gcc/ChangeLog

   1 2021-01-06  Gerald Pfeifer  <gerald@pfeifer.com>
   2
   3         Revert:
   4         2020-12-28  Gerald Pfeifer  <gerald@pfeifer.com>
   5
   6         * doc/standards.texi (HSAIL): Remove section.
   7
   8 2021-01-05  Samuel Thibault  <samuel.thibault@ens-lyon.org>
   9
  10         * configure: Re-generate.
  11
  12 2021-01-05  Jakub Jelinek  <jakub@redhat.com>
  13
  14         * doc/invoke.texi (-std=c++20): Adjust for the publication of
  15         ISO 14882:2020 standard.
  16         * doc/standards.texi: Likewise.
  17
  18 2021-01-05  Jakub Jelinek  <jakub@redhat.com>
  19
  20         PR tree-optimization/94802
  21         * expr.h (maybe_optimize_sub_cmp_0): Declare.
  22         * expr.c: Include tree-pretty-print.h and flags.h.
  23         (maybe_optimize_sub_cmp_0): New function.
  24         (do_store_flag): Use it.
  25         * cfgexpand.c (expand_gimple_cond): Likewise.
  26
  27 2021-01-05  Richard Sandiford  <richard.sandiford@arm.com>
  28
  29         * mux-utils.h (pointer_mux::m_ptr): Tweak description of contents.
  30         * rtlanal.c (simple_regno_set): Tweak description to clarify the
  31         RMW condition.
  32
  33 2021-01-05  Richard Biener  <rguenther@suse.de>
  34
  35         PR tree-optimization/98516
  36         * tree-vect-slp.c (vect_optimize_slp): Permute the incoming
  37         lanes when materializing on a VEC_PERM node.
  38         (vectorizable_slp_permutation): Dump the permute properly.
  39
  40 2021-01-05  Richard Biener  <rguenther@suse.de>
  41
  42         * tree-vect-slp.c (vect_slp_region): Move debug counter
  43         to cover individual subgraphs.
  44
  45 2021-01-05  Richard Biener  <rguenther@suse.de>
  46
  47         PR tree-optimization/98428
  48         * tree-vect-slp.c (vect_build_slp_tree_1): Properly reject
  49         vector lane extracts for loop vectorization.
  50
  51 2021-01-05  Jakub Jelinek  <jakub@redhat.com>
  52
  53         PR tree-optimization/98514
  54         * tree-ssa-reassoc.c (bb_rank): Change type from long * to
  55         int64_t *.
  56         (operand_rank): Change type from hash_map<tree, long> to
  57         hash_map<tree, int64_t>.
  58         (phi_rank): Change return type from long to int64_t.
  59         (loop_carried_phi): Change block_rank variable type from long to
  60         int64_t.
  61         (propagate_rank): Change return type, rank parameter type and
  62         op_rank variable type from long to int64_t.
  63         (find_operand_rank): Change return type from long to int64_t
  64         and change slot variable type from long * to int64_t *.
  65         (insert_operand_rank): Change rank parameter type from long to
  66         int64_t.
  67         (get_rank): Change return type and rank variable type from long to
  68         int64_t.  Use PRId64 instead of ld to print the rank.
  69         (init_reassoc): Change rank variable type from long to int64_t
  70         and adjust correspondingly bb_rank and operand_rank initialization.
  71
  72 2021-01-05  Jakub Jelinek  <jakub@redhat.com>
  73
  74         PR tree-optimization/96928
  75         * tree-ssa-phiopt.c (xor_replacement): New function.
  76         (tree_ssa_phiopt_worker): Call it.
  77
  78 2021-01-05  Jakub Jelinek  <jakub@redhat.com>
  79
  80         PR tree-optimization/96930
  81         * match.pd ((A / (1 << B)) -> (A >> B)): If A is extended
  82         from narrower value which has the same type as 1 << B, perform
  83         the right shift on the narrower value followed by extension.
  84
  85 2021-01-05  Jakub Jelinek  <jakub@redhat.com>
  86
  87         PR tree-optimization/96239
  88         * gimple-ssa-store-merging.c (maybe_optimize_vector_constructor): New
  89         function.
  90         (get_status_for_store_merging): Don't return BB_INVALID for blocks
  91         with potential bswap optimizable CONSTRUCTORs.
  92         (pass_store_merging::execute): Optimize vector CONSTRUCTORs with bswap
  93         if possible.
  94
  95 2021-01-05  Richard Biener  <rguenther@suse.de>
  96
  97         PR tree-optimization/98381
  98         * tree.c (vector_element_bits): Properly compute bool vector
  99         element size.
 100         * tree-vect-loop.c (vectorizable_live_operation): Properly
 101         compute the last lane bit offset.
 102
 103 2021-01-05  Uroš Bizjak  <ubizjak@gmail.com>
 104
 105         PR target/98522
 106         * config/i386/sse.md (sse_cvtps2pi): Redefine as define_insn_and_split.
 107         Clear the top 64 bytes of the input XMM register.
 108         (sse_cvttps2pi): Ditto.
 109
 110 2021-01-05  Uroš Bizjak  <ubizjak@gmail.com>
 111
 112         PR target/98521
 113         * config/i386/xopintrin.h (_mm256_cmov_si256): New.
 114
 115 2021-01-05  H.J. Lu  <hjl.tools@gmail.com>
 116
 117         PR target/98495
 118         * config/i386/xmmintrin.h (_mm_extract_pi16): Cast to unsigned
 119         short first.
 120
 121 2021-01-05  Claudiu Zissulescu  <claziss@synopsys.com>
 122
 123         * config/arc/arc.md (maddsidi4_split): Use ACC_REG_FIRST.
 124         (umaddsidi4_split): Likewise.
 125
 126 2021-01-05  liuhongt  <hongtao.liu@intel.com>
 127
 128         PR target/98461
 129         * config/i386/sse.md (*sse2_pmovskb_zexthisi): New
 130         define_insn_and_split for zero_extend of subreg HI of pmovskb
 131         result.
 132         (*sse2_pmovskb_zexthisi): Add new combine splitters for
 133         zero_extend of not of subreg HI of pmovskb result.
 134
 135 2021-01-05  Richard Sandiford  <richard.sandiford@arm.com>
 136
 137         PR target/97269
 138         * explow.c (convert_memory_address_addr_space_1): Handle UNSPECs
 139         nested in CONSTs.
 140         * config/aarch64/aarch64.c (aarch64_expand_mov_immediate): Use
 141         convert_memory_address to convert symbolic immediates to ptr_mode
 142         before forcing them to memory.
 143
 144 2021-01-05  Richard Sandiford  <richard.sandiford@arm.com>
 145
 146         PR rtl-optimization/97144
 147         * recog.c (constrain_operands): Initialize matching_operand
 148         for each alternative, rather than only doing it once.
 149
 150 2021-01-05  Richard Sandiford  <richard.sandiford@arm.com>
 151
 152         PR rtl-optimization/98403
 153         * rtl-ssa/changes.cc (function_info::finalize_new_accesses): Explain
 154         why we don't remove call clobbers.
 155         (function_info::apply_changes_to_insn): Don't attempt to add
 156         call clobbers here.
 157
 158 2021-01-05  Richard Sandiford  <richard.sandiford@arm.com>
 159
 160         PR tree-optimization/98371
 161         * tree-vect-loop.c (vect_reanalyze_as_main_loop): New function.
 162         (vect_analyze_loop): If an epilogue loop appears to be cheaper
 163         than the main loop, re-analyze it as a main loop before adopting
 164         it as a main loop.
 165
 166 2021-01-05  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
 167
 168         PR c++/98316
 169         * configure.ac (NETLIBS): Determine using AX_LIB_SOCKET_NSL.
 170         * aclocal.m4, configure: Regenerate.
 171         * Makefile.in (NETLIBS): Define.
 172         (BACKEND): Remove $(CODYLIB).
 173
 174 2021-01-05  Jakub Jelinek  <jakub@redhat.com>
 175
 176         PR rtl-optimization/98334
 177         * simplify-rtx.c (simplify_context::simplify_binary_operation_1):
 178         Optimize (X - 1) * Y + Y to X * Y or (X + 1) * Y - Y to X * Y.
 179
 180 2021-01-05  Bernd Edlinger  <bernd.edlinger@hotmail.de>
 181
 182         * tree-inline.c (expand_call_inline): Restore input_location.
 183         Return result from recursive call.
 184
 185 2021-01-04  Richard Sandiford  <richard.sandiford@arm.com>
 186
 187         PR tree-optimization/95401
 188         * config/aarch64/aarch64-sve-builtins.cc
 189         (gimple_folder::load_store_cookie): Use bits rather than bytes
 190         for the alignment argument to IFN_MASK_LOAD and IFN_MASK_STORE.
 191         * gimple-fold.c (gimple_fold_mask_load_store_mem_ref): Likewise.
 192         * tree-vect-stmts.c (vectorizable_store): Likewise.
 193         (vectorizable_load): Likewise.
 194
 195 2021-01-04  Richard Biener  <rguenther@suse.de>
 196
 197         PR tree-optimization/98308
 198         * tree-vect-stmts.c (vectorizable_load): Set invariant mask
 199         SLP vectype.
 200
 201 2021-01-04  Jakub Jelinek  <jakub@redhat.com>
 202
 203         PR tree-optimization/95771
 204         * tree-ssa-loop-niter.c (number_of_iterations_popcount): Handle types
 205         with precision smaller than int's precision and types with precision
 206         twice as large as long long.  Formatting fixes.
 207
 208 2021-01-04  Richard Biener  <rguenther@suse.de>
 209
 210         PR tree-optimization/98464
 211         * tree-ssa-sccvn.c (vn_valueize_for_srt): Rename from ...
 212         (vn_valueize_wrapper): ... this.  Temporarily adjust vn_context_bb.
 213         (process_bb): Adjust.
 214
 215 2021-01-04  Matthew Malcomson  <matthew.malcomson@arm.com>
 216
 217         PR other/98437
 218         * doc/invoke.texi (-fsanitize=address): Fix wording describing
 219         clash with -fsanitize=hwaddress.
 220
 221 2021-01-04  Richard Biener  <rguenther@suse.de>
 222
 223         PR tree-optimization/98282
 224         * tree-ssa-sccvn.c (vn_get_stmt_kind): Classify tcc_reference on
 225         invariants as VN_NARY.
 226
 227 2021-01-04  Richard Sandiford  <richard.sandiford@arm.com>
 228
 229         PR target/89057
 230         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Accept
 231         aarch64_simd_reg_or_zero for operand 2.  Use the combinez patterns
 232         to handle zero operands.
 233
 234 2021-01-04  Richard Sandiford  <richard.sandiford@arm.com>
 235
 236         * config/aarch64/aarch64.c (offset_6bit_signed_scaled_p): New function.
 237         (offset_6bit_unsigned_scaled_p): Fix typo in comment.
 238         (aarch64_sve_prefetch_operand_p): Accept MUL VLs in the range
 239         [-32, 31].
 240
 241 2021-01-04  Richard Biener  <rguenther@suse.de>
 242
 243         PR tree-optimization/98393
 244         * tree-vect-slp.c (vect_build_slp_tree): Properly zero matches
 245         when hitting the limit.
 246
 247 2021-01-04  Richard Biener  <rguenther@suse.de>
 248
 249         PR tree-optimization/98291
 250         * tree-vect-loop.c (vectorizable_reduction): Bypass
 251         associativity check for SLP reductions with VF 1.
 252
 253 2021-01-04  Jakub Jelinek  <jakub@redhat.com>
 254
 255         PR tree-optimization/96782
 256         * match.pd (x == ~x -> false, x != ~x -> true): New simplifications.
 257
 258 2021-01-04  Bernd Edlinger  <bernd.edlinger@hotmail.de>
 259
 260         * collect-utils.c (collect_execute): Check dumppfx.
 261         * collect2.c (maybe_run_lto_and_relink, do_link): Pass atsuffix
 262         to collect_execute.
 263         (do_link): Add new parameter atsuffix.
 264         (main): Handle -dumpdir option.  Skip one argument for
 265         -o, -isystem and -B options.
 266         * gcc.c (make_at_file): New helper function.
 267         (close_at_file): Use it.
 268
 269 2021-01-02  Iain Sandoe  <iain@sandoe.co.uk>
 270
 271         * config/darwin.h (MIN_LD64_NO_COAL_SECTS): Adjust.
 272         Amend handling for LD64_VERSION fallback defaults.
 273
 274 2021-01-02  Iain Sandoe  <iain@sandoe.co.uk>
 275
 276         * config.gcc: Compute default version information
 277         from the configured target.  Likewise defaults for
 278         ld64.
 279         * config/darwin10.h: Removed.
 280         * config/darwin12.h: Removed.
 281         * config/darwin9.h: Removed.
 282         * config/rs6000/darwin8.h: Removed.
 283
 284 2021-01-02  Iain Sandoe  <iain@sandoe.co.uk>
 285
 286         * config/darwin9.h (ASM_OUTPUT_ALIGNED_COMMON): Delete.
 287
 288 2021-01-02  Iain Sandoe  <iain@sandoe.co.uk>
 289
 290         * config/darwin9.h (STACK_CHECK_STATIC_BUILTIN): Move from here..
 291         * config/darwin.h (STACK_CHECK_STATIC_BUILTIN): .. to here.
 292
 293 2021-01-02  Iain Sandoe  <iain@sandoe.co.uk>
 294
 295         * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Move from
 296         here...
 297         * config/darwin.h (LINK_GCC_C_SEQUENCE_SPEC): ... to here.
 298
 299 2021-01-02  Iain Sandoe  <iain@sandoe.co.uk>
 300
 301         * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Move the spec
 302         for the Darwin10 unwinder stub from here ...
 303         * config/darwin.h (LINK_COMMAND_SPEC_A): ... to here.
 304
 305 2021-01-02  Iain Sandoe  <iain@sandoe.co.uk>
 306
 307         * config/darwin.h (DSYMUTIL_SPEC): Default to DWARF
 308         (ASM_DEBUG_SPEC):Only define if the assembler supports
 309         stabs.
 310         (PREFERRED_DEBUGGING_TYPE): Default to DWARF.
 311         (DARWIN_PREFER_DWARF): Define.
 312         * config/darwin9.h (PREFERRED_DEBUGGING_TYPE): Remove.
 313         (DARWIN_PREFER_DWARF): Likewise
 314         (DSYMUTIL_SPEC): Likewise.
 315         (COLLECT_RUN_DSYMUTIL): Likewise.
 316         (ASM_DEBUG_SPEC): Likewise.
 317         (ASM_DEBUG_OPTION_SPEC): Likewise.
 318
 319 2021-01-02  Jan Hubicka  <jh@suse.cz>
 320
 321         * cfg.c (free_block): ggc_free bb.
 322
 323 2021-01-01  Jakub Jelinek  <jakub@redhat.com>
 324
 325         * gcc.c (process_command): Update copyright notice dates.
 326         * gcov-dump.c (print_version): Ditto.
 327         * gcov.c (print_version): Ditto.
 328         * gcov-tool.c (print_version): Ditto.
 329         * gengtype.c (create_file): Ditto.
 330         * doc/cpp.texi: Bump @copying's copyright year.
 331         * doc/cppinternals.texi: Ditto.
 332         * doc/gcc.texi: Ditto.
 333         * doc/gccint.texi: Ditto.
 334         * doc/gcov.texi: Ditto.
 335         * doc/install.texi: Ditto.
 336         * doc/invoke.texi: Ditto.
 337
 338 2021-01-01  Jakub Jelinek  <jakub@redhat.com>
 339
 340         * ChangeLog-2020: Rotate ChangeLog.  New file.
 341
 342 \f
 343 Copyright (C) 2021 Free Software Foundation, Inc.
 344
 345 Copying and distribution of this file, with or without modification,
 346 are permitted in any medium without royalty provided the copyright
 347 notice and this notice are preserved.