gcc/ChangeLog

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