gcc/ChangeLog

   1 2020-01-07  Martin Liska  <mliska@suse.cz>
   2
   3         PR tree-optimization/92860
   4         * common.opt: Make flag_ree as optimization
   5         attribute.
   6
   7 2020-01-07  Martin Liska  <mliska@suse.cz>
   8
   9         PR optimization/92860
  10         * params.opt: Mark param_min_crossjump_insns with Optimization
  11         keyword.
  12
  13 2020-01-07  Luo Xiong Hu  <luoxhu@linux.ibm.com>
  14
  15         * ipa-inline-analysis.c (estimate_growth): Fix typo.
  16         * ipa-inline.c (caller_growth_limits): Use OR instead of AND.
  17
  18 2020-01-06  Michael Meissner  <meissner@linux.ibm.com>
  19
  20         * config/rs6000/rs6000.c (hard_reg_and_mode_to_addr_mask): New
  21         helper function to return the valid addressing formats for a given
  22         hard register and mode.
  23         (rs6000_adjust_vec_address): Call hard_reg_and_mode_to_addr_mask.
  24
  25         * config/rs6000/constraints.md (Q constraint): Update
  26         documentation.
  27         * doc/md.texi (RS/6000 constraints): Update 'Q' cosntraint
  28         documentation.
  29
  30         * config/rs6000/vsx.md (vsx_extract_<mode>_var, VSX_D iterator):
  31         Use 'Q' for doing vector extract from memory.
  32         (vsx_extract_v4sf_var): Use 'Q' for doing vector extract from
  33         memory.
  34         (vsx_extract_<mode>_var, VSX_EXTRACT_I iterator): Use 'Q' for
  35         doing vector extract from memory.
  36         (vsx_extract_<mode>_<VS_scalar>mode_var): Use 'Q' for doing vector
  37         extract from memory.
  38
  39         * config/rs6000/rs6000.c (rs6000_adjust_vec_address): Add support
  40         for the offset being 34-bits when -mcpu=future is used.
  41
  42 2020-01-06  John David Anglin  <danglin@gcc.gnu.org>
  43
  44         * config/pa/pa.md: Revert change to use ordered_comparison_operator
  45         instead of cmpib_comparison_operator in cmpib patterns.
  46         * config/pa/predicates.md (cmpib_comparison_operator): Revert removal
  47         of cmpib_comparison_operator.  Revise comment.
  48
  49 2020-01-06  Richard Sandiford  <richard.sandiford@arm.com>
  50
  51         * tree-vect-slp.c (vect_build_slp_tree_1): Require all shifts
  52         in an IFN_DIV_POW2 node to be equal.
  53
  54 2020-01-06  Richard Sandiford  <richard.sandiford@arm.com>
  55
  56         * tree-vect-stmts.c (vect_check_load_store_mask): Rename to...
  57         (vect_check_scalar_mask): ...this.
  58         (vectorizable_store, vectorizable_load): Update call accordingly.
  59         (vectorizable_call): Use vect_check_scalar_mask to check the mask
  60         argument in calls to conditional internal functions.
  61
  62 2020-01-06  Andrew Stubbs  <ams@codesourcery.com>
  63
  64         * config/gcn/gcn-valu.md (subv64di3): Use separate alternatives for
  65         '0' matching inputs.
  66         (subv64di3_exec): Likewise.
  67
  68 2020-01-06  Bryan Stenson  <bryan@siliconvortex.com>
  69
  70         * config/mips/mips.c (vr4130_align_insns): Fix typo.
  71         * doc/md.texi (movstr): Likewise.
  72
  73 2020-01-06  Andrew Stubbs  <ams@codesourcery.com>
  74
  75         * config/gcn/gcn-valu.md (vec_extract<mode><scalar_mode>): Add early
  76         clobber.
  77
  78 2020-01-06  Richard Sandiford  <richard.sandiford@arm.com>
  79
  80         * config/aarch64/t-aarch64 ($(srcdir)/config/aarch64/aarch64-tune.md):
  81         Depend on...
  82         (s-aarch64-tune-md): ...this new stamp file.  Pipe the new contents
  83         to a temporary file and use move-if-change to update the real
  84         file where necessary.
  85
  86 2020-01-06  Richard Sandiford  <richard.sandiford@arm.com>
  87
  88         * config/aarch64/aarch64-sve.md (@aarch64_sel_dup<mode>): Use Upl
  89         rather than Upa for CPY /M.
  90
  91 2020-01-06  Andrew Stubbs  <ams@codesourcery.com>
  92
  93         * config/gcn/gcn.c (gcn_inline_constant_p): Allow 64 as an inline
  94         immediate.
  95
  96 2020-01-06  Martin Liska  <mliska@suse.cz>
  97
  98     PR tree-optimization/92860
  99     * params.opt: Mark param_max_combine_insns with Optimization
 100     keyword.
 101
 102 2020-01-05  Jakub Jelinek  <jakub@redhat.com>
 103
 104         PR target/93141
 105         * config/i386/i386.md (SWIDWI): New mode iterator.
 106         (DWI, dwi): Add TImode variants.
 107         (addv<mode>4): Use SWIDWI iterator instead of SWI.  Use
 108         <general_hilo_operand> instead of <general_operand>.  Use
 109         CONST_SCALAR_INT_P instead of CONST_INT_P.
 110         (*addv<mode>4_1): Rename to ...
 111         (addv<mode>4_1): ... this.
 112         (QWI): New mode attribute.
 113         (*addv<dwi>4_doubleword, *addv<dwi>4_doubleword_1): New
 114         define_insn_and_split patterns.
 115         (*addv<mode>4_overflow_1, *addv<mode>4_overflow_2): New define_insn
 116         patterns.
 117         (uaddv<mode>4): Use SWIDWI iterator instead of SWI.  Use
 118         <general_hilo_operand> instead of <general_operand>.
 119         (*addcarry<mode>_1): New define_insn.
 120         (*add<dwi>3_doubleword_cc_overflow_1): New define_insn_and_split.
 121
 122 2020-01-03  Konstantin Kharlamov  <Hi-Angel@yandex.ru>
 123
 124         * gdbinit.in (pr, prl, pt, pct, pgg, pgq, pgs, pge, pmz, pdd, pbs, pbm):
 125         Use "call" instead of "set".
 126
 127 2020-01-03  Martin Jambor  <mjambor@suse.cz>
 128
 129         PR ipa/92917
 130         * ipa-cp.c (print_all_lattices): Skip functions without info.
 131
 132 2020-01-03  Jakub Jelinek  <jakub@redhat.com>
 133
 134         PR target/93089
 135         * config/i386/i386-options.c (ix86_simd_clone_adjust): If
 136         TARGET_PREFER_AVX128, use prefer-vector-width=256 for 'c' and 'd'
 137         simd clones.  If TARGET_PREFER_AVX256, use prefer-vector-width=512
 138         for 'e' simd clones.
 139
 140         PR target/93089
 141         * config/i386/i386.opt (x_prefer_vector_width_type): Remove TargetSave
 142         entry.
 143         (mprefer-vector-width=): Add Save.
 144         * config/i386/i386-options.c (ix86_target_string): Add PVW argument, print
 145         -mprefer-vector-width= if non-zero.  Fix up -mfpmath= comment.
 146         (ix86_debug_options, ix86_function_specific_print): Adjust
 147         ix86_target_string callers.
 148         (ix86_valid_target_attribute_inner_p): Handle prefer-vector-width=.
 149         (ix86_valid_target_attribute_tree): Likewise.
 150         * config/i386/i386-options.h (ix86_target_string): Add PVW argument.
 151         * config/i386/i386-expand.c (ix86_expand_builtin): Adjust
 152         ix86_target_string caller.
 153
 154         PR target/93110
 155         * config/i386/i386.md (abs<mode>2): Use expand_simple_binop instead of
 156         emitting ASHIFTRT, XOR and MINUS by hand.  Use gen_int_mode with QImode
 157         instead of gen_int_shift_amount + convert_modes.
 158
 159         PR rtl-optimization/93088
 160         * loop-iv.c (find_single_def_src): Punt after looking through
 161         128 reg copies for regs with single definitions.  Move definitions
 162         to first uses.
 163
 164 2020-01-02  Dennis Zhang  <dennis.zhang@arm.com>
 165
 166         * config/arm/arm-c.c (arm_cpu_builtins): Define
 167         __ARM_FEATURE_MATMUL_INT8, __ARM_FEATURE_BF16_VECTOR_ARITHMETIC,
 168         __ARM_FEATURE_BF16_SCALAR_ARITHMETIC, and
 169         __ARM_BF16_FORMAT_ALTERNATIVE when enabled.
 170         * config/arm/arm-cpus.in (armv8_6, i8mm, bf16): New features.
 171         * config/arm/arm-tables.opt: Regenerated.
 172         * config/arm/arm.c (arm_option_reconfigure_globals): Initialize
 173         arm_arch_i8mm and arm_arch_bf16 when enabled.
 174         * config/arm/arm.h (TARGET_I8MM): New macro.
 175         (TARGET_BF16_FP, TARGET_BF16_SIMD): Likewise.
 176         * config/arm/t-aprofile: Add matching rules for -march=armv8.6-a.
 177         * config/arm/t-arm-elf (all_v8_archs): Add armv8.6-a.
 178         * config/arm/t-multilib: Add matching rules for -march=armv8.6-a.
 179         (v8_6_a_simd_variants): New.
 180         (v8_*_a_simd_variants): Add i8mm and bf16.
 181         * doc/invoke.texi (armv8.6-a, i8mm, bf16): Document new options.
 182
 183 2020-01-02  Jakub Jelinek  <jakub@redhat.com>
 184
 185         PR ipa/93087
 186         * predict.c (compute_function_frequency): Don't call
 187         warn_function_cold on functions that already have cold attribute.
 188
 189 2020-01-01  John David Anglin  <danglin@gcc.gnu.org>
 190
 191         PR target/67834
 192         * config/pa/pa.c (pa_elf_select_rtx_section): New.  Put references to
 193         COMDAT group function labels in .data.rel.ro.local section.
 194         * config/pa/pa32-linux.h (TARGET_ASM_SELECT_RTX_SECTION): Define.
 195
 196         PR target/93111
 197         * config/pa/pa.md (scc): Use ordered_comparison_operator instead of
 198         comparison_operator in B and S integer comparisons.  Likewise, use
 199         ordered_comparison_operator instead of cmpib_comparison_operator in
 200         cmpib patterns.
 201         * config/pa/predicates.md (cmpib_comparison_operator): Remove.
 202
 203 2020-01-01  Jakub Jelinek  <jakub@redhat.com>
 204
 205         Update copyright years.
 206
 207         * gcc.c (process_command): Update copyright notice dates.
 208         * gcov-dump.c (print_version): Ditto.
 209         * gcov.c (print_version): Ditto.
 210         * gcov-tool.c (print_version): Ditto.
 211         * gengtype.c (create_file): Ditto.
 212         * doc/cpp.texi: Bump @copying's copyright year.
 213         * doc/cppinternals.texi: Ditto.
 214         * doc/gcc.texi: Ditto.
 215         * doc/gccint.texi: Ditto.
 216         * doc/gcov.texi: Ditto.
 217         * doc/install.texi: Ditto.
 218         * doc/invoke.texi: Ditto.
 219
 220 2020-01-01  Jan Hubicka  <hubicka@ucw.cz>
 221
 222         * ipa.c (walk_polymorphic_call_targets): Fix updating of overall
 223         summary.
 224
 225 2020-01-01  Jakub Jelinek  <jakub@redhat.com>
 226
 227         PR tree-optimization/93098
 228         * match.pd (popcount): For shift amounts, use integer_onep
 229         or wi::to_widest () == cst instead of tree_to_uhwi () == cst
 230         tests.  Make sure that precision is power of two larger than or equal
 231         to 16.  Ensure shift is never negative.  Use HOST_WIDE_INT_UC macro
 232         instead of ULL suffixed constants.  Formatting fixes.
 233 \f
 234 Copyright (C) 2020 Free Software Foundation, Inc.
 235
 236 Copying and distribution of this file, with or without modification,
 237 are permitted in any medium without royalty provided the copyright
 238 notice and this notice are preserved.