpret_s8_s64): Likewise.
5132         (vreinterpret_s8_f32): Likewise.
5133         (vreinterpret_s8_u8): Likewise.
5134         (vreinterpret_s8_u16): Likewise.
5135         (vreinterpret_s8_u32): Likewise.
5136         (vreinterpret_s8_u64): Likewise.
5137         (vreinterpret_s8_p8): Likewise.
5138         (vreinterpret_s8_p16): Likewise.
5139         (vreinterpretq_s8_s16): Likewise.
5140         (vreinterpretq_s8_s32): Likewise.
5141         (vreinterpretq_s8_s64): Likewise.
5142         (vreinterpretq_s8_f32): Likewise.
5143         (vreinterpretq_s8_u8): Likewise.
5144         (vreinterpretq_s8_u16): Likewise.
5145         (vreinterpretq_s8_u32): Likewise.
5146         (vreinterpretq_s8_u64): Likewise.
5147         (vreinterpretq_s8_p8): Likewise.
5148         (vreinterpretq_s8_p16): Likewise.
5149         (vreinterpret_s16_s8): Likewise.
5150         (vreinterpret_s16_s32): Likewise.
5151         (vreinterpret_s16_s64): Likewise.
5152         (vreinterpret_s16_f32): Likewise.
5153         (vreinterpret_s16_u8): Likewise.
5154         (vreinterpret_s16_u16): Likewise.
5155         (vreinterpret_s16_u32): Likewise.
5156         (vreinterpret_s16_u64): Likewise.
5157         (vreinterpret_s16_p8): Likewise.
5158         (vreinterpret_s16_p16): Likewise.
5159         (vreinterpretq_s16_s8): Likewise.
5160         (vreinterpretq_s16_s32): Likewise.
5161         (vreinterpretq_s16_s64): Likewise.
5162         (vreinterpretq_s16_f32): Likewise.
5163         (vreinterpretq_s16_u8): Likewise.
5164         (vreinterpretq_s16_u16): Likewise.
5165         (vreinterpretq_s16_u32): Likewise.
5166         (vreinterpretq_s16_u64): Likewise.
5167         (vreinterpretq_s16_p8): Likewise.
5168         (vreinterpretq_s16_p16): Likewise.
5169         (vreinterpret_s32_s8): Likewise.
5170         (vreinterpret_s32_s16): Likewise.
5171         (vreinterpret_s32_s64): Likewise.
5172         (vreinterpret_s32_f32): Likewise.
5173         (vreinterpret_s32_u8): Likewise.
5174         (vreinterpret_s32_u16): Likewise.
5175         (vreinterpret_s32_u32): Likewise.
5176         (vreinterpret_s32_u64): Likewise.
5177         (vreinterpret_s32_p8): Likewise.
5178         (vreinterpret_s32_p16): Likewise.
5179         (vreinterpretq_s32_s8): Likewise.
5180         (vreinterpretq_s32_s16): Likewise.
5181         (vreinterpretq_s32_s64): Likewise.
5182         (vreinterpretq_s32_f32): Likewise.
5183         (vreinterpretq_s32_u8): Likewise.
5184         (vreinterpretq_s32_u16): Likewise.
5185         (vreinterpretq_s32_u32): Likewise.
5186         (vreinterpretq_s32_u64): Likewise.
5187         (vreinterpretq_s32_p8): Likewise.
5188         (vreinterpretq_s32_p16): Likewise.
5189         (vreinterpret_u8_s8): Likewise.
5190         (vreinterpret_u8_s16): Likewise.
5191         (vreinterpret_u8_s32): Likewise.
5192         (vreinterpret_u8_s64): Likewise.
5193         (vreinterpret_u8_f32): Likewise.
5194         (vreinterpret_u8_u16): Likewise.
5195         (vreinterpret_u8_u32): Likewise.
5196         (vreinterpret_u8_u64): Likewise.
5197         (vreinterpret_u8_p8): Likewise.
5198         (vreinterpret_u8_p16): Likewise.
5199         (vreinterpretq_u8_s8): Likewise.
5200         (vreinterpretq_u8_s16): Likewise.
5201         (vreinterpretq_u8_s32): Likewise.
5202         (vreinterpretq_u8_s64): Likewise.
5203         (vreinterpretq_u8_f32): Likewise.
5204         (vreinterpretq_u8_u16): Likewise.
5205         (vreinterpretq_u8_u32): Likewise.
5206         (vreinterpretq_u8_u64): Likewise.
5207         (vreinterpretq_u8_p8): Likewise.
5208         (vreinterpretq_u8_p16): Likewise.
5209         (vreinterpret_u16_s8): Likewise.
5210         (vreinterpret_u16_s16): Likewise.
5211         (vreinterpret_u16_s32): Likewise.
5212         (vreinterpret_u16_s64): Likewise.
5213         (vreinterpret_u16_f32): Likewise.
5214         (vreinterpret_u16_u8): Likewise.
5215         (vreinterpret_u16_u32): Likewise.
5216         (vreinterpret_u16_u64): Likewise.
5217         (vreinterpret_u16_p8): Likewise.
5218         (vreinterpret_u16_p16): Likewise.
5219         (vreinterpretq_u16_s8): Likewise.
5220         (vreinterpretq_u16_s16): Likewise.
5221         (vreinterpretq_u16_s32): Likewise.
5222         (vreinterpretq_u16_s64): Likewise.
5223         (vreinterpretq_u16_f32): Likewise.
5224         (vreinterpretq_u16_u8): Likewise.
5225         (vreinterpretq_u16_u32): Likewise.
5226         (vreinterpretq_u16_u64): Likewise.
5227         (vreinterpretq_u16_p8): Likewise.
5228         (vreinterpretq_u16_p16): Likewise.
5229         (vreinterpret_u32_s8): Likewise.
5230         (vreinterpret_u32_s16): Likewise.
5231         (vreinterpret_u32_s32): Likewise.
5232         (vreinterpret_u32_s64): Likewise.
5233         (vreinterpret_u32_f32): Likewise.
5234         (vreinterpret_u32_u8): Likewise.
5235         (vreinterpret_u32_u16): Likewise.
5236         (vreinterpret_u32_u64): Likewise.
5237         (vreinterpret_u32_p8): Likewise.
5238         (vreinterpret_u32_p16): Likewise.
5239         (vreinterpretq_u32_s8): Likewise.
5240         (vreinterpretq_u32_s16): Likewise.
5241         (vreinterpretq_u32_s32): Likewise.
5242         (vreinterpretq_u32_s64): Likewise.
5243         (vreinterpretq_u32_f32): Likewise.
5244         (vreinterpretq_u32_u8): Likewise.
5245         (vreinterpretq_u32_u16): Likewise.
5246         (vreinterpretq_u32_u64): Likewise.
5247         (vreinterpretq_u32_p8): Likewise.
5248         (vreinterpretq_u32_p16): Likewise.
5249
5250 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
5251
5252         * gcc/config/aarch64/aarch64-simd.md (aarch64_s<optab><mode>):
5253         Pattern extended.
5254         * config/aarch64/aarch64-simd-builtins.def (sqneg): Iterator extended.
5255         (sqabs): Likewise.
5256         * config/aarch64/arm_neon.h (vqneg_s64): New intrinsic.
5257         (vqnegd_s64): Likewise.
5258         (vqabs_s64): Likewise.
5259         (vqabsd_s64): Likewise.
5260
5261 2014-04-22  Richard Henderson  <rth@redhat.com>
5262
5263         * config/sparc/sparc.c (sparc_init_modes): Hoist GET_MODE_SIZE
5264         computation to the top of the loop.
5265
5266 2014-04-22  Renlin  <renlin.li@arm.com>
5267             Jiong Wang  <jiong.wang@arm.com>
5268
5269         * config/aarch64/aarch64.h (aarch64_frame): Delete "fp_lr_offset".
5270         * config/aarch64/aarch64.c (aarch64_layout_frame)
5271         (aarch64_initial_elimination_offset): Likewise.
5272
5273 2014-04-22  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5274
5275         * config/aarch64/aarch64.c (aarch64_initial_elimination_offset):
5276         Fix indentation.
5277
5278 2014-04-22  Richard Sandiford  <rdsandiford@googlemail.com>
5279
5280         * machmode.h (bitwise_mode_for_mode): Declare.
5281         * stor-layout.h (bitwise_type_for_mode): Likewise.
5282         * stor-layout.c (bitwise_mode_for_mode): New function.
5283         (bitwise_type_for_mode): Likewise.
5284         * builtins.c (fold_builtin_memory_op): Use it instead of
5285         int_mode_for_mode and build_nonstandard_integer_type.
5286
5287 2014-04-22  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
5288
5289         * config.gcc (enable_obsolete): Remove *-*-solaris2.9*.
5290         (*-*-solaris2.[0-9] | *-*-solaris2.[0-9].*): Mark unsupported.
5291         (*-*-solaris2*): Simplify.
5292         (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Likewise.
5293         (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Remove
5294         *-*-solaris2.9* handling.
5295
5296         * configure.ac (gcc_cv_as_hidden): Remove test for Solaris 9/x86
5297         as bug.
5298         (gcc_cv_ld_hidden): Remove *-*-solaris2.9* handling.
5299         (ld_tls_support): Remove i?86-*-solaris2.9, sparc*-*-solaris2.9
5300         handling, simplify.
5301         (gcc_cv_as_gstabs_flag): Remove workaround for Solaris 9/x86 as bug.
5302         * configure: Regenerate.
5303
5304         * config/i386/sol2-9.h: Remove.
5305
5306         * doc/install.texi (Specific, i?86-*-solaris2.9): Remove.
5307         (Specific, *-*-solaris2*): Mention Solaris 9 support removal.
5308         Remove Solaris 9 references.
5309
5310 2014-04-22  Vidya Praveen  <vidyapraveen@arm.com>
5311
5312         * aarch64.md (float<GPI:mode><GPF:mode>2): Remove.
5313         (floatuns<GPI:mode><GPF:mode>2): Remove.
5314         (<optab><fcvt_target><GPF:mode>2): New pattern for equal width float
5315         and floatuns conversions.
5316         (<optab><fcvt_iesize><GPF:mode>2): New pattern for inequal width float
5317         and floatuns conversions.
5318         * iterators.md (fcvt_target, FCVT_TARGET): Support SF and DF modes.
5319         (w1,w2): New mode attributes for inequal width conversions.
5320
5321 2014-04-22  Renlin Li  <Renlin.Li@arm.com>
5322
5323         * config/aarch64/aarch64.c (aarch64_print_operand_address): Adjust
5324         the output asm format.
5325
5326 2014-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
5327
5328         * config/aarch64/aarch64-simd.md
5329         (aarch64_cm<optab>di): Always split.
5330         (*aarch64_cm<optab>di): New.
5331         (aarch64_cmtstdi): Always split.
5332         (*aarch64_cmtstdi): New.
5333
5334 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
5335
5336         PR tree-optimization/60823
5337         * omp-low.c (ipa_simd_modify_function_body): Go through
5338         all SSA_NAMEs and for those refering to vector arguments
5339         which are going to be replaced adjust SSA_NAME_VAR and,
5340         if it is a default definition, change it into a non-default
5341         definition assigned at the beginning of function from new_decl.
5342         (ipa_simd_modify_stmt_ops): Rewritten.
5343         * tree-dfa.c (set_ssa_default_def): When removing default def,
5344         check for NULL loc instead of NULL *loc.
5345
5346 2014-04-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
5347
5348         * config/arm/arm.c (arm_hard_regno_mode_ok): Loosen
5349         restrictions on core registers for DImode values in Thumb2.
5350
5351 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
5352
5353         * config/arm/arm.md (*anddi_notdi_zesidi): New pattern.
5354         * config/arm/thumb2.md (*iordi_notdi_zesidi): New pattern.
5355
5356 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
5357
5358         * config/arm/thumb2.md (*iordi_notdi_di): New pattern.
5359         (*iordi_notzesidi_di): Likewise.
5360         (*iordi_notsesidi_di): Likewise.
5361
5362 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
5363
5364         * config/arm/arm-protos.h (tune_params): New struct members.
5365         * config/arm/arm.c: Initialise tune_params per processor.
5366         (thumb2_reorg): Suppress conversion from t32 to t16 when optimizing
5367         for speed, based on new tune_params.
5368
5369 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
5370
5371         * config/aarch64/aarch64-builtins.c (BUILTIN_VDQF_DF): Macro added.
5372         * config/aarch64/aarch64-simd-builtins.def (frintn): Use added macro.
5373         * config/aarch64/aarch64-simd.md (<frint_pattern>): Comment corrected.
5374         * config/aarch64/aarch64.md (<frint_pattern>): Likewise.
5375         * config/aarch64/arm_neon.h (vrnd_f64): Added.
5376         (vrnda_f64): Likewise.
5377         (vrndi_f64): Likewise.
5378         (vrndm_f64): Likewise.
5379         (vrndn_f64): Likewise.
5380         (vrndp_f64): Likewise.
5381         (vrndx_f64): Likewise.
5382
5383 2014-04-22  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
5384
5385         * config/arm/arm.c (arm_print_operand, thumb_exit): Make sure
5386         GET_MODE_SIZE argument is enum machine_mode.
5387
5388 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
5389
5390         PR target/60910
5391         * config/sparc/sparc.c (sparc_init_modes): Pass enum machine_mode
5392         value instead of int to GET_MODE_CLASS and GET_MODE_SIZE macros.
5393
5394 2014-04-22  Lin Zuojian  <manjian2006@gmail.com>
5395
5396         PR middle-end/60281
5397         * asan.c (asan_emit_stack_protection): Force the base to align to
5398         appropriate bits if STRICT_ALIGNMENT.  Set shadow_mem align to
5399         appropriate bits if STRICT_ALIGNMENT.
5400         * cfgexpand.c (expand_stack_vars): Set base_align appropriately
5401         when asan is on.
5402         (expand_used_vars): Leave a space in the stack frame for alignment
5403         if STRICT_ALIGNMENT.
5404
5405 2014-04-21  David Malcolm  <dmalcolm@redhat.com>
5406
5407         * gimple.h (gimple_assign_single_p): Accept a const_gimple rather
5408         than a gimple.
5409         (gimple_store_p): Likewise.
5410         (gimple_assign_load_p): Likewise.
5411         (gimple_assign_cast_p): Likewise.
5412         (gimple_clobber_p): Likewise.
5413
5414         * doc/gimple.texi (gimple_assign_cast_p): Accept a const_gimple
5415         rather than a gimple.
5416         (gimple_assign_cast_p): Likewise.
5417
5418 2014-04-21  Michael Meissner  <meissner@linux.vnet.ibm.com>
5419
5420         PR target/60735
5421         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64 case):
5422         If mode is DDmode and TARGET_E500_DOUBLE allow move.
5423
5424         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print some
5425         more debug information for E500 if -mdebug=reg.
5426
5427 2014-04-21  Uros Bizjak  <ubizjak@gmail.com>
5428
5429         PR target/60909
5430         * config/i386/i386.c (ix86_expand_builtin)
5431         <case IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Use temporary
5432         register for target RTX.
5433         <case IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Ditto.
5434
5435 2014-04-18  Cong Hou  <congh@google.com>
5436
5437         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Enhance
5438         the widen-mult pattern by handling two operands with different sizes,
5439         and operands whose size is smaller than half of the result type.
5440
5441 2014-04-18  Jan Hubicka  <hubicka@ucw.cz>
5442
5443         * ipa-inline.h (INLINE_HINT_known_hot): New hint.
5444         * ipa-inline-analysis.c (dump_inline_hints): Dump it.
5445         (do_estimate_edge_time): Compute it.
5446         * ipa-inline.c (want_inline_small_function_p): Bypass
5447         INLINE_INSNS_AUTO/SINGLE limits for calls that are known to be hot.
5448
5449 2014-04-18  Jan Hubicka  <hubicka@ucw.cz>
5450
5451         * ipa-inline.c (spec_rem): New static variable.
5452         (dump_overall_stats): New function.
5453         (dump_inline_stats): New function.
5454
5455 2014-04-18  Richard Henderson  <rth@redhat.com>
5456
5457         * config/aarch64/aarch64.c (aarch64_register_move_cost): Pass a mode
5458         to GET_MODE_SIZE, not a reg_class_t.
5459
5460 2014-04-18  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5461
5462         * config/rs6000/vsx.md (vsx_xxmrghw_<mode>): Adjust for little-endian.
5463         (vsx_xxmrglw_<mode>): Likewise.
5464
5465 2014-04-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
5466
5467         PR target/60876
5468         * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Make sure
5469         GET_MODE_SIZE gets passed an enum machine_mode type and not integer.
5470         (rs6000_init_hard_regno_mode_ok): Likewise.
5471
5472 2014-04-17  Jan Hubicka  <hubicka@ucw.cz>
5473
5474         * ipa-inline.c (inline_small_functions): Account only non-cold
5475         functions.
5476         * doc/invoke.texi (inline-unit-growth): Update documentation.
5477
5478 2014-04-17  Pat Haugen  <pthaugen@us.ibm.com>
5479
5480         * config/rs6000/rs6000.md (addti3, subti3): New.
5481
5482 2014-04-17  H.J. Lu  <hongjiu.lu@intel.com>
5483
5484         PR target/60863
5485         * config/i386/i386.c (ix86_expand_clear): Remove outdated
5486         comment.  Check optimize_insn_for_size_p instead of
5487         optimize_insn_for_speed_p.
5488
5489 2014-04-17  Martin Jambor  <mjambor@suse.cz>
5490
5491         * gimple-iterator.c (gsi_start_edge): New function.
5492         * gimple-iterator.h (gsi_start_edge): Declare.
5493         * tree-sra.c (single_non_eh_succ): New function.
5494         (disqualify_ops_if_throwing_stmt): Renamed to
5495         disqualify_if_bad_bb_terminating_stmt.  Allow throwing statements
5496         having one non-EH successor BB.
5497         (sra_modify_expr): If stmt ends bb, use single non-EH successor to
5498         generate loads into replacements.
5499         (sra_modify_assign): Likewise and and also use the simple path for
5500         such statements.
5501         (sra_modify_function_body): Commit statements on edges.
5502
5503 2014-04-17  Richard Biener  <rguenther@suse.de>
5504
5505         PR middle-end/60849
5506         * tree-ssa-propagate.c (valid_gimple_rhs_p): Allow vector
5507         comparison results and add clarifying comment.
5508
5509 2014-04-17  Jakub Jelinek  <jakub@redhat.com>
5510
5511         * genmodes.c (struct mode_data): Add need_bytesize_adj field.
5512         (blank_mode): Initialize it.
5513         (emit_mode_size_inline, emit_mode_nunits_inline,
5514         emit_mode_inner_inline): New functions.
5515         (emit_insn_modes_h): Call them and surround their output with
5516         #if GCC_VERSION >= 4001 ... #endif.
5517         * machmode.h (GET_MODE_SIZE, GET_MODE_NUNITS, GET_MODE_INNER):
5518         For GCC_VERSION >= 4001 use mode_*_inline routines instead of
5519         mode_* arrays if the argument is __builtin_constant_p.
5520         * lower-subreg.c (dump_choices): Make sure GET_MODE_SIZE argument
5521         is enum machine_mode.
5522
5523 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
5524
5525         * passes.c (opt_pass::execute): Adjust.
5526         (pass_manager::execute_pass_mode_switching): Likewise.
5527         (early_local_passes::execute): Likewise.
5528         (execute_one_pass): Pass cfun to the pass's execute method.
5529         * tree-pass.h (opt_pass::execute): Add function * argument.
5530         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
5531         cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
5532         compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
5533         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
5534         config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
5535         config/sparc/sparc.c, cprop.c, dce.c, df-core.c, dse.c, dwarf2cfi.c,
5536         except.c, final.c, function.c, fwprop.c, gcse.c, gimple-low.c,
5537         gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
5538         graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
5539         ipa-inline-analysis.c, ipa-inline.c, ipa-profile.c, ipa-pure-const.c,
5540         ipa-reference.c, ipa-split.c, ipa.c, ira.c, jump.c, loop-init.c,
5541         lower-subreg.c, mode-switching.c, omp-low.c, postreload-gcse.c,
5542         postreload.c, predict.c, recog.c, ree.c, reg-stack.c, regcprop.c,
5543         reginfo.c, regrename.c, reorg.c, sched-rgn.c, stack-ptr-mod.c,
5544         store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
5545         tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
5546         tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
5547         tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-ssa-ccp.c,
5548         tree-ssa-copy.c, tree-ssa-copyrename.c, tree-ssa-dce.c,
5549         tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
5550         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
5551         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
5552         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
5553         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
5554         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
5555         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
5556         tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
5557         tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
5558         tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
5559         Adjust.
5560
5561 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
5562
5563         * passes.c (opt_pass::gate): Take function * argument.
5564         (gate_all_early_local_passes): Merge into
5565         (early_local_passes::gate): this.
5566         (gate_all_early_optimizations): Merge into
5567         (all_early_optimizations::gate): this.
5568         (gate_all_optimizations): Mege into
5569         (all_optimizations::gate): this.
5570         (gate_all_optimizations_g): Merge into
5571         (all_optimizations_g::gate): this.
5572         (gate_rest_of_compilation): Mege into
5573         (rest_of_compilation::gate): this.
5574         (gate_postreload): Merge into
5575         (postreload::gate): this.
5576         (dump_one_pass): Pass cfun to the pass's gate method.
5577         (execute_ipa_summary_passes): Likewise.
5578         (execute_one_pass): Likewise.
5579         (ipa_write_summaries_2): Likewise.
5580         (ipa_write_optimization_summaries_1): Likewise.
5581         (ipa_read_summaries_1): Likewise.
5582         (ipa_read_optimization_summaries_1): Likewise.
5583         (execute_ipa_stmt_fixups): Likewise.
5584         * tree-pass.h (opt_pass::gate): Add function * argument.
5585         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c,
5586         combine-stack-adj.c, combine.c, compare-elim.c,
5587         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
5588         config/rl78/rl78.c, config/sh/sh_optimize_sett_clrt.cc,
5589         config/sh/sh_treg_combine.cc, config/sparc/sparc.c, cprop.c, cse.c,
5590         dce.c, df-core.c, dse.c, dwarf2cfi.c, except.c,  fwprop.c, gcse.c,
5591         gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
5592         graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
5593         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
5594         loop-init.c, lower-subreg.c, mode-switching.c, modulo-sched.c,
5595         omp-low.c, postreload-gcse.c, postreload.c, predict.c, recog.c, ree.c,
5596         reg-stack.c, regcprop.c, regrename.c, reorg.c, sched-rgn.c,
5597         store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
5598         tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
5599         tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c,
5600         tree-nrv.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
5601         tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
5602         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
5603         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
5604         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
5605         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
5606         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
5607         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
5608         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
5609         tree-ssa.c, tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
5610         tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c, tsan.c, ubsan.c,
5611         var-tracking.c, vtable-verify.c, web.c: Adjust.
5612
5613 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
5614
5615         * configure.ac: Check for -Woverloaded-virtual and enable it if found.
5616         * configure: Regenerate.
5617
5618 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
5619
5620         * passes.c (dump_one_pass): don't check pass->has_gate.
5621         (execute_ipa_summary_passes): Likewise.
5622         (execute_one_pass): Likewise.
5623         (ipa_write_summaries_2): Likewise.
5624         (ipa_write_optimization_summaries_1): Likewise.
5625         (ipa_read_optimization_summaries_1): Likewise.
5626         (execute_ipa_stmt_fixups): Likewise.
5627         * tree-pass.h (pass_data::has_gate): Remove.
5628         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
5629         cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
5630         compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
5631         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
5632         config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
5633         config/sh/sh_optimize_sett_clrt.cc, config/sh/sh_treg_combine.cc,
5634         config/sparc/sparc.c, cprop.c, cse.c, dce.c, df-core.c, dse.c,
5635         dwarf2cfi.c, except.c, final.c, function.c, fwprop.c, gcse.c,
5636         gimple-low.c, gimple-ssa-isolate-paths.c,
5637         gimple-ssa-strength-reduction.c, graphite.c, ifcvt.c, init-regs.c,
5638         ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
5639         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
5640         ira.c, jump.c, loop-init.c, lower-subreg.c, mode-switching.c,
5641         modulo-sched.c, omp-low.c, postreload-gcse.c, postreload.c, predict.c,
5642         recog.c, ree.c, reg-stack.c, regcprop.c, reginfo.c, regrename.c,
5643         reorg.c, sched-rgn.c, stack-ptr-mod.c, store-motion.c, tracer.c,
5644         trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
5645         tree-complex.c, tree-eh.c, tree-emutls.c, tree-if-conv.c,
5646         tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
5647         tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
5648         tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
5649         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
5650         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
5651         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
5652         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
5653         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
5654         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
5655         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
5656         tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
5657         tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
5658         tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
5659         Adjust.
5660
5661 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
5662
5663         * pass_manager.h (pass_manager::register_dump_files_1): Remove
5664         declaration.
5665         * passes.c (pass_manager::register_dump_files_1): Merge into
5666         (pass_manager::register_dump_files): this, and remove its handling of
5667         properties since the pass always has the properties anyway.
5668         (pass_manager::pass_manager): Adjust.
5669
5670 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
5671
5672         * pass_manager.h (pass_manager::register_dump_files_1): Adjust.
5673         * passes.c (pass_manager::register_dump_files_1): Remove dead code
5674         dealing with properties.
5675         (pass_manager::register_dump_files): Adjust.
5676
5677 2014-03-20  Mark Wielaard  <mjw@redhat.com>
5678
5679         * dwarf2out.c (add_bound_info): If HOST_WIDE_INT is big enough,
5680         then represent the bound as normal constant value.
5681
5682 2014-04-17  Jakub Jelinek  <jakub@redhat.com>
5683
5684         PR target/60847
5685         Forward port from 4.8 branch
5686         2013-07-19  Kirill Yukhin  <kirill.yukhin@intel.com>
5687
5688         * config/i386/bmiintrin.h (_blsi_u32): New.
5689         (_blsi_u64): Ditto.
5690         (_blsr_u32): Ditto.
5691         (_blsr_u64): Ditto.
5692         (_blsmsk_u32): Ditto.
5693         (_blsmsk_u64): Ditto.
5694         (_tzcnt_u32): Ditto.
5695         (_tzcnt_u64): Ditto.
5696
5697 2014-04-17  Kito Cheng  <kito@0xlab.org>
5698
5699         * gcc.c (used_arg): Prevent out of bound access for multilib_options.
5700
5701 2014-04-17  Richard Biener  <rguenther@suse.de>
5702
5703         PR middle-end/60849
5704         * tree-ssa-propagate.c (valid_gimple_rhs_p): Only allow effective
5705         boolean results for comparisons.
5706
5707 2014-04-17  Richard Biener  <rguenther@suse.de>
5708
5709         PR tree-optimization/60836
5710         * tree-vect-loop.c (vect_create_epilog_for_reduction): Force
5711         initial PHI args to be gimple values.
5712
5713 2014-04-17   Richard Biener  <rguenther@suse.de>
5714
5715         PR tree-optimization/60841
5716         * tree-vect-data-refs.c (vect_analyze_data_refs): Count stmts.
5717         * tree-vect-loop.c (vect_analyze_loop_2): Pass down number
5718         of stmts to SLP build.
5719         * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
5720         (vect_analyze_slp): Likewise.
5721         (vect_analyze_slp_instance): Likewise.
5722         (vect_build_slp_tree): Limit overall SLP tree growth.
5723         * tree-vectorizer.h (vect_analyze_data_refs,
5724         vect_analyze_slp): Adjust prototypes.
5725
5726 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
5727
5728         * config/i386/i386.c (x86_add_stmt_cost): Fix vector cost model for
5729         Silvermont.
5730
5731 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
5732
5733         * config/i386/x86-tune.def (TARGET_SLOW_PSHUFB): New tune definition.
5734         * config/i386/i386.h (TARGET_SLOW_PSHUFB): New tune flag.
5735         * config/i386/i386.c (expand_vec_perm_even_odd_1): Avoid byte shuffles
5736         for TARGET_SLOW_PSHUFB
5737
5738 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
5739
5740         * config/i386/i386.c (slm_cost): Adjust vec_to_scalar_cost.
5741         * config/i386/i386.c (intel_cost): Ditto.
5742
5743 2014-04-17  Joey Ye  <joey.ye@arm.com>
5744
5745         * opts.c (OPT_fif_conversion, OPT_fif_conversion2): Disable for Og.
5746
5747 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
5748
5749         * opts.c (common_handle_option): Disable -fipa-reference coorectly
5750         with -fuse-profile.
5751
5752 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
5753
5754         * ipa-devirt.c (odr_type_d): Add field all_derivations_known.
5755         (type_all_derivations_known_p): New predicate.
5756         (type_all_ctors_visible_p): New predicate.
5757         (type_possibly_instantiated_p): New predicate.
5758         (get_odr_type): Compute all_derivations_known.
5759         (dump_odr_type): Dump the flag.
5760         (maybe_record_type): Cleanup.
5761         (record_target_from_binfo): Add bases_to_consider array;
5762         record bases for types w/o instances and skip CXX destructor.
5763         (possible_polymorphic_call_targets_1): Add bases_to_consider
5764         and consider_construction parameters; check if type may have instance.
5765         (get_polymorphic_call_info): Set maybe_in_construction to true
5766         when we know nothing.
5767         (record_targets_from_bases): Skip CXX destructors; they are
5768         never called for types in construction.
5769         (possible_polymorphic_call_targets): Do not record target when
5770         type may not have instance.
5771
5772 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
5773
5774         PR ipa/60854
5775         * ipa.c (symtab_remove_unreachable_nodes): Mark targets of
5776         external aliases alive, too.
5777
5778 2014-04-16  Andrew  Pinski  <apinski@cavium.com>
5779
5780         * config/host-linux.c (TRY_EMPTY_VM_SPACE): Change aarch64 ilp32
5781         definition.
5782
5783 2014-04-16  Eric Botcazou  <ebotcazou@adacore.com>
5784
5785         * final.c (compute_alignments): Do not apply loop alignment to a block
5786         falling through to the exit.
5787
5788 2014-04-16  Catherine Moore  <clm@codesourcery.com>
5789
5790         * mips.md (*mov<mode>_internal, *movhi_internal, *movqi_internal):
5791         Adjust constraints for microMIPS store patterns.
5792
5793 2014-04-16  Pitchumani Sivanupandi  <Pitchumani.S@atmel.com>
5794
5795         * config/avr/avr-mcus.def: Correct typo for atxmega256a3bu macro.
5796
5797 2014-04-16  Eric Botcazou  <ebotcazou@adacore.com>
5798
5799         * tree-ssa-operands.c (create_vop_var): Set DECL_IGNORED_P.
5800         (append_use): Run at -O0.
5801         (append_vdef): Likewise.
5802         * tree-ssa-ter.c (ter_is_replaceable_p): Do not special-case -O0.
5803         * tree-ssa-uninit.c (warn_uninitialized_vars): Remove obsolete comment.
5804
5805 2014-04-16  Jakub Jelinek  <jakub@redhat.com>
5806
5807         PR tree-optimization/60844
5808         * tree-ssa-reassoc.c (reassoc_remove_stmt): New function.
5809         (propagate_op_to_single_use, remove_visited_stmt_chain,
5810         linearize_expr, repropagate_negates, reassociate_bb): Use it
5811         instead of gsi_remove.
5812
5813 2014-04-16  Martin Jambor  <mjambor@suse.cz>
5814
5815         * cgraphclones.c (cgraph_create_virtual_clone): Duplicate
5816         ipa_transforms_to_apply.
5817         (cgraph_function_versioning): Assert that old_node has empty
5818         ipa_transforms_to_apply.
5819         * trans-mem.c (ipa_tm_create_version): Likewise.
5820         * tree-inline.c (tree_function_versioning): Do not duplicate
5821         ipa_transforms_to_apply.
5822
5823 2014-04-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
5824
5825         PR target/60817
5826         * configure.ac (set_have_as_tls): Merge i[34567]86-*-* and
5827         x86_64-*-* cases.
5828         Pass necessary as flags on 64-bit Solaris/x86.
5829         Use lowercase relocs for x86_64-*-*.
5830         * configure: Regenerate.
5831
5832 2014-04-15  Jan Hubicka  <jh@suse.cz>
5833
5834         * ipa-devirt.c (referenced_from_vtable_p): New predicate.
5835         (maybe_record_node, likely_target_p): Use it.
5836
5837 2014-04-15  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5838
5839         PR target/60839
5840         Revert following patch
5841
5842         2014-04-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
5843
5844         PR target/60735
5845         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
5846         software floating point or no floating point registers, do not
5847         allow any type in the FPRs.  Eliminate a test for SPE SIMD types
5848         in GPRs that occurs after we tested for GPRs that would never be
5849         true.
5850
5851         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
5852         Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
5853         since the FMOVE64 type is DFmode/DDmode.  If TARGET_E500_DOUBLE,
5854         specifically allow DDmode, since that does not use the SPE SIMD
5855         instructions.
5856
5857 2014-03-21  Mark Wielaard  <mjw@redhat.com>
5858
5859         * dwarf2out.c (gen_enumeration_type_die): Add DW_AT_const_value
5860         as unsigned or int depending on type and value used.
5861
5862 2014-04-15  Richard Biener  <rguenther@suse.de>
5863
5864         PR rtl-optimization/56965
5865         * alias.c (ncr_compar, nonoverlapping_component_refs_p): Move ...
5866         * tree-ssa-alias.c (ncr_compar, nonoverlapping_component_refs_p):
5867         ... here.
5868         * alias.c (true_dependence_1): Do not call
5869         nonoverlapping_component_refs_p.
5870         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Call
5871         nonoverlapping_component_refs_p.
5872         (indirect_refs_may_alias_p): Likewise.
5873
5874 2014-04-15  Teresa Johnson  <tejohnson@google.com>
5875
5876         * cfg.c (dump_bb_info): Fix flags check.
5877         * tree-cfg.c (remove_bb): Only dump TDF_BLOCKS when removing.
5878
5879 2014-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5880
5881         PR rtl-optimization/60663
5882         * config/arm/arm.c (arm_new_rtx_costs): Improve ASM_OPERANDS case,
5883         avoid 0 cost.
5884
5885 2014-04-15  Richard Biener  <rguenther@suse.de>
5886
5887         * lto-streamer.h (LTO_major_version): Bump to 4.
5888
5889 2014-04-15  Richard Biener  <rguenther@suse.de>
5890
5891         * common.opt (lto_partition_model): New enum.
5892         (flto-partition=): Merge separate options with a single with argument,
5893         add -flto-partition=one support.
5894         * flag-types.h (enum lto_partition_model): Declare.
5895         * opts.c (finish_options): Remove duplicate -flto-partition=
5896         option check.
5897         * lto-wrapper.c (run_gcc): Adjust.
5898
5899 2014-04-15  Richard Biener  <rguenther@suse.de>
5900
5901         * alias.c (ncr_compar): New function.
5902         (nonoverlapping_component_refs_p): Re-implement in O (n log n).
5903
5904 2014-04-15  Richard Biener  <rguenther@suse.de>
5905
5906         * alias.c (record_component_aliases): Do not walk BINFOs.
5907
5908 2014-04-15  Richard Biener  <rguenther@suse.de>
5909
5910         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
5911         Add struct function argument and adjust.
5912         (find_func_aliases_for_call): Likewise.
5913         (find_func_aliases): Likewise.
5914         (find_func_clobbers): Likewise.
5915         (intra_create_variable_infos): Likewise.
5916         (compute_points_to_sets): Likewise.
5917         (ipa_pta_execute): Adjust.  Do not push/pop cfun.
5918
5919 2014-04-15  Richard Biener  <rguenther@suse.de>
5920
5921         * tree.c (iterative_hash_expr): Use enum tree_code_class
5922         to store TREE_CODE_CLASS.
5923         (tree_block): Likewise.
5924         (tree_set_block): Likewise.
5925         * tree.h (fold_build_pointer_plus_loc): Use
5926         convert_to_ptrofftype_loc.
5927
5928 2014-04-15  Jakub Jelinek  <jakub@redhat.com>
5929
5930         PR plugins/59335
5931         * Makefile.in (PLUGIN_HEADERS): Add various headers that have been
5932         added in 4.9.
5933
5934 2014-04-15  Eric Botcazou  <ebotcazou@adacore.com>
5935
5936         * cfgloop.h (struct loop): Move force_vectorize down.
5937         * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Handle new kinds.
5938         (gimplify_expr) <ANNOTATE_EXPR>: Minor tweak.
5939         * lto-streamer-in.c (input_cfg): Read dont_vectorize field.
5940         * lto-streamer-out.c (output_cfg): Write dont_vectorize field.
5941         * tree-cfg.c (replace_loop_annotate): Revamp and handle new kinds.
5942         * tree-core.h (enum annot_expr_kind): Add new kind values.
5943         * tree-inline.c (copy_loops): Copy dont_vectorize field and reorder.
5944         * tree-pretty-print.c (dump_generic_node) <ANNOTATE_EXPR>: Handle new
5945         kinds.
5946         * tree.def (ANNOTATE_EXPR): Tweak comment.
5947
5948 2014-04-14  Jan Hubicka  <hubicka@ucw.cz>
5949
5950         * ipa-devirt.c (maybe_record_node): Ignore all non-methods (including
5951         cxa_pure_virtual).
5952
5953 2014-04-14  Paolo Carlini  <paolo.carlini@oracle.com>
5954
5955         * tree.h (TYPE_IDENTIFIER): Declare.
5956         * tree.c (subrange_type_for_debug_p): Use it.
5957         * godump.c (go_format_type): Likewise.
5958         * dwarf2out.c (is_cxx_auto, modified_type_die,
5959         gen_type_die_with_usage, gen_type_die_with_usage): Likewise.
5960         * dbxout.c (dbxout_type, dbxout_symbol): Likewise.
5961
5962 2014-04-14  Jan Hubicka   <hubicka@ucw.cz>
5963
5964         PR lto/60820
5965         * varpool.c (varpool_remove_node): Do not alter decls when streaming.
5966
5967 2014-04-14  Uros Bizjak  <ubizjak@gmail.com>
5968
5969         * config/i386/i386.c (examine_argument): Return bool.  Return true if
5970         parameter should be passed in memory.
5971         <case X86_64_COMPLEX_X87_CLASS>: Adjust.
5972         (construct_container): Update calls to examine_argument.
5973         (function_arg_advance_64): Ditto.
5974         (return_in_memory_32): Merge with ix86_return_in_memory.
5975         (return_in_memory_64): Ditto.
5976         (return_in_memory_ms_64): Ditto.
5977
5978 2014-04-14  Jan Hubicka  <hubicka@ucw.cz>
5979
5980         * ipa-utils.c (ipa_merge_profiles): Merge profile_id.
5981         * coverage.c (coverage_compute_profile_id): Handle externally visible
5982         symbols.
5983
5984 2014-04-14  Martin Jambor  <mjambor@suse.cz>
5985
5986         * tree-sra.c (ipa_sra_preliminary_function_checks): Skip
5987         DECL_DISREGARD_INLINE_LIMITS functions.
5988
5989 2014-04-14  H.J. Lu  <hongjiu.lu@intel.com>
5990
5991         PR target/60827
5992         * config/i386/i386.md (*fixuns_trunc<mode>_1): Revert the last change.
5993
5994 2014-04-14  H.J. Lu  <hongjiu.lu@intel.com>
5995
5996         PR target/60827
5997         * config/i386/i386.md (*fixuns_trunc<mode>_1): Check
5998         optimize_insn_for_speed_p instead of
5999         optimize_function_for_speed_p.
6000
6001 2014-04-14  Yufeng Zhang  <yufeng.zhang@arm.com>
6002
6003         * doc/invoke.texi (free): Document AArch64.
6004
6005 2014-04-14  Richard Biener  <rguenther@suse.de>
6006
6007         PR tree-optimization/60042
6008         * tree-ssa-pre.c (inhibit_phi_insertion): Remove.
6009         (insert_into_preds_of_block): Do not prevent PHI insertion
6010         for REFERENCE exprs here ...
6011         (eliminate_dom_walker::before_dom_children): ... but prevent
6012         their use here under similar conditions when applied to the
6013         IL after PRE optimizations.
6014
6015 2014-04-14  Richard Biener  <rguenther@suse.de>
6016
6017         * passes.def: Move early points-to after early SRA.
6018
6019 2014-04-14  Richard Biener  <rguenther@suse.de>
6020
6021         * tree-ssa-forwprop.c (simplify_gimple_switch): Enhance
6022         check for which sign-changes we allow when forwarding
6023         a converted value into a switch.
6024
6025 2014-04-14  Eric Botcazou  <ebotcazou@adacore.com>
6026
6027         * stor-layout.c (place_field): Finalize non-constant offset for the
6028         field, if any.
6029
6030 2014-04-14  Richard Biener  <rguenther@suse.de>
6031
6032         * tree-switch-conversion.c (lshift_cheap_p): Get speed_p
6033         as argument.
6034         (expand_switch_using_bit_tests_p): Likewise.
6035         (process_switch): Compute and pass on speed_p based on the
6036         switch stmt.
6037         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Use
6038         optimize_bb_for_speed_p.
6039
6040 2014-04-14  Eric Botcazou  <ebotcazou@adacore.com>
6041
6042         * cfgloop.h (struct loop): Rename force_vect into force_vectorize.
6043         * function.h (struct function): Rename has_force_vect_loops into
6044         has_force_vectorize_loops.
6045         * lto-streamer-in.c (input_cfg): Adjust for renaming.
6046         (input_struct_function_base): Likewise.
6047         * lto-streamer-out.c (output_cfg): Likewise.
6048         (output_struct_function_base): Likewise.
6049         * omp-low.c (expand_omp_simd): Likewise.
6050         * tree-cfg.c (move_sese_region_to_fn): Likewise.
6051         * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
6052         (version_loop_for_if_conversion): Likewise.
6053         (tree_if_conversion): Likewise.
6054         (main_tree_if_conversion): Likewise.
6055         (gate_tree_if_conversion): Likewise.
6056         * tree-inline.c (copy_loops): Likewise.
6057         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Likewise.
6058         * tree-ssa-loop.c (tree_loop_vectorize): Likewise.
6059         * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Likewise.
6060         * tree-vect-loop.c (vect_estimate_min_profitable_iters): Likewise.
6061         * tree-vectorizer.c (vectorize_loops): Likewise.
6062         * tree-vectorizer.h (unlimited_cost_model): Likewise.
6063
6064 2014-04-14  Richard Biener  <rguenther@suse.de>
6065
6066         PR lto/60720
6067         * lto-streamer-out.c (wrap_refs): New function.
6068         (lto_output): Wrap symbol references in global initializes in
6069         type-preserving MEM_REFs.
6070
6071 2014-04-14  Christian Bruel  <christian.bruel@st.com>
6072
6073         * config/sh/sh-mem.cc (sh_expand_strlen): Unroll last word.
6074
6075 2014-04-14  Christian Bruel  <christian.bruel@st.com>
6076
6077         * config/sh/sh.md (setmemqi): New expand pattern.
6078         * config/sh/sh.h (CLEAR_RATIO): Define.
6079         * config/sh/sh-mem.cc (sh_expand_setmem): Define.
6080         * config/sh/sh-protos.h (sh_expand_setmem): Declare.
6081
6082 2014-04-14  Richard Biener  <rguenther@suse.de>
6083
6084         PR middle-end/55022
6085         * fold-const.c (negate_expr_p): Don't negate directional rounding
6086         division.
6087         (fold_negate_expr): Likewise.
6088
6089 2014-04-14  Richard Biener  <rguenther@suse.de>
6090
6091         PR tree-optimization/59817
6092         PR tree-optimization/60453
6093         * graphite-scop-detection.c (graphite_can_represent_scev): Complete
6094         recursion to catch all CHRECs in the scalar evolution and restrict
6095         the predicate for the remains appropriately.
6096
6097 2014-04-12  Catherine Moore  <clm@codesourcery.com>
6098
6099         * config/mips/constraints.md: Add new register constraint "kb".
6100         * config/mips/mips.md (*mov<mode>_internal): Use constraint "kb".
6101         (*movhi_internal): Likewise.
6102         (*movqi_internal): Likewise.
6103         * config/mips/mips.h (M16_STORE_REGS): New register class.
6104         (REG_CLASS_NAMES): Add M16_STORE_REGS.
6105         (REG_CLASS_CONTENTS): Likewise.
6106         * config/mips/mips.c (mips_regno_to_class): Add M16_STORE_REGS.
6107
6108 2014-04-11  Tobias Burnus  <burnus@net-b.de>
6109
6110         PR c/60194
6111         * doc/invoke.texi (-Wformat-signedness): Document it.
6112         (Wformat=2): Mention that this enables -Wformat-signedness.
6113
6114 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
6115
6116         * common/config/epiphany/epiphany-common.c
6117         (epiphany_option_optimization_table): Enable section anchors by
6118         default at -O1 or higher.
6119         * config/epiphany/epiphany.c (TARGET_MAX_ANCHOR_OFFSET): Define.
6120         (TARGET_MIN_ANCHOR_OFFSET): Likewise.
6121         (epiphany_rtx_costs) <SET>: For binary operators, the set as such
6122         carries no extra cost.
6123         (epiphany_legitimate_address_p): For BLKmode, apply SImode check.
6124         * config/epiphany/epiphany.h (ASM_OUTPUT_DEF): Define.
6125         * config/epiphany/predicates.md (memclob_operand): New predicate.
6126         * config/epiphany/epiphany.md (stack_adjust_add, stack_adjust_str):
6127         Use memclob_operand predicate and X constraint for operand 3.
6128
6129 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
6130
6131         * config/epiphany/epiphany.c (epiphany_rtx_cost): Compare
6132         with CC_N_NE / CC_C_LTU / CC_C_GTU carries no extra cost for
6133         its operands.
6134
6135 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
6136
6137         PR rtl-optimization/60651
6138         * mode-switching.c (optimize_mode_switching): Make sure to emit
6139         sets of a lower numbered entity before sets of a higher numbered
6140         entity to a mode of the same or lower priority.
6141         When creating a seginfo for a basic block that starts with a code
6142         label, move the insertion point past the code label.
6143         (new_seginfo): Document and enforce requirement that
6144         NOTE_INSN_BASIC_BLOCK only appears for empty blocks.
6145         * doc/tm.texi.in: Document ordering constraint for emitted mode sets.
6146         * doc/tm.texi: Regenerate.
6147
6148 2014-01-11  Joern Rennecke  <joern.rennecke@embecosm.com>
6149
6150         PR target/60811
6151         * config/arc/arc.c (arc_save_restore): Fix assert typo.
6152
6153 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
6154
6155         * BASE-VER: Set to 4.10.0.
6156
6157 2014-04-11  Tobias Burnus  <burnus@net-b.de>
6158
6159         PR other/59055
6160         * doc/bugreport.texi (Bugs): Remove nodes pointing to the nirvana.
6161         * doc/gcc.texi (Service): Update description in the @menu
6162         * doc/invoke.texi (Option Summary): Remove misplaced and
6163         duplicated @menu.
6164
6165 2014-04-11  Steve Ellcey  <sellcey@mips.com>
6166             Jakub Jelinek  <jakub@redhat.com>
6167
6168         PR middle-end/60556
6169         * expr.c (convert_move): Use emit_store_flag_force instead of
6170         emit_store_flag.  Pass lowpart_mode instead of VOIDmode as 5th
6171         argument to it.
6172
6173 2014-04-11  Richard Biener  <rguenther@suse.de>
6174
6175         PR middle-end/60797
6176         * varasm.c (assemble_alias): Avoid endless error reporting
6177         recursion by setting TREE_ASM_WRITTEN.
6178
6179 2014-04-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6180
6181         * config/s390/s390.md: Add a splitter for NOT rtx.
6182
6183 2014-04-11  Jakub Jelinek  <jakub@redhat.com>
6184
6185         PR rtl-optimization/60663
6186         * cse.c (cse_insn): Set src_volatile on ASM_OPERANDS in PARALLEL.
6187
6188 2014-04-10  Jan Hubicka  <hubicka@ucw.cz>
6189             Jakub Jelinek  <jakub@redhat.com>
6190
6191         PR lto/60567
6192         * ipa.c (function_and_variable_visibility): Copy forced_by_abi
6193         flag from decl_node to node.
6194
6195 2014-04-10  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
6196
6197         PR debug/60655
6198         * config/arm/arm.c (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Define
6199         (arm_const_not_ok_for_debug_p): Reject MINUS with SYM_REF's
6200         ameliorating the cases where it can be.
6201
6202 2014-04-09  David Edelsohn  <dje.gcc@gmail.com>
6203
6204         Revert
6205         2014-04-08  Pat Haugen  <pthaugen@us.ibm.com>
6206
6207         * config/rs6000/sync.md (AINT mode_iterator): Move definition.
6208         (loadsync_<mode>): Change mode.
6209         (load_quadpti, store_quadpti): New.
6210         (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
6211         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
6212         * config/rs6000/predicates.md (quad_memory_operand): !TARGET_SYNC_TI.
6213
6214 2014-04-09  Cong Hou  <congh@google.com>
6215
6216         PR testsuite/60773
6217         * doc/sourcebuild.texi (vect_widen_mult_si_to_di_pattern): Add
6218         documentation.
6219
6220 2014-04-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6221
6222         * config/rs6000/rs6000.c (rs6000_expand_vector_set): Use vnand
6223         instead of vnor to exploit possible fusion opportunity in the
6224         future.
6225         (altivec_expand_vec_perm_const_le): Likewise.
6226
6227 2014-04-08  Pat Haugen  <pthaugen@us.ibm.com>
6228
6229         * config/rs6000/sync.md (AINT mode_iterator): Move definition.
6230         (loadsync_<mode>): Change mode.
6231         (load_quadpti, store_quadpti): New.
6232         (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
6233         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
6234
6235 2014-04-08  Richard Sandiford  <rdsandiford@googlemail.com>
6236
6237         PR target/60763
6238         * config/rs6000/vsx.md (vsx_xscvdpspn_scalar): Change input to DImode.
6239         * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Update accordingly.
6240         Use gen_rtx_REG rather than simplify_gen_subreg for op0_di.
6241
6242 2014-04-08  Richard Biener  <rguenther@suse.de>
6243
6244         PR middle-end/60706
6245         * tree-pretty-print.c (pp_double_int): For HWI32 hosts with
6246         a 64bit widest int print double-int similar to on HWI64 hosts.
6247
6248 2014-04-08  Richard Biener  <rguenther@suse.de>
6249
6250         PR tree-optimization/60785
6251         * graphite-sese-to-poly.c (rewrite_phi_out_of_ssa): Treat
6252         default defs properly.
6253
6254 2014-04-08  Nathan Sidwell  <nathan@codesourcery.com>
6255
6256         * doc/invoke (Wnon-virtual-dtor): Update to match implementation.
6257         (Weffc++): Likewise.
6258
6259 2014-04-07  Jan Hubicka  <hubcika@ucw.cz>
6260
6261         * ipa-devirt.c (maybe_record_node): When node is not recorded,
6262         set completep to false rather than true.
6263
6264 2014-04-07  Douglas B Rupp  <rupp@adacore.com>
6265
6266         PR target/60504
6267         * config/arm/arm.h (ASM_PREFERRED_EH_DATA_FORMAT): Expose from
6268         ARM_TARGET2_DWARF_FORMAT.
6269
6270 2014-04-07  Charles Baylis  <charles.baylis@linaro.org>
6271
6272         PR target/60609
6273         * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
6274         (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
6275         ADDR_DIFF_VEC.
6276
6277 2014-04-07  Richard Biener  <rguenther@suse.de>
6278
6279         PR tree-optimization/60766
6280         * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an unsigned type.
6281         (may_eliminate_iv): Convert cand_value_at result to desired type.
6282
6283 2014-04-07  Jason Merrill  <jason@redhat.com>
6284
6285         PR c++/60731
6286         * common.opt (-fno-gnu-unique): Add.
6287         * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
6288
6289 2014-04-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6290
6291         * haifa-sched.c: Fix outdated function reference and minor
6292         grammar errors in introductory comment.
6293
6294 2014-04-07  Richard Biener  <rguenther@suse.de>
6295
6296         PR middle-end/60750
6297         * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
6298         for noreturn calls.
6299         * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
6300
6301 2014-04-06  John David Anglin  <danglin@gcc.gnu.org>
6302
6303         PR debug/55794
6304         * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
6305         size accounting for thunks.
6306         (pa_asm_output_mi_thunk): Use final_start_function() and
6307         final_end_function() to output function start and end directives.
6308
6309 2014-04-05  Pitchumani Sivanupandi  <Pitchumani.S@atmel.com>
6310
6311         * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have
6312         device specific ISA/ feature information. Remove short_sp and
6313         errata_skip ds.  Add avr_device_specific_features enum to have device
6314         specific info.
6315         * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
6316         errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
6317         * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
6318         updated device specific info.
6319         * config/avr/avr-mcus.def: Merge device specific details to
6320         dev_attribute field.
6321         * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
6322         errata_skip.
6323         * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
6324         * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
6325         assembler if RMW isa supported by current device.
6326         * config/avr/genmultilib.awk: Update as device info structure changed.
6327         * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
6328
6329 2014-04-04  Cong Hou  <congh@google.com>
6330
6331         PR tree-optimization/60656
6332         * tree-vect-stmts.c (supportable_widening_operation):
6333         Fix a bug that elements in a vector with vect_used_by_reduction
6334         property are incorrectly reordered when the operation on it is not
6335         consistant with the one in reduction operation.
6336
6337 2014-04-04  John David Anglin  <danglin@gcc.gnu.org>
6338
6339         PR rtl-optimization/60155
6340         * gcse.c (record_set_data): New function.
6341         (single_set_gcse): New function.
6342         (gcse_emit_move_after): Use single_set_gcse instead of single_set.
6343         (hoist_code): Likewise.
6344         (get_pressure_class_and_nregs): Likewise.
6345
6346 2014-04-04  Eric Botcazou  <ebotcazou@adacore.com>
6347
6348         * explow.c (probe_stack_range): Emit a final optimization blockage.
6349
6350 2014-04-04  Anthony Green  <green@moxielogic.com>
6351
6352         * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
6353         typos.
6354
6355 2014-04-04  Jan Hubicka  <hubicka@ucw.cz>
6356
6357         PR ipa/59626
6358         * lto-cgraph.c (input_overwrite_node): Check that partitioning
6359         flags are set only during streaming.
6360         * ipa.c (process_references, walk_polymorphic_call_targets,
6361         symtab_remove_unreachable_nodes): Drop bodies of always inline
6362         after early inlining.
6363         (symtab_remove_unreachable_nodes): Remove always_inline attribute.
6364
6365 2014-04-04  Jakub Jelinek  <jakub@redhat.com>
6366         Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
6367
6368         PR debug/60655
6369         * dwarf2out.c (const_ok_for_output_1): Reject expressions
6370         containing a NOT.
6371
6372 2014-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6373
6374         PR bootstrap/60743
6375         * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
6376         duration.
6377         (cortex_a53_fdivd): Likewise.
6378
6379 2014-04-04  Martin Jambor  <mjambor@suse.cz>
6380
6381         PR ipa/60640
6382         * cgraph.h (cgraph_clone_node): New parameter added to declaration.
6383         Adjust all callers.
6384         * cgraph.c (clone_of_p): Also return true if thunks match.
6385         (verify_edge_corresponds_to_fndecl): Removed extraneous call to
6386         cgraph_function_or_thunk_node and an obsolete comment.
6387         * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
6388         file.
6389         (build_function_decl_skip_args): Likewise.
6390         (set_new_clone_decl_and_node_flags): New function.
6391         (duplicate_thunk_for_node): Likewise.
6392         (redirect_edge_duplicating_thunks): Likewise.
6393         (cgraph_clone_node): New parameter args_to_skip, pass it to
6394         redirect_edge_duplicating_thunks which is called instead of
6395         cgraph_redirect_edge_callee.
6396         (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
6397         moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
6398
6399 2014-04-04  Jeff Law  <law@redhat.com>
6400
6401         PR target/60657
6402         * config/arm/predicates.md (const_int_I_operand): New predicate.
6403         (const_int_M_operand): Similarly.
6404         * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
6405         const_int_operand.
6406         (insv_t2, extv_reg, extzv_t2): Likewise.
6407         (load_multiple_with_writeback): Similarly for const_int_I_operand.
6408         (pop_multiple_with_writeback_and_return): Likewise.
6409         (vfp_pop_multiple_with_writeback): Likewise
6410
6411 2014-04-04  Richard Biener  <rguenther@suse.de>
6412
6413         PR ipa/60746
6414         * tree-ssanames.c (make_ssa_name_fn): Fix assert.
6415         * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
6416         non-GIMPLE_LABELs.
6417         * gimplify.h (gimple_add_tmp_var_fn): Declare.
6418         * gimplify.c (gimple_add_tmp_var_fn): New function.
6419         * gimple-expr.h (create_tmp_reg_fn): Declare.
6420         * gimple-expr.c (create_tmp_reg_fn): New function.
6421         * gimple-low.c (record_vars_into): Don't change cfun.
6422         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
6423         code generation without cfun.
6424
6425 2014-04-04  Thomas Schwinge  <thomas@codesourcery.com>
6426
6427         PR bootstrap/60719
6428         * Makefile.in (install-driver): Fix shell scripting.
6429
6430 2014-04-03  Cong Hou  <congh@google.com>
6431
6432         PR tree-optimization/60505
6433         * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
6434         threshold of number of iterations below which no vectorization
6435         will be done.
6436         * tree-vect-loop.c (new_loop_vec_info):
6437         Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
6438         * tree-vect-loop.c (vect_analyze_loop_operations):
6439         Set LOOP_VINFO_COST_MODEL_THRESHOLD.
6440         * tree-vect-loop.c (vect_transform_loop):
6441         Use LOOP_VINFO_COST_MODEL_THRESHOLD.
6442         * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
6443         of iterations of the loop and see if we should build the epilogue.
6444
6445 2014-04-03  Richard Biener  <rguenther@suse.de>
6446
6447         * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx member.
6448         (streamer_tree_cache_create): Adjust.
6449         * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
6450         to allow optional nodes array.
6451         (streamer_tree_cache_insert_1): Use next_idx to assign idx.
6452         (streamer_tree_cache_append): Likewise.
6453         (streamer_tree_cache_create): Create nodes array optionally
6454         as specified by parameter.
6455         * lto-streamer-out.c (create_output_block): Avoid maintaining
6456         the node array in the writer cache.
6457         (DFS_write_tree): Remove assertion.
6458         (produce_asm_for_decls): Free the out decl state hash table early.
6459         * lto-streamer-in.c (lto_data_in_create): Adjust for
6460         streamer_tree_cache_create prototype change.
6461
6462 2014-04-03  Richard Biener  <rguenther@suse.de>
6463
6464         * tree-streamer-out.c (streamer_write_chain): Do not temporarily
6465         set TREE_CHAIN to NULL_TREE.
6466
6467 2014-04-03  Richard Biener  <rguenther@suse.de>
6468
6469         PR tree-optimization/60740
6470         * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
6471         over all GIMPLE_COND operands.
6472
6473 2014-04-03  Nathan Sidwell  <nathan@codesourcery.com>
6474
6475         * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
6476         (Weffc++): Remove Scott's numbering, merge lists and reference
6477         Wnon-virtual-dtor.
6478
6479 2014-04-03  Nick Clifton  <nickc@redhat.com>
6480
6481         * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
6482         properly.
6483
6484 2014-04-03  Martin Jambor  <mjambor@suse.cz>
6485
6486         * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
6487         mention gcc_unreachable before failing.
6488         * ipa.c (symtab_remove_unreachable_nodes): Also print order of
6489         removed symbols.
6490
6491 2014-04-02  Jan Hubicka  <hubicka@ucw.cz>
6492
6493         PR ipa/60659
6494         * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type
6495         inconsistent code and instead mark the context inconsistent.
6496         (possible_polymorphic_call_targets): For inconsistent contexts
6497         return empty complete list.
6498
6499 2014-04-02  Anthony Green  <green@moxielogic.com>
6500
6501         * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
6502         (extendqisi2, extendhisi2): Define.
6503         * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
6504         (WCHAR_TYPE): Change to unsigned int.
6505
6506 2014-04-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6507
6508         PR tree-optimization/60733
6509         * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
6510         insertion point for PHI candidates to be the end of the feeding
6511         block for the PHI argument.
6512
6513 2014-04-02  Vladimir Makarov  <vmakarov@redhat.com>
6514
6515         PR rtl-optimization/60650
6516         * lra-constraints.c (process_alt_operands): Decrease reject for
6517         earlyclobber matching.
6518
6519 2014-04-02  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6520
6521         * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
6522
6523 2014-04-02  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
6524
6525         * config/spu/spu.c (pad_bb): Do not crash when the last
6526         insn is CODE_FOR_blockage.
6527
6528 2014-04-02  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
6529
6530         * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
6531         lies outside the target mode.
6532
6533 2014-04-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
6534
6535         PR target/60735
6536         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
6537         software floating point or no floating point registers, do not
6538         allow any type in the FPRs.  Eliminate a test for SPE SIMD types
6539         in GPRs that occurs after we tested for GPRs that would never be
6540         true.
6541
6542         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
6543         Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
6544         since the FMOVE64 type is DFmode/DDmode.  If TARGET_E500_DOUBLE,
6545         specifically allow DDmode, since that does not use the SPE SIMD
6546         instructions.
6547
6548 2014-04-02  Richard Biener  <rguenther@suse.de>
6549
6550         PR middle-end/60729
6551         * optabs.c (expand_abs_nojump): Honor flag_trapv only for
6552         MODE_INTs.  Properly use negv_optab.
6553         (expand_abs): Likewise.
6554
6555 2014-04-02  Richard Biener  <rguenther@suse.de>
6556
6557         PR bootstrap/60719
6558         * Makefile.in (install-driver): Guard extra installs with special
6559         names properly.
6560
6561 2014-04-01  Michael Meissner  <meissner@linux.vnet.ibm.com>
6562
6563         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
6564         Document vec_vgbbd.
6565
6566 2014-04-01  Richard Henderson  <rth@redhat.com>
6567
6568         PR target/60704
6569         * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
6570         alternative enabled before register allocation.
6571
6572 2014-04-01  Chung-Lin Tang  <cltang@codesourcery.com>
6573
6574         * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
6575         * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
6576         typo.
6577         (nios2_large_got_address): Remove unneeded 'sym' parameter.
6578         (nios2_got_address): Update nios2_large_got_address call site.
6579         (nios2_delegitimize_address): New function.
6580         (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
6581         * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
6582         (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
6583
6584 2014-04-01  Martin Husemann  <martin@duskware.de>
6585
6586         * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
6587         for -mabi=32.
6588
6589 2014-04-01  Richard Sandiford  <rdsandiford@googlemail.com>
6590
6591         PR rtl-optimization/60604
6592         * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
6593         check from register_operand.
6594         (register_operand): Redefine in terms of general_operand.
6595         (nonmemory_operand): Use register_operand for the non-constant cases.
6596
6597 2014-04-01  Richard Biener  <rguenther@suse.de>
6598
6599         * gimple.h (struct gimple_statement_base): Align subcode to 16 bits.
6600
6601 2014-04-01  Sebastian Huber  <sebastian.huber@embedded-brains.de>
6602
6603         * doc/invoke.texi (mapp-regs): Clarify.
6604
6605 2014-03-31  Ulrich Drepper  <drepper@gmail.com>
6606
6607         * config/i386/avx512fintrin.h (__v32hi): Define type.
6608         (__v64qi): Likewise.
6609         (_mm512_set1_epi8): Define.
6610         (_mm512_set1_epi16): Define.
6611         (_mm512_set4_epi32): Define.
6612         (_mm512_set4_epi64): Define.
6613         (_mm512_set4_pd): Define.
6614         (_mm512_set4_ps): Define.
6615         (_mm512_setr4_epi64): Define.
6616         (_mm512_setr4_epi32): Define.
6617         (_mm512_setr4_pd): Define.
6618         (_mm512_setr4_ps): Define.
6619         (_mm512_setzero_epi32): Define.
6620
6621 2014-03-31  Martin Jambor  <mjambor@suse.cz>
6622
6623         PR middle-end/60647
6624         * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
6625         callsite_arguments_match_p.  Updated all callers.  Also check types of
6626         corresponding formal parameters and actual arguments.
6627         (not_all_callers_have_enough_arguments_p) Renamed to
6628         some_callers_have_mismatched_arguments_p.
6629
6630 2014-03-31  Yuri Rumyantsev  <ysrumyan@gmail.com>
6631
6632         * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
6633
6634 2014-03-31  Kugan Vivekanandarajah  <kuganv@linaro.org>
6635
6636         PR target/60034
6637         * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
6638         section anchor.
6639
6640 2014-03-30  Uros Bizjak  <ubizjak@gmail.com>
6641
6642         * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
6643         (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
6644         Split out
6645         <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
6646         Use FMAMODE_NOVF512 mode iterator.
6647         (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
6648         (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
6649         (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
6650         (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
6651         Split out
6652         <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
6653         Use VF_128_256 mode iterator.
6654         (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
6655         Ditto.
6656
6657 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
6658
6659         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
6660         static chain if needed.
6661
6662 2014-03-28  Vladimir Makarov  <vmakarov@redhat.com>
6663
6664         PR target/60697
6665         * lra-constraints.c (index_part_to_reg): New.
6666         (process_address): Use it.
6667
6668 2014-03-27  Jeff Law  <law@redhat.com>
6669             Jakub Jelinek  <jakub@redhat.com>
6670
6671         PR target/60648
6672         * expr.c (do_tablejump): Use simplify_gen_binary rather than
6673         gen_rtx_{PLUS,MULT} to build up the address expression.
6674
6675         * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
6676         creating non-canonical RTL.
6677
6678 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
6679
6680         PR ipa/60243
6681         * ipa-inline.c (want_inline_small_function_p): Short circuit large
6682         functions; reorganize to make cheap checks first.
6683         (inline_small_functions): Do not estimate growth when dumping;
6684         it is expensive.
6685         * ipa-inline.h (inline_summary): Add min_size.
6686         (growth_likely_positive): New function.
6687         * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
6688         (set_cond_stmt_execution_predicate): Cleanup.
6689         (estimate_edge_size_and_time): Compute min_size.
6690         (estimate_calls_size_and_time): Likewise.
6691         (estimate_node_size_and_time): Likewise.
6692         (inline_update_overall_summary): Update min_size.
6693         (do_estimate_edge_time): Likewise.
6694         (do_estimate_edge_size): Update.
6695         (do_estimate_edge_hints): Update.
6696         (growth_likely_positive): New function.
6697
6698 2014-03-28  Jakub Jelinek  <jakub@redhat.com>
6699
6700         PR target/60693
6701         * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
6702         also if addr has VOIDmode.
6703
6704 2014-03-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6705
6706         * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
6707         * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
6708         Declare extern.
6709         * config/arm/cortex-a53.md: Add reservations and bypass for crypto
6710         instructions as well as AdvancedSIMD loads.
6711
6712 2014-03-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6713
6714         * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
6715         Use crypto_aese type.
6716         (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
6717         * config/arm/arm.md (is_neon_type): Replace crypto_aes with
6718         crypto_aese, crypto_aesmc.  Move to types.md.
6719         * config/arm/types.md (crypto_aes): Split into crypto_aese,
6720         crypto_aesmc.
6721         * config/arm/iterators.md (crypto_type): Likewise.
6722
6723 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
6724
6725         * cgraph.c: Include expr.h and tree-dfa.h.
6726         (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
6727         remove LHS.
6728
6729 2014-03-28  Vladimir Makarov  <vmakarov@redhat.com>
6730
6731         PR target/60675
6732         * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
6733         regs from checking multi-reg pseudos.
6734
6735 2014-03-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
6736
6737         * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
6738
6739 2014-03-28  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
6740
6741         * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
6742         if it would clobber the stack pointer, even temporarily.
6743
6744 2014-03-28  Eric Botcazou  <ebotcazou@adacore.com>
6745
6746         * mode-switching.c: Make small adjustments to the top comment.
6747
6748 2014-03-27  Michael Meissner  <meissner@linux.vnet.ibm.com>
6749
6750         * config/rs6000/constraints.md (wD constraint): New constraint to
6751         match the constant integer to get the top DImode/DFmode out of a
6752         vector in a VSX register.
6753
6754         * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
6755         match the constant integer to get the top DImode/DFmode out of a
6756         vector in a VSX register.
6757
6758         * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
6759         for ISA 2.07.
6760
6761         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6762         vbpermq builtins.
6763
6764         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
6765         -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
6766
6767         * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
6768         Optimize vec_extract of 64-bit values, where the value being
6769         extracted is in the top word, where we can use scalar
6770         instructions.  Add direct move and store support.  Combine the big
6771         endian/little endian vector select load support into a single insn.
6772         (vsx_extract_<mode>_internal1): Likewise.
6773         (vsx_extract_<mode>_internal2): Likewise.
6774         (vsx_extract_<mode>_load): Likewise.
6775         (vsx_extract_<mode>_store): Likewise.
6776         (vsx_extract_<mode>_zero): Delete, big and little endian insns are
6777         combined into vsx_extract_<mode>_load.
6778         (vsx_extract_<mode>_one_le): Likewise.
6779
6780         * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
6781         define the top 64-bit vector element.
6782
6783         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
6784         constraint.
6785
6786         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
6787         Document vec_vbpermq builtin.
6788
6789         PR target/60672
6790         * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
6791         enable use of xxsldwi and xxpermdi builtin functions.
6792         (vec_xxpermdi): Likewise.
6793
6794         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
6795         Document use of vec_xxsldwi and vec_xxpermdi builtins.
6796
6797 2014-03-27  Vladimir Makarov  <vmakarov@redhat.com>
6798
6799         PR rtl-optimization/60650
6800         * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
6801         first_p.  Use it.
6802         (find_spills_for): New.
6803         (assign_by_spills): Pass the new parameter to find_hard_regno_for.
6804         Spill all pseudos on the second iteration.
6805
6806 2014-03-27  Marek Polacek  <polacek@redhat.com>
6807
6808         PR c/50347
6809         * doc/extend.texi (ffs Builtins): Change unsigned types to signed
6810         types.
6811
6812 2014-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6813
6814         * config/s390/s390.c (s390_can_use_return_insn): Check for
6815         call-saved FPRs on 31 bit.
6816
6817 2014-03-27  Jakub Jelinek  <jakub@redhat.com>
6818
6819         PR middle-end/60682
6820         * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
6821         if they need regimplification, just drop them instead of
6822         calling gimple_regimplify_operands on them.
6823
6824 2014-03-27  Marcus Shawcroft  <marcus.shawcroft@arm.com>
6825
6826         PR target/60580
6827         * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
6828         (aarch64_frame_pointer_required): Adjust logic.
6829         (aarch64_can_eliminate): Adjust logic.
6830         (aarch64_override_options_after_change): Adjust logic.
6831
6832 2014-03-27  Dehao Chen  <dehao@google.com>
6833
6834         * ipa-inline.c (early_inliner): Update node's inline info.
6835
6836 2014-03-26  Dehao Chen  <dehao@google.com>
6837
6838         * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
6839         compiler inserted conditional jumps for NAN float check.
6840
6841 2014-03-26  Jakub Jelinek  <jakub@redhat.com>
6842
6843         * ubsan.h (ubsan_create_data): Change second argument's type
6844         to const location_t *.
6845         * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
6846         _("<unknown>").
6847         (ubsan_create_data): Change second argument to const location_t *PLOC.
6848         Create Loc field whenever PLOC is non-NULL.
6849         (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
6850         ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
6851         callers.
6852
6853         PR other/59545
6854         * real.c (real_to_integer2): Change type of low to UHWI.
6855
6856 2014-03-26  Tobias Burnus  <burnus@net-b.de>
6857
6858         * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
6859         (CILK_SELF_SPECS): New define.
6860         (driver_self_specs): Use it.
6861
6862 2014-03-26  Richard Biener  <rguenther@suse.de>
6863
6864         * tree-pretty-print.c (percent_K_format): Implement special
6865         case for LTO and its stripped down BLOCK tree.
6866
6867 2014-03-26  Jakub Jelinek  <jakub@redhat.com>
6868
6869         PR sanitizer/60636
6870         * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
6871
6872         * tree-vrp.c (simplify_internal_call_using_ranges): If only
6873         one range is range_int_cst_p, but not both, at least optimize
6874         addition/subtraction of 0 and multiplication by 0 or 1.
6875         * gimple-fold.c (gimple_fold_call): Fold
6876         IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
6877         (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
6878         INTEGER_CSTs, try to fold at least x * 0 and y - y.
6879
6880 2014-03-26  Eric Botcazou  <ebotcazou@adacore.com>
6881
6882         PR rtl-optimization/60452
6883         * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
6884         <case REG>: Return 1 for invalid offsets from the frame pointer.
6885
6886 2014-03-26  Marek Polacek  <polacek@redhat.com>
6887
6888         PR c/37428
6889         * doc/extend.texi (C Extensions): Mention variable-length arrays in
6890         a structure/union.
6891
6892 2014-03-26  Marek Polacek  <polacek@redhat.com>
6893
6894         PR c/39525
6895         * doc/extend.texi (Designated Inits): Describe what happens to omitted
6896         field members.
6897
6898 2014-03-26  Marek Polacek  <polacek@redhat.com>
6899
6900         PR other/59545
6901         * ira-color.c (update_conflict_hard_regno_costs): Perform the
6902         multiplication in unsigned type.
6903
6904 2014-03-26  Chung-Ju Wu  <jasonwucj@gmail.com>
6905
6906         * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
6907
6908 2014-03-26  Chung-Ju Wu  <jasonwucj@gmail.com>
6909
6910         * doc/contrib.texi: Add myself as Andes nds32 port contributor.
6911
6912 2014-03-25  Jan Hubicka  <hubicka@ucw.cz>
6913
6914         PR ipa/60315
6915         * cif-code.def (UNREACHABLE) New code.
6916         * ipa-inline.c (inline_small_functions): Skip edges to
6917         __builtlin_unreachable.
6918         (estimate_edge_growth): Allow edges to __builtlin_unreachable.
6919         * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
6920         predicate to __bulitin_unreachable.
6921         (set_cond_stmt_execution_predicate): Fix issue when
6922         invert_tree_comparison returns ERROR_MARK.
6923         * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
6924         propagate to inline clones.
6925         * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
6926         to unreachable.
6927         * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
6928         * cgraphclones.c (cgraph_clone_node): If call destination is already
6929         ureachable, do not redirect it back.
6930         * tree-inline.c (fold_marked_statements): Hanlde calls becoming
6931         unreachable.
6932
6933 2014-03-25  Jan Hubicka  <hubicka@ucw.cz>
6934
6935         * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
6936         Do not modify inline clones.
6937
6938 2014-03-25  Jakub Jelinek  <jakub@redhat.com>
6939
6940         * config/i386/i386.md (general_sext_operand): New mode attr.
6941         (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
6942         don't generate (sign_extend (const_int)).
6943         (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
6944         operands[2].  Use We constraint instead of <i> and
6945         <general_sext_operand> predicate instead of <general_operand>.
6946         (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
6947         * config/i386/constraints.md (We): New constraint.
6948         * config/i386/predicates.md (x86_64_sext_operand,
6949         sext_operand): New predicates.
6950
6951 2014-03-25  Martin Jambor  <mjambor@suse.cz>
6952
6953         PR ipa/60600
6954         * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
6955         inconsistent devirtualizations to __builtin_unreachable.
6956
6957 2014-03-25  Marek Polacek  <polacek@redhat.com>
6958
6959         PR c/35449
6960         * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
6961
6962 2014-03-25  Alan Lawrence  <alan.lawrence@arm.com>
6963
6964         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
6965         order of elements for big-endian.
6966
6967 2014-03-25  Richard Biener  <rguenther@suse.de>
6968
6969         PR middle-end/60635
6970         * gimplify-me.c (gimple_regimplify_operands): Update the
6971         re-gimplifed stmt.
6972
6973 2014-03-25  Martin Jambor  <mjambor@suse.cz>
6974
6975         PR ipa/59176
6976         * lto-cgraph.c (lto_output_node): Stream body_removed flag.
6977         (lto_output_varpool_node): Likewise.
6978         (input_overwrite_node): Likewise.
6979         (input_varpool_node): Likewise.
6980
6981 2014-03-25  Richard Biener  <rguenther@suse.de>
6982
6983         * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
6984         (run_gcc): Likewise.
6985
6986 2014-03-25  Jakub Jelinek  <jakub@redhat.com>
6987
6988         * combine.c (simplify_compare_const): Add MODE argument.
6989         Handle mode_width 0 as very large mode_width.
6990         (try_combine, simplify_comparison): Adjust callers.
6991
6992         * cselib.c (cselib_hash_rtx): Perform addition in unsigned
6993         type to avoid signed integer overflow.
6994         * explow.c (plus_constant): Likewise.
6995
6996 2014-03-25  Dominik Vogt  <vogt@linux.vnet.ibm.com>
6997
6998         * doc/generic.texi: Correct typos.
6999
7000 2014-03-24  Tobias Burnus  <burnus@net-b.de>
7001
7002         * doc/invoke.texi (-flto): Expand section about
7003         using static libraries with LTO.
7004
7005 2014-03-24  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7006
7007         PR rtl-optimization/60501
7008         * optabs.def (addptr3_optab): New optab.
7009         * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
7010         * doc/md.texi ("addptrm3"): Document new RTL standard expander.
7011         * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
7012
7013         * lra.c (emit_add3_insn): Use the addptr pattern if available.
7014
7015         * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
7016
7017 2014-03-24  Ulrich Drepper  <drepper@gmail.com>
7018
7019         * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
7020         _mm512_set1_pd.
7021
7022         * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
7023         (_mm256_undefined_ps): Define.
7024         (_mm256_undefined_pd): Define.
7025         * config/i386/emmintrin.h (_mm_undefined_si128): Define.
7026         (_mm_undefined_pd): Define.
7027         * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
7028         * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
7029         (_mm512_undefined_ps): Define.
7030         (_mm512_undefined_pd): Define.
7031         Use _mm*_undefined_*.
7032         * config/i386/avx2intrin.h: Use _mm*_undefined_*.
7033
7034 2014-03-24  Alex Velenko  <Alex.Velenko@arm.com>
7035
7036         * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
7037         (lshr_simd): DI mode added.
7038         * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
7039         (aarch64_ushr_simddi): Likewise.
7040         * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
7041         * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
7042         (vshrd_n_u64): Likewise.
7043
7044 2014-03-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
7045
7046         * Makefile.in (s-macro_list): Depend on cc1.
7047
7048 2014-03-23  Teresa Johnson  <tejohnson@google.com>
7049
7050         * ipa-utils.c (ipa_print_order): Use specified dump file.
7051
7052 2014-03-23  Eric Botcazou  <ebotcazou@adacore.com>
7053
7054         PR rtl-optimization/60601
7055         * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
7056
7057         * gcc.c (eval_spec_function): Initialize save_growing_value.
7058
7059 2014-03-22  Jakub Jelinek  <jakub@redhat.com>
7060
7061         PR sanitizer/60613
7062         * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
7063         code == MINUS_EXPR, never swap op0 with op1.
7064
7065         * toplev.c (init_local_tick): Avoid signed integer multiplication
7066         overflow.
7067         * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
7068         shift by first operand's bitsize.
7069
7070 2014-03-21  Jakub Jelinek  <jakub@redhat.com>
7071
7072         PR target/60610
7073         * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
7074         redefine to 1 or 0.
7075         * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
7076         TARGET_ISA_64BIT_P(x).
7077
7078 2014-03-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7079
7080         * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
7081         pattern for vector nor instead of subtract from splat(-1).
7082         (altivec_expand_vec_perm_const_le): Likewise.
7083
7084 2014-03-21  Richard Henderson  <rth@twiddle.net>
7085
7086         PR target/60598
7087         * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
7088         related insns after epilogue_completed.
7089
7090 2014-03-21  Martin Jambor  <mjambor@suse.cz>
7091
7092         PR ipa/59176
7093         * cgraph.h (symtab_node): New flag body_removed.
7094         * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
7095         when removing bodies.
7096         * symtab.c (dump_symtab_base): Dump body_removed flag.
7097         * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
7098         had their bodies removed.
7099
7100 2014-03-21  Martin Jambor  <mjambor@suse.cz>
7101
7102         PR ipa/60419
7103         * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
7104         in the border.
7105
7106 2014-03-21  Richard Biener  <rguenther@suse.de>
7107
7108         PR tree-optimization/60577
7109         * tree-core.h (struct tree_base): Document nothrow_flag use
7110         in VAR_DECL_NONALIASED.
7111         * tree.h (VAR_DECL_NONALIASED): New.
7112         (may_be_aliased): Adjust.
7113         * coverage.c (build_var): Set VAR_DECL_NONALIASED.
7114
7115 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
7116
7117         * expr.c (expand_expr_real_1): Remove outdated comment.
7118
7119 2014-03-20  Jakub Jelinek  <jakub@redhat.com>
7120
7121         PR middle-end/60597
7122         * ira.c (adjust_cleared_regs): Call copy_rtx on
7123         *reg_equiv[REGNO (loc)].src_p before passing it to
7124         simplify_replace_fn_rtx.
7125
7126         PR target/60568
7127         * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
7128         into CONST, put pic register as first operand of PLUS.  Use
7129         gen_const_mem for both 32-bit and 64-bit PIC got loads.
7130
7131 2014-03-20  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
7132
7133         * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
7134
7135 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
7136
7137         * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
7138         around for store forwarding issue in the FPU on the UT699.
7139         * config/sparc/sparc.md (in_branch_delay): Return false for single FP
7140         loads and operations if -mfix-ut699 is specified.
7141         (divtf3_hq): Tweak attribute.
7142         (sqrttf2_hq): Likewise.
7143
7144 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
7145
7146         * calls.c (store_one_arg): Remove incorrect const qualification on the
7147         type of the temporary.
7148         * cfgexpand.c (expand_return): Likewise.
7149         * expr.c (expand_constructor): Likewise.
7150         (expand_expr_real_1): Likewise.
7151
7152 2014-03-20  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
7153
7154         * config/arm/arm.c (arm_dwarf_register_span): Update the element number
7155         of parts.
7156
7157 2014-03-19  Kaz Kojima  <kkojima@gcc.gnu.org>
7158
7159         PR target/60039
7160         * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
7161
7162 2014-03-19  James Greenhalgh  <james.greenhalgh@arm.com>
7163
7164         * config/arm/aarch-common-protos.h
7165         (alu_cost_table): Fix spelling of "extend".
7166         * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
7167
7168 2014-03-19  Richard Biener  <rguenther@suse.de>
7169
7170         PR middle-end/60553
7171         * tree-core.h (tree_type_common): Re-order pointer members
7172         to reduce recursion depth during GC walks.
7173
7174 2014-03-19  Marek Polacek  <polacek@redhat.com>
7175
7176         PR sanitizer/60569
7177         * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
7178         before accessing it.
7179
7180 2014-03-19  Richard Biener  <rguenther@suse.de>
7181
7182         PR lto/59543
7183         * lto-streamer-in.c (input_function): In WPA stage do not drop
7184         debug stmts.
7185
7186 2014-03-19  Jakub Jelinek  <jakub@redhat.com>
7187
7188         PR tree-optimization/60559
7189         * vectorizable_mask_load_store): Replace scalar MASK_LOAD
7190         with build_zero_cst assignment.
7191
7192 2014-03-18  Kai Tietz  <ktietz@redhat.com>
7193
7194         PR rtl-optimization/56356
7195         * sdbout.c (sdbout_parms): Verify that parms'
7196         incoming argument is valid.
7197         (sdbout_reg_parms): Likewise.
7198
7199 2014-03-18  Richard Henderson  <rth@redhat.com>
7200
7201         PR target/60562
7202         * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
7203         be shadowed by *float<SWI48><MODEF>2_sse.  Test X87_ENABLE_FLOAT.
7204         (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
7205
7206 2014-03-18  Basile Starynkevitch  <basile@starynkevitch.net>
7207
7208         * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
7209         * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
7210         Italicize plugin event names in description.  Explain that
7211         PLUGIN_PRAGMAS has no sense for lto1.  Explain PLUGIN_INCLUDE_FILE.
7212         Remind that no GCC functions should be called after PLUGIN_FINISH.
7213         Explain what pragmas with expansion are.
7214
7215 2014-03-18  Martin Liska  <mliska@suse.cz>
7216
7217         * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
7218         gimple call statement is update.
7219         * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
7220         GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
7221
7222 2014-03-18  Jakub Jelinek  <jakub@redhat.com>
7223
7224         PR sanitizer/60557
7225         * ubsan.c (ubsan_instrument_unreachable): Call
7226         initialize_sanitizer_builtins.
7227         (ubsan_pass): Likewise.
7228
7229         PR sanitizer/60535
7230         * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
7231         varpool_finalize_decl instead of rest_of_decl_compilation.
7232
7233 2014-03-18  Richard Biener  <rguenther@suse.de>
7234
7235         * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
7236         by using bitmap_and_compl instead of bitmap_and_compl_into.
7237         (df_rd_transfer_function): Likewise.
7238
7239 2014-03-18  Richard Biener  <rguenther@suse.de>
7240
7241         * doc/lto.texi (fresolution): Fix typo.
7242
7243 2014-03-18  Richard Biener  <rguenther@suse.de>
7244
7245         * doc/invoke.texi (flto): Update for changes in 4.9.
7246
7247 2014-03-18  Richard Biener  <rguenther@suse.de>
7248
7249         * doc/loop.texi: Remove section on the removed lambda framework.
7250         Update loop docs with recent changes in preserving loop structure.
7251
7252 2014-03-18  Richard Biener  <rguenther@suse.de>
7253
7254         * doc/lto.texi (-fresolution): Document.
7255
7256 2014-03-18  Richard Biener  <rguenther@suse.de>
7257
7258         * doc/contrib.texi: Adjust my name.
7259
7260 2014-03-18  Jakub Jelinek  <jakub@redhat.com>
7261
7262         PR ipa/58721
7263         * internal-fn.c: Include diagnostic-core.h.
7264         (expand_BUILTIN_EXPECT): New function.
7265         * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
7266         (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
7267         IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
7268         * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
7269         IFN_BUILTIN_EXPECT.
7270         * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
7271         Revert 3 argument __builtin_expect code.
7272         (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
7273         * gimple-fold.c (gimple_fold_call): Likewise.
7274         * tree.h (fold_builtin_expect): New prototype.
7275         * builtins.c (build_builtin_expect_predicate): Add predictor
7276         argument, if non-NULL, create 3 argument __builtin_expect.
7277         (fold_builtin_expect): No longer static.  Add ARG2 argument,
7278         pass it through to build_builtin_expect_predicate.
7279         (fold_builtin_2): Adjust caller.
7280         (fold_builtin_3): Handle BUILT_IN_EXPECT.
7281         * internal-fn.def (BUILTIN_EXPECT): New.
7282
7283 2014-03-18  Tobias Burnus  <burnus@net-b.de>
7284
7285         PR ipa/58721
7286         * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
7287         PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
7288         PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
7289
7290 2014-03-18  Jan Hubicka  <hubicka@ucw.cz>
7291
7292         PR ipa/58721
7293         * predict.c (combine_predictions_for_bb): Fix up formatting.
7294         (expr_expected_value_1, expr_expected_value): Add predictor argument,
7295         fill what it points to if non-NULL.
7296         (tree_predict_by_opcode): Adjust caller, use the predictor.
7297         * predict.def (PRED_COMPARE_AND_SWAP): Add.
7298
7299 2014-03-18  Eric Botcazou  <ebotcazou@adacore.com>
7300
7301         * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
7302         proper constant for the store mode.
7303
7304 2014-03-18  Ilya Enkovich  <ilya.enkovich@intel.com>
7305
7306         * symtab.c (change_decl_assembler_name): Fix transparent alias
7307         chain construction.
7308
7309 2014-03-16  Renlin Li  <Renlin.Li@arm.com>
7310
7311         * config/aarch64/aarch64.c: Correct the comments about the
7312         aarch64 stack layout.
7313
7314 2014-03-18  Thomas Schwinge  <thomas@codesourcery.com>
7315
7316         * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
7317         check for GF_OMP_FOR_KIND_FOR.
7318
7319 2013-03-18  Kirill Yukhin  <kirill.yukhin@intel.com>
7320
7321         * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
7322         ymm and zmm register names.
7323
7324 2014-03-17  Jakub Jelinek  <jakub@redhat.com>
7325
7326         PR target/60516
7327         * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
7328         note creation for the 2010-08-31 changes.
7329
7330 2014-03-17  Marek Polacek  <polacek@redhat.com>
7331
7332         PR middle-end/60534
7333         * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
7334         as -fno-tree-loop-vectorize.
7335         (expand_omp_simd): Likewise.
7336
7337 2014-03-15  Eric Botcazou  <ebotcazou@adacore.com>
7338
7339         * config/sparc/sparc-protos.h (tls_call_delay): Delete.
7340         (eligible_for_call_delay): New prototype.
7341         * config/sparc/sparc.c (tls_call_delay): Rename into...
7342         (eligible_for_call_delay): ...this.  Return false if the instruction
7343         cannot be put in the delay slot of a branch.
7344         (eligible_for_restore_insn): Simplify.
7345         (eligible_for_return_delay): Return false if the instruction cannot be
7346         put in the delay slot of a branch and simplify.
7347         (eligible_for_sibcall_delay): Return false if the instruction cannot be
7348         put in the delay slot of a branch.
7349         * config/sparc/sparc.md (fix_ut699): New attribute.
7350         (tls_call_delay): Delete.
7351         (in_call_delay): Reimplement.
7352         (eligible_for_sibcall_delay): Rename into...
7353         (in_sibcall_delay): ...this.
7354         (eligible_for_return_delay): Rename into...
7355         (in_return_delay): ...this.
7356         (in_branch_delay): Reimplement.
7357         (in_uncond_branch_delay): Delete.
7358         (in_annul_branch_delay): Delete.
7359
7360 2014-03-14  Richard Henderson  <rth@redhat.com>
7361
7362         PR target/60525
7363         * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
7364         define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
7365         (*floathi<X87MODEF>2_i387_with_temp): Remove.
7366         (floathi splitters): Remove.
7367         (float<SWI48x>xf2): New pattern.
7368         (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2.  Drop
7369         code that tried to handle DImode for 32-bit, but which was excluded
7370         by the pattern's condition.  Drop allocation of stack temporary.
7371         (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
7372         (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
7373         (*float<SWI48><MODEF>2_mixed_interunit): Remove.
7374         (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
7375         (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
7376         (*float<SWI48><MODEF>2_sse_with_temp): Remove.
7377         (*float<SWI48><MODEF>2_sse_interunit): Remove.
7378         (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
7379         (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
7380         (*float<SWI48x><X87MODEF>2_i387): Remove.
7381         (all float _with_temp splitters): Remove.
7382         (*float<SWI48x><MODEF>2_i387): New pattern.
7383         (*float<SWI48><MODEF>2_sse): New pattern.
7384         (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
7385         (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
7386
7387 2014-03-14  Jakub Jelinek  <jakub@redhat.com>
7388             Marek Polacek  <polacek@redhat.com>
7389
7390         PR middle-end/60484
7391         * common.opt (dump_base_name_prefixed): New Variable.
7392         * opts.c (finish_options): Don't prepend directory to x_dump_base_name
7393         if x_dump_base_name_prefixed is already set, set it at the end.
7394
7395 2014-03-14  Vladimir Makarov  <vmakarov@redhat.com>
7396
7397         PR rtl-optimization/60508
7398         * lra-constraints.c (get_reload_reg): Add new parameter
7399         in_subreg_p.
7400         (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
7401         Pass the new parameter values.
7402
7403 2014-03-14  Richard Biener  <rguenther@suse.de>
7404
7405         * common.opt: Revert unintented changes from r205065.
7406         * opts.c: Likewise.
7407
7408 2014-03-14  Richard Biener  <rguenther@suse.de>
7409
7410         PR middle-end/60518
7411         * cfghooks.c (split_block): Properly adjust all loops the
7412         block was a latch of.
7413
7414 2014-03-14  Martin Jambor  <mjambor@suse.cz>
7415
7416         PR lto/60461
7417         * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
7418         and simplify it.
7419
7420 2014-03-14  Georg-Johann Lay  <avr@gjlay.de>
7421
7422         PR target/59396
7423         * config/avr/avr.c (avr_set_current_function): Pass function name
7424         through default_strip_name_encoding before sanity checking instead
7425         of skipping the first char of the assembler name.
7426
7427 2014-03-13  Richard Henderson  <rth@redhat.com>
7428
7429         PR debug/60438
7430         * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
7431         (ix86_force_to_memory, ix86_free_from_memory): Remove.
7432         * config/i386/i386-protos.h: Likewise.
7433         * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
7434         in the expander instead of a splitter.
7435         (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
7436         any possibility of requiring a memory.
7437         (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
7438         (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
7439         (fp branch splitters): Update for ix86_split_fp_branch.
7440         (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
7441         (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
7442         (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
7443         (*fop_<MODEF>_2_i387): Remove f/r alternative.
7444         (*fop_<MODEF>_3_i387): Likewise.
7445         (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
7446         (splitters for the fop_* register patterns): Remove.
7447         (fscalexf4_i387): Rename from *fscalexf4_i387.
7448         (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
7449
7450 2014-03-13  Jakub Jelinek  <jakub@redhat.com>
7451
7452         PR tree-optimization/59779
7453         * tree-dfa.c (get_ref_base_and_extent): Use double_int
7454         type for bitsize and maxsize instead of HOST_WIDE_INT.
7455
7456 2014-03-13  Steven Bosscher  <steven@gcc.gnu.org>
7457
7458         PR rtl-optimization/57320
7459         * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
7460         the CFG after thread_prologue_and_epilogue_insns.
7461
7462 2014-03-13  Vladimir Makarov  <vmakarov@redhat.com>
7463
7464         PR rtl-optimization/57189
7465         * lra-constraints.c (process_alt_operands): Disfavor spilling
7466         vector pseudos.
7467
7468 2014-03-13  Cesar Philippidis  <cesar@codesourcery.com>
7469
7470         * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
7471
7472 2014-03-13  Jakub Jelinek  <jakub@redhat.com>
7473
7474         PR tree-optimization/59025
7475         PR middle-end/60418
7476         * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
7477         same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
7478
7479 2014-03-13  Georg-Johann Lay  <avr@gjlay.de>
7480
7481         PR target/60486
7482         * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
7483         calls of avr_out_plus_1.
7484
7485 2014-03-13  Bin Cheng  <bin.cheng@arm.com>
7486
7487         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
7488         BB's single pred and update the father loop's latch info later.
7489
7490 2014-03-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
7491
7492         * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
7493         (VEC_M): Likewise.
7494         (VEC_N): Likewise.
7495         (VEC_R): Likewise.
7496         (VEC_base): Likewise.
7497         (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
7498         registers, we need to swap double words in little endian mode.
7499
7500         * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
7501         to be a container mode for 128-bit integer operations added in ISA
7502         2.07.  Unlike TImode and PTImode, the preferred register set is
7503         the Altivec/VMX registers for the 128-bit operations.
7504
7505         * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
7506         declarations.
7507         (rs6000_split_128bit_ok_p): Likewise.
7508
7509         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
7510         macros for creating ISA 2.07 normal and overloaded builtin
7511         functions with 3 arguments.
7512         (BU_P8V_OVERLOAD_3): Likewise.
7513         (VPERM_1T): Add support for V1TImode in 128-bit vector operations
7514         for use as overloaded functions.
7515         (VPERM_1TI_UNS): Likewise.
7516         (VSEL_1TI): Likewise.
7517         (VSEL_1TI_UNS): Likewise.
7518         (ST_INTERNAL_1ti): Likewise.
7519         (LD_INTERNAL_1ti): Likewise.
7520         (XXSEL_1TI): Likewise.
7521         (XXSEL_1TI_UNS): Likewise.
7522         (VPERM_1TI): Likewise.
7523         (VPERM_1TI_UNS): Likewise.
7524         (XXPERMDI_1TI): Likewise.
7525         (SET_1TI): Likewise.
7526         (LXVD2X_V1TI): Likewise.
7527         (STXVD2X_V1TI): Likewise.
7528         (VEC_INIT_V1TI): Likewise.
7529         (VEC_SET_V1TI): Likewise.
7530         (VEC_EXT_V1TI): Likewise.
7531         (EQV_V1TI): Likewise.
7532         (NAND_V1TI): Likewise.
7533         (ORC_V1TI): Likewise.
7534         (VADDCUQ): Add support for 128-bit integer arithmetic instructions
7535         added in ISA 2.07.  Add both normal 'altivec' builtins, and the
7536         overloaded builtin.
7537         (VADDUQM): Likewise.
7538         (VSUBCUQ): Likewise.
7539         (VADDEUQM): Likewise.
7540         (VADDECUQ): Likewise.
7541         (VSUBEUQM): Likewise.
7542         (VSUBECUQ): Likewise.
7543
7544         * config/rs6000/rs6000-c.c (__int128_type): New static to hold
7545         __int128_t and __uint128_t types.
7546         (__uint128_type): Likewise.
7547         (altivec_categorize_keyword): Add support for vector __int128_t,
7548         vector __uint128_t, vector __int128, and vector unsigned __int128
7549         as a container type for TImode operations that need to be done in
7550         VSX/Altivec registers.
7551         (rs6000_macro_to_expand): Likewise.
7552         (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
7553         to support 128-bit integer instructions vaddcuq, vadduqm,
7554         vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
7555         (altivec_resolve_overloaded_builtin): Add support for V1TImode.
7556
7557         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
7558         for V1TImode, and set up preferences to use VSX/Altivec registers.
7559         Setup VSX reload handlers.
7560         (rs6000_debug_reg_global): Likewise.
7561         (rs6000_init_hard_regno_mode_ok): Likewise.
7562         (rs6000_preferred_simd_mode): Likewise.
7563         (vspltis_constant): Do not allow V1TImode as easy altivec constants.
7564         (easy_altivec_constant): Likewise.
7565         (output_vec_const_move): Likewise.
7566         (rs6000_expand_vector_set): Convert V1TImode set and extract to
7567         simple move.
7568         (rs6000_expand_vector_extract): Likewise.
7569         (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
7570         addressing.
7571         (rs6000_const_vec): Add support for V1TImode.
7572         (rs6000_emit_le_vsx_load): Swap double words when loading or
7573         storing TImode/V1TImode.
7574         (rs6000_emit_le_vsx_store): Likewise.
7575         (rs6000_emit_le_vsx_move): Likewise.
7576         (rs6000_emit_move): Add support for V1TImode.
7577         (altivec_expand_ld_builtin): Likewise.
7578         (altivec_expand_st_builtin): Likewise.
7579         (altivec_expand_vec_init_builtin): Likewise.
7580         (altivec_expand_builtin): Likewise.
7581         (rs6000_init_builtins): Add support for V1TImode type.  Add
7582         support for ISA 2.07 128-bit integer builtins.  Define type names
7583         for the VSX/Altivec vector types.
7584         (altivec_init_builtins): Add support for overloaded vector
7585         functions with V1TImode type.
7586         (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
7587         (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
7588         external function.
7589         (rs6000_split_128bit_ok_p): Likewise.
7590         (rs6000_handle_altivec_attribute): Create V1TImode from vector
7591         __int128_t and vector __uint128_t.
7592
7593         * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
7594         and mode attributes.
7595         (VSX_M): Likewise.
7596         (VSX_M2): Likewise.
7597         (VSm): Likewise.
7598         (VSs): Likewise.
7599         (VSr): Likewise.
7600         (VSv): Likewise.
7601         (VS_scalar): Likewise.
7602         (VS_double): Likewise.
7603         (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
7604
7605         * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
7606         we support the ISA 2.07 128-bit integer arithmetic instructions.
7607         (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
7608         (enum rs6000_builtin_type_index): Add fields to hold V1TImode
7609         and TImode types for use with the builtin functions.
7610         (V1TI_type_node): Likewise.
7611         (unsigned_V1TI_type_node): Likewise.
7612         (intTI_type_internal_node): Likewise.
7613         (uintTI_type_internal_node): Likewise.
7614
7615         * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
7616         128-bit builtin functions.
7617         (UNSPEC_VADDEUQM): Likewise.
7618         (UNSPEC_VADDECUQ): Likewise.
7619         (UNSPEC_VSUBCUQ): Likewise.
7620         (UNSPEC_VSUBEUQM): Likewise.
7621         (UNSPEC_VSUBECUQ): Likewise.
7622         (VM): Add V1TImode to vector mode iterators.
7623         (VM2): Likewise.
7624         (VI_unit): Likewise.
7625         (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
7626         (altivec_vaddcuq): Likewise.
7627         (altivec_vsubuqm): Likewise.
7628         (altivec_vsubcuq): Likewise.
7629         (altivec_vaddeuqm): Likewise.
7630         (altivec_vaddecuq): Likewise.
7631         (altivec_vsubeuqm): Likewise.
7632         (altivec_vsubecuq): Likewise.
7633
7634         * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
7635         mode iterators.
7636         (BOOL_128): Likewise.
7637         (BOOL_REGS_OUTPUT): Likewise.
7638         (BOOL_REGS_OP1): Likewise.
7639         (BOOL_REGS_OP2): Likewise.
7640         (BOOL_REGS_UNARY): Likewise.
7641         (BOOL_REGS_AND_CR0): Likewise.
7642
7643         * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
7644         128-bit integer builtin support.
7645         (vec_vadduqm): Likewise.
7646         (vec_vaddecuq): Likewise.
7647         (vec_vaddeuqm): Likewise.
7648         (vec_vsubecuq): Likewise.
7649         (vec_vsubeuqm): Likewise.
7650         (vec_vsubcuq): Likewise.
7651         (vec_vsubuqm): Likewise.
7652
7653         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
7654         Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
7655         vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
7656         128-bit integer add/subtract to ISA 2.07.
7657
7658 2014-03-12  Joern Rennecke  <joern.rennecke@embecosm.com>
7659
7660         * config/arc/arc.c (arc_predicate_delay_insns):
7661         Fix third argument passed to conditionalize_nonjump.
7662
7663 2014-03-12  Yufeng Zhang  <yufeng.zhang@arm.com>
7664
7665         * config/aarch64/aarch64-builtins.c
7666         (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
7667         BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
7668         * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
7669         instead of __builtin_lfloor.
7670         (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
7671
7672 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
7673
7674         * tree-ssa-ifcombine.c (forwarder_block_to): New function.
7675         (tree_ssa_ifcombine_bb_1): New function.
7676         (tree_ssa_ifcombine_bb): Use it.  Handle also cases where else_bb
7677         is an empty forwarder block to then_bb or vice versa and then_bb
7678         and else_bb are effectively swapped.
7679
7680 2014-03-12  Christian Bruel  <christian.bruel@st.com>
7681
7682         PR target/60264
7683         * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
7684         REG_CFA_DEF_CFA note.
7685         (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
7686         (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
7687
7688 2014-03-12  Thomas Preud'homme  <thomas.preudhomme@arm.com>
7689
7690         PR tree-optimization/60454
7691         * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
7692
7693 2014-03-12  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7694
7695         * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
7696         Do not define target_cpu_default2 to generic.
7697         * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
7698         * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
7699         * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
7700
7701 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
7702             Marc Glisse  <marc.glisse@inria.fr>
7703
7704         PR tree-optimization/60502
7705         * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
7706         instead of build_low_bits_mask.
7707
7708 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
7709
7710         PR middle-end/60482
7711         * tree-vrp.c (register_edge_assert_for_1): Don't add assert
7712         if there are multiple uses, but op doesn't live on E edge.
7713         * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
7714         clobber stmts before __builtin_unreachable.
7715
7716 2014-03-11  Richard Sandiford  <rdsandiford@googlemail.com>
7717
7718         * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
7719         hard_frame_pointer_rtx.
7720         * cse.c (cse_insn): Remove volatile check.
7721         * cselib.c (cselib_process_insn): Likewise.
7722         * dse.c (scan_insn): Likewise.
7723
7724 2014-03-11  Joern Rennecke  <joern.rennecke@embecosm.com>
7725
7726         * config/arc/arc.c (conditionalize_nonjump): New function,
7727         broken out of ...
7728         (arc_ifcvt): ... this.
7729         (arc_predicate_delay_insns): Use it.
7730
7731 2014-03-11  Joern Rennecke  <joern.rennecke@embecosm.com>
7732
7733         * config/arc/predicates.md (extend_operand): During/after reload,
7734         allow const_int_operand.
7735         * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
7736         (umulsidi3_700): Likewise.  Change operand 2 constraint back to "cL".
7737         (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
7738         to "i".
7739         (umulsi3_highpart_i): Likewise.
7740
7741 2014-03-11  Richard Biener  <rguenther@suse.de>
7742
7743         * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
7744         Add asserts to guard possible wrong-code bugs.
7745
7746 2014-03-11  Richard Biener  <rguenther@suse.de>
7747
7748         PR tree-optimization/60429
7749         PR tree-optimization/60485
7750         * tree-ssa-structalias.c (set_union_with_increment): Properly
7751         take into account all fields that overlap the shifted vars.
7752         (do_sd_constraint): Likewise.
7753         (do_ds_constraint): Likewise.
7754         (get_constraint_for_ptr_offset): Likewise.
7755
7756 2014-03-11  Chung-Lin Tang  <cltang@codesourcery.com>
7757
7758         * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
7759         (nios2_compute_frame_layout):
7760         Add calculation of cfun->machine->fp_save_offset.
7761         (nios2_expand_prologue): Correct setting of frame pointer register
7762         in prologue.
7763         (nios2_expand_epilogue): Update recovery of stack pointer from
7764         frame pointer accordingly.
7765         (nios2_initial_elimination_offset): Update calculation of offset
7766         for eliminating to HARD_FRAME_POINTER_REGNUM.
7767
7768 2014-03-10  Jakub Jelinek  <jakub@redhat.com>
7769
7770         PR ipa/60457
7771         * ipa.c (symtab_remove_unreachable_nodes): Don't call
7772         cgraph_get_create_node on VAR_DECLs.
7773
7774 2014-03-10  Richard Biener  <rguenther@suse.de>
7775
7776         PR middle-end/60474
7777         * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
7778
7779 2014-03-08  Douglas B Rupp  <rupp@gnat.com>
7780
7781         * config/vms/vms.opt (vms_float_format): New variable.
7782
7783 2014-03-08  Tobias Burnus  <burnus@net-b.de>
7784
7785         * doc/invoke.texi (-fcilkplus): Update implementation status.
7786
7787 2014-03-08  Paulo Matos  <paulo@matos-sorge.com>
7788             Richard Biener  <rguenther@suse.de>
7789
7790         * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
7791         consistently accross all TUs.
7792         (run_gcc): Enable -fshort-double automatically at link at link-time
7793         and disallow override.
7794
7795 2014-03-08  Richard Sandiford  <rdsandiford@googlemail.com>
7796
7797         PR target/58271
7798         * config/mips/mips.c (mips_option_override): Promote -mpaired-single
7799         warning to an error.  Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
7800         if they can't be used.
7801
7802 2014-03-07  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
7803
7804         * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
7805         for Solaris 11/x86 ld.
7806         * configure: Regenerate.
7807
7808 2014-03-07  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
7809
7810         * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
7811         (LIB_TLS_SPEC): Save as ld_tls_libs.
7812         (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
7813         (HAVE_AS_IX86_TLSLDM): New test.
7814         * configure, config.in: Regenerate.
7815         * config/i386/i386.c (legitimize_tls_address): Fall back to
7816         TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
7817         cannot support TLS_MODEL_LOCAL_DYNAMIC.
7818         * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
7819         instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
7820
7821 2014-03-07  Paulo Matos  <paulo@matos-sorge.com>
7822
7823         * common.opt (fira-loop-pressure): Mark as optimization.
7824
7825 2014-03-07  Thomas Schwinge  <thomas@codesourcery.com>
7826
7827         * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
7828         an OpenMP mappable type.
7829
7830 2014-03-06  Matthias Klose  <doko@ubuntu.com>
7831
7832         * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
7833         MULTILIB_OSDIRNAMES is not defined.
7834
7835 2014-03-06  Jakub Jelinek  <jakub@redhat.com>
7836             Meador Inge  <meadori@codesourcery.com>
7837
7838         PR target/58595
7839         * config/arm/arm.c (arm_tls_symbol_p): Remove.
7840         (arm_legitimize_address): Call legitimize_tls_address for any
7841         arm_tls_referenced_p expression, handle constant addend.  Call it
7842         before testing for !TARGET_ARM.
7843         (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
7844
7845 2014-03-06  Richard Biener  <rguenther@suse.de>
7846
7847         PR middle-end/60445
7848         PR lto/60424
7849         PR lto/60427
7850         Revert
7851         2014-03-04  Paulo Matos  <paulo@matos-sorge.com>
7852
7853         * tree-streamer.c (record_common_node): Assert we don't record
7854         nodes with type double.
7855         (preload_common_node): Skip type double, complex double and double
7856         pointer since it is now frontend dependent due to fshort-double option.
7857
7858 2014-03-06  Richard Biener  <rguenther@suse.de>
7859
7860         * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
7861         or -fno-lto is specified and the linker has full plugin support.
7862         * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
7863         (main): Remove -flto processing, adjust lto_mode using use_plugin late.
7864         * lto-wrapper.c (merge_and_complain): Merge compile-time
7865         optimization levels.
7866         (run_gcc): And pass it through to the link options.
7867
7868 2014-03-06  Alexandre Oliva  <aoliva@redhat.com>
7869
7870         PR debug/60381
7871         Revert:
7872         2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
7873         PR debug/59992
7874         * cselib.c (remove_useless_values): Skip to avoid quadratic
7875         behavior if the condition moved from...
7876         (cselib_process_insn): ... here holds.
7877
7878 2014-03-05  Jakub Jelinek  <jakub@redhat.com>
7879
7880         PR plugins/59335
7881         * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
7882         ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
7883
7884         PR plugins/59335
7885         * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
7886         (TM_H): Add x86-tune.def.
7887
7888 2014-03-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7889
7890         * config/aarch64/aarch64.c (generic_tunings):
7891         Use cortexa57_extra_costs.
7892
7893 2014-03-05  Jakub Jelinek  <jakub@redhat.com>
7894
7895         PR lto/60404
7896         * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
7897         of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
7898         * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
7899         cost for in_lto_p.
7900
7901 2014-03-04  Heiher  <r@hev.cc>
7902
7903         * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
7904         * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
7905
7906 2014-03-04  Uros Bizjak  <ubizjak@gmail.com>
7907
7908         * config/i386/predicates.md (const2356_operand): Change to ...
7909         (const2367_operand): ... this.
7910         * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
7911         const2367_operand.
7912         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
7913         (*avx512pf_scatterpf<mode>sf): Ditto.
7914         (avx512pf_scatterpf<mode>df): Ditto.
7915         (*avx512pf_scatterpf<mode>df_mask): Ditto.
7916         (*avx512pf_scatterpf<mode>df): Ditto.
7917         * config/i386/i386.c (ix86_expand_builtin): Update
7918         incorrect hint operand error message.
7919
7920 2014-03-04  Richard Biener  <rguenther@suse.de>
7921
7922         * lto-section-in.c (lto_get_section_data): Fix const cast.
7923
7924 2014-03-04  Paulo Matos  <paulo@matos-sorge.com>
7925
7926         * tree-streamer.c (record_common_node): Assert we don't record
7927         nodes with type double.
7928         (preload_common_node): Skip type double, complex double and double
7929         pointer since it is now frontend dependent due to fshort-double option.
7930
7931 2014-03-04  Richard Biener  <rguenther@suse.de>
7932
7933         PR lto/60405
7934         * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
7935         (lto_input_toplevel_asms): Likewise.
7936         * lto-section-in.c (lto_get_section_data): Instead do it here
7937         for every section.
7938
7939 2014-03-04  Richard Biener  <rguenther@suse.de>
7940
7941         PR tree-optimization/60382
7942         * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
7943         dead PHIs a reduction.
7944
7945 2014-03-03  Uros Bizjak  <ubizjak@gmail.com>
7946
7947         * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
7948         hint value.
7949         (_mm_prefetch): Move out of GCC target("sse") pragma.
7950         * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
7951         GCC target("prfchw") pragma.
7952         * config/i386/i386.md (prefetch): Emit prefetchwt1 only
7953         for locality <= 2.
7954         * config/i386/i386.c (ix86_option_override_internal): Enable
7955         -mprfchw with -mprefetchwt1.
7956
7957 2014-03-03  Joern Rennecke  <joern.rennecke@embecosm.com>
7958
7959         * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
7960         Mark as varying.
7961
7962 2014-03-03  Joern Rennecke  <joern.rennecke@embecosm.com>
7963
7964         * opts.h (CL_PCH_IGNORE): Define.
7965         * targhooks.c (option_affects_pch_p):
7966         Return false for options that have CL_PCH_IGNORE set.
7967         * opt-functions.awk: Process PchIgnore.
7968         * doc/options.texi: Document PchIgnore.
7969
7970         * config/arc/arc.opt (misize): Add PchIgnore property.
7971
7972 2014-03-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7973
7974         * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
7975         reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
7976         constraint on constants to permit them being loaded into
7977         GENERAL_REGS or BASE_REGS.
7978
7979 2014-03-03  Nick Clifton  <nickc@redhat.com>
7980
7981         * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
7982         anti-cacnonical alternatives.
7983         (negandhi3_real): New pattern.
7984         * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
7985
7986 2014-03-03  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
7987
7988         * config/avr/avr-mcus.def: Remove atxmega16x1.
7989         * config/avr/avr-tables.opt: Regenerate.
7990         * config/avr/t-multilib: Regenerate.
7991         * doc/avr-mmcu.texi: Regenerate.
7992
7993 2014-03-03  Tobias Grosser  <tobias@grosser.es>
7994             Mircea Namolaru  <mircea.namolaru@inria.fr>
7995
7996         PR tree-optimization/58028
7997         * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
7998         scalar dimensions.
7999
8000 2014-03-03  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
8001
8002         * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
8003         not handled by recognizers.
8004
8005 2014-03-03  Jakub Jelinek  <jakub@redhat.com>
8006
8007         PR middle-end/60175
8008         * function.c (expand_function_end): Don't emit
8009         clobber_return_register sequence if clobber_after is a BARRIER.
8010         * cfgexpand.c (construct_exit_block): Append instructions before
8011         return_label to prev_bb.
8012
8013 2014-03-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8014
8015         * config/rs6000/constraints.md: Document reserved use of "wc".
8016
8017 2014-03-02  Jan Hubicka  <hubicka@ucw.cz>
8018
8019         PR ipa/60150
8020         * ipa.c (function_and_variable_visibility): When dissolving comdat
8021         group, also set all symbols to local.
8022
8023 2014-03-02  Jan Hubicka  <hubicka@ucw.cz>
8024
8025         PR ipa/60306
8026
8027         Revert:
8028         2013-12-14   Jan Hubicka  <jh@suse.cz>
8029         PR middle-end/58477
8030         * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
8031
8032 2014-03-02  Jon Beniston  <jon@beniston.com>
8033
8034         PR bootstrap/48230
8035         PR bootstrap/50927
8036         PR bootstrap/52466
8037         PR target/46898
8038         * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
8039           (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
8040         * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
8041         (simple_return, *simple_return): New patterns
8042         * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
8043         * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
8044
8045 2014-03-01  Paolo Carlini  <paolo.carlini@oracle.com>
8046
8047         * dwarf2out.c (gen_subprogram_die): Tidy.
8048
8049 2014-03-01  Oleg Endo  <olegendo@gcc.gnu.org>
8050
8051         PR target/60071
8052         * config/sh/sh.md (*mov_t_msb_neg): Split into ...
8053         (*mov_t_msb_neg_negc): ... this new insn.
8054
8055 2014-02-28  Jason Merrill  <jason@redhat.com>
8056
8057         PR c++/58678
8058         * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
8059         function.
8060
8061 2014-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
8062
8063         PR c++/60314
8064         * dwarf2out.c (decltype_auto_die): New static.
8065         (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
8066         (gen_type_die_with_usage): Handle 'decltype(auto)'.
8067         (is_cxx_auto): Likewise.
8068
8069 2014-02-28  Ian Bolton  <ian.bolton@arm.com>
8070
8071         * config/aarch64/aarch64.h: Define __ARM_NEON by default if
8072         we are not using general regs only.
8073
8074 2014-02-28  Richard Biener  <rguenther@suse.de>
8075
8076         PR target/60280
8077         * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
8078         previous fix and only allow to remove trivial pre-headers
8079         and latches.  Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
8080         (remove_forwarder_block): Properly update the latch of a loop.
8081
8082 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
8083
8084         PR debug/59992
8085         * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
8086         (cselib_preserved_hash_table): New.
8087         (preserve_constants_and_equivs): Move preserved vals to it.
8088         (cselib_find_slot): Look it up first.
8089         (cselib_init): Initialize it.
8090         (cselib_finish): Release it.
8091         (dump_cselib_table): Dump it.
8092
8093 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
8094
8095         PR debug/59992
8096         * cselib.c (remove_useless_values): Skip to avoid quadratic
8097         behavior if the condition moved from...
8098         (cselib_process_insn): ... here holds.
8099
8100 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
8101
8102         PR debug/57232
8103         * var-tracking.c (vt_initialize): Apply the same condition to
8104         preserve the CFA base value.
8105
8106 2014-02-28  Joey Ye  <joey.ye@arm.com>
8107
8108         PR target/PR60169
8109         * config/arm/arm.c (thumb_far_jump_used_p): Don't change
8110         if reload in progress or completed.
8111
8112 2014-02-28  Tobias Burnus  <burnus@net-b.de>
8113
8114         PR middle-end/60147
8115         * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
8116         NAMELIST_DECL.
8117
8118 2014-02-27  H.J. Lu  <hongjiu.lu@intel.com>
8119
8120         * doc/tm.texi.in (Condition Code Status): Update documention for
8121         relative locations of cc0-setter and cc0-user.
8122
8123 2014-02-27  Jeff Law  <law@redhat.com>
8124
8125         PR rtl-optimization/52714
8126         * combine.c (try_combine): When splitting an unrecognized PARALLEL
8127         into two independent simple sets, if I3 is a jump, ensure the
8128         pattern we place into I3 is a (set (pc) ...).
8129
8130 2014-02-27  Mikael Pettersson  <mikpe@it.uu.se>
8131             Jeff Law  <law@redhat.com>
8132
8133         PR rtl-optimization/49847
8134         * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
8135         are in different blocks.
8136         * doc/tm.texi (Condition Code Status): Update documention for
8137         relative locations of cc0-setter and cc0-user.
8138
8139 2014-02-27  Vladimir Makarov  <vmakarov@redhat.com>
8140
8141         PR target/59222
8142         * lra.c (lra_emit_add): Check SUBREG too.
8143
8144 2014-02-27  Andreas Schwab  <schwab@suse.de>
8145
8146         * config/m68k/m68k.c (m68k_option_override): Disable
8147         -flive-range-shrinkage for classic m68k.
8148         (m68k_override_options_after_change): Likewise.
8149
8150 2014-02-27  Marek Polacek  <polacek@redhat.com>
8151
8152         PR middle-end/59223
8153         * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
8154         -Wmaybe-uninitialized.
8155
8156 2014-02-27  Alan Modra  <amodra@gmail.com>
8157
8158         PR target/57936
8159         * reload1.c (emit_input_reload_insns): When reload_override_in,
8160         set old to rl->in_reg when rl->in_reg is a subreg.
8161
8162 2014-02-26  Richard Biener  <rguenther@suse.de>
8163
8164         PR bootstrap/60343
8165         * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
8166
8167 2014-02-25  Ilya Tocar  <ilya.tocar@intel.com>
8168
8169         * common/config/i386/predicates.md (const1256_operand): Remove.
8170         (const2356_operand): New.
8171         (const_1_to_2_operand): Remove.
8172         * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
8173         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
8174         (*avx512pf_gatherpf<mode>sf): Ditto.
8175         (avx512pf_gatherpf<mode>df): Ditto.
8176         (*avx512pf_gatherpf<mode>df_mask): Ditto.
8177         (*avx512pf_gatherpf<mode>df): Ditto.
8178         (avx512pf_scatterpf<mode>sf): Ditto.
8179         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
8180         (*avx512pf_scatterpf<mode>sf): Ditto.
8181         (avx512pf_scatterpf<mode>df): Ditto.
8182         (*avx512pf_scatterpf<mode>df_mask): Ditto.
8183         (*avx512pf_scatterpf<mode>df): Ditto.
8184         * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
8185
8186 2014-02-26  Ilya Tocar  <ilya.tocar@intel.com>
8187
8188         * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
8189         (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
8190         (_mm512_mask_testn_epi64_mask): Move to ...
8191         * config/i386/avx512cdintrin.h: Here.
8192         * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
8193         * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
8194         (avx512f_scalef<mode><mask_name><round_name>): Ditto.
8195         (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
8196         TARGET_AVX512F from TARGET_AVX512CD.
8197
8198 2014-02-26  Richard Biener  <rguenther@suse.de>
8199
8200         PR ipa/60327
8201         * ipa.c (walk_polymorphic_call_targets): Properly guard
8202         call to inline_update_overall_summary.
8203
8204 2014-02-26  Bin Cheng  <bin.cheng@arm.com>
8205
8206         PR target/60280
8207         * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
8208         and latches only if requested.  Fix latch if it is removed.
8209         * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
8210         LOOPS_HAVE_PREHEADERS.
8211
8212 2014-02-25  Andrew Pinski  <apinski@cavium.com>
8213
8214         * builtins.c (expand_builtin_thread_pointer): Create a new target
8215         when the target is NULL.
8216
8217 2014-02-25  Vladimir Makarov  <vmakarov@redhat.com>
8218
8219         PR rtl-optimization/60317
8220         * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
8221         * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
8222         * lra-assigns.c: Include params.h.
8223         (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
8224         other reload pseudos considerations.
8225
8226 2014-02-25  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8227
8228         * config/rs6000/vector.md (*vector_unordered<mode>): Change split
8229         to use canonical form for nor<mode>3.
8230
8231 2014-02-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8232
8233         PR target/55426
8234         * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
8235         conversions.
8236
8237 2014-02-25  Ilya Tocar  <ilya.tocar@intel.com>
8238
8239         * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
8240         (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
8241         (ix86_handle_option): Handle OPT_mprefetchwt1.
8242         * config/i386/cpuid.h (bit_PREFETCHWT1): New.
8243         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
8244         PREFETCHWT1 CPUID.
8245         * config/i386/i386-c.c (ix86_target_macros_internal): Handle
8246         OPTION_MASK_ISA_PREFETCHWT1.
8247         * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
8248         (PTA_PREFETCHWT1): New.
8249         (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
8250         (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
8251         * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
8252         * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
8253         (*prefetch_avx512pf_<mode>_: Change into ...
8254         (*prefetch_prefetchwt1_<mode>: This.
8255         * config/i386/i386.opt (mprefetchwt1): New.
8256         * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
8257         (_mm_prefetch): Handle intent to write.
8258         * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
8259
8260 2014-02-25  Richard Biener  <rguenther@suse.de>
8261
8262         PR middle-end/60291
8263         * emit-rtl.c (mem_attrs_htab): Remove.
8264         (mem_attrs_htab_hash): Likewise.
8265         (mem_attrs_htab_eq): Likewise.
8266         (set_mem_attrs): Always allocate new mem-attrs when something changed.
8267         (init_emit_once): Do not allocate mem_attrs_htab.
8268
8269 2014-02-25  Richard Biener  <rguenther@suse.de>
8270
8271         PR lto/60319
8272         * lto-opts.c (lto_write_options): Output non-explicit conservative
8273         -fwrapv, -fno-trapv and -fno-strict-overflow.
8274         * lto-wrapper.c (merge_and_complain): Handle merging those options.
8275         (run_gcc): And pass them through.
8276
8277 2014-02-25  Andrey Belevantsev  <abel@ispras.ru>
8278
8279         * sel-sched.c (calculate_new_fences): New parameter ptime.
8280         Calculate it as a maximum over all fence cycles.
8281         (sel_sched_region_2): Adjust the call to calculate_new_fences.
8282         Print the final schedule timing when sched_verbose.
8283
8284 2014-02-25  Andrey Belevantsev  <abel@ispras.ru>
8285
8286         PR rtl-optimization/60292
8287         * sel-sched.c (fill_vec_av_set): Do not reset target availability
8288         bit fot the fence instruction.
8289
8290 2014-02-24  Alangi Derick  <alangiderick@gmail.com>
8291
8292         * calls.h: Fix typo in comment.
8293
8294 2014-02-24  John David Anglin  <danglin@gcc.gnu.org>
8295
8296         * config/pa/pa.c (pa_output_move_double): Don't valididate when
8297         adjusting offsetable addresses.
8298
8299 2014-02-24  Guozhi Wei  <carrot@google.com>
8300
8301         * sparseset.h (sparseset_pop): Fix the wrong index.
8302
8303 2014-02-24  Walter Lee  <walt@tilera.com>
8304
8305         * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
8306         (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
8307         triplet.
8308         * common/config/tilegx/tilegx-common.c
8309         (TARGET_DEFAULT_TARGET_FLAGS): Define.
8310         * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
8311         (LINK_SPEC): Ditto.
8312         * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
8313         * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
8314         (tilegx_gimplify_va_arg_expr): Handle big endian.
8315         (tilegx_expand_unaligned_load): Ditto.
8316         (tilegx_expand_unaligned_store): Ditto.
8317         (TARGET_RETURN_IN_MSB): New.
8318         * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
8319         (TARGET_ENDIAN_DEFAULT): New.
8320         (TARGET_BIG_ENDIAN): Handle big endian.
8321         (BYTES_BIG_ENDIAN): Ditto.
8322         (WORDS_BIG_ENDIAN): Ditto.
8323         (FLOAT_WORDS_BIG_ENDIAN): Ditto.
8324         (ENDIAN_SPEC): New.
8325         (EXTRA_SPECS): New.
8326         * config/tilegx/tilegx.md (extv): Handle big endian.
8327         (extzv): Ditto.
8328         (insn_st<n>): Ditto.
8329         (insn_st<n>_add<bitsuffix>): Ditto.
8330         (insn_stnt<n>): Ditto.
8331         (insn_stnt<n>_add<bitsuffix>):Ditto.
8332         (vec_interleave_highv8qi): Handle big endian.
8333         (vec_interleave_highv8qi_be): New.
8334         (vec_interleave_highv8qi_le): New.
8335         (insn_v1int_h): Handle big endian.
8336         (vec_interleave_lowv8qi): Handle big endian.
8337         (vec_interleave_lowv8qi_be): New.
8338         (vec_interleave_lowv8qi_le): New.
8339         (insn_v1int_l): Handle big endian.
8340         (vec_interleave_highv4hi): Handle big endian.
8341         (vec_interleave_highv4hi_be): New.
8342         (vec_interleave_highv4hi_le): New.
8343         (insn_v2int_h): Handle big endian.
8344         (vec_interleave_lowv4hi): Handle big endian.
8345         (vec_interleave_lowv4hi_be): New.
8346         (vec_interleave_lowv4hi_le): New.
8347         (insn_v2int_l): Handle big endian.
8348         (vec_interleave_highv2si): Handle big endian.
8349         (vec_interleave_highv2si_be): New.
8350         (vec_interleave_highv2si_le): New.
8351         (insn_v4int_h): Handle big endian.
8352         (vec_interleave_lowv2si): Handle big endian.
8353         (vec_interleave_lowv2si_be): New.
8354         (vec_interleave_lowv2si_le): New.
8355         (insn_v4int_l): Handle big endian.
8356         * config/tilegx/tilegx.opt (mbig-endian): New option.
8357         (mlittle-endian): New option.
8358         * doc/install.texi: Document tilegxbe-linux.
8359         * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
8360
8361 2014-02-24  Martin Jambor  <mjambor@suse.cz>
8362
8363         PR ipa/60266
8364         * ipa-cp.c (propagate_constants_accross_call): Bail out early if
8365         there are no parameter descriptors.
8366
8367 2014-02-24  Andrey Belevantsev  <abel@ispras.ru>
8368
8369         PR rtl-optimization/60268
8370         * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
8371         initialization to ...
8372         (sched_rgn_init): ... here.
8373         (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
8374
8375 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
8376
8377         * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
8378         names.
8379
8380 2014-02-23  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
8381
8382         * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
8383         definition.
8384
8385 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
8386
8387         * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
8388         define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
8389
8390 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
8391
8392         * config/microblaze/predicates.md: Add cmp_op predicate.
8393         * config/microblaze/microblaze.md: Add branch_compare instruction
8394         which uses cmp_op predicate and emits cmp insn before branch.
8395         * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
8396         to microblaze_expand_conditional_branch and consolidate logic.
8397         (microblaze_expand_conditional_branch): emit branch_compare
8398         insn instead of handling cmp op separate from branch insn.
8399
8400 2014-02-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8401
8402         * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
8403         to permit subregs.
8404
8405 2014-02-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8406
8407         * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
8408         define_insn with define_expand and new define_insn
8409         *altivec_lve<VI_char>x_internal.
8410         (altivec_stve<VI_char>x): Replace define_insn with define_expand
8411         and new define_insn *altivec_stve<VI_char>x_internal.
8412         * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
8413         prototype.
8414         * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
8415         lve*x built-ins.
8416         (altivec_expand_stvex_be): New function.
8417
8418 2014-02-22  Joern Rennecke  <joern.rennecke@embecosm.com>
8419
8420         * config/avr/avr.c (avr_can_eliminate): Allow elimination from
8421         ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
8422         * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
8423         ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
8424
8425 2014-02-21  Vladimir Makarov  <vmakarov@redhat.com>
8426
8427         PR target/60298
8428         * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
8429         instead of emit_move_insn.
8430
8431 2014-02-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8432
8433         * config/rs6000/altivec.md (altivec_vsumsws): Replace second
8434         vspltw with vsldoi.
8435         (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
8436         gen_altivec_vsumsws.
8437
8438 2014-02-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8439
8440         * config/rs6000/altivec.md (altivec_lvxl): Rename as
8441         *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
8442         (altivec_lvxl_<mode>): New define_expand incorporating
8443         -maltivec=be semantics where needed.
8444         (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
8445         (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
8446         semantics where needed.
8447         (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
8448         (altivec_stvx_<mode>): New define_expand incorporating
8449         -maltivec=be semantics where needed.
8450         (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
8451         VM2 iterator instead of V4SI.
8452         (altivec_stvxl_<mode>): New define_expand incorporating
8453         -maltivec=be semantics where needed.
8454         * config/rs6000/rs6000-builtin.def: Add new built-in definitions
8455         LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
8456         LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
8457         STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
8458         STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
8459         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
8460         ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
8461         similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
8462         ALTIVEC_BUILTIN_STVXL.
8463         * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
8464         (altivec_expand_stvx_be): Likewise.
8465         * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
8466         (altivec_expand_lvx_be): Likewise.
8467         (altivec_expand_stvx_be): Likewise.
8468         (altivec_expand_builtin): Add cases for
8469         ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
8470         ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
8471         (altivec_init_builtins): Add definitions for
8472         __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
8473         __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
8474
8475 2014-02-21  Catherine Moore  <clm@codesourcery.com>
8476
8477         * doc/invoke.texi (mvirt, mno-virt): Document.
8478         * config/mips/mips.opt (mvirt): New option.
8479         * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
8480
8481 2014-02-21  Richard Biener  <rguenther@suse.de>
8482
8483         PR tree-optimization/60276
8484         * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
8485         (STMT_VINFO_MIN_NEG_DIST): New macro.
8486         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
8487         STMT_VINFO_MIN_NEG_DIST.
8488         * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
8489         made for negative dependence distances still hold.
8490
8491 2014-02-21  Richard Biener  <rguenther@suse.de>
8492
8493         PR middle-end/60291
8494         * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
8495         DECL_INITIAL for globals not in the current function context.
8496
8497 2014-02-21  Jakub Jelinek  <jakub@redhat.com>
8498
8499         PR tree-optimization/56490
8500         * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
8501         * tree-ssa-uninit.c: Include params.h.
8502         (compute_control_dep_chain): Add num_calls argument, return false
8503         if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
8504         num_calls to recursive call.
8505         (find_predicates): Change dep_chain into normal array,
8506         cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
8507         variable and adjust compute_control_dep_chain caller.
8508         (find_def_preds): Likewise.
8509
8510 2014-02-21  Thomas Schwinge  <thomas@codesourcery.com>
8511
8512         * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
8513         <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
8514
8515 2014-02-21  Nick Clifton  <nickc@redhat.com>
8516
8517         * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
8518         (pushhi1): Likewise.
8519         (popqi1): Add mode to pre_dec.
8520         (pophi1): Likewise.
8521
8522 2014-02-21  Jakub Jelinek  <jakub@redhat.com>
8523
8524         * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
8525         mode for mask of V8SFmode permutation.
8526
8527 2014-02-20  Richard Henderson  <rth@redhat.com>
8528
8529         PR c++/60272
8530         * builtins.c (expand_builtin_atomic_compare_exchange): Always make
8531         a new pseudo for OLDVAL.
8532
8533 2014-02-20  Jakub Jelinek  <jakub@redhat.com>
8534
8535         PR target/57896
8536         * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
8537         gen_reg_rtx if d->testing_p.
8538         (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
8539         if d->testing_p and we will certainly return true.
8540         (expand_vec_perm_even_odd_1): Likewise.  Don't call gen_reg_rtx
8541         if d->testing_p.
8542
8543 2014-02-20  Uros Bizjak  <ubizjak@gmail.com>
8544
8545         * emit-rtl.c (gen_reg_rtx): Assert that
8546         crtl->emit.regno_pointer_align_length is non-zero.
8547
8548 2014-02-20  Richard Henderson  <rth@redhat.com>
8549
8550         PR c++/60272
8551         * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
8552         on failure the store back into EXPECT.
8553
8554 2014-02-20  Chung-Lin Tang  <cltang@codesourcery.com>
8555             Sandra Loosemore  <sandra@codesourcery.com>
8556
8557         * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
8558         * config/nios2/nios2.c (nios2_function_profiler): Add
8559         -fPIC (flag_pic == 2) support.
8560         (nios2_handle_custom_fpu_cfg): Fix warning parameter.
8561         (nios2_large_offset_p): New function.
8562         (nios2_unspec_reloc_p): Move up position, update to use
8563         nios2_large_offset_p.
8564         (nios2_unspec_address): Remove function.
8565         (nios2_unspec_offset): New function.
8566         (nios2_large_got_address): New function.
8567         (nios2_got_address): Add large offset support.
8568         (nios2_legitimize_tls_address): Update usage of removed and new
8569         functions.
8570         (nios2_symbol_binds_local_p): New function.
8571         (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
8572         (nios2_legitimize_address): Update to use nios2_large_offset_p.
8573         (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
8574         (nios2_print_operand): Merge H/L processing, add hiadj/lo
8575         processing for (const (unspec ...)).
8576         (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
8577
8578 2014-02-20  Richard Biener  <rguenther@suse.de>
8579
8580         * tree-cfg.c (replace_uses_by): Mark altered BBs before
8581         doing the substitution.
8582         (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
8583
8584 2014-02-20  Martin Jambor  <mjambor@suse.cz>
8585
8586         PR ipa/55260
8587         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
8588         info when checking whether lattices are bottom.
8589
8590 2014-02-20  Richard Biener  <rguenther@suse.de>
8591
8592         PR middle-end/60221
8593         * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
8594         regions at -O0.
8595
8596 2014-02-20  Jan Hubicka  <hubicka@ucw.cz>
8597
8598         PR ipa/58555
8599         * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
8600         parameter specifying the scaling.
8601         (inline_call): Update.
8602         (want_inline_recursively): Guard division by zero.
8603         (recursive_inlining): Update.
8604         * ipa-inline.h (clone_inlined_nodes): Update.
8605
8606 2014-02-20  Ilya Tocar  <ilya.tocar@intel.com>
8607
8608         PR target/60204
8609         * config/i386/i386.c (classify_argument): Pass structures of size
8610         64 bytes or less in register.
8611
8612 2014-02-20  Ilya Tocar  <ilya.tocar@intel.com>
8613             Kirill Yukhin  <kirill.yukhin@intel.com>
8614
8615         * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
8616         (_mm_rcp28_round_ss): Ditto.
8617         (_mm_rsqrt28_round_sd): Ditto.
8618         (_mm_rsqrt28_round_ss): Ditto.
8619         * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
8620         (_mm_rcp14_round_ss): Ditto.
8621         (_mm_rsqrt14_round_sd): Ditto.
8622         (_mm_rsqrt14_round_ss): Ditto.
8623         * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
8624         the first input operand, get rid of match_dup.
8625         (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
8626         attribute to sse.
8627         (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
8628         Ditto.
8629         (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
8630         operand as the first input operand, set type attribute.
8631         (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
8632         Set type attribute.
8633         (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
8634         operand as the first input operand, set type attribute.
8635
8636 2014-02-19  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8637
8638         * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
8639         bit of zero.
8640
8641 2014-02-19  H.J. Lu  <hongjiu.lu@intel.com>
8642
8643         PR target/60207
8644         * config/i386/i386.c (construct_container): Remove TFmode check
8645         for X86_64_INTEGER_CLASS.
8646
8647 2014-02-19  Uros Bizjak  <ubizjak@gmail.com>
8648
8649         PR target/59794
8650         * config/i386/i386.c (type_natural_mode): Warn for ABI changes
8651         only when -Wpsabi is enabled.
8652
8653 2014-02-19  Michael Hudson-Doyle  <michael.hudson@linaro.org>
8654
8655          PR target/59799
8656         * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
8657         passing arrays in registers are the same as for structs, so remove the
8658         special case for them.
8659
8660 2014-02-19  Eric Botcazou  <ebotcazou@adacore.com>
8661
8662         * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
8663         destination type, extract only the valid bits if the source type is not
8664         integral and has a different mode.
8665
8666 2014-02-19  Richard Biener  <rguenther@suse.de>
8667
8668         PR ipa/60243
8669         * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
8670         for all calls.
8671
8672 2014-02-19  Richard Biener  <rguenther@suse.de>
8673
8674         PR ipa/60243
8675         * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
8676         (ipa_modify_call_arguments): Emit an argument load explicitely and
8677         preserve virtual SSA form there and for the replacement call.
8678         Do not update SSA form nor free dominance info.
8679
8680 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
8681
8682         * ipa.c (function_and_variable_visibility): Also clear WEAK
8683         flag when disolving COMDAT_GROUP.
8684
8685 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
8686
8687         * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
8688         * ipa-prop.c (ipa_set_jf_known_type): Return early when
8689         not devirtualizing.
8690         (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
8691         do more sanity checks.
8692         (detect_type_change): Return true when giving up early.
8693         (compute_complex_assign_jump_func): Fix type parameter of
8694         ipa_set_ancestor_jf.
8695         (compute_complex_ancestor_jump_func): Likewise.
8696         (update_jump_functions_after_inlining): Fix updating of
8697         ancestor function.
8698         * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
8699
8700 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
8701
8702         * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
8703         inline clones when edge disappears.
8704
8705 2014-02-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
8706
8707         PR target/60203
8708         * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
8709         Split 64-bit moves into 2 patterns.  Do not allow the use of
8710         direct move for TDmode in little endian, since the decimal value
8711         has little endian bytes within a word, but the 64-bit pieces are
8712         ordered in a big endian fashion, and normal subreg's of TDmode are
8713         not allowed.
8714         (mov<mode>_64bit_dm): Likewise.
8715         (movtd_64bit_nodm): Likewise.
8716
8717 2014-02-18  Eric Botcazou  <ebotcazou@adacore.com>
8718
8719         PR tree-optimization/60174
8720         * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
8721         statement of an SSA_NAME that occurs in an abnormal PHI node.
8722
8723 2014-02-18  Jakub Jelinek  <jakub@redhat.com>
8724
8725         PR sanitizer/60142
8726         * final.c (SEEN_BB): Remove.
8727         (SEEN_NOTE, SEEN_EMITTED): Renumber.
8728         (final_scan_insn): Don't force_source_line on second
8729         NOTE_INSN_BASIC_BLOCK.
8730
8731 2014-02-18  Uros Bizjak  <ubizjak@gmail.com>
8732
8733         PR target/60205
8734         * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
8735         * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
8736         (type_natural_mode): Warn ABI change when %zmm register is not
8737         available for AVX512F vector value passing.
8738
8739 2014-02-18  Kai Tietz  <ktietz@redhat.com>
8740
8741         PR target/60193
8742         * config/i386/i386.c (ix86_expand_prologue): Use value in
8743         rax register as displacement when restoring %r10 or %rax.
8744         Fix wrong offset when restoring both registers.
8745
8746 2014-02-18  Eric Botcazou  <ebotcazou@adacore.com>
8747
8748         * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
8749         assertion with conditional return.
8750
8751 2014-02-18  Jakub Jelinek  <jakub@redhat.com>
8752             Uros Bizjak  <ubizjak@gmail.com>
8753
8754         PR driver/60233
8755         * config/i386/driver-i386.c (host_detect_local_cpu): If
8756         YMM state is not saved by the OS, also clear has_f16c.  Move
8757         CPUID 0x80000001 handling before YMM state saving checking.
8758
8759 2014-02-18  Andrey Belevantsev  <abel@ispras.ru>
8760
8761         PR rtl-optimization/58960
8762         * haifa-sched.c (alloc_global_sched_pressure_data): New,
8763         factored out from ...
8764         (sched_init): ... here.
8765         (free_global_sched_pressure_data): New, factored out from ...
8766         (sched_finish): ... here.
8767         * sched-int.h (free_global_sched_pressure_data): Declare.
8768         * sched-rgn.c (nr_regions_initial): New static global.
8769         (haifa_find_rgns): Initialize it.
8770         (schedule_region): Disable sched-pressure for the newly
8771         generated regions.
8772
8773 2014-02-17  Richard Biener  <rguenther@suse.de>
8774
8775         * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
8776         release SSA defs of pattern stmts.
8777
8778 2014-02-17  Richard Biener  <rguenther@suse.de>
8779
8780         * tree-inline.c (expand_call_inline): Release the virtual
8781         operand defined by the call we are about to inline.
8782
8783 2014-02-17  Richard Biener  <rguenther@suse.de>
8784
8785         * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
8786
8787 2014-02-17  Kirill Yukhin  <kirill.yukhin@intel.com>
8788             Ilya Tocar  <ilya.tocar@intel.com>
8789
8790         * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
8791         arguments order in builtin.
8792         (_mm512_permutexvar_epi64): Ditto.
8793         (_mm512_mask_permutexvar_epi64): Ditto
8794         (_mm512_maskz_permutexvar_epi32): Ditto
8795         (_mm512_permutexvar_epi32): Ditto
8796         (_mm512_mask_permutexvar_epi32): Ditto
8797
8798 2014-02-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8799
8800         * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
8801         (p8_vmrgow): Likewise.
8802
8803 2014-02-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8804
8805         * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
8806         endian targets.
8807
8808 2014-02-15  Michael Meissner  <meissner@linux.vnet.ibm.com>
8809
8810         PR target/60203
8811         * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
8812         (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
8813         into 64-bit and 32-bit moves.  On 64-bit moves, add support for
8814         using direct move instructions on ISA 2.07.  Also adjust
8815         instruction length for 64-bit.
8816         (mov<mode>_64bit, TFmode/TDmode): Likewise.
8817         (mov<mode>_32bit, TFmode/TDmode): Likewise.
8818
8819 2014-02-15  Alan Modra  <amodra@gmail.com>
8820
8821         PR target/58675
8822         PR target/57935
8823         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
8824         find_replacement on parts of insn rtl that might be reloaded.
8825
8826 2014-02-15  Richard Biener  <rguenther@suse.de>
8827
8828         PR tree-optimization/60183
8829         * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
8830         (tree_ssa_phiprop): Calculate and free post-dominators.
8831
8832 2014-02-14  Jeff Law  <law@redhat.com>
8833
8834         PR rtl-optimization/60131
8835         * ree.c (get_extended_src_reg): New function.
8836         (combine_reaching_defs): Use it rather than assuming location of REG.
8837         (find_and_remove_re): Verify first operand of extension is
8838         a REG before adding the insns to the copy list.
8839
8840 2014-02-14  Roland McGrath  <mcgrathr@google.com>
8841
8842         * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
8843         * configure: Regenerated.
8844         * config.in: Regenerated.
8845         * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
8846         instead of ASM_SHORT.
8847
8848 2014-02-14  Vladimir Makarov  <vmakarov@redhat.com>
8849             Richard Earnshaw  <rearnsha@arm.com>
8850
8851         PR rtl-optimization/59535
8852         * lra-constraints.c (process_alt_operands): Encourage alternative
8853         when unassigned pseudo class is superset of the alternative class.
8854         (inherit_reload_reg): Don't inherit when optimizing for code size.
8855         * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
8856         returning CORE_REGS for anything but Thumb1 and BASE_REGS for
8857         modes not less than 4 for Thumb1.
8858
8859 2014-02-14  Kyle McMartin  <kyle@redhat.com>
8860
8861         PR pch/60010
8862         * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
8863
8864 2014-02-14  Richard Biener  <rguenther@suse.de>
8865
8866         * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
8867         (get_frame_arg): Drop the assert with langhook types_compatible_p.
8868         Do not strip INDIRECT_REFs.
8869
8870 2014-02-14  Richard Biener  <rguenther@suse.de>
8871
8872         PR lto/60179
8873         * lto-streamer-out.c (DFS_write_tree_body): Do not follow
8874         DECL_FUNCTION_SPECIFIC_TARGET.
8875         (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
8876         * tree-streamer-out.c (pack_ts_target_option): Remove.
8877         (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
8878         (write_ts_function_decl_tree_pointers): Do not stream
8879         DECL_FUNCTION_SPECIFIC_TARGET.
8880         * tree-streamer-in.c (unpack_ts_target_option): Remove.
8881         (unpack_value_fields): Do not stream TS_TARGET_OPTION.
8882         (lto_input_ts_function_decl_tree_pointers): Do not stream
8883         DECL_FUNCTION_SPECIFIC_TARGET.
8884
8885 2014-02-14  Jakub Jelinek  <jakub@redhat.com>
8886
8887         * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
8888         (get_initial_def_for_induction, vectorizable_induction): Ignore
8889         debug stmts when looking for exit_phi.
8890         (vectorizable_live_operation): Fix up condition.
8891
8892 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
8893
8894         * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
8895         nreverse() because it changes the content of original tree list.
8896
8897 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
8898
8899         * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
8900         * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
8901
8902 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
8903
8904         * config/nds32/nds32.c (nds32_naked_function_p): Follow the
8905         GNU coding standards.
8906
8907 2014-02-13  Jakub Jelinek  <jakub@redhat.com>
8908
8909         PR debug/60152
8910         * dwarf2out.c (gen_subprogram_die): Don't call
8911         add_calling_convention_attribute if subr_die is old_die.
8912
8913 2014-02-13  Sharad Singhai  <singhai@google.com>
8914
8915         * doc/optinfo.texi: Fix order of nodes.
8916
8917 2014-02-13  Uros Bizjak  <ubizjak@gmail.com>
8918
8919         * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
8920         operands[2], not operands[3].
8921
8922 2014-02-13  Richard Biener  <rguenther@suse.de>
8923
8924         PR bootstrap/59878
8925         * doc/install.texi (ISL): Update recommended version to 0.12.2,
8926         mention the possibility of an in-tree build.
8927         (CLooG): Update recommended version to 0.18.1, mention the
8928         possibility of an in-tree build and clarify that the ISL
8929         bundled with CLooG does not work.
8930
8931 2014-02-13  Jakub Jelinek  <jakub@redhat.com>
8932
8933         PR target/43546
8934         * expr.c (compress_float_constant): If x is a hard register,
8935         extend into a pseudo and then move to x.
8936
8937 2014-02-13  Dominik Vogt  <vogt@linux.vnet.ibm.com>
8938
8939         * config/s390/s390.c (s390_asm_output_function_label): Fix crash
8940         caused by bad second argument to warning_at() with -mhotpatch and
8941         nested functions (e.g. with gfortran).
8942
8943 2014-02-13  Richard Sandiford  <rdsandiford@googlemail.com>
8944
8945         * opts.c (option_name): Remove "enabled by default" rider.
8946
8947 2014-02-12  John David Anglin  <danglin@gcc.gnu.org>
8948
8949         * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
8950
8951 2014-02-12  H.J. Lu  <hongjiu.lu@intel.com>
8952             Uros Bizjak  <ubizjak@gmail.com>
8953
8954         PR target/60151
8955         * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
8956         * configure: Regenerated.
8957
8958 2014-02-12  Richard Biener  <rguenther@suse.de>
8959
8960         * vec.c (vec_prefix::calculate_allocation): Move as
8961         inline variant to vec.h.
8962         (vec_prefix::calculate_allocation_1): New out-of-line version.
8963         * vec.h (vec_prefix::calculate_allocation_1): Declare.
8964         (vec_prefix::m_has_auto_buf): Rename to ...
8965         (vec_prefix::m_using_auto_storage): ... this.
8966         (vec_prefix::calculate_allocation): Inline the easy cases
8967         and dispatch to calculate_allocation_1 which doesn't need the
8968         prefix address.
8969         (va_heap::reserve): Use gcc_checking_assert.
8970         (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
8971         m_using_auto_storage.
8972         (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
8973         member and adjust.
8974         (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
8975         (vec<T, va_heap, vl_ptr>::release): Avoid casting.
8976         (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
8977
8978 2014-02-12  Richard Biener  <rguenther@suse.de>
8979
8980         * gcse.c (compute_transp): break from loop over canon_modify_mem_list
8981         when we found a dependence.
8982
8983 2014-02-12  Thomas Schwinge  <thomas@codesourcery.com>
8984
8985         * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
8986         common code...
8987         (maybe_fold_stmt): ... into this new function.
8988         * omp-low.c (lower_omp): Update comment.
8989
8990         * omp-low.c (lower_omp_target): Add clobber for sizes array, after
8991         last use.
8992
8993         * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
8994         dereference.
8995
8996 2014-02-12  James Greenhalgh  <james.greenhalgh@arm.com>
8997
8998         * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
8999         identifiers in comments.
9000         (cortexa53_extra_costs): Likewise.
9001         * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
9002         (cortexa7_extra_costs): Likewise.
9003         (cortexa12_extra_costs): Likewise.
9004         (cortexa15_extra_costs): Likewise.
9005         (v7m_extra_costs): Likewise.
9006
9007 2014-02-12  Richard Biener  <rguenther@suse.de>
9008
9009         PR middle-end/60092
9010         * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
9011         of posix_memalign being successful.
9012         (lower_stmt): Restrict lowering of posix_memalign to when
9013         -ftree-bit-ccp is enabled.
9014
9015 2014-02-12  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
9016
9017         * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
9018         arg_loc.
9019         * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
9020
9021 2014-02-12  Eric Botcazou  <ebotcazou@adacore.com>
9022
9023         PR rtl-optimization/60116
9024         * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
9025         other_insn once the combination has been validated.
9026
9027 2014-02-11  Jan Hubicka  <hubicka@ucw.cz>
9028
9029         PR lto/59468
9030         * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
9031         and wrapper.
9032         * ipa-devirt.c: Include demangle.h
9033         (odr_violation_reported): New static variable.
9034         (add_type_duplicate): Update odr_violations.
9035         (maybe_record_node): Add completep parameter; update it.
9036         (record_target_from_binfo): Add COMPLETEP parameter;
9037         update it as needed.
9038         (possible_polymorphic_call_targets_1): Likewise.
9039         (struct polymorphic_call_target_d): Add nonconstruction_targets;
9040         rename FINAL to COMPLETE.
9041         (record_targets_from_bases): Sanity check we found the binfo;
9042         fix COMPLETEP updating.
9043         (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
9044         parameter, fix computing of COMPLETEP.
9045         (dump_possible_polymorphic_call_targets): Imrove readability of dump;
9046         at LTO time do demangling.
9047         (ipa_devirt): Use nonconstruction_targets; Improve dumps.
9048         * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
9049         parameter.
9050         (gimple_get_virt_method_for_binfo): Likewise.
9051         * gimple-fold.h (gimple_get_virt_method_for_binfo,
9052         gimple_get_virt_method_for_vtable): Update prototypes.
9053
9054 2014-02-11  Vladimir Makarov  <vmakarov@redhat.com>
9055
9056         PR target/49008
9057         * genautomata.c (add_presence_absence): Fix typo with
9058         {final_}presence_list.
9059
9060 2014-02-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
9061
9062         PR target/60137
9063         * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
9064         for VSX/Altivec vectors that land in GPR registers.
9065
9066 2014-02-11  Richard Henderson  <rth@redhat.com>
9067             Jakub Jelinek  <jakub@redhat.com>
9068
9069         PR debug/59776
9070         * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
9071         around drhs if type conversion to lacc->type is not useless.
9072
9073 2014-02-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9074
9075         * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
9076         tuning struct.
9077         (cortex-a57.cortex-a53): Likewise.
9078         * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
9079
9080 2014-02-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9081
9082         * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
9083         arm_restrict_it.
9084
9085 2014-02-11  Renlin Li  <Renlin.Li@arm.com>
9086
9087         * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
9088         add_options_for_arm_vfp3.
9089
9090 2014-02-11  Jeff Law  <law@redhat.com>
9091
9092         PR middle-end/54041
9093         * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
9094         object with an undesirable mode.
9095
9096 2014-02-11  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
9097
9098         PR libgomp/60107
9099         * config/i386/sol2-9.h: New file.
9100         * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
9101         *-*-solaris2.9*): Use it.
9102
9103 2014-02-10  Nagaraju Mekala  <nagaraju.mekala@xilinx.com>
9104
9105         * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
9106         * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
9107
9108 2014-02-10  Nagaraju Mekala  <nagaraju.mekala@xilinx.com>
9109
9110         * config/microblaze/microblaze.c: Extend mcpu version format
9111
9112 2014-02-10  David Holsgrove  <david.holsgrove@xilinx.com>
9113
9114         * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
9115
9116 2014-02-10  Richard Henderson  <rth@redhat.com>
9117
9118         PR target/59927
9119         * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
9120         * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
9121         ms-abi vs -mno-accumulate-outgoing-args.
9122         (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
9123         * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
9124         respect to ms-abi.
9125
9126 2014-02-10  Bernd Edlinger  <bernd.edlinger@hotmail.de>
9127
9128         PR middle-end/60080
9129         * cfgexpand.c (expand_asm_operands): Attach source location to
9130         ASM_INPUT rtx objects.
9131         * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
9132
9133 2014-02-10  Nick Clifton  <nickc@redhat.com>
9134
9135         * config/mn10300/mn10300.c (popcount): New function.
9136         (mn10300_expand_prologue): Include saved registers in stack usage
9137         count.
9138
9139 2014-02-10  Jeff Law  <law@redhat.com>
9140
9141         PR middle-end/52306
9142         * reload1.c (emit_input_reload_insns): Do not create invalid RTL
9143         when changing the SET_DEST of a prior insn to avoid an input reload.
9144
9145 2014-02-10  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
9146
9147         * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
9148         big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
9149         -mcall-openbsd, or -mcall-linux.
9150         (CC1_ENDIAN_BIG_SPEC): Remove.
9151         (CC1_ENDIAN_LITTLE_SPEC): Remove.
9152         (CC1_ENDIAN_DEFAULT_SPEC): Remove.
9153         (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
9154         (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
9155         and %cc1_endian_default.
9156         * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
9157
9158 2014-02-10  Richard Biener  <rguenther@suse.de>
9159
9160         PR tree-optimization/60115
9161         * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
9162         MEM_REF handling.  Properly verify that the accesses are not
9163         out of the objects bound.
9164
9165 2014-02-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9166
9167         * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
9168         coretex to cortex.
9169
9170 2014-02-10  Eric Botcazou  <ebotcazou@adacore.com>
9171
9172         * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
9173         proper constants and fix formatting.
9174         (possible_polymorphic_call_targets): Fix formatting.
9175
9176 2014-02-10  Kirill Yukhin  <kirill.yukhin@intel.com>
9177             Ilya Tocar  <ilya.tocar@intel.com>
9178
9179         * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
9180         (_mm512_loadu_epi32): Renamed into...
9181         (_mm512_loadu_si512): This.
9182         (_mm512_storeu_epi32): Renamed into...
9183         (_mm512_storeu_si512): This.
9184         (_mm512_maskz_ceil_ps): Removed.
9185         (_mm512_maskz_ceil_pd): Ditto.
9186         (_mm512_maskz_floor_ps): Ditto.
9187         (_mm512_maskz_floor_pd): Ditto.
9188         (_mm512_floor_round_ps): Ditto.
9189         (_mm512_floor_round_pd): Ditto.
9190         (_mm512_ceil_round_ps): Ditto.
9191         (_mm512_ceil_round_pd): Ditto.
9192         (_mm512_mask_floor_round_ps): Ditto.
9193         (_mm512_mask_floor_round_pd): Ditto.
9194         (_mm512_mask_ceil_round_ps): Ditto.
9195         (_mm512_mask_ceil_round_pd): Ditto.
9196         (_mm512_maskz_floor_round_ps): Ditto.
9197         (_mm512_maskz_floor_round_pd): Ditto.
9198         (_mm512_maskz_ceil_round_ps): Ditto.
9199         (_mm512_maskz_ceil_round_pd): Ditto.
9200         (_mm512_expand_pd): Ditto.
9201         (_mm512_expand_ps): Ditto.
9202         * config/i386/i386.c (ix86_builtins): Remove
9203         IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
9204         (bdesc_args): Ditto.
9205         * config/i386/predicates.md (const1256_operand): New.
9206         (const_1_to_2_operand): Ditto.
9207         * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
9208         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
9209         (*avx512pf_gatherpf<mode>sf): Ditto.
9210         (avx512pf_gatherpf<mode>df): Ditto.
9211         (*avx512pf_gatherpf<mode>df_mask): Ditto.
9212         (*avx512pf_gatherpf<mode>df): Ditto.
9213         (avx512pf_scatterpf<mode>sf): Ditto.
9214         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
9215         (*avx512pf_scatterpf<mode>sf): Ditto.
9216         (avx512pf_scatterpf<mode>df): Ditto.
9217         (*avx512pf_scatterpf<mode>df_mask): Ditto.
9218         (*avx512pf_scatterpf<mode>df): Ditto.
9219         (avx512f_expand<mode>): Removed.
9220         (<shift_insn><mode>3<mask_name>): Change predicate type.
9221
9222 2014-02-08  Jakub Jelinek  <jakub@redhat.com>
9223
9224         * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
9225         not at the end of datarefs vector use ordered_remove to avoid
9226         reordering datarefs vector.
9227
9228         PR c/59984
9229         * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
9230         mark local addressable non-static vars as GOVD_PRIVATE
9231         instead of GOVD_LOCAL.
9232         * omp-low.c (lower_omp_for): Move gimple_bind_vars
9233         and BLOCK_VARS of gimple_bind_block to new_stmt rather
9234         than copying them.
9235
9236         PR middle-end/60092
9237         * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
9238         if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
9239         assume_aligned or alloc_align attributes.
9240         (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
9241         arguments.  Handle also assume_aligned and alloc_align attributes.
9242         (evaluate_stmt): Adjust bit_value_assume_aligned caller.  Handle
9243         calls to functions with assume_aligned or alloc_align attributes.
9244         * doc/extend.texi: Document assume_aligned and alloc_align attributes.
9245
9246 2014-02-08  Terry Guo  <terry.guo@arm.com>
9247
9248         * doc/invoke.texi: Document ARM -march=armv7e-m.
9249
9250 2014-02-08  Jakub Jelinek  <jakub@redhat.com>
9251
9252         * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
9253         flag on __cilkrts_rethrow builtin.
9254
9255         PR ipa/60026
9256         * ipa-cp.c (determine_versionability): Fail at -O0
9257         or __attribute__((optimize (0))) or -fno-ipa-cp functions.
9258         * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
9259
9260         Revert:
9261         2014-02-04  Jakub Jelinek  <jakub@redhat.com>
9262
9263         PR ipa/60026
9264         * tree-inline.c (copy_forbidden): Fail for
9265         __attribute__((optimize (0))) functions.
9266
9267 2014-02-07  Jan Hubicka  <hubicka@ucw.cz>
9268
9269         * varpool.c: Include pointer-set.h.
9270         (varpool_remove_unreferenced_decls): Variables in other partitions
9271         will not be output; be however careful to not lose information
9272         about partitioning.
9273
9274 2014-02-07  Jan Hubicka  <hubicka@ucw.cz>
9275
9276         * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
9277         lookup in the vtable constructor.
9278
9279 2014-02-07  Jeff Law  <law@redhat.com>
9280
9281         PR target/40977
9282         * config/m68k/m68k.md (ashldi_extsi): Turn into a
9283         define_insn_and_split.
9284
9285         * ipa-inline.c (inline_small_functions): Fix typos.
9286
9287 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
9288
9289         * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
9290         (s390_can_use_return_insn): Declare.
9291         * config/s390/s390.h (EPILOGUE_USES): Define.
9292         * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
9293         instructions.
9294         (s390_chunkify_start): Handle return JUMP_LABELs.
9295         (s390_early_mach): Emit a main_pool instruction on the entry edge.
9296         (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
9297         (s390_can_use_return_insn): New functions.
9298         (s390_fix_long_loop_prediction): Handle conditional returns.
9299         (TARGET_SET_UP_BY_PROLOGUE): Define.
9300         * config/s390/s390.md (ANY_RETURN): New code iterator.
9301         (*creturn, *csimple_return, return, simple_return): New patterns.
9302
9303 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
9304
9305         * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
9306         notes to each restore.  Also add REG_CFA_DEF_CFA when restoring %r15.
9307         (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P.  Update the
9308         REG_CFA_RESTORE list when deciding not to restore a register.
9309
9310 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
9311
9312         * config/s390/s390.c: Include tree-pass.h and context.h.
9313         (s390_early_mach): New function, split out from...
9314         (s390_emit_prologue): ...here.
9315         (pass_data_s390_early_mach): New pass structure.
9316         (pass_s390_early_mach): New class.
9317         (s390_option_override): Create and register early_mach pass.
9318         Move to end of file.
9319
9320 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
9321
9322         * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
9323         to match for the exit block.
9324
9325 2014-02-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9326
9327         * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
9328         ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
9329         Reject misaligned operands.
9330
9331 2014-02-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9332
9333         * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
9334
9335 2014-02-07  Richard Biener  <rguenther@suse.de>
9336
9337         PR middle-end/60092
9338         * gimple-low.c (lower_builtin_posix_memalign): New function.
9339         (lower_stmt): Call it to lower posix_memalign in a way
9340         to make alignment info accessible.
9341
9342 2014-02-07  Jakub Jelinek  <jakub@redhat.com>
9343
9344         PR c++/60082
9345         * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
9346         __builtin_setjmp_receiver.
9347
9348 2014-02-07  Richard Biener  <rguenther@suse.de>
9349
9350         PR middle-end/60092
9351         * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
9352         * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
9353         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
9354         Handle BUILT_IN_POSIX_MEMALIGN.
9355         (find_func_clobbers): Likewise.
9356         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
9357         (call_may_clobber_ref_p_1): Likewise.
9358
9359 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
9360
9361         PR ipa/59918
9362         * ipa-devirt.c (record_target_from_binfo): Remove overactive
9363         sanity check.
9364
9365 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
9366
9367         PR ipa/59469
9368         * lto-cgraph.c (lto_output_node): Use
9369         symtab_get_symbol_partitioning_class.
9370         (lto_output_varpool_node): likewise.
9371         (symtab_get_symbol_partitioning_class): Move here from
9372         lto/lto-partition.c
9373         * cgraph.h (symbol_partitioning_class): Likewise.
9374         (symtab_get_symbol_partitioning_class): Declare.
9375
9376 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
9377
9378         * ggc.h (ggc_internal_cleared_alloc): New macro.
9379         * vec.h (vec_safe_copy): Handle memory stats.
9380         * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
9381         * target-globals.c (save_target_globals): Likewise.
9382
9383 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
9384
9385         PR target/60077
9386         * expr.c (emit_move_resolve_push): Export; be bit more selective
9387         on when to clear alias set.
9388         * expr.h (emit_move_resolve_push): Declare.
9389         * function.h (struct function): Add tail_call_marked.
9390         * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
9391         * config/i386/i386-protos.h (ix86_expand_push): Remove.
9392         * config/i386/i386.md (TImode move expander): De not call
9393         ix86_expand_push.
9394         (FP push expanders): Preserve memory attributes.
9395         * config/i386/sse.md (push<mode>1): Remove.
9396         * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
9397         (ix86_expand_push): Remove.
9398         * config/i386/mmx.md (push<mode>1): Remove.
9399
9400 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
9401
9402         PR rtl-optimization/60030
9403         * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
9404         lopart with paradoxical subreg before shifting it up by hprec.
9405
9406 2014-02-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9407
9408         * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
9409         Remove extra newline at end of file.
9410         * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
9411         (arm_issue_rate): Handle cortexa57.
9412         * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
9413         (cortex-a57.cortex-a53): Likewise.
9414
9415 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
9416
9417         PR target/59575
9418         * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
9419         don't record in REG_FRAME_RELATED_EXPR registers not set in that
9420         bitmask.
9421         (arm_expand_prologue): Adjust all callers.
9422         (arm_unwind_emit_sequence): Allow saved, but not important for unwind
9423         info, registers also at the lowest numbered registers side.  Use
9424         gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
9425         XEXP.
9426
9427         PR debug/59992
9428         * var-tracking.c (adjust_mems): Before adding a SET to
9429         amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
9430
9431 2014-02-06  Alan Modra  <amodra@gmail.com>
9432
9433         PR target/60032
9434         * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
9435         change SDmode to DDmode when lra_in_progress.
9436
9437 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
9438
9439         PR middle-end/59150
9440         * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
9441         free_data_ref on the dr first, and before goto again also set dr
9442         to the next dr.  For simd_lane_access, free old datarefs[i] before
9443         overwriting it.  For get_vectype_for_scalar_type failure, don't
9444         free_data_ref if simd_lane_access.
9445
9446         * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
9447
9448         PR target/60062
9449         * tree.h (opts_for_fn): New inline function.
9450         (opt_for_fn): Define.
9451         * config/i386/i386.c (ix86_function_regparm): Use
9452         opt_for_fn (decl, optimize) instead of optimize.
9453
9454 2014-02-06  Marcus Shawcroft  <marcus.shawcroft@arm.com>
9455
9456         * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
9457         for SYMBOL_REF in large memory model.
9458
9459 2014-02-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9460
9461         * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
9462         and crypto support.
9463         (cortex-a57): Likewise.
9464         (cortex-a57.cortex-a53): Likewise.
9465
9466 2014-02-06  Yury Gribov  <y.gribov@samsung.com>
9467             Kugan Vivekanandarajah  <kuganv@linaro.org>
9468
9469         * config/arm/arm.c (arm_vector_alignment_reachable): Check
9470         unaligned_access.
9471         * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
9472
9473 2014-02-06  Richard Biener  <rguenther@suse.de>
9474
9475         * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
9476         set_loop_copy and initialize_original_copy_tables.
9477
9478 2014-02-06  Alex Velenko  <Alex.Velenko@arm.com>
9479
9480         * config/aarch64/aarch64-simd.md
9481         (aarch64_ashr_simddi): Change QI to SI.
9482
9483 2014-02-05  Jan Hubicka  <hubicka@ucw.cz>
9484             Jakub Jelinek  <jakub@redhat.com>
9485
9486         PR middle-end/60013
9487         * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
9488         of the dataflow.
9489
9490 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9491
9492         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
9493         CODE_FOR_altivec_vpku[hw]um to
9494         CODE_FOR_altivec_vpku[hw]um_direct.
9495         * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
9496         UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
9497         (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
9498         UNSPEC_VUNPACK_LO_SIGN_DIRECT.
9499
9500 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9501
9502         * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
9503         generation for -maltivec=be.
9504         (altivec_vsumsws): Simplify redundant test.
9505
9506 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9507
9508         * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
9509         (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
9510         (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
9511         (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
9512         gen_altivec_vpkuwum.
9513         (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
9514         BYTES_BIG_ENDIAN.
9515         (altivec_vpks<VI_char>ss): Likewise.
9516         (altivec_vpks<VI_char>us): Likewise.
9517         (altivec_vpku<VI_char>us): Likewise.
9518         (altivec_vpku<VI_char>um): Likewise.
9519         (altivec_vpku<VI_char>um_direct): New (copy of
9520         altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
9521         internal use).
9522         (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
9523         target is little endian and -maltivec=be is not specified.
9524         (*altivec_vupkhs<VU_char>_direct): New (copy of
9525         altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
9526         (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
9527         target is little endian and -maltivec=be is not specified.
9528         (*altivec_vupkls<VU_char>_direct): New (copy of
9529         altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
9530         (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
9531         little endian and -maltivec=be is not specified.
9532         (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
9533         little endian and -maltivec=be is not specified.
9534
9535 2014-02-05  Richard Henderson  <rth@redhat.com>
9536
9537         PR debug/52727
9538         * combine-stack-adj.c: Revert r206943.
9539         * sched-int.h (struct deps_desc): Add last_args_size.
9540         * sched-deps.c (init_deps): Initialize it.
9541         (sched_analyze_insn): Add OUTPUT dependencies between insns that
9542         contain REG_ARGS_SIZE notes.
9543
9544 2014-02-05  Jan Hubicka  <hubicka@ucw.cz>
9545
9546         * lto-cgraph.c (asm_nodes_output): Make global.
9547         * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
9548         * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
9549         (driver_handle_option): Handle OPT_fwpa.
9550
9551 2014-02-05  Jakub Jelinek  <jakub@redhat.com>
9552
9553         PR ipa/59947
9554         * ipa-devirt.c (possible_polymorphic_call_targets): Fix
9555         a comment typo and formatting issue.  If odr_hash hasn't been
9556         created, return vNULL and set *completep to false.
9557
9558         PR middle-end/57499
9559         * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
9560         bb with no successors.
9561
9562 2014-02-05  James Greenhalgh  <james.greenhalgh@arm.com>
9563
9564         PR target/59718
9565         * doc/invoke.texi (-march): Clarify documentation for ARM.
9566         (-mtune): Likewise.
9567         (-mcpu): Likewise.
9568
9569 2014-02-05  Richard Biener  <rguenther@suse.de>
9570
9571         * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
9572         when not vectorizing because of too many alias checks.
9573         * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
9574         Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
9575
9576 2014-02-05  Nick Clifton  <nickc@redhat.com>
9577
9578         * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
9579         accept extended registers in any mode when compiling for the MN10300.
9580
9581 2014-02-05  Yury Gribov  <y.gribov@samsung.com>
9582
9583         * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
9584         * ipa-inline.c (report_inline_failed_reason): Handle mismatched
9585         sanitization attributes.
9586         (can_inline_edge_p): Likewise.
9587         (sanitize_attrs_match_for_inline_p): New function.
9588
9589 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
9590
9591         * ipa-prop.c (detect_type_change): Shor circuit testing of
9592         type changes on THIS pointer.
9593
9594 2014-02-04  John David Anglin  <danglin@gcc.gnu.org>
9595
9596         PR target/59777
9597         * config/pa/pa.c (legitimize_tls_address): Return original address
9598         if not passed a SYMBOL_REF rtx.
9599         (hppa_legitimize_address): Call legitimize_tls_address for all TLS
9600         addresses.
9601         (pa_emit_move_sequence): Simplify TLS source operands.
9602         (pa_legitimate_constant_p): Reject all TLS constants.
9603         * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
9604         (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
9605
9606 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
9607
9608         * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
9609         groups when we know they are controlled by LTO.
9610         * varasm.c (default_binds_local_p_1): If object is in other partition,
9611         it will be resolved locally.
9612
9613 2014-02-04  Bernd Edlinger  <bernd.edlinger@hotmail.de>
9614
9615         * config/host-linux.c (linux_gt_pch_use_address): Don't
9616         use SSIZE_MAX because it is not always defined.
9617
9618 2014-02-04  Vladimir Makarov  <vmakarov@redhat.com>
9619
9620         PR bootstrap/59913
9621         * lra-constraints.c (need_for_split_p): Use more 3 reloads as
9622         threshold for pseudo splitting.
9623         (update_ebb_live_info): Process call argument hard registers and
9624         hard registers from insn definition too.
9625         (max_small_class_regs_num): New constant.
9626         (inherit_in_ebb): Update live hard regs through EBBs.  Update
9627         reloads_num only for small register classes.  Don't split for
9628         outputs of jumps.
9629
9630 2014-02-04  Markus Trippelsdorf  <markus@trippelsdorf.de>
9631
9632         PR ipa/60058
9633         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
9634         is non-null.
9635
9636 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
9637
9638         * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
9639         visibility is safe.
9640
9641 2014-02-04  Marek Polacek  <polacek@redhat.com>
9642
9643         * gdbinit.in (pel): Define.
9644
9645 2014-02-04  Bernd Edlinger  <bernd.edlinger@hotmail.de>
9646
9647         * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
9648         behavior.
9649
9650 2014-02-04  Richard Biener  <rguenther@suse.de>
9651
9652         PR lto/59723
9653         * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
9654         in function context local.
9655         (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
9656         * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
9657         similar to LTO_imported_decl_ref.
9658
9659 2014-02-04  Jakub Jelinek  <jakub@redhat.com>
9660
9661         PR tree-optimization/60002
9662         * cgraphclones.c (build_function_decl_skip_args): Clear
9663         DECL_LANG_SPECIFIC.
9664
9665         PR tree-optimization/60023
9666         * tree-if-conv.c (predicate_mem_writes): Pass true instead of
9667         false to gsi_replace.
9668         * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
9669         has been in some EH region and vec_stmt could throw, add
9670         vec_stmt into the same EH region.
9671         * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
9672         has no lhs, ignore it.
9673         * internal-fn.c (expand_MASK_LOAD): Likewise.
9674
9675         PR ipa/60026
9676         * tree-inline.c (copy_forbidden): Fail for
9677         __attribute__((optimize (0))) functions.
9678
9679         PR other/58712
9680         * omp-low.c (simd_clone_struct_copy): If from->inbranch
9681         is set, copy one less argument.
9682         (expand_simd_clones): Don't subtract clone_info->inbranch
9683         from simd_clone_struct_alloc argument.
9684
9685         PR rtl-optimization/57915
9686         * recog.c (simplify_while_replacing): If all unary/binary/relational
9687         operation arguments are constant, attempt to simplify those.
9688
9689         PR middle-end/59261
9690         * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
9691         if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
9692
9693 2014-02-04  Richard Biener  <rguenther@suse.de>
9694
9695         PR tree-optimization/60012
9696         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
9697         TBAA disambiguation to all DDRs.
9698
9699 2014-02-04  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
9700
9701         PR target/59788
9702         * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
9703         (LINK_SPEC): Use it for -shared, -shared-libgcc.
9704
9705 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
9706
9707         PR ipa/59882
9708         * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
9709
9710 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
9711
9712         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
9713         * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
9714
9715 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
9716
9717         PR ipa/59831
9718         * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
9719         to figure out targets of polymorphic calls with known decl.
9720         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
9721         * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
9722         * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
9723         (get_polymorphic_call_info): ... here.
9724         (get_polymorphic_call_info_from_invariant): New function.
9725
9726 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
9727
9728         * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
9729         lookup via vtable pointer; check for type consistency
9730         and turn inconsitent facts into UNREACHABLE.
9731         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
9732         * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
9733         type inconsistent querries; return UNREACHABLE instead.
9734
9735 2014-02-03  Richard Henderson  <rth@twiddle.net>
9736
9737         PR tree-opt/59924
9738         * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
9739         already processed this node.
9740         (normalize_one_pred_1): Pass along mark_set.
9741         (normalize_one_pred): Create and destroy a pointer_set_t.
9742         (normalize_one_pred_chain): Likewise.
9743
9744 2014-02-03  Laurent Aflonsi  <laurent.alfonsi@st.com>
9745
9746         PR gcov-profile/58602
9747         * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
9748
9749 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
9750
9751         PR ipa/59831
9752         * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
9753         -fno-devirtualize; try to devirtualize by the knowledge of
9754         virtual table pointer given by aggregate propagation.
9755         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
9756         (ipa_print_node_jump_functions): Dump also offset that
9757         is relevant for polymorphic calls.
9758         (determine_known_aggregate_parts): Add arg_type parameter; use it
9759         instead of determining the type from pointer type.
9760         (ipa_compute_jump_functions_for_edge): Update call of
9761         determine_known_aggregate_parts.
9762         * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
9763         (gimple_get_virt_method_for_binfo): ... here; simplify using
9764         vtable_pointer_value_to_vtable.
9765         * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
9766         * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
9767         to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
9768         (vtable_pointer_value_to_vtable): Break out from ...; handle also
9769         POINTER_PLUS_EXPR.
9770         (vtable_pointer_value_to_binfo): ... here.
9771         * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
9772
9773 2014-02-03  Teresa Johnson  <tejohnson@google.com>
9774
9775         * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
9776         redef of outer loop index variable.
9777
9778 2014-02-03  Marc Glisse  <marc.glisse@inria.fr>
9779
9780         PR c++/53017
9781         PR c++/59211
9782         * doc/extend.texi (Function Attributes): Typo.
9783
9784 2014-02-03  Cong Hou  <congh@google.com>
9785
9786         PR tree-optimization/60000
9787         * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
9788         if the vectorized statement is a store.  A store statement can only
9789         appear at the end of pattern statements.
9790
9791 2014-02-03  H.J. Lu  <hongjiu.lu@intel.com>
9792
9793         * config/i386/i386.c (flag_opts): Add -mlong-double-128.
9794         (ix86_option_override_internal): Default long double to 64-bit for
9795         32-bit Bionic and to 128-bit for 64-bit Bionic.
9796
9797         * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
9798         TARGET_LONG_DOUBLE_128 is true.
9799         (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
9800
9801         * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
9802         (mlong-double-64): Negate -mlong-double-128.
9803         (mlong-double-128): New option.
9804
9805         * config/i386/i386-c.c (ix86_target_macros): Define
9806         __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
9807
9808         * doc/invoke.texi: Document -mlong-double-128.
9809
9810 2014-02-03  H.J. Lu  <hongjiu.lu@intel.com>
9811
9812         PR rtl-optimization/60024
9813         * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
9814
9815 2014-02-03  Markus Trippelsdorf  <markus@trippelsdorf.de>
9816
9817         * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
9818
9819 2014-02-03  Andrey Belevantsev  <abel@ispras.ru>
9820
9821         PR rtl-optimization/57662
9822         * sel-sched.c (code_motion_path_driver): Do not mark already not
9823         existing blocks in the visiting bitmap.
9824
9825 2014-02-03  Andrey Belevantsev  <abel@ispras.ru>
9826
9827         * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
9828         on the insn being emitted.
9829
9830 2014-02-03  James Greenhalgh  <james.greenhalgh@arm.com>
9831             Will Deacon  <will.deacon@arm.com>
9832
9833         * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
9834
9835 2014-02-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9836
9837         * config/arm/arm-tables.opt: Regenerate.
9838
9839 2014-02-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9840
9841         * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
9842         for vector types other than V16QImode.
9843         * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
9844         define_expand, and call altivec_expand_vec_perm_le when producing
9845         code with little endian element order.
9846         (*altivec_vperm_<mode>_internal): New insn having previous
9847         behavior of altivec_vperm_<mode>.
9848         (altivec_vperm_<mode>_uns): Change to a define_expand, and call
9849         altivec_expand_vec_perm_le when producing code with little endian
9850         element order.
9851         (*altivec_vperm_<mode>_uns_internal): New insn having previous
9852         behavior of altivec_vperm_<mode>_uns.
9853
9854 2014-02-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9855
9856         * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
9857         (altivec_vsumsws): Add handling for -maltivec=be with a little
9858         endian target.
9859         (altivec_vsumsws_direct): New.
9860         (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
9861         gen_altivec_vsumsws.
9862
9863 2014-02-02  Jan Hubicka  <hubicka@ucw.cz>
9864
9865         * ipa-devirt.c (subbinfo_with_vtable_at_offset,
9866         vtable_pointer_value_to_binfo): New functions.
9867         * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
9868         * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
9869
9870 2014-02-02  Sandra Loosemore  <sandra@codesourcery.com>
9871
9872         * config/nios2/nios2.md (load_got_register): Initialize GOT
9873         pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
9874         * config/nios2/nios2.c (nios2_function_profiler): Likewise.
9875
9876 2014-02-02  Jan Hubicka  <hubicka@ucw.cz>
9877
9878         * ipa-prop.c (update_jump_functions_after_inlining): When type is not
9879         preserverd by passthrough, do not propagate the type.
9880
9881 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
9882
9883         * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
9884         (mips_atomic_assign_expand_fenv): New function.
9885         (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
9886
9887 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
9888
9889         * doc/extend.texi (__builtin_mips_get_fcsr): Document.
9890         (__builtin_mips_set_fcsr): Likewise.
9891         * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
9892         MIPS_USI_FTYPE_VOID.
9893         * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
9894         (mips16_expand_set_fcsr): Likewise.
9895         * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
9896         (mips16_set_fcsr_stub): Likewise.
9897         (mips16_get_fcsr_one_only_stub): New class.
9898         (mips16_set_fcsr_one_only_stub): Likewise.
9899         (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
9900         (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
9901         (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
9902         (hard_float): New availability predicate.
9903         (mips_builtins): Add get_fcsr and set_fcsr.
9904         (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
9905         * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
9906         (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
9907         (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
9908         (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
9909         patterns.
9910
9911 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
9912
9913         * config/mips/mips.c (mips_one_only_stub): New class.
9914         (mips_need_mips16_rdhwr_p): Replace with...
9915         (mips16_rdhwr_stub): ...this new variable.
9916         (mips16_stub_call_address): New function.
9917         (mips16_rdhwr_one_only_stub): New class.
9918         (mips_expand_thread_pointer): Use mips16_stub_call_address.
9919         (mips_output_mips16_rdhwr): Delete.
9920         (mips_finish_stub): New function.
9921         (mips_code_end): Use it to handle rdhwr stubs.
9922
9923 2014-02-02  Uros Bizjak  <ubizjak@gmail.com>
9924
9925         PR target/60017
9926         * config/i386/i386.c (classify_argument): Fix handling of bit_offset
9927         when calculating size of integer atomic types.
9928
9929 2014-02-02  H.J. Lu  <hongjiu.lu@intel.com>
9930
9931         * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
9932
9933 2014-02-01  Jakub Jelinek  <jakub@redhat.com>
9934
9935         PR tree-optimization/60003
9936         * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
9937         * profile.c (branch_prob): Use gimple_call_builtin_p
9938         to check for BUILT_IN_SETJMP_RECEIVER.
9939         * tree-inline.c (copy_bb): Call notice_special_calls.
9940
9941 2014-01-31  Vladimir Makarov  <vmakarov@redhat.com>
9942
9943         PR bootstrap/59985
9944         * lra-constraints.c (process_alt_operands): Update reload_sum only
9945         on the first pass.
9946
9947 2014-01-31  Richard Henderson  <rth@redhat.com>
9948
9949         PR middle-end/60004
9950         * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
9951         until after else_eh is processed.
9952
9953 2014-01-31  Ilya Tocar  <ilya.tocar@intel.com>
9954
9955         * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
9956         (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
9957         (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
9958         in smmintrin.h, remove them.
9959         (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
9960         * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
9961         * config/i386/i386.md (ROUND_SAE): Fix value.
9962         * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
9963         (const48_operand): New.
9964         * config/i386/subst.md (round), (round_expand): Use
9965         const_4_or_8_to_11_operand.
9966         (round_saeonly), (round_saeonly_expand): Use const48_operand.
9967
9968 2014-01-31  Ilya Tocar  <ilya.tocar@intel.com>
9969
9970         * config/i386/constraints.md (Yk): Swap meaning with k.
9971         * config/i386/i386.md (movhi_internal): Change Yk to k.
9972         (movqi_internal): Ditto.
9973         (*k<logic><mode>): Ditto.
9974         (*andhi_1): Ditto.
9975         (*andqi_1): Ditto.
9976         (kandn<mode>): Ditto.
9977         (*<code>hi_1): Ditto.
9978         (*<code>qi_1): Ditto.
9979         (kxnor<mode>): Ditto.
9980         (kortestzhi): Ditto.
9981         (kortestchi): Ditto.
9982         (kunpckhi): Ditto.
9983         (*one_cmplhi2_1): Ditto.
9984         (*one_cmplqi2_1): Ditto.
9985         * config/i386/sse.md (): Change k to Yk.
9986         (avx512f_load<mode>_mask): Ditto.
9987         (avx512f_blendm<mode>): Ditto.
9988         (avx512f_store<mode>_mask): Ditto.
9989         (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
9990         (avx512f_storedqu<mode>_mask): Ditto.
9991         (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
9992         Ditto.
9993         (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
9994         (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
9995         (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
9996         (avx512f_maskcmp<mode>3): Ditto.
9997         (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
9998         (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
9999         (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
10000         (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
10001         (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
10002         (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
10003         (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
10004         (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
10005         (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
10006         (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
10007         (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
10008         (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
10009         (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
10010         (vec_extract_lo_<mode>_maskm): Ditto.
10011         (vec_extract_hi_<mode>_maskm): Ditto.
10012         (avx512f_vternlog<mode>_mask): Ditto.
10013         (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
10014         (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
10015         (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
10016         (avx512f_<code>v8div16qi2_mask): Ditto.
10017         (avx512f_<code>v8div16qi2_mask_store): Ditto.
10018         (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
10019         (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
10020         (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
10021         (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
10022         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
10023         (*avx512pf_gatherpf<mode>df_mask): Ditto.
10024         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
10025         (*avx512pf_scatterpf<mode>df_mask): Ditto.
10026         (avx512cd_maskb_vec_dupv8di): Ditto.
10027         (avx512cd_maskw_vec_dupv16si): Ditto.
10028         (avx512f_vpermi2var<mode>3_maskz): Ditto.
10029         (avx512f_vpermi2var<mode>3_mask): Ditto.
10030         (avx512f_vpermi2var<mode>3_mask): Ditto.
10031         (avx512f_vpermt2var<mode>3_maskz): Ditto.
10032         (*avx512f_gathersi<mode>): Ditto.
10033         (*avx512f_gathersi<mode>_2): Ditto.
10034         (*avx512f_gatherdi<mode>): Ditto.
10035         (*avx512f_gatherdi<mode>_2): Ditto.
10036         (*avx512f_scattersi<mode>): Ditto.
10037         (*avx512f_scatterdi<mode>): Ditto.
10038         (avx512f_compress<mode>_mask): Ditto.
10039         (avx512f_compressstore<mode>_mask): Ditto.
10040         (avx512f_expand<mode>_mask): Ditto.
10041         * config/i386/subst.md (mask): Change k to Yk.
10042         (mask_scalar_merge): Ditto.
10043         (sd): Ditto.
10044
10045 2014-01-31  Marc Glisse  <marc.glisse@inria.fr>
10046
10047         * doc/extend.texi (Vector Extensions): Document ?: in C++.
10048
10049 2014-01-31  Richard Biener  <rguenther@suse.de>
10050
10051         PR middle-end/59990
10052         * builtins.c (fold_builtin_memory_op): Make sure to not
10053         use a floating-point mode or a boolean or enumeral type for
10054         the copy operation.
10055
10056 2014-01-30  DJ Delorie  <dj@redhat.com>
10057
10058         * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
10059         * config/msp430/msp430.md (msp430_refsym_need_exit): New.
10060         * config/msp430/msp430.c (msp430_expand_epilogue): Call it
10061         whenever main() has an epilogue.
10062
10063 2014-01-30  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10064
10065         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
10066         unused variable "field".
10067         * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
10068         (vsx_mergeh_<mode>): Likewise.
10069         * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
10070         (altivec_vmrghh): Likewise.
10071         (altivec_vmrghw): Likewise.
10072         (altivec_vmrglb): Likewise.
10073         (altivec_vmrglh): Likewise.
10074         (altivec_vmrglw): Likewise.
10075         (altivec_vspltb): Add missing uses.
10076         (altivec_vsplth): Likewise.
10077         (altivec_vspltw): Likewise.
10078         (altivec_vspltsf): Likewise.
10079
10080 2014-01-30  Jakub Jelinek  <jakub@redhat.com>
10081
10082         PR target/59923
10083         * ifcvt.c (cond_exec_process_insns): Don't conditionalize
10084         frame related instructions.
10085
10086 2014-01-30  Vladimir Makarov  <vmakarov@redhat.com>
10087
10088         PR rtl-optimization/59959
10089         * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
10090         any reload of register whose subreg is invalid.
10091
10092 2014-01-30  Jakub Jelinek  <jakub@redhat.com>
10093
10094         * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
10095         * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
10096         Add missing return type - void.
10097
10098 2014-01-30  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10099
10100         * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
10101         gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
10102         remove element index adjustment for endian (now handled in vsx.md
10103         and altivec.md).
10104         (altivec_expand_vec_perm_const): Use
10105         gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
10106         * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
10107         (vsx_xxspltw_<mode>): Adjust element index for little endian.
10108         * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
10109         define_expand and a new define_insn *altivec_vspltb_internal;
10110         adjust for -maltivec=be on a little endian target.
10111         (altivec_vspltb_direct): New.
10112         (altivec_vsplth): Divide into a define_expand and a new
10113         define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
10114         little endian target.
10115         (altivec_vsplth_direct): New.
10116         (altivec_vspltw): Divide into a define_expand and a new
10117         define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
10118         little endian target.
10119         (altivec_vspltw_direct): New.
10120         (altivec_vspltsf): Divide into a define_expand and a new
10121         define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
10122         a little endian target.
10123
10124 2014-01-30  Richard Biener  <rguenther@suse.de>
10125
10126         PR tree-optimization/59993
10127         * tree-ssa-forwprop.c (associate_pointerplus): Check we
10128         can propagate form the earlier stmt and avoid the transform
10129         when the intermediate result is needed.
10130
10131 2014-01-30  Alangi Derick  <alangiderick@gmail.com>
10132
10133         * README.Portability: Fix typo.
10134
10135 2014-01-30  David Holsgrove  <david.holsgrove@xilinx.com>
10136
10137         * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
10138         comparison_operator with ordered_comparison_operator.
10139
10140 2014-01-30  Nick Clifton  <nickc@redhat.com>
10141
10142         * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
10143         Rename to mn10300_store_multiple_regs.
10144         * config/mn10300/mn10300.c: Likewise.
10145         * config/mn10300/mn10300.md (store_movm): Fix typo: call
10146         store_multiple_regs.
10147         * config/mn10300/predicates.md (mn10300_store_multiple_operation):
10148         Call mn10300_store_multiple_regs.
10149
10150 2014-01-30  Nick Clifton  <nickc@redhat.com>
10151             DJ Delorie  <dj@redhat.com>
10152
10153         * config/rl78/rl78.c (register_sizes): Make the "upper half" of
10154         %fp 2 to keep registers after it properly word-aligned.
10155         (rl78_alloc_physical_registers_umul): Handle the case where both
10156         input operands are the same.
10157
10158 2014-01-30  Richard Biener  <rguenther@suse.de>
10159
10160         PR tree-optimization/59903
10161         * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
10162         check properly.
10163
10164 2014-01-30  Jason Merrill  <jason@redhat.com>
10165
10166         PR c++/59633
10167         * tree.c (walk_type_fields): Handle VECTOR_TYPE.
10168
10169         PR c++/59645
10170         * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
10171
10172 2014-01-30  Richard Biener  <rguenther@suse.de>
10173
10174         PR tree-optimization/59951
10175         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
10176
10177 2014-01-30  Savin Zlobec  <savin.zlobec@gmail.com>
10178
10179         PR target/59784
10180         * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
10181         SFmode to DFmode case.
10182
10183 2014-01-29  DJ Delorie  <dj@redhat.com>
10184
10185         * config/msp430/msp430.opt (-minrt): New.
10186         * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
10187         if -minrt given.
10188         (ENDFILE_SPEC): Likewise.
10189
10190 2014-01-29  Jan Hubicka  <hubicka@ucw.cz>
10191
10192         * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
10193         (estimate_function_body_sizes): Use it.
10194
10195 2014-01-29  Paolo Carlini  <paolo.carlini@oracle.com>
10196
10197         PR c++/58561
10198         * dwarf2out.c (is_cxx_auto): New.
10199         (is_base_type): Use it.
10200         (gen_type_die_with_usage): Likewise.
10201
10202 2014-01-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10203
10204         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const):  Use
10205         CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
10206         * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
10207         -maltivec=be with LE targets.
10208         (vsx_mergeh_<mode>): Likewise.
10209         * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
10210         (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
10211         (altivec_vmrghb): Replace with define_expand and new
10212         *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
10213         (altivec_vmrghb_direct): New define_insn.
10214         (altivec_vmrghh): Replace with define_expand and new
10215         *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
10216         (altivec_vmrghh_direct): New define_insn.
10217         (altivec_vmrghw): Replace with define_expand and new
10218         *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
10219         (altivec_vmrghw_direct): New define_insn.
10220         (*altivec_vmrghsf): Adjust for endianness.
10221         (altivec_vmrglb): Replace with define_expand and new
10222         *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
10223         (altivec_vmrglb_direct): New define_insn.
10224         (altivec_vmrglh): Replace with define_expand and new
10225         *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
10226         (altivec_vmrglh_direct): New define_insn.
10227         (altivec_vmrglw): Replace with define_expand and new
10228         *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
10229         (altivec_vmrglw_direct): New define_insn.
10230         (*altivec_vmrglsf): Adjust for endianness.
10231         (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
10232         (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
10233         (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
10234         (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
10235         (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
10236         (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
10237         (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
10238         (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
10239
10240 2014-01-29  Marcus Shawcroft  <marcus.shawcroft@arm.com>
10241
10242         * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
10243         (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
10244         whitespace.
10245
10246 2014-01-29  Richard Biener  <rguenther@suse.de>
10247
10248         PR tree-optimization/58742
10249         * tree-ssa-forwprop.c (associate_pointerplus): Rename to
10250         associate_pointerplus_align.
10251         (associate_pointerplus_diff): New function.
10252         (associate_pointerplus): Likewise.  Call associate_pointerplus_align
10253         and associate_pointerplus_diff.
10254
10255 2014-01-29  Richard Biener  <rguenther@suse.de>
10256
10257         * lto-streamer.h (LTO_major_version): Bump to 3.
10258         (LTO_minor_version): Reset to 0.
10259
10260 2014-01-29  Renlin Li  <Renlin.Li@arm.com>
10261
10262         * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
10263         * config/arm/arm.c (FL_FOR_ARCH7VE): New.
10264         (arm_file_start): Generate correct asm header for armv7ve.
10265         * config/arm/bpabi.h: Add multilib support for armv7ve.
10266         * config/arm/driver-arm.c: Change the architectures of cortex-a7
10267         and cortex-a15 to armv7ve.
10268         * config/arm/t-aprofile: Add multilib support for armv7ve.
10269         * doc/invoke.texi: Document -march=armv7ve.
10270
10271 2014-01-29  Richard Biener  <rguenther@suse.de>
10272
10273         PR tree-optimization/58742
10274         * tree-ssa-forwprop.c (associate_plusminus): Return true
10275         if we changed sth, defer EH cleanup to ...
10276         (ssa_forward_propagate_and_combine): ... here.  Call simplify_mult.
10277         (simplify_mult): New function.
10278
10279 2014-01-29  Jakub Jelinek  <jakub@redhat.com>
10280
10281         PR middle-end/59917
10282         PR tree-optimization/59920
10283         * tree.c (build_common_builtin_nodes): Remove
10284         __builtin_setjmp_dispatcher initialization.
10285         * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
10286         * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
10287         instead of gsi_after_labels + manually skipping debug stmts.
10288         Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
10289         ignore bbs with IFN_ABNORMAL_DISPATCHER.
10290         * tree-inline.c (copy_edges_for_bb): Remove
10291         can_make_abnormal_goto argument, instead add abnormal_goto_dest
10292         argument.  Ignore computed_goto_p stmts.  Don't call
10293         make_abnormal_goto_edges.  If a call might need abnormal edges
10294         for non-local gotos, see if it already has an edge to
10295         IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
10296         with true argument, don't do anything then, otherwise add
10297         EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
10298         (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
10299         caller.
10300         * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
10301         (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
10302         (lower_stmt): Don't set data->calls_builtin_setjmp.
10303         (lower_builtin_setjmp): Adjust comment.
10304         * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
10305         * tree-cfg.c (found_computed_goto): Remove.
10306         (factor_computed_gotos): Remove.
10307         (make_goto_expr_edges): Return bool, true for computed gotos.
10308         Don't call make_abnormal_goto_edges.
10309         (build_gimple_cfg): Don't set found_computed_goto, don't call
10310         factor_computed_gotos.
10311         (computed_goto_p): No longer static.
10312         (make_blocks): Don't set found_computed_goto.
10313         (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
10314         (make_edges): If make_goto_expr_edges returns true, push bb
10315         into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
10316         instead of calling make_abnormal_goto_edges push bb into ab_edge_call
10317         vector.  Record mapping between bbs and OpenMP regions if there
10318         are any, adjust make_gimple_omp_edges caller.  Call
10319         handle_abnormal_edges.
10320         (make_abnormal_goto_edges): Remove.
10321         * tree-cfg.h (make_abnormal_goto_edges): Remove.
10322         (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
10323         * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
10324         * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
10325         * internal-fn.def (ABNORMAL_DISPATCHER): New.
10326         * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
10327         filling *region also set *region_idx to (*region)->entry->index.
10328
10329         PR other/58712
10330         * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
10331         For REGs set ORIGINAL_REGNO.
10332
10333 2014-01-29  Bingfeng Mei  <bmei@broadcom.com>
10334
10335         * doc/md.texi: Mention that a target shouldn't implement
10336         vec_widen_(s|u)mul_even/odd pair if it is less efficient
10337         than hi/lo pair.
10338
10339 2014-01-29  Jakub Jelinek  <jakub@redhat.com>
10340
10341         PR tree-optimization/59594
10342         * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
10343         a copy of the datarefs vector rather than the vector itself.
10344
10345 2014-01-28  Jason Merrill  <jason@redhat.com>
10346
10347         PR c++/53756
10348         * dwarf2out.c (auto_die): New static.
10349         (gen_type_die_with_usage): Handle C++1y 'auto'.
10350         (gen_subprogram_die): If in-class DIE had 'auto', emit type again
10351         on definition.
10352
10353 2014-01-28  H.J. Lu  <hongjiu.lu@intel.com>
10354
10355         PR target/59672
10356         * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
10357         (SPEC_X32): Likewise.
10358         (SPEC_64): Likewise.
10359         * config/i386/i386.c (ix86_option_override_internal): Turn off
10360         OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
10361         for TARGET_16BIT.
10362         (x86_file_start): Output .code16gcc for TARGET_16BIT.
10363         * config/i386/i386.h (TARGET_16BIT): New macro.
10364         (TARGET_16BIT_P): Likewise.
10365         * config/i386/i386.opt: Add m16.
10366         * doc/invoke.texi: Document -m16.
10367
10368 2014-01-28  Jakub Jelinek  <jakub@redhat.com>
10369
10370         PR preprocessor/59935
10371         * input.c (location_get_source_line): Bail out on when line number
10372         is zero, and test the return value of lookup_or_add_file_to_cache_tab.
10373
10374 2014-01-28  Richard Biener  <rguenther@suse.de>
10375
10376         PR tree-optimization/58742
10377         * tree-ssa-forwprop.c (associate_plusminus): Handle
10378         pointer subtraction of the form (T)(P + A) - (T)P.
10379
10380 2014-01-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10381
10382         * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
10383         at const_int_cost.
10384
10385 2014-01-28  Richard Biener  <rguenther@suse.de>
10386
10387         Revert
10388         2014-01-28  Richard Biener  <rguenther@suse.de>
10389
10390         PR rtl-optimization/45364
10391         PR rtl-optimization/59890
10392         * var-tracking.c (local_get_addr_clear_given_value): Handle
10393         already cleared slot.
10394         (val_reset): Handle not allocated local_get_addr_cache.
10395         (vt_find_locations): Use post-order on the inverted CFG.
10396
10397 2014-01-28  Richard Biener  <rguenther@suse.de>
10398
10399         * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
10400
10401 2014-01-28  Richard Biener  <rguenther@suse.de>
10402
10403         PR rtl-optimization/45364
10404         PR rtl-optimization/59890
10405         * var-tracking.c (local_get_addr_clear_given_value): Handle
10406         already cleared slot.
10407         (val_reset): Handle not allocated local_get_addr_cache.
10408         (vt_find_locations): Use post-order on the inverted CFG.
10409
10410 2014-01-28  Alan Modra  <amodra@gmail.com>
10411
10412         * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
10413         * configure.ac <recursive call for build != host>: Define
10414         GENERATOR_FILE.  Comment.  Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
10415         and LD_FOR_BUILD too.
10416         * configure: Regenerate.
10417
10418 2014-01-27  Allan Sandfeld Jensen  <sandfeld@kde.org>
10419
10420         * config/i386/i386.c (get_builtin_code_for_version): Separate
10421         Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
10422         Broadwell from Haswell.
10423
10424 2014-01-27  Steve Ellcey  <sellcey@mips.com>
10425
10426         * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
10427         Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
10428         * config/mips/mips.c (mips_option_override): Change setting
10429         of TARGET_DSP.
10430         * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
10431         * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
10432         Change from Mask to Var.
10433
10434 2014-01-27  Jeff Law  <law@redhat.com>
10435
10436         * ipa-inline.c (inline_small_functions): Fix typo.
10437
10438 2014-01-27  Ilya Tocar  <ilya.tocar@intel.com>
10439
10440         * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
10441         (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
10442         (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
10443         (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
10444         (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
10445         (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
10446         (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
10447         (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
10448         (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
10449         (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
10450         (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
10451         (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
10452         (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
10453         (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
10454         (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
10455         (_mm512_storeu_epi64): Ditto.
10456         (_mm512_cmpge_epi32_mask): Ditto.
10457         (_mm512_cmpge_epu32_mask): Ditto.
10458         (_mm512_cmpge_epi64_mask): Ditto.
10459         (_mm512_cmpge_epu64_mask): Ditto.
10460         (_mm512_cmple_epi32_mask): Ditto.
10461         (_mm512_cmple_epu32_mask): Ditto.
10462         (_mm512_cmple_epi64_mask): Ditto.
10463         (_mm512_cmple_epu64_mask): Ditto.
10464         (_mm512_cmplt_epi32_mask): Ditto.
10465         (_mm512_cmplt_epu32_mask): Ditto.
10466         (_mm512_cmplt_epi64_mask): Ditto.
10467         (_mm512_cmplt_epu64_mask): Ditto.
10468         (_mm512_cmpneq_epi32_mask): Ditto.
10469         (_mm512_cmpneq_epu32_mask): Ditto.
10470         (_mm512_cmpneq_epi64_mask): Ditto.
10471         (_mm512_cmpneq_epu64_mask): Ditto.
10472         (_mm512_expand_pd): Ditto.
10473         (_mm512_expand_ps): Ditto.
10474         * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
10475         VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
10476         VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
10477         * config/i386/i386.c (ix86_builtins): Add
10478         IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
10479         IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
10480         IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
10481         IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
10482         IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
10483         IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
10484         IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
10485         IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
10486         IX86_BUILTIN_PMOVUSQW512_MEM.
10487         (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
10488         __builtin_ia32_pmovsqd512mem_mask,
10489         __builtin_ia32_pmovqd512mem_mask,
10490         __builtin_ia32_pmovusqw512mem_mask,
10491         __builtin_ia32_pmovsqw512mem_mask,
10492         __builtin_ia32_pmovqw512mem_mask,
10493         __builtin_ia32_pmovusdw512mem_mask,
10494         __builtin_ia32_pmovsdw512mem_mask,
10495         __builtin_ia32_pmovdw512mem_mask,
10496         __builtin_ia32_pmovqb512mem_mask,
10497         __builtin_ia32_pmovusqb512mem_mask,
10498         __builtin_ia32_pmovsqb512mem_mask,
10499         __builtin_ia32_pmovusdb512mem_mask,
10500         __builtin_ia32_pmovsdb512mem_mask,
10501         __builtin_ia32_pmovdb512mem_mask.
10502         (bdesc_args): Add __builtin_ia32_expanddf512,
10503         __builtin_ia32_expandsf512.
10504         (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
10505         VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
10506         VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
10507         * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
10508         (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
10509         (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
10510         (avx512f_<code>v8div16qi2_mask_store): This.
10511         (avx512f_expand<mode>): New.
10512
10513 2014-01-27  Kirill Yukhin  <kirill.yukhin@intel.com>
10514
10515         * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
10516         New.
10517         (_mm512_mask_prefetch_i64gather_pd): Ditto.
10518         (_mm512_prefetch_i32scatter_pd): Ditto.
10519         (_mm512_mask_prefetch_i32scatter_pd): Ditto.
10520         (_mm512_prefetch_i64scatter_pd): Ditto.
10521         (_mm512_mask_prefetch_i64scatter_pd): Ditto.
10522         (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
10523         (_mm512_mask_prefetch_i64gather_ps): Ditto.
10524         (_mm512_prefetch_i32scatter_ps): Ditto.
10525         (_mm512_mask_prefetch_i32scatter_ps): Ditto.
10526         (_mm512_prefetch_i64scatter_ps): Ditto.
10527         (_mm512_mask_prefetch_i64scatter_ps): Ditto.
10528         * config/i386/i386-builtin-types.def: Define
10529         VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
10530         and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
10531         * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
10532         IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
10533         IX86_BUILTIN_SCATTERPFQPD.
10534         (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
10535         __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
10536         __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
10537         __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
10538         __builtin_ia32_scatterpfqps.
10539         (ix86_expand_builtin): Expand new built-ins.
10540         * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
10541         fix memory access data type.
10542         (*avx512pf_gatherpf<mode>_mask): Ditto.
10543         (*avx512pf_gatherpf<mode>): Ditto.
10544         (avx512pf_scatterpf<mode>): Ditto.
10545         (*avx512pf_scatterpf<mode>_mask): Ditto.
10546         (*avx512pf_scatterpf<mode>): Ditto.
10547         (GATHER_SCATTER_SF_MEM_MODE): New.
10548         (avx512pf_gatherpf<mode>df): Ditto.
10549         (*avx512pf_gatherpf<mode>df_mask): Ditto.
10550         (*avx512pf_scatterpf<mode>df): Ditto.
10551
10552 2014-01-27  Jakub Jelinek  <jakub@redhat.com>
10553
10554         PR bootstrap/59934
10555         * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
10556         know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
10557         reached.
10558
10559 2014-01-27  James Greenhalgh  <james.greenhalgh@arm.com>
10560
10561         * common/config/arm/arm-common.c
10562         (arm_rewrite_mcpu): Handle multiple names.
10563         * config/arm/arm.h
10564         (BIG_LITTLE_SPEC): Do not discard mcpu switches.
10565
10566 2014-01-27  James Greenhalgh  <james.greenhalgh@arm.com>
10567
10568         * gimple-builder.h (create_gimple_tmp): Delete.
10569
10570 2014-01-27  Christian Bruel  <christian.bruel@st.com>
10571
10572         * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
10573         words comparisons.
10574
10575 2014-01-26  John David Anglin  <danglin@gcc.gnu.org>
10576
10577         * config/pa/pa.md (call): Generate indirect long calls to non-local
10578         functions when outputing 32-bit code.
10579         (call_value): Likewise except for special call to buggy powf function.
10580
10581         * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
10582         portable runtime and PIC indirect calls.
10583         (pa_output_indirect_call): Remove unnecessary nop from portable runtime
10584         and PIC call sequences.  Use ldo instead of blr to set return register
10585         in PIC call sequence.
10586
10587 2014-01-25  Walter Lee  <walt@tilera.com>
10588
10589         * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
10590         avoid clobbering a live register.
10591
10592 2014-01-25  Walter Lee  <walt@tilera.com>
10593
10594         * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
10595         Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
10596         * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
10597         Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
10598
10599 2014-01-25  Walter Lee  <walt@tilera.com>
10600
10601         * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
10602         arguments on even registers.
10603         (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
10604         STACK_BOUNDARY.
10605         * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
10606         (BIGGEST_ALIGNMENT): Ditto.
10607         (BIGGEST_FIELD_ALIGNMENT): Ditto.
10608
10609 2014-01-25  Walter Lee  <walt@tilera.com>
10610
10611         * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
10612         insns before bundling.
10613         * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
10614
10615 2014-01-25  Walter Lee  <walt@tilera.com>
10616
10617         * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
10618         PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
10619         * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
10620
10621 2014-01-25  Richard Sandiford  <rdsandiford@googlemail.com>
10622
10623         * config/mips/constraints.md (kl): Delete.
10624         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
10625         define expands, using...
10626         (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
10627         instructions for MIPS16.
10628         (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
10629         non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
10630
10631 2014-01-25  Walter Lee  <walt@tilera.com>
10632
10633         * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
10634         (clzdi2): Ditto.
10635         (ffsdi2): Ditto.
10636
10637 2014-01-25  Walter Lee  <walt@tilera.com>
10638
10639         * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
10640         (TARGET_EXPAND_TO_RTL_HOOK): Define.
10641
10642 2014-01-25  Richard Sandiford  <rdsandiford@googlemail.com>
10643
10644         * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
10645         Handle XOR.
10646
10647 2014-01-25  Jakub Jelinek  <jakub@redhat.com>
10648
10649         * print-rtl.c (in_call_function_usage): New var.
10650         (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
10651         EXPR_LIST mode as mode and not as reg note name.
10652
10653         PR middle-end/59561
10654         * cfgloopmanip.c (copy_loop_info): If
10655         loop->warned_aggressive_loop_optimizations, make sure
10656         the flag is set in target loop too.
10657
10658 2014-01-24  Balaji V. Iyer  <balaji.v.iyer@intel.com>
10659
10660         * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
10661         flag_cilkplus.
10662         * builtins.def: Likewise.
10663         * cilk.h (fn_contains_cilk_spawn_p): Likewise.
10664         * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
10665         * ira.c (ira_setup_eliminable_regset): Likewise.
10666         * omp-low.c (gate_expand_omp): Likewise.
10667         (execute_lower_omp): Likewise.
10668         (diagnose_sb_0): Likewise.
10669         (gate_diagnose_omp_blocks): Likewise.
10670         (simd_clone_clauses_extract): Likewise.
10671         (gate): Likewise.
10672
10673 2014-01-24  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10674
10675         * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
10676         correction for little endian...
10677         * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
10678         here.
10679
10680 2014-01-24  Jeff Law  <law@redhat.com>
10681
10682         PR tree-optimization/59919
10683         * tree-vrp.c (find_assert_locations_1): Do not register asserts
10684         for non-returning calls.
10685
10686 2014-01-24  James Greenhalgh  <james.greenhalgh@arm.com>
10687
10688         * common/config/aarch64/aarch64-common.c
10689         (aarch64_rewrite_mcpu): Handle multiple names.
10690         * config/aarch64/aarch64.h
10691         (BIG_LITTLE_SPEC): Do not discard mcpu switches.
10692
10693 2014-01-24  Dodji Seketeli  <dodji@redhat.com>
10694
10695         * input.c (add_file_to_cache_tab): Handle the case where fopen
10696         returns NULL.
10697
10698 2014-01-23  H.J. Lu  <hongjiu.lu@intel.com>
10699
10700         PR target/59929
10701         * config/i386/i386.md (pushsf splitter): Get stack adjustment
10702         from push operand if code of push isn't PRE_DEC.
10703
10704 2014-01-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
10705
10706         PR target/59909
10707         * doc/invoke.texi (RS/6000 and PowerPC Options): Document
10708         -mquad-memory-atomic.  Update -mquad-memory documentation to say
10709         it is only used for non-atomic loads/stores.
10710
10711         * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
10712         -mquad-memory or -mquad-memory-atomic switches.
10713
10714         * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
10715         -mquad-memory-atomic to ISA 2.07 support.
10716
10717         * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
10718         to separate support of normal quad word memory operations (ldq, stq)
10719         from the atomic quad word memory operations.
10720
10721         * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
10722         support to separate non-atomic quad word operations from atomic
10723         quad word operations.  Disable non-atomic quad word operations in
10724         little endian mode so that we don't have to swap words after the
10725         load and before the store.
10726         (quad_load_store_p): Add comment about atomic quad word support.
10727         (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
10728         options printed with -mdebug=reg.
10729
10730         * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
10731         -mquad-memory-atomic as the test for whether we have quad word
10732         atomic instructions.
10733         (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
10734         or -mp8-vector are used, allow byte/half-word atomic operations.
10735
10736         * config/rs6000/sync.md (load_lockedti): Insure that the address
10737         is a proper indexed or indirect address for the lqarx instruction.
10738         On little endian systems, swap the hi/lo registers after the lqarx
10739         instruction.
10740         (load_lockedpti): Use indexed_or_indirect_operand predicate to
10741         insure the address is valid for the lqarx instruction.
10742         (store_conditionalti): Insure that the address is a proper indexed
10743         or indirect address for the stqcrx. instruction.  On little endian
10744         systems, swap the hi/lo registers before doing the stqcrx.
10745         instruction.
10746         (store_conditionalpti): Use indexed_or_indirect_operand predicate to
10747         insure the address is valid for the stqcrx. instruction.
10748
10749         * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
10750         Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
10751         type of quad memory support is available.
10752
10753 2014-01-23  Vladimir Makarov  <vmakarov@redhat.com>
10754
10755         PR regression/59915
10756         * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
10757         there is a danger of looping.
10758
10759 2014-01-23  Pat Haugen  <pthaugen@us.ibm.com>
10760
10761         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
10762         force flag_ira_loop_pressure if set via command line.
10763
10764 2014-01-23  Alex Velenko  <Alex.Velenko@arm.com>
10765
10766         * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
10767         (ashr_simd): New builtin handling DI mode.
10768         * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
10769         (aarch64_sshr_simddi): New match pattern.
10770         * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
10771         (vshrd_n_s64): Likewise.
10772         * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
10773
10774 2014-01-23  Nick Clifton  <nickc@redhat.com>
10775
10776         * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
10777         (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
10778         favour of mcu specific scripts.
10779         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
10780         430x multilibs.
10781
10782 2014-01-23  James Greenhalgh  <james.greenhalgh@arm.com>
10783             Alex Velenko  <Alex.Velenko@arm.com>
10784
10785         * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
10786         (vaddv_s16): Likewise.
10787         (vaddv_s32): Likewise.
10788         (vaddv_u8): Likewise.
10789         (vaddv_u16): Likewise.
10790         (vaddv_u32): Likewise.
10791         (vaddvq_s8): Likewise.
10792         (vaddvq_s16): Likewise.
10793         (vaddvq_s32): Likewise.
10794         (vaddvq_s64): Likewise.
10795         (vaddvq_u8): Likewise.
10796         (vaddvq_u16): Likewise.
10797         (vaddvq_u32): Likewise.
10798         (vaddvq_u64): Likewise.
10799         (vaddv_f32): Likewise.
10800         (vaddvq_f32): Likewise.
10801         (vaddvq_f64): Likewise.
10802         (vmaxv_f32): Likewise.
10803         (vmaxv_s8): Likewise.
10804         (vmaxv_s16): Likewise.
10805         (vmaxv_s32): Likewise.
10806         (vmaxv_u8): Likewise.
10807         (vmaxv_u16): Likewise.
10808         (vmaxv_u32): Likewise.
10809         (vmaxvq_f32): Likewise.
10810         (vmaxvq_f64): Likewise.
10811         (vmaxvq_s8): Likewise.
10812         (vmaxvq_s16): Likewise.
10813         (vmaxvq_s32): Likewise.
10814         (vmaxvq_u8): Likewise.
10815         (vmaxvq_u16): Likewise.
10816         (vmaxvq_u32): Likewise.
10817         (vmaxnmv_f32): Likewise.
10818         (vmaxnmvq_f32): Likewise.
10819         (vmaxnmvq_f64): Likewise.
10820         (vminv_f32): Likewise.
10821         (vminv_s8): Likewise.
10822         (vminv_s16): Likewise.
10823         (vminv_s32): Likewise.
10824         (vminv_u8): Likewise.
10825         (vminv_u16): Likewise.
10826         (vminv_u32): Likewise.
10827         (vminvq_f32): Likewise.
10828         (vminvq_f64): Likewise.
10829         (vminvq_s8): Likewise.
10830         (vminvq_s16): Likewise.
10831         (vminvq_s32): Likewise.
10832         (vminvq_u8): Likewise.
10833         (vminvq_u16): Likewise.
10834         (vminvq_u32): Likewise.
10835         (vminnmv_f32): Likewise.
10836         (vminnmvq_f32): Likewise.
10837         (vminnmvq_f64): Likewise.
10838
10839 2014-01-23  James Greenhalgh  <james.greenhalgh@arm.com>
10840
10841         * config/aarch64/aarch64-simd.md
10842         (aarch64_dup_lane<mode>): Correct lane number on big-endian.
10843         (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
10844         (*aarch64_mul3_elt<mode>): Likewise.
10845         (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
10846         (*aarch64_mul3_elt_to_64v2df): Likewise.
10847         (*aarch64_mla_elt<mode>): Likewise.
10848         (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
10849         (*aarch64_mls_elt<mode>): Likewise.
10850         (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
10851         (*aarch64_fma4_elt<mode>): Likewise.
10852         (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
10853         (*aarch64_fma4_elt_to_64v2df): Likewise.
10854         (*aarch64_fnma4_elt<mode>): Likewise.
10855         (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
10856         (*aarch64_fnma4_elt_to_64v2df): Likewise.
10857         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
10858         (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
10859         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
10860         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
10861         (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
10862         (aarch64_sqdmull_lane<mode>_internal): Likewise.
10863         (aarch64_sqdmull2_lane<mode>_internal): Likewise.
10864
10865 2013-01-23  Alex Velenko  <Alex.Velenko@arm.com>
10866
10867         * config/aarch64/aarch64-simd.md
10868         (aarch64_be_checked_get_lane<mode>): New define_expand.
10869         * config/aarch64/aarch64-simd-builtins.def
10870         (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
10871         New builtin definition.
10872         * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
10873         Use new safe be builtin.
10874
10875 2014-01-23  Alex Velenko  <Alex.Velenko@arm.com>
10876
10877         * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
10878         New define_insn.
10879         (aarch64_be_st1<mode>): Likewise.
10880         (aarch_ld1<VALL:mode>): Define_expand modified.
10881         (aarch_st1<VALL:mode>): Likewise.
10882         * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
10883         (UNSPEC_ST1): Likewise.
10884
10885 2014-01-23  David Holsgrove  <david.holsgrove@xilinx.com>
10886
10887         * config/microblaze/microblaze.md: Add trap insn and attribute
10888
10889 2014-01-23  Dodji Seketeli  <dodji@redhat.com>
10890
10891         PR preprocessor/58580
10892         * input.h (location_get_source_line): Take an additional line_size
10893         parameter.
10894         (void diagnostics_file_cache_fini): Declare new function.
10895         * input.c (struct fcache): New type.
10896         (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
10897         New static constants.
10898         (diagnostic_file_cache_init, total_lines_num)
10899         (lookup_file_in_cache_tab, evicted_cache_tab_entry)
10900         (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
10901         (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
10902         (get_next_line, read_next_line, goto_next_line, read_line_num):
10903         New static function definitions.
10904         (diagnostic_file_cache_fini): New function.
10905         (location_get_source_line): Take an additional output line_len
10906         parameter.  Re-write using lookup_or_add_file_to_cache_tab and
10907         read_line_num.
10908         * diagnostic.c (diagnostic_finish): Call
10909         diagnostic_file_cache_fini.
10910         (adjust_line): Take an additional input parameter for the length
10911         of the line, rather than calculating it with strlen.
10912         (diagnostic_show_locus): Adjust the use of
10913         location_get_source_line and adjust_line with respect to their new
10914         signature.  While displaying a line now, do not stop at the first
10915         null byte.  Rather, display the zero byte as a space and keep
10916         going until we reach the size of the line.
10917         * Makefile.in: Add vec.o to OBJS-libcommon
10918
10919 2014-01-23  Kirill Yukhin  <kirill.yukhin@intel.com>
10920             Ilya Tocar     <ilya.tocar@intel.com>
10921
10922         * config/i386/avx512fintrin.h (_mm512_kmov): New.
10923         * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
10924         (__builtin_ia32_kmov16): Ditto.
10925         * config/i386/i386.md (UNSPEC_KMOV): New.
10926         (kmovw): Ditto.
10927
10928 2014-01-23  Kirill Yukhin  <kirill.yukhin@intel.com>
10929
10930         * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
10931         (_mm512_storeu_si512): Ditto.
10932
10933 2014-01-23  Richard Sandiford  <rdsandiford@googlemail.com>
10934
10935         PR target/52125
10936         * rtl.h (get_referenced_operands): Declare.
10937         * recog.c (get_referenced_operands): New function.
10938         * config/mips/mips.c (mips_reorg_process_insns): Check which asm
10939         operands have been referenced when recording LO_SUM references.
10940
10941 2014-01-22  David Holsgrove  <david.holsgrove@xilinx.com>
10942
10943         * config/microblaze/microblaze.md: Correct bswaphi2 insn.
10944
10945 2014-01-22  Jan Hubicka  <hubicka@ucw.cz>
10946
10947         * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
10948         Enable for generic and recent AMD targets.
10949
10950 2014-01-22  Jan Hubicka  <hubicka@ucw.cz>
10951
10952         * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
10953         ARG_SIZE note when adjustment was eliminated.
10954
10955 2014-01-22  Jeff Law  <law@redhat.com>
10956
10957         PR tree-optimization/59597
10958         * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
10959         in file.  Accept new argument REGISTERING and use it to modify
10960         dump output appropriately.
10961         (register_jump_thread): Corresponding changes.
10962         (mark_threaded_blocks): Reinstate code to cancel unprofitable
10963         thread paths involving joiner blocks.  Add code to dump cancelled
10964         jump threading paths.
10965
10966 2014-01-22  Vladimir Makarov  <vmakarov@redhat.com>
10967
10968         PR rtl-optimization/59477
10969         * lra-constraints.c (inherit_in_ebb): Process call for living hard
10970         regs.  Update reloads_num and potential_reload_hard_regs for all insns.
10971
10972 2014-01-22  Tom Tromey  <tromey@redhat.com>
10973
10974         * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
10975         PARAMS.
10976         * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
10977
10978 2014-01-21  Vladimir Makarov  <vmakarov@redhat.com>
10979
10980         PR rtl-optimization/59896
10981         * lra-constraints.c (process_alt_operands): Check unused note for
10982         matched operands of insn with no output reloads.
10983
10984 2014-01-21  Richard Sandiford  <rdsandiford@googlemail.com>
10985
10986         * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
10987         (mips_move_from_gpr_cost): Likewise.
10988
10989 2014-01-21  Vladimir Makarov  <vmakarov@redhat.com>
10990
10991         PR rtl-optimization/59858
10992         * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
10993         ira_class_hard_regs_num.
10994         (process_alt_operands): Increase reject for dying matched operand.
10995
10996 2014-01-21  Jakub Jelinek  <jakub@redhat.com>
10997
10998         PR target/59003
10999         * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
11000         smaller than size, perform several stores or loads and stores
11001         at dst + count - size to store or copy all of size bytes, rather
11002         than just last modesize bytes.
11003
11004 2014-01-20  DJ Delorie  <dj@redhat.com>
11005
11006         * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
11007         that CLOBBERs are REGs before propogating their values.
11008
11009 2014-01-20  H.J. Lu  <hongjiu.lu@intel.com>
11010
11011         PR middle-end/59789
11012         * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
11013         (cgraph_inline_failed_type): New function.
11014         * cgraph.h (DEFCIFCODE): Add type.
11015         (cgraph_inline_failed_type_t): New enum.
11016         (cgraph_inline_failed_type): New prototype.
11017         * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
11018         FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
11019         FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
11020         LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
11021         MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
11022         RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
11023         OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
11024         INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
11025         Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
11026         FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
11027         EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
11028         OPTIMIZATION_MISMATCH.
11029         * tree-inline.c (expand_call_inline): Emit errors during
11030         early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
11031
11032 2014-01-20  Uros Bizjak  <ubizjak@gmail.com>
11033
11034         PR target/59685
11035         * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
11036         mode attribute in insn output.
11037
11038 2014-01-20  Eric Botcazou  <ebotcazou@adacore.com>
11039
11040         * output.h (output_constant): Delete.
11041         * varasm.c (output_constant): Make private.
11042
11043 2014-01-20  Alex Velenko  <Alex.Velenko@arm.com>
11044
11045         * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
11046
11047 2014-01-20  Jakub Jelinek  <jakub@redhat.com>
11048
11049         PR middle-end/59860
11050         * tree.h (fold_builtin_strcat): New prototype.
11051         * builtins.c (fold_builtin_strcat): No longer static.  Add len
11052         argument, if non-NULL, don't call c_strlen.  Optimize
11053         directly into __builtin_memcpy instead of __builtin_strcpy.
11054         (fold_builtin_2): Adjust fold_builtin_strcat caller.
11055         * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
11056
11057 2014-01-20  Uros Bizjak  <ubizjak@gmail.com>
11058
11059         * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
11060         for SImode_address_operand operands, having only a REG argument.
11061
11062 2014-01-20  Marcus Shawcroft  <marcus.shawcroft@arm.com>
11063
11064         * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
11065         loader name using mbig-endian.
11066         (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
11067
11068 2014-01-20  James Greenhalgh  <james.greenhalgh@arm.com>
11069
11070         * doc/invoke.texi (-march): Clarify documentation for AArch64.
11071         (-mtune): Likewise.
11072         (-mcpu): Likewise.
11073
11074 2014-01-20  Tejas Belagod  <tejas.belagod@arm.com>
11075
11076         * config/aarch64/aarch64-protos.h
11077         (aarch64_cannot_change_mode_class_ptr): Declare.
11078         * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
11079         aarch64_cannot_change_mode_class_ptr): New.
11080         * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
11081         backend hook aarch64_cannot_change_mode_class.
11082
11083 2014-01-20  James Greenhalgh  <james.greenhalgh@arm.com>
11084
11085         * common/config/aarch64/aarch64-common.c
11086         (aarch64_handle_option): Don't handle any option order logic here.
11087         * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
11088         selected_cpu, warn on architecture version mismatch.
11089         (aarch64_override_options): Fix parsing order for option strings.
11090
11091 2014-01-20  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
11092             Iain Sandoe  <iain@codesourcery.com>
11093
11094         PR bootstrap/59496
11095         * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
11096         warning.  Amend comment to reflect current functionality.
11097
11098 2014-01-20  Richard Biener  <rguenther@suse.de>
11099
11100         PR middle-end/59860
11101         * builtins.c (fold_builtin_strcat): Remove case better handled
11102         by tree-ssa-strlen.c.
11103
11104 2014-01-20  Alan Lawrence  <alan.lawrence@arm.com>
11105
11106         * config/aarch64/aarch64.opt
11107         (mcpu, march, mtune): Make case-insensitive.
11108
11109 2014-01-20  Jakub Jelinek  <jakub@redhat.com>
11110
11111         PR target/59880
11112         * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
11113         if operands[1] is a REG or ZERO_EXTEND of a REG.
11114
11115 2014-01-19  Jan Hubicka  <hubicka@ucw.cz>
11116
11117         * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
11118
11119 2014-01-19  John David Anglin  <danglin@gcc.gnu.org>
11120
11121         * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
11122         long non-pic millicode calls.
11123
11124 2014-01-19  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
11125
11126         * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
11127
11128 2014-01-19  Kito Cheng  <kito@0xlab.org>
11129
11130         * builtins.c (expand_movstr): Check movstr expand done or fail.
11131
11132 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
11133             H.J. Lu  <hongjiu.lu@intel.com>
11134
11135         PR target/59379
11136         * config/i386/i386.md (*lea<mode>): Zero-extend return register
11137         to DImode for zero-extended addresses.
11138
11139 2014-01-19  Jakub Jelinek  <jakub@redhat.com>
11140
11141         PR rtl-optimization/57763
11142         * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
11143         on the new indirect jump_insn and increment LABEL_NUSES (label).
11144
11145 2014-01-18  H.J. Lu  <hongjiu.lu@intel.com>
11146
11147         PR bootstrap/59580
11148         PR bootstrap/59583
11149         * config.gcc (x86_archs): New variable.
11150         (x86_64_archs): Likewise.
11151         (x86_cpus): Likewise.
11152         Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
11153         --with-arch/--with-cpu= options.
11154         Support --with-arch=/--with-cpu={nehalem,westmere,
11155         sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
11156
11157 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
11158
11159         * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
11160         and PROCESSOR_ATHLON to simplify code.  Move "memory" calculation.
11161
11162 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
11163
11164         * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
11165
11166 2014-01-18  Jakub Jelinek  <jakub@redhat.com>
11167
11168         PR target/58944
11169         * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
11170         clear cpp_get_options (parse_in)->warn_unused_macros for
11171         ix86_target_macros_internal with cpp_define.
11172
11173 2014-01-18  Richard Sandiford  <rdsandiford@googlemail.com>
11174
11175         * jump.c (delete_related_insns): Keep (use (insn))s.
11176         * reorg.c (redundant_insn): Check for barriers too.
11177
11178 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
11179
11180         * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
11181
11182 2014-01-17  John David Anglin  <danglin@gcc.gnu.org>
11183
11184         * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
11185         call to $$dyncall when TARGET_LONG_CALLS is true.
11186
11187 2014-01-17  Jeff Law  <law@redhat.com>
11188
11189         * ree.c (combine_set_extension): Temporarily disable test for
11190         changing number of hard registers.
11191
11192 2014-01-17  Jan Hubicka  <hubicka@ucw.cz>
11193
11194         PR middle-end/58125
11195         * ipa-inline-analysis.c (inline_free_summary):
11196         Do not free summary of aliases.
11197
11198 2014-01-17  Jakub Jelinek  <jakub@redhat.com>
11199
11200         PR middle-end/59706
11201         * gimplify.c (gimplify_expr): Use create_tmp_var
11202         instead of create_tmp_var_raw.  If cond doesn't have
11203         integral type, don't add the IFN_ANNOTATE builtin at all.
11204
11205 2014-01-17  Martin Jambor  <mjambor@suse.cz>
11206
11207         PR ipa/59736
11208         * ipa-cp.c (prev_edge_clone): New variable.
11209         (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
11210         Also resize prev_edge_clone vector.
11211         (ipcp_edge_duplication_hook): Also update prev_edge_clone.
11212         (ipcp_edge_removal_hook): New function.
11213         (ipcp_driver): Register ipcp_edge_removal_hook.
11214
11215 2014-01-17  Andrew Pinski  <apinski@cavium.com>
11216             Steve Ellcey  <sellcey@mips.com>
11217
11218         PR target/59462
11219         * config/mips/mips.c (mips_print_operand): Check operand mode instead
11220         of operator mode.
11221
11222 2014-01-17  Jeff Law  <law@redhat.com>
11223
11224         PR middle-end/57904
11225         * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
11226         so that pass_ccp runs first.
11227
11228 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
11229
11230         * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
11231         (ix86_adjust_cost): Use !TARGET_XXX.
11232         (do_reorder_for_imul): Likewise.
11233         (swap_top_of_ready_list): Likewise.
11234         (ix86_sched_reorder): Likewise.
11235
11236 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
11237
11238         * config/i386/i386-c.c (ix86_target_macros_internal): Handle
11239         PROCESSOR_INTEL.  Treat like PROCESSOR_GENERIC.
11240         * config/i386/i386.c (intel_memcpy): New.  Duplicate slm_memcpy.
11241         (intel_memset): New.  Duplicate slm_memset.
11242         (intel_cost): New.  Duplicate slm_cost.
11243         (m_INTEL): New macro.
11244         (processor_target_table): Add "intel".
11245         (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
11246         with PROCESSOR_INTEL for "intel".
11247         (ix86_lea_outperforms): Support PROCESSOR_INTEL.  Duplicate
11248         PROCESSOR_SILVERMONT.
11249         (ix86_issue_rate): Likewise.
11250         (ix86_adjust_cost): Likewise.
11251         (ia32_multipass_dfa_lookahead): Likewise.
11252         (swap_top_of_ready_list): Likewise.
11253         (ix86_sched_reorder): Likewise.
11254         (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
11255         instead of TARGET_OPT_AGU.
11256         * config/i386/i386.h (TARGET_INTEL): New.
11257         (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
11258         (processor_type): Add PROCESSOR_INTEL.
11259         * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
11260         Add X86_TUNE_AVOID_LEA_FOR_ADDR.
11261
11262 2014-01-17  Marek Polacek  <polacek@redhat.com>
11263
11264         PR c/58346
11265         * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
11266         size is zero.
11267
11268 2014-01-17  Richard Biener  <rguenther@suse.de>
11269
11270         PR tree-optimization/46590
11271         * opts.c (default_options_table): Add entries for
11272         OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
11273         all enabled at -O1 but not for -Og.
11274         * common.opt (fbranch-count-reg): Remove Init(1).
11275         (fmove-loop-invariants): Likewise.
11276         (ftree-pta): Likewise.
11277
11278 2014-01-17  Jakub Jelinek  <jakub@redhat.com>
11279
11280         * config/i386/i386.c (ix86_data_alignment): For compatibility with
11281         (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
11282         decls to at least the GCC 4.8 used alignments.
11283
11284         PR fortran/59440
11285         * tree-nested.c (convert_nonlocal_reference_stmt,
11286         convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
11287         of GIMPLE_BIND stmts, adjust associated decls.
11288
11289 2014-01-17  Richard Biener  <rguenther@suse.de>
11290
11291         PR tree-optimization/46590
11292         * vec.h (vec<>::bseach): New member function implementing
11293         binary search according to C89 bsearch.
11294         (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
11295         * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
11296         bitmap pointer again.  Make accesses_in_loop a flat array.
11297         (mem_ref_obstack): New global.
11298         (outermost_indep_loop): Adjust for mem_ref->stored changes.
11299         (mark_ref_stored): Likewise.
11300         (ref_indep_loop_p_2): Likewise.
11301         (set_ref_stored_in_loop): New helper function.
11302         (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
11303         (memref_free): Adjust.
11304         (record_mem_ref_loc): Simplify.
11305         (gather_mem_refs_stmt): Adjust.
11306         (sort_locs_in_loop_postorder_cmp): New function.
11307         (analyze_memory_references): Sort accesses_in_loop after
11308         loop postorder number.
11309         (find_ref_loc_in_loop_cmp): New function.
11310         (for_all_locs_in_loop): Find relevant cluster of locs in
11311         accesses_in_loop and iterate without recursion.
11312         (execute_sm): Avoid uninit warning.
11313         (struct ref_always_accessed): Simplify.
11314         (ref_always_accessed::operator ()): Likewise.
11315         (ref_always_accessed_p): Likewise.
11316         (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
11317         loop postorder numbers here.
11318         (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
11319         numbers.
11320
11321 2014-01-17  Jan Hubicka  <hubicka@ucw.cz>
11322
11323         PR c++/57945
11324         * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
11325         on decls for which assemble_alias has been called.
11326
11327 2014-01-17  Nick Clifton  <nickc@redhat.com>
11328
11329         * config/msp430/msp430.opt: (mcpu): New option.
11330         * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
11331         (msp430_option_override): Parse target_cpu.  If the MCU name
11332         matches a generic string, clear target_mcu.
11333         (msp430_attr): Allow numeric interrupt values up to 63.
11334         (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
11335         * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
11336         option.
11337         * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
11338         Add mcpu matches.
11339         * config/msp430/msp430.md (popm): Use %J rather than %I.
11340         (addsi3): Use msp430_nonimmediate_operand for operand 2.
11341         (addhi_cy_i): Use immediate_operand for operand 2.
11342         * doc/invoke.texi: Document -mcpu option.
11343
11344 2014-01-17  Richard Biener  <rguenther@suse.de>
11345
11346         PR rtl-optimization/38518
11347         * df.h (df_analyze_loop): Declare.
11348         * df-core.c: Include cfgloop.h.
11349         (df_analyze_1): Split out main part of df_analyze.
11350         (df_analyze): Adjust.
11351         (loop_inverted_post_order_compute): New function.
11352         (loop_post_order_compute): Likewise.
11353         (df_analyze_loop): New function avoiding whole-function
11354         postorder computes.
11355         * loop-invariant.c (find_defs): Use df_analyze_loop.
11356         (find_invariants): Adjust.
11357         * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
11358
11359 2014-01-17  Zhenqiang Chen  <zhenqiang.chen@arm.com>
11360
11361         * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
11362         (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
11363
11364 2014-01-16  Ilya Enkovich  <ilya.enkovich@intel.com>
11365
11366         * ipa-ref.c (ipa_remove_stmt_references): Fix references
11367         traversal when removing references.
11368
11369 2014-01-16  Jan Hubicka  <hubicka@ucw.cz>
11370
11371         PR ipa/59775
11372         * tree.c (get_binfo_at_offset): Look harder for virtual bases.
11373
11374 2014-01-16  Bernd Schmidt  <bernds@codesourcery.com>
11375
11376         PR middle-end/56791
11377         * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
11378         pushing a reload for an autoinc when we had previously reloaded an
11379         inner part of the address.
11380
11381 2014-01-16  Jakub Jelinek  <jakub@redhat.com>
11382
11383         * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
11384         field.
11385         (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
11386         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
11387         when not giving up or versioning for alias only because of
11388         loop->safelen.
11389         (vect_analyze_data_ref_dependences): Set to true.
11390         * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
11391         is a GIMPLE_PHI.
11392         (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
11393         LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
11394         to the condition.
11395
11396         PR middle-end/58344
11397         * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
11398
11399         PR target/59839
11400         * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
11401         operand 0 predicate for gathers, use a new pseudo as subtarget.
11402
11403 2014-01-16  Vladimir Makarov  <vmakarov@redhat.com>
11404
11405         PR middle-end/59609
11406         * lra-constraints.c (process_alt_operands): Add printing debug info.
11407         Check absence of input/output reloads for matched operands too.
11408
11409 2014-01-16  Vladimir Makarov  <vmakarov@redhat.com>
11410
11411         PR rtl-optimization/59835
11412         * ira.c (ira_init_register_move_cost): Increase cost for
11413         impossible modes.
11414
11415 2014-01-16  Alan Lawrence  <alan.lawrence@arm.com>
11416
11417         * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
11418
11419 2014-01-16  Richard Earnshaw  <rearnsha@arm.com>
11420
11421         PR target/59780
11422         * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
11423         non-register objects.  Use gen_(high/low)part more consistently.
11424         Fix assertions.
11425
11426 2014-01-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
11427
11428         PR target/59844
11429         * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
11430         endian support, remove tests for WORDS_BIG_ENDIAN.
11431         (p8_mfvsrd_3_<mode>): Likewise.
11432         (reload_gpr_from_vsx<mode>): Likewise.
11433         (reload_gpr_from_vsxsf): Likewise.
11434         (p8_mfvsrd_4_disf): Likewise.
11435
11436 2014-01-16  Richard Biener  <rguenther@suse.de>
11437
11438         PR rtl-optimization/46590
11439         * lcm.c (compute_antinout_edge): Use postorder iteration.
11440         (compute_laterin): Use inverted postorder iteration.
11441
11442 2014-01-16  Nick Clifton  <nickc@redhat.com>
11443
11444         PR middle-end/28865
11445         * varasm.c (output_constant): Return the number of bytes actually
11446         emitted.
11447         (output_constructor_array_range): Update the field size with the
11448         number of bytes emitted by output_constant.
11449         (output_constructor_regular_field): Likewise.  Also do not
11450         complain if the total number of bytes emitted is now greater
11451         than the expected fieldpos.
11452         * output.h (output_constant): Update prototype and descriptive comment.
11453
11454 2014-01-16  Marek Polacek  <polacek@redhat.com>
11455
11456         PR middle-end/59827
11457         * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
11458         it is error_mark_node.
11459
11460 2014-01-15  Uros Bizjak  <ubizjak@gmail.com>
11461
11462         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
11463         VALID_AVX256_REG_OR_OI_MODE.
11464
11465 2014-01-15  Pat Haugen  <pthaugen@us.ibm.com>
11466
11467         * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
11468         current procedure should be profiled.
11469
11470 2014-01-15  Andrew Pinski  <apinski@cavium.com>
11471
11472         * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
11473         of moving from/to the STACK_REG register class.
11474
11475 2014-01-15  Richard Henderson  <rth@redhat.com>
11476
11477         PR debug/54694
11478         * reginfo.c (global_regs_decl): Globalize.
11479         * rtl.h (global_regs_decl): Declare.
11480         * ira.c (do_reload): Diagnose frame_pointer_needed and it
11481         reserved via global_regs.
11482
11483 2014-01-15  Teresa Johnson  <tejohnson@google.com>
11484
11485         * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
11486
11487 2014-01-15  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11488
11489         * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
11490         and vmulosh rather than call gen_vec_widen_smult_*.
11491         (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
11492         than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
11493         (vec_widen_smult_even_v16qi): Likewise.
11494         (vec_widen_umult_even_v8hi): Likewise.
11495         (vec_widen_smult_even_v8hi): Likewise.
11496         (vec_widen_umult_odd_v16qi): Likewise.
11497         (vec_widen_smult_odd_v16qi): Likewise.
11498         (vec_widen_umult_odd_v8hi): Likewise.
11499         (vec_widen_smult_odd_v8hi): Likewise.
11500         (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
11501         vmuloub rather than call gen_vec_widen_umult_*.
11502         (vec_widen_umult_lo_v16qi): Likewise.
11503         (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
11504         vmulosb rather than call gen_vec_widen_smult_*.
11505         (vec_widen_smult_lo_v16qi): Likewise.
11506         (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
11507         rather than call gen_vec_widen_umult_*.
11508         (vec_widen_umult_lo_v8hi): Likewise.
11509         (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
11510         rather than call gen_vec_widen_smult_*.
11511         (vec_widen_smult_lo_v8hi): Likewise.
11512
11513 2014-01-15  Jeff Law  <law@redhat.com>
11514
11515         PR tree-optimization/59747
11516         * ree.c (find_and_remove_re): Properly handle case where a second
11517         eliminated extension requires widening a copy created for elimination
11518         of a prior extension.
11519         (combine_set_extension): Ensure that the number of hard regs needed
11520         for a destination register does not change when we widen it.
11521
11522 2014-01-15  Sebastian Huber  <sebastian.huber@embedded-brains.de>
11523
11524         * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
11525         (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
11526         (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
11527         (arm*-*-rtems*): Use t-rtems from existing tmake_file.
11528         (avr-*-rtems*): Likewise.
11529         (bfin*-rtems*): Likewise.
11530         (moxie-*-rtems*): Likewise.
11531         (h8300-*-rtems*): Likewise.
11532         (i[34567]86-*-rtems*): Likewise.
11533         (lm32-*-rtems*): Likewise.
11534         (m32r-*-rtems*): Likewise.
11535         (m68k-*-rtems*): Likewise.
11536         (microblaze*-*-rtems*): Likewise.
11537         (mips*-*-rtems*): Likewise.
11538         (powerpc-*-rtems*): Likewise.
11539         (sh-*-rtems*): Likewise.
11540         (sparc-*-rtems*): Likewise.
11541         (sparc64-*-rtems*): Likewise.
11542         (v850-*-rtems*): Likewise.
11543         (m32c-*-rtems*): Likewise.
11544
11545 2014-01-15  Vladimir Makarov  <vmakarov@redhat.com>
11546
11547         PR rtl-optimization/59511
11548         * ira.c (ira_init_register_move_cost): Use memory costs for some
11549         cases of register move cost calculations.
11550         * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
11551         instead of BB frequency.
11552         * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
11553         * lra-assigns.c (find_hard_regno_for): Ditto.
11554
11555 2014-01-15  Richard Biener  <rguenther@suse.de>
11556
11557         PR tree-optimization/59822
11558         * tree-vect-stmts.c (hoist_defs_of_uses): New function.
11559         (vectorizable_load): Use it to hoist defs of uses of invariant
11560         loads out of the loop.
11561
11562 2014-01-15  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
11563             Kugan Vivekanandarajah  <kuganv@linaro.org>
11564
11565         PR target/59695
11566         * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
11567         truncation.
11568
11569 2014-01-15  Richard Biener  <rguenther@suse.de>
11570
11571         PR rtl-optimization/59802
11572         * lcm.c (compute_available): Use inverted postorder to seed
11573         the initial worklist.
11574
11575 2014-01-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
11576
11577         PR target/59803
11578         * config/s390/s390.c (s390_preferred_reload_class): Don't return
11579         ADDR_REGS for invalid symrefs in non-PIC code.
11580
11581 2014-01-15  Jakub Jelinek  <jakub@redhat.com>
11582
11583         PR other/58712
11584         * builtins.c (determine_block_size): Initialize *probable_max_size
11585         even if len_rtx is CONST_INT.
11586
11587 2014-01-14  Andrew Pinski  <apinski@cavium.com>
11588
11589         * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
11590         * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
11591         (cortexa53_tunings): Likewise.
11592         (aarch64_sched_issue_rate): New function.
11593         (TARGET_SCHED_ISSUE_RATE): Define.
11594
11595 2014-01-14  Vladimir Makarov  <vmakarov@redhat.com>
11596
11597         * ira-costs.c (find_costs_and_classes): Add missed
11598         ira_init_register_move_cost_if_necessary.
11599
11600 2014-01-14  Vladimir Makarov  <vmakarov@redhat.com>
11601
11602         PR target/59787
11603         * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
11604
11605 2014-01-14  H.J. Lu  <hongjiu.lu@intel.com>
11606
11607         PR target/59794
11608         * config/i386/i386.c (type_natural_mode): Add a bool parameter
11609         to indicate if type is used for function return value.  Warn ABI
11610         change if the vector mode isn't available for function return value.
11611         (ix86_function_arg_advance): Pass false to type_natural_mode.
11612         (ix86_function_arg): Likewise.
11613         (ix86_gimplify_va_arg): Likewise.
11614         (function_arg_32): Don't warn ABI change.
11615         (ix86_function_value): Pass true to type_natural_mode.
11616         (ix86_return_in_memory): Likewise.
11617         (ix86_struct_value_rtx): Removed.
11618         (TARGET_STRUCT_VALUE_RTX): Likewise.
11619
11620 2014-01-14  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
11621
11622         * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
11623         converting a conditional jump into a conditional return.
11624
11625 2014-01-14  Richard Biener  <rguenther@suse.de>
11626
11627         PR tree-optimization/58921
11628         PR tree-optimization/59006
11629         * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
11630         hoisting invariant stmts.
11631         * tree-vect-stmts.c (vectorizable_load): Insert the splat of
11632         invariant loads on the preheader edge if possible.
11633
11634 2014-01-14  Joey Ye  <joey.ye@arm.com>
11635
11636         * doc/plugin.texi (Building GCC plugins): Update to C++.
11637
11638 2014-01-14  Kirill Yukhin  <kirill.yukhin@intel.com>
11639
11640         * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
11641         (_mm_rcp28_round_ss): Ditto.
11642         (_mm_rsqrt28_round_sd): Ditto.
11643         (_mm_rsqrt28_round_ss): Ditto.
11644         (_mm_rcp28_sd): Ditto.
11645         (_mm_rcp28_ss): Ditto.
11646         (_mm_rsqrt28_sd): Ditto.
11647         (_mm_rsqrt28_ss): Ditto.
11648         * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
11649         * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
11650         * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
11651         (IX86_BUILTIN_RCP28SD): Ditto.
11652         (IX86_BUILTIN_RCP28SS): Ditto.
11653         (IX86_BUILTIN_RSQRT28SD): Ditto.
11654         (IX86_BUILTIN_RSQRT28SS): Ditto.
11655         (bdesc_special_args): Define __builtin_ia32_movntdqa512,
11656         __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
11657         __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
11658         (ix86_expand_special_args_builtin): Expand new FTYPE.
11659         * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
11660         (srcp14<mode>): Make insn unary.
11661         (avx512f_vmscalef<mode><round_name>): Use substed predicate.
11662         (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
11663         (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
11664         (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
11665         (avx512er_exp2<mode><mask_name><round_saeonly_name>):
11666         Fix rounding: make it SAE only.
11667         (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
11668         Ditto.
11669         (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
11670         Ditto.
11671         (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
11672         (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
11673         (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
11674         * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
11675         (round_saeonly_mask_scalar_operand4): Ditto.
11676         (round_saeonly_mask_scalar_op3): Ditto.
11677         (round_saeonly_mask_scalar_op4): Ditto.
11678
11679 2014-01-13  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11680
11681         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11682         Implement -maltivec=be for vec_insert and vec_extract.
11683
11684 2014-01-10  DJ Delorie  <dj@redhat.com>
11685
11686         * config/msp430/msp430.md (call_internal): Don't allow memory
11687         references with SP as the base register.
11688         (call_value_internal): Likewise.
11689         * config/msp430/constraints.md (Yc): New.  For memory references
11690         that don't use SP as a base register.
11691
11692         * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
11693         "an integer without a # prefix"
11694         * config/msp430/msp430.md (epilogue_helper): Use it.
11695
11696 2014-01-13  Jakub Jelinek  <jakub@redhat.com>
11697
11698         PR target/59617
11699         * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
11700         AVX512F gather builtins.
11701         * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
11702         on gather decls with INTEGER_TYPE masktype.
11703         (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
11704         directly into the builtin rather than hoisting it before loop.
11705
11706         PR tree-optimization/59387
11707         * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
11708         (scev_const_prop): If folded_casts and type has undefined overflow,
11709         use force_gimple_operand instead of force_gimple_operand_gsi and
11710         for each added stmt if it is assign with
11711         arith_code_with_undefined_signed_overflow, call
11712         rewrite_to_defined_overflow.
11713         * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
11714         gimple-fold.h instead.
11715         (arith_code_with_undefined_signed_overflow,
11716         rewrite_to_defined_overflow): Moved to ...
11717         * gimple-fold.c (arith_code_with_undefined_signed_overflow,
11718         rewrite_to_defined_overflow): ... here.  No longer static.
11719         Include gimplify-me.h.
11720         * gimple-fold.h (arith_code_with_undefined_signed_overflow,
11721         rewrite_to_defined_overflow): New prototypes.
11722
11723 2014-01-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11724
11725         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
11726
11727 2014-01-13  Eric Botcazou  <ebotcazou@adacore.com>
11728
11729         * builtins.c (get_object_alignment_2): Minor tweak.
11730         * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
11731
11732 2014-01-13  Christian Bruel  <christian.bruel@st.com>
11733
11734         * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
11735         optimized non constant lengths.
11736
11737 2014-01-13  Jakub Jelinek  <jakub@redhat.com>
11738
11739         PR libgomp/59194
11740         * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
11741         load as __atomic_load_N if possible.
11742
11743 2014-01-11  David Edelsohn  <dje.gcc@gmail.com>
11744
11745         * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
11746         target parameter.
11747         (rs6000_expand_builtin): Adjust call.
11748
11749 2014-01-11  David Edelsohn  <dje.gcc@gmail.com>
11750
11751         PR target/58115
11752         * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
11753         * config/rs6000/rs6000.c: Include target-globals.h.
11754         (rs6000_set_current_function): Instead of doing target_reinit
11755         unconditionally, use save_target_globals_default_opts and
11756         restore_target_globals.
11757
11758         * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
11759         FPSCR.
11760         * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
11761         (rs6000_expand_builtin): Handle mffs and mtfsf.
11762         (rs6000_init_builtins): Define mffs and mtfsf.
11763         * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
11764         (rs6000_mffs): New pattern.
11765         (rs6000_mtfsf): New pattern.
11766
11767 2014-01-11  Bin Cheng  <bin.cheng@arm.com>
11768
11769         * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
11770         Start narrowing with START.  Apply candidate-use pair
11771         and check overall cost in narrowing.
11772         (iv_ca_prune): Pass new argument.
11773
11774 2014-01-10  Jeff Law  <law@redhat.com>
11775
11776         PR middle-end/59743
11777         * ree.c (combine_reaching_defs): Ensure the defining statement
11778         occurs before the extension when optimizing extensions with
11779         different source and destination hard registers.
11780
11781 2014-01-10  Jan Hubicka  <hubicka@ucw.cz>
11782
11783         PR ipa/58585
11784         * ipa-devirt.c (build_type_inheritance_graph): Also add types of
11785         vtables into the type inheritance graph.
11786
11787 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
11788
11789         PR rtl-optimization/59754
11790         * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
11791         modes in the REGNO != REGNO case.
11792
11793 2014-01-10  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11794
11795         * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
11796
11797 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
11798
11799         PR tree-optimization/59745
11800         * tree-predcom.c (tree_predictive_commoning_loop): Call
11801         free_affine_expand_cache if giving up because components is NULL.
11802
11803         * target-globals.c (save_target_globals): Allocate < 4KB structs using
11804         GC in payload of target_globals struct instead of allocating them on
11805         the heap and the larger structs separately using GC.
11806         * target-globals.h (struct target_globals): Make regs, hard_regs,
11807         reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
11808         of GTY((skip)) and change type to void *.
11809         (reset_target_globals): Cast loads from those fields to corresponding
11810         types.
11811
11812 2014-01-10  Steve Ellcey  <sellcey@mips.com>
11813
11814         PR plugins/59335
11815         * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
11816         gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
11817         tree-ssanames.h, print-tree.h, varasm.h, and context.h.
11818
11819 2014-01-10  Richard Earnshaw  <rearnsha@arm.com>
11820
11821         PR target/59744
11822         * aarch64-modes.def (CC_Zmode): New flags mode.
11823         * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
11824         represents an equality.
11825         (aarch64_get_condition_code): Handle CC_Zmode.
11826         * aarch64.md (compare_neg<mode>): Restrict to equality operations.
11827
11828 2014-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
11829
11830         * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
11831         extraction in good case.
11832
11833 2014-01-10  Richard Biener  <rguenther@suse.de>
11834
11835         PR tree-optimization/59374
11836         * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
11837         checking after SLP discovery.  Mark stmts not participating
11838         in any SLP instance properly.
11839
11840 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11841
11842         * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
11843         when handling a SET rtx.
11844
11845 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11846
11847         * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
11848         (cortex-a57): Likewise.
11849         (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
11850
11851 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11852
11853         * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
11854         non-iwmmxt builtins.
11855
11856 2014-01-10  Jan Hubicka  <hubicka@ucw.cz>
11857
11858         PR ipa/58252
11859         PR ipa/59226
11860         * ipa-devirt.c record_target_from_binfo): Take as argument
11861         stack of binfos and lookup matching one for virtual inheritance.
11862         (possible_polymorphic_call_targets_1): Update.
11863
11864 2014-01-10  Huacai Chen  <chenhc@lemote.com>
11865
11866         * config/mips/driver-native.c (host_detect_local_cpu): Handle new
11867         kernel strings for Loongson-2E/2F/3A.
11868
11869 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
11870
11871         PR middle-end/59670
11872         * tree-vect-data-refs.c (vect_analyze_data_refs): Check
11873         is_gimple_call before calling gimple_call_internal_p.
11874
11875 2014-01-09  Steve Ellcey  <sellcey@mips.com>
11876
11877         * Makefile.in (TREE_FLOW_H): Remove.
11878         (TREE_SSA_H): Add file names from tree-flow.h.
11879         * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
11880         * tree.h: Remove tree-flow.h reference.
11881         * hash-table.h: Remove tree-flow.h reference.
11882         * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
11883         reference with tree-ssa-loop.h.
11884
11885 2014-01-09  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11886
11887         * doc/invoke.texi: Add -maltivec={be,le} options, and document
11888         default element-order behavior for -maltivec.
11889         * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
11890         * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
11891         that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
11892         when targeting big endian, at least for now.
11893         * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
11894
11895 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
11896
11897         PR middle-end/47735
11898         * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
11899         var satisfies use_register_for_decl, just take into account type
11900         alignment, rather than decl alignment.
11901
11902         PR tree-optimization/59622
11903         * gimple-fold.c (gimple_fold_call): Fix a typo in message.  For
11904         __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
11905         __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
11906         Don't devirtualize for inplace at all.  For targets.length () == 1,
11907         if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
11908
11909 2014-01-09  H.J. Lu  <hongjiu.lu@intel.com>
11910
11911         * config/i386/i386.md (cpu): Remove the unused btver1.
11912
11913 2014-01-09  H.J. Lu  <hongjiu.lu@intel.com>
11914
11915         * gdbasan.in: Put a breakpoint on __sanitizer::Report.
11916
11917 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
11918
11919         PR target/58115
11920         * tree-core.h (struct target_globals): New forward declaration.
11921         (struct tree_target_option): Add globals field.
11922         * tree.h (TREE_TARGET_GLOBALS): Define.
11923         (prepare_target_option_nodes_for_pch): New prototype.
11924         * target-globals.h (struct target_globals): Define even if
11925         !SWITCHABLE_TARGET.
11926         * tree.c (prepare_target_option_node_for_pch,
11927         prepare_target_option_nodes_for_pch): New functions.
11928         * config/i386/i386.h (SWITCHABLE_TARGET): Define.
11929         * config/i386/i386.c: Include target-globals.h.
11930         (ix86_set_current_function): Instead of doing target_reinit
11931         unconditionally, use save_target_globals_default_opts and
11932         restore_target_globals.
11933
11934 2014-01-09  Richard Biener  <rguenther@suse.de>
11935
11936         PR tree-optimization/59715
11937         * tree-cfg.h (split_critical_edges): Declare.
11938         * tree-cfg.c (split_critical_edges): Export.
11939         * tree-ssa-sink.c (execute_sink_code): Split critical edges.
11940
11941 2014-01-09  Max Ostapenko  <m.ostapenko@partner.samsung.com>
11942
11943         * cfgexpand.c (expand_stack_vars): Optionally disable
11944         asan stack protection.
11945         (expand_used_vars): Likewise.
11946         (partition_stack_vars): Likewise.
11947         * asan.c (asan_emit_stack_protection): Optionally disable
11948         after return stack usage.
11949         (instrument_derefs): Optionally disable memory access instrumentation.
11950         (instrument_builtin_call): Likewise.
11951         (instrument_strlen_call): Likewise.
11952         (asan_protect_global): Optionally disable global variables protection.
11953         * doc/invoke.texi: Added doc for new options.
11954         * params.def: Added new options.
11955         * params.h: Likewise.
11956
11957 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
11958
11959         PR rtl-optimization/59724
11960         * ifcvt.c (cond_exec_process_if_block): Don't call
11961         flow_find_head_matching_sequence with 0 longest_match.
11962         * cfgcleanup.c (flow_find_head_matching_sequence): Count even
11963         non-active insns if !stop_after.
11964         (try_head_merge_bb): Revert 2014-01-07 changes.
11965
11966 2014-01-08  Jeff Law  <law@redhat.com>
11967
11968         * ree.c (get_sub_rtx): New function, extracted from...
11969         (merge_def_and_ext): Here.
11970         (combine_reaching_defs): Use get_sub_rtx.
11971
11972 2014-01-08  Eric Botcazou  <ebotcazou@adacore.com>
11973
11974         * cgraph.h (varpool_variable_node): Do not choke on null node.
11975
11976 2014-01-08  Catherine Moore  <clm@codesourcery.com>
11977
11978         * config/mips/mips.md (simple_return): Attempt to use JRC
11979         for microMIPS.
11980         * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
11981
11982 2014-01-08  Richard Sandiford  <rdsandiford@googlemail.com>
11983
11984         PR rtl-optimization/59137
11985         * reorg.c (steal_delay_list_from_target): Call update_block for
11986         elided insns.
11987         (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
11988
11989 2014-01-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11990
11991         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
11992         two duplicate entries.
11993
11994 2014-01-08  Richard Sandiford  <rdsandiford@googlemail.com>
11995
11996         Revert:
11997         2012-10-07  Richard Sandiford  <rdsandiford@googlemail.com>
11998
11999         * config/mips/mips.c (mips_truncated_op_cost): New function.
12000         (mips_rtx_costs): Adjust test for BADDU.
12001         * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
12002
12003         2012-10-02  Richard Sandiford  <rdsandiford@googlemail.com>
12004
12005         * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
12006         (*baddu_si): ...this new pattern.
12007
12008 2014-01-08  Jakub Jelinek  <jakub@redhat.com>
12009
12010         PR ipa/59722
12011         * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
12012
12013 2014-01-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
12014
12015         PR middle-end/57748
12016         * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
12017         inner_reference_p.
12018         (expand_expr, expand_normal): Adjust.
12019         * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
12020         inner_reference_p. Use inner_reference_p to expand inner references.
12021         (store_expr): Adjust.
12022         * cfgexpand.c (expand_call_stmt): Adjust.
12023
12024 2014-01-08  Rong Xu  <xur@google.com>
12025
12026         * gcov-io.c (gcov_var): Move from gcov-io.h.
12027         (gcov_position): Ditto.
12028         (gcov_is_error): Ditto.
12029         (gcov_rewrite): Ditto.
12030         * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
12031         only part to libgcc/libgcov.h.
12032
12033 2014-01-08  Marek Polacek  <polacek@redhat.com>
12034
12035         PR middle-end/59669
12036         * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
12037
12038 2014-01-08  Marek Polacek  <polacek@redhat.com>
12039
12040         PR sanitizer/59667
12041         * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
12042
12043 2014-01-08  Jakub Jelinek  <jakub@redhat.com>
12044
12045         PR rtl-optimization/59649
12046         * stor-layout.c (get_mode_bounds): For BImode return
12047         0 and STORE_FLAG_VALUE.
12048
12049 2014-01-08  Richard Biener  <rguenther@suse.de>
12050
12051         PR middle-end/59630
12052         * gimple.h (is_gimple_builtin_call): Remove.
12053         (gimple_builtin_call_types_compatible_p): New.
12054         (gimple_call_builtin_p): New overload.
12055         * gimple.c (is_gimple_builtin_call): Remove.
12056         (validate_call): Rename to ...
12057         (gimple_builtin_call_types_compatible_p): ... this and export.  Also
12058         check return types.
12059         (validate_type): New static function.
12060         (gimple_call_builtin_p): New overload and adjust.
12061         * gimple-fold.c (gimple_fold_builtin): Fold the return value.
12062         (gimple_fold_call): Likewise.  Use gimple_call_builtin_p.
12063         (gimple_fold_stmt_to_constant_1): Likewise.
12064         * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
12065
12066 2014-01-08  Richard Biener  <rguenther@suse.de>
12067
12068         PR middle-end/59471
12069         * gimplify.c (gimplify_expr): Gimplify register-register type
12070         VIEW_CONVERT_EXPRs to separate stmts.
12071
12072 2014-01-07  Jeff Law  <law@redhat.com>
12073
12074         PR middle-end/53623
12075         * ree.c (combine_set_extension): Handle case where source
12076         and destination registers in an extension insn are different.
12077         (combine_reaching_defs): Allow source and destination registers
12078         in extension to be different under limited circumstances.
12079         (add_removable_extension): Remove restriction that the
12080         source and destination registers in the extension are the same.
12081         (find_and_remove_re): Emit a copy from the extension's
12082         destination to its source after the defining insn if
12083         the source and destination registers are different.
12084
12085         PR middle-end/59285
12086         * ifcvt.c (merge_if_block): If we are merging a block with more than
12087         one successor with a block with no successors, remove any BARRIER
12088         after the second block.
12089
12090 2014-01-07  Dan Xio Qiang  <ziyan01@163.com>
12091
12092         * hw-doloop.c (reorg_loops): Release the bitmap obstack.
12093
12094 2014-01-07  John David Anglin  <danglin@gcc.gnu.org>
12095
12096         PR target/59652
12097         * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
12098         for 14-bit register offsets when INT14_OK_STRICT is false.
12099
12100 2014-01-07  Roland Stigge  <stigge@antcom.de>
12101             Michael Meissner  <meissner@linux.vnet.ibm.com>
12102
12103         PR 57386/target
12104         * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
12105         Only check TFmode for SPE constants.  Don't check TImode or TDmode.
12106
12107 2014-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
12108
12109         * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
12110         -mcpu.
12111
12112 2014-01-07  Yufeng Zhang  <yufeng.zhang@arm.com>
12113
12114         * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
12115         with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
12116         rtx is const0_rtx or not.
12117
12118 2014-01-07  Richard Sandiford  <rdsandiford@googlemail.com>
12119
12120         PR target/58115
12121         * target-globals.c (save_target_globals): Remove this_fn_optab
12122         handling.
12123         * toplev.c: Include optabs.h.
12124         (target_reinit): Temporarily restore the global options if another
12125         set of options are in force.
12126
12127 2014-01-07  Jakub Jelinek  <jakub@redhat.com>
12128
12129         PR rtl-optimization/58668
12130         * cfgcleanup.c (flow_find_cross_jump): Don't count
12131         any jumps if dir_p is NULL.  Remove p1 variable, use active_insn_p
12132         to determine what is counted.
12133         (flow_find_head_matching_sequence): Use active_insn_p to determine
12134         what is counted.
12135         (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
12136         counting change.
12137         * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
12138         determine what is counted.
12139
12140         PR tree-optimization/59643
12141         * tree-predcom.c (split_data_refs_to_components): If one dr is
12142         read and one write, determine_offset fails and the write isn't
12143         in the bad component, just put the read into the bad component.
12144
12145 2014-01-07  Mike Stump  <mikestump@comcast.net>
12146             Jakub Jelinek  <jakub@redhat.com>
12147
12148         PR pch/59436
12149         * tree-core.h (struct tree_optimization_option): Change optabs
12150         type from unsigned char * to void *.
12151         * optabs.c (init_tree_optimization_optabs): Adjust
12152         TREE_OPTIMIZATION_OPTABS initialization.
12153
12154 2014-01-06  Jakub Jelinek  <jakub@redhat.com>
12155
12156         PR target/59644
12157         * config/i386/i386.h (struct machine_function): Add
12158         no_drap_save_restore field.
12159         * config/i386/i386.c (ix86_save_reg): Use
12160         !cfun->machine->no_drap_save_restore instead of
12161         crtl->stack_realign_needed.
12162         (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
12163         this function clears frame_pointer_needed.  Set
12164         cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
12165         and DRAP reg is needed.
12166
12167 2014-01-06  Marek Polacek  <polacek@redhat.com>
12168
12169         PR c/57773
12170         * doc/implement-c.texi: Mention that other integer types are
12171         permitted as bit-field types in strictly conforming mode.
12172
12173 2014-01-06  Felix Yang  <fei.yang0953@gmail.com>
12174
12175         * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
12176         is newly allocated.
12177
12178 2014-01-06  Richard Earnshaw  <rearnsha@arm.com>
12179
12180         * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
12181
12182 2014-01-06  Martin Jambor  <mjambor@suse.cz>
12183
12184         PR ipa/59008
12185         * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
12186         to int.
12187         * ipa-prop.c (ipa_print_node_params): Fix indentation.
12188
12189 2014-01-06  Eric Botcazou  <ebotcazou@adacore.com>
12190
12191         PR debug/59350
12192         PR debug/59510
12193         * var-tracking.c (add_stores): Preserve the value of the source even if
12194         we don't record the store.
12195
12196 2014-01-06  Terry Guo  <terry.guo@arm.com>
12197
12198         * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
12199
12200 2014-01-05  Iain Sandoe  <iain@codesourcery.com>
12201
12202         PR bootstrap/59541
12203         * config/darwin.c (darwin_function_section): Adjust return values to
12204         correspond to optimisation changes made in r206070.
12205
12206 2014-01-05  Uros Bizjak  <ubizjak@gmail.com>
12207
12208         * config/i386/i386.c (ix86_data_alignment): Calculate max_align
12209         from prefetch_block tune setting.
12210         (nocona_cost): Correct size of prefetch block to 64.
12211
12212 2014-01-04  Eric Botcazou  <ebotcazou@adacore.com>
12213
12214         * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
12215         (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
12216         used to save the static chain register in the computation of the offset
12217         from which the FP registers need to be restored.
12218
12219 2014-01-04  Jakub Jelinek  <jakub@redhat.com>
12220
12221         PR tree-optimization/59519
12222         * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
12223         ICE if get_current_def (current_new_name) is already non-NULL, as long
12224         as it is a phi result of some other phi in *new_exit_bb that has
12225         the same argument.
12226
12227         * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
12228         or vmovdqu* for misaligned_operand.
12229         (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
12230         <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
12231         * config/i386/i386.c (ix86_expand_special_args_builtin): Set
12232         aligned_mem for AVX512F masked aligned load and store builtins and for
12233         non-temporal moves.
12234
12235 2014-01-03  Bingfeng Mei  <bmei@broadcom.com>
12236
12237         PR tree-optimization/59651
12238         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
12239         Address range for negative step should be added by TYPE_SIZE_UNIT.
12240
12241 2014-01-03  Andreas Schwab  <schwab@linux-m68k.org>
12242
12243         * config/m68k/m68k.c (handle_move_double): Handle pushes with
12244         overlapping registers also for registers other than the stack pointer.
12245
12246 2014-01-03  Marek Polacek  <polacek@redhat.com>
12247
12248         PR other/59661
12249         * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
12250         __builtin_FILE.
12251
12252 2014-01-03  Jakub Jelinek  <jakub@redhat.com>
12253
12254         PR target/59625
12255         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
12256         asm goto as jump.
12257
12258         * config/i386/i386.md (MODE_SIZE): New mode attribute.
12259         (push splitter): Use <P:MODE_SIZE> instead of
12260         GET_MODE_SIZE (<P:MODE>mode).
12261         (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
12262         (mov -1, reg peephole2): Likewise.
12263         * config/i386/sse.md (*mov<mode>_internal,
12264         <sse>_storeu<ssemodesuffix><avxsizesuffix>,
12265         <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
12266         *<code><mode>3, *andnot<mode>3<mask_name>,
12267         <mask_codefor><code><mode>3<mask_name>): Likewise.
12268         * config/i386/subst.md (mask_mode512bit_condition,
12269         sd_mask_mode512bit_condition): Likewise.
12270
12271 2014-01-02  Xinliang David Li  <davidxl@google.com>
12272
12273         PR tree-optimization/59303
12274         * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
12275         (dump_predicates): Better output format.
12276         (pred_equal_p): New function.
12277         (is_neq_relop_p): Ditto.
12278         (is_neq_zero_form_p): Ditto.
12279         (pred_expr_equal_p): Ditto.
12280         (pred_neg_p): Ditto.
12281         (simplify_pred): Ditto.
12282         (simplify_preds_2): Ditto.
12283         (simplify_preds_3): Ditto.
12284         (simplify_preds_4): Ditto.
12285         (simplify_preds): Ditto.
12286         (push_pred): Ditto.
12287         (push_to_worklist): Ditto.
12288         (get_pred_info_from_cmp): Ditto.
12289         (is_degenerated_phi): Ditto.
12290         (normalize_one_pred_1): Ditto.
12291         (normalize_one_pred): Ditto.
12292         (normalize_one_pred_chain): Ditto.
12293         (normalize_preds): Ditto.
12294         (normalize_cond_1): Remove function.
12295         (normalize_cond): Ditto.
12296         (is_gcond_subset_of): Ditto.
12297         (is_subset_of_any): Ditto.
12298         (is_or_set_subset_of): Ditto.
12299         (is_and_set_subset_of): Ditto.
12300         (is_norm_cond_subset_of): Ditto.
12301         (pred_chain_length_cmp): Ditto.
12302         (convert_control_dep_chain_into_preds): Type change.
12303         (find_predicates): Ditto.
12304         (find_def_preds): Ditto.
12305         (destroy_predicates_vecs): Ditto.
12306         (find_matching_predicates_in_rest_chains): Ditto.
12307         (use_pred_not_overlap_with_undef_path_pred): Ditto.
12308         (is_pred_expr_subset): Ditto.
12309         (is_pred_chain_subset_of): Ditto.
12310         (is_included_in): Ditto.
12311         (is_superset_of): Ditto.
12312
12313 2014-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
12314
12315         Update copyright years.
12316
12317 2014-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
12318
12319         * common/config/arc/arc-common.c, config/arc/arc-modes.def,
12320         config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
12321         config/arc/arc.md, config/arc/arc.opt,
12322         config/arm/arm_neon_builtins.def, config/arm/crypto.def,
12323         config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
12324         config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
12325         config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
12326         config/linux-protos.h, config/linux.c, config/winnt-c.c,
12327         diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
12328         vtable-verify.c, vtable-verify.h: Use the standard form for the
12329         copyright notice.
12330
12331 2014-01-02  Tobias Burnus  <burnus@net-b.de>
12332
12333         * gcc.c (process_command): Update copyright notice dates.
12334         * gcov-dump.c: Ditto.
12335         * gcov.c: Ditto.
12336         * doc/cpp.texi: Bump @copying's copyright year.
12337         * doc/cppinternals.texi: Ditto.
12338         * doc/gcc.texi: Ditto.
12339         * doc/gccint.texi: Ditto.
12340         * doc/gcov.texi: Ditto.
12341         * doc/install.texi: Ditto.
12342         * doc/invoke.texi: Ditto.
12343
12344 2014-01-01  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
12345
12346         * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
12347
12348 2014-01-01  Jakub Jelinek  <jakub@redhat.com>
12349
12350         * config/i386/sse.md (*mov<mode>_internal): Guard
12351         EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
12352
12353         PR rtl-optimization/59647
12354         * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
12355         new_rtx into UNSIGNED_FLOAT rtxes.
12356 \f
12357 Copyright (C) 2014 Free Software Foundation, Inc.
12358
12359 Copying and distribution of this file, with or without modification,
12360 are permitted in any medium without royalty provided the copyright
12361 notice and this notice are preserved.