1 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
3 * configure.ac (--enable-libssp): New.
4 (gcc_cv_libc_provides_ssp): Check for explicit setting before
5 trying to determine target-specific default. Adjust indentation.
6 * configure: Regenerated.
7 * doc/install.texi (Configuration): Expand --disable-libssp
10 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
12 config/i386/i386.c (ix86_expand_epilogue): Correct stack
15 2017-10-31 Martin Jambor <mjambor@suse.cz>
18 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
20 2017-10-31 David Malcolm <dmalcolm@redhat.com>
22 * auto-profile.c (autofdo_source_profile::read): Use
23 UNKNOWN_LOCATION rather than 0.
24 * diagnostic-core.h (warning_at_rich_loc): Rename to...
25 (warning_at): ...this overload.
26 (warning_at_rich_loc_n): Rename to...
27 (warning_n): ...this overload.
28 (error_at_rich_loc): Rename to...
29 (error_at): ...this overload.
30 (pedwarn_at_rich_loc): Rename to...
31 (pedwarn): ...this overload.
32 (permerror_at_rich_loc): Rename to...
33 (permerror): ...this overload.
34 (inform_at_rich_loc): Rename to...
35 (inform): ...this overload.
36 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
37 (diagnostic_n_impl_richloc): Rename to...
38 (diagnostic_n_impl): ...this rich_location *-based decl.
39 (inform_at_rich_loc): Rename to...
40 (inform): ...this, and add an assertion.
41 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
42 (warning_at_rich_loc): Rename to...
43 (warning_at): ...this, and add an assertion.
44 (warning_at_rich_loc_n): Rename to...
45 (warning_n): ...this, and add an assertion.
46 (warning_n): Update location_t-based implementation for removal of
47 location_t-based diagnostic_n_impl.
48 (pedwarn_at_rich_loc): Rename to...
49 (pedwarn): ...this, and add an assertion.
50 (permerror_at_rich_loc): Rename to...
51 (permerror): ...this, and add an assertion.
52 (error_n): Update for removal of location_t-based diagnostic_n_impl.
53 (error_at_rich_loc): Rename to...
54 (error_at): ...this, and add an assertion.
55 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
56 (driver::do_spec_on_infiles): Likewise.
57 * substring-locations.c (format_warning_va): Update for renaming
58 of inform_at_rich_loc.
60 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
62 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
63 _Float<N>X built-in functions so that the variant without the
64 "__builtin_" prefix is only enabled for the GNU C and Objective C
65 languages when they are in non-strict ANSI/ISO mode.
66 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
67 * target.def (floatn_builtin_p): Add a target hook to control
68 whether _Float<N> and _Float<N>X built-in functions without the
69 "__builtin_" prefix are enabled, and return true for C and
70 Objective C in the default hook. Include langhooks.h in
72 * targhooks.h (default_floatn_builtin_p): Likewise.
73 * targhooks.c (default_floatn_builtin_p): Likewise.
74 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
75 floatn_builtin_p target hook.
76 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
78 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
79 Eric Botcazou <ebotcazou@adacore.com>
81 PR rtl-optimization/81803
82 * lra-constraints.c (curr_insn_transform): Also reload the whole
83 register for a strict subreg no wider than a word if this is for
84 a WORD_REGISTER_OPERATIONS target.
86 2017-10-31 Jason Merrill <jason@redhat.com>
88 * gdbinit.in: Skip over inlines from timevar.h.
90 2017-10-31 Martin Liska <mliska@suse.cz>
92 * doc/gcov.texi: Document new option.
93 * gcov.c (print_usage): Likewise print it.
94 (process_args): Support the argument.
95 (format_count): New function.
96 (format_gcov): Use the function.
98 2017-10-31 Martin Liska <mliska@suse.cz>
100 * gcov.c (struct name_map): do not use typedef.
101 Define operator== and operator<.
102 (name_search): Remove.
104 (main): Do not allocate names.
105 (process_file): Add vertical space.
106 (generate_results): Use std::find.
107 (release_structures): Do not release memory.
108 (find_source): Use std::find.
110 2017-10-31 Martin Liska <mliska@suse.cz>
112 * gcov.c (struct line_info): Remove it's typedef.
113 (line_info::line_info): Add proper ctor.
114 (line_info::has_block): Do not use a typedef.
115 (struct source_info): Do not use typedef.
117 (get_cycles_count): Likewise.
118 (output_intermediate_file): Iterate via vector iterator.
119 (add_line_counts): Use std::vector methods.
120 (accumulate_line_counts): Likewise.
121 (output_lines): Likewise.
123 2017-10-31 Martin Liska <mliska@suse.cz>
125 * gcov.c (struct source_info): Remove typedef.
126 (source_info::source_info): Add proper ctor.
127 (accumulate_line_counts): Use struct, not it's typedef.
128 (output_gcov_file): Likewise.
129 (output_lines): Likewise.
130 (main): Do not allocate an array.
131 (output_intermediate_file): Use size of vector container.
132 (process_file): Resize the vector.
133 (generate_results): Do not preallocate, use newly added vector
135 (release_structures): Do not release sources.
136 (find_source): Use vector methods.
137 (add_line_counts): Do not use typedef.
139 2017-10-31 Martin Liska <mliska@suse.cz>
141 * doc/gcov.texi: Document that.
142 * gcov.c (add_line_counts): Mark lines with a non-executed
144 (output_line_beginning): Handle such lines.
145 (output_lines): Pass new argument.
146 (output_intermediate_file): Print it in intermediate format.
148 2017-10-31 Martin Liska <mliska@suse.cz>
150 * color-macros.h: New file.
151 * diagnostic-color.c: Factor out color related to macros to
153 * doc/gcov.texi: Document -k option.
154 * gcov.c (INCLUDE_STRING): Include string.h.
155 (print_usage): Add -k option.
156 (process_args): Parse it.
157 (pad_count_string): New function.
158 (output_line_beginning): Likewise.
159 (DEFAULT_LINE_START): New macro.
160 (output_lines): Support color output.
162 2017-10-31 Martin Liska <mliska@suse.cz>
164 PR gcov-profile/82633
165 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
166 their interaction with GCOV infrastructure.
167 * configure.ac: Add -fkeep-{inline,static}-functions to
169 * configure: Regenerate.
171 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
174 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
176 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
179 * config/rs6000/rs6000.md (allocate_stack): Force update interval
180 into a register if it does not fit into an immediate offset field.
182 2017-10-31 Olivier Hainque <hainque@adacore.com>
184 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
186 2017-10-31 Julia Koval <julia.koval@intel.com>
188 * config.gcc: Add gfniintrin.h.
189 * config/i386/gfniintrin.h: New.
190 * config/i386/i386-builtin-types.def
191 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
192 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
193 __builtin_ia32_vgf2p8affineinvqb_v32qi,
194 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
195 __builtin_ia32_vgf2p8affineinvqb_v16qi,
196 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
197 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
198 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
199 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
200 V64QI_FTYPE_V64QI_V64QI_INT): New types.
201 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
202 * config/i386/immintrin.h: Include gfniintrin.h.
203 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
205 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
207 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
209 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
211 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
212 (arm_ashldi3_1bit): Remove pattern.
213 (ashrdi3): Remove shift by 1 expansion.
214 (arm_ashrdi3_1bit): Remove pattern.
215 (lshrdi3): Remove shift by 1 expansion.
216 (arm_lshrdi3_1bit): Remove pattern.
217 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
218 cost of ashldi3 by 1.
219 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
220 (<shift>di3_neon): Likewise.
222 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
224 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
225 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
226 and (*aarch64_simd_mov<VQ:mode>).
227 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
229 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
230 attributes to match pattern alternative.
232 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
234 * config.gcc (powerpc*-*-*): Add emmintrin.h.
235 * config/rs6000/emmintrin.h: New file.
236 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
238 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
240 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
241 * (movdi_vfp_cortexa8): Remove pattern.
243 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
245 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
247 (Specific, alpha*-dec-osf5.1): Remove.
248 (Specific, mips-sgi-irix5): Remove.
249 (Specific, mips-sgi-irix6): Remove.
251 2017-10-30 Jakub Jelinek <jakub@redhat.com>
254 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
255 arguments to clear_bit_region_be.
257 2017-10-30 Jim Wilson <wilson@tuliptree.org>
259 * gimplify.c: Include memmodel.h.
261 2017-10-30 Martin Jambor <mjambor@suse.cz>
263 * omp-grid.c (grid_attempt_target_gridification): Also insert a
264 condition whether loop should be executed at all.
266 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
268 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
269 gimple folding of vec_madd() intrinsics.
270 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
271 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
272 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
274 2017-10-30 Richard Biener <rguenther@suse.de>
276 PR tree-optimization/82762
278 2017-10-23 Richard Biener <rguenther@suse.de>
280 PR tree-optimization/82129
282 2017-08-01 Richard Biener <rguenther@suse.de>
284 PR tree-optimization/81181
285 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
286 (compute_antic): ... end of iteration here.
288 2017-10-30 Joseph Myers <joseph@codesourcery.com>
290 * doc/invoke.texi (C Dialect Options): Document -std=c17,
291 -std=iso9899:2017 and -std=gnu17.
292 * doc/standards.texi (C Language): Document C17 support.
293 * doc/cpp.texi (Overview): Mention -std=c17.
294 (Standard Predefined Macros): Document C11 and C17 values of
295 __STDC_VERSION__. Do not refer to C99 support as incomplete.
296 * doc/extend.texi (Inline): Do not list individual options for
297 standards newer than C99.
298 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
300 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
303 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
305 * asan.c (asan_finish_file): Align asan globals array by shadow
308 2017-10-30 Jakub Jelinek <jakub@redhat.com>
311 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
312 (struct store_immediate_info): Add bitregion_start and bitregion_end
314 (store_immediate_info::store_immediate_info): Add brs and bre
315 arguments and initialize bitregion_{start,end} from those.
316 (struct merged_store_group): Add bitregion_start, bitregion_end,
317 align_base and mask fields. Drop unnecessary struct keyword from
318 struct store_immediate_info. Add do_merge method.
319 (clear_bit_region_be): Use memset instead of loop storing zeros.
320 (merged_store_group::do_merge): New method.
321 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
322 stores as long as the surrounding bitregions have no gaps.
323 (merged_store_group::merge_overlapping): Use do_merge.
324 (merged_store_group::apply_stores): Test that bitregion_{start,end}
325 is byte aligned, rather than requiring that start and width are
326 byte aligned. Drop unnecessary struct keyword from
327 struct store_immediate_info. Allocate and populate also mask array.
328 Make start of the arrays relative to bitregion_start rather than
329 start and size them according to bitregion_{end,start} difference.
330 (struct imm_store_chain_info): Drop unnecessary struct keyword from
331 struct store_immediate_info.
332 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
333 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
334 struct keyword from struct store_immediate_info.
335 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
336 between stores as long as the surrounding bitregions have no gaps.
338 (struct split_store): Add orig non-static data member.
339 (split_store::split_store): Initialize orig to false.
340 (find_constituent_stmts): Return store_immediate_info *, non-NULL
341 if there is exactly a single original stmt. Change stmts argument
342 to pointer from reference, if NULL, don't push anything to it. Add
343 first argument, use it to optimize skipping over orig stmts that
344 are known to be before bitpos already. Simplify.
345 (split_group): Return unsigned int count how many stores are or
346 would be needed rather than a bool. Add allow_unaligned argument.
347 Change split_stores argument from reference to pointer, if NULL,
348 only do a dry run computing how many stores would be produced.
349 Rewritten algorithm to use both alignment and misalign if
350 !allow_unaligned and handle bitfield stores with gaps.
351 (imm_store_chain_info::output_merged_store): Set start_byte_pos
352 from bitregion_start instead of start. Compute allow_unaligned
353 here, if true, do 2 split_group dry runs to compute which one
354 produces fewer stores and prefer aligned if equal. Punt if
355 new count is bigger or equal than original before emitting any
356 statements, rather than during that. Remove no longer needed
357 new_ssa_names tracking. Replace num_stmts with
358 split_stores.length (). Use 32-bit stack allocated entries
359 in split_stores auto_vec. Try to reuse original store lhs/rhs1
360 if possible. Handle bitfields with gaps.
361 (pass_store_merging::execute): Ignore bitsize == 0 stores.
362 Compute bitregion_{start,end} for the stores and construct
363 store_immediate_info with that. Formatting fixes.
365 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
368 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
369 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
371 2017-10-29 Jim Wilson <wilson@tuliptree.org>
373 * gimplify.c: Include tm_p.h.
375 * common.opt (gcoff): Re-add as ignored option.
376 (gcoff1, gcoff2, gcoff3): Likewise.
378 * Makefile.in (OBJS): Delete sdbout.o.
379 (GTFILES): Delete $(srcdir)/sdbout.c.
380 * debug.h: Delete sdb_debug_hooks.
381 * final.c: Delete sdbout.h include.
382 (final_scan_insn): Delete SDB_DEBUG check.
383 (rest_of_clean_state): Likewise.
384 * output.h: Delete sdb_begin_function_line.
387 * toplev.c: Delete sdbout.h include.
388 (process_options): Delete SDB_DEBUG check.
389 * tree-core.h (tree_type_common): Delete pointer field of
391 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
393 * tree.h (TYPE_SYMTAB_POINTER): Delete.
394 (TYPE_SYMTAB_IS_POINTER): Delete.
395 (TYPE_SYMTAB_IS_DIE): Renumber.
396 * xcoffout.c: Refer to former sdbout.c file.
397 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
399 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
400 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
401 (Debugging Options): Delete -gcoff.
402 (-gstabs): Delete SDB reference.
404 (-gcoff@var{level}): Delete.
405 * doc/passes.texi (Debugging information output): Delete SDB and
407 * doc/tm.texi: Regenerate.
408 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
409 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
411 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
412 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
413 (SDB_DEBUGGING_INFO): Delete.
414 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
415 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
416 * target.def (output_source_filename): Delete COFF reference.
418 * common.opt (gcoff): Delete.
419 (gxcoff+): Update Negative chain.
420 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
422 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
423 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
424 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
426 (expand_function_start): Change sdb reference to past tense.
427 (expand_function_end): Change sdb reference to past tense.
428 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
429 * opts.c (debug_type_names): Delete coff entry.
430 (common_handle_option): Delete OPT_gcoff case.
431 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
433 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
434 * config/cris/cris.h: Delete SDB reference in comment.
435 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
436 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
437 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
438 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
440 (ix86_expand_prologue): Likewise.
441 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
442 * config/ia64/ia64.h: Likewise.
443 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
444 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
446 * config/mmix/mmix.h: Likewise.
447 * config/nds32/nds32.c: Likewise.
448 * config/stormy/storym16.h: Likewise.
449 * config/visium/visium.h: Likewise.
450 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
452 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
454 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
455 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
456 FRAME_POINTER_REGNUM point at high end of local var area.
458 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
460 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
461 Move comment around. Do not reset best_edge for a copiable
462 destination if the copy would cause a partition change.
463 (better_edge_p): Remove redundant check.
465 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
467 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
469 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
471 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
472 for math functions that have _Float<N> and _Float<N>X variants.
473 (mathfn_built_in_2): Add support for math functions that have
474 _Float<N> and _Float<N>X variants.
475 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
476 (expand_builtin_mathfn_ternary): Add support for fma with
477 _Float<N> and _Float<N>X variants.
478 (expand_builtin): Likewise.
479 (fold_builtin_3): Likewise.
480 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
481 create math function _Float<N> and _Float<N>X variants as external
483 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
484 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
485 the __builtin_ prefix and if not strict ansi, without the prefix.
486 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
487 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
488 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
489 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
490 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
491 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
492 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
493 function signatures for fma _Float<N> and _Float<N>X variants.
494 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
495 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
496 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
497 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
498 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
499 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
500 * gencfn-macros.c (print_case_cfn): Add support for math functions
501 that have _Float<N> and _Float<N>X variants.
502 (print_define_operator_list): Likewise.
503 (fltfn_suffixes): Likewise.
505 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
506 for math functions that have _Float<N> and _Float<N>X variants.
507 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
508 and _Float<N>X variants.
509 (COPYSIGN): Likewise.
512 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
513 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
515 (integer_valued_read_call_p): Likewise.
516 * fold-const-call.c (fold_const_call_ss): Likewise.
517 (fold_const_call_sss): Add support for copysign, fmin, and fmax
518 _Float<N> and _Float<N>X variants.
519 (fold_const_call_ssss): Add support for fma _Float<N> and
521 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
522 support for copysign and fma _Float<N> and _Float<N>X variants.
523 (backprop::process_builtin_call_use): Likewise.
524 * tree-call-cdce.c (can_test_argument_range); Add support for
525 sqrt _Float<N> and _Float<N>X variants.
526 (edom_only_function): Likewise.
527 (get_no_error_domain): Likewise.
528 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
529 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
530 copysign _Float<N> and _Float<N>X variants.
531 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
532 handled by machine independent code.
534 * doc/cpp.texi (Common Predefined Macros): Document defining
535 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
536 fma _Float<N> and _Float<N>X variants.
538 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
541 * config/i386/i386-modes.def (CCFPU): Remove definition.
542 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
543 (ix86_cc_modes_compatible): Ditto.
544 (ix86_expand_carry_flag_compare): Ditto.
545 (ix86_expand_int_movcc): Ditto.
546 (ix86_expand_int_addcc): Ditto.
547 (ix86_reverse_condition): Ditto.
548 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
549 Return true/false for unordered/ordered fp comparisons.
550 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
551 (ix86_prepare_fp_compare_args): Update for rename.
552 (ix86_expand_fp_compare): Update for rename. Generate unordered
553 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
554 (ix86_expand_sse_compare_and_jump): Ditto.
555 * config/i386/predicates.md (fcmov_comparison_operator):
556 Remove CCFPU mode handling.
557 (ix86_comparison_operator): Ditto.
558 (ix86_carry_flag_operator): Ditto.
559 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
560 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
561 (*cmpu<mode>_cc_i387): Ditto.
562 (FPCMP): Remove mode iterator.
563 (unord): Remove mode attribute.
564 (unord_subst): New define_subst transformation
565 (unord): New define_subst attribute.
567 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
568 (*cmpi<unord>xf_i387): Ditto.
569 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
570 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
571 using unord_subst transformation.
572 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
573 (round_saeonly): Also handle CCFP mode.
574 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
575 Remove UNSPEC_SAHF unspec handling.
577 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
579 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
581 2017-10-27 Jeff Law <law@redhat.com>
583 * gimple-ssa-sprintf.c: Include domwalk.h.
584 (class sprintf_dom_walker): New class, derived from dom_walker.
585 (sprintf_dom_walker::before_dom_children): New function.
586 (struct call_info): Moved into sprintf_dom_walker class
587 (compute_formath_length, handle_gimple_call): Likewise.
588 (sprintf_length::execute): Call the dominator walker rather
589 than walking the statements.
591 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
592 gimple statement locations.
593 (check_array_bounds): Corresponding changes. Get the statement's
594 location directly from wi->stmt.
596 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
599 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
601 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
603 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
604 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
606 2017-10-27 Jakub Jelinek <jakub@redhat.com>
609 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
610 * config/i386/i386.c (maybe_get_pool_constant): Removed.
611 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
612 maybe_get_pool_constant.
613 * config/i386/predicates.md (zero_extended_scalar_load_operand):
616 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
618 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
619 2.26 caveat. Update gas and gld versions.
620 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
623 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
625 * cgraph.h (set_malloc_flag): Declare.
626 * cgraph.c (set_malloc_flag_1): New function.
627 (set_malloc_flag): Likewise.
628 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
629 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
631 (read_ipa_call_summary): Add support for reading is_return_callee.
632 (write_ipa_call_summary): Stream is_return_callee.
633 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
634 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
635 ipa-prop.h, ipa-fnsummary.h.
636 (pure_const_names): Change to static.
637 (malloc_state_e): Define.
638 (malloc_state_names): Define.
639 (funct_state_d): Add field malloc_state.
640 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
641 (check_retval_uses): New function.
642 (malloc_candidate_p): Likewise.
643 (analyze_function): Add support for malloc attribute.
644 (pure_const_write_summary): Stream malloc_state.
645 (pure_const_read_summary): Add support for reading malloc_state.
646 (dump_malloc_lattice): New function.
647 (propagate_malloc): New function.
648 (warn_function_malloc): New function.
649 (ipa_pure_const::execute): Call propagate_malloc and
651 (pass_local_pure_const::execute): Add support for malloc attribute.
652 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
653 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
655 2017-10-27 Martin Liska <mliska@suse.cz>
657 PR gcov-profile/82457
658 * doc/invoke.texi: Document that one needs a non-strict ISO mode
659 for fork-like functions to be properly instrumented.
661 2017-10-27 Richard Biener <rguenther@suse.de>
664 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
665 info when we redirected EH.
667 2017-10-26 Michael Collison <michael.collison@arm.com>
669 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
671 (<optab>_trunchf<GPI:mode>2: New pattern.
672 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
673 * config/aarch64/iterators.md (wv): New mode attribute.
674 (vf, VF): New mode attributes.
675 (vgp, VGP): New mode attributes.
676 (s): Update attribute with SImode and DImode prefixes.
678 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
680 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
681 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
682 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
683 (nios2_option_overide): Initialize it. Don't allow R0-relative
685 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
686 (nios2_symbolic_constant_p): Likewise.
687 (nios2_legitimate_address_p): Likewise.
688 (nios2_r0rel_section_name_p): New.
689 (nios2_symbol_ref_in_r0rel_data_p): New.
690 (nios2_emit_move_sequence): Handle r0rel_constant_p.
691 (r0rel_constant_p): New.
692 (nios2_print_operand_address): Handle r0rel_constant_p.
693 (nios2_cdx_narrow_form_p): Likewise.
694 * config/nios2/nios2.opt (mr0rel-sec=): New option.
695 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
696 (Nios II Options): Document -mr0rel-sec.
698 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
700 * config/nios2/nios2.c: Include xregex.h.
701 (nios2_gprel_sec_regex): New.
702 (nios2_option_overide): Initialize it. Don't allow GP-relative
704 (nios2_small_section_name_p): Check for regex match.
705 * config/nios2/nios2.opt (mgprel-sec=): New option.
706 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
707 (Nios II Options): Document -mgprel-sec.
709 2017-10-26 Jim Wilson <wilson@tuliptree.org>
711 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
713 2017-10-26 Tom de Vries <tom@codesourcery.com>
715 PR tree-optimization/82707
716 * gimple.c (gimple_copy): Fix unsharing of
717 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
719 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
721 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
722 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
723 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
724 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
725 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
726 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
727 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
728 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
729 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
730 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
731 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
732 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
733 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
734 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
735 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
736 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
737 _mm512_mask_cmpunord_ps_mask): New intrinsics.
739 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
741 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
743 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
744 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
745 warning to rs6000.c. Remove the Undocumented flag, since it has
747 (-mabi=ibmlongdouble): Likewise.
748 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
749 already set, set the default format for long double.
750 (rs6000_debug_reg_global): Print whether long double is IBM or
752 (rs6000_option_override_internal): Rework setting long double
753 format. Only warn if the user is changing the long double default
754 and they did not use -Wno-psabi.
755 * doc/invoke.texi (PowerPC options): Update the documentation for
756 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
758 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
759 Alan Hayward <alan.hayward@arm.com>
760 David Sherwood <david.sherwood@arm.com>
762 * rtl.h (wider_subreg_mode): New function.
763 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
764 rather than an unsigned int *.
765 * ira-color.c (regno_max_ref_width): Replace with...
766 (regno_max_ref_mode): ...this new variable.
767 (coalesced_pseudo_reg_slot_compare): Update accordingly.
768 Use wider_subreg_mode.
769 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
770 rather than an unsigned int *.
771 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
772 (process_alt_operands): Likewise.
773 (invariant_p): Likewise.
774 * lra-spills.c (assign_mem_slot): Likewise.
775 (add_pseudo_to_slot): Likewise.
776 * lra.c (collect_non_operand_hard_regs): Likewise.
777 (add_regs_to_insn_regno_info): Likewise.
778 * reload1.c (regno_max_ref_width): Replace with...
779 (regno_max_ref_mode): ...this new variable.
780 (reload): Update accordingly. Update call to
781 ira_sort_regnos_for_alter_reg.
782 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
783 (init_eliminable_invariants): Update to use regno_max_ref_mode.
784 (scan_paradoxical_subregs): Likewise.
786 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
788 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
789 (aarch64_frame): Add emit_frame_chain boolean.
790 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
791 Move eh_return case to aarch64_layout_frame.
792 (aarch64_layout_frame): Initialize emit_frame_chain.
793 (aarch64_expand_prologue): Use emit_frame_chain.
795 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
797 * config/aarch64/aarch64.c (aarch64_layout_frame):
798 Ensure LR is always stored at the bottom of the callee-saves.
799 Remove rarely used frame layout which saves callee-saves at top of
800 frame, so the store of LR can be used as a valid probe in all cases.
802 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
804 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
805 Improve unaligned TImode/TFmode base/offset split.
807 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
808 Alan Hayward <alan.hayward@arm.com>
809 David Sherwood <david.sherwood@arm.com>
811 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
812 * combine.c (find_single_use_1): Likewise.
813 (expand_field_assignment): Likewise.
814 (move_deaths): Likewise.
815 * lra-constraints.c (simplify_operand_subreg): Likewise.
816 (curr_insn_transform): Likewise.
817 * lra.c (collect_non_operand_hard_regs): Likewise.
818 (add_regs_to_insn_regno_info): Likewise.
819 * rtlanal.c (reg_referenced_p): Likewise.
820 (covers_regno_no_parallel_p): Likewise.
822 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
824 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
825 Don't print any bits outside the precision of the value.
826 * wide-int.cc (test_printing): Add some new tests.
828 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
830 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
831 supports -xbrace_comment option.
832 * configure: Regenerate.
833 * config.in: Regenerate.
834 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
835 (ASM_CPU_SPEC): Use it.
837 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
839 * target.def (static_rtx_alignment): New hook.
840 * targhooks.h (default_static_rtx_alignment): Declare.
841 * targhooks.c (default_static_rtx_alignment): New function.
842 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
843 * doc/tm.texi: Regenerate.
844 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
845 instead of targetm.constant_alignment. Remove call to
847 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
848 (cris_preferred_mininum_alignment): New function, split out from...
849 (cris_constant_alignment): ...here.
850 (cris_static_rtx_alignment): New function.
851 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
853 (ix86_constant_alignment): ...here.
854 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
855 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
856 (mmix_static_rtx_alignment): New function.
857 * config/spu/spu.c (spu_static_rtx_alignment): New function.
858 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
860 2017-10-26 Tamar Christina <tamar.christina@arm.com>
863 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
864 Add flag_trapping_math and flag_fp_int_builtin_inexact.
866 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
868 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
869 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
870 mark as a sign-extending load.
871 (local_pic_load_u): Define.
873 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
876 * fold-const.c (operand_equal_for_comparison_p): Also return true
877 if ARG0 is a simple variant of ARG1 with narrower precision.
878 (fold_ternary_loc): Always pass unstripped operands to the predicate.
880 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
882 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
884 * i386.h (processor_costs): Add gather_static, gather_per_elt,
885 scatter_static, scatter_per_elt.
886 * x86-tune-costs.h: Add new cost entries.
888 2017-10-25 Richard Biener <rguenther@suse.de>
890 * tree-ssa-sccvn.h (vn_eliminate): Declare.
891 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
892 class pass_fre): Move to ...
893 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
894 class pass_fre): ... here and adjust for statistics.
896 2017-10-25 Jakub Jelinek <jakub@redhat.com>
899 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
900 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
901 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
902 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
905 2017-10-25 Richard Biener <rguenther@suse.de>
907 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
908 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
909 eliminate_push_avail, eliminate_insert): Move inside...
910 (class eliminate_dom_walker): ... this class in preparation
912 (fini_eliminate): Remove by merging with ...
913 (eliminate): ... this function. Adjust for class changes.
914 (pass_pre::execute): Remove fini_eliminate call.
915 (pass_fre::execute): Likewise.
917 2017-10-24 Jakub Jelinek <jakub@redhat.com>
920 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
921 (VPERMI2, VPERMI2I): New mode iterators.
922 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
923 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
925 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
926 mode iterator. Remove 3 old define_insn patterns.
927 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
928 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
929 VPERMI2 mode iterator, remove the other two expanders.
930 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
931 to use VPERMI2 mode iterator, add another alternative for vpermi2*
932 instructions, remove the other two patterns.
933 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
934 mode iterator, remove the other two patterns.
935 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
936 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
937 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
938 and adjust argument order accordingly.
939 (ix86_expand_vec_perm): Adjust caller.
940 (expand_vec_perm_1): Likewise.
941 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
942 (expand_vec_perm_vpermt2_vpshub2): ... this.
943 (ix86_expand_vec_perm_const_1): Adjust caller.
944 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
947 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
948 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
949 (vec_shl_<mode>): Remove unused expander.
950 (avx512bw_<shift_insn><mode>3): New define_insn.
951 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
952 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
954 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
957 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
960 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
962 PR rtl-optimization/82396
963 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
964 (autopref_multipass_init): Simplify initialization.
965 (autopref_rank_data): Simplify sort order.
966 * gcc/sched-int.h (autopref_multipass_data_): Remove
967 multi_mem_insn_p, min_offset and max_offset.
969 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
972 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
973 Check special value of flag_omit_frame_pointer.
974 (aarch64_can_eliminate): Likewise.
975 (aarch64_override_options_after_change_1): Simplify handling of
976 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
978 2017-10-24 Richard Biener <rguenther@suse.de>
980 PR tree-optimization/82697
981 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
982 zero for conditional load and unconditional store.
984 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
986 * doc/install.texi: Document bootstrap-cet.
988 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
991 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
992 ENDBR instruction at function entrance if function is only
995 2017-10-24 Jakub Jelinek <jakub@redhat.com>
998 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
999 patterns to better describe from which operation the CF is computed.
1000 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
1001 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
1002 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
1003 is 0, use _0 suffixed expanders instead of emitting a comparison
1006 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
1008 * config/i386/i386.md(*movsf_internal, *movdf_internal):
1009 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
1011 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
1014 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
1015 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
1016 * loop-iv.c (iv_get_reaching_def): Likewise.
1017 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
1018 variable is promoted and the partition contains undefined values.
1020 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1022 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
1024 (nios2_address_cost): Define.
1025 (nios2_legitimize_address): Recognize (exp + constant) directly.
1026 (TARGET_ADDRESS_COST): Define.
1028 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1030 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
1031 (nios2_symbolic_memory_operand_p): Declare.
1032 (nios2_split_large_constant): Declare.
1033 (nios2_split_symbolic_memory_operand): Declare.
1034 * config/nios2/nios2.c: Adjust includes.
1035 (nios2_symbolic_constant_allowed): New.
1036 (nios2_symbolic_constant_p): New.
1037 (nios2_plus_symbolic_constant_p): New.
1038 (nios2_valid_addr_expr_p): Recognize addresses involving
1040 (nios2_legitimate_address_p): Likewise, also LO_SUM.
1041 (nios2_symbolic_memory_operand_p): New.
1042 (nios2_large_constant_p): New.
1043 (nios2_split_large_constant): New.
1044 (nios2_split_plus_large_constant): New.
1045 (nios2_split_symbolic_memory_operand): New.
1046 (nios2_legitimize_address): Code refactoring. Handle addresses
1047 involving symbolic constants.
1048 (nios2_emit_move_sequence): Likewise.
1049 (nios2_print_operand): Improve error output.
1050 (nios2_print_operand_address): Handle LO_SUM.
1051 (nios2_cdx_narrow_form_p): Likewise.
1052 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
1053 operands involving symbolic constants.
1054 (movhi_internal, movsi_internal): Likewise.
1055 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
1056 (extendhisi2, extendqi<mode>2): Likewise.
1058 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1060 * tree-pass.h (PROP_rtl_split_insns): Define.
1061 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
1063 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1065 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
1067 2017-10-23 Jakub Jelinek <jakub@redhat.com>
1070 * target.def (const_not_ok_for_debug_p): Default to
1071 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
1072 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
1073 * targhooks.c (default_const_not_ok_for_debug_p): New function.
1074 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
1075 which targetm.const_not_ok_for_debug_p returned true.
1076 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
1078 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
1080 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
1081 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
1082 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
1084 (ix86_const_not_ok_for_debug_p): New function.
1085 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
1086 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
1088 2017-10-23 David Malcolm <dmalcolm@redhat.com>
1091 * system.h: Conditionally include "unique-ptr.h" if
1092 INCLUDE_UNIQUE_PTR is defined.
1093 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
1094 of defining INCLUDE_UNIQUE_PTR before including "system.h".
1096 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
1098 * config/rl78/rl78.md: New define_expand "subdi3".
1100 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
1103 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
1104 DF_REF_INSN if DF_REF_INSN_INFO is false.
1106 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
1108 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
1109 xmm_move instead of sse_move.
1110 (sse_store_index): New function.
1111 (ix86_register_move_cost): Be more sensible about mismatch stall;
1112 model AVX moves correctly; make difference between sse->integer and
1114 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
1115 moves; make difference between SSE and AVX.
1116 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
1117 and zmm_move. Increase size of sse load and store tables;
1118 add unaligned load and store tables; add ssemmx_to_integer.
1119 * x86-tune-costs.h: Update all entries according to real
1120 move latencies from Agner Fog's manual and chip documentation.
1122 2017-10-23 Jakub Jelinek <jakub@redhat.com>
1125 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
1126 * config/i386/constraints.md (Wf): New constraint.
1127 * config/i386/i386.md (UNSPEC_SBB): New unspec.
1128 (cmp<dwi>_doubleword): Removed.
1129 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
1130 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
1131 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
1132 expand with cmp<dwi>_doubleword. For LTU and GEU use
1133 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
1135 * common.opt (gcolumn-info): Enable by default.
1136 * doc/invoke.texi (gcolumn-info): Document new default.
1138 2017-10-23 Richard Biener <rguenther@suse.de>
1140 PR tree-optimization/82672
1141 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
1142 Fold the stmt if we propagated into it.
1144 2017-10-23 Richard Biener <rguenther@suse.de>
1146 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
1147 (bitmap_remove_expr_from_set): ... this. All callers call this
1148 for non-constant values.
1149 (bitmap_set_subtract): Rename to...
1150 (bitmap_set_subtract_expressions): ... this. Adjust and
1152 (bitmap_set_contains_value): Remove superfluous check.
1153 (bitmap_set_replace_value): Inline into single caller ...
1154 (bitmap_value_replace_in_set): ... here and simplify.
1155 (dependent_clean): Merge into ...
1156 (clean): ... this using an overload. Adjust.
1157 (prune_clobbered_mems): Adjust.
1158 (compute_antic_aux): Likewise.
1159 (compute_partial_antic_aux): Likewise.
1161 2017-10-23 Richard Biener <rguenther@suse.de>
1163 PR tree-optimization/82129
1165 2017-08-01 Richard Biener <rguenther@suse.de>
1167 PR tree-optimization/81181
1168 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
1169 (compute_antic): ... end of iteration here.
1171 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1173 * target.def (starting_frame_offset): New hook.
1174 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
1175 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
1176 * doc/tm.texi.in: Regenerate.
1177 * hooks.h (hook_hwi_void_0): Declare.
1178 * hooks.c (hook_hwi_void_0): New function.
1179 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
1180 STARTING_FRAME_OFFSET.
1181 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
1182 * reload1.c (reload): Likewise.
1183 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
1184 instead of STARTING_FRAME_OFFSET.
1185 * function.c (try_fit_stack_local): Likewise.
1186 (assign_stack_local_1): Likewise
1187 (instantiate_virtual_regs): Likewise.
1188 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
1189 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
1190 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
1191 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
1192 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
1193 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
1194 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
1195 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
1196 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
1197 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
1198 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
1199 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
1200 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
1201 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
1202 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
1203 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
1204 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
1205 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
1206 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
1207 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
1208 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
1209 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
1210 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
1211 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
1212 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
1213 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
1214 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
1215 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
1216 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
1217 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
1218 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
1219 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
1220 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
1221 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
1222 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
1223 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
1224 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
1225 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
1226 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
1227 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
1228 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
1229 * config/avr/avr.c (avr_starting_frame_offset): Make static and
1230 return a HOST_WIDE_INT.
1231 (avr_builtin_setjmp_frame_value): Use it instead of
1232 STARTING_FRAME_OFFSET.
1233 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1234 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
1235 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
1237 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1238 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
1239 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
1240 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1241 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
1242 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
1243 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1244 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
1245 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
1246 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1247 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
1248 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
1250 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1251 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
1252 * config/mips/mips.c (mips_compute_frame_info): Refer to
1253 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
1254 (mips_starting_frame_offset): New function.
1255 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1256 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
1257 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
1258 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
1259 and return a HOST_WIDE_INT.
1260 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1261 (mmix_initial_elimination_offset): Refer to
1262 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
1263 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
1264 * config/pa/pa.c (pa_starting_frame_offset): New function.
1265 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
1266 (pa_expand_prologue): Likewise.
1267 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1268 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
1269 !FRAME_GROWS_DOWNWARD handling to...
1270 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1271 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
1272 !FRAME_GROWS_DOWNWARD handling to...
1273 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1274 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
1275 !FRAME_GROWS_DOWNWARD handling to...
1276 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1277 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
1279 (rs6000_starting_frame_offset): New function.
1280 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
1281 !FRAME_GROWS_DOWNWARD handling to...
1282 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1283 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
1284 !FRAME_GROWS_DOWNWARD handling to...
1285 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1286 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
1287 !FRAME_GROWS_DOWNWARD handling to...
1288 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1289 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
1290 (rs6000_starting_frame_offset): New function.
1291 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
1292 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
1293 * config/vax/vax.c (vax_starting_frame_offset): New function.
1294 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
1295 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1296 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
1297 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
1298 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1299 * system.h (STARTING_FRAME_OFFSET): Poison.
1301 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1303 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
1304 SCALAR_TYPE_MODE instead of TYPE_MODE.
1306 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1307 Alan Hayward <alan.hayward@arm.com>
1308 David Sherwood <david.sherwood@arm.com>
1310 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
1312 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1313 Alan Hayward <alan.hayward@arm.com>
1314 David Sherwood <david.sherwood@arm.com>
1316 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
1318 2017-10-23 Richard Biener <rguenther@suse.de>
1320 PR tree-optimization/82129
1321 * tree-ssa-pre.c (bitmap_set_and): Remove.
1322 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
1323 canonicalizing expressions in the set to those with lowest
1324 ID rather than taking that from the first edge.
1326 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1328 * combine.c (rtx_equal_for_field_assignment_p): Use
1329 byte_lowpart_offset.
1331 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1332 Alan Hayward <alan.hayward@arm.com>
1333 David Sherwood <david.sherwood@arm.com>
1335 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
1336 to a SUBREG_PROMOTED_VAR.
1338 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1339 Alan Hayward <alan.hayward@arm.com>
1340 David Sherwood <david.sherwood@arm.com>
1342 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
1343 (expand_debug_source_expr): Likewise.
1344 * combine.c (combine_simplify_rtx): Likewise.
1345 * cse.c (fold_rtx): Likewise.
1346 * optabs.c (expand_float): Likewise.
1347 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1348 (simplify_binary_operation_1): Likewise.
1350 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1351 Alan Hayward <alan.hayward@arm.com>
1352 David Sherwood <david.sherwood@arm.com>
1354 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
1355 (record_promoted_value): Likewise.
1356 * expr.c (expand_expr_real_2): Likewise.
1357 * ree.c (update_reg_equal_equiv_notes): Likewise.
1358 (combine_set_extension): Likewise.
1359 * rtlanal.c (low_bitmask_len): Likewise.
1360 * simplify-rtx.c (neg_const_int): Likewise.
1361 (simplify_binary_operation_1): Likewise.
1363 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1364 Alan Hayward <alan.hayward@arm.com>
1365 David Sherwood <david.sherwood@arm.com>
1367 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
1368 * regcprop.c (maybe_mode_change): Likewise.
1369 * reload1.c (alter_reg): Likewise.
1371 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1373 * inchash.h (inchash::hash::add_wide_int): New function.
1374 * lto-streamer-out.c (hash_tree): Use it.
1376 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1378 * inchash.h (inchash::hash::add_wide_int): Rename to...
1379 (inchash::hash::add_hwi): ...this.
1380 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
1381 (polymorphic_call_target_hasher::hash): Likewise.
1382 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
1383 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
1384 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
1385 * lto-streamer-out.c (hash_tree): Likewise.
1386 * optc-save-gen.awk: Likewise.
1387 * tree.c (add_expr): Likewise.
1389 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
1392 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
1393 for ordered inequality comparisons even with TARGET_IEEE_FP.
1395 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
1398 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
1399 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
1400 Expand with cmp<dwi>_doubleword.
1402 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1404 * extend.texi: Add x86 specific to 'nocf_check' attribute.
1405 List CET intrinsics.
1406 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
1407 specific to -fcf-protection option.
1409 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1411 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
1412 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
1413 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
1414 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
1415 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
1416 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
1417 (extra_objs): Add cet.o for Linux/x86 targets.
1418 (tmake_file): Add i386/t-cet for Linux/x86 targets.
1419 * config/i386/cet.c: New file.
1420 * config/i386/cetintrin.h: Likewise.
1421 * config/i386/t-cet: Likewise.
1422 * config/i386/cpuid.h (bit_SHSTK): New.
1423 (bit_IBT): Likewise.
1424 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
1425 pass IBT and SHSTK bits.
1426 * config/i386/i386-builtin-types.def
1427 (VOID_FTYPE_UNSIGNED_PVOID): New.
1428 (VOID_FTYPE_UINT64_PVOID): Likewise.
1429 * config/i386/i386-builtin.def: Add CET intrinsics.
1430 * config/i386/i386-c.c (ix86_target_macros_internal): Add
1431 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
1432 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
1433 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
1435 * config/i386/i386.c (rest_of_insert_endbranch): New.
1436 (pass_data_insert_endbranch): Likewise.
1437 (pass_insert_endbranch): Likewise.
1438 (make_pass_insert_endbranch): Likewise.
1439 (ix86_notrack_prefixed_insn_p): Likewise.
1440 (ix86_target_string): Add -mibt, -mshstk flags.
1441 (ix86_option_override_internal): Add flag_cf_protection
1443 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
1444 (ix86_print_operand): Add 'notrack' prefix output.
1445 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
1446 (ix86_expand_builtin): Expand CET intrinsics.
1447 (x86_output_mi_thunk): Add 'endbranch' instruction.
1448 * config/i386/i386.h (TARGET_IBT): New.
1449 (TARGET_IBT_P): Likewise.
1450 (TARGET_SHSTK): Likewise.
1451 (TARGET_SHSTK_P): Likewise.
1452 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
1453 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
1454 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
1455 (builtin_setjmp_setup): New pattern.
1456 (builtin_longjmp): Likewise.
1457 (rdssp<mode>): Likewise.
1458 (incssp<mode>): Likewise.
1459 (saveprevssp): Likewise.
1460 (rstorssp): Likewise.
1461 (wrss<mode>): Likewise.
1462 (wruss<mode>): Likewise.
1463 (setssbsy): Likewise.
1464 (clrssbsy): Likewise.
1465 (nop_endbr): Likewise.
1466 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
1468 * config/i386/immintrin.h: Include <cetintrin.h>.
1469 * config/i386/linux-common.h
1470 (file_end_indicate_exec_stack_and_cet): New prototype.
1471 (TARGET_ASM_FILE_END): New.
1473 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
1475 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
1476 latencies instead of having separate table; make difference between
1477 integer and float costs.
1478 * i386.h (processor_costs): Remove scalar_stmt_cost,
1479 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
1480 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
1482 * x86-tune-costs.h: Remove entries which has been removed in
1483 procesor_costs from all tables; make cond_taken_branch_cost
1484 and cond_not_taken_branch_cost COST_N_INSNS based.
1486 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
1488 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
1490 2017-10-20 Jakub Jelinek <jakub@redhat.com>
1492 * config/i386/i386.md (isa): Remove fma_avx512f.
1493 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
1494 <avx512>_fmadd_<mode>_mask3<round_name>,
1495 <avx512>_fmsub_<mode>_mask<round_name>,
1496 <avx512>_fmsub_<mode>_mask3<round_name>,
1497 <avx512>_fnmadd_<mode>_mask<round_name>,
1498 <avx512>_fnmadd_<mode>_mask3<round_name>,
1499 <avx512>_fnmsub_<mode>_mask<round_name>,
1500 <avx512>_fnmsub_<mode>_mask3<round_name>,
1501 <avx512>_fmaddsub_<mode>_mask<round_name>,
1502 <avx512>_fmaddsub_<mode>_mask3<round_name>,
1503 <avx512>_fmsubadd_<mode>_mask<round_name>,
1504 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
1505 (*vec_widen_umult_even_v16si<mask_name>,
1506 *vec_widen_smult_even_v16si<mask_name>): Likewise.
1507 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
1509 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1511 * extend.texi: Add 'nocf_check' documentation.
1512 * gimple.texi: Add second parameter to
1513 gimple_build_call_from_tree.
1514 * invoke.texi: Add -fcf-protection documentation.
1515 * rtl.texi: Add REG_CALL_NOTRACK documenation.
1517 2017-10-20 Richard Biener <rguenther@suse.de>
1519 PR tree-optimization/82473
1520 * tree-vect-loop.c (vectorizable_reduction): Properly get at
1521 the largest input type.
1523 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1525 * c-attribs.c (handle_nocf_check_attribute): New function.
1526 (c_common_attribute_table): Add 'nocf_check' handling.
1527 * gimple-parser.c: Add second argument NULL to
1528 gimple_build_call_from_tree.
1529 * attrib.c (comp_type_attributes): Check nocf_check attribute.
1530 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
1532 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
1533 * common.opt: Add fcf-protection flag.
1534 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
1535 * flag-types.h: Add enum cf_protection_level.
1536 * gimple.c (gimple_build_call_from_tree): Add second parameter.
1537 Add 'nocf_check' attribute propagation to gimple call.
1538 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
1539 (gimple_build_call_from_tree): Update prototype.
1540 (gimple_call_nocf_check_p): New function.
1541 (gimple_call_set_nocf_check): Likewise.
1542 * gimplify.c: Add second argument to gimple_build_call_from_tree.
1543 * ipa-icf.c: Add nocf_check attribute in statement hash.
1544 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
1545 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
1546 * toplev.c (process_options): Add flag_cf_protection handling.
1548 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
1550 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
1552 2017-10-20 Richard Biener <rguenther@suse.de>
1554 PR tree-optimization/82603
1555 * tree-if-conv.c (predicate_mem_writes): Make sure to only
1556 remove false predicated stores.
1558 2017-10-20 Richard Biener <rguenther@suse.de>
1560 * graphite-isl-ast-to-gimple.c
1561 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
1562 Remove return value and simplify, dump copied stmt after lhs
1564 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
1565 Reduce dump verbosity.
1566 (gsi_insert_earliest): Likewise.
1567 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
1568 * graphite.c (print_global_statistics): Adjust dumping.
1569 (print_graphite_scop_statistics): Likewise.
1570 (print_graphite_statistics): Do not dump loops here.
1571 (graphite_transform_loops): But here.
1573 2017-10-20 Nicolas Roche <roche@adacore.com>
1575 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
1576 * configure: Regenerate.
1578 2017-10-20 Jakub Jelinek <jakub@redhat.com>
1581 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
1582 functions when optimizing replace GIMPLE_RETURN stmts with
1583 calls to __builtin_unreachable ().
1586 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
1587 for -fsanitize=thread link of executables.
1588 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
1589 link of executables.
1592 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
1594 (<shift_insn><mode>3<mask_name>): Change the last of the 3
1595 define_insns for logical vector shifts to use VI248_AVX512BW
1596 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
1597 condition, useless isa and prefix attributes. Change the first
1598 2 of these define_insns to ...
1599 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
1600 define_insn for avx512vl.
1601 (<shift_insn><mode>3): ... and this, new define_insn without
1602 masking for non-avx512vl.
1605 * config/i386/sse.md (*andnot<mode>3,
1606 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
1607 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
1608 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
1609 not applied use empty suffix even for TARGET_AVX512VL.
1610 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
1611 is applied, supply evex,evex or evex,evex,evex instead of just
1614 2017-10-20 Julia Koval <julia.koval@intel.com>
1616 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
1617 (OPTION_MASK_ISA_GFNI_UNSET): New.
1618 (ix86_handle_option): Handle OPT_mgfni.
1619 * config/i386/cpuid.h (bit_GFNI): New.
1620 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
1621 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
1622 * config/i386/i386.c (ix86_target_string): Add -mgfni.
1623 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
1624 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
1625 * config/i386/i386.opt: Add mgfni.
1627 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
1629 * config/msp430/msp430.c (msp430_option_override): Disable
1630 -fdelete-null-pointer-checks.
1631 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
1633 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
1635 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
1636 of x87 and SSE instructions.
1638 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
1640 * asan.c (create_cond_insert_point): Do not update edge count.
1641 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
1642 (afdo_propagate_circuit): Likewise.
1643 (afdo_calculate_branch_prob): Likewise.
1644 (afdo_annotate_cfg): Likewise.
1645 * basic-block.h (struct edge_def): Remove count.
1646 (edge_def::count): New accessor.
1647 * bb-reorder.c (rotate_loop): Update.
1648 (find_traces_1_round): Update.
1649 (connect_traces): Update.
1650 (sanitize_hot_paths): Update.
1651 * cfg.c (unchecked_make_edge): Update.
1652 (make_single_succ_edge): Update.
1653 (check_bb_profile): Update.
1654 (dump_edge_info): Update.
1655 (update_bb_profile_for_threading): Update.
1656 (scale_bbs_frequencies_int): Update.
1657 (scale_bbs_frequencies_gcov_type): Update.
1658 (scale_bbs_frequencies_profile_count): Update.
1659 (scale_bbs_frequencies): Update.
1660 * cfganal.c (connect_infinite_loops_to_exit): Update.
1661 * cfgbuild.c (compute_outgoing_frequencies): Update.
1662 (find_many_sub_basic_blocks): Update.
1663 * cfgcleanup.c (try_forward_edges): Update.
1664 (try_crossjump_to_edge): Update
1665 * cfgexpand.c (expand_gimple_cond): Update
1666 (expand_gimple_tailcall): Update
1667 (construct_exit_block): Update
1668 * cfghooks.c (verify_flow_info): Update
1669 (redirect_edge_succ_nodup): Update
1670 (split_edge): Update
1671 (make_forwarder_block): Update
1672 (duplicate_block): Update
1673 (account_profile_record): Update
1674 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
1675 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
1676 * cfgloopmanip.c (scale_loop_profile): Update.
1678 (lv_adjust_loop_entry_edge): Update.
1679 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
1680 (force_nonfallthru_and_redirect): Update.
1681 (purge_dead_edges): Update.
1682 (rtl_flow_call_edges_add): Update.
1683 * cgraphunit.c (init_lowered_empty_function): Update.
1684 (cgraph_node::expand_thunk): Update.
1685 * gimple-pretty-print.c (dump_probability): Update.
1686 (dump_edge_probability): Update.
1687 * gimple-ssa-isolate-paths.c (isolate_path): Update.
1688 * haifa-sched.c (sched_create_recovery_edges): Update.
1689 * hsa-gen.c (convert_switch_statements): Update.
1690 * ifcvt.c (dead_or_predicable): Update.
1691 * ipa-inline-transform.c (inline_transform): Update.
1692 * ipa-split.c (split_function): Update.
1693 * ipa-utils.c (ipa_merge_profiles): Update.
1694 * loop-doloop.c (add_test): Update.
1695 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
1696 * lto-streamer-in.c (input_cfg): Update.
1697 (input_function): Update.
1698 * lto-streamer-out.c (output_cfg): Update.
1699 * modulo-sched.c (sms_schedule): Update.
1700 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
1701 * predict.c (maybe_hot_edge_p): Update.
1702 (unlikely_executed_edge_p): Update.
1703 (probably_never_executed_edge_p): Update.
1704 (dump_prediction): Update.
1705 (drop_profile): Update.
1706 (propagate_unlikely_bbs_forward): Update.
1707 (determine_unlikely_bbs): Update.
1708 (force_edge_cold): Update.
1709 * profile.c (compute_branch_probabilities): Update.
1710 * reg-stack.c (better_edge): Update.
1711 * shrink-wrap.c (handle_simple_exit): Update.
1712 * tracer.c (better_p): Update.
1713 * trans-mem.c (expand_transaction): Update.
1714 (split_bb_make_tm_edge): Update.
1715 * tree-call-cdce.c: Update.
1716 * tree-cfg.c (gimple_find_sub_bbs): Update.
1717 (gimple_split_edge): Update.
1718 (gimple_duplicate_sese_region): Update.
1719 (gimple_duplicate_sese_tail): Update.
1720 (gimple_flow_call_edges_add): Update.
1721 (insert_cond_bb): Update.
1722 (execute_fixup_cfg): Update.
1723 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
1724 * tree-complex.c (expand_complex_div_wide): Update.
1725 * tree-eh.c (lower_resx): Update.
1726 (unsplit_eh): Update.
1727 (cleanup_empty_eh_move_lp): Update.
1728 * tree-inline.c (copy_edges_for_bb): Update.
1729 (freqs_to_counts): Update.
1730 (copy_cfg_body): Update.
1731 * tree-ssa-dce.c (remove_dead_stmt): Update.
1732 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
1733 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
1734 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
1735 (unloop_loops): Update.
1736 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
1737 * tree-ssa-loop-split.c (connect_loops): Update.
1738 (split_loop): Update.
1739 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
1740 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
1741 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
1742 * tree-ssa-reassoc.c (branch_fixup): Update.
1743 * tree-ssa-tail-merge.c (replace_block_by): Update.
1744 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
1745 (compute_path_counts): Update.
1746 (update_profile): Update.
1747 (recompute_probabilities): Update.
1748 (update_joiner_offpath_counts): Update.
1749 (estimated_freqs_path): Update.
1750 (freqs_to_counts_path): Update.
1751 (clear_counts_path): Update.
1752 (ssa_fix_duplicate_block_edges): Update.
1753 (duplicate_thread_path): Update.
1754 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
1755 (case_bit_test_cmp): Update.
1756 (collect_switch_conv_info): Update.
1757 (gen_inbound_check): Update.
1758 (do_jump_if_equal): Update.
1759 (emit_cmp_and_jump_insns): Update.
1760 * tree-tailcall.c (decrease_profile): Update.
1761 (eliminate_tail_call): Update.
1762 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
1763 (vect_do_peeling): Update.
1764 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
1765 * ubsan.c (ubsan_expand_null_ifn): Update.
1766 (ubsan_expand_ptr_ifn): Update.
1767 * value-prof.c (gimple_divmod_fixed_value): Update.
1768 (gimple_mod_pow2): Update.
1769 (gimple_mod_subtract): Update.
1770 (gimple_ic): Update.
1771 (gimple_stringop_fixed_value): Update.
1773 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
1776 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
1778 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
1780 PR rtl-optimization/82395
1781 * ira-color.c (allocno_priority_compare_func): Fix comparison step
1782 based on non_spilled_static_chain_regno_p.
1784 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
1786 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
1787 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
1788 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
1790 2017-10-19 Martin Sebor <msebor@redhat.com>
1792 PR tree-optimization/82596
1793 * tree.c (array_at_struct_end_p): Handle STRING_CST.
1795 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
1797 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
1798 (get_mem_refs_of_builtin_call): Likewise.
1799 * builtins.c (expand_builtin_apply): Adjust call to
1800 allocate_dynamic_stack_space.
1801 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
1802 the third argument to allocate_dynamic_stack_space, otherwise -1.
1803 (expand_builtin): Deal with all alloca variants.
1804 (is_inexpensive_builtin): Likewise.
1805 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
1806 * calls.c (special_function_p): Deal with all alloca variants.
1807 (initialize_argument_information): Adjust call to
1808 allocate_dynamic_stack_space.
1809 (expand_call): Likewise.
1810 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
1811 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
1812 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
1813 use it for the stack usage computation.
1814 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
1815 * function.c (gimplify_parameters): Call build_alloca_call_expr.
1816 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
1817 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
1818 (in_loop_p): Remove first argument and useless check.
1819 (pass_walloca::execute): Remove useless test and adjust call to above.
1820 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
1821 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
1822 (gimplify_call_expr): Deal with all alloca variants.
1823 * hsa-gen.c (gen_hsa_alloca): Likewise.
1824 (gen_hsa_insns_for_call): Likewise.
1825 * ipa-pure-const.c (special_builtin_state): Likewise.
1826 * tree-chkp.c (chkp_build_returned_bound): Likewise.
1827 * tree-object-size.c (alloc_object_size): Likewise.
1828 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
1829 (call_may_clobber_ref_p_1): Likewise.
1830 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
1831 (ccp_fold_stmt): Likewise.
1832 (optimize_stack_restore): Likewise.
1833 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
1834 (mark_all_reaching_defs_necessary_1): Likewise.
1835 (propagate_necessity): Likewise.
1836 (eliminate_unnecessary_stmts): Likewise.
1837 * tree.c (build_common_builtin_nodes): Build
1838 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
1839 (build_alloca_call_expr): New function.
1840 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
1841 (CASE_BUILT_IN_ALLOCA): Likewise.
1842 (build_alloca_call_expr): Declare.
1843 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
1845 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
1848 * dwarf2out.c (new_die_raw): New static inline function.
1849 (new_die): Use it to create the DIE.
1850 (add_AT_external_die_ref): Likewise.
1851 (clone_die): Likewise.
1852 (clone_as_declaration): Likewise.
1853 (dwarf2out_vms_debug_main_pointer): Likewise.
1854 (base_type_die): Likewise. Remove early return for corner cases.
1855 Do not call add_pubtype on the DIE here.
1856 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
1857 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
1858 typedefs for base types with DW_AT_endianity. Make sure a DIE with
1859 native order exists for base types, attach the DIE manually and call
1860 add_pubtype on it. Do not equate a reverse order DIE to the type.
1862 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
1864 * config/arm/arm.c (align_ok_ldrd_strd): New function.
1865 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
1867 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
1869 2017-10-19 Jakub Jelinek <jakub@redhat.com>
1871 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
1872 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
1873 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
1874 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
1875 * opts.c (sanitizer_opts): Add builtin.
1876 * ubsan.c (instrument_builtin): New function.
1877 (pass_ubsan::execute): Call it.
1878 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
1879 * doc/invoke.texi: Document -fsanitize=builtin.
1881 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
1882 builtins, store max (log2 (align), 0) into uchar field instead of
1883 align into uptr field.
1884 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
1885 store uchar 0 field instead of uptr 0 field.
1886 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
1887 instead of passing one address of struct with 2 locations pass
1888 two addresses of structs with 1 location each.
1889 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
1890 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
1891 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
1892 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
1893 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
1894 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
1895 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
1896 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
1898 2017-10-19 Martin Liska <mliska@suse.cz>
1901 * file-find.c (remove_prefix): Remove.
1902 * file-find.h (remove_prefix): Likewise.
1903 * gcc-ar.c: Remove smartness of lookup.
1905 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
1907 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
1908 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
1909 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
1911 2017-10-19 Jakub Jelinek <jakub@redhat.com>
1914 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
1915 (setcc + and to xor + setcc): New peephole2.
1917 2017-10-19 Tom de Vries <tom@codesourcery.com>
1919 * doc/sourcebuild.texi (Test Directives, Variants of
1920 dg-require-support): Add dg-require-stack-size.
1922 2017-10-19 Martin Liska <mliska@suse.cz>
1925 * gimplify.c (gimplify_decl_expr): Do not instrument variables
1926 that have a large alignment.
1927 (gimplify_target_expr): Likewise.
1929 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
1931 PR rtl-optimization/82602
1932 * ira.c (rtx_moveable_p): Return false for volatile asm.
1934 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
1937 * config/i386/i386-modes.def (CCGZ): New CC mode.
1938 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
1939 * config/i386/predicates.md (ix86_comparison_operator):
1941 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
1942 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
1943 with double-word subtraction.
1944 (put_condition_code): Handle CCGZmode.
1946 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
1948 * wide-int.cc (debug (const wide_int &)): New.
1949 (debug (const wide_int *)): New.
1950 (debug (const widest_int &)): New.
1951 (debug (const widest_int *)): New.
1953 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
1956 * lra-constraints.c (curr_insn_transform): Use non-input operand
1957 instead of output one for matched reload.
1959 2017-10-18 Bin Cheng <bin.cheng@arm.com>
1961 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
1962 (tree-ssa-loop-ivopts.h): New header file.
1963 (struct builtin_info): New fields.
1964 (classify_builtin_1): Compute and record base and offset parts for
1965 memset builtin partition by calling strip_offset.
1966 (offset_cmp, fuse_memset_builtins): New functions.
1967 (finalize_partitions): Fuse adjacent memset partitions by calling
1969 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
1970 Expose the interface.
1971 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
1973 2017-10-18 Bin Cheng <bin.cheng@arm.com>
1975 PR tree-optimization/82574
1976 * tree-loop-distribution.c (find_single_drs): New parameter. Check
1977 that data reference must be executed exactly once per iteration
1978 against the outermost loop in nest.
1979 (classify_partition): Update call to above function.
1981 2017-10-18 Richard Biener <rguenther@suse.de>
1983 PR tree-optimization/82591
1984 * graphite.c (graphite_transform_loops): Move code gen message
1986 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1987 Here. Handle scop_to_isl_ast failing.
1988 (scop_to_isl_ast): Limit the number of ISL operations.
1990 2017-10-18 Richard Biener <rguenther@suse.de>
1992 * graphite-isl-ast-to-gimple.c
1993 (translate_isl_ast_to_gimple::set_rename): Simplify.
1994 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
1995 (graphite_copy_stmts_from_block): ... here.
1996 (copy_bb_and_scalar_dependences): Simplify.
1997 (add_parameters_to_ivs_params): Canonicalize.
1998 (generate_entry_out_of_ssa_copies): Simplify.
1999 * graphite-sese-to-poly.c (extract_affine_name): Simplify
2000 by passing in ISL dimension.
2001 (parameter_index_in_region_1): Rename to ...
2002 (parameter_index_in_region): ... this.
2003 (extract_affine): Adjust assert, pass down parameter index.
2004 (add_param_constraints): Use range-info when available.
2005 (build_scop_context): Adjust.
2006 * sese.c (new_sese_info): Adjust.
2007 (free_sese_info): Likewise.
2008 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
2009 Remove unused typedefs.
2010 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
2012 2017-10-18 Martin Liska <mliska@suse.cz>
2014 * combine.c (simplify_compare_const): Add gcc_fallthrough.
2016 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
2018 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
2019 (s390_sched_init): Do not reset s390_sched_state if we entered the
2020 current basic block via a fallthru edge and all others are unlikely.
2022 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
2024 * config/s390/s390.c (NUM_SIDES): New variable.
2025 (LONGRUNNING_THRESHOLD): New variable.
2026 (LATENCY_FACTOR): New variable.
2027 (s390_sched_score): Decrease score for long-running instructions on
2029 (s390_sched_variable_issue): Perform bookkeeping for long-running
2032 2017-10-18 Richard Biener <rguenther@suse.de>
2034 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
2035 Simplify with removal of the parameter rename map.
2036 (set_rename): Likewise.
2037 (should_copy_to_new_region): Likewise.
2038 (graphite_copy_stmts_from_block): Likewise.
2039 (copy_bb_and_scalar_dependences): Remove initialization of
2040 unused copied_bb_map.
2042 (copy_internal_parameters): Likewise.
2043 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
2044 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
2045 Use INTEGRAL_TYPE_P.
2046 (parameter_index_in_region_1): Rename to ...
2047 (assign_parameter_index_in_region): ... this. Assert we have
2048 a parameter we handle.
2049 (scan_tree_for_params): Adjust.
2050 * sese.h (parameter_rename_map_t): Remove.
2051 (struct sese_info_t): Remove unused parameter_rename_map and
2052 copied_bb_map members.
2053 * sese.c (new_sese_info): Adjust.
2054 (free_sese_info): Likewise.
2056 2017-10-18 Martin Liska <mliska@suse.cz>
2059 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
2060 on an abnormal edge.
2062 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
2064 * doc/invoke.texi (ffunction-sections and fdata-sections):
2067 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
2069 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
2070 the use statement can throw internally.
2072 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
2074 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
2075 any RTX present on the RHS of a SET.
2076 * compare-elim.c (try_eliminate_compare): Restore comment.
2078 2017-10-17 Jakub Jelinek <jakub@redhat.com>
2080 * langhooks.h (struct lang_hooks): Document that tree_size langhook
2081 may be also called on tcc_type nodes.
2082 * langhooks.c (lhd_tree_size): Likewise.
2084 2017-10-17 David Malcolm <dmalcolm@redhat.com>
2086 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
2087 format_warning_at_substring.
2088 (maybe_warn): Convert source_range * param to a location_t. Pass
2089 UNKNOWN_LOCATION rather than NULL to fmtwarn.
2090 (format_directive): Remove code to extract source_ranges and
2091 source_range * in favor of just a location_t.
2092 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
2094 * substring-locations.c (format_warning_va): Convert
2095 source_range * param to a location_t.
2096 (format_warning_at_substring): Likewise.
2097 * substring-locations.h (format_warning_va): Likewise.
2098 (format_warning_at_substring): Likewise.
2100 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
2102 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
2104 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
2105 and scatter/gather ops.
2107 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
2108 vec_gather_load and vec_scatter_store.
2109 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
2110 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
2111 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
2112 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
2113 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
2114 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
2116 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
2118 * reg-stack.c (compare_for_stack_reg): Add bool argument.
2119 Detect FTST instruction and handle its register pops. Only pop
2120 second operand if can_pop_second_op is true.
2121 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
2122 set can_pop_second_op to false in the compare_for_stack_reg call.
2124 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
2125 output_fp_compare for stack register operands.
2126 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
2127 instructions here. Do not emit stack register pops here. Assert
2128 that FCOMPP pops next to top stack register. Rewrite function.
2130 2017-10-17 Nathan Sidwell <nathan@acm.org>
2133 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
2134 use DECL_ASSEMBLER_NAME_RAW.
2137 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
2140 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
2141 Wilco Dijkstra <wilco.dijkstra@arm.com>
2143 * builtins.c (expand_builtin_update_setjmp_buf): Add a
2144 converstion to Pmode from the buf_addr.
2146 2017-10-17 Richard Biener <rguenther@suse.de>
2148 * graphite-dependences.c (scop_get_reads_and_writes): Change
2149 output parameters to references.
2151 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
2153 PR 71026/tree-optimization
2154 * fold-const.c (distribute_real_division): Removed.
2155 (fold_binary_loc): Remove calls to distribute_real_divison.
2157 2017-10-17 Richard Biener <rguenther@suse.de>
2159 * graphite-scop-detection.c
2160 (scop_detection::stmt_has_simple_data_refs_p): Always use
2161 the full nest as region.
2162 (try_generate_gimple_bb): Likewise.
2163 * sese.c (scalar_evolution_in_region): Simplify now that
2164 SCEV can handle instantiation in regions.
2165 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
2166 in the non-loop part of a function if requested.
2168 2017-10-17 Richard Biener <rguenther@suse.de>
2170 PR tree-optimization/82563
2171 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
2173 (graphite_regenerate_ast_isl): Call it.
2174 * graphite-scop-detection.c (build_scops): Remove entry edge split.
2176 2017-10-17 Jakub Jelinek <jakub@redhat.com>
2178 PR tree-optimization/82549
2179 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
2180 Formatting fixes. Instead of calling make_bit_field_ref with negative
2183 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
2185 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
2186 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
2187 _mm_maskz_reduce_ss): New.
2188 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
2189 __builtin_ia32_reducess_mask): Ditto..
2190 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
2191 * config/i386/sse.md (reduces<mode>): Renamed to ...
2192 (reduces<mode><mask_scalar_name>): ... this.
2193 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
2195 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
2196 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
2198 2017-10-16 David Malcolm <dmalcolm@redhat.com>
2200 * Makefile.in (OBJS): Add unique-ptr-tests.o.
2201 * selftest-run-tests.c (selftest::run_tests): Call
2202 selftest::unique_ptr_tests_cc_tests.
2203 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
2204 * unique-ptr-tests.cc: New file.
2206 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
2209 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
2211 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
2212 (make_hard_regno_born, make_hard_regno_dead): Update
2213 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
2215 2017-10-16 Jeff Law <law@redhat.com>
2217 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
2219 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
2221 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
2223 2017-10-16 Olivier Hainque <hainque@adacore.com>
2225 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
2226 with_cpu if we were configured for an e500v2 target cpu name.
2228 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
2230 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
2231 * doc/invoke.texi: Document +nodsp as a valid extension for
2234 2017-10-16 Martin Liska <mliska@suse.cz>
2236 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
2237 (test_set_range): Likewise.
2238 (test_range_functions): Rename to ...
2239 (test_bit_in_range): ... this.
2240 (sbitmap_c_tests): Add new test.
2242 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2244 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
2246 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
2247 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
2249 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2251 * config/aarch64/aarch64-builtins.c
2252 (aarch64_types_quadopu_lane_qualifiers): New.
2253 (TYPES_QUADOPU_LANE): New.
2254 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
2255 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
2256 (aarch64_<sur>dot_laneq<vsi2qi>): New.
2257 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
2258 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
2259 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
2260 (Vdottype, DOTPROD): New.
2261 (sur): Add SDOT and UDOT.
2263 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2265 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
2266 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
2267 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
2269 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
2270 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
2271 Enable TARGET_DOTPROD.
2272 (cortex-a75.cortex-a55): Likewise.
2273 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
2275 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2277 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
2278 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
2280 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
2282 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
2283 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
2284 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
2285 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
2286 * config/arm/types.md (neon_dot, neon_dot_q): New.
2287 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
2289 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2291 * config/arm/arm.h (TARGET_DOTPROD): New.
2292 * config/arm/arm.c (arm_arch_dotprod): New.
2293 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
2294 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
2295 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
2296 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
2297 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
2298 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
2299 * doc/invoke.texi (armv8.2-a): Document dotprod
2301 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
2303 * i386.c (ix86_vec_cost): New function.
2304 (ix86_rtx_costs): Handle vector operations better.
2305 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
2306 * x86-tune-costs.h: Add new costs to all tables.
2308 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
2310 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
2312 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
2313 divsd, sqrtss and sqrtsd
2314 * x86-tune-costs.h: Add new entries to all costs.
2315 (znver1_cost): Fix to match real instruction latencies.
2317 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2318 Michael Collison <michael.collison@arm.com>
2320 * compare-elim.c: Include emit-rtl.h.
2321 (can_merge_compare_into_arith): New function.
2322 (try_validate_parallel): Likewise.
2323 (try_merge_compare): Likewise.
2324 (try_eliminate_compare): Call the above when no previous clobber
2326 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
2329 2017-10-14 Jakub Jelinek <jakub@redhat.com>
2333 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
2334 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
2338 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
2339 to be any operand_equal_p operands. For & (B - 1) require
2340 B to be power of 2. Recognize
2341 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
2343 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
2346 * optabs.c (expand_memory_blockage): Fix call of
2347 targetm.have_memory_blockage.
2349 2017-10-14 Jakub Jelinek <jakub@redhat.com>
2352 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
2353 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
2354 objects to extra_objs instead of overwriting it.
2356 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
2358 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
2359 Use any_fp_register_operand as operand[3] predicate. Simplify
2360 equality test for operands[2] and operands[4] memory location.
2361 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
2362 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
2363 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
2364 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
2365 any_fp_register_operand as operand[1] predicate. Simplify
2366 equality test for operands[0] and operands[3] memory location.
2367 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
2368 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
2369 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
2371 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
2373 * target-insns.def: Add memory_blockage.
2374 * optabs.c (expand_memory_blockage): New function.
2375 (expand_asm_memory_barrier): Rename ...
2376 (expand_asm_memory_blockage): ... to this.
2377 (expand_mem_thread_fence): Call expand_memory_blockage
2378 instead of expand_asm_memory_barrier.
2379 (expand_mem_singnal_fence): Ditto.
2380 (expand_atomic_load): Ditto.
2381 (expand_atomic_store): Ditto.
2382 * doc/md.texi (Standard Pattern Names For Generation):
2383 Document memory_blockage instruction pattern.
2385 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
2387 * config/rl78/rl78.c (rl78_emit_libcall): New function.
2388 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
2389 * config/rl78/rl78.md: New define_expand "adddi3".
2391 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
2393 * cfghooks.c (verify_flow_info): Disable check that all probabilities
2396 2017-10-13 Jeff Law <law@redhat.com>
2398 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
2400 2017-10-13 Jakub Jelinek <jakub@redhat.com>
2403 * internal-fn.c (expand_mul_overflow): If both operands have
2404 the same highpart of -1 or 0 and the topmost bit of lowpart
2405 is different, overflow is if res <= 0 rather than res < 0.
2407 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
2409 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
2410 TARGET_P9_VECTOR code for unaligned_load case.
2412 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
2414 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
2416 2017-10-13 Nathan Sidwell <nathan@acm.org>
2418 * tree-core.h (tree_contains_struct): Make bool.
2419 * tree.c (tree_contains_struct): Likewise.
2420 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
2421 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
2422 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
2423 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
2424 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
2426 2017-10-13 Richard Biener <rguenther@suse.de>
2428 * graphite-isl-ast-to-gimple.c
2429 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
2430 parameters and dominance check.
2431 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
2432 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
2433 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
2434 Do not update SSA form here or do intermediate IL verification.
2435 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
2436 (graphite_initialize): Remove check on the number of loops in
2437 the function and inline into graphite_transform_loops.
2438 (graphite_finalize): Inline into graphite_transform_loops.
2439 (graphite_transform_loops): Perform SSA update and IL verification
2441 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
2443 2017-10-13 Richard Biener <rguenther@suse.de>
2445 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
2446 graphite_expression_type_precision): Avoid global constructor
2448 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
2449 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
2450 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
2451 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
2452 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
2453 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
2455 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
2458 * config/i386/i386.h (ix86_red_zone_size): New.
2459 * config/i386/i386.md (push peephole2s): Replace
2460 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
2462 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2463 Alan Hayward <alan.hayward@arm.com>
2464 David Sherwood <david.sherwood@arm.com>
2466 * combine.c (can_change_dest_mode): Reject changes in
2467 REGMODE_NATURAL_SIZE.
2469 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2470 Alan Hayward <alan.hayward@arm.com>
2471 David Sherwood <david.sherwood@arm.com>
2473 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
2474 (expand_debug_source_expr): Likewise.
2475 * combine.c (combine_simplify_rtx): Likewise.
2476 * cse.c (fold_rtx): Likewise.
2477 * fwprop.c (canonicalize_address): Likewise.
2478 * targhooks.c (default_shift_truncation_mask): Likewise.
2480 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2481 Alan Hayward <alan.hayward@arm.com>
2482 David Sherwood <david.sherwood@arm.com>
2484 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
2485 (widened_mode): Likewise.
2486 (expand_unop): Likewise.
2487 * ree.c (transform_ifelse): Likewise.
2488 (merge_def_and_ext): Likewise.
2489 (combine_reaching_defs): Likewise.
2490 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2492 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2493 Alan Hayward <alan.hayward@arm.com>
2494 David Sherwood <david.sherwood@arm.com>
2496 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
2497 * combine.c (gen_lowpart_for_combine): Likewise.
2498 * dwarf2out.c (rtl_for_decl_location): Likewise.
2499 * final.c (alter_subreg): Likewise.
2500 * rtlhooks.c (gen_lowpart_general): Likewise.
2501 (gen_lowpart_if_possible): Likewise.
2503 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2504 Alan Hayward <alan.hayward@arm.com>
2505 David Sherwood <david.sherwood@arm.com>
2507 * calls.c (expand_call): Use subreg_lowpart_offset.
2508 * cse.c (cse_insn): Likewise.
2509 * regcprop.c (copy_value): Likewise.
2510 (copyprop_hardreg_forward_1): Likewise.
2512 2017-10-13 Jakub Jelinek <jakub@redhat.com>
2515 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
2516 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
2517 =Q constraints to +Q and into insn condition add check
2518 that operands[0] and operands[1] are equal.
2519 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
2520 =Q constraints to +Q and into insn condition add check
2521 that operands[0] is equal to either operands[1] or operands[2].
2524 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
2525 instead of handling MINUS_EXPR twice (once for each argument),
2526 canonicalize operand order and handle just once, use rtype where
2527 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
2530 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
2531 any values of __C while still being pattern recognizable as a simple
2534 2017-10-13 Richard Biener <rguenther@suse.de>
2536 PR tree-optimization/82451
2538 2017-10-02 Richard Biener <rguenther@suse.de>
2540 PR tree-optimization/82355
2541 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
2542 a mapping for the enclosing loop but avoid generating one for
2544 (copy_bb_and_scalar_dependences): Remove premature codegen
2545 error on PHIs in blocks duplicated into multiple places.
2546 * graphite-scop-detection.c
2547 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
2548 in the region use it as loop and nest to analyze the DR in.
2549 (try_generate_gimple_bb): Likewise.
2550 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
2551 (add_loop_constraints): For blocks in a loop not in the region
2552 create a dimension with a single iteration.
2553 * sese.h (gbb_loop_at_index): Remove assert.
2555 * cfgloop.c (loop_preheader_edge): For the loop tree root
2556 return the single successor of the entry block.
2557 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
2558 Reset the SCEV hashtable and niters.
2559 * graphite-scop-detection.c
2560 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
2561 assert that we only have POLYNOMIAL_CHREC that vary in loops
2562 contained in the region.
2563 (scop_detection::graphite_can_represent_expr): Adjust.
2564 (scop_detection::stmt_has_simple_data_refs_p): For loops
2565 not in the region set loop to NULL. The nest is now the
2566 entry edge to the region.
2567 (try_generate_gimple_bb): Likewise.
2568 * sese.c (scalar_evolution_in_region): Adjust for
2569 instantiate_scev change.
2570 * tree-data-ref.h (graphite_find_data_references_in_stmt):
2571 Make nest parameter the edge into the region.
2572 (create_data_ref): Likewise.
2573 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
2574 entry edge into a region and adjust instantiate_scev calls.
2575 (create_data_ref): Likewise.
2576 (graphite_find_data_references_in_stmt): Likewise.
2577 (find_data_references_in_stmt): Pass the loop preheader edge
2578 from the nest argument.
2579 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
2580 parameter the edge into the region.
2581 (instantiate_parameters): Use the loop preheader edge as entry.
2582 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
2584 (get_instantiated_value_entry): Make instantiate_below parameter
2585 the edge into the region.
2586 (instantiate_scev_name): Likewise. Adjust dominance checks,
2587 when we cannot use loop-based instantiation instantiate by
2588 walking use-def chains.
2589 (instantiate_scev_poly): Adjust.
2590 (instantiate_scev_binary): Likewise.
2591 (instantiate_scev_convert): Likewise.
2592 (instantiate_scev_not): Likewise.
2593 (instantiate_array_ref): Remove.
2594 (instantiate_scev_3): Likewise.
2595 (instantiate_scev_2): Likewise.
2596 (instantiate_scev_1): Likewise.
2597 (instantiate_scev_r): Do not blindly handle N-operand trees.
2598 Do not instantiate array-refs. Handle all constants and invariants.
2599 (instantiate_scev): Make instantiate_below parameter
2600 the edge into the region.
2601 (resolve_mixers): Use the loop preheader edge for the region
2602 parameter to instantiate_scev_r.
2603 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
2605 2017-10-13 Richard Biener <rguenther@suse.de>
2607 PR tree-optimization/82525
2608 * graphite-isl-ast-to-gimple.c
2609 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
2611 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
2612 Fail code generation when we cannot represent the isl integer.
2613 (binary_op_to_tree): Elide modulo operations that are no-ops
2614 in the type we code generate. Remove now superfluous code
2617 2017-10-13 Richard Biener <rguenther@suse.de>
2619 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
2620 (scop_detection::harmful_loop_in_region): Remove premature
2621 IV type restriction.
2622 (scop_detection::graphite_can_represent_scev): We can handle
2623 pointer IVs just fine.
2625 2017-10-13 Alan Modra <amodra@gmail.com>
2627 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
2628 "Clobbers and Scratch Registers". Add paragraph on
2629 alternative to clobbers for scratch registers and OpenBLAS
2632 2017-10-13 Alan Modra <amodra@gmail.com>
2634 * doc/extend.texi (Clobbers): Correct vax example. Delete old
2635 example of a memory input for a string of known length. Move
2636 commentary out of table. Add a number of new examples
2637 covering array memory inputs.
2639 2017-10-12 Martin Liska <mliska@suse.cz>
2641 PR tree-optimization/82493
2642 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
2643 (test_range_functions): New function.
2644 (sbitmap_c_tests): Likewise.
2645 * selftest-run-tests.c (selftest::run_tests): Run new tests.
2646 * selftest.h (sbitmap_c_tests): New function.
2648 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
2650 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
2652 * config/rs6000/amo.h: Fix spacing issue.
2654 2017-10-12 Jakub Jelinek <jakub@redhat.com>
2657 * config/i386/i386.md (*ashl<mode>3_mask_1,
2658 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
2659 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
2662 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
2664 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
2665 (profile_probability): Set max_probability
2666 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
2668 * profile-count.c (profile_probability::differs_from_p): Do not
2669 rely on max_probaiblity == 10000
2671 2017-10-12 Jeff Law <law@redhat.com>
2673 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
2676 2017-10-12 Martin Sebor <msebor@redhat.com>
2680 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
2681 (handle_alias_pairs): Call it.
2682 * common.opt (-Wattribute-alias): New option.
2683 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
2684 * doc/invoke.texi (-Wattribute-alias): Document.
2686 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
2689 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
2691 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
2693 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
2694 (make_hard_regno_born, make_hard_regno_dead): Update
2695 bb_killed_pseudos and bb_gen_pseudos.
2697 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
2699 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
2701 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
2703 * config/alpha/alpha.c (alpha_split_conditional_move):
2704 Use std::swap instead of manually swapping.
2705 (alpha_stdarg_optimize_hook): Ditto.
2706 (alpha_canonicalize_comparison): Ditto.
2708 2017-10-12 Bin Cheng <bin.cheng@arm.com>
2710 * tree-loop-distribution.c (struct builtin_info): New struct.
2711 (struct partition): Refactor fields into struct builtin_info.
2712 (partition_free): Free struct builtin_info.
2713 (build_size_arg_loc, build_addr_arg_loc): Delete.
2714 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
2715 information from struct builtin_info.
2716 (find_single_drs): New function refactored from classify_partition.
2717 Also moved builtin validity checks to this function.
2718 (compute_access_range, alloc_builtin): New functions.
2719 (classify_builtin_st, classify_builtin_ldst): New functions.
2720 (classify_partition): Refactor code into functions find_single_drs,
2721 classify_builtin_st and classify_builtin_ldst.
2722 (distribute_loop): Don't do runtime alias check when distributing
2724 (find_seed_stmts_for_distribution): New function.
2725 (pass_loop_distribution::execute): Refactor code finding seed
2726 stmts into above function. Support distribution for the innermost
2727 two-level loop nest. Adjust dump information.
2729 2017-10-12 Bin Cheng <bin.cheng@arm.com>
2731 * tree-loop-distribution.c: Adjust the general comment.
2732 (NUM_PARTITION_THRESHOLD): New macro.
2733 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
2734 (classify_partition): Skip builtin pattern of loop nest's inner loop.
2735 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
2736 in call to build_partition_graph.
2737 (finalize_partitions): New parameter. Make loop distribution more
2738 conservative by fusing more partitions.
2739 (distribute_loop): Don't do runtime alias check in case of loop nest
2741 (find_seed_stmts_for_distribution): New function.
2742 (prepare_perfect_loop_nest): New function.
2743 (pass_loop_distribution::execute): Refactor code finding seed stmts
2744 and loop nest into above functions. Support loop nest distribution.
2745 Adjust dump information accordingly.
2747 2017-10-12 Bin Cheng <bin.cheng@arm.com>
2749 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
2750 and set PTYPE_SEQUENTIAL for merged partition.
2752 2017-10-12 Richard Biener <rguenther@suse.de>
2754 PR tree-optimization/69728
2756 2017-09-19 Richard Biener <rguenther@suse.de>
2758 PR tree-optimization/69728
2759 * graphite-sese-to-poly.c (schedule_error): New global.
2760 (add_loop_schedule): Handle empty domain by failing the
2762 (build_original_schedule): Handle schedule_error.
2764 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
2765 domain by returning an unchanged schedule.
2767 2017-10-12 Jakub Jelinek <jakub@redhat.com>
2769 * genrecog.c (validate_pattern): For VEC_SELECT verify that
2770 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
2772 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
2774 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
2777 2017-10-12 Jakub Jelinek <jakub@redhat.com>
2780 * expr.c (store_field): Don't optimize away bitsize == 0 store
2781 from CALL_EXPR with addressable return type.
2783 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
2785 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
2786 * config/rs6000/rs6000.md (sel): Delete mode attribute.
2787 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
2788 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
2789 TARGET_ISEL instead of TARGET_ISEL<sel>.
2791 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
2793 * config/rs6000/rs6000.c
2794 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
2796 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
2798 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
2800 (reg_or_cint_operand): Fix comment.
2801 (reg_or_zero_operand): New predicate.
2802 * config/rs6000/rs6000-protos.h (output_isel): Delete.
2803 * config/rs6000/rs6000.c (output_isel): Delete.
2804 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
2805 instead of reg_or_cint_operand. Output instruction directly (not via
2807 (isel_unsigned_<mode>): Ditto.
2808 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
2809 gpc_reg_operand. Add an instruction alternative for this. Output
2810 instruction directly.
2811 (*isel_reversed_unsigned_<mode>): Ditto.
2813 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
2815 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
2816 (TARGET_CANONICALIZE_COMPARISON): Define.
2818 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
2821 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
2822 Check whether the dest is REG before adding REG_EQUIV note.
2824 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
2827 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
2829 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
2830 (make_hard_regno_born, make_hard_regno_dead): Update
2831 bb_killed_pseudos and bb_gen_pseudos.
2833 2017-10-11 Nathan Sidwell <nathan@acm.org>
2835 * incpath.h (enum incpath_kind): Name enum, prefix values.
2836 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
2837 * incpath.c (heads, tails): Use INC_MAX.
2838 (add_env_var_paths, add_standard_paths): Use incpath_kind.
2839 (merge_include_chains, split_quote_chain,
2840 register_include_chains): Update incpath_kind names.
2841 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
2842 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
2844 (add_framework_path, darwin_register_objc_includes): Likewise.
2845 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
2847 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
2849 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
2850 Do not use float_operator operator predicate.
2851 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
2852 * config/i386/predicates.md (float_operator): Remove predicate.
2854 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
2856 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
2857 (*jccxf_i387): Ditto.
2858 (*jcc<mode>_i387): Ditto.
2859 (*jccu<mode>_i387): Ditto.
2860 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
2861 (*jcc_*_i387 splitters): Remove.
2862 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
2863 * config/i386/i386.c (ix86_split_fp_branch): Remove.
2864 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
2867 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
2869 * profile-count.h (slow_safe_scale_64bit): New function.
2870 (safe_scale_64bit): New inline.
2871 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
2872 * profile-count.c: Include wide-int.h
2873 (slow_safe_scale_64bit): New.
2875 2017-10-11 Nathan Sidwell <nathan@acm.org>
2877 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
2878 HAS_DECL_ASSEMBLER_NAME_P.
2879 * gimple-expr.c (gimple_decl_printable_name: Check
2880 HAS_DECL_ASSEMBLER_NAME_P too.
2881 * ipa-utils.h (type_in_anonymous_namespace_p): Check
2882 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
2883 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
2884 * passes.c (rest_of_decl_compilation): Check
2885 HAS_DECL_ASSEMBLER_NAME_P too.
2886 * recog.c (verify_changes): Likewise.
2887 * tree-pretty-print.c (dump_decl_name): Likewise.
2888 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
2890 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
2891 (SET_DECL_ASSEMBLER_NAME): Use it.
2892 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
2893 (COPY_DECL_ASSEMBLER_NAME): Likewise.
2894 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
2896 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
2898 * config.gcc (i386, x86_64): Add extra objects.
2899 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
2900 (ix86_min_insn_size): Declare.
2901 (ix86_issue_rate): Declare.
2902 (ix86_adjust_cost): Declare.
2903 (ia32_multipass_dfa_lookahead): Declare.
2904 (ix86_macro_fusion_p): Declare.
2905 (ix86_macro_fusion_pair_p): Declare.
2906 (ix86_bd_has_dispatch): Declare.
2907 (ix86_bd_do_dispatch): Declare.
2908 (ix86_core2i7_init_hooks): Declare.
2909 (ix86_atom_sched_reorder): Declare.
2910 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
2911 (COSTS_N_BYTES): Move to x86-tune-costs.h.
2912 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
2913 (rip_relative_addr_p): Rename to ...
2914 (ix86_rip_relative_addr_p): ... this one; export.
2915 (memory_address_length): Update.
2916 (ix86_issue_rate): Move to x86-tune-sched.c.
2917 (ix86_flags_dependent): Move to x86-tune-sched.c.
2918 (ix86_agi_dependent): Move to x86-tune-sched.c.
2919 (exact_dependency_1): Move to x86-tune-sched.c.
2920 (exact_store_load_dependency): Move to x86-tune-sched.c.
2921 (ix86_adjust_cost): Move to x86-tune-sched.c.
2922 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
2923 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
2924 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
2925 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
2926 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
2927 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
2928 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
2929 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
2930 (min_insn_size): Rename to ...
2931 (ix86_min_insn_size): ... this one; export.
2932 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
2933 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
2934 (core2i7_first_cycle_multipass_backtrack): Move to
2935 x86-tune-sched-core.c.
2936 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
2937 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
2938 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
2939 (ix86_avoid_jump_mispredicts): Update.
2940 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
2941 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
2942 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
2943 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
2944 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
2945 (MAX_INSN): Move to ix86-tune-sched-bd.c.
2946 (MAX_IMM): Move to ix86-tune-sched-bd.c.
2947 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
2948 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
2949 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
2950 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
2951 (MAX_STORE): Move to ix86-tune-sched-bd.c.
2952 (BIG): Move to ix86-tune-sched-bd.c.
2953 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
2954 (enum insn_path): Move to ix86-tune-sched-bd.c.
2955 (get_mem_group): Move to ix86-tune-sched-bd.c.
2956 (is_cmp): Move to ix86-tune-sched-bd.c.
2957 (dispatch_violation): Move to ix86-tune-sched-bd.c.
2958 (is_branch): Move to ix86-tune-sched-bd.c.
2959 (is_prefetch): Move to ix86-tune-sched-bd.c.
2960 (init_window): Move to ix86-tune-sched-bd.c.
2961 (allocate_window): Move to ix86-tune-sched-bd.c.
2962 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
2963 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
2964 (process_end_window): Move to ix86-tune-sched-bd.c.
2965 (allocate_next_window): Move to ix86-tune-sched-bd.c.
2966 (find_constant): Move to ix86-tune-sched-bd.c.
2967 (get_num_immediates): Move to ix86-tune-sched-bd.c.
2968 (has_immediate): Move to ix86-tune-sched-bd.c.
2969 (get_insn_path): Move to ix86-tune-sched-bd.c.
2970 (get_insn_group): Move to ix86-tune-sched-bd.c.
2971 (count_num_restricted): Move to ix86-tune-sched-bd.c.
2972 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
2973 (add_insn_window): Move to ix86-tune-sched-bd.c.
2974 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
2975 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
2976 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
2977 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
2978 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
2979 (do_dispatch): Move to ix86-tune-sched-bd.c.
2980 (has_dispatch): Move to ix86-tune-sched-bd.c.
2981 * config/i386/t-i386: Add new object files.
2982 * config/i386/x86-tune-costs.h: New file.
2983 * config/i386/x86-tune-sched-atom.c: New file.
2984 * config/i386/x86-tune-sched-bd.c: New file.
2985 * config/i386/x86-tune-sched-core.c: New file.
2986 * config/i386/x86-tune-sched.c: New file.
2988 2017-10-11 Liu Hao <lh_mouse@126.com>
2990 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
2991 the generic version below instead.
2992 (should_colorize): Recognize Windows consoles as terminals
2994 * pretty-print.c [__MINGW32__] (write_all): New function.
2995 [__MINGW32__] (find_esc_head): Likewise.
2996 [__MINGW32__] (find_esc_terminator): Likewise.
2997 [__MINGW32__] (eat_esc_sequence): Likewise.
2998 [__MINGW32__] (mingw_ansi_fputs): New function that handles
3000 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
3003 2017-10-11 Richard Biener <rguenther@suse.de>
3005 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
3006 Properly call analyze_scalar_evolution with the loop of the stmt.
3008 2017-10-11 Richard Biener <rguenther@suse.de>
3010 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
3011 * tree-core.h (tree_base): Add chrec_var union member.
3012 * tree.h (CHREC_VAR): Remove.
3013 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
3014 * tree-chrec.h (build_polynomial_chrec): Adjust.
3015 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
3016 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
3018 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
3020 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
3021 * match.pd: ... here.
3022 ((T) X == (T) Y): Relax condition.
3024 2017-10-11 Bin Cheng <bin.cheng@arm.com>
3026 PR tree-optimization/82472
3027 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
3029 (break_alias_scc_partitions): Update postorder number.
3031 2017-10-11 Martin Liska <mliska@suse.cz>
3034 * opts.c (parse_no_sanitize_attribute): Do not use error_value
3036 * opts.h (parse_no_sanitize_attribute): Remove last argument.
3038 2017-10-11 Martin Liska <mliska@suse.cz>
3040 * print-rtl.c (print_insn): Move declaration of idbuf
3041 to same scope as name.
3043 2017-10-11 Martin Liska <mliska@suse.cz>
3048 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
3051 2017-10-11 Martin Liska <mliska@suse.cz>
3054 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
3057 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
3059 * config.gcc (default_gnu_indirect_function): Default to yes for
3060 arm*-*-linux* with glibc.
3062 2017-10-11 Richard Biener <rguenther@suse.de>
3064 * tree-scalar-evolution.c (get_scalar_evolution): Handle
3065 default-defs and types we do not want to analyze.
3066 (interpret_loop_phi): Replace unreachable code with an assert.
3067 (compute_scalar_evolution_in_loop): Remove and inline ...
3068 (analyze_scalar_evolution_1): ... here, replacing condition with
3069 what makes the intent clearer. Remove handling of cases
3070 get_scalar_evolution now handles.
3072 2017-10-10 Jim Wilson <wilson@tuliptree.org>
3074 PR rtl-optimization/81434
3075 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
3076 comment for main loop. In sched_group_found if, also add checks for
3077 pass and min_cost_group.
3079 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
3081 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
3082 (rs6000_insn_cost): New function.
3083 * config/rs6000/rs6000.md (cost): New attribute.
3085 2017-10-10 Jakub Jelinek <jakub@redhat.com>
3086 H.J. Lu <hongjiu.lu@intel.com>
3090 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
3091 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
3092 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
3093 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
3094 to OPTION_MASK_ISA_AVX512VL - builtins that have both
3095 OPTION_MASK_ISA_MMX and some other bit set require both
3096 mmx and the ISAs without the mmx bit.
3097 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
3098 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
3099 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
3100 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
3101 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
3102 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
3103 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
3104 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
3105 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
3106 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
3107 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
3108 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
3109 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
3110 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
3111 Add OPTION_MASK_ISA_MMX.
3113 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
3115 * config.gcc (armv7*-*-freebsd*): New target.
3116 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
3118 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
3120 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
3121 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
3124 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
3126 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
3127 a template parameter.
3128 (WIDE_INT_REF_FOR): Update accordingly.
3129 * tree.h (wi::int_traits <const_tree>): Delete.
3130 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
3131 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
3132 (wi::tree_to_wide_ref): New typedef.
3133 (wi::to_wide): New function.
3134 * calls.c (get_size_range): Use wi::to_wide when operating on
3136 * cgraph.c (cgraph_node::create_thunk): Likewise.
3137 * config/i386/i386.c (ix86_data_alignment): Likewise.
3138 (ix86_local_alignment): Likewise.
3139 * dbxout.c (stabstr_O): Likewise.
3140 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
3141 * expr.c (const_vector_from_tree): Likewise.
3142 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
3143 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
3144 (fold_negate_expr_1, int_const_binop_1, const_binop)
3145 (fold_convert_const_int_from_real, optimize_bit_field_compare)
3146 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
3147 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
3148 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
3149 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
3150 (fold_not_const, round_up_loc): Likewise.
3151 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
3152 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
3153 (alloca_call_type): Likewise.
3154 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
3155 * godump.c (go_output_typedef): Likewise.
3156 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
3157 * internal-fn.c (get_min_precision): Likewise.
3158 * ipa-cp.c (ipcp_store_vr_results): Likewise.
3159 * ipa-polymorphic-call.c
3160 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
3161 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
3162 (ipa_modify_call_arguments): Likewise.
3163 * match.pd: Likewise.
3164 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
3165 * print-tree.c (print_node_brief, print_node): Likewise.
3166 * stmt.c (expand_case): Likewise.
3167 * stor-layout.c (layout_type): Likewise.
3168 * tree-affine.c (tree_to_aff_combination): Likewise.
3169 * tree-cfg.c (group_case_labels_stmt): Likewise.
3170 * tree-data-ref.c (dr_analyze_indices): Likewise.
3171 (prune_runtime_alias_test_list): Likewise.
3172 * tree-dump.c (dequeue_and_dump): Likewise.
3173 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
3174 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
3175 * tree-pretty-print.c (dump_generic_node): Likewise.
3176 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
3177 (simple_iv_with_niters): Likewise.
3178 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
3179 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
3180 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
3181 * tree-ssa-loop-niter.c (split_to_var_and_offset)
3182 (refine_value_range_using_guard, number_of_iterations_ne_max)
3183 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
3184 (get_cst_init_from_scev, record_nonwrapping_iv)
3185 (scev_var_range_cant_overflow): Likewise.
3186 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
3187 * tree-ssa-pre.c (compute_avail): Likewise.
3188 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
3189 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
3190 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
3191 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
3192 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
3193 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
3194 (dump_case_nodes, try_switch_expansion): Likewise.
3195 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
3196 (vect_do_peeling): Likewise.
3197 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
3198 * tree-vect-stmts.c (vectorizable_load): Likewise.
3199 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
3200 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
3201 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
3202 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
3203 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
3204 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
3205 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
3206 (evrp_dom_walker::before_dom_children): Likewise.
3207 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
3208 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
3209 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
3210 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
3211 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
3212 (get_range_pos_neg): Likewise.
3213 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
3214 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
3215 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
3216 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
3217 * config/avr/avr.c (avr_fold_builtin): Likewise.
3218 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
3219 * config/msp430/msp430.c (msp430_attr): Likewise.
3220 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
3221 * config/powerpcspe/powerpcspe-c.c
3222 (altivec_resolve_overloaded_builtin): Likewise.
3223 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
3224 (rs6000_expand_ternop_builtin): Likewise.
3225 * config/rs6000/rs6000-c.c
3226 (altivec_resolve_overloaded_builtin): Likewise.
3227 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
3228 (rs6000_expand_ternop_builtin): Likewise.
3229 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
3231 2017-10-10 Bin Cheng <bin.cheng@arm.com>
3233 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
3234 when copying loop nest with only one inner loop.
3236 2017-10-10 Richard Biener <rguenther@suse.de>
3238 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
3239 blocks if SCEV is active.
3240 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
3242 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
3243 (scev_initialize): Assert we are not yet initialized.
3245 2017-10-10 Bin Cheng <bin.cheng@arm.com>
3247 * tree-loop-distribution.c (generate_loops_for_partition): Remove
3248 inner loop's exit stmt by making it always exit the loop, otherwise
3249 we would generate an infinite empty loop.
3251 2017-10-10 Bin Cheng <bin.cheng@arm.com>
3253 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
3254 renaming variables in new preheader if it's deleted.
3256 2017-10-10 Bin Cheng <bin.cheng@arm.com>
3258 * tree-loop-distribution.c (struct partition): Remove unused field
3259 loops of the structure.
3260 (partition_alloc, partition_free): Ditto.
3261 (build_rdg_partition_for_vertex): Ditto.
3263 2017-10-09 Jeff Law <law@redhat.com>
3265 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
3266 return type to match prototype and documentation.
3268 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
3270 * config/rs6000/rs6000.c (processor_costs): Move to ...
3271 * config/rs6000/rs6000.h: ... here.
3272 (rs6000_cost): Declare.
3274 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
3276 * except.c (setjmp_fn): New global variable.
3277 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
3278 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
3279 if DONT_USE_BUILTIN_SETJMP is defined.
3281 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
3283 * target.def (insn_cost): New hook.
3284 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
3285 * doc/tm.texi: Regenerate.
3286 * rtlanal.c (insn_cost): Use the new hook.
3288 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
3290 * combine.c (combine_validate_cost): Compute the new insn_cost,
3291 not just pattern_cost.
3292 (try_combine): Adjust comment.
3294 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
3296 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
3298 * combine.c (uid_insn_cost): Adjust comment.
3299 (combine_validate_cost): Adjust comment. Use pattern_cost instead
3301 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
3302 * dse.c (find_shift_sequence): Ditto.
3303 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
3304 (bb_valid_for_noce_process_p): Use pattern_cost.
3305 * rtl.h (insn_rtx_cost): Delete.
3306 (pattern_cost): New prototype.
3307 (insn_cost): New prototype.
3308 * rtlanal.c (insn_rtx_cost): Rename to...
3309 (pattern_cost): ... this.
3312 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
3314 * config/i386/i386.md (*jcc_2): Remove insn pattern.
3315 (*jcc<mode>_0_r_i387): Ditto.
3316 (*jccxf_r_i387): Ditto.
3317 (*jcc<mode>_r_i387): Ditto.
3318 (*jccu<mode>_r_i387): Ditto.
3319 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
3320 (*jcc): Rename from *jcc_1.
3322 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3324 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
3325 deferred rescans after the lvx/stvx recombination pre-pass.
3327 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
3329 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
3330 memory operation instruction support.
3331 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
3332 (rs6000-ibm-aix[789]*): Likewise.
3333 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
3334 Document new functions.
3336 2017-10-09 Richard Biener <rguenther@suse.de>
3338 PR tree-optimization/82397
3339 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
3340 equality only for semantically equal trees.
3342 2017-10-09 Richard Biener <rguenther@suse.de>
3344 PR tree-optimization/82449
3345 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
3346 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
3347 allow constant addresses.
3348 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
3351 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3353 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
3356 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3359 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
3362 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3365 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
3367 2017-10-09 Jakub Jelinek <jakub@redhat.com>
3370 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
3371 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
3373 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
3375 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
3376 (WI_BINARY_PREDICATE_RESULT): Likewise.
3377 (wi::binary_traits::operator_result): New type.
3378 (wi::binary_traits::predicate_result): Likewise.
3379 (generic_wide_int::operator~, unary generic_wide_int::operator-)
3380 (generic_wide_int::operator==, generic_wide_int::operator!=)
3381 (generic_wide_int::operator&, generic_wide_int::and_not)
3382 (generic_wide_int::operator|, generic_wide_int::or_not)
3383 (generic_wide_int::operator^, generic_wide_int::operator+
3384 (binary generic_wide_int::operator-, generic_wide_int::operator*):
3386 (operator~, unary operator-, operator==, operator!=, operator&)
3387 (operator|, operator^, operator+, binary operator-, operator*): New
3389 * expr.c (get_inner_reference): Use wi::bit_and_not.
3390 * fold-const.c (fold_binary_loc): Likewise.
3391 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
3392 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
3393 (bit_value_binop): Likewise.
3394 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
3395 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
3396 (extract_range_from_binary_expr_1): Likewise.
3397 (masked_increment): Likewise.
3398 (simplify_bit_ops_using_ranges): Likewise.
3400 2017-10-09 Martin Jambor <mjambor@suse.cz>
3403 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
3404 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
3405 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
3407 (hsa_fixup_mov_insn_type): New function.
3408 (hsa_op_with_type::get_in_type): Use it.
3409 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
3410 immediates in an assert.
3411 (hsa_op_with_type::extend_int_to_32bit): New method.
3412 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
3413 types. Convert to dest type if necessary.
3414 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
3415 (reg_for_gimple_ssa): Pass false as min32int to
3416 hsa_type_for_scalar_tree_type.
3417 (gen_hsa_addr): Fixup type when creating addresable temporary.
3418 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
3419 (gen_hsa_unary_operation): Extend operands and convert to dest type if
3420 necessary. Call hsa_fixup_mov_insn_type.
3421 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
3422 extend operands and convert to dest type if necessary.
3423 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
3424 to dest type if necessary.
3425 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
3426 if conversion nt necessary and size matches.
3427 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
3428 to dest type if necessary.
3429 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
3430 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
3432 (gen_hsa_clrsb): Likewise.
3433 (gen_hsa_ffs): Likewise.
3434 (gen_hsa_divmod): Extend operands and convert to dest type if
3436 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
3438 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
3440 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
3441 Remove empty default arguments. Use a brace block as output
3443 (conditional return): Ditto.
3445 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
3446 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
3448 (group_ending_nop): Ditto.
3449 (doloop_end): Ditto.
3450 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
3451 (splitters for those): Ditto.
3453 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
3455 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
3456 a conditional jump (and the compare for it) so that pc_rtx is the
3458 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
3459 for the deleted and renamed ctr<mode>_internal[234] patterns.
3460 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
3461 Delete second conditional return pattern.
3462 (ctr<mode>_internal2): Delete this second bdnz pattern.
3463 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
3464 (ctr<mode>_internal4): Delete this second bdz pattern.
3466 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
3468 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
3469 (always_initialized_rtx_for_ssa_name_p): New predicate.
3470 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
3471 (finish_out_of_ssa): Free new field of SA.
3472 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
3473 * tree-ssa-coalesce.c: Include tree-ssa.h.
3474 (get_parm_default_def_partitions): Remove extern keyword.
3475 (get_undefined_value_partitions): New function.
3476 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
3477 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
3479 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
3481 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
3483 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
3485 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
3487 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
3488 for targets that preffer 128bit.
3490 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
3492 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
3494 2017-10-08 Olivier Hainque <hainque@adacore.com>
3496 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
3497 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
3499 (thumb_set_return_address): Likewise.
3501 2017-10-08 Olivier Hainque <hainque@adacore.com>
3503 * common/config/arm/arm-common.c (arm_except_unwind_info):
3504 Handle DWARF2_UNWIND_INFO.
3506 2017-10-07 Michael Collison <michael.collison@arm.com>
3508 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
3511 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
3513 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
3514 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
3515 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
3516 defined, force the creation of a new block for a dispatch label.
3518 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
3520 * invoke.texi (Wsuggest-attribute=cold): Document.
3521 * common.opt (Wsuggest-attribute=cold): New
3522 * ipa-pure-const.c (warn_function_cold): New function.
3523 * predict.c (compute_function_frequency): Use it.
3524 * predict.h (warn_function_cold): Declare.
3526 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
3528 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
3531 2017-10-06 Martin Liska <mliska@suse.cz>
3533 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
3534 keyword for member functions.
3535 (struct sanopt_tree_couple): New struct.
3536 (struct sanopt_tree_couple_hash): New function.
3537 (struct sanopt_ctx): Add new hash_map.
3538 (has_dominating_ubsan_ptr_check): New function.
3539 (record_ubsan_ptr_check_stmt): Likewise.
3540 (maybe_optimize_ubsan_ptr_ifn): Likewise.
3541 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
3542 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
3544 2017-10-06 Sudakshina Das <sudi.das@arm.com>
3547 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
3548 aarch64_simd_valid_immediate on CONST_VECTORs.
3549 (aarch64_reg_or_bic_imm): Likewise.
3551 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
3553 PR rtl-optimization/82396
3554 * haifa-sched.c (ready_sort_real): Disable qsort checking.
3556 2017-10-06 Sebastian Pop <sebpop@gmail.com>
3558 * graphite-dependences.c (scop_get_reads): Move code to...
3559 (scop_get_must_writes): Move code to...
3560 (scop_get_may_writes): Move code to...
3561 (scop_get_reads_and_writes): ... here.
3562 (scop_get_dependences): Call scop_get_reads_and_writes.
3564 2017-10-06 Jakub Jelinek <jakub@redhat.com>
3566 PR tree-optimization/82434
3567 * fold-const.h (can_native_encode_type_p,
3568 can_native_encode_string_p): Remove.
3569 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
3570 don't encode anything, just return what would be otherwise returned.
3571 (native_encode_fixed, native_encode_complex, native_encode_vector):
3573 (native_encode_string): Likewise. Inline by hand
3574 can_native_encode_string_p.
3575 (can_native_encode_type_p): Remove.
3576 (can_native_encode_string_p): Remove.
3577 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
3578 STRING_CSTs using can_native_encode_string_p, test all
3579 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
3580 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
3581 argument from native_encode_expr.
3582 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
3583 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
3586 2017-10-06 Richard Biener <rguenther@suse.de>
3588 PR tree-optimization/82397
3589 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
3590 operand_equal_p but rely on data_ref_compare_tree for detecting
3592 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
3593 to match up with dr_group_sort_cmp.
3595 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3598 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
3600 * config/s390/s390-builtin-types.def: Regenerate.
3602 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3605 * config/s390/s390-builtin-types.def: Regenerate.
3606 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
3607 Change flag from B_VXE to B_VX.
3608 (s390_vec_min_dbl): Remove B_VXE flag.
3610 2017-10-06 Richard Biener <rguenther@suse.de>
3612 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
3613 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
3614 translate_isl_ast_to_gimple::is_valid_rename,
3615 translate_isl_ast_to_gimple::get_rename,
3616 translate_isl_ast_to_gimple::get_def_bb_for_const,
3617 translate_isl_ast_to_gimple::get_new_name,
3618 translate_isl_ast_to_gimple::collect_all_ssa_names,
3619 translate_isl_ast_to_gimple::copy_loop_phi_args,
3620 translate_isl_ast_to_gimple::collect_all_ssa_names,
3621 translate_isl_ast_to_gimple::copy_loop_phi_args,
3622 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
3623 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
3624 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
3625 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
3626 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
3627 translate_isl_ast_to_gimple::copy_cond_phi_args,
3628 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
3629 translate_isl_ast_to_gimple::edge_for_new_close_phis,
3630 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
3631 translate_isl_ast_to_gimple::rename_uses,
3632 translate_isl_ast_to_gimple::rename_all_uses): Remove.
3633 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
3634 (set_rename_for_each_def): Likewise.
3635 (graphite_copy_stmts_from_block): Handle debug stmt resetting
3636 here. Handle rewriting SCEV analyzable uses here.
3637 (copy_bb_and_scalar_dependences): Generate code for PHI
3639 (graphite_regenerate_ast_isl): Adjust.
3640 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
3641 (add_write, add_read): New functions.
3642 (build_cross_bb_scalars_def): Use it and simplify.
3643 (build_cross_bb_scalars_use): Likewise.
3644 (graphite_find_cross_bb_scalar_vars): Inline into...
3645 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
3646 simulating out-of-SSA. Compute liveout and add dependencies.
3647 (build_scops): Force an empty entry block.
3648 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
3650 (sese_build_liveouts): Declare.
3651 (sese_trivially_empty_bb_p): Likewise.
3652 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
3653 compute liveout and debug_liveout.
3654 (sese_bad_liveouts_use): Remove.
3655 (sese_reset_debug_liveouts_bb): Likewise.
3656 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
3657 (sese_build_liveouts): Build liveout and debug_liveout and store
3659 (new_sese_info): Adjust.
3660 (free_sese_info): Likewise.
3661 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
3662 do not build liveout here.
3663 (move_sese_in_condition): Adjust region entry.
3664 (scev_analyzable_p): Match up with chrec_apply requirements.
3665 (sese_trivially_empty_bb_p): New.
3666 * tree-into-ssa.c (get_reaching_def): Properly support generating
3667 default-defs for incremental rewrite of anonymous names.
3669 2017-10-06 Richard Biener <rguenther@suse.de>
3671 * graphite-sese-to-poly.c (extract_affine): For casts increasing
3672 precision do not perform modulo reduction.
3674 2017-10-06 Richard Biener <rguenther@suse.de>
3676 PR tree-optimization/82436
3677 * tree-vect-slp.c (vect_supported_load_permutation_p): More
3678 conservatively choose the vectorization factor when checking
3679 whether we can perform the required load permutation.
3680 (vect_transform_slp_perm_load): Assert when we may not fail.
3682 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
3684 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
3685 message for incompatible -msdata=* and -mcall-* options.
3687 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
3689 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
3690 rate for post-reload scheduling.
3692 2017-10-05 Tamar Christina <tamar.christina@arm.com>
3694 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
3696 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
3698 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
3699 to improve monte carlo in scimark.
3701 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
3703 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
3704 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
3705 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
3706 pentium4_cost, nocona_cost): Set reassociation width to 1.
3707 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
3708 width to 2 for fp operations and 1 otherwise.
3709 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
3711 (atom_cost): Set reassociation width to 2.
3712 (slm_cost, generic_cost): Set fp reassociation width
3713 to 2 and 1 otherwise.
3714 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
3715 (core_cost): Set fp reassociation width to 4 and vector to 2.
3716 (ix86_reassociation_width): Rewrite using cost table; special case
3717 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
3718 and TARGET_AVX128_OPTIMAL.
3719 * config/i386/i386.h (processor_costs): Add
3720 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
3721 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
3722 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
3723 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
3724 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
3725 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
3727 2017-10-05 Nathan Sidwell <nathan@acm.org>
3729 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
3731 2017-10-05 Tamar Christina <tamar.christina@arm.com>
3733 * config/arm/arm.c (arm_test_fpu_data): New.
3734 (arm_run_selftests): Call arm_test_fpu_data.
3736 2017-10-04 Nathan Sidwell <nathan@acm.org>
3738 * toplev.c (toplev::main): Remove excess parens on pretty_printer
3740 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
3742 2017-10-04 Sudakshina Das <sudi.das@arm.com>
3744 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
3745 check type for aarch64_simd_valid_immediate.
3746 (aarch64_output_simd_mov_immediate): Update prototype.
3747 (aarch64_simd_valid_immediate): Update prototype.
3748 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
3749 support for ORR-immediate.
3750 (and<mode>3): modified pattern to add support for BIC-immediate.
3751 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
3752 now checks for valid immediate for BIC and ORR based on new enum
3754 (aarch64_output_simd_mov_immediate): Function now used to output
3755 BIC/ORR imm as well based on new enum argument.
3756 * config/aarch64/constraints.md (Do): New vector immediate constraint.
3758 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
3759 (aarch64_reg_or_bic_imm): Likewise.
3761 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3763 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
3764 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
3766 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
3770 PR rtl-optimization/82396
3771 * haifa-sched.c (autopref_multipass_init): Simplify
3773 (autopref_rank_data): Simplify sort order.
3774 * sched-int.h (autopref_multipass_data_): Remove
3775 multi_mem_insn_p, min_offset and max_offset.
3777 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3779 * doc/sourcebuild.texi: Document vect_peeling_profitable.
3781 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3783 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
3786 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3788 * doc/sourcebuild.texi: Document vect_long_mult.
3790 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
3792 PR tree-optimization/82413
3793 * fold-const.c (build_range_check): Use widest_int when comparing
3794 the maximum ETYPE value with HIGH.
3796 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
3798 PR rtl-optimization/82396
3799 * haifa-sched.c (autopref_multipass_init): Simplify
3801 (autopref_rank_data): Simplify sort order.
3802 * sched-int.h (autopref_multipass_data_): Remove
3803 multi_mem_insn_p, min_offset and max_offset.
3805 2017-10-04 Jakub Jelinek <jakub@redhat.com>
3807 PR tree-optimization/82381
3808 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
3809 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
3812 PR tree-optimization/82374
3813 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
3814 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
3815 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
3816 current_function_decl to the new decl.
3818 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
3820 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
3821 helper macro for IEEE float128 hardware built-in functions.
3822 (SQRTF128_ODD): Add built-in functions with the round-to-odd
3824 (TRUNCF128_ODD): Likewise.
3825 (ADDF128_ODD): Likewise.
3826 (SUBF128_ODD): Likewise.
3827 (MULF128_ODD): Likewise.
3828 (DIVF128_ODD): Likewise.
3829 (FMAF128_ODD): Likewise.
3830 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
3831 UNSPEC_TRUNC_ROUND_TO_ODD.
3832 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
3833 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
3834 floating point round to odd instructions.
3835 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
3836 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
3837 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
3838 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
3839 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
3840 (trunc<mode>sf2_hw): Change the truncate with round to odd
3841 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
3842 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
3843 to odd hardware instructions.
3844 (sub<mode>3_odd): Likewise.
3845 (mul<mode>3_odd): Likewise.
3846 (div<mode>3_odd): Likewise.
3847 (sqrt<mode>2_odd): Likewise.
3848 (fma<mode>4_odd): Likewise.
3849 (fms<mode>4_odd): Likewise.
3850 (nfma<mode>4_odd): Likewise.
3851 (nfms<mode>4_odd): Likewise.
3852 (trunc<mode>df2_odd): Change the truncate with round to odd
3853 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
3855 * doc/extend.texi (PowerPC built-in functions): Update documentation
3856 for existing IEEE float128-bit built-in functions. Add built-in
3857 functions that generate the IEEE 128-bit floating point round to
3860 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
3862 PR rtl-optimization/77729
3863 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
3864 to (X&(C1&~C2))|C2 transformations.
3866 2017-10-03 Martin Jambor <mjambor@suse.cz>
3868 PR tree-optimization/82363
3869 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
3870 mismatch, mark lacc written regardless of racc.
3872 2017-10-03 Jakub Jelinek <jakub@redhat.com>
3874 PR tree-optimization/82381
3875 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
3876 stmt_to_insert nor wheather SSA_NAMEs are default defs.
3877 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
3878 fallthrough into reassoc_stmt_dominates_stmt_p.
3881 * combine.c (combine_instructions): Don't combine in unreachable
3884 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
3887 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
3888 function to not use the have_cpu variable. Do not set cpu_index,
3889 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
3891 (rs6000_valid_attribute_p): Remove duplicate initializations of
3892 old_optimize and func_optimize.
3893 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
3894 (rs6000_activate_target_options): Make global.
3895 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
3898 2017-10-02 Jakub Jelinek <jakub@redhat.com>
3900 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
3901 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
3902 Set *poffset to 0 and *psize and *pmax_size to -1 if
3903 *poffset + *psize overflows in HOST_WIDE_INT.
3905 PR tree-optimization/82387
3906 PR tree-optimization/82388
3907 PR tree-optimization/82389
3908 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
3909 instead of live_bytes non-NULL.
3911 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
3914 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
3917 2017-10-02 Richard Biener <rguenther@suse.de>
3919 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
3920 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
3921 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
3923 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
3925 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
3926 requested precision matches the type's.
3927 * calls.c (alloc_max_size): Calculate the new candidate size as
3928 a widest_int and use wi::to_widest when comparing it with the
3929 current candidate size.
3930 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
3931 zero rather than integer_zero_node.
3932 * match.pd: Check for a no-op conversion before using wi::add
3933 rather than after. Use tree_to_uhwi when summing small shift
3934 counts into an unsigned int.
3936 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
3937 Alan Hayward <alan.hayward@arm.com>
3938 David Sherwood <david.sherwood@arm.com>
3941 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
3942 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
3943 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
3944 POINTER_AND_FP_REGS.
3946 2017-10-02 Richard Biener <rguenther@suse.de>
3948 PR tree-optimization/82355
3949 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
3950 a mapping for the enclosing loop but avoid generating one for
3952 (copy_bb_and_scalar_dependences): Remove premature codegen
3953 error on PHIs in blocks duplicated into multiple places.
3954 * graphite-scop-detection.c
3955 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
3956 in the region use it as loop and nest to analyze the DR in.
3957 (try_generate_gimple_bb): Likewise.
3958 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
3959 (add_loop_constraints): For blocks in a loop not in the region
3960 create a dimension with a single iteration.
3961 * sese.h (gbb_loop_at_index): Remove assert.
3963 2017-10-01 Kevin Buettner <kevinb@redhat.com>
3965 * omp-expand.c (adjust_context_scope): New function.
3966 (expand_parallel_call): Call adjust_context_scope.
3968 2017-10-01 Jeff Law <law@redhat.com>
3970 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
3971 dom_opt_dom_walker class with direct access to private members.
3972 Add comments. Call test_for_singularity.
3973 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
3974 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
3975 m_dummy_cond anymore.
3976 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
3978 (pass_dominator:execute): Build the dummy_cond here and pass it
3979 to the dom_opt_dom_walker ctor.
3980 (test_for_singularity): New function.
3982 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
3983 Maya Rashish <coypu@sdf.org>
3985 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
3986 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
3987 (alpha*-*-netbsd*) Use nbsd_tm_file.
3988 (arm*-*-netbsdelf*) Likewise.
3989 (i[34567]86-*-netbsdelf*) Likewise.
3990 (x86_64-*-netbsd*) Likewise.
3991 (mips*-*-netbsd*) Likewise.
3992 (powerpc-*-netbsd*) Likewise.
3993 (sh*-*-netbsd*) Likewise.
3994 (sparc-*-netbsdelf*) Likewise.
3995 (sparc64-*-netbsd*) Likewise.
3996 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
3998 (vax-*-netbsdelf*) Likewise.
3999 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
4000 (UINT_FAST8_TYPE) Likewise.
4001 (INT_FAST16_TYPE) Check CHAR_FAST16.
4002 (UINT_FAST16_TYPE) Likewise.
4004 2017-09-30 Jakub Jelinek <jakub@redhat.com>
4007 * config/i386/i386.md
4008 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
4009 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
4010 *divmodsi4_zext_2): New define_insn_and_split.
4011 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
4012 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
4013 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
4014 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
4015 New define_insn_and_split.
4016 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
4017 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
4018 operands[1] having DImode when mode is SImode.
4020 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
4021 always SImode for DIV and MOD in REG_EQUAL notes.
4023 2017-09-29 Yury Gribov <tetra2005@gmail.com>
4026 * match.pd: Fix handling of NaNs in pattern.
4028 2017-09-29 Jeff Law <law@redhat.com>
4030 * sbitmap.c (bitmap_bit_in_range_p): New function.
4031 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
4032 * tree-ssa-dse.c (live_bytes_read): New function.
4033 (dse_classify_store): Ignore reads of dead bytes.
4035 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
4036 typos and whitespace errors.
4037 * config/i386/predicates.md (address_no_seg_operand): Likewise.
4038 * config/s390/s390.c (s390_emit_prologue): Likewise.
4040 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
4043 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
4044 with a symbol for LRA.
4046 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
4048 PR rtl-optimization/82338
4049 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
4051 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
4053 * genmodes.c (calc_wider_mode): Suppress qsort macro.
4054 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
4055 (qsort_chk): Declare.
4056 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
4057 (qsort_chk): New function.
4059 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4061 PR tree-optimization/82337
4062 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
4063 phi definition if the PHI result appears in an abnormal PHI.
4064 (find_basis_for_base_expr): Don't record a basis if the LHS of the
4065 basis appears in an abnormal PHI.
4067 2017-09-29 Richard Biener <rguenther@suse.de>
4069 * graphite-isl-ast-to-gimple.c
4070 (translate_isl_ast_to_gimple::set_codegen_error): New function.
4071 (binary_op_to_tree): Use it.
4072 (get_rename_from_scev): Likewise.
4073 (copy_loop_phi_nodes): Likewise.
4074 (copy_bb_and_scalar_dependences): Likewise.
4075 (translate_pending_phi_nodes): Likewise.
4077 2017-09-29 Jakub Jelinek <jakub@redhat.com>
4080 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
4081 for movabsq $(i32 << shift), r64.
4083 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
4085 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
4086 index when encoding %esp as %rsp to avoid 0x67 prefix.
4088 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
4090 * config/i386/i386.md (*movsf_internal, *movdf_internal):
4091 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
4093 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
4095 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
4096 Extensions with more than 16 double VFP registers.
4097 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
4098 to_clear_mask and all code related to it. Replace the remaining
4099 entry by a sbitmap and adapt code accordingly.
4101 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
4103 * brig-builtins.def: Change pure attributes to const.
4105 2017-09-28 Joseph Myers <joseph@codesourcery.com>
4107 * config.gcc (default_gnu_indirect_function): Default to yes for
4108 sparc*-*-linux* with glibc.
4110 2017-09-28 Joseph Myers <joseph@codesourcery.com>
4112 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
4113 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
4114 when creating .init_array and .fini_array sections with priority
4117 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
4120 * config/aarch64/aarch64.c
4121 (aarch64_builtin_support_vector_misalignment): Always return false
4122 when misalignment is unknown.
4124 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
4126 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
4127 this function to return false if the definition used by the swap
4128 instruction is artificial, or if the memory address from which the
4129 constant value is loaded is not represented by a base address held
4130 in a register or if the base address register is a frame or stack
4131 pointer. Additionally, return false if the base address of the
4132 loaded constant is a SYMBOL_REF but is not considered to be a
4134 (replace_swapped_load_constant): New function.
4135 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
4136 loaded constant vector with a load of a swapped constant vector.
4138 2017-09-27 Carl Love <cel@us.ibm.com>
4140 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
4141 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
4142 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
4145 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
4147 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
4148 first, always call autopref_rank_data otherwise.
4150 2017-09-27 Richard Biener <rguenther@suse.de>
4152 * graphite-scop-detection.c (find_scop_parameters): Move
4153 loop bound handling ...
4154 (gather_bbs::before_dom_children): ... here, avoiding the need
4155 to build scop_info->loop_nest.
4156 (record_loop_in_sese): Remove.
4157 * sese.h (sese_info_t::loop_nest): Remove.
4158 * sese.c (new_sese_info): Do not allocate loop_nest.
4159 (free_sese_info): Do not free loop_nest.
4161 2017-09-27 Jakub Jelinek <jakub@redhat.com>
4164 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
4165 lhs from calls if the lhs has addressable type.
4167 2017-09-27 Richard Biener <rguenther@suse.de>
4169 * graphite.h (scop::max_alias_set): New member.
4170 * graphite-scop-detection.c: Remove references to non-existing
4171 --param in comments.
4172 (build_alias_sets): Record the maximum alias set used for drs.
4173 (build_scops): Support zero as unlimited for
4174 --param graphite-max-arrays-per-scop.
4175 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
4177 (build_poly_sr_1): ... here. Compute alias set based on the
4178 maximum alias set used for drs rather than
4179 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
4181 2017-09-27 Richard Biener <rguenther@suse.de>
4183 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
4184 --param loop-block-tile-size=0 to disable tiling.
4186 2017-09-27 Richard Biener <rguenther@suse.de>
4188 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
4189 (graphite-max-nb-scop-params): Document special value zero.
4190 * domwalk.h (dom_walker::STOP): New symbolical constant.
4191 (dom_walker::dom_walker): Add optional parameter for bb to
4193 (dom_walker::~dom_walker): Declare.
4194 (dom_walker::before_dom_children): Document STOP return value.
4195 (dom_walker::m_user_bb_to_rpo): New member.
4196 (dom_walker::m_bb_to_rpo): Likewise.
4197 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
4198 mapping here if not provided by the user.
4199 (dom_walker::~dom_walker): Free bb to RPO mapping if not
4200 provided by the user.
4201 (dom_walker::STOP): Define.
4202 (dom_walker::walk): Do not compute bb to RPO mapping here.
4203 Support STOP return value from before_dom_children to stop
4205 * graphite-optimize-isl.c (optimize_isl): If the schedule
4206 is the same still generate code if -fgraphite-identity
4207 or -floop-parallelize-all are given.
4208 * graphite-scop-detection.c: Include cfganal.h.
4209 (gather_bbs::gather_bbs): Get and pass through bb to RPO
4211 (gather_bbs::before_dom_children): Return STOP for BBs
4213 (build_scops): Compute bb to RPO mapping and pass it to
4214 the domwalk. Treat --param graphite-max-nb-scop-params=0
4215 as not limiting the number of params.
4216 * graphite.c (graphite_initialize): Remove limit on the
4217 number of basic-blocks in a function.
4218 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
4219 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
4220 default value of 10.
4222 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4224 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
4225 Adjust code to eliminate needing to do the shift right 32-bits
4226 operation after XSCVDPSPN.
4228 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4230 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
4231 ((X / Y) != 0 -> X >= Y): Likewise.
4233 2017-09-26 Carl Love <cel@us.ibm.com>
4235 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
4236 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
4237 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
4238 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
4239 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
4240 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
4241 definitions and overloading.
4242 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
4243 statement for P9V_BUILTIN_XST_LEN_R.
4244 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
4245 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
4246 define_expand and define_insn for the instructions and builtins.
4247 * doc/extend.texi: Update the built-in documentation file for the new
4249 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
4250 define_insn for the instructions
4252 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
4255 * gcc/config/netbsd-protos.h: New file.
4256 * gcc/config/netbsd.c: New file.
4257 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
4258 * gcc/config/t-netbsd: New file.
4259 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
4260 (tmake_file) Add t-netbsd.
4261 (extra_objs) Add netbsd.o.
4263 2017-09-26 Janus Weil <janus@gcc.gnu.org>
4267 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
4269 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4271 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
4272 sign extension from a vector register to a GPR by doing a 32-bit
4273 direct move and then an EXTSW.
4274 (extendsi<mode>2 splitter): Likewise.
4275 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
4276 right or vector extract after doing XSCVDPSPN. Use
4277 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
4279 (movdi_from_sf_zero_ext): Likewise.
4280 (reload_gpr_from_vsxsf): Likewise.
4281 (p8_mfvsrd_4_disf): Delete, no longer used.
4282 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
4283 then needing to move the SFmode to a GPR to use the XSCVDPSP
4284 instruction instead of FRSP and XSCVDPSPN.
4285 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
4286 it is adjacent to the other XSCVSPDP insns.
4287 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
4288 SFmode to be in traditional Altivec registers.
4289 (vsx_xscvdpspn): Eliminate useless alternative constraint.
4290 (vsx_xscvspdpn): Likewise.
4291 (vsx_xscvspdpn_scalar): Likewise.
4293 2017-09-26 Martin Jambor <mjambor@suse.cz>
4295 * tree-sra.c (compare_access_positions): Put integral types first,
4296 stabilize sorting of integral types, remove conditions putting
4297 non-full-precision integers last.
4298 (sort_and_splice_var_accesses): Disable scalarization if a
4299 non-integert would be represented by a non-full-precision integer.
4301 2017-09-26 Joseph Myers <joseph@codesourcery.com>
4303 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
4304 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
4305 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
4306 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
4307 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
4308 conditionals inside the function instead of around it. Call
4309 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
4310 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
4312 2017-09-26 Richard Biener <rguenther@suse.de>
4314 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
4316 (scop_detection::build_scop_breadth): ... this. Removed.
4317 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
4318 (scop_detection::harmful_stmt_in_bb): Likewise.
4319 (scop_detection::graphite_can_represent_stmt): Likewise.
4320 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
4321 (scop_detection::can_represent_loop): Remove recursion, fold in ...
4322 (scop_detection::can_represent_loop_1): ... this. Removed.
4323 (scop_detection::harmful_loop_in_region): Simplify after inlining
4324 the above and remove more quadraticness.
4325 (build_scops): Adjust.
4326 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
4329 2017-09-26 Jakub Jelinek <jakub@redhat.com>
4332 * config/i386/i386.c (ix86_print_operand_address_as): Only test
4333 REGNO (base) == SP_REG if base is a REG.
4336 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
4337 if it is different SSA_NAME.
4338 (optimize_range_tests_cmp_bitwise): New function.
4339 (optimize_range_tests): Call it.
4341 2017-09-26 Richard Biener <rguenther@suse.de>
4343 PR tree-optimization/82321
4344 * graphite.c (canonicalize_loop_closed_ssa): Properly check
4345 for the def being inside the loop.
4347 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4349 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
4351 * config/s390/s390-builtins.def: Fix constraint on op4.
4353 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4355 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
4356 independent expanders.
4357 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
4358 ("vec_ordered", "vec_unordered"): New expanders.
4360 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4362 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
4365 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4367 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
4368 vec_unpacks_lo_v16qi.
4369 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
4371 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4373 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
4374 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
4375 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
4377 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4379 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
4381 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
4383 ("*vec_slb<mode>"): New insn pattern.
4384 ("vec_shr_<mode>"): New expander.
4385 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
4386 and force the shift count operand to V16QImode.
4387 ("vec_srb<mode>"): Set shift count mode to V16QI.
4389 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4391 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
4392 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
4393 ("vec_widen_smult_hi_<mode>"): New expander definitions.
4395 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
4398 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
4400 2017-09-26 Richard Biener <rguenther@suse.de>
4402 PR tree-optimization/82320
4403 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
4406 2017-09-25 Jeff Law <law@redhat.com>
4408 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
4409 prototype for new argument.
4410 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
4411 mostly extracted from rs6000_emit_allocate_stack.
4412 (rs6000_emit_probe_stack_range_stack_clash): New function.
4413 (rs6000_emit_allocate_stack): Call
4414 rs6000_emit_probe_stack_range_stack_clash as needed.
4415 (rs6000_emit_probe_stack_range): Add additional argument
4416 to call to gen_probe_stack_range{si,di}.
4417 (output_probe_stack_range): New.
4418 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
4419 (output_probe_stack_range_stack_clash): New.
4420 (rs6000_emit_prologue): Emit notes into dump file as requested.
4421 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
4422 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
4423 Add additional operand and pass it to output_probe_stack_range.
4425 2017-09-25 Bin Cheng <bin.cheng@arm.com>
4427 PR tree-optimization/82163
4428 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
4429 (checking_verify_loop_closed_ssa): New parameter.
4430 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
4431 (check_loop_closed_ssa_stmt): Delete.
4432 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
4433 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
4434 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
4437 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
4439 * brig-builtins.def: Treat HSAIL barrier builtins as
4440 setjmp/longjump style functions.
4442 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
4444 * target.def (constant_alignment): New hook.
4445 * defaults.h (CONSTANT_ALIGNMENT): Delete.
4446 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
4447 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
4448 * doc/tm.texi: Regenerate.
4449 * targhooks.h (default_constant_alignment): Declare.
4450 (constant_alignment_word_strings): Likewise.
4451 * targhooks.c (default_constant_alignment): New function.
4452 (constant_alignment_word_strings): Likewise.
4453 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
4454 instead of CONSTANT_ALIGNMENT.
4455 * varasm.c (align_variable, get_variable_align, build_constant_desc)
4456 (force_const_mem): Likewise.
4457 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
4458 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
4459 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
4460 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4461 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
4463 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
4464 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4465 constant_alignment_word_strings.
4466 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
4467 (CONSTANT_ALIGNMENT): Likewise.
4468 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4469 (arm_constant_alignment): New function.
4470 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
4471 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4472 constant_alignment_word_strings.
4473 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
4474 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4475 constant_alignment_word_strings.
4476 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
4477 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4478 (cris_constant_alignment): New function.
4479 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
4480 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4481 (epiphany_constant_alignment): New function.
4482 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
4483 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4484 constant_alignment_word_strings.
4485 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
4486 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4487 constant_alignment_word_strings.
4488 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
4489 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4490 constant_alignment_word_strings.
4491 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
4492 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
4493 * config/i386/i386.c (ix86_constant_alignment): Make static.
4494 Use the same interface as the target hook.
4495 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4496 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
4497 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4498 constant_alignment_word_strings.
4499 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
4500 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
4501 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4502 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
4503 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4504 constant_alignment_word_strings.
4505 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
4506 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4507 constant_alignment_word_strings.
4508 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
4509 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4510 constant_alignment_word_strings.
4511 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
4512 * config/microblaze/microblaze.c (microblaze_constant_alignment):
4514 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4515 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
4516 * config/mips/mips.c (mips_constant_alignment): New function.
4517 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4518 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
4519 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
4520 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4521 (mmix_constant_alignment): Make static. Use the same interface
4523 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
4524 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4525 constant_alignment_word_strings.
4526 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
4527 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4528 constant_alignment_word_strings.
4529 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
4530 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4531 constant_alignment_word_strings.
4532 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
4533 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4534 (rs6000_constant_alignment): New function.
4535 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
4536 * config/riscv/riscv.c (riscv_constant_alignment): New function.
4537 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4538 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
4539 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4540 (rs6000_constant_alignment): New function.
4541 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
4542 * config/s390/s390.c (s390_constant_alignment): New function.
4543 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4544 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
4545 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4546 constant_alignment_word_strings.
4547 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
4548 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4549 (sparc_constant_alignment): New function.
4550 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
4551 * config/spu/spu.c (spu_constant_alignment): New function.
4552 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4553 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
4554 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4555 constant_alignment_word_strings.
4556 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
4557 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4558 constant_alignment_word_strings.
4559 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
4560 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4561 constant_alignment_word_strings.
4562 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
4563 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4564 (visium_constant_alignment): New function.
4565 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
4566 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4567 (xtensa_constant_alignment): New function.
4568 * system.h (CONSTANT_ALIGNMENT): Poison.
4570 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
4572 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
4573 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
4574 (rs6000_builtin_valid_without_lhs): New helper function.
4575 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4576 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
4578 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
4580 * target.h (vec_perm_indices): Use unsigned short rather than
4582 (auto_vec_perm_indices): Likewise.
4583 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
4584 Use unsigned int rather than unsigned char.
4585 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
4587 2017-09-25 Richard Biener <rguenther@suse.de>
4589 * cfgloop.h (sort_sibling_loops): Declare.
4590 * cfgloop.c (sort_sibling_loops_cmp): New helper.
4591 (sort_sibling_loops): New function sorting the sibling loop list
4593 * graphite.c (graphite_transform_loops): Sort sibling loops.
4595 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
4597 * target.def (vec_perm_const_ok): Change sel parameter to
4599 * optabs-query.c (can_vec_perm_p): Update accordingly.
4600 * doc/tm.texi: Regenerate.
4601 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
4602 auto_vec_perm_indices and remove separate nelt field.
4603 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
4604 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
4605 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
4606 (aarch64_expand_vec_perm_const): Update accordingly.
4607 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
4608 to vec_perm_indices.
4609 * config/arm/arm.c (expand_vec_perm_d): Change perm to
4610 auto_vec_perm_indices and remove separate nelt field.
4611 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
4612 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
4613 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
4615 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
4616 to vec_perm_indices.
4617 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
4618 sel to vec_perm_indices.
4619 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
4620 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
4621 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
4623 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
4626 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
4629 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
4630 on the FUNCTION_DECL function context if it has a DIE that is a
4633 2017-09-25 Richard Biener <rguenther@suse.de>
4635 PR tree-optimization/82285
4636 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
4639 2017-09-25 Tom de Vries <tom@codesourcery.com>
4643 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
4646 2017-09-25 Richard Biener <rguenther@suse.de>
4648 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
4649 ISL errors other than isl_error_quota happen. Dump if the
4650 schedule is the same.
4651 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
4652 errors instead of aborting inside ISL.
4654 2017-09-25 Iain Sandoe <iain@codesourcery.com>
4657 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
4658 of libgcc_eh for m64.
4659 * config/i386/darwin64.h: Likewise.
4661 2017-09-25 Richard Biener <rguenther@suse.de>
4664 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
4665 attribute for incomplete types nor twice for complete ones.
4667 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
4670 * config/i386/i386.c (ix86_print_operand_address_as): Encode
4671 %esp as %rsp to avoid 0x67 prefix if there is no index or base
4674 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
4677 * config/i386/i386.opt (mprefer-avx256): Use
4678 ix86_target_flags variable.
4679 * config/i386/i386.c (ix86_target_string): Move
4680 -mprefer-avx256 to flag2_opts.
4682 2017-09-22 Jakub Jelinek <jakub@redhat.com>
4685 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
4686 and x != -1 | y != -1 into (x & y) != -1.
4688 2017-09-22 Steve Ellcey <sellcey@cavium.com>
4690 * config.gcc: Add new case statement to set
4691 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
4692 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
4693 s390x-*-linux* case statements. Added aarch64 to the list of
4694 supported architectures.
4696 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4698 PR tree-optimization/82289
4699 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
4700 STMT_VINFO_RELEVANT_P.
4702 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4703 Alan Hayward <alan.hayward@arm.com>
4704 David Sherwood <david.sherwood@arm.com>
4706 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
4707 for VR_RANGE only; don't allow VR_ANTI_RANGE.
4708 (extract_range_from_binary_expr_1): Don't call
4709 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
4711 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4712 Alan Hayward <alan.hayward@arm.com>
4713 David Sherwood <david.sherwood@arm.com>
4715 * target.def (preferred_vector_alignment): New hook.
4716 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
4718 * doc/tm.texi: Regenerate.
4719 * targhooks.h (default_preferred_vector_alignment): Declare.
4720 * targhooks.c (default_preferred_vector_alignment): New function.
4721 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
4723 (DR_TARGET_ALIGNMENT): New macro.
4724 (aligned_access_p): Update commentary.
4725 (vect_known_alignment_in_bytes): New function.
4726 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
4728 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
4729 Calculate the misalignment based on the target alignment rather than
4731 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
4732 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
4733 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
4734 the target alignment, rather than masking the element misalignment
4735 with the number of elements in a vector. Also use the target
4736 alignment when calculating the maximum number of peels.
4737 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
4738 instead of TYPE_ALIGN_UNIT.
4739 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
4740 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
4741 (vect_create_addr_base_for_vector_ref): Update call accordingly.
4742 (vect_create_data_ref_ptr): Likewise.
4743 (vect_setup_realignment): Realign by ANDing with
4744 -DR_TARGET_MISALIGNMENT.
4745 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
4746 the number of peels based on DR_TARGET_ALIGNMENT.
4747 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
4748 with the guaranteed alignment boundary when deciding whether
4750 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
4751 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
4752 (ensure_base_align): Remove stmt_info parameter. Get the
4753 target base alignment from DR_TARGET_ALIGNMENT.
4754 (vectorizable_store): Update call accordingly. Interpret
4755 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
4757 (vectorizable_load): Likewise.
4759 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4760 Alan Hayward <alan.hayward@arm.com>
4761 David Sherwood <david.sherwood@arm.com>
4763 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
4764 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
4765 (vect_enhance_data_refs_alignment): Likewise.
4767 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
4769 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
4770 error. Only quit immediately if parsing is complete.
4771 (BEGIN): Initialize fatal_err and parse_done.
4772 (begin fpu, end fpu): Check number of arguments.
4773 (begin arch, end arch): Likewise.
4774 (begin cpu, end cpu): Likewise.
4775 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
4776 (optalias): Likewise.
4778 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
4780 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
4781 * config/arm/arm-isa.h: Delete. Move definitions to ...
4782 * arm-cpus.in: ... here. Use new feature and fgroup values.
4783 * config/arm/arm.c (arm_option_override): Use lower case for feature
4785 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
4786 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
4787 * config/arm/parsecpu.awk (END): Add new command 'isa'.
4789 (print_isa_bits_for): New function.
4790 (gen_isa): New function.
4791 (gen_comm_data): Use print_isa_bits_for.
4792 (define feature): New keyword.
4793 (define fgroup): New keyword.
4794 * config/arm/t-arm (TM_H): Remove.
4795 (GTM_H): Add arm-isa.h.
4796 (arm-isa.h): Add rule to generate file.
4797 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
4798 case for feature bit names.
4800 2017-09-22 Richard Biener <rguenther@suse.de>
4802 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
4804 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
4805 print of no dependency loops ...
4806 * graphite.c (graphite_transform_loops): ... here.
4807 (canonicalize_loop_closed_ssa_form): Work from inner to outer
4809 (same_close_phi_node, remove_duplicate_close_phi,
4810 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
4811 (canonicalize_loop_closed_ssa): ... here and simplify.
4812 * graphite-optimize-isl.c: Include tree-vectorizer.h.
4813 (optimize_isl): Use dump_printf_loc to tell when we stopped
4814 optimizing because of an ISL timeout.
4816 2017-09-22 Richard Biener <rguenther@suse.de>
4818 PR tree-optimization/82291
4819 * tree-if-conv.c (predicate_mem_writes): Make sure to
4820 remove writes in blocks predicated with false.
4822 2017-09-22 Richard Biener <rguenther@suse.de>
4824 * sese.c: Include cfganal.h.
4825 (if_region_set_false_region): Remove.
4826 (create_if_region_on_edge): Likewise.
4827 (move_sese_in_condition): Re-implement without destroying
4830 2017-09-22 Richard Biener <rguenther@suse.de>
4832 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
4833 Verify both BBs contain loop PHI nodes before dispatching to
4835 (graphite_regenerate_ast_isl): Do not recompute dominators,
4836 do not verify three times. Restructure for clarity.
4837 * graphite-scop-detection.c (same_close_phi_node,
4838 remove_duplicate_close_phi, make_close_phi_nodes_unique,
4839 defined_in_loop_p, canonicalize_loop_closed_ssa,
4840 canonicalize_loop_closed_ssa_form): Simplify, remove excess
4841 checking and SSA rewrite, move to ...
4842 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
4843 (graphite_initialize): Do not pass in ctx, do not reset the
4844 SCEV cache, compute only dominators.
4845 (graphite_transform_loops): Allocate ISL ctx after
4846 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
4847 Maintain post-dominators only around build_scops.
4848 * sese.c (if_region_set_false_region): Make static. Free
4849 and recompute dominators.
4850 (move_sese_in_condition): Assert we don't get called with
4851 post-dominators computed.
4852 * sese.h (if_region_set_false_region): Remove.
4854 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
4856 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
4857 mode attribute for TARGET_AVX512VL.
4859 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
4861 * config/i386/i386.opt (mprefer-avx256): New option.
4862 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
4864 (ix86_preferred_simd_mode): Return 256-bit AVX modes
4865 for TARGET_PREFER_AVX256.
4866 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
4868 2017-09-21 Jeff Law <law@redhat.com>
4870 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
4871 Fix dump output if the only stack space is for pushed registers.
4873 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4875 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
4878 2017-09-21 Martin Sebor <msebor@redhat.com>
4881 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
4882 code (in C++) or code that triggers warnings.
4884 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
4886 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
4888 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
4890 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
4891 * sched-rgn.c: Ditto.
4892 * sel-sched-ir.c: Ditto.
4894 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
4896 * toplev.h (set_random_seed): Adjust return type.
4897 * toplev.c (init_local_tick): Move eager initialization of random_seed
4898 to get_random_seed. Adjust comment.
4899 (init_random_seed): Inline to get_random_seed, delete.
4900 (get_random_seed): Initialize random_seed lazily.
4901 (set_random_seed): Do not return previous value.
4902 (print_switch_value): Do not call get_random_seed.
4904 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
4906 * cgraph.c (delete_function_version): New, broken out from...
4907 (cgraph_node::delete_function_version): ...here. Rename to
4908 cgraph_node::delete_function_version_by_decl. Update all uses.
4909 (cgraph_node::remove): Call delete_function_version.
4911 2017-09-21 Jakub Jelinek <jakub@redhat.com>
4914 * tree-inline.c (expand_call_inline): Emit clobber stmts for
4915 VAR_DECLs to which addressable non-volatile parameters are mapped
4916 and for id->retvar after the return value assignment. Clear
4917 id->retval and id->retbnd after inlining.
4919 2017-09-21 Richard Biener <rguenther@suse.de>
4921 PR tree-optimization/82276
4922 PR tree-optimization/82244
4923 * tree-vrp.c (build_assert_expr_for): Set
4924 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
4926 (remove_range_assertions): Revert earlier change.
4928 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
4931 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
4933 2017-09-21 Richard Biener <rguenther@suse.de>
4935 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4936 Restore valid IL after code generation errors.
4937 * graphite.c (graphite_transform_loops): Diagnose code
4938 generation issues as MSG_MISSED_OPTIMIZATION and continue
4939 with processing SCOPs.
4941 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4942 Alan Hayward <alan.hayward@arm.com>
4943 David Sherwood <david.sherwood@arm.com>
4945 * calls.c (compute_argument_addresses): Use simplify_gen_binary
4946 rather than choosing between plus_constant and gen_rtx_<CODE>.
4947 * expr.c (emit_push_insn): Likewise.
4948 (expand_expr_real_2): Likewise.
4950 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4951 Alan Hayward <alan.hayward@arm.com>
4952 David Sherwood <david.sherwood@arm.com>
4954 * loop-unroll.c (split_iv): Call copy_rtx on the step.
4956 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4957 Alan Hayward <alan.hayward@arm.com>
4958 David Sherwood <david.sherwood@arm.com>
4960 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
4961 calling tree_to_uhwi.
4963 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4964 Alan Hayward <alan.hayward@arm.com>
4965 David Sherwood <david.sherwood@arm.com>
4967 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
4968 INTEGER_CST rather than a negative test for ADDR_EXPR.
4970 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4971 Alan Hayward <alan.hayward@arm.com>
4972 David Sherwood <david.sherwood@arm.com>
4974 * tree-vrp.c (extract_range_from_binary_expr_1): Check
4975 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
4977 2017-09-21 Richard Biener <rguenther@suse.de>
4979 PR tree-optimization/71351
4980 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
4981 graphite_create_new_loop_guard): Remove, fold remaining parts
4983 (translate_isl_ast_node_for): ... here and simplify.
4985 2017-09-21 Jakub Jelinek <jakub@redhat.com>
4988 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
4989 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
4990 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
4991 alternative always use QI mode, for -Os imov (=R,R) alternative
4992 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
4995 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4996 Jeff Law <law@redhat.com>
4998 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
4999 (allocate_stack_space): New function, partially extracted from
5001 (s390_emit_prologue): Track offset to most recent stack probe.
5002 Code to allocate space moved into allocate_stack_space.
5003 Dump actions when no stack is allocated.
5004 (s390_prologue_plus_offset): New function.
5005 (s390_emit_stack_probe): Likewise.
5007 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
5009 * common.opt (Wa, Wl, Wp, g, gz=): Add
5011 (gno-column-info): Remove.
5012 (gcolumn-info): Drop RejectNegative.
5014 (gno-record-gcc-switches): Remove.
5015 (grecord-gcc-switches): Drop RejectNegative.
5016 (gno-split-dwarf): Remove.
5017 (gsplit-dwarf): Drop RejectNegative.
5018 (gno-strict-dwarf): Remove.
5019 (gstrict-dwarf): Drop RejectNegative.
5020 * config/darwin.opt (gfull, gused): Add RejectNegative.
5021 * dwarf2out.c (gen_producer_string): Drop
5022 gno-record-gcc-switches handler.
5023 * optc-gen.awk: Add g to prefixes with negative forms.
5024 * opts-common.c (remapping_prefix_p): New.
5025 (find_opt): Check it.
5026 (generate_canonical_option): Test g prefix.
5027 (option_map): Add -gno- mapping.
5028 (add_misspelling_candidates): Check remapping_prefix_p.
5030 2017-09-20 Jeff Law <law@redhat.com>
5032 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
5033 thinko in stack clash protection support.
5035 * explow.c (compute_stack_clash_protection_loop_data): Use
5036 CONST_INT_P instead of explicit test. Verify object is a
5037 CONST_INT_P before looking at INTVAL.
5038 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
5039 instead of explicit test.
5041 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
5044 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
5045 address instead of to r1 and r11.
5047 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
5049 * config.gcc: Support "knm".
5050 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
5051 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
5053 * config/i386/i386.c (m_KNM): Define.
5054 (processor_target_table): Add "knm".
5056 (ix86_option_override_internal): Add "knm".
5057 (ix86_issue_rate): Add PROCESSOR_KNM.
5058 (ix86_adjust_cost): Ditto.
5059 (ia32_multipass_dfa_lookahead): Ditto.
5060 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
5061 (fold_builtin_cpu): Add M_INTEL_KNM.
5062 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
5063 (processor_type): Add PROCESSOR_KNM.
5064 * config/i386/x86-tune.def: Add m_KNM.
5065 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
5067 2017-09-20 Richard Biener <rguenther@suse.de>
5069 PR tree-optimization/80213
5070 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
5071 are allowed in empty BBs as well.
5072 (canonicalize_loop_closed_ssa): Also look for other complex
5074 (scop_detection::get_sese): Include the loop-closed PHI block
5076 (scop_detection::merge_sese): Remove code adding extra blocks.
5077 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
5078 (build_scops): Assert the final returned scop is invalid.
5080 2017-09-20 Richard Biener <rguenther@suse.de>
5082 PR tree-optimization/82264
5083 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
5085 (vn_phi_lookup): Likewise.
5086 (vn_phi_insert): Likewise.
5088 2017-09-20 Jakub Jelinek <jakub@redhat.com>
5090 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
5091 that fits into uhwi or shwi, add DW_AT_const_value regardless
5092 of early_dwarf without going through RTL, using add_AT_unsigned
5095 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
5096 (DEBUG_LTO_ABBREV_SECTION): Likewise.
5097 (DEBUG_LTO_MACINFO_SECTION): Likewise.
5098 (DEBUG_MACRO_SECTION): Likewise.
5099 (DEBUG_LTO_MACRO_SECTION): Likewise.
5100 (DEBUG_STR_DWO_SECTION): Likewise.
5101 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
5102 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
5103 (DEBUG_LTO_DWO_LINE_SECTION): Define.
5104 (DEBUG_LTO_LINE_STR_SECTION): Define.
5105 (init_sections_and_labels): Initialize debug_line_str_section
5106 variable. Initialize debug_loc_section for -gdwarf-5 to
5107 DEBUG_LOCLISTS_SECTION. Formatting fixes.
5109 2017-09-20 Richard Biener <rguenther@suse.de>
5111 * graphite-sese-to-poly.c (extract_affine): Properly handle
5112 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
5114 2017-09-20 Richard Biener <rguenther@suse.de>
5116 PR tree-optimization/81373
5117 * graphite-scop-detection.c (build_cross_bb_scalars_def):
5118 Force SESE live-out defs to be handled even if they are
5121 2017-09-19 Jeff Law <law@redhat.com>
5123 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
5124 nothing for stack adjustments with REG_STACK_CHECK.
5125 * sched-deps.c (parse_add_or_inc): Reject insns with
5126 REG_STACK_CHECK from dependency breaking.
5127 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
5128 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
5129 * reg-notes.def (STACK_CHECK): New note.
5131 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
5132 (ix86_expand_prologue): Dump stack clash info as needed.
5133 Call ix86_adjust_stack_and_probe_stack_clash as needed.
5135 * function.c (dump_stack_clash_frame_info): New function.
5136 * function.h (dump_stack_clash_frame_info): Prototype.
5137 (enum stack_clash_probes): New enum.
5139 * config/alpha/alpha.c (alpha_expand_prologue): Also check
5140 flag_stack_clash_protection.
5141 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
5142 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
5143 (arm_frame_pointer_required): Likewise.
5144 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
5145 (ia64_expand_prologue): Likewise.
5146 * config/mips/mips.c (mips_expand_prologue): Likewise.
5147 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
5148 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
5149 (sparc_flat_expand_prologue): Likewise.
5150 * config/spu/spu.c (spu_expand_prologue): Likewise.
5152 * explow.c: Include "params.h".
5153 (anti_adjust_stack_and_probe_stack_clash): New function.
5154 (get_stack_check_protect): Likewise.
5155 (compute_stack_clash_protection_loop_data): Likewise.
5156 (emit_stack_clash_protection_loop_start): Likewise.
5157 (emit_stack_clash_protection_loop_end): Likewise.
5158 (allocate_dynamic_stack_space): Use get_stack_check_protect.
5159 Use anti_adjust_stack_and_probe_stack_clash.
5160 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
5161 (emit_stack_clash_protection_loop_start): Likewise.
5162 (emit_stack_clash_protection_loop_end): Likewise.
5163 * rtl.h (get_stack_check_protect): Prototype.
5164 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
5165 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
5166 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
5168 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
5170 * doc/tm.texi: Rebuilt.
5171 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
5172 get_stack_check_protect.
5173 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
5174 * config/arm/arm.c (arm_expand_prologue): Likewise.
5175 (arm_frame_pointer_required): Likewise.
5176 * config/i386/i386.c (ix86_expand_prologue): Likewise.
5177 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
5178 * config/mips/mips.c (mips_expand_prologue): Likewise.
5179 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
5180 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
5181 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
5182 (sparc_flat_expand_prologue): Likewise.
5184 * common.opt (-fstack-clash-protection): New option.
5185 * flag-types.h (enum stack_check_type): Note difference between
5186 -fstack-check= and -fstack-clash-protection.
5187 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
5188 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
5189 * toplev.c (process_options): Issue warnings/errors for cases
5190 not handled with -fstack-clash-protection.
5191 * doc/invoke.texi (-fstack-clash-protection): Document new option.
5192 (-fstack-check): Note additional problem with -fstack-check=generic.
5193 Note that -fstack-check is primarily for Ada and refer users
5194 to -fstack-clash-protection for stack-clash-protection.
5195 Document new params for stack clash protection.
5197 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
5199 * config/i386/i386.c (ix86_split_long_move): Do not handle
5200 address used for LEA in a special way.
5202 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
5204 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
5206 2017-09-19 Martin Sebor <msebor@redhat.com>
5209 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
5210 of incompatible types.
5212 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
5214 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5215 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
5216 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5217 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
5219 2017-09-19 Richard Biener <rguenther@suse.de>
5221 PR tree-optimization/82244
5222 * tree-vrp.c (remove_range_assertions): Do not propagate
5223 a constant to abnormals but replace the assert with a copy.
5225 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
5227 PR rtl-optimization/57878
5228 PR rtl-optimization/68988
5229 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
5230 avoidance test involving non_reload_pseudos. Move frequency test
5231 below the general fragmentation avoidance test.
5233 2017-09-19 Richard Biener <rguenther@suse.de>
5235 PR tree-optimization/69728
5236 * graphite-sese-to-poly.c (schedule_error): New global.
5237 (add_loop_schedule): Handle empty domain by failing the
5239 (build_original_schedule): Handle schedule_error.
5241 2017-09-19 Richard Biener <rguenther@suse.de>
5243 * graphite-scop-detection.c (scop_detection::can_represent_loop):
5244 Do not iterate to sibling loops but only to siblings of inner
5247 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
5250 * config/m68k/m68k.md (moveq feeding equality comparison): Check
5251 that the registers are different.
5253 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
5255 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
5256 to processor_model and "amdfam17h" to arch_names_table.
5257 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
5259 2017-09-18 Jakub Jelinek <jakub@redhat.com>
5262 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
5264 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5265 Alan Hayward <alan.hayward@arm.com>
5266 David Sherwood <david.sherwood@arm.com>
5268 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
5270 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
5272 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
5273 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
5274 vect_schedule_slp_instance.
5275 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
5276 Update call to vect_slp_analyze_node_operations. Simplify return
5278 (vect_slp_analyze_bb_1): Update call accordingly.
5279 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
5280 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
5281 (vect_schedule_slp): Update call accordingly.
5283 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5284 Alan Hayward <alan.hayward@arm.com>
5285 David Sherwood <david.sherwood@arm.com>
5287 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
5288 with types that aren't in fact scalar.
5290 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5292 * tree-vect-slp.c (vect_record_max_nunits): New function,
5294 (vect_build_slp_tree_1): ...here.
5295 (vect_build_slp_tree_2): Call it for phis too.
5297 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5299 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
5300 to vect_get_vec_def_for_operand when getting the mask operand.
5302 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5303 Alan Hayward <alan.hayward@arm.com>
5304 David Sherwood <david.sherwood@arm.com>
5306 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
5309 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5310 Alan Hayward <alan.hayward@arm.com>
5311 David Sherwood <david.sherwood@arm.com>
5313 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
5314 calculation for vector booleans.
5316 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5317 Alan Hayward <alan.hayward@arm.com>
5318 David Sherwood <david.sherwood@arm.com>
5320 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
5322 (vect_transform_stmt): ...here.
5323 (vect_analyze_stmt): Use it instead of calling
5324 vectorizable_live_operation directly.
5326 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
5328 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
5329 non-SIMT targets in acc vector loops.
5331 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
5333 * configure.ac: Add arc and check if assembler supports gdwarf2.
5334 * configure: Regenerate.
5336 2017-09-18 Richard Biener <rguenther@suse.de>
5338 PR tree-optimization/82220
5339 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
5340 epilogue niters from the min_profitable_iters compute.
5342 2017-09-18 Jakub Jelinek <jakub@redhat.com>
5345 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
5346 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
5347 (ix86_init_pic_reg): Revert 2017-09-01 changes.
5349 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
5352 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
5353 switching to a new text section.
5355 2017-09-18 Richard Biener <rguenther@suse.de>
5357 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
5359 (build_alias_set): Reject aliases with no access function.
5361 2017-09-18 Richard Biener <rguenther@suse.de>
5363 PR tree-optimization/79622
5364 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
5366 (build_cross_bb_scalars_use): Likewise.
5368 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
5370 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
5372 2017-09-18 Alan Modra <amodra@gmail.com>
5375 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
5376 stack_pointer_rtx for count 0. Update comments. Break up
5377 large rtl expression.
5379 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
5381 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
5382 Increase to 20 bytes.
5383 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
5384 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
5385 or avx version of the stub.
5387 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
5390 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
5391 compute the minimum stack alignment. Also update preferred stack
5392 boundary for leaf functions.
5394 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
5396 PR tree-optimization/82228
5397 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
5400 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
5402 * common/config/nds32/nds32-common.c
5403 (nds32_option_optimization_table): Refine formatting.
5404 (nds32_option_optimization_table): Use -fsched-pressure and
5405 -fomit-frame-pointer for specific optimization level.
5407 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
5409 * config/nds32/nds32.c: Refine formatting and comments.
5410 * config/nds32/nds32.h: Likewise.
5411 * config/nds32/nds32.md: Likewise.
5412 * config/nds32/nds32-cost.c: Likewise.
5413 * config/nds32/nds32-isr.c: Likewise.
5414 * config/nds32/nds32-md-auxiliary.c: Likewise.
5415 * config/nds32/nds32-multiple.md: Likewise.
5416 * config/nds32/nds32-predicates.c: Likewise.
5418 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
5419 Jakub Jelinek <jakub@redhat.com>
5421 Add support for -std=c++2a.
5422 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
5424 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
5426 2017-09-15 Steve Ellcey <sellcey@cavium.com>
5429 * doc/extend.texi (Common Function Attributes): Add
5430 references to ARM, AArch64, and S/390 specific attributes.
5431 (Function Specific Option Pragmas): Add AArch64 and S/390
5432 to list of back ends that support the target pragma.
5434 2017-09-15 Nathan Sidwell <nathan@acm.org>
5436 * doc/standards.texi: Fix C++17 description. Update URLs for
5439 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
5441 * common.opt (Wcast-align=strict): New warning option.
5442 * doc/invoke.texi: Document -Wcast-align=strict.
5444 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
5446 * cgraph.h (cgraph_thunk_info): Add comments.
5447 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
5448 assert for VIRTUAL_* arguments stricter.
5450 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
5452 PR tree-optimization/71026
5453 * match.pd: Move RDIV patterns from fold-const.c
5454 * fold-const.c (distribute_real_division): Removed.
5455 (fold_binary_loc): Remove calls to distribute_real_divison.
5457 2017-09-15 Jakub Jelinek <jakub@redhat.com>
5459 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
5460 c++1z and gnu++1z as deprecated. Change other references to
5461 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
5462 Change -Wc++1z-compat to -Wc++17-compat.
5463 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
5464 * dwarf2out.c (highest_c_language): Handle C++17.
5465 (gen_compile_unit_die): Likewise.
5467 2017-09-15 Jakub Jelinek <jakub@redhat.com>
5469 PR rtl-optimization/82192
5470 * combine.c (make_extraction): Don't look through non-paradoxical
5471 SUBREGs or TRUNCATE if pos + len is or might be bigger than
5474 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
5475 Alan Hayward <alan.hayward@arm.com>
5476 David Sherwood <david.sherwood@arm.com>
5478 * target.def (function_arg_offset): New hook.
5479 * targhooks.h (default_function_arg_offset): Declare.
5480 * targhooks.c (default_function_arg_offset): New function.
5481 * function.c (locate_and_pad_parm): Use
5482 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
5483 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
5484 (TARGET_FUNCTION_ARG_OFFSET): ...this.
5485 * doc/tm.texi: Regenerate.
5486 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
5487 * config/spu/spu.c (spu_function_arg_offset): New function.
5488 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
5489 * system.h (FUNCTION_ARG_OFFSET): Poison.
5491 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
5492 Alan Hayard <alan.hayward@arm.com>
5493 David Sherwood <david.sherwood@arm.com>
5495 * target.def (truly_noop_truncation): New hook.
5496 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
5497 than TRULY_NOOP_TRUNCATION.
5498 * hooks.h (hook_bool_uint_uint_true): Declare.
5499 * hooks.c (hook_bool_uint_uint_true): New function.
5500 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
5501 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
5502 * doc/tm.texi: Regenerate.
5503 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
5504 rather than TRULY_NOOP_TRUNCATION in comments.
5505 (simplify_comparison): Likewise.
5506 (record_truncated_value): Likewise.
5507 * expmed.c (extract_bit_field_1): Likewise.
5508 (extract_split_bit_field): Likewise.
5509 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
5510 instead of TRULY_NOOP_TRUNCATION.
5511 * function.c (assign_parm_setup_block): Likewise.
5512 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
5513 * rtlhooks.c: Include target.h.
5514 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
5515 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
5516 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
5517 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
5518 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
5519 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
5520 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
5521 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
5522 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
5523 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
5524 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
5525 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
5526 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
5527 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
5528 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
5529 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
5530 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
5531 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
5532 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
5533 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
5534 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
5535 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
5536 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
5537 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
5538 * config/mips/mips.c (mips_truly_noop_truncation): New function.
5539 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
5540 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
5541 rather than TRULY_NOOP_TRUNCATION in comments.
5542 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
5543 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
5544 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
5545 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
5546 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
5547 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
5548 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
5549 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
5550 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
5551 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
5552 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
5553 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
5554 rather than TRULY_NOOP_TRUNCATION in comments.
5555 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
5556 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
5557 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
5558 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
5559 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
5560 TRULY_NOOP_TRUNCATION condition.
5561 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
5562 (TRULY_NOOP_TRUNCATION): Delete.
5563 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
5564 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
5565 * config/spu/spu.c (spu_truly_noop_truncation): New function.
5566 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
5567 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
5568 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
5569 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
5570 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
5571 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
5572 rather than TRULY_NOOP_TRUNCATION in comments.
5573 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
5574 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
5575 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
5576 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
5577 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
5578 * system.h (TRULY_NOOP_TRUNCATION): Poison.
5580 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
5583 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
5584 (*cmp_ior): Likewise.
5585 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
5586 (*ior_scc_scc_cmp): Likewise.
5587 (*and_scc_scc): Likewise.
5588 (*and_scc_scc_cmp): Likewise.
5590 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
5591 Alan Hayard <alan.hayward@arm.com>
5592 David Sherwood <david.sherwood@arm.com>
5594 * target.def (can_change_mode_class): New hook.
5595 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
5596 (hard_regno_nregs): Likewise.
5597 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
5598 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
5599 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
5600 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
5601 (LOAD_EXTEND_OP): Update accordingly.
5602 * doc/tm.texi: Regenerate.
5603 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
5604 CANNOT_CHANGE_MODE_CLASS.
5605 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
5606 (REG_CAN_CHANGE_MODE_P): ...this new macro.
5607 * combine.c (simplify_set): Update accordingly.
5608 * emit-rtl.c (validate_subreg): Likewise.
5609 * recog.c (general_operand): Likewise.
5610 * regcprop.c (mode_change_ok): Likewise.
5611 * reload1.c (choose_reload_regs): Likewise.
5612 (inherit_piecemeal_p): Likewise.
5613 * rtlanal.c (simplify_subreg_regno): Likewise.
5614 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
5615 instead of CANNOT_CHANGE_MODE_CLASS.
5616 (reload_cse_simplify_operands): Likewise.
5617 * reload.c (push_reload): Use targetm.can_change_mode_class
5618 instead of CANNOT_CHANGE_MODE_CLASS.
5619 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
5620 REG_CANNOT_CHANGE_MODE_P.
5621 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5622 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
5623 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5624 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5625 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5626 (arm_can_change_mode_class): New function.
5627 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
5628 than CANNOT_CHANGE_MODE_CLASS in comments.
5629 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5630 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
5631 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
5632 (ix86_can_change_mode_class): ...this new function, inverting the
5633 sense of the return value.
5634 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5635 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5636 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5637 (ia64_can_change_mode_class): New function.
5638 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5639 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
5640 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
5641 (m32c_can_change_mode_class): ...this new function, inverting the
5642 sense of the return value.
5643 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5644 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5645 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
5646 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
5647 (mips_can_change_mode_class): ...this new function, inverting the
5648 sense of the return value.
5649 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5650 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5651 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5652 (msp430_can_change_mode_class): New function.
5653 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5654 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
5655 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5656 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5657 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5658 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
5659 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5660 (pa_cannot_change_mode_class): Replace with...
5661 (pa_can_change_mode_class): ...this new function, inverting the
5662 sense of the return value.
5663 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
5664 than CANNOT_CHANGE_MODE_CLASS in comments.
5665 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5666 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
5667 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5668 (pdp11_cannot_change_mode_class): Replace with...
5669 (pdp11_can_change_mode_class): ...this new function, inverting the
5670 sense of the return value.
5671 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5672 * config/powerpcspe/powerpcspe-protos.h
5673 (rs6000_cannot_change_mode_class_ptr): Delete.
5674 * config/powerpcspe/powerpcspe.c
5675 (rs6000_cannot_change_mode_class_ptr): Delete.
5676 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5677 (rs6000_option_override_internal): Assign to
5678 targetm.can_change_mode_class instead of
5679 rs6000_cannot_change_mode_class_ptr.
5680 (rs6000_cannot_change_mode_class): Replace with...
5681 (rs6000_can_change_mode_class): ...this new function, inverting the
5682 sense of the return value.
5683 (rs6000_debug_cannot_change_mode_class): Replace with...
5684 (rs6000_debug_can_change_mode_class): ...this new function.
5685 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5686 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
5687 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5688 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5689 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
5691 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
5692 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5693 (rs6000_option_override_internal): Assign to
5694 targetm.can_change_mode_class instead of
5695 rs6000_cannot_change_mode_class_ptr.
5696 (rs6000_cannot_change_mode_class): Replace with...
5697 (rs6000_can_change_mode_class): ...this new function, inverting the
5698 sense of the return value.
5699 (rs6000_debug_cannot_change_mode_class): Replace with...
5700 (rs6000_debug_can_change_mode_class): ...this new function.
5701 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5702 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
5703 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
5704 (s390_can_change_mode_class): ...this new function, inverting the
5705 sense of the return value.
5706 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5707 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5708 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
5709 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5710 (sh_cannot_change_mode_class): Replace with...
5711 (sh_can_change_mode_class): ...this new function, inverting the
5712 sense of the return value.
5713 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5714 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5715 (sparc_can_change_mode_class): New function.
5716 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5717 * config/spu/spu.c (spu_can_change_mode_class): New function.
5718 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5719 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5720 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5721 (visium_can_change_mode_class): New function.
5722 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
5724 2017-09-15 Richard Biener <rguenther@suse.de>
5726 PR tree-optimization/82217
5727 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
5728 but not undefined case.
5730 2017-09-15 Jakub Jelinek <jakub@redhat.com>
5733 * postreload.c (reload_cse_simplify_operands): Skip
5734 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
5736 2017-09-15 Richard Biener <rguenther@suse.de>
5738 PR tree-optimization/68823
5739 * graphite-scop-detection.c (build_alias_set): If we have a
5740 possible dependence check whether we can handle them by just
5741 looking at the DRs DR_ACCESS_FNs.
5742 (build_scops): If build_alias_set fails, fail the SCOP.
5744 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
5746 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
5747 to support float128 built-in functions that require the ISA 3.0
5749 (BU_FLOAT128_3_HW): Likewise.
5750 (SQRTF128): Add support for the IEEE 128-bit square root and fma
5752 (FMAF128): Likewise.
5754 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
5755 support for built-in functions that need the ISA 3.0 IEEE 128-bit
5756 floating point instructions.
5757 (rs6000_invalid_builtin): Likewise.
5758 (rs6000_builtin_mask_names): Likewise.
5759 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
5760 (RS6000_BTM_FLOAT128_HW): Likewise.
5761 (RS6000_BTM_COMMON): Likewise.
5762 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
5764 * doc/extend.texi (RS/6000 built-in functions): Document the
5765 IEEE 128-bit floating point square root and fused multiply-add
5768 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
5770 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
5771 reg (r2) isn't in the set of registers defined in the prologue.
5773 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5774 Alan Hayward <alan.hayward@arm.com>
5775 David Sherwood <david.sherwood@arm.com>
5777 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
5778 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
5779 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
5780 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
5781 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
5783 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
5784 max_vectorization_factor.
5785 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
5787 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5788 Alan Hayward <alan.hayward@arm.com>
5789 David Sherwood <david.sherwood@arm.com>
5791 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
5792 (vect_worthwhile_without_simd_p): Declare.
5793 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
5794 (vectorizable_reduction): Use it.
5795 * tree-vect-stmts.c (vectorizable_shift): Likewise.
5796 (vectorizable_operation): Likewise.
5798 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5799 Alan Hayward <alan.hayward@arm.com>
5800 David Sherwood <david.sherwood@arm.com>
5802 * tree-vectorizer.h (vect_get_num_copies): New function.
5803 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
5804 * tree-vect-loop.c (vectorizable_reduction): Likewise.
5805 (vectorizable_induction): Likewise.
5806 (vectorizable_live_operation): Likewise.
5807 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
5808 (vectorizable_bswap): Likewise.
5809 (vectorizable_call): Likewise.
5810 (vectorizable_conversion): Likewise.
5811 (vectorizable_assignment): Likewise.
5812 (vectorizable_shift): Likewise.
5813 (vectorizable_operation): Likewise.
5814 (vectorizable_store): Likewise.
5815 (vectorizable_load): Likewise.
5816 (vectorizable_condition): Likewise.
5817 (vectorizable_comparison): Likewise.
5818 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
5820 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5821 Alan Hayward <alan.hayward@arm.com>
5822 David Sherwood <david.sherwood@arm.com>
5824 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
5825 of vect_init_vector.
5827 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5828 Alan Hayward <alan.hayward@arm.com>
5829 David Sherwood <david.sherwood@arm.com>
5831 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
5832 an inline wrapper that provides a location.
5833 (gimple_build_vector): Likewise.
5834 * gimple-fold.c (gimple_build_vector_from_val): New function.
5835 (gimple_build_vector): Likewise.
5836 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
5837 functions to build the initial value. Always return a gimple value.
5838 (get_initial_defs_for_reduction): Likewise. Only compute
5840 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
5841 vect_init_vector on the results from get_initial_def(s)_for_reduction.
5842 (vectorizable_induction): Use gimple_build_vector rather than
5845 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5846 Alan Hayward <alan.hayward@arm.com>
5847 David Sherwood <david.sherwood@arm.com>
5849 * target.h (vec_perm_indices): New typedef.
5850 (auto_vec_perm_indices): Likewise.
5851 * optabs-query.h: Include target.h
5852 (can_vec_perm_p): Take a vec_perm_indices *.
5853 * optabs-query.c (can_vec_perm_p): Likewise.
5854 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
5855 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
5856 * tree-vect-generic.c (lower_vec_perm): Likewise.
5857 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
5858 (vect_grouped_load_supported): Likewise.
5859 (vect_shift_permute_load_chain): Likewise.
5860 (vect_permute_store_chain): Use auto_vec_perm_indices.
5861 (vect_permute_load_chain): Likewise.
5862 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
5863 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
5864 Update uses of can_vec_perm_p.
5865 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
5866 mode with a number of elements. Take a vec_perm_indices *.
5867 (vect_create_epilog_for_reduction): Update accordingly.
5868 Use auto_vec_perm_indices.
5869 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
5870 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
5871 (vect_transform_slp_perm_load): Likewise.
5872 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
5873 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
5874 (vect_gen_perm_mask_checked): Likewise.
5875 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
5876 (vect_gen_perm_mask_checked): Likewise.
5877 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
5878 (vectorizable_store): Likewise.
5879 (vectorizable_load): Likewise.
5880 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
5881 (vectorizable_bswap): Likewise.
5883 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5884 Alan Hayward <alan.hayward@arm.com>
5885 David Sherwood <david.sherwood@arm.com>
5887 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
5888 * tree.c (build_vector): Likewise.
5889 (build_vector_from_ctor): Update accordingly.
5890 (build_vector_from_val): Likewise.
5891 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
5892 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
5893 * tree-vect-generic.c (add_rshift): Likewise.
5894 (expand_vector_divmod): Likewise.
5895 (optimize_vector_constructor): Likewise.
5896 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
5897 (vect_transform_slp_perm_load): Likewise.
5898 (vect_schedule_slp_instance): Likewise.
5899 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
5900 (vectorizable_call): Likewise.
5901 (vect_gen_perm_mask_any): Likewise. Add elements in order.
5902 * expmed.c (make_tree): Likewise.
5903 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
5904 a vector passed to build_vector.
5905 (fold_convert_const): Likewise.
5906 (exact_inverse): Likewise.
5907 (fold_ternary_loc): Likewise.
5908 (fold_relational_const): Likewise.
5909 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
5910 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
5911 (const_unop): Likewise. Store the reduction accumulator in a
5912 variable rather than an array.
5913 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
5914 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
5915 the new vector, rather than constructing it after the input arrays.
5916 (native_interpret_vector): Use auto_vec<tree> when building
5917 a vector passed to build_vector. Add elements in order.
5918 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
5919 auto_vec<tree> when building a vector passed to build_vector.
5920 (vect_create_epilog_for_reduction): Likewise.
5921 (vectorizable_induction): Likewise.
5922 (get_initial_def_for_reduction): Likewise. Fix indentation of
5924 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
5926 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
5927 passed to build_vector.
5929 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5930 Alan Hayward <alan.hayward@arm.com>
5931 David Sherwood <david.sherwood@arm.com>
5933 * tree-core.h (tree_base::u): Add an "nelts" field.
5934 (tree_vector): Use VECTOR_CST_NELTS as the length.
5935 * tree.c (tree_size): Likewise.
5936 (make_vector): Initialize VECTOR_CST_NELTS.
5937 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
5938 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
5939 TYPE_VECTOR_SUBPARTS.
5940 * expr.c (const_vector_mask_from_tree): Consistently use "units"
5941 as the number of units, setting it from VECTOR_CST_NELTS.
5942 (const_vector_from_tree): Likewise.
5943 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
5944 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
5945 (fold_negate_expr_1): Likewise.
5946 (fold_convert_const): Likewise.
5947 (const_binop): Likewise. Differentiate the number of output and
5949 (const_unop): Likewise.
5950 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
5951 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
5952 in cases that did the opposite.
5954 2017-09-14 Richard Biener <rguenther@suse.de>
5956 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
5959 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
5961 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
5963 2017-09-14 Jakub Jelinek <jakub@redhat.com>
5966 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
5967 if and where to split a bb, except for splitting before debug insn
5968 sequences followed by non-label real insn. Delete debug insns
5969 in between basic blocks.
5971 * combine.c (make_compound_operation_int): Formatting fixes.
5973 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
5974 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5975 * config/netbsd.h (LINK_EH_SPEC): Likewise.
5976 * config/sol2.h (LINK_EH_SPEC): Likewise.
5977 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5978 * config/s390/linux.h (LINK_SPEC): Likewise.
5979 * config/freebsd.h (LINK_EH_SPEC): Likewise.
5980 * config/openbsd.h (LINK_EH_SPEC): Likewise.
5981 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5982 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
5983 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
5984 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5985 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
5986 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
5988 2017-09-13 Jakub Jelinek <jakub@redhat.com>
5990 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
5992 (ENDFILE_LINUX_SPEC): Likewise.
5993 (LINK_EH_SPEC): Likewise.
5994 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
5995 (LINK_OS_LINUX_SPEC32): Likewise.
5996 (LINK_OS_LINUX_SPEC64): Likewise.
5997 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
5998 (LINK_OS_LINUX_SPEC): Likewise.
6000 2017-09-13 Martin Liska <mliska@suse.cz>
6003 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
6004 CASE_HIGH is NULL_TREE.
6006 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
6007 Alan Hayward <alan.hayward@arm.com>
6008 David Sherwood <david.sherwood@arm.com>
6010 * target.def (secondary_memory_needed): New hook.
6011 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
6012 instead of SECONDARY_MEMORY_NEEDED.
6013 (secondary_memory_needed_mode): Likewise.
6014 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
6015 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
6016 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
6017 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
6018 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
6019 * doc/tm.texi: Regenerate.
6020 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
6021 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
6022 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6023 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
6024 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
6025 * config/i386/i386.c (inline_secondary_memory_needed): Put the
6026 mode argument first and change the reg_class arguments to reg_class_t.
6027 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
6028 Make static. Update the call to inline_secondary_memory_needed.
6029 (ix86_register_move_cost): Update the call to
6030 inline_secondary_memory_needed.
6031 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6032 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
6034 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
6035 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
6037 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
6038 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
6039 * config/mips/mips.c (mips_secondary_memory_needed): Make static
6040 and match hook interface. Add comment from mips.h.
6041 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6042 * config/mmix/mmix.md (truncdfsf2): Refer to
6043 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
6045 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
6046 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
6047 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6048 (pa_secondary_memory_needed): New function.
6049 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
6050 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
6051 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6052 (pdp11_secondary_memory_needed): Make static and match hook interface.
6053 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
6054 * config/powerpcspe/powerpcspe-protos.h
6055 (rs6000_secondary_memory_needed_ptr): Delete.
6056 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
6058 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6059 (rs6000_option_override_internal): Assign to
6060 targetm.secondary_memory_needed rather than
6061 rs6000_secondary_memory_needed_ptr.
6062 (rs6000_secondary_memory_needed): Match hook interface.
6063 (rs6000_debug_secondary_memory_needed): Likewise.
6064 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
6065 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
6066 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
6067 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6068 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
6069 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
6071 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
6072 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6073 (rs6000_option_override_internal): Assign to
6074 targetm.secondary_memory_needed rather than
6075 rs6000_secondary_memory_needed_ptr.
6076 (rs6000_secondary_memory_needed): Match hook interface.
6077 (rs6000_debug_secondary_memory_needed): Likewise.
6078 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
6079 * config/s390/s390.c (s390_secondary_memory_needed): New function.
6080 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6081 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
6082 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6083 (sparc_secondary_memory_needed): New function.
6084 * lra-constraints.c (check_and_process_move): Refer to
6085 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
6087 (curr_insn_transform): Likewise.
6088 (process_alt_operands): Use targetm.secondary_memory_needed
6089 instead of TARGET_SECONDARY_MEMORY_NEEDED.
6090 (check_secondary_memory_needed_p): Likewise.
6091 (choose_split_class): Likewise.
6092 * reload.c: Unconditionally include code that was previously
6093 conditional on SECONDARY_MEMORY_NEEDED.
6094 (push_secondary_reload): Use targetm.secondary_memory_needed
6095 instead of TARGET_SECONDARY_MEMORY_NEEDED.
6096 (push_reload): Likewise.
6097 * reload1.c: Unconditionally include code that was previously
6098 conditional on SECONDARY_MEMORY_NEEDED.
6099 (choose_reload_regs): Use targetm.secondary_memory_needed
6100 instead of TARGET_SECONDARY_MEMORY_NEEDED.
6101 (gen_reload): Likewise.
6102 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
6104 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
6105 Alan Hayward <alan.hayward@arm.com>
6106 David Sherwood <david.sherwood@arm.com>
6108 * target.def (secondary_memory_needed_mode): New hook:
6109 * targhooks.c (default_secondary_memory_needed_mode): Declare.
6110 * targhooks.h (default_secondary_memory_needed_mode): New function.
6111 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
6112 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
6113 * doc/tm.texi: Regenerate.
6114 * lra-constraints.c (check_and_process_move): Use
6115 targetm.secondary_memory_needed_mode instead of
6116 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
6117 (curr_insn_transform): Likewise.
6118 * reload.c (get_secondary_mem): Likewise.
6119 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6120 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
6122 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
6123 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6124 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
6125 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
6126 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
6128 * config/powerpcspe/powerpcspe-protos.h
6129 (rs6000_secondary_memory_needed_mode): Delete.
6130 * config/powerpcspe/powerpcspe.c
6131 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
6132 (rs6000_secondary_memory_needed_mode): Make static.
6133 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6134 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
6136 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
6138 (rs6000_secondary_memory_needed_mode): Make static.
6139 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6140 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
6141 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
6142 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6143 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
6145 (sparc_secondary_memory_needed_mode): New function.
6146 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
6148 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
6150 * config/aarch64/constraints.md (Umq): New constraint.
6151 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
6153 (mov<mode>): Update condition.
6155 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6157 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
6158 when bitposition is the same.
6160 2017-09-13 Richard Biener <rguenther@suse.de>
6162 * dwarf2out.c (output_die_symbol): Remove.
6163 (output_die): Do not output a DIEs symbol.
6165 2017-09-13 Richard Biener <rguenther@suse.de>
6168 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
6169 default-def to avoid breaking iterator update with the weird
6170 interaction with cgraph_update_edges_for_call_stmt_node.
6172 2017-09-13 Richard Biener <rguenther@suse.de>
6174 * tree-cfg.c (verify_gimple_assign_binary): Add verification
6175 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
6176 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
6177 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
6179 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
6181 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
6182 Disable pc relative literal load irrespective of
6183 TARGET_FIX_ERR_A53_84341 for default.
6185 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
6187 * config/sparc/sparc.c (output_return): Output the source location of
6188 the insn in the delay slot, if any.
6189 (output_sibcall): Likewise.
6191 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
6194 * common.opt (-static-pie): New alias.
6195 (shared): Negate static-pie.
6196 (-no-pie): Update help text.
6198 (static-pie): New option.
6199 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
6200 -static-pie support.
6201 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
6202 (LINK_EH_SPEC): Likewise.
6203 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6204 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
6205 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
6206 * gcc.c (LINK_COMMAND_SPEC): Likewise.
6207 (init_gcc_specs): Likewise.
6208 (init_spec): Likewise.
6209 (display_help): Update help message for -pie.
6210 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
6213 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
6215 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
6216 (movdi_aarch64): Likewise.
6217 (movti_aarch64): Likewise.
6219 2017-09-12 Simon Wright <simon@pushface.org>
6222 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
6223 calculation of the minor version, always output as 0.
6225 2017-09-12 Jakub Jelinek <jakub@redhat.com>
6228 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
6229 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
6230 on it early, rather than manual conversion late. For
6231 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
6232 instead of performing manual conversion.
6234 2017-09-12 Carl Love <cel@us.ibm.com>
6236 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
6237 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
6239 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
6240 VMULOSW): Add definitions.
6241 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6242 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
6243 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
6244 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
6245 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
6247 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
6249 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
6251 (movti_aarch64): Likewise.
6252 (movdf_aarch64): Likewise.
6253 (movtf_aarch64): Likewise.
6254 (load_pairdi): Likewise.
6255 (store_pairdi): Likewise.
6256 (load_pairdf): Likewise.
6257 (store_pairdf): Likewise.
6258 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
6259 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
6260 (ldr_got_small_<mode>): Likewise.
6261 (ldr_got_small_28k_<mode>): Likewise.
6262 (ldr_got_tiny): Likewise.
6263 * config/aarch64/iterators.md (ldst_sz): New.
6264 (ldpstp_sz): Likewise.
6265 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
6267 (thunderx_load): Split load_8 to load_16.
6268 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
6270 (thunderx2t99_storepair_basic): Split store_8 to store_16.
6271 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
6272 (xgene1_store_pair): Split store_8 to store_16.
6273 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
6274 (falkor_st_0_st_sd): Split store_8 to store_16.
6276 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
6278 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
6279 and store1/2/3/4 to store_4/8/12/16.
6280 * config/aarch64/aarch64.md: Update for rename.
6281 * config/arm/arm.md: Likewise.: Likewise.
6282 * config/arm/arm.c: Likewise.
6283 * config/arm/thumb1.md: Likewise.
6284 * config/arm/thumb2.md: Likewise.
6285 * config/arm/vfp.md: Likewise.
6286 * config/arm/arm-generic.md: Likewise.
6287 * config/arm/arm1020e.md: Likewise.
6288 * config/arm/arm1026ejs.md: Likewise.
6289 * config/arm/arm1136jfs.md: Likewise.
6290 * config/arm/arm926ejs.md: Likewise.
6291 * config/arm/cortex-a15.md: Likewise.
6292 * config/arm/cortex-a17.md: Likewise.
6293 * config/arm/cortex-a5.md: Likewise.
6294 * config/arm/cortex-a53.md: Likewise.
6295 * config/arm/cortex-a57.md: Likewise.
6296 * config/arm/cortex-a7.md: Likewise.
6297 * config/arm/cortex-a8.md: Likewise.
6298 * config/arm/cortex-a9.md: Likewise.
6299 * config/arm/cortex-m4.md: Likewise.
6300 * config/arm/cortex-m7.md: Likewise.
6301 * config/arm/cortex-r4.md: Likewise.
6302 * config/arm/exynos-m1.md: Likewise.
6303 * config/arm/fa526.md: Likewise.
6304 * config/arm/fa606te.md: Likewise.
6305 * config/arm/fa626te.md: Likewise.
6306 * config/arm/fa726te.md: Likewise.
6307 * config/arm/fmp626.md: Likewise.
6308 * config/arm/iwmmxt.md: Likewise.
6309 * config/arm/ldmstm.md: Likewise.
6310 * config/arm/marvell-pj4.md: Likewise.
6311 * config/arm/xgene1.md: Likewise.
6312 * config/aarch64/thunderx.md: Likewise.
6313 * config/aarch64/thunderx2t99.md: Likewise.
6314 * config/aarch64/falkor.md: Likewise.
6316 2017-09-12 Martin Liska <mliska@suse.cz>
6318 * attribs.c (private_lookup_attribute): New function.
6319 * attribs.h (private_lookup_attribute): Declared here.
6320 (lookup_attribute): Called from this place.
6322 2017-09-12 Richard Biener <rguenther@suse.de>
6324 PR tree-optimization/82157
6325 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
6326 stmts with side-effects.
6328 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6329 Alan Hayward <alan.hayward@arm.com>
6330 David Sherwood <david.sherwood@arm.com>
6332 * target.def (hard_regno_nregs): New hook.
6333 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
6334 * targhooks.h (default_hard_regno_nregs): Declare.
6335 * targhooks.c (default_hard_regno_nregs): New function.
6336 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
6337 (TARGET_HARD_REGNO_NREGS): ...this hook.
6338 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
6339 (CLASS_MAX_NREGS): Likewise.
6340 * doc/tm.texi: Regenerate.
6341 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
6342 instead of HARD_REGNO_NREGS.
6343 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
6344 HARD_REGNO_NREGS in the comment.
6345 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
6346 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
6347 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
6348 Return an unsigned int.
6349 (TARGET_HARD_REGNO_NREGS): Redefine.
6350 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
6351 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
6352 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
6353 (arc_hard_regno_nregs): New function.
6354 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
6355 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
6356 (arm_hard_regno_nregs): New function.
6357 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
6358 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
6359 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
6360 (TARGET_HARD_REGNO_NREGS): Redefine.
6361 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
6362 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
6363 (HARD_REGNO_NREGS): Delete.
6364 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
6365 (cr16_hard_regno_nregs): New function.
6366 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
6367 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
6368 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
6369 (cris_hard_regno_nregs): New function.
6370 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
6371 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
6372 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
6373 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
6374 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
6375 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
6376 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
6377 (frv_hard_regno_nregs): Make static. Take and return an
6379 (frv_class_max_nregs): Remove outdated copy of documentation.
6380 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
6381 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
6382 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
6383 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
6384 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
6385 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
6386 (TARGET_HARD_REGNO_NREGS): Redefine.
6387 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
6388 (CLASS_MAX_NREGS): Update comment.
6389 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
6390 (ia64_hard_regno_nregs): New function.
6391 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
6392 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
6393 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
6394 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
6395 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
6397 (m32c_hard_regno_nregs): Likewise. Make static.
6398 (TARGET_HARD_REGNO_NREGS): Redefine.
6399 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
6400 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
6401 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
6402 (m68k_hard_regno_nregs): New function.
6403 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
6404 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
6405 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
6406 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
6407 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
6408 Take and return an unsigned int.
6409 (TARGET_HARD_REGNO_NREGS): Redefine.
6410 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
6411 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
6412 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
6413 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
6414 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
6415 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
6416 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
6417 (msp430_hard_regno_nregs): Make static. Take and return an
6419 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
6420 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
6421 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
6422 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
6423 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
6424 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
6425 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
6426 (TARGET_HARD_REGNO_NREGS): Redefine.
6427 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
6428 (PA_HARD_REGNO_NREGS): ...this.
6429 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
6430 (PA_HARD_REGNO_NREGS): ...this.
6431 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
6432 (pa_hard_regno_nregs): New function.
6433 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
6434 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
6435 (pdp11_hard_regno_nregs): New function.
6436 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
6437 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
6438 (rs6000_hard_regno_nregs_hook): New function.
6439 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
6440 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
6441 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
6442 Take and return an unsigned int. Move earlier in file.
6443 (TARGET_HARD_REGNO_NREGS): Redefine.
6444 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
6445 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
6446 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
6447 (rl78_hard_regno_nregs): Make static. Take and return an
6449 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
6450 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
6451 (rs6000_hard_regno_nregs_hook): New function.
6452 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
6453 * config/rx/rx.c (rx_hard_regno_nregs): New function.
6454 (TARGET_HARD_REGNO_NREGS): Redefine.
6455 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
6456 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
6457 instead of HARD_REGNO_NREGS.
6458 (s390_hard_regno_nregs): New function.
6459 (s390_hard_regno_mode_ok): Add comment from s390.h.
6460 (TARGET_HARD_REGNO_NREGS): Redefine.
6461 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
6462 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
6463 (sh_hard_regno_nregs): New function.
6464 (sh_pass_in_reg_p): Use it.
6465 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
6466 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
6467 (sparc_hard_regno_nregs): New function.
6468 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
6469 * config/spu/spu.c (spu_hard_regno_nregs): New function.
6470 (spu_function_arg_advance): Use it, supplying a valid register number.
6471 (TARGET_HARD_REGNO_NREGS): Redefine.
6472 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
6473 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
6474 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
6475 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
6476 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
6477 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
6478 (CLASS_MAX_NREGS): Remove copy of old documentation.
6479 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
6480 (visium_hard_regno_nregs): New function.
6481 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
6482 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
6483 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
6484 xtensa_hard_regno_nregs): New function.
6485 * system.h (HARD_REGNO_NREGS): Poison.
6487 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6489 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
6490 hard_regno_nregs instead of HARD_REGNO_NREGS.
6491 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
6492 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
6493 (c6x_expand_epilogue): Likewise.
6494 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
6495 (frv_read_iacc_argument): Likewise.
6496 * config/sh/sh.c: Include regs.h.
6497 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
6498 (regs_used): Likewise.
6499 (output_stack_adjust): Likewise.
6500 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
6501 * expmed.c: Include regs.h.
6502 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
6503 * ree.c: Include regs.h.
6504 (combine_reaching_defs): Use hard_regno_nregs instead of
6506 (add_removable_extension): Likewise.
6508 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6510 * regs.h (hard_regno_nregs): Turn into a function.
6511 (end_hard_regno): Update accordingly.
6512 * caller-save.c (setup_save_areas): Likewise.
6513 (save_call_clobbered_regs): Likewise.
6514 (replace_reg_with_saved_mem): Likewise.
6515 (insert_restore): Likewise.
6516 (insert_save): Likewise.
6517 * combine.c (can_change_dest_mode): Likewise.
6518 (move_deaths): Likewise.
6519 (distribute_notes): Likewise.
6520 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
6521 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
6522 (rs6000_split_multireg_move): Likewise.
6523 (rs6000_register_move_cost): Likewise.
6524 (rs6000_memory_move_cost): Likewise.
6525 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
6526 (rs6000_split_multireg_move): Likewise.
6527 (rs6000_register_move_cost): Likewise.
6528 (rs6000_memory_move_cost): Likewise.
6529 * cselib.c (cselib_reset_table): Likewise.
6530 (cselib_lookup_1): Likewise.
6531 * emit-rtl.c (set_mode_and_regno): Likewise.
6532 * function.c (aggregate_value_p): Likewise.
6533 * ira-color.c (setup_profitable_hard_regs): Likewise.
6534 (check_hard_reg_p): Likewise.
6535 (calculate_saved_nregs): Likewise.
6536 (assign_hard_reg): Likewise.
6537 (improve_allocation): Likewise.
6538 (calculate_spill_cost): Likewise.
6539 * ira-emit.c (modify_move_list): Likewise.
6540 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
6541 (ira_hard_reg_in_set_p): Likewise.
6542 * ira.c (setup_reg_mode_hard_regset): Likewise.
6543 (clarify_prohibited_class_mode_regs): Likewise.
6544 (check_allocation): Likewise.
6545 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6546 (lra_setup_reg_renumber): Likewise.
6547 (setup_try_hard_regno_pseudos): Likewise.
6548 (spill_for): Likewise.
6549 (assign_hard_regno): Likewise.
6550 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
6551 * lra-constraints.c (in_class_p): Likewise.
6552 (lra_constraint_offset): Likewise.
6553 (simplify_operand_subreg): Likewise.
6554 (lra_constraints): Likewise.
6555 (split_reg): Likewise.
6556 (split_if_necessary): Likewise.
6557 (invariant_p): Likewise.
6558 (inherit_in_ebb): Likewise.
6559 * lra-lives.c (process_bb_lives): Likewise.
6560 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
6561 (get_hard_regs): Likewise.
6562 (do_remat): Likewise.
6563 * lra-spills.c (assign_spill_hard_regs): Likewise.
6564 * mode-switching.c (create_pre_exit): Likewise.
6565 * postreload.c (reload_combine_recognize_pattern): Likewise.
6566 * recog.c (peep2_find_free_register): Likewise.
6567 * regcprop.c (kill_value_regno): Likewise.
6568 (set_value_regno): Likewise.
6569 (copy_value): Likewise.
6570 (maybe_mode_change): Likewise.
6571 (find_oldest_value_reg): Likewise.
6572 (copyprop_hardreg_forward_1): Likewise.
6573 * regrename.c (check_new_reg_p): Likewise.
6574 (regrename_do_replace): Likewise.
6575 * reload.c (push_reload): Likewise.
6576 (combine_reloads): Likewise.
6577 (find_dummy_reload): Likewise.
6578 (operands_match_p): Likewise.
6579 (find_reloads): Likewise.
6580 (find_equiv_reg): Likewise.
6581 (reload_adjust_reg_for_mode): Likewise.
6582 * reload1.c (count_pseudo): Likewise.
6583 (count_spilled_pseudo): Likewise.
6584 (find_reg): Likewise.
6585 (clear_reload_reg_in_use): Likewise.
6586 (free_for_value_p): Likewise.
6587 (allocate_reload_reg): Likewise.
6588 (choose_reload_regs): Likewise.
6589 (reload_adjust_reg_for_temp): Likewise.
6590 (emit_reload_insns): Likewise.
6591 (delete_output_reload): Likewise.
6592 * rtlanal.c (subreg_get_info): Likewise.
6593 * sched-deps.c (sched_analyze_reg): Likewise.
6594 * sel-sched.c (init_regs_for_mode): Likewise.
6595 (mark_unavailable_hard_regs): Likewise.
6596 (choose_best_reg_1): Likewise.
6597 (verify_target_availability): Likewise.
6598 * valtrack.c (dead_debug_insert_temp): Likewise.
6599 * var-tracking.c (track_loc_p): Likewise.
6600 (emit_note_insn_var_location): Likewise.
6601 * varasm.c (make_decl_rtl): Likewise.
6602 * reginfo.c (choose_hard_reg_mode): Likewise.
6603 (init_reg_modes_target): Refer directly to
6604 this_target_regs->x_hard_regno_nregs.
6606 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6608 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
6609 instead of hard_regno_nregs.
6611 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6613 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
6614 end_hard_regno instead of hard_regno_nregs.
6615 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
6616 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
6617 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
6618 * ira-color.c (improve_allocation): Likewise.
6619 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6620 * lra-lives.c (mark_regno_live): Likewise.
6621 (mark_regno_dead): Likewise.
6622 * lra-remat.c (operand_to_remat): Likewise.
6623 * lra.c (collect_non_operand_hard_regs): Likewise.
6624 * postreload.c (reload_combine_note_store): Likewise.
6625 (move2add_valid_value_p): Likewise.
6626 * reload.c (regno_clobbered_p): Likewise.
6628 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6630 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
6632 * config/v850/v850.c (v850_reorg): Likewise.
6633 * reload.c (refers_to_regno_for_reload_p): Likewise.
6634 (find_equiv_reg): Likewise.
6635 * reload1.c (reload_reg_reaches_end_p): Likewise.
6637 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6639 * caller-save.c (add_used_regs): Use REG_NREGS instead of
6641 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
6642 * config/arm/arm.c (output_move_neon): Likewise.
6643 (arm_attr_length_move_neon): Likewise.
6644 (neon_split_vcombine): Likewise.
6645 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
6646 (c6x_mark_reg_written): Likewise.
6647 (c6x_dwarf_register_span): Likewise.
6648 * config/i386/i386.c (ix86_save_reg): Likewise.
6649 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
6650 (rws_access_reg): Likewise.
6651 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
6652 * mode-switching.c (create_pre_exit): Likewise.
6653 * ree.c (combine_reaching_defs): Likewise.
6654 (add_removable_extension): Likewise.
6655 * regcprop.c (find_oldest_value_reg): Likewise.
6656 (copyprop_hardreg_forward_1): Likewise.
6657 * reload.c (reload_inner_reg_of_subreg): Likewise.
6658 (push_reload): Likewise.
6659 (combine_reloads): Likewise.
6660 (find_dummy_reload): Likewise.
6661 (reload_adjust_reg_for_mode): Likewise.
6662 * reload1.c (find_reload_regs): Likewise.
6663 (forget_old_reloads_1): Likewise.
6664 (reload_reg_free_for_value_p): Likewise.
6665 (reload_adjust_reg_for_temp): Likewise.
6666 (emit_reload_insns): Likewise.
6667 (delete_output_reload): Likewise.
6668 * sel-sched.c (choose_best_reg_1): Likewise.
6669 (choose_best_pseudo_reg): Likewise.
6671 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6672 Alan Hayward <alan.hayward@arm.com>
6673 David Sherwood <david.sherwood@arm.com>
6675 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
6676 * target.def (slow_unaligned_access): New hook.
6677 * targhooks.h (default_slow_unaligned_access): Declare.
6678 * targhooks.c (default_slow_unaligned_access): New function.
6679 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
6680 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
6681 * doc/tm.texi: Regenerate.
6682 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
6683 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
6684 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
6686 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
6687 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
6689 (rs6000_slow_unaligned_access): New function.
6690 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
6691 (expand_block_compare): Likewise.
6692 (expand_strn_compare): Likewise.
6693 (rs6000_rtx_costs): Likewise.
6694 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
6695 (riscv_slow_unaligned_access): Likewise.
6696 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
6697 (riscv_slow_unaligned_access_p): ...this and make static.
6698 (riscv_option_override): Update accordingly.
6699 (riscv_slow_unaligned_access): New function.
6700 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
6701 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
6702 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
6703 (rs6000_slow_unaligned_access): New function.
6704 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
6705 (rs6000_rtx_costs): Likewise.
6706 * config/rs6000/rs6000-string.c (expand_block_compare)
6707 (expand_strn_compare): Use targetm.slow_unaligned_access instead
6708 of SLOW_UNALIGNED_ACCESS.
6709 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
6710 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
6711 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
6712 of SLOW_UNALIGNED_ACCESS.
6713 * expmed.c (simple_mem_bitfield_p): Likewise.
6714 * expr.c (alignment_for_piecewise_move): Likewise.
6715 (emit_group_load_1): Likewise.
6716 (emit_group_store): Likewise.
6717 (copy_blkmode_from_reg): Likewise.
6718 (emit_push_insn): Likewise.
6719 (expand_assignment): Likewise.
6720 (store_field): Likewise.
6721 (expand_expr_real_1): Likewise.
6722 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
6723 * lra-constraints.c (simplify_operand_subreg): Likewise.
6724 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
6725 * gimple-ssa-store-merging.c: Likewise in block comment at start
6727 * tree-ssa-strlen.c: Include target.h.
6728 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
6729 of SLOW_UNALIGNED_ACCESS.
6730 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
6732 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6734 PR rtl-optimization/82185
6735 * expmed.c (emit_store_flag_int): Only test tem if it has been
6738 2017-09-12 Richard Biener <rguenther@suse.de>
6741 * match.pd ((FTYPE) N CMP CST): Fix typo.
6743 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
6745 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
6747 (mips_near_type_p): Add 'short_call' attribute as a synonym
6749 * doc/extend.texi (short_call): Document new function attribute.
6751 2017-09-12 Jakub Jelinek <jakub@redhat.com>
6754 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
6755 assertion check that in the condition.
6756 (get_atomic_generic_size): Likewise. Before testing if parameter
6757 has pointer type, if it has array type, call for C++
6758 default_conversion to perform array-to-pointer conversion.
6760 2017-09-12 Richard Biener <rguenther@suse.de>
6762 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
6763 for operations we cannot scalarize.
6765 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
6767 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
6768 vectors heap vectors. Clean up comments.
6769 Make visited_bbs a reference.
6770 (profitable_jump_thread_path): Make GC
6771 vectors heap vectors. Clean up comments.
6773 (convert_and_register_jump_thread_path): Make GC vectors heap
6775 (check_subpath_and_update_thread_path): Same. Clean up comments.
6776 Make visited_bbs a reference.
6777 (handle_phi): Abstract common code to to
6778 register_jump_thread_path_if_profitable.
6779 Rename VAR_BB to DEF_BB.
6781 Make GC vectors heap vectors.
6782 Make visited_bbs a reference.
6783 (handle_assignment): Same.
6784 (register_jump_thread_path_if_profitable): New.
6785 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
6787 Make GC vectors heap vectors. Clean up comments.
6788 Make visited_bbs a reference.
6789 (find_jump_threads_backwards): Make visited_bbs live in the stack.
6790 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
6793 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
6796 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
6797 words of E_DImode object are reachable by xtensa_uimm8x4 access.
6799 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
6801 Revert r251800 and r251799.
6803 2017-09-11 Martin Jambor <mjambor@suse.cz>
6806 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
6807 arguments in advance.
6808 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
6809 use it to find predecessor edges.
6810 (naive_outof_ssa): Collect vector of predecessors.
6812 2017-09-08 Jason Merrill <jason@redhat.com>
6814 PR c++/70029 - ICE with ref-qualifier and -flto
6815 * langhooks.h (struct lang_hooks_for_types): Add
6816 copy_lang_qualifiers.
6817 * attribs.c (build_type_attribute_qual_variant): Use it.
6818 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
6820 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
6821 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
6823 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
6826 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
6827 (*mulsi3_sp64): New instruction.
6828 (mulsi3): New expander.
6830 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
6832 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
6834 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6836 * sancov.c: Include memmodel.h.
6838 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
6841 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
6844 2017-09-07 Carl Love <cel@us.ibm.com>
6846 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
6847 the sldi instruction.
6849 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
6851 * sancov.c: Include tm_p.h.
6853 2017-09-07 Jakub Jelinek <jakub@redhat.com>
6856 * output.h (switch_to_other_text_partition): New declaration.
6857 * varasm.c (switch_to_other_text_partition): New function.
6858 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
6859 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
6860 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
6861 to the other text partition before emitting LCL label and switch back
6862 after emitting the word after it.
6864 2017-09-07 Richard Biener <rguenther@suse.de>
6866 * passes.def (pass_split_crit_edges): Remove instance before PRE.
6867 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
6868 critical edges here, after loop init.
6869 (pass_data_pre): Remove PROP_no_crit_edges flags.
6870 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
6871 for valueization of call args to avoid leaking VN_TOP.
6872 (visit_use): Assert we do not visit default defs.
6873 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
6874 Use error_mark_node to more easily detect leaking VN_TOP.
6875 All default-defs are varying, not VN_TOP. Mark them visited.
6876 (run_scc_vn): Make code match comment.
6878 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
6880 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
6881 OPTION_MASK_FLOAT128_KEYWORD.
6882 (POWERPC_MASKS): Likewise.
6883 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
6884 support for the -mfloat128-type option, and make -mfloat128
6885 default on PowerPC Linux systems. Define or undefine
6886 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
6887 Define __float128 to be __ieee128 if IEEE 128-bit support is
6888 enabled, or undefine it.
6889 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
6890 Delete defining __FLOAT128_TYPE__.
6891 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
6892 -mfloat128-type option and make -mfloat128 default on PowerPC
6894 (TARGET_FLOAT128_TYPE): Likewise.
6895 (-mfloat128-type): Likewise.
6896 * config/rs6000/rs6000.c (rs6000_option_override_internal):
6897 Delete the -mfloat128-type option and make -mfloat128 default on
6898 PowerPC Linux systems. Always use __ieee128 to be the keyword for
6899 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
6900 128-bit floating point is enabled. Change tests from using
6901 -mfloat128-type to -mfloat128.
6902 (rs6000_mangle_type): Use the correct mangling for the __float128
6903 type even if normal long double is restricted to 64-bits.
6904 (floatn_mode): Enable the _Float128 type by default on VSX Linux
6906 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
6907 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
6908 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
6910 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
6911 documentation for -mfloat128.
6913 2017-09-06 Olivier Hainque <hainque@adacore.com>
6915 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
6917 2017-09-06 Wish Wu <wishwu007@gmail.com>
6918 Jakub Jelinek <jakub@redhat.com>
6920 * asan.c (initialize_sanitizer_builtins): Add
6921 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
6922 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
6923 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
6924 BT_FN_VOID_UINT64_PTR variables.
6925 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
6926 (BT_FN_VOID_UINT16_UINT16): Likewise.
6927 (BT_FN_VOID_UINT32_UINT32): Likewise.
6928 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
6929 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
6930 (BT_FN_VOID_UINT64_PTR): Likewise.
6931 * common.opt (flag_sanitize_coverage): New variable.
6932 (fsanitize-coverage=trace-pc): Remove.
6933 (fsanitize-coverage=): Add.
6934 * flag-types.h (enum sanitize_coverage_code): New enum.
6935 * fold-const.c (fold_range_test): Disable non-short-circuit
6936 optimization if flag_sanitize_coverage.
6937 (fold_truth_andor): Likewise.
6938 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
6939 * opts.c (COVERAGE_SANITIZER_OPT): Define.
6940 (coverage_sanitizer_opts): New array.
6941 (get_closest_sanitizer_option): Add OPTS argument, handle also
6942 OPT_fsanitize_coverage_.
6943 (parse_sanitizer_options): Adjusted to also handle
6944 OPT_fsanitize_coverage_.
6945 (common_handle_option): Add OPT_fsanitize_coverage_.
6946 * sancov.c (instrument_comparison, instrument_switch): New function.
6947 (sancov_pass): Add trace-cmp support.
6948 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
6949 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
6950 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
6951 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
6952 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
6953 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
6954 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
6955 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
6956 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
6957 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
6959 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
6961 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
6962 error. Only quit immediately if parsing is complete.
6963 (BEGIN): Initialize fatal_err and parse_done.
6964 (begin fpu, end fpu): Check number of arguments.
6965 (begin arch, end arch): Likewise.
6966 (begin cpu, end cpu): Likewise.
6967 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
6968 (optalias): Likewise.
6970 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
6972 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
6973 * config/arm/arm-isa.h: Delete. Move definitions to ...
6974 * arm-cpus.in: ... here. Use new feature and fgroup values.
6975 * config/arm/arm.c (arm_option_override): Use lower case for feature
6977 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
6978 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
6979 * config/arm/parsecpu.awk (END): Add new command 'isa'.
6981 (print_isa_bits_for): New function.
6982 (gen_isa): New function.
6983 (gen_comm_data): Use print_isa_bits_for.
6984 (define feature): New keyword.
6985 (define fgroup): New keyword.
6986 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
6987 (arm-isa.h): Add rule to generate file.
6988 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
6989 case for feature bit names.
6991 2017-09-06 Richard Biener <rguenther@suse.de>
6993 * tree-ssa-pre.c (NECESSARY): Remove.
6994 (create_expression_by_pieces): Do not touch pass-local flags.
6995 (insert_into_preds_of_block): Likewise.
6996 (do_pre_regular_insertion): Likewise.
6997 (eliminate_insert): Likewise.
6998 (eliminate_dom_walker::before_dom_children): Likewise.
6999 (fini_eliminate): Do not look at inserted_exprs.
7000 (mark_operand_necessary): Remove.
7001 (remove_dead_inserted_code): Replace with simple work-list
7002 algorithm based on inserted_exprs and SSA uses.
7003 (pass_pre::execute): Re-order fini_eliminate and
7004 remove_dead_inserted_code.
7006 2017-09-06 Olivier Hainque <hainque@adacore.com>
7008 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
7009 for VxWorks 7. Adjust surrounding comments.
7011 2017-09-06 Richard Biener <rguenther@suse.de>
7013 * gimple-ssa-strength-reduction.c
7014 (find_candidates_dom_walker::before_dom_children): Also allow
7017 2017-09-06 Richard Biener <rguenther@suse.de>
7019 PR tree-optimization/82108
7020 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
7021 for gap in the non-permutation SLP case.
7023 2017-09-06 Martin Jambor <mjambor@suse.cz>
7025 PR tree-optimization/82078
7026 * tree-sra.c (sort_and_splice_var_accesses): Move call to
7027 add_access_to_work_queue...
7028 (build_accesses_from_assign): ...here.
7029 (propagate_all_subaccesses): Make sure racc is the group
7030 representative, if there is one.
7032 2017-09-06 Jakub Jelinek <jakub@redhat.com>
7035 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
7038 2017-09-06 Richard Biener <rguenther@suse.de>
7040 * gimple-ssa-strength-reduction.c
7041 (find_candidates_dom_walker::before_doom_children): Use a
7042 type and not a mode check.
7044 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
7047 * config/arm/predicates.md (arm_general_adddi_operand): Create new
7049 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
7051 2017-09-05 Jeff Law <law@redhat.com>
7053 PR tree-optimization/64910
7054 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
7055 cases where we have 3 or more operands.
7057 2017-09-05 Jakub Jelinek <jakub@redhat.com>
7060 * omp-low.c (lower_omp_for): Recompute tree invariant if
7061 gimple_omp_for_initial/final is ADDR_EXPR.
7064 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
7065 into gimple val before gimplification fo the COND_EXPR.
7067 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
7069 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
7070 REGION_COPY argument.
7071 (thread_through_all_blocks): Remove unused argument to
7072 duplicate_thread_path.
7074 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7075 Alan Hayward <alan.hayward@arm.com>
7076 David Sherwood <david.sherwood@arm.com>
7078 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
7079 Take a scalar_mode rather than a machine_mode.
7080 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
7081 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
7082 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
7083 (aarch64_gen_adjusted_ldpstp): Likewise.
7084 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
7086 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7087 Alan Hayward <alan.hayward@arm.com>
7088 David Sherwood <david.sherwood@arm.com>
7090 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
7091 Take a scalar_int_mode instead of a machine_mode.
7092 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
7093 (aarch64_output_scalar_simd_mov_immediate): Likewise.
7094 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
7095 (aarch64_simd_attr_length_rglist): Delete.
7096 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
7097 a scalar_int_mode instead of a machine_mode.
7098 (aarch64_add_offset): Likewise.
7099 (aarch64_internal_mov_immediate): Likewise
7100 (aarch64_add_constant_internal): Likewise.
7101 (aarch64_add_constant): Likewise.
7102 (aarch64_movw_imm): Likewise.
7103 (aarch64_rtx_arith_op_extract_p): Likewise.
7104 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
7105 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
7106 Remove assert that the mode isn't a vector.
7107 (aarch64_output_scalar_simd_mov_immediate): Likewise.
7108 (aarch64_expand_mov_immediate): Update calls after above changes.
7109 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
7110 (aarch64_and_bitmask_imm): Check for scalar integer modes.
7111 (aarch64_move_imm): Likewise.
7112 (aarch64_can_const_movi_rtx_p): Likewise.
7113 (aarch64_strip_extend): Likewise.
7114 (aarch64_extr_rtx_p): Likewise.
7115 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
7116 a CONST_INT when the mode parameter is VOIDmode.
7117 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
7119 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7121 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
7122 * stor-layout.c (bitwise_mode_for_mode): Likewise.
7123 (bitwise_type_for_mode): Update accordingly.
7125 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7127 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
7128 * stor-layout.c (mode_for_size_tree): Likewise.
7129 (mode_for_array): Update accordingly.
7130 (layout_decl): Likewise.
7131 (compute_record_mode): Likewise. Only set the mode once.
7133 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7135 * target.def (get_mask_mode): Change return type to opt_mode.
7137 * doc/tm.texi: Regenerate.
7138 * targhooks.h (default_get_mask_mode): Return an opt_mode.
7139 * targhooks.c (default_get_mask_mode): Likewise.
7140 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
7141 * optabs-query.c (can_vec_mask_load_store_p): Update use of
7142 targetm.get_mask_mode.
7143 * tree.c (build_truth_vector_type): Likewise.
7145 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7147 * machmode.h (mode_for_vector): Return an opt_mode.
7148 * stor-layout.c (mode_for_vector): Likewise.
7149 (mode_for_int_vector): Update accordingly.
7150 (layout_type): Likewise.
7151 * config/i386/i386.c (emit_memmov): Likewise.
7152 (ix86_expand_set_or_movmem): Likewise.
7153 (ix86_expand_vector_init): Likewise.
7154 (ix86_get_mask_mode): Likewise.
7155 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
7157 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
7158 * expmed.c (extract_bit_field_1): Likewise.
7159 * expr.c (expand_expr_real_2): Likewise.
7160 * optabs-query.c (can_vec_perm_p): Likewise.
7161 (can_vec_mask_load_store_p): Likewise.
7162 * optabs.c (expand_vec_perm): Likewise.
7163 * targhooks.c (default_get_mask_mode): Likewise.
7164 * tree-vect-stmts.c (vectorizable_store): Likewise.
7165 (vectorizable_load): Likewise.
7166 (get_vectype_for_scalar_type_and_size): Likewise.
7168 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7170 * machmode.h (mode_for_int_vector): New function.
7171 * stor-layout.c (mode_for_int_vector): Likewise.
7172 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
7173 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
7174 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
7175 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
7176 (s390_expand_vcond): Likewise.
7178 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7180 * machmode.h (opt_machine_mode): New type.
7181 (opt_mode<T>): Allow construction from anything that can be
7183 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
7184 (mode_for_size): Return an opt_machine_mode.
7185 * stor-layout.c (mode_for_size): Likewise.
7186 (mode_for_size_tree): Update call accordingly.
7187 (bitwise_mode_for_mode): Likewise.
7188 (make_fract_type): Likewise.
7189 (make_accum_type): Likewise.
7190 * caller-save.c (replace_reg_with_saved_mem): Update call
7192 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
7193 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
7194 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
7195 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
7196 * expmed.c (extract_bit_field_1): Likewise.
7197 * reload.c (get_secondary_mem): Likewise.
7198 * varasm.c (assemble_integer): Likewise.
7199 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
7202 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7204 * machmode.h (decimal_float_mode_for_size): New function.
7205 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
7206 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
7207 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
7208 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
7209 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
7210 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
7212 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7214 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
7215 (get_builtin_sync_mode): Likewise.
7216 (expand_ifn_atomic_compare_exchange): Likewise.
7217 (expand_builtin_atomic_clear): Likewise.
7218 (expand_builtin_atomic_test_and_set): Likewise.
7219 (fold_builtin_atomic_always_lock_free): Likewise.
7220 * calls.c (compute_argument_addresses): Likewise.
7221 (emit_library_call_value_1): Likewise.
7222 (store_one_arg): Likewise.
7223 * combine.c (combine_instructions): Likewise.
7224 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
7225 * config/arm/arm.c (arm_function_value): Likewise.
7226 (aapcs_allocate_return_reg): Likewise.
7227 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
7228 * config/i386/i386.c (construct_container): Likewise.
7229 (ix86_gimplify_va_arg): Likewise.
7230 (ix86_expand_sse_cmp): Likewise.
7231 (emit_memmov): Likewise.
7232 (emit_memset): Likewise.
7233 (expand_small_movmem_or_setmem): Likewise.
7234 (ix86_expand_pextr): Likewise.
7235 (ix86_expand_pinsr): Likewise.
7236 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
7237 * config/microblaze/microblaze.c (microblaze_block_move_straight):
7239 * config/mips/mips.c (mips_function_value_1) Likewise.
7240 (mips_block_move_straight): Likewise.
7241 (mips_expand_ins_as_unaligned_store): Likewise.
7242 * config/powerpcspe/powerpcspe.c
7243 (rs6000_darwin64_record_arg_advance_flush): Likewise.
7244 (rs6000_darwin64_record_arg_flush): Likewise.
7245 * config/rs6000/rs6000.c
7246 (rs6000_darwin64_record_arg_advance_flush): Likewise.
7247 (rs6000_darwin64_record_arg_flush): Likewise.
7248 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
7249 (sparc_function_value_1): Likewise.
7250 * config/spu/spu.c (adjust_operand): Likewise.
7251 (spu_emit_branch_or_set): Likewise.
7252 (arith_immediate_p): Likewise.
7253 * emit-rtl.c (gen_lowpart_common): Likewise.
7254 * expr.c (expand_expr_real_1): Likewise.
7255 * function.c (assign_parm_setup_block): Likewise.
7256 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
7257 * reload1.c (alter_reg): Likewise.
7258 * stor-layout.c (mode_for_vector): Likewise.
7259 (layout_type): Likewise.
7261 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7263 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
7264 (spu_convert_move): Likewise.
7265 * lower-subreg.c (resolve_simple_move): Likewise.
7267 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7270 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
7271 define_insn to a define_expand.
7272 (altivec_vsum2sws_direct): New define_insn.
7273 (altivec_vsumsws): Convert from a define_insn to a define_expand.
7275 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
7277 * config/arm/arm.c (arm_option_params_internal): Improve setting of
7280 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
7285 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
7286 realign stack if stack alignment needed is less than incoming
7289 2017-09-05 Marek Polacek <polacek@redhat.com>
7292 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
7295 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
7297 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
7299 2017-09-05 Richard Biener <rguenther@suse.de>
7301 PR tree-optimization/82084
7302 * fold-const.c (can_native_encode_string_p): Handle wide characters.
7304 2017-09-05 Richard Biener <rguenther@suse.de>
7306 PR tree-optimization/82102
7307 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
7309 2017-09-05 Martin Liska <mliska@suse.cz>
7311 PR tree-optimization/82032
7312 * tree-cfg.c (generate_range_test): New function.
7313 * tree-cfg.h (generate_range_test): Declared here.
7314 * tree-cfgcleanup.c (convert_single_case_switch): New function.
7315 (cleanup_control_expr_graph): Use it.
7316 * tree-switch-conversion.c (try_switch_expansion): Remove
7318 (emit_case_nodes): Use generate_range_test.
7320 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
7323 * config/i386/i386.md (*<btsc><mode>_mask): Add
7324 TARGET_USE_BT to insn constraint.
7325 (*btr<mode>_mask): Ditto.
7327 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
7329 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
7330 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
7332 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
7335 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
7336 TARGET_NEON and TARGET_IWMMXT.
7337 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
7338 TARGET_NEON and TARGET_IWMMXT.
7339 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
7341 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
7343 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
7344 (ix86_rewrite_tls_address): Ditto.
7345 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
7346 (ix86_rewrite_tls_address_1): Ditto.
7347 (ix86_rewrite_tls_address): Ditto.
7348 * config/i386/predicates.md (tls_address_pattern): New predicate.
7349 * config/i386/i386.md (TLS address splitter): New splitter.
7351 2017-09-04 Richard Biener <rguenther@suse.de>
7353 PR tree-optimization/82084
7354 * fold-const.h (can_native_encode_string_p): Declare.
7355 * fold-const.c (can_native_encode_string_p): Factor out from ...
7356 (native_encode_string): ... here.
7357 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
7358 vectorizing stores from constants we later cannot handle.
7360 2017-09-04 Marek Polacek <polacek@redhat.com>
7363 * doc/invoke.texi: Update -Wtautological-compare documentation.
7365 2017-09-04 Jeff Law <law@redhat.com>
7367 PR tree-optimization/64910
7368 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
7369 swap the first and last operand if the last is a constant.
7371 2017-09-04 Marek Polacek <polacek@redhat.com>
7374 * convert.c (do_narrow): When sanitizing signed integer overflows,
7375 bail out for signed types.
7376 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
7378 2017-09-04 Richard Biener <rguenther@suse.de>
7380 PR tree-optimization/82060
7381 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
7382 Move devirtualization after stmt folding and before EH/AB/noreturn
7383 cleanup to get the stmt refs canonicalized. Use a bool instead
7384 of gimple_modified_p since that doesn't work for NOPs. Schedule
7385 NOPs generated by folding for removal.
7387 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7388 Alan Hayward <alan.hayward@arm.com>
7389 David Sherwood <david.sherwood@arm.com>
7391 * coretypes.h (pad_direction): New enum.
7392 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
7393 (FUNCTION_ARG_PADDING): Likewise.
7394 * target.def (function_arg_padding): New hook.
7395 * targhooks.h (default_function_arg_padding): Declare.
7396 * targhooks.c (default_function_arg_padding): New function.
7397 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
7398 (TARGET_FUNCTION_ARG_PADDING): ...this.
7399 * doc/tm.texi: Regenerate.
7400 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
7401 instead of direction.
7402 (compute_argument_addresses): Likewise.
7403 (load_register_parameters): Likewise.
7404 (emit_library_call_value_1): Likewise.
7405 (store_one_arg): Use targetm.calls.function_arg_padding instead
7406 of FUNCTION_ARG_PADDING.
7407 (must_pass_in_stack_var_size_or_pad): Likewise.
7408 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
7409 (emit_group_store): Likewise.
7410 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
7411 instead of FUNCTION_ARG_PADDING.
7412 (emit_push_insn): Likewise, and propagate enum change throughout
7414 * function.h (direction): Delete.
7415 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
7417 * function.c (assign_parm_find_stack_rtl): Likewise.
7418 (assign_parm_setup_block_p): Likewise.
7419 (assign_parm_setup_block): Likewise.
7420 (gimplify_parameters): Likewise.
7421 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
7422 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
7424 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
7425 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
7426 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
7427 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
7428 (aarch64_function_arg_padding): ...this new function.
7429 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
7430 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7431 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
7432 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
7433 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
7434 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7435 (arm_pad_arg_upward): Replace with...
7436 (arm_function_arg_padding): ...this new function.
7437 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
7439 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
7440 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
7441 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7442 (ia64_hpux_function_arg_padding): Replace with...
7443 (ia64_function_arg_padding): ...this new function. Use pad_direction
7444 instead of direction. Check for TARGET_HPUX.
7445 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
7446 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7447 (iq2000_function_arg_padding): New function.
7448 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
7449 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
7450 (mips_function_arg_padding): ...this new function.
7451 (mips_pad_reg_upward): Update accordingly.
7452 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7453 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
7454 targetm.calls.function_arg_padding.
7455 (FUNCTION_ARG_PADDING): Delete.
7456 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
7457 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
7458 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
7459 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
7460 (nios2_block_reg_padding): Return pad_direction instead of direction.
7461 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
7462 instead of direction.
7463 (nios2_function_arg_padding): Likewise. Make static.
7464 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7465 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
7466 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
7467 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
7468 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7469 (pa_function_arg_padding): Make static. Return pad_direction instead
7471 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
7472 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
7473 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
7474 instead of direction. Use targetm.calls.function_arg_padding.
7475 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
7476 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
7477 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
7478 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
7479 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
7481 (function_arg_padding): Rename to...
7482 (rs6000_function_arg_padding): ...this. Make static. Return
7483 pad_direction instead of direction.
7484 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
7485 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
7486 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
7487 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
7488 instead of direction. Use targetm.calls.function_arg_padding.
7489 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
7490 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
7491 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
7492 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
7493 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7494 (function_arg_padding): Rename to...
7495 (rs6000_function_arg_padding): ...this. Make static. Return
7496 pad_direction instead of direction.
7497 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
7498 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
7499 * config/s390/s390.c (s390_function_arg_padding): New function.
7500 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7501 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
7502 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
7503 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7504 (function_arg_padding): Rename to...
7505 (sparc_function_arg_padding): ...this. Make static. Return
7506 pad_direction instead of direction.
7507 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
7508 * config/spu/spu.c (spu_function_arg_padding): New function.
7509 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7510 * system.h (FUNCTION_ARG_PADDING): Poison.
7512 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7513 Alan Hayward <alan.hayward@arm.com>
7514 David Sherwood <david.sherwood@arm.com>
7516 * target.def (modes_tieable_p): New hook.
7517 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
7518 (TARGET_MODES_TIEABLE_P): ...this.
7519 * doc/tm.texi.in: Regenerate.
7520 * hooks.h (hook_bool_mode_mode_true): Declare.
7521 * hooks.c (hook_bool_mode_mode_true): New function.
7522 * combine.c (subst): Use targetm.modes_tieable_p instead of
7524 * dse.c (find_shift_sequence): Likewise.
7525 * expmed.c (extract_low_bits): Likewise.
7526 * lower-subreg.c: Include target.h.
7527 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
7529 * rtlanal.c (rtx_cost): Likewise.
7530 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
7531 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
7532 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
7533 (TARGET_MODES_TIEABLE_P): Redefine.
7534 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
7535 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
7536 (TARGET_MODES_TIEABLE_P): Redefine.
7537 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
7538 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
7539 (arc_modes_tieable_p): New function.
7540 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
7541 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
7542 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
7543 (arm_modes_tieable_p): Make static.
7544 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
7545 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
7546 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
7547 (TARGET_MODES_TIEABLE_P): Redefine.
7548 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
7549 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
7550 (TARGET_MODES_TIEABLE_P): Redefine.
7551 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
7552 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
7553 (cr16_modes_tieable_p): New function.
7554 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
7555 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
7556 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
7557 (TRULY_NOOP_TRUNCATION): Update comment.
7558 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
7559 (TRULY_NOOP_TRUNCATION): Update comment.
7560 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
7561 (frv_modes_tieable_p): New function.
7562 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
7563 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
7564 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
7565 (TARGET_MODES_TIEABLE_P): Redefine.
7566 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
7567 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
7568 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
7569 (TARGET_MODES_TIEABLE_P): Redefine.
7570 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
7571 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
7572 (ia64_modes_tieable_p): New function.
7573 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
7574 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
7575 (iq2000_modes_tieable_p): New function.
7576 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
7577 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
7578 (lm32_modes_tieable_p): New function.
7579 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
7580 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
7581 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
7582 (TARGET_MODES_TIEABLE_P): Redefine.
7583 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
7584 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
7585 (m32r_modes_tieable_p): New function.
7586 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
7587 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
7588 (m68k_modes_tieable_p): New function.
7589 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
7590 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
7591 (mcore_modes_tieable_p): New function.
7592 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
7593 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
7595 (TARGET_MODES_TIEABLE_P): Redefine.
7596 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
7597 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
7598 * config/mips/mips.c (mips_modes_tieable_p): Make static.
7599 (TARGET_MODES_TIEABLE_P): Redefine.
7600 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
7601 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
7602 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
7603 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
7604 (mn10300_modes_tieable_p): ...this and make static.
7605 (TARGET_MODES_TIEABLE_P): Redefine.
7606 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
7607 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
7608 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
7609 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
7610 (msp430_modes_tieable_p): Make static.
7611 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
7612 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
7613 (TARGET_MODES_TIEABLE_P): Redefine.
7614 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
7615 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
7616 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
7617 (TARGET_MODES_TIEABLE_P): Redefine.
7618 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
7619 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
7620 * config/pa/pa.c (pa_modes_tieable_p): Make static.
7621 (TARGET_MODES_TIEABLE_P): Redefine.
7622 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
7623 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
7624 (pdp11_modes_tieable_p): New function.
7625 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
7626 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
7627 (rs6000_modes_tieable_p): New function.
7628 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
7629 * config/powerpcspe/powerpcspe.md: Update comment.
7630 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
7631 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
7632 (TARGET_MODES_TIEABLE_P): Redefine.
7633 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
7634 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
7635 (rl78_modes_tieable_p): New function.
7636 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
7637 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
7638 (rs6000_modes_tieable_p): New function.
7639 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
7640 * config/rs6000/rs6000.md: Update comment.
7641 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
7642 * config/rx/rx.c (rx_modes_tieable_p): New function.
7643 (TARGET_MODES_TIEABLE_P): Redefine.
7644 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
7645 * config/s390/s390.c (s390_modes_tieable_p): New function.
7646 (TARGET_MODES_TIEABLE_P): Redefine.
7647 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
7648 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
7649 (sh_modes_tieable_p): New function.
7650 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
7651 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
7652 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
7653 (sparc_modes_tieable_p): Make static.
7654 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
7655 * config/spu/spu.c (spu_modes_tieable_p): New function.
7656 (TARGET_MODES_TIEABLE_P): Redefine.
7657 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
7658 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
7659 (TARGET_MODES_TIEABLE_P): Redefine.
7660 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
7661 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
7662 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
7663 * config/v850/v850.c (v850_modes_tieable_p): New function.
7664 (TARGET_MODES_TIEABLE_P): Redefine.
7665 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
7666 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
7667 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
7668 (visium_modes_tieable_p): New function.
7669 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
7670 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
7671 (xtensa_modes_tieable_p): New function.
7672 * system.h (MODES_TIEABLE_P): Poison.
7674 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7675 Alan Hayward <alan.hayward@arm.com>
7676 David Sherwood <david.sherwood@arm.com>
7678 * target.def (hard_regno_mode_ok): New hook.
7679 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
7680 (TARGET_HARD_REGNO_MODE_OK): ...this.
7681 * doc/tm.texi.in: Regenerate.
7682 * hooks.h (hook_bool_uint_mode_true): Declare.
7683 * hooks.c (hook_bool_uint_mode_true): New function.
7684 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
7686 * genpreds.c (write_insn_preds_c): Add an include of target.h.
7687 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
7688 instead of HARD_REGNO_MODE_OK.
7689 * caller-save.c: Include target.h.
7690 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
7692 * combine.c (can_combine_p): Likewise.
7693 (combinable_i3pat): Likewise.
7694 (can_change_dest_mode): Likewise.
7695 * expr.c (init_expr_target): Likewise.
7696 (convert_move): Likewise.
7697 (convert_modes): Likewise.
7698 * ira.c (setup_prohibited_class_mode_regs): Likewise.
7699 (setup_prohibited_mode_move_regs): Likewise.
7700 * ira.h (target_ira): Likewise.
7701 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7702 * lra-constraints.c (process_alt_operands): Likewise.
7703 (split_reg): Likewise.
7704 * recog.c (peep2_find_free_register): Likewise.
7705 * ree.c (combine_reaching_defs): Likewise.
7706 * regcprop.c (maybe_mode_change): Likewise.
7707 * reginfo.c (init_reg_sets_1): Likewise.
7708 (choose_hard_reg_mode): Likewise.
7709 (simplifiable_subregs): Likewise.
7710 * regrename.c (check_new_reg_p): Likewise.
7711 * reload.c (find_valid_class): Likewise.
7712 (find_valid_class_1): Likewise.
7713 (reload_inner_reg_of_subreg): Likewise.
7714 (push_reload): Likewise.
7715 (combine_reloads): Likewise.
7716 (find_dummy_reload): Likewise.
7717 (find_reloads): Likewise.
7718 * reload1.c (find_reg): Likewise.
7719 (set_reload_reg): Likewise.
7720 (allocate_reload_reg): Likewise.
7721 (choose_reload_regs): Likewise.
7722 (reload_adjust_reg_for_temp): Likewise.
7723 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
7724 (simplify_subreg_regno): Likewise.
7725 * sel-sched.c (init_regs_for_mode): Likewise.
7726 * varasm.c (make_decl_rtl): Likewise.
7727 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
7728 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
7730 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
7731 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
7732 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7733 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
7734 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
7735 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7736 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
7737 (arc_mode_class): Delete.
7738 (HARD_REGNO_MODE_OK): Delete.
7739 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7740 (arc_hard_regno_mode_ok): Rename old array to...
7741 (arc_hard_regno_mode_ok_modes): ...this.
7742 (arc_conditional_register_usage): Update accordingly.
7743 (arc_mode_class): Make static.
7744 (arc_hard_regno_mode_ok): New function.
7745 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
7746 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
7747 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7748 (arm_hard_regno_mode_ok): Make static.
7749 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
7751 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
7752 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
7753 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
7755 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7756 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
7757 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
7758 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
7759 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
7760 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7761 * config/bfin/predicates.md (valid_reg_operand): Use
7762 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
7763 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
7764 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
7765 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7766 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
7767 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
7768 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7769 (cr16_hard_regno_mode_ok): Make static and return a bool.
7770 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
7771 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7772 (cris_hard_regno_mode_ok): New function.
7773 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
7774 (epiphany_mode_class): Delete.
7775 (HARD_REGNO_MODE_OK): Delete.
7776 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
7777 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7778 (hard_regno_mode_ok): Rename to...
7779 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
7780 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
7781 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
7783 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
7784 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
7785 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7786 (frv_hard_regno_mode_ok): Make static and return a bool.
7787 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
7789 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
7790 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
7791 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
7792 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
7794 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7795 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
7796 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
7797 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
7799 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7800 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
7801 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7802 (ia64_hard_regno_mode_ok): New function.
7803 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
7804 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7805 (iq2000_hard_regno_mode_ok): New function.
7806 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
7807 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7808 (lm32_hard_regno_mode_ok): New function.
7809 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
7810 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
7811 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
7812 instead of HARD_REGNO_MODE_OK.
7813 (m32c_hard_regno_ok): Rename to...
7814 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
7815 (m32c_cannot_change_mode_class): Update accordingly.
7816 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7817 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
7818 (m32r_mode_class): Delete.
7819 (HARD_REGNO_MODE_OK): Delete.
7820 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7821 (m32r_hard_regno_mode_ok): Rename to...
7822 (m32r_hard_regno_modes): ...this.
7823 (m32r_mode_class): Make static.
7824 (m32r_hard_regno_mode_ok): New function.
7825 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
7826 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
7827 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7828 (m68k_hard_regno_mode_ok): Make static.
7829 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
7830 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7831 (mcore_hard_regno_mode_ok): New function.
7832 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
7833 (HARD_REGNO_MODE_OK): Delete.
7834 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
7836 (microblaze_hard_regno_mode_ok_p): ...this and make static.
7837 (microblaze_hard_regno_mode_ok): New function.
7838 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7839 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
7840 (mips_hard_regno_mode_ok): Delete.
7841 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
7842 (mips_hard_regno_mode_ok_p): ...this and make static.
7843 (mips_hard_regno_mode_ok_p): Rename to...
7844 (mips_hard_regno_mode_ok_uncached): ...this.
7845 (mips_hard_regno_mode_ok): New function.
7846 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
7847 of HARD_REGNO_MODE_OK.
7848 (mips_option_override): Update after above name changes.
7849 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7850 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
7851 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
7852 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
7853 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
7854 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7855 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
7856 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
7857 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
7858 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7859 (msp430_hard_regno_mode_ok): Make static and return a bool.
7860 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
7861 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
7862 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
7864 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7865 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
7866 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
7867 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
7868 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
7869 (PA_HARD_REGNO_MODE_OK): ...this
7870 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
7871 (PA_HARD_REGNO_MODE_OK): ...this.
7872 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7873 (pa_hard_regno_mode_ok): New function.
7874 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
7875 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7876 (pdp11_hard_regno_mode_ok): New function.
7877 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
7878 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
7880 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
7882 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7883 (rs6000_hard_regno_mode_ok): Rename to...
7884 (rs6000_hard_regno_mode_ok_uncached): ...this.
7885 (rs6000_init_hard_regno_mode_ok): Update accordingly.
7886 (rs6000_hard_regno_mode_ok): New function.
7887 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
7888 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
7889 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
7890 (riscv_hard_regno_mode_ok): ...this and make static.
7891 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7892 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
7893 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
7894 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7895 (rl78_hard_regno_mode_ok): Make static and return bool.
7896 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
7897 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
7899 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
7900 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7901 (rs6000_hard_regno_mode_ok): Rename to...
7902 (rs6000_hard_regno_mode_ok_uncached): ...this.
7903 (rs6000_init_hard_regno_mode_ok): Update accordingly.
7904 (rs6000_hard_regno_mode_ok): New function.
7905 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
7906 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
7907 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7908 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
7909 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
7910 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
7911 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7912 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
7913 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
7914 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7915 (sh_hard_regno_mode_ok): Make static.
7916 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
7917 instead of HARD_REGNO_MODE_OK.
7918 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
7919 (sparc_mode_class): Delete.
7920 (HARD_REGNO_MODE_OK): Delete.
7921 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7922 (hard_regno_mode_classes): Make static.
7923 (sparc_mode_class): Likewise.
7924 (sparc_hard_regno_mode_ok): New function.
7925 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
7926 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
7927 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
7929 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7930 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
7931 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
7932 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
7933 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
7934 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7935 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
7936 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
7937 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7938 (visium_hard_regno_mode_ok): New function.
7939 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
7940 instead of HARD_REGNO_MODE_OK.
7941 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
7942 (HARD_REGNO_MODE_OK): Delete.
7943 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
7944 (xtensa_hard_regno_mode_ok_p): ...this and make static.
7945 (xtensa_option_override): Update accordingly.
7946 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7947 (xtensa_hard_regno_mode_ok): New function.
7948 * system.h (HARD_REGNO_MODE_OK): Poison.
7950 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7951 Alan Hayward <alan.hayward@arm.com>
7952 David Sherwood <david.sherwood@arm.com>
7954 * target.def (hard_regno_call_part_clobbered): New hook.
7955 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
7956 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
7957 * doc/tm.texi: Regenerate.
7958 * hooks.h (hook_bool_uint_mode_false): Declare.
7959 * hooks.c (hook_bool_uint_mode_false): New function.
7960 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7961 * cselib.c (cselib_process_insn): Use
7962 targetm.hard_regno_call_part_clobbered instead of
7963 HARD_REGNO_CALL_PART_CLOBBERED.
7964 * ira-conflicts.c (ira_build_conflicts): Likewise.
7965 * ira-costs.c (ira_tune_allocno_costs): Likewise.
7966 * lra-constraints.c (need_for_call_save_p): Likewise.
7967 * lra-lives.c: Include target.h.
7968 (check_pseudos_live_through_calls): Use
7969 targetm.hard_regno_call_part_clobbered instead of
7970 HARD_REGNO_CALL_PART_CLOBBERED.
7971 * regcprop.c: Include target.h.
7972 (copyprop_hardreg_forward_1): Use
7973 targetm.hard_regno_call_part_clobbered instead of
7974 HARD_REGNO_CALL_PART_CLOBBERED.
7975 * reginfo.c (choose_hard_reg_mode): Likewise.
7976 * regrename.c (check_new_reg_p): Likewise.
7977 * reload.c (find_equiv_reg): Likewise.
7978 * reload1.c (emit_reload_insns): Likewise.
7979 * sched-deps.c (deps_analyze_insn): Likewise.
7980 * sel-sched.c (init_regs_for_mode): Likewise.
7981 (mark_unavailable_hard_regs): Likewise.
7982 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
7983 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7984 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
7986 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7987 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7988 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
7990 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
7992 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7993 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7994 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
7996 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7997 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7998 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
8000 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8001 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
8003 * config/powerpcspe/powerpcspe.c
8004 (rs6000_hard_regno_call_part_clobbered): New function.
8005 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8006 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8007 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
8009 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8010 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8011 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
8013 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8014 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8015 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
8017 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8018 Alan Hayward <alan.hayward@arm.com>
8019 David Sherwood <david.sherwood@arm.com>
8021 * rtl.h (subreg_memory_offset): Declare.
8022 * emit-rtl.c (subreg_memory_offset): New function.
8023 * expmed.c (store_bit_field_1): Use it.
8024 * expr.c (undefined_operand_subword_p): Likewise.
8025 * simplify-rtx.c (simplify_subreg): Likewise.
8027 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
8029 PR rtl-optimization/57448
8032 * optabs.c (expand_atomic_load): Place compiler memory barriers if
8033 using atomic_load pattern.
8034 (expand_atomic_store): Likewise.
8036 2017-09-04 Jakub Jelinek <jakub@redhat.com>
8039 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
8040 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
8041 handling. Use replace_call_with_value with NULL instead of
8042 gsi_replace, unlink_stmt_vdef and release_defs.
8044 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
8047 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
8049 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8052 * rtl.h (emit_library_call_value_1): Declare.
8053 (emit_library_call): Replace declaration with a series of overloads.
8054 Remove the parameter count argument.
8055 (emit_library_call_value): Likewise.
8056 * calls.c (emit_library_call_value_1): Make global. Replace varargs
8057 with an "rtx_mode_t *".
8058 (emit_library_call_value): Delete.
8059 (emit_library_call): Likewise.
8060 * asan.c (asan_emit_stack_protection): Update calls accordingly.
8061 (asan_emit_allocas_unpoison): Likewise.
8062 * builtins.c (expand_builtin_powi): Likewise.
8063 (expand_asan_emit_allocas_unpoison): Likewise.
8064 * cfgexpand.c (expand_main_function): Likewise.
8065 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
8066 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
8067 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
8068 * config/arm/arm.c (arm_trampoline_init): Likewise.
8069 (arm_call_tls_get_addr): Likewise.
8070 (arm_expand_divmod_libfunc): Likewise.
8071 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
8072 (smulsi3_highpart): Likewise.
8073 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
8074 (c6x_expand_compare): Likewise.
8075 (c6x_expand_movmem): Likewise.
8076 * config/frv/frv.c (frv_trampoline_init): Likewise.
8077 * config/i386/i386.c (ix86_trampoline_init): Likewise.
8078 (ix86_expand_divmod_libfunc): Likewise.
8079 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
8080 (ia64_expand_compare): Likewise.
8081 (ia64_profile_hook): Likewise.
8082 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
8083 (nonlocal_goto): Likewise.
8084 (restore_stack_nonlocal): Likewise.
8085 * config/m32r/m32r.c (block_move_call): Likewise.
8086 (m32r_trampoline_init): Likewise.
8087 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
8088 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
8089 (m68k_call_m68k_read_tp): Likewise.
8090 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
8091 (microblaze_expand_divide): Likewise.
8092 * config/mips/mips.h (mips_args): Likewise.
8093 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
8094 (MIPS_ICACHE_SYNC): Likewise.
8095 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
8096 (nios2_trampoline_init): Likewise.
8097 * config/pa/pa.c (hppa_tls_call): Likewise.
8098 (pa_trampoline_init): Likewise.
8099 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
8100 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
8101 (expand_strn_compare): Likewise.
8102 (rs6000_generate_compare): Likewise.
8103 (rs6000_expand_float128_convert): Likewise.
8104 (output_profile_hook): Likewise.
8105 (rs6000_trampoline_init): Likewise.
8106 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
8107 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
8108 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
8109 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
8110 (rs6000_generate_compare): Likewise.
8111 (rs6000_expand_float128_convert): Likewise.
8112 (output_profile_hook): Likewise.
8113 (rs6000_trampoline_init): Likewise.
8114 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
8115 * config/sh/sh.c (sh_trampoline_init): Likewise.
8116 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
8117 (sparc_emit_float_lib_cmp): Likewise.
8118 (sparc32_initialize_trampoline): Likewise.
8119 (sparc64_initialize_trampoline): Likewise.
8120 (sparc_profile_hook): Likewise.
8121 * config/spu/spu.c (ea_load_store): Likewise.
8122 * config/spu/spu.md (floatunssidf2): Likewise.
8123 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
8124 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
8125 * config/visium/visium.c (expand_block_move_4): Likewise.
8126 (expand_block_move_2): Likewise.
8127 (expand_block_move_1): Likewise.
8128 (expand_block_set_4): Likewise.
8129 (expand_block_set_2): Likewise.
8130 (expand_block_set_1): Likewise.
8131 (visium_trampoline_init): Likewise.
8132 (visium_profile_hook): Likewise.
8133 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
8134 (xtensa_setup_frame_addresses): Likewise.
8135 (xtensa_trampoline_init): Likewise.
8136 * except.c (sjlj_emit_function_enter): Likewise.
8137 (sjlj_emit_function_exit): Likewise.
8138 * explow.c (allocate_dynamic_stack_space): Likewise.
8139 (probe_stack_range): Likewise.
8140 * expr.c (convert_mode_scalar): Likewise.
8141 * optabs.c (expand_binop): Likewise.
8142 (expand_twoval_binop_libfunc): Likewise.
8143 (expand_unop): Likewise.
8144 (prepare_cmp_insn): Likewise.
8145 (prepare_float_lib_cmp): Likewise.
8146 (expand_float): Likewise.
8147 (expand_fix): Likewise.
8148 (expand_fixed_convert): Likewise.
8149 (maybe_emit_sync_lock_test_and_set): Likewise.
8150 (expand_atomic_compare_and_swap): Likewise.
8151 (expand_mem_thread_fence): Likewise.
8152 (expand_atomic_fetch_op): Likewise.
8154 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
8156 * doc/generic.texi (OpenACC): Adjust URL.
8157 * doc/invoke.texi (C Dialect Options): Ditto.
8159 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
8161 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
8162 predicate for operand 1. Add (m,<S>) constraint.
8163 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
8164 Prevent memory operand 1 with register operand 2.
8166 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
8168 PR rtl-optimization/82024
8169 * combine.c (try_combine): If the combination result is a PARALLEL,
8170 and we only need to retain the SET in there that would be placed
8171 at I2, check that we can place that at I3 instead, before doing so.
8173 2017-09-01 Jakub Jelinek <jakub@redhat.com>
8176 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
8178 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
8179 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
8182 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
8183 Jeff Law <law@redhat.com>
8185 * varasm.c (bss_initializer_p): Do not put constants into .bss
8186 (categorize_decl_for_section): Handle bss_initializer_p returning
8187 false when DECL_INITIAL is NULL.
8189 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8192 * config/s390/s390.c (s390_can_inline_p): New function.
8194 2017-09-01 Jeff Law <law@redhat.com>
8196 PR tree-optimization/82052
8197 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
8198 Always initialize the returned slot after a hash table miss
8199 when INSERT is true.
8201 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
8203 * config/s390/s390.md (mem_signal_fence): Remove.
8204 * doc/md.texi (mem_signal_fence): Remove.
8205 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
8207 * target-insns.def (mem_signal_fence): Remove.
8209 2017-09-01 Jakub Jelinek <jakub@redhat.com>
8212 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
8215 * asan.c (create_odr_indicator): Strip name encoding from assembler
8216 name before appending it after __odr_asan_.
8218 2017-09-01 Martin Liska <mliska@suse.cz>
8220 PR tree-optimization/82059
8221 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
8222 frequency only when an edge is redirected.
8224 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
8226 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
8227 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
8228 (arc_conditional_register_usage): Remove ARC600 lp_count
8230 (arc_file_start): Emit Tag_ARC_CPU_variation.
8231 (arc_can_use_doloop_p): New conditions to use ZOLs.
8232 (hwloop_fail): New function.
8233 (hwloop_optimize): Likewise.
8234 (hwloop_pattern_reg): Likewise.
8235 (arc_doloop_hooks): New struct, to be used with reorg_loops.
8236 (arc_reorg_loops): New function, calls reorg_loops.
8237 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
8238 (arc600_corereg_hazard): Remove ZOL checking, case handled by
8240 (arc_loop_hazard): Remove function, functionality moved into
8242 (arc_hazard): Remove arc_loop_hazard call.
8243 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
8244 into hwloop_optimize.
8245 (arc_label_align): Remove ZOL handling.
8246 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
8247 * config/arc/arc.md (doloop_begin): Remove pattern.
8248 (doloop_begin_i): Likewise.
8249 (doloop_end_i): Likewise.
8250 (doloop_fallback): Likewise.
8251 (doloop_fallback_m): Likewise.
8252 (doloop_end): Reimplement expand.
8253 (arc_lp): New pattern for LP instruction.
8254 (loop_end): New pattern.
8255 (loop_fail): Likewise.
8256 (decrement_and_branch_until_zero): Likewise.
8257 * config/arc/arc.opt (mlpc-width): New option.
8258 * doc/invoke.texi (mlpc-width): Document option.
8260 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
8262 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
8263 (arc_ccfsm_advance): Fix checking for delay slots.
8264 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
8266 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
8268 * config/arc/arc.md (movqi_insn): Add stores to save constant long
8270 (movhi_insn): Update store instruction constraint which are saving
8271 6-bit short immediates.
8272 (movsi_insn): Consider also short scaled load operations.
8273 (zero_extendhisi2_i): Use Usd constraint instead of T.
8274 (extendhisi2_i): Add q constraint.
8275 (arc_clzsi2): Add type and length attributes.
8276 (arc_ctzsi2): Likewise.
8277 * config/arc/constraints.md (Usc): Update constraint, the
8278 assembler can parse two relocations for a single instruction.
8280 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
8282 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
8283 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
8285 2017-08-31 Olivier Hainque <hainque@adacore.com>
8287 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
8288 match as powerpc-wrs-vxworks*.
8290 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
8292 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
8293 register constraint for by-element operand.
8294 (aarch64_mls_elt_merge<mode>): Likewise.
8296 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
8298 * config/arc/arc.c (arc_can_follow_jump): Check for short
8301 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
8303 * config.gcc: Use g.opt for arc.
8304 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
8305 functionality moved to ...
8306 (legitimate_scaled_address_p): New function, ...here.
8307 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
8308 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
8309 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
8311 (arc_override_options): Handle G option.
8312 (arc_output_pic_addr_const): Correct function definition.
8313 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
8314 (arc_decl_anon_ns_mem_p): Delete.
8315 (arc_in_small_data_p): Overhaul this function to take into
8316 consideration the value given via G option.
8317 (arc_rewrite_small_data_1): Renamed and corrected old
8318 arc_rewrite_small_data function.
8319 (arc_rewrite_small_data): New function.
8320 (small_data_pattern): Don't use pic_offset_table_rtx.
8321 * config/arc/arc.h (CC1_SPEC): Recognize G option.
8322 * config/arc/simdext.md (movmisalignv2hi): Use
8323 prepare_move_operands function.
8325 (movmisalign*): Likewise.
8326 * doc/invoke.texi (ARC options): Document -G option.
8328 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
8330 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
8332 * config/arc/arc.c (arc_print_operand): Output scalled address for
8333 sdata whenever is possible.
8334 (arc_in_small_data_p): Allow sdata for 64bit datum when double
8335 load/stores are available.
8336 (compact_sda_memory_operand): Check for the alignment required by
8337 code density instructions.
8338 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
8340 * config/arc/constraints.md (Usd): Update constraint.
8341 (Us0): New constraint.
8342 (Usc): Update constraint.
8344 2017-08-31 Richard Biener <rguenther@suse.de>
8347 * dwarf2out.c (dwarf2out_early_global_decl): Process each
8350 2017-08-31 Tamar Christina <tamar.christina@arm.com>
8352 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
8353 Resize type_signature.
8355 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
8356 Alan Hayward <alan.hayward@arm.com>
8357 David Sherwood <david.sherwood@arm.com>
8359 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
8360 subregs whose inner modes can be stored in GPRs.
8361 (aarch64_classify_index): Likewise.
8363 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
8364 Alan Hayward <alan.hayward@arm.com>
8365 David Sherwood <david.sherwood@arm.com>
8367 * config/aarch64/iterators.md (V_cmp_result): Rename to...
8368 (V_INT_EQUIV): ...this.
8369 (v_cmp_result): Rename to...
8370 (v_int_equiv): ...this.
8371 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
8372 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
8373 (copysign<mode>3): Likewise.
8374 (aarch64_simd_bsl<mode>_internal): Likewise.
8375 (aarch64_simd_bsl<mode>): Likewise.
8376 (vec_cmp<mode><mode>): Likewise.
8377 (vcond<mode><mode>): Likewise.
8378 (vcond<v_cmp_mixed><mode>): Likewise.
8379 (vcondu<mode><v_cmp_mixed>): Likewise.
8380 (aarch64_cm<optab><mode>): Likewise.
8381 (aarch64_cmtst<mode>): Likewise.
8382 (aarch64_fac<optab><mode>): Likewise.
8383 (vec_perm_const<mode>): Likewise.
8384 (vcond_mask_<mode><v_cmp_result>): Rename to...
8385 (vcond_mask_<mode><v_int_equiv>): ...this.
8386 (vec_cmp<mode><v_cmp_result>): Rename to...
8387 (vec_cmp<mode><v_int_equiv>): ...this.
8389 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
8390 Alan Hayward <alan.hayward@arm.com>
8391 David Sherwood <david.sherwood@arm.com>
8393 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
8395 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
8396 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
8397 (UNSPEC_LD4_DREG): New unspecs.
8398 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
8399 (aarch64_ld2<mode>_dreg_be): Replace with...
8400 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
8402 (aarch64_ld3<mode>_dreg_le)
8403 (aarch64_ld3<mode>_dreg_be): Replace with...
8404 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
8406 (aarch64_ld4<mode>_dreg_le)
8407 (aarch64_ld4<mode>_dreg_be): Replace with...
8408 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
8411 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8413 PR tree-optimization/81987
8414 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
8415 insert an initializer in a location not dominated by the stride
8418 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
8420 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
8421 on the entire header of the finally block in the fallthru case.
8423 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
8425 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
8427 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
8429 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
8430 rs6000_emit_move_from_cr and call renamed function.
8431 (rs6000_emit_prologue): Call renamed functions.
8432 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
8433 movesi_from_cr, remove volatile CRs.
8435 2017-08-30 Jon Beniston <jon@beniston.com>
8436 Richard Biener <rguenther@suse.de>
8438 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
8439 of VECTOR_MODE_P check.
8440 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
8441 element vector types.
8443 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8445 * df.h (df_read_modify_subreg_p): Remove in favor of...
8446 * rtl.h (read_modify_subreg_p): ...this new function. Take a
8447 const_rtx instead of an rtx.
8448 * cprop.c (local_cprop_find_used_regs): Update accordingly.
8449 * df-problems.c (df_word_lr_mark_ref): Likewise.
8450 * ira-lives.c (mark_pseudo_reg_live): Likewise.
8451 (mark_pseudo_reg_dead): Likewise.
8452 (mark_ref_dead): Likewise.
8453 * reginfo.c (init_subregs_of_mode): Likewise.
8454 * sched-deps.c (sched_analyze_1): Likewise.
8455 * df-scan.c (df_def_record_1): Likewise.
8456 (df_uses_record): Likewise.
8457 (df_read_modify_subreg_p): Remove in favor of...
8458 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
8459 const_rtx instead of an rtx.
8461 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8462 Alan Hayward <alan.hayward@arm.com>
8463 David Sherwood <david.sherwood@arm.com>
8465 * rtl.h (partial_subreg_p): New function.
8466 * caller-save.c (save_call_clobbered_regs): Use it.
8467 * calls.c (expand_call): Likewise.
8468 * combine.c (combinable_i3pat): Likewise.
8469 (simplify_set): Likewise.
8470 (make_extraction): Likewise.
8471 (make_compound_operation_int): Likewise.
8472 (gen_lowpart_or_truncate): Likewise.
8473 (force_to_mode): Likewise.
8474 (make_field_assignment): Likewise.
8475 (reg_truncated_to_mode): Likewise.
8476 (record_truncated_value): Likewise.
8477 (move_deaths): Likewise.
8478 * cse.c (record_jump_cond): Likewise.
8479 (cse_insn): Likewise.
8480 * cselib.c (cselib_lookup_1): Likewise.
8481 * expmed.c (extract_bit_field_using_extv): Likewise.
8482 * function.c (assign_parm_setup_reg): Likewise.
8483 * ifcvt.c (noce_convert_multiple_sets): Likewise.
8484 * ira-build.c (create_insn_allocnos): Likewise.
8485 * lra-coalesce.c (merge_pseudos): Likewise.
8486 * lra-constraints.c (match_reload): Likewise.
8487 (simplify_operand_subreg): Likewise.
8488 (curr_insn_transform): Likewise.
8489 * lra-lives.c (process_bb_lives): Likewise.
8490 * lra.c (new_insn_reg): Likewise.
8491 (lra_substitute_pseudo): Likewise.
8492 * regcprop.c (mode_change_ok): Likewise.
8493 (maybe_mode_change): Likewise.
8494 (copyprop_hardreg_forward_1): Likewise.
8495 * reload.c (push_reload): Likewise.
8496 (find_reloads): Likewise.
8497 (find_reloads_subreg_address): Likewise.
8498 * reload1.c (alter_reg): Likewise.
8499 (eliminate_regs_1): Likewise.
8500 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8502 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
8504 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
8505 back to if statements, including unpack.
8507 2017-08-30 Martin Liska <mliska@suse.cz>
8510 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
8512 (func_checker::compare_asm_inputs_outputs): ... this function.
8513 (func_checker::compare_gimple_asm): Use the function to compare
8514 also ASM constrains.
8515 * ipa-icf-gimple.h: Rename the function.
8517 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8518 Alan Hayward <alan.hayward@arm.com>
8519 David Sherwood <david.sherwood@arm.com>
8521 * coretypes.h (complex_mode): New type.
8522 * gdbhooks.py (build_pretty_printer): Handle it.
8523 * machmode.h (complex_mode): New class.
8524 (complex_mode::includes_p): New function.
8525 (is_complex_int_mode): Likewise.
8526 (is_complex_float_mode): Likewise.
8527 * genmodes.c (get_mode_class): Handle complex mode classes.
8528 * function.c (expand_function_end): Use is_complex_int_mode.
8530 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8531 Alan Hayward <alan.hayward@arm.com>
8532 David Sherwood <david.sherwood@arm.com>
8534 * coretypes.h (scalar_mode_pod): New typedef.
8535 * gdbhooks.py (build_pretty_printer): Handle it.
8536 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
8537 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
8538 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
8539 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
8542 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8543 Alan Hayward <alan.hayward@arm.com>
8544 David Sherwood <david.sherwood@arm.com>
8546 * machmode.h (mode_for_vector): Take a scalar_mode instead
8548 * stor-layout.c (mode_for_vector): Likewise.
8549 * explow.c (promote_mode): Use as_a <scalar_mode>.
8550 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
8552 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8553 Alan Hayward <alan.hayward@arm.com>
8554 David Sherwood <david.sherwood@arm.com>
8556 * target.def (preferred_simd_mode): Take a scalar_mode
8557 instead of a machine_mode.
8558 * targhooks.h (default_preferred_simd_mode): Likewise.
8559 * targhooks.c (default_preferred_simd_mode): Likewise.
8560 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
8561 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
8562 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
8563 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
8564 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
8565 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
8566 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
8567 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
8569 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
8570 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
8571 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
8572 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
8573 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
8574 * doc/tm.texi: Regenerate.
8575 * optabs-query.c (can_vec_mask_load_store_p): Return false for
8578 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8579 Alan Hayward <alan.hayward@arm.com>
8580 David Sherwood <david.sherwood@arm.com>
8582 * target.def (scalar_mode_supported_p): Take a scalar_mode
8583 instead of a machine_mode.
8584 * targhooks.h (default_scalar_mode_supported_p): Likewise.
8585 * targhooks.c (default_scalar_mode_supported_p): Likewise.
8586 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
8587 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
8588 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
8589 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
8590 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
8591 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
8592 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
8593 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
8594 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
8595 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
8596 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
8597 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
8599 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
8600 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
8601 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
8602 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
8603 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
8605 * doc/tm.texi: Regenerate.
8607 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8608 Alan Hayward <alan.hayward@arm.com>
8609 David Sherwood <david.sherwood@arm.com>
8611 * coretypes.h (opt_scalar_mode): New typedef.
8612 * gdbhooks.py (build_pretty_printers): Handle it.
8613 * machmode.h (mode_iterator::get_2xwider): Add overload for
8615 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
8617 * expr.c (convert_mode_scalar): Likewise.
8618 * omp-low.c (omp_clause_aligned_alignment): Likewise.
8619 * optabs.c (expand_float): Likewise.
8620 (expand_fix): Likewise.
8621 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8623 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8624 Alan Hayward <alan.hayward@arm.com>
8625 David Sherwood <david.sherwood@arm.com>
8627 * optabs.c (expand_float): Explicitly check for scalars before
8628 using a branching expansion.
8629 (expand_fix): Likewise.
8631 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8632 Alan Hayward <alan.hayward@arm.com>
8633 David Sherwood <david.sherwood@arm.com>
8635 * expr.c (convert_mode): Split scalar handling out into...
8636 (convert_mode_scalar): ...this new function. Treat the modes
8639 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8640 Alan Hayward <alan.hayward@arm.com>
8641 David Sherwood <david.sherwood@arm.com>
8643 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
8645 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
8647 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8648 Alan Hayward <alan.hayward@arm.com>
8649 David Sherwood <david.sherwood@arm.com>
8651 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
8652 rather than a machine_mode.
8653 (fixed_from_string): Likewise.
8654 (fixed_convert): Likewise.
8655 (fixed_convert_from_int): Likewise.
8656 (fixed_convert_from_real): Likewise.
8657 (real_convert_from_fixed): Likewise.
8658 * fixed-value.c (fixed_from_double_int): Likewise.
8659 (fixed_from_string): Likewise.
8660 (fixed_convert): Likewise.
8661 (fixed_convert_from_int): Likewise.
8662 (fixed_convert_from_real): Likewise.
8663 (real_convert_from_fixed): Likewise.
8664 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
8666 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8667 Alan Hayward <alan.hayward@arm.com>
8668 David Sherwood <david.sherwood@arm.com>
8670 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
8671 of separate mode class checks. Do not allow vector modes here.
8672 (immed_wide_int_const): Use as_a <scalar_mode>.
8673 * explow.c (trunc_int_for_mode): Likewise.
8674 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
8675 (wi::shwi): Likewise.
8676 (wi::min_value): Likewise.
8677 (wi::max_value): Likewise.
8678 * dwarf2out.c (loc_descriptor): Likewise.
8679 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
8682 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8683 Alan Hayward <alan.hayward@arm.com>
8684 David Sherwood <david.sherwood@arm.com>
8686 * tree.h (SCALAR_TYPE_MODE): New macro.
8687 * expr.c (expand_expr_addr_expr_1): Use it.
8688 (expand_expr_real_2): Likewise.
8689 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
8690 (fold_convert_const_fixed_from_int): Likewise.
8691 (fold_convert_const_fixed_from_real): Likewise.
8692 (native_encode_fixed): Likewise
8693 (native_encode_complex): Likewise
8694 (native_encode_vector): Likewise.
8695 (native_interpret_fixed): Likewise.
8696 (native_interpret_real): Likewise.
8697 (native_interpret_complex): Likewise.
8698 (native_interpret_vector): Likewise.
8699 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
8700 (simd_clone_adjust_argument_types): Likewise.
8701 (simd_clone_init_simd_arrays): Likewise.
8702 (simd_clone_adjust): Likewise.
8703 * stor-layout.c (layout_type): Likewise.
8704 * tree.c (build_minus_one_cst): Likewise.
8705 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
8706 * tree-inline.c (estimate_move_cost): Likewise.
8707 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
8708 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
8709 (vectorizable_reduction): Likewise.
8710 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
8711 (vect_recog_mixed_size_cond_pattern): Likewise.
8712 (check_bool_pattern): Likewise.
8713 (adjust_bool_pattern): Likewise.
8714 (search_type_for_mask_1): Likewise.
8715 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
8716 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8717 (vectorizable_load): Likewise.
8718 (vectorizable_store): Likewise.
8719 * ubsan.c (ubsan_encode_value): Likewise.
8720 * varasm.c (output_constant): Likewise.
8722 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8723 Alan Hayward <alan.hayward@arm.com>
8724 David Sherwood <david.sherwood@arm.com>
8726 * coretypes.h (scalar_mode): New class.
8727 * machmode.h (scalar_mode): Likewise.
8728 (scalar_mode::includes_p): New function.
8729 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
8730 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
8731 * genmodes.c (get_mode_class): Handle remaining scalar modes.
8732 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
8733 * expmed.c (store_bit_field_1): Likewise.
8734 (extract_bit_field_1): Likewise.
8735 * expr.c (write_complex_part): Likewise.
8736 (read_complex_part): Likewise.
8737 (emit_move_complex_push): Likewise.
8738 (expand_expr_real_2): Likewise.
8739 * function.c (assign_parm_setup_reg): Likewise.
8740 (assign_parms_unsplit_complex): Likewise.
8741 * optabs.c (expand_binop): Likewise.
8742 * rtlanal.c (subreg_get_info): Likewise.
8743 * simplify-rtx.c (simplify_immed_subreg): Likewise.
8744 * varasm.c (output_constant_pool_2): Likewise.
8746 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8747 Alan Hayward <alan.hayward@arm.com>
8748 David Sherwood <david.sherwood@arm.com>
8750 * expmed.c (extract_high_half): Use scalar_int_mode and remove
8752 (expmed_mult_highpart_optab): Likewise.
8753 (expmed_mult_highpart): Likewise.
8755 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8756 Alan Hayward <alan.hayward@arm.com>
8757 David Sherwood <david.sherwood@arm.com>
8759 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
8760 instead of a machine_mode.
8761 (builtin_memset_read_str): Likewise.
8762 * builtins.c (c_readstr): Likewise.
8763 (builtin_memcpy_read_str): Likewise.
8764 (builtin_strncpy_read_str): Likewise.
8765 (builtin_memset_read_str): Likewise.
8766 (builtin_memset_gen_str): Likewise.
8767 (expand_builtin_signbit): Use scalar_int_mode for local variables.
8768 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
8769 instead of a machine_mode.
8770 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
8772 (make_extraction): Likewise.
8773 (try_widen_shift_mode): Take and return scalar_int_modes instead
8775 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
8776 a scalar_int_mode instead of a machine_mode.
8777 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
8778 (avr_addr_space_pointer_mode): Likewise.
8779 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
8780 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
8781 (msp430_unwind_word_mode): Likewise.
8782 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
8783 (spu_addr_space_pointer_mode): Likewise.
8784 (spu_addr_space_address_mode): Likewise.
8785 (spu_libgcc_cmp_return_mode): Likewise.
8786 (spu_libgcc_shift_count_mode): Likewise.
8787 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
8788 (rl78_addr_space_pointer_mode): Likewise.
8789 (fl78_unwind_word_mode): Likewise.
8790 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
8792 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
8793 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
8794 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
8795 (mips_valid_pointer_mode): Likewise.
8796 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
8797 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
8798 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
8800 (ft32_addr_space_address_mode): Likewise.
8801 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
8802 scalar_int_mode instead of a machine_mode.
8803 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
8805 (m32c_addr_space_address_mode): Likewise.
8806 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
8807 (rs6000_eh_return_filter_mode): Likewise.
8808 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
8809 (rs6000_eh_return_filter_mode): Likewise.
8810 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
8811 (s390_libgcc_shift_count_mode): Likewise.
8812 (s390_unwind_word_mode): Likewise.
8813 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
8815 * target.def (mode_rep_extended): Likewise.
8816 (valid_pointer_mode): Likewise.
8817 (addr_space.valid_pointer_mode): Likewise.
8818 (eh_return_filter_mode): Return a scalar_int_mode rather than
8820 (libgcc_cmp_return_mode): Likewise.
8821 (libgcc_shift_count_mode): Likewise.
8822 (unwind_word_mode): Likewise.
8823 (addr_space.pointer_mode): Likewise.
8824 (addr_space.address_mode): Likewise.
8825 * doc/tm.texi: Regenerate.
8826 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
8828 (do_jump): Use scalar_int_mode for local variables.
8829 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
8830 rather than a machine_mode.
8831 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
8832 (scompare_loc_descriptor_wide): Likewise.
8833 (scompare_loc_descriptor_narrow): Likewise.
8834 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
8836 * except.c (sjlj_emit_dispatch_table): Likewise.
8837 (expand_builtin_eh_copy_values): Likewise.
8838 * explow.c (convert_memory_address_addr_space_1): Likewise.
8839 Take a scalar_int_mode rather than a machine_mode.
8840 (convert_memory_address_addr_space): Take a scalar_int_mode rather
8841 than a machine_mode.
8842 (memory_address_addr_space): Use scalar_int_mode for local variables.
8843 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
8844 rather than a machine_mode.
8845 * expmed.c (mask_rtx): Likewise.
8846 (init_expmed_one_conv): Likewise.
8847 (expand_mult_highpart_adjust): Likewise.
8848 (extract_high_half): Likewise.
8849 (expmed_mult_highpart_optab): Likewise.
8850 (expmed_mult_highpart): Likewise.
8851 (expand_smod_pow2): Likewise.
8852 (expand_sdiv_pow2): Likewise.
8853 (emit_store_flag_int): Likewise.
8854 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
8856 (extract_low_bits): Likewise.
8857 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
8859 * expr.c (pieces_addr::adjust): Likewise.
8860 (can_store_by_pieces): Likewise.
8861 (store_by_pieces): Likewise.
8862 (clear_by_pieces_1): Likewise.
8863 (expand_expr_addr_expr_1): Likewise.
8864 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
8865 (expand_expr_real_1): Likewise.
8866 (try_casesi): Likewise.
8867 * final.c (shorten_branches): Likewise.
8868 * fold-const.c (fold_convert_const_int_from_fixed): Change the
8869 type of "mode" to machine_mode.
8870 * internal-fn.c (expand_arith_overflow_result_store): Take a
8871 scalar_int_mode rather than a machine_mode.
8872 (expand_mul_overflow): Use scalar_int_mode for local variables.
8873 * loop-doloop.c (doloop_modify): Likewise.
8874 (doloop_optimize): Likewise.
8875 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
8876 than a machine_mode.
8877 (expand_doubleword_shift_condmove): Likewise.
8878 (expand_doubleword_shift): Likewise.
8879 (expand_doubleword_clz): Likewise.
8880 (expand_doubleword_popcount): Likewise.
8881 (expand_doubleword_parity): Likewise.
8882 (expand_absneg_bit): Use scalar_int_mode for local variables.
8883 (prepare_float_lib_cmp): Likewise.
8884 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
8885 rather than a machine_mode.
8886 (convert_memory_address_addr_space): Likewise.
8887 (get_mode_bounds): Likewise.
8888 (get_address_mode): Return a scalar_int_mode rather than a
8890 * rtlanal.c (get_address_mode): Likewise.
8891 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
8892 than a machine_mode.
8893 * targhooks.c (default_mode_rep_extended): Likewise.
8894 (default_valid_pointer_mode): Likewise.
8895 (default_addr_space_valid_pointer_mode): Likewise.
8896 (default_eh_return_filter_mode): Return a scalar_int_mode rather
8897 than a machine_mode.
8898 (default_libgcc_cmp_return_mode): Likewise.
8899 (default_libgcc_shift_count_mode): Likewise.
8900 (default_unwind_word_mode): Likewise.
8901 (default_addr_space_pointer_mode): Likewise.
8902 (default_addr_space_address_mode): Likewise.
8903 * targhooks.h (default_eh_return_filter_mode): Likewise.
8904 (default_libgcc_cmp_return_mode): Likewise.
8905 (default_libgcc_shift_count_mode): Likewise.
8906 (default_unwind_word_mode): Likewise.
8907 (default_addr_space_pointer_mode): Likewise.
8908 (default_addr_space_address_mode): Likewise.
8909 (default_mode_rep_extended): Take a scalar_int_mode rather than
8911 (default_valid_pointer_mode): Likewise.
8912 (default_addr_space_valid_pointer_mode): Likewise.
8913 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
8915 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
8916 rather than a machine_mode.
8917 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
8918 for local variables.
8919 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8920 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
8921 than a machine_mode.
8923 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8924 Alan Hayward <alan.hayward@arm.com>
8925 David Sherwood <david.sherwood@arm.com>
8927 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
8928 the mode argument to scalar_int_mode.
8929 (do_jump_by_parts_zero_rtx): Likewise.
8930 (do_jump_by_parts_equality_rtx): Likewise.
8931 (do_jump_by_parts_greater): Take a mode argument.
8932 (do_jump_by_parts_equality): Likewise.
8933 (do_jump_1): Update calls accordingly.
8935 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8936 Alan Hayward <alan.hayward@arm.com>
8937 David Sherwood <david.sherwood@arm.com>
8939 * is-a.h (safe_dyn_cast): New function.
8940 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
8941 (jump_table_for_label): Likewise.
8942 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
8943 instead of an rtx_insn *.
8944 (shorten_branches): Use dyn_cast instead of LABEL_P and
8945 JUMP_TABLE_DATA_P. Use jump_table_for_label and
8946 rtx_jump_table_data::get_data_mode.
8947 (final_scan_insn): Likewise.
8949 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8950 Alan Hayward <alan.hayward@arm.com>
8951 David Sherwood <david.sherwood@arm.com>
8953 * combine.c (try_combine): Use is_a <scalar_int_mode> when
8954 trying to combine a full-register integer set with a subreg
8957 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8958 Alan Hayward <alan.hayward@arm.com>
8959 David Sherwood <david.sherwood@arm.com>
8961 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
8962 that is always either address_mode or pointer_mode.
8964 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8965 Alan Hayward <alan.hayward@arm.com>
8966 David Sherwood <david.sherwood@arm.com>
8968 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
8969 when the two are known to be equal.
8971 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8972 Alan Hayward <alan.hayward@arm.com>
8973 David Sherwood <david.sherwood@arm.com>
8975 * simplify-rtx.c (simplify_const_unary_operation): Use
8976 is_a <scalar_int_mode> instead of checking for a nonzero
8977 precision. Forcibly convert op_mode to a scalar_int_mode
8978 in that case. More clearly differentiate the operand and
8979 result modes and use the former when deciding what the value
8980 of a count-bits operation should be. Use is_int_mode instead
8981 of checking for a MODE_INT. Remove redundant check for whether
8982 this mode has a zero precision.
8984 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8985 Alan Hayward <alan.hayward@arm.com>
8986 David Sherwood <david.sherwood@arm.com>
8988 * optabs.c (widen_leading): Change the type of the mode argument
8989 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
8990 (widen_bswap): Likewise.
8991 (expand_parity): Likewise.
8992 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
8993 (expand_ffs): Likewise.
8994 (epand_unop): Check for scalar integer modes before calling the
8997 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8998 Alan Hayward <alan.hayward@arm.com>
8999 David Sherwood <david.sherwood@arm.com>
9001 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
9003 (expand_expr_real_1): Update call accordingly.
9005 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9006 Alan Hayward <alan.hayward@arm.com>
9007 David Sherwood <david.sherwood@arm.com>
9009 * expmed.c (store_bit_field_using_insv): Add op0_mode and
9010 value_mode arguments. Use scalar_int_mode internally.
9011 (store_bit_field_1): Rename the new integer mode from imode
9012 to op0_mode and use it instead of GET_MODE (op0). Update calls
9013 to store_split_bit_field, store_bit_field_using_insv and
9014 store_fixed_bit_field.
9015 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
9016 Use scalar_int_mode internally. Use a bit count rather than a mode
9017 when calculating the largest bit size for get_best_mode.
9018 Update calls to store_split_bit_field and store_fixed_bit_field_1.
9019 (store_fixed_bit_field_1): Add mode and value_mode arguments.
9020 Remove assertion that OP0 has a scalar integer mode.
9021 (store_split_bit_field): Add op0_mode and value_mode arguments.
9022 Update calls to extract_fixed_bit_field.
9023 (extract_bit_field_using_extv): Add an op0_mode argument.
9024 Use scalar_int_mode internally.
9025 (extract_bit_field_1): Rename the new integer mode from imode to
9026 op0_mode and use it instead of GET_MODE (op0). Update calls to
9027 extract_split_bit_field, extract_bit_field_using_extv and
9028 extract_fixed_bit_field.
9029 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
9030 to extract_split_bit_field and extract_fixed_bit_field_1.
9031 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
9032 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
9034 (extract_split_bit_field): Add an op0_mode argument. Update call
9035 to extract_fixed_bit_field.
9037 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9038 Alan Hayward <alan.hayward@arm.com>
9039 David Sherwood <david.sherwood@arm.com>
9041 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
9042 * explow.c (hard_function_value): Likewise.
9043 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
9044 convert_to_mode call outside the loop.
9045 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
9046 for the mode iterator. Require the mode specified by max_pieces
9048 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
9050 (copy_blkmode_to_reg): Likewise.
9051 (set_storage_via_setmem): Likewise.
9052 * optabs.c (prepare_cmp_insn): Likewise.
9053 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
9054 * stor-layout.c (finish_bitfield_representative): Likewise.
9056 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9057 Alan Hayward <alan.hayward@arm.com>
9058 David Sherwood <david.sherwood@arm.com>
9060 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
9061 * expr.c (convert_move): Use them.
9062 (convert_modes): Likewise.
9063 (store_expr_with_bounds): Likewise.
9065 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9066 Alan Hayward <alan.hayward@arm.com>
9067 David Sherwood <david.sherwood@arm.com>
9069 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
9070 parameter for the mode of "x". Remove the "known_x", "known_mode"
9071 and "known_ret" arguments. Change the type of the mode argument
9073 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
9074 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
9075 (reg_num_sign_bit_copies_for_combine): Likewise.
9076 * rtlanal.c (nonzero_bits1): Likewise.
9077 (num_sign_bit_copies1): Likewise.
9078 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
9079 (reg_num_sign_bit_copies_general): Likewise.
9080 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
9081 (reg_nonzero_bits_general): Likewise.
9083 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9084 Alan Hayward <alan.hayward@arm.com>
9085 David Sherwood <david.sherwood@arm.com>
9087 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
9088 than in subroutines. Return 1 for non-integer modes.
9089 (cached_num_sign_bit_copies): Change the type of the mode parameter
9091 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
9092 classes. Handle CONST_INT_P first and then check whether X also
9093 has a scalar integer mode. Check the same thing for inner registers
9094 of a SUBREG and for values that are being extended or truncated.
9096 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9097 Alan Hayward <alan.hayward@arm.com>
9098 David Sherwood <david.sherwood@arm.com>
9100 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
9101 in subroutines. Return the mode mask for non-integer modes.
9102 (cached_nonzero_bits): Change the type of the mode parameter
9104 (nonzero_bits1): Likewise. Remove early exit for other mode
9105 classes. Handle CONST_INT_P first and then check whether X
9106 also has a scalar integer mode.
9108 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9109 Alan Hayward <alan.hayward@arm.com>
9110 David Sherwood <david.sherwood@arm.com>
9112 * expr.c (widest_int_mode_for_size): Make the comment match the code.
9113 Return a scalar_int_mode and assert that the size is greater than
9115 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
9116 (op_by_pieces_d::op_by_pieces_d): Likewise.
9117 (op_by_pieces_d::run): Likewise.
9118 (can_store_by_pieces): Likewise.
9120 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9121 Alan Hayward <alan.hayward@arm.com>
9122 David Sherwood <david.sherwood@arm.com>
9124 * combine.c (extract_left_shift): Add a mode argument and update
9126 (make_compound_operation_int): Change the type of the mode parameter
9127 to scalar_int_mode and update the call to extract_left_shift.
9129 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9130 Alan Hayward <alan.hayward@arm.com>
9131 David Sherwood <david.sherwood@arm.com>
9133 * combine.c (simplify_and_const_int): Change the type of the mode
9134 parameter to scalar_int_mode.
9135 (simplify_and_const_int_1): Likewise. Update recursive call.
9137 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9138 Alan Hayward <alan.hayward@arm.com>
9139 David Sherwood <david.sherwood@arm.com>
9141 * combine.c (simplify_compare_const): Check that the mode is a
9142 scalar_int_mode (rather than VOIDmode) before testing its
9144 (simplify_comparison): Move COMPARISON_P handling out of the
9145 loop and restrict the latter part of the loop to scalar_int_modes.
9146 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
9147 and when considering SUBREG_REGs. Use is_int_mode instead of
9148 checking GET_MODE_CLASS against MODE_INT.
9150 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9151 Alan Hayward <alan.hayward@arm.com>
9152 David Sherwood <david.sherwood@arm.com>
9154 * combine.c (try_widen_shift_mode): Move check for equal modes to...
9155 (simplify_shift_const_1): ...here. Use scalar_int_mode for
9156 shift_unit_mode and for modes involved in scalar shifts.
9158 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9159 Alan Hayward <alan.hayward@arm.com>
9160 David Sherwood <david.sherwood@arm.com>
9162 * combine.c (force_int_to_mode): New function, split out from...
9163 (force_to_mode): ...here. Keep xmode up-to-date and use it
9164 instead of GET_MODE (x).
9166 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9167 Alan Hayward <alan.hayward@arm.com>
9168 David Sherwood <david.sherwood@arm.com>
9170 * optabs-query.h (extraction_insn::struct_mode): Change type to
9171 opt_scalar_int_mode and update comment.
9172 (extraction_insn::field_mode): Change type to scalar_int_mode.
9173 (extraction_insn::pos_mode): Likewise.
9174 * combine.c (make_extraction): Update accordingly.
9175 * optabs-query.c (get_traditional_extraction_insn): Likewise.
9176 (get_optab_extraction_insn): Likewise.
9177 * recog.c (simplify_while_replacing): Likewise.
9178 * expmed.c (narrow_bit_field_mem): Change the type of the mode
9179 parameter to opt_scalar_int_mode.
9181 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9182 Alan Hayward <alan.hayward@arm.com>
9183 David Sherwood <david.sherwood@arm.com>
9185 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
9186 to a scalar_int_mode instead of a machine_mode.
9187 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
9188 (get_best_mode): Return a boolean and use a pointer argument to store
9189 the selected mode. Replace the limit mode parameter with a bit limit.
9190 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
9191 for the values returned by bit_field_mode_iterator::next_mode.
9192 (store_bit_field): Update call to get_best_mode.
9193 (store_fixed_bit_field): Likewise.
9194 (extract_fixed_bit_field): Likewise.
9195 * expr.c (optimize_bitfield_assignment_op): Likewise.
9196 * fold-const.c (optimize_bit_field_compare): Likewise.
9197 (fold_truth_andor_1): Likewise.
9198 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
9199 Update for new type of m_mode.
9200 (get_best_mode): As above.
9202 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9203 Alan Hayward <alan.hayward@arm.com>
9204 David Sherwood <david.sherwood@arm.com>
9206 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
9207 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
9208 (store_bit_field): Check is_a <scalar_int_mode> before calling
9209 strict_volatile_bitfield_p.
9210 (extract_bit_field): Likewise.
9212 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9213 Alan Hayward <alan.hayward@arm.com>
9214 David Sherwood <david.sherwood@arm.com>
9216 * target.def (cstore_mode): Return a scalar_int_mode.
9217 * doc/tm.texi: Regenerate.
9218 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
9219 * targhooks.h (default_cstore_mode): Likewise.
9220 * targhooks.c (default_cstore_mode): Likewise, using a forced
9222 * expmed.c (emit_cstore): Expect the target of the cstore to be
9225 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9226 Alan Hayward <alan.hayward@arm.com>
9227 David Sherwood <david.sherwood@arm.com>
9229 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
9231 (niter_desc): Likewise mode.
9232 (iv_analyze): Add a mode parameter.
9234 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
9235 and change its type to scalar_int_mode.
9236 * loop-iv.c: Update commentary at head of file.
9237 (iv_constant): Pass the mode paraeter before the rtx it describes
9238 and change its type to scalar_int_mode. Remove VOIDmode handling.
9239 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
9240 (iv_extend): Likewise.
9241 (shorten_into_mode): Likewise.
9242 (iv_add): Use scalar_int_mode.
9243 (iv_mult): Likewise.
9244 (iv_shift): Likewise.
9245 (canonicalize_iv_subregs): Likewise.
9246 (get_biv_step_1): Pass the outer_mode parameter before the rtx
9247 it describes and change its mode to scalar_int_mode. Also change
9248 the type of the returned inner_mode to scalar_int_mode.
9249 (get_biv_step): Likewise, turning outer_mode from a pointer
9250 into a direct parameter. Update call to get_biv_step_1.
9251 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
9252 iv_constant and get_biv_step.
9253 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
9254 and change its type to scalar_int_mode. Don't initialise iv->mode
9255 to VOIDmode and remove later checks for its still being VOIDmode.
9256 Update calls to iv_analyze_op and iv_analyze_expr. Check
9257 is_a <scalar_int_mode> when changing the mode under consideration.
9258 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
9259 Update call to iv_analyze_expr.
9260 (iv_analyze_op): Add a mode parameter. Reject subregs whose
9261 inner register is not also a scalar_int_mode. Update call to
9263 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
9264 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
9265 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
9266 separate mode class checks. Update calls to iv_analyze. Remove
9267 fix-up of VOIDmodes after iv_analyze_biv.
9268 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
9269 don't have a scalar_int_mode. Update call to biv_p.
9271 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9272 Alan Hayward <alan.hayward@arm.com>
9273 David Sherwood <david.sherwood@arm.com>
9275 * cfgexpand.c (convert_debug_memory_address): Use
9276 as_a <scalar_int_mode>.
9277 * combine.c (expand_compound_operation): Likewise.
9278 (make_extraction): Likewise.
9279 (change_zero_ext): Likewise.
9280 (simplify_comparison): Likewise.
9281 * cse.c (cse_insn): Likewise.
9282 * dwarf2out.c (minmax_loc_descriptor): Likewise.
9283 (mem_loc_descriptor): Likewise.
9284 (loc_descriptor): Likewise.
9285 * expmed.c (init_expmed_one_mode): Likewise.
9286 (synth_mult): Likewise.
9287 (emit_store_flag_1): Likewise.
9288 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
9289 of a comparison with size.
9290 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
9291 (reduce_to_bit_field_precision): Likewise.
9292 * function.c (expand_function_end): Likewise.
9293 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
9294 * loop-doloop.c (doloop_modify): Likewise.
9295 * optabs.c (expand_binop): Likewise.
9296 (expand_unop): Likewise.
9297 (expand_copysign_absneg): Likewise.
9298 (prepare_cmp_insn): Likewise.
9299 (maybe_legitimize_operand): Likewise.
9300 * recog.c (const_scalar_int_operand): Likewise.
9301 * rtlanal.c (get_address_mode): Likewise.
9302 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9303 (simplify_cond_clz_ctz): Likewise.
9304 * tree-nested.c (get_nl_goto_field): Likewise.
9305 * tree.c (build_vector_type_for_mode): Likewise.
9306 * var-tracking.c (use_narrower_mode): Likewise.
9308 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9309 Alan Hayward <alan.hayward@arm.com>
9310 David Sherwood <david.sherwood@arm.com>
9312 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
9313 * builtins.c (expand_builtin_signbit): Use it.
9314 * cfgexpand.c (expand_debug_expr): Likewise.
9315 * dojump.c (do_jump): Likewise.
9316 (do_compare_and_jump): Likewise.
9317 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
9318 * expmed.c (make_tree): Likewise.
9319 * expr.c (expand_expr_real_2): Likewise.
9320 (expand_expr_real_1): Likewise.
9321 (try_casesi): Likewise.
9322 * fold-const-call.c (fold_const_call_ss): Likewise.
9323 * fold-const.c (unextend): Likewise.
9324 (extract_muldiv_1): Likewise.
9325 (fold_single_bit_test): Likewise.
9326 (native_encode_int): Likewise.
9327 (native_encode_string): Likewise.
9328 (native_interpret_int): Likewise.
9329 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
9330 * internal-fn.c (expand_addsub_overflow): Likewise.
9331 (expand_neg_overflow): Likewise.
9332 (expand_mul_overflow): Likewise.
9333 (expand_arith_overflow): Likewise.
9334 * match.pd: Likewise.
9335 * stor-layout.c (layout_type): Likewise.
9336 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
9337 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
9338 * tree-ssanames.c (get_range_info): Likewise.
9339 * tree-switch-conversion.c (array_value_type) Likewise.
9340 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
9341 (vect_recog_divmod_pattern): Likewise.
9342 (vect_recog_mixed_size_cond_pattern): Likewise.
9343 * tree-vrp.c (extract_range_basic): Likewise.
9344 (simplify_float_conversion_using_ranges): Likewise.
9345 * tree.c (int_fits_type_p): Likewise.
9346 * ubsan.c (instrument_bool_enum_load): Likewise.
9347 * varasm.c (mergeable_string_section): Likewise.
9348 (narrowing_initializer_constant_valid_p): Likewise.
9349 (output_constant): Likewise.
9351 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9352 Alan Hayward <alan.hayward@arm.com>
9353 David Sherwood <david.sherwood@arm.com>
9355 * machmode.h (NARROWEST_INT_MODE): New macro.
9356 * expr.c (alignment_for_piecewise_move): Use it instead of
9357 GET_CLASS_NARROWEST_MODE (MODE_INT).
9358 (push_block): Likewise.
9359 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
9361 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9363 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9364 Alan Hayward <alan.hayward@arm.com>
9365 David Sherwood <david.sherwood@arm.com>
9367 * postreload.c (move2add_valid_value_p): Change the type of the
9368 mode parameter to scalar_int_mode.
9369 (move2add_use_add2_insn): Add a mode parameter and use it instead
9371 (move2add_use_add3_insn): Likewise.
9372 (reload_cse_move2add): Update accordingly.
9374 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9375 Alan Hayward <alan.hayward@arm.com>
9376 David Sherwood <david.sherwood@arm.com>
9378 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
9380 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
9381 * optabs.c (expand_unop): Likewise.
9383 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9384 Alan Hayward <alan.hayward@arm.com>
9385 David Sherwood <david.sherwood@arm.com>
9387 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
9388 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
9389 (popcount_loc_descriptor): Likewise.
9390 (bswap_loc_descriptor): Likewise.
9391 (rotate_loc_descriptor): Likewise.
9392 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
9393 calling the functions above.
9395 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9396 Alan Hayward <alan.hayward@arm.com>
9397 David Sherwood <david.sherwood@arm.com>
9399 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
9401 (try_combine): Likewise.
9402 (simplify_if_then_else): Likewise.
9403 * cse.c (cse_insn): Likewise.
9404 * dwarf2out.c (mem_loc_descriptor): Likewise.
9405 * emit-rtl.c (gen_lowpart_common): Likewise.
9406 * simplify-rtx.c (simplify_truncation): Likewise.
9407 (simplify_binary_operation_1): Likewise.
9408 (simplify_const_relational_operation): Likewise.
9409 (simplify_ternary_operation): Likewise.
9410 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
9412 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9413 Alan Hayward <alan.hayward@arm.com>
9414 David Sherwood <david.sherwood@arm.com>
9416 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
9417 * reload.c (push_reload): Likewise.
9418 (find_reloads): Likewise.
9420 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9421 Alan Hayward <alan.hayward@arm.com>
9422 David Sherwood <david.sherwood@arm.com>
9424 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
9425 (make_compound_operation_int): Likewise.
9426 (change_zero_ext): Likewise.
9427 * expr.c (convert_move): Likewise.
9428 (convert_modes): Likewise.
9429 * fwprop.c (forward_propagate_subreg): Likewise.
9430 * loop-iv.c (get_biv_step_1): Likewise.
9431 * optabs.c (widen_operand): Likewise.
9432 * postreload.c (move2add_valid_value_p): Likewise.
9433 * recog.c (simplify_while_replacing): Likewise.
9434 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9435 (simplify_binary_operation_1): Likewise. Remove redundant
9436 mode equality check.
9438 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9439 Alan Hayward <alan.hayward@arm.com>
9440 David Sherwood <david.sherwood@arm.com>
9442 * combine.c (combine_simplify_rtx): Add checks for
9443 is_a <scalar_int_mode>.
9444 (simplify_if_then_else): Likewise.
9445 (make_field_assignment): Likewise.
9446 (simplify_comparison): Likewise.
9447 * ifcvt.c (noce_try_bitop): Likewise.
9448 * loop-invariant.c (canonicalize_address_mult): Likewise.
9449 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9451 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9452 Alan Hayward <alan.hayward@arm.com>
9453 David Sherwood <david.sherwood@arm.com>
9455 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
9456 is_a <scalar_int_mode> instead of != BLKmode.
9458 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9459 Alan Hayward <alan.hayward@arm.com>
9460 David Sherwood <david.sherwood@arm.com>
9462 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
9463 instead of != VOIDmode.
9464 * combine.c (if_then_else_cond): Likewise.
9465 (change_zero_ext): Likewise.
9466 * dwarf2out.c (mem_loc_descriptor): Likewise.
9467 (loc_descriptor): Likewise.
9468 * rtlanal.c (canonicalize_condition): Likewise.
9469 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
9471 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9472 Alan Hayward <alan.hayward@arm.com>
9473 David Sherwood <david.sherwood@arm.com>
9475 * simplify-rtx.c (simplify_binary_operation_1): Use
9476 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
9478 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9479 Alan Hayward <alan.hayward@arm.com>
9480 David Sherwood <david.sherwood@arm.com>
9482 * wide-int.h (int_traits<unsigned char>) New class.
9483 (int_traits<unsigned short>) Likewise.
9484 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
9485 Use GET_MODE_UNIT_PRECISION and remove redundant test for
9487 * combine.c (set_nonzero_bits_and_sign_copies): Use
9488 is_a <scalar_int_mode>.
9489 (find_split_point): Likewise.
9490 (combine_simplify_rtx): Likewise.
9491 (simplify_logical): Likewise.
9492 (expand_compound_operation): Likewise.
9493 (expand_field_assignment): Likewise.
9494 (make_compound_operation): Likewise.
9495 (extended_count): Likewise.
9496 (change_zero_ext): Likewise.
9497 (simplify_comparison): Likewise.
9498 * dwarf2out.c (scompare_loc_descriptor): Likewise.
9499 (ucompare_loc_descriptor): Likewise.
9500 (minmax_loc_descriptor): Likewise.
9501 (mem_loc_descriptor): Likewise.
9502 (loc_descriptor): Likewise.
9503 * expmed.c (init_expmed_one_mode): Likewise.
9504 * lra-constraints.c (lra_constraint_offset): Likewise.
9505 * optabs.c (prepare_libcall_arg): Likewise.
9506 * postreload.c (move2add_note_store): Likewise.
9507 * reload.c (operands_match_p): Likewise.
9508 * rtl.h (load_extend_op): Likewise.
9509 * rtlhooks.c (gen_lowpart_general): Likewise.
9510 * simplify-rtx.c (simplify_truncation): Likewise.
9511 (simplify_unary_operation_1): Likewise.
9512 (simplify_binary_operation_1): Likewise.
9513 (simplify_const_binary_operation): Likewise.
9514 (simplify_const_relational_operation): Likewise.
9515 (simplify_subreg): Likewise.
9516 * stor-layout.c (bitwise_mode_for_mode): Likewise.
9517 * var-tracking.c (adjust_mems): Likewise.
9518 (prepare_call_arguments): Likewise.
9520 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9521 Alan Hayward <alan.hayward@arm.com>
9522 David Sherwood <david.sherwood@arm.com>
9524 * machmode.h (is_int_mode): New fuction.
9525 * combine.c (find_split_point): Use it.
9526 (combine_simplify_rtx): Likewise.
9527 (simplify_if_then_else): Likewise.
9528 (simplify_set): Likewise.
9529 (simplify_shift_const_1): Likewise.
9530 (simplify_comparison): Likewise.
9531 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
9532 * cse.c (notreg_cost): Likewise.
9533 (cse_insn): Likewise.
9534 * cselib.c (cselib_lookup_1): Likewise.
9535 * dojump.c (do_jump_1): Likewise.
9536 (do_compare_rtx_and_jump): Likewise.
9537 * dse.c (get_call_args): Likewise.
9538 * dwarf2out.c (rtl_for_decl_init): Likewise.
9539 (native_encode_initializer): Likewise.
9540 * expmed.c (emit_store_flag_1): Likewise.
9541 (emit_store_flag): Likewise.
9542 * expr.c (convert_modes): Likewise.
9543 (store_field): Likewise.
9544 (expand_expr_real_1): Likewise.
9545 * fold-const.c (fold_read_from_constant_string): Likewise.
9546 * gimple-ssa-sprintf.c (get_format_string): Likewise.
9547 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
9548 * optabs.c (expand_binop): Likewise.
9549 (expand_unop): Likewise.
9550 (expand_abs_nojump): Likewise.
9551 (expand_one_cmpl_abs_nojump): Likewise.
9552 * simplify-rtx.c (mode_signbit_p): Likewise.
9553 (val_signbit_p): Likewise.
9554 (val_signbit_known_set_p): Likewise.
9555 (val_signbit_known_clear_p): Likewise.
9556 (simplify_relational_operation_1): Likewise.
9557 * tree.c (vector_type_mode): Likewise.
9559 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9560 Alan Hayward <alan.hayward@arm.com>
9561 David Sherwood <david.sherwood@arm.com>
9563 * machmode.h (smallest_mode_for_size): Fix formatting.
9564 (smallest_int_mode_for_size): New function.
9565 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
9566 instead of smallest_mode_for_size.
9567 * combine.c (make_extraction): Likewise.
9568 * config/arc/arc.c (arc_expand_movmem): Likewise.
9569 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
9570 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
9571 * config/s390/s390.c (s390_expand_insv): Likewise.
9572 * config/sparc/sparc.c (assign_int_registers): Likewise.
9573 * config/spu/spu.c (spu_function_value): Likewise.
9574 (spu_function_arg): Likewise.
9575 * coverage.c (get_gcov_type): Likewise.
9576 (get_gcov_unsigned_t): Likewise.
9577 * dse.c (find_shift_sequence): Likewise.
9578 * expmed.c (store_bit_field_1): Likewise.
9579 * expr.c (convert_move): Likewise.
9580 (store_field): Likewise.
9581 * internal-fn.c (expand_arith_overflow): Likewise.
9582 * optabs-query.c (get_best_extraction_insn): Likewise.
9583 * optabs.c (expand_twoval_binop_libfunc): Likewise.
9584 * stor-layout.c (layout_type): Likewise.
9585 (initialize_sizetypes): Likewise.
9586 * targhooks.c (default_get_mask_mode): Likewise.
9587 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
9589 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9590 Alan Hayward <alan.hayward@arm.com>
9591 David Sherwood <david.sherwood@arm.com>
9593 * machmode.h (opt_mode::else_blk): New function.
9594 (int_mode_for_mode): Declare.
9595 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
9596 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
9598 * cfgexpand.c (expand_debug_expr): Likewise.
9599 * combine.c (gen_lowpart_or_truncate): Likewise.
9600 (gen_lowpart_for_combine): Likewise.
9601 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
9602 * config/avr/avr.c (avr_to_int_mode): Likewise.
9603 (avr_out_plus_1): Likewise.
9604 (avr_out_plus): Likewise.
9605 (avr_out_round): Likewise.
9606 * config/i386/i386.c (ix86_split_to_parts): Likewise.
9607 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
9608 (s390_expand_vcond): Likewise.
9609 * config/spu/spu.c (spu_split_immediate): Likewise.
9610 (spu_expand_mov): Likewise.
9611 * dse.c (get_stored_val): Likewise.
9612 * expmed.c (store_bit_field_1): Likewise.
9613 (convert_extracted_bit_field): Use int_mode_for_mode instead of
9615 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
9616 (extract_low_bits): Likewise.
9617 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
9618 handling rather than repeating the check.
9619 (emit_group_store): Likewise.
9620 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
9621 * optabs.c (expand_absneg_bit): Likewise.
9622 (expand_copysign_absneg): Likewise.
9623 (expand_copysign_bit): Likewise.
9624 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
9625 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
9626 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
9627 * var-tracking.c (prepare_call_arguments): Likewise.
9628 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
9629 int_mode_for_mode instead of mode_for_size.
9630 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
9632 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9633 Alan Hayward <alan.hayward@arm.com>
9634 David Sherwood <david.sherwood@arm.com>
9636 * machmode.h (int_mode_for_size): New function.
9637 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
9638 instead of mode_for_size.
9639 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
9641 * combine.c (expand_field_assignment): Use int_mode_for_size
9642 instead of mode_for_size.
9643 (make_extraction): Likewise.
9644 (simplify_shift_const_1): Likewise.
9645 (simplify_comparison): Likewise.
9646 * dojump.c (do_jump): Likewise.
9647 * dwarf2out.c (mem_loc_descriptor): Likewise.
9648 * emit-rtl.c (init_derived_machine_modes): Likewise.
9649 * expmed.c (flip_storage_order): Likewise.
9650 (convert_extracted_bit_field): Likewise.
9651 * expr.c (copy_blkmode_from_reg): Likewise.
9652 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
9653 * internal-fn.c (expand_mul_overflow): Likewise.
9654 * lower-subreg.c (simple_move): Likewise.
9655 * optabs-libfuncs.c (init_optabs): Likewise.
9656 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9657 * tree.c (vector_type_mode): Likewise.
9658 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
9659 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
9660 * tree-vect-generic.c (expand_vector_parallel): Likewise.
9661 * tree-vect-stmts.c (vectorizable_load): Likewise.
9662 (vectorizable_store): Likewise.
9664 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9665 Alan Hayward <alan.hayward@arm.com>
9666 David Sherwood <david.sherwood@arm.com>
9668 * coretypes.h (pod_mode): New type.
9669 (scalar_int_mode_pod): New typedef.
9670 * machmode.h (pod_mode): New class.
9671 (int_n_data_t::m): Change type to scalar_int_mode_pod.
9672 * genmodes.c (emit_mode_int_n): Update accordingly.
9673 * lower-subreg.h (target_lower_subreg): Change type to
9674 scalar_int_mode_pod.
9675 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
9676 scalar_int_mode_pod.
9678 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9679 Alan Hayward <alan.hayward@arm.com>
9680 David Sherwood <david.sherwood@arm.com>
9682 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
9683 machine_mode to scalar_int_mode.
9684 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
9685 (rs6000_option_override_internal): Remove cast to int.
9686 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
9687 machine_mode to scalar_int_mode.
9688 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
9689 (rs6000_option_override_internal): Remove cast to int.
9690 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
9691 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
9693 * config/s390/s390.c (s390_expand_builtin): Likewise.
9694 * coretypes.h (scalar_int_mode): New type.
9695 (opt_scalar_int_mode): New typedef.
9696 * machmode.h (scalar_int_mode): New class.
9697 (scalar_int_mode::includes_p): New function.
9698 (byte_mode): Change type to scalar_int_mode.
9699 (word_mode): Likewise.
9700 (ptr_mode): Likewise.
9701 * emit-rtl.c (byte_mode): Likewise.
9702 (word_mode): Likewise.
9703 (ptr_mode): Likewise.
9704 (init_derived_machine_modes): Update accordingly.
9705 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
9706 and MODE_PARTIAL_INT.
9707 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
9708 opt_scalar_int_mode.
9710 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9711 Alan Hayward <alan.hayward@arm.com>
9712 David Sherwood <david.sherwood@arm.com>
9714 * target.def (libgcc_floating_mode_supported_p): Take a
9716 * doc/tm.texi: Regenerate.
9717 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
9719 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
9720 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
9723 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9724 Alan Hayward <alan.hayward@arm.com>
9725 David Sherwood <david.sherwood@arm.com>
9727 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
9728 * doc/tm.texi: Regenerate.
9729 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
9730 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
9731 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
9732 * targhooks.h (default_floatn_mode): Likewise.
9733 * targhooks.c (default_floatn_mode): Likewise.
9734 * tree.c (build_common_tree_nodes): Update accordingly.
9736 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9737 Alan Hayward <alan.hayward@arm.com>
9738 David Sherwood <david.sherwood@arm.com>
9740 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
9741 (mode_iterator::iterate_p): Likewise.
9742 (mode_iterator::get_wider): Likewise.
9743 * expr.c (init_expr_target): Use opt_scalar_float_mode.
9745 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9746 Alan Hayward <alan.hayward@arm.com>
9747 David Sherwood <david.sherwood@arm.com>
9749 * coretypes.h (opt_scalar_float_mode): New typedef.
9750 * machmode.h (float_mode_for_size): New function.
9751 * emit-rtl.c (double_mode): Delete.
9752 (init_emit_once): Use float_mode_for_size.
9753 * stor-layout.c (layout_type): Likewise.
9754 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
9756 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9757 Alan Hayward <alan.hayward@arm.com>
9758 David Sherwood <david.sherwood@arm.com>
9760 * output.h (assemble_real): Take a scalar_float_mode.
9761 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
9762 * config/arm/arm.md (consttable_4): Likewise.
9763 (consttable_8): Likewise.
9764 (consttable_16): Likewise.
9765 * config/mips/mips.md (consttable_float): Likewise.
9766 * config/s390/s390.c (s390_output_pool_entry): Likewise.
9767 * varasm.c (assemble_real): Take a scalar_float_mode.
9768 (output_constant_pool_2): Update accordingly.
9769 (output_constant): Likewise.
9771 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9772 Alan Hayward <alan.hayward@arm.com>
9773 David Sherwood <david.sherwood@arm.com>
9775 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
9776 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
9777 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
9778 (native_encode_real): Likewise.
9779 (native_interpret_real): Likewise.
9780 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
9781 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9783 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9784 Alan Hayward <alan.hayward@arm.com>
9785 David Sherwood <david.sherwood@arm.com>
9787 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
9788 <scalar_float_mode>. Simplify.
9789 (gen_extend_conv_libfunc): Likewise.
9791 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9792 Alan Hayward <alan.hayward@arm.com>
9793 David Sherwood <david.sherwood@arm.com>
9795 * coretypes.h (scalar_float_mode): New type.
9796 * machmode.h (mode_traits::from_int): Use machine_mode if
9797 USE_ENUM_MODES is defined.
9798 (is_a): New function.
9800 (dyn_cast): Likewise.
9801 (scalar_float_mode): New class.
9802 (scalar_float_mode::includes_p): New function.
9803 (is_float_mode): Likewise.
9804 * gdbhooks.py (MachineModePrinter): New class.
9805 (build_pretty_printer): Use it for scalar_float_mode.
9806 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
9807 (format_helper::format_helper): Turn into a template.
9808 * genmodes.c (get_mode_class): New function.
9809 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
9810 or machine_mode if none.
9811 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
9812 as_a <scalar_float_mode>.
9813 * dwarf2out.c (mem_loc_descriptor): Likewise.
9814 (insert_float): Likewise.
9815 (add_const_value_attribute): Likewise.
9816 * simplify-rtx.c (simplify_immed_subreg): Likewise.
9817 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
9818 (expand_unop): Update accordingly.
9819 (expand_abs_nojump): Likewise.
9820 (expand_copysign_absneg): Take a scalar_float_mode.
9821 (expand_copysign_bit): Likewise.
9822 (expand_copysign): Update accordingly.
9824 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9825 Alan Hayward <alan.hayward@arm.com>
9826 David Sherwood <david.sherwood@arm.com>
9828 * coretypes.h (opt_mode): New class.
9829 * machmode.h (opt_mode): Likewise.
9830 (opt_mode::else_void): New function.
9831 (opt_mode::require): Likewise.
9832 (opt_mode::exists): Likewise.
9833 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
9834 (GET_MODE_2XWIDER_MODE): Likewise.
9835 (mode_iterator::get_wider): Update accordingly.
9836 (mode_iterator::get_2xwider): Likewise.
9837 (mode_iterator::get_known_wider): Likewise, turning into a template.
9838 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
9839 forcing a wider mode to exist.
9840 * config/cr16/cr16.h (LONG_REG_P): Likewise.
9841 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
9842 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
9843 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
9844 * lower-subreg.c (init_lower_subreg): Likewise.
9845 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
9846 on the final iteration.
9847 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
9848 a wider mode exists before asking for a move pattern.
9849 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
9850 forcing a wider mode to exist.
9851 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
9852 returning false if no such mode exists.
9853 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
9854 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
9855 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
9856 Avoid checking for a MODE_INT if we already know the mode is not a
9858 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
9859 forcing a wider mode to exist.
9860 (expmed_mult_highpart_optab): Likewise.
9861 (expmed_mult_highpart): Likewise.
9862 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
9864 * lto-streamer-in.c (lto_input_mode_table): Likewise.
9865 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
9866 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9867 * internal-fn.c (expand_mul_overflow): Update use of
9868 GET_MODE_2XWIDER_MODE.
9869 * omp-low.c (omp_clause_aligned_alignment): Likewise.
9870 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
9871 GET_MODE_WIDER_MODE.
9872 (convert_plusminus_to_widen): Likewise.
9873 * tree-switch-conversion.c (array_value_type): Likewise.
9874 * var-tracking.c (emit_note_insn_var_location): Likewise.
9875 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9876 Return false inside rather than outside the loop if no wider mode
9878 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
9879 and GET_MODE_2XWIDER_MODE
9880 (can_compare_p): Use else_void.
9881 * gdbhooks.py (OptMachineModePrinter): New class.
9882 (build_pretty_printer): Use it for opt_mode.
9884 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9885 Alan Hayward <alan.hayward@arm.com>
9886 David Sherwood <david.sherwood@arm.com>
9888 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
9889 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
9891 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9892 Alan Hayward <alan.hayward@arm.com>
9893 David Sherwood <david.sherwood@arm.com>
9895 * machmode.h (mode_traits): New structure.
9896 (get_narrowest_mode): New function.
9897 (mode_iterator::start): Likewise.
9898 (mode_iterator::iterate_p): Likewise.
9899 (mode_iterator::get_wider): Likewise.
9900 (mode_iterator::get_known_wider): Likewise.
9901 (mode_iterator::get_2xwider): Likewise.
9902 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
9903 (FOR_EACH_MODE): Likewise.
9904 (FOR_EACH_MODE_FROM): Likewise.
9905 (FOR_EACH_MODE_UNTIL): Likewise.
9906 (FOR_EACH_WIDER_MODE): Likewise.
9907 (FOR_EACH_2XWIDER_MODE): Likewise.
9908 * builtins.c (expand_builtin_strlen): Use new mode iterators.
9909 * combine.c (simplify_comparison): Likewise
9910 * config/i386/i386.c (type_natural_mode): Likewise.
9911 * cse.c (cse_insn): Likewise.
9912 * dse.c (find_shift_sequence): Likewise.
9913 * emit-rtl.c (init_derived_machine_modes): Likewise.
9914 (init_emit_once): Likewise.
9915 * explow.c (hard_function_value): Likewise.
9916 * expmed.c (extract_fixed_bit_field_1): Likewise.
9917 (extract_bit_field_1): Likewise.
9918 (expand_divmod): Likewise.
9919 (emit_store_flag_1): Likewise.
9920 * expr.c (init_expr_target): Likewise.
9921 (convert_move): Likewise.
9922 (alignment_for_piecewise_move): Likewise.
9923 (widest_int_mode_for_size): Likewise.
9924 (emit_block_move_via_movmem): Likewise.
9925 (copy_blkmode_to_reg): Likewise.
9926 (set_storage_via_setmem): Likewise.
9927 (compress_float_constant): Likewise.
9928 * omp-low.c (omp_clause_aligned_alignment): Likewise.
9929 * optabs-query.c (get_best_extraction_insn): Likewise.
9930 * optabs.c (expand_binop): Likewise.
9931 (expand_twoval_unop): Likewise.
9932 (expand_twoval_binop): Likewise.
9933 (widen_leading): Likewise.
9934 (widen_bswap): Likewise.
9935 (expand_parity): Likewise.
9936 (expand_unop): Likewise.
9937 (prepare_cmp_insn): Likewise.
9938 (prepare_float_lib_cmp): Likewise.
9939 (expand_float): Likewise.
9940 (expand_fix): Likewise.
9941 (expand_sfix_optab): Likewise.
9942 * postreload.c (move2add_use_add2_insn): Likewise.
9943 * reg-stack.c (reg_to_stack): Likewise.
9944 * reginfo.c (choose_hard_reg_mode): Likewise.
9945 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
9946 * stor-layout.c (mode_for_size): Likewise.
9947 (smallest_mode_for_size): Likewise.
9948 (mode_for_vector): Likewise.
9949 (finish_bitfield_representative): Likewise.
9950 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
9951 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
9952 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9953 * var-tracking.c (prepare_call_arguments): Likewise.
9955 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9956 Alan Hayward <alan.hayward@arm.com>
9957 David Sherwood <david.sherwood@arm.com>
9959 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
9960 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
9961 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
9962 * machmode.h (mode_size): Move earlier in file.
9963 (mode_precision): Likewise.
9964 (mode_inner): Likewise.
9965 (mode_nunits): Likewise.
9966 (mode_unit_size): Likewise.
9967 (unit_unit_precision): Likewise.
9968 (mode_wider): Likewise.
9969 (mode_2xwider): Likewise.
9970 (machine_mode): New class.
9971 (mode_to_bytes): New function.
9972 (mode_to_bits): Likewise.
9973 (mode_to_precision): Likewise.
9974 (mode_to_inner): Likewise.
9975 (mode_to_unit_size): Likewise.
9976 (mode_to_unit_precision): Likewise.
9977 (mode_to_nunits): Likewise.
9978 (GET_MODE_SIZE): Use mode_to_bytes.
9979 (GET_MODE_BITSIZE): Use mode_to_bits.
9980 (GET_MODE_PRECISION): Use mode_to_precision.
9981 (GET_MODE_INNER): Use mode_to_inner.
9982 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
9983 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
9984 (GET_MODE_NUNITS): Use mode_to_nunits.
9985 * system.h (ALWAYS_INLINE): New macro.
9986 * config/powerpcspe/powerpcspe-c.c
9987 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
9988 int for arg1_mode and arg2_mode.
9990 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9991 Alan Hayward <alan.hayward@arm.com>
9992 David Sherwood <david.sherwood@arm.com>
9994 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
9995 Prefix mode names with E_ in case statements.
9996 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9997 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
9998 (aarch64_split_simd_move): Likewise.
9999 (aarch64_gen_storewb_pair): Likewise.
10000 (aarch64_gen_loadwb_pair): Likewise.
10001 (aarch64_gen_store_pair): Likewise.
10002 (aarch64_gen_load_pair): Likewise.
10003 (aarch64_get_condition_code_1): Likewise.
10004 (aarch64_constant_pool_reload_icode): Likewise.
10005 (get_rsqrte_type): Likewise.
10006 (get_rsqrts_type): Likewise.
10007 (get_recpe_type): Likewise.
10008 (get_recps_type): Likewise.
10009 (aarch64_gimplify_va_arg_expr): Likewise.
10010 (aarch64_simd_container_mode): Likewise.
10011 (aarch64_emit_load_exclusive): Likewise.
10012 (aarch64_emit_store_exclusive): Likewise.
10013 (aarch64_expand_compare_and_swap): Likewise.
10014 (aarch64_gen_atomic_cas): Likewise.
10015 (aarch64_emit_bic): Likewise.
10016 (aarch64_emit_atomic_swap): Likewise.
10017 (aarch64_emit_atomic_load_op): Likewise.
10018 (aarch64_evpc_trn): Likewise.
10019 (aarch64_evpc_uzp): Likewise.
10020 (aarch64_evpc_zip): Likewise.
10021 (aarch64_evpc_ext): Likewise.
10022 (aarch64_evpc_rev): Likewise.
10023 (aarch64_evpc_dup): Likewise.
10024 (aarch64_gen_ccmp_first): Likewise.
10025 (aarch64_gen_ccmp_next): Likewise.
10026 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
10027 (alpha_emit_xfloating_libcall): Likewise.
10028 (emit_insxl): Likewise.
10029 (alpha_arg_type): Likewise.
10030 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
10031 (arc_preferred_simd_mode): Likewise.
10032 (arc_secondary_reload): Likewise.
10033 (get_arc_condition_code): Likewise.
10034 (arc_print_operand): Likewise.
10035 (arc_legitimate_constant_p): Likewise.
10036 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10037 * config/arc/arc.md (casesi_load): Likewise.
10038 (casesi_compact_jump): Likewise.
10039 * config/arc/predicates.md (proper_comparison_operator): Likewise.
10040 (cc_use_register): Likewise.
10041 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10042 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
10043 (arm_init_iwmmxt_builtins): Likewise.
10044 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
10045 (neon_expand_vector_init): Likewise.
10046 (arm_attr_length_move_neon): Likewise.
10047 (maybe_get_arm_condition_code): Likewise.
10048 (arm_emit_vector_const): Likewise.
10049 (arm_preferred_simd_mode): Likewise.
10050 (arm_output_iwmmxt_tinsr): Likewise.
10051 (thumb1_output_casesi): Likewise.
10052 (thumb2_output_casesi): Likewise.
10053 (arm_emit_load_exclusive): Likewise.
10054 (arm_emit_store_exclusive): Likewise.
10055 (arm_expand_compare_and_swap): Likewise.
10056 (arm_evpc_neon_vuzp): Likewise.
10057 (arm_evpc_neon_vzip): Likewise.
10058 (arm_evpc_neon_vrev): Likewise.
10059 (arm_evpc_neon_vtrn): Likewise.
10060 (arm_evpc_neon_vext): Likewise.
10061 (arm_validize_comparison): Likewise.
10062 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
10063 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
10064 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
10065 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
10066 (c6x_preferred_simd_mode): Likewise.
10067 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
10068 (epiphany_rtx_costs): Likewise.
10069 * config/epiphany/predicates.md (proper_comparison_operator):
10071 * config/frv/frv.c (condexec_memory_operand): Likewise.
10072 (frv_emit_move): Likewise.
10073 (output_move_single): Likewise.
10074 (output_condmove_single): Likewise.
10075 (frv_hard_regno_mode_ok): Likewise.
10076 (frv_matching_accg_mode): Likewise.
10077 * config/h8300/h8300.c (split_adds_subs): Likewise.
10078 (h8300_rtx_costs): Likewise.
10079 (h8300_print_operand): Likewise.
10080 (compute_mov_length): Likewise.
10081 (output_logical_op): Likewise.
10082 (compute_logical_op_length): Likewise.
10083 (compute_logical_op_cc): Likewise.
10084 (h8300_shift_needs_scratch_p): Likewise.
10085 (output_a_shift): Likewise.
10086 (compute_a_shift_length): Likewise.
10087 (compute_a_shift_cc): Likewise.
10088 (expand_a_rotate): Likewise.
10089 (output_a_rotate): Likewise.
10090 * config/i386/i386.c (classify_argument): Likewise.
10091 (function_arg_advance_32): Likewise.
10092 (function_arg_32): Likewise.
10093 (function_arg_64): Likewise.
10094 (function_value_64): Likewise.
10095 (ix86_gimplify_va_arg): Likewise.
10096 (ix86_legitimate_constant_p): Likewise.
10097 (put_condition_code): Likewise.
10098 (split_double_mode): Likewise.
10099 (ix86_avx256_split_vector_move_misalign): Likewise.
10100 (ix86_expand_vector_logical_operator): Likewise.
10101 (ix86_split_idivmod): Likewise.
10102 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
10103 (ix86_build_const_vector): Likewise.
10104 (ix86_build_signbit_mask): Likewise.
10105 (ix86_match_ccmode): Likewise.
10106 (ix86_cc_modes_compatible): Likewise.
10107 (ix86_expand_branch): Likewise.
10108 (ix86_expand_sse_cmp): Likewise.
10109 (ix86_expand_sse_movcc): Likewise.
10110 (ix86_expand_int_sse_cmp): Likewise.
10111 (ix86_expand_vec_perm_vpermi2): Likewise.
10112 (ix86_expand_vec_perm): Likewise.
10113 (ix86_expand_sse_unpack): Likewise.
10114 (ix86_expand_int_addcc): Likewise.
10115 (ix86_split_to_parts): Likewise.
10116 (ix86_vectorize_builtin_gather): Likewise.
10117 (ix86_vectorize_builtin_scatter): Likewise.
10118 (avx_vpermilp_parallel): Likewise.
10119 (inline_memory_move_cost): Likewise.
10120 (ix86_tieable_integer_mode_p): Likewise.
10121 (x86_maybe_negate_const_int): Likewise.
10122 (ix86_expand_vector_init_duplicate): Likewise.
10123 (ix86_expand_vector_init_one_nonzero): Likewise.
10124 (ix86_expand_vector_init_one_var): Likewise.
10125 (ix86_expand_vector_init_concat): Likewise.
10126 (ix86_expand_vector_init_interleave): Likewise.
10127 (ix86_expand_vector_init_general): Likewise.
10128 (ix86_expand_vector_set): Likewise.
10129 (ix86_expand_vector_extract): Likewise.
10130 (emit_reduc_half): Likewise.
10131 (ix86_emit_i387_round): Likewise.
10132 (ix86_mangle_type): Likewise.
10133 (ix86_expand_round_sse4): Likewise.
10134 (expand_vec_perm_blend): Likewise.
10135 (canonicalize_vector_int_perm): Likewise.
10136 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
10137 (expand_vec_perm_1): Likewise.
10138 (expand_vec_perm_interleave3): Likewise.
10139 (expand_vec_perm_even_odd_pack): Likewise.
10140 (expand_vec_perm_even_odd_1): Likewise.
10141 (expand_vec_perm_broadcast_1): Likewise.
10142 (ix86_vectorize_vec_perm_const_ok): Likewise.
10143 (ix86_expand_vecop_qihi): Likewise.
10144 (ix86_expand_mul_widen_hilo): Likewise.
10145 (ix86_expand_sse2_abs): Likewise.
10146 (ix86_expand_pextr): Likewise.
10147 (ix86_expand_pinsr): Likewise.
10148 (ix86_preferred_simd_mode): Likewise.
10149 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
10150 * config/i386/sse.md (*andnot<mode>3): Likewise.
10151 (<mask_codefor><code><mode>3<mask_name>): Likewise.
10152 (*<code><mode>3): Likewise.
10153 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
10154 (ia64_expand_atomic_op): Likewise.
10155 (ia64_arg_type): Likewise.
10156 (ia64_mode_to_int): Likewise.
10157 (ia64_scalar_mode_supported_p): Likewise.
10158 (ia64_vector_mode_supported_p): Likewise.
10159 (expand_vec_perm_broadcast): Likewise.
10160 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
10161 (iq2000_function_arg_advance): Likewise.
10162 (iq2000_function_arg): Likewise.
10163 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
10164 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
10165 (m68k_libcall_value): Likewise.
10166 (m68k_function_value): Likewise.
10167 (sched_attr_op_type): Likewise.
10168 * config/mcore/mcore.c (mcore_output_move): Likewise.
10169 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
10171 (microblaze_function_arg): Likewise.
10172 * config/mips/mips.c (mips16_build_call_stub): Likewise.
10173 (mips_print_operand): Likewise.
10174 (mips_mode_ok_for_mov_fmt_p): Likewise.
10175 (mips_vector_mode_supported_p): Likewise.
10176 (mips_preferred_simd_mode): Likewise.
10177 (mips_expand_vpc_loongson_even_odd): Likewise.
10178 (mips_expand_vec_unpack): Likewise.
10179 (mips_expand_vi_broadcast): Likewise.
10180 (mips_expand_vector_init): Likewise.
10181 (mips_expand_vec_reduc): Likewise.
10182 (mips_expand_msa_cmp): Likewise.
10183 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
10184 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
10185 (cc_flags_for_mode): Likewise.
10186 * config/msp430/msp430.c (msp430_print_operand): Likewise.
10187 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
10188 (nds32_output_casesi_pc_relative): Likewise.
10189 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10190 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
10191 (nvptx_gen_unpack): Likewise.
10192 (nvptx_gen_pack): Likewise.
10193 (nvptx_gen_shuffle): Likewise.
10194 (nvptx_gen_wcast): Likewise.
10195 (nvptx_preferred_simd_mode): Likewise.
10196 * config/pa/pa.c (pa_secondary_reload): Likewise.
10197 * config/pa/predicates.md (base14_operand): Likewise.
10198 * config/powerpcspe/powerpcspe-c.c
10199 (altivec_resolve_overloaded_builtin): Likewise.
10200 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
10202 (rs6000_preferred_simd_mode): Likewise.
10203 (output_vec_const_move): Likewise.
10204 (rs6000_expand_vector_extract): Likewise.
10205 (rs6000_split_vec_extract_var): Likewise.
10206 (reg_offset_addressing_ok_p): Likewise.
10207 (rs6000_legitimate_offset_address_p): Likewise.
10208 (rs6000_legitimize_address): Likewise.
10209 (rs6000_emit_set_const): Likewise.
10210 (rs6000_const_vec): Likewise.
10211 (rs6000_emit_move): Likewise.
10212 (spe_build_register_parallel): Likewise.
10213 (rs6000_darwin64_record_arg_recurse): Likewise.
10214 (swap_selector_for_mode): Likewise.
10215 (spe_init_builtins): Likewise.
10216 (paired_init_builtins): Likewise.
10217 (altivec_init_builtins): Likewise.
10218 (do_load_for_compare): Likewise.
10219 (rs6000_generate_compare): Likewise.
10220 (rs6000_expand_float128_convert): Likewise.
10221 (emit_load_locked): Likewise.
10222 (emit_store_conditional): Likewise.
10223 (rs6000_output_function_epilogue): Likewise.
10224 (rs6000_handle_altivec_attribute): Likewise.
10225 (rs6000_function_value): Likewise.
10226 (emit_fusion_gpr_load): Likewise.
10227 (emit_fusion_p9_load): Likewise.
10228 (emit_fusion_p9_store): Likewise.
10229 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
10230 (fusion_gpr_mem_load): Likewise.
10231 (fusion_addis_mem_combo_load): Likewise.
10232 (fusion_addis_mem_combo_store): Likewise.
10233 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
10234 (fusion_gpr_mem_load): Likewise.
10235 (fusion_addis_mem_combo_load): Likewise.
10236 (fusion_addis_mem_combo_store): Likewise.
10237 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10239 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
10240 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
10241 (rs6000_preferred_simd_mode): Likewise.
10242 (output_vec_const_move): Likewise.
10243 (rs6000_expand_vector_extract): Likewise.
10244 (rs6000_split_vec_extract_var): Likewise.
10245 (reg_offset_addressing_ok_p): Likewise.
10246 (rs6000_legitimate_offset_address_p): Likewise.
10247 (rs6000_legitimize_address): Likewise.
10248 (rs6000_emit_set_const): Likewise.
10249 (rs6000_const_vec): Likewise.
10250 (rs6000_emit_move): Likewise.
10251 (rs6000_darwin64_record_arg_recurse): Likewise.
10252 (swap_selector_for_mode): Likewise.
10253 (paired_init_builtins): Likewise.
10254 (altivec_init_builtins): Likewise.
10255 (rs6000_expand_float128_convert): Likewise.
10256 (emit_load_locked): Likewise.
10257 (emit_store_conditional): Likewise.
10258 (rs6000_output_function_epilogue): Likewise.
10259 (rs6000_handle_altivec_attribute): Likewise.
10260 (rs6000_function_value): Likewise.
10261 (emit_fusion_gpr_load): Likewise.
10262 (emit_fusion_p9_load): Likewise.
10263 (emit_fusion_p9_store): Likewise.
10264 * config/rx/rx.c (rx_gen_move_template): Likewise.
10265 (flags_from_mode): Likewise.
10266 * config/s390/predicates.md (s390_alc_comparison): Likewise.
10267 (s390_slb_comparison): Likewise.
10268 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
10269 (s390_vector_mode_supported_p): Likewise.
10270 (s390_cc_modes_compatible): Likewise.
10271 (s390_match_ccmode_set): Likewise.
10272 (s390_canonicalize_comparison): Likewise.
10273 (s390_emit_compare_and_swap): Likewise.
10274 (s390_branch_condition_mask): Likewise.
10275 (s390_rtx_costs): Likewise.
10276 (s390_secondary_reload): Likewise.
10277 (__SECONDARY_RELOAD_CASE): Likewise.
10278 (s390_expand_cs): Likewise.
10279 (s390_preferred_simd_mode): Likewise.
10280 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
10281 * config/sh/sh.c (sh_print_operand): Likewise.
10282 (dump_table): Likewise.
10283 (sh_secondary_reload): Likewise.
10284 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10285 * config/sh/sh.md (casesi_worker_1): Likewise.
10286 (casesi_worker_2): Likewise.
10287 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
10288 (fcc_comparison_operator): Likewise.
10289 * config/sparc/sparc.c (sparc_expand_move): Likewise.
10290 (emit_soft_tfmode_cvt): Likewise.
10291 (sparc_preferred_simd_mode): Likewise.
10292 (output_cbranch): Likewise.
10293 (sparc_print_operand): Likewise.
10294 (sparc_expand_vec_perm_bmask): Likewise.
10295 (vector_init_bshuffle): Likewise.
10296 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
10297 (spu_vector_mode_supported_p): Likewise.
10298 (spu_expand_insv): Likewise.
10299 (spu_emit_branch_or_set): Likewise.
10300 (spu_handle_vector_attribute): Likewise.
10301 (spu_builtin_splats): Likewise.
10302 (spu_builtin_extract): Likewise.
10303 (spu_builtin_promote): Likewise.
10304 (spu_expand_sign_extend): Likewise.
10305 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
10306 (tilegx_simd_int): Likewise.
10307 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
10308 (tilepro_simd_int): Likewise.
10309 * config/v850/v850.c (const_double_split): Likewise.
10310 (v850_print_operand): Likewise.
10311 (ep_memory_offset): Likewise.
10312 * config/vax/vax.c (vax_rtx_costs): Likewise.
10313 (vax_output_int_move): Likewise.
10314 (vax_output_int_add): Likewise.
10315 (vax_output_int_subtract): Likewise.
10316 * config/visium/predicates.md (visium_branch_operator): Likewise.
10317 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
10318 (visium_print_operand_address): Likewise.
10319 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10320 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
10321 (xtensa_expand_conditional_branch): Likewise.
10322 (xtensa_copy_incoming_a7): Likewise.
10323 (xtensa_output_literal): Likewise.
10324 * dfp.c (decimal_real_maxval): Likewise.
10325 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
10327 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10328 Alan Hayward <alan.hayward@arm.com>
10329 David Sherwood <david.sherwood@arm.com>
10331 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
10332 (mode_nunits_inline): Likewise.
10333 (mode_inner_inline): Likewise.
10334 (mode_unit_size_inline): Likewise.
10335 (mode_unit_precision_inline): Likewise.
10336 (emit_insn_modes_h): Likewise. Also emit a #define of the
10338 (emit_mode_wider): Add an E_ prefix to mode names.
10339 (emit_mode_complex): Likewise.
10340 (emit_mode_inner): Likewise.
10341 (emit_mode_adjustments): Likewise.
10342 (emit_mode_int_n): Likewise.
10343 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
10344 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
10345 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
10346 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
10347 (CRC32_BUILTIN, ENTRY): Likewise.
10348 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
10349 (aarch64_pop_regs): Likewise.
10350 (aarch64_process_components): Likewise.
10351 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
10352 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
10353 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
10354 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
10355 * config/arm/arm.c (arm_init_libfuncs): Likewise.
10356 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
10358 * config/i386/i386-builtin.def (pcmpestr): Likewise.
10359 (pcmpistr): Likewise.
10360 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
10361 * config/mmix/mmix.c (mmix_output_condition): Likewise.
10362 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
10364 * config/rl78/rl78.c (mduc_regs): Likewise.
10365 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
10366 (htm_expand_builtin): Likewise.
10367 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
10368 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
10369 * config/xtensa/xtensa.c (print_operand): Likewise.
10370 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
10371 (NUM_MODE_VECTOR_INT): Likewise.
10372 * genoutput.c (null_operand): Likewise.
10373 (output_operand_data): Likewise.
10374 * genrecog.c (print_parameter_value): Likewise.
10375 * lra.c (debug_operand_data): Likewise.
10377 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10379 * dwarf2out.c (scompare_loc_descriptor_wide)
10380 (scompare_loc_descriptor_narrow): New functions, split out from...
10381 (scompare_loc_descriptor): ...here.
10382 * expmed.c (emit_store_flag_int): New function, split out from...
10383 (emit_store_flag): ...here.
10385 2017-08-30 Richard Biener <rguenther@suse.de>
10387 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
10388 (dwarf2out_early_finish): Move setting of AT_pubnames from
10389 early debug output to early finish.
10391 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
10393 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
10394 and -mdata-region to the assembler.
10396 2017-08-30 Richard Biener <rguenther@suse.de>
10398 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
10400 (gen_subprogram_die): Add DW_AT_object_pointer only early.
10401 (dwarf2out_early_global_decl): Only generate a DIE for the
10402 abstract origin if it doesn't already exist or is a declaration DIE.
10403 (resolve_addr): Do not add the linkage name twice when
10404 generating a stub DIE for the DW_TAG_GNU_call_site target.
10406 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10408 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10409 Use machine_mode rather than int for arg1_mode.
10411 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
10414 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
10415 that the second argument of the built-in functions to unpack
10416 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
10417 switch statement instead a lot of if statements.
10418 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
10419 Allow 64-bit values to be in Altivec registers as well as
10420 traditional floating point registers.
10421 (pack<mode>, FMOVE128_VSX iterator): Likewise.
10423 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
10425 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
10426 MAX_REGS_PER_ADDRESS == 1.
10428 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
10430 * config/i386/i386.opt (flag_fentry): Do not init to -1.
10431 * config/i386/i386.c (ix86_option_override_internal): Simplify
10432 setting of opts->x_flag_entry.
10434 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10435 Jakub Jelinek <jakub@redhat.com>
10436 Richard Biener <rguenther@suse.de>
10438 PR tree-optimization/81503
10439 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
10440 folded constant fits in the target type; reorder tests for clarity.
10442 2017-08-29 Martin Liska <mliska@suse.cz>
10444 * passes.def: Include pass_lower_switch.
10445 * stmt.c (dump_case_nodes): Remove and move to
10446 tree-switch-conversion.
10447 (case_values_threshold): Likewise.
10448 (expand_switch_as_decision_tree_p): Likewise.
10449 (emit_case_decision_tree): Likewise.
10450 (expand_case): Likewise.
10451 (balance_case_nodes): Likewise.
10452 (node_has_low_bound): Likewise.
10453 (node_has_high_bound): Likewise.
10454 (node_is_bounded): Likewise.
10455 (emit_case_nodes): Likewise.
10456 (struct simple_case_node): New struct.
10457 (add_case_node): Remove.
10458 (emit_case_dispatch_table): Use vector instead of case_list.
10459 (reset_out_edges_aux): Remove.
10460 (compute_cases_per_edge): Likewise.
10461 (expand_case): Build list of simple_case_node.
10462 (expand_sjlj_dispatch_table): Use it.
10463 * tree-switch-conversion.c (struct case_node): Moved from
10464 stmt.c and adjusted.
10465 (emit_case_nodes): Likewise.
10466 (node_has_low_bound): Likewise.
10467 (node_has_high_bound): Likewise.
10468 (node_is_bounded): Likewise.
10469 (case_values_threshold): Likewise.
10470 (reset_out_edges_aux): Likewise.
10471 (compute_cases_per_edge): Likewise.
10472 (add_case_node): Likewise.
10473 (dump_case_nodes): Likewise.
10474 (balance_case_nodes): Likewise.
10475 (expand_switch_as_decision_tree_p): Likewise.
10476 (emit_jump): Likewise.
10477 (emit_case_decision_tree): Likewise.
10478 (try_switch_expansion): Likewise.
10479 (do_jump_if_equal): Likewise.
10480 (emit_cmp_and_jump_insns): Likewise.
10481 (fix_phi_operands_for_edge): New function.
10482 (record_phi_operand_mapping): Likewise.
10483 (class pass_lower_switch): New pass.
10484 (pass_lower_switch::execute): New function.
10485 (make_pass_lower_switch): Likewise.
10486 (conditional_probability):
10487 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
10488 * tree-pass.h: Add make_pass_lower_switch.
10490 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
10493 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
10496 2017-08-29 Richard Biener <rguenther@suse.de>
10498 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
10499 we do not add a DW_AT_inline attribute twice.
10500 (gen_subprogram_die): Remove code setting DW_AT_inline on
10501 DECL_ABSTRACT_P nodes.
10503 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
10505 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
10506 calls to internal functions.
10507 (gimplify_modify_expr): Likewise.
10508 * tree-call-cdce.c (use_internal_fn): Likewise.
10509 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
10510 (convert_to_divmod): Set the nothrow flag.
10511 * tree-if-conv.c (predicate_mem_writes): Likewise.
10512 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
10513 (vectorizable_call): Likewise.
10514 (vectorizable_store): Likewise.
10515 (vectorizable_load): Likewise.
10516 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
10517 (vect_recog_mask_conversion_pattern): Likewise.
10519 2017-08-29 Martin Liska <mliska@suse.cz>
10522 * gcc.c (driver_handle_option): Add new argument.
10523 * opts-common.c (handle_option): Pass
10524 target_option_override_hook.
10525 * opts-global.c (lang_handle_option): Add new option.
10526 (set_default_handlers): Add new argument.
10527 (decode_options): Likewise.
10528 * opts.c (target_handle_option): Likewise.
10529 (common_handle_option): Call target_option_override_hook.
10530 * opts.h (struct cl_option_handler_func): Add hook for
10531 target option override.
10532 (struct cl_option_handlers): Likewise.
10533 (set_default_handlers): Add new argument.
10534 (decode_options): Likewise.
10535 (common_handle_option): Likewise.
10536 (target_handle_option): Likewise.
10537 * toplev.c (toplev::main): Pass targetm.target_option.override
10540 2017-08-29 Richard Biener <rguenther@suse.de>
10541 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
10543 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
10544 life to the active subtree.
10546 2017-08-28 Jeff Law <law@redhat.com>
10548 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
10549 derive_equivalences.
10550 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
10551 Code moved into....
10552 (edge_info::derive_equivalences): New private member function
10554 * tree-ssa-dom.c (class edge_info): Changed from a struct
10555 to a class. Add ctor/dtor, methods and data members.
10556 (edge_info::edge_info): Renamed from allocate_edge_info.
10557 Initialize additional members.
10558 (edge_info::~edge_info): New.
10559 (free_dom_edge_info): Delete the edge info.
10560 (record_edge_info): Use new class & associated member functions.
10561 Tighten forms for testing for edge equivalences.
10562 (record_temporary_equivalences): Iterate over the simple
10563 equivalences rather than assuming there's only one per edge.
10564 (cprop_into_successor_phis): Iterate over the simple
10565 equivalences rather than assuming there's only one per edge.
10566 (optimize_stmt): Use operand_equal_p rather than pointer
10567 equality for mini-DSE code.
10569 2017-08-28 Nathan Sidwell <nathan@acm.org>
10571 * gcc.c (execute): Fold SIGPIPE handling into switch
10572 statement. Adjust internal error message.
10574 2017-08-28 Richard Biener <rguenther@suse.de>
10577 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
10578 Do nothing for removed DIEs.
10580 2017-08-28 Richard Biener <rguenther@suse.de>
10582 PR tree-optimization/81977
10583 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
10586 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
10589 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
10590 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
10591 using targetm.gen_mem_thread_fence.
10593 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
10596 * config/i386/i386.md (*<btsc><mode>): Change operand 2
10597 predicate to register_operand. Reorder operands.
10598 (*btr<mode>): Ditto.
10599 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
10600 (*btr<mode>_mask): Ditto.
10602 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
10604 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
10605 * config/rs6000/xmmintrin.h: New file.
10606 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
10608 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10611 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
10612 parameter and_insn and return it.
10613 (recombine_lvx_pattern): Insert a copy to ensure availability of
10614 the base register of the copied masking operation at the point of
10615 the instruction replacement.
10616 (recombine_stvx_pattern): Likewise.
10618 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
10620 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
10621 undocumented switches.
10622 (-mpower9-dform-vector): Likewise.
10623 (-mpower9-dform): Likewise.
10624 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
10625 comments to delete references to -mpower9-dform* switches.
10626 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
10627 Delete reference to -mpower9-dform* switches, test for
10628 -mpower9-vector instead.
10629 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
10630 (OTHER_P9_VECTOR_MASKS): Likewise.
10631 (POWERPC_MASKS): Likewise.
10632 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
10633 tests against -mpower9-dform* to -mpower9-vector. Delete code
10634 that checked for -mpower9-dform* consistancy with other options.
10635 Add test for -mpower9-misc to enable other power9 switches.
10636 (rs6000_init_hard_regno_mode_ok): Likewise.
10637 (rs6000_option_override_internal): Likewise.
10638 (rs6000_emit_prologue): Likewise.
10639 (rs6000_emit_epilogue): Likewise.
10640 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
10641 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
10642 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
10644 (emit_fusion_p9_store): Likewise.
10645 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
10646 resetting these macros if the assembler does not support ISA 3.0
10648 (TARGET_P9_DFORM_VECTOR): Likewise.
10649 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
10650 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
10652 2017-08-25 Alan Modra <amodra@gmail.com>
10654 PR rtl-optimization/81747
10655 * cse.c (cse_extended_basic_block): Don't attempt to record
10656 equivalences for degenerate conditional branches that branch
10657 to their fall-through.
10659 2017-08-24 Martin Sebor <msebor@redhat.com>
10661 PR middle-end/81908
10662 * gimple-fold.c (size_must_be_zero_p): New function.
10663 (gimple_fold_builtin_memory_op): Call it.
10665 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
10667 * config/rs6000/mm_malloc.h: New file.
10669 2017-08-24 Bin Cheng <bin.cheng@arm.com>
10671 PR tree-optimization/81913
10672 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
10673 analysis when either IVs in condition can wrap.
10675 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
10677 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
10678 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
10680 2017-08-24 Richard Biener <rguenther@suse.de>
10683 * targhooks.c (default_target_can_inline_p): Properly
10684 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
10685 is present and always compare.
10686 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
10687 infer -mfpmath=sse from TARGET_SSE_P.
10688 (ix86_can_inline_p): Properly use target_option_default_node when
10689 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
10691 2017-08-24 Richard Biener <rguenther@suse.de>
10694 * dwarf2out.c (output_die): Handle flag_generate_offload like
10696 (output_comp_unit): Likewise.
10697 (gen_array_type_die): Likewise.
10698 (dwarf2out_early_finish): Likewise.
10699 (note_variable_value_in_expr): Likewise.
10700 (dwarf2out_finish): Likewise. Adjust assert.
10701 * cgraphunit.c (symbol_table::compile): Move setting of
10702 flag_generate_offload earlier ...
10703 (symbol_table::finalize_compilation_unit): ... here, before
10704 early debug finalization.
10706 2017-08-24 Richard Biener <rguenther@suse.de>
10708 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
10709 and ipa-fnsummary.h.
10710 (ix86_can_inline_p): When ix86_fpmath flags do not match
10711 check whether the callee uses FP math at all.
10713 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
10715 PR middle-end/81931
10716 * tree-ssanames.c (get_nonzero_bits): Use element_precision
10717 instead of TYPE_PRECISION.
10719 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
10720 Alan Hayward <alan.hayward@arm.com>
10721 David Sherwood <david.sherwood@arm.com>
10723 * combine.c (make_extraction): Use subreg_offset_from_lsb.
10725 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
10727 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
10729 (ix86_frame::stack_realign_allocate): New field.
10730 (struct machine_frame_state): Modify comments.
10731 (machine_frame_state::sp_realigned_fp_end): New field.
10732 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
10733 layout calculation.
10734 (sp_valid_at): Add assertion to assure no attempt to access invalid
10735 offset of a realigned stack.
10736 (fp_valid_at): Likewise.
10737 (choose_baseaddr): Modify comments.
10738 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
10739 ix86_expand_prologue.
10740 (ix86_expand_prologue): Modify stack realignment and allocation.
10741 (ix86_expand_epilogue): Modify comments.
10742 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
10743 avx2_runtime, avx512f, and avx512f_runtime.
10745 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
10747 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
10748 (mstackrealign): Do not init to -1.
10749 * config/i386/i386.c (ix86_option_override_internal):
10750 Check opts_set, not opts when setting default value of
10751 opts->x_ix86_force_align_arg_pointer.
10753 2017-08-23 Richard Biener <rguenther@suse.de>
10755 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
10756 lang_hooks.decl_printable_name.
10757 * print-rtl-function.c (print_rtx_function): Likewise.
10758 * tree-pretty-print.c (dump_function_header): Likewise.
10760 2017-08-23 Richard Biener <rguenther@suse.de>
10763 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
10764 -g0 at compile-time.
10766 2017-08-23 Tamar Christina <tamar.christina@arm.com>
10768 PR middle-end/19706
10769 * doc/sourcebuild.texi (Other hardware attributes):
10772 2017-08-23 Tamar Christina <tamar.christina@arm.com>
10774 PR middle-end/19706
10775 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
10776 Fix single-use check.
10778 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
10780 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
10782 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
10784 * doc/install.texi: Modify to add more details on running selected
10787 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
10789 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
10790 is combined with -mabi=ms.
10791 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
10794 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10796 PR tree-optimization/81488
10797 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
10798 and cached_basis fields.
10799 (MAX_SPREAD): New constant.
10800 (alloc_cand_and_find_basis): Initialize new fields.
10801 (clear_visited): New function.
10802 (create_phi_basis_1): Rename from create_phi_basis, set visited
10803 and cached_basis fields.
10804 (create_phi_basis): New wrapper function.
10805 (phi_add_costs_1): Rename from phi_add_costs, add spread
10806 parameter, set visited field, short-circuit when limits reached.
10807 (phi_add_costs): New wrapper function.
10808 (record_phi_increments_1): Rename from record_phi_increments, set
10810 (record_phi_increments): New wrapper function.
10811 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
10812 (phi_incr_cost): New wrapper function.
10813 (all_phi_incrs_profitable_1): Rename from
10814 all_phi_incrs_profitable, set visited field.
10815 (all_phi_incrs_profitable): New wrapper function.
10817 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
10818 Alan Hayward <alan.hayward@arm.com>
10819 David Sherwood <david.sherwood@arm.com>
10821 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
10822 that takes the outer and inner modes.
10823 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
10824 comparison as the canonical test for a paradoxical subreg.
10825 * combine.c (simplify_set): Use paradoxical_subreg_p.
10826 (make_extraction): Likewise.
10827 (force_to_mode): Likewise.
10828 (rtx_equal_for_field_assignment_p): Likewise.
10829 (gen_lowpart_for_combine): Likewise.
10830 (simplify_comparison): Likewise.
10831 * cse.c (equiv_constant): Likewise.
10832 * expmed.c (store_bit_field_1): Likewise.
10833 * final.c (alter_subreg): Likewise.
10834 * fwprop.c (propagate_rtx): Likewise.
10835 (forward_propagate_subreg): Likewise.
10836 * ira-conflicts.c (ira_build_conflicts): Likewise.
10837 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
10838 * lra-constraints.c (curr_insn_transform): Likewise.
10839 (split_reg): Likewise.
10840 * lra-eliminations.c (move_plus_up): Likewise.
10841 (lra_eliminate_regs_1): Likewise.
10842 * recog.c (general_operand): Likewise.
10843 * ree.c (combine_reaching_defs): Likewise.
10844 * reload.c (push_reload): Likewise.
10845 (find_reloads): Likewise.
10846 * reload1.c (elimination_effects): Likewise.
10847 (compute_reload_subreg_offset): Likewise.
10848 (choose_reload_regs): Likewise.
10849 * rtlanal.c (subreg_lsb_1): Likewise.
10850 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10851 (simplify_subreg): Likewise.
10852 * var-tracking.c (track_loc_p): Likewise.
10853 * emit-rtl.c (byte_lowpart_offset): Likewise.
10854 (paradoxical_subreg_p): Delete out-of-line definition.
10856 2017-08-22 Jeff Law <law@redhat.com>
10858 PR tree-optimization/81741
10859 PR tree-optimization/71947
10860 * tree-ssa-dom.c: Include tree-inline.h.
10861 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
10862 equivalences if one is more expensive to compute than the other.
10863 * tree-ssa-scopedtables.h (class const_or_copies): Make
10864 record_const_or_copy_raw method private.
10865 (class avail_exprs_stack): New method simplify_binary_operation.
10866 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
10867 avail_exprs_stack::simplify_binary_operation as needed.
10868 (avail_exprs_stack::simplify_binary_operation): New function.
10870 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
10872 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
10873 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
10874 (DOT_SYMBOLS): Likewise.
10875 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
10876 (RELOCATABLE_NEEDS_FIXUP): Likewise.
10877 (RS6000_ABI_NAME): Likewise.
10878 (TARGET_CMODEL): Likewise.
10879 (TOC_SECTION_ASM_OP): Likewise.
10880 (SET_CMODEL): New macro.
10881 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
10883 2017-08-22 Richard Biener <rguenther@suse.de>
10885 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
10886 to allow for free-lang-data replacements similar to verify_type_variant.
10888 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
10889 Alan Hayward <alan.hayward@arm.com>
10890 David Sherwood <david.sherwood@arm.com>
10892 * config/aarch64/aarch64.md (casesi): Use DImode rather than
10893 VOIDmode for the LABEL_REF.
10895 2017-08-22 Richard Biener <rguenther@suse.de>
10897 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
10899 2017-08-22 Richard Biener <rguenther@suse.de>
10901 * common.opt (feliminate-dwarf2-dups): Ignore.
10902 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
10903 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
10904 same_die_p_wrap, compute_section_prefix,
10905 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
10906 (comdat_symbol_id, comdat_symbol_number): Likewise.
10907 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
10909 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
10910 (output_die): Mark unreachable path unreachable.
10911 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
10912 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
10913 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
10914 (dwarf2out_early_finish): Likewise.
10916 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
10918 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
10920 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
10923 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
10924 not VAR_P. Filter attribute warnings with OPT_Wattributes.
10925 (avr_attribute_table) <io, io_low, address>: Initialize
10926 .decl_required with true.
10928 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
10930 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
10931 undocumented debugging options.
10932 (-mvsx-scalar-double): Likewise.
10933 (-mallow-df-permute): Likewise.
10934 (-mvectorize-builtins): Likewise.
10935 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
10936 (rs6000_builtin_vectorized_function): Likewise.
10937 (rs6000_builtin_md_vectorized_function): Likewise.
10938 (rs6000_opt_vars): Likewise.
10940 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
10943 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
10944 (*btrq_imm): Rename from *btrq.
10945 (*btcq_imm): Rename from *btcq.
10946 (btsc): New code attribute.
10947 (*<btsc><mode>): New insn pattern.
10948 (*btr<mode>): Ditto.
10949 (*<btsc><mode>_mask): New insn_and_split pattern.
10950 (*btr<mode>_mask): Ditto.
10952 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10953 Alan Hayward <alan.hayward@arm.com>
10954 David Sherwood <david.sherwood@arm.com>
10956 * function.c (pad_below): Simplify padding calculation.
10958 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10959 Alan Hayward <alan.hayward@arm.com>
10960 David Sherwood <david.sherwood@arm.com>
10962 * target.def (function_prologue): Remove frame size argument.
10963 (function_epilogue): Likewise.
10964 * doc/tm.texi: Regenerate.
10965 * final.c (final_start_function): Update call to function_prologue.
10966 (final_end_function): Update call to function_epilogue.
10967 (default_function_pro_epilogue): Remove frame size argument.
10968 * output.h (default_function_pro_epilogue): Likewise.
10969 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
10970 (arm_output_function_prologue): Likewise.
10971 * config/frv/frv.c (frv_function_prologue): Likewise.
10972 (frv_function_epilogue): Likewise.
10973 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
10974 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
10975 (ia64_output_function_epilogue): Likewise.
10976 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
10977 (m32r_output_function_epilogue): Likewise.
10978 * config/microblaze/microblaze.c (microblaze_function_prologue)
10979 (microblaze_function_epilogue): Likewise.
10980 * config/mips/mips.c (mips_output_function_prologue): Likewise.
10981 (mips_output_function_epilogue): Likewise.
10982 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
10983 (mmix_target_asm_function_epilogue): Likewise.
10984 * config/msp430/msp430.c (msp430_start_function): Likewise.
10985 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
10986 (nds32_asm_function_epilogue): Likewise.
10987 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
10988 * config/pa/pa.c (pa_output_function_prologue): Likewise.
10989 (pa_output_function_epilogue): Likewise.
10990 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
10991 (rs6000_output_function_epilogue): Likewise.
10992 * config/rl78/rl78.c (rl78_start_function): Likewise.
10993 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
10994 (rs6000_output_function_epilogue): Likewise.
10995 * config/rx/rx.c (rx_output_function_prologue): Likewise.
10996 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
10997 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
10998 (sparc_asm_function_epilogue): Likewise.
11000 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11002 * tree.h (type_has_mode_precision_p): New function.
11003 * convert.c (convert_to_integer_1): Use it.
11004 * expr.c (expand_expr_real_2): Likewise.
11005 (expand_expr_real_1): Likewise.
11006 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
11007 * match.pd: Likewise.
11008 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
11009 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
11010 * tree-tailcall.c (process_assignment): Likewise.
11011 * tree-vect-loop.c (vectorizable_reduction): Likewise.
11012 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
11013 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
11014 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11015 (vectorizable_assignment): Likewise.
11016 (vectorizable_shift): Likewise.
11017 (vectorizable_operation): Likewise.
11018 * tree-vrp.c (register_edge_assert_for_2): Likewise.
11020 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
11022 * match.pd: Add pow (C, x) simplification.
11024 2017-08-21 Richard Biener <rguenther@suse.de>
11026 PR tree-optimization/81900
11027 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
11028 for blocks with abnormal predecessors.
11029 (compute_antic): Do not set visited flag prematurely.
11031 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
11034 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
11036 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11038 * stor-layout.h (vector_type_mode): Move to...
11039 * tree.h (vector_type_mode): ...here.
11040 * stor-layout.c (vector_type_mode): Move to...
11041 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
11043 2017-08-21 Richard Biener <rguenther@suse.de>
11045 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
11046 register_external_die hooks.
11047 (debug_false_tree_charstarstar_uhwistar): Declare.
11048 (debug_nothing_tree_charstar_uhwi): Likewise.
11049 * debug.c (do_nothing_debug_hooks): Adjust.
11050 (debug_false_tree_charstarstar_uhwistar): New do nothing.
11051 (debug_nothing_tree_charstar_uhwi): Likewise.
11052 * dbxout.c (dbx_debug_hooks): Adjust.
11053 (xcoff_debug_hooks): Likewise.
11054 * sdbout.c (sdb_debug_hooks): Likewise.
11055 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
11056 * dwarf2out.c (macinfo_label_base): New global.
11057 (dwarf2out_register_external_die): New function for the
11058 register_external_die hook.
11059 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
11060 (dwarf2_debug_hooks): Use them.
11061 (dwarf2_lineno_debug_hooks): Adjust.
11062 (struct die_struct): Add with_offset flag.
11063 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
11064 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
11065 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
11066 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
11067 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
11068 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
11069 defining section names for the early LTO debug variants.
11070 (reset_indirect_string): New helper.
11071 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
11072 (print_dw_val): Add support for offsetted symbol references.
11073 (get_ultimate_context): Split out from is_cxx.
11074 (is_cxx): Use get_ultimate_context.
11075 (is_fortran): Add decl overload.
11076 (compute_comp_unit_symbol): Split out worker from
11077 compute_section_prefix.
11078 (compute_section_prefix): Call compute_comp_unit_symbol and
11079 set comdat_type_p here.
11080 (output_die): Skip DIE symbol output for the LTO added one.
11081 Handle DIE symbol references with offset.
11082 (output_comp_unit): Guard section name mangling properly.
11083 For LTO debug sections emit a symbol at the section beginning
11084 which we use to refer to its DIEs.
11085 (add_abstract_origin_attribute): For DIEs registered via
11086 dwarf2out_register_external_die directly refer to the early
11087 DIE rather than indirectly through the shadow one we created.
11088 Remove obsolete call to dwarf2out_abstract_function for
11089 non-function/block origins.
11090 (gen_array_type_die): When generating early LTO debug do
11091 not emit DW_AT_string_length.
11092 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
11093 late when in LTO. As suggested place a gcc_unreachable for
11094 the DECL_ABSTRACT_P case.
11095 (gen_subprogram_die): Avoid another specification DIE
11096 for early built declarations/definitions for the late LTO case.
11097 (gen_variable_die): Add type references for late duplicated VLA dies
11099 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
11100 we have the abstract instance already.
11101 (process_scope_var): Adjust decl DIE contexts in LTO which
11102 first puts them in limbo.
11103 (gen_decl_die): Do not generate type DIEs late apart from
11104 types for VLAs or for decls we do not yet have a DIE. Do not
11105 call dwarf2out_abstract_function late.
11106 (dwarf2out_early_global_decl): Make sure to create DIEs
11107 for abstract instances of a decl first.
11108 (dwarf2out_late_global_decl): Adjust comment.
11109 (output_macinfo_op): With multiple macro sections use
11110 macinfo_label_base to distinguish labels.
11111 (output_macinfo): Likewise. Update macinfo_label_base.
11112 Pass in the line info label.
11113 (note_variable_value_in_expr): When generating LTO resolve
11114 all variable values here by generating DIEs as needed.
11115 (init_sections_and_labels): Add early LTO debug flag parameter
11116 and generate different sections and names if set. Add generation
11117 counter for the labels so we can have multiple of them.
11118 (reset_dies): Helper to allow DIEs to be output multiple times.
11119 (dwarf2out_finish): When outputting DIEs to the fat part of an
11120 LTO object first reset DIEs.
11121 (dwarf2out_early_finish): Output early DIEs when generating LTO.
11122 (modified_type_die): Check for decl_ultimate_origin being self
11124 (gen_type_die_with_usage): Likewise.
11125 (gen_typedef_die): Allow decl_ultimate_origin being self.
11126 (set_decl_abstract_flags): Remove.
11127 (set_block_abstract_flags): Likewise.
11128 (dwarf2out_abstract_function): Treat the early generated DIEs
11129 as the abstract copy and only add DW_AT_inline and
11130 DW_AT_artificial here and call set_decl_origin_self.
11131 If the DIE has an abstract origin don't do anything.
11132 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
11133 if we have none yet (Go fails to build one, PR78628).
11134 (variably_modified_type_p): Prevent endless recursion for Ada
11135 cyclic pointer types.
11136 * lto-streamer-in.c: Include debug.h.
11137 (dref_queue): New global.
11138 (lto_read_tree_1): Stream in DIE references.
11139 (lto_input_tree): Register DIE references.
11140 (input_function): Stream DECL_DEBUG_ARGS.
11141 * lto-streamer-out.c: Include debug.h.
11142 (lto_write_tree_1): Output DIE references.
11143 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
11144 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
11145 (output_function): Stream DECL_DEBUG_ARGS.
11146 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
11147 Stream DECL_ABSTRACT_ORIGIN.
11148 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
11149 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
11150 DECL_CONTEXT for file-scope decls.
11151 * lto-streamer.h (struct dref_entry): Declare.
11152 (dref_queue): Likewise.
11153 * cfgexpand.c (pass_expand::execute): Do not call the
11154 outlining_inline_function hook here.
11155 * lto-wrapper.c (debug_obj): New global.
11156 (tool_cleanup): Unlink it if required.
11157 (debug_objcopy): New function.
11158 (run_gcc): Handle early debug sections in the IL files by
11159 extracting them to separate files, partially linkin them and
11160 feeding the result back as result to the linker.
11161 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
11162 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
11163 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
11164 sections into a separate segment.
11165 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
11167 (darwin_asm_dwarf_section): Likewise.
11168 (darwin_asm_output_dwarf_offset): Likewise.
11169 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
11171 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11172 Alan Hayward <alan.hayward@arm.com>
11173 David Sherwood <david.sherwood@arm.com>
11175 * read-md.h (md_reader::record_potential_iterator_use): Replace
11176 pointer argument with an rtx and an index.
11177 * read-rtl.c (iterator_group::apply_iterator): Likewise.
11178 (apply_mode_iterator): Likewise.
11179 (apply_code_iterator): Likewise.
11180 (apply_int_iterator): Likewise.
11181 (apply_subst_iterator): Likewise.
11182 (record_iterator_use): Likewise.
11183 (record_attribute_use): Likewise.
11184 (md_reader::record_potential_iterator_use): Likewise. Update calls
11185 to record_iterator_use and apply_iterator.
11186 (iterator_use): Replace ptr with x and index.
11187 (attribute_use): Likewise.
11188 (apply_attribute_uses): Update calls to apply_iterator.
11189 (apply_iterators): Likewise. Update initialization of iterator_use.
11190 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
11191 and record_potential_iterator_use.
11192 (rtx_reader::read_rtx_operand): Likewise.
11194 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11195 Alan Hayward <alan.hayward@arm.com>
11196 David Sherwood <david.sherwood@arm.com>
11198 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
11201 2017-08-21 Richard Biener <rguenther@suse.de>
11203 PR middle-end/81884
11204 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
11205 at struct end conservatively when comparing common bases.
11207 2017-08-21 Richard Biener <rguenther@suse.de>
11209 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
11210 (mem_ref_in_stmt): Remove.
11211 (determine_max_movement): Use ref index to get at the reference.
11212 (invariantness_dom_walker::before_dom_children): Deal with
11213 lim data already initialized.
11214 (gather_mem_refs_stmt): Initialize lim data and record ref index.
11216 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
11218 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
11219 (TARGET_ISA_ROUND): Ditto.
11220 (TARGET_ROUND): Ditto.
11221 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
11222 * config/i386/i386.md: Ditto.
11223 * config/i386/sse.md: Ditto.
11224 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
11225 with OPTION_MASK_ISA_SSE4_1.
11227 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
11230 * doc/extend.texi (x86 Built-in Functions): Correct the name of
11231 __builtin_ia32_lzcnt_u16.
11233 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
11236 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
11237 (rs6000_set_current_function): Rewrite function to use it.
11239 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
11242 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
11243 and TYPE_WARN_IF_NOT_ALIGN.
11244 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
11245 (handle_warn_if_not_align): New.
11246 (place_union_field): Call handle_warn_if_not_align.
11247 (place_field): Call handle_warn_if_not_align.
11248 Copy TYPE_WARN_IF_NOT_ALIGN.
11249 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
11250 (layout_type): Likewise.
11251 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
11253 (tree_decl_common): Add warn_if_not_align.
11254 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
11255 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
11256 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
11257 (DECL_WARN_IF_NOT_ALIGN): Likewise.
11258 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
11259 * doc/extend.texi: Document warn_if_not_aligned attribute.
11260 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
11262 2017-08-17 Martin Liska <mliska@suse.cz>
11265 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
11266 (get_data_dependence): Use it as pointer type.
11267 (distribute_loop): Likewise.
11269 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11271 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
11272 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
11273 (p8_vmrgow_<mode>_direct): New define_insn.
11274 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
11275 handle endianness for vmrgew and vmrgow permute patterns.
11277 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
11279 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
11280 * config/rs6000/rs6000-cpus.def: Remove comment.
11281 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
11282 (POWERPC_MASKS): Likewise.
11283 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
11284 use of TARGET_VSX_TIMODE.
11285 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
11286 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
11287 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
11288 (rs6000_option_override_internal): Remove dead code.
11289 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
11290 (rs6000_legitimize_reload_address): Likewise.
11291 (rs6000_legitimate_address_p): Likewise.
11292 (rs6000_opt_masks): Delete "vsx-timode".
11293 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
11294 from function comment.
11295 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
11296 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
11297 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
11299 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
11300 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
11301 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
11302 (VSX_TI): Likewise.
11304 (define_peephole2): Likewise.
11306 2017-08-17 Martin Sebor <msebor@redhat.com>
11309 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
11310 past the end of an array.
11311 (test_pp_format): Add test cases.
11313 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
11315 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
11316 missing ECF_NOTHROW flags.
11318 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
11321 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
11322 operands residing in integer registers.
11323 (*vsx_le_perm_load_<mode>): Likewise.
11324 (*vsx_le_perm_store_<mode>): Likewise.
11325 (define_peephole2): Add peepholes to optimize the above.
11327 2017-08-17 Marek Polacek <polacek@redhat.com>
11329 PR middle-end/81814
11330 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
11331 to mimic what shorten_compare did. Change the return type to bool.
11332 (fold_cond_expr_with_comparison): Update call to
11333 operand_equal_for_comparison_p.
11334 (fold_ternary_loc): Likewise.
11336 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
11338 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
11340 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
11342 2017-08-17 Richard Biener <rguenther@suse.de>
11344 * tree-ssa-structalias.c (solve_graph): When propagating
11345 to successors update the graphs succ edges and avoid duplicate work.
11347 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
11350 * config/i386/i386.c (ix86_option_override_internal): Save target
11351 specific options after ix86_stack_protector_guard_reg was changed.
11353 2017-08-17 Richard Biener <rguenther@suse.de>
11355 PR tree-optimization/81827
11356 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
11357 (new_var_info): Initialize it conservatively.
11358 (get_call_vi): Mark register vars.
11359 (new_scalar_tmp_constraint_exp): Likewise.
11360 (handle_rhs_call): Likewise.
11361 (handle_const_call): Likewise.
11362 (create_function_info_for): Likewise.
11363 (solve_constraints): Sort varinfos to separate register from
11364 non-register vars to pack points-to solution bitmaps during
11367 2017-08-17 Marek Polacek <polacek@redhat.com>
11369 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
11371 2017-08-17 Richard Biener <rguenther@suse.de>
11373 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
11374 to true when overflow is undefined and we saturated the result.
11376 2017-08-17 Alan Modra <amodra@gmail.com>
11379 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
11380 Don't use store multiple if only one reg needs saving.
11381 (interesting_frame_related_regno): New function.
11382 (rs6000_frame_related): Don't emit frame info for regs that
11384 (rs6000_emit_epilogue): Likewise.
11386 2017-08-16 Nathan Sidwell <nathan@acm.org>
11388 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
11389 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
11390 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
11391 * tree.c (free_lang_data_in_type): Use else-if chain. Always
11392 clear TYPE_LANG_1. Remove obsolete member-function stripping.
11393 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
11394 (verify_type): Adjust for TYPE_BINFO move.
11395 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
11396 process TYPE_BINFO directly.
11397 (hash_tree): Likewise.
11398 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
11400 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
11403 2017-08-16 David Malcolm <dmalcolm@redhat.com>
11405 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
11407 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
11410 * config/i386/i386.md (*anddi_1_btr): Change predicates of
11411 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
11412 Add ix86_binary_operator_ok to insn constraint.
11413 (*iordi_1_bts): Ditto.
11414 (*xordi_1_btc): Ditto.
11415 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
11416 Update corresponding peephole2 pattern.
11420 2017-08-16 Bin Cheng <bin.cheng@arm.com>
11422 PR tree-optimization/81832
11423 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
11424 copy loop header which has IFN_LOOP_DIST_ALIAS call.
11426 2017-08-16 Marek Polacek <polacek@redhat.com>
11429 * fold-const.c (fold_indirect_ref_1): Restore original behavior
11430 regarding size_zero_node.
11432 2017-08-16 Martin Liska <mliska@suse.cz>
11435 * config.gcc: Respect previously set extra_objs in case
11438 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
11440 PR tree-optimization/81835
11441 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
11442 the phi SSA_NAME. Check that the condition in a COND_EXPR does
11443 not depend on the phi.
11445 2017-08-16 Alan Modra <amodra@gmail.com>
11447 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
11450 2017-08-16 Alan Modra <amodra@gmail.com>
11452 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
11453 (save_reg_p): ..into this. Update all callers.
11454 (first_reg_to_save): Simplify.
11456 2017-08-16 Alan Modra <amodra@gmail.com>
11458 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
11461 2017-08-15 Joseph Myers <joseph@codesourcery.com>
11465 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
11466 constant count if that count is less than 32.
11468 2017-08-15 Nathan Sidwell <nathan@acm.org>
11470 * gcc.c (execute): Emit friendlier message if inferior is killed
11471 by an external cause.
11473 2017-08-15 Richard Biener <rguenther@suse.de>
11475 PR tree-optimization/81790
11476 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
11477 CONSTRUCTORs from simplifying and VN.
11479 2017-08-14 Martin Sebor <msebor@redhat.com>
11481 * builtin-attrs.def: Add comments.
11483 2017-08-14 Martin Sebor <msebor@redhat.com>
11486 * doc/extend.texi (attribute nonstring): Document new attribute.
11488 2017-08-14 Martin Sebor <msebor@redhat.com>
11491 * tree-diagnostic.c (default_tree_printer): Handle %G.
11492 * gimple-pretty-print.h (percent_G_format): Declare new function.
11493 * gimple-pretty-print.c (percent_G_format): Define.
11494 * tree-pretty-print.c (percent_K_format): Add argument.
11496 2017-08-14 Martin Sebor <msebor@redhat.com>
11498 PR translation/79998
11499 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
11500 Remove a stray space.
11502 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
11505 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
11506 (*iordi_1_bts): Ditto.
11507 (*xordi_1_btc): Ditto.
11509 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11512 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
11513 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11515 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
11516 quoted strings, and make more translator-friendly.
11517 (darwin_rs6000_override_options): Likewise.
11518 (rs6000_option_override_internal): Likewise.
11519 (rs6000_return_in_memory): Fix overlong line.
11520 (init_cmulative_args): Use quoted strings, and make more
11521 translator-friendly.
11522 (rs6000_pass_by_reference): Fix overlong line.
11523 (def_builtin): Use quoted strings.
11524 (altivec_expand_predicate_builtin): Use quoted strings, and make
11525 more translator-friendly.
11526 (htm_expand_builtin): Use quoted strings.
11527 (cpu_expand_builtin): Use quoted strings, and make more
11528 translator-friendly.
11529 (altivec_expand_builtin): Likewise.
11530 (paired_expand_predicate_builtin): Likewise.
11531 (rs6000_invalid_builtin): Likewise.
11532 (builtin_function_type): Use quoted strings.
11533 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
11534 more translator-friendly.
11535 (rs6000_trampoline_init): Likewise.
11536 (rs6000_handle_altivec_attribute): Likewise.
11537 (rs6000_inner_target_options): Use quoted strings.
11538 (rs6000_disable_incompatible_switches): Likewise.
11539 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
11540 strings, and make more translator-friendly.
11541 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
11543 2017-08-14 Bin Cheng <bin.cheng@arm.com>
11545 PR tree-optimization/81799
11546 * tree-loop-distribution.c (version_loop_by_alias_check): Force
11547 cond_expr to simple gimple operand.
11549 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
11551 PR middle-end/46932
11552 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
11554 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
11558 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
11559 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
11560 TARGET_GASISR_PROLOGUES.
11561 * config/avr/avr.c (avr_option_override): Same.
11562 (avr_pass_pre_proep::execute): Same.
11564 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
11567 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
11568 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
11570 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
11572 * config/i386/i386.md (*load_tp_<mode>): Redefine as
11573 define_insn_and_split. Split to a memory load from 0 in
11574 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
11575 using PTR mode iterator.
11576 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
11577 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
11578 (*add_tp_<mode>): Redefine as define_insn_and_split.
11579 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
11580 address space. Merge with *add_tp_x32 using PTR mode iterator.
11581 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
11582 Split to an add with a memory load from 0 in
11583 DEFAULT_TLS_SEG_REG address space.
11585 2017-08-12 Andrew Pinski <apinski@cavium.com>
11587 * config/aarch64/aarch64-option-extensions.def (rdma):
11588 Fix feature string to what Linux prints out in /proc/cpuinfo.
11590 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
11593 * dwarf2out.c (modified_type_die): For C typedef types that have
11594 an ultimate origin, process the ultimate origin instead of the
11596 (gen_typedef_die): Assert that input DECLs have no ultimate
11598 (gen_type_die_with_usage): For typedef variants that have an
11599 ultimate origin, just call gen_decl_die on the original DECL.
11600 (process_scope_var): Avoid creating DIEs for local typedefs and
11601 concrete static variables.
11603 2017-08-12 Alan Modra <amodra@gmail.com>
11607 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
11608 match gnu-user.h startfile.
11609 (ENDFILE_LINUX_SPEC): Similarly.
11611 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
11614 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
11616 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
11618 2017-08-11 Tamar Christina <tamar.christina@arm.com>
11619 * config/aarch64/aarch64.md (mov<mode>): Change.
11620 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
11621 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
11622 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
11624 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
11626 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
11627 for storage order barriers.
11629 2017-08-11 Martin Liska <mliska@suse.cz>
11632 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
11633 variables of void type.
11635 2017-08-11 Martin Liska <mliska@suse.cz>
11637 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
11638 TARGET_SUPPORTS_ALIASES.
11639 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
11640 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
11641 (optimize_weakref): Likewise.
11642 * symtab.c (symtab_node::noninterposable_alias): Likewise.
11643 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
11644 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
11646 2017-08-11 Martin Liska <mliska@suse.cz>
11649 * config/i386/i386.c (make_resolver_func): Do complete
11650 refactoring of the function.
11652 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
11655 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
11656 * config/i386/i386.c (ix86_stack_protect_guard): Use
11657 ix86_stack_protect_guard_symbol_str to generate varible declaration.
11658 * doc/invoke.texi (x86 Options): Document
11659 -mstack-protector-guard-symbol= option.
11661 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
11663 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
11664 * config/i386/i386.c (ix86_split_stack_guard): New function.
11665 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
11666 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
11667 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
11668 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
11669 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
11670 (split_stack_space_check): Call ix86_split_stack_guard.
11672 2017-08-10 Martin Sebor <msebor@redhat.com>
11674 * print-tree.c (print_node): Print location using the established
11676 Replace spaces with colons.
11677 (debug_raw, debug): Ditto.
11679 2017-08-10 Martin Sebor <msebor@redhat.com>
11682 * pretty-print.c (pp_format): Correct the handling of %s precision.
11684 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
11687 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
11689 (ix86_finalize_stack_frame_flags): This. Also clear
11690 frame_pointer_needed if -fno-omit-frame-pointer is used without
11692 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
11693 with ix86_finalize_stack_frame_flags.
11694 (ix86_expand_epilogue): Likewise.
11695 (ix86_expand_split_stack_prologue): Likewise.
11696 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
11698 2017-08-10 Martin Liska <mliska@suse.cz>
11701 * c-attribs.c (handle_target_attribute):
11702 Report warning for an empty string argument of target attribute.
11704 2017-08-09 Jakub Jelinek <jakub@redhat.com>
11707 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
11709 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
11710 or DECL_NONLOCAL labels.
11711 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
11712 or DECL_NONLOCAL labels here.
11714 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
11716 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
11717 to indicate when early gimple folding has been disabled.
11718 (rs6000_gimple_fold_builtin): Add debug content.
11719 (rs6000_invalid_builtin): Fix whitespace.
11720 (rs6000_expand_builtin): Fix whitespace.
11721 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
11723 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
11726 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
11727 SAVE_MULTIPLE if not all the registers that saves, should be saved.
11729 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
11731 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
11732 (qdf24xx): Likewise.
11733 * config/aarch64/aarch64.md: Include falkor.md.
11734 * config/aarch64/falkor.md: New.
11736 2017-08-09 Marek Polacek <polacek@redhat.com>
11739 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
11740 * diagnostic.c (emit_diagnostic): Add a comment.
11741 (emit_diagnostic_valist): New function.
11743 2017-08-09 Marek Polacek <polacek@redhat.com>
11746 * input.c (make_location): New overload.
11747 * input.h (make_location): Declare.
11749 2017-08-08 Alan Modra <amodra@gmail.com>
11750 H.J. Lu <hongjiu.lu@intel.com>
11753 * gcc.c (NO_PIE_SPEC): Delete.
11754 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
11756 (LINK_PIE_SPEC): ..to here.
11757 (LINK_COMMAND_SPEC): Support -no-pie.
11758 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
11759 chain of crtbegin*.o selection, update for PIE_SPEC changes and
11761 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
11762 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
11763 (ENDFILE_CRTEND_SPEC): Similarly.
11765 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
11768 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
11769 (mstack-protector-guard-offset=): Ditto.
11770 * config/i386/i386.c (ix86_option_override): Handle
11771 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
11773 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
11774 ix86_stack_protect_guard_offset variables.
11775 (TARGET_STACK_PROTECT_GUARD): Always define.
11776 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
11777 and -mstack-protector-guard-offset= options.
11779 2017-08-08 Bin Cheng <bin.cheng@arm.com>
11781 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
11782 boundary case for the last candidate.
11784 2017-08-08 Bin Cheng <bin.cheng@arm.com>
11786 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
11787 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
11789 2017-08-08 Tamar Christina <tamar.christina@arm.com>
11791 PR middle-end/19706
11792 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
11793 * config/aarch64/aarch64-builtins.c
11794 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
11795 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
11796 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
11798 2017-08-08 Tamar Christina <tamar.christina@arm.com>
11799 Andrew Pinski <pinskia@gmail.com>
11801 PR middle-end/19706
11802 * internal-fn.def (XORSIGN): New.
11803 * optabs.def (xorsign_optab): New.
11804 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
11805 (convert_expand_mult_copysign): New.
11806 (pass_optimize_widening_mul::execute): Call
11807 convert_expand_mult_copysign.
11809 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11811 PR tree-optimization/81354
11812 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
11813 Insert on edges rather than explicitly creating landing pads.
11814 (analyze_candidates_and_replace): Commit edge inserts.
11816 2017-08-08 Richard Biener <rguenther@suse.de>
11818 PR middle-end/81719
11819 * tree-ssa-loop-niter.c: Include tree-dfa.h.
11820 (expand_simple_operations): Also look through ADDR_EXPRs with
11821 MEM_REF bases treating them as POINTER_PLUS_EXPR.
11823 2017-08-08 Richard Biener <rguenther@suse.de>
11825 PR tree-optimization/81723
11826 * tree-vect-slp.c (struct bst_traits): New hash traits.
11827 (bst_fail): New global.
11828 (vect_build_slp_tree_2): New worker, split out from ...
11829 (vect_build_slp_tree): ... this now wrapping it with using
11830 bst_fail set to cache SLP tree build fails. Properly handle
11832 (vect_analyze_slp_instance): Allocate and free bst_fail.
11834 2017-08-08 Martin Liska <mliska@suse.cz>
11837 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
11838 LABEL_DECLs that can be from a different function.
11840 2017-08-08 Bin Cheng <bin.cheng@arm.com>
11842 PR tree-optimization/81744
11843 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
11844 loop's number of iterations.
11846 2017-08-08 Martin Liska <mliska@suse.cz>
11848 * asan.c: Include header files.
11849 * attribs.c (build_decl_attribute_variant): New function moved
11851 (build_type_attribute_qual_variant): Likewise.
11852 (cmp_attrib_identifiers): Likewise.
11853 (simple_cst_list_equal): Likewise.
11854 (omp_declare_simd_clauses_equal): Likewise.
11855 (attribute_value_equal): Likewise.
11856 (comp_type_attributes): Likewise.
11857 (build_type_attribute_variant): Likewise.
11858 (lookup_ident_attribute): Likewise.
11859 (remove_attribute): Likewise.
11860 (merge_attributes): Likewise.
11861 (merge_type_attributes): Likewise.
11862 (merge_decl_attributes): Likewise.
11863 (merge_dllimport_decl_attributes): Likewise.
11864 (handle_dll_attribute): Likewise.
11865 (attribute_list_equal): Likewise.
11866 (attribute_list_contained): Likewise.
11867 * attribs.h (lookup_attribute): New function moved from tree.[ch].
11868 (lookup_attribute_by_prefix): Likewise.
11869 * bb-reorder.c: Include header files.
11870 * builtins.c: Likewise.
11871 * calls.c: Likewise.
11872 * cfgexpand.c: Likewise.
11873 * cgraph.c: Likewise.
11874 * cgraphunit.c: Likewise.
11875 * convert.c: Likewise.
11876 * dwarf2out.c: Likewise.
11877 * final.c: Likewise.
11878 * fold-const.c: Likewise.
11879 * function.c: Likewise.
11880 * gimple-expr.c: Likewise.
11881 * gimple-fold.c: Likewise.
11882 * gimple-pretty-print.c: Likewise.
11883 * gimple.c: Likewise.
11884 * gimplify.c: Likewise.
11885 * hsa-common.c: Likewise.
11886 * hsa-gen.c: Likewise.
11887 * internal-fn.c: Likewise.
11888 * ipa-chkp.c: Likewise.
11889 * ipa-cp.c: Likewise.
11890 * ipa-devirt.c: Likewise.
11891 * ipa-fnsummary.c: Likewise.
11892 * ipa-inline.c: Likewise.
11893 * ipa-visibility.c: Likewise.
11895 * lto-cgraph.c: Likewise.
11896 * omp-expand.c: Likewise.
11897 * omp-general.c: Likewise.
11898 * omp-low.c: Likewise.
11899 * omp-offload.c: Likewise.
11900 * omp-simd-clone.c: Likewise.
11901 * opts-global.c: Likewise.
11902 * passes.c: Likewise.
11903 * predict.c: Likewise.
11904 * sancov.c: Likewise.
11905 * sanopt.c: Likewise.
11906 * symtab.c: Likewise.
11907 * toplev.c: Likewise.
11908 * trans-mem.c: Likewise.
11909 * tree-chkp.c: Likewise.
11910 * tree-eh.c: Likewise.
11911 * tree-into-ssa.c: Likewise.
11912 * tree-object-size.c: Likewise.
11913 * tree-parloops.c: Likewise.
11914 * tree-profile.c: Likewise.
11915 * tree-ssa-ccp.c: Likewise.
11916 * tree-ssa-live.c: Likewise.
11917 * tree-ssa-loop.c: Likewise.
11918 * tree-ssa-sccvn.c: Likewise.
11919 * tree-ssa-structalias.c: Likewise.
11920 * tree-ssa.c: Likewise.
11921 * tree-streamer-in.c: Likewise.
11922 * tree-vectorizer.c: Likewise.
11923 * tree-vrp.c: Likewise.
11924 * tsan.c: Likewise.
11925 * ubsan.c: Likewise.
11926 * varasm.c: Likewise.
11927 * varpool.c: Likewise.
11928 * tree.c: Remove functions moved to attribs.[ch].
11929 * tree.h: Likewise.
11930 * config/aarch64/aarch64.c: Add attrs.h header file.
11931 * config/alpha/alpha.c: Likewise.
11932 * config/arc/arc.c: Likewise.
11933 * config/arm/arm.c: Likewise.
11934 * config/avr/avr.c: Likewise.
11935 * config/bfin/bfin.c: Likewise.
11936 * config/c6x/c6x.c: Likewise.
11937 * config/cr16/cr16.c: Likewise.
11938 * config/cris/cris.c: Likewise.
11939 * config/darwin.c: Likewise.
11940 * config/epiphany/epiphany.c: Likewise.
11941 * config/fr30/fr30.c: Likewise.
11942 * config/frv/frv.c: Likewise.
11943 * config/ft32/ft32.c: Likewise.
11944 * config/h8300/h8300.c: Likewise.
11945 * config/i386/winnt.c: Likewise.
11946 * config/ia64/ia64.c: Likewise.
11947 * config/iq2000/iq2000.c: Likewise.
11948 * config/lm32/lm32.c: Likewise.
11949 * config/m32c/m32c.c: Likewise.
11950 * config/m32r/m32r.c: Likewise.
11951 * config/m68k/m68k.c: Likewise.
11952 * config/mcore/mcore.c: Likewise.
11953 * config/microblaze/microblaze.c: Likewise.
11954 * config/mips/mips.c: Likewise.
11955 * config/mmix/mmix.c: Likewise.
11956 * config/mn10300/mn10300.c: Likewise.
11957 * config/moxie/moxie.c: Likewise.
11958 * config/msp430/msp430.c: Likewise.
11959 * config/nds32/nds32-isr.c: Likewise.
11960 * config/nds32/nds32.c: Likewise.
11961 * config/nios2/nios2.c: Likewise.
11962 * config/nvptx/nvptx.c: Likewise.
11963 * config/pa/pa.c: Likewise.
11964 * config/pdp11/pdp11.c: Likewise.
11965 * config/powerpcspe/powerpcspe.c: Likewise.
11966 * config/riscv/riscv.c: Likewise.
11967 * config/rl78/rl78.c: Likewise.
11968 * config/rx/rx.c: Likewise.
11969 * config/s390/s390.c: Likewise.
11970 * config/sh/sh.c: Likewise.
11971 * config/sol2.c: Likewise.
11972 * config/sparc/sparc.c: Likewise.
11973 * config/spu/spu.c: Likewise.
11974 * config/stormy16/stormy16.c: Likewise.
11975 * config/tilegx/tilegx.c: Likewise.
11976 * config/tilepro/tilepro.c: Likewise.
11977 * config/v850/v850.c: Likewise.
11978 * config/vax/vax.c: Likewise.
11979 * config/visium/visium.c: Likewise.
11980 * config/xtensa/xtensa.c: Likewise.
11982 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
11985 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
11986 constraints since the -mupper-regs-* switches have been
11988 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
11989 into a vector from a double word element that was extracted from
11990 another vector, and eliminate extra XXPERMDI instructions.
11991 (vsx_concat_<mode>_2): Likewise.
11992 (vsx_concat_<mode>_3): Likewise.
11993 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
11994 concat to allow optimizing inserts from previous extracts.
11996 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
11998 * config/i386/i386.c (ix86_stack_protect_guard): Generate
11999 memory reference to a SSP offset in TLS address space.
12000 (ix86_print_operand) <case '@'>: Remove.
12001 (ix86_print_operand_punct_valid_p): Remove '@' code.
12002 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
12003 UNSPEC_SP_TLS_TEST.
12004 (stack_tls_protect_set_<mode>): Remove.
12005 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
12006 (stack_tls_protect_test_<mode>): Remove.
12007 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
12009 2017-08-07 Olivier Hainque <hainque@adacore.com>
12012 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
12014 2017-08-07 Douglas Rupp <rupp@adacore.com>
12016 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
12017 variable was referenced as multidir in command.
12019 2017-08-07 Jakub Jelinek <jakub@redhat.com>
12022 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
12025 2017-08-07 Martin Liska <mliska@suse.cz>
12027 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
12028 * config/rl78/rl78.c: Add include of attribs.h.
12029 * config/sh/sh.c: Likewise.
12030 * config/v850/v850.c: Likewise.
12032 2017-08-07 Tom de Vries <tom@codesourcery.com>
12034 PR middle-end/78266
12035 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
12037 2017-08-07 Martin Liska <mliska@suse.cz>
12039 * config/mips/mips.c: Include attribs.h.
12041 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
12044 * doc/invoke.texi: Document change in behvaior for -Ofast for
12047 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
12049 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
12051 (aarch64_pop_regs): Likewise.
12052 (aarch64_gen_load_pair): Likewise.
12053 (aarch64_save_callee_saves): Likewise.
12054 (aarch64_restore_callee_saves): Likewise.
12056 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
12058 * config/i386/i386.c: Revert the last change.
12060 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
12063 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
12065 (ix86_finalize_stack_frame_flags): This. Also clear
12066 frame_pointer_needed if -fno-omit-frame-pointer is used without
12068 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
12069 with ix86_finalize_stack_frame_flags.
12070 (ix86_expand_epilogue): Likewise.
12071 (ix86_expand_split_stack_prologue): Likewise.
12073 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
12076 * config/i386/i386.c (get_builtin_code_for_version): Set priority
12077 to P_AES for Westmere.
12079 2017-08-07 Jonathan Yong <10walls@gmail.com>
12081 * config/i386/mingw.opt (fset-stack-executable): Removed.
12082 * config/i386/cygming.opt (fset-stack-executable): Moved
12084 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
12086 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
12088 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
12090 2017-08-07 Marek Polacek <polacek@redhat.com>
12092 PR middle-end/81737
12093 * fold-const.c (fold_indirect_ref_1): Check type_domain.
12095 2017-08-07 Martin Liska <mliska@suse.cz>
12097 * attribs.h (canonicalize_attr_name): New function.
12098 (cmp_attribs): Move from c-format.c and adjusted.
12099 (is_attribute_p): Moved from tree.h.
12100 * tree-inline.c: Add new includes.
12101 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
12102 (private_is_attribute_p): Remove.
12103 (private_lookup_attribute): Likewise.
12104 (private_lookup_attribute_by_prefix): Simplify.
12105 (remove_attribute): Use is_attribute_p.
12106 * tree.h: Remove removed declarations.
12108 2017-08-07 Jakub Jelinek <jakub@redhat.com>
12110 PR middle-end/81698
12111 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
12112 instead of computing it in the function. Formatting fix.
12113 (expand_case): Don't rely on default_edge being the first edge,
12114 clear it if removing it, pass default_edge to
12115 emit_case_dispatch_table.
12116 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
12119 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
12121 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
12122 insn in the function, emit NOP after the insn.
12124 2017-08-06 Tom de Vries <tom@codesourcery.com>
12126 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
12129 2017-08-06 Tom de Vries <tom@codesourcery.com>
12131 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
12134 2017-08-04 Yury Gribov <tetra2005@gmail.com>
12136 PR tree-optimization/57371
12137 * match.pd: New pattern.
12139 2017-08-04 Marek Polacek <polacek@redhat.com>
12141 PR middle-end/81695
12142 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
12143 perform the computation in offset_int.
12145 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
12147 PR tree-optimization/81136
12148 * tree-vectorizer.h: Include tree-hash-traits.h.
12149 (vec_base_alignments): New typedef.
12150 (vec_info): Add a base_alignments field.
12151 (vect_record_base_alignments): Declare.
12152 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
12154 (DR_IS_CONDITIONAL_IN_STMT): New macro.
12155 (create_data_ref): Add an is_conditional_in_stmt argument.
12156 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
12157 the is_conditional_in_stmt field.
12158 (data_ref_loc): Add an is_conditional_in_stmt field.
12159 (get_references_in_stmt): Set the is_conditional_in_stmt field.
12160 (find_data_references_in_stmt): Update call to create_data_ref.
12161 (graphite_find_data_references_in_stmt): Likewise.
12162 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
12163 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
12164 (vect_record_base_alignment): New function.
12165 (vect_record_base_alignments): Likewise.
12166 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
12167 for nested statements even if we fail to compute a misalignment.
12168 Use pooled base alignments for unconditional references.
12169 (vect_find_same_alignment_drs): Compare base addresses instead
12171 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
12172 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
12174 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
12176 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
12177 Add an explicit name for the enum. Use auto_vec for slp_instances
12178 and grouped_stores.
12179 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
12181 (_bb_vec_info): Add a constructor and destructor.
12182 (vinfo_for_stmt): Return NULL for uids of -1 as well.
12183 (destroy_loop_vec_info): Delete.
12184 (vect_destroy_datarefs): Likewise.
12185 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
12186 (vec_info::vec_info): New function.
12187 (vec_info::~vec_info): Likewise.
12188 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
12189 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
12190 destroy_loop_vec_info.
12191 * tree-vect-loop.c (new_loop_vec_info): Replace with...
12192 (_loop_vec_info::_loop_vec_info): ...this.
12193 (destroy_loop_vec_info): Replace with...
12194 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
12195 the stmt_vec_infos. Leave handling of vec_info information to its
12196 destructor. Remove explicit vector releases.
12197 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
12198 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
12199 * tree-vect-slp.c (new_bb_vec_info): Replace with...
12200 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
12201 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
12202 (destroy_bb_vec_info): Replace with...
12203 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
12204 information to its destructor.
12205 (vect_slp_analyze_bb_1): Use new and delete instead of
12206 new_bb_vec_info and destroy_bb_vec_info.
12207 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
12210 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
12212 * tree-data-ref.h (subscript): Add access_fn field.
12213 (data_dependence_relation): Add could_be_independent_p.
12214 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
12215 (same_access_functions): Move to tree-data-ref.c.
12216 * tree-data-ref.c (ref_contains_union_access_p): New function.
12217 (access_fn_component_p): Likewise.
12218 (access_fn_components_comparable_p): Likewise.
12219 (dr_analyze_indices): Add a reference to access_fn_component_p.
12220 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
12222 (constant_access_functions): Likewise.
12223 (add_other_self_distances): Likewise.
12224 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
12225 (initialize_data_dependence_relation): Use XCNEW and remove
12226 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
12227 of access functions that have the same type. Allow the
12228 subsequence to end with different bases in some circumstances.
12229 Record the chosen access functions in SUB_ACCESS_FN.
12230 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
12231 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
12232 (subscript_dependence_tester_1): Likewise dra and drb.
12233 (build_classic_dist_vector): Update calls accordingly.
12234 (subscript_dependence_tester): Likewise.
12235 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
12236 DDR_COULD_BE_INDEPENDENT_P.
12237 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
12238 comp_alias_ddrs instead of may_alias_ddrs.
12239 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
12241 (vect_analyze_data_ref_dependence): Use it if
12242 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
12243 distance vectors if that fails.
12244 (dependence_distance_ge_vf): New function.
12245 (vect_prune_runtime_alias_test_list): Use it. Don't clear
12246 LOOP_VINFO_MAY_ALIAS_DDRS.
12248 2017-08-04 Richard Biener <rguenther@suse.de>
12250 PR middle-end/81705
12251 * fold-const.c (fold_binary_loc): Properly restrict
12252 minus_var0 && minus_var1 case when associating undefined overflow
12255 2017-08-04 Tom de Vries <tom@codesourcery.com>
12257 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
12259 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12261 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12262 Don't start diagnostic messages with a capital letter.
12263 * config/rs6000/rs6000.c (rs6000_option_override_internal):
12265 (rs6000_invalid_builtin): Likewise.
12266 (rs6000_trampoline_init): Likewise.
12268 2017-08-03 Jakub Jelinek <jakub@redhat.com>
12271 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
12272 after setting changeable df flags.
12274 2017-08-03 Richard Biener <rguenther@suse.de>
12276 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
12277 up if the use is in USE - X.
12279 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
12281 * toplev.c (dumpfile.h): New include.
12282 (internal_error_reentered): New static function. Use it...
12283 (internal_error_function): ...here to handle reentered internal_error.
12285 2017-08-03 Richard Biener <rguenther@suse.de>
12287 PR middle-end/81148
12288 * fold-const.c (split_tree): Add minus_var and minus_con
12289 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
12290 here but always use minus_*.
12291 (associate_trees): Assert we never associate with MINUS_EXPR
12292 and NULL first operand. Do not recurse for PLUS_EXPR operands
12293 when associating as MINUS_EXPR either.
12294 (fold_binary_loc): Track minus_var and minus_con.
12296 2017-08-03 Tom de Vries <tom@codesourcery.com>
12299 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
12300 ACCEL_COMPILER, apply finish_options on
12301 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
12303 2017-08-03 Tom de Vries <tom@codesourcery.com>
12306 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
12307 function_entry_patch_area_size > 0.
12309 2017-08-03 Jakub Jelinek <jakub@redhat.com>
12312 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
12313 instead of 10??LU, perform unit multiplication in wide_int,
12314 don't change alloc_object_size_limit if the limit is larger
12317 PR tree-optimization/81655
12318 PR tree-optimization/81588
12319 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
12320 the case when ranges[i].low and high are 1 for unsigned type with
12323 PR middle-end/81052
12324 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
12325 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
12327 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12329 * tree-vrp.h: Add include guard.
12331 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
12334 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
12335 (ud2): New insn pattern.
12336 * config/i386/i386.c (ix86_expand_epilogue):
12337 For naked functions, generate ud2 instead of trap insn.
12339 2017-08-02 Marek Polacek <polacek@redhat.com>
12342 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
12344 2017-08-02 Tom de Vries <tom@codesourcery.com>
12345 Cesar Philippidis <cesar@codesourcery.com>
12347 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
12348 Add missing edge probabilities.
12350 2017-08-02 Tamar Christina <tamar.christina@arm.com>
12352 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
12353 Correct endianness.
12355 2017-08-02 Jakub Jelinek <jakub@redhat.com>
12357 PR middle-end/79499
12358 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
12359 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
12360 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
12361 prologue_seq sequences - if any.
12363 2017-08-02 Richard Biener <rguenther@suse.de>
12365 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
12366 via vectors if supported, integer extracts via punning if supported
12367 or otherwise vector extracts.
12369 2017-08-02 Richard Biener <rguenther@suse.de>
12371 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
12373 (bitmap_insert_into_set): ... this.
12375 2017-08-02 Richard Biener <rguenther@suse.de>
12377 PR tree-optimization/81633
12379 2015-08-17 Alan Hayward <alan.hayward@arm.com>
12381 PR tree-optimization/71752
12382 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
12384 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
12386 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
12387 (machine_function::call_ms2sysv_pad_out): Remove field.
12388 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
12389 (ix86_compute_frame_layout): Likewise.
12391 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
12394 * config/i386/i386.c (ix86_set_func_type): Disallow naked
12395 attribute with interrupt attribute.
12397 2017-08-01 Andrew Pinski <apinski@cavium.com>
12399 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
12400 BIT_INSERT_EXPR's operand 1
12401 to see if the types precision matches.
12403 2017-08-01 Martin Liska <mliska@suse.cz>
12405 PR middle-end/70140
12406 * builtins.c (expand_builtin_memcpy_args): Remove.
12407 (expand_builtin_memcpy): Call newly added function
12408 expand_builtin_memory_copy_args.
12409 (expand_builtin_memcpy_with_bounds): Likewise.
12410 (expand_builtin_mempcpy): Remove last argument.
12411 (expand_builtin_mempcpy_with_bounds): Likewise.
12412 (expand_builtin_memory_copy_args): New function created from
12413 expand_builtin_mempcpy_args with small modifications.
12414 (expand_builtin_mempcpy_args): Remove.
12415 (expand_builtin_stpcpy): Remove unused argument.
12416 (expand_builtin): Likewise.
12417 (expand_builtin_with_bounds): Likewise.
12419 2017-08-01 Martin Liska <mliska@suse.cz>
12422 Make mempcpy more optimal (PR middle-end/70140).
12424 2017-08-01 Jakub Jelinek <jakub@redhat.com>
12427 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
12428 __builtin_vec_cmpne verify both arguments are compatible vectors
12429 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
12430 verify arg1_type is a pointer or array type. For __builtin_vec_st,
12431 move computation of aligned to after checking the argument types.
12435 * config/rs6000/vsx.md (vextract_fp_from_shorth,
12436 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
12439 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
12440 Doug Rupp <rupp@adacore.com>
12441 Olivier Hainque <hainque@adacore.com>
12443 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
12444 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
12446 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
12447 for TARGET_OS_CPP_BUILTIN.
12448 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
12449 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
12451 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
12452 passing required abi options to the assembler for EABI configurations.
12453 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
12454 of .text.hot and .text.unlikely sections for kernel modules when
12455 using ARM style exceptions.
12456 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
12457 options. Add EXTRA_CC1_SPEC.
12458 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
12460 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
12462 (ARM_TARGET2_DWARF_FORMAT): Define.
12463 * config/arm/t-vxworks: Adjust multilib control to removal of the
12464 Diab command line options.
12466 2017-08-01 Martin Liska <mliska@suse.cz>
12468 PR gcov-profile/81561
12469 * gcov.c (unblock): Make unblocking safe as we need to preserve
12470 index correspondence of blocks and block_lists.
12472 2017-08-01 Richard Biener <rguenther@suse.de>
12474 PR tree-optimization/81181
12475 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
12476 (compute_antic): ... end of iteration here.
12478 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
12480 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
12481 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
12482 (ftree-slp-vectorize): Likewise.
12483 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
12484 can no longer be set independent of flag_tree_loop_vectorize.
12485 * omp-general.c (emp_max_vf): Likewise.
12486 * opts.c (enable_fdo_optimizations): Remove references to
12487 flag_tree_vectorize, these are now implicit.
12488 (common_handle_option): Remove handling for OPT_ftree_vectorize,
12489 and leave it for the options machinery.
12491 2017-08-01 Martin Liska <mliska@suse.cz>
12493 PR middle-end/70140
12494 * builtins.c (expand_builtin_memcpy_args): Remove.
12495 (expand_builtin_memcpy): Call newly added function
12496 expand_builtin_memory_copy_args.
12497 (expand_builtin_memcpy_with_bounds): Likewise.
12498 (expand_builtin_mempcpy): Remove last argument.
12499 (expand_builtin_mempcpy_with_bounds): Likewise.
12500 (expand_builtin_memory_copy_args): New function created from
12501 expand_builtin_mempcpy_args with small modifications.
12502 (expand_builtin_mempcpy_args): Remove.
12503 (expand_builtin_stpcpy): Remove unused argument.
12504 (expand_builtin): Likewise.
12505 (expand_builtin_with_bounds): Likewise.
12507 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
12510 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
12511 print "ds:" only for immediates in generic address space.
12513 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
12516 * config/i386/i386.c (ix86_funciton_naked): New prototype.
12517 (ix86_function_ok_for_sibcall): Return false for naked functions.
12519 2017-08-01 Richard Biener <rguenther@suse.de>
12521 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
12522 (compute_antic): Seed worklist with exit block predecessors.
12523 * cfganal.c (dfs_find_deadend): For a cycle return the source
12524 of the edge closing it.
12526 2017-08-01 Tamar Christina <tamar.christina@arm.com>
12528 * config/aarch64/aarch64.c
12529 (aarch64_can_const_movi_rtx_p): Move 0 check.
12531 2017-08-01 Bin Cheng <bin.cheng@arm.com>
12533 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
12534 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
12536 * match.pd: Ditto in address comparison pattern.
12538 2017-08-01 Bin Cheng <bin.cheng@arm.com>
12540 PR tree-optimization/81627
12541 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
12542 closed ssa form for store-store chain.
12544 2017-08-01 Bin Cheng <bin.cheng@arm.com>
12546 PR tree-optimization/81620
12547 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
12548 for store-store chain.
12550 2017-08-01 Jakub Jelinek <jakub@redhat.com>
12552 PR tree-optimization/81588
12553 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
12554 ranges[i].in_p, invert comparison code ccode. For >/>=,
12555 swap rhs1 and rhs2 and comparison code unconditionally,
12556 for </<= don't do that. Don't swap rhs1/rhs2 again if
12557 ranges[i].in_p, instead invert comparison code ccode if
12558 opcode or oe->rank is BIT_IOR_EXPR.
12561 * optabs.def (vec_extract_optab, vec_init_optab): Change from
12562 a direct optab to conversion optab.
12563 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
12564 with GET_MODE_INNER as last argument instead of optab_handler.
12565 * expmed.c (extract_bit_field_1): Likewise. Use vector from
12566 vector extraction if possible and optab is available.
12567 * expr.c (store_constructor): Use convert_optab_handler instead
12568 of optab_handler. Use vector initialization from smaller
12569 vectors if possible and optab is available.
12570 * tree-vect-stmts.c (vectorizable_load): Likewise.
12571 * doc/md.texi (vec_extract, vec_init): Document that the optabs
12572 now have two modes.
12573 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
12574 of vec_init from half-sized vectors with the same element mode.
12575 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
12576 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
12577 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
12578 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
12579 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
12580 after mode in gen_vec_extract* calls.
12581 (vec_extract<mode>): Renamed to ...
12582 (vec_extract<mode><ssescalarmodelower>): ... this.
12583 (vec_extract<mode><ssehalfvecmodelower>): New expander.
12584 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
12585 element mode after mode in gen_vec_init* calls.
12586 (VEC_INIT_HALF_MODE): New mode iterator.
12587 (vec_init<mode>): Renamed to ...
12588 (vec_init<mode><ssescalarmodelower>): ... this.
12589 (vec_init<mode><ssehalfvecmodelower>): New expander.
12590 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
12591 (vec_extractv2sfsf): ... this.
12592 (vec_initv2sf): Renamed to ...
12593 (vec_initv2sfsf): ... this.
12594 (vec_extractv2si): Renamed to ...
12595 (vec_extractv2sisi): ... this.
12596 (vec_initv2si): Renamed to ...
12597 (vec_initv2sisi): ... this.
12598 (vec_extractv4hi): Renamed to ...
12599 (vec_extractv4hihi): ... this.
12600 (vec_initv4hi): Renamed to ...
12601 (vec_initv4hihi): ... this.
12602 (vec_extractv8qi): Renamed to ...
12603 (vec_extractv8qiqi): ... this.
12604 (vec_initv8qi): Renamed to ...
12605 (vec_initv8qiqi): ... this.
12606 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
12607 (vec_init<mode>): Renamed to ...
12608 (vec_init<mode><VEC_base_l>): ... this.
12609 (vec_extract<mode>): Renamed to ...
12610 (vec_extract<mode><VEC_base_l>): ... this.
12611 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
12612 (vec_initv2sfsf): ... this.
12613 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
12614 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
12615 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
12616 element mode after mode in gen_vec_init* calls.
12617 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
12618 (vec_init<mode><Vel>): ... this.
12619 (vec_extract<mode>): Renamed to ...
12620 (vec_extract<mode><Vel>): ... this.
12621 * config/aarch64/iterators.md (Vel): New mode attribute.
12622 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
12623 Add element mode after mode in gen_vec_extract* calls.
12624 * config/s390/vector.md (non_vec_l): New mode attribute.
12625 (vec_extract<mode>): Renamed to ...
12626 (vec_extract<mode><non_vec_l>): ... this.
12627 (vec_init<mode>): Renamed to ...
12628 (vec_init<mode><non_vec_l>): ... this.
12629 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
12630 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
12632 * config/arm/iterators.md (V_elem_l): New mode attribute.
12633 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
12634 (vec_extract<mode><V_elem_l>): ... this.
12635 (vec_extractv2di): Renamed to ...
12636 (vec_extractv2didi): ... this.
12637 (vec_init<mode>): Renamed to ...
12638 (vec_init<mode><V_elem_l>): ... this.
12639 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
12640 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
12641 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
12642 Add element mode after gen_vec_extract* calls.
12643 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
12644 (vec_init<mode><unitmode>): ... this.
12645 (vec_extract<mode>): Renamed to ...
12646 (vec_extract<mode><unitmode>): ... this.
12647 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
12648 (vec_init<mode><unitmode>): ... this.
12649 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
12650 (vec_initv2sfsf): ... this.
12651 (vec_extractv2sf): Renamed to ...
12652 (vec_extractv2sfsf): ... this.
12653 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
12654 Add element mode after gen_vec_extract* calls.
12655 * config/mips/mips.md (unitmode): New mode iterator.
12656 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
12657 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
12658 * config/spu/spu.md (inner_l): New mode attribute.
12659 (vec_init<mode>): Renamed to ...
12660 (vec_init<mode><inner_l>): ... this.
12661 (vec_extract<mode>): Renamed to ...
12662 (vec_extract<mode><inner_l>): ... this.
12663 * config/sparc/sparc.md (veltmode): New mode iterator.
12664 (vec_init<VMALL:mode>): Renamed to ...
12665 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
12666 * config/ia64/vect.md (vec_initv2si): Renamed to ...
12667 (vec_initv2sisi): ... this.
12668 (vec_initv2sf): Renamed to ...
12669 (vec_initv2sfsf): ... this.
12670 (vec_extractv2sf): Renamed to ...
12671 (vec_extractv2sfsf): ... this.
12672 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
12673 (vec_init<mode>): Renamed to ...
12674 (vec_init<mode><VEC_base_l>): ... this.
12675 (vec_extract<mode>): Renamed to ...
12676 (vec_extract<mode><VEC_base_l>): ... this.
12677 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
12678 (vec_initv2sfsf): ... this.
12679 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
12680 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
12681 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
12682 gen_vec_init* calls.
12684 2017-08-01 Richard Biener <rguenther@suse.de>
12686 PR tree-optimization/81297
12687 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
12688 TREE_OVERFLOW from INTEGER_CSTs.
12690 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
12692 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
12694 2017-07-31 Carl Love <cel@us.ibm.com>
12696 * config/rs6000/rs6000-c: Add support for built-in functions
12697 vector signed char vec_xl_be (signed long long, signed char *);
12698 vector unsigned char vec_xl_be (signed long long, unsigned char *);
12699 vector signed int vec_xl_be (signed long long, signed int *);
12700 vector unsigned int vec_xl_be (signed long long, unsigned int *);
12701 vector signed long long vec_xl_be (signed long long, signed long long *);
12702 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
12703 vector signed short vec_xl_be (signed long long, signed short *);
12704 vector unsigned short vec_xl_be (signed long long, unsigned short *);
12705 vector double vec_xl_be (signed long long, double *);
12706 vector float vec_xl_be (signed long long, float *);
12707 * config/rs6000/altivec.h (vec_xl_be): Add #define.
12708 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
12709 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
12711 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
12712 (altivec_expand_builtin): Add switch statement to call
12713 altivec_expand_xl_be for each builtin.
12714 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
12715 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
12716 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
12717 __builtin_vsx_le_be_v16qi.
12718 * doc/extend.texi: Update the built-in documentation file for the
12719 new built-in functions.
12721 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
12724 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
12726 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
12728 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12730 * config.gcc: Add z14.
12731 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
12732 CPU model numbers for z13s and z14.
12733 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
12735 * config/s390/s390-opts.h (enum processor_type): Rename
12736 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
12737 * config/s390/s390.c (processor_table): Add field for CPU name to
12738 be passed to Binutils.
12739 (s390_asm_output_machine_for_arch): Use the new field in
12740 processor_table for Binutils.
12741 (s390_expand_builtin): Replace arch12 with z14.
12742 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
12743 (s390_get_sched_attrmask): Likewise.
12744 (s390_get_unit_mask): Likewise.
12745 * config/s390/s390.opt: Add z14 to processor_type enum.
12747 2017-07-31 Martin Jambor <mjambor@suse.cz>
12750 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
12751 regardless of optimization level.
12753 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
12754 Martin Liska <mliska@suse.cz>
12756 * predict.def: Remove old comment and adjust probability.
12757 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
12758 PREDICT statements.
12760 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
12763 * config/i386/i386.c (ix86_function_naked): New function.
12764 (ix86_can_use_return_insn_p): Return false for naked functions.
12765 (ix86_expand_prologue): Skip prologue for naked functions.
12766 (ix86_expand_epilogue): Skip epilogue for naked functions
12767 and emit trap instruction.
12768 (ix86_warn_func_return): New function.
12769 (ix86_attribute_table): Add "naked" attribute specification.
12770 (TARGET_WARN_FUNC_RETURN): Define.
12771 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
12773 2017-07-31 Martin Liska <mliska@suse.cz>
12775 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
12776 (dump_gimple_bb_header): Always dump BB info.
12777 (pp_cfg_jump): Do not append info about BB when dumping a jump.
12779 2017-07-31 Martin Liska <mliska@suse.cz>
12782 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
12783 also with current_function_decl non-null equality.
12785 2017-07-31 Jakub Jelinek <jakub@redhat.com>
12788 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
12789 change type to the element type, instead add eltype variable and
12790 use it where we are interested in the element type.
12792 PR tree-optimization/81603
12793 * ipa-polymorphic-call.c
12794 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
12795 offset arithmetic in offset_int, bail out if the resulting bit offset
12796 doesn't fit into shwi.
12798 2017-07-31 Martin Liska <mliska@suse.cz>
12800 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
12801 (gimplify_save_expr): Fix comment.
12803 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
12806 * config/i386/i386.c (ix86_function_arg): Update arguments for
12808 (ix86_compute_frame_layout): Set the initial stack offset to
12809 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
12810 INCOMING_FRAME_SP_OFFSET.
12811 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
12812 stack before exception handler returns.
12813 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
12814 the 'ERROR_CODE' for exception handler.
12816 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
12818 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
12819 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
12820 (ASM_OUTPUT_REG_POP): Ditto.
12821 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
12822 instead of asm_fprintf to output pure string.
12824 2017-07-29 Jakub Jelinek <jakub@redhat.com>
12826 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
12827 to imported_module_or_decl hook.
12828 (debug_nothing_tree_tree_tree_bool): Remove.
12829 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
12830 * debug.c (do_nothing_debug_hooks): Use
12831 debug_nothing_tree_tree_tree_bool_bool instead of
12832 debug_nothing_tree_tree_tree_bool.
12833 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
12834 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
12835 * sdbout.c (sdb_debug_hooks): Likewise.
12836 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
12837 (gen_namespace_die): Add DW_AT_export_symbols attribute if
12839 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
12840 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
12841 attribute, don't add anything.
12843 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12845 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
12846 (fold_build2_stat_loc): Likewise.
12847 (fold_build3_stat_loc): Likewise.
12848 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
12849 (fold_build1_loc): Remove macro.
12850 (fold_build2_loc): Likewise.
12851 (fold_build3_loc): Likewise.
12853 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12855 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
12856 (gimple_build_debug_bind_source_stat): Likewise.
12857 * gimple.h (gimple_build_debug_bind): Remove macro.
12858 (gimple_build_debug_bind_source): Likewise.
12860 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12862 * bitmap.c (bitmap_alloc): Adjust.
12863 (bitmap_gc_alloc): Likewise.
12864 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
12866 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12868 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
12869 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
12870 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
12871 (bitmap_gc_alloc_stat): Likewise.
12872 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
12874 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12876 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
12877 * rtl.h (shallow_copy_rtx): Remove macro.
12879 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12881 * emit-rtl.c (gen_raw_REG): Adjust.
12882 * gengenrtl.c (gendef): Likewise.
12883 * rtl.c (rtx_alloc_stat): Remove _stat from name.
12884 * rtl.h (rtx_alloc): Remove macro.
12886 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12888 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
12889 (build_tree_list_stat): Likewise.
12890 * tree.h (build_tree_list): Remove macro.
12891 (build_tree_list_vec): Likewise.
12893 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12895 * tree.c (make_vector_stat): Remove _stat from name.
12896 (build_vector_stat): Likewise.
12897 * tree.h (make_vector_stat): Remove macro.
12898 (build_vector_stat): Likewise.
12900 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12902 * tree.h (build_var_debug_value): Remove prototype.
12904 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12906 * tree.c (tree_cons_stat): Remove _stat from name.
12907 * tree.h (tree_cons): Remove macro.
12909 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12911 * tree.c (build_vl_exp_stat): Remove _stat from name.
12912 * tree.h (build_vl_exp): Remove macro.
12914 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12916 * tree.c (build_decl_stat): Remove _stat from name.
12917 * tree.h (build_decl): Remove macro.
12919 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12921 * gimple.c (gimple_build_with_ops_stat): Adjust.
12922 (gimple_alloc_stat): Remove _stat from name.
12923 * gimple.h (gimple_alloc): Remove macro.
12925 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12927 * tree.c (make_tree_vec_stat): Remove _stat from name.
12928 (grow_tree_vec_stat): Likewise.
12929 * tree.h (make_tree_vec_stat): Adjust prototype.
12930 (grow_tree_vec_stat): Likewise.
12931 (make_tree_vec): Remove macro.
12932 (grow_tree_vec): Likewise.
12934 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12936 * fold-const.c (fold_build1_stat_loc): Adjust.
12937 (fold_build2_stat_loc): Likewise.
12938 (fold_build3_stat_loc): Likewise.
12939 * tree.c (build0_stat): Remove _stat from name.
12940 (build1_stat): Likewise.
12941 (build2_stat): Likewise.
12942 (build3_stat): Likewise.
12943 (build4_stat): Likewise.
12944 (build5_stat): Likewise.
12945 * tree.h (build1_loc): Remove macro, and rename _stat function
12947 (build2_loc): Likewise.
12948 (build3_loc): Likewise.
12949 (build4_loc): Likewise.
12950 (build5_loc): Likewise.
12952 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12954 * tree.c (make_int_cst_stat): Remove _stat from name.
12955 * tree.h (make_int_cst_stat): Adjust prototype.
12956 (make_int_cst): Remove macro.
12958 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12960 * tree.c (make_tre_binfo_stat): Remove _stat from name.
12961 * tree.h (make_tree_binfo_stat): Adjust prototype.
12962 (make_tree_binfo): Remove.
12964 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12966 * tree.c (copy_node_stat): Rename to copy_node.
12967 (build_distinct_type_copy): Adjust.
12968 * tree.h (copy_node_stat): Adjust prototype.
12969 (copy_node): Remove macro.
12971 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12973 * tree.c (make_node_stat): rename to make_node.
12974 (build_tree_list_stat): Adjust.
12975 (build0_stat): Likewise.
12976 (build2_stat): Likewise.
12977 (build3_stat): Likewise.
12978 (build4_stat): Likewise.
12979 (build5_stat): Likewise.
12980 (build_decl_stat): Likewise.
12981 * tree.h (make_node_stat): Adjust prototype.
12982 (make_node): remove macro.
12984 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
12986 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
12987 (PPC_FEATURE2_SCV): Likewise.
12988 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
12990 2017-07-28 Tamar Christina <tamar.christina@arm.com>
12992 * config/aarch64/aarch64.c
12993 (aarch64_internal_mov_immediate): Add new special pattern.
12994 * config/aarch64/aarch64.md (*movdi_aarch64):
12995 Add reg/32bit const mov case.
12997 2017-07-28 Tamar Christina <tamar.christina@arm.com>
12998 Richard Sandiford <richard.sandiford@linaro.org>
13000 * config/aarch64/aarch64.md (mov<mode>): Generalize.
13001 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
13002 Add integer and movi cases.
13003 (movi-split-hf-df-sf split, fp16): New.
13004 (enabled): Added TARGET_FP_F16INST.
13005 * config/aarch64/iterators.md (GPF_HF): New.
13006 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
13008 2017-07-28 Tamar Christina <tamar.christina@arm.com>
13010 * config/aarch64/aarch64.c
13011 (aarch64_simd_container_mode): Add prototype.
13012 (aarch64_expand_mov_immediate): Add HI support.
13013 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
13014 (aarch64_can_const_movi_rtx_p): New.
13015 (aarch64_preferred_reload_class):
13016 Remove restrictions of using FP registers for certain SIMD operations.
13017 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
13018 (aarch64_valid_floating_const): Add integer move validation.
13019 (aarch64_simd_imm_scalar_p): Remove.
13020 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
13021 (aarch64_legitimate_constant_p): Expand list of supported cases.
13022 * config/aarch64/aarch64-protos.h
13023 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
13024 (aarch64_reinterpret_float_as_int): New.
13025 (aarch64_simd_imm_scalar_p): Remove.
13026 * config/aarch64/constraints.md (Uvi): New.
13027 (Dd): Split into Ds and new Dd.
13028 * config/aarch64/aarch64.md (*movsi_aarch64):
13030 (*movdi_aarch64): Add SIMD mov case.
13032 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13034 * tree-predcom.c: (struct chain): Handle store-store chain in which
13035 stores for elimination only store loop invariant values.
13036 (execute_pred_commoning_chain): Ditto.
13037 (prepare_initializers_chain_store_elim): Ditto.
13038 (prepare_finalizers): Ditto.
13039 (is_inv_store_elimination_chain): New function.
13040 (initialize_root_vars_store_elim_1): New function.
13042 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13044 * tree-predcom.c: Revise general description of the pass.
13045 (enum chain_type): New enum type for store elimination.
13046 (struct chain): New field supporting store elimination.
13047 (struct component): Ditto.
13048 (dump_chain): Dump store-stores chain.
13049 (release_chain): Release resources.
13050 (split_data_refs_to_components): Compute and create component
13051 contains only stores for elimination.
13052 (get_chain_last_ref_at): New function.
13053 (make_invariant_chain): Initialization.
13054 (make_rooted_chain): Specify chain type in parameter and record it.
13055 (add_looparound_copies): Skip for store-stores chain.
13056 (determine_roots_comp): Compute type of chain and pass it to
13058 (initialize_root_vars_store_elim_2): New function.
13059 (finalize_eliminated_stores): New function.
13060 (remove_stmt): Handle store for elimination.
13061 (execute_pred_commoning_chain): Execute predictive commoning on
13062 store-store chains.
13063 (determine_unroll_factor): Skip unroll for store-stores chain.
13064 (prepare_initializers_chain_store_elim): New function.
13065 (prepare_initializers_chain): Hanlde store-store chain.
13066 (prepare_finalizers_chain, prepare_finalizers): New function.
13067 (tree_predictive_commoning_loop): Return integer value indicating
13068 if loop is unrolled or lcssa form is corrupted.
13069 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
13071 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13073 * tree-predcom.c (initialize_root): Delete.
13074 (execute_pred_commoning_chain): Initialize root vars and replace
13075 reference of non-combined chain directly, rather than call above
13078 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13080 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
13081 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
13083 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13085 * tree-predcom.c (struct chain): New field init_seq.
13086 (release_chain): Release init_seq.
13087 (prepare_initializers_chain): Record intialization stmts in above
13089 (insert_init_seqs): New function.
13090 (tree_predictive_commoning_loop): Call insert_init_seqs.
13092 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13094 * tree-predcom.c (determine_roots_comp): Skip trivial components.
13096 2017-07-28 Richard Biener <rguenther@suse.de>
13098 * match.pd: Remove superfluous :c.
13099 * genmatch.c (simplify::id): Add member.
13100 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
13102 (current_id): New global.
13103 (dt_node::parent): Move from ...
13104 (dt_operand::parent): ... here. Add for_id member.
13105 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
13106 (decision_tree::find_node): Relax order requirement when
13107 merging DT_TRUE nodes to ones inbetween the current simplify
13108 and the one we try to merge with. Add diagnostic whenever
13109 we need to enforce pattern order by not merging.
13110 (decision_tree::insert): Set current_id.
13111 (decision_tree::print_node): Dump parent node and for_id.
13112 (parser::last_id): Add member.
13113 (parser::push_simplify): Assign unique id.
13114 (parser::parser): Initialize last_id.
13116 2017-07-28 Martin Liska <mliska@suse.cz>
13119 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
13120 gimple_build_debug_bind.
13122 2017-07-28 Richard Biener <rguenther@suse.de>
13124 PR tree-optimization/81502
13125 * match.pd: Add pattern combining BIT_INSERT_EXPR with
13127 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
13129 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
13130 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
13131 for BIT_FIELD_REF args.
13132 * fold-const.c (make_bit_field_ref): Likewise.
13133 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
13135 2017-07-28 Jakub Jelinek <jakub@redhat.com>
13138 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
13139 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
13140 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
13141 Or it into SANITIZER_UNDEFINED.
13142 * ubsan.c: Include gimple-fold.h and varasm.h.
13143 (ubsan_expand_ptr_ifn): New function.
13144 (instrument_pointer_overflow): New function.
13145 (maybe_instrument_pointer_overflow): New function.
13146 (instrument_object_size): Formatting fix.
13147 (pass_ubsan::execute): Call instrument_pointer_overflow
13148 and maybe_instrument_pointer_overflow.
13149 * internal-fn.c (expand_UBSAN_PTR): New function.
13150 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
13151 * sanitizer.def (__ubsan_handle_pointer_overflow,
13152 __ubsan_handle_pointer_overflow_abort): New builtins.
13153 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
13154 * internal-fn.def (UBSAN_PTR): New internal function.
13155 * opts.c (sanitizer_opts): Add pointer-overflow.
13156 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
13157 * fold-const.c (build_range_check): Compute pointer range check in
13158 integral type if pointer arithmetics would be needed. Formatting
13161 2017-07-28 Martin Liska <mliska@suse.cz>
13164 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
13165 parameters that are of a variable-length.
13167 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
13169 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
13171 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
13172 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
13173 (CRT_CALL_STATIC_FUNCTION): Likewise.
13174 (ASM_DEFAULT_SPEC): New define.
13175 (ASM_SPEC32): Likewise.
13176 (ASM_SPEC64): Likewise.
13177 (ASM_SPEC_COMMON): Likewise.
13178 (ASM_SPEC): Likewise.
13179 (INVALID_64BIT): Likewise.
13180 (LINK_OS_DEFAULT_SPEC): Likewise.
13181 (LINK_OS_SPEC32): Likewise.
13182 (LINK_OS_SPEC64): Likewise.
13183 (POWERPC_LINUX): Likewise.
13184 (PTRDIFF_TYPE): Likewise.
13185 (RESTORE_FP_PREFIX): Likewise.
13186 (RESTORE_FP_SUFFIX): Likewise.
13187 (SAVE_FP_PREFIX): Likewise.
13188 (SAVE_FP_SUFFIX): Likewise.
13189 (SIZE_TYPE): Likewise.
13190 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
13191 (TARGET_64BIT): Likewise.
13192 (TARGET_64BIT): Likewise.
13193 (TARGET_AIX): Likewise.
13194 (WCHAR_TYPE_SIZE): Likewise.
13195 (WCHAR_TYPE): Undefine.
13196 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
13197 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
13198 (CPP_OS_RTEMS_SPEC): Delete.
13199 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
13200 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
13202 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
13204 2017-07-28 Jakub Jelinek <jakub@redhat.com>
13206 PR tree-optimization/81578
13207 * tree-parloops.c (build_new_reduction): Bail out if
13208 reduction_code isn't one of the standard OpenMP reductions.
13209 Move the details printing after that decision.
13211 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
13213 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
13214 related to reload_in_progress.
13215 (splat_input_operand): Likewise.
13216 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
13218 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
13220 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
13221 (TARGET_INSTANTIATE_DECLS): Likewise.
13222 (legitimate_indexed_address_p): Delete reload_in_progress code.
13223 (rs6000_debug_legitimate_address_p): Likewise.
13224 (rs6000_eliminate_indexed_memrefs): Likewise.
13225 (rs6000_emit_le_vsx_store): Likewise.
13226 (rs6000_emit_move_si_sf_subreg): Likewise.
13227 (rs6000_emit_move): Likewise.
13228 (register_to_reg_type): Likewise.
13229 (rs6000_pre_atomic_barrier): Likewise.
13230 (rs6000_machopic_legitimize_pic_address): Likewise.
13231 (rs6000_allocate_stack_temp): Likewise.
13232 (rs6000_address_for_fpconvert): Likewise.
13233 (rs6000_address_for_altivec): Likewise.
13234 (rs6000_secondary_memory_needed_rtx): Delete function.
13235 (rs6000_check_sdmode): Likewise.
13236 (rs6000_alloc_sdmode_stack_slot): Likewise.
13237 (rs6000_instantiate_decls): Likewise.
13238 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
13239 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
13240 Delete reload_in_progress.
13241 (*vec_reload_and_plus_<mptrsize>): Likewise.
13242 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
13243 (vsx_div_v2di): Likewise.
13244 (vsx_udiv_v2di): Likewise.
13246 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
13248 * config/rs6000/rs6000.opt (mlra): Replace with stub.
13249 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
13250 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
13251 (rs6000_debug_reg_global): Delete print of LRA status.
13252 (rs6000_option_override_internal): Delete dead LRA related code.
13253 (rs6000_lra_p): Delete function.
13254 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
13256 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
13258 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
13259 * config/riscv/rtems.h: New file.
13261 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13262 Sudakshina Das <sudi.das@arm.com>
13264 * config/aarch64/aarch64.md
13265 (define_split for and<mode>3nr_compare): Move
13266 non aarch64_logical_operand to a register.
13267 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
13268 register immediate operand to a register.
13269 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
13271 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
13273 PR middle-end/81564
13274 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
13276 2017-07-27 Richard Biener <rguenther@suse.de>
13278 PR tree-optimization/81573
13279 PR tree-optimization/81494
13280 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
13281 multi defuse cycle case.
13283 2017-07-27 Richard Biener <rguenther@suse.de>
13285 PR tree-optimization/81571
13286 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
13289 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
13291 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
13292 earlier and only if MASK_FPU is set. Adjust formatting.
13294 2017-07-27 Martin Liska <mliska@suse.cz>
13296 * opt-functions.awk: Add validation of value of Init.
13297 * optc-gen.awk: Pass new argument.
13299 2017-07-27 Martin Liska <mliska@suse.cz>
13301 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
13302 Fix wrong condition.
13304 2017-07-27 Martin Liska <mliska@suse.cz>
13306 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
13307 BBs and edges seen by autoFDO.
13309 2017-07-27 Richard Biener <rguenther@suse.de>
13311 PR tree-optimization/81502
13312 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
13313 with incompatible but same sized type.
13314 (execute_update_addresses_taken): Likewise.
13316 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
13318 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
13319 flag_tree_loop_vectorize rather than flag_tree_vectorize.
13321 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13324 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
13325 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
13326 Change s_operand to memory_operand.
13328 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
13330 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
13331 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
13332 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
13333 Emit instructions rather than returning an expression. Handle TFmode
13334 and KFmode by casting to TImode.
13335 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
13336 (rs6000_emit_le_vsx_store): Likewise.
13337 * config/rs6000/vsx.md (VSX_TI): New iterator.
13338 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
13339 (*vsx_le_undo_permute_<mode>): Likewise.
13340 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
13341 emit the split sequence.
13342 (*vsx_le_perm_store_<mode>): Likewise.
13344 2017-07-27 Jakub Jelinek <jakub@redhat.com>
13346 PR tree-optimization/81555
13347 PR tree-optimization/81556
13348 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
13349 if true, force CHANGED for the recursive invocation.
13350 (reassociate_bb): Remember original length of ops array, pass
13351 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
13353 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
13354 attributes for noipa attribute. For naked attribute use
13355 lookup_attribute first before lookup_attribute_spec.
13356 * final.c (rest_of_handle_final): Disable IPA RA for functions with
13358 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
13359 for functions with noipa attribute.
13360 (cgraph_externally_visible_p): Return true for functions with noipa
13362 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
13363 for functions with noipa attribute.
13364 * doc/extend.texi: Document noipa function attribute.
13365 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
13366 also for functions with noipa attribute.
13367 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
13369 2017-07-26 Andrew Pinski <apinski@cavium.com>
13371 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
13372 vec_unalign_load_cost and vec_unalign_store_cost.
13374 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13376 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
13377 -mvsx-small-integer option.
13378 (ISA_3_0_MASKS_IEEE): Likewise.
13379 (OTHER_VSX_VECTOR_MASKS): Likewise.
13380 (POWERPC_MASKS): Likewise.
13381 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
13382 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
13383 code, only testing for DImode being allowed in non-VSX floating
13385 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
13386 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
13388 (rs6000_option_override_internal): Delete -mvsx-small-integer.
13389 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
13390 TARGET_P8_VECTOR test.
13391 (rs6000_secondary_reload_simple_move): Likewise.
13392 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
13393 since TARGET_P9_VECTOR was already tested.
13394 (rs6000_opt_masks): Remove -mvsx-small-integer.
13395 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
13396 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
13398 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
13399 test for TARGET_VEXTRACTUB was used, and that uses
13401 (p9 extract splitter): Likewise.
13402 (vsx_extract_<mode>_di_p9): Likewise.
13403 (vsx_extract_<mode>_store_p9): Likewise.
13404 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
13405 for TARGET_P9_VECTOR was used. Delete code that is now dead with
13406 the elimination of TARGET_VSX_SMALL_INTEGER.
13407 (vsx_extract_<mode>_p8): Likewise.
13408 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
13409 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
13410 (vsx_set_<mode>_p9): Likewise.
13411 (vsx_set_v4sf_p9): Likewise.
13412 (vsx_set_v4sf_p9_zero): Likewise.
13413 (vsx_insert_extract_v4sf_p9): Likewise.
13414 (vsx_insert_extract_v4sf_p9_2): Likewise.
13415 * config/rs6000/rs6000.md (sign extend splitter): Change
13416 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
13417 (floatsi<mode>2_lfiwax_mem): Likewise.
13418 (floatunssi<mode>2_lfiwzx_mem): Likewise.
13419 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
13420 since a test for TARGET_P9_VECTOR was used.
13421 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
13422 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
13423 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
13424 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
13425 TARGET_P8_VECTOR test.
13426 (fix_trunc<mode>si2_stfiwx): Likewise.
13427 (fix_trunc<mode>si2_internal): Likewise.
13428 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
13429 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
13431 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
13432 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
13433 TARGET_P8_VECTOR test.
13434 (fixuns_trunc<mode>si2_stfiwx): Likewise.
13435 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
13436 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
13438 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
13439 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
13440 since a test for TARGET_P9_VECTOR was used.
13441 (splitter for loading small constants): Likewise.
13443 2017-07-26 Andrew Pinski <apinski@cavium.com>
13445 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
13448 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
13451 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
13452 (fp_valid_at): Likewise.
13454 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
13456 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
13457 (qdf24xx_addrcost_table): Likewise.
13458 (cortexa57_tunings): Update to use generic_branch_cost.
13459 (cortexa72_tunings): Likewise.
13460 (cortexa73_tunings): Likewise.
13461 (qdf24xx_tunings): Likewise.
13463 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
13465 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
13466 (thunderx2t99_branch_cost): Likewise.
13467 (cortexa35_tunings): Update to use generic_branch_cost.
13468 (cortexa53_tunings): Likewise.
13469 (cortexa57_tunings): Likewise.
13470 (cortexa72_tunings): Likewise.
13471 (cortexa73_tunings): Likewise.
13472 (thunderx2t99_tunings): Likewise.
13474 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
13476 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
13477 (sparc_option_override): Honour MASK_FSMULD.
13478 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
13479 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
13480 * config/sparc/sparc.opt (mfsmuld): New option.
13481 * doc/invoke.texi (mfsmuld): Document option.
13483 2017-07-26 Marek Polacek <polacek@redhat.com>
13485 PR middle-end/70992
13486 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
13488 2017-07-26 Richard Biener <rguenther@suse.de>
13490 * gimple-match-head.c (do_valueize): Return OP if valueize
13492 (get_def): New helper to get at the def stmt of a SSA name
13493 if valueize allows.
13494 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
13495 do_valueize to get at the def stmt.
13496 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
13498 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
13500 PR middle-end/46932
13501 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
13503 2017-07-26 Martin Liska <mliska@suse.cz>
13506 * function.c (expand_function_start): Make expansion of
13507 nonlocal_goto_save_area after parm_birth_insn.
13509 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
13511 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
13512 from all CPU target flags enable members.
13514 2017-07-26 Richard Biener <rguenther@suse.de>
13516 * genmatch.c (dt_simplify::gen): Make iterator vars const.
13517 (decision_tree::gen): Make 'type' const.
13518 (write_predicate): Likewise.
13520 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
13522 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
13523 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
13524 (rs6000_option_override_internal): Likewise.
13525 (rs6000_expand_vector_set): Likewise.
13526 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
13527 (TARGET_UPPER_REGS_SF): Likewise.
13528 (TARGET_UPPER_REGS_DI): Likewise.
13529 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
13530 (TARGET_DIRECT_MOVE_64BIT): Likewise.
13531 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
13532 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
13533 (Splitters for DI constants in Altivec registers): Likewise.
13534 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
13535 (vsx_set_v4sf_p9): Likewise.
13536 (vsx_set_v4sf_p9_zero): Likewise.
13537 (vsx_insert_extract_v4sf_p9): Likewise.
13538 (vsx_insert_extract_v4sf_p9_2): Likewise.
13540 2017-07-25 Carl Love <cel@us.ibm.com>
13542 * doc/extend.texi: Update the built-in documentation file for the
13543 existing built-in functions
13544 vector signed char vec_cnttz (vector signed char);
13545 vector unsigned char vec_cnttz (vector unsigned char);
13546 vector signed short vec_cnttz (vector signed short);
13547 vector unsigned short vec_cnttz (vector unsigned short);
13548 vector signed int vec_cnttz (vector signed int);
13549 vector unsigned int vec_cnttz (vector unsigned int);
13550 vector signed long long vec_cnttz (vector signed long long);
13551 vector unsigned long long vec_cnttz (vector unsigned long long);
13553 2017-07-25 Andrew Pinski <apinski@cavium.com>
13555 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
13556 accesses where the use is for the first operand of a BIT_INSERT.
13558 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
13561 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
13562 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
13564 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
13566 * config/i386/gstabs.h: Delete.
13567 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
13569 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
13571 * config/i386/i386.c (ix86_decompose_address): Do not check for
13572 register RTX when looking at index_reg or base_reg.
13573 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
13575 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
13577 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
13578 to update EH info here.
13580 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
13582 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
13584 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
13586 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
13588 2017-07-25 Torsten Duwe <duwe@suse.de>
13590 * common.opt: Introduce -fpatchable-function-entry
13591 command line option, and its variables function_entry_patch_area_size
13592 and function_entry_patch_area_start.
13593 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
13594 including a two-value parser.
13595 * target.def (print_patchable_function_entry): New target hook.
13596 * targhooks.h (default_print_patchable_function_entry): New function.
13597 * targhooks.c (default_print_patchable_function_entry): Likewise.
13598 * toplev.c (process_options): Switch off IPA-RA if
13599 patchable function entries are being generated.
13600 * varasm.c (assemble_start_function): Look at the
13601 patchable-function-entry command line switch and current
13602 function attributes and maybe generate NOP instructions by
13603 calling the print_patchable_function_entry hook.
13604 * doc/extend.texi: Document patchable_function_entry attribute.
13605 * doc/invoke.texi: Document -fpatchable_function_entry
13606 command line option.
13607 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
13609 * doc/tm.texi: Re-generate.
13611 2017-07-25 Jakub Jelinek <jakub@redhat.com>
13614 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
13615 TARGET_AVX512DQ rather than TARGET_AVX512BW.
13617 2017-07-25 Tamar Christina <tamar.christina@arm.com>
13619 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
13621 2017-07-25 Richard Biener <rguenther@suse.de>
13623 PR tree-optimization/81455
13624 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
13625 not walk in cycles when looking for guards.
13627 2017-07-25 Richard Biener <rguenther@suse.de>
13629 PR tree-optimization/81529
13630 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
13631 when optimizing backedge uses.
13633 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
13635 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
13637 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
13638 to dl_section_ref. On AIX, append an expression to subtract
13639 the size of the section length to dl_section_ref.
13641 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
13643 * configure.ac: If any of the config.* scripts fail, exit 1.
13644 * configure: Regenerate.
13646 2017-07-25 Richard Biener <rguenther@suse.de>
13648 PR middle-end/81546
13649 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
13650 of immediate uses, be more verbose on errors.
13652 2017-07-25 Richard Biener <rguenther@suse.de>
13654 PR tree-optimization/81510
13655 * tree-vect-loop.c (vect_is_simple_reduction): When the
13656 reduction stmt is not inside the loop bail out.
13658 2017-07-25 Richard Biener <rguenther@suse.de>
13660 PR tree-optimization/81303
13661 * tree-vect-loop-manip.c (vect_loop_versioning): Build
13662 profitability check against LOOP_VINFO_NITERSM1.
13664 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
13666 * domwalk.c (cmp_bb_postorder): Simplify.
13667 (sort_bbs_postorder): New function. Use it...
13668 (dom_walker::walk): ...here to optimize common cases.
13670 2017-07-25 Martin Liska <mliska@suse.cz>
13673 * ipa-visibility.c (function_and_variable_visibility): Make the
13674 redirection just on target that supports aliasing.
13675 Fix GNU coding style.
13677 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
13680 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
13682 * config/arm/rtems.h: Likewise.
13683 * config/bfin/rtems.h: Likewise.
13684 * config/i386/rtemself.h: Likewise.
13685 * config/lm32/rtems.h: Likewise.
13686 * config/m32c/rtems.h: Likewise.
13687 * config/m68k/rtemself.h: Likewise.
13688 * config/microblaze/rtems.h: Likewise.
13689 * config/mips/rtems.h: Likewise.
13690 * config/moxie/rtems.h: Likewise.
13691 * config/nios2/rtems.h: Likewise.
13692 * config/powerpcspe/rtems.h: Likewise.
13693 * config/rs6000/rtems.h: Likewise.
13694 * config/rtems.h: Likewise.
13695 * config/sh/rtems.h: Likewise.
13696 * config/sh/rtemself.h: Likewise.
13697 * config/sparc/rtemself.h: Likewise.
13699 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
13702 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
13703 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
13704 * tree-ssa-structalias.c (alias_get_name): Same.
13706 2017-07-25 Bin Cheng <bin.cheng@arm.com>
13709 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
13710 instructions if no du chain is found.
13712 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
13714 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
13716 2017-07-25 Richard Biener <rguenther@suse.de>
13718 PR middle-end/81505
13719 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
13722 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
13724 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
13725 upper-regs options.
13726 (ISA_2_7_MASKS_SERVER): Likewise.
13727 (ISA_3_0_MASKS_IEEE): Likewise.
13728 (OTHER_P8_VECTOR_MASKS): Likewise.
13729 (OTHER_VSX_VECTOR_MASKS): Likewise.
13730 (POWERPC_MASKS): Likewise.
13731 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
13732 duplicate list of options.
13733 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
13734 explicit -mupper-regs options.
13735 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
13736 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
13737 alias for -mupper-regs-df.
13738 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
13739 (rs6000_init_hard_regno_mode_ok): Likewise.
13740 (rs6000_option_override_internal): Likewise.
13741 (rs6000_opt_masks): Likewise.
13742 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
13743 options in terms of whether -mvsx or -mpower8-vector was used.
13744 (TARGET_UPPER_REGS_DI): Likewise.
13745 (TARGET_UPPER_REGS_SF): Likewise.
13746 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
13747 -mupper-regs-* options.
13749 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
13751 * passes.c (emergency_dump_function): Print some empty lines and a
13752 header before the RTL dump.
13754 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
13756 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
13758 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
13761 * config/aarch64/aarch64.c (aarch64_classify_symbol):
13762 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
13764 2017-07-24 Carl Love <cel@us.ibm.com>
13766 * config/rs6000/rs6000-c.c: Add support for built-in functions
13767 vector float vec_extract_fp32_from_shorth (vector unsigned short);
13768 vector float vec_extract_fp32_from_shortl (vector unsigned short);
13769 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
13770 vec_extract_fp_from_shortl): Add defines for the two builtins.
13771 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
13772 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
13774 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
13775 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
13776 * doc/extend.texi: Update the built-in documentation file for the
13777 new built-in function.
13779 2017-07-24 Jakub Jelinek <jakub@redhat.com>
13782 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
13784 * doc/generic.texi: Likewise.
13785 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
13786 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
13788 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
13790 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
13791 (aarch64_mls_elt_merge<mode>): Likewise.
13793 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
13795 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
13796 having __cxa_atexit.
13798 2017-07-23 Michael Collison <michael.collison@arm.com>
13800 * config/arm/arm.c (arm_option_override): Deprecate
13801 use of -mstructure-size-boundary.
13802 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
13803 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
13805 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13808 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
13809 Reduce cost estimate for direct moves.
13811 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
13814 * config/i386/i386.c (ix86_option_override_internal): Disable
13815 BMI, BMI2 and TBM instructions for -m16.
13817 2017-07-21 Carl Love <cel@us.ibm.com>
13819 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13820 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13821 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13822 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13823 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13824 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
13825 VMULOSW): New enum "unspec" values.
13826 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
13827 altivec_vmulosw): New patterns.
13828 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13829 VMULOSW): Add definitions.
13831 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
13833 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
13834 (qdf24xx): Likewise.
13835 * config/aarch64/aarch64-options-extensions.def (rdma); New.
13836 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
13837 (AARCH64_FL_V8_1): Renumber.
13838 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
13839 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
13840 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
13841 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
13842 rdma to feature modifiers list.
13844 2017-07-21 Yury Gribov <tetra2005@gmail.com>
13846 PR middle-end/56727
13847 * ipa-visibility (function_and_variable_visibility): Convert
13848 recursive PLT call to direct call if appropriate.
13850 2017-07-21 Andrew Pinski <apinski@cavium.com>
13852 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
13853 operand 1 to see if the types precision matches.
13854 * fold-const.c (operand_equal_p): Likewise.
13856 2017-07-21 Richard Biener <rguenther@suse.de>
13858 PR tree-optimization/81303
13859 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
13860 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
13861 (vect_peeling_hash_get_lowest_cost): Adjust.
13862 (vect_enhance_data_refs_alignment): Likewise. Use
13863 vect_get_peeling_costs_all_drs to compute the penalty for no
13864 peeling to match up costs.
13866 2017-07-21 Richard Biener <rguenther@suse.de>
13868 PR tree-optimization/81500
13869 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
13870 we didn't identify a reduction path.
13872 2017-07-21 Tom de Vries <tom@codesourcery.com>
13873 Cesar Philippidis <cesar@codesourcery.com>
13875 PR gcov-profile/81442
13876 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
13879 2017-07-21 Tom de Vries <tom@codesourcery.com>
13882 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
13884 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
13885 nvptx_override_options_after_change.
13887 2017-07-21 Ulrich Drepper <drepper@redhat.com>
13889 * dwarf2out.c (output_file_names): Avoid double testing for
13890 dwarf_version >= 5.
13892 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
13894 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
13896 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
13898 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
13900 (try_crossjump_to_edge): Do not punt on partitioned functions.
13902 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
13904 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
13905 Put all BBs reachable only via paths crossing cold region to cold
13907 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
13909 2016-07-21 Richard Biener <rguenther@suse.de>
13911 PR tree-optimization/81303
13912 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
13913 into account prologue and epilogue iterations when raising
13914 min_profitable_iters to sth at least covering one vector iteration.
13916 2017-07-21 Tamar Christina <tamar.christina@arm.com>
13918 * config/arm/arm.c (arm_test_cpu_arch_dat):
13921 2017-07-20 Nathan Sidwell <nathan@acm.org>
13923 Remove TYPE_METHODS.
13924 * tree.h (TYPE_METHODS): Delete.
13925 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
13926 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
13927 (dbxout_type_methods): Scan TYPE_FIELDS.
13928 (dbxout_type): Don't check TYPE_METHODS here.
13929 * function.c (use_register_for_decl): Always ignore register for
13930 class types when not optimizing.
13931 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
13932 * tree.c (free_lang_data_in_type): Stitch out member functions and
13933 templates from TYPE_FIELDS.
13934 (build_distinct_type_copy, verify_type_variant,
13935 verify_type): Member fns are on TYPE_FIELDS.
13936 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
13937 * tree-pretty-print.c (dump_generic_node): Likewise.
13939 2017-07-20 Jakub Jelinek <jakub@redhat.com>
13942 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
13943 V2TImode and V4TImode.
13944 (ix86_expand_vector_extract): Likewise.
13945 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
13946 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
13947 (ssescalarmode): Handle V4TImode and V2TImode.
13948 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
13949 (*vec_extractv2ti, *vec_extractv4ti): New insns.
13950 (VEXTRACTI128_MODE): New mode iterator.
13951 (splitter for *vec_extractv?ti first element): New.
13952 (VEC_INIT_MODE): New mode iterator.
13953 (vec_init<mode>): Consolidate 3 expanders into one using
13954 VEC_INIT_MODE mode iterator.
13956 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
13958 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
13959 non_spilled_static_chain_regno_p.
13961 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
13963 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
13965 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
13967 * bb-reorder.c (connect_traces): Allow copying of blocks within
13970 2017-07-20 Richard Biener <rguenther@suse.de>
13972 * gimple.h (gimple_phi_result): Add gphi * overload.
13973 (gimple_phi_result_ptr): Likewise.
13974 (gimple_phi_arg): Likewise. Adjust index assert to only
13975 allow actual argument accesses rather than all slots available
13977 (gimple_phi_arg_def): Add gphi * overload.
13978 * tree-phinodes.c (make_phi_node): Initialize only actual
13980 (resize_phi_node): Clear memory not covered by old node,
13981 do not initialize excess argument slots.
13982 (reserve_phi_args_for_new_edge): Initialize new argument slot
13985 2017-07-20 Bin Cheng <bin.cheng@arm.com>
13987 PR tree-optimization/81388
13989 2016-07-21 Bin Cheng <bin.cheng@arm.com>
13991 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
13992 by removing computation of may_be_zero.
13994 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
13995 Tom de Vries <tom@codesourcery.com>
13997 PR middle-end/81030
13998 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
13999 when gimple level profile disagrees with what RTL expander did.
14001 2017-07-20 Richard Biener <rguenther@suse.de>
14003 PR tree-optimization/61171
14004 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
14005 (vect_analyze_stmt): Add slp instance parameter.
14006 (vectorizable_reduction): Likewise.
14007 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
14008 (vect_is_simple_reduction): Deal with chains not detected
14009 as SLP reduction chain, specifically not properly associated
14010 chains containing a mix of plus/minus.
14011 (get_reduction_op): Remove.
14012 (get_initial_defs_for_reduction): Simplify, pass in whether
14013 this is a reduction chain, pass in the SLP node for the PHIs.
14014 (vect_create_epilog_for_reduction): Get the SLP instance as
14016 (vectorizable_reduction): Get the SLP instance as arg.
14017 During analysis remember the SLP node with the PHIs in the
14018 instance. Simplify getting at the vectorized reduction PHIs.
14019 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
14020 through SLP instance.
14021 (vect_slp_analyze_operations): Likewise.
14022 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
14023 (vect_transform_stmt): Likewise.
14025 2017-07-20 Tom de Vries <tom@codesourcery.com>
14027 PR tree-optimization/81489
14028 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
14029 read of phi arg location to before loop that modifies phi.
14031 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14033 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
14036 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
14038 PR middle-end/81331
14039 * except.c (execute): Fix ordering issue.
14041 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
14043 PR rtl-optimization/81423
14044 * combine.c (make_compound_operation_int): Don't try to optimize
14045 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
14047 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
14049 PR rtl-optimization/81423
14050 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
14051 with a constant that is -1 in the truncated to mode.
14053 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
14055 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
14056 (determine_unlikely_bbs): ... here.
14057 * predict.h (propagate_unlikely_bbs_forward): Declare.
14058 * cfgexpand.c (pass_expand::execute): Use it.
14059 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
14061 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
14062 propagate_unlikely_bbs_forward.
14064 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
14066 PR middle-end/81331
14067 * except.c (maybe_add_nop_after_section_switch): New function.
14070 2017-07-19 Tom de Vries <tom@codesourcery.com>
14072 * gimple.h (gimple_phi_set_arg): Make assert more strict.
14074 2017-07-19 Tom de Vries <tom@codesourcery.com>
14076 * gimple.h (gimple_phi_arg): Make assert more strict.
14078 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
14080 * config.gcc (powerpc*-*-*): Add mmintrin.h.
14081 * config/rs6000/mmintrin.h: New file.
14082 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
14084 2017-07-19 Jakub Jelinek <jakub@redhat.com>
14086 PR tree-optimization/81346
14087 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
14089 2017-07-19 Tom de Vries <tom@codesourcery.com>
14091 * config/nvptx/nvptx.md (VECIM): Add V2DI.
14093 2017-07-19 Tom de Vries <tom@codesourcery.com>
14095 * config/nvptx/nvptx-modes.def: Add V2DImode.
14096 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
14097 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
14098 (nvptx_output_mov_insn): Handle lack of mov.b128.
14099 (nvptx_print_operand): Handle 'H' and 'L' codes.
14100 (nvptx_vector_mode_supported): Allow V2DImode.
14101 (nvptx_preferred_simd_mode): New function.
14102 (nvptx_data_alignment): New function.
14103 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
14104 nvptx_preferred_simd_mode.
14105 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
14107 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
14109 2017-07-19 Tom de Vries <tom@codesourcery.com>
14111 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
14112 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
14113 (nvptx_vector_mode_supported): New function. Allow V2SImode.
14114 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
14115 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
14116 (mov<VECIM>_insn): New define_insn.
14117 (define_expand "mov<VECIM>): New define_expand.
14119 2017-07-19 Tom de Vries <tom@codesourcery.com>
14121 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
14123 2017-07-19 Jakub Jelinek <jakub@redhat.com>
14125 PR tree-optimization/81346
14126 * fold-const.h (fold_div_compare, range_check_type): Declare.
14127 * fold-const.c (range_check_type): New function.
14128 (build_range_check): Use range_check_type.
14129 (fold_div_compare): No longer static, rewritten into
14130 a match.pd helper function.
14131 (fold_comparison): Don't call fold_div_compare here.
14132 * match.pd (X / C1 op C2): New optimization using fold_div_compare
14133 as helper function.
14135 2017-07-19 Nathan Sidwell <nathan@acm.org>
14137 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
14138 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
14139 * tree.c (find_decls_types_r, verify_type): Use
14140 TYPE_{MIN,MAX}_VALUE_RAW.
14141 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
14142 (hash_tree): Likewise.
14143 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
14145 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
14148 2017-07-18 Tom de Vries <tom@codesourcery.com>
14150 PR middle-end/81464
14151 * omp-expand.c (expand_omp_for_static_chunk): Handle
14152 equal-argument loop exit phi.
14154 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
14157 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
14158 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
14159 operand 2 predicate.
14160 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
14161 operand 2 predicate.
14162 (ror,rol -> rorx splitters): Use const_int_operand as
14163 operand 2 predicate.
14165 2017-06-18 Richard Biener <rguenther@suse.de>
14167 PR tree-optimization/81410
14168 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
14169 the gap in the ! slp_perm SLP case after each group.
14171 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
14173 PR middle-end/81463
14174 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
14177 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
14179 PR middle-end/81462
14180 * predict.c (set_even_probabilities): Cleanup; do not affect
14181 probabilities that are already known.
14182 (combine_predictions_for_bb): Call even when count is set.
14184 2017-07-18 Nathan Sidwell <nathan@acm.org>
14186 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
14189 2017-07-18 Bin Cheng <bin.cheng@arm.com>
14192 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
14193 optimization for loop niter analysis.
14195 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
14198 * config/avr/avr.c (avr_optimize_casesi): Don't use
14199 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
14201 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
14203 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
14204 body_cost_vec from _vect_peel_extended_info.
14205 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
14206 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
14209 2017-07-18 Bin Cheng <bin.cheng@arm.com>
14211 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
14213 2017-07-18 Richard Biener <rguenther@suse.de>
14215 PR tree-optimization/80620
14216 PR tree-optimization/81403
14217 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
14218 info when re-using a VN table entry.
14220 2017-07-18 Richard Biener <rguenther@suse.de>
14222 PR tree-optimization/81418
14223 * tree-vect-loop.c (vectorizable_reduction): Properly compute
14224 vectype_in. Verify that with lane-reducing reduction operations
14225 we have a single def-use cycle.
14227 2017-07-17 Carl Love <cel@us.ibm.com>
14229 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
14231 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14232 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14233 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14234 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14235 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14236 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
14237 VMULOSW): New enum "unspec" values.
14238 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
14239 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
14240 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
14241 altivec_vmulosw): New patterns.
14242 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14243 VMULOSW): Add definitions.
14244 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
14246 * config/alpha/alpha.c: Include predict.h.
14248 2017-07-17 Yury Gribov <tetra2005@gmail.com>
14250 * tree-vrp.c (compare_assert_loc): Fix comparison function
14251 to return predictable results.
14253 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14255 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
14257 (subdi3): Likewise.
14258 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
14259 * doc/invoke.texi (mexpand-adddi): Update text.
14261 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14263 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
14264 that also clobbers the CC register. The old expand code is moved
14266 (*arc_clzsi2): ... here.
14267 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
14268 the CC register. The old expand code is moved to ...
14269 (arc_ctzsi2): ... here.
14271 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14273 * config/arc/arc.opt (mindexed-loads): Use initial value
14274 TARGET_INDEXED_LOADS_DEFAULT.
14275 (mauto-modify-reg): Use initial value
14276 TARGET_AUTO_MODIFY_REG_DEFAULT.
14277 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
14278 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
14279 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
14280 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
14282 2017-07-17 Martin Liska <mliska@suse.cz>
14285 * opts.c (finish_options): Do not allow -fgnu-tm
14286 w/ -fsanitize={kernel-,}address. Say sorry.
14288 2017-07-17 Bin Cheng <bin.cheng@arm.com>
14291 * tree-loop-distribution.c (classify_partition): Only assert on
14292 numer of iterations.
14293 (merge_dep_scc_partitions): Delete prameter. Update function call.
14294 (distribute_loop): Remove code handling loop with unknown niters.
14295 (pass_loop_distribution::execute): Skip loop with unknown niters.
14297 2017-07-17 Bin Cheng <bin.cheng@arm.com>
14300 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
14301 function sort_partitions_by_post_order.
14303 2017-07-17 Bin Cheng <bin.cheng@arm.com>
14305 PR tree-optimization/81374
14306 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
14307 the max index of basic blocks, rather than number of basic blocks.
14309 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14311 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
14313 (arc_legitimate_pic_operand_p): Likewise.
14314 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
14316 (arc_needs_pcl_p): Likewise.
14317 (arc_legitimate_pc_offset_p): Likewise.
14318 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
14319 function is also used in constrains.md.
14320 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
14321 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
14322 PLUS. Only return true/false in known cases, otherwise assert.
14323 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
14324 is already called in arc_legitimate_constant_p.
14325 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
14327 (LEGITIMATE_PIC_OPERAND_P): Use
14328 arc_raw_symbolic_reference_mentioned_p function.
14329 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
14334 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14335 Andrew Burgess <andrew.burgess@embecosm.com>
14337 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
14338 (arc_return_address_register): New function.
14339 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
14340 (arc_handle_fndecl_attribute): Add naked attribute.
14341 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
14342 (TARGET_WARN_FUNC_RETURN): Likewise.
14343 (arc_allocate_stack_slots_for_args): New function.
14344 (arc_warn_func_return): Likewise.
14345 (machine_function): Change type fn_type.
14346 (arc_compute_function_type): Consider new naked function type,
14347 change function return type.
14348 (arc_must_save_register): Adapt to handle new
14349 arc_compute_function_type's return type.
14350 (arc_expand_prologue): Likewise.
14351 (arc_expand_epilogue): Likewise.
14352 (arc_return_address_regs): Delete.
14353 (arc_return_address_register): New function.
14354 (arc_epilogue_uses): Use above function.
14355 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
14356 (arc_function_type): Change encoding, add naked type.
14357 (ARC_INTERRUPT_P): Change to handle the new encoding.
14358 (ARC_FAST_INTERRUPT_P): Likewise.
14359 (ARC_NORMAL_P): Define.
14360 (ARC_NAKED_P): Likewise.
14361 (arc_compute_function_type): Delete prototype.
14362 * config/arc/arc.md (in_ret_delay_slot): Use
14363 arc_return_address_register function.
14364 (simple_return): Likewise.
14365 (p_return_i): Likewise.
14367 2017-07-17 Jakub Jelinek <jakub@redhat.com>
14369 PR tree-optimization/81428
14370 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
14371 can't be built for those types.
14373 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
14375 Remove stuff dead since r239246.
14377 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
14378 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
14379 (avr_inform_devices): Remove dead stuff.
14381 2017-07-17 Tamar Christina <tamar.christina@arm.com>
14383 * config/arm/arm_neon.h: Fix softp typo.
14385 2017-07-17 Jakub Jelinek <jakub@redhat.com>
14387 PR tree-optimization/81365
14388 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
14389 aggregate moves onto bb predecessor edges, make sure there are no
14390 loads that could alias the lhs in between the start of bb and the
14393 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
14396 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
14397 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
14398 [LSHIFTRT, outer_code = TRUNCATE]: Same.
14400 2017-07-17 Jakub Jelinek <jakub@redhat.com>
14402 PR tree-optimization/81396
14403 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
14404 (init_symbolic_number): Initialize it to 1.
14405 (perform_symbolic_merge): Add n_ops from both operands into the new
14407 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
14408 without base_addr as useless if they need more than one operation.
14409 (bswap_replace): Handle !bswap case for NULL base_addr.
14411 2017-07-17 Tom de Vries <tom@codesourcery.com>
14414 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
14417 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
14419 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
14420 conditional builtin define __FIX_LEON3FT_B2BST.
14422 2017-07-17 Daniel Cederman <cederman@gaisler.com>
14424 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
14425 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
14428 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
14430 PR rtl-optimization/81424
14431 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
14432 to remove potential trapping from operands if -fnon-call-exceptions.
14434 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14436 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
14437 profile_proability for scalling.
14438 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
14440 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14442 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
14444 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14446 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
14447 fixpoint arithmetics.
14449 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14451 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
14452 fixpoint arithmetics.
14454 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14456 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
14457 fixpoint arithmetics.
14459 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14461 * profile-count.h (profile_probability::from_reg_br_prob_note,
14462 profile_probability::to_reg_br_prob_note): New functions.
14463 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
14464 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
14465 * predict.c (probability_reliable_p): Update.
14466 (edge_probability_reliable_p): Update.
14467 (br_prob_note_reliable_p): Update.
14468 (invert_br_probabilities): Update.
14469 (add_reg_br_prob_note): New function.
14470 (combine_predictions_for_insn): Update.
14471 * asan.c (asan_clear_shadow): Update.
14472 * cfgbuild.c (compute_outgoing_frequencies): Update.
14473 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
14474 (update_br_prob_note): Update.
14475 (rtl_verify_edges): Update.
14476 (purge_dead_edges): Update.
14477 (fixup_reorder_chain): Update.
14478 * emit-rtl.c (try_split): Update.
14479 * ifcvt.c (cond_exec_process_insns): Update.
14480 (cond_exec_process_if_block): Update.
14481 (dead_or_predicable): Update.
14482 * internal-fn.c (expand_addsub_overflow): Update.
14483 (expand_neg_overflow): Update.
14484 (expand_mul_overflow): Update.
14485 * loop-doloop.c (doloop_modify): Update.
14486 * loop-unroll.c (compare_and_jump_seq): Update.
14487 * optabs.c (emit_cmp_and_jump_insn_1): Update.
14488 * predict.h: Update.
14489 * reorg.c (mostly_true_jump): Update.
14491 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
14492 * config/alpha/alpha.c (emit_unlikely_jump): Update.
14493 * config/arc/arc.c: (emit_unlikely_jump): Update.
14494 * config/arm/arm.c: (emit_unlikely_jump): Update.
14495 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
14496 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
14497 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
14498 (ix86_print_operand): Update.
14499 (ix86_split_fp_branch): Update.
14500 (predict_jump): Update.
14501 * config/ia64/ia64.c (ia64_print_operand): Update.
14502 * config/mmix/mmix.c (mmix_print_operand): Update.
14503 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
14504 (rs6000_expand_split_stack_prologue): Update.
14505 * config/rs6000/rs6000.c: Update.
14506 * config/s390/s390.c (s390_expand_vec_strlen): Update.
14507 (s390_expand_vec_movstr): Update.
14508 (s390_expand_cs_tdsi): Update.
14509 (s390_expand_split_stack_prologue): Update.
14510 * config/sh/sh.c (sh_print_operand): Update.
14511 (expand_cbranchsi4): Update.
14512 (expand_cbranchdi4): Update.
14513 * config/sparc/sparc.c (output_v9branch): Update.
14514 * config/spu/spu.c (get_branch_target): Update.
14515 (ea_load_store_inline): Update.
14516 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
14517 * config/tilepro/tilepro.c: Update.
14519 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
14521 * gimplify.c (mostly_copy_tree_r): Revert latest change.
14522 (gimplify_save_expr): Likewise.
14524 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14526 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
14528 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14530 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
14532 * timevar.def (TV_IPA_FNSUMMARY): Define.
14534 2017-07-16 Daniel Cederman <cederman@gaisler.com>
14536 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
14537 to back store errata sensitive sequence from being generated.
14538 (sqrtdf2_fix): Likewise.
14540 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14542 * tree-ssa-threadupdate.c (compute_path_counts,
14543 update_joiner_offpath_counts): Use profile_probability.
14545 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
14548 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
14550 * config/arm/arm-c.c (arm_cpu_builtins): Define
14551 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
14553 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
14555 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14556 array entries to represent __ieee128 versions of the
14557 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
14558 scalar_extract_sig, and scalar_insert_exp built-in functions.
14559 (altivec_resolve_overloaded_builtin): Add special case handling
14560 for the __builtin_scalar_insert_exp function, as represented by
14561 the P9V_BUILTIN_VEC_VSIEDP constant.
14562 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
14563 exponent support for __ieee128 argument.
14564 (VSESQP): Add scalar extract signature support for __ieee128
14566 (VSTDCNQP): Add scalar test negative support for __ieee128
14568 (VSIEQP): Add scalar insert exponent support for __int128 argument
14569 with __ieee128 result.
14570 (VSIEQPF): Add scalar insert exponent support for __ieee128
14571 argument with __ieee128 result.
14572 (VSTDCQP): Add scalar test data class support for __ieee128
14574 (VSTDCNQP): Add overload support for scalar test negative with
14575 __ieee128 argument.
14576 (VSTDCQP): Add overload support for scalar test data class
14577 __ieee128 argument.
14578 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
14579 UNSPEC_VSX_SXSIGDP.
14580 (UNSPEC_VSX_SIEXPQP): New constant.
14581 (xsxexpqp): New insn for VSX scalar extract exponent quad
14583 (xsxsigqp): New insn for VSX scalar extract significand quad
14585 (xsiexpqpf): New insn for VSX scalar insert exponent quad
14586 precision with floating point argument.
14587 (xststdcqp): New expand for VSX scalar test data class quad
14589 (xststdcnegqp): New expand for VSX scalar test negative quad
14591 (xststdcqp): New insn to match expansions for VSX scalar test data
14592 class quad precision and VSX scalar test negative quad precision.
14593 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
14594 special case operand checking to enforce that second operand of
14595 VSX scalar test data class with quad precision argument is a 7-bit
14597 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
14598 prototypes and descriptions of __ieee128 versions of
14599 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
14600 scalar_test_data_class, and scalar_test_neg built-in functions.
14602 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14604 PR tree-optimization/81162
14605 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
14606 replace a negate with an add.
14608 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
14610 * doc/invoke.texi (arm/-mcpu): Document +crypto.
14612 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
14614 * config/arm/arm-c.c (arm_cpu_builtins): Define
14615 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
14617 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
14619 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
14620 (armv8-r): Set ARM Cortex-R52 as default CPU.
14621 * config/arm/arm-tables.opt: Regenerate.
14622 * config/arm/arm-tune.md: Regenerate.
14623 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
14625 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
14626 extension for -mcpu=cortex-r52.
14628 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
14630 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
14631 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
14632 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
14633 (fp-armv8): Define it as FP_ARMv8 only.
14634 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
14635 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
14637 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
14638 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
14639 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
14640 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
14641 than TARGET_FPU_ARMV8.
14642 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
14643 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
14644 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
14646 * config/arm/neon.md (neon_vrint): Likewise.
14647 (neon_vcvt): Likewise.
14648 (neon_<fmaxmin_op><mode>): Likewise.
14649 (<fmaxmin><mode>3): Likewise.
14650 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
14651 * config/arm/predicates.md (arm_cond_move_operator): Check against
14652 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
14654 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
14656 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
14657 to top of function.
14659 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14661 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
14662 loop in comment with memset.
14664 2017-07-14 Martin Liska <mliska@suse.cz>
14666 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
14667 * dwarf2out.c (is_java): Remove the function.
14668 (output_pubname): Remove usage of the function.
14669 (lower_bound_default): Remove usage of DW_LANG_Java.
14670 (gen_compile_unit_die): Likewise.
14671 * gcc.c: Remove compiler defaults for .java and .zip files.
14672 * gimple-expr.c (remove_suffix): Change as there's no longer
14673 extension than 4-letter one.
14674 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
14675 (gimplify_save_expr): Likewise.
14676 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
14677 as it's possible even for other languages than Java.
14678 * langhooks.h (struct lang_hooks): Remove Java from a comment.
14679 * lto-opts.c (lto_write_options): Remove reference to Java.
14680 * opts.c (strip_off_ending): Update file extension handling.
14681 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
14682 * tree-eh.c (lower_resx): Likewise.
14683 * tree.c (free_lang_data_in_type): Remove dead code.
14684 (find_decls_types_r): Likewise.
14685 (build_common_builtin_nodes): Remove Java from a comment.
14686 (verify_type): Remove dead code.
14687 * varasm.c (assemble_external): Remove Java from a comment.
14689 2017-07-14 Martin Liska <mliska@suse.cz>
14691 * opts.c (finish_options): Add quotes.
14692 (common_handle_option): Likewise.
14694 2017-07-14 Martin Liska <mliska@suse.cz>
14696 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
14697 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
14698 Remove N_SO_PASCAL.
14699 * dwarf2out.c (lower_bound_default): Do not handle
14701 (gen_compile_unit_die): Likewise.
14702 * gcc.c: Remove default extension binding for GNU Pascal.
14703 * stmt.c: Remove Pascal language from a comment.
14704 * xcoffout.c: Likewise.
14706 2017-07-13 David Malcolm <dmalcolm@redhat.com>
14709 * diagnostic-show-locus.c (fixit_cmp): New function.
14710 (layout::layout): Sort m_fixit_hints.
14711 (column_range::column_range): Assert that the values are valid.
14712 (struct char_span): New struct.
14713 (correction::overwrite): New method.
14714 (struct source_line): New struct.
14715 (line_corrections::add_hint): Add assertions. Reimplement memcpy
14716 calls in terms of classes source_line and char_span, and
14717 correction::overwrite.
14718 (selftest::test_overlapped_fixit_printing_2): New function.
14719 (selftest::diagnostic_show_locus_c_tests): Call it.
14721 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
14723 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
14724 early if there is no lhs.
14726 2017-07-13 Martin Liska <mliska@suse.cz>
14728 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
14729 (gen_reference_type_die): Likewise.
14730 * stor-layout.c: Remove Pascal-related comment.
14732 2017-07-13 Martin Liska <mliska@suse.cz>
14734 * opts.c (finish_options): Add quotes to error messages.
14735 (parse_sanitizer_options): Likewise.
14737 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
14739 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
14741 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
14743 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
14745 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
14747 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
14749 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
14751 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
14754 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
14755 provides the hardware capability bits, define the macro
14756 __BUILTIN_CPU_SUPPORTS__.
14757 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
14758 if GLIBC does not provide the hardware capability bits. Add a
14759 gcc_unreachable call if the built-in cpu function is neither
14760 __builtin_cpu_is nor __builtin_cpu_supports.
14761 (rs6000_get_function_versions_dispatcher): Change the warning
14762 that an old GLIBC is used which does not export the capability
14763 bits to be an error.
14764 * doc/extend.texi (target_clones attribute): Document the
14765 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
14766 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
14767 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
14768 the macros defined by GCC if the newer GLIBC is available.
14770 2017-07-12 Jeff Law <law@redhat.com>
14772 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
14773 remaining includes slightly.
14774 * config/riscv/riscv-builtins.c: Include profile-count.h.
14776 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
14779 * config/avr/avr.c (avr_set_current_function): In diagnostic
14780 messages: Quote keywords and (parts of) identifiers.
14781 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
14784 2017-07-12 Carl Love <cel@us.ibm.com>
14786 * config/rs6000/rs6000-c.c: Add support for built-in functions
14787 vector bool char vec_revb (vector bool char);
14788 vector bool short vec_revb (vector short char);
14789 vector bool int vec_revb (vector bool int);
14790 vector bool long long vec_revb (vector bool long long);
14791 * doc/extend.texi: Update the built-in documentation file for the
14792 new built-in functions.
14794 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14796 * config/s390/s390.md: Remove movcc splitter.
14798 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14800 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
14801 load/store on condition.
14803 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
14806 * config/avr/avr.c (avr_encode_section_info)
14807 [progmem && !TREE_READONLY]: Error if progmem object needs
14810 2017-07-11 Michael Collison <michael.collison@arm.com>
14812 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
14815 2017-07-11 Carl Love <cel@us.ibm.com>
14817 * config/rs6000/rs6000-c.c: Add support for builtins
14818 vector unsigned int vec_parity_lsbb (vector signed int);
14819 vector unsigned int vec_parity_lsbb (vector unsigned int);
14820 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
14821 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
14822 vector unsigned long long vec_parity_lsbb (vector signed long long);
14823 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
14824 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
14825 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
14826 * doc/extend.texi: Update the built-in documentation file for the
14827 new built-in functions.
14829 2017-07-11 David Malcolm <dmalcolm@redhat.com>
14831 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
14832 (layout::m_primary_loc): New field.
14833 (layout::layout): Initialize new field. Move location filtering
14834 logic from here to...
14835 (layout::maybe_add_location_range): ...this new method. Add
14836 support for filtering to just the lines already specified by other
14838 (layout::will_show_line_p): New method.
14839 (gcc_rich_location::add_location_if_nearby): New method.
14840 (selftest::test_add_location_if_nearby): New test function.
14841 (selftest::diagnostic_show_locus_c_tests): Call it.
14842 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
14845 2017-07-11 Tom de Vries <tom@codesourcery.com>
14847 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
14848 (bb_first_real_insn): New function.
14849 (nvptx_single): Add extra initialization of broadcasted condition
14852 2017-07-11 Nathan Sidwell <nathan@acm.org>
14854 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
14856 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
14858 * doc/extend.texi (AVR Function Attributes): Remove weblink to
14859 Binutils doc as TEXI will mess them up.
14860 * doc/invoke.texi (AVR Options): Same here.
14862 2017-07-11 Daniel Cederman <cederman@gaisler.com>
14864 * config/sparc/sparc.opt (mfix-ut700): New option.
14865 (mfix-gr712rc): Likewise.
14866 (sparc_fix_b2bst): New variable.
14867 * doc/invoke.texi (SPARC options): Document them.
14868 (ARM options): Fix warnings.
14869 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
14870 instructions to prevent sequences that can trigger the store-store
14871 errata for certain LEON3FT processors.
14872 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
14873 (sparc_option_override): Set sparc_fix_b2bst appropriately.
14874 * config/sparc/sparc.md (fix_b2bst): New attribute.
14875 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
14877 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
14880 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
14882 (*rsqrtsf2_sse): Ditto.
14884 (div<mode>3): Macroize insn from divdf3 and divsf3
14885 using MODEF mode iterator.
14887 2017-07-10 Martin Sebor <msebor@redhat.com>
14889 PR tree-optimization/80397
14890 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
14891 instead of testing for equality to INTEGER_TYPE.
14893 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
14895 * config.gcc: Remove uclibc from arc target spec.
14897 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
14899 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
14901 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14904 * lto-wrapper.c (remove_option): New function.
14905 (merge_and_complain): Merge PIC/PIE options more realistically.
14907 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
14909 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
14913 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
14914 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
14915 * config.in: Regenerate.
14916 * configure: Regenerate.
14917 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
14918 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
14919 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
14920 (TARGET_GASISR_PROLOGUES): ...target mask.
14921 * common/config/avr/avr-common.c
14922 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
14923 Set -mgas-isr-prologues.
14924 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
14925 INSERT_PASS_BEFORE for it.
14926 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
14927 * config/avr/avr.c (avr_option_override)
14928 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
14929 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
14930 (avr_attribute_table) <no_gccisr>: Add new function attribute.
14931 (avr_set_current_function) <is_no_gccisr>: Init machine field.
14932 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
14934 (make_avr_pass_pre_proep): New function.
14935 (emit_push_sfr) <treg>: Add argument to function and use it
14936 instead of TMP_REG.
14937 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
14938 and set machine->gasisr.yes.
14939 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
14940 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
14941 __gcc_isr.n_pushed to .L__stack_usage.
14942 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
14943 (avr_asm_final_postscan_insn): ...this new static function.
14944 * config/avr/avr.h (machine_function)
14945 <is_no_gccisr, use_L__stack_usage>: New fields.
14946 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
14947 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
14948 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
14949 (gasisr, *gasisr): New expander and insn.
14950 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
14951 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
14952 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
14954 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
14956 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
14959 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
14961 Move jump-tables out of .text again.
14964 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
14965 (ASM_OUTPUT_ADDR_VEC): New function.
14966 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
14967 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
14968 INSN_ADDRESSes as asm comment.
14969 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
14970 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
14971 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
14972 * config/avr/avr.md (*tablejump): Adjust comment.
14973 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
14974 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
14976 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
14977 (avr_output_addr_vec): New proto.
14978 (avr_log_t) <insn_addresses>: New field.
14980 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
14983 * config/i386/i386.c (ix86_function_arg_advance): Set
14984 outgoing_args_on_stack to true if there are outgoing arguments
14986 (ix86_function_arg): Likewise.
14987 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
14988 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
14989 * config/i386/i386.h (machine_function): Add
14990 outgoing_args_on_stack.
14992 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
14994 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
14995 supporting pthreds.
14996 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
14998 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
15000 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
15001 (REAL_H): Remove $(MACHMODE_H).
15002 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
15004 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
15005 $(MACHMODE_H) and double-int.h.
15006 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
15008 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
15011 2017-07-07 Andrew Pinski <apinski@cavium.com>
15013 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
15014 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
15016 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15018 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
15019 Add warning if GCC was not configured to link against a GLIBC that
15020 exports the hardware capability bits.
15021 (make_resolver_func): Make resolver function private and not a
15022 COMDAT function. Create the name with clone_function_name instead
15023 of make_unique_name.
15026 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
15027 correct operand in doing the split.
15029 2017-07-07 Carl Love <cel@us.ibm.com>
15031 * config/rs6000/rs6000-c: Add support for built-in function
15032 vector unsigned short vec_pack_to_short_fp32 (vector float,
15034 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
15035 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
15036 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
15037 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
15038 (convert_4f32_8i16): Add define_expand.
15039 * doc/extend.texi: Update the built-in documentation file for the
15040 new built-in function.
15042 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15044 * config/sparc/m8.md: New file.
15045 * config/sparc/sparc.md: Include m8.md.
15047 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15049 * config/sparc/sparc.opt: New option -mvis4b.
15050 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
15051 (sparc_option_override): Handle VIS4B.
15052 (enum sparc_builtins): Define
15053 SPARC_BUILTIN_DICTUNPACK{8,16,32},
15054 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
15055 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
15056 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
15057 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
15058 (check_constant_argument): New function.
15059 (sparc_vis_init_builtins): Define builtins
15060 __builtin_vis_dictunpack{8,16,32},
15061 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
15062 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
15063 __builtin_vis_fpcmpde{8,16,32}shl and
15064 __builtin_vis_fpcmpur{8,16,32}shl.
15065 (sparc_expand_builtin): Check that the constant operands to
15066 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
15067 constant and in range.
15068 * config/sparc/sparc-c.c (sparc_target_macros): Handle
15070 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
15071 (SPARC_IMM5_P): Likewise.
15072 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
15073 (enabled): Handle vis4b.
15074 (UNSPEC_DICTUNPACK): New unspec.
15075 (UNSPEC_FPCMPSHL): Likewise.
15076 (UNSPEC_FPUCMPSHL): Likewise.
15077 (UNSPEC_FPCMPDESHL): Likewise.
15078 (UNSPEC_FPCMPURSHL): Likewise.
15079 (cpu_feature): New CPU feature `vis4b'.
15080 (dictunpack{8,16,32}): New insns.
15081 (FPCSMODE): New mode iterator.
15082 (fpcscond): New code iterator.
15083 (fpcsucond): Likewise.
15084 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
15085 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
15086 (fpcmpde{8,16,32}{si,di}shl): Likewise.
15087 (fpcmpur{8,16,32}{si,di}shl): Likewise.
15088 * config/sparc/constraints.md: Define constraints `q' for unsigned
15089 2-bit integer constants and `t' for unsigned 5-bit integer
15091 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
15093 (imm5_operand_dictunpack16): Likewise.
15094 (imm5_operand_dictunpack32): Likewise.
15095 (imm2_operand): Likewise.
15096 * doc/invoke.texi (SPARC Options): Document -mvis4b.
15097 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
15098 ditunpack* and fpcmp*shl builtins.
15100 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15102 * config.gcc: Handle m8 in --with-{cpu,tune} options.
15103 * config.in: Add HAVE_AS_SPARC6 define.
15104 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
15106 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
15108 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
15109 (CPP_CPU_SPEC): Handle m8.
15110 (ASM_CPU_SPEC): Likewise.
15111 * config/sparc/sparc-opts.h (enum processor_type): Add
15113 * config/sparc/sparc.c (m8_costs): New struct.
15114 (sparc_option_override): Handle TARGET_CPU_m8.
15115 (sparc32_initialize_trampoline): Likewise.
15116 (sparc64_initialize_trampoline): Likewise.
15117 (sparc_issue_rate): Likewise.
15118 (sparc_register_move_cost): Likewise.
15119 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
15120 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
15121 (ASM_CPU64_DEFAULT_SPEC): Likewise.
15122 (CPP_CPU_SPEC): Handle M8.
15123 (ASM_CPU_SPEC): Likewise.
15124 (AS_M8_FLAG): Define.
15125 * config/sparc/sparc.md: Add m8 to the cpu attribute.
15126 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
15127 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
15129 * configure: Regenerate.
15130 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
15133 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15135 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
15137 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
15138 ("*movdi_insn_sp32"): Do not set v3pipe.
15139 ("*movsi_insn"): Likewise.
15140 ("*movdi_insn_sp64"): Likewise.
15141 ("*movsf_insn"): Likewise.
15142 ("*movdf_insn_sp32"): Likewise.
15143 ("*movdf_insn_sp64"): Likewise.
15144 ("*zero_extendsidi2_insn_sp64"): Likewise.
15145 ("*sign_extendsidi2_insn"): Likewise.
15146 ("*mov<VM32:mode>_insn"): Likewise.
15147 ("*mov<VM64:mode>_insn_sp64"): Likewise.
15148 ("*mov<VM64:mode>_insn_sp32"): Likewise.
15149 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
15150 ("<vlop:code><VL:mode>3"): Likewise.
15151 ("*not_<vlop:code><VL:mode>3"): Likewise.
15152 ("*nand<VL:mode>_vis"): Likewise.
15153 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
15154 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
15155 ("one_cmpl<VL:mode>2"): Likewise.
15156 ("faligndata<VM64:mode>_vis"): Likewise.
15157 ("alignaddrsi_vis"): Likewise.
15158 ("alignaddrdi_vis"): Likweise.
15159 ("alignaddrlsi_vis"): Likewise.
15160 ("alignaddrldi_vis"): Likewise.
15161 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
15162 ("bmaskdi_vis"): Likewise.
15163 ("bmasksi_vis"): Likewise.
15164 ("bshuffle<VM64:mode>_vis"): Likewise.
15165 ("cmask8<P:mode>_vis"): Likewise.
15166 ("cmask16<P:mode>_vis"): Likewise.
15167 ("cmask32<P:mode>_vis"): Likewise.
15168 ("pdistn<P:mode>_vis"): Likewise.
15169 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
15171 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15173 * config/sparc/sparc.md ("subtype"): New insn attribute.
15174 ("*wrgsr_sp64"): Set insn subtype.
15175 ("*rdgsr_sp64"): Likewise.
15176 ("alignaddrsi_vis"): Likewise.
15177 ("alignaddrdi_vis"): Likewise.
15178 ("alignaddrlsi_vis"): Likewise.
15179 ("alignaddrldi_vis"): Likewise.
15180 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
15181 ("fexpand_vis"): Likewise.
15182 ("fpmerge_vis"): Likewise.
15183 ("faligndata<VM64:mode>_vis"): Likewise.
15184 ("bshuffle<VM64:mode>_vis"): Likewise.
15185 ("cmask8<P:mode>_vis"): Likewise.
15186 ("cmask16<P:mode>_vis"): Likewise.
15187 ("cmask32<P:mode>_vis"): Likewise.
15188 ("fchksm16_vis"): Likewise.
15189 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
15190 ("fmean16_vis"): Likewise.
15191 ("fp<plusminus_insn>64_vis"): Likewise.
15192 ("<plusminus_insn>v8qi3"): Likewise.
15193 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
15194 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
15195 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
15196 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
15197 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
15198 ("*movqi_insn"): Likewise.
15199 ("*movhi_insn"): Likewise.
15200 ("*movsi_insn"): Likewise.
15201 ("movsi_pic_gotdata_op"): Likewise.
15202 ("*movdi_insn_sp32"): Likewise.
15203 ("*movdi_insn_sp64"): Likewise.
15204 ("movdi_pic_gotdata_op"): Likewise.
15205 ("*movsf_insn"): Likewise.
15206 ("*movdf_insn_sp32"): Likewise.
15207 ("*movdf_insn_sp64"): Likewise.
15208 ("*zero_extendhisi2_insn"): Likewise.
15209 ("*zero_extendqihi2_insn"): Likewise.
15210 ("*zero_extendqisi2_insn"): Likewise.
15211 ("*zero_extendqidi2_insn"): Likewise.
15212 ("*zero_extendhidi2_insn"): Likewise.
15213 ("*zero_extendsidi2_insn_sp64"): Likewise.
15214 ("ldfsr"): Likewise.
15215 ("prefetch_64"): Likewise.
15216 ("prefetch_32"): Likewise.
15217 ("tie_ld32"): Likewise.
15218 ("tie_ld64"): Likewise.
15219 ("*tldo_ldub_sp32"): Likewise.
15220 ("*tldo_ldub1_sp32"): Likewise.
15221 ("*tldo_ldub2_sp32"): Likewise.
15222 ("*tldo_ldub_sp64"): Likewise.
15223 ("*tldo_ldub1_sp64"): Likewise.
15224 ("*tldo_ldub2_sp64"): Likewise.
15225 ("*tldo_ldub3_sp64"): Likewise.
15226 ("*tldo_lduh_sp32"): Likewise.
15227 ("*tldo_lduh1_sp32"): Likewise.
15228 ("*tldo_lduh_sp64"): Likewise.
15229 ("*tldo_lduh1_sp64"): Likewise.
15230 ("*tldo_lduh2_sp64"): Likewise.
15231 ("*tldo_lduw_sp32"): Likewise.
15232 ("*tldo_lduw_sp64"): Likewise.
15233 ("*tldo_lduw1_sp64"): Likewise.
15234 ("*tldo_ldx_sp64"): Likewise.
15235 ("*mov<VM32:mode>_insn"): Likewise.
15236 ("*mov<VM64:mode>_insn_sp64"): Likewise.
15237 ("*mov<VM64:mode>_insn_sp32"): Likewise.
15239 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15241 * config/sparc/sparc.md ("type"): New insn type viscmp.
15242 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
15244 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
15245 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
15246 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
15247 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
15249 ("n7_vis_logical_11cycle"): Likewise.
15250 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
15251 * config/sparc/niagara2.md ("niag3_vis": Likewise.
15252 * config/sparc/niagara.md ("niag_vis"): Likewise.
15253 * config/sparc/ultra3.md ("us3_fga"): Likewise.
15254 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
15256 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15258 * config/sparc/sparc.md: New instruction type `bmask'.
15259 (bmaskdi_vis): Use the `bmask' type.
15260 (bmasksi_vis): Likewise.
15261 * config/sparc/ultra3.md (us3_array): Likewise.
15262 * config/sparc/niagara7.md (n7_array): Likewise.
15263 * config/sparc/niagara4.md (n4_array): Likewise.
15264 * config/sparc/niagara2.md (niag2_vis): Likewise.
15265 (niag3_vis): Likewise.
15266 * config/sparc/niagara.md (niag_vis): Likewise.
15268 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15270 * ipa-comdats.c: Remove optimize check from gate.
15271 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
15272 for functions not optimized.
15273 (ipa_fn_summary_read): Skip optimize check.
15274 (ipa_fn_summary_write): Likewise.
15275 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
15277 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
15279 (can_inline_edge_p): Check flag_pcc_struct_return for match.
15280 (check_callers): Give up on caller which is not optimized.
15281 (inline_small_functions): Likewise.
15282 (ipa_inline): Do not give up when not optimizing.
15283 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
15284 away unoptimizes cdtors.
15285 (whole_program_function_and_variable_visibility): Do
15286 ipa_discover_readonly_nonaddressable_vars in LTO mode.
15287 * ipa.c (process_references): Do not check optimize.
15288 (symbol_table::remove_unreachable_nodes): Update optimize check.
15289 (set_writeonly_bit): Update optimize check.
15290 (pass_ipa_cdtor_merge::gate): Do not check optimize.
15291 (pass_ipa_single_use::gate): Remove.
15293 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15295 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
15296 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
15297 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
15298 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
15299 permute_load, permute_store, adjust_extract, adjust_splat,
15300 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
15301 replace_swap_with_copy, dump_swap_insn_table,
15302 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
15303 recombine_lvx_pattern, recombine_stvx_pattern,
15304 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
15305 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
15306 to file rs6000-p8swap.c.
15307 * config/rs6000/rs6000-p8swap.c: New file.
15308 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
15309 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
15310 and rs6000*-*-* targets.
15312 2017-07-06 David Malcolm <dmalcolm@redhat.com>
15314 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
15316 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15318 * lto-wrapper.c (merge_and_complain): Do not merge
15319 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
15320 fsigned_zeros, ftrapping_math, fwrapv.
15321 (append_compiler_options): Do not track these options.
15322 (append_linker_options): Likewie
15324 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15326 * cgraphunit.c (cgraph_node::finalize_function): When
15327 !flag_toplevel_reorde set no_reorder flag.
15328 (varpool_node::finalize_decl): Likewise.
15329 (symbol_table::compile): Drop no toplevel reorder path.
15331 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15333 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
15334 edges; zero probability is not better than uninitialized.
15336 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
15338 * asan.h (asan_sanitize_allocas_p): Declare.
15339 * asan.c (asan_sanitize_allocas_p): New function.
15340 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
15341 (handle_builtin_alloca): Likewise.
15342 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
15343 if !asan_sanitize_allocas_p.
15344 * params.def (asan-instrument-allocas): Add new option.
15345 * params.h (ASAN_PROTECT_ALLOCAS): Define.
15346 * opts.c (common_handle_option): Disable allocas sanitization for
15349 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
15351 * asan.c: Include gimple-fold.h.
15352 (get_last_alloca_addr): New function.
15353 (handle_builtin_stackrestore): Likewise.
15354 (handle_builtin_alloca): Likewise.
15355 (asan_emit_allocas_unpoison): Likewise.
15356 (get_mem_refs_of_builtin_call): Add new parameter, remove const
15357 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
15358 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
15359 (instrument_builtin_call): Pass gimple iterator to
15360 get_mem_refs_of_builtin_call.
15361 (last_alloca_addr): New global.
15362 * asan.h (asan_emit_allocas_unpoison): Declare.
15363 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
15364 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
15365 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
15366 if function calls alloca.
15367 * gimple-fold.c (replace_call_with_value): Remove static keyword.
15368 * gimple-fold.h (replace_call_with_value): Declare.
15369 * internal-fn.c: Include asan.h.
15370 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
15371 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
15373 2017-07-06 David Malcolm <dmalcolm@redhat.com>
15375 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
15376 (C_SELFTEST_FLAGS): New.
15377 (CPP_SELFTEST_FLAGS): New.
15378 (SELFTEST_DEPS): New, from deps of s-selftest.
15379 (C_SELFTEST_DEPS): New, from deps of s-selftest.
15380 (CPP_SELFTEST_DEPS): New.
15381 (selftest): Add dependency on s-selftest-c++.
15382 (s-selftest): Rename to...
15383 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
15384 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
15385 than SELFTEST_FLAGS.
15386 (selftest-gdb): Rename to...
15387 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
15389 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
15390 (selftest-valgrind): Rename to...
15391 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
15393 (selftest-valgrind): Reintroduce as an alias for
15394 selftest-c-valgrind.
15395 (s-selftest-c++): New.
15396 (selftest-c++-gdb): New.
15397 (selftest-c++-valgrind): New.
15399 2017-07-06 Olivier Hainque <hainque@adacore.com>
15401 * gcc.c (process_command): When deciding if undefined variables
15402 should be ignored when processing specs, accept "gcc -v" as well.
15404 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15406 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
15407 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
15409 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
15411 * config/arm/arm-cpus.in (armv8-r): Add new entry.
15412 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
15413 * config/arm/arm-tables.opt: Regenerate.
15414 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
15416 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
15418 2017-07-06 Carl Love <cel@us.ibm.com>
15420 * ChangeLog: Clean up from mid air collision
15422 2017-07-06 Carl Love <cel@us.ibm.com>
15424 * config/rs6000/rs6000-c.c: Add support for built-in functions
15425 vector signed int vec_subc (vector signed int, vector signed int);
15426 vector signed __int128 vec_subc (vector signed __int128,
15427 vector signed __int128);
15428 vector unsigned __int128 vec_subc (vector unsigned __int128,
15429 vector unsigned __int128);
15430 vector signed int vec_sube (vector signed int, vector signed int,
15431 vector signed int);
15432 vector unsigned int vec_sube (vector unsigned int,
15433 vector unsigned int,
15434 vector unsigned int);
15435 vector signed __int128 vec_sube (vector signed __int128,
15436 vector signed __int128,
15437 vector signed__int128);
15438 vector unsigned __int128 vec_sube (vector unsigned __int128,
15439 vector unsigned __int128,
15440 vector unsigned __int128);
15441 vector signed int vec_subec (vector signed int, vector signed int,
15442 vector signed int);
15443 vector unsigned int vec_subec (vector unsigned int,
15444 vector unsigned int,
15445 vector unsigned int);
15446 vector signed __int128 vec_subec (vector signed __int128,
15447 vector signed __int128,
15448 vector signed__int128);
15449 vector unsigned __int128 vec_subec (vector unsigned __int128,
15450 vector unsigned __int128,
15451 vector unsigned __int128);
15452 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
15453 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
15454 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
15455 BU_ALTIVEC_OVERLOAD_X definitions.
15456 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
15457 * doc/extend.texi: Update the built-in documentation file for the new
15458 built-in functions.
15460 2017-07-06 David Malcolm <dmalcolm@redhat.com>
15463 * diagnostic-show-locus.c (layout::layout): Use start and finish
15464 spelling location for the start and finish of each range.
15465 * genmatch.c (linemap_client_expand_location_to_spelling_point):
15466 Add unused aspect param.
15467 * input.c (expand_location_1): Add "aspect" param, and use it
15468 to access the correct part of the location.
15469 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
15471 (expand_location_to_spelling_point): Likewise.
15472 (linemap_client_expand_location_to_spelling_point): Add "aspect"
15473 param, and pass it to expand_location_1.
15475 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
15477 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
15478 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
15479 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
15480 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
15481 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
15482 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
15483 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
15484 _mm_maskz_getmant_ss): New intrinsics.
15485 (__builtin_ia32_getexpss128_mask): Changed to ...
15486 __builtin_ia32_getexpss128_round ... this.
15487 (__builtin_ia32_getexpsd128_mask): Changed to ...
15488 __builtin_ia32_getexpsd128_round ... this.
15489 * config/i386/i386-builtin-types.def
15490 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
15491 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
15492 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
15493 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
15494 __builtin_ia32_getmantss_mask_round): New builtins.
15495 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
15496 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
15497 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
15498 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
15499 * config/i386/sse.md
15500 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
15501 avx512f_sgetexp<mode><mask_scalar_name>
15502 <round_saeonly_scalar_name> ... this.
15503 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
15504 %0, %1, %2<round_saeonly_op3>}): Changed to ...
15505 vgetexp<ssescalarmodesuffix>
15506 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
15507 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
15508 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
15509 avx512f_vgetmant<mode><mask_scalar_name>
15510 <round_saeonly_scalar_name> ... this.
15511 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
15512 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
15513 vgetmant<ssescalarmodesuffix>
15514 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
15515 %0<mask_scalar_operand4>, %1, %2
15516 <round_saeonly_scalar_mask_op4>, %3} ... this.
15517 * config/i386/subst.md (mask_scalar_operand4,
15518 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
15519 round_saeonly_scalar_nimm_predicate): New subst attributes.
15521 2017-07-06 Julia Koval <julia.koval@intel.com>
15523 * config/i386/i386.c (ix86_erase_embedded_rounding):
15524 Remove code for old rounding pattern.
15526 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
15528 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
15530 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
15532 * doc/sourcebuild.texi (Test Directives, Variants of
15533 dg-require-support): Add documentation for dg-require-stack-check.
15535 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
15537 * config/i386/subst.md (mask_scalar, round_scalar,
15538 round_saeonly_scalar): New meta-templates.
15539 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
15540 round_scalar_mask_operand3, round_scalar_mask_op3,
15541 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
15542 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
15543 round_saeonly_scalar_constraint,
15544 round_saeonly_scalar_prefix): New subst attribute.
15545 * config/i386/sse.md
15546 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
15547 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
15548 <round_scalar_name> ... this.
15549 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
15550 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
15551 <round_scalar_name> ... this.
15552 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
15553 <sse>_vm<code><mode>3<mask_scalar_name>
15554 <round_saeonly_scalar_name> ... this.
15555 (v<plusminus_mnemonic><ssescalarmodesuffix>
15556 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
15557 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
15558 v<plusminus_mnemonic><ssescalarmodesuffix>
15559 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
15560 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
15561 (v<multdiv_mnemonic><ssescalarmodesuffix>
15562 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
15563 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
15564 v<multdiv_mnemonic><ssescalarmodesuffix>
15565 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
15566 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
15567 (v<maxmin_float><ssescalarmodesuffix>
15568 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
15569 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
15570 v<maxmin_float><ssescalarmodesuffix>
15571 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
15572 %0<mask_scalar_operand3>, %1, %<iptr>2
15573 <round_saeonly_scalar_mask_op3>} ... this.
15575 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
15577 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
15578 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
15580 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
15581 Alan Hayward <alan.hayward@arm.com>
15582 David Sherwood <david.sherwood@arm.com>
15584 * combine.c (simplify_if_then_else): Remove "enum" before
15586 * compare-elim.c (can_eliminate_compare): Likewise.
15587 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
15589 (aarch64_lookup_simd_builtin_type): Likewise.
15590 (aarch64_simd_builtin_type): Likewise.
15591 (aarch64_init_simd_builtin_types): Likewise.
15592 (aarch64_simd_expand_args): Likewise.
15593 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
15595 (aarch64_reverse_mask): Likewise.
15596 (aarch64_simd_emit_reg_reg_move): Likewise.
15597 (aarch64_gen_adjusted_ldpstp): Likewise.
15598 (aarch64_ccmp_mode_to_code): Likewise.
15599 (aarch64_operands_ok_for_ldpstp): Likewise.
15600 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15601 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
15603 (aarch64_min_divisions_for_recip_mul): Likewise.
15604 (aarch64_reassociation_width): Likewise.
15605 (aarch64_get_condition_code_1): Likewise.
15606 (aarch64_simd_emit_reg_reg_move): Likewise.
15607 (aarch64_simd_attr_length_rglist): Likewise.
15608 (aarch64_reverse_mask): Likewise.
15609 (aarch64_operands_ok_for_ldpstp): Likewise.
15610 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15611 (aarch64_gen_adjusted_ldpstp): Likewise.
15612 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
15614 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
15615 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
15616 (arm_lookup_simd_builtin_type): Likewise.
15617 (arm_simd_builtin_type): Likewise.
15618 (arm_init_simd_builtin_types): Likewise.
15619 (arm_expand_builtin_args): Likewise.
15620 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
15621 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
15622 (ft32_setup_incoming_varargs): Likewise.
15623 (ft32_function_arg): Likewise.
15624 (ft32_function_arg_advance): Likewise.
15625 (ft32_pass_by_reference): Likewise.
15626 (ft32_arg_partial_bytes): Likewise.
15627 (ft32_valid_pointer_mode): Likewise.
15628 (ft32_addr_space_pointer_mode): Likewise.
15629 (ft32_addr_space_legitimate_address_p): Likewise.
15630 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
15632 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
15633 (ix86_emit_outlined_ms2sysv_restore): Likewise.
15634 (iamcu_alignment): Likewise.
15635 (canonicalize_vector_int_perm): Likewise.
15636 (ix86_noce_conversion_profitable_p): Likewise.
15637 (ix86_mpx_bound_mode): Likewise.
15638 (ix86_operands_ok_for_move_multiple): Likewise.
15639 * config/microblaze/microblaze-protos.h
15640 (microblaze_expand_conditional_branch_reg): Likewise.
15641 * config/microblaze/microblaze.c
15642 (microblaze_expand_conditional_branch_reg): Likewise.
15643 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
15645 (rs6000_reassociation_width): Likewise.
15646 (rs6000_invalid_binary_op): Likewise.
15647 (fusion_p9_p): Likewise.
15648 (emit_fusion_p9_load): Likewise.
15649 (emit_fusion_p9_store): Likewise.
15650 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
15652 (riscv_hard_regno_mode_ok_p): Likewise.
15653 (riscv_address_insns): Likewise.
15654 (riscv_split_symbol): Likewise.
15655 (riscv_legitimize_move): Likewise.
15656 (riscv_function_value): Likewise.
15657 (riscv_hard_regno_nregs): Likewise.
15658 (riscv_expand_builtin): Likewise.
15659 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
15660 (riscv_build_integer): Likewise.
15661 (riscv_split_integer): Likewise.
15662 (riscv_legitimate_constant_p): Likewise.
15663 (riscv_cannot_force_const_mem): Likewise.
15664 (riscv_regno_mode_ok_for_base_p): Likewise.
15665 (riscv_valid_base_register_p): Likewise.
15666 (riscv_valid_offset_p): Likewise.
15667 (riscv_valid_lo_sum_p): Likewise.
15668 (riscv_classify_address): Likewise.
15669 (riscv_legitimate_address_p): Likewise.
15670 (riscv_address_insns): Likewise.
15671 (riscv_load_store_insns): Likewise.
15672 (riscv_force_binary): Likewise.
15673 (riscv_split_symbol): Likewise.
15674 (riscv_force_address): Likewise.
15675 (riscv_legitimize_address): Likewise.
15676 (riscv_move_integer): Likewise.
15677 (riscv_legitimize_const_move): Likewise.
15678 (riscv_legitimize_move): Likewise.
15679 (riscv_address_cost): Likewise.
15680 (riscv_subword): Likewise.
15681 (riscv_output_move): Likewise.
15682 (riscv_canonicalize_int_order_test): Likewise.
15683 (riscv_emit_int_order_test): Likewise.
15684 (riscv_function_arg_boundary): Likewise.
15685 (riscv_pass_mode_in_fpr_p): Likewise.
15686 (riscv_pass_fpr_single): Likewise.
15687 (riscv_pass_fpr_pair): Likewise.
15688 (riscv_get_arg_info): Likewise.
15689 (riscv_function_arg): Likewise.
15690 (riscv_function_arg_advance): Likewise.
15691 (riscv_arg_partial_bytes): Likewise.
15692 (riscv_function_value): Likewise.
15693 (riscv_pass_by_reference): Likewise.
15694 (riscv_setup_incoming_varargs): Likewise.
15695 (riscv_print_operand): Likewise.
15696 (riscv_elf_select_rtx_section): Likewise.
15697 (riscv_save_restore_reg): Likewise.
15698 (riscv_for_each_saved_reg): Likewise.
15699 (riscv_register_move_cost): Likewise.
15700 (riscv_hard_regno_mode_ok_p): Likewise.
15701 (riscv_hard_regno_nregs): Likewise.
15702 (riscv_class_max_nregs): Likewise.
15703 (riscv_memory_move_cost): Likewise.
15704 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
15705 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
15706 (rl78_addr_space_address_mode): Likewise.
15707 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15709 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
15710 (rs6000_reassociation_width): Likewise.
15711 (rs6000_invalid_binary_op): Likewise.
15712 (fusion_p9_p): Likewise.
15713 (emit_fusion_p9_load): Likewise.
15714 (emit_fusion_p9_store): Likewise.
15715 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
15716 (ok_for_simple_move_operands): Likewise.
15717 (ok_for_simple_move_strict_operands): Likewise.
15718 (ok_for_simple_arith_logic_operands): Likewise.
15719 (visium_legitimize_reload_address): Likewise.
15720 (visium_select_cc_mode): Likewise.
15721 (output_cbranch): Likewise.
15722 (visium_split_double_move): Likewise.
15723 (visium_expand_copysign): Likewise.
15724 (visium_expand_int_cstore): Likewise.
15725 (visium_expand_fp_cstore): Likewise.
15726 * config/visium/visium.c (visium_pass_by_reference): Likewise.
15727 (visium_function_arg): Likewise.
15728 (visium_function_arg_advance): Likewise.
15729 (visium_libcall_value): Likewise.
15730 (visium_setup_incoming_varargs): Likewise.
15731 (visium_legitimate_constant_p): Likewise.
15732 (visium_legitimate_address_p): Likewise.
15733 (visium_legitimize_address): Likewise.
15734 (visium_secondary_reload): Likewise.
15735 (visium_register_move_cost): Likewise.
15736 (visium_memory_move_cost): Likewise.
15737 (prepare_move_operands): Likewise.
15738 (ok_for_simple_move_operands): Likewise.
15739 (ok_for_simple_move_strict_operands): Likewise.
15740 (ok_for_simple_arith_logic_operands): Likewise.
15741 (visium_function_value_1): Likewise.
15742 (rtx_ok_for_offset_p): Likewise.
15743 (visium_legitimize_reload_address): Likewise.
15744 (visium_split_double_move): Likewise.
15745 (visium_expand_copysign): Likewise.
15746 (visium_expand_int_cstore): Likewise.
15747 (visium_expand_fp_cstore): Likewise.
15748 (visium_split_cstore): Likewise.
15749 (visium_select_cc_mode): Likewise.
15750 (visium_split_cbranch): Likewise.
15751 (output_cbranch): Likewise.
15752 (visium_print_operand_address): Likewise.
15753 * expmed.c (flip_storage_order): Likewise.
15754 * expmed.h (emit_cstore): Likewise.
15755 (flip_storage_order): Likewise.
15756 * genrecog.c (validate_pattern): Likewise.
15757 * hsa-gen.c (gen_hsa_addr): Likewise.
15758 * internal-fn.c (expand_arith_overflow): Likewise.
15759 * ira-color.c (allocno_copy_cost_saving): Likewise.
15760 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15761 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
15762 (process_invariant_for_inheritance): Likewise.
15763 * lra-eliminations.c (move_plus_up): Likewise.
15764 * omp-low.c (lower_oacc_reductions): Likewise.
15765 * simplify-rtx.c (simplify_subreg): Likewise.
15766 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
15767 (TARGET_CHKP_BOUND_MODE): Likewise..
15768 * targhooks.c (default_chkp_bound_mode): Likewise.
15769 (default_setup_incoming_vararg_bounds): Likewise.
15770 * targhooks.h (default_chkp_bound_mode): Likewise.
15771 (default_setup_incoming_vararg_bounds): Likewise.
15772 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
15773 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
15774 (have_whole_vector_shift): Likewise.
15775 * tree-vect-stmts.c (vectorizable_load): Likewise.
15776 * doc/tm.texi: Regenerate.
15778 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
15780 Graceful degrade if Binutils PR21472 is not available.
15783 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
15784 .rodata in flash test fails.
15785 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
15786 * confgure: Regenerate.
15787 * config.in: Regenerate.
15788 * config/avr/avr.c (avr_asm_named_section)
15789 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
15790 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
15791 (avr_asm_init_sections): Same.
15793 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15795 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
15796 (fma<VH:mode>4_intrinsic): Likewise.
15797 (*fmsub<VCVTF:mode>4): Likewise.
15798 (*fmsub<VH:mode>4_intrinsic): Likewise.
15800 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
15803 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
15804 Don't depend on "optimize > 0".
15805 (out_movhi_r_mr, out_movqi_mr_r): Same.
15806 (out_movhi_mr_r, out_movqi_r_mr): Same.
15807 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
15808 io_address_operand on "optimize > 0".
15810 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15812 * tree-loop-distribution.c: Add general explanantion on the pass.
15813 (generate_loops_for_partition): Mark distributed loop.
15814 (pg_add_dependence_edges): New parameter. Handle alias data
15815 dependence specially and record it in the parameter if asked.
15816 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
15817 (init_partition_graph_vertices, add_partition_graph_edge): New.
15818 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
15819 (free_partition_graph_vdata, build_partition_graph): New.
15820 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
15821 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
15822 (data_ref_segment_size, latch_dominated_by_data_ref): New.
15823 (compute_alias_check_pairs, version_loop_by_alias_check): New.
15824 (version_for_distribution_p, finalize_partitions): New.
15825 (distribute_loop): Handle alias data dependence specially. Factor
15826 out loop fusion code as functions and call these functions.
15828 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15830 * tree-loop-distribution.c (classify_partition): New parameter and
15831 better handle reduction statement.
15832 (rdg_build_partitions): Revise comment.
15833 (distribute_loop): Compute statements in all partitions and pass it
15834 to classify_partition.
15836 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15838 * tree-loop-distribution.c (enum partition_type): New.
15839 (struct partition): New field type.
15840 (partition_merge_into): Add parameter. Update partition type.
15841 (data_dep_in_cycle_p, update_type_for_merge): New functions.
15842 (build_rdg_partition_for_vertex): Compute partition type.
15843 (rdg_build_partitions): Dump partition type.
15844 (distribute_loop): Update calls to partition_merge_into.
15846 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15848 * tree-loop-distribution.c (struct ddr_hasher): New.
15849 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
15851 (classify_partition): Call get_data_dependence.
15852 (pg_add_dependence_edges): Ditto.
15853 (distribute_loop): Release data dependence hash table.
15855 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15857 * tree-loop-distribution.c (ref_base_address): Delete.
15858 (similar_memory_accesses): Rename ...
15859 (share_memory_accesses): ... to this. Check if partitions access
15860 the same memory reference.
15861 (distribute_loop): Call share_memory_accesses.
15863 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15865 * tree-loop-distribution.c (struct partition): New field recording
15866 its data reference.
15867 (partition_alloc, partition_free): Init and release data refs.
15868 (partition_merge_into): Merge data refs.
15869 (build_rdg_partition_for_vertex): Collect data refs for partition.
15870 (pg_add_dependence_edges): Change parameters from vector to bitmap.
15872 (distribute_loop): Remve data refs from vertice data of partition
15875 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15877 * tree-loop-distribution.c (params.h): Include header file.
15878 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
15879 (datarefs_vec): New global var.
15880 (create_rdg_vertices): Use datarefs_vec directly.
15881 (free_rdg): Don't free data references.
15882 (build_rdg): Update use. Don't free data references.
15883 (distribute_loop): Compute global variable for data references.
15884 Bail out if there are too many data references.
15886 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15888 * tree-loop-distribution.c (loop_nest): New global var.
15889 (build_rdg): Use loop directly, rather than loop nest.
15890 (pg_add_dependence_edges): Remove loop nest parameter. Use global
15892 (distribute_loop): Compute global variable loop nest. Update use.
15894 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15896 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
15897 (partition_merge_into): New parameter. Dump reason for fusion.
15898 (distribute_loop): Update use of partition_merge_into.
15900 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15902 * tree-loop-distribution.c (bb_top_order_index): New.
15903 (bb_top_order_index_size, bb_top_order_cmp): New.
15904 (stmts_from_loop): Use topological order.
15905 (pass_loop_distribution::execute): Compute and release topological
15906 order for basic blocks.
15908 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15910 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
15913 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15915 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
15916 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
15917 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
15918 * internal-fn.def (LOOP_DIST_ALIAS): New.
15919 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
15920 (fold_loop_internal_call): ... this.
15921 (vect_loop_dist_alias_call): New function.
15922 (set_uid_loop_bbs): Call fold_loop_internal_call.
15923 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
15926 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
15929 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
15930 Require dead FLAGS_REG at the beginning of a peephole.
15932 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
15935 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
15936 arguments in the call to __builtin_ia32_sbb_u32.
15937 (_subborrow_u64): Swap _X and _Y arguments in the call to
15938 __builtin_ia32_sbb_u64.
15940 2017-07-04 Jakub Jelinek <jakub@redhat.com>
15943 * tree-vrp.c (compare_assert_loc): Turn into a function template
15944 with stable template parameter. Only test if a->e is NULL,
15945 !a->e == !b->e has been verified already. Use e == NULL or
15946 e != NULL instead of e or ! e tests. If stable is true, don't use
15947 iterative_hash_expr, on the other side allow a or b or both NULL
15948 and sort the NULLs last.
15949 (process_assert_insertions): Sort using compare_assert_loc<false>
15950 instead of compare_assert_loc, later sort using
15951 compare_assert_loc<true> before calling process_assert_insertions_for
15952 in a loop. Use break instead of continue once seen NULL pointer.
15954 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
15956 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
15957 Cortex-R7 and Cortex-R8 processors.
15959 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
15961 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
15962 uninitialized while src is not.
15964 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
15966 * common/config/arm/arm-common.c: Adjust include path for
15968 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
15969 (arm-cpu.h): Create in build directory. Adjust dependency rules.
15970 (arm-cpu-data.h): Likewise.
15971 (arm-cpu-cdata.h): Likewise.
15972 * config/arm/arm-cpu.h: Delete.
15973 * config/arm/arm-cpu-cdata.h: Delete.
15974 * config/arm/arm-cpu-data.h: Delete.
15976 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
15978 * config/arm/arm-cpus.in (cortex-a55): New.
15979 (cortex-a75): Likewise.
15980 (cortex-a75.cortex-a55): Likewise.
15981 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
15983 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
15984 * config/arm/arm-cpu-cdata.h: Regenerate.
15985 * config/arm/arm-cpu-data.h: Regenerate.
15986 * config/arm/arm-cpu.h: Regenerate.
15987 * config/arm/arm-tables.opt: Regenerate.
15988 * config/arm/arm-tune.md: Regenerate.
15990 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
15992 * haifa-sched.c (sched_create_recovery_edges): Update profile.
15994 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
15996 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
15999 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
16001 PR tree-optimization/81292
16002 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
16003 full_string_p, also call adjust_related_strinfos if the adjustment
16004 is simple, otherwise invalidate related strinfos.
16006 2017-07-04 Martin Liska <mliska@suse.cz>
16009 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
16010 newly created variable as DECL_IGNORED_P.
16012 2017-07-04 Martin Liska <mliska@suse.cz>
16015 * ipa-inline.c (inline_small_functions):
16016 Use xstrdup_for_dump.
16018 2017-07-04 Tom de Vries <tom@codesourcery.com>
16020 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
16022 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
16025 * config/darwin.c (darwin_function_switched_text_sections):
16028 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
16030 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
16032 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
16034 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
16036 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16038 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
16039 min_profitable_iters, and th as inclusive lower bounds.
16040 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
16041 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
16042 for min_profitable_iters and min_profitable_estimate.
16043 (vect_transform_loop): Treat th as an inclusive lower bound.
16044 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
16046 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
16049 * config/darwin.c (darwin_function_switched_text_sections):
16050 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
16051 in two pieces, and suppress the use of buf.
16053 2017-07-03 Nathan Sidwell <nathan@acm.org>
16055 * hash-table.h (hash_table_mod1): Fix indentation.
16057 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16059 PR middle-end/81290
16060 * predict.c (force_edge_cold): Be more careful about propagation
16062 * profile-count.h (profile_probability::guessed,
16063 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
16065 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
16067 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
16069 * doc/invoke.texi (rcpc architecture extension): Document it.
16071 2017-07-03 Richard Biener <rguenther@suse.de>
16073 PR tree-optimization/60510
16074 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
16075 the scalar reduction PHI and use it.
16076 (vectorizable_reduction): Properly guard the single_defuse_cycle
16077 path for non-SLP reduction chains where we cannot use it.
16078 Rework reduc_def/index and vector type deduction. Rework
16079 vector operand gathering during reduction op code-gen.
16080 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
16081 chains dissolve the chain and leave it to non-SLP reduction
16084 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16086 * tree-data-ref.h (dr_alignment): Declare.
16087 * tree-data-ref.c (dr_alignment): New function.
16088 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
16089 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
16091 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
16093 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16095 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
16096 and base_misalignment fields.
16097 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
16098 * tree-data-ref.c: Include builtins.h.
16099 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
16100 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
16101 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
16102 * tree-vect-data-refs.c: Include tree-cfg.h.
16103 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
16104 fields instead of calculating an alignment here.
16105 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
16106 innermost_loop_behavior fields.
16108 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16110 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
16112 (DR_STEP_ALIGNMENT): New macro.
16113 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
16114 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
16115 (create_data_ref): Print it.
16116 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
16117 to tell whether the step preserves vector (mis)alignment.
16118 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
16119 Move the check for an integer step and generalise to all INTEGER_CST.
16120 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
16121 Print the outer step alignment.
16123 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16125 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
16126 with offset_alignment.
16127 (DR_ALIGNED_TO): Delete.
16128 (DR_OFFSET_ALIGNMENT): New macro.
16129 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
16130 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
16131 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
16132 (create_data_ref): Likewise.
16133 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
16134 (vect_analyze_data_refs): Likewise.
16135 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
16136 creating dummy innermost behavior.
16138 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16140 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
16141 with a "innermost_loop_behavior *" and refeence tree.
16142 * tree-data-ref.c (dr_analyze_innermost): Likewise.
16143 (create_data_ref): Update call accordingly.
16144 * tree-predcom.c (find_looparound_phi): Likewise.
16146 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16148 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
16149 fields with dr_wrt_vec_loop.
16150 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
16151 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
16152 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
16153 (vect_dr_behavior): New function.
16154 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
16155 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
16156 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
16157 track whether the step preserves the misalignment.
16158 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
16159 Use vect_dr_behavior.
16160 (vect_setup_realignment): Update call accordingly.
16161 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
16162 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
16163 call to vect_create_addr_base_for_vector_ref.
16164 (vect_create_cond_for_align_checks): Likewise.
16165 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
16166 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
16167 (vect_recog_mask_conversion_pattern): Likewise.
16168 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
16169 (new_stmt_vec_info): Remove redundant zeroing.
16171 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
16173 * common/config/arm/arm-common.c (arm_be8_option): New function.
16174 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
16175 (ISA_ARMv6): Add isa_bit_be8.
16176 * config/arm/arm.h (arm_be8_option): Add prototype.
16177 (BE8_SPEC_FUNCTION): New define.
16178 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
16179 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
16180 (mlittle-endian): Similarly.
16181 (mbe8, mbe32): New options.
16182 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
16183 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
16185 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16187 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
16189 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16191 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
16192 (cleanup_tree_cfg_bb): Use it.
16193 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
16195 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
16197 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16200 * loop-doloop.c (add_test): Update profile.
16202 2017-07-03 Martin Liska <mliska@suse.cz>
16205 * sanopt.c (rewrite_usage_of_param): New function.
16206 (sanitize_rewrite_addressable_params): Likewise.
16207 (pass_sanopt::execute): Call rewrite_usage_of_param.
16209 2017-07-03 Richard Biener <rguenther@suse.de>
16211 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
16212 back to using VIEW_CONVERT_EXPR.
16214 2017-07-03 Martin Liska <mliska@suse.cz>
16217 * doc/extend.texi: Document when a resolver function is
16218 generated for target_clones.
16220 2017-07-03 Martin Liska <mliska@suse.cz>
16222 * asan.c (asan_emit_stack_protection): Unpoison just red zones
16223 and shadow memory of auto variables which are subject of
16224 use-after-scope sanitization.
16225 (asan_expand_mark_ifn): Add do set only when is_poison.
16227 2016-07-03 Richard Biener <rguenther@suse.de>
16229 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
16231 (vect_force_simple_reduction): Record reduction def -> phi mapping.
16232 (vectorizable_reduction): Perform reduction PHI creation when
16233 visiting a reduction PHI and adjust and simplify code generation
16234 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
16235 (vect_transform_loop): Visit reduction PHIs.
16236 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
16237 defs into the SLP tree.
16238 (vect_build_slp_tree): Reduction defs terminate the recursion.
16239 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
16241 (vect_get_vec_defs_for_stmt_copy): Export.
16242 (vect_get_vec_defs): Likewise.
16243 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
16245 (vect_get_vec_defs_for_stmt_copy): Declare.
16246 (vect_get_vec_defs): Likewise.
16248 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16250 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
16251 parameter with a "loop" parameter and use it instead of the
16252 loop containing DR_STMT. Don't check simple_iv when doing
16253 BB analysis. Describe the two analysis modes in the comment.
16255 2017-07-03 Tom de Vries <tom@codesourcery.com>
16257 PR tree-optimization/69468
16258 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
16259 (find_same_succ_bb): Handle ignore_edge_flags.
16261 2017-07-03 Tom de Vries <tom@codesourcery.com>
16263 PR tree-optimization/81192
16264 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
16266 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
16268 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
16270 2017-07-03 Tom de Vries <tom@codesourcery.com>
16272 PR tree-optimization/81192
16273 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
16274 BB_SAME_SUCC (bb) == NULL.
16276 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16278 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
16281 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16283 * dumpfile.c: Include profile-count.h
16284 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
16286 (insert_cond_bb): Update profile.
16287 * tree-cfg.h (insert_cond_bb): Update prototype.
16288 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
16289 * tree-dump.c: Do not include tree-cfg.
16291 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16293 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
16295 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16297 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
16300 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16302 * tree-complex.c (expand_complex_div_wide): update profile.
16304 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16305 Alan Hayward <alan.hayward@arm.com>
16306 David Sherwood <david.sherwood@arm.com>
16308 * Makefile.in (MACHMODE_H): Remove insn-modes.h
16309 (CORETYPES_H): New define.
16310 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
16311 (insn-modes-inline.h, s-modes-inline-h): New rules.
16312 (generated_files): Add insn-modes-inline.h.
16313 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
16314 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
16315 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
16316 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
16317 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
16318 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
16319 (build/gencodes.o, build/genconditions.o): Likewise.
16320 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
16321 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
16322 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
16323 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
16324 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
16325 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
16326 * coretypes.h: Include everything up to real.h for generators.
16327 Include insn-modes.h first. Include wide-int-print.h after
16328 wide-int.h. Include insn-modes-inline.h and then machmode.h.
16329 * machmode.h: Don't include insn-modes.h here.
16330 * function-tests.c: Remove includes of signop.h, machmode.h,
16331 double-int.h and wide-int.h.
16333 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
16335 * optc-save-gen.awk: Likewise.
16336 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
16337 * godump.c: Remove include of wide-int-print.h.
16338 * pretty-print.h: Likewise.
16339 * wide-int-print.cc: Likewise.
16340 * wide-int.cc: Likewise.
16341 * hash-map-tests.c: Remove include of signop.h.
16342 * hash-set-tests.c: Likewise.
16343 * rtl-tests.c: Likewise.
16344 * mkconfig.sh: Remove include of machmode.h.
16345 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
16347 (emit_insn_modes_inline_h): ...this new function. Emit the code
16348 into an insn-modes-inline.h header file, adding appropriate
16349 include guards and end comments.
16350 (emit_insn_modes_c_header): Remove include of machmode.h.
16351 (emit_min_insn_modes_c_header): Include coretypes.h rather than
16353 (main): Handle -i flag and call emit_insn_modes_inline_h when
16356 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16358 * tree-ssa-strlen.c (strinfo): Rename the length field to
16359 nonzero_chars. Add a full_string_p field.
16360 (compare_nonzero_chars, zero_length_string_p): New functions.
16361 (get_addr_stridx): Add an offset_out parameter.
16362 Use compare_nonzero_chars.
16363 (get_stridx): Update accordingly. Use compare_nonzero_chars.
16364 (new_strinfo): Update after above changes to strinfo.
16365 (set_endptr_and_length): Set full_string_p.
16366 (get_string_length): Update after above changes to strinfo.
16367 (unshare_strinfo): Update call to new_strinfo.
16368 (maybe_invalidate): Likewise.
16369 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
16370 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
16371 as a uhwi instead of an shwi. Update after above changes to
16372 strinfo and new_strinfo.
16373 (zero_length_string): Assert that chainsi contains full strings.
16374 Use zero_length_string_p. Update call to new_strinfo.
16375 (adjust_related_strinfos): Update after above changes to strinfo.
16376 Copy full_string_p from origsi.
16377 (adjust_last_stmt): Use zero_length_string_p.
16378 (handle_builtin_strlen): Update after above changes to strinfo and
16379 new_strinfo. Install the lhs as the string length if the previous
16380 entry didn't describe a full string.
16381 (handle_builtin_strchr): Update after above changes to strinfo
16383 (handle_builtin_strcpy): Likewise.
16384 (handle_builtin_strcat): Likewise.
16385 (handle_builtin_malloc): Likewise.
16386 (handle_pointer_plus): Likewise.
16387 (handle_builtin_memcpy): Likewise. Track nonzero characters
16388 that aren't necessarily followed by a nul terminator.
16389 (handle_char_store): Likewise.
16391 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16393 PR tree-optimization/80769
16394 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
16395 for malloc and calloc. Document the new invariant that all related
16396 strinfos have delayed lengths or none do.
16397 (verify_related_strinfos): Move earlier in file.
16398 (set_endptr_and_length): New function, split out from...
16399 (get_string_length): ...here. Also set the lengths of related
16401 (zero_length_string): Assert that chainsi has known (rather than
16403 (adjust_related_strinfos): Likewise.
16405 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16407 PR tree-optimization/81136
16408 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
16409 assert that two references with the same misalignment have the same
16410 compile-time misalignment if those compile-time misalignments
16413 2017-07-01 Andi Kleen <ak@linux.intel.com>
16415 * print-tree.c (print_node): Print all attributes.
16417 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16419 * cfg.c (scale_bbs_frequencies): New function.
16420 * cfg.h (scale_bbs_frequencies): Declare it.
16421 * cfgloopanal.c (single_likely_exit): Cleanup.
16422 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
16424 (scale_loop_profile): Likewise.
16425 (loop_version): Likewise.
16426 (create_empty_loop_on_edge): Update.
16427 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
16428 scale_loop_frequencies, scale_loop_profile, loopify,
16429 loop_version): Update prototypes.
16430 * modulo-sched.c (sms_schedule): Update.
16431 * predict.c (unlikely_executed_edge_p): Also check probability.
16432 (probably_never_executed_edge_p): Fix typo.
16433 * tree-if-conv.c (version_loop_for_if_conversion): Update.
16434 * tree-parloops.c (gen_parallel_loop): Update.
16435 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
16436 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
16437 * tree-ssa-loop-split.c (split_loop): Update.
16438 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
16439 * tree-vect-loop-manip.c (vect_do_peeling): Update.
16440 (vect_loop_versioning): Update.
16441 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
16443 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16445 * trans-mem.c (split_bb_make_tm_edge): Update profile.
16447 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16449 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
16450 to keep profile consistent.
16452 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16454 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
16455 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
16456 * profile-count.h (max_safe_multiplier): Make unsigned.
16457 (profile_count::guessed_zero): New.
16459 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16461 * bb-reorder.c (fix_up_crossing_landing_pad,
16462 fix_crossing_conditional_branches): Use make_single_succ_edge
16463 to keep profile consistent.
16465 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16467 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
16470 2017-07-01 Jakub Jelinek <jakub@redhat.com>
16473 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
16474 the right scopes, make sure cond_jump isn't preserved between multiple
16475 iterations. Search for fallthru edge whenever there are 3+ edges and
16476 use find_fallthru_edge for it.
16478 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16480 Patch by Alexander Monakov <amonakov@ispras.ru>
16481 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
16482 probabilities consistently.
16484 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16486 * pa.c (pa_expand_compare_and_swap_loop): Update call of
16487 emit_cmp_and_jump_insns.
16489 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16492 * tree-inline.c (expand_call_inline): Combine profile statuses.
16494 2017-06-30 Andrew Pinski <apinski@cavium.com>
16496 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
16497 fold_stmt returned true.
16499 2017-06-30 Nathan Sidwell <nathan@acm.org>
16501 * ggc.h (empty_string): Delete.
16502 * cfgexpand.c (expand_asm_stmt): Use plain "".
16503 * optabs.c (expand_asm_memory_barrier): Likewise.
16504 * stringpool.c (empty_string): Delete.
16505 (digit_vector, digit_string): Delete.
16506 (ggc_alloc_string): Use plain "", don't optimize single digit
16507 strings. Use ggc_alloc_atomic.
16509 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
16511 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
16512 comparison set and one other set, use the cost of the non-comparison
16515 2017-06-30 Nathan Sidwell <nathan@acm.org>
16517 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
16520 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
16522 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
16523 loops. Remove now unneeded calls to gimple_switch_set_label() that
16524 just set removed labels to NULL_TREE.
16526 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
16528 * tree-ssanames.c (set_range_info_raw): Abstract from ...
16529 (set_range_info): ...here. Only call set_range_info_raw if domain
16531 (set_nonzero_bits): Call set_range_info_raw.
16532 * tree-ssanames.h (set_range_info_raw): New.
16534 2017-06-30 Jakub Jelinek <jakub@redhat.com>
16537 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
16538 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
16539 of nonimmediate_operand and <store_mask_constraint> instead of m
16540 for the input operand. For V8FI iterator, always split if input
16541 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
16542 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
16543 <store_mask_predicate> instead of register_operand and
16544 <store_mask_constraint> instead of v for the input operand. Make
16545 sure both operands aren't MEMs for if not <mask_applied>.
16547 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
16549 * lto-wrapper.c (copy_file) Close both file descriptors before
16552 2017-06-30 Martin Liska <mliska@suse.cz>
16555 * multiple_target.c (create_dispatcher_calls): Make ifunc
16556 also for function that don't have calls or are not referenced.
16558 2017-06-30 Richard Biener <rguenther@suse.de>
16560 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
16561 analyze the first scalar stmt. Move vector type computation
16562 for the BB case here from ...
16563 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
16564 live operation processing in the SLP case properly.
16566 2017-06-30 Richard Biener <rguenther@suse.de>
16568 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
16570 2017-06-30 Martin Liska <mliska@suse.cz>
16573 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
16574 before BUILT_IN_UNWIND_RESUME when ASAN is used.
16576 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
16578 * doc/invoke.texi (AArch64): Add missing options and remove redundant
16581 2017-06-30 Richard Biener <rguenther@suse.de>
16583 PR tree-optimization/81249
16584 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
16585 condition reduction result to original scalar type.
16587 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16589 * profile-count.h (enum profile_quality): Fix typos and whitespace
16592 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16594 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
16595 type for branch probabilities.
16597 2017-06-29 Julian Brown <julian@codesourcery.com>
16598 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16600 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
16601 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
16602 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
16603 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
16605 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16607 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
16608 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
16609 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
16610 CC usage from generic code to here.
16611 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
16612 CC usage into the target macros.
16614 2017-06-29 Maya Rashish <coypu@sdf.org>
16616 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
16619 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16621 * arm/arm-builtins.c: Include profile-count.h
16622 * except.c (sjlj_emit_function_enter): Use
16623 profile_probability::unlikely.
16625 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16627 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
16628 and tocrel_offset be pointer args rather than implicitly using
16630 (legitimate_constant_pool_address_p, rs6000_emit_move,
16631 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
16632 tocrel_offset and use in toc_relative_expr_p call.
16633 (print_operand, print_operand_address): Use static tocrel_base_oac
16634 and tocrel_offset_oac.
16635 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
16638 2017-06-29 Maya Rashish <coypu@sdf.org>
16640 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
16642 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
16644 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
16645 objects, take into account only the alignment of 'op0' and 'mode1' if
16648 2017-06-29 Steve Ellcey <sellcey@cavium.com>
16650 * ccmp.c (ccmp_tree_comparison_p): New function.
16651 (ccmp_candidate_p): Update to use above function.
16652 (get_compare_parts): New function.
16653 (expand_ccmp_next): Update to use new functions.
16654 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
16656 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
16657 take mode as argument.
16658 * ccmp.h (expand_ccmp_expr): Add mode as argument.
16659 * expr.c (expand_expr_real_1): Pass mode as argument.
16661 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
16663 * combine.c (combine_instructions): Print insns to dump_file, together
16666 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16668 * asan.c (asan_emit_stack_protection): Update.
16669 (create_cond_insert_point): Update.
16670 * auto-profile.c (afdo_propagate_circuit): Update.
16671 * basic-block.h (struct edge_def): Turn probability to
16672 profile_probability.
16673 (EDGE_FREQUENCY): Update.
16674 * bb-reorder.c (find_traces_1_round): Update.
16675 (better_edge_p): Update.
16676 (sanitize_hot_paths): Update.
16677 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
16678 (make_single_succ_edge): Update.
16679 (check_bb_profile): Update.
16680 (dump_edge_info): Update.
16681 (update_bb_profile_for_threading): Update.
16682 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
16683 probabilitycount to 0.
16684 * cfgbuild.c (compute_outgoing_frequencies): Update.
16685 * cfgcleanup.c (try_forward_edges): Update.
16686 (outgoing_edges_match): Update.
16687 (try_crossjump_to_edge): Update.
16688 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
16689 (expand_gimple_tailcall): Update.
16690 (construct_init_block): Use make_single_succ_edge.
16691 (construct_exit_block): Use make_single_succ_edge.
16692 * cfghooks.c (verify_flow_info): Update.
16693 (redirect_edge_succ_nodup): Update.
16694 (split_edge): Update.
16695 (account_profile_record): Update.
16696 * cfgloopanal.c (single_likely_exit): Update.
16697 * cfgloopmanip.c (scale_loop_profile): Update.
16698 (set_zero_probability): Remove.
16699 (duplicate_loop_to_header_edge): Update.
16700 * cfgloopmanip.h (loop_version): Update prototype.
16701 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
16702 (force_nonfallthru_and_redirect): Update.
16703 (update_br_prob_note): Update.
16704 (rtl_verify_edges): Update.
16705 (purge_dead_edges): Update.
16706 (rtl_lv_add_condition_to_bb): Update.
16707 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
16708 * cgraphunit.c (init_lowered_empty_function): Update.
16709 (cgraph_node::expand_thunk): Update.
16710 * cilk-common.c: Include profile-count.h
16711 * dojump.c (inv): Remove.
16712 (jumpifnot): Update.
16713 (jumpifnot_1): Update.
16714 (do_jump_1): Update.
16716 (do_jump_by_parts_greater_rtx): Update.
16717 (do_compare_rtx_and_jump): Update.
16718 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
16719 do_jump_1. do_compare_rtx_and_jump): Update prototype.
16720 * dwarf2cfi.c: Include profile-count.h
16721 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
16722 (sjlj_emit_dispatch_table): Likewise.
16723 * explow.c: Include profile-count.h
16724 * expmed.c (emit_store_flag_force): Update.
16725 (do_cmp_and_jump): Update.
16726 * expr.c (compare_by_pieces_d::generate): Update.
16727 (compare_by_pieces_d::finish_mode): Update.
16728 (emit_block_move_via_loop): Update.
16729 (store_expr_with_bounds): Update.
16730 (store_constructor): Update.
16731 (expand_expr_real_2): Update.
16732 (expand_expr_real_1): Update.
16733 * expr.h (try_casesi, try_tablejump): Update prototypes.
16734 * gimple-pretty-print.c (dump_probability): Update.
16735 (dump_profile): New.
16736 (dump_gimple_label): Update.
16737 (dump_gimple_bb_header): Update.
16738 * graph.c (draw_cfg_node_succ_edges): Update.
16739 * hsa-gen.c (convert_switch_statements): Update.
16740 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
16741 (find_if_case_1): Update.
16742 (find_if_case_2): Update.
16743 * internal-fn.c (expand_arith_overflow_result_store): Update.
16744 (expand_addsub_overflow): Update.
16745 (expand_neg_overflow): Update.
16746 (expand_mul_overflow): Update.
16747 (expand_vector_ubsan_overflow): Update.
16748 * ipa-cp.c (good_cloning_opportunity_p): Update.
16749 * ipa-split.c (split_function): Use make_single_succ_edge.
16750 * ipa-utils.c (ipa_merge_profiles): Update.
16751 * loop-doloop.c (add_test): Update.
16752 (doloop_modify): Update.
16753 * loop-unroll.c (compare_and_jump_seq): Update.
16754 (unroll_loop_runtime_iterations): Update.
16755 * lra-constraints.c (lra_inheritance): Update.
16756 * lto-streamer-in.c (input_cfg): Update.
16757 * lto-streamer-out.c (output_cfg): Update.
16758 * mcf.c (adjust_cfg_counts): Update.
16759 * modulo-sched.c (sms_schedule): Update.
16760 * omp-expand.c (expand_omp_for_init_counts): Update.
16761 (extract_omp_for_update_vars): Update.
16762 (expand_omp_ordered_sink): Update.
16763 (expand_omp_for_ordered_loops): Update.
16764 (expand_omp_for_generic): Update.
16765 (expand_omp_for_static_nochunk): Update.
16766 (expand_omp_for_static_chunk): Update.
16767 (expand_cilk_for): Update.
16768 (expand_omp_simd): Update.
16769 (expand_omp_taskloop_for_outer): Update.
16770 (expand_omp_taskloop_for_inner): Update.
16771 * omp-simd-clone.c (simd_clone_adjust): Update.
16772 * optabs.c (expand_doubleword_shift): Update.
16773 (expand_abs): Update.
16774 (emit_cmp_and_jump_insn_1): Update.
16775 (expand_compare_and_swap_loop): Update.
16776 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
16777 * predict.c (predictable_edge_p): Update.
16778 (edge_probability_reliable_p): Update.
16779 (set_even_probabilities): Update.
16780 (combine_predictions_for_insn): Update.
16781 (combine_predictions_for_bb): Update.
16782 (propagate_freq): Update.
16783 (estimate_bb_frequencies): Update.
16784 (force_edge_cold): Update.
16785 * profile-count.c (profile_count::dump): Add missing space into dump.
16786 (profile_count::debug): Add newline.
16787 (profile_count::differs_from_p): Explicitly convert to unsigned.
16788 (profile_count::stream_in): Update.
16789 (profile_probability::dump): New member function.
16790 (profile_probability::debug): New member function.
16791 (profile_probability::differs_from_p): New member function.
16792 (profile_probability::differs_lot_from_p): New member function.
16793 (profile_probability::stream_in): New member function.
16794 (profile_probability::stream_out): New member function.
16795 * profile-count.h (profile_count_quality): Rename to ...
16796 (profile_quality): ... this one.
16797 (profile_probability): New.
16798 (profile_count): Update.
16799 * profile.c (compute_branch_probabilities): Update.
16800 * recog.c (peep2_attempt): Update.
16801 * sched-ebb.c (schedule_ebbs): Update.
16802 * sched-rgn.c (find_single_block_region): Update.
16803 (compute_dom_prob_ps): Update.
16804 (schedule_region): Update.
16805 * sel-sched-ir.c (compute_succs_info): Update.
16806 * stmt.c (struct case_node): Update.
16807 (do_jump_if_equal): Update.
16808 (get_outgoing_edge_probs): Update.
16809 (conditional_probability): Update.
16810 (emit_case_dispatch_table): Update.
16811 (expand_case): Update.
16812 (expand_sjlj_dispatch_table): Update.
16813 (emit_case_nodes): Update.
16814 * targhooks.c: Update.
16815 * tracer.c (better_p): Update.
16816 (find_best_successor): Update.
16817 * trans-mem.c (expand_transaction): Update.
16818 * tree-call-cdce.c: Update.
16819 * tree-cfg.c (gimple_split_edge): Upate.
16820 (move_sese_region_to_fn): Upate.
16821 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
16822 * tree-eh.c (lower_resx): Upate.
16823 (cleanup_empty_eh_move_lp): Upate.
16824 * tree-if-conv.c (version_loop_for_if_conversion): Update.
16825 * tree-inline.c (copy_edges_for_bb): Update.
16826 (copy_cfg_body): Update.
16827 * tree-parloops.c (gen_parallel_loop): Update.
16828 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
16829 (gimple_gen_time_profiler): Update.
16830 * tree-ssa-dce.c (remove_dead_stmt): Update.
16831 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
16832 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
16833 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
16834 (unloop_loops): Update.
16835 (try_peel_loop): Update.
16836 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
16837 * tree-ssa-loop-split.c (connect_loops): Update.
16838 (split_loop): Update.
16839 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
16840 (hoist_guard): Update.
16841 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
16842 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
16843 (value_replacement): Update.
16844 * tree-ssa-reassoc.c (branch_fixup): Update.
16845 * tree-ssa-tail-merge.c (replace_block_by): Update.
16846 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
16847 (create_edge_and_update_destination_phis): Update.
16848 (compute_path_counts): Update.
16849 (recompute_probabilities): Update.
16850 (update_joiner_offpath_counts): Update.
16851 (freqs_to_counts_path): Update.
16852 (duplicate_thread_path): Update.
16853 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
16854 (struct switch_conv_info): Update.
16855 (gen_inbound_check): Update.
16856 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
16857 (vect_do_peeling): Update.
16858 (vect_loop_versioning): Update.
16859 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
16860 (optimize_mask_stores): Update.
16861 * ubsan.c (ubsan_expand_null_ifn): Update.
16862 * value-prof.c (gimple_divmod_fixed_value): Update.
16863 (gimple_divmod_fixed_value_transform): Update.
16864 (gimple_mod_pow2): Update.
16865 (gimple_mod_pow2_value_transform): Update.
16866 (gimple_mod_subtract): Update.
16867 (gimple_mod_subtract_transform): Update.
16868 (gimple_ic): Update.
16869 (gimple_stringop_fixed_value): Update.
16870 (gimple_stringops_transform): Update.
16871 * value-prof.h: Update.
16873 2017-06-29 Carl Love <cel@us.ibm.com>
16875 * config/rs6000/rs6000-c.c: Add support for built-in functions
16876 vector signed int vec_signed (vector float);
16877 vector signed long long vec_signed (vector double);
16878 vector signed int vec_signed2 (vector double, vector double);
16879 vector signed int vec_signede (vector double);
16880 vector signed int vec_signedo (vector double);
16881 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
16882 instruction generator.
16883 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
16884 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
16885 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
16887 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
16888 vunsignede_v2df): Add define_expands.
16889 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
16890 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
16891 VEC_UNSIGNEDO): Add definitions.
16892 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
16893 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
16894 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
16895 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
16896 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
16897 * config/rs6000/altivec.h (vec_signed, vec_signed2,
16898 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
16899 vec_unsignede, vec_unsignedo): Add builtin defines.
16900 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
16902 * doc/extend.texi: Update the built-in documentation file for the
16903 new built-in functions.
16905 2017-06-29 Richard Biener <rguenther@suse.de>
16907 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
16908 reduction chains to LOOP_VINFO_REDUCTIONS.
16909 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
16910 SLP reductions after processing reduction chains.
16912 2017-06-29 Nathan Sidwell <nathan@acm.org>
16914 * builtins.c (fold_builtin_FUNCTION): Use
16915 lang_hooks.decl_printable_name.
16917 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
16919 PR middle-end/81194
16920 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
16921 with only one label.
16922 * stmt.c (expand_case): Assert NCASES is greater than one.
16924 2017-06-29 Richard Biener <rguenther@suse.de>
16926 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
16928 (group_case_labels): Likewise.
16929 (find_taken_edge): Push sanity checking on val to workers...
16930 (find_taken_edge_cond_expr): ... here
16931 (find_taken_edge_switch_expr): ... and here, handle cases
16932 with just a default label.
16933 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
16934 (group_case_labels): Likewise.
16935 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
16936 group_case_labels does anything cleanup the CFG again.
16938 2017-06-29 Bin Cheng <bin.cheng@arm.com>
16940 PR tree-optimization/81196
16941 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
16942 exit condition comparing two IVs.
16944 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
16946 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
16947 profile to the dummy entry at the end of the list of architectures.
16948 * config/arm/arm-cpu-cdata.h: Regenerated.
16950 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16951 Michael Collison <michael.collison@arm.com>
16954 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
16956 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
16957 (*aarch64_reg_<mode>3_minus_mask): New pattern.
16958 (*aarch64_<optab>_reg_di3_mask2): New pattern.
16959 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
16960 of shift when the shift amount is masked with constant equal to
16961 the size of the mode.
16962 * config/aarch64/predicates.md (subreg_lowpart_operator): New
16965 2017-06-29 Martin Liska <mliska@suse.cz>
16967 * config/i386/i386.opt: Change range from [1,5] to [0,5].
16969 2017-06-29 Yury Gribov <tetra2005@gmail.com>
16972 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
16975 (edge_growth_cache_entry::edge_growth_cache_entry): New
16977 (reset_edge_growth_cache): Update to use constructor.
16979 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16981 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
16982 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
16983 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
16985 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
16987 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
16988 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
16990 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
16992 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
16993 (*-linux-uclibc*): Add t-uclibc tmake_file.
16994 * config/t-musl: New.
16995 * config/t-uclibc: New.
16997 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
16999 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
17001 (gen_comm_data): Emit architectural setting of arch_prof.
17002 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
17004 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
17005 (armv8-m.base, armv8-m.main): Likewise.
17006 * arm-protos.h (arm_build_target): Add profile field.
17007 (arch_option): Likewise.
17008 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
17010 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
17011 arm_active_target.profile.
17013 2017-06-28 Richard Biener <rguenther@suse.de>
17015 PR middle-end/81227
17016 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
17017 TYPE_OVERFLOW_WRAPS.
17018 * match.pd (negate_expr_p): Likewise.
17019 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
17020 fold_build2, not fold_binary.
17022 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
17024 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
17025 Convert memory address to Pmode.
17026 (aarch64_print_operand): Assert MEM operands are always Pmode.
17028 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
17031 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
17032 Remove redundant if.
17033 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
17034 * config/arm/aarch-common-protos.h
17035 (aarch_forward_to_shift_is_not_shifted_re): Remove.
17036 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
17038 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
17041 * multiple_target.c (create_dispatcher_calls): Set the default
17042 clone to be static, not public.
17044 2017-06-28 Richard Biener <rguenther@suse.de>
17046 * tree-vect-loop.c (vectorizable_reduction): Move special
17047 cond reduction IV var creation ...
17048 (vect_create_epilog_for_reduction): ... here. Remove induction_index
17049 parameter. Use STMT_VINFO_VECTYPE.
17050 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
17053 2017-06-28 Martin Liska <mliska@suse.cz>
17056 * ipa-visibility.c (non_local_p): Handle visibility.
17058 2017-06-28 Martin Liska <mliska@suse.cz>
17061 * common.opt: Add IntegerRange to various options.
17062 * opt-functions.awk (integer_range_info): New function.
17063 * optc-gen.awk: Add integer_range_info to cl_options struct.
17064 * opts-common.c (decode_cmdline_option): Handle
17065 CL_ERR_INT_RANGE_ARG.
17066 (cmdline_handle_error): Likewise.
17067 * opts.c (print_filtered_help): Show valid interval in
17068 when --help is provided.
17069 * opts.h (struct cl_option): Add range_min and range_max fields.
17070 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
17072 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
17074 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
17075 (x * C EQ/NE y * C): New transformation.
17077 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
17079 * genmultilib (combination_space): Accept '+' in option names.
17081 2017-06-28 Martin Liska <mliska@suse.cz>
17084 * asan.c (instrument_derefs): Bail out inner references
17085 that are hard register variables.
17087 2017-06-28 Jakub Jelinek <jakub@redhat.com>
17090 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
17091 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
17093 2017-06-28 Richard Biener <rguenther@suse.de>
17095 * tree-vectorizer.h (vect_get_vec_defs): Remove.
17096 (vect_get_slp_defs): Adjust.
17097 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
17099 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
17101 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
17102 get_initial_defs_for_reduction instead of vect_get_vec_defs.
17103 (vectorizable_reduction): Adjust.
17104 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
17106 (vect_get_slp_defs): Likewise.
17107 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
17108 (vectorizable_bswap): Adjust.
17109 (vectorizable_call): Likewise.
17110 (vectorizable_conversion): Likewise.
17111 (vectorizable_assignment): Likewise.
17112 (vectorizable_shift): Likewise.
17113 (vectorizable_operation): Likewise.
17114 (vectorizable_store): Likewise.
17115 (vectorizable_condition): Likewise.
17116 (vectorizable_comparison): Likewise.
17118 2017-06-28 Michael Collison <michael.collison@arm.com>
17121 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
17123 (arm_gen_movmemqi): Removed unused variable 'i'.
17124 Convert 'for' loop into 'while' loop.
17125 (arm_expand_prologue): Remove last unnecessary set of insn.
17126 (thumb_pop): Remove unused variable 'pushed_words'.
17127 (thumb_exit): Remove last unnecessary set of regs_to_pop.
17129 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17131 * config/s390/predicates.md: Use s390_rel_address_ok_p.
17132 * config/s390/s390-protos.h: Add prototype of
17133 s390_rel_address_ok_p.
17134 * config/s390/s390.c (s390_got_symbol): New function.
17135 (s390_rel_address_ok_p): New function.
17136 (legitimize_pic_address): Use s390_rel_address_ok_p.
17137 (s390_load_got): Use s390_got_symbol.
17138 (s390_option_override): Issue error if
17139 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
17140 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
17142 * config/s390/s390.opt: New option mpic-data-is-text-relative.
17144 2017-06-27 Andrew Pinski <apinski@cavium.com>
17146 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
17147 (X * copysign (1.0, X)): New pattern.
17148 (X * copysign (1.0, -X)): New pattern.
17149 (copysign (-1.0, CST)): New pattern.
17151 2017-06-27 Joseph Myers <joseph@codesourcery.com>
17153 * genmultilib (combination_space): Remove variable.
17154 Validate reuse rules against regular expression for any sequence
17155 of multilib options in any order.
17157 2017-06-27 Michael Collison <michael.collison@arm.com>
17159 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
17160 call aarch64_split_simd_combine.
17161 * (aarch64_combine_internal<mode>): Delete pattern.
17162 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
17163 Allow register and subreg operands.
17165 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
17167 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
17168 specific need, just fallback on defaults.
17169 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
17171 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
17172 Olivier Hainque <hainque@adacore.com>
17174 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
17176 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
17177 targets. Pick a default if no particular attempt applied.
17178 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
17181 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
17183 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
17184 (x86_64-wrs-vxworks7): Likewise.
17186 2017-06-27 Marek Polacek <polacek@redhat.com>
17189 * ubsan.c (instrument_null): Check get_base_address's result for null.
17191 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
17193 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
17195 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
17197 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
17198 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
17199 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
17200 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
17201 New function types.
17202 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
17203 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
17204 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
17205 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
17206 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
17207 BUILT_IN_FEUPDATEENV): New builtins.
17208 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
17209 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
17210 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
17211 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
17213 (builtin_structptr_types): Adjust size.
17214 * tree.c (builtin_structptr_types): Add four entries.
17216 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
17217 Olivier Hainque <hainque@adacore.com>
17219 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
17220 (TLS_SYM): New local macro, forcing reference to __tls__ on
17221 link command lines for VxWorks 7 RTPs, triggering initialization
17223 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
17224 OS features TLS support, true for RTPs on VxWorks 7.
17225 * config/vxworks.c (vxworks_override_options): Setup emutls
17228 2017-06-27 Jakub Jelinek <jakub@redhat.com>
17230 * predict.c (test_prediction_value_range): Use -1U instead of -1
17231 to avoid narrowing conversion warning.
17232 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
17233 to avoid narrowing conversion warning.
17234 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
17236 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
17238 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
17240 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
17241 64bit configurations.
17242 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
17243 (SIZE_TYPE): Likewise.
17244 * config/vxworks.c (vxworks_emutls_var_fields): Use
17245 long_unsigned_type_node instead of unsigned_type_node as the offset
17246 field type, which is "pointer" mode in emutls.c.
17248 2017-06-27 Jakub Jelinek <jakub@redhat.com>
17251 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
17253 PR middle-end/81207
17254 * gimple-fold.c (replace_call_with_call_and_fold): Handle
17255 gimple_vuse copying separately from gimple_vdef copying.
17257 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17259 * value-prof.c (free_hist): Remove call to memset and the enclosing if
17262 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
17263 Olivier Hainque <hainque@adacore.com>
17265 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
17266 for all vxworks7 targets.
17267 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
17268 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
17269 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
17270 variations for VX6/VX7 and 32/64bits later on in ...
17271 (VXWORKS_LIB_SPEC): Leverage new macros.
17272 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
17273 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
17275 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
17277 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
17278 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
17280 2017-06-26 Carl Love <cel@us.ibm.com>
17282 * config/rs6000/rs6000-c.c: Add support for built-in functions
17283 vector bool char vec_reve (vector bool char);
17284 vector signed char vec_reve (vector signed char);
17285 vector unsigned char vec_reve (vector unsigned char);
17286 vector bool int vec_reve (vector bool int);
17287 vector signed int vec_reve (vector signed int);
17288 vector unsigned int vec_reve (vector unsigned int);
17289 vector bool long long vec_reve (vector bool long long);
17290 vector signed long long vec_reve (vector signed long long);
17291 vector unsigned long long vec_reve (vector unsigned long long);
17292 vector bool short vec_reve (vector bool short);
17293 vector signed short vec_reve (vector signed short);
17294 vector double vec_reve (vector double);
17295 vector float vec_reve (vector float);
17296 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
17297 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
17298 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
17299 (altivec_vreve): New pattern.
17300 * config/rs6000/altivec.h (vec_reve): New define.
17301 * doc/extend.texi (vec_rev): Update the built-in documentation file
17302 for the new built-in functions.
17304 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17306 PR tree-optimization/71815
17307 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
17309 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
17311 (slsr_process_phi): Likewise.
17312 (replace_uncond_cands_and_profitable_phis): Don't replace a
17313 multiply candidate with a stride of 1 (copy or cast).
17314 (phi_incr_cost): Call uses_consumed_by_stmt rather than
17316 (lowest_cost_path): Likewise.
17317 (total_savings): Likewise.
17319 2017-06-26 Richard Biener <rguenther@suse.de>
17322 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
17323 Use def_builtin_pure for all gather builtins.
17325 2017-06-26 Richard Biener <rguenther@suse.de>
17327 PR tree-optimization/81203
17328 * tree-tailcall.c (find_tail_calls): Do not move stmts into
17329 non-dominating BBs.
17331 2017-06-26 Marek Polacek <polacek@redhat.com>
17334 * doc/invoke.texi: Document -Wmultistatement-macros.
17336 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
17338 * doc/sourcebuild.texi (ARM-specific attributes): Document new
17339 arm_neon_ok_no_float_abi effective target.
17341 2017-06-26 Richard Biener <rguenther@suse.de>
17343 PR tree-optimization/80928
17344 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
17345 (copy_bbs): Set BB_DUPLICATED flag early.
17346 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
17348 (execute_on_shrinking_pred): Likewise.
17349 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
17350 BB_DUPLICATED blocks.
17351 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
17352 iterate over all PHIs considering removal of *gsi.
17354 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
17356 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
17359 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17361 * config/rs6000/rs6000-string.c: (expand_block_clear,
17362 do_load_for_compare, select_block_compare_mode,
17363 compute_current_alignment, expand_block_compare,
17364 expand_strncmp_align_check, expand_strn_compare,
17365 expand_block_move, rs6000_output_load_multiple)
17366 Move functions related to string/block move/compare
17367 to a separate file.
17368 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
17369 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
17370 for this function which is now used in two files.
17371 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
17372 * config.gcc: Add rs6000-string.o to extra_objs for
17373 targets powerpc*-*-* and rs6000*-*-*.
17375 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
17378 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
17379 32-bit, since indexed is not valid for DImode.
17380 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
17381 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
17382 (define_peephole2 for Altivec d-form load): Add 32-bit support.
17383 (define_peephole2 for Altivec d-form store): Likewise.
17386 * multiple_target.c (create_dispatcher_calls): Only create the
17387 dispatcher call if the function is the default clone of a
17388 versioned function.
17390 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
17392 PR middle-end/80902
17393 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
17394 a call, force the call to not be a tail call.
17396 2017-06-23 Jeff Law <law@redhat.com>
17398 * doc/contrib.texi: Add entry for Steven Pemberton's work on
17401 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
17403 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
17404 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
17405 handling for early expansion of vector shifts (sl,sr,sra,rl).
17406 (builtin_function_type): Add vector shift right instructions
17407 to the unsigned argument list.
17409 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
17411 rtl-optimizatoin/79286
17412 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
17413 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
17414 trap. PIC register plus a const unspec without offset can never trap.
17416 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
17418 * tree.h (builtin_structptr_type): New type.
17419 (builtin_structptr_types): Declare new array.
17420 * tree.c (builtin_structptr_types): New array.
17421 (free_lang_data, build_common_tree_nodes): Use it.
17423 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
17426 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
17429 2017-06-22 Matt Turner <mattst88@gmail.com>
17431 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
17432 Lake models to skylake case. Assume skylake for unknown
17433 models with clflushopt.
17435 2017-06-22 Jeff Law <law@redhat.com>
17437 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
17438 frame sizes that do not satisfy aarch64_uimm12_shift.
17440 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
17442 * profile-count.h (apply_probability,
17443 apply_scale, probability_in): Fix checks for zero.
17445 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17447 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
17448 * doc/cppdiropts.texi (-I @var{dir}): Document it.
17450 2016-06-22 Richard Biener <rguenther@suse.de>
17452 * tree-vect-loop.c (vect_model_reduction_cost): Handle
17453 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
17454 REDUC_MAX_EXPR support.
17455 (vectorizable_reduction): Likewise.
17456 (vect_create_epilog_for_reduction): Likewise.
17458 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
17460 * match.pd (A / (1 << B) -> A >> B): New.
17461 * generic-match-head.c: Include optabs-tree.h.
17462 * gimple-match-head.c: Likewise.
17463 * optabs-tree.h (target_supports_op_p): New.
17464 * optabs-tree.c (target_supports_op_p): New.
17466 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17468 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
17469 $gcc_cv_ld --help output.
17470 (gcc_cv_ld_demangle): Likewise.
17471 (gcc_cv_ld_eh_frame_hdr): Likewise.
17472 (gcc_cv_ld_pie): Likewise.
17473 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
17474 (gcc_cv_ld_buildid): Likewise.
17475 (gcc_cv_ld_sysroot): Likewise.
17476 (ld_bndplt_support): Likewise.
17477 (ld_pushpopstate_support): Likewise.
17478 * configure: Regenerate.
17479 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
17481 2017-06-21 Jakub Jelinek <jakub@redhat.com>
17484 * config/i386/sse.md (round<mode>2): Renumber match_dup and
17485 operands indexes to avoid gap between operands and match_dups.
17487 2017-06-21 Andrew Pinski <apinski@cavium.com>
17489 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
17490 Increment Arith_shift and Arith_shift_reg by 1.
17491 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
17493 * config/aarch64/aarch64.c (thunderx_tunings): Enable
17494 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
17495 (aarch64_strip_extend): Add new argument and test for it.
17496 (aarch64_cheap_mult_shift_p): New function.
17497 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
17498 add a cost if it is true.
17499 Update calls to aarch64_strip_extend.
17500 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
17502 2017-06-21 Andrew Pinski <apinski@cavium.com>
17504 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
17506 (thunderxt88): Likewise.
17507 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
17508 (thunderx_prefetch_tune): New variable.
17509 (thunderx2t99_prefetch_tune): Update for the correct values.
17510 (thunderxt88_tunings): New variable.
17511 (thunderx_tunings): Use thunderx_prefetch_tune instead of
17512 generic_prefetch_tune.
17513 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
17515 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17517 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
17518 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
17519 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
17520 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
17521 (aarch64_atomic_cas<mode>, GPI): Likewise.
17523 2017-06-21 Martin Liska <mliska@suse.cz>
17525 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
17526 statements on cold and hot labels.
17527 * predict.c (tree_estimate_probability_bb): Remove the
17528 prediction from this place.
17530 2017-06-21 Martin Liska <mliska@suse.cz>
17532 PR tree-optimization/79489
17533 * gimplify.c (maybe_add_early_return_predict_stmt): New
17535 (gimplify_return_expr): Call the function.
17536 * predict.c (tree_estimate_probability_bb): Remove handling
17538 * predict.def: Update comment about early return predictor.
17539 * gimple-predict.h (is_gimple_predict): New function.
17540 * predict.def: Change default value of early return to 66.
17541 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
17543 * passes.def: Put pass_strip_predict_hints to the beginning of
17546 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
17548 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
17549 FUNCTION_DECL declarations.
17550 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
17552 (dwaf2out_decl): Likewise.
17553 * godump.c (go_early_global_decl): Skip call to the real debug hook
17554 for FUNCTION_DECL declarations.
17555 * passes.c (rest_of_decl_compilation): Skip call to the
17556 early_global_decl debug hook for FUNCTION_DECL declarations, unless
17557 -fdump-go-spec is passed.
17559 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
17561 * config/i386/i386.c (struct builtin_isa): New field pure_p.
17562 Reorder for compactness.
17563 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
17564 (def_builtin_pure, def_builtin_pure2): New functions.
17565 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
17567 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
17569 * match.pd (nop_convert): New predicate.
17570 ((A +- CST1) +- CST2): Allow some NOP conversions.
17572 2017-06-21 Jakub Jelinek <jakub@redhat.com>
17575 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
17576 with ctors/dtors if GOVD_SHARED is set.
17578 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
17580 * config/aarch64/aarch64.md (movti_aarch64):
17581 Emit mov rather than orr.
17582 (movtf_aarch64): Likewise.
17583 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
17584 Emit mov rather than orr.
17586 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
17588 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
17589 Swap alternatives, make integer dup more expensive.
17591 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
17593 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
17594 Return true for non-tls symbols.
17596 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
17598 * config/aarch64/aarch64-cores.def (cortex-a55): New.
17599 (cortex-a75): Likewise.
17600 (cortex-a75.cortex-a55): Likewise.
17601 * config/aarch64/aarch64-tune.md: Regenerate.
17602 * doc/invoke.texi (-mtune): Document new values for -mtune.
17604 2017-06-21 Tom de Vries <tom@codesourcery.com>
17606 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
17607 stack_size feature.
17608 (Effective-Target Keywords, Other attributes): Suggest using
17609 dg-add-options stack_size feature to get stack limit in stack_size
17610 effective target documentation.
17612 2017-06-21 Julian Brown <julian@codesourcery.com>
17613 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17615 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
17616 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
17617 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
17619 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
17620 attribute type list for neon_multiply.
17621 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
17622 attribute type list for neon_multiply.
17623 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
17624 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
17625 attribute type list for neon_multiply.
17626 * config/arm/types.md (crypto_pmull): Add.
17627 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
17628 attribute type list.
17630 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
17632 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
17635 2017-06-20 Jakub Jelinek <jakub@redhat.com>
17637 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
17638 to make sure not to dereference a NULL cost_classes_ptr pointer.
17640 2017-06-20 Carl Love <cel@us.ibm.com>
17642 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17643 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17644 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17645 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17646 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17647 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17648 VMULOSW): New enum "unspec" values.
17649 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17650 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
17651 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17652 altivec_vmulosw): New patterns.
17653 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17654 VMULOSW): Add definitions.
17656 2017-06-20 Julia Koval <julia.koval@intel.com>
17658 * config/i386/i386.c: Fix rounding expand for new pattern.
17659 * config/i386/subst.md: Fix pattern (parallel -> unspec).
17661 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
17663 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
17664 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
17666 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
17668 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
17671 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
17673 * config/aarch64/aarch64-cores.def: Rearrange to sort by
17674 architecture, then by implementer ID.
17675 * config/aarch64/aarch64-tune.md: Regenerate.
17677 2017-06-20 Richard Biener <rguenther@suse.de>
17679 PR middle-end/81097
17680 * fold-const.c (split_tree): Fold to type before negating.
17682 2017-06-20 David Malcolm <dmalcolm@redhat.com>
17684 * diagnostic-show-locus.c
17685 (selftest::test_fixit_deletion_affecting_newline): New function.
17686 (selftest::diagnostic_show_locus_c_tests): Call it.
17688 2017-06-20 Andreas Schwab <schwab@suse.de>
17691 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
17694 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17696 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
17697 __ARM_FEATURE_COPROC according to support.
17699 2017-06-20 Jakub Jelinek <jakub@redhat.com>
17701 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
17702 Rewritten to avoid overflow for > 32-bit pointers.
17705 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
17706 by removing enum keyword.
17707 (ubsan_type_descriptor): Likewise. Formatting fix.
17710 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
17711 splitter): Require TARGET_SSE2 in the condition.
17713 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
17716 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
17717 for doing vector set of SFmode on ISA 3.0.
17718 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
17719 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
17721 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
17722 SFmode value into a V4SF variable that was extracted from another
17723 V4SF variable without converting the element to double precision
17724 and back to single precision vector format.
17725 (vsx_insert_extract_v4sf_p9_2): Likewise.
17727 2017-06-19 Jakub Jelinek <jakub@redhat.com>
17729 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
17730 in UWHI to avoid undefined overflow.
17733 * ubsan.h (enum ubsan_encode_value_phase): New.
17734 (ubsan_encode_value): Change second argument to
17735 enum ubsan_encode_value_phase with default value of
17736 UBSAN_ENCODE_VALUE_GENERIC.
17737 * ubsan.c (ubsan_encode_value): Change second argument to
17738 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
17739 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
17740 create_tmp_var_raw instead of create_tmp_var and use a
17742 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
17743 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
17744 ubsan_encode_value callers.
17747 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
17748 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
17749 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
17751 2017-06-19 Richard Biener <rguenther@suse.de>
17753 PR middle-end/81118
17754 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
17755 estimates if we changed anything.
17757 2017-06-19 Richard Biener <rguenther@suse.de>
17759 PR tree-optimization/80887
17760 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
17761 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
17762 simplified lookups, then reset mprts_hook.
17763 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
17765 (try_to_simplify): Likewise.
17767 2017-06-19 Martin Liska <mliska@suse.cz>
17770 * gimplify.c (gimplify_switch_expr):
17771 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
17773 2017-06-19 Martin Liska <mliska@suse.cz>
17775 * doc/install.texi: Document that PGO runs in 4 stages.
17777 2017-06-19 Martin Liska <mliska@suse.cz>
17780 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
17781 to dispatcher function name.
17782 * multiple_target.c (replace_function_decl): New function.
17783 (create_dispatcher_calls): Redirect both edges and references.
17785 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
17787 * profile-count.c (profile_count::dump): Dump quality.
17788 (profile_count::differs_from_p): Update for unsigned val.
17789 * profile-count.h (profile_count_quality): New enum.
17790 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
17792 2017-06-19 Richard Biener <rguenther@suse.de>
17794 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
17795 struct function as arg.
17796 (estimate_numbers_of_iterations): Export overload with loop arg.
17797 (free_numbers_of_iterations_estimates_loop): Use an overload of
17798 free_numbers_of_iterations_estimates instead.
17799 * tree-cfg.c (remove_bb): Adjust.
17800 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
17801 * tree-parloops.c (gen_parallel_loop): Likewise.
17802 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
17804 (tree_unroll_loops_completely): Likewise.
17805 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
17806 Use an overload instead and export.
17807 (estimated_loop_iterations): Adjust.
17808 (max_loop_iterations): Likewise.
17809 (likely_max_loop_iterations): Likewise.
17810 (estimate_numbers_of_iterations): Take struct function as arg
17812 (loop_exits_before_overflow): Adjust.
17813 (free_numbers_of_iterations_estimates_loop): Use an overload.
17814 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
17815 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
17817 2017-06-19 Richard Biener <rguenther@suse.de>
17820 * ipa-prop.c (find_constructor_constant_at_offset): Handle
17821 RANGE_EXPR conservatively.
17823 2017-06-16 Carl Love <cel@us.ibm.com>
17825 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17826 definitions for vec_float, vec_float2, vec_floato,
17827 vec_floate built-ins.
17828 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
17829 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
17831 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
17832 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
17833 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
17834 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
17835 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
17836 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
17837 vec_floato): Add builtin defines.
17838 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
17839 Update the built-in documentation file for the new built-in
17842 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17844 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
17845 (mthumb): Mark as the negative of -marm.
17847 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17849 * doc/invoke.texi (ARM Options, -mcpu): Document supported
17851 (ARM Options, -mtune): Document that this accepts the same
17852 extension options as -mcpu.
17853 (ARM Options, -mfpu): Document addition of -mfpu=auto.
17855 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17857 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
17858 permitted extensions.
17860 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17862 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
17863 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
17864 (armv8-m.main): Add option +nodsp.
17865 * config/arm/arm-cpu-cdata.h: Regenerated.
17867 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17869 * config/arm/t-fuchsia: New file.
17870 * config.gcc (arm*-*-fuchsia*): Use it.
17872 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17874 * config/arm/t-symbian: Rewrite for new option infrastructure.
17876 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17878 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
17879 (MULTILIB_REQUIRED): Likewise.
17881 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17883 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
17884 (MULTILIB_RESUE): Likewise.
17885 (MULTILIB_MATCHES): Likewise.
17886 (MULTLIB_REQUIRED): Likewise.
17888 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17890 * config/arm/t-rtems: Rewrite for new option framework.
17892 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17894 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
17895 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
17896 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
17897 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
17898 * config/arm/t-multilib: ... here.
17899 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
17900 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
17901 armv7-a and armv8*-a when A-profile libraries have not been built.
17902 * config/arm/t-rmprofile: Rewrite.
17904 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17906 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
17907 with a backslash. Remove the backslash after substituting unescaped
17909 * doc/fragments.texi (MULTILIB_REUSE): Document it.
17911 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17913 * config.gcc: (arm*-*-*): When building a-profile libraries, force
17914 the driver to pass through the default setting of -mfloat-abi.
17915 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
17917 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
17918 (all_feat_combs): New rule.
17919 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
17921 * config/arm/t-aprofile: Rewrite.
17923 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17925 * config/arm/arm.h (FPUTYPE_AUTO): Define.
17926 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
17927 fpu is not specified by the user/command-line.
17928 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
17929 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
17930 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
17931 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
17932 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
17933 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
17935 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17937 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
17938 * config/arm/t-arm-elf: Rewritten.
17940 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17942 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
17943 have some floating-point instructions.
17944 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
17945 (TARGET_MAYBE_HARD_FLOAT): New macro.
17946 * config/arm/arm-builtins.c (arm_init_builtins): Use
17947 TARGET_MAYBE_HARD_FLOAT.
17948 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
17950 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17952 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
17953 (configargs.h): Include it.
17954 (arm_print_hint_for_fpu_option): New function.
17955 (arm_parse_fpu_option): New function.
17956 (candidate_extension): New class.
17957 (arm_canon_for_multilib): New function.
17958 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
17959 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
17960 (ARCH_CANONICAL_SPECS): New macro.
17961 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
17963 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17965 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
17966 are set after handling multilib fragments. Set target_cpu_default2
17969 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17971 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
17973 (arm*-*-*): Set target_cpu_default2 to a quoted string.
17974 * config/arm/parsecpu.awk (check_cpu): Validate any extension
17976 (check_arch): Likewise.
17977 * config/arm/arm.c (arm_configure_build_target): Handle
17978 TARGET_CPU_DEFAULT being a string constant. Scan any feature
17979 options in the default.
17981 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17983 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
17984 when an option is an alias of another.
17985 * config/arm/parsecpu.awk (optalias): New parser token.
17986 (gen_comm_data): Mark non-alias options as such. Emit entries
17987 for extension aliases.
17988 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
17989 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
17990 (armv6kz, armv6zk, armv6t2): Likewise.
17991 (armv7): Make vfpv3-d16 an alias.
17992 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
17994 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
17995 Sort in canonical order.
17996 (armv8-a): Sort in canonical order.
17997 (armv8.1-a, armv8.2-a): Likewise.
17998 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
18000 (cortex-a9): Sort in canonical order.
18001 * config/arm/arm.c (selftests.h): Include it.
18002 (arm_test_cpu_arch_data): New function.
18003 (arm_run_self_tests): New function.
18004 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
18005 (targetm): Move declaration to the end of the file.
18006 * arm-cpu-cdata.h: Regenerated.
18008 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18010 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
18011 call to target_mode_check describing the type of option passed.
18012 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
18013 (arm_target_thumb_only): Use arm_parse_arch_option_name or
18014 arm_parse_cpu_option_name to match parameters against list of
18016 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
18017 arm_arch_core_flags data structure.
18018 * config/arm/arm-cpu_cdata.h: Regenerated.
18020 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18022 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
18024 (arm_print_hint_for_cpu_option): Likewise.
18025 (arm_print_hint_for_arch_option): Likewise.
18026 (arm_parse_cpu_option_name): Likewise.
18027 (arm_parse_arch_option_name): Likewise.
18028 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
18029 of entries in the all_fpus list.
18030 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
18031 (arm_parse_cpu_option_name): Declare.
18032 (arm_parse_arch_option_name): Declare.
18033 (arm_parse_option_features): Declare.
18034 (arm_intialize_isa): Declare.
18035 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
18037 (gen_comm_data): ... here. Make definitions non-static.
18038 * config/arm/arm-cpu-data.h: Regenerated.
18039 * config/arm/arm-cpu-cdata.h: Regenerated.
18041 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18043 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
18044 (cpu_arch_extension): New structure.
18045 (cpu_arch_option, arch_option, cpu_option): New structures.
18046 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
18047 architecture types.
18048 (gen_data): Generate new format data tables.
18049 * config/arm/arm.c (cpu_tune): New structure.
18050 (cpu_option, processors): Delete.
18051 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
18052 (arm_print_hint_for_cpu_option): ... this and ...
18053 (arm_print_hint_for_arch_option): ... this.
18054 (arm_parse_arch_cpu_name): Delete. Replace with ...
18055 (arm_parse_cpu_option_name): ... this and ...
18056 (arm_parse_arch_option_name): ... this.
18057 (arm_unrecognized_feature): Change type of target parameter to
18059 (arm_parse_arch_cpu_features): Delete. Replace with ...
18060 (arm_parse_option_features): ... this.
18061 (arm_configure_build_target): Rework to use new configuration data
18063 (arm_print_tune_info): Rework for new configuration data tables.
18064 * config/arm/arm-cpu-data.h: Regenerated.
18065 * config/arm/arm-cpu.h: Regenerated.
18067 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18069 * Makefile.in (OBJS): Move sbitmap.o from here ...
18070 (OBJS-libcommon): ... to here.
18072 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18074 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
18075 (ISA_ALL_CRYPTO): New macro.
18076 (ISA_ALL_SIMD): New macro
18077 (ISA_ALL_FP): New macro.
18078 * config/arm/arm.c (fpu_bitlist): Update initializer.
18079 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
18081 (arm9e): Add fpu. Add option for nofp
18082 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
18083 (arm926ej-s, arm1026ej-s): Likewise.
18084 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
18085 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
18086 neon-fp16, neon-vfpv4, nofp and nosimd.
18087 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
18088 (cortex-a8): Add fpu. Add option for nofp.
18089 (cortex-a9): Add fpu. Add options for nosimd and nofp.
18090 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
18091 (cortex-r4f): Add fpu.
18092 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
18093 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
18095 (cortex-r8): Likewise.
18096 (cortex-m4): Add fpu. Add option for nofp.
18097 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
18098 (cortex-a17.cortex-a7): Likewise.
18099 (cortex-a32): Add fpu. Add options for crypto and nofp.
18100 (cortex-a35, cortex-a53): Likewise.
18101 (cortex-a57): Add fpu. Add option for crypto.
18102 (cortex-a72, cortex-a73): Likewise.
18103 (exynos-m1): Likewise.
18104 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
18105 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
18106 (cortex-m33): Add fpu. Add option for nofp.
18107 * config/arm/arm-cpu-cdata.h: Regenerated
18108 * config/arm/arm-cpu-data.h: Regenerated.
18110 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18112 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
18113 (armv5te, armv5tej): Likewise.
18114 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
18115 (armv7): Add options fp and vfpv3-d16.
18116 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
18117 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
18119 (armv7ve): Likewise.
18120 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
18121 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
18122 (armv8-a): Add nocrypto option.
18123 (armv8.1-a, armv8.2-a): Likewise.
18124 (armv8-m.main): add options fp, fp.dp and nofp.
18126 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18128 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
18130 (armv8-a+crc): Delete.
18131 (armv8.1-a): Add options simd, crypto and nofp.
18132 (armv8.2-a): Add options fp16, simd, crypto and nofp.
18133 (armv8.2-a+fp16): Delete.
18134 (armv8-m.main): Add option dsp.
18135 (armv8-m.main+dsp): Delete.
18136 (cortex-a8): Add fpu. Add nofp option.
18137 (cortex-a9): Add fpu. Add nofp and nosimd options.
18138 * config/arm/parsecpu.awk (gen_data): Generate option tables and
18139 link to main cpu and architecture data structures.
18140 (gen_comm_data): Only put isa attributes from the main architecture
18142 (option): New statement for architecture and CPU entries.
18143 * arm.c (struct cpu_option): New structure.
18144 (struct processors): Add entry for options.
18145 (arm_unrecognized_feature): New function.
18146 (arm_parse_arch_cpu_name): Ignore any characters after the first
18148 (arm_parse_arch_cpu_feature): New function.
18149 (arm_configure_build_target): Separate out any CPU and architecture
18150 features and parse separately. Don't error out if -mfpu=auto is
18151 used with only an architecture string.
18152 (arm_print_asm_arch_directives): New function.
18153 (arm_file_start): Call it.
18154 * config/arm/arm-cpu-cdata.h: Regenerated.
18155 * config/arm/arm-cpu-data.h: Likewise.
18156 * config/arm/arm-tables.opt: Likewise.
18158 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18160 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
18161 assembler when it is not -mfpu=auto.
18163 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18165 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
18166 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
18167 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
18168 (ASM_CPU_SPEC): Rewrite.
18169 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
18170 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
18171 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
18172 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
18173 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
18174 copied string is NUL-terminated. Also strip any characters prefixed
18176 (arm_rewrite_selected_arch): New function.
18177 (arm_rewrite_march): New function.
18179 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18181 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
18182 (x_arm_cpu_string, x_arm_tune_string): Likewise.
18183 (march, mcpu, mtune): Convert to string-based options.
18184 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
18185 (arm_parse_arch_cpu_name): New function.
18186 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
18187 identify selected architecture or CPU.
18188 (arm_option_save): New function.
18189 (TARGET_OPTION_SAVE): Redefine.
18190 (arm_option_restore): Restore string options.
18191 (arm_option_print): Print string options.
18193 2017-06-16 Martin Sebor <msebor@redhat.com>
18195 PR tree-optimization/80933
18196 PR tree-optimization/80934
18197 * builtins.c (fold_builtin_3): Do not handle bcmp here.
18198 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
18199 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
18200 (gimple_fold_builtin): Call them.
18202 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18204 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
18205 as unlikely; update profile.
18207 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18209 * predict.c (force_edge_cold): Handle declaring edges impossible
18212 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18214 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
18216 (try_unroll_loop_completely): Fix reporting.
18218 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18220 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
18222 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
18225 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
18226 if given a non-constant argument for an intrinsic which requires a
18229 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18231 * profile.c (compare_freqs): New function.
18232 (branch_prob): Sort edge list.
18233 (find_spanning_tree): Assume that the list is priority sorted.
18235 2017-06-16 Richard Biener <rguenther@suse.de>
18237 PR tree-optimization/81090
18238 * passes.def (pass_record_bounds): Remove.
18239 * tree-pass.h (make_pass_record_bounds): Likewise.
18240 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
18241 make_pass_record_bounds): Likewise.
18242 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
18243 not free niter estimates at the beginning but at the end.
18244 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
18246 2017-06-16 Richard Biener <rguenther@suse.de>
18248 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
18249 initializer to workaround ICE in host GCC 4.8.
18251 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18253 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
18255 (clone_inlined_nodes): Update.
18257 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
18259 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
18260 prefetch settings, and enable prefetching by default at -O3.
18262 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
18264 * config/aarch64/aarch64.c (aarch64_override_options_internal):
18265 Set flag_prefetch_loop_arrays according to tuning data.
18267 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
18269 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
18270 New tune structure.
18271 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
18272 [Unrelated to main purpose of the patch] Place the pointer field last
18273 to enable type checking errors when tune structure are wrongly merged.
18274 * config/aarch64/aarch64.c (generic_prefetch_tune,)
18275 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
18276 (thunderx2t99_prefetch_tune): New tune constants.
18277 (tune_params *_tunings): Update all tunings (no functional change).
18278 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
18279 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
18280 from tunings structures.
18282 2017-06-16 Jakub Jelinek <jakub@redhat.com>
18285 * ubsan.c (instrument_null): Add T argument, use it instead
18286 of computing it based on IS_LHS.
18287 (instrument_object_size): Likewise.
18288 (pass_ubsan::execute): Adjust instrument_null and
18289 instrument_object_size callers to pass gimple_get_lhs or
18290 gimple_assign_rhs1 result to it. Use instrument_null instead of
18291 calling get_base_address and instrument_mem_ref. Handle
18292 aggregate call arguments for object-size sanitization.
18294 2017-06-16 Yury Gribov <tetra2005@gmail.com>
18296 PR tree-optimization/81089
18297 * tree-vrp.c (is_masked_range_test): Validate operands of
18300 2017-06-15 Martin Sebor <msebor@redhat.com>
18303 * dumpfile.c (dump_register): Avoid calling memset to initialize
18304 a class with a default ctor.
18305 * gcc.c (struct compiler): Remove const qualification.
18306 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
18307 * hash-table.h: Ditto.
18308 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
18310 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
18311 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
18313 * params.h (struct param_info): Make struct members non-const.
18314 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
18315 with default initialization.
18316 * vec.h (vec_copy_construct, vec_default_construct): New helper
18318 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
18319 with vec_copy_construct.
18320 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
18321 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
18322 * doc/invoke.texi (-Wclass-memaccess): Document.
18324 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18326 * emit-rtl.h (is_leaf): Update comment about local
18327 register allocator.
18329 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18332 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
18333 for a variable to have a section before checking if the section has a
18335 Set section to.persistent if persistent attribute is set.
18336 Warn if .persistent attribute is used on an automatic variable.
18338 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
18340 PR rtl-optimization/80474
18341 * reorg.c (update_block): Do not ignore instructions in a delay slot.
18343 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
18345 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
18348 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
18350 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
18351 (casesi): Emit bounds checking as RTL.
18352 (casesi_internal_mips16_<mode>): Remove bounds checking.
18354 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
18356 * config/xtensa/xtensa.c (xtensa_option_override): Append
18357 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
18358 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
18359 xtensa_doloop_hooks): Define unconditionally.
18360 (xtensa_reorg_loops): Only call reorg_loops in the presence of
18362 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
18363 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
18364 for it in xtensa_option_override.
18365 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
18366 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
18368 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
18370 * doc/cppopts.texi: Document '-' special value to -MF.
18372 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
18374 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
18375 (cortex_a53_fconst): Likewise.
18376 (cortex_a53_fpmul): Likewise.
18377 (cortex_a53_f_load_64): Likewise.
18378 (cortex_a53_f_load_many): Likewise.
18379 (cortex_a53_advsimd_alu): Likewise.
18380 (cortex_a53_advsimd_alu_q): Likewise.
18381 (cortex_a53_advsimd_mul): Likewise.
18382 (cortex_a53_advsimd_mul_q): Likewise.
18383 (fpmac bypass): Add new bypass for fpmac-fpmac case.
18384 Add missing fmul, r2f_cvt and fconst cases.
18386 2017-06-14 Richard Biener <rguenther@suse.de>
18388 PR middle-end/81088
18389 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
18391 (fold_binary_loc): When associating do not treat pre-existing
18392 TREE_OVERFLOW on literal constants as a reason to allow
18393 TREE_OVERFLOW on associated literal constants.
18395 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
18397 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
18398 (MASK_FEATURES): New macro.
18399 * config/sparc/sparc.c (sparc_option_override): Remove the special
18400 handling of -mfpu and generalize it to all MASK_FEATURES switches.
18402 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
18404 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
18405 a division of 0 if non-call exceptions are enabled.
18407 2017-06-14 Andrew Pinski <apinski@cavium.com>
18408 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18411 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
18412 Improve vector initialization code gen for only variable case.
18414 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
18416 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
18418 2017-06-14 Richard Biener <rguenther@suse.de>
18420 PR tree-optimization/81083
18421 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
18424 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18426 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
18427 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
18428 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
18429 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
18430 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
18431 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
18433 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18435 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
18436 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
18438 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18440 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
18442 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18444 * config/rs6000/t-rtems: Don't handle SPE.
18446 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18448 * config/rs6000/t-linux: Don't handle SPE.
18450 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18452 * config/rs6000/eabispe.h: Delete file.
18454 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18456 * config/rs6000/t-spe: Delete file.
18458 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18460 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
18461 (rs6000_legitimate_offset_address_p): Return false for anything in
18462 V2SImode or V2SFmode.
18464 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18466 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
18467 except V2SF and V2SI. Rearrange the vector modes, and add comments.
18468 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
18470 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
18471 (rs6000_legitimate_offset_address_p): Ditto.
18472 (rs6000_emit_move): Ditto.
18473 (rs6000_init_builtins): Remove V4HI_type_node.
18475 2017-06-13 Martin Liska <mliska@suse.cz>
18478 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
18479 (gate_asan): Likewise.
18480 * asan.h (asan_no_sanitize_address_p): Remove the function.
18481 (sanitize_flags_p): New function.
18482 * builtins.def: Fix coding style.
18483 * common.opt: Use renamed enum value.
18484 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
18485 * doc/extend.texi: Document no_sanitize attribute.
18486 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
18487 to SANITIZE_UNDEFINED_NONDEFAULT.
18488 * gcc.c (sanitize_spec_function): Use the renamed enum value.
18489 * gimple-fold.c (optimize_atomic_compare_exchange_p):
18490 Use sanitize_flags_p.
18491 * gimplify.c (gimplify_function_tree): Likewise.
18492 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
18493 * opts.c (parse_no_sanitize_attribute): New function.
18494 (common_handle_option): Use renamed enum value.
18495 * opts.h (parse_no_sanitize_attribute): Declare.
18496 * tree.c (sanitize_flags_p): New function.
18497 * tree.h: Declared here.
18498 * tsan.c: Use sanitize_flags_p.
18499 * ubsan.c (ubsan_expand_null_ifn): Likewise.
18500 (instrument_mem_ref): Likewise.
18501 (instrument_bool_enum_load): Likewise.
18502 (do_ubsan_in_current_function): Remove the function.
18503 (pass_ubsan::execute): Use sanitize_flags_p.
18504 * ubsan.h: Remove do_ubsan_in_current_function
18505 * tree-cfg.c (print_no_sanitize_attr_value): New function.
18506 (dump_function_to_file): Use it here.
18508 2017-06-13 Martin Jambor <mjambor@suse.cz>
18510 PR tree-optimization/80803
18511 PR tree-optimization/81063
18512 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
18513 (propagate_subaccesses_across_link): Enqueue subtree whenever
18514 necessary instead of relying on the caller.
18516 2017-06-13 Martin Jambor <mjambor@suse.cz>
18518 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
18519 that have a first_link.
18520 (sort_and_splice_var_accesses): Do not check first_link before
18522 (subtree_mark_written_and_enqueue): Likewise.
18523 (propagate_all_subaccesses): Likewise and do not stop at first
18524 parent with a first_link.
18526 2017-06-13 Martin Jambor <mjambor@suse.cz>
18528 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
18531 2017-06-13 Yury Gribov <tetra2005@gmail.com>
18533 * match.pd: New pattern.
18535 2017-06-13 Yury Gribov <tetra2005@gmail.com>
18537 * tree-vrp.c (is_masked_range_test): New function.
18538 (register_edge_assert_for): Determine ranges for
18541 2017-06-13 Yury Gribov <tetra2005@gmail.com>
18543 PR tree-optimization/67328
18544 * fold-const.c (maskable_range_p): New function.
18545 (build_range_check): Generate bittests if possible.
18547 2017-06-13 Martin Liska <mliska@suse.cz>
18549 * gimple-pretty-print.c (dump_probability): Add new argument.
18550 (dump_edge_probability): Dump both probability and count.
18551 (dump_gimple_label): Likewise.
18552 (dump_gimple_bb_header): Likewise.
18554 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
18557 * config/avr/avr-devices.c: Fix indentation.
18558 * config/avr/gen-avr-mmcu-specs.c: Dito.
18560 2017-06-13 Richard Biener <rguenther@suse.de>
18562 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
18563 instead get vector type from stmt_info.
18564 (vectorizable_reduction): Adjust. Remove dead code.
18566 2017-06-13 Richard Biener <rguenther@suse.de>
18568 PR middle-end/81065
18569 * fold-const.c (extract_muldiv_1): Remove bogus distribution
18570 case of C * (x * C2 + C3).
18571 (fold_addr_of_array_ref_difference): Properly fold index difference.
18573 2017-06-12 David S. Miller <davem@davemloft.net>
18576 * config/sparc/sparc.md (return expander): Emit frame blockage if
18577 function uses alloca.
18579 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
18581 * combine.c (make_field_assignment): Check len rather than the mode
18582 precision when calling force_to_mode.
18584 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
18586 Support multilibs and devices that see flash in RAM address range.
18589 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
18590 (avr_mcu_t) <flash_pm_offset>: New field.
18591 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
18592 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
18593 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
18594 (AVR_TINY_PM_OFFSET): Remove macro.
18595 * config/avr/avr.opt (-mshort-calls): New option.
18596 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
18597 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
18598 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
18599 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
18600 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
18601 instead of avr_arch->have_jmp_call.
18602 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
18603 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
18604 avr_arch->flash_pm_offset to define.
18605 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
18606 new field flash_pm_offset. Add entry for avrxmega3.
18607 (avr_texinfo): Add entry for avrxmega3.
18608 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
18609 attiny212, attiny214,
18610 attiny412, attiny414, attiny416, attiny417,
18611 attiny814, attiny816, attiny817,
18612 attiny1614, attiny1616, attiny1617,
18613 attiny3214, attiny3216, attiny3217.
18614 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
18615 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
18616 (avr_print_operand_address) [AVR_TINY]: Same.
18617 (avr_asm_init_sections) <readonly_data_section>: Only patch
18618 callback if avr_arch->flash_pm_offset = 0.
18619 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
18620 for rodata if avr_arch->flash_pm_offset != 0.
18621 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
18622 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
18623 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
18624 (m_options): Append opt_rcall.
18625 (m_dirnames): Append dir_rcall.
18626 * config/avr/t-multilib: Regenerate.
18628 * configure.ac [target=avr]: Check whether avrxmega3 default
18629 linker description file works as needed.
18630 * configure: Regenerate.
18631 * doc/avr-mmcu.texi: Regenerate.
18632 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
18633 <__AVR_ARCH__>: Document avrxmega3 and 103.
18634 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
18635 <__AVR_SHORT_CALLS__>: Document it.
18636 <__AVR_PM_BASE_ADDRESS__>: Document it.
18637 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
18638 (AVR Variable Attributes) <progmem>: Document this is
18639 not needed for avrxmega3.
18640 (AVR Named Address Spaces) <__flash>: Dito.
18642 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
18644 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
18646 2017-06-12 Doug Rupp <rupp@adacore.com>
18648 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
18649 Append vxworks-stdint.h to the tm_file list.
18650 * config/vxworks-stdint.h: New file.
18652 2017-06-12 Martin Liska <mliska@suse.cz>
18654 PR tree-optimization/81041
18655 * tree-profile.c (gimple_gen_ic_func_profiler):
18656 Create an extra BB in profile-generate
18657 (gimple_gen_time_profiler): Likewise.
18659 2017-06-12 Jakub Jelinek <jakub@redhat.com>
18661 PR tree-optimization/81003
18662 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
18663 (update_range_test): Use it instead of force_gimple_operand_gsi.
18665 2017-06-12 Richard Biener <rguenther@suse.de>
18667 PR tree-optimization/81053
18668 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
18669 with backedge value not defined in loop. Simplify def stmt
18672 2017-06-11 Tom de Vries <tom@codesourcery.com>
18675 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
18677 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
18678 nvptx_cannot_force_const_mem.
18680 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18682 * opts.c (finish_options): Move test for flag_split_stack after
18683 it has been initialized.
18685 2017-06-11 Jason Merrill <jason@redhat.com>
18687 * tree.h (id_equal): New.
18688 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
18689 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
18690 instead of strcmp of IDENTIFIER_POINTER.
18692 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18694 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
18695 (mark_all_inlined_calls_cdtor): Fix formating.
18696 (inline_transform): Rescale profile before inlining.
18698 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18700 * cgraph.h (cgraph_edge::clone): Update prototype.
18701 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
18702 (cgraph_node::create_clone): Update.
18703 (cgraph_node::create_version_clone): Update.
18704 * tree-inline.c (copy_bb): Update.
18705 (expand_call_inline): Update.
18707 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
18709 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
18710 factored out from ...
18711 (rs6000_emit_prologue): ... here.
18713 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
18715 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
18716 factored out from ...
18717 (rs6000_emit_prologue): ... here.
18719 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18721 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
18723 (handle_missing_profiles): Fix computation of tp_first_run.
18724 (counts_to_freqs): Do not touch freqs when count is 0.
18726 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18728 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
18731 2017-06-10 Tom de Vries <tom@codesourcery.com>
18733 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
18734 attributes): Document signal effective target.
18736 2017-06-10 Tom de Vries <tom@codesourcery.com>
18738 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
18739 Document effective target stack_size.
18741 2017-06-09 David Malcolm <dmalcolm@redhat.com>
18743 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
18744 to the edit_context if they can be auto-applied.
18746 2017-06-9 Ian Lance Taylor <iant@golang.org>
18748 * opts.c (finish_options): If -fsplit-stack, disable implicit
18749 -forder-blocks-and-partition.
18750 * doc/invoke.texi (Optimize Options): Document that when using
18751 -fsplit-stack -forder-blocks-and-partition is not implicitly
18754 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18756 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
18757 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
18758 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
18759 * builtins.def (abort, trap, unreachable): Declare cold.
18760 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
18761 * tree-core.h (ECF_COLD): New.
18762 * tree.c (set_call_expr_flags): Handle ECF_COLD.
18763 (build_common_builtin_nodes): Mark unreachable and abort as cold.
18765 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18767 * predict.c (unlikely_executed_stmt_p): Cleanup.
18769 2017-06-09 Richard Biener <rguenther@suse.de>
18771 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
18772 model if the ref is always written to.
18774 2017-06-09 Tamar Christina <tamar.christina@arm.com>
18776 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
18778 2017-06-09 Tamar Christina <tamar.christina@arm.com>
18780 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
18783 2017-06-09 Tom de Vries <tom@codesourcery.com>
18786 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
18787 "target cannot support label values" when encountering LABEL_REF.
18789 2017-06-09 Martin Liska <mliska@suse.cz>
18791 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
18792 (gimple_gen_ic_func_profiler): Emit direct comparison
18793 of __gcov_indirect_call_callee with NULL.
18794 (gimple_gen_time_profiler): Change probability from
18795 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
18797 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18799 * profile.c (edge_gcov_counts): Turn to pointer.
18800 (compute_branch_probabilities, compute_branch_probabilities): Update.
18801 (branch_prob): Do not clear edge_gcov_count.
18802 * profile.h (edge_gcov_counts): Turn to pointer.
18803 (edge_gcov_count): Update.
18805 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18807 * gimple.h (gimple_check_failed): Mark cold.
18809 2017-06-09 Richard Biener <rguenther@suse.de>
18811 PR tree-optimization/66623
18812 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
18813 refactor check_reduction into two parts, properly computing
18814 whether we have to check reduction validity for outer loop
18817 2017-06-09 Richard Biener <rguenther@suse.de>
18819 PR tree-optimization/79483
18820 * graphite-scop-detection.c (order): New global.
18821 (get_order): Compute bb to order mapping that satisfies code
18822 generation constraints.
18823 (cmp_pbbs): New helper.
18824 (build_scops): Start domwalk at entry block, sort generated
18827 2017-06-09 Richard Biener <rguenther@suse.de>
18829 PR middle-end/81007
18830 * ipa-polymorphic-call.c
18831 (ipa_polymorphic_call_context::restrict_to_inner_class):
18832 Skip FIELD_DECLs with error_mark_node type.
18833 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
18836 2017-06-09 Martin Liska <mliska@suse.cz>
18838 * predict.c (struct branch_predictor): New struct.
18839 (test_prediction_value_range): New test.
18840 (predict_c_tests): New function.
18841 * selftest-run-tests.c (selftest::run_tests): Run the function.
18842 * selftest.h: Declare new tests.
18844 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
18847 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
18848 gen_add3_insn did not fail.
18849 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
18850 r0, construct that number in a temporary reg and add that reg to r0.
18851 If asked to put the result in r0 as well, fail.
18853 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
18855 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
18856 for early expansion of vec_eqv.
18858 2017-06-08 Jakub Jelinek <jakub@redhat.com>
18860 PR middle-end/81005
18861 * ubsan.c (instrument_null): Avoid pointless code temporary.
18862 (pass_ubsan::execute): Instrument aggregate arguments of calls.
18864 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
18868 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
18871 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
18874 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18876 * predict.c (unlikely_executed_edge_p): Move ahead.
18877 (probably_never_executed_edge_p): Use it.
18879 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
18881 PR middle-end/79988
18882 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
18883 gimple_call_builtin_p call.
18885 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18887 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
18888 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
18889 rtl_check_failed_type2, rtl_check_failed_code1,
18890 rtl_check_failed_code2, rtl_check_failed_code_mode,
18891 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
18892 rtvec_check_failed_bounds, rtl_check_failed_flag,
18893 _fatal_insn_not_found, _fatal_insn): Likewise.
18894 * tree.h (tree_contains_struct_check_failed,
18895 tree_check_failed, tree_not_check_failed,
18896 tree_class_check_failed, tree_range_check_failed,
18897 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
18898 tree_vec_elt_check_failed, phi_node_elt_check_failed,
18899 tree_operand_check_failed, omp_clause_check_failed,
18900 omp_clause_operand_check_failed, omp_clause_range_check_failed):
18903 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18905 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
18906 flag_branch_probabilities.
18907 * ipa-inline.c (edge_badness): Likewise.
18908 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
18909 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
18910 * predict.c (maybe_hot_frequency_p): Likewise.
18911 (probably_never_executed): Likewise.
18912 * sched-ebb.c (schedule_ebbs): Likewise.
18913 * sched-rgn.c (find_single_block_region): Likewise.
18914 * tracer.c (tail_duplicate): Likewise.
18916 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18918 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
18919 longer requires x_flag_profile_use.
18921 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18923 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
18924 instead of flag_reorder_blocks_and_partition.
18925 * dbxout.c (dbxout_function_end): Likewise.
18926 * dwarf2out.c (gen_subprogram_die): Likewise.
18927 * haifa-sched.c (sched_create_recovery_edges): Likewise.
18928 * hw-doloop.c (reorg_loops): Likewise.
18929 * varasm.c (assemble_start_function,
18930 assemble_end_function): Likewise.
18931 (decide_function_section): Do not check for
18932 flag_reorder_blocks_and_partition.
18934 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
18936 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
18938 (chkp_get_hard_register_fake_addr_expr): Ditto.
18939 (chkp_build_addr_expr): Add check for hard reg case.
18940 (chkp_parse_array_and_component_ref): Ditto.
18941 (chkp_find_bounds_1): Ditto.
18942 (chkp_process_stmt): Don't generate bounds store for
18945 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18947 * predict.c (maybe_hot_bb_p): Do not check profile status.
18948 (maybe_hot_edge_p): Likewise.
18949 (probably_never_executed): Check for zero counts even if profile
18951 (unlikely_executed_edge_p): New function.
18952 (unlikely_executed_stmt_p): New function.
18953 (unlikely_executed_bb_p): New function.
18954 (set_even_probabilities): Use unlikely predicates.
18955 (combine_predictions_for_bb): Likewise.
18956 (predict_paths_for_bb): Likewise.
18957 (predict_paths_leading_to_edge): Likewise.
18958 (determine_unlikely_bbs): New function.
18959 (estimate_bb_frequencies): Use it.
18960 (compute_function_frequency): Use zero counts even if profile is
18962 * profile-count.h: Fix typo.
18964 2017-08-08 Julia Koval <julia.koval@intel.com>
18966 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
18967 _mm512_mask_cvtsepi16_storeu_epi8,
18968 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
18969 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
18970 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
18971 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
18972 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
18973 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
18974 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
18975 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
18976 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
18977 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
18978 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
18979 __builtin_ia32_pmovuswb256mem_mask,
18980 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
18981 __builtin_ia32_pmovwb512mem_mask): New builtins.
18983 2017-08-08 Julia Koval <julia.koval@intel.com>
18985 PR target/73350,80862
18986 * config/i386/subst.md (round): Fix round pattern.
18987 * config/i386/i386.c (ix86_erase_embedded_rounding):
18988 Fix erasing rounding for the fixed pattern.
18990 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18992 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
18994 2017-06-08 Martin Liska <mliska@suse.cz>
18996 PR gcov-profile/80911
18997 * gcov.c (block_info::block_info): New constructor.
18999 2017-06-07 Carl Love <cel@us.ibm.com>
19001 * config/rs6000/rs6000-c: The return type of the following
19002 built-in functions was implemented as int not long long. Fix sign
19003 of return value for the unsigned version of vec_mulo and vec_mule.
19004 vector unsigned long long vec_bperm (vector unsigned long long,
19005 vector unsigned char)
19006 vector signed long long vec_mule (vector signed int,
19008 vector unsigned long long vec_mule (vector unsigned int,
19009 vector unsigned int)
19010 vector signed long long vec_mulo (vector signed int,
19012 vector unsigned long long vec_mulo (vector unsigned int,
19013 vector unsigned int)
19014 * doc/extend.texi: Fix the documentation for the built-in
19017 2017-06-07 Carl Love <cel@us.ibm.com>
19020 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
19023 2017-06-07 Carl Love <cel@us.ibm.com>
19025 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
19026 support, Generate doublehv for signed int/float for BE case only.
19028 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
19030 * doc/invoke.texi (mcx16): Rewrite.
19032 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19034 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
19035 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
19036 *mov<mode>_softfloat, and an anonymous splitter): Use
19037 nonimmediate_operand instead of rs6000_nonimmediate_operand.
19039 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19041 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
19043 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
19044 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
19045 (rs6000_debug_reg_global): Adjust.
19046 (rs6000_init_hard_regno_mode_ok): Adjust.
19047 (rs6000_dbx_register_number): Adjust.
19048 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
19049 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
19050 Remove SPE_ACC and SPEFSCR.
19051 (REG_ALLOC_ORDER): Ditto.
19052 (FRAME_POINTER_REGNUM): Change to 111.
19053 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
19054 (REG_CLASS_NAMES): Ditto.
19055 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
19056 (REGISTER_NAMES): Ditto.
19057 (ADDITIONAL_REG_NAMES): Ditto.
19058 (rs6000_reg_names): Ditto.
19059 * config/rs6000/rs6000.md: Renumber some register number
19062 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19064 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
19066 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
19067 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
19069 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
19070 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
19071 Delete the SPE high registers.
19072 (REG_ALLOC_ORDER): Ditto.
19073 (enum reg_class): Remove SPE_HIGH_REGS.
19074 (REG_CLASS_NAMES): Ditto.
19075 (REG_CLASS_CONTENTS): Delete the SPE high registers.
19076 (REGISTER_NAMES): Ditto.
19077 (rs6000_reg_names): Ditto.
19078 * doc/tm.texi.in: Remove SPE as example.
19079 * doc/tm.texi: Regenerate.
19081 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19083 * config/rs6000/8540.md (ppc8540_brinc): Delete.
19084 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
19085 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
19086 * config/rs6000/rs6000.md (type): Remove "brinc".
19088 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19090 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
19091 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
19092 * config/rs6000/linuxspe.h: Delete file.
19093 * config/rs6000/rs6000.md: Don't include spe.md.
19094 * config/rs6000/spe.h: Delete file.
19095 * config/rs6000/spe.md: Delete file.
19096 * config/rs6000/t-rs6000: Remove spe.md.
19098 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19100 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
19101 (reg_or_none500mem_operand): Delete.
19102 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
19103 instead of reg_or_none500mem_operand.
19105 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19107 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
19108 handling of SPE flags.
19109 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
19111 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19113 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
19115 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
19116 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
19117 paired_divv2sf3): Similar.
19118 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
19119 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
19120 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
19122 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
19123 Rename the paired_* instruction patterns.
19124 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
19126 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
19127 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
19128 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
19129 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
19130 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
19131 PAIRED_VECTOR_MODE.
19132 (struct machine_function): Delete field spe_insn_chain_scanned_p.
19133 (spe_func_has_64bit_regs_p): Delete.
19134 (spe_expand_predicate_builtin): Delete.
19135 (spe_expand_evsel_builtin): Delete.
19136 (TARGET_DWARF_REGISTER_SPAN): Do not define.
19137 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
19138 (invalid_e500_subreg): Delete.
19139 (rs6000_legitimize_address): Always force_reg op2 as well, for
19140 paired single memory accesses.
19141 (rs6000_member_type_forces_blk): Delete.
19142 (rs6000_spe_function_arg): Delete.
19143 (rs6000_expand_unop_builtin): Delete SPE handling.
19144 (rs6000_expand_binop_builtin): Ditto.
19145 (spe_expand_stv_builtin): Delete.
19146 (bdesc_2arg_spe): Delete.
19147 (spe_expand_builtin): Delete.
19148 (spe_expand_predicate_builtin): Delete.
19149 (spe_expand_evsel_builtin): Delete.
19150 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
19151 (spe_init_builtins): Delete.
19152 (spe_func_has_64bit_regs_p): Delete.
19153 (savres_routine_name): Delete "info" parameter. Adjust callers.
19154 (rs6000_emit_stack_reset): Ditto.
19155 (rs6000_dwarf_register_span): Delete.
19156 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
19157 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
19158 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
19160 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
19162 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
19163 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
19164 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
19165 mulv2sf3, divv2sf3): Delete expanders.
19167 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19169 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
19171 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19173 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
19174 * config/rs6000/rs6000.c: Ditto.
19176 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19178 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
19179 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
19180 comparison_operator.
19182 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19184 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
19185 * config/rs6000/rs6000.opt: Ditto.
19186 * config/rs6000/t-rtems: Ditto.
19188 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19190 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
19191 TARGET_E500_SINGLE by 0, simplify.
19192 * config/rs6000/rs6000.c: Ditto.
19193 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
19194 (spe_build_register_parallel): Delete.
19195 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
19196 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
19197 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
19198 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
19199 (E500_CONVERT): Delete.
19200 * config/rs6000/spe.md: Remove many patterns and all define_constants.
19202 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19204 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
19205 * config/rs6000/dfp.md: Ditto.
19206 (negdd2, *negdd2_fpr): Merge.
19207 (absdd2, *absdd2_fpr): Merge.
19208 (negtd2, *negtd2_fpr): Merge.
19209 (abstd2, *abstd2_fpr): Merge.
19210 * config/rs6000/e500.h: Delete file.
19211 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
19212 TARGET_FPRS by 1 and simplify.
19213 * config/rs6000/rs6000-c.c: Ditto.
19214 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
19215 TARGET_DF_SPE by 0.
19216 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
19218 * config/rs6000/rs6000.md: Ditto.
19219 (floatdidf2, *floatdidf2_fpr): Merge.
19220 (move_from_CR_gt_bit): Delete.
19221 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
19222 (E500_CR_IOR_COMPARE): Delete.
19223 (All patterns that require !TARGET_FPRS): Delete.
19224 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
19226 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19228 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
19230 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19232 * graphds.c (add_edge): Intitialize edge's attached data.
19233 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
19234 pointer parameter. Call pointed function on each edge during
19235 graph traversing. Skip traversing the edge when the function
19237 (graphds_dfs, graphds_scc): Ditto.
19238 (for_each_edge): New parameter. Pass the new parameter to callback
19240 * graphds.h (skip_edge_callback): New function pointer type.
19241 (graphds_dfs, graphds_scc): New function pointer parameter.
19242 (graphds_edge_callback, for_each_edge): New parameter.
19244 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19246 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
19247 out code checking if runtime alias check is possible to below ...
19248 Call the new function.
19249 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
19250 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
19252 2017-06-07 Marek Polacek <polacek@redhat.com>
19255 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
19256 TYPE_OVERFLOW_WRAPS check.
19258 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19260 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
19261 if versioning is required.
19262 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
19263 peeling with the check for versioning.
19265 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19267 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
19268 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
19269 Set true to new parameter if new ssa variable is defined.
19270 (vect_gen_vector_loop_niters): Refactor. Set range information
19271 for the new vector loop bound variable.
19272 (vect_do_peeling): Ditto.
19274 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19276 * tree-affine.c (ssa.h): Include header file.
19277 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
19278 has wrapping overflow behavior.
19280 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19282 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
19284 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19286 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
19287 (tree_to_aff_combination): ... here.
19289 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19291 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
19292 reg_pressure model function.
19293 (ivopts_global_cost_for_size): Delete.
19294 (determine_set_costs, iv_ca_recount_cost): Call new model function
19295 ivopts_estimate_reg_pressure.
19297 2017-06-07 Tamar Christina <tamar.christina@arm.com>
19299 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
19300 expensive than udiv. Remove floating point cases from mod.
19302 2017-06-07 Tamar Christina <tamar.christina@arm.com>
19304 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
19305 Increase idiv cost.
19307 2017-06-07 Tamar Christina <tamar.christina@arm.com>
19309 * config/aarch64/aarch64.md
19310 (copysignsf3): Fix mask generation.
19312 2017-06-07 Jakub Jelinek <jakub@redhat.com>
19314 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
19316 (class dump_manager): Add register_dumps method.
19317 * dumpfile.c: Include langhooks.h.
19318 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
19319 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
19320 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
19321 (dump_manager::dump_register): Start with 512 entries instead of 32.
19322 (dump_manager::register_dumps): New method.
19323 * toplev.c (general_init): Instead of invoking register_dumps
19324 langhook, invoke register_dumps method on the dump manager.
19325 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
19328 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
19330 * doc/md.texi: Clarify the restrictions on a define_insn condition.
19331 Say that # requires an associated define_split to exist, and that
19332 the define_split must be suitable for use after register allocation.
19334 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19336 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
19337 (compute_outgoing_frequencies): Also initialize zero counts.
19338 (find_many_sub_basic_blocks): Do not produce uninitialized profile
19339 around loops; preserve more of profile when nothing changes.
19341 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
19343 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
19345 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
19346 * config/arm/arm-cpu-cdata.h: Regenerate.
19347 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
19348 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
19349 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
19350 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
19351 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
19353 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
19355 * config/arm/t-rmprofile: Likewise.
19356 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
19358 2017-06-06 David S. Miller <davem@davemloft.net>
19361 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
19362 blockage if function uses alloca.
19364 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19366 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
19367 New "uid" fields to hold pretty-print IDs of group and ref.
19368 Memory references are now identified as <group_id>:<ref_id>
19369 instead of using [random] addresses.
19370 (dump_mem_details): Simplify, no functional change.
19371 (dump_mem_ref): Simplify and make output more concise.
19372 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
19373 (find_or_create_group): Initialize group uid.
19374 (record_ref): Initialize ref uid. Improve debug output.
19375 (prune_group_by_reuse, should_issue_prefetch_p,)
19376 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
19377 (mark_nontemporal_store, determine_loop_nest_reuse):
19378 Improve debug output.
19380 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19382 * dbgcnt.def (prefetch): New debug counter.
19383 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
19384 (schedule_prefetches): Stop issueing prefetches if debug counter
19387 2017-06-06 Tom de Vries <tom@codesourcery.com>
19389 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
19390 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
19392 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19394 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
19395 Use aarch64_reg_or_zero predicate for operand 4.
19396 (aarch64_compare_and_swap<mode> define_insn_and_split):
19397 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
19398 (aarch64_store_exclusive<mode>): Likewise for operand 2.
19400 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
19402 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
19403 (arm_compute_save_core_reg_mask): This.
19404 (thumb1_compute_save_reg_mask): Rename into ...
19405 (thumb1_compute_save_core_reg_mask): This.
19406 (arm_compute_save_reg0_reg12_mask): Adapt comment.
19407 (arm_compute_frame_layout): Likewise.
19409 2017-06-06 Richard Biener <rguenther@suse.de>
19411 PR tree-optimization/80974
19412 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
19413 keep or clear leaders SSA info.
19415 2017-06-06 Tom de Vries <tom@codesourcery.com>
19417 * config/nvptx/nvptx.c (split_mode_p): New function.
19418 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
19420 2017-06-06 Tom de Vries <tom@codesourcery.com>
19422 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
19424 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19427 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
19430 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19432 * shrink-wrap.c (handle_simple_exit): Update profile.
19433 (try_shrink_wrapping): Upate profile.
19435 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19437 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
19438 (tree_guess_outgoing_edge_probabilities): New.
19439 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
19440 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
19442 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19444 * ipa-split.c (split_function): Initialize return bb profile.
19446 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19448 * profile.c (compute_branch_probabilities): Also initialize
19449 EXIT_BLOCK profile.
19451 2017-06-06 Richard Biener <rguenther@suse.de>
19453 PR tree-optimization/80928
19454 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
19455 (vect_analyze_loop_operations): Properly guard analysis for
19457 (vect_transform_loop): Likewise.
19458 (vect_analyze_loop_2): Also reset SLP type on PHIs.
19459 (vect_model_induction_cost): Do not cost for pure SLP.
19460 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
19461 of induction in inner loop vectorization.
19462 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
19463 (vect_get_and_check_slp_defs): Handle vect_induction_def.
19464 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
19466 (vect_analyze_slp_cost_1): Cost induction.
19467 (vect_detect_hybrid_slp_stmts): Handle PHIs.
19468 (vect_get_slp_vect_defs): Likewise.
19469 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
19470 (vect_transform_stmt): Handle SLP reductions.
19471 * tree-vectorizer.h (vectorizable_induction): Adjust.
19473 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
19475 * config/rs6000/rs6000.c (make_resolver_func): Update
19476 init_lowered_empty_function call.
19478 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
19480 * doc/invoke.texi: Document the -fprofile-abs-path option.
19481 * common.opt (fprofile-abs-path): New option.
19482 * gcov-io.h (gcov_write_filename): Declare.
19483 * gcov-io.c (gcov_write_filename): New function.
19484 * coverage.c (coverage_begin_function): Use gcov_write_filename.
19485 * profile.c (output_location): Likewise.
19487 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19489 * shring-wrap.c: Revert accidental commit.
19491 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
19493 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
19495 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19497 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
19499 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
19500 profile in callgraph edge.
19501 * profile-count.h (apply_probability): If THIS is 0, then result is 0
19502 (apply_scale): Likewise.
19503 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
19504 Also scale profile when inlining function with zero profile.
19505 (initialize_cfun): Update exit block profile even when it is zero.
19506 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
19507 when profile is read.
19509 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
19511 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
19512 (CLONE_*): New constants to define the processors we can generate
19513 code for with the target_clone attribute.
19514 (rs6000_clone_map): New array to identify which clone processors
19515 the current program is running on.
19516 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
19517 target_clone attribute.
19518 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
19519 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
19520 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
19521 (cpu_expand_builtin): Add support for target_clone attribute.
19522 (rs6000_valid_attribute_p): Allow "default" attribute.
19523 (get_decl_name): New debug function to simplify printing the
19524 current function name in debugging statements.
19525 (rs6000_clone_priority): New functions to support the target_clone
19526 attribute, and be able to generate code to switch between ISA 2.05
19527 through ISA 3.0 (power6 through power9).
19528 (rs6000_compare_version_priority): Likewise.
19529 (rs6000_get_function_versions_dispatcher): Likewise.
19530 (make_resolver_func): Likewise.
19531 (add_condition_to_bb): Likewise.
19532 (dispatch_function_versions): Likewise.
19533 (rs6000_generate_version_dispatcher_body): Likewise.
19534 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
19535 (fusion_gpr_load_p): Fix a spacing issue.
19536 * doc/extend.texi (Common Function Attributes): Document that the
19537 PowerPC supports the target_clone attribute.
19539 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
19541 * config/arm/arm.h: explain F symbol found in description of ARM
19542 register allocation in its legend.
19544 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19546 * config/mips/frame-header-opt.c: Include profile-count.h.
19547 * config/riscv/riscv.c: Include profile-count.h
19549 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19551 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
19553 (sm_set_flag_if_changed): Add bbs field.
19554 (execute_sm_if_changed_flag_set): Pass BBS.
19555 (execute_sm): Update.
19557 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19559 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
19562 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19564 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
19565 (peephole2): New peephole2 to emit the above.
19566 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
19568 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19570 * config/aarch64/aarch64.c (define_peephole2 above
19571 *sub_<shift>_<mode>): New peephole.
19573 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19575 * config/i386/i386.c (make_resolver_func): Update.
19576 * Makefile.in: Add profile-count.h and profile-count.o
19577 * auto-profile.c (afdo_indirect_call): Update to new API.
19578 (afdo_set_bb_count): Update.
19579 (afdo_propagate_edge): Update.
19580 (afdo_propagate_circuit): Update.
19581 (afdo_calculate_branch_prob): Update.
19582 (afdo_annotate_cfg): Update.
19583 * basic-block.h: Include profile-count.h
19584 (struct edge_def): Turn count to profile_count.
19585 (struct basic_block_def): Likewie.
19586 (REG_BR_PROB_BASE): Move to profile-count.h
19587 (RDIV): Move to profile-count.h
19588 * bb-reorder.c (max_entry_count): Turn to profile_count.
19589 (find_traces): Update.
19590 (rotate_loop):Update.
19591 (connect_traces):Update.
19592 (sanitize_hot_paths):Update.
19593 * bt-load.c (migrate_btr_defs): Update.
19594 * cfg.c (RDIV): Remove.
19595 (init_flow): Use alloc_block.
19596 (alloc_block): Uninitialize count.
19597 (unchecked_make_edge): Uninitialize count.
19598 (check_bb_profile): Update.
19599 (dump_edge_info): Update.
19600 (dump_bb_info): Update.
19601 (update_bb_profile_for_threading): Update.
19602 (scale_bbs_frequencies_int): Update.
19603 (scale_bbs_frequencies_gcov_type): Update.
19604 (scale_bbs_frequencies_profile_count): New.
19605 * cfg.h (update_bb_profile_for_threading): Update.
19606 (scale_bbs_frequencies_profile_count): Declare.
19607 * cfgbuild.c (compute_outgoing_frequencies): Update.
19608 (find_many_sub_basic_blocks): Update.
19609 * cfgcleanup.c (try_forward_edges): Update.
19610 (try_crossjump_to_edge): Update.
19611 * cfgexpand.c (expand_gimple_tailcall): Update.
19612 (construct_exit_block): Update.
19613 * cfghooks.c (verify_flow_info): Update.
19614 (dump_bb_for_graph): Update.
19615 (split_edge): Update.
19616 (make_forwarder_block): Update.
19617 (duplicate_block): Update.
19618 (account_profile_record): Update.
19619 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
19620 (get_estimated_loop_iterations): Update.
19621 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
19622 (single_likely_exit): Update.
19623 * cfgloopmanip.c (scale_loop_profile): Update.
19625 (set_zero_probability): Update.
19626 (lv_adjust_loop_entry_edge): Update.
19627 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
19628 (purge_dead_edges): Update.
19629 (rtl_account_profile_record): Update.
19630 * cgraph.c (cgraph_node::create): Uninitialize count.
19631 (symbol_table::create_edge): Uninitialize count.
19632 (cgraph_update_edges_for_call_stmt_node): Update.
19633 (cgraph_edge::dump_edge_flags): Update.
19634 (cgraph_node::dump): Update.
19635 (cgraph_edge::maybe_hot_p): Update.
19636 * cgraph.h: Include profile-count.h
19637 (create_clone), create_edge, create_indirect_edge): Update.
19638 (cgraph_node): Turn count to profile_count.
19639 (cgraph_edge0: Likewise.
19640 (make_speculative, clone): Update.
19641 (create_edge): Update.
19642 (init_lowered_empty_function): Update.
19643 * cgraphclones.c (cgraph_edge::clone): Update.
19644 (duplicate_thunk_for_node): Update.
19645 (cgraph_node::create_clone): Update.
19646 * cgraphunit.c (cgraph_node::analyze): Update.
19647 (cgraph_node::expand_thunk): Update.
19648 * final.c (dump_basic_block_info): Update.
19649 * gimple-streamer-in.c (input_bb): Update.
19650 * gimple-streamer-out.c (output_bb): Update.
19651 * graphite.c (print_global_statistics): Update.
19652 (print_graphite_scop_statistics): Update.
19653 * hsa-brig.c: Include basic-block.h.
19654 * hsa-dump.c: Include basic-block.h.
19655 * hsa-gen.c (T sum_slice): Update.
19656 (convert_switch_statements):Update.
19657 * hsa-regalloc.c: Include basic-block.h.
19658 * ipa-chkp.c (chkp_produce_thunks): Update.
19659 * ipa-cp.c (struct caller_statistics): Update.
19660 (init_caller_stats): Update.
19661 (gather_caller_stats): Update.
19662 (ipcp_cloning_candidate_p): Update.
19663 (good_cloning_opportunity_p): Update.
19664 (get_info_about_necessary_edges): Update.
19665 (dump_profile_updates): Update.
19666 (update_profiling_info): Update.
19667 (update_specialized_profile): Update.
19668 (perhaps_add_new_callers): Update.
19669 (decide_about_value): Update.
19670 (ipa_cp_c_finalize): Update.
19671 * ipa-devirt.c (struct odr_type_warn_count): Update.
19672 (struct decl_warn_count): Update.
19673 (struct final_warning_record): Update.
19674 (possible_polymorphic_call_targets): Update.
19675 (ipa_devirt): Update.
19676 * ipa-fnsummary.c (redirect_to_unreachable): Update.
19677 * ipa-icf.c (sem_function::merge): Update.
19678 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
19679 * ipa-inline.c (compute_uninlined_call_time): Update.
19680 (compute_inlined_call_time): Update.
19681 (want_inline_small_function_p): Update.
19682 (want_inline_self_recursive_call_p): Update.
19683 (edge_badness): Update.
19684 (lookup_recursive_calls): Update.
19685 (recursive_inlining): Update.
19686 (inline_small_functions): Update.
19687 (dump_overall_stats): Update.
19688 (dump_inline_stats): Update.
19689 * ipa-profile.c (ipa_profile_generate_summary): Update.
19690 (ipa_propagate_frequency): Update.
19691 (ipa_profile): Update.
19692 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
19693 * ipa-utils.c (ipa_merge_profiles): Update.
19694 * loop-doloop.c (doloop_modify): Update.
19695 * loop-unroll.c (report_unroll): Update.
19696 (unroll_loop_runtime_iterations): Update.
19697 * lto-cgraph.c (lto_output_edge): Update.
19698 (lto_output_node): Update.
19699 (input_node): Update.
19700 (input_edge): Update.
19701 (merge_profile_summaries): Update.
19702 * lto-streamer-in.c (input_cfg): Update.
19703 * lto-streamer-out.c (output_cfg): Update.
19704 * mcf.c (create_fixup_graph): Update.
19705 (adjust_cfg_counts): Update.
19706 (sum_edge_counts): Update.
19707 * modulo-sched.c (sms_schedule): Update.
19708 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
19709 * predict.c (maybe_hot_count_p): Update.
19710 (probably_never_executed): Update.
19711 (dump_prediction): Update.
19712 (combine_predictions_for_bb): Update.
19713 (propagate_freq): Update.
19714 (handle_missing_profiles): Update.
19715 (counts_to_freqs): Update.
19716 (rebuild_frequencies): Update.
19717 (force_edge_cold): Update.
19718 * predict.h: Include profile-count.h
19719 (maybe_hot_count_p, counts_to_freqs): UPdate.
19720 * print-rtl-function.c: Do not include cfg.h
19721 * print-rtl.c: Include basic-block.h
19722 * profile-count.c: New file.
19723 * profile-count.h: New file.
19724 * profile.c (is_edge_inconsistent): Update.
19725 (correct_negative_edge_counts): Update.
19726 (is_inconsistent): Update.
19727 (set_bb_counts): Update.
19728 (read_profile_edge_counts): Update.
19729 (compute_frequency_overlap): Update.
19730 (compute_branch_probabilities): Update; Initialize and deinitialize
19732 (branch_prob): Update.
19733 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
19734 (edge_gcov_count): New.
19735 (bb_gcov_count): New.
19736 * shrink-wrap.c (try_shrink_wrapping): Update.
19737 * tracer.c (better_p): Update.
19738 * trans-mem.c (expand_transaction): Update.
19739 (ipa_tm_insert_irr_call): Update.
19740 (ipa_tm_insert_gettmclone_call): Update.
19741 * tree-call-cdce.c: Update.
19742 * tree-cfg.c (gimple_duplicate_sese_region): Update.
19743 (gimple_duplicate_sese_tail): Update.
19744 (gimple_account_profile_record): Update.
19745 (execute_fixup_cfg): Update.
19746 * tree-inline.c (copy_bb): Update.
19747 (copy_edges_for_bb): Update.
19748 (initialize_cfun): Update.
19749 (freqs_to_counts): Update.
19750 (copy_cfg_body): Update.
19751 (expand_call_inline): Update.
19752 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
19753 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
19754 (try_unroll_loop_completely): Update.
19755 (try_peel_loop): Update.
19756 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19757 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
19758 * tree-ssa-loop-split.c (connect_loops): Update.
19759 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
19760 * tree-ssa-reassoc.c (branch_fixup): Update.
19761 * tree-ssa-tail-merge.c (replace_block_by): Update.
19762 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
19763 (compute_path_counts): Update.
19764 (update_profile): Update.
19765 (recompute_probabilities): Update.
19766 (update_joiner_offpath_counts): Update.
19767 (estimated_freqs_path): Update.
19768 (freqs_to_counts_path): Update.
19769 (clear_counts_path): Update.
19770 (ssa_fix_duplicate_block_edges): Update.
19771 (duplicate_thread_path): Update.
19772 * tree-switch-conversion.c (case_bit_test_cmp): Update.
19773 (struct switch_conv_info): Update.
19774 * tree-tailcall.c (decrease_profile): Update.
19775 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
19776 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19777 * value-prof.c (check_counter): Update.
19778 (gimple_divmod_fixed_value): Update.
19779 (gimple_mod_pow2): Update.
19780 (gimple_mod_subtract): Update.
19781 (gimple_ic_transform): Update.
19782 (gimple_stringop_fixed_value): Update.
19783 * value-prof.h (gimple_ic): Update.
19785 2017-06-02 Carl Love <cel@us.ibm.com>
19787 * config/rs6000/rs6000-c: Add support for built-in functions
19788 vector double vec_doublee (vector signed int);
19789 vector double vec_doublee (vector unsigned int);
19790 vector double vec_doublee (vector float);
19791 vector double vec_doubleh (vector signed int);
19792 vector double vec_doubleh (vector unsigned int);
19793 vector double vec_doubleh (vector float);
19794 vector double vec_doublel (vector signed int);
19795 vector double vec_doublel (vector unsigned int);
19796 vector double vec_doublel (vector float);
19797 vector double vec_doubleo (vector signed int);
19798 vector double vec_doubleo (vector unsigned int);
19799 vector double vec_doubleo (vector float);.
19800 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
19801 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
19803 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
19804 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
19805 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
19807 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
19808 vec_doublel, vec_doubleh.
19809 * doc/extend.texi: Update the built-in documentation file for the
19810 new built-in functions.
19812 2017-06-02 David Malcolm <dmalcolm@redhat.com>
19815 * ipa-inline-analysis.c (free_growth_caches): Set
19816 edge_removal_hook_holder to NULL after removing it.
19818 2017-06-02 Sudakshina Das <sudi.das@arm.com>
19820 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
19821 comparision with zero.
19823 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
19824 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
19825 for early expansion of vec_min and vec_max builtins.
19826 (builtin_function_type): Add min/max unsigned variants to those
19827 identified as having unsigned arguments.
19829 2017-06-02 Olivier Hainque <hainque@adacore.com>
19831 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
19833 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19835 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
19836 Use VALL_F16 iterator rather than VALL.
19838 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19840 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
19841 Emit CBNZ inside loop when doing a strong exchange and comparing
19842 against zero. Generate the CC flags after the loop.
19844 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
19846 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
19847 (dl_section_ref): New.
19848 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
19849 On AIX, append an expression to subtract the size of the
19850 section length to dl_section_ref.
19852 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
19854 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
19855 for early expansion of vector absolute builtins.
19857 2017-06-02 Richard Biener <rguenther@suse.de>
19859 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
19860 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
19862 2017-06-02 Richard Biener <rguenther@suse.de>
19864 PR tree-optimization/80948
19865 * tree-tailcall.c (find_tail_calls): Track stmts to move in
19866 stmt order as well.
19868 2017-06-02 Richard Biener <rguenther@suse.de>
19870 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
19872 * tree-vect-stmts.c (process_use): Do not mark backedge defs
19873 for inductions as relevant.
19875 2017-06-02 Richard Biener <rguenther@suse.de>
19877 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
19878 (vectorizable_induction): ... this. Remove dead code.
19880 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
19882 * builtins. (expand_builtin_alloca): Remove second parameter and
19883 infer its value from the first parameter instead.
19884 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
19886 2017-06-02 Jakub Jelinek <jakub@redhat.com>
19888 PR rtl-optimization/80903
19889 * loop-doloop.c (add_test): Unshare sequence.
19891 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
19893 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
19895 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
19897 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
19899 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
19900 xlogue_layout::get_instance, logue_layout::xlogue_layout,
19901 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
19902 (xlogue_layout::get_stub_rtx): Make static.
19903 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
19904 (xlogue_layout::compute_stub_managed_regs): Rename to...
19905 (xlogue_layout::count_stub_managed_regs): ...this.
19906 (xlogue_layout::is_stub_managed_reg): New function.
19907 (xlogue_layout::m_stub_names): Rename to...
19908 (xlogue_layout::s_stub_names): ...this, make static.
19909 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
19910 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
19911 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
19912 xlogue_layout::s_stub_names): Instantiate statics.
19913 (stub_managed_regs): Remove.
19914 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
19915 (disable_call_ms2sysv_xlogues): Rename to...
19916 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
19917 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
19919 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
19920 change after reload_completed.
19921 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
19923 (ix86_expand_prologue): Likewise.
19924 (ix86_expand_epilogue): Likewise.
19925 (ix86_expand_split_stack_prologue): Likewise.
19926 (ix86_compute_frame_layout): Remove frame parameter ...
19927 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
19928 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
19930 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
19931 (ix86_frame): Move from here ...
19932 * config/i386/i386.h (ix86_frame): ... to here.
19933 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
19934 complete ix86_frame data structure instead. Remove some_ld_name.
19936 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
19938 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
19939 symbols that hold a DECL_VALUE_EXPR.
19941 2017-06-01 Martin Jambor <mjambor@suse.cz>
19943 PR tree-optimization/80898
19944 * tree-sra.c (process_subtree_disqualification): Removed.
19945 (disqualify_candidate): Do not acll
19946 process_subtree_disqualification.
19947 (subtree_mark_written_and_enqueue): New function.
19948 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
19949 RHS has been disqualified and re-queue LHS if necessary. Apart
19950 from that, ignore disqualified RHS.
19952 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19954 * config/s390/s390.c (s390_emit_epilogue): Disable early return
19955 address fetch for z10 or later.
19957 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19959 * config/arc/arc.md (tst_movb): Add guard when splitting.
19961 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19963 * config/arc/arc.c (arc_can_eliminate): Test against
19964 arc_frame_pointer_needed.
19966 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19968 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
19969 to prevent store reordering.
19970 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
19971 (type): Add block type.
19972 (stack_tie): Define special instruction to be used in
19975 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19977 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
19978 constraint. It is not valid for the pattern.
19979 (noncommutative_binary_comparison): Likewise.
19981 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19983 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
19986 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19988 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
19989 be used by the reg-alloc.
19991 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19993 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
19994 reg-alloc when having mul64 or mul32x16 instructions.
19995 (mulsidi3): Likewise.
19996 (umulsidi3): Likewise.
19997 (mulsi32x16): New pattern.
19998 (mulsi64): Likewise.
19999 (mulsidi64): Likewise.
20000 (umulsidi64): Likewise.
20001 (MUL32x16_REG): Define.
20002 (mul64_600): Use MUL32x16_REG.
20003 (mac64_600): Likewise.
20004 (umul64_600): Likewise.
20005 (umac64_600): Likewise.
20007 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20009 * config/arc/arc.md (mulsi3_700): Make it commutative.
20011 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
20013 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
20015 (*sign_extendsidi2_insn): Likewise for movstosw.
20017 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
20019 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
20020 the type of the input discriminant value. Convert the
20021 discriminant value of signedness vary.
20023 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
20025 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
20026 Add to -Wall section.
20028 2017-06-01 Richard Biener <rguenther@suse.de>
20030 PR middle-end/66313
20031 * fold-const.c (fold_plusminus_mult_expr): If the factored
20032 factor may be zero use a wrapping type for the inner operation.
20033 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
20034 and handle moved defs.
20035 (process_assignment): Properly guard the unary op case. Return a
20036 tri-state indicating that moving the stmt before the call may allow
20037 to continue. Pass through to_move.
20038 (find_tail_calls): Handle moving unrelated defs before
20041 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
20044 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
20045 splitter result in the canonical way.
20047 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
20049 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
20050 also for 32bit target. Update insn attributes.
20051 (zero-extendsidi2 splitter): Allow all registers for operand 1.
20053 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
20055 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
20056 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
20057 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
20058 (_mm_maskz_min_ss): New intrinsics.
20060 2017-05-31 Martin Liska <mliska@suse.cz>
20062 * tree-vect-loop.c (vect_create_epilog_for_reduction):
20063 Change comment style to one we normally use.
20064 (vectorizable_reduction): Likewise.
20065 (vectorizable_induction): Likewise.
20066 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
20067 (vectorizable_call): Likewise.
20068 (vectorizable_simd_clone_call): Likewise.
20069 (vectorizable_conversion): Likewise.
20070 (vectorizable_assignment): Likewise.
20071 (vectorizable_shift): Likewise.
20072 (vectorizable_operation): Likewise.
20073 (vectorizable_store): Likewise.
20074 (vectorizable_load): Likewise.
20075 * tree-vectorizer.h: Likewise.
20077 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
20079 * passes.c (emergency_dump_function): New.
20080 * tree-pass.h (emergency_dump_function): Declare.
20081 * plugin.c (plugins_internal_error_function): Remove.
20082 * plugin.h (plugins_internal_error_function): Remove declaration.
20083 * toplev.c (internal_error_function): New static function. Use it...
20084 (general_init): ...here.
20086 2017-05-31 Graham Markall <graham.markall@embecosm.com>
20088 * config/arc/arc.c (arc_print_operand): Handle constant operands.
20089 (arc_rtx_costs): Add costs for new patterns.
20090 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
20091 * config/arc/predicates.md: Add _1_2_3_operand predicate.
20093 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
20095 * tree-ssa-strlen.c (get_next_strinfo): New function.
20096 (get_stridx_plus_constant): Use it.
20097 (zero_length_string): Likewise.
20098 (adjust_related_strinfos): Likewise.
20099 (adjust_last_stmt): Likewise.
20101 2017-05-31 Richard Biener <rguenther@suse.de>
20104 * config/i386/i386.c (ix86_expand_builtin): Remove assert
20105 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
20107 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
20109 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
20110 loop_vinfo argument and use of dependence distance vectors.
20111 Check instead whether the two references differ only in their
20112 initial value and assume that they have the same alignment if the
20113 difference is a multiple of the vector alignment.
20114 (vect_analyze_data_refs_alignment): Update call accordingly.
20116 2017-05-31 Martin Liska <mliska@suse.cz>
20119 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
20121 2017-05-31 Bin Cheng <bin.cheng@arm.com>
20123 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
20124 (create_intersect_range_checks): Move from ...
20125 * tree-data-ref.c (create_intersect_range_checks_index)
20126 (create_intersect_range_checks): ... to here.
20127 (create_runtime_alias_checks): New function factored from ...
20128 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
20129 here. Call above function.
20130 * tree-data-ref.h (create_runtime_alias_checks): New function.
20132 2017-05-31 Bin Cheng <bin.cheng@arm.com>
20134 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
20135 segment length for dr_b and compute it in wide_int.
20137 2017-05-31 Richard Biener <rguenther@suse.de>
20139 PR tree-optimization/80906
20140 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
20141 and pass through iv_map.
20142 (copy_bb_and_scalar_dependences): Adjust.
20143 (translate_pending_phi_nodes): Likewise.
20144 (copy_loop_close_phi_args): Handle code-generating IVs instead
20147 2017-05-30 David Malcolm <dmalcolm@redhat.com>
20149 * diagnostic-color.c (color_dict): Add "type-diff".
20150 (parse_gcc_colors): Update comment.
20151 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
20152 -fdiagnostics-show-template-tree and -fno-elide-type.
20153 (GCC_COLORS): Add type-diff to example.
20155 (-fdiagnostics-show-template-tree): New.
20156 (-fno-elide-type): New.
20157 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
20158 the pp_format_decoder callback. Call any m_format_postprocessor's
20160 (pretty_printer::pretty_printer): Initialize
20161 m_format_postprocessor.
20162 (pretty_printer::~pretty_printer): Delete any
20163 m_format_postprocessor.
20164 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
20165 (class format_postprocessor): New class.
20166 (struct pretty_printer::format_decoder): Document the new parameters.
20167 (struct pretty_printer::m_format_postprocessor): New field.
20168 * tree-diagnostic.c (default_tree_printer): Update for new
20169 bool and const char ** params.
20170 * tree-diagnostic.h (default_tree_printer): Likewise.
20172 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
20174 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
20175 (lwa_operand): Delete rs6000_gen_cell_microcode test.
20176 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
20177 rs6000_gen_cell_microcode code.
20178 (rs6000_final_prescan_insn): Delete.
20179 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
20180 "warn-cell-microcode" entries.
20181 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
20182 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
20183 throughout. Change cc_reg_not_micro_cr0_operand to
20184 cc_reg_not_cr0_operand throughout.
20185 (*extendhi<mode>2_noload): Delete.
20186 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
20187 (mwarn-cell-microcode): Delete.
20188 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
20189 -mgen-cell-microcode and -mwarn-cell-microcode.
20191 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
20194 * config/i386/constraints.md (Yd): New constraint.
20196 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
20197 and (?Yd, r) alternatives. Update insn attributes.
20198 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
20199 and (?*Yd, r) alternatives. Update insn attributes.
20200 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
20202 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
20204 * gimplify.c (gimplify_modify_expr): Don't create a
20205 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
20208 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
20210 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
20212 2017-05-30 Richard Biener <rguenther@suse.de>
20214 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
20215 and reduc_def fields.
20216 (STMT_VINFO_REDUC_TYPE): New define.
20217 (STMT_VINFO_REDUC_DEF): Likewise.
20218 (vect_force_simple_reduction): Adjust prototype.
20219 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
20220 (vect_is_simple_reduction): Remove check_reduction argument.
20221 (vect_force_simple_reduction): Adjust and set
20222 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
20223 (vectorizable_reduction): Do not re-do reduction analysis
20224 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
20225 * tree-parloops.c (gather_scalar_reductions): Adjust.
20227 2017-05-30 Richard Biener <rguenther@suse.de>
20229 PR middle-end/80901
20230 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
20233 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
20235 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
20236 Introduce unknown_misalignment parameter and remove vf.
20237 (vect_peeling_hash_get_lowest_cost):
20238 Pass unknown_misalignment parameter.
20239 (vect_enhance_data_refs_alignment):
20240 Fix unsupportable data ref treatment.
20242 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
20244 * tree-vect-data-refs.c (vect_get_data_access_cost):
20245 Workaround for SLP handling.
20246 (vect_enhance_data_refs_alignment):
20247 Compute costs for doing no peeling at all, compare to the best
20248 peeling costs so far and avoid peeling if cheaper.
20250 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
20252 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
20253 Return peeling info and set costs to zero for unlimited cost
20255 (vect_enhance_data_refs_alignment): Also inspect all datarefs
20256 with unknown misalignment. Compute and costs for unknown
20257 misalignment, compare them to the costs for known misalignment
20258 and choose the cheapest for peeling.
20260 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
20262 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
20263 (vect_get_peeling_costs_all_drs): Create function.
20264 (vect_peeling_hash_get_lowest_cost):
20265 Use vect_get_peeling_costs_all_drs.
20266 (vect_peeling_supportable): Create function.
20267 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
20269 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
20271 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
20272 DR_HAS_NEGATIVE_STEP.
20273 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
20274 (vect_enhance_data_refs_alignment): Use.
20275 (vect_duplicate_ssa_name_ptr_info): Use.
20276 * tree-vectorizer.h (dr_misalignment): Use.
20277 (known_alignment_for_access_p): Use.
20279 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20282 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
20284 (has_section_name): New function.
20286 2017-05-30 Martin Liska <mliska@suse.cz>
20289 * auto-profile.c (get_function_decl_from_block): Fix
20292 2017-05-30 Richard Biener <rguenther@suse.de>
20294 PR middle-end/80876
20295 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
20297 2017-05-30 Martin Liska <mliska@suse.cz>
20299 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
20300 * dumpfile.h (struct dump_file_info): Remove ctors.
20302 2017-05-30 Martin Liska <mliska@suse.cz>
20304 * predict.def: Fix GNU coding style.
20306 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
20308 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
20309 Mark 'to' argument with ATTRIBUTE_UNUSED.
20311 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
20313 * config/xtensa/xtensa.c (xtensa_emit_call): Use
20314 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
20315 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
20318 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
20320 * doc/install.texi (Options specification): Restore entry of
20321 --enable-sjlj-exceptions.
20323 2017-05-27 Michael Eager <eager@eagercon.com>
20326 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
20328 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
20330 * config/microblaze/microblaze.h
20331 (FIXED_REGISTERS): Update in macro.
20332 (CALL_USED_REGISTERS): Update in macro.
20334 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
20336 * doc/install.texi: Add links to macOS binary distributions.
20338 2017-05-27 Jakub Jelinek <jakub@redhat.com>
20342 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20344 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
20346 2017-05-26 Martin Liska <mliska@suse.cz>
20348 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
20350 2017-05-26 Martin Liska <mliska@suse.cz>
20352 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
20353 always leading ';; '.
20354 (dump_bb_info): Likewise.
20355 (brief_dump_cfg): Likewise.
20356 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
20357 * dumpfile.c: Remove usage of TDF_VERBOSE.
20358 * dumpfile.h (enum dump_kind): Likewise.
20359 (dump_gimple_bb_header): Do not use TDF_COMMENT.
20360 * print-tree.c (debug_verbose): Remove.
20361 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
20362 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
20363 * tree-diagnostic.c (default_tree_printer): Replace
20364 TDF_DIAGNOSTIC with TDF_SLIM.
20366 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20368 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
20369 in parameter loop, rather than loop_vinfo.
20370 (create_intersect_range_checks): Ditto.
20371 (vect_create_cond_for_alias_checks): Update call to above functions.
20373 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20375 PR tree-optimization/80815
20376 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
20377 for merging runtime alias checks. Handle negative DR_STEPs.
20379 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20381 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
20383 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
20384 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
20385 out code pruning runtime alias checks.
20386 * tree-data-ref.c (prune_runtime_alias_test_list): New function
20387 factored out from above.
20388 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
20390 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
20392 (prune_runtime_alias_test_list): New decalaration.
20394 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20396 * tree-vect-data-refs.c (compare_tree): Rename and move ...
20397 * tree-data-ref.c (data_ref_compare_tree): ... to here.
20398 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
20399 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
20400 (operator==, comp_dr_with_seg_len_pair): Ditto.
20401 (vect_prune_runtime_alias_test_list): Ditto.
20403 2017-05-26 Martin Liska <mliska@suse.cz>
20406 * params.def: Bound partial-inlining-entry-probability param.
20408 2017-05-26 Marek Polacek <polacek@redhat.com>
20411 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
20414 2017-05-26 Richard Biener <rguenther@suse.de>
20416 PR tree-optimization/80842
20417 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
20420 2017-05-26 Richard Biener <rguenther@suse.de>
20422 PR tree-optimization/80844
20423 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
20425 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
20427 * doc/md.texi (Machine Constraints): Update x86 family
20428 machine constraints section to match 'config/i386/constraints.md'.
20430 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
20432 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
20434 2017-05-25 Nathan Sidwell <nathan@acm.org>
20436 * doc/invoke.texi (--enable-languages): Update documentation.
20438 2017-05-25 Martin Liska <mliska@suse.cz>
20440 * dumpfile.c: Add TDF_FOLDING.
20441 * dumpfile.h (enum dump_kind): Likewise.
20442 * genmatch.c (dt_simplify::gen_1): Use it.
20444 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20446 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
20448 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20450 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
20451 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
20453 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20455 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
20456 * match.pd (X == C): Rewrite it here.
20457 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
20458 with_certain_nonzero_bits2): New predicates.
20459 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
20461 2017-05-24 Nathan Sidwell <nathan@acm.org>
20463 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
20466 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
20469 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
20471 * config/powerpcspe: New port. Files are copied from the rs6000
20472 port, with "rs6000" in filenames replaced by "powerpcspe".
20474 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
20476 PR rtl-optimization/80754
20477 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
20479 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
20481 * config/sparc/sparc.md (length): Return the correct value for -mflat
20482 sibcalls to match output_sibcall.
20484 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
20488 * config/rs6000/rs6000.c (struct machine_function): Add new field
20490 (rs6000_get_separate_components): Init that field, use it.
20491 (rs6000_components_for_bb): Use the field.
20493 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
20495 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
20497 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
20499 PR middle-end/80823
20500 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
20502 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20505 * config/s390/s390.c (s390_check_qrst_address): Check incoming
20506 address against address_operand predicate.
20507 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
20509 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
20511 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
20512 parameters passed indirectly.
20514 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
20516 * config/i386/i386.md (*movdi_internal): Remove SSE4
20517 alternative 18 (?r, *v). Update insn attributes.
20518 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
20519 Update insn attributes.
20520 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
20521 Update insn attributes.
20522 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
20523 alternative 1 (r, v). Remove isa attribute.
20524 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
20525 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
20526 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
20528 2017-05-23 Tom de Vries <tom@codesourcery.com>
20530 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
20533 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20535 * cgraphunit.c (symbol_table::process_new_functions): Update.
20536 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
20537 (inline_generate_summary): Rename to ...
20538 (ipa_fn_summary_generate): ... this one.
20539 (inline_read_summary): Rename to ...
20540 (ipa_fn_summary_read): ... this one.
20541 (inline_write_summary): Rename to ...
20542 (ipa_fn_summary_write): ... this one.
20543 (inline_free_summary): Rename to ...
20544 (ipa_free_fn_summary): ... this one.
20545 (pass_data_local_fn_summary, pass_local_fn_summary,
20546 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
20547 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
20548 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
20549 make_pass_ipa_fn_summary): New.
20550 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
20551 inline_write_summary, inline_free_summary): Remove.
20552 (ipa_free_fn_summary) : New.
20553 * ipa-inline.c (ipa_inline): Update.
20554 (pass_ipa_inline): Do not generate summaries.
20555 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
20557 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
20558 and add pass_ipa_fn_summary.
20559 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
20561 (make_pass_inline_parameters): Remove.
20563 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
20565 * omp-low.c (struct omp_context): Remove "default_kind" member.
20568 * omp-offload.c (execute_oacc_device_lower): Remove the
20569 parallelism dimensions function attributes for unparallelized
20570 OpenACC kernels constructs.
20572 2017-05-23 Martin Liska <mliska@suse.cz>
20574 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
20576 (cgraph_edge::make_speculative): Likewise.
20577 (cgraph_edge::resolve_speculation): Likewise.
20578 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
20579 (cgraph_node::dump): Likewise.
20580 * cgraph.h: Likewise.
20581 * cgraphunit.c (analyze_functions): Likewise.
20582 (symbol_table::compile): Likewise.
20583 * ipa-cp.c (print_all_lattices): Likewise.
20584 (determine_versionability): Likewise.
20585 (initialize_node_lattices): Likewise.
20586 (ipcp_verify_propagated_values): Likewise.
20587 (estimate_local_effects): Likewise.
20588 (update_profiling_info): Likewise.
20589 (create_specialized_node): Likewise.
20590 (perhaps_add_new_callers): Likewise.
20591 (decide_about_value): Likewise.
20592 (decide_whether_version_node): Likewise.
20593 (identify_dead_nodes): Likewise.
20594 (ipcp_store_bits_results): Likewise.
20595 * ipa-devirt.c (dump_targets): Likewise.
20596 (ipa_devirt): Likewise.
20597 * ipa-icf.c (sem_item::dump): Likewise.
20598 (sem_function::equals): Likewise.
20599 (sem_variable::equals): Likewise.
20600 (sem_item_optimizer::read_section): Likewise.
20601 (sem_item_optimizer::execute): Likewise.
20602 (congruence_class::dump): Likewise.
20603 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
20604 (dump_inline_summary): Likewise.
20605 (estimate_node_size_and_time): Likewise.
20606 (inline_analyze_function): Likewise.
20607 * ipa-inline-transform.c (inline_call): Likewise.
20608 * ipa-inline.c (report_inline_failed_reason): Likewise.
20609 (want_early_inline_function_p): Likewise.
20610 (edge_badness): Likewise.
20611 (update_edge_key): Likewise.
20612 (inline_small_functions): Likewise.
20613 * ipa-profile.c (ipa_profile): Likewise.
20614 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
20615 (ipa_make_edge_direct_to_target): Likewise.
20616 (remove_described_reference): Likewise.
20617 (ipa_impossible_devirt_target): Likewise.
20618 (propagate_controlled_uses): Likewise.
20619 (ipa_print_node_params): Likewise.
20620 (ipcp_transform_function): Likewise.
20621 * ipa-pure-const.c (pure_const_read_summary): Likewise.
20622 (propagate_pure_const): Likewise.
20623 * ipa-reference.c (generate_summary): Likewise.
20624 (read_write_all_from_decl): Likewise.
20625 (propagate): Likewise.
20626 (ipa_reference_read_optimization_summary): Likewise.
20627 * ipa-utils.c (ipa_merge_profiles): Likewise.
20628 * ipa.c (walk_polymorphic_call_targets): Likewise.
20629 (symbol_table::remove_unreachable_nodes): Likewise.
20630 (ipa_single_use): Likewise.
20631 * passes.c (execute_todo): Likewise.
20632 * predict.c (drop_profile): Likewise.
20633 * symtab.c (symtab_node::get_dump_name): New function.
20634 (symtab_node::dump_name): Likewise.
20635 (symtab_node::dump_asm_name): Likewise.
20636 (symtab_node::dump_references): Likewise.
20637 (symtab_node::dump_referring): Likewise.
20638 (symtab_node::dump_base): Likewise.
20639 (symtab_node::debug_symtab): Likewise.
20640 * tree-sra.c (convert_callers_for_node): Likewise.
20641 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
20642 * value-prof.c (init_node_map): Likewise.
20644 2017-05-23 Martin Liska <mliska@suse.cz>
20646 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
20647 and symtab_node::debug_symtab to symbol_table::debug.
20648 * cgraphunit.c (analyze_functions): Use the renamed function.
20649 (symbol_table::compile): Likewise.
20650 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
20651 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
20652 * passes.c (execute_todo): Likewise.
20653 * symtab.c (symbol_table::dump): New function.
20654 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
20656 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20658 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
20659 that nonconst implies exec.
20661 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20663 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
20664 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
20665 (inline_edge_summary_vec): Turn into ...
20666 (ipa_call_summaries): ... this one.
20667 (redirect_to_unreachable, edge_set_predicate,
20668 evaluate_properties_for_edge, inline_summary_alloc,
20669 reset_ipa_call_summary, reset_inline_summary,
20670 inline_summary_t::duplicate): Update.
20671 (inline_edge_duplication_hook): Turn to ...
20672 (ipa_call_summary_t::duplicate): ... this one.
20673 (inline_edge_removal_hook): Turn to ...
20674 (ipa_call_summary_t::remove): ... this one.
20675 (dump_inline_edge_summary): Turn to ...
20676 (dump_ipa_call_summary): ... this one.
20677 (estimate_function_body_sizes): Update.
20678 (inline_update_callee_summaries): Update.
20679 (remap_edge_change_prob): Update.
20680 (remap_edge_summaries): Update.
20681 (inline_merge_summary): Update.
20682 (do_estimate_edge_time): Update.
20683 (inline_generate_summary): Update.
20684 (inline_read_section): Update.
20685 (inline_read_summary): Update.
20686 (inline_free_summary): Update.
20687 * ipa-inline.c (can_inline_edge_p): Update.
20688 (compute_inlined_call_time): Update.
20689 (want_inline_small_function_p): Update.
20690 (edge_badness): Update.
20691 (early_inliner): Update.
20692 * ipa-inline.h (inline_edge_summary): Turn to ...
20693 (ipa_call_summary): ... this one.
20694 (ipa_call_summary_t): New class.
20695 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
20696 (ipa_call_summaries): New.
20697 (inline_edge_summary): Remove.
20698 (estimate_edge_growth): Update.
20699 * ipa-profile.c (ipa_propagate_frequency_1): Update.
20700 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
20701 * ipa-split.c (execute_split_functions): Update.
20702 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
20704 2017-05-23 Tom de Vries <tom@codesourcery.com>
20706 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
20707 attributes): Document rdrand effective target.
20709 2017-05-23 Tom de Vries <tom@codesourcery.com>
20711 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
20712 attributes): Sort alphabetically.
20714 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
20716 * config/avr/genmultilib.awk: Use gsub instead of gensub.
20718 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
20721 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
20722 V2DF/V2DI splat into two separate patterns, one that handles
20723 registers, and the other that only handles memory. Drop support
20724 for splatting from a GPR on ISA 2.07 and then splitting the
20725 splat into direct move and splat.
20726 (vsx_splat_<mode>_reg): Likewise.
20727 (vsx_splat_<mode>_mem): Likewise.
20729 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
20731 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
20733 2017-05-22 Jakub Jelinek <jakub@redhat.com>
20735 PR middle-end/80809
20736 * omp-low.c (finish_taskreg_remap): New function.
20737 (finish_taskreg_scan): If unit size of ctx->record_type
20738 is non-constant, unshare the size expression and replace
20739 decls in it with possible outer var refs.
20741 PR middle-end/80809
20742 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
20743 GOVD_SHARED rather than GOVD_PRIVATE with it.
20744 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
20745 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
20747 PR middle-end/80853
20748 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
20749 as last argument to build_outer_var_ref for pointer bases of array
20750 section reductions.
20752 2017-05-19 Martin Sebor <msebor@redhat.com>
20754 * print-tree.c (print_node): Print DECL_READ_P flag.
20756 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20758 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
20759 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
20760 * cgraph.c: Likewise.
20761 * cgraphunit.c: Likewise.
20762 * gengtype.c: Likewise.
20763 * ipa-cp.c: Likewise.
20764 * ipa-devirt.c: Likewise.
20765 * ipa-icf.c: Likewise.
20766 * ipa-predicate.c: Likewise.
20767 * ipa-profile.c: Likewise.
20768 * ipa-prop.c: Likewise.
20769 * ipa-split.c: Likewise.
20771 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
20772 edge_predicate_pool, dump_inline_hints,
20773 inline_summary::account_size_time, redirect_to_unreachable,
20774 edge_set_predicate, set_hint_predicate,
20775 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
20776 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
20777 inline_summary_t::remove, remap_hint_predicate_after_duplication,
20778 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
20779 ipa_call_summary_t::remove, initialize_growth_caches,
20780 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
20781 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
20782 mark_modified, unmodified_parm_1, unmodified_parm,
20783 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
20784 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
20785 compute_bb_predicates, will_be_nonconstant_expr_predicate,
20786 will_be_nonconstant_predicate, record_modified_bb_info,
20787 get_minimal_bb, record_modified, param_change_prob,
20788 phi_result_unknown_predicate, predicate_for_phi_result,
20789 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
20790 estimate_function_body_sizes, compute_inline_parameters,
20791 compute_inline_parameters_for_curren, pass_data_inline_parameters,
20792 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
20793 inline_update_callee_summaries, remap_edge_change_prob,
20794 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
20795 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
20796 inline_analyze_function, inline_summary_t::insert,
20797 inline_generate_summary, read_ipa_call_summary, inline_read_section,
20798 inline_read_summary, write_ipa_call_summary, inline_write_summary,
20799 inline_free_summary): Move to ipa-fnsummary.h
20800 (predicate_t): Remove.
20801 * ipa-fnsummary.c: New file.
20802 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
20803 (enum inline_hints_vals, inline_hints, agg_position_info,
20804 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
20805 inline_summaries, ipa_call_summary, ipa_call_summary_t,
20806 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
20807 dump_inline_summary, dump_inline_hints, inline_generate_summary,
20808 inline_read_summary, inline_write_summary, inline_free_summary,
20809 inline_analyze_function, initialize_inline_failed,
20810 inline_merge_summary, inline_update_overall_summary,
20811 compute_inline_parameters): Move to ipa-fnsummary.h
20812 * ipa-fnsummary.h: New file.
20813 * ipa-inline-transform.h: Include ipa-inline.h.
20814 * ipa-inline.c: LIkewise.
20816 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20818 * ipa-inline.c (edge_badness): Use inlined_time instead of
20819 inline_summaries->get.
20821 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20823 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
20825 2017-05-22 Nathan Sidwell <nathan@acm.org>
20827 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
20828 (fdump-lang): Document 'raw' option.
20829 * dumpfile.h (TDI_tu): Delete.
20830 * dumpfile.c (dump_files): Remove translation-unit.
20831 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
20833 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
20835 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
20836 command option from $(AWK) call.
20837 * config/avr/genmultilib.awk: Simplify and rewrite so that it
20838 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
20839 [FORMAT]: Remove handling of variable.
20840 * config/avr/t-multilib: Regenerate.
20842 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20844 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
20846 (dump_inline_summary): Do not print self_time.
20847 (estimate_function_body_sizes): Do not set self_time.
20848 (compute_inline_parameters): Likewise.
20849 (inline_read_section, inline_write_summary): Do not stream self_time.
20850 * ipa-inline.h (inline_summary): Drop self_time.
20852 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20854 * ipa-inline-analysis.c (account_size_time): Rename to ...
20855 (inline_summary::account_size_time): ... this one.
20856 (reset_ipa_call_summary): Turn to ...
20857 (ipa_call_summary::reset): ... this one.
20858 (reset_inline_summary): Turn to ...
20859 (inline_summary::reset): ... this one.
20860 (inline_summary_t::remove): Update.
20861 (inline_summary_t::duplicate): Update.
20862 (ipa_call_summary_t::remove): Update.
20863 (dump_inline_summary): Update.
20864 (estimate_function_body_sizes): Update.
20865 (compute_inline_parameters): Update.
20866 (estimate_node_size_and_time): Update.
20867 (inline_merge_summary): Update.
20868 (inline_update_overall_summary): Update.
20869 (inline_read_section): Update.
20870 (inline_write_summary): Update.
20871 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
20872 add account_size_time and reset member functions.
20873 (ipa_call_summary): Add reset function.
20874 * ipa-predicate.h (predicate::operator &): Constify.
20876 2017-05-22 Richard Biener <rguenther@suse.de>
20878 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
20880 2017-05-19 Jason Merrill <jason@redhat.com>
20882 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
20884 2017-05-19 Marek Polacek <polacek@redhat.com>
20887 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
20888 TYPE_OVERFLOW_WRAPS checks.
20890 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
20892 * tree-core.h (enum omp_clause_default_kind): Add
20893 "OMP_CLAUSE_DEFAULT_PRESENT".
20894 * tree-pretty-print.c (dump_omp_clause): Handle it.
20895 * gimplify.c (enum gimplify_omp_var_data): Add
20896 "GOVD_MAP_FORCE_PRESENT".
20897 (gimplify_adjust_omp_clauses_1): Map it to
20898 "GOMP_MAP_FORCE_PRESENT".
20899 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
20901 * gimplify.c (oacc_default_clause): Clarify.
20903 2017-05-19 Nathan Sidwell <nathan@acm.org>
20905 LANG_HOOK_REGISTER_DUMPS
20906 * toplev.c (general_init): Call register dump lang hook.
20907 * doc/invoke.texi: Document -fdump-lang option family.
20908 * dumpfile.c (dump_files): Remove class dump here.
20909 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
20910 * dumpfile.h (tree_dump_index): Remove TDI_class.
20911 * langhooks-def.h (lhd_register_dumps): Declare.
20912 (LANG_HOOKS_REGISTER_DUMPS): Define.
20913 (LANG_HOOKS_INITIALIZER): Add it.
20914 * langhooks.c (lhd_register_dumps): Define.
20915 * langhooks.h (struct lang_hooks): Add register_dumps.
20917 2017-05-19 Nathan Sidwell <nathan@acm.org>
20919 * context.h (context::set_passes): New.
20920 * context.c (context::context): Do not create pass manager.
20921 * toplev.c (general_init): Create pass manager here.
20923 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
20925 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
20926 use this splitter if two add or or instructions would also work for
20927 the constant we want to generate.
20929 2017-05-19 Richard Biener <rguenther@suse.de>
20932 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
20933 predicate evaluation.
20935 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
20937 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
20939 * ipa-inline.c (want_inline_small_function_p): Do not cast to
20942 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
20944 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
20945 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
20946 (inline_edge_summary_vec): Turn into ...
20947 (ipa_call_summaries): ... this one.
20948 (redirect_to_unreachable, edge_set_predicate,
20949 evaluate_properties_for_edge, inline_summary_alloc,
20950 reset_ipa_call_summary, reset_inline_summary,
20951 inline_summary_t::duplicate): Update.
20952 (inline_edge_duplication_hook): Turn to ...
20953 (ipa_call_summary_t::duplicate): ... this one.
20954 (inline_edge_removal_hook): Turn to ...
20955 (ipa_call_summary_t::remove): ... this one.
20956 (dump_inline_edge_summary): Turn to ...
20957 (dump_ipa_call_summary): ... this one.
20958 (estimate_function_body_sizes): Update.
20959 (inline_update_callee_summaries): Update.
20960 (remap_edge_change_prob): Update.
20961 (remap_edge_summaries): Update.
20962 (inline_merge_summary): Update.
20963 (do_estimate_edge_time): Update.
20964 (inline_generate_summary): Update.
20965 (inline_read_section): Update.
20966 (inline_read_summary): Update.
20967 (inline_free_summary): Update.
20968 * ipa-inline.c (can_inline_edge_p): Update.
20969 (compute_inlined_call_time): Update.
20970 (want_inline_small_function_p): Update.
20971 (edge_badness): Update.
20972 (early_inliner): Update.
20973 * ipa-inline.h (inline_edge_summary): Turn to ...
20974 (ipa_call_summary): ... this one.
20975 (ipa_call_summary_t): New class.
20976 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
20977 (ipa_call_summaries): New.
20978 (inline_edge_summary): Remove.
20979 (estimate_edge_growth): Update.
20980 * ipa-profile.c (ipa_propagate_frequency_1): Update.
20981 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
20982 * ipa-split.c (execute_split_functions): Update.
20983 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
20985 2017-05-19 Richard Biener <rguenther@suse.de>
20987 PR middle-end/80764
20988 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
20990 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
20992 * config/rs6000/rs6000.c (struct machine_function): Add field
20993 fpr_is_wrapped_separately.
20994 (rs6000_get_separate_components): Use 64 components. Handle the
20995 new FPR components.
20996 (rs6000_components_for_bb): Handle the FPR components.
20997 (rs6000_emit_prologue_components): Handle the FPR components.
20998 (rs6000_emit_epilogue_components): Handle the FPR components.
20999 (rs6000_set_handled_components): Handle the FPR components.
21000 (rs6000_emit_prologue): Don't output prologue code for those FPRs
21001 that are already separately shrink-wrapped.
21002 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
21003 that are already separately shrink-wrapped.
21005 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
21008 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
21011 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
21012 (define_peephole2 for Altivec d-form load): Add peepholes to catch
21013 cases where the register allocator uses a move and an offsettable
21014 memory operation to/from a FPR register on ISA 2.06/2.07.
21015 (define_peephole2 for Altivec d-form store): Likewise.
21017 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
21020 * config/i386/mmx.md (*mov<mode>_internal): Enable
21021 alternatives 11, 12, 13 and 14 also for 32bit targets.
21022 Remove alternatives 15, 16, 17 and 18.
21023 * config/i386/sse.md (vec_concatv2di): Change
21024 alternative (!x, *y) to (x, ?!*Yn).
21026 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
21028 * dumpfile.h (enum dump_kind): Remove stray comma.
21030 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
21032 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
21033 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
21034 predicate::num_conditions
21035 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
21036 (CHANGED): turn into predicate::changed.
21037 (agg_position_info): Move to ipa-predicate.h
21038 (add_condition, predicate::add_clause, predicate::operator &=,
21039 predicate::or_with, predicate::evaluate, predicate::probability,
21040 dump_condition, dump_clause, predicate::dump,
21041 predicate::remap_after_duplication, predicate::remap_after_inlining,
21042 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
21043 (evaluate_conditions_for_known_args): Update.
21044 (set_cond_stmt_execution_predicate): Update.
21045 * ipa-inline.h: Include ipa-predicate.h
21046 (condition, inline_param_summary, conditions, agg_position_info,
21047 predicate): Move to ipa-predicate.h
21048 * ipa-predicate.c: New file.
21049 * ipa-predicate.h: New file.
21051 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
21053 * final.c (leaf_function_p): Check we are not in a sequence.
21055 2017-05-18 Martin Liska <mliska@suse.cz>
21057 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
21058 * dumpfile.c (dump_register): Use new enum dump_kind.
21059 (get_dump_file_name): Likewise.
21060 (dump_enable_all): Likewise.
21061 (dump_switch_p_1): Likewise.
21062 (enable_rtl_dump_file): Remove usage of TDF_RTL.
21063 * dumpfile.h (enum dump_kind): New enum type.
21064 (struct dump_file_info): Create constructor and
21065 format fields and comments.
21066 * passes.c (pass_manager::register_one_dump_file):
21068 * statistics.c (statistics_early_init): Likewise.
21069 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
21070 TDF_TREE with TDF_SLIM.
21071 (gather_memory_references_ref): Likewise.
21073 2017-05-18 Martin Liska <mliska@suse.cz>
21075 * vec.h (struct vnull): Use it.
21077 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
21079 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
21080 (true_predicate, false_predicate, true_predicate_p,
21081 false_predicate_p): Remove.
21082 (single_cond_predicate, not_inlined_predicate): Turn to member function
21084 (add_condition): Update.
21085 (add_clause): Turn to...
21086 (predicate::add_clause): ... this one; update; allow passing NULL
21088 (and_predicates): Turn to ...
21089 (predicate::operator &=): ... this one.
21090 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
21091 (or_predicates): Turn to ...
21092 (predicate::or_with): ... this one.
21093 (evaluate_predicate): Turn to ...
21094 (predicate::evaluate): ... this one.
21095 (predicate_probability): Turn to ...
21096 (predicate::probability): ... this one.
21097 (dump_condition): Update.
21098 (dump_predicate): Turn to ...
21099 (predicate::dump): ... this one.
21100 (account_size_time): Update.
21101 (edge_set_predicate): Update.
21102 (set_hint_predicate): UPdate.
21103 (evaluate_conditions_for_known_args): Update.
21104 (evaluate_properties_for_edge): Update.
21105 (remap_predicate_after_duplication): Turn to...
21106 (predicate::remap_after_duplication): ... this one.
21107 (remap_hint_predicate_after_duplication): Update.
21108 (inline_summary_t::duplicate): UPdate.
21109 (dump_inline_edge_summary): Update.
21110 (dump_inline_summary): Update.
21111 (set_cond_stmt_execution_predicate): Update.
21112 (set_switch_stmt_execution_predicate): Update.
21113 (compute_bb_predicates): Update.
21114 (will_be_nonconstant_expr_predicate): Update.
21115 (will_be_nonconstant_predicate): Update.
21116 (phi_result_unknown_predicate): Update.
21117 (predicate_for_phi_result): Update.
21118 (array_index_predicate): Update.
21119 (estimate_function_body_sizes): Update.
21120 (estimate_node_size_and_time): Update.
21121 (estimate_ipcp_clone_size_and_time): Update.
21122 (remap_predicate): Rename to ...
21123 (predicate::remap_after_inlining): ... this one.
21124 (remap_hint_predicate): Update.
21125 (inline_merge_summary): Update.
21126 (inline_update_overall_summary): Update.
21127 (estimate_size_after_inlining): Update.
21128 (read_predicate): Rename to ...
21129 (predicate::stream_in): ... this one.
21130 (read_inline_edge_summary): Update.
21131 (write_predicate): Rename to ...
21132 (predicate::stream_out): ... this one.
21133 (write_inline_edge_summary): Update.
21134 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
21135 (clause_t): Turn to uint32_t
21136 (predicate): Turn to class; implement constructor and operators
21138 (size_time_entry): Update.
21139 (inline_summary): Update.
21140 (inline_edge_summary): Update.
21142 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
21144 * fold-const.c (fold_binary_loc): Move transformation...
21145 * match.pd (C - X CMP X): ... here.
21147 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
21149 * config/sparc/sparc.c (sparc_option_override): Set function
21150 alignment for -mcpu=niagara7 to 64 to match the I$ line.
21151 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
21153 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
21155 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
21157 2017-05-18 Marek Polacek <polacek@redhat.com>
21160 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
21161 (pass_ubsan::execute): Call gimple_assign_single_p instead of
21162 gimple_assign_load_p.
21164 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
21166 PR middle-end/80692
21167 * real.c (do_compare): Give decimal_do_compare preference over
21168 comparing just the signs.
21170 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
21172 * doc/md.texi (Canonicalization of Instructions): Describe the
21173 canonical form of instructions that inherently set a condition
21176 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
21178 PR middle-end/80775
21179 * tree-cfg.c: Move deletion of unreachable case statements to after
21180 the merging of consecutive case labels.
21182 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
21184 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
21185 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
21186 restoring of callee-saved registers.
21188 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
21190 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
21191 * config/visium/visium.c (single_set_and_flags): Likewise.
21192 * config/visium/visium.md (Substitutions): Likewise.
21194 2017-05-17 Martin Liska <mliska@suse.cz>
21196 * cfg.c: Introduce dump_flags_t type and
21197 use it instead of int type.
21199 * cfghooks.c: Likewise.
21200 * cfghooks.h (struct cfg_hooks): Likewise.
21201 * cfgrtl.c: Likewise.
21202 * cfgrtl.h: Likewise.
21203 * cgraph.c (cgraph_node::get_body): Likewise.
21204 * coretypes.h: Likewise.
21205 * domwalk.c: Likewise.
21206 * domwalk.h: Likewise.
21207 * dumpfile.c (struct dump_option_value_info): Likewise.
21208 (dump_enable_all): Likewise.
21209 (dump_switch_p_1): Likewise.
21210 (opt_info_switch_p): Likewise.
21211 * dumpfile.h (enum tree_dump_index): Likewise.
21212 (struct dump_file_info): Likewise.
21213 * genemit.c: Likewise.
21214 * generic-match-head.c: Likewise.
21215 * gengtype.c (open_base_files): Likewise.
21216 * gimple-pretty-print.c: Likewise.
21217 * gimple-pretty-print.h: Likewise.
21218 * graph.c (print_graph_cfg): Likewise.
21219 * graphite-scop-detection.c (dot_all_sese): Likewise.
21220 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
21221 * loop-unroll.c (report_unroll): Likewise.
21222 * passes.c (pass_manager::register_one_dump_file): Likewise.
21223 * print-tree.c: Likewise.
21224 * statistics.c: Likewise.
21225 * tree-cfg.c: Likewise.
21226 * tree-cfg.h: Likewise.
21227 * tree-dfa.c: Likewise.
21228 * tree-dfa.h: Likewise.
21229 * tree-dump.c (dump_function): Likewise.
21230 * tree-dump.h (struct dump_info): Likewise.
21231 * tree-pretty-print.c: Likewise.
21232 * tree-pretty-print.h: Likewise.
21233 * tree-ssa-live.c: Likewise.
21234 * tree-ssa-live.h: Likewise.
21235 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
21236 * tree-vect-loop.c: Likewise.
21237 * tree-vect-slp.c: Likewise.
21239 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
21240 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21242 PR tree-optimization/80457
21243 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
21244 of all arguments to a statement as scalar_to_vec operations.
21245 (vectorizable_call): Adjust call to vect_model_simple_cost for
21247 (vectorizable_conversion): Likewise.
21248 (vectorizable_assignment): Likewise.
21249 (vectorizable_shift): Likewise.
21250 (vectorizable_operation): Likewise.
21251 (vectorizable_comparison): Likewise.
21252 (vect_is_simple_cond): Record the def types for operands.
21253 (vectorizable_condition): Likewise, call vect_model_simple_cost.
21254 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
21255 for statement argument count.
21257 2017-05-16 Carl Love <cel@us.ibm.com>
21259 * config/rs6000/rs6000-c: Add support for built-in functions
21260 vector unsigned long long vec_bperm (vector unsigned long long,
21261 vector unsigned char)
21262 vector signed long long vec_mule (vector signed int,
21264 vector unsigned long long vec_mule (vector unsigned int,
21265 vector unsigned int)
21266 vector signed long long vec_mulo (vector signed int,
21268 vector unsigned long long vec_mulo (vector unsigned int,
21269 vector unsigned int)
21270 vector signed char vec_sldw (vector signed char,
21271 vector signed char,
21273 vector unsigned char vec_sldw (vector unsigned char,
21274 vector unsigned char,
21276 vector signed short vec_sldw (vector signed short,
21277 vector signed short,
21279 vector unsigned short vec_sldw (vector unsigned short,
21280 vector unsigned short,
21282 vector signed int vec_sldw (vector signed int,
21285 vector unsigned int vec_sldw (vector unsigned int,
21286 vector unsigned int,
21288 vector signed long long vec_sldw (vector signed long long,
21289 vector signed long long,
21291 vector unsigned long long vec_sldw (vector unsigned long long,
21292 vector unsigned long long,
21294 * config/rs6000/rs6000-c: Add support for built-in functions
21295 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
21296 * config/rs6000/altivec.h: Add defintion for vec_sldw.
21297 * doc/extend.texi: Update the built-in documentation for the
21298 new built-in functions.
21300 2017-05-16 Marek Polacek <polacek@redhat.com>
21304 * tree.c (save_expr): Don't fold the expression.
21306 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
21308 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
21309 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
21310 and (?*y,m). Update insn attributes.
21312 2017-05-16 Martin Liska <mliska@suse.cz>
21314 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
21315 flags argument of print_gimple_stmt, print_gimple_expr,
21316 print_generic_stmt and print_generic_expr.
21317 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
21318 * coretypes.h: Likewise.
21319 * except.c (dump_eh_tree): Likewise.
21320 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
21321 * gimple-pretty-print.h: Likewise.
21322 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
21323 (backprop::push_to_worklist): Likewise.
21324 (backprop::pop_from_worklist): Likewise.
21325 (backprop::process_use): Likewise.
21326 (backprop::intersect_uses): Likewise.
21327 (note_replacement): Likewise.
21328 * gimple-ssa-store-merging.c
21329 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
21330 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
21331 (pass_store_merging::execute): Likewise.
21332 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
21333 (ssa_base_cand_dump_callback): Likewise.
21334 (dump_incr_vec): Likewise.
21335 (replace_refs): Likewise.
21336 (replace_mult_candidate): Likewise.
21337 (create_add_on_incoming_edge): Likewise.
21338 (create_phi_basis): Likewise.
21339 (insert_initializers): Likewise.
21340 (all_phi_incrs_profitable): Likewise.
21341 (introduce_cast_before_cand): Likewise.
21342 (replace_one_candidate): Likewise.
21343 * gimplify.c (gimplify_expr): Likewise.
21344 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
21345 (set_rename): Likewise.
21346 (rename_uses): Likewise.
21347 (copy_loop_phi_nodes): Likewise.
21348 (add_close_phis_to_merge_points): Likewise.
21349 (copy_loop_close_phi_args): Likewise.
21350 (copy_cond_phi_args): Likewise.
21351 (graphite_copy_stmts_from_block): Likewise.
21352 (translate_pending_phi_nodes): Likewise.
21353 * graphite-poly.c (print_pdr): Likewise.
21354 (dump_gbb_cases): Likewise.
21355 (dump_gbb_conditions): Likewise.
21356 (print_scop_params): Likewise.
21357 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
21358 (build_cross_bb_scalars_use): Likewise.
21359 (gather_bbs::before_dom_children): Likewise.
21360 * hsa-dump.c (dump_hsa_immed): Likewise.
21361 * ipa-cp.c (print_ipcp_constant_value): Likewise.
21362 (get_replacement_map): Likewise.
21363 * ipa-inline-analysis.c (dump_condition): Likewise.
21364 (estimate_function_body_sizes): Likewise.
21365 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
21366 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
21367 * ipa-prop.c (ipa_dump_param): Likewise.
21368 (ipa_print_node_jump_functions_for_edge): Likewise.
21369 (ipa_modify_call_arguments): Likewise.
21370 (ipa_modify_expr): Likewise.
21371 (ipa_dump_param_adjustments): Likewise.
21372 (ipa_dump_agg_replacement_values): Likewise.
21373 (ipcp_modif_dom_walker::before_dom_children): Likewise.
21374 * ipa-pure-const.c (check_stmt): Likewise.
21375 (pass_nothrow::execute): Likewise.
21376 * ipa-split.c (execute_split_functions): Likewise.
21377 * omp-offload.c (dump_oacc_loop_part): Likewise.
21378 (dump_oacc_loop): Likewise.
21379 * trans-mem.c (tm_log_emit): Likewise.
21380 (tm_memopt_accumulate_memops): Likewise.
21381 (dump_tm_memopt_set): Likewise.
21382 (dump_tm_memopt_transform): Likewise.
21383 * tree-cfg.c (gimple_verify_flow_info): Likewise.
21384 (print_loop): Likewise.
21385 * tree-chkp-opt.c (chkp_print_addr): Likewise.
21386 (chkp_gather_checks_info): Likewise.
21387 (chkp_get_check_result): Likewise.
21388 (chkp_remove_check_if_pass): Likewise.
21389 (chkp_use_outer_bounds_if_possible): Likewise.
21390 (chkp_reduce_bounds_lifetime): Likewise.
21391 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
21392 (chkp_mark_completed_bounds): Likewise.
21393 (chkp_register_incomplete_bounds): Likewise.
21394 (chkp_mark_invalid_bounds): Likewise.
21395 (chkp_maybe_copy_and_register_bounds): Likewise.
21396 (chkp_build_returned_bound): Likewise.
21397 (chkp_get_bound_for_parm): Likewise.
21398 (chkp_build_bndldx): Likewise.
21399 (chkp_get_bounds_by_definition): Likewise.
21400 (chkp_generate_extern_var_bounds): Likewise.
21401 (chkp_get_bounds_for_decl_addr): Likewise.
21402 * tree-chrec.c (chrec_apply): Likewise.
21403 * tree-data-ref.c (dump_data_reference): Likewise.
21404 (dump_subscript): Likewise.
21405 (dump_data_dependence_relation): Likewise.
21406 (analyze_overlapping_iterations): Likewise.
21407 * tree-inline.c (expand_call_inline): Likewise.
21408 (tree_function_versioning): Likewise.
21409 * tree-into-ssa.c (dump_defs_stack): Likewise.
21410 (dump_currdefs): Likewise.
21411 (dump_names_replaced_by): Likewise.
21412 (dump_update_ssa): Likewise.
21413 (update_ssa): Likewise.
21414 * tree-object-size.c (pass_object_sizes::execute): Likewise.
21415 * tree-parloops.c (build_new_reduction): Likewise.
21416 (try_create_reduction_list): Likewise.
21417 (ref_conflicts_with_region): Likewise.
21418 (oacc_entry_exit_ok_1): Likewise.
21419 (oacc_entry_exit_single_gang): Likewise.
21420 * tree-pretty-print.h: Likewise.
21421 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
21422 (get_scalar_evolution): Likewise.
21423 (add_to_evolution): Likewise.
21424 (get_loop_exit_condition): Likewise.
21425 (analyze_evolution_in_loop): Likewise.
21426 (analyze_initial_condition): Likewise.
21427 (analyze_scalar_evolution): Likewise.
21428 (instantiate_scev): Likewise.
21429 (number_of_latch_executions): Likewise.
21430 (gather_chrec_stats): Likewise.
21431 (final_value_replacement_loop): Likewise.
21432 (scev_const_prop): Likewise.
21433 * tree-sra.c (dump_access): Likewise.
21434 (disqualify_candidate): Likewise.
21435 (create_access): Likewise.
21436 (reject): Likewise.
21437 (maybe_add_sra_candidate): Likewise.
21438 (create_access_replacement): Likewise.
21439 (analyze_access_subtree): Likewise.
21440 (analyze_all_variable_accesses): Likewise.
21441 (sra_modify_assign): Likewise.
21442 (initialize_constant_pool_replacements): Likewise.
21443 (find_param_candidates): Likewise.
21444 (decide_one_param_reduction): Likewise.
21445 (replace_removed_params_ssa_names): Likewise.
21446 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
21447 * tree-ssa-copy.c (dump_copy_of): Likewise.
21448 (copy_prop_visit_cond_stmt): Likewise.
21449 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
21450 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
21451 (record_equivalences_from_stmt): Likewise.
21452 * tree-ssa-dse.c (compute_trims): Likewise.
21453 (delete_dead_call): Likewise.
21454 (delete_dead_assignment): Likewise.
21455 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
21456 (forward_propagate_into_cond): Likewise.
21457 (pass_forwprop::execute): Likewise.
21458 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
21459 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
21461 (move_computations_worker): Likewise.
21462 (execute_sm): Likewise.
21463 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
21464 (remove_exits_and_undefined_stmts): Likewise.
21465 (remove_redundant_iv_tests): Likewise.
21466 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
21467 (adjust_iv_update_pos): Likewise.
21468 * tree-ssa-math-opts.c (bswap_replace): Likewise.
21469 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
21470 (value_replacement): Likewise.
21471 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
21472 * tree-ssa-pre.c (print_pre_expr): Likewise.
21473 (get_representative_for): Likewise.
21474 (create_expression_by_pieces): Likewise.
21475 (insert_into_preds_of_block): Likewise.
21476 (eliminate_insert): Likewise.
21477 (eliminate_dom_walker::before_dom_children): Likewise.
21478 (eliminate): Likewise.
21479 (remove_dead_inserted_code): Likewise.
21480 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
21481 * tree-ssa-reassoc.c (get_rank): Likewise.
21482 (eliminate_duplicate_pair): Likewise.
21483 (eliminate_plus_minus_pair): Likewise.
21484 (eliminate_not_pairs): Likewise.
21485 (undistribute_ops_list): Likewise.
21486 (eliminate_redundant_comparison): Likewise.
21487 (update_range_test): Likewise.
21488 (optimize_range_tests_var_bound): Likewise.
21489 (optimize_vec_cond_expr): Likewise.
21490 (rewrite_expr_tree): Likewise.
21491 (rewrite_expr_tree_parallel): Likewise.
21492 (linearize_expr): Likewise.
21493 (break_up_subtract): Likewise.
21494 (linearize_expr_tree): Likewise.
21495 (attempt_builtin_powi): Likewise.
21496 (attempt_builtin_copysign): Likewise.
21497 (transform_stmt_to_copy): Likewise.
21498 (transform_stmt_to_multiply): Likewise.
21499 (dump_ops_vector): Likewise.
21500 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
21501 (print_scc): Likewise.
21502 (set_ssa_val_to): Likewise.
21503 (visit_reference_op_store): Likewise.
21504 (visit_use): Likewise.
21505 (sccvn_dom_walker::before_dom_children): Likewise.
21506 (run_scc_vn): Likewise.
21507 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
21509 (expr_hash_elt::print): Likewise.
21510 (const_and_copies::pop_to_marker): Likewise.
21511 (const_and_copies::record_const_or_copy_raw): Likewise.
21512 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
21513 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
21514 (dump_predicates): Likewise.
21515 (find_uninit_use): Likewise.
21516 (warn_uninitialized_phi): Likewise.
21517 (pass_late_warn_uninitialized::execute): Likewise.
21518 * tree-ssa.c (verify_vssa): Likewise.
21519 (verify_ssa): Likewise.
21520 (maybe_optimize_var): Likewise.
21521 * tree-vrp.c (dump_value_range): Likewise.
21522 (dump_all_value_ranges): Likewise.
21523 (dump_asserts_for): Likewise.
21524 (register_edge_assert_for_2): Likewise.
21525 (vrp_visit_cond_stmt): Likewise.
21526 (vrp_visit_switch_stmt): Likewise.
21527 (vrp_visit_stmt): Likewise.
21528 (vrp_visit_phi_node): Likewise.
21529 (simplify_cond_using_ranges_1): Likewise.
21530 (fold_predicate_in): Likewise.
21531 (evrp_dom_walker::before_dom_children): Likewise.
21532 (evrp_dom_walker::push_value_range): Likewise.
21533 (evrp_dom_walker::pop_value_range): Likewise.
21534 (execute_early_vrp): Likewise.
21536 2017-05-16 Richard Biener <rguenther@suse.de>
21538 * dwarf2out.c (loc_list_from_tree_1): Do not create
21539 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
21541 2017-05-16 Richard Biener <rguenther@suse.de>
21543 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
21545 (note_variable_value_in_expr): If we resolved the decl ref
21546 do not push to the stack.
21548 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
21550 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
21551 operations in fast-math mode.
21552 (vaddq_f16): Likewise.
21553 (vmul_f16): Likewise.
21554 (vmulq_f16): Likewise.
21555 (vsub_f16): Likewise.
21556 (vsubq_f16): Likewise.
21557 * config/arm/neon.md (add<mode>3): New.
21559 (fma:<VH:mode>3): New. Also remove outdated comment.
21562 2017-05-16 Martin Liska <mliska@suse.cz>
21566 * ipa-devirt.c (warn_types_mismatch): Fix typo.
21567 (odr_types_equivalent_p): Likewise.
21569 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
21571 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
21573 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
21576 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
21577 non-interunit SSE move alternatives with '?'.
21578 (zero-extendsidi peephole2): New peephole to skip intermediate
21579 general register in SSE zero-extend sequence.
21581 2017-05-15 Jeff Law <law@redhat.com>
21583 * reorg.c (relax_delay_slots): Create a new variable to hold
21584 the temporary target rather than clobbering TARGET_LABEL.
21586 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
21587 missing argument to extract_bit_field call.
21588 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
21590 2017-05-15 Martin Liska <mliska@suse.cz>
21593 * gcc.c (process_command): Do not allow empty argument of -o option.
21595 2017-05-15 Renlin Li <renlin.li@arm.com>
21597 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
21598 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
21599 * config/aarch64/constraints.md (Usf): Add long call check.
21600 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
21601 (call_value): Likewise.
21602 (sibcall): Likewise.
21603 (sibcall_value): Likewise.
21605 (call_value_insn): New.
21606 (sibcall_insn): Update rtx pattern.
21607 (sibcall_value_insn): Likewise.
21608 (call_internal): Remove.
21609 (call_value_internal): Likewise.
21610 (sibcall_internal): Likewise.
21611 (sibcall_value_internal): Likewise.
21612 (call_reg): Likewise.
21613 (call_symbol): Likewise.
21614 (call_value_reg): Likewise.
21615 (call_value_symbol): Likewise.
21617 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
21620 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
21622 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
21624 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
21625 compatible with CCGOCmode and with CCZmode.
21627 2017-05-14 Martin Sebor <msebor@redhat.com>
21629 PR middle-end/77671
21630 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
21631 (gimple_fold_builtin_snprintf): Same.
21632 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
21633 (gimple_fold_builtin_snprintf): Same.
21634 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
21635 of character types.
21636 (is_call_safe): New function.
21637 (try_substitute_return_value): Call it.
21638 (try_simplify_call): New function.
21639 (pass_sprintf_length::handle_gimple_call): Call it.
21641 2017-05-14 Martin Sebor <msebor@redhat.com>
21643 PR middle-end/80669
21644 * builtins.c (expand_builtin_stpncpy): Simplify.
21646 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
21648 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
21649 * config/i386/i386.h
21650 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
21651 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
21652 (struct machine_function): Add new members call_ms2sysv,
21653 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
21654 (struct machine_frame_state): New fields sp_realigned and
21655 sp_realigned_offset.
21656 * config/i386/i386.c
21657 (enum xlogue_stub): New enum.
21658 (enum xlogue_stub_sets): New enum.
21659 (class xlogue_layout): New class.
21660 (struct ix86_frame): New fields stack_realign_allocate_offset,
21661 stack_realign_offset and outlined_save_offset. Modify comments to
21662 detail stack layout when using out-of-line stubs.
21663 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
21664 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
21665 -mcall-ms2sysv-xlogues.
21666 (stub_managed_regs): New static variable.
21667 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
21668 registers managed by out-of-line stub.
21669 (disable_call_ms2sysv_xlogues): New function.
21670 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
21671 m->call_ms2sysv when appropriate and compute frame layout for
21673 (sp_valid_at, fp_valid_at): New inline functions.
21674 (choose_basereg): New function.
21675 (choose_baseaddr): Add align parameter, use choose_basereg and modify
21677 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
21678 Use align parameter of choose_baseaddr to generated aligned SSE movs
21680 (pro_epilogue_adjust_stack): Modify to track
21681 machine_frame_state::sp_realigned.
21682 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
21683 (ix86_nsaved_sseregs): Likewise.
21684 (ix86_emit_save_regs): Likewise.
21685 (ix86_emit_save_regs_using_mov): Likewise.
21686 (ix86_emit_save_sse_regs_using_mov): Likewise.
21687 (get_scratch_register_on_entry): Likewise.
21688 (gen_frame_set): New function.
21689 (gen_frame_load): Likewise.
21690 (gen_frame_store): Likewise.
21691 (emit_outlined_ms2sysv_save): Likewise.
21692 (emit_outlined_ms2sysv_restore): Likewise.
21693 (ix86_expand_prologue): Modify stack re-alignment code and call
21694 emit_outlined_ms2sysv_save when appropriate.
21695 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
21696 parameter rtx_insn *insn, which allows the function to be used to only
21697 generate the notes.
21698 (ix86_expand_epilogue): Modify validity checks of frame and stack
21699 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
21700 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
21701 * config/i386/predicates.md
21702 (save_multiple): New predicate.
21703 (restore_multiple): Likewise.
21704 * config/i386/sse.md
21705 (save_multiple<mode>): New pattern.
21706 (save_multiple_realign<mode>): Likewise.
21707 (restore_multiple<mode>): Likewise.
21708 (restore_multiple_and_return<mode>): Likewise.
21709 (restore_multiple_leave_return<mode>): Likewise.
21710 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
21712 2017-05-14 Julia Koval <julia.koval@intel.com>
21714 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
21715 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
21716 (__builtin_ia32_xsetbv): New builtins.
21717 * config/i386/i386.c (ix86_expand_special_args_builtin):
21719 (ix86_expand_builtin): Special expand for new intrinsics.
21720 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
21721 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
21722 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
21724 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21726 * cfganal.c (inverted_post_order_compute): Change argument type
21728 * cfganal.h (inverted_post_order_compute): Adjust prototype.
21729 * df-core.c (rest_of_handle_df_initialize): Adjust.
21730 (rest_of_handle_df_finish): Likewise.
21731 (df_analyze_1): Likewise.
21732 (df_analyze): Likewise.
21733 (loop_inverted_post_order_compute): Change argument to be a vec *.
21734 (df_analyze_loop): Adjust.
21735 (df_get_n_blocks): Likewise.
21736 (df_get_postorder): Likewise.
21737 * df.h (struct df_d): Change field to be a vec.
21738 * lcm.c (compute_laterin): Adjust.
21739 (compute_available): Likewise.
21740 * lra-lives.c (lra_create_live_ranges_1): Likewise.
21741 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
21742 * tree-ssa-pre.c (compute_antic): Likewise.
21744 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21746 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
21747 (depth_first_search::depth_first_search): Change structure init
21748 function to this constructor.
21749 (depth_first_search::add_bb): Rename function to this member.
21750 (depth_first_search::execute): Likewise.
21751 (flow_dfs_compute_reverse_finish): Adjust.
21753 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21755 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
21756 (longest_simple_path): Likewise.
21757 * shrink-wrap.c (spread_components): Likewise.
21758 (disqualify_problematic_components): Likewise.
21759 (emit_common_heads_for_components): Likewise.
21760 (emit_common_tails_for_components): Likewise.
21761 (insert_prologue_epilogue_for_components): Likewise.
21763 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21765 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
21768 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21770 * df-core.c (df_set_blocks): Start using auto_bitmap.
21771 (df_compact_blocks): Likewise.
21772 * df-problems.c (df_rd_confluence_n): Likewise.
21773 * df-scan.c (df_insn_rescan_all): Likewise.
21774 (df_process_deferred_rescans): Likewise.
21775 (df_update_entry_block_defs): Likewise.
21776 (df_update_exit_block_uses): Likewise.
21777 (df_entry_block_bitmap_verify): Likewise.
21778 (df_exit_block_bitmap_verify): Likewise.
21779 (df_scan_verify): Likewise.
21780 * lra-constraints.c (lra_constraints): Likewise.
21781 (undo_optional_reloads): Likewise.
21782 (lra_undo_inheritance): Likewise.
21783 * lra-remat.c (calculate_gen_cands): Likewise.
21784 (do_remat): Likewise.
21785 * lra-spills.c (assign_spill_hard_regs): Likewise.
21786 (spill_pseudos): Likewise.
21787 * tree-ssa-pre.c (bitmap_set_and): Likewise.
21788 (bitmap_set_subtract_values): Likewise.
21790 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21792 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
21793 management with auto_bitmap.
21794 (fix_inter_tick): Likewise.
21795 (fix_recovery_deps): Likewise.
21796 * ira.c (add_store_equivs): Likewise.
21797 (find_moveable_pseudos): Likewise.
21798 (split_live_ranges_for_shrink_wrap): Likewise.
21799 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
21800 (rtx_reuse_manager::seen_def_p): Likewise.
21801 (rtx_reuse_manager::set_seen_def): Likewise.
21802 * print-rtl.h (class rtx_reuse_manager): Likewise.
21804 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21806 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
21808 (migrate_btr_def): Likewise.
21809 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
21810 * df-core.c (loop_post_order_compute): Likewise.
21811 (loop_inverted_post_order_compute): Likewise.
21812 * hsa-common.h: Likewise.
21813 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
21814 * init-regs.c (initialize_uninitialized_regs): Likewise.
21815 * ipa-inline.c (resolve_noninline_speculation): Likewise.
21816 (inline_small_functions): Likewise.
21817 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
21818 * ira.c (combine_and_move_insns): Likewise.
21819 (build_insn_chain): Likewise.
21820 * loop-invariant.c (find_invariants): Likewise.
21821 * lower-subreg.c (propagate_pseudo_copies): Likewise.
21822 * predict.c (tree_predict_by_opcode): Likewise.
21823 (predict_paths_leading_to): Likewise.
21824 (predict_paths_leading_to_edge): Likewise.
21825 (estimate_loops_at_level): Likewise.
21826 (estimate_loops): Likewise.
21827 * shrink-wrap.c (try_shrink_wrapping): Likewise.
21828 (spread_components): Likewise.
21829 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
21830 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
21831 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
21832 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
21833 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
21834 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
21835 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
21836 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
21837 (mark_threaded_blocks): Likewise.
21838 (thread_through_all_blocks): Likewise.
21839 * tree-ssa.c (verify_ssa): Likewise.
21840 (execute_update_addresses_taken): Likewise.
21841 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
21843 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21845 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
21847 (post_order_compute): Likewise.
21848 (inverted_post_order_compute): Likewise.
21849 (pre_and_rev_post_order_compute_fn): Likewise.
21851 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21853 * genrecog.c (int_set::int_set): Explicitly construct our
21854 auto_vec base class.
21855 * vec.h (auto_vec::auto_vec): New constructor.
21857 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21859 * bitmap.h (class auto_bitmap): New constructor taking
21860 bitmap_obstack * argument.
21862 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21864 * bitmap.h (class auto_bitmap): Change type of m_bits to
21865 bitmap_head, and adjust ctor / dtor and member operators.
21867 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
21869 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
21870 when returned register mode doesn't match original mode.
21872 2017-05-12 Jeff Law <law@redhat.com>
21873 Jakub Jelinek <jakub@redhat.com>
21875 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
21876 we look for cc setter after the compare-elim changes.
21877 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
21878 within the vector to match what compare-elim now expects.
21879 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
21880 (xorsi3_flags, one_cmplsi2_flags): Likewise.
21882 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
21883 after the compare-elim changes.
21884 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
21885 the vector to match what compare-elim now expects.
21886 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
21887 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
21888 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
21889 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
21890 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
21892 * config/visium/visium.c (single_set_and_flags): Fix where
21893 we look for cc setter after the compare-elim changes.
21894 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
21895 with the vector to match what compare-elim now expects.
21896 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
21897 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
21898 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
21899 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
21900 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
21901 (neg<mode>2_insn_set_overflow): Likewise.
21903 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
21905 PR middle-end/79794
21906 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
21907 maybe_expand_insn call, set ops[0].target. If still set after call,
21908 set alt_rtl. Add extra arg to recursive calls.
21909 (extract_bit_field): Add alt_rtl argument. Pass to
21911 * expmed.h (extract_bit_field): Fix prototype.
21912 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
21913 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
21914 to extract_bit_field_calls.
21915 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
21916 Pass alt_rtl to extract_bit_field calls.
21917 * calls.c (store_unaligned_arguments_into_psuedos)
21918 load_register_parameters): Pass extra NULL to extract_bit_field calls.
21919 * optabs.c (maybe_legitimize_operand): Clear op->target when call
21921 * optabs.h (struct expand_operand): Add target bitfield.
21923 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
21925 * compare-elim.c (try_eliminate_compare): Canonicalize
21926 operation with embedded compare to
21927 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
21928 (set (reg) (operation)].
21930 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
21932 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
21935 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
21936 cost of adding a carry flag for ADC instruction.
21937 [case MINUS]: Ignore the cost of subtracting a carry flag
21938 for SBB instruction.
21940 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
21942 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
21944 * config/rs6000/bmiintrin.h: New file.
21945 * config/rs6000/bmi2intrin.h: New file.
21946 * config/rs6000/x86intrin.h: New file.
21948 2017-05-12 Jeff Law <law@redhat.com>
21950 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
21953 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
21955 PR middle-end/80707
21956 * tree-cfg.c: Remove cfg edges of unreachable case statements.
21958 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21960 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
21961 early expansion of vector divide builtins.
21962 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
21963 builtins identified as having unsigned arguments.
21965 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21967 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
21968 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
21969 expansion of vector logical operations (and, andc, or, xor,
21972 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21974 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
21975 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
21977 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21979 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
21980 early GIMPLE expansion of vector multiplies.
21982 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21984 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
21985 TARGET_HAVE_MOVT conditional.
21986 (movt splitter): Likewise.
21988 2017-05-12 Richard Biener <rguenther@suse.de>
21990 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
21991 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
21992 Fold all stmts not inplace.
21994 2017-05-12 Richard Biener <rguenther@suse.de>
21996 PR tree-optimization/80713
21997 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
21998 inserted_exprs bit for not removed stmts.
22000 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
22002 PR middle-end/69921
22003 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
22004 parallelized" attribute for parallelized OpenACC kernels.
22005 * omp-offload.c (execute_oacc_device_lower): Use it.
22007 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
22008 Set "oacc kernels" attribute.
22009 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
22010 parameter. Adjust all users.
22011 (oacc_fn_attrib_kernels_p): Remove function.
22012 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
22013 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
22014 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
22015 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
22016 assert "oacc kernels" attribute is set.
22018 2017-05-11 Carl Love <cel@us.ibm.com>
22020 * config/rs6000/rs6000-c: Add support for built-in functions
22021 vector unsigned char vec_popcnt (vector signed char)
22022 vector unsigned char vec_popcnt (vector unsigned char)
22023 vector unsigned short vec_popcnt (vector signed short)
22024 vector unsigned short vec_popcnt (vector unsigned short)
22025 vector unsigned int vec_popcnt (vector signed int)
22026 vector unsigned int vec_popcnt (vector unsigned int)
22027 vector unsigned long long vec_popcnt (vector signed long long)
22028 vector unsigned long long vec_popcnt (vector unsigned long long)
22029 vector signed long long vec_slo (vector signed long long,
22030 vector signed char)
22031 vector signed long long vec_slo (vector signed long long,
22032 vector unsigned char)
22033 vector unsigned long long vec_slo (vector unsigned long long,
22034 vector signed char)
22035 vector unsigned long long vec_slo (vector unsigned long long,
22036 vector unsigned char)
22037 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
22038 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
22039 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
22040 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
22041 * doc/extend.texi: Update the built-in documentation file for the
22042 new built-in functions.
22044 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
22046 * attribs.h (sorted_attr_string): Move machine independent
22047 functions for target clone support from the i386 port to common
22048 code. Rename ix86_function_versions to common_function_versions.
22049 Rename make_name to make_unique_name.
22050 (common_function_versions): Likewise.
22051 (make_unique_name): Likewise.
22052 (make_dispatcher_decl): Likewise.
22053 (is_function_default_version): Likewise.
22054 * attribs.c (attr_strcmp): Likewise.
22055 (sorted_attr_string): Likewise.
22056 (common_function_versions): Likewise.
22057 (make_unique_name): Likewise.
22058 (make_dispatcher_decl): Likewise.
22059 (is_function_default_version): Likewise.
22060 * config/i386/i386.c (attr_strcmp): Likewise.
22061 (sorted_attr_string): Likewise.
22062 (ix86_function_versions): Likewise.
22063 (make_name): Likewise.
22064 (make_dispatcher_decl): Likewise.
22065 (is_function_default_version): Likewise.
22066 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
22068 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22071 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
22072 Account for direct move costs for vec_construct of integer
22075 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
22078 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
22079 (UNSPEC_STX_ATOMIC): Ditto.
22080 (loaddi_via_sse): New insn.
22081 (storedi_via_sse): Ditto.
22082 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
22083 Update corresponding peephole2 patterns.
22084 (atomic_storedi_fpu): Ditto.
22086 2017-05-11 Julia Koval <julia.koval@intel.com>
22088 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
22089 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
22091 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
22092 (__builtin_ia32_rsqrt14ss_mask): New builtins.
22093 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
22095 2017-05-11 Nathan Sidwell <nathan@acm.org>
22097 * graphite-poly.c: Include dumpfile.h.
22099 * dumpfle.h (dump_function): Declare here ...
22100 * tree-dump.h (dump_function): ... not here.
22101 * dumpfile.c: #include tree-cfg.h.
22102 (dump_function): Move here from ...
22103 * tree-dump.c (dump_function): ... here.
22104 * gimplify.c: #include splay-tree.h, not tree-dump.h.
22105 * graphite-poly.c: Don't include tree-dump.h.
22106 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
22107 * print-tree.c: Likewise.
22108 * stor-layout.c: Likewise.
22109 * tree-nested.c: Likewise.
22111 * dumpfile.c (dump_start): Use TDF_FLAGS.
22112 (dump_enable_all): Fix TDF_KIND check thinko.
22114 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
22116 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22117 array entries to represent two legal parameterizations of the
22118 overloaded __builtin_cmpb function, as represented by the
22119 P6_OV_BUILTIN_CMPB constant.
22120 (altivec_resolve_overloaded_builtin): Add special case handling
22121 for the __builtin_cmpb function, as represented by the
22122 P6_OV_BUILTIN_CMPB constant.
22123 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
22124 (BU_P6_64BIT_2): New macro.
22125 (BU_P6_OVERLOAD_2): New macro
22126 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
22127 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
22128 (CMPB): Add overload support to represent both 32-bit and 64-bit
22129 compare-bytes function.
22130 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
22131 support for TARGET_CMPB.
22132 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
22133 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
22134 documentation of the __builtin_cmpb overloaded built-in function.
22136 2017-05-11 Richard Biener <rguenther@suse.de>
22138 PR tree-optimization/80705
22139 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
22140 bases are not vectorizable.
22142 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22144 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
22145 when counting register pressure.
22147 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22149 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
22150 (iv_ca_more_deps): Renamed to ...
22151 (iv_ca_compare_deps): ... this.
22152 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
22154 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22156 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
22158 (determine_group_iv_costs): ... here.
22159 (find_inv_vars_cb): Record inv var if it's not recorded before.
22161 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22163 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
22164 (get_shiftadd_cost): Ditto.
22166 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22168 * tree-ssa-address.c: Include header file.
22169 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
22171 (add_to_parts): Refactor.
22172 (addr_to_parts): New parameter. Update use of move_hint_to_base.
22173 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
22176 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22178 PR tree-optimization/53090
22179 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
22181 (extract_cond_operands): Detect condition with IV on both sides
22182 and return COMP_IV_EXPR_2.
22183 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
22184 (rewrite_use_compare): Simplify by removing call to function
22185 extract_cond_operands.
22187 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22189 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
22190 (extract_cond_operands): Detect condition comparing against non-
22191 invariant bound and return appropriate enum value.
22192 (find_interesting_uses_cond): Update use of extract_cond_operands.
22193 Handle its return value accordingly.
22194 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
22196 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22198 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
22199 nonlinear iv_use computation in loop invariant sensitive way.
22201 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22203 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
22204 (find_iv_candidates): Call relate_compare_use_with_all_cands.
22206 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22208 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
22209 (dump_cand): Support iv_cand.inv_exprs.
22210 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
22212 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
22215 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22217 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
22219 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
22220 as local function. Include necessary header files.
22221 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
22223 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22225 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
22227 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22229 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
22230 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
22231 RSHIFT_EXPR and BIT_NOT_EXPR.
22233 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22235 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
22236 (adjust_setup_cost): New parameter supporting round up adjustment.
22237 (struct address_cost_data): Delete.
22238 (force_expr_to_var_cost): Don't bound cost with spill_cost.
22239 (split_address_cost, ptr_difference_cost): Delete.
22240 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
22241 (struct ainc_cost_data): New struct.
22242 (get_address_cost_ainc): New function.
22243 (get_address_cost, get_computation_cost): Reimplement.
22244 (determine_group_iv_cost_address): Record inv_expr for all uses of
22246 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
22247 (iv_ca_has_deps): Reimplemented to ...
22248 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
22250 (iv_ca_extend): Call iv_ca_more_deps.
22252 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22254 * tree-ssa-address.c (struct mem_address): Move to header file.
22255 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
22256 * tree-ssa-address.h (struct mem_address): Move from C file.
22257 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
22259 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22261 * tree-affine.h (aff_combination_type): New interface.
22262 (aff_combination_zero_p): Remove static.
22263 (aff_combination_const_p): New interface.
22264 (aff_combination_singleton_var_p): New interfaces.
22266 2017-05-11 Richard Biener <rguenther@suse.de>
22268 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
22269 Skip unreachable blocks and destinations.
22270 (eliminate): Move stmt removal and fixup ...
22271 (fini_eliminate): ... here. Skip inserted exprs.
22272 (pass_pre::execute): Move fini_pre after fini_eliminate.
22273 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
22274 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
22275 PRE to get rid of dead code that has invalid SSA form and
22276 split critical edges again.
22278 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22280 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
22282 2017-05-11 Richard Biener <rguenther@suse.de>
22284 * passes.c (execute_function_todo): Verify loops if they are
22285 said to be up-to-date.
22286 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
22287 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
22289 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
22292 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
22293 handle calling assemble_external ourself.
22296 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
22297 modes with zero size. Enhance comment.
22299 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22301 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
22302 built-ins for vec_xl and vec_xst with short and char pointer
22305 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
22307 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
22308 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
22309 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
22310 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
22311 (_mm_maskz_min_round_ss): New intrinsics.
22312 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
22313 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
22314 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
22315 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
22316 (__builtin_ia32_minss_mask_round): New builtins.
22317 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22318 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
22319 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
22321 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
22322 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
22324 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
22327 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
22329 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
22330 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
22331 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
22332 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
22333 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
22334 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
22335 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
22336 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22337 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
22338 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
22339 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
22340 (__builtin_ia32_mulss_mask_round): New builtins.
22341 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22342 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
22343 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
22345 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
22346 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
22348 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
22351 2017-05-10 Julia Koval <julia.koval@intel.com>
22353 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
22354 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
22355 (_mm256_setr_m128i): New intrinsics.
22357 2017-05-10 Julia Koval <julia.koval@intel.com>
22359 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
22360 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
22361 (_mm_maskz_rcp14_ss): New intrinsics.
22362 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
22363 (__builtin_ia32_rcp14ss_mask): New builtins.
22364 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
22366 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
22368 PR tree-optimization/51513
22369 * tree-cfg.c (gimple_seq_unreachable_p): New function.
22370 (assert_unreachable_fallthru_edge_p): Use it.
22371 (group_case_labels_stmt): Likewise.
22372 * tree-cfg.h: Prototype it.
22373 * stmt.c: Include cfghooks.h and tree-cfg.h.
22374 (emit_case_dispatch_table) <gap_label>: New local variable.
22375 Use it to fill dispatch table gaps.
22376 Test for default_label before updating probabilities.
22377 (expand_case) <default_label>: Remove unneeded initialization.
22378 Test for unreachable default case statement and remove its edge.
22379 Set default_label accordingly.
22380 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
22382 2017-05-10 Carl Love <cel@us.ibm.com>
22384 * config/rs6000/rs6000-c: Add support for built-in functions
22385 vector signed char vec_neg (vector signed char)
22386 vector signed short int vec_neg (vector short int)
22387 vector signed int vec_neg (vector signed int)
22388 vector signed long long vec_neg (vector signed long long)
22389 vector float vec_neg (vector float)
22390 vector double vec_neg (vector double)
22391 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
22393 * config/rs6000/altivec.h: Add define for vec_neg
22394 * doc/extend.texi: Update the built-in documentation for the
22395 new built-in functions.
22397 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22399 PR tree-optimization/77644
22400 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
22402 2017-05-10 Nathan Sidwell <nathan@acm.org>
22404 * dumpfile.h (TDI_lang_all): New.
22405 (TDF_KIND): New. Renumber others
22406 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
22408 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
22410 (get_dump_file_name): Adjust suffix generation.
22411 (dump_enable_all): Use TDF_KIND.
22412 * doc/invoke.texi (-fdump-lang-all): Document.
22414 * dumpfile.h: Tabify.
22416 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
22419 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
22420 Move member access before delete.
22422 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
22424 * tree-inline.c (expand_call_inline): Split block at stmt
22427 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
22430 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
22431 are now unused after splitting mov{sf,sd}_hardfloat.
22432 (f32_lr2): Likewise.
22433 (f32_lm): Likewise.
22434 (f32_lm2): Likewise.
22435 (f32_li): Likewise.
22436 (f32_li2): Likewise.
22437 (f32_lv): Likewise.
22438 (f32_sr): Likewise.
22439 (f32_sr2): Likewise.
22440 (f32_sm): Likewise.
22441 (f32_sm2): Likewise.
22442 (f32_si): Likewise.
22443 (f32_si2): Likewise.
22444 (f32_sv): Likewise.
22445 (f32_dm): Likewise.
22446 (f32_vsx): Likewise.
22447 (f32_av): Likewise.
22448 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
22449 For movsf, order stores so the VSX stores occur before the GPR
22450 store which encourages the register allocator to use a traditional
22451 FPR instead of a GPR. For movsd, order the stores so that the GPR
22452 store comes before the VSX stores to allow the power6 to work.
22453 This is due to the power6 not having a 32-bit integer store
22454 instruction from a FPR.
22455 (movsf_hardfloat): Likewise.
22456 (movsd_hardfloat): Likewise.
22458 2017-05-09 Martin Sebor <msebor@redhat.com>
22460 PR translation/80280
22461 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
22464 PR translation/80280
22465 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
22466 data member added in r247778.
22467 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
22469 2017-05-09 Nathan Sidwell <nathan@acm.org>
22471 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
22473 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
22476 2017-05-09 Marek Polacek <polacek@redhat.com>
22478 * doc/invoke.texi: Fix typo.
22480 2017-05-09 Richard Biener <rguenther@suse.de>
22482 * tree-vrp.c (vrp_val_is_max): Adjust comment.
22483 (vrp_val_is_min): Likewise.
22484 (set_value_range_to_value): Likewise.
22485 (set_value_range_to_nonnegative): Likewise.
22486 (gimple_assign_nonzero_p): Likewise.
22487 (gimple_stmt_nonzero_p): Likewise.
22488 (vrp_int_const_binop): Likewise. Remove unreachable case.
22489 (adjust_range_with_scev): Adjust comments.
22490 (compare_range_with_value): Likewise.
22491 (extract_range_from_phi_node): Likewise.
22492 (test_for_singularity): Likewise.
22494 2017-05-09 Richard Biener <rguenther@suse.de>
22496 * tree-vrp.c (get_single_symbol): Add assert that we don't
22497 get overflowed constants as invariant part.
22498 (compare_values_warnv): Add comment before the TREE_NO_WARNING
22499 checks. Use wi::cmp instead of recursing for integer constants.
22500 (compare_values): Just ignore whether we assumed undefined
22501 overflow instead of failing the compare.
22502 (extract_range_for_var_from_comparison_expr): Add comment before the
22503 TREE_NO_WARNING sets.
22504 (test_for_singularity): Likewise.
22505 (extract_range_from_comparison): Do not disable optimization
22506 when we assumed undefined overflow.
22507 (extract_range_basic): Remove init of unused var.
22509 2017-05-09 Richard Biener <rguenther@suse.de>
22511 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
22512 (extract_range_from_multiplicative_op_1): Adjust.
22513 (extract_range_from_binary_expr_1): Use int_const_binop.
22515 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
22518 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
22519 rs6000_store_data_bypass_p in seven define_bypass directives and
22520 in several comments.
22521 * config/rs6000/rs6000-protos.h: Add prototype for
22522 rs6000_store_data_bypass_p function.
22523 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
22524 function implements slightly different (rs6000-specific) semantics
22525 than store_data_bypass_p, returning false rather than aborting
22526 with assertion error when arguments do not satisfy the
22527 requirements of store data bypass.
22528 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
22529 rs6000_store_data_bypass_p.
22531 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
22533 * config/xtensa/xtensa-protos.h
22534 (xtensa_initial_elimination_offset): New declaration.
22535 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
22536 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
22537 macro definition, add case for FRAME_POINTER_REGNUM when
22538 FRAME_GROWS_DOWNWARD.
22539 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
22540 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
22541 xtensa_initial_elimination_offset.
22543 2017-05-08 Nathan Sidwell <nathan@acm.org>
22545 * doc/invoke.texi: Alphabetize -fdump options.
22547 2017-05-08 Martin Sebor <msebor@redhat.com>
22549 PR translation/80280
22550 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
22552 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
22554 * target.def (compute_frame_layout): New optional target hook.
22555 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
22556 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
22557 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
22559 * reload1.c (verify_initial_elim_offsets): Likewise.
22560 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
22561 (use_simple_return_p): Call arm_compute_frame_layout if needed.
22562 (arm_get_frame_offsets): Split up into this ...
22563 (arm_compute_frame_layout): ... and this function.
22565 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
22567 * config/aarch64/constraints.md (Usa): New constraint.
22568 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
22570 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
22572 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
22573 with_multilib_list after it has been checked.
22575 2017-05-08 Richard Biener <rguenther@suse.de>
22577 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
22578 (bitmap_set_subtract_values): Likewise.
22580 2017-05-08 Richard Biener <rguenther@suse.de>
22582 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
22583 (gimple_assign_nonzero): ... this and remove strict_overflow_p
22585 (gimple_stmt_nonzero_warnv_p): Rename to ...
22586 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
22588 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
22589 (extract_range_basic): Adjust, do not disable propagation on
22590 strict overflow sensitive simplification.
22591 (vrp_visit_cond_stmt): Likewise.
22593 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
22595 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
22596 body size unconditionally.
22598 2017-05-07 Jeff Law <law@redhat.com>
22601 2017-05-06 Jeff Law <law@redhat.com>
22602 PR tree-optimization/78496
22603 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
22606 PR tree-optimization/78496
22607 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
22608 (simplify_stmt_using_ranges): Call it.
22609 (vrp_dom_walker::before_dom_children): Extract equivalences
22610 from an ASSERT_EXPR with an equality comparison against a
22613 2017-05-06 Jeff Law <law@redhat.com>
22615 PR tree-optimization/78496
22616 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
22619 PR tree-optimization/78496
22620 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
22621 (simplify_stmt_using_ranges): Call it.
22622 (vrp_dom_walker::before_dom_children): Extract equivalences
22623 from an ASSERT_EXPR with an equality comparison against a
22626 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
22628 * lra-constraints.c (lra_copy_reg_equiv): New function.
22629 (split_reg): Use it to copy equivalence information from the
22630 original register to the spill register.
22632 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
22634 PR rtl-optimization/75964
22635 * simplify-rtx.c (simplify_const_relational_operation): Remove
22636 invalid handling of comparisons of integer ABS.
22638 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
22640 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
22641 initialize to zero.
22642 (init_regs): Remove declaration.
22643 (function_arg_advance_32): Initialize error_p as boolean variable.
22645 2017-05-05 Nathan Sidwell <nathan@acm.org>
22647 * store-motion.c (remove_reachable_equiv_notes): Reformat long
22648 lines. Use for (;;).
22650 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22652 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
22653 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
22654 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
22655 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
22656 VF=2 that require versioning.
22658 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22660 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
22663 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22665 * diagnostic.h (diagnostic_override_option_index): Convert from
22666 macro to inline function.
22668 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22670 * diagnostic.c (last_module_changed_p): New function.
22671 (set_last_module): New function.
22672 (diagnostic_report_current_module): Convert macro usage to
22673 the above functions.
22674 * diagnostic.h (diagnostic_context::last_module): Strengthen
22675 from const line_map * to const line_map_ordinary *.
22676 (diagnostic_last_module_changed): Delete macro.
22677 (diagnostic_set_last_module): Delete macro.
22679 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22681 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
22682 with diagnostic_report_diagnostic.
22683 (diagnostic_n_impl_richloc): Likewise.
22684 * diagnostic.h (report_diagnostic): Delete macro.
22685 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
22686 with diagnostic_report_diagnostic.
22687 * substring-locations.c (format_warning_va): Likewise.
22689 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22691 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
22692 save/restor of format_spec. Move option-printing code to...
22693 (print_option_information): ...this new function, and
22694 reimplement by simply printing to the pretty_printer,
22695 rather than appending to the format string.
22697 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22699 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
22700 handling logic into...
22701 (update_effective_level_from_pragmas): ...this new function.
22703 2017-05-04 Andrew Waterman <andrew@sifive.com>
22705 * config/riscv/riscv.opt (mstrict-align): New option.
22706 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
22707 (SLOW_UNALIGNED_ACCESS): Define.
22708 (riscv_slow_unaligned_access): Declare.
22709 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
22711 (riscv_slow_unaligned_access): New variable.
22712 (rocket_tune_info): Set slow_unaligned_access to true.
22713 (optimize_size_tune_info): Set slow_unaligned_access to false.
22714 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
22715 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
22716 (riscv_option_override): Set riscv_slow_unaligned_access.
22717 * doc/invoke.texi: Add -mstrict-align to RISC-V.
22719 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
22721 * config/riscv/riscv.md: Unify indentation.
22723 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22728 * config/rs6000/rs6000.md (u code attribute): Add FIX and
22730 (extendsi<mode>2): Add support for doing sign extension via
22731 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
22732 don't have ISA 3.0 instructions.
22733 (extendsi<mode>2 splitter): Likewise.
22734 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
22735 generate the normal insns since SImode can now go in vector
22736 registers. Disallow the special UNSPECs needed for previous
22737 machines to hide SImode being used. Add new insns
22738 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
22739 (fix_trunc<mode>si2_stfiwx): Likewise.
22740 (fix_trunc<mode>si2_internal): Likewise.
22741 (fixuns_trunc<mode>si2): Likewise.
22742 (fixuns_trunc<mode>si2_stfiwx): Likewise.
22743 (fctiw<u>z_<mode>_smallint): Likewise.
22744 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
22745 of floating point to 32-bit integer from doing a direct move to
22746 the GPR registers to do a store.
22747 (fctiwz_<mode>): Break long line.
22749 2017-05-05 Bin Cheng <bin.cheng@arm.com>
22751 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
22752 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
22753 (addr_list, addr_offset_valid_p): New.
22754 (split_address_groups): Check offset validity with above function.
22755 (gt-tree-ssa-loop-ivopts.h): Include header file.
22757 2017-05-05 Nathan Sidwell <nathan@acm.org>
22759 * config.gcc (arm*-*-*): Add missing 'fi'.
22761 2017-05-05 Steve Ellcey <sellcey@cavium.com>
22763 * doc/invoke.texi (-fopt-info): Explicitly say order of options
22764 included in -fopt-info does not matter.
22765 * doc/optinfo.texi (-fopt-info): Fix description of default
22766 behavour. Explicitly say order of options included in -fopt-info
22769 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22771 * config.gcc: Allow combinations of aprofile and rmprofile values for
22772 --with-multilib-list.
22773 * config/arm/t-multilib: New file.
22774 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
22775 variables. Remove setting of ISA and floating-point ABI in
22776 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
22777 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
22778 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
22779 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
22781 * config/arm/t-rmprofile: Likewise except for the matches changes.
22782 * doc/install.texi (--with-multilib-list): Document the combination of
22783 aprofile and rmprofile values and warn about pitfalls in doing that.
22785 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
22787 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
22788 (movdi_aarch64): Likewise.
22790 2017-05-05 Jakub Jelinek <jakub@redhat.com>
22792 PR tree-optimization/80632
22793 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
22795 (build_arrays): Initialize it for virtual phis.
22796 (fix_phi_nodes): Use it for virtual phis.
22798 PR tree-optimization/80558
22799 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
22800 [x, y] op z into [x op, y op z] for op & or | if conditions
22803 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
22804 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22807 * config/arm/arm.md (use_literal_pool): Remove.
22808 (64-bit immediate split): No longer takes cost into consideration
22809 if arm_disable_literal_pool is enabled.
22810 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
22811 used when arm_disable_literal_pool is enabled.
22812 (arm_max_const_double_inline_cost): Remove use of
22813 arm_disable_literal_pool.
22814 (push_minipool_fix): Add assert.
22815 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
22816 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
22817 (no_literal_pool_sf_immediate): New.
22819 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22821 PR tree-optimization/80613
22822 * tree-ssa-dce.c (propagate_necessity): Remove cases for
22823 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
22825 2017-05-05 Richard Biener <rguenther@suse.de>
22827 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
22829 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
22831 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
22832 of this flag from insn conditions due to removal from r247495.
22834 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
22836 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
22838 (arm_early_store_addr_dep_ptr): Likewise.
22839 * config/arm/aarch-common-protos.h
22840 (arm_early_load_addr_dep_ptr): Add prototype.
22841 (arm_early_store_addr_dep_ptr): Likewise.
22842 * config/arm/cortex-a53.md: Add new bypasses.
22844 2017-05-05 Jakub Jelinek <jakub@redhat.com>
22846 * tree.c (next_type_uid): Change type to unsigned.
22847 (type_hash_canon): Decrement back next_type_uid if
22848 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
22849 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
22852 2017-05-04 Martin Sebor <msebor@redhat.com>
22854 * builtins.c: Fix a trivial typo in a comment.
22856 PR middle-end/79234
22857 * builtins.c (check_sizes): Adjust to handle reading past the end.
22858 Avoid printing excessive upper bound of ranges. Use %E to print
22859 tree nodes instead of converting them to %wu.
22860 (expand_builtin_memchr): New function.
22861 (compute_dest_size): Rename...
22862 (compute_objsize): ...to this.
22863 (expand_builtin_memcpy): Adjust.
22864 (expand_builtin_mempcpy): Adjust.
22865 (expand_builtin_strcat): Adjust.
22866 (expand_builtin_strcpy): Adjust.
22867 (check_strncat_sizes): Adjust.
22868 (expand_builtin_strncat): Adjust.
22869 (expand_builtin_strncpy): Adjust and simplify.
22870 (expand_builtin_memset): Adjust.
22871 (expand_builtin_bzero): Adjust.
22872 (expand_builtin_memcmp): Adjust.
22873 (expand_builtin): Handle memcmp.
22874 (maybe_emit_chk_warning): Check strncat just once.
22876 2017-05-04 Martin Sebor <msebor@redhat.com>
22878 PR preprocessor/79214
22879 PR middle-end/79222
22880 PR middle-end/79223
22881 * builtins.c (check_sizes): Add inlining context and issue
22882 warnings even when -Wno-system-headers is set.
22883 (check_strncat_sizes): Same.
22884 (expand_builtin_strncat): Same.
22885 (expand_builtin_memmove): New function.
22886 (expand_builtin_stpncpy): Same.
22887 (expand_builtin): Handle memmove and stpncpy.
22889 2017-05-04 Bin Cheng <bin.cheng@arm.com>
22891 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
22892 which is not used any more.
22894 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22896 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
22898 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22900 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
22901 (cortexa53_tunings): Likewise.
22902 (cortexa57_tunings): Likewise.
22903 (cortexa72_tunings): Likewise.
22904 (cortexa73_tunings): Likewise.
22906 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22908 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
22909 Set loop alignment to 8.
22911 2017-05-04 Martin Sebor <msebor@redhat.com>
22913 PR translation/80280
22914 * builtins.c (expand_builtin_object_size): Add missing quoting to
22915 %D and like directives.
22916 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
22917 (hsa_type_for_tree_type): Same.
22918 (verify_function_arguments): Same.
22919 * symtab.c (symbol_table::change_decl_assembler_name): Same.
22920 * varasm.c (get_section): Same.
22923 2017-05-04 Martin Sebor <msebor@redhat.com>
22925 PR translation/80280
22926 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
22928 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22930 * config/aarch64/aarch64.c (generic_addrcost_table):
22931 Change HI/TI mode setting.
22933 2017-05-04 Martin Jambor <mjambor@suse.cz>
22935 PR tree-optimization/80622
22936 * tree-sra.c (comes_initialized_p): New function.
22937 (build_accesses_from_assign): Only set write lazily when
22938 comes_initialized_p is false.
22939 (analyze_access_subtree): Use comes_initialized_p.
22940 (propagate_subaccesses_across_link): Assert !comes_initialized_p
22941 instead of testing for PARM_DECL.
22943 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22945 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
22946 constraint on operand 0 to allow more general addressing modes.
22947 Adjust output template.
22948 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
22950 * config/aarch64/aarch64-protos.h
22951 (aarch64_address_valid_for_prefetch_p): Declare prototype.
22952 * config/aarch64/constraints.md (Dp): New address constraint.
22953 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
22956 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
22958 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
22959 update use of estimate_ipcp_clone_size_and_time.
22960 (estimate_local_effects): Update use of
22961 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
22962 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
22963 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
22964 Return nonspecialized time.
22966 2017-05-04 Richard Biener <rguenther@suse.de>
22968 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
22969 for the last VUSE which def dominates the PHI. Directly call
22971 (get_continuation_for_phi_1): Remove.
22973 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
22975 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
22976 to explain the use of truncating division. Cap the number of
22977 iterations to the maximum given by nb_iterations_upper_bound,
22980 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22982 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
22983 * configure: Regenerate.
22984 * config.in: Regenerate.
22985 * config/i386/driver-mingw32.c: new file.
22986 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
22987 * config.host: Link driver-mingw32.o on MinGW host.
22988 * doc/install.texi: Document new --enable-mingw-wildcard configure
22991 2017-05-04 Marek Polacek <polacek@redhat.com>
22993 PR tree-optimization/80612
22994 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
22996 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22997 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
22999 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
23000 (movt splitter): Likewise.
23001 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
23002 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
23003 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
23004 block for Thumb-1 with MOVT.
23005 (thumb2_legitimate_address_p): Move code block ...
23006 (can_avoid_literal_pool_for_label_p): ... into this new function.
23007 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
23009 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
23010 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
23011 "M-profile targets with the MOVT instruction".
23013 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23015 * config/arm/arm-builtins.c (arm_init_builtins): Rename
23016 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
23017 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
23019 2017-05-04 Martin Liska <mliska@suse.cz>
23021 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
23022 variable cond_code.
23024 2017-05-04 Richard Biener <rguenther@suse.de>
23026 * tree.c (array_at_struct_end_p): Handle arrays at struct
23027 end with flexarrays more conservatively. Refactor and treat
23028 arrays of arrays or aggregates more strict. Fix
23029 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
23030 * tree.c (array_at_struct_end_p): Adjust prototype.
23031 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
23032 * gimple-fold.c (get_range_strlen): Likewise.
23033 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
23035 2017-05-04 Richard Biener <rguenther@suse.de>
23037 PR tree-optimization/31130
23038 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
23040 (supports_overflow_infinity): Likewise.
23041 (is_negative_overflow_infinity): Likewise.
23042 (is_positive_overflow_infinity): Likewise.
23043 (is_overflow_infinity): Likewise.
23044 (stmt_overflow_infinity): Likewise.
23045 (overflow_infinity_range_p): Likewise.
23046 (usable_range_p): Remove as always returning true.
23047 (make_overflow_infinity): Remove.
23048 (negative_overflow_infinity): Likewise.
23049 (positive_overflow_infinity): Likewise.
23050 (avoid_overflow_infinity): Likewise.
23051 (set_value_range): Adjust accordingly.
23052 (set_value_range_to_nonnegative): Likewise, remove now unused
23053 overflow_infinity arg.
23054 (vrp_operand_equal_p): Adjust.
23055 (update_value_range): Likewise.
23056 (range_int_cst_singleton_p): Likewise.
23057 (operand_less_p): Likewise.
23058 (compare_values_warnv): Likewise.
23059 (extract_range_for_var_from_comparison_expr): Likewise.
23060 (vrp_int_const_binop): Likewise.
23061 (zero_nonzero_bits_from_vr): Likewise.
23062 (extract_range_from_multiplicative_op_1): Likewise.
23063 (extract_range_from_binary_expr_1): Likewise.
23064 (extract_range_from_unary_expr): Likewise.
23065 (extract_range_from_comparison): Likewise.
23066 (extract_range_basic): Likewise.
23067 (adjust_range_with_scev): Likewise.
23068 (compare_ranges): Likewise.
23069 (compare_range_with_value): Likewise.
23070 (dump_value_range): Likewise.
23071 (test_for_singularity): Likewise, remove strict_overflow_p parameter
23073 (simplify_cond_using_ranges): Adjust.
23075 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
23077 * brig-builtins.def: Added a builtin for class_f64.
23078 * builtin-types.def: Added a builtin type needed by class_f64.
23080 2017-05-03 Jason Merrill <jason@redhat.com>
23082 * timevar.def: Add TV_CONSTEXPR.
23084 2017-05-03 David Malcolm <dmalcolm@redhat.com>
23086 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
23088 2017-05-03 Martin Jambor <mjambor@suse.cz>
23090 * ipa-prop.c (ipa_update_after_lto_read): Removed.
23091 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
23092 * ipa-cp.c (ipcp_propagate_stage): Do not call
23093 ipa_update_after_lto_read.
23094 * ipa-inline.c (ipa_inline): Likewise.
23096 2017-05-03 Martin Jambor <mjambor@suse.cz>
23098 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
23099 tag. Added a default constructor and a destructor.
23100 (ipa_edge_args_sum_t): New class;
23101 (ipa_edge_args_sum): Declare.
23102 (ipa_edge_args_vector): Remove declaration.
23103 (IPA_EDGE_REF): Use ipa_edge_args_sum.
23104 (ipa_free_edge_args_substructures): Remove declaration.
23105 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
23106 (ipa_edge_args_info_available_for_edge_p): Likewise.
23107 * ipa-prop.c (ipa_edge_args_vector): Removed.
23108 (edge_removal_hook_holder): Likewise.
23109 (edge_duplication_hook_holder): Likewise.
23110 (ipa_edge_args_sum): New variable.
23111 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
23112 ipa_edge_args_vector.
23113 (ipa_free_edge_args_substructures): Likewise.
23114 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
23115 ipa_edge_args_vector.
23116 (ipa_edge_removal_hook): Turned into method
23117 ipa_edge_args_sum_t::remove.
23118 (ipa_edge_duplication_hook): Turned into method
23119 ipa_edge_args_sum_t::duplicate.
23120 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
23121 registering edge hooks.
23122 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
23123 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
23124 ipa_edge_args_sum instead of ipa_edge_args_vector.
23125 * ipa-profile.c (ipa_profile): Likewise.
23127 2017-05-03 Martin Jambor <mjambor@suse.cz>
23129 * symbol-summary.h (function_summary): New method exists.
23130 (function_summary::symtab_removal): Deallocate through release.
23131 (call_summary): New class.
23132 (gt_ggc_mx): New overload.
23133 (gt_pch_nx): Likewise.
23134 (gt_pch_nx): Likewise.
23136 2017-05-03 Jeff Law <law@redhat.com>
23138 PR tree-optimization/78496
23139 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
23140 from simplify_cond_using_ranges. Split off code to walk
23141 backwards through casts into ...
23142 (simplify_cond_using_ranges_2): New function.
23143 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
23144 (execute_vrp): After identifying jump threads, call
23145 simplify_cond_using_ranges_2.
23147 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
23150 * ipa-inline.h (inline_summary): Add ctor.
23151 (create_ggc): Do not use ggc_cleared_alloc.
23153 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
23154 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23156 * gcc.c (handle_braces): Support escaping in switch matching
23158 * doc/invoke.texi (Spec Files): Document it.
23159 Remove superfluous @code markup in items.
23161 2017-05-03 David Malcolm <dmalcolm@redhat.com>
23163 * diagnostic-show-locus.c (struct column_range): New struct.
23164 (get_affected_columns): New function.
23165 (get_printed_columns): New function.
23166 (struct correction): New struct.
23167 (correction::ensure_capacity): New function.
23168 (correction::ensure_terminated): New function.
23169 (struct line_corrections): New struct.
23170 (line_corrections::~line_corrections): New dtor.
23171 (line_corrections::add_hint): New function.
23172 (layout::print_trailing_fixits): Reimplement in terms of the new
23174 (selftest::test_overlapped_fixit_printing): New function.
23175 (selftest::diagnostic_show_locus_c_tests): Call it.
23177 2017-05-03 Nathan Sidwell <nathan@acm.org>
23179 Canonicalize canonical type hashing
23180 * tree.h (type_hash_canon_hash): Declare.
23181 * tree.c (type_hash_list, attribute_hash_list): Move into
23182 type_hash_canon_hash.
23183 (build_type_attribute_qual_variant): Break out hash code calc into
23184 type_hash_canon_hash.
23185 (type_hash_canon_hash): New. Generic type hash computation.
23186 (build_range_type_1, build_array_type_1, build_function_type,
23187 build_method_type_directly, build_offset_type, build_complex_type,
23188 make_vector_type): Call it.
23190 2017-05-03 Richard Biener <rguenther@suse.de>
23192 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
23193 When all DRs have unknown misaligned do not always peel
23194 when there is a store but apply the same costing model as if
23195 there were only loads.
23197 2017-05-03 Richard Biener <rguenther@suse.de>
23200 PR tree-optimization/80492
23201 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
23202 compare_base_decls returning dont-know properly.
23204 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
23206 * config/arm/iterators.md (CCSI): New mode iterator.
23207 (arch): New mode attribute.
23208 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
23209 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
23210 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
23211 code iterator for success result mode.
23212 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
23213 the corresponding new insn generators.
23215 2017-05-03 Bin Cheng <bin.cheng@arm.com>
23218 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23219 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23221 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
23223 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
23224 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
23225 (DDR_A): Wrap DDR argument in brackets.
23226 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
23227 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
23228 (DDR_REVERSED_P): Likewise.
23230 2017-05-03 Jakub Jelinek <jakub@redhat.com>
23232 PR tree-optimization/79472
23233 * tree-switch-conversion.c (struct switch_conv_info): Add
23234 contiguous_range and default_case_nonstandard fields.
23235 (collect_switch_conv_info): Compute contiguous_range and
23236 default_case_nonstandard fields, don't clear final_bb if
23237 contiguous_range and only the default case doesn't have the required
23239 (check_all_empty_except_final): Set default_case_nonstandard instead
23240 of failing if contiguous_range and the default case doesn't have empty
23242 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
23243 and only the default case doesn't have the required constants. Skip
23245 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
23246 if default_case_nonstandard.
23247 (build_constructors): Build constant 1 just once. Assert that default
23248 values aren't inserted in between cases if contiguous_range. Skip
23250 (build_arrays): Skip virtual phis.
23251 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
23252 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
23253 Handle virtual phis.
23254 (gen_inbound_check): Handle default_case_nonstandard case.
23255 (process_switch): Adjust check_final_bb caller. Call
23256 gather_default_values with the first non-default case instead of
23257 default case if default_case_nonstandard.
23259 2017-05-02 Nathan Sidwell <nathan@acm.org>
23261 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
23262 check. Fix formatting.
23264 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
23266 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
23267 errors when comparing specialized and unspecialized times.
23269 2017-05-02 David Malcolm <dmalcolm@redhat.com>
23271 * diagnostic-show-locus.c
23272 (layout::should_print_annotation_line_p): Make private.
23273 (layout::print_annotation_line): Make private.
23274 (layout::annotation_line_showed_range_p): Make private.
23275 (layout::show_ruler): Make private.
23276 (layout::print_source_line): Make private. Pass in line and
23277 line_width, rather than calling location_get_source_line. Drop
23279 (layout::print_leading_fixits): New method.
23280 (layout::print_any_fixits): Rename to...
23281 (layout::print_trailing_fixits): ...this, and make private.
23282 Don't print newline fixits.
23283 (diagnostic_show_locus): Move logic for printing one row into...
23284 (layout::print_line): ...this new function. Move the
23285 location_get_source_line call and error-handling from
23286 print_source_line to here. Call print_leading_fixits, and rename
23287 print_any_fixits to print_trailing_fixits.
23288 (selftest::test_fixit_insert_containing_newline): Update now that
23289 newlines are partially supported.
23290 (selftest::test_fixit_insert_containing_newline_2): New test.
23291 (selftest::test_fixit_replace_containing_newline): Update comments.
23292 (selftest::diagnostic_show_locus_c_tests): Call the new test.
23293 * edit-context.c (class added_line): New class.
23294 (class edited_line): Describe newline handling in comment.
23295 (edited_line::actually_edited_p): New method.
23296 (edited_line::print_content): Delete redundant decl.
23297 (edited_line::m_predecessors): New field.
23298 (edited_file::print_content): Call edited_line::print_content.
23299 (edited_file::print_diff): Update to support newlines.
23300 (edited_file::print_diff_hunk): Likewise.
23301 (edited_file::print_run_of_changed_lines): New function.
23302 (edited_file::print_diff_line): Convert to...
23303 (print_diff_line): ...this.
23304 (edited_file::get_effective_line_count): New function.
23305 (edited_line::edited_line): Initialize new field m_predecessors.
23306 (edited_line::~edited_line): Clean up m_predecessors.
23307 (edited_line::apply_fixit): Handle newlines.
23308 (edited_line::get_effective_line_count): New function.
23309 (edited_line::print_content): New function.
23310 (edited_line::print_diff_lines): New function.
23311 (selftest::test_applying_fixits_insert_containing_newline): New
23313 (selftest::test_applying_fixits_replace_containing_newline): New
23315 (selftest::insert_line): New function.
23316 (selftest::test_applying_fixits_multiple_lines): Add example of
23318 (selftest::edit_context_c_tests): Call the new tests.
23320 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23322 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
23323 parameter cand. Update dump information.
23324 (get_computation_cost): Update uses.
23326 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23328 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
23329 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
23330 (get_computation_at, rewrite_use_address): Update use of
23331 get_computation_aff.
23333 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23335 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
23336 (get_computation): Delete.
23337 (get_computation_cost): Implement like get_computation_cost_at.
23338 Use get_computation_at.
23339 (get_computation_cost_at): Delete.
23340 (rewrite_use_nonlinear_expr): Use get_computation_at.
23341 (rewrite_use_compare, remove_unused_ivs): Ditto.
23343 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23345 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
23347 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23349 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
23350 (ivopts_global_cost_for_size): Rename parameter and update uses.
23351 (iv_ca_recount_cost): Update uses.
23352 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
23353 candidates seperately in n_invs and n_cands.
23354 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
23356 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23358 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
23359 (find_inv_vars_cb): New.
23360 (find_depends): Renamed to ...
23361 (find_inv_vars): ... this.
23362 (add_candidate_1, force_var_cost): Call find_inv_vars.
23363 (split_address_cost, determine_group_iv_cost_cond): Ditto.
23365 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23367 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
23368 inv_vars. Add inv_exprs.
23369 (struct iv_cand): Rename depends_on to inv_vars.
23370 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
23371 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
23372 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
23373 (dump_cand): Dump inv_vars.
23374 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
23375 (record_invariant, find_depends, add_candidate_1): Ditto.
23376 (set_group_iv_cost, force_var_cost): Ditto.
23377 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
23378 (get_computation_cost_at, get_computation_cost): Ditto.
23379 (determine_group_iv_cost_generic): Ditto.
23380 (determine_group_iv_cost_address): Ditto.
23381 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
23382 (determine_group_iv_costs): Ditto.
23383 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
23384 (iv_ca_set_remove_invariants): Renamed to ...
23385 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
23386 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
23387 (iv_ca_set_add_invariants): Renamed to ...
23388 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
23389 (iv_ca_set_cp): Use iv_ca_set_add_invs.
23390 (iv_ca_has_deps): Support inv_vars and inv_exprs.
23391 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
23392 (create_new_ivs): Remove useless dump.
23394 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23396 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
23398 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
23399 (iv_ca_set_no_cp, create_new_iv): Ditto.
23401 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23403 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23405 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
23407 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
23408 function tree_check2.
23410 2017-05-02 Martin Liska <mliska@suse.cz>
23412 * doc/gcov.texi: Add missing preposition.
23413 * gcov.c (function_info::function_info): Properly fill up
23414 all member variables.
23416 2017-05-02 Tamar Christina <tamar.christina@arm.com>
23418 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
23420 2017-05-02 Tamar Christina <tamar.christina@arm.com>
23422 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
23424 2017-05-02 Martin Liska <mliska@suse.cz>
23427 * lto-streamer-in.c (lto_read_tree_1): Remove
23428 LTO_STREAMER_DEBUG.
23429 * lto-streamer.c (struct tree_hash_entry): Likewise.
23430 (struct tree_entry_hasher): Likewise.
23431 (tree_entry_hasher::hash): Likewise.
23432 (tree_entry_hasher::equal): Likewise.
23433 (lto_streamer_init): Likewise.
23434 (lto_orig_address_map): Likewise.
23435 (lto_orig_address_get): Likewise.
23436 (lto_orig_address_remove): Likewise.
23437 * lto-streamer.h: Likewise.
23438 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
23439 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
23441 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
23443 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
23444 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
23445 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
23446 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
23447 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
23448 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
23449 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
23450 (mm_maskz_sub_ss): New intrinsics.
23451 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23452 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
23453 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
23454 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
23455 (__builtin_ia32_subss_mask_round): New builtins.
23456 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23457 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23458 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
23460 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
23461 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
23463 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
23466 2017-05-02 Martin Jambor <mjambor@suse.cz>
23468 PR tree-optimization/78687
23469 * tree-sra.c (access): New field parent.
23470 (process_subtree_disqualification): New function.
23471 (disqualify_candidate): Call it.
23472 (build_accesses_from_assign): Reset write flag if creating an
23474 (build_access_subtree): Fill in parent field and also prpagate
23475 down grp_write flag.
23476 (create_artificial_child_access): New parameter set_grp_write, set
23477 grp_write to its value.
23478 (propagate_subaccesses_across_link): Also propagate grp_write flag
23480 (propagate_all_subaccesses): Push the closest parent back to work
23481 queue if add_access_to_work_queue returned true.
23483 2017-05-02 Richard Biener <rguenther@suse.de>
23485 * common.opt (fstrict-overflow): Alias negative to fwrapv.
23486 * doc/invoke.texi (fstrict-overflow): Remove all traces of
23487 -fstrict-overflow documentation.
23488 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
23489 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
23490 flag_strict_overflow.
23491 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
23492 * lto-opts.c (lto_write_options): Do not stream it.
23493 * lto-wrapper.c (merge_and_complain): Do not handle it.
23494 * opts.c (default_options_table): Do not set -fstrict-overflow.
23495 (finish_options): Likewise do not clear it when sanitizing.
23496 * simplify-rtx.c (simplify_const_relational_operation): Do not
23497 test flag_strict_overflow.
23499 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
23501 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
23502 using enabled attribute.
23503 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
23504 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
23505 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
23506 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
23507 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
23508 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
23509 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
23510 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
23511 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
23512 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
23514 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
23516 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
23518 2017-05-02 Richard Biener <rguenther@suse.de>
23520 PR tree-optimization/80591
23522 2017-04-10 Richard Biener <rguenther@suse.de>
23524 * tree-ssa-structalias.c (find_func_aliases): Properly handle
23527 2017-05-02 Richard Biener <rguenther@suse.de>
23529 PR tree-optimization/80549
23530 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
23531 (cleanup_tree_cfg_noloop): Create forwarders to known loop
23532 headers if they do not have a preheader.
23534 2017-05-02 Martin Liska <mliska@suse.cz>
23537 * common.opt: Fix typo.
23538 * doc/invoke.texi: Likewise.
23540 2017-05-01 Jan Beulich <jbeulich@suse.com>
23542 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
23543 swapping, add (x,x,m,x,n) alternative.
23545 2017-05-01 Nathan Sidwell <nathan@acm.org>
23547 * calls.c (combine_pending_stack_adjustment_and_call): Remove
23548 unnecessary unadjusted_alignment check.
23550 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
23553 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
23555 * gimplify.c (gimplify_cilk_detach): New function.
23556 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
23557 * tree-core.h: Document EXPR_CILK_SPAWN.
23558 * tree.h (EXPR_CILK_SPAWN): Define.
23560 2017-05-01 David Malcolm <dmalcolm@redhat.com>
23562 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
23563 to use new fixit_hint representation, using the "replace" logic.
23564 (get_line_span_for_fixit_hint): Likewise.
23565 (layout::print_any_fixits): Likewise.
23566 (selftest::test_one_liner_many_fixits): Rename to...
23567 (selftest::test_one_liner_many_fixits_1): ...this, and update
23568 comment and expected output to reflect that the multiple fix-it
23569 hints are now consolidated into one insertion.
23570 (selftest::test_one_liner_many_fixits_2): New test.
23571 (selftest::test_diagnostic_show_locus_one_liner): Update for
23573 (selftest::test_fixit_consolidation): Update for fix-it API
23575 * diagnostic.c (print_parseable_fixits): Likewise.
23576 * edit-context.c (edited_line::m_line_events): Convert from
23577 auto_vec <line_event *> to auto_vec <line_event>.
23578 (class line_event): Convert from abstract base class to a concrete
23579 class, taking over the role of replace_event.
23580 (class insert_event): Delete.
23581 (class replace_event): Rename to class line_event. Convert to
23583 (edit_context::add_fixits): Reimplement.
23584 (edit_context::apply_insert): Delete.
23585 (edit_context::apply_replace): Rename to...
23586 (edit_context::apply_fixit): ...this. Convert to half-open range.
23587 (edited_file::apply_insert): Delete.
23588 (edited_file::apply_replace): Rename to...
23589 (edited_file::apply_fixit): ...this.
23590 (edited_line::~edited_line): Drop deletion of events.
23591 (edited_line::apply_insert): Delete.
23592 (edited_line::apply_replace): Rename to...
23593 (edited_line::apply_fixit): ...this. Convert to half-open range.
23594 Update for change to type of m_line_events.
23595 * edit-context.h (edit_context::apply_insert): Delete.
23596 (edit_context::apply_replace): Rename to...
23597 (edit_context::apply_fixit): ...this.
23599 2017-05-01 Martin Sebor <msebor@redhat.com>
23601 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
23604 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
23607 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
23608 __get_cpuid_max returns 0.
23609 (__get_cpuid_count): Ditto.
23611 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
23613 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
23614 replacement expression is another instance of one of its arguments.
23616 2017-05-01 Jakub Jelinek <jakub@redhat.com>
23619 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
23620 check for stack push/pop autoinc.
23621 * config/i386/i386.c (ix86_agi_dependent): Return false
23622 if the only reason why modified_in_p returned true is that
23623 addr is SP based and set_insn is a push or pop.
23625 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
23627 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
23630 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
23633 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
23634 (account_size_time): Use two predicates - exec_pred and
23636 (evaluate_conditions_for_known_args): Compute both clause and
23638 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
23639 (inline_summary_t::duplicate): Update.
23640 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
23642 (compute_inline_parameters): Likewise.
23643 (estimate_edge_size_and_time): Update caluclation of time.
23644 (estimate_node_size_and_time): Compute both time and nonspecialized
23646 (estimate_ipcp_clone_size_and_time): Update.
23647 (inline_merge_summary): Update.
23648 (do_estimate_edge_time): Update.
23649 (do_estimate_edge_size): Update.
23650 (do_estimate_edge_hints): Update.
23651 (inline_read_section, inline_write_summary): Stream both new predicates.
23652 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
23654 (compute_inlined_call_time): Cleanup.
23655 (big_speedup_p): Update.
23656 (edge_badness): Update.
23657 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
23658 (size_time_entry): Replace predicate by exec_predicate and
23659 nonconst_predicate.
23660 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
23661 (estimate_edge_time): Return also nonspec_time.
23662 (reset_edge_growth_cache): Update.
23664 2017-04-29 Jakub Jelinek <jakub@redhat.com>
23666 PR rtl-optimization/80491
23667 * ifcvt.c (noce_process_if_block): When looking for x setter
23668 with missing else_bb, don't check only the insn right before
23669 cond_earliest, but look for the last insn that x is modified in
23670 within the same bb.
23672 PR rtl-optimization/80491
23673 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
23675 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
23677 PR tree-optimization/80487
23678 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
23680 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23682 PR tree-optimization/79697
23683 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
23684 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
23685 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
23687 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
23688 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
23690 2017-04-28 Martin Sebor <msebor@redhat.com>
23692 PR tree-optimization/80523
23693 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
23694 (init_target_to_host_charmap, target_to_host, target_strtol10): New
23696 (maybe_warn, format_directive, parse_directive): Use new functions.
23697 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
23699 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
23701 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
23703 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
23705 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
23706 target_header_dir): Set correctly.
23707 * configure: Regenerated.
23708 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
23709 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
23710 instead of SYSTEM_HEADER_DIR.
23712 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
23714 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
23715 (estimate_local_effects): Likewise.
23716 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
23717 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
23718 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
23719 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
23720 do_estimate_edge_time, estimate_edge_time): Likewise.
23721 * ipa-inline-analysis.c (estimate_node_size_and_time,
23722 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
23723 (estimate_time_after_inlining): Remove.
23725 2017-04-28 Martin Liska <mliska@suse.cz>
23727 * doc/gcov.texi: Enhance documentation of gcov.
23729 2017-04-28 Martin Liska <mliska@suse.cz>
23731 * doc/gcov.texi: Sort options in alphabetic order.
23732 * doc/gcov-dump.texi: Likewise.
23733 * doc/gcov-tool.texi: Likewise.
23734 * gcov.c (print_usage): Likewise.
23735 * gcov-dump.c (print_usage): Likewise.
23736 * gcov-tool.c (print_merge_usage_message): Likewise.
23737 (print_rewrite_usage_message): Likewise.
23738 (print_overlap_usage_message): Likewise.
23740 2017-04-28 Martin Liska <mliska@suse.cz>
23742 PR gcov-profile/53915
23743 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
23745 2017-04-28 Martin Liska <mliska@suse.cz>
23747 PR gcov-profile/79891
23748 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
23749 is marked by compiler as living on a line.
23750 (get_cycles_count): Remove usage of the union.
23751 (output_intermediate_file): Likewise.
23752 (find_source): Fix GNU coding style.
23753 (accumulate_line_counts): Remove old non-all block mode.
23754 (output_lines): Remove usage of the union.
23755 * profile.c (output_location): Include all BBs, even if
23756 belonging to a same line (and file) as a previous BB.
23758 2017-04-28 Martin Liska <mliska@suse.cz>
23760 * gcov.c (process_args): Handle new argument 'w'.
23761 (read_graph_file): Assign ID to BBs.
23762 (output_branch_count): Display BB # if verbose flag is set.
23763 (output_lines): Likewise for arcs.
23764 (print_usage): Add '--verbose' option help.
23765 * doc/gcov.texi: Document --verbose (-w) option.
23767 2017-04-28 Martin Liska <mliska@suse.cz>
23769 * gcov.c (struct block_location_info): New struct.
23770 (process_file): Fill up the new structure.
23771 (read_graph_file): Replace usage of encoding by the newly added
23773 (add_line_counts): Likewise.
23774 (accumulate_line_counts): Remove usage of the union.
23775 (function_info::function_info): New function.
23776 (function_info::~function_info): Likewise.
23777 (process_file): Call delete instead of release_function.
23778 (release_function): Release the function.
23779 (release_structures): Call delete instead of release_function.
23780 (solve_flow_graph): Replace usage of num_blocks.
23781 (find_exception_blocks): Likewise.
23782 (output_lines): Fix GNU coding style.
23784 2017-04-28 Martin Liska <mliska@suse.cz>
23787 * coverage.c (coverage_remove_note_file): New function.
23788 * coverage.h: Declare the function.
23789 * toplev.c (finalize): Clean if an error has been seen.
23791 2017-04-28 Martin Liska <mliska@suse.cz>
23793 PR gcov-profile/80031
23794 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
23795 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
23796 * gcov.c (read_graph_file): Read just number of blocks.
23797 * profile.c (branch_prob): Do not stream 0 flags per a basic
23800 2017-04-28 Martin Liska <mliska@suse.cz>
23802 * gcov-dump.c (tag_*): Add new argument to declarations.
23803 (dump_gcov_file): Likewise.
23804 (tag_blocks): Add and use new argument depth.
23805 (tag_arcs): Likewise.
23806 (tag_lines): Likewise.
23807 (tag_counters): Likewise.
23808 (tag_summary): Likewise.
23809 (dump_working_sets): Use depth to do a proper indentation.
23811 2017-04-28 Jakub Jelinek <jakub@redhat.com>
23814 * cgraph.h (symtab_node::debug_symtab): No longer inline.
23815 * symtab.c (symtab_node::debug_symtab): Move definition here.
23817 2017-04-28 Richard Biener <rguenther@suse.de>
23819 * lto-streamer.h (LTO_major_version): Bump to 7.
23821 2017-04-28 Richard Biener <rguenther@suse.de>
23823 * tree-vrp.c (assert_info): New struct.
23824 (add_assert_info): New helper.
23825 (register_edge_assert_for_2): Refactor to add asserts to a vector
23827 (register_edge_assert_for_1): Likewise.
23828 (register_edge_assert_for): Likewise.
23829 (finish_register_edge_assert_for): New helper actually registering
23830 asserts where live on edge.
23831 (find_conditional_asserts): Adjust.
23832 (find_switch_asserts): Likewise.
23833 (evrp_dom_walker::try_find_new_range): Generalize.
23834 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
23836 2017-04-27 Marek Polacek <polacek@redhat.com>
23839 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
23840 arg10 and arg11 to itype.
23842 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
23844 * doc/extend.texi (Object Size Checking): Improve grammar.
23846 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
23849 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
23850 that the logic for permitting reciprocal estimates matches that
23853 2017-04-27 Jakub Jelinek <jakub@redhat.com>
23856 * tree.c (type_cache_hasher::equal): Only compare
23857 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
23858 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
23859 non-aggregate element types.
23860 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
23861 about the flag on ARRAY_TYPEs in the comment, formatting fix.
23863 2017-04-27 Richard Biener <rguenther@suse.de>
23865 PR middle-end/80533
23866 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
23867 stripping ARRAY_REFs from MEM_EXPR make sure we're not
23868 keeping a reference to a trailing array.
23870 2017-04-27 Richard Biener <rguenther@suse.de>
23872 PR middle-end/80539
23873 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
23874 being in loop-closed SSA form conservatively.
23875 (chrec_fold_multiply_poly_poly): Likewise.
23877 2017-04-27 Tamar Christina <tamar.christina@arm.com>
23879 PR middle-end/79665
23880 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
23881 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
23883 2017-04-27 Jakub Jelinek <jakub@redhat.com>
23886 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
23887 (aarch64_function_arg_alignment): Return unsigned int again, but still
23888 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
23889 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
23890 Don't emit -Wpsabi note.
23891 (aarch64_function_arg_boundary): Likewise.
23892 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
23895 2017-04-26 Nathan Sidwell <nathan@acm.org>
23897 * tree.h (crc32_unsigned_n): Declare.
23898 (crc32_unsigned, crc32_unsigned): Make inline.
23899 * tree.c (crc32_unsigned_bits): Replace with ...
23900 (crc32_unsigned_n): ... this.
23901 (crc32_unsigned, crc32_byte): Remove.
23902 (crc32_string): Remove unnecessary braces.
23904 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
23906 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
23907 * ipa-inline-analysis.c (MAX_TIME): Remove.
23908 (account_size_time): Use sreal for time.
23909 (dump_inline_summary): Update.
23910 (estimate_function_body_sizes): Update.
23911 (estimate_edge_size_and_time): Update.
23912 (estimate_calls_size_and_time): Update.
23913 (estimate_node_size_and_time): Update.
23914 (inline_merge_summary): Update.
23915 (inline_update_overall_summary): Update.
23916 (estimate_time_after_inlining): Update.
23917 (inline_read_section): Update.
23918 (inline_write_summary): Update.
23919 * ipa-inline.c (compute_uninlined_call_time): Update.
23920 (compute_inlined_call_time): Update.
23921 (recursive_inlining): Update.
23922 (inline_small_functions): Update.
23923 (dump_overall_stats): Update.
23924 * ipa-inline.h: Include sreal.h.
23925 (size_time_entry): Turn time to sreal.
23926 (inline_summary): Turn self_time nad time to sreal.
23928 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
23930 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
23932 (sreal::stream_out, sreal::stream_in): New.
23933 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
23935 2017-04-25 Jakub Jelinek <jakub@redhat.com>
23937 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
23940 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
23943 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
23945 (dimode_scalar_chain::compute_convert_gain): Ditto.
23946 (dimode_scalar_chain::make_vector_copies): Ditto.
23947 (dimode_scalar_chain::convert_reg): Ditto.
23948 (dimode_scalar_chain::convert_insn): Ditto.
23949 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
23950 (VI248_AVX512BW_1): New mode iterator.
23951 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
23952 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
23955 2017-04-25 Martin Sebor <msebor@redhat.com>
23957 PR tree-optimization/80497
23958 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
23959 constants are representable in HOST_WIDE_INT.
23960 (parse_directive): Ditto.
23962 2017-04-25 Martin Sebor <msebor@redhat.com>
23965 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
23966 (new_zero_array): Adjust signature.
23967 (dom_info::dom_init): Used unsigned rather that size_t.
23968 (dom_info::dom_info): Same.
23970 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23971 Jakub Jelinek <jakub@redhat.com>
23974 * config/arm/arm.c: Include gimple.h.
23975 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
23976 returns negative, increment ncrn only if it returned positive.
23977 (arm_needs_doubleword_align): Return int instead of bool,
23978 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
23979 members, but if there is any such non-FIELD_DECL
23980 > PARM_BOUNDARY aligned decl, return -1 instead of false.
23981 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
23982 returns negative, increment nregs only if it returned positive.
23983 (arm_setup_incoming_varargs): Likewise.
23984 (arm_function_arg_boundary): Emit -Wpsabi note if
23985 arm_needs_doubleword_align returns negative, return
23986 DOUBLEWORD_ALIGNMENT only if it returned positive.
23988 2017-04-25 Marek Polacek <polacek@redhat.com>
23991 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
23992 first argument to type.
23994 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
23997 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
23998 type checks to test for compatibility instead of equality.
24000 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24001 Jakub Jelinek <jakub@redhat.com>
24004 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
24006 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
24007 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
24008 the alignment computation, but return their maximum in warn_alignment.
24009 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
24010 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
24012 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
24013 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
24016 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24018 * config/arc/simdext.md (dmpyh): Fix typo.
24020 2017-04-25 Richard Biener <rguenther@suse.de>
24022 PR tree-optimization/80492
24023 * alias.c (compare_base_decls): Handle registers with asm
24024 specification conservatively.
24025 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24026 compare_base_decls returning dont-know properly.
24028 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24030 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
24031 (legitimate_offset_address_p): New function.
24032 (arc_legitimate_address_p): Use above function.
24034 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24036 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
24038 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24040 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
24041 ACCH registers whenever they are available.
24043 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24045 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
24046 double regs fix when not used.
24048 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24050 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
24052 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
24053 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
24055 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24057 * config/arc/arc.c (arc_output_addsi): Check for h-register class
24058 when emitting short ADD instructions.
24060 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24062 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
24064 (cmpsi_cc_c_insn): Likewise.
24065 (cbranchsi4_scratch): Compute proper instruction length using
24066 compact_hreg_operand.
24067 * config/arc/predicates.md (compact_hreg_operand): New predicate.
24069 2017-04-25 Richard Biener <rguenther@suse.de>
24071 PR middle-end/80509
24072 * passes.c (pass_manager::pass_manager): Initialize
24073 m_name_to_pass_map.
24075 2017-04-25 Richard Biener <rguenther@suse.de>
24077 PR tree-optimization/79201
24078 * tree-ssa-sink.c (statement_sink_location): Handle calls.
24080 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24083 * config/s390/vector.md: Split MEM->GPR vector moves for
24084 non-s_operand addresses.
24086 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24089 * config/s390/predicates.md (reload_const_wide_int_operand): New
24091 * config/s390/s390.md ("movti"): Remove d/P alternative.
24092 ("movti_bigconst"): New pattern definition.
24094 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
24097 * s390-protos.h (s390_expand_cs_hqi): Removed.
24098 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
24099 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
24100 modes as well as CCZ1mode and CCZmode.
24101 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
24102 signature of s390_emit_compare_and_swap.
24103 (s390_expand_cs_hqi): Likewise, make static.
24104 (s390_expand_cs_tdsi): Generate an explicit compare before trying
24105 compare-and-swap, in some cases.
24106 (s390_expand_cs): Wrapper function.
24107 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
24109 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
24110 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
24111 patterns for small and large integers. Forbid symref memory operands.
24112 Move expander to s390.c. Require cc register.
24113 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
24114 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
24115 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
24116 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
24117 symref memory operands. Remove CC mode and call s390_match_ccmode
24119 ("atomic_exchange<mode>"): Allow and implement all integer modes.
24121 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
24123 * config/s390/s390.md (define_peephole2): New peephole to help
24124 combining the load-and-test pattern with volatile memory.
24126 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
24128 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
24129 with CCZmode for TARGET_Z196.
24131 2017-04-25 Jakub Jelinek <jakub@redhat.com>
24133 PR rtl-optimization/80501
24134 * combine.c (make_compound_operation_int): Set subreg_code to SET
24135 even for AND with mask of the sign bit of mode.
24137 PR rtl-optimization/80500
24138 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
24139 sum's initial value.
24141 2017-04-25 Julian Brown <julian@codesourcery.com>
24142 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24144 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
24146 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
24148 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
24150 2017-04-25 Julian Brown <julian@codesourcery.com>
24151 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24153 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
24154 (thunderx2t99_sha): New Reservation.
24156 2017-04-25 Julian Brown <julian@codesourcery.com>
24157 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24159 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
24160 type for 1-element load.
24162 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
24164 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
24166 2017-04-24 Martin Jambor <mjambor@suse.cz>
24168 PR tree-optimization/80293
24169 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
24170 char arrays not totally scalarizable if it is false.
24171 (analyze_all_variable_accesses): Pass correct value in the new
24172 parameter. Add a statistics counter.
24174 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
24176 PR middle-end/79931
24177 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
24179 2017-04-24 Richard Biener <rguenther@suse.de>
24181 PR tree-optimization/80494
24182 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
24183 out for complex types.
24185 2017-04-24 Richard Biener <rguenther@suse.de>
24187 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
24188 * tree-ssa-sccvn.c (print_scc): Print SCC size.
24189 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
24190 (DFS): Adjust and never fail.
24191 (sccvn_dom_walker::fail): Remove.
24192 (sccvn_dom_walker::before_dom_children): Adjust.
24193 (run_scc_vn): Likewise and never fail.
24194 * tree-ssa-pre.c (pass_pre::execute): Adjust.
24195 (pass_fre::execute): Likewise.
24197 2017-04-24 Richard Biener <rguenther@suse.de>
24199 PR tree-optimization/79725
24200 * tree-ssa-sink.c (statement_sink_location): Return whether
24201 failure reason was zero uses. Move that check later.
24202 (sink_code_in_bb): Deal with zero uses by removing the stmt
24205 2017-04-24 Richard Biener <rguenther@suse.de>
24208 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
24209 pointer-based references.
24211 2017-04-24 Richard Biener <rguenther@suse.de>
24214 * pass_manager.h (pass_manager::operator new): Remove.
24215 (pass_manager::operator delete): Likewise.
24216 * passes.c (pass_manager::operator new): Remove.
24217 (pass_manager::operator delete): Likewise.
24218 (pass_manager::pass_manager): Zero individual pass members.
24220 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
24223 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
24224 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
24225 Check "XEXP (src, 1)" operand here.
24226 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
24227 Check "XEXP (src, 1)" operand here.
24228 (dimode_scalar_chain::make_vector_copies): Detect count register
24229 of a shift instruction. Zero extend count register from QImode
24230 to DImode to satisfy vector shift pattern count operand predicate.
24231 Substitute vector shift count operand with a DImode copy.
24232 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
24235 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
24237 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
24238 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
24239 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
24240 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
24241 (UNSPEC_NOREX_MEM): Remove definition.
24243 2017-04-21 Richard Biener <rguenther@suse.de>
24245 PR tree-optimization/79547
24246 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
24247 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
24248 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
24249 without any constraints.
24251 2017-04-21 Richard Biener <rguenther@suse.de>
24253 PR tree-optimization/78847
24254 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
24256 2017-04-21 Richard Biener <rguenther@suse.de>
24258 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
24259 (build_distinct_type_copy): Likewise.
24260 (build_variant_type_copy): Likewise.
24261 * tree.c (build_qualified_type): Pass down mem-stat info.
24262 (build_distinct_type_copy): Likewise.
24263 (build_variant_type_copy): Likewise.
24265 2017-04-21 Richard Biener <rguenther@suse.de>
24267 PR tree-optimization/80237
24268 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
24270 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
24271 for a simplified result.
24273 2016-04-21 Richard Biener <rguenther@suse.de>
24275 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
24276 sth as strict as a simple_iv but a chrec without symbols and an
24277 operand defined in the loop we are peeling (and not some subloop).
24278 (propagate_constants_for_unrolling): Propagate all constants.
24280 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
24283 * config/i386/i386.c (print_reg): Remove assert for disalowed
24284 regno values, call output_operand_lossage instead.
24286 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
24289 * config/i386/constraints.md (Yc): New register constraint.
24290 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
24291 Use Yc constraint for alternative 2 of operand 0. Remove
24292 preferred_for_speed attribute.
24294 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
24296 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
24297 lastprivate clauses in SIMT case.
24299 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
24301 * doc/invoke.texi (-Wextra-semi): Document new warning option.
24303 2017-04-20 Richard Biener <rguenther@suse.de>
24305 PR tree-optimization/57796
24306 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
24307 as N scalar stores.
24308 (vect_model_load_cost): Cost gathers as N scalar loads.
24310 2017-04-20 Richard Biener <rguenther@suse.de>
24312 * ggc-page.c (ggc_allocated_p): Rename to ...
24313 (safe_lookup_page_table_entry): ... this and return the lookup
24315 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
24317 2017-04-20 Richard Biener <rguenther@suse.de>
24319 PR tree-optimization/80453
24320 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
24321 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
24322 from the conditions.
24323 (vn_phi_eq): Pass them down.
24324 (vn_phi_lookup): Record them.
24325 (vn_phi_insert): Likewise.
24327 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
24329 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
24330 uninitialized variable warning to avoid buffer overrun.
24332 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
24335 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
24336 is suppressed for '{ 0 }' in C.
24338 2017-04-20 Jakub Jelinek <jakub@redhat.com>
24340 * BASE-VER: Set to 8.0.0.
24342 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
24344 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
24345 priority .init_array and .fini_array section with SECTION_NOTYPE
24348 2017-04-20 Jakub Jelinek <jakub@redhat.com>
24350 PR middle-end/80423
24351 * tree.h (build_array_type): Add typeless_storage default argument.
24352 * tree.c (type_cache_hasher::equal): Also compare
24353 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
24354 (build_array_type): Add typeless_storage argument, set
24355 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
24357 (build_nonshared_array_type): Adjust build_array_type_1 caller.
24358 (build_array_type): Likewise. Add typeless_storage argument.
24360 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
24361 Jakub Jelinek <jakub@redhat.com>
24363 PR tree-optimization/80426
24364 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
24365 operation on symbolic operands, also compute the overflow for the
24366 invariant part when the operation degenerates into a negation.
24368 2017-04-19 Jakub Jelinek <jakub@redhat.com>
24371 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
24372 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
24375 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
24377 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
24380 * config/avr/avr.c (tree.h): Include it.
24381 (cgraph.h): Include it.
24382 (avr_encode_section_info): Don't warn for uninitialized progmem
24383 variable if it's just an alias.
24385 2017-04-19 Richard Biener <rguenther@suse.de>
24388 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
24389 when needed by AutoPGO.
24391 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
24394 * doc/lto.texi: Remove an extra 'that'.
24396 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
24398 PR rtl-optimization/80429
24399 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
24400 are only used in debug insns.
24402 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
24403 Vladimir Makarov <vmakarov@redhat.com>
24405 * config/sparc/predicates.md (input_operand): Add comment. Return
24406 true for any memory operand when LRA is in progress.
24407 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
24409 2017-04-18 Jeff Law <law@redhat.com>
24412 * mips.md ({return,simple_return}_internal): Do not overwrite
24415 2017-04-18 Jakub Jelinek <jakub@redhat.com>
24417 PR tree-optimization/80443
24418 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
24419 instead of adding 1, subtract -1 and similarly instead of subtracting
24422 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
24424 PR rtl-optimization/80357
24425 * haifa-sched.c (tmp_bitmap): New variable.
24426 (model_recompute): Handle duplicate use records.
24427 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
24428 (free_global_sched_pressure_data): Free it.
24430 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
24433 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
24434 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
24435 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
24436 instead of SYSTEM_HEADER_DIR.
24438 2017-04-18 Jeff Law <law@redhat.com>
24440 PR middle-end/80422
24441 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
24442 predecessors after walking up the insn chain.
24444 2017-04-18 Jakub Jelinek <jakub@redhat.com>
24447 * dwarf2out.c (modified_type_die): Try harder not to emit internal
24448 sizetype type into debug info.
24450 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24453 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
24454 unneeded test for TARGET_UPPER_REGS_SF.
24455 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
24457 2017-04-18 Jakub Jelinek <jakub@redhat.com>
24460 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
24461 instead of gsi_after_labels.
24463 2017-04-18 Jeff Law <law@redhat.com>
24465 * regcprop.c (maybe_mode_change): Avoid creating copies of the
24469 2017-04-13 Jeff Law <law@redhat.com>
24470 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
24471 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
24473 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
24476 * config/avr/avr.c (intl.h): Include it.
24477 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
24479 2017-04-18 Martin Liska <mliska@suse.cz>
24481 PR gcov-profile/78783
24482 * gcov-tool.c (gcov_output_files): Validate that destination
24483 file is either removed by the tool or by a user.
24485 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
24486 Guy Benyei <guybe@mellanox.com>
24488 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
24489 block, and do not negate it, the stored id is already negative.
24491 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
24493 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
24495 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
24498 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
24499 masks of options that should be turned off if the VSX vector
24500 options are turned off.
24501 (OTHER_P8_VECTOR_MASKS): Likewise.
24502 (OTHER_VSX_VECTOR_MASKS): Likewise.
24503 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
24504 rs6000_disable_incompatible_switches to validate no type switches
24506 (rs6000_incompatible_switch): New function to disallow turning on
24507 other vector options if -mno-vsx, -mno-power8-vector, or
24508 -mno-power9-vector are specified.
24510 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
24512 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
24514 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
24516 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
24517 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
24518 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
24519 (ARG_POINTER_CFA_OFFSET): Likewise.
24521 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
24523 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
24524 conditions to take advantage of various optimizations.
24526 2017-04-13 Jeff Law <law@redhat.com>
24528 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
24529 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
24530 (zero_extendsidi2_dext): Likewise.
24532 2017-04-13 Jakub Jelinek <jakub@redhat.com>
24535 * fold-const.c (fold_ternary_loc): Revert
24536 use op0 instead of fold_convert_loc (loc, type, arg0) part of
24539 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
24541 PR rtl-optimization/80343
24542 * lra-remat.c (update_scratch_ops): Assign original hard reg to
24543 new scratch pseudo.
24545 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
24548 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
24549 to ubsan_encode_value.
24551 2017-04-13 Jeff Law <law@redhat.com>
24553 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
24554 appearing in DEBUG_INSNs.
24556 2017-04-13 Martin Liska <mliska@suse.cz>
24558 PR gcov-profile/80413
24559 * gcov-io.c (gcov_write_string): Copy to buffer just when
24560 allocated size is greater than zero.
24562 2017-04-13 Jakub Jelinek <jakub@redhat.com>
24565 * dwarf2out.c (decls_for_scope): Ignore declarations of
24566 current_function_decl in BLOCK_NONLOCALIZED_VARS.
24568 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
24571 * ipa-visibility.c (non_local_p): Fix typos.
24572 (localize_node): When localizing symbol in same comdat group,
24573 dissolve the group only when we know external symbols are going
24575 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
24577 2017-04-12 Jakub Jelinek <jakub@redhat.com>
24579 PR tree-optimization/79390
24580 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
24581 order does not result in usable sequence, retry with reversed operand
24587 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
24588 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
24589 op0 instead of fold_convert_loc (loc, type, arg0).
24591 2017-04-12 Jeff Law <law@redhat.com>
24593 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
24594 has a delay slot in the generated code.
24596 * config/cris/cris.md (cris_preferred_reload_class): Return
24597 GENNONACR_REGS rather than GENERAL_REGS.
24599 2017-04-12 Jakub Jelinek <jakub@redhat.com>
24602 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
24603 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
24604 signedness of the result type.
24606 2017-04-12 Richard Biener <rguenther@suse.de>
24607 Jeff Law <law@redhat.com>
24609 PR tree-optimization/80359
24610 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
24611 trim stores to TARGET_MEM_REFs.
24613 2017-04-12 Richard Biener <rguenther@suse.de>
24615 PR tree-optimization/79390
24616 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
24617 threading case even more.
24619 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
24622 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
24623 for quad_address_p for TImode, instead of just not indexed_address.
24625 2017-04-12 Richard Biener <rguenther@suse.de>
24626 Bernd Edlinger <bernd.edlinger@hotmail.de>
24628 PR middle-end/79671
24629 * alias.c (component_uses_parent_alias_set_from): Handle
24630 TYPE_TYPELESS_STORAGE.
24631 (get_alias_set): Likewise.
24632 * tree-core.h (tree_type_common): Add typeless_storage flag.
24633 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
24634 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
24635 for types containing members with TYPE_TYPELESS_STORAGE.
24636 (place_field): Likewise.
24637 (layout_type): Likewise for ARRAY_TYPE.
24638 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
24639 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
24640 TYPE_TYPELESS_STORAGE.
24641 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
24643 2017-04-12 Jakub Jelinek <jakub@redhat.com>
24646 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
24647 first argument to type.
24649 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24653 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
24654 CONST0_RTX (mode) rather than const0_rtx where appropriate.
24655 (rs6000_expand_binop_builtin): Likewise.
24656 (rs6000_expand_ternop_builtin): Likewise; also add missing
24657 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
24658 vshasigma built-ins.
24659 * doc/extend.texi: Document that vec_xxpermdi's third argument
24660 must be a constant.
24662 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
24664 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
24665 Use shift_const cost parameter when calculating gain of STV shifts.
24667 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
24669 PR rtl-optimization/70478
24670 * lra-constraints.c (process_alt_operands): Check memory for
24671 disfavoring memory insn operand.
24673 2017-04-11 Jakub Jelinek <jakub@redhat.com>
24675 PR middle-end/80100
24676 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
24677 left shift in unsigned HOST_WIDE_INT type.
24679 PR rtl-optimization/80385
24680 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
24681 (not (neg X)) into (plus X -1) for complex or non-integral modes.
24684 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
24685 if they have any depend clauses.
24687 2017-04-11 Martin Liska <mliska@suse.cz>
24690 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
24691 * ipa-split.c (split_function): Create a local comdat symbol
24692 if caller is in a comdat group.
24694 2017-04-11 Martin Liska <mliska@suse.cz>
24697 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
24700 2017-04-11 Martin Sebor <msebor@redhat.com>
24702 PR middle-end/80364
24703 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
24704 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
24706 (directive::set_width, directive::set_precision, format_character):
24708 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
24711 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
24714 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
24715 conflict, set target->arch_name instead of target->cpu_name.
24717 2017-04-11 Richard Biener <rguenther@suse.de>
24719 PR tree-optimization/80374
24720 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
24721 build_zero_cst, remove fold_convertible_p check again.
24723 2017-04-11 Martin Liska <mliska@suse.cz>
24726 * ubsan.c (instrument_object_size): Do not instrument register
24729 2017-04-11 Jakub Jelinek <jakub@redhat.com>
24732 * config/i386/i386-builtin-types.def
24733 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
24734 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
24735 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
24736 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
24737 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
24738 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
24739 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
24740 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
24741 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
24742 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
24743 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
24744 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
24745 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
24746 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
24747 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
24748 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
24749 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
24750 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
24751 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
24752 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
24753 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
24754 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
24755 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
24756 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
24757 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
24758 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
24759 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
24760 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
24761 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
24762 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
24763 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
24764 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
24765 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
24766 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
24767 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
24768 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
24769 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
24770 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
24771 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
24772 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
24773 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
24774 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
24775 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
24776 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
24777 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
24778 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
24780 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
24781 flag to second_arg_count, handle 4 argument function type _COUNT
24782 aliases, handle second_arg_count on second argument rather than last.
24784 2017-04-10 Jeff Law <law@redhat.com>
24786 PR tree-optimization/80374
24787 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
24788 record anything if we can not convert integer_zero_node to the
24791 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
24794 * config/rs6000/rs6000.c (rs6000_option_override_internal):
24795 Enhance special handling given to the TARGET_P9_MINMAX option in
24796 relation to certain other options.
24798 2017-04-10 Bin Cheng <bin.cheng@arm.com>
24800 PR tree-optimization/80153
24801 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
24802 remove POINTER_PLUS_EXPR's base part directly, rather than through
24805 2017-04-10 Richard Biener <rguenther@suse.de>
24806 Bin Cheng <bin.cheng@arm.com>
24808 PR tree-optimization/80153
24809 * tree-affine.c (aff_combination_to_tree): Get base pointer from
24810 the first element of pointer type aff_tree. Build result expr in
24812 (add_elt_to_tree): Convert to type unconditionally. Remove other
24813 fold_convert calls.
24814 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
24815 (rewrite_use_nonlinear_expr): Check invariant using iv information.
24817 2017-04-10 Richard Biener <rguenther@suse.de>
24819 * tree-ssa-structalias.c (find_func_aliases): Properly handle
24822 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
24824 PR rtl-optimization/70478
24825 * lra-constraints.c (curr_small_class_check): New.
24826 (update_and_check_small_class_inputs): New.
24827 (process_alt_operands): Update curr_small_class_check. Disfavor
24828 alternative insn memory operands. Check available regs for small
24831 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
24834 * config/mips/mips.opt (-mvirt): Update description.
24835 * doc/invoke.texi (-mvirt): Likewise.
24837 2017-04-10 Richard Biener <rguenther@suse.de>
24839 PR middle-end/80362
24840 * fold-const.c (fold_binary_loc): Look at unstripped ops when
24841 looking for NEGATE_EXPR in -A / -B to A / B folding.
24843 2017-04-10 Martin Liska <mliska@suse.cz>
24845 PR gcov-profile/80224
24846 * gcov.c (print_usage): Fix usage string.
24847 (get_gcov_intermediate_filename): Remove.
24848 (output_gcov_file): Use both for normal and intermediate format.
24849 (generate_results): Do not initialize special file for
24850 intermediate format.
24852 2017-04-10 Richard Biener <rguenther@suse.de>
24854 PR tree-optimization/80304
24855 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
24858 2017-04-10 Nathan Sidwell <nathan@acm.org>
24861 * config/rs6000/rs6000.c (rs6000_vector_type): New.
24862 (rs6000_init_builtins): Use it.
24864 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24866 * config/arm/arm.md (<mrc>): Add mode to SET source.
24867 (<mrrc>): Likewise.
24869 2017-04-10 Richard Biener <rguenther@suse.de>
24871 PR middle-end/80344
24872 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
24874 2017-04-10 Jakub Jelinek <jakub@redhat.com>
24877 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
24878 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
24879 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
24880 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
24881 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
24882 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
24883 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
24884 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
24885 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
24886 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
24887 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
24888 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
24889 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
24890 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
24891 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
24892 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
24893 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
24894 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
24895 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
24896 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
24897 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
24898 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
24899 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
24901 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
24903 PR rtl-optimization/70478
24904 * lra-constraints.c: Reverse the last patch.
24906 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
24908 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
24909 Add comment for WCHAR_T.
24911 2017-04-08 Martin Liska <mliska@suse.cz>
24914 2017-04-07 Martin Liska <mliska@suse.cz>
24917 * ipa-split.c (split_function): Add function part to a same comdat
24920 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24923 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
24925 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
24927 * rs6000/rs6000.c (vec_load_pendulum): Rename...
24928 (vec_pairing): ...to this.
24929 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
24930 (rs6000_sched_init): Adjust for name change.
24931 (struct rs6000_sched_context): Likewise.
24932 (rs6000_init_sched_context): Likewise.
24933 (rs6000_set_sched_context): Likewise.
24935 2017-04-07 Jakub Jelinek <jakub@redhat.com>
24941 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
24943 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
24944 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
24945 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
24947 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
24949 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
24951 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
24953 PR rtl-optimization/70703
24954 * ira-color.c (update_conflict_hard_regno_costs): Use
24955 int64_t instead of HOST_WIDE_INT.
24957 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
24959 PR rtl-optimization/70478
24960 * lra-constraints.c (process_alt_operands): Disfavor alternative
24961 insn memory operands.
24963 2017-04-07 Jeff Law <law@redhat.com>
24965 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
24966 CALL and NOTE_INSN_CALL_ARG_LOCATION.
24968 2017-04-07 Martin Liska <mliska@suse.cz>
24971 * config/aarch64/aarch64.c (aarch64_process_target_attr):
24972 Show error message instead of an ICE.
24974 2017-04-07 Martin Liska <mliska@suse.cz>
24977 * ipa-split.c (split_function): Add function part to a same comdat
24980 2017-04-07 Richard Biener <rguenther@suse.de>
24982 PR middle-end/80341
24983 * tree.c (get_unwidened): Also handle ! for_type case for
24985 * convert.c (do_narrow): Split out from ...
24986 (convert_to_integer_1): ... here. Do not pass final truncation
24987 type to get_unwidened for TRUNC_DIV_EXPR.
24989 2017-04-07 Richard Biener <rguenther@suse.de>
24991 * tree-affine.c (wide_int_ext_for_comb): Take type rather
24993 (aff_combination_const): Adjust.
24994 (aff_combination_scale): Likewise.
24995 (aff_combination_add_elt): Likewise.
24996 (aff_combination_add_cst): Likewise.
24997 (aff_combination_convert): Likewise.
24998 (add_elt_to_tree): Likewise. Remove unused argument.
24999 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
25001 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
25003 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
25005 * config/arm/arm.c (arm_default_short_enums): Use
25006 ARM_DEFAULT_SHORT_ENUMS.
25007 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
25009 2017-04-06 Jakub Jelinek <jakub@redhat.com>
25012 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
25013 members with redundant out-of-class redeclaration.
25015 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
25018 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
25019 * config/i386/i386.md (*zero_extendsidi2):
25020 Add (?*x,*x) and (?*v,*v) alternatives.
25022 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
25025 * config/i386/i386.c (ix86_expand_builtin)
25026 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
25027 mode from insn data. Convert operands to insn operand mode.
25028 Copy operands that don't satisfy insn predicate to a register.
25030 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
25032 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
25035 2017-04-06 Richard Biener <rguenther@suse.de>
25037 PR tree-optimization/80334
25038 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
25039 preserve alignment of accesses.
25041 2017-04-06 Richard Biener <rguenther@suse.de>
25043 PR tree-optimization/80262
25044 * tree-sra.c (build_ref_for_offset): Preserve address-space
25046 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
25047 Drop useless address-space information on MEM_REF offsets.
25049 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
25051 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
25053 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
25055 PR rtl-optimization/70703
25056 * ira-color.c (update_conflict_hard_regno_costs): Use
25057 HOST_WIDE_INT instead of long.
25059 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
25062 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
25063 not defined for x86_64 target. Add -mmmx target option when __SSE2__
25065 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
25066 for x86_64 target. Handle -m3dnowa option.
25068 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
25070 PR rtl-optimization/70703
25071 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
25072 (update_conflict_hard_regno_costs): Use long instead of unsigned
25073 arithmetic for cost calculation.
25075 2017-04-05 Jakub Jelinek <jakub@redhat.com>
25076 Bernd Edlinger <bernd.edlinger@hotmail.de>
25079 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
25082 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
25085 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
25086 ptr_mode with Pmode throughout.
25087 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
25088 into probe_stack_range and use DImode.
25090 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
25093 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
25094 call_eh_return is true.
25096 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25098 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
25099 Initialize last_match_fntype_index.
25101 2017-04-05 Jakub Jelinek <jakub@redhat.com>
25104 * tree-nvr.c: Include internal-fn.h.
25105 (pass_return_slot::execute): Ignore internal calls without
25108 2017-04-04 Jakub Jelinek <jakub@redhat.com>
25109 Richard Biener <rguenther@suse.de>
25112 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
25113 captures used multiple times, except for the last use.
25114 * generic-match-head.c: Include gimplify.h.
25116 2017-04-04 Jakub Jelinek <jakub@redhat.com>
25118 PR tree-optimization/79390
25119 * target.h (struct noce_if_info): Declare.
25120 * targhooks.h (default_noce_conversion_profitable_p): Declare.
25121 * target.def (noce_conversion_profitable_p): New target hook.
25122 * ifcvt.h (struct noce_if_info): New type, moved from ...
25123 * ifcvt.c (struct noce_if_info): ... here.
25124 (noce_conversion_profitable_p): Renamed to ...
25125 (default_noce_conversion_profitable_p): ... this. No longer
25127 (noce_try_store_flag_constants, noce_try_addcc,
25128 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
25129 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
25130 instead of noce_conversion_profitable_p.
25131 * config/i386/i386.c: Include ifcvt.h.
25132 (ix86_option_override_internal): Don't override
25133 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
25134 (ix86_noce_conversion_profitable_p): New function.
25135 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
25136 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
25137 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
25138 * doc/tm.texi: Regenerated.
25140 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25142 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
25145 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25148 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
25149 instructions for small multiply cores.
25151 2017-04-04 Jeff Law <law@redhat.com>
25153 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
25155 (mips_expand_vec_perm_const): Initialize elements in orig_perm
25156 that are not set by the loop over the elements.
25158 2017-04-04 Jakub Jelinek <jakub@redhat.com>
25161 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
25162 int mode, convert_modes it to mode as unsigned, otherwise use
25163 lowpart_subreg to mode rather than SImode.
25164 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
25165 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
25166 Use DImode instead of SImode for the shift count operand.
25167 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
25170 2017-04-04 Richard Biener <rguenther@suse.de>
25172 PR middle-end/80281
25173 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
25174 arithmetic done for the negate or the plus. Simplify.
25175 (A - (-B) -> A + B): Likewise.
25176 * fold-const.c (split_tree): Make sure to not negate pointers.
25178 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
25180 PR rtl-optimization/60818
25181 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
25182 a compare of comparisons with the thing compared if this results
25183 in a different machine mode.
25185 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
25187 * alias.c (base_alias_check): Fix typo in comment.
25188 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
25189 * cgraphunit.c (symbol_table::compile): Likewise.
25190 * collect2.c (maybe_run_lto_and_relink): Likewise.
25191 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
25192 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
25193 * config/avr/avr.c (avr_map_op_t): Likewise.
25194 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
25195 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
25196 * config/epiphany/epiphany.md (movcc): Likewise.
25197 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
25198 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
25200 * config/mips/mips.c (mips_save_restore_reg): Likewise.
25201 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
25202 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
25203 * config/sh/sh.c (sh_rtx_costs): Likewise.
25204 * fold-const.c (fold_truth_andor): Likewise.
25205 * genautomata.c (collapse_flag): Likewise.
25206 * gengtype.h (struct type::u::s): Likewise.
25207 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
25208 * input.c (FORMAT_AMOUNT): Likewise.
25209 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
25210 (known_aggs_to_agg_replacement_list): Likewise.
25211 * ipa-inline-analysis.c: Likewise.
25212 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
25213 * ipa-polymorphic-call.c
25214 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
25215 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
25216 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
25218 * modulo-sched.c (apply_reg_moves): Likewise.
25219 * omp-expand.c (build_omp_regions_1): Likewise.
25220 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
25221 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
25222 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
25223 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
25224 * value-prof.c: Likewise.
25225 * var-tracking.c (val_reset): Likewise.
25227 2017-04-03 Richard Biener <rguenther@suse.de>
25229 PR tree-optimization/80275
25230 * fold-const.c (split_address_to_core_and_offset): Handle
25233 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
25235 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
25236 descriptors is at least equal to that of functions.
25238 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
25240 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
25242 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
25245 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
25246 (mov<IMOD4:mode>): New expander.
25247 (*mov<IMOD4:mode>_internal): New insn and split pattern.
25249 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
25251 PR rtl-optimization/79405
25252 * fwprop.c (propagations_left): New variable.
25253 (forward_propagate_into): Decrement it.
25254 (fwprop_init): Initialize it.
25255 (fw_prop): If the variable has reached zero, stop propagating.
25256 (fwprop_addr): Ditto.
25258 2017-03-31 Jakub Jelinek <jakub@redhat.com>
25261 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
25262 a FUNCTION_DECL, pass it as decl instead of origin to
25265 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
25267 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
25270 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
25273 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
25274 TARGET_VSX_SMALL_INTEGER.
25276 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25278 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
25279 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
25281 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
25283 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
25284 extraction from odd-numbered MSA register.
25286 2017-03-31 Jakub Jelinek <jakub@redhat.com>
25288 PR middle-end/80173
25289 * expmed.c (store_bit_field_1): Don't attempt to create
25290 a word subreg out of hard registers wider than word if they
25291 have HARD_REGNO_NREGS of 1 for their mode.
25293 PR middle-end/80163
25294 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
25295 conversions to integer types wider than word and pointer.
25298 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
25299 (rtx_equal_for_cselib_p): Pass 0 to it.
25300 * cselib.c (cselib_hasher::equal): Likewise.
25301 (rtx_equal_for_cselib_1): Add depth argument. If depth
25302 is 128, don't look up VALUE locs and punt. Increment
25303 depth in recursive calls when walking VALUE locs.
25305 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
25307 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
25308 (make_gcov_file_name): Use the canonical path name for generating
25310 (read_line): Fix handling of files with ascii null bytes.
25312 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
25314 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
25315 to initialise a vector register instead
25316 of using a const_int.
25318 2017-03-30 Jakub Jelinek <jakub@redhat.com>
25320 PR translation/80189
25321 * gimplify.c (omp_default_clause): Use %qs instead of %s in
25322 diagnostic messages.
25324 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
25327 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
25328 (dfp_diex_<mode>): Update mode of operand 1.
25329 * doc/extend.texi (dxex, dxexq): Document change to return type.
25330 (diex, diexq): Document change to argument type.
25332 2017-03-30 Martin Jambor <mjambor@suse.cz>
25335 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
25336 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
25337 it reflects the signature changes performed at the callee side.
25338 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
25339 to cgraph_build_function_type_skip_args.
25340 (build_function_decl_skip_args): Adjust call to the above function.
25342 2017-03-30 Jakub Jelinek <jakub@redhat.com>
25345 * config/i386/sse.md
25346 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
25347 register as dest whenever it is a MEM not rtx_equal_p to the
25348 corresponding dup operand, and when forcing into reg move the
25349 reg into the memory afterwards.
25350 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
25351 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
25352 for the force_reg mode.
25353 (avx512vl_vextractf128<mode>): Use register as dest either
25354 always when a MEM, or when it is a MEM not rtx_equal_p to the
25355 corresponding dup operand, or even not when it is a CONST_VECTOR
25356 depending on the mode and lo vs. hi.
25357 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
25359 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
25360 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
25361 Likewise. Require that operands[2] is even.
25362 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
25363 Remove extraneous parens. Require that operands[2] is a multiple
25365 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
25366 operands[0] is a MEM if <mask_applied>, the predicates/constraints
25367 disallow memory then.
25369 2017-03-30 Richard Biener <rguenther@suse.de>
25371 PR tree-optimization/77498
25372 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
25373 to non-constants over backedges.
25375 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
25377 PR rtl-optimization/80233
25378 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
25379 as last_combined_insn. Do not test for BARRIER_P separately.
25381 2017-03-29 Andreas Schwab <schwab@suse.de>
25384 * calls.c (prepare_call_address): Convert funexp to Pmode before
25385 copying to temp reg.
25387 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25389 PR tree-optimization/80158
25390 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
25391 Handle possible future case of more than one alternate
25393 (replace_rhs_if_not_dup): Likewise.
25394 (replace_one_candidate): Likewise.
25396 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
25398 PR rtl-optimization/80193
25399 * ira.c (ira): Do not check allocation for LRA.
25401 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
25403 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
25404 (nvptx_output_simt_exit): Declare.
25405 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
25406 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
25407 (init_softstack_frame): Move initialization of crtl->is_leaf to...
25408 (nvptx_declare_function_name): ...here. Emit declaration of local
25409 memory space buffer for omp_simt_enter insn.
25410 (nvptx_output_unisimt_switch): New.
25411 (nvptx_output_softstack_switch): New.
25412 (nvptx_output_simt_enter): New.
25413 (nvptx_output_simt_exit): New.
25414 * config/nvptx/nvptx.h (struct machine_function): New fields
25415 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
25416 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
25417 (UNSPECV_SIMT_EXIT): Ditto.
25418 (omp_simt_enter_insn): New insn.
25419 (omp_simt_enter): New expansion.
25420 (omp_simt_exit): New insn.
25421 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
25423 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
25424 (expand_GOMP_SIMT_ENTER_ALLOC): New.
25425 (expand_GOMP_SIMT_EXIT): New.
25426 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
25427 (GOMP_SIMT_ENTER_ALLOC): Ditto.
25428 (GOMP_SIMT_EXIT): Ditto.
25429 * target-insns.def (omp_simt_enter): New insn.
25430 (omp_simt_exit): Ditto.
25431 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
25433 (lower_rec_simd_input_clauses): Implement SIMT privatization.
25434 (lower_rec_input_clauses): Likewise.
25435 (lower_lastprivate_clauses): Handle SIMT privatization.
25437 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
25438 (ompdevlow_adjust_simt_enter): New.
25439 (find_simtpriv_var_op): New.
25440 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
25441 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
25443 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
25444 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
25445 (copy_decl_for_dup_finish): Ditto.
25447 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
25449 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
25452 * config/i386/i386.c (ix86_option_override_internal): Always
25453 allow -mpreferred-stack-boundary=3 for 64-bit targets.
25455 2017-03-28 Bin Cheng <bin.cheng@arm.com>
25457 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
25459 2017-03-28 Bin Cheng <bin.cheng@arm.com>
25461 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
25462 mark new edge's irreducible flag accordign to it.
25463 (vect_do_peeling): Check loop preheader edge's irreducible flag
25464 and pass it to function slpeel_add_loop_guard.
25466 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
25468 PR tree-optimization/80218
25469 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
25470 Update block frequencies and counts.
25472 2017-03-28 Richard Biener <rguenther@suse.de>
25474 PR tree-optimization/78644
25475 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
25476 of a simplification result we may not use it at all.
25478 2017-03-28 Richard Biener <rguenther@suse.de>
25481 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
25482 without arguments, generate default definition of a SSA name.
25484 2017-03-28 Richard Biener <rguenther@suse.de>
25486 PR middle-end/80222
25487 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
25488 TYPE_REF_CAN_ALIAS_ALL references.
25489 * fold-const.c (fold_indirect_ref_1): Likewise.
25491 2017-03-28 Martin Liska <mliska@suse.cz>
25494 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
25495 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
25497 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
25498 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
25500 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
25501 (EXTRA_SPECS): Define.
25502 (SUBTARGET_EXTRA_SPECS): Likewise.
25503 (SUBTARGET_CPP_SPEC): Likewise.
25504 * config/arc/elf.h (EXTRA_SPECS): Renamed to
25505 SUBTARGET_EXTRA_SPECS.
25506 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
25508 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
25510 * config/arc/simdext.md (vst64_insn): Update pattern.
25511 (vld32wh_insn): Likewise.
25512 (vld32wl_insn): Likewise.
25513 (vld64_insn): Likewise.
25514 (vld32_insn): Likewise.
25516 2017-03-28 Marek Polacek <polacek@redhat.com>
25519 * fold-const.c (fold_comparison): Use protected_set_expr_location
25520 instead of SET_EXPR_LOCATION.
25522 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
25524 * tree.c (add_expr): Avoid name lookup warning.
25526 2017-03-27 Jeff Law <law@redhat.com>
25528 PR tree-optimization/80216
25529 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
25530 function name. Limit recursion depth.
25531 (record_temporary_equivalences): Corresponding changes.
25533 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
25535 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
25538 2017-03-27 Jakub Jelinek <jakub@redhat.com>
25541 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
25543 * cfgcleanup.c (reg_note_cfa_p): New array.
25544 (insns_have_identical_cfa_notes): New function.
25545 (old_insns_match_p): Don't cross-jump in between /f
25546 and non-/f instructions. If both i1 and i2 are frame related,
25547 verify all CFA notes, their order and content.
25549 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
25552 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
25553 HImode and SImode with zero extend to DImode to one insn.
25554 (bswap<mode>2_extenddi): Likewise.
25555 (bswapsi2_extenddi): Likewise.
25556 (bswaphi2_extendsi): Likewise.
25557 (bswaphi2): Combine bswap HImode and SImode into one insn.
25558 Separate memory insns from swapping register.
25559 (bswapsi2): Likewise.
25560 (bswap<mode>2): Likewise.
25561 (bswaphi2_internal): Delete, no longer used.
25562 (bswapsi2_internal): Likewise.
25563 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
25564 store, and gpr<-gpr swap insns.
25565 (bswap<mode>2_store): Likewise.
25566 (bswaphi2_reg): Register only splitter, combine with the splitter.
25567 (bswaphi2 splitter): Likewise.
25568 (bswapsi2_reg): Likewise.
25569 (bswapsi2 splitter): Likewise.
25570 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
25571 the insns into load, store, and register/register insns.
25572 (bswapdi2_ldbrx): Likewise.
25573 (bswapdi2_load): Likewise.
25574 (bswapdi2_store): Likewise.
25575 (bswapdi2_reg): Likewise.
25577 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
25579 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
25580 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
25582 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
25585 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
25587 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
25588 special handling for target option conflicts between dform
25589 options (-mpower9-dform, -mpower9-dform-vector,
25590 -mpower9-dform-scalar) and -mno-direct-move.
25592 2017-03-27 Richard Biener <rguenther@suse.de>
25594 PR tree-optimization/80181
25595 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
25597 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
25599 * config/arc/predicates.md (move_double_src_operand): Replace the
25600 call to move_double_src_operand with a call to address_operand.
25602 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
25604 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
25605 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
25606 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
25608 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
25610 * config/arc/predicates.md (long_immediate_loadstore_operand):
25611 Consider scaled addresses cases.
25613 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
25615 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
25616 restored when in interrupt.
25617 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
25618 doesn't have delay slot.
25620 2017-03-27 Richard Biener <rguenther@suse.de>
25623 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
25624 inlined thunk clones.
25626 2017-03-27 Jakub Jelinek <jakub@redhat.com>
25629 * asan.c (instrument_derefs): Copy over last operand from
25630 original COMPONENT_REF to the new COMPONENT_REF with
25631 DECL_BIT_FIELD_REPRESENTATIVE.
25632 * ubsan.c (instrument_object_size): Likewise.
25634 2017-03-27 Richard Biener <rguenther@suse.de>
25636 PR tree-optimization/80170
25637 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
25638 sure DR/SCEV didnt fold in constants we do not see when looking
25639 at the reference base alignment.
25641 2017-03-27 Richard Biener <rguenther@suse.de>
25643 PR middle-end/80171
25644 * gimple-fold.c (fold_ctor_reference): Properly guard against
25645 NULL return value from canonicalize_constructor_val.
25647 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
25650 * config/i386/i386.c (ix86_expand_builtin)
25651 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
25652 flags reg setting and flags reg using instructions.
25653 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
25654 clobbering instructions to zero extend op2.
25656 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
25658 * doc/install.texi (Configuration) <--with-aix-soname>:
25659 Update link to AIX ld.
25661 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
25663 PR rtl-optimization/80160
25664 PR rtl-optimization/80159
25665 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
25666 reg_alternate_class into account.
25668 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
25671 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
25672 to consider in curr_insn_transform.
25674 2017-03-24 Jakub Jelinek <jakub@redhat.com>
25676 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
25677 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
25678 and emit_mode_inner.
25680 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25682 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
25683 argument to the overloaded builtin variants. Use the new flag to
25684 deprecate certain builtin variants.
25685 * config/s390/s390-builtin-types.def: Add new builtin types.
25686 * config/s390/s390-builtins.h: Support new flags field for
25687 overloaded builtins.
25688 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
25689 (s390_macro_to_expand): Enable vector float data type.
25690 (s390_cpu_cpp_builtins_internal): Indicate support of the new
25691 builtins by incrementing the __VEC__ version number.
25692 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
25694 (s390_resolve_overloaded_builtin): Emit error messages depending
25695 on the builtin flags.
25696 * config/s390/s390.c (s390_expand_builtin): Support additional
25697 flags argument. Change error message to match the messages
25698 emitted in s390-c.c.
25699 * config/s390/s390.md: New UNSPEC_* constants.
25700 (op_type): Add new instruction types.
25701 * config/s390/vecintrin.h: Add new builtins and test data class
25703 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
25704 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
25705 (VEC_INEXACT, VEC_NOINEXACT): New constants.
25706 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
25707 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
25708 ("vec_mergel<mode>"): V_HW -> VEC_HW.
25710 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
25711 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
25712 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
25713 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
25715 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
25716 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
25717 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
25718 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
25720 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
25721 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
25722 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
25723 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
25724 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
25725 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
25726 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
25728 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
25729 ("vec_scatter_element<V_HW_4:mode>_DI")
25730 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
25731 ("vec_fpint<mode>", "vflls")
25732 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
25733 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
25734 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
25735 ("*vec_cmphe<mode>_cc"): ... these.
25737 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
25738 mode constant instead of magic value.
25740 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25742 * config/s390/s390.c (s390_expand_vec_compare): Support other
25743 vector floating point modes than just V2DF.
25744 (s390_expand_vcond): Likewise.
25745 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
25746 (s390_cannot_change_mode_class): Prevent mode changes between TF
25747 and V1TF in vector registers.
25748 * config/s390/s390.md (DF, SF): New mode attributes.
25749 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
25750 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
25751 SFmode support for VRs.
25752 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
25754 (VFT, VF_HW): New mode iterators.
25755 (vw, sdx): New mode attributes.
25756 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
25757 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
25758 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
25759 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
25760 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
25761 also the new vector floating point modes. Renaming to ...
25763 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
25764 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
25765 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
25766 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
25767 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
25768 ("vec_unordered<mode>"): ... these.
25770 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
25771 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
25772 ("*vec_extendv2df"): New insn definitions.
25774 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25776 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
25777 ("mulditi3_2", "*muldi3_sign"): New patterns.
25778 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
25779 rename the pattern definition.
25781 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25783 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
25785 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
25787 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25789 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
25790 instruction if possible.
25791 * config/s390/vector.md (vec_halfnumelts): New mode
25793 ("*vec_vllezlf<mode>"): New pattern.
25795 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25797 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
25798 ("popcountv4si2", "popcountv2di2"): Rename to ...
25799 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
25800 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
25802 ("popcount<mode>2_vxe"): New pattern.
25804 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25806 * common/config/s390/s390-common.c (processor_flags_table): Add
25808 * config.gcc: Add arch12.
25809 * config/s390/driver-native.c (s390_host_detect_local_cpu):
25810 Default to arch12 for unknown CPU model numbers.
25811 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
25812 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
25813 PROCESSOR_max sanity check.
25814 * config/s390/s390-opts.h (enum processor_type): Add
25816 * config/s390/s390.c (processor_table): Add arch12.
25817 (s390_expand_builtin): Add check for B_VXE flag.
25818 (s390_issue_rate): Add PROCESSOR_ARCH12.
25819 (s390_get_sched_attrmask): Likewise.
25820 (s390_get_unit_mask): Likewise.
25821 (s390_sched_score): Enable z13 scheduling for arch12.
25822 (s390_sched_reorder): Likewise.
25823 (s390_sched_variable_issue): Likewise.
25824 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
25826 (s390_tune_attr): Use z13 scheduling also for arch12.
25827 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
25828 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
25829 (TARGET_VXE_P): New macros.
25830 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
25831 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
25832 * config/s390/s390.opt: Add arch12 as processor_type.
25834 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25836 * config/s390/s390.md
25837 ("fixuns_truncdddi2", "fixuns_trunctddi2")
25838 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
25839 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
25841 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
25842 Rename expanders to ...
25844 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
25845 ("fixuns_truncdddi2_emu"): ... these.
25847 ("fixuns_trunc<mode>si2_emu"): New expander.
25849 ("*fixuns_truncdfdi2_z13"): Rename to ...
25850 ("*fixuns_truncdfdi2_vx"): ... this.
25852 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25854 * config/s390/2964.md: Remove the single element vector compare
25855 instructions which are no longer used.
25856 * config/s390/s390.c (s390_select_ccmode): Remove handling of
25858 (s390_canonicalize_comparison): Remove handling of DFmode
25860 (s390_expand_vec_compare_scalar): Remove function.
25861 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
25862 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
25864 ("*cmp<mode>_ccs"): Add wfcdb instruction.
25866 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25868 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
25870 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
25871 will anyway by matched by mov<mode>_64dfp.
25873 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25875 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
25876 vlef/vstef. Add missing operand to vleif.
25878 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25880 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
25881 pair for all vector types with 64 bit elements.
25882 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
25883 * config/s390/vector.md (V_HW_64): ... here.
25884 (V_128_NOSINGLE): New mode iterator.
25885 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
25886 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
25887 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
25888 ("*vec_load_pairv2di"): Change to ...
25889 ("*vec_load_pair<mode>"): ... this one.
25891 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25893 * config/s390/constraints.md: Add comments.
25894 (jKK): Reject element sizes > 8 bytes.
25895 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
25897 * config/s390/s390.md: Add the s_operand checks formerly in
25898 s390_split_ok_p to various splitters where they are still
25900 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
25901 for 128 bit vectors. Plus two splitters.
25903 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25905 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
25908 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25911 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
25912 error if the boundary argument is not constant.
25914 2017-03-24 Jakub Jelinek <jakub@redhat.com>
25916 PR rtl-optimization/80112
25917 * loop-doloop.c (doloop_condition_get): Don't check condition
25918 if cmp isn't SET with IF_THEN_ELSE src.
25920 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25922 PR tree-optimization/80158
25923 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
25924 replacing a candidate statement, also replace it for the
25925 candidate's alternate interpretation.
25926 (replace_rhs_if_not_dup): Likewise.
25927 (replace_one_candidate): Likewise.
25929 2017-03-24 Richard Biener <rguenther@suse.de>
25931 PR tree-optimization/80167
25932 * graphite-isl-ast-to-gimple.c
25933 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
25935 (translate_isl_ast_to_gimple::get_rename): Likewise.
25937 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
25939 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
25940 handling of certain combinations of target options, including the
25941 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
25942 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
25944 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25947 * config/arm/arm.md (*load_multiple): Add reload_completed to
25948 matching condition.
25950 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25951 Richard Biener <rguenth@suse.de>
25953 PR tree-optimization/79908
25954 PR tree-optimization/80136
25955 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
25956 been cast away, gimplify_and_add suffices.
25958 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
25960 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
25962 2017-03-23 Richard Biener <rguenther@suse.de>
25964 PR tree-optimization/80032
25965 * gimplify.c (gimple_push_cleanup): Forced unconditional
25966 cleanups still have to go to the conditional_cleanups
25969 2017-03-22 Jakub Jelinek <jakub@redhat.com>
25971 PR tree-optimization/80072
25972 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
25974 (next_operand_entry_id): Change type to unsigned int.
25975 (sort_by_operand_rank): Make sure to return the right return value
25976 even if unsigned fields are bigger than INT_MAX.
25977 (struct oecount): Change cnt and id type to unsigned int.
25978 (oecount_hasher::equal): Formatting fix.
25979 (oecount_cmp): Make sure to return the right return value
25980 even if unsigned fields are bigger than INT_MAX.
25981 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
25984 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
25985 TREE_READONLY on result if writing it more than once.
25988 * doc/invoke.texi (-fsanitize=thread): Document that with
25989 -fnon-call-exceptions atomics are not able to throw
25993 * tsan.c: Include tree-eh.h.
25994 (instrument_builtin_call): Call maybe_clean_eh_stmt or
25995 maybe_clean_or_replace_eh_stmt where needed.
25996 (instrument_memory_accesses): Add cfg_changed argument.
25997 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
25998 if it returned true.
25999 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
26001 PR rtl-optimization/63191
26002 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
26003 wrapper function, moved the whole old content into ...
26004 (ix86_delegitimize_address_1): ... this. New inline function.
26005 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
26006 true as last argument instead of ix86_delegitimize_address.
26008 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
26010 * config/aarch64/aarch64.c (generic_branch_cost): Copy
26011 cortexa57_branch_cost.
26013 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
26015 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
26017 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26020 * doc/md.texi (Constraints): Document wA constraint.
26021 * config/rs6000/constraints.md (wA): New.
26022 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
26023 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
26024 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
26025 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
26027 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
26030 * gimplify.c (is_oacc_declared): New function.
26031 (oacc_default_clause): Use it to set default flags for acc declared
26032 variables inside parallel regions.
26033 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
26034 declared variables.
26035 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
26036 declare attribute to any decl as necessary.
26038 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
26041 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
26042 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
26043 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
26044 (arm_arch_lpae): This.
26045 * config/arm/arm.c (arm_arch7ve): Rename into ...
26046 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
26047 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
26050 2017-03-22 Martin Liska <mliska@suse.cz>
26053 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
26054 error message instead of an ICE.
26056 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26058 * doc/extend.texi (6.11 Additional Floating Types): Revise.
26060 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
26062 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
26064 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
26067 2017-03-21 Martin Sebor <msebor@redhat.com>
26069 * doc/extend.texi: Use "cannot" instead of "can't."
26070 * doc/hostconfig.texi: Same.
26071 * doc/install.texi: Same.
26072 * doc/invoke.texi: Same.
26073 * doc/loop.texi: Same.
26074 * doc/md.texi: Same.
26075 * doc/objc.texi: Same.
26076 * doc/rtl.texi: Same.
26077 * doc/tm.texi: Same.
26078 * doc/tm.texi.in: Same.
26079 * doc/trouble.texi: Same.
26081 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
26084 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
26085 (collect_checksum_attributes): Set it.
26086 (die_checksum_ordered): Use it.
26088 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26090 PR tree-optimization/79908
26091 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
26092 change: For a VA_ARG whose LHS has been cast away, use
26093 force_gimple_operand to construct the side effects.
26095 2017-03-21 David Malcolm <dmalcolm@redhat.com>
26097 PR translation/80001
26098 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
26099 more amenable to translation.
26100 (oacc_loop_auto_partitions): Likewise.
26102 2017-03-21 Marek Polacek <polacek@redhat.com>
26103 Martin Sebor <msebor@redhat.com>
26105 PR tree-optimization/80109
26106 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
26107 on INTEGRAL_TYPE_P.
26109 2017-03-21 Jakub Jelinek <jakub@redhat.com>
26110 Segher Boessenkool <segher@kernel.crashing.org>
26113 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
26114 check reg_used_between_p between insn and one of succ or succ2
26115 depending on if succ is artificial insn not inserted into insn
26118 2017-03-21 Martin Liska <mliska@suse.cz>
26120 PR gcov-profile/80081
26121 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
26122 * doc/gcc.texi: Include gcov-dump stuff.
26123 * doc/gcov-dump.texi: New file.
26125 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
26127 PR rtl-optimization/79150
26128 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
26129 conditional jump, if the jump is the last insn of the loop.
26131 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26132 Richard Biener <rguenth@suse.de>
26134 PR tree-optimization/79908
26135 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
26136 been cast away, use force_gimple_operand to construct the side
26139 2017-03-21 Martin Liska <mliska@suse.cz>
26141 PR libfortran/79956
26142 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
26145 2017-03-21 Brad Spengler <spender@grsecurity.net>
26148 * plugin.c (htab_hash_plugin): New function.
26149 (add_new_plugin): Use it and adjust.
26150 (parse_plugin_arg_opt): Adjust.
26151 (init_one_plugin): Likewise.
26153 2017-03-21 Richard Biener <rguenther@suse.de>
26155 PR tree-optimization/80032
26156 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
26157 if set force the cleanup to happen unconditionally.
26158 (gimplify_target_expr): Push inserted clobbers with force_uncond
26159 to avoid them being removed by control-dependent DCE.
26161 2017-03-21 Richard Biener <rguenther@suse.de>
26163 PR tree-optimization/80122
26164 * tree-inline.c (copy_bb): Do not expans va-arg packs or
26165 va_arg_pack_len when the inlined call stmt requires pack
26167 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
26169 2017-03-21 Jakub Jelinek <jakub@redhat.com>
26172 * tsan.c (instrument_builtin_call): If the memory model argument
26173 is not a constant, assume it is valid.
26176 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
26179 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
26181 PR rtl-optimization/79910
26182 * combine.c (can_combine_p): Do not allow combining an I0 or I1
26183 if its dest is used by an insn before I2 (other than the combined
26184 insns themselves, which are properly handled already).
26186 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
26189 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
26191 * combine.c (record_used_regs): New static function.
26192 (try_combine): Handle situations where there is an additional
26193 instruction between I2 and I3 which needs to have a LOG_LINK
26197 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
26199 * combine.c (try_combine): Delete redundant i1 test. Call
26200 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
26202 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26205 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
26206 alternatives 13/14.
26208 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26210 PR tree-optimization/80054
26211 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
26212 the optimization if a PHI or any of its arguments is not dominated
26213 by the candidate's basis. Use gphi* rather than gimple* as
26215 (replace_profitable_candidates): Clean up a gimple* variable that
26216 should be a gphi* variable.
26218 2017-03-20 Martin Sebor <msebor@redhat.com>
26221 * doc/extend.texi (attribute constructor): Document present limitation.
26223 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
26226 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
26227 __POWER9_VECTOR__ #ifdef control, change template definition to
26228 use Power9-specific built-in function.
26229 (vec_any_eq): Likewise.
26230 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
26231 to control outcomes from this test.
26232 (vector_ae_<mode>p): For VEC_F modes, likewise.
26234 2017-03-20 Ian Lance Taylor <iant@google.com>
26236 * config/i386/i386.c (ix86_function_regparm): Save an extra
26237 register for -fsplit-stack with DECL_STATIC_CHAIN.
26239 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
26242 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
26243 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
26245 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
26247 * config/riscv/riscv.c (riscv_print_operand): Use "fence
26249 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
26252 2017-03-20 Marek Polacek <polacek@redhat.com>
26255 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
26257 2017-03-20 Richard Biener <rguenther@suse.de>
26259 PR tree-optimization/80113
26260 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
26261 allocate extra SSA name for PHI def.
26262 (add_close_phis_to_outer_loops): Likewise.
26263 (add_close_phis_to_merge_points): Likewise.
26264 (copy_loop_close_phi_args): Likewise.
26265 (copy_cond_phi_nodes): Likewise.
26267 2017-03-20 Martin Liska <mliska@suse.cz>
26269 PR middle-end/79753
26270 * tree-chkp.c (chkp_build_returned_bound): Do not build
26271 returned bounds for a LHS that's not a BOUNDED_P type.
26273 2017-03-20 Martin Liska <mliska@suse.cz>
26277 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
26278 COMPLEX_CST and VECTOR_CST.
26280 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26283 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
26284 target operand. A new splitter adds the clobber statement in case
26285 the target operand is dead anyway.
26287 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
26289 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
26290 to age-old versions of binutils and glibc.
26292 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
26294 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
26296 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
26298 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
26300 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
26302 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
26303 requirement for binutils 2.13.
26305 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
26307 * combine.c (try_combine): Delete redundant i1 test. Call
26308 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
26310 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
26312 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
26313 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
26315 <riscv64-*-elf>: Re-arrange section
26316 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
26317 <riscv32-*-linux>: Likewise.
26318 <riscv64-*-elf>: Likewise
26319 <riscv64-*-linux>: Likewise.
26321 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
26324 * aarch64.opt(verbose-cost-dump): Fix typo.
26326 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
26329 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
26330 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
26332 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
26334 * reload.c (find_reloads): When reloading a nonoffsettable address,
26335 use RELOAD_OTHER for it and its address reloads.
26337 PR rtl-optimization/79910
26338 * combine.c (record_used_regs): New static function.
26339 (try_combine): Handle situations where there is an additional
26340 instruction between I2 and I3 which needs to have a LOG_LINK
26343 2017-03-17 Jeff Law <law@redhat.com>
26345 PR tree-optimization/71437
26346 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
26347 conditional in the hash table first.
26348 (vrp_dom_walker::before_dom_children): Extract condition from
26349 ASSERT_EXPR. Record condition, its inverion and any implied
26350 conditions as well.
26352 2017-03-17 Marek Polacek <polacek@redhat.com>
26353 Markus Trippelsdorf <markus@trippelsdorf.de>
26355 PR tree-optimization/80079
26356 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
26359 2017-03-17 Richard Biener <rguenther@suse.de>
26361 PR middle-end/80075
26362 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
26363 Properly verify the LHS before the RHS possibly claims to be
26365 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
26368 2017-03-17 Martin Jambor <mjambor@suse.cz>
26370 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
26371 (List of -O2 options): Likewise.
26372 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
26375 2017-03-17 Tom de Vries <tom@codesourcery.com>
26377 * gcov-dump.c (print_usage): Print bug_report_url.
26379 2017-03-17 Richard Biener <rguenther@suse.de>
26381 PR middle-end/80050
26382 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
26383 (parser::peek): Likewise.
26385 2017-03-17 Richard Biener <rguenther@suse.de>
26387 PR tree-optimization/80048
26388 * sese.c (free_sese_info): Properly release rename_map and
26389 copied_bb_map elements.
26391 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
26393 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
26394 Add linked-list forward and backlinks. Insert on
26395 construction, remove on destruction.
26396 (class pass_store_merging): Add m_stores_head field.
26397 (pass_store_merging::terminate_and_process_all_chains):
26398 Iterate over m_stores_head list.
26399 (pass_store_merging::terminate_all_aliasing_chains):
26401 (pass_store_merging::execute): Check for debug stmts first.
26402 Push new chains onto the m_stores_head stack.
26404 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
26407 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
26408 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
26409 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
26411 2017-03-16 Jeff Law <law@redhat.com>
26413 PR tree-optimization/71437
26414 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
26415 member function. Implementation moved into after_dom_children
26416 member function and into the threader's thread_outgoing_edges
26418 (dom_opt_dom_walker::after_dom_children): Simplify by moving
26419 some code into new thread_outgoing_edges.
26420 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
26421 definition. Simplify marker handling (do it here). Assume we always
26422 have the available expression and the const/copies tables.
26423 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
26425 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
26426 * tree-vrp.c (equiv_stack): No longer file scoped.
26427 (vrp_dom_walker): New class.
26428 (vrp_dom_walker::before_dom_children): New member function.
26429 (vrp_dom_walker::after_dom_children): Likewise.
26430 (identify_jump_threads): Setup domwalker. Use it rather than
26431 walking edges in a random order by hand. Simplify setup/finalization.
26432 (finalize_jump_threads): Remove.
26433 (vrp_finalize): Do not call identify_jump_threads here.
26434 (execute_vrp): Do it here instead and call thread_through_all_blocks
26437 PR tree-optimization/71437
26438 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
26440 (simplify_stmt_for_jump_threading): Add basic_block argument. All
26442 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
26443 (dom_opt_dom_walker::thread_across_edge): Remove
26444 handle_dominating_asserts argument. All callers changed.
26445 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
26446 changes. Remove calls to lhs_of_dominating_assert. Other
26447 uses of handle_dominating_asserts turn into unconditional code
26448 (simplify_control_stmt_condition_1): Likewise.
26449 (simplify_control_stmt_condition): Likewise.
26450 (thread_through_normal_block, thread_across_edge): Likewise.
26451 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
26452 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
26453 object if it is not an SSA_NAME.
26454 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
26455 before calling into the VRP specific simplifiers.
26456 (identify_jump_threads): Remove handle_dominating_asserts
26459 2017-03-16 Jakub Jelinek <jakub@redhat.com>
26462 * tree-diagnostic.c (default_tree_printer): No longer static.
26463 * tree-diagnostic.h (default_tree_printer): New prototype.
26465 2017-03-16 Tamar Christina <tamar.christina@arm.com>
26467 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
26468 Change ins into fmov.
26470 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26472 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
26473 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
26474 Use h_con constraint for operand 1.
26475 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
26476 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
26478 2017-03-15 Jeff Law <law@redhat.com>
26480 PR tree-optimization/71437
26481 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
26482 (record_temporary_equivalences): Use it.
26484 PR tree-optimization/71437
26485 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
26486 tree-ssa-scopedtables.
26487 (lookup_avail_expr, build_and_record_new_cond): Likewise.
26488 (record_conditions, record_cond, vuse_eq): Likewise.
26489 (record_edge_info): Adjust to API tweak of record_conditions.
26490 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
26491 (record_temporary_equivalences, optimize_stmt): Likewise.
26492 (eliminate_redundant_computations): Likewise.
26493 (record_equivalences_from_stmt): Likewise.
26494 * tree-ssa-scopedtables.c: Include options.h and params.h.
26495 (vuse_eq): New function, moved from tree-ssa-dom.c
26496 (build_and_record_new_cond): Likewise.
26497 (record_conditions): Likewise. Accept vector of conditions rather
26498 than edge_equivalence structure for first argument.
26499 for the first argument.
26500 (avail_exprs_stack::lookup_avail_expr): New member function, moved
26501 from tree-ssa-dom.c.
26502 (avail_exprs_stack::record_cond): Likewise.
26503 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
26504 from tree-ssa-dom.c.
26505 (avail_exprs_stack): Add new member functions lookup_avail_expr
26507 (record_conditions): Declare.
26509 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
26512 * lra-constraints.c (process_alt_operands): Increase reject for
26513 reloading an input/output operand.
26515 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
26518 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
26519 insns to convert from signed/unsigned char/short to IEEE 128-bit
26521 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
26523 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
26526 * config/i386/i386.c (ix86_vector_duplicate_value): Create
26527 subreg of inner mode for values already in registers.
26529 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
26531 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
26532 iteration reg is used after the loop.
26534 2017-03-14 Martin Sebor <msebor@redhat.com>
26536 PR tree-optimization/79800
26537 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
26538 precision in negative-positive range.
26539 (format_floating): Call non-const overload with adjusted precision.
26541 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
26544 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
26547 2017-03-14 Martin Sebor <msebor@redhat.com>
26549 PR middle-end/80020
26550 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
26551 * builtins.def (aligned_alloc): Use it.
26554 * Makefile.in (GTFILES): Add calls.c.
26555 * calls.c: Include "gt-calls.h".
26557 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
26559 PR rtl-optimization/79728
26560 * regs.h (struct target_regs): New field
26561 x_contains_allocatable_regs_of_mode.
26562 (contains_allocatable_regs_of_mode): New macro.
26563 * reginfo.c (init_reg_sets_1): Initialize it, and change
26564 contains_reg_of_mode so it includes global regs as well.
26565 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
26566 rather than contains_regs_of_mode.
26568 2017-03-14 Martin Liska <mliska@suse.cz>
26570 * doc/invoke.texi: Document options that can't be combined with
26571 -fcheck-pointer-bounds.
26573 2017-03-14 Martin Liska <mliska@suse.cz>
26575 PR middle-end/79831
26576 * doc/invoke.texi (-Wchkp): Document the option.
26578 2017-03-14 Martin Liska <mliska@suse.cz>
26580 * Makefile.in: Install gcov-dump.
26582 2017-03-14 Martin Liska <mliska@suse.cz>
26584 * multiple_target.c (expand_target_clones): Bail out for
26585 an invalid attribute.
26587 2017-03-14 Richard Biener <rguenther@suse.de>
26589 * alias.c (struct alias_set_entry): Pack properly.
26590 * cfgloop.h (struct loop): Likewise.
26591 * cse.c (struct set): Likewise.
26592 * ipa-utils.c (struct searchc_env): Likewise.
26593 * loop-invariant.c (struct invariant): Likewise.
26594 * lra-remat.c (struct cand): Likewise.
26595 * recog.c (struct change_t): Likewise.
26596 * rtl.h (struct address_info): Likewise.
26597 * symbol-summary.h (function_summary): Likewise.
26598 * tree-loop-distribution.c (struct partition): Likewise.
26599 * tree-object-size.c (struct object_size_info): Likewise.
26600 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
26601 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
26602 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
26603 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
26604 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
26605 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
26606 (struct _stmt_vec_info): Likewise.
26608 2017-03-14 Martin Liska <mliska@suse.cz>
26611 * multiple_target.c (create_dispatcher_calls): Check that
26612 a target can create a function dispatcher.
26614 2017-03-14 Martin Liska <mliska@suse.cz>
26617 * multiple_target.c (expand_target_clones): Drop local.local
26618 flag for default implementation.
26620 2017-03-14 Richard Biener <rguenther@suse.de>
26622 PR tree-optimization/80030
26623 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
26625 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
26627 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
26628 gcc_fallthrough() instead of __attribute__((fallthrough));
26630 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
26632 * doc/gcc.texi: Remove "up" link to (DIR).
26633 * doc/gccint.texi: Ditto.
26635 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
26637 * doc/install.texi (Specific) <avr>: Remove reference to
26640 2017-03-13 Jeff Law <law@redhat.com>
26642 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
26643 attribute rather than comments.
26645 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
26646 match_scratch operand is highest.
26648 2017-03-13 Martin Liska <mliska@suse.cz>
26650 PR middle-end/78339
26651 * ipa-pure-const.c (warn_function_noreturn): If the declarations
26652 is a CHKP clone, use original declaration.
26654 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26656 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
26657 (arc_conditional_register_usage): Use a different allocation order
26658 when optimizing for size.
26659 * common/config/arc/arc-common.c (arc_option_optimization_table):
26660 Section anchors default on when optimizing for size.
26662 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26664 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
26666 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26668 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
26669 * config/arc/arc.md (cpu_facility): Add cd variant.
26670 (*movqi_insn): Add code density variant.
26671 (*movhi_insn): Likewise.
26672 (*movqi_insn): Likewise.
26673 (*addsi3_mixed): Likewise.
26674 (subsi3_insn): Likewise.
26676 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26678 * config/arc/arc.md (movsi_cond_exec): Update constraint.
26680 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26682 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
26683 expressions with MINUS and UNARY ops.
26685 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26688 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
26690 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
26691 between vec_select and vector argument.
26692 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
26693 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
26694 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
26695 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
26696 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
26697 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
26699 2017-03-13 Richard Biener <rguenther@suse.de>
26702 * params.def (vect-max-peeling-for-alignment): Fix typo.
26704 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
26706 * doc/install.texi (Specific) <mips-*-*>: Remove description of
26707 issue that only occurred with binutils below 2.18.
26709 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
26711 * doc/install.texi (Specific) <cris-axis-elf>: No longer
26712 refer to binutils 2.11/2.12 minimum.
26714 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
26716 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
26717 ftp.kernel.org and simplify binutils requirement.
26719 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
26721 * doc/invoke.texi (Warning Options): Fix spelling of link-time
26723 (Optimize Options): Ditto. Also remove redundancy.
26725 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26727 PR translation/79848
26728 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
26730 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
26731 to G_ to avoid double translation.
26733 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26735 PR translation/79923
26736 * auto-profile.c (get_combined_location): Convert leading
26737 character of diagnostics to lower case and remove trailing period.
26738 (read_profile): Likewise for various diagnostics.
26739 * config/arm/arm.c (arm_option_override): Remove trailing period
26740 from various diagnostics.
26741 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
26742 (msp430_expand_delay_cycles): Likewise.
26744 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26747 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
26748 full command-line argument, rather than just "str".
26749 (aarch64_validate_march): Likewise.
26750 (aarch64_validate_mtune): Likewise.
26752 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
26754 PR rtl-optimization/78911
26755 * lra-assigns.c (must_not_spill_p): New function.
26756 (spill_for): Use it.
26758 2017-03-10 Jakub Jelinek <jakub@redhat.com>
26760 PR tree-optimization/79981
26761 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
26762 ATOMIC_COMPARE_EXCHANGE ifn result.
26763 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
26764 IFN_ATOMIC_COMPARE_EXCHANGE.
26766 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26769 * opts.c (parse_sanitizer_options): Add missing question mark to
26770 "did you mean" message.
26772 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26774 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
26776 (VMULEUH_UNS): Likewise.
26777 (VMULOUB_UNS): Likewise.
26778 (VMULOUH_UNS): Likewise.
26779 * config/rs6000/rs6000.c (builtin_function_type): Remove
26780 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
26782 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26785 * read-rtl-function.c (function_reader::read_rtx_operand): Update
26786 x with result of extra_parsing_for_operand_code_0.
26787 (function_reader::extra_parsing_for_operand_code_0): Convert
26788 return type from void to rtx, returning x. When reading
26789 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
26790 larger size containing struct block_symbol.
26792 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
26794 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
26795 -mfloat128-hardware without -m64.
26797 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
26800 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
26801 entries to the case statement that marks unsigned arguments to
26802 overloaded functions.
26804 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26806 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
26807 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
26809 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
26812 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
26813 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
26815 2017-03-10 Martin Liska <mliska@suse.cz>
26819 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
26820 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
26823 2017-03-10 Olivier Hainque <hainque@adacore.com>
26825 * tree-switch-conversion (array_value_type): Start by resetting
26826 candidate type to it's main variant.
26828 2017-03-10 Jakub Jelinek <jakub@redhat.com>
26830 PR rtl-optimization/79909
26831 * combine.c (try_combine): Use simplify_replace_rtx on individual
26832 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
26833 of the whole CALL_INSN_FUNCTION_USAGE.
26835 PR tree-optimization/79972
26836 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
26837 get_range_info on SSA_NAMEs. Formatting fixes.
26839 2017-03-10 Richard Biener <rguenther@suse.de>
26840 Jakub Jelinek <jakub@redhat.com>
26842 PR tree-optimization/77975
26843 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
26844 edge to be constant.
26845 (get_val_for): For constant x return it. Formatting fix.
26846 (loop_niter_by_eval): Avoid pointless looping if the next iteration
26847 would use the same bases as the current one.
26849 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26851 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
26852 instead of vec_select for V1TImode.
26853 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
26855 (VSX_LE_128): Add V1TI to this mode iterator.
26856 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
26857 (*vsx_le_perm_store_<mode>): Likewise.
26858 (pre-reload splitter for VSX stores): Likewise.
26859 (post-reload splitter for VSX stores): Likewise.
26860 (*vsx_xxpermdi2_le_<mode>): Likewise.
26861 (*vsx_lxvd2x2_le_<mode>): Likewise.
26862 (*vsx_stxvd2x2_le_<mode>): Likewise.
26864 2017-03-09 Michael Eager <eager@eagercon.com>
26866 Correct failures with --enable-checking=yes,rtl.
26868 * config/microblaze/microblaze.c (microblaze_expand_shift):
26869 Replace GET_CODE test with CONST_INT_P and INTVAL test with
26870 test for const0_rtx.
26871 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
26872 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
26874 2017-03-09 Richard Biener <rguenther@suse.de>
26876 PR tree-optimization/79977
26877 * graphite-scop-detection.c (scop_detection::merge_sese):
26878 Handle the case of extra exits to blocks dominating the entry.
26880 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
26882 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
26885 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
26887 PR rtl-optimization/79949
26888 * lra-constraints.c (process_alt_operands): Check memory when
26889 trying to predict a cycle. Print about the overall increase.
26891 2017-03-09 Richard Biener <rguenther@suse.de>
26893 PR middle-end/79971
26894 * gimple-expr.c (useless_type_conversion_p): Preserve
26895 TYPE_SATURATING for fixed-point types.
26897 2017-03-09 Richard Biener <rguenther@suse.de>
26900 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
26901 alignment of BLKmode params.
26903 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26906 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
26907 (VALL_NO_V2Q): Likewise.
26909 * config/aarch64/aarch64-simd.md
26910 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
26912 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
26913 VALL_NO_V2Q mode iterator.
26914 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
26916 2017-03-09 Martin Liska <mliska@suse.cz>
26918 PR tree-optimization/79631
26919 * tree-chkp-opt.c (chkp_is_constant_addr): Call
26920 tree_int_cst_sign_bit just for INTEGER constants.
26922 2017-03-09 Martin Liska <mliska@suse.cz>
26926 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
26929 2017-03-09 Marek Polacek <polacek@redhat.com>
26932 * tree.c (inchash::add_expr): Handle TREE_VEC.
26934 2017-03-09 Martin Liska <mliska@suse.cz>
26937 (chkp_narrow_size_and_offset): New function.
26938 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
26939 (void chkp_parse_bit_field_ref): New function.
26940 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
26941 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
26943 2017-03-09 Martin Liska <mliska@suse.cz>
26946 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
26947 (chkp_find_bounds_1): Remove gcc_unreachable.
26949 2017-03-09 Jakub Jelinek <jakub@redhat.com>
26952 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
26953 BUILT_IN_SYNC*, determine the access type from the size suffix and
26954 always build a MEM_REF with that type. Handle forgotten
26955 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
26958 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
26959 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
26960 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
26961 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
26962 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
26963 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
26964 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
26965 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
26966 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
26967 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
26968 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
26969 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
26970 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
26971 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
26972 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
26973 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
26974 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
26975 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
26976 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
26977 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
26978 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
26979 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
26980 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
26981 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
26982 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
26983 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
26984 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
26985 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
26986 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
26987 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
26988 definitions outside of __OPTIMIZE__ guarded section.
26991 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
26992 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
26993 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
26994 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
26997 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26999 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
27000 ("vfenez<mode>"): Add missing constraints.
27002 2017-03-08 Martin Sebor <msebor@redhat.com>
27005 * config/nds32/nds32.c (nds32_option_override):
27006 Fix misspelled diagnostic.
27008 2017-03-08 Jakub Jelinek <jakub@redhat.com>
27011 * gimplify.c (gimplify_omp_for): Replace index var in outer
27012 taskloop statement with an artificial variable and add
27013 OMP_CLAUSE_PRIVATE clause for it.
27015 2017-03-08 Richard Biener <rguenther@suse.de>
27017 PR tree-optimization/79955
27018 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
27019 for accesses that are completely outside of the variable.
27021 2017-03-08 Andrew Haley <aph@redhat.com>
27023 PR tree-optimization/79943
27024 * tree-ssa-loop-split.c (compute_new_first_bound): When
27025 calculating the new upper bound, (END-BEG) should be added, not
27028 2017-03-08 Jakub Jelinek <jakub@redhat.com>
27030 * config/avr/avr.md (setmemhi): Make sure match_dup
27031 operand number comes before match_scratch.
27033 2017-03-08 Richard Biener <rguenther@suse.de>
27035 PR tree-optimization/79920
27036 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
27037 with ncopies == 1 to ...
27038 (vect_transform_slp_perm_load): ... here. Properly compute
27039 all element loads by iterating VF times over the group. Do
27040 not handle ncopies (computed in a broken way) in
27041 vect_create_mask_and_perm.
27043 2017-03-08 Jakub Jelinek <jakub@redhat.com>
27046 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
27047 is a uniform vector, use uniform_vector_p return value instead of
27048 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
27050 2017-03-07 Marek Polacek <polacek@redhat.com>
27052 PR middle-end/79809
27053 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
27054 (alloca_call_type): Likewise.
27056 2017-03-07 Martin Liska <mliska@suse.cz>
27058 * gcov.c (process_args): Put comment to correct location.
27060 2017-03-07 Martin Liska <mliska@suse.cz>
27062 PR middle-end/68270
27063 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
27064 Use array_at_struct_end_p instead of DECL_CHAIN (field).
27065 (chkp_narrow_bounds_for_field): Likewise.
27066 (chkp_parse_array_and_component_ref): Pass one more argument to
27069 2017-03-07 Richard Biener <rguenther@suse.de>
27071 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
27074 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
27076 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
27077 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
27079 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27082 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
27083 to end of description.
27084 (PARAM_MAX_STORES_TO_MERGE): Likewise.
27086 2017-03-07 Jakub Jelinek <jakub@redhat.com>
27088 PR rtl-optimization/79901
27089 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
27091 (*avx512f_<code><mode>3<mask_name>): ... this.
27092 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
27093 iterator instead of VI8_AVX2_AVX512BW.
27095 PR rtl-optimization/79901
27096 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
27097 min/max expander, expand it using expand_vec_cond_expr.
27100 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
27103 2017-03-06 Jakub Jelinek <jakub@redhat.com>
27106 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
27107 to void * for PCH reasons.
27108 * dwarf2out.c (output_loc_operands, output_die): Cast
27109 v.val_vec.array to unsigned char *.
27111 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
27114 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
27117 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
27119 PR rtl-optimization/79571
27120 * lra-constraints.c (process_alt_operands): Calculate static
27121 reject and subtract it from overall when only addresses will be
27124 2017-03-06 Julia Koval <julia.koval@intel.com>
27127 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
27128 incoming stack boundary to 128 for 64-bit targets.
27130 2017-03-06 Richard Biener <rguenther@suse.de>
27132 PR tree-optimization/79894
27133 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
27134 to NULL after folding it.
27136 2017-03-06 Richard Biener <rguenther@suse.de>
27138 PR tree-optimization/79824
27139 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
27140 check disabling peeling for gaps.
27142 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
27144 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
27145 attributes): Document gettimeofday.
27147 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
27149 * config/s390/s390.c (s390_option_override_internal): Set
27150 PARAM_MIN_VECT_LOOP_BOUND
27152 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
27154 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
27155 * config/s390/s390.md: Likewise.
27157 2017-03-06 Jakub Jelinek <jakub@redhat.com>
27160 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
27161 (<avx2_avx512>_perm<mode>): Rename to ...
27162 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
27164 (<avx512>_perm<mode>_mask): Rename to ...
27165 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
27167 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
27168 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
27169 instead of VI8F_256_512.
27170 (avx512f_perm<mode>): New define_expand.
27171 (avx512f_perm<mode>_mask): Likewise.
27172 (avx512f_perm<mode>_1<mask_name>): New define_insn.
27173 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
27175 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
27177 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
27178 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
27180 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
27182 2017-03-06 Martin Liska <mliska@suse.cz>
27185 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
27186 when having a SSA NAME w/o VAR_DECL assigned to it.
27188 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
27190 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
27191 msa_dpsub_<su>_d): Fix MODE for vec_select.
27193 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
27195 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
27197 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
27199 2017-03-06 Richard Biener <rguenther@suse.de>
27201 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
27202 * plugin.c (register_plugin_info): Likewise.
27203 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
27205 2017-03-05 Jakub Jelinek <jakub@redhat.com>
27207 * config/i386/sse.md (sse_storehps, sse_storelps,
27208 avx_<castmode><avxsizesuffix>_<castmode>,
27209 avx512f_<castmode><avxsizesuffix>_<castmode>,
27210 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
27211 in condition that at least one operand is not a MEM.
27213 2017-03-03 Jakub Jelinek <jakub@redhat.com>
27215 PR middle-end/79805
27216 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
27217 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
27219 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
27220 gimple_call_nothrow_p flag based on whether original builtin can throw.
27221 If it can, emit following stmts on the fallthrough edge.
27222 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
27223 don't create new bb if inserting just debug stmts on the edge, try to
27224 insert them on the fallthru bb or just reset debug stmts.
27226 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
27229 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
27230 restore recog_data (including the operand rtxes inside it) around
27231 the call to get_insn_template.
27233 2017-03-03 Martin Sebor <msebor@redhat.com>
27235 PR tree-optimization/79699
27236 * context.c (context::~context): Free MPFR caches to avoid
27237 a memory leak on program exit.
27239 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27241 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
27242 Use wide_int::ulow () instead of .elt (0).
27244 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
27246 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
27247 (*pushxf): Limit oF constraint to 32bit targets and add oC
27248 constraint for 64bit targets.
27249 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
27250 (*pushdf): Change rmF constraint to rmC.
27252 2017-03-03 Martin Liska <mliska@suse.cz>
27254 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
27255 Remove unused variable.
27257 2017-03-03 Jakub Jelinek <jakub@redhat.com>
27260 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
27261 is a memory operand, increase num_memory.
27262 (ix86_expand_args_builtin): Likewise.
27264 2017-03-03 Jan Hubicka <jh@suse.cz>
27267 * ipa-devirt.c (maybe_record_node): Properly handle
27268 __cxa_pure_virtual visibility.
27270 2017-03-03 Martin Liska <mliska@suse.cz>
27272 PR tree-optimization/79803
27273 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
27275 (pass_loop_prefetch::execute): Disabled optimization if an
27276 assumption about L1 cache size is not met.
27278 2017-03-03 Martin Liska <mliska@suse.cz>
27280 PR rtl-optimization/79574
27281 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
27282 (hash_scan_set): Likewise.
27283 (dump_hash_table): Likewise.
27284 (hoist_code): Likewise.
27286 2017-03-03 Richard Biener <rguenther@suse.de>
27288 * fixed-value.c (fixed_from_string): Restore use of elt (1)
27289 in place of uhigh ().
27290 (fixed_convert_from_real): Likewise.
27292 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
27295 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
27297 2017-03-03 Richard Biener <rguenther@suse.de>
27299 PR middle-end/79818
27300 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
27301 TYPE_OVERFLOW_UNDEFINED check.
27303 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27305 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
27307 (vector_ae_<mode>_p): Likewise.
27308 (vector_nez_<mode>_p): Likewise.
27309 (vector_ne_v2di_p): Likewise.
27310 (vector_ae_v2di_p): Likewise.
27311 (vector_ne_<mode>_p): Likewise.
27312 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
27314 (vsx_tsqrt<mode>2_fe): Likewise.
27316 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
27319 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
27321 2017-03-02 Jakub Jelinek <jakub@redhat.com>
27323 PR rtl-optimization/79780
27324 * cprop.c (one_cprop_pass): When second and further conditional trap
27325 in a single basic block is turned into an unconditional trap, turn it
27326 into a deleted note to avoid RTL verification failures.
27328 2017-03-02 Richard Biener <rguenther@suse.de>
27330 * fold-const.c (const_binop): Use ulow () instead of elt (0).
27332 2017-03-02 Richard Biener <rguenther@suse.de>
27334 PR tree-optimization/79345
27336 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
27337 param and abort the walk, returning -1 if it is hit.
27338 (walk_aliased_vdefs): Take a limit param and pass it on.
27339 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
27340 defaulting to 0 and return a signed int.
27341 * tree-ssa-uninit.c (struct check_defs_data): New struct.
27342 (check_defs): New helper.
27343 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
27344 about uninitialized memory.
27345 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
27346 bogus uninitialized warning.
27347 (fixed_convert_from_real): Likewise.
27349 2017-03-02 Bin Cheng <bin.cheng@arm.com>
27351 PR tree-optimization/66768
27352 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
27353 iv_use if base object can't be determined.
27355 2017-03-02 Jakub Jelinek <jakub@redhat.com>
27357 PR tree-optimization/79345
27358 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
27359 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
27360 (get_pattern_stats): Initialize it.
27361 * genemit.c (gen_expand): Verify match_scratch numbers come after
27362 match_operand/match_dup numbers.
27363 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
27364 match_scratch numbers.
27365 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
27367 * config/s390/s390.md (trunctdsd2): Likewise.
27369 2017-03-02 Richard Biener <rguenther@suse.de>
27371 * wide-int.h (wide_int_storage::operator=): Implement in terms
27374 2017-03-02 Richard Biener <rguenther@suse.de>
27376 PR tree-optimization/79777
27377 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
27378 the to insert expression to sth existing.
27380 2017-03-01 Martin Sebor <msebor@redhat.com>
27382 PR middle-end/79692
27383 * gimple-ssa-sprintf.c
27384 (directive::known_width_and_precision): New function.
27385 (format_integer): Use it.
27386 (get_mpfr_format_length): Consider the full range of precision
27387 when computing %g output with the # flag. Set the likely byte
27388 count to 3 rather than 1 when precision is indeterminate.
27389 (format_floating): Correct the lower bound of precision.
27391 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27393 * doc/invoke.texi: Document default code model for 64-bit Linux.
27395 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27398 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
27399 udiv rather than div since input pattern is unsigned.
27401 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
27403 * config/i386/i386.c (print_reg): Warn for values of
27404 unsupported size in integer register.
27406 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
27409 * config/rs6000/predicates.md (current_file_function_operand): Do
27410 not allow self calls to be local if the function is replaceable.
27412 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
27415 * config/rs6000/altivec.h (vec_ctz and others): Change the
27416 preprocessor macro that controls conditional compilation from
27417 _ARCH_PWR9 to __POWER9_VECTOR__.
27418 (vec_all_ne): Change parameterization of __altivec_scalar_pred
27419 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
27420 control (instead of _ARCH_PWR9 control) so that template
27421 definition uses power9-specific function.
27422 (vec_any_eq): Likewise.
27423 (vec_all_ne): Change macro definition to use a power9-specific
27424 expansion under #ifdef __POWER9_VECTOR__ control (instead of
27425 _ARCH_PWR9 control).
27426 (vec_any_eq) Likewise.
27427 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
27428 expansion for CMPNEF to remove support for xvcmpnesp instruction.
27429 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
27430 support for xvcmpnedp instruction.
27431 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
27432 macro expansion so that Power9 implementation of vec_all_ne does
27433 not use the AltiVec predicate framework.
27434 (VCMPNEH_P): Likewise.
27435 (VCMPNEW_P): Likewise.
27436 (VCMPNED_P): Likewise.
27437 (VCMPNEFP_P): Likewise.
27438 (VCMPNEDP_P): Likewise.
27439 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
27440 implementation of vec_any_eq to not use AltiVec predicate
27442 (VCMPAEH_P): Likewise.
27443 (VCMPAEW_P): Likewise.
27444 (VCMPAED_P): Likewise.
27445 (VCMPAEFP_P): Likewise.
27446 (VCMPAEDP_P): Likewise.
27447 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
27448 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
27449 not use the AltiVec predicate framework.
27450 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
27451 of vec_any_eq to not use AltiVec predicate framework.
27452 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
27453 support for predefined __POWER9_VECTOR__ macro to indicate that
27454 Power9 instruction selection is enabled.
27455 (altivec_overloaded_builtins): Remove extraneous
27456 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
27457 function argument types RS6000_BTI_bool_V16QI and
27458 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
27459 entry for overloaded function argument types RS6000_BTI_bool_V4SI
27460 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
27461 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
27462 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
27463 Power9 for implementations of vec_cmpne. Change the signature for
27464 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
27465 (representing vec_all_ne) to remove the previously described first
27466 argument of type RS6000_BTI_INTSI, as this was an artifact of
27467 reliance on the AltiVec predicate framework, which is no longer
27468 used in the implementation of these functions. Add
27469 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
27470 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
27471 since, unlike the AltiVec predicate framework implementation, we
27472 do not share function descriptors between vec_alle and vec_anyeq.
27473 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
27474 set of modes that receive special treatment even when
27475 TARGET_P9_VECTOR is true. The special treatment emits code that
27476 does not depend on Power9 instructions.
27477 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
27478 define_expand to not rely on AltiVec predicate framework.
27479 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
27481 (vector_ne_v2di_p): Change this define_expand to not rely on
27482 AltiVec predicate framework.
27483 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
27485 (vector_ne_<mode>_p): Change this define_expand to not rely on
27486 AltiVec predicate framework.
27487 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
27489 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
27490 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
27491 define_insn pattern.
27492 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
27493 define_insn pattern because the xvcmpne<VSs>. instruction is not
27495 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
27496 instruction is not supported.
27498 2017-03-01 Jakub Jelinek <jakub@redhat.com>
27500 * config/nvptx/nvptx.c: Include intl.h.
27502 2017-03-01 Martin Jambor <mjambor@suse.cz>
27505 * ipa-prop.h (ipa_bits): Removed field known.
27506 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
27507 to pointers. Adjusted their comments to warn about their sharing.
27508 (ipcp_transformation_summary): Change bits to a vector of pointers.
27509 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
27510 (ipa_get_ipa_bits_for_value): Declare.
27511 * tree-vrp.h (value_range): Mark as GTY((for_user)).
27512 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
27513 (ipa_bits_hash_table): Likewise.
27514 (ipa_vr_ggc_hash_traits): Likewise.
27515 (ipa_vr_hash_table): Likewise.
27516 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
27517 being pointers and vr_known being removed.
27518 (ipa_set_jf_unknown): Likewise.
27519 (ipa_get_ipa_bits_for_value): New function.
27520 (ipa_set_jfunc_bits): Likewise.
27521 (ipa_get_value_range): New overloaded functions.
27522 (ipa_set_jfunc_vr): Likewise.
27523 (ipa_compute_jump_functions_for_edge): Use the above functions to
27524 construct bits and vr parts of jump functions.
27525 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
27526 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
27528 (ipcp_grow_transformations_if_necessary): Also allocate
27529 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
27531 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
27532 them. Fix too long lines.
27533 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
27534 vr_known being removed.
27535 (ipa_read_jump_function): Use new setter functions to construct bits
27536 and vr parts of jump functions or set them to NULL.
27537 (write_ipcp_transformation_info): Adjust for bits being pointers.
27538 (read_ipcp_transformation_info): Likewise.
27539 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
27541 Include gt-ipa-prop.h.
27542 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
27544 (ipcp_store_bits_results): Likewise.
27545 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
27546 Do not write to existing jump functions but use a temporary instead.
27548 2017-03-01 Jakub Jelinek <jakub@redhat.com>
27551 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
27552 attempt to use its first operand as BIT_FIELD_REF base.
27554 2017-03-01 Richard Biener <rguenther@suse.de>
27556 PR middle-end/79721
27557 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
27558 interpolating formula in wrapping arithmetic.
27559 (chrec_apply): Convert chrec_evaluate return value to wanted type.
27561 2017-03-01 Jakub Jelinek <jakub@redhat.com>
27563 PR tree-optimization/79734
27564 * tree-vect-generic.c (expand_vector_condition): Optimize
27565 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
27566 Handle VEC_COND_EXPR where comparison has different inner width from
27567 type's inner width.
27569 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
27571 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
27572 markup, and similar issues. Remove @opindex entries for things
27573 that aren't options. Add missing -mmpy-option entries.
27575 2017-02-28 Jakub Jelinek <jakub@redhat.com>
27577 PR tree-optimization/79737
27578 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
27579 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
27580 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
27581 instead of byte_size. Formatting fix.
27582 (shift_bytes_in_array_right): Formatting fix.
27584 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
27587 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
27588 condition on optimize for the leaf function test.
27590 2017-02-28 Martin Liska <mliska@suse.cz>
27593 * read-rtl-function.c (function_reader::handle_unknown_directive):
27594 Bail out when one uses -flto.
27596 2017-02-28 Martin Liska <mliska@suse.cz>
27598 * common.opt: Replace space with tabular for options of <number>
27600 * config/i386/i386.opt: Show <number> value for
27601 -mlarge-data-threshold.
27602 * opts.c (print_filtered_help): Do not display number in hexadecimal
27605 2017-02-28 Martin Liska <mliska@suse.cz>
27607 * common.opt: Fix --help=option -Q for options which are of
27610 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
27612 * config/i386/i386.c (print_reg): Error out for values
27613 of 8-bit size in invalid integer register.
27615 2017-02-28 Martin Sebor <msebor@redhat.com>
27617 PR tree-optimization/79691
27618 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
27620 2017-02-28 Jakub Jelinek <jakub@redhat.com>
27623 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
27624 gcc_unreachable with output_operand_lossage.
27626 2017-02-28 Richard Biener <rguenther@suse.de>
27628 PR tree-optimization/79740
27629 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
27631 (visit_nary_op): Insert the nary into the hashtable if we
27632 pattern-matched sth.
27633 * tree-ssa-pre.c (eliminate_insert): Robustify.
27635 2017-02-28 Richard Biener <rguenther@suse.de>
27637 PR middle-end/79731
27638 * fold-const.c (decode_field_reference): Reject out-of-bound
27641 2017-02-28 Jakub Jelinek <jakub@redhat.com>
27643 * config/i386/i386.c: Include intl.h.
27644 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
27645 instead of just cond ? "..." : "...".
27646 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
27647 * coverage.c (read_counts_file): Likewise.
27648 * omp-offload.c: Include intl.h.
27649 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
27650 of just cond ? "..." : "...".
27651 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
27652 of just cond ? "..." : "...".
27654 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
27657 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
27659 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
27660 'tune for' CPU name.
27661 * config/arm/arm-cpu-data.h: Regenerated.
27663 2017-02-28 Richard Biener <rguenther@suse.de>
27665 PR tree-optimization/79732
27666 * tree-inline.c (expand_call_inline): Do not shadow var.
27668 2017-02-28 Richard Biener <rguenther@suse.de>
27670 PR tree-optimization/79723
27671 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
27672 address-space properly.
27674 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
27676 * doc/optinfo.texi (Optimization groups): Fix option used for
27678 * doc/invoke.texi (-fopt-info): Document "omp".
27679 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
27680 (OPTGROUP_ALL): Add OPTGROUP_OMP.
27681 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
27682 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
27683 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
27685 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
27687 * dumpfile.c (optgroup_options): Instead of "openmp", associate
27688 OPTGROUP_OMP with "omp".
27690 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
27693 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
27694 for arithmetic shift of unsigned V2DI.
27696 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
27698 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
27699 arc/linux.h headers.
27700 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
27701 (LINK_SPEC): Likewise.
27702 (ARC_TLS_EXTRA_START_SPEC): Likewise.
27703 (EXTRA_SPECS): Likewise.
27704 (STARTFILE_SPEC): Likewise.
27705 (ENDFILE_SPEC): Likewise.
27706 (LIB_SPEC): Likewise.
27707 (TARGET_SDATA_DEFAULT): Likewise.
27708 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
27709 (MULTILIB_DEFAULTS): Likewise.
27710 (DWARF2_UNWIND_INFO): Likewise.
27711 * config/arc/big.h: New file.
27712 * config/arc/elf.h: Likewise.
27713 * config/arc/linux.h: Likewise.
27714 * config/arc/t-uClibc: Remove.
27716 2017-02-27 Bin Cheng <bin.cheng@arm.com>
27718 PR tree-optimization/77536
27719 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
27720 (tree_transform_and_unroll_loop): Use above function to compute the
27721 estimated niter of unrolled loop and use it when scaling profile.
27722 Also use count info rather than frequency if it's non-zero.
27723 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
27724 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
27725 (vect_transform_loop): Call above function.
27727 2017-02-27 Richard Biener <rguenther@suse.de>
27729 PR tree-optimization/45397
27730 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
27731 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
27732 (visit_nary_op): Add pattern matching for CSEing sign-changed
27733 or truncated operations with wider ones.
27735 2017-02-27 Richard Biener <rguenther@suse.de>
27737 PR tree-optimization/79690
27738 * tree-vect-stmts.c (vectorizable_store): Use vector type
27739 built from the DR with address-space.
27741 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
27743 * doc/invoke.texi (Optimize Options): Refine the description
27744 of asan-use-after-return.
27746 2017-02-25 Alan Modra <amodra@gmail.com>
27748 PR rtl-optimization/79584
27749 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
27750 base, not ad->base_term, the reg within base. Remove assertion
27751 that ad->base == ad->base_term. Replace gen_int_mode using
27752 bogus mode with const0_rtx.
27754 2017-02-25 Jakub Jelinek <jakub@redhat.com>
27756 PR middle-end/79396
27757 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
27758 FMA_EXPR like tcc_binary or tcc_unary.
27760 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
27763 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
27765 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
27766 (output_loc_operands): Handle DW_OP_call_ref and
27767 DW_OP_GNU_variable_value.
27768 (struct variable_value_struct): New type.
27769 (struct variable_value_hasher): Likewise.
27770 (variable_value_hash): New variable.
27771 (string_types): Remove.
27772 (copy_loc_descr): New function.
27773 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
27774 (prepend_loc_descr_to_each): New function.
27775 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
27776 instead of add_loc_descr_to_each if the first argument is single
27777 location list and the second has multiple.
27778 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
27779 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
27780 when looking for variable value which doesn't have other location info.
27781 (loc_list_from_tree): Formatting fix.
27782 (gen_array_type_die): Simplify DW_AT_string_length handling.
27783 (adjust_string_types): Remove.
27784 (gen_subprogram_die): Don't call adjust_string_types nor test/set
27785 string_types. Call resolve_variable_values.
27786 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
27787 (resolve_addr_in_expr): Likewise. Add A argument.
27788 (copy_deref_exprloc): Remove deref argument. Adjust for the
27789 original expression being DW_OP_GNU_variable_value with optionally
27790 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
27791 optionally after it.
27792 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
27793 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
27794 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
27795 (variable_value_hasher::hash, variable_value_hasher::equal): New
27797 (resolve_variable_value_in_expr, resolve_variable_value,
27798 resolve_variable_values, note_variable_value_in_expr,
27799 note_variable_value): New functions.
27800 (dwarf2out_early_finish): Call note_variable_value on all toplevel
27803 2017-02-24 Jakub Jelinek <jakub@redhat.com>
27806 * opts.h (handle_generated_option): Add GENERATED_P argument.
27807 * opts-common.c (handle_option): Adjust function comment.
27808 (handle_generated_option): Add GENERATED_P argument, pass it to
27810 (control_warning_option): Pass false to handle_generated_option
27812 * opts.c (maybe_default_option): Pass true to handle_generated_option
27814 * optc-gen.awk: Likewise.
27816 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27818 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
27819 a REG, look at the REG it is a SUBREG of.
27820 (splitter for cmpeqsi_t): Ditto.
27822 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27824 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
27825 the special USEs with the pattern of the insn, not the insn itself.
27827 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
27830 * doc/invoke.texi: Document -mload-store-pairs.
27832 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27833 Sandra Loosemore <sandra@codesourcery.com>
27835 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
27836 argument isn't a CONST_INT.
27837 (nios2_alternate_compare_const): Assert op is a CONST_INT.
27838 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
27839 (nios2_validate_compare): Bypass alternate compare logic if *op2
27840 is not a CONST_INT.
27841 (ldstwm_operation_p): Return false if first_base is not a REG or
27842 if first_offset is not a CONST_INT.
27844 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27846 * config/cris/cris.md: Use correct operand in a define_peephole2.
27848 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27850 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
27852 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27854 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
27855 this_insn if it is an INSN or JUMP_INSN.
27856 (force_offsettable): Look at base, not at addr.
27857 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
27858 on things that aren't necessarily CONST_INTs.
27860 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
27862 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
27863 -mfpmath=sse is the default also for x86-32 targets with SSE2
27864 instruction set when @option{-ffast-math} is enabled
27866 2017-02-24 Jeff Law <law@redhat.com>
27868 PR rtl-optimizatoin/79286
27869 * ira.c (update_equiv_regs): Drop may_trap_p exception to
27872 2017-02-24 Richard Biener <rguenther@suse.de>
27874 PR tree-optimization/79389
27875 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
27878 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
27880 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
27881 function comment to reflect reality.
27882 (loop_exits_before_overflow): Fix typo in function description.
27884 2017-02-24 Richard Biener <rguenther@suse.de>
27886 PR tree-optimization/79389
27887 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
27888 properly that a threading opportunity exists. Detect conditional
27889 copy/constant propagation opportunities.
27891 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
27893 * config/visium/visium.md (type): Add trap.
27894 (b): New mode attribute.
27895 (*btst): Rename into...
27896 (*btst<mode>): ...this and adjust.
27897 (*cbranchsi4_btst_insn): Rename into...
27898 (*cbranch<mode>4_btst_insn): ...this and adjust.
27899 (trap): New define_insn.
27901 2017-02-23 Jakub Jelinek <jakub@redhat.com>
27903 PR tree-optimization/79389
27904 * ifcvt.c (struct noce_if_info): Add rev_cond field.
27905 (noce_reversed_cond_code): New function.
27906 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
27907 reversed_comparison_code. Formatting fix.
27908 (noce_try_store_flag): Test rev_cond != NULL in addition to
27909 reversed_comparison_code.
27910 (noce_try_store_flag_constants): Likewise.
27911 (noce_try_store_flag_mask): Likewise.
27912 (noce_try_addcc): Use rev_cond if non-NULL instead of
27913 reversed_comparison_code.
27914 (noce_try_cmove_arith): Likewise. Formatting fixes.
27915 (noce_try_minmax, noce_try_abs): Clear rev_cond.
27916 (noce_find_if_block): Initialize rev_cond.
27917 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
27918 instead of false as last argument never attempt to reverse it
27921 2017-02-23 Bin Cheng <bin.cheng@arm.com>
27923 PR tree-optimization/79663
27924 * tree-predcom.c (combine_chains): Process refs in reverse order
27925 only for ZERO length chains, and add explaining comment.
27927 2017-02-23 Jeff Law <law@redhat.com>
27929 PR tree-optimization/79578
27930 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
27931 in call to operand_equal_p.
27933 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
27936 * config/i386/cpuid.h: Fix another undefined behavior.
27938 2017-02-23 Richard Biener <rguenther@suse.de>
27940 PR tree-optimization/79683
27941 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
27942 vector types for data-refs.
27944 2017-02-23 Martin Liska <mliska@suse.cz>
27946 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
27948 2017-02-23 Jakub Jelinek <jakub@redhat.com>
27950 PR middle-end/79665
27951 * internal-fn.c (get_range_pos_neg): Moved to ...
27952 * tree.c (get_range_pos_neg): ... here. No longer static.
27953 * tree.h (get_range_pos_neg): New prototype.
27954 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
27955 are known to be in between 0 and signed maximum inclusive, try to
27956 expand both unsigned and signed divmod and use the cheaper one from
27959 2017-02-22 Jeff Law <law@redhat.com>
27961 PR tree-optimization/79578
27962 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
27963 to compare base operands.
27965 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
27968 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
27969 gpc_reg_operand instead of fpr_reg_operand.
27971 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
27973 * config/mips/mips.c (mips_return_in_memory): Force FP
27974 vector types to be returned in memory for o32 ABI.
27976 2017-02-22 Jakub Jelinek <jakub@redhat.com>
27978 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
27979 instead of DW_TAG_member for static data member declarations and don't
27980 set no_linkage_name for static inline data members.
27981 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
27984 2017-02-22 Martin Liska <mliska@suse.cz>
27986 * doc/invoke.texi: Replace inequality signs with square brackets
27989 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27991 PR tree-optimization/68644
27992 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
27994 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
27997 * lra-constraints.c (simplify_operand_subreg): Handle
27998 WORD_REGISTER_OPERATIONS targets.
28000 2017-02-22 Jakub Jelinek <jakub@redhat.com>
28003 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
28004 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
28005 elimination by swapping fld*.
28007 2017-02-22 Richard Biener <rguenther@suse.de>
28009 PR tree-optimization/79673
28010 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
28011 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
28012 irrelevant address-space qualifiers and avoiding a
28013 ADDR_SPACE_CONVERT_EXPR from fold_convert.
28015 2017-02-22 Richard Biener <rguenther@suse.de>
28017 PR tree-optimization/79666
28018 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
28019 to not symbolically negate if that may introduce undefined
28022 2017-02-22 Martin Liska <mliska@suse.cz>
28025 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
28026 * data-streamer-out.c (streamer_write_gcov_count_stream):
28028 * value-prof.c (stream_out_histogram_value): Make assert more
28029 precise based on type of counter.
28031 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
28034 * config/i386/i386.md (standard_x87sse_constant_load splitter):
28035 Use nonimmediate_operand instead of memory_operand for operand 1.
28036 (float-extend standard_x87sse_constant_load splitter): Ditto.
28038 2017-02-21 Jeff Law <law@redhat.com>
28040 PR tree-optimization/79621
28041 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
28042 blocks with edges to themselves.
28044 2017-02-21 Jakub Jelinek <jakub@redhat.com>
28047 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
28048 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
28049 Use gimple_call_builtin_p.
28052 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
28053 on temporarily removed DEBUG_INSNs.
28055 PR tree-optimization/79649
28056 * tree-loop-distribution.c (classify_partition): Give up on
28057 non-generic address space loads/stores.
28059 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
28061 * doc/loop.texi (Loop manipulation): Remove nonexistent
28062 tree_ssa_loop_version from the documentation.
28063 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
28065 2017-02-21 Jakub Jelinek <jakub@redhat.com>
28068 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
28069 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
28070 * config/rs6000/rs6000.c: Include except.h.
28071 (rs6000_expand_split_stack_prologue): Call
28072 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
28074 2017-02-21 Martin Jambor <mjambor@suse.cz>
28077 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
28078 have been analyzed.
28080 2017-02-21 Martin Jambor <mjambor@suse.cz>
28082 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
28083 for backward compatibility only.
28084 * doc/invoke.texi (Option Summary): Remove all references to
28085 -fipa-cp-alignment.
28087 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
28091 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28093 * lra-constraints.c (curr_insn_transform): Handle
28094 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
28096 2017-02-21 Martin Liska <mliska@suse.cz>
28098 * config/i386/i386.opt: Replace -masm-dialect with -masm.
28100 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
28102 PR translation/79638
28103 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
28105 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
28108 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
28109 (arm_function_ok_for_sibcall): Return false for an indirect call by
28110 descriptor if all the argument registers are used.
28111 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
28112 alignment of the function.
28114 2017-02-21 Jakub Jelinek <jakub@redhat.com>
28116 PR tree-optimization/61441
28117 * simplify-rtx.c (simplify_const_unary_operation): For
28118 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
28119 the sNaN unmodified.
28121 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
28123 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
28124 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
28125 instead of SYSTEM_HEADER_DIR.
28127 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
28128 Martin Liška <mliska@suse.cz>
28130 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
28131 Fix typos and grammar, use active voice, and clarify.
28133 2017-02-20 Marek Polacek <polacek@redhat.com>
28135 PR middle-end/79537
28136 * gimplify.c (gimplify_expr): Handle unused *&&L;.
28139 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
28141 2017-02-20 Jakub Jelinek <jakub@redhat.com>
28144 * config/i386/i386.c (ix86_expand_builtin): Handle
28145 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
28146 ix86_builtins_isa[fcode].isa as a requirement of those
28147 flags and any other flag in the bitmask.
28148 (ix86_init_mmx_sse_builtins): Use 0 instead of
28149 ~OPTION_MASK_ISA_64BIT as mask.
28150 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
28151 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
28152 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
28153 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
28155 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28158 * lra-constraints.c (split_reg): Check requested split mode
28159 is supported by the register.
28161 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28163 * lra-constraints.c (simplify_operand_subreg): Remove early
28166 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28169 * lra-constraints.c (curr_insn_transform): Tighten condition
28170 for converting SUBREG reloads from OP_OUT to OP_INOUT.
28172 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28175 * lra-constraints.c (curr_insn_transform): Handle
28176 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
28178 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
28181 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
28183 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
28185 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
28188 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
28191 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28193 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
28196 2017-02-18 Jakub Jelinek <jakub@redhat.com>
28198 * final.c (last_columnnum, override_columnnum): New variables.
28199 (final_start_function): Set last_columnnum, pass it to begin_prologue
28200 hook and pass 0 to dwarf2out_begin_prologue.
28201 (final_scan_insn): Update override_columnnum. Pass last_columnnum
28202 to source_line debug hook.
28203 (notice_source_line): Compute last_columnnum and for debug_column_info
28204 return true on column changes.
28205 * debug.h (struct gcc_debug_hooks): Add column argument to
28206 source_line and begin_prologue hooks.
28207 (debug_nothing_int_charstar_int_bool): Remove prototype.
28208 (debug_nothing_int_int_charstar,
28209 debug_nothing_int_int_charstar_int_bool): New prototypes.
28210 (dwarf2out_begin_prologue): Add column argument.
28211 * debug.c (do_nothing_debug_hooks): Adjust source_line and
28212 begin_prologue hooks.
28213 (debug_nothing_int_charstar_int_bool): Remove.
28214 (debug_nothing_int_int_charstar,
28215 debug_nothing_int_int_charstar_int_bool): New functions.
28216 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
28217 through to dwarf2out_source_line.
28218 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
28219 (dwarf2out_source_line): Add column argument, emit it if requested.
28220 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
28222 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
28223 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
28224 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
28225 through to dwarf2out_begin_prologue.
28226 (vmsdbgout_source_line): Add column argument, pass it through to
28227 dwarf2out_source_line.
28228 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
28229 dbxout_source_line caller.
28230 (dbxout_source_line): Add column argument.
28232 * common.opt (gno-column-info, gcolumn-info): New options.
28233 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
28234 (check_die): Also test for multiple DW_AT_decl_column attributes.
28235 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
28236 DW_AT_decl_column if requested.
28237 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
28239 (gen_variable_die): Likewise.
28240 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
28241 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
28244 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
28245 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
28246 (ix86_handle_option): Handle OPT_m3dnowa.
28247 * doc/invoke.texi (-m3dnowa): Document.
28248 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
28249 -m3dnowa instead of -m3dnow -march=athlon.
28252 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
28253 instead of gcc_assert for K, r and R code checks. Formatting fixes.
28255 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28258 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
28259 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
28260 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
28261 generator for vsx_xxpermdi_<mode>_be.
28262 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
28263 force big-endian semantics.
28264 (vsx_xxpermdi_<mode>_be): New define_expand with same
28265 implementation as previous version of vsx_xxpermdi_<mode>.
28267 2017-02-17 Jakub Jelinek <jakub@redhat.com>
28269 PR tree-optimization/79327
28270 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
28271 variable, its initialization and use.
28273 2017-02-17 Julia Koval <julia.koval@intel.com>
28275 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
28276 (OPTION_MASK_ISA_PKU_UNSET): New.
28277 (ix86_handle_option): Handle -mrdpid.
28278 * config/i386/cpuid.h (bit_RDPID): New.
28279 * config/i386/driver-i386.c (host_detect_local_cpu):
28280 Detect RDPID feature.
28281 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
28282 * config/i386/i386-c.c (ix86_target_macros_internal):
28284 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
28285 (ix86_valid_target_attribute_inner_p): Add "rdpid".
28286 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
28287 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
28288 * config/i386/i386.md (define_insn "rdpid"): New.
28289 * config/i386/i386.opt Add -mrdpid.
28290 * config/i386/immintrin.h (_rdpid_u32): New.
28292 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
28294 PR rtl-optimization/79541
28295 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
28296 instead of transforming it into USE.
28298 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
28300 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
28301 If HONOR_SNANS (SFmode) force the input to a register.
28302 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
28303 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
28304 an frsp or similar insn.
28306 2017-02-17 Martin Liska <mliska@suse.cz>
28308 PR rtl-optimization/79577
28309 * params.def (selsched-max-sched-times): Increase minimum to 1.
28311 2017-02-17 Martin Liska <mliska@suse.cz>
28313 PR rtl-optimization/79574
28314 * gcse.c (want_to_gcse_p): Prevent integer overflow.
28316 2017-02-17 Martin Liska <mliska@suse.cz>
28318 PR tree-optimization/79529
28319 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
28320 ssa_defined_default_def_p to handle cases which are implicitly
28322 * tree-ssa.c (ssa_defined_default_def_p): New function.
28323 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
28324 which are implicitly defined.
28325 * tree-ssa.h (ssa_defined_default_def_p): Declare.
28327 2017-02-17 Richard Biener <rguenther@suse.de>
28329 PR middle-end/79576
28330 * params.def (max-ssa-name-query-depth): Limit to 10.
28332 2017-02-17 Richard Biener <rguenther@suse.de>
28334 PR tree-optimization/79552
28335 * tree-ssa-structalias.c (visit_loadstore): Properly verify
28338 2017-02-17 Richard Biener <rguenther@suse.de>
28341 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
28343 2017-02-17 Marek Polacek <polacek@redhat.com>
28345 PR middle-end/79536
28346 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
28347 (fold_negate_expr): New wrapper.
28349 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
28351 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
28352 Correct terminology and de-emphasize pre-standard behavior.
28354 2017-02-16 Alan Modra <amodra@gmail.com>
28356 PR rtl-optimization/79286
28357 * ira.c (def_dominates_uses): New function.
28358 (update_equiv_regs): Don't create an equivalence for insns that
28359 may trap where the register def does not dominate the use.
28361 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
28363 PR rtl-optimization/78127
28364 * lra.c (lra): Call lra_eliminate before finish the loop after
28367 2017-02-16 Richard Biener <rguenther@suse.de>
28369 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
28371 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
28372 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
28373 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
28374 (isl_val_int_from_wi): New function.
28375 (extract_affine_gmp): Rename to ...
28376 (extract_affine_wi): ... this, take a widest_int.
28377 (extract_affine_int): Just wrap extract_affine_wi.
28378 (add_param_constraints): Use isl_val_int_from_wi.
28379 (add_loop_constraints): Likewise, and extract_affine_wi.
28381 2017-02-15 Jeff Law <law@redhat.com>
28383 PR middle-end/79521
28384 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
28385 ira_init_register_move_cost_if_necessary.
28387 2017-02-15 Martin Sebor <msebor@redhat.com>
28389 PR middle-end/32003
28390 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
28391 removed in a prior commit.
28393 2017-02-15 Bin Cheng <bin.cheng@arm.com>
28395 PR tree-optimization/79347
28396 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
28397 counters during peeling.
28399 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
28401 * Makefile.in (site.exp): Remove "set ISLVER".
28403 2017-02-15 Jakub Jelinek <jakub@redhat.com>
28406 * real.c (real_from_integer): Call real_convert even for decimal.
28408 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
28411 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
28413 2017-02-14 Andrew Pinski <apinski@cavium.com>
28415 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
28416 cores and change the partno/implementer to be correct.
28417 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
28418 the 'B" as the implementer.
28419 * config/aarch64/aarch64-tune.md: Regenerate.
28421 2017-02-14 Carl Love <cel@us.ibm.com>
28423 * config/rs6000/rs6000.c: Add case statement entry to make the
28424 xvcvuxdsp built-in argument unsigned.
28425 * config/rs6000/vsx.md: Fix the source and return operand types so they
28426 match the instruction definitions from the ISA document. Fix typo
28427 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
28430 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
28433 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
28434 member early_clobber_alts.
28435 * lra-lives.c (reg_early_clobber_p): New.
28436 (process_bb_lives): Use it.
28437 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
28438 (debug_operand_data): Initialize early_clobber_alts.
28439 (setup_operand_alternative): Set up early_clobber_alts.
28440 (collect_non_operand_hard_regs): Ditto. Pass early clobber
28441 alternatives to new_insn_reg.
28442 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
28444 (lra_update_insn_regno_info): Pass the new arg.
28446 2017-02-14 Jakub Jelinek <jakub@redhat.com>
28448 PR middle-end/79505
28449 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
28450 (new_oacc_loop_raw): Don't clear already cleared fields.
28453 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
28454 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
28455 _mm512_prefetch_i64gather_ps): New inline functions and macros.
28457 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
28460 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
28462 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
28465 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
28466 the extra instruction to the right place to store 128-bit constant
28469 2017-02-14 Martin Sebor <msebor@redhat.com>
28471 PR middle-end/79448
28472 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
28473 warning for strings of unknown length.
28475 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
28477 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
28479 2017-02-14 Jeff Law <law@redhat.com>
28482 * ira-costs.c (scan_one_insn): Initialize register move costs
28483 for pseudos seen in USE/CLOBBER insns.
28485 PR tree-optimization/79095
28486 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
28487 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
28488 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
28489 if the operands are known to be not equal, then the resulting range
28491 (intersect_ranges): If the new range is ~[0,0] and the old range is
28492 wide, then prefer ~[0,0].
28493 * tree-vrp.c (overflow_comparison_p_1): New function.
28494 (overflow_comparison_p): New function.
28495 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
28496 if NAME is used in an overflow test.
28497 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
28498 overflow check that can be expressed as an equality test, then adjust
28499 ops to be that equality test.
28501 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28503 * config/s390/s390-builtin-types.def: Remove flags argument.
28504 * config/s390/s390.c (s390_init_builtins): Likewise.
28506 2017-02-14 Martin Liska <mliska@suse.cz>
28508 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
28509 vector. Fix trailing white spaces.
28511 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
28513 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
28516 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28518 PR rtl-optimization/68664
28519 * config/arm/arm.c (arm_sched_can_speculate_insn):
28520 New function. Declare prototype.
28521 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
28523 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28525 PR rtl-optimization/68664
28526 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
28528 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
28530 2017-02-14 Amit Pawar <amit.pawar@amd.com>
28532 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
28533 max skip bytes for function, loop and jump.
28535 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28537 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
28538 ABS_EXPR for gimple dump.
28540 2017-02-14 Jakub Jelinek <jakub@redhat.com>
28543 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
28545 PR tree-optimization/79408
28546 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
28547 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
28548 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
28549 also if rhs1 is INTEGER_CST.
28551 2017-02-14 Richard Biener <rguenther@suse.de>
28553 PR middle-end/79432
28554 * tree-into-ssa.c (insert_phi_nodes): When the function can
28555 have abnormal edges rewrite SSA names with broken use-def
28556 dominance out of SSA and register them for PHI insertion.
28558 2017-02-13 Martin Sebor <msebor@redhat.com>
28560 PR middle-end/79496
28561 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
28562 clearing info.nowrite flag when snprintf size argument is a range.
28564 2017-02-13 Jakub Jelinek <jakub@redhat.com>
28566 * cprop.c (cprop_jump): Add missing space in string literal.
28567 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
28568 (get_constraint_for_component_ref): Likewise.
28569 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
28570 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
28571 * lra-constraints.c (process_alt_operands): Likewise.
28572 * ipa-inline.c (inline_small_functions): Likewise.
28573 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
28574 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
28575 * trans-mem.c (diagnose_tm_1_op): Likewise.
28576 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
28577 (grid_parallel_clauses_gridifiable): Likewise.
28579 * config/nvptx/mkoffload.c (process): Add space in between
28582 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
28583 "MOD4_SSE_REGS" and "ALL_REGS".
28585 * spellcheck.c (test_data): Add , in between "foo" and "food".
28587 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28590 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
28591 boundary crossing check and subsequent code generation agree.
28593 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28595 * config/aarch64/aarch64.c (has_memory_op): Delete.
28596 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
28599 2017-02-13 Jakub Jelinek <jakub@redhat.com>
28601 PR rtl-optimization/79388
28602 PR rtl-optimization/79450
28603 * combine.c (distribute_notes): When removing TEM_INSN for which
28604 corresponding dest has last value recorded, invalidate that last
28607 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28609 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
28610 of explicit '@'. Add missing assembly comment marker on branch costs
28613 2017-02-13 Nathan Sidwell <nathan@acm.org>
28615 * gengtype-lex.l (<in_struct>): Add '/'.
28617 2017-02-13 Martin Liska <mliska@suse.cz>
28620 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
28622 2017-02-13 Richard Biener <rguenther@suse.de>
28624 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
28626 * configure: Re-generate.
28627 * config.in: Likewise.
28628 * graphite-dependences.c: Simplify as if
28629 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
28630 * graphite-isl-ast-to-gimple.c: Likewise.
28631 * graphite-optimize-isl.c: Likewise.
28632 * graphite-poly.c: Likewise.
28633 * graphite-sese-to-poly.c: Likewise.
28634 * graphite.h: Likewise.
28635 * toplev.c: Include isl/version.h and use isl_version () for
28636 printing the ISL version.
28637 * doc/install.texi: Update ISL requirement.
28639 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
28641 * doc/standards.texi (Standards): Update reference to
28644 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
28646 * doc/extend.texi (Named Address Spaces): sourceware.org now
28648 * doc/install.texi (Binaries): Ditto.
28651 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
28653 * doc/cpp.texi: Replace "stringify"/"stringification" with C
28654 standard terminology "stringize"/"stringizing" throughout.
28655 * doc/cppinternals.texi: Likewise.
28657 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
28659 * doc/extend.texi: Fix some spelling mistakes and typos.
28660 * doc/invoke.texi: Likewise.
28662 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
28665 * params.def (inline-min-speedup) Change from 10 to 8.
28667 2017-02-11 Jakub Jelinek <jakub@redhat.com>
28669 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
28672 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
28675 * ipa-inline-analysis.c (get_minimal_bb): New function.
28676 (record_modified): Use it.
28677 (remap_edge_change_prob): Handle also ancestor functions.
28679 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
28681 * doc/contrib.texi (Contributors): Remove broken link into
28682 the Mauve CVS repository.
28684 2017-02-11 Jakub Jelinek <jakub@redhat.com>
28686 PR middle-end/79454
28687 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
28688 result computation whenever lhs doesn't have vector mode, not
28689 just when it has BLKmode.
28691 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
28693 * doc/makefile.texi (profiledbootstrap): Refer to the
28694 installation instructions only in textual form.
28696 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28699 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
28701 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
28703 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
28704 (Specific): Update mingw-w64 reference.
28706 (Specific): Remove broken link to Renesas RX processor.
28708 2017-02-10 Richard Biener <rguenther@suse.de>
28710 * toplev.c (process_options): Do not mention obsolete graphite
28711 options when printing sorry message about missing graphite support.
28712 Mention -floop-nest-optimize.
28714 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
28716 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
28717 (vtst_p16): Likewise.
28718 (vtstq_p8): Likewise.
28719 (vtstq_p16): Likewise.
28721 (vtstq_p64): Likewise.
28722 * config/arm/arm_neon.h (vgetq_lane_p64): New.
28723 (vset_lane_p64): New.
28724 (vsetq_lane_p64): New.
28726 2017-02-10 Jakub Jelinek <jakub@redhat.com>
28728 PR tree-optimization/79411
28729 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
28730 stmt operands are SSA_NAMEs used in abnormal phis.
28731 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
28734 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
28737 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
28740 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
28742 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
28744 2017-02-09 Jakub Jelinek <jakub@redhat.com>
28746 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
28750 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
28751 not arbitrary TREE_CONSTANT.
28754 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
28755 "omp declare target link" attribute unless is_global_var.
28756 * omp-offload.c (find_link_var_op): Likewise.
28758 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
28759 Chung-Lin Tang <cltang@codesourcery.com>
28761 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
28763 (gimplify_adjust_omp_clauses): Don't delete TILE.
28764 (gimplify_omp_for): Deal with TILE.
28765 * internal-fn.c (expand_GOACC_TILE): New function.
28766 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
28768 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
28769 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
28771 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
28772 avoid DIV for outermost collapse var.
28773 (expand_oacc_for): Insert tile element loop as needed. Adjust.
28774 Remove out of date comments, fix whitespace.
28775 * omp-general.c (omp_extract_for_data): Deal with tiling.
28776 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
28777 adjust OLF_DIM_BASE value.
28778 (struct omp_for_data): Add tiling field.
28779 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
28780 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
28781 for auto loops. Remove default auto determining, moved to
28782 oacc_loop_fixed_partitions.
28783 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
28784 stmts, add e_mask field.
28785 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
28786 (oacc_thread_numbers): Use oacc_dim_call.
28787 (oacc_xform_tile): New.
28788 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
28789 (finish_oacc_loop): Adjust for ifns vector.
28790 (oacc_loop_discover_walk): Append loop abstraction sites to list,
28791 add case for GOACC_TILE fns.
28792 (oacc_loop_xform_loop): Delete.
28793 (oacc_loop_process): Iterate over call list directly, and add
28794 handling for GOACC_TILE fns.
28795 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
28797 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
28798 vector partitioning to outer loops. Assign 2 partitions to loops
28799 when available. Add TILE handling.
28800 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
28801 (execite_oacc_device_lower): Process GOACC_TILE fns,
28802 ignore unknown specs.
28803 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
28804 * tree.c (omp_clause_num_ops): Adjust TILE ops.
28805 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
28807 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
28809 * configure.ac (ACX_BUGURL): Update.
28810 * configure: Regenerate.
28812 2017-02-09 Richard Biener <rguenther@suse.de>
28814 PR tree-optimization/69823
28815 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
28816 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
28818 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
28820 * config/arc/arc-c.def: Add __NPS400__ definition.
28821 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
28822 (TARGET_NPS400): Define.
28824 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
28826 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
28828 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
28829 pointer, arch_info.
28830 (arc_cpu_types): Fill the arch_info field with a pointer into the
28831 arc_arch_types table.
28832 (arc_selected_cpu): Declare.
28833 * config/arc/arc.c (arc_selected_cpu): Make global.
28834 (arc_selected_arch): Delete.
28835 (arc_base_cpu): Delete.
28836 (arc_override_options): Remove references to deleted variables,
28837 update access to arch information.
28838 (ARC_OPT): Update access to arch information.
28839 (ARC_OPTX): Likewise.
28840 * config/arc/arc.h (arc_base_cpu): Remove declaration.
28841 (TARGET_ARC600): Update access to arch information.
28842 (TARGET_ARC601): Likewise.
28843 (TARGET_ARC700): Likewise.
28844 (TARGET_EM): Likewise.
28845 (TARGET_HS): Likewise.
28846 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
28849 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
28852 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
28853 condition/operands for integer GE/LE/GEU/LEU operations.
28855 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
28857 PR translation/79397
28858 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
28861 2017-02-08 Martin Jambor <mjambor@suse.cz>
28864 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
28865 whether allocation happened.
28866 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
28867 nothing was allocated.
28869 2017-02-08 Jakub Jelinek <jakub@redhat.com>
28871 PR tree-optimization/79408
28872 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
28873 constant, but SSA_NAME with a known integer range, use the minimum
28874 of that range instead of op1 to determine if modulo can be replaced
28875 with its first operand.
28877 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28879 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
28881 2017-02-08 Richard Biener <rguenther@suse.de>
28883 PR tree-optimization/71824
28884 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
28885 Check all loops contained in the merged region.
28887 2017-02-07 Andrew Pinski <apinski@cavium.com>
28889 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
28891 2017-02-07 Andrew Pinski <apinski@cavium.com>
28893 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
28894 (thunderxt88): Likewise.
28895 (thunderxt81): Disable LSE and change v8.1 to v8.
28896 (thunderxt83): Likewise.
28898 2017-02-07 Jakub Jelinek <jakub@redhat.com>
28899 Richard Biener <rguenther@suse.de>
28901 PR middle-end/79399
28902 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
28903 type from int to size_t.
28904 * ira-costs.c (struct_costs_size): Change type from int to size_t.
28906 2017-02-07 Jakub Jelinek <jakub@redhat.com>
28908 PR rtl-optimization/79386
28909 * cprop.c (bypass_conditional_jumps): Initialize
28910 bypass_last_basic_block already before splitting bbs after
28911 unconditional traps...
28912 (bypass_conditional_jumps): ... rather than here.
28915 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
28916 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
28917 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
28918 fix -masm=intel patterns.
28920 2017-02-07 Richard Biener <rguenther@suse.de>
28922 PR tree-optimization/79256
28923 PR middle-end/79278
28924 * builtins.c (get_object_alignment_2): Use min_align_of_type
28925 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
28926 and ADJUST_FIELD_ALIGN.
28928 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
28930 * doc/tm.texi: Regenerate.
28931 * stor-layout.c (layout_decl): Adjust.
28932 (update_alignment_for_field): Likewise.
28933 (place_field): Likewise.
28934 (min_align_of_type): Likewise.
28935 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
28936 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
28937 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
28938 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
28939 * config/frv/frv.c (frv_adjust_field_align): Likewise.
28940 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
28941 * config/i386/i386.c (x86_field_alignment): Likewise.
28942 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
28943 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
28944 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
28945 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
28946 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
28947 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
28951 2017-01-30 Richard Biener <rguenther@suse.de>
28953 PR tree-optimization/79256
28954 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
28955 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
28958 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
28960 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
28961 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
28962 builtins to SImode and emit a zero-extend, if necessary.
28964 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
28966 * docs/invoke.texi (RISC-V Options): Alphabetize.
28968 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
28970 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
28973 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
28975 * config/riscv/riscv.c: New file.
28976 * common/config/riscv/riscv-common.c: Likewise.
28977 * config.gcc: Likewise.
28978 * config/riscv/constraints.md: Likewise.
28979 * config/riscv/elf.h: Likewise.
28980 * config/riscv/generic.md: Likewise.
28981 * config/riscv/linux.h: Likewise.
28982 * config/riscv/multilib-generator: Likewise.
28983 * config/riscv/peephole.md: Likewise.
28984 * config/riscv/pic.md: Likewise.
28985 * config/riscv/predicates.md: Likewise.
28986 * config/riscv/riscv-builtins.c: Likewise.
28987 * config/riscv/riscv-c.c: Likewise.
28988 * config/riscv/riscv-ftypes.def: Likewise.
28989 * config/riscv/riscv-modes.def: Likewise.
28990 * config/riscv/riscv-opts.h: Likewise.
28991 * config/riscv/riscv-protos.h: Likewise.
28992 * config/riscv/riscv.h: Likewise.
28993 * config/riscv/riscv.md: Likewise.
28994 * config/riscv/riscv.opt: Likewise.
28995 * config/riscv/sync.md: Likewise.
28996 * config/riscv/t-elf-multilib: Likewise.
28997 * config/riscv/t-linux: Likewise.
28998 * config/riscv/t-linux-multilib: Likewise.
28999 * config/riscv/t-riscv: Likewise.
29000 * configure.ac: Likewise.
29001 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
29002 Waterman as RISC-V maintainers.
29003 * doc/install.texi: Add RISC-V entries.
29004 * doc/invoke.texi: Add RISC-V options section.
29005 * doc/md.texi: Add RISC-V constraints section.
29006 * configure: Regenerated.
29008 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
29011 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
29012 false values to be constant vectors with all 0 or all 1 bits set.
29013 (vcondu<mode><mode>): Likewise.
29014 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
29016 (fpmask_comparison_operator): Update comment.
29017 (vecint_comparison_operator): New predicate.
29018 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
29019 vector conditionals when the true and false values are constant
29020 vectors with all 0 bits or all 1 bits set.
29022 2017-02-06 Martin Sebor <msebor@redhat.com>
29024 PR tree-optimization/79376
29025 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
29027 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
29029 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
29030 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
29031 to simplify split condition.
29033 2017-02-06 Jakub Jelinek <jakub@redhat.com>
29035 * omp-expand.c (oxpand_omp_atomic_fetch_op,
29036 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
29039 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
29041 PR rtl-optimization/68664
29042 * target.def (can_speculate_insn): New hook.
29043 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
29044 * doc/tm.texi: Regenerate.
29045 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
29046 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
29047 (rs6000_sched_can_speculate_insn): New function.
29049 2017-02-06 Jakub Jelinek <jakub@redhat.com>
29051 PR tree-optimization/79284
29052 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
29053 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
29054 vectorizable_mask_load_store, vectorizable_operation,
29055 vect_is_simple_cond, get_same_sized_vectype): Use it instead
29056 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
29057 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
29058 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
29059 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
29060 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
29061 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
29062 is_gimple_assign (stmt). Replace another such test with
29063 is_gimple_assign (stmt).
29065 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
29068 * config/avr/avr.c (rtl-iter.h): Include it.
29069 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
29070 (avr_legitimate_combined_insn): ...and implementation.
29072 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
29074 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
29075 * config/s390/s390.c (s390_const_operand_ok)
29076 (s390_canonicalize_comparison, s390_extract_part)
29077 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
29078 (s390_contiguous_bitmask_p, s390_rtx_costs)
29079 (legitimize_pic_address): Likewise.
29080 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
29081 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
29082 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
29083 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
29084 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
29086 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
29088 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
29089 REGNO($0) == REGNO($1).
29091 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29093 * config/s390/linux.h(SIZE_TYPE): Add comment.
29095 2017-02-06 Julian Brown <julian@codesourcery.com>
29096 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29097 Virendra Pathak <virendra.pathak@broadcom.com>
29099 * config/aarch64/aarch64-cores.def: Change the scheduler
29101 * config/aarch64/aarch64.md: Include thunderx2t99.md.
29102 * config/aarch64/thunderx2t99.md: New file.
29104 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
29106 * doc/standards.texi (Go Language): Update link to language
29109 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
29111 * tree-eh.c (lower_resx): Sanitize profile.
29112 (cleanup_empty_eh_move_lp): Likewise.
29114 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
29117 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
29119 * cfgloopmanip.h (loop_version): Update prototype.
29120 * modulo-sched.c (sms_schedule): Update call of loop_version.
29121 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
29122 * tree-parloops.c (gen_parallel_loop): Likewise.
29123 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
29124 * tree-ssa-loop-split.c (split_loop): Likewise.
29125 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
29126 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
29128 2017-02-05 Martin Liska <mliska@suse.cz>
29131 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
29133 (print_operand_address): Initialize a struct to zero.
29135 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
29137 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
29138 garbage collector only in textual form.
29140 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
29142 * doc/extend.texi (x86 specific memory model extensions for
29143 transactional memory): Simplify a phrase.
29145 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
29148 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
29149 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
29150 (atomic_storedi_1): Likewise.
29152 2017-02-04 Jakub Jelinek <jakub@redhat.com>
29154 PR tree-optimization/79338
29155 * tree-parloops.c (gather_scalar_reductions): Don't call
29156 vect_analyze_loop_form for loop->inner before destroying loop's
29159 2017-02-03 Martin Sebor <msebor@redhat.com>
29161 PR tree-optimization/79327
29162 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
29163 when precision has resulted in leading zeros.
29164 (format_integer): Adjust the likely counter to assume an unknown
29165 argument that may be zero is non-zero.
29167 2017-02-03 Jason Merrill <jason@redhat.com>
29170 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
29171 avoid copying non-taken branch.
29173 2017-02-03 Jakub Jelinek <jakub@redhat.com>
29175 PR tree-optimization/79340
29176 * tree-vect-loop.c (vectorizable_reduction): Release
29177 vec_defs elements after safe_splicing them into other vectors.
29180 PR tree-optimization/79327
29181 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
29182 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
29184 (format_integer): Use wide_int_to_tree instead of build_int_cst
29185 + to_?hwi. If argmin is NULL, just set argmin and argmax to
29186 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
29187 of shortest and longest sequence.
29189 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
29191 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
29192 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
29194 2017-02-03 Walter Lee <walt@tilera.com>
29197 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
29198 after initial stackframe link reg save.
29199 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
29201 2017-02-03 Jakub Jelinek <jakub@redhat.com>
29204 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
29205 wu for stxssp alternative.
29207 2017-02-03 Martin Sebor <msebor@redhat.com>
29209 PR tree-optimization/79352
29210 * gimple-fold.c (get_range_strlen): Add argument.
29211 (get_range_strlen): Change return type to bool.
29212 (get_maxval_strlen): Pass in a dummy argument.
29213 * gimple-fold.h (get_range_strlen): Change return type to bool.
29214 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
29215 * tree.h (array_at_struct_end_p): Add argument.
29216 * tree.c (array_at_struct_end_p): Handle it.
29218 2017-02-03 Martin Liska <mliska@suse.cz>
29221 * multiple_target.c (create_dispatcher_calls): Redirect edge
29222 from a caller of a dispatcher.
29223 (expand_target_clones): Make the clones local.
29224 (ipa_target_clone): Do both target clones and resolvers.
29225 (ipa_dispatcher_calls): Remove the pass.
29226 (pass_dispatcher_calls::gate): Likewise.
29227 (make_pass_dispatcher_calls): Likewise.
29228 * passes.def (pass_target_clone): Put as very first IPA early
29231 2017-02-03 Martin Liska <mliska@suse.cz>
29233 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
29234 in case of a function with ifunc attribute.
29236 2017-02-03 Martin Liska <mliska@suse.cz>
29238 * cgraph.c (cgraph_node::dump): Dump function version info.
29239 * symtab.c (symtab_node::dump_base): Add missing new line.
29241 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
29243 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
29244 (ifcombine_ifandif): Use it.
29246 2017-02-03 Martin Liska <mliska@suse.cz>
29248 * doc/invoke.texi: Document default value for
29249 use-after-scope-direct-emission-threshold.
29251 2017-02-03 Martin Liska <mliska@suse.cz>
29253 PR tree-optimization/79339
29254 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
29255 (format_floating): Likewise.
29257 2017-02-03 Martin Liska <mliska@suse.cz>
29260 * ipa-prop.c (ipa_node_params_t::insert): Remove current
29262 (ipa_node_params_t::remove): Likewise.
29263 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
29264 initialization from removed ipa_node_params_t::insert.
29265 (ipa_node_params::~ipa_node_params): Move from removed
29266 ipa_node_params_t::release.
29267 * symbol-summary.h (symbol_summary::m_released): New member.
29268 Do not release a summary twice. Do not allow to call finalizer
29269 for types of a summary that live in GGC memory.
29271 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29273 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
29276 2017-02-02 Martin Sebor <msebor@redhat.com>
29278 PR middle-end/79275
29279 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
29280 (format_string): Tighten up the range of output for non-constant
29281 strings and correct the expected range for wide non-constant strings.
29283 2017-02-02 Martin Sebor <msebor@redhat.com>
29285 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
29287 PR middle-end/32003
29288 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
29290 (-fdump-tree-@var): Add to index and document how to come up
29291 with pass-specific option and dump file names.
29292 (-fdump-passes): Clarify where to look for output.
29294 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
29296 PR middle-end/77445
29297 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
29298 statistics of the analyzed path; allow threading for speed when
29299 any of BBs along the path are optimized for speed.
29301 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
29303 PR middle-end/78468
29304 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
29305 settings of the virtual registers.
29308 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
29310 * explow.c (get_dynamic_stack_size): Take known alignment of stack
29311 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
29314 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29316 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
29317 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
29319 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29321 * config/s390/s390.md: Add missing comments with the expanded
29323 * config/s390/vector.md: Likewise.
29324 * config/s390/vx-builtins.md: Likewise.
29326 2017-02-02 Jakub Jelinek <jakub@redhat.com>
29329 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
29330 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
29331 conditions on a single line.
29333 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29335 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
29336 __S390_VX__ to __VX__.
29338 2017-02-01 Andrew Pinski <apinski@cavium.com>
29340 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
29341 stmt_info to record_stmt_cost.
29342 (vect_get_known_peeling_cost): Pass stmt_info if known to
29344 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
29345 cpu_vector_cost field into
29346 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
29347 field into vec_int_stmt_cost and vec_fp_stmt_cost.
29348 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
29349 splitting of scalar_stmt_cost and vec_stmt_cost.
29350 (thunderx_vector_cost): Likewise.
29351 (cortexa57_vector_cost): LIkewise.
29352 (exynosm1_vector_cost): Likewise.
29353 (xgene1_vector_cost): Likewise.
29354 (thunderx2t99_vector_cost): Improve after the splitting of the two
29356 (aarch64_builtin_vectorization_cost): Update for the splitting of
29357 scalar_stmt_cost and vec_stmt_cost.
29359 2017-02-01 Torvald Riegel <triegel@redhat.com>
29360 Richard Henderson <rth@redhat.com>
29362 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
29363 conditional on existance of a fast atomic load.
29364 * optabs-query.c (can_atomic_load_p): New function.
29365 * optabs-query.h (can_atomic_load_p): Declare it.
29366 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
29367 no fast atomic load is available for the particular size of access.
29368 (expand_atomic_compare_and_swap): Likewise.
29369 (expand_atomic_load): Likewise.
29370 (expand_atomic_store): Likewise.
29371 (expand_atomic_fetch_op): Likewise.
29372 * testsuite/lib/target-supports.exp
29373 (check_effective_target_sync_int_128): Remove x86 because it provides
29374 no fast atomic load.
29375 (check_effective_target_sync_int_128_runtime): Likewise.
29377 2017-02-01 Richard Biener <rguenther@suse.de>
29379 * graphite.c: Include tree-vectorizer.h for find_loop_location.
29380 (graphite_transform_loops): Provide opt-info for optimized nests.
29381 * tree-parloop.c (parallelize_loops): Provide opt-info for
29382 parallelized loops.
29384 2017-02-01 Richard Biener <rguenther@suse.de>
29386 PR middle-end/79315
29387 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
29388 was not set before.
29390 2017-02-01 Richard Biener <rguenther@suse.de>
29392 PR tree-optimization/71824
29393 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
29394 Verify the loops are valid in the merged SESE region.
29395 (scop_detection::can_represent_loop_1): Check analyzing the
29396 evolution of the number of iterations in the region succeeds.
29398 2017-01-31 Ian Lance Taylor <iant@golang.org>
29400 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
29401 REG_ARGS_SIZE note to 32-bit push insns and call insn.
29403 2017-01-31 David Malcolm <dmalcolm@redhat.com>
29405 PR preprocessor/79210
29406 * input.c (get_substring_ranges_for_loc): Replace line_width
29407 assertion with error-handling.
29409 2017-01-31 Richard Biener <rguenther@suse.de>
29411 PR tree-optimization/77318
29412 * graphite-sese-to-poly.c (extract_affine): Fix assert.
29413 (create_pw_aff_from_tree): Take loop parameter.
29414 (add_condition_to_pbb): Pass loop of the condition to
29415 create_pw_aff_from_tree.
29417 2017-01-31 Jakub Jelinek <jakub@redhat.com>
29419 * config/s390/s390.c (s390_asan_shadow_offset): New function.
29420 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
29422 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
29426 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
29428 (convert_int_to_float128): Likewise.
29429 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
29430 (convert_int_to_float128): Likewise.
29431 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
29432 (UNSPEC_IEEE128_CONVERT): Likewise.
29433 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
29434 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
29435 Use local variables for IBM extended format.
29436 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
29437 (fix_trunc<mode>si2_fprs): Likewise.
29438 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
29439 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
29440 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
29441 to know that we can now have integers of all sizes in vector
29443 (fix<uns>_<mode>di2_hw): Likewise.
29444 (float<uns>_<mode>si2_hw): Likewise.
29445 (fix_<mode>si2_hw): Likewise.
29446 (fixuns_<mode>si2_hw): Likewise.
29447 (float<uns>_<mode>di2_hw): Likewise.
29448 (float_<mode>di2_hw): Likewise.
29449 (float_<mode>si2_hw): Likewise.
29450 (floatuns_<mode>di2_hw): Likewise.
29451 (floatuns_<mode>si2_hw): Likewise.
29452 (xscvqp<su>wz_<mode>): Delete, no longer used.
29453 (xscvqp<su>dz_<mode>): Likewise.
29454 (xscv<su>dqp_<mode>): Likewise.
29455 (ieee128_mfvsrd_64bit): Likewise.
29456 (ieee128_mfvsrd_32bit): Likewise.
29457 (ieee128_mfvsrwz): Likewise.
29458 (ieee128_mtvsrw): Likewise.
29459 (ieee128_mtvsrd_64bit): Likewise.
29460 (ieee128_mtvsrd_32bit): Likewise.
29462 2017-01-31 Martin Liska <mliska@suse.cz>
29465 * ipa-prop.c (ipa_free_all_node_params): Call release method
29466 instead of ~sumbol_summary to not to trigger double times
29469 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
29471 PR tree-optimization/71691
29472 * bitmap.h (class auto_bitmap): New.
29473 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
29474 is_maybe_undefined instead of ssa_undefined_value_p.
29476 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29478 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
29479 __S390_ARCH_LEVEL__ to __ARCH__.
29481 2017-01-31 Jakub Jelinek <jakub@redhat.com>
29483 PR tree-optimization/79267
29484 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
29485 if should_remove_lhs_p is true.
29487 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
29490 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
29491 (add_alignment_attribute): New.
29492 (base_type_die): Add alignment attribute.
29493 (subrange_type_die): Likewise.
29494 (modified_type_die): Likewise.
29495 (gen_array_type_die): Likewise.
29496 (gen_descr_array_type_die: Likewise.
29497 (gen_enumeration_type_die): Likewise.
29498 (gen_subprogram_die): Likewise.
29499 (gen_variable_die): Likewise.
29500 (gen_field_die): Likewise.
29501 (gen_ptr_to_mbr_type_die): Likewise.
29502 (gen_struct_or_union_type_die): Likewise.
29503 (gen_subroutine_type_die): Likewise.
29504 (gen_typedef_die): Likewise.
29505 (base_type_cmp): Compare alignment attribute.
29507 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29510 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
29511 (setb_unsigned) New pattern for setb with CCUNS.
29512 * config/rs6000/rs6000.c (expand_block_compare): Use a different
29513 subfc./subfe sequence to avoid overflow problems. Generate a
29514 shorter sequence with cmpld/setb for power9.
29515 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
29516 for generating subfc. instruction.
29517 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
29518 now uses this instruction.
29520 2017-01-30 Ian Lance Taylor <iant@google.com>
29523 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
29524 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
29526 2017-01-30 Martin Sebor <msebor@redhat.com>
29528 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
29529 Move constant to the right of a relational operator.
29530 (get_mpfr_format_length, format_character, format_string): Ditto.
29531 (should_warn_p, maybe_warn): Same.
29533 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
29535 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
29538 * asan.c (get_translation_unit_decl): Remove function.
29539 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
29541 2017-01-30 Martin Liska <mliska@suse.cz>
29543 PR gcov-profile/79259
29544 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
29545 -fprofile-generate.
29547 2017-01-30 Martin Liska <mliska@suse.cz>
29550 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
29551 Initialize variables with NULL value.
29553 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
29556 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
29558 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
29560 2017-01-30 Richard Biener <rguenther@suse.de>
29562 PR tree-optimization/79276
29563 * tree-vrp.c (process_assert_insertions): Properly adjust common
29564 when removing a duplicate.
29566 * gcc.dg/torture/pr79276.c: New testcase.
29568 2017-01-30 Richard Biener <rguenther@suse.de>
29570 PR tree-optimization/79256
29571 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
29572 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
29574 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
29576 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
29579 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
29580 ("*r<noxa>sbg_<mode>_sll_bitmask")
29581 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
29582 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
29583 Use contiguous_bitmask_nowrap_operand.
29585 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29588 * config/rs6000/altivec.h (vec_xl): Revise #define.
29589 (vec_xst): Likewise.
29591 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
29593 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
29595 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
29597 PR rtl-optimization/79194
29598 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
29599 traps before call to bypass_conditional_jumps.
29601 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
29603 PR tree-optimization/71374
29604 * lra-constraints.c (check_conflict_input_operands): New.
29605 (match_reload): Use it.
29607 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
29610 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
29611 account to calculate conflict_set.
29613 2017-01-27 Bin Cheng <bin.cheng@arm.com>
29615 PR rtl-optimization/78559
29616 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
29617 other_insn in combine.
29619 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
29621 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
29622 uint16_type_node for BT_UINT16.
29624 2017-01-27 David Malcolm <dmalcolm@redhat.com>
29626 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
29627 "RTL Tests" to menu.
29628 (GIMPLE Tests): New node.
29629 (RTL Tests): New node.
29631 2017-01-27 Richard Biener <rguenther@suse.de>
29633 PR tree-optimization/79245
29634 * tree-loop-distribution.c (distribute_loop): Apply cost
29635 modeling also to detected patterns.
29637 2017-01-27 Richard Biener <rguenther@suse.de>
29639 PR tree-optimization/71433
29640 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
29641 (compare_assert_loc): New function.
29642 (process_assert_insertions): Sort and optimize assert locations
29643 to remove duplicates and push down identical assertions on
29644 edges to their destination block.
29646 2017-01-27 Richard Biener <rguenther@suse.de>
29648 PR tree-optimization/79244
29649 * tree-vrp.c (remove_range_assertions): Forcefully propagate
29650 out SSA names even if abnormal.
29652 2017-01-27 Jakub Jelinek <jakub@redhat.com>
29654 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
29655 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
29656 instead of MPFR_RNDN.
29658 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
29661 * arm.c (arm_option_override): Don't call build_target_option_node
29662 until after doing all option overrides.
29663 (arm_valid_target_attribute_tree): Likewise.
29665 2017-01-27 Martin Liska <mliska@suse.cz>
29667 * doc/invoke.texi (-fprofile-arcs): Document profiling support
29668 for {cd}tors and C++ {cd}tors.
29670 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
29672 * config/s390/s390.md ("*setmem_long_and")
29673 ("*setmem_long_and_31z"): Use zero_extend instead of and.
29675 2017-01-26 Martin Sebor <msebor@redhat.com>
29677 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
29680 2017-01-26 Martin Sebor <msebor@redhat.com>
29682 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
29683 HAVE_DFmode before using XFmode or DFmode.
29684 (parse_directive): Avoid using the z length modifier to avoid
29685 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
29687 PR middle-end/78703
29688 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
29689 to accept adjustment as an array.
29690 (get_int_range): New function.
29691 (struct directive): Make width and prec arrays.
29692 (directive::set_width, directive::set_precision): Call get_int_range.
29693 (format_integer, format_floating): Handle width and precision ranges.
29694 (format_string, parse_directive): Same.
29696 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29699 * dwarf2out.c (generate_skeleton_bottom_up): For children with
29700 comdat_type_p set, just clone them, but keep the children in the
29704 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
29705 which have direct callers with -fvar-tracking-assignments enabled
29707 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
29708 inside of type units.
29710 2017-01-26 Martin Sebor <msebor@redhat.com>
29712 PR middle-end/78703
29713 * gimple-ssa-sprintf.c (struct result_range): Add likely and
29715 (struct format_result): Replace number_chars, number_chars_min,
29716 and number_chars_max with a single member of struct result_range.
29718 (format_result::operator+=): Adjust.
29719 (struct fmtresult): Remove bounded. Handle likely and unlikely
29721 (fmtresult::adjust_for_width_or_precision): New function.
29722 (fmtresult:type_max_digits): New function.
29723 (bytes_remaining): Handle likely and unlikely counters.
29724 (min_bytes_remaining): Remove.
29725 (format_percent): Simplify.
29726 (format_integer, format_floating): Set likely and unlikely counters.
29727 (get_string_length, format_character, format_string): Same.
29728 (format_plain, should_warn_p): New function.
29729 (maybe_warn): Call should_warn_p. Update diagnostic messages
29730 and handle those for all directives, including plain strings.
29731 (format_directive): Handle likely and unlikely counters.
29732 Remove unnecessary quoting from diagnostics. Add an informational
29734 (add_bytes): Remove.
29735 (pass_sprintf_length::compute_format_length): Simplify.
29736 (try_substitute_return_value): Handle likely and unlikely counters.
29738 2017-01-26 Carl Love <cel@us.ibm.com>
29740 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
29741 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
29743 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
29746 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
29747 endianess for subregs into account.
29748 * lra-constraints.c (lra_constraints): Do risky transformations
29749 always on the first iteration.
29750 * lra-lives.c (check_pseudos_live_through_calls): Add arg
29751 last_call_used_reg_set.
29752 (process_bb_lives): Define and use last_call_used_reg_set.
29753 * lra.c (lra): Always continue after lra_constraints on the first
29756 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
29758 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
29760 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
29762 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29764 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
29765 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
29766 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
29767 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
29768 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
29769 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
29770 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
29771 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
29772 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
29774 2017-01-26 Marek Polacek <polacek@redhat.com>
29777 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
29778 for the third operand.
29780 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29782 PR middle-end/79236
29783 * omp-low.c (struct omp_context): Add simt_stmt field.
29784 (scan_omp_for): Return omp_context *.
29785 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
29786 context to the _simt_ SIMD stmt.
29787 (lower_omp_for): For combined SIMD with sibling _simt_
29788 SIMD, make sure to use the same decls in _looptemp_
29789 clauses as in the sibling.
29791 2017-01-26 David Sherwood <david.sherwood@arm.com>
29793 PR middle-end/79212
29794 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
29797 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29800 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
29801 emit fld b; fld a; if possible.
29803 * brig-builtins.def: Update copyright years.
29804 * config/arm/arm_acle_builtins.def: Update copyright years.
29806 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
29809 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
29810 constraint instead of o for the stxsd instruction.
29812 2017-01-25 Carl Love <cel@us.ibm.com>
29814 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
29815 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
29817 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
29819 * doc/invoke.texi (C++ Dialect Options): Fix typo.
29821 2017-01-25 Richard Biener <rguenther@suse.de>
29823 PR tree-optimization/69264
29824 * target.def (vector_alignment_reachable): Improve documentation.
29825 * doc/tm.texi: Regenerate.
29826 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
29828 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
29829 earlier changes with respect to TYPE_USER_ALIGN.
29830 (vector_alignment_reachable_p): Likewise. Improve dumping.
29832 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29835 * config/arm/arm.md (xordi3): Force constant operand into a register
29838 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29840 * doc/invoke.texi (-fstore-merging): Correct default optimization
29841 levels at which it is enabled.
29842 (-O): Move -fstore-merging from list to...
29845 2017-01-25 Richard Biener <rguenther@suse.de>
29848 * omp-expand.c: Include debug.h.
29849 (expand_omp_taskreg): Make sure to generate early debug before
29850 outlining anything from a function.
29851 (expand_omp_target): Likewise.
29852 (grid_expand_target_grid_body): Likewise.
29854 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
29857 * asan.c (get_translation_unit_decl): New function.
29858 (asan_add_global): Extract modules file name from globals
29859 TRANSLATION_UNIT_DECL name.
29861 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
29864 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
29865 for long calls with APCS frame and VFP.
29867 2017-01-24 David Malcolm <dmalcolm@redhat.com>
29869 * cfg.c (original_copy_tables_initialized_p): New function.
29870 * cfg.h (original_copy_tables_initialized_p): New decl.
29871 * cfgrtl.c (relink_block_chain): Guard the call to
29872 free_original_copy_tables with a call to
29873 original_copy_tables_initialized_p.
29874 * cgraph.h (symtab_node::native_rtl_p): New decl.
29875 * cgraphunit.c (symtab_node::native_rtl_p): New function.
29876 (symtab_node::needed_p): Don't assert for early assembly output
29877 for __RTL functions.
29878 (cgraph_node::finalize_function): Set "force_output" for __RTL
29880 (cgraph_node::analyze): Bail out early for __RTL functions.
29881 (analyze_functions): Update assertion to support __RTL functions.
29882 (cgraph_node::expand): Bail out early for __RTL functions.
29883 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
29885 * function.h (struct function): Update comment for field
29887 * gimple-expr.c: Include "tree-pass.h".
29888 (gimple_has_body_p): Return false for __RTL functions.
29889 * Makefile.in (OBJS): Add run-rtl-passes.o.
29890 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
29892 (gcc::pass_manager::get_clean_slate): New accessor.
29893 * passes.c: Include "insn-addr.h".
29894 (should_skip_pass_p): Add logging. Update logic for running
29895 "expand" to be compatible with both __GIMPLE and __RTL. Guard
29896 property-provider override so it is only done for gimple passes.
29898 (skip_pass): New function.
29899 (execute_one_pass): Call skip_pass when skipping passes.
29900 * read-md.c (md_reader::read_char): Support filtering
29901 the input to a subset of line numbers.
29902 (md_reader::md_reader): Initialize fields
29903 m_first_line and m_last_line.
29904 (md_reader::read_file_fragment): New function.
29905 * read-md.h (md_reader::read_file_fragment): New decl.
29906 (md_reader::m_first_line): New field.
29907 (md_reader::m_last_line): New field.
29908 * read-rtl-function.c (function_reader::create_function): Only
29909 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
29910 curr_properties. Set DECL_INITIAL to a dummy block.
29911 (read_rtl_function_body_from_file_range): New function.
29912 * read-rtl-function.h (read_rtl_function_body_from_file_range):
29914 * run-rtl-passes.c: New file.
29915 * run-rtl-passes.h: New file.
29917 2017-01-24 Jeff Law <law@redhat.com>
29919 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
29922 2017-01-24 Bin Cheng <bin.cheng@arm.com>
29924 PR tree-optimization/79159
29925 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
29926 (record_nonwrapping_iv): Improve boundary using above function if no
29927 value range information.
29929 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
29930 Martin Jambor <mjambor@suse.cz>
29932 * brig-builtins.def: New file.
29933 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
29934 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
29935 (DEF_HSAIL_SAT_BUILTIN): Likewise.
29936 (DEF_HSAIL_INTR_BUILTIN): Likewise.
29937 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
29938 * builtin-types.def (BT_INT8): New.
29939 (BT_INT16): Likewise.
29940 (BT_UINT8): Likewise.
29941 (BT_UINT16): Likewise.
29942 (BT_FN_ULONG): Likewise.
29943 (BT_FN_UINT_INT): Likewise.
29944 (BT_FN_UINT_ULONG): Likewise.
29945 (BT_FN_UINT_LONG): Likewise.
29946 (BT_FN_UINT_PTR): Likewise.
29947 (BT_FN_ULONG_PTR): Likewise.
29948 (BT_FN_INT8_FLOAT): Likewise.
29949 (BT_FN_INT16_FLOAT): Likewise.
29950 (BT_FN_UINT32_FLOAT): Likewise.
29951 (BT_FN_UINT16_FLOAT): Likewise.
29952 (BT_FN_UINT8_FLOAT): Likewise.
29953 (BT_FN_UINT64_FLOAT): Likewise.
29954 (BT_FN_UINT16_UINT32): Likewise.
29955 (BT_FN_UINT32_UINT16): Likewise.
29956 (BT_FN_UINT16_UINT16_UINT16): Likewise.
29957 (BT_FN_INT_PTR_INT): Likewise.
29958 (BT_FN_UINT_PTR_UINT): Likewise.
29959 (BT_FN_LONG_PTR_LONG): Likewise.
29960 (BT_FN_ULONG_PTR_ULONG): Likewise.
29961 (BT_FN_VOID_UINT64_UINT64): Likewise.
29962 (BT_FN_UINT8_UINT8_UINT8): Likewise.
29963 (BT_FN_INT8_INT8_INT8): Likewise.
29964 (BT_FN_INT16_INT16_INT16): Likewise.
29965 (BT_FN_INT_INT_INT): Likewise.
29966 (BT_FN_UINT_FLOAT_UINT): Likewise.
29967 (BT_FN_FLOAT_UINT_UINT): Likewise.
29968 (BT_FN_ULONG_UINT_UINT): Likewise.
29969 (BT_FN_ULONG_UINT_PTR): Likewise.
29970 (BT_FN_ULONG_ULONG_ULONG): Likewise.
29971 (BT_FN_UINT_UINT_UINT): Likewise.
29972 (BT_FN_VOID_UINT_PTR): Likewise.
29973 (BT_FN_UINT_UINT_PTR: Likewise.
29974 (BT_FN_UINT32_UINT64_PTR): Likewise.
29975 (BT_FN_INT_INT_UINT_UINT): Likewise.
29976 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
29977 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
29978 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
29979 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
29980 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
29981 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
29982 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
29983 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
29984 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
29985 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
29986 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
29987 * doc/frontends.texi: List BRIG FE.
29988 * doc/install.texi (Testing): Add BRIG tesring requirements.
29989 * doc/invoke.texi (Overall Options): Mention BRIG.
29990 * doc/standards.texi (Standards): Doucment BRIG HSA version.
29992 2017-01-24 Richard Biener <rguenther@suse.de>
29994 PR translation/79208
29995 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
29997 2017-01-24 Martin Jambor <mjambor@suse.cz>
30000 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
30001 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
30002 and known_contexts.
30004 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
30006 PR middle-end/79123
30007 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
30008 casts from signed to unsigned really don't have a range.
30010 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
30012 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
30013 GMP_RNDx for compatiblity.
30015 2017-01-24 Martin Liska <mliska@suse.cz>
30018 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
30019 that would prevent us to call alloca with -1 as argument.
30021 2017-01-24 Jakub Jelinek <jakub@redhat.com>
30023 * dwarf2out.c (output_compilation_unit_header, output_file_names):
30024 Avoid -Wformat-security warning.
30026 2017-01-23 Andrew Pinski <apinski@cavium.com>
30028 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
30031 2017-01-23 Martin Sebor <msebor@redhat.com>
30033 PR middle-end/78703
30034 * gimple-ssa-sprintf.c (warn_level): New global.
30035 (format_integer): Use it here and throughout the rest of the file.
30036 Use the same switch to compute sign as base.
30037 (maybe_warn): New function.
30038 (format_directive): Factor out warnings into maybe_warn.
30039 Add debugging output. Use warn_level.
30040 (add_bytes): Use warn_level.
30041 (pass_sprintf_length::compute_format_length): Add debugging output.
30042 (try_substitute_return_value): Same.
30043 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
30045 PR middle-end/78703
30046 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
30047 (struct fmtresult, format_integer, format_floating): Adjust.
30048 (fmtresult::fmtresult): Set max correctly in two argument ctor.
30049 (get_string_length, format_string,format_directive): Same.
30050 (pass_sprintf_length::compute_format_length): Same.
30051 (try_substitute_return_value): Simplify slightly.
30053 PR middle-end/78703
30054 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
30055 (fmtresult::operator+=): Outlined.
30056 (struct fmtresult): Add ctors.
30057 (struct conversion_spec): Rename...
30058 (struct directive): ...to this. Add and remove data members.
30059 (directive::set_width, directive::set_precision): New functions.
30060 (format_percent): Use fmtresult ctor.
30061 (get_width_and_precision): Remove.
30062 (format_integer): Make naming changes. Avoid computing width and
30064 (format_floating): Same. Adjust indentation.
30065 (format_character, format_none): New functions.
30066 (format_string): Moved character handling to format_character.
30067 (format_directive): Remove arguments, change return type.
30068 (parse_directive): New function.
30069 (pass_sprintf_length::compute_format_length): Move directive
30070 parsing to parse_directive.
30072 2017-01-23 Jakub Jelinek <jakub@redhat.com>
30074 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
30075 (assign_assembler_name_if_needed): ... this.
30076 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
30077 (assign_assembler_name_if_needed): ... this.
30078 (free_lang_data_in_cgraph): Adjust callers.
30079 * cgraphunit.c (cgraph_node::analyze): Likewise.
30080 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
30083 2017-01-23 Richard Biener <rguenther@suse.de>
30085 PR tree-optimization/79088
30086 PR tree-optimization/79188
30087 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
30088 resetting loop bounds after last path deletion. Reset loop
30089 bounds of the target loop, make code match the comments.
30090 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
30091 Make sure loops need no fixups.
30093 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
30095 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
30096 exponent support with double type for first argument.
30097 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
30098 type returned by __builtin_vec_extract_sig,
30099 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
30100 functions from "vector int" to "vector unsigned int" or from
30101 "vector long long int" to "vector unsigned long long int".
30102 Changed type returned by __builtin_vec_extract_exp,
30103 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
30104 functions from "vector int" to "vector unsigned int" or from
30105 "vector long long int" to "vector unsigned long long int".
30106 Changed return type of __builtin_vec_test_data_class,
30107 __builtin_vec_test_data_class_sp, and
30108 __builtin_vec_test_data_class_dp from "vector int" to
30109 "vector bool int" or from "vector long long int" to "vector bool
30110 long long int" and changed second argument type from "unsigned
30111 int" to "int". Added new overloaded function forms "vector float
30112 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
30113 "vector float __builtin_vec_insert_exp_sp (vector float, vector
30114 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
30115 double, vector unsigned long long int)" and "vector double
30116 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
30117 long int)". Changed return type of
30118 __builtin_scalar_test_data_class and
30119 __builtin_scalar_test_data_class_sp and
30120 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
30121 int" and changed second argument from "unsigned int" to "int".
30122 Changed type returned by __builtin_scalar_test_neg,
30123 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
30124 from "int" to "bool int". Added new overloaded function form
30125 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
30126 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
30127 exponent double-precision with floating point first argument.
30128 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
30129 documentation of scalar_test_data_class, scalar_test_neg,
30130 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
30131 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
30132 vec_test_data_class built-in functions to reflect refinements in
30133 their type signatures.
30135 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
30137 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
30139 (aarch64_elf_asm_destructor): Likewise.
30141 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
30143 PR rtl-optimization/78634
30144 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
30145 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
30146 * ifcvt.c (noce_try_cmove): Add missing cost check.
30148 PR rtl-optimization/71724
30149 * combine.c (if_then_else_cond): Look for situations where it is
30150 beneficial to undo the work of one of the recursive calls.
30152 2017-01-23 Bin Cheng <bin.cheng@arm.com>
30154 PR tree-optimization/70754
30155 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
30156 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
30157 combined stmt before it if not NULL.
30158 (combine_chains): Process refs reversely and compute dominance point
30161 2017-01-23 Martin Liska <mliska@suse.cz>
30163 PR tree-optimization/79196
30164 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
30165 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
30167 (strlen_optimize_stmt): Call the renamed function.
30169 2017-01-23 Michael Matz <matz@suse.de>
30171 PR tree-optimization/78384
30172 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
30174 2017-01-23 Richard Biener <rguenther@suse.de>
30176 PR tree-optimization/79186
30177 * tree-vrp.c (register_new_assert_for): Make sure we've seen
30178 both incoming edges before moving an assert.
30180 2017-01-23 Martin Jambor <mjambor@suse.cz>
30182 * ipa-prop.c (load_from_param_1): Removed.
30183 (load_from_unmodified_param): Bits from load_from_param_1 put back
30185 (load_from_param): Removed.
30186 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
30187 with stmt. Reverted back to use of load_from_unmodified_param.
30189 2017-01-23 Martin Jambor <mjambor@suse.cz>
30192 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
30193 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
30194 field a pointer to garbage collected vector, mark lattices and
30195 ipcp_orig_node with GTY((skip)).
30196 (ipa_get_param_count): Adjust to descriptors being a pointer.
30197 (ipa_get_param): Likewise.
30198 (ipa_get_type): Likewise.
30199 (ipa_get_param_move_cost): Likewise.
30200 (ipa_set_param_used): Likewise.
30201 (ipa_get_controlled_uses): Likewise.
30202 (ipa_set_controlled_uses): Likewise.
30203 (ipa_is_param_used): Likewise.
30204 (ipa_node_params_t): Move into garbage collector. New methods insert
30206 (ipa_node_params_sum): Annotate wth GTY(()).
30207 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
30209 (ipa_load_from_parm_agg): Adjust declaration.
30210 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
30211 * ipa-profile.c (ipa_profile): Likewise.
30212 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
30213 (ipa_populate_param_decls): Make descriptors parameter garbage
30215 (ipa_dump_param): Adjust to descriptors being a pointer.
30216 (ipa_alloc_node_params): Likewise.
30217 (ipa_initialize_node_params): Likewise.
30218 (load_from_param_1): Make descriptors parameter garbage collected.
30219 (load_from_unmodified_param): Likewise.
30220 (load_from_param): Likewise.
30221 (ipa_load_from_parm_agg): Likewise.
30222 (ipa_node_params::~ipa_node_params): Removed.
30223 (ipa_free_all_node_params): Remove call to delete operator.
30224 (ipa_node_params_t::insert): New.
30225 (ipa_node_params_t::remove): Likewise.
30226 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
30227 copy known_csts and known_contexts vectors.
30228 (ipa_read_node_info): Adjust to descriptors being a pointer.
30229 (ipcp_modif_dom_walker): Make m_descriptors field garbage
30231 (ipcp_transform_function): Make descriptors variable garbage
30234 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
30236 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
30237 * config/i386/avx512dqintrin.h: Ditto.
30238 * config/i386/avx512fintrin.h: Ditto.
30239 * config/i386/i386.c: Handle new builtins.
30240 * config/i386/i386-builtin.def: Add new builtins.
30241 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
30242 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
30244 2017-01-23 Jakub Jelinek <jakub@redhat.com>
30245 Martin Liska <mliska@suse.cz>
30247 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
30248 * asan.c (asan_expand_poison_ifn): Support stores and use
30249 appropriate ASAN report function.
30250 * internal-fn.c (expand_ASAN_POISON_USE): New function.
30251 * internal-fn.def (ASAN_POISON_USE): Declare.
30252 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
30253 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
30254 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
30255 ASAN_POISON calls w/o LHS.
30256 * tree-ssa.c (execute_update_addresses_taken): Create clobber
30257 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
30258 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
30259 * gimplify.c (asan_poison_variables): Add attribute
30260 use_after_scope_memory to variables that really needs to live
30262 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
30263 having the attribute.
30265 2017-01-23 Martin Liska <mliska@suse.cz>
30267 * asan.c (create_asan_shadow_var): New function.
30268 (asan_expand_poison_ifn): Likewise.
30269 * asan.h (asan_expand_poison_ifn): New declaration.
30270 * internal-fn.c (expand_ASAN_POISON): Likewise.
30271 * internal-fn.def (ASAN_POISON): New builtin.
30272 * sanopt.c (pass_sanopt::execute): Expand
30273 asan_expand_poison_ifn.
30274 * tree-inline.c (copy_decl_for_dup_finish): Make function
30276 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
30277 * tree-ssa.c (is_asan_mark_p): New function.
30278 (execute_update_addresses_taken): Rewrite local variables
30279 (identified just by use-after-scope as addressable) into SSA.
30281 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
30283 * doc/install.texi (Specific): opensource.apple.com uses https
30284 now. Remove trailing slash.
30286 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
30288 * README.Portability: Remove note on an Irix compatibility issue.
30290 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
30292 * gcov.c (INCLUDE_ALGORITHM): Define.
30293 (INCLUDE_VECTOR): Define.
30294 No longer include <vector> and <algorithm> directly.
30296 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
30298 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
30300 * doc/invoke.texi (Code Gen Options): Ditto.
30302 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
30305 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
30307 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
30309 rtl-optimization/79125
30310 * cprop.c (local_cprop_pass): Handle cases where we make an
30311 unconditional trap.
30313 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
30317 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
30318 read from, for big endian.
30320 2017-01-20 Jiong Wang <jiong.wang@arm.com>
30322 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
30323 register pauth builtins for LP64 only.
30325 2017-01-20 Marek Polacek <polacek@redhat.com>
30328 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
30331 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
30333 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
30335 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
30336 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
30337 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
30339 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30342 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
30343 in big-endian mode when they are not a single duplicated value.
30345 2017-01-20 Richard Biener <rguenther@suse.de>
30347 * BASE-VER: Bump to 7.0.1.
30349 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
30351 * omp-low.c (omplow_simd_context): New struct. Use it...
30352 (lower_rec_simd_input_clauses): ...here and...
30353 (lower_rec_input_clauses): ...here to hold common data. Adjust all
30354 references to idx, lane, max_vf, is_simt.
30356 2017-01-20 Graham Markall <graham.markall@embecosm.com>
30358 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
30361 2017-01-20 Martin Jambor <mjambor@suse.cz>
30363 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
30364 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
30366 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
30367 (GTFILES): Rename hsa.c to hsa-common.c.
30368 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
30369 * hsa-dump.c: Likewise.
30370 * hsa-gen.c: Likewise.
30371 * hsa-regalloc.c: Likewise.
30372 * ipa-hsa.c: Likewise.
30373 * omp-expand.c: Likewise.
30374 * omp-low.c: Likewise.
30375 * toplev.c: Likewise.
30377 2017-01-20 Marek Polacek <polacek@redhat.com>
30380 * doc/invoke.texi: Document -Wduplicated-branches.
30381 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
30382 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
30383 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
30384 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
30385 return 0 only when not OEP_LEXICOGRAPHIC.
30386 (fold_build_cleanup_point_expr): Use the expression
30387 location when building CLEANUP_POINT_EXPR.
30388 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
30389 * tree.c (add_expr): Handle error_mark_node.
30391 2017-01-20 Martin Liska <mliska@suse.cz>
30394 * tree-profile.c (init_ic_make_global_vars): Do not call
30396 (gimple_init_gcov_profiler): Likewise.
30398 2017-01-20 Martin Liska <mliska@suse.cz>
30401 * cgraph.h (maybe_create_reference): Remove argument and
30403 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
30405 * ipa-cp.c (create_specialized_node): Likewise.
30406 * symtab.c (symtab_node::maybe_create_reference): Handle
30407 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
30409 2017-01-20 Martin Liska <mliska@suse.cz>
30411 * read-rtl-function.c (function_reader::create_function): Use
30412 build_decl instread of build_decl_stat.
30414 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
30416 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
30417 * config/i386/avx512dqintrin.h: Ditto.
30418 * config/i386/avx512fintrin.h: Ditto.
30419 * config/i386/i386-builtin-types.def: Add new types.
30420 * config/i386/i386.c: Handle new types.
30421 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
30422 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
30423 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
30424 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
30425 (__builtin_ia32_kshiftridi): New.
30426 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
30428 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
30432 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
30433 define to rs6000_init_stack_protect_guard.
30434 (rs6000_init_stack_protect_guard): New function.
30436 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
30437 Yunqiang Su <yunqiang.su@imgtec.com>
30439 * config.gcc (supported_defaults): Add madd4.
30440 (with_madd4): Add validation.
30441 (all_defaults): Add madd4.
30442 * config/mips/mips.opt (mmadd4): New option.
30443 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
30445 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
30447 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
30448 (ISA_HAS_FUSED_MADD4): Likewise.
30449 * doc/invoke.texi (-mmadd4): Document the new option.
30450 * doc/install.texi (--with-madd4): Document the new option.
30452 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30454 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
30455 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
30456 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
30457 (aarch64_init_pauth_hint_builtins): New.
30458 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
30459 (aarch64_expand_builtin): Expand new builtins.
30461 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30463 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
30464 * combine-stack-adj.c (no_unhandled_cfa): Handle
30465 REG_CFA_TOGGLE_RA_MANGLE.
30466 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
30467 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
30468 info for return address signing.
30469 (aarch64_expand_epilogue): Likewise.
30471 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30473 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
30474 * config/aarch64/aarch64-protos.h
30475 (aarch64_return_address_signing_enabled): New declaration.
30476 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
30478 (aarch64_expand_prologue): Sign return address before it's pushed onto
30480 (aarch64_expand_epilogue): Authenticate return address fetched from
30482 (aarch64_override_options): Sanity check for ILP32 and ISA level.
30483 (aarch64_attributes): New function attributes for "sign-return-address".
30484 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
30485 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
30486 ("*do_return"): Generate combined instructions according to key index.
30487 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
30488 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
30490 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
30491 * config/aarch64/aarch64.opt (msign-return-address=): New.
30492 * doc/extend.texi (AArch64 Function Attributes): Documents
30493 "sign-return-address=".
30494 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
30496 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
30498 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
30499 overall option summary.
30501 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30503 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
30504 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
30505 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
30506 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
30508 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
30510 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
30511 -mpower9-minmax by default for -mcpu=power9.
30512 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
30513 128-bit floating point.
30515 2017-01-20 Alan Modra <amodra@gmail.com>
30517 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
30518 optimizing for size.
30520 2017-01-20 Alan Modra <amodra@gmail.com>
30523 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
30524 for strcmp and strncmp from corresponding builtin decl.
30526 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
30528 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
30529 instead of i386/rtems-64.h.
30530 * config/i386/rtems-64.h: Remove.
30532 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
30536 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
30538 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
30540 2017-01-19 Tamar Christina <tamar.christina@arm.com>
30542 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
30543 Change int to HOST_WIDE_INT.
30544 * config/aarch64/aarch64-protos.h
30545 (aarch64_simd_gen_const_vector_dup): Likewise.
30546 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
30548 2017-01-19 David Malcolm <dmalcolm@redhat.com>
30550 * langhooks-def.h (lhd_type_for_size): New decl.
30551 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
30552 * langhooks.c (lhd_type_for_size): New function, taken from
30555 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
30557 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
30558 define_bypass for CR latency.
30559 (power9-cracked-alu): Update bypass latency and remove power9-branch.
30560 (power9-alu2): Add define_bypass for CR latency.
30562 (power9-mul): Update insn latency.
30563 (power9-mul-compare): Update insn latency, bypass latency and remove
30566 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30568 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
30570 * config/aarch64/aarch64.md
30571 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
30572 aarch64_nopcrelative_literal_loads.
30573 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
30575 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
30577 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
30578 TARGET_LOONGSON_3A.
30579 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
30581 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
30584 * config.gcc (supported_defaults): Add lxc1-sxc1.
30585 (with_lxc1_sxc1): Add validation.
30586 (all_defaults): Add lxc1-sxc1.
30587 * config/mips/mips.opt (mlxc1-sxc1): New option.
30588 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
30590 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
30591 __mips_no_lxc1_sxc1.
30592 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
30593 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
30594 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
30596 2017-01-19 Richard Biener <rguenther@suse.de>
30598 PR tree-optimization/72488
30599 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
30600 sure to restore SSA info.
30601 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
30603 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
30605 PR rtl-optimization/79121
30606 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
30607 of the inner type when shifting an extended value.
30609 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
30612 * symtab.c (symtab_node::equal_address_to): Fix comparing of
30613 interposable aliases.
30615 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
30618 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
30619 Use the evmergelohi instruction.
30620 (mov_si<mode>_e500_subreg4_2_le): Likewise.
30621 (mov_sitf_e500_subreg8_2_be): Likewise.
30622 (mov_sitf_e500_subreg12_2_le): Likewise.
30623 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
30624 (mov_si<mode>_e500_subreg4_2_be): Likewise.
30625 (mov_sitf_e500_subreg8_2_le): Likewise.
30626 (mov_sitf_e500_subreg12_2_be): Likewise.
30628 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30630 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
30631 attribute from vecsimple to vecperm.
30632 (altivec_vbpermq2): Likewise.
30634 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30637 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
30639 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30640 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
30641 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
30642 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
30643 case where N arg is SIZE_MAX.
30644 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
30645 (cmpstrsi): Add pattern.
30647 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30649 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
30650 __builtin_vec_revb builtins.
30651 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
30652 built-in functions to support generation of the ISA 3.0 XXBR<x>
30653 vector byte reverse instructions.
30654 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
30655 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
30656 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
30657 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
30658 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
30659 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
30660 (P9V_BUILTIN_VEC_REVB): Likewise.
30661 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
30662 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
30663 (p9_xxbrq_v16qi): Likewise.
30664 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
30665 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
30666 (p9_xxbrh_v8hi): Likewise.
30667 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
30668 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
30669 vec_revb built-in functions.
30671 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
30673 PR rtl-optimization/78952
30674 * config/i386/i386.md (any_extract): New code iterator.
30675 (*insvqi_2): Use any_extract for source operand.
30676 (*insvqi_3): Use any_shiftrt for source operand.
30678 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
30680 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
30682 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
30684 2017-01-18 Matthias Klose <doko@ubuntu.com>
30686 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
30688 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30690 * config/rs6000/altivec.h (vec_bperm): Change #define.
30691 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
30692 (altivec_vbpermq2): New define_insn.
30693 (altivec_vbpermd): Likewise.
30694 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
30695 function interface.
30696 (VBPERMD): Likewise.
30697 (VBPERM): New polymorphic function interface.
30698 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
30699 Add entries for P9V_BUILTIN_VEC_VBPERM.
30700 * doc/extend.texi: Add interfaces for vec_bperm.
30702 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30704 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
30705 first letter of error messages.
30706 (s390_resolve_overloaded_builtin): Likewise.
30707 * config/s390/s390.c (s390_expand_builtin): Likewise.
30708 (s390_invalid_arg_for_unprototyped_fn): Likewise.
30709 (s390_valid_target_attribute_inner_p): Likewise.
30710 * config/s390/s390.md ("tabort"): Likewise.
30712 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
30714 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
30715 (ISA_AVOID_DIV_HILO): New macro.
30716 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
30717 (ISA_HAS_DDIV): Likewise.
30719 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30721 * doc/invoke.texi (fabi-version): Correct number of occurrences.
30723 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30725 * doc/invoke.texi (fabi-version): Spelling fix.
30727 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30730 * doc/invoke.texi (fabi-version): Mention mangling fix for
30733 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30736 * doc/invoke.texi (fabi-version): Document discriminator mangling.
30738 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
30741 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
30742 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
30744 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
30746 (stack_protect_test): Ditto.
30747 * config/rs6000/rs6000.opt (mstack-protector-guard=,
30748 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
30750 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
30751 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
30752 -mstack-protector-guard-offset=.
30753 (RS/6000 and PowerPC Options): Ditto.
30755 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
30757 * config/i386/i386.h (MASK_CLASS_P): New define.
30758 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
30759 there are no registers from different register sets also when
30760 mask registers are used. Update function comment.
30761 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
30762 to (*k/*r) and (*k/*km) alternatives.
30764 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
30766 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
30767 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
30768 (EH_RETURN_HANDLER_RTX): New define.
30769 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
30770 Force frame pointer in EH return functions.
30771 (aarch64_expand_epilogue): Add barrier for eh_return.
30772 (aarch64_final_eh_return_addr): Remove.
30773 (aarch64_eh_return_handler_rtx): New function.
30774 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
30776 (aarch64_eh_return_handler_rtx): New prototype.
30778 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30780 * config/rs6000/altivec.h (vec_rlmi): New #define.
30781 (vec_vrlnm): Likewise.
30782 (vec_rlnm): Likewise.
30783 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
30784 (UNSPEC_VRLNM): Likewise.
30785 (VIlong): New mode iterator.
30786 (altivec_vrl<VI_char>mi): New define_insn.
30787 (altivec_vrl<VI_char>nm): Likewise.
30788 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
30790 (VRLDNM): Likewise.
30791 (RLNM): New polymorphic function entry.
30792 (VRLWMI): New monomorphic function entry.
30793 (VRLDMI): Likewise.
30794 (RLMI): New polymorphic function entry.
30795 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
30796 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
30797 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
30800 2017-01-17 Jakub Jelinek <jakub@redhat.com>
30803 * dwarf2out.c (field_byte_offset): Restore the
30804 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
30805 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
30806 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
30809 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
30812 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
30814 2017-01-17 Jakub Jelinek <jakub@redhat.com>
30817 * dwarf2out.c (add_data_member_location_attribute): For constant
30818 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
30819 instead of DW_AT_data_member_location, DW_AT_bit_offset and
30820 DW_AT_byte_size attributes.
30822 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
30824 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
30825 after forcing to constant memory when the code model is medium.
30827 2017-01-17 Julia Koval <julia.koval@intel.com>
30830 * config/i386/avx512fintrin.h
30831 (_mm512_i32gather_ps): Change __addr type to void const*.
30832 (_mm512_mask_i32gather_ps): Ditto.
30833 (_mm512_i32gather_pd): Ditto.
30834 (_mm512_mask_i32gather_pd): Ditto.
30835 (_mm512_i64gather_ps): Ditto.
30836 (_mm512_mask_i64gather_ps): Ditto.
30837 (_mm512_i64gather_pd): Ditto.
30838 (_mm512_mask_i64gather_pd): Ditto.
30839 (_mm512_i32gather_epi32): Ditto.
30840 (_mm512_mask_i32gather_epi32): Ditto.
30841 (_mm512_i32gather_epi64): Ditto.
30842 (_mm512_mask_i32gather_epi64): Ditto.
30843 (_mm512_i64gather_epi32): Ditto.
30844 (_mm512_mask_i64gather_epi32): Ditto.
30845 (_mm512_i64gather_epi64): Ditto.
30846 (_mm512_mask_i64gather_epi64): Ditto.
30847 (_mm512_i32scatter_ps): Change __addr type to void*.
30848 (_mm512_mask_i32scatter_ps): Ditto.
30849 (_mm512_i32scatter_pd): Ditto.
30850 (_mm512_mask_i32scatter_pd): Ditto.
30851 (_mm512_i64scatter_ps): Ditto.
30852 (_mm512_mask_i64scatter_ps): Ditto.
30853 (_mm512_i64scatter_pd): Ditto.
30854 (_mm512_mask_i64scatter_pd): Ditto.
30855 (_mm512_i32scatter_epi32): Ditto.
30856 (_mm512_mask_i32scatter_epi32): Ditto.
30857 (_mm512_i32scatter_epi64): Ditto.
30858 (_mm512_mask_i32scatter_epi64): Ditto.
30859 (_mm512_i64scatter_epi32): Ditto.
30860 (_mm512_mask_i64scatter_epi32): Ditto.
30861 (_mm512_i64scatter_epi64): Ditto.
30862 (_mm512_mask_i64scatter_epi64): Ditto.
30863 * config/i386/avx512pfintrin.h
30864 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
30865 (_mm512_mask_prefetch_i32gather_ps): Ditto.
30866 (_mm512_mask_prefetch_i64gather_pd): Ditto.
30867 (_mm512_mask_prefetch_i64gather_ps): Ditto.
30868 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
30869 (_mm512_prefetch_i32scatter_ps): Ditto.
30870 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
30871 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
30872 (_mm512_prefetch_i64scatter_pd): Ditto.
30873 (_mm512_prefetch_i64scatter_ps): Ditto.
30874 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
30875 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
30876 * config/i386/avx512vlintrin.h
30877 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
30878 (_mm_mmask_i32gather_ps): Ditto.
30879 (_mm256_mmask_i32gather_pd): Ditto.
30880 (_mm_mmask_i32gather_pd): Ditto.
30881 (_mm256_mmask_i64gather_ps): Ditto.
30882 (_mm_mmask_i64gather_ps): Ditto.
30883 (_mm256_mmask_i64gather_pd): Ditto.
30884 (_mm_mmask_i64gather_pd): Ditto.
30885 (_mm256_mmask_i32gather_epi32): Ditto.
30886 (_mm_mmask_i32gather_epi32): Ditto.
30887 (_mm256_mmask_i32gather_epi64): Ditto.
30888 (_mm_mmask_i32gather_epi64): Ditto.
30889 (_mm256_mmask_i64gather_epi32): Ditto.
30890 (_mm_mmask_i64gather_epi32): Ditto.
30891 (_mm256_mmask_i64gather_epi64): Ditto.
30892 (_mm_mmask_i64gather_epi64): Ditto.
30893 (_mm256_i32scatter_ps): Change __addr type to void*.
30894 (_mm256_mask_i32scatter_ps): Ditto.
30895 (_mm_i32scatter_ps): Ditto.
30896 (_mm_mask_i32scatter_ps): Ditto.
30897 (_mm256_i32scatter_pd): Ditto.
30898 (_mm256_mask_i32scatter_pd): Ditto.
30899 (_mm_i32scatter_pd): Ditto.
30900 (_mm_mask_i32scatter_pd): Ditto.
30901 (_mm256_i64scatter_ps): Ditto.
30902 (_mm256_mask_i64scatter_ps): Ditto.
30903 (_mm_i64scatter_ps): Ditto.
30904 (_mm_mask_i64scatter_ps): Ditto.
30905 (_mm256_i64scatter_pd): Ditto.
30906 (_mm256_mask_i64scatter_pd): Ditto.
30907 (_mm_i64scatter_pd): Ditto.
30908 (_mm_mask_i64scatter_pd): Ditto.
30909 (_mm256_i32scatter_epi32): Ditto.
30910 (_mm256_mask_i32scatter_epi32): Ditto.
30911 (_mm_i32scatter_epi32): Ditto.
30912 (_mm_mask_i32scatter_epi32): Ditto.
30913 (_mm256_i32scatter_epi64): Ditto.
30914 (_mm256_mask_i32scatter_epi64): Ditto.
30915 (_mm_i32scatter_epi64): Ditto.
30916 (_mm_mask_i32scatter_epi64): Ditto.
30917 (_mm256_i64scatter_epi32): Ditto.
30918 (_mm256_mask_i64scatter_epi32): Ditto.
30919 (_mm_i64scatter_epi32): Ditto.
30920 (_mm_mask_i64scatter_epi32): Ditto.
30921 (_mm256_i64scatter_epi64): Ditto.
30922 (_mm256_mask_i64scatter_epi64): Ditto.
30923 (_mm_i64scatter_epi64): Ditto.
30924 (_mm_mask_i64scatter_epi64): Ditto.
30925 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
30926 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
30927 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
30928 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
30929 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
30930 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
30931 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
30932 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
30933 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
30934 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
30935 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
30936 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
30937 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
30938 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
30939 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
30940 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
30941 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
30942 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
30943 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
30944 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
30945 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
30946 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
30947 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
30948 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
30949 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
30950 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
30951 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
30952 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
30953 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
30954 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
30955 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
30956 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
30957 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
30958 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
30959 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
30960 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
30961 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
30962 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
30963 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
30964 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
30965 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
30966 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
30967 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
30968 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
30969 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
30970 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
30971 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
30972 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
30973 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
30974 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
30975 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
30976 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
30977 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
30978 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
30979 definitions accordingly.
30981 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
30982 Kuan-Lin Chen <kuanlinchentw@gmail.com>
30985 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
30988 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
30991 * ira-conflicts.c (ira_build_conflicts): Update total conflict
30992 hard regs for inner regno.
30994 2017-01-17 Martin Liska <mliska@suse.cz>
30997 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
30998 assumption and add comment.
31000 2017-01-17 Nathan Sidwell <nathan@acm.org>
31002 * ipa-visibility.c (localize_node): New function, broken out of ...
31003 (function_and_variable_visibility): ... here. Call it.
31005 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
31007 PR middle-end/77445
31008 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
31009 correctly set frequency of oudgoing edge.
31010 (duplicate_thread_path): Fix profile updating.
31012 2017-01-17 Jakub Jelinek <jakub@redhat.com>
31015 * configure.ac: Add GCC_BASE_VER.
31016 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
31017 version from BASE-VER file.
31018 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
31019 (gcc.o): Depend on $(BASEVER).
31020 * common.opt (dumpfullversion): New option.
31021 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
31022 * doc/invoke.texi: Document -dumpfullversion.
31023 * doc/install.texi: Document --with-gcc-major-version-only.
31024 * configure: Regenerated.
31026 2017-01-17 Richard Biener <rguenther@suse.de>
31028 PR tree-optimization/71433
31029 * tree-vrp.c (register_new_assert_for): Merge same asserts
31030 on all incoming edges.
31031 (process_assert_insertions_for): Handle insertions at the
31034 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
31036 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
31037 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
31039 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
31042 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
31045 2017-01-17 Alan Modra <amodra@gmail.com>
31048 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
31049 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
31050 symbolic stack limit when pic.
31052 2017-01-16 Martin Sebor <msebor@redhat.com>
31054 PR tree-optimization/78608
31055 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
31057 2017-01-16 Jeff Law <law@redhat.com>
31060 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
31061 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
31062 for several include directories that may be relative to sysroot.
31063 * config/i386/x-mingw32 (gplus_includedir): Define.
31064 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
31065 (native_system_includedir): Likewise.
31066 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
31067 override if TARGET_SYSTEM_ROOT is defined.
31068 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
31070 PR tree-optimization/79090
31071 PR tree-optimization/33562
31072 PR tree-optimization/61912
31073 PR tree-optimization/77485
31074 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
31075 and computed trims into the dump file.
31077 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
31079 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
31081 2017-01-16 Jakub Jelinek <jakub@redhat.com>
31084 * gimplify.c (gimplify_init_constructor): If want_value and
31085 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
31089 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
31090 sequence. Formatting fixes.
31091 (doloop_optimize): Formatting fixes.
31094 * gcc.c (debug_level_greater_than_spec_func): New function.
31095 (static_spec_functions): Add debug-level-gt spec function.
31096 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
31098 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
31099 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
31100 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
31101 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
31102 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
31103 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
31105 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
31107 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
31108 QImode fixups to general and mask registers only.
31110 2017-01-16 Carl Love <cel@us.ibm.com>
31112 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
31113 for built-in functions
31114 vector signed char vec_nabs (vector signed char)
31115 vector signed short vec_nabs (vector signed short)
31116 vector signed int vec_nabs (vector signed int)
31117 vector signed long long vec_nabs (vector signed long long)
31118 vector float vec_nabs (vector float)
31119 vector double vec_nabs (vector double)
31120 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
31122 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
31123 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
31124 * doc/extend.texi: Update the documentation file for the new built-in
31127 2017-01-16 Martin Sebor <msebor@redhat.com>
31129 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
31132 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31134 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
31135 UNSPEC_VSX__XXSPLTD to require special splat handling.
31137 2017-01-16 David Malcolm <dmalcolm@redhat.com>
31140 * system.h: Poison strndup.
31142 2017-01-16 Alan Modra <amodra@gmail.com>
31145 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
31148 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
31150 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
31152 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
31154 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
31155 call recog here. Assert that INSN_CODE (insn) is non-negative.
31157 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
31160 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
31162 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
31163 in the currently scheduled RTL fragment.
31165 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
31167 PR rtl-optimization/78751
31168 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
31171 2017-01-14 Jeff Law <law@redhat.com>
31173 PR tree-optimization/79090
31174 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
31175 variable length stores.
31176 (compute_trims): Delete dead assignment to *trim_tail.
31177 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
31180 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
31182 PR rtl-optimization/78626
31183 PR rtl-optimization/78727
31184 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
31185 of a block, and split such blocks after everything else is finished.
31187 2017-01-14 Alan Modra <amodra@gmail.com>
31190 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
31191 target legitimate_combined_insn.
31192 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
31193 (rs6000_legitimate_combined_insn): New function.
31194 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
31196 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
31197 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
31198 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
31200 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
31202 * doc/frontends.texi (G++ and GCC): Remove references to Java.
31204 2017-01-13 Jeff Law <law@redhat.com>
31206 PR tree-optimization/33562
31207 PR tree-optimization/61912
31208 PR tree-optimization/77485
31209 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
31211 (delete_dead_assignment): Likewise.
31212 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
31213 statement to delete_dead_call and delete_dead_assignment.
31215 2017-01-13 David Malcolm <dmalcolm@redhat.com>
31218 * substring-locations.c (format_warning_va): Strengthen case 1 so
31219 that both endpoints of the substring must be within the format
31220 range for just the substring to be printed.
31222 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
31224 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
31225 * config/i386/i386.c (ix86_target_string): Add missing options
31226 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
31227 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
31228 flags_other and ix86_target_other to flags2_other. Display unknown
31230 (ix86_valid_target_attribute_inner_p): Add missing options and
31231 reorder options by implied ISAs, as in ix86_target_string.
31233 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
31235 * hash-table.h (hash_table::too_empty_p): New function.
31236 (hash_table::expand): Use it.
31237 (hash_table::traverse): Likewise.
31238 (hash_table::empty_slot): Use sizeof (value_type) instead of
31239 sizeof (PTR) to convert bytes to elements. Shrink the table
31240 if the current size is excessive for the current number of
31243 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
31245 * ira-costs.c (record_reg_classes): Break from the inner loop
31246 early once alt_fail is known to be true. Update outer loop
31247 handling accordingly.
31249 2017-01-13 Jeff Law <law@redhat.com>
31251 * tree-ssa-dse.c (decrement_count): New function.
31252 (increment_start_addr, maybe_trim_memstar_call): Likewise.
31253 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
31254 when we know the partially dead statement is a mem* function.
31256 PR tree-optimization/61912
31257 PR tree-optimization/77485
31258 * tree-ssa-dse.c: Include expr.h.
31259 (maybe_trim_constructor_store): New function.
31260 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
31262 PR tree-optimization/33562
31263 PR tree-optimization/61912
31264 PR tree-optimization/77485
31265 * doc/invoke.texi: Document new dse-max-object-size param.
31266 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
31267 * tree-ssa-dse.c: Include params.h.
31268 (dse_store_status): New enum.
31269 (initialize_ao_ref_for_dse): New, partially extracted from
31271 (valid_ao_ref_for_dse, normalize_ref): New.
31272 (setup_live_bytes_from_ref, compute_trims): Likewise.
31273 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
31274 (maybe_trim_partially_dead_store): Likewise.
31275 (maybe_trim_complex_store): Likewise.
31276 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
31277 Track what bytes live from the original store. Return tri-state
31278 for dead, partially dead or live.
31279 (dse_dom_walker): Add constructor, destructor and new private members.
31280 (delete_dead_call, delete_dead_assignment): New extracted from
31282 (dse_optimize_stmt): Make a member of dse_dom_walker.
31283 Use initialize_ao_ref_for_dse.
31285 PR tree-optimization/33562
31286 PR tree-optimization/61912
31287 PR tree-optimization/77485
31288 * sbitmap.h (bitmap_count_bits): Prototype.
31289 (bitmap_clear_range, bitmap_set_range): Likewise.
31290 * sbitmap.c (bitmap_clear_range): New function.
31291 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
31293 2017-01-13 Martin Liska <mliska@suse.cz>
31296 * function.c (set_cfun): Add new argument force.
31297 * function.h (set_cfun): Likewise.
31298 * ipa-inline-transform.c (inline_call): Use the function when
31299 strict alising from is dropped for function we inline to.
31301 2017-01-13 Richard Biener <rguenther@suse.de>
31303 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
31304 for dumping GIMPLE INTEGER_CSTs.
31306 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31308 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
31309 to 201112L since C++17.
31311 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
31314 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
31315 if -fsanitize=kernel-address is present.
31317 2017-01-13 Richard Biener <rguenther@suse.de>
31319 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
31320 as _Literal ( type ) number in case usual suffixes do not
31321 preserve all information.
31323 2017-01-13 Richard Biener <rguenther@suse.de>
31325 PR tree-optimization/77283
31326 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
31327 and ssa-iterators.h.
31328 (is_feasible_trace): Implement a cost model based on joiner
31331 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
31334 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
31335 char or short to __float128/_Float128 directly.
31337 2017-01-12 Martin Sebor <msebor@redhat.com>
31339 to -Wformat-overflow.
31340 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
31341 (min_bytes_remaining): Same.
31342 (get_string_length): Same.
31343 (format_string): Same.
31344 (format_directive): Same.
31346 (pass_sprintf_length::handle_gimple_call): Same.
31348 2017-01-12 Jakub Jelinek <jakub@redhat.com>
31350 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
31351 info.nowrite calls with no lhs that can't throw. Return bool
31352 whether gsi_remove has been called or not.
31353 (pass_sprintf_length::handle_gimple_call): Return bool whether
31354 try_substitute_return_value called gsi_remove. Formatting fix.
31355 (pass_sprintf_length::execute): Don't use gsi_remove if
31356 handle_gimple_call returned true.
31359 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
31360 be removed due to side-effects, don't remove following barrier nor
31361 turn the successor edge into fallthru edge.
31363 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31366 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
31367 element-reversing loads and stores as not swappable.
31369 2017-01-12 Nathan Sidwell <nathan@acm.org>
31370 Nicolai Stange <nicstange@gmail.com>
31372 * combine.c (try_combine): Don't ignore result of overlap checking
31373 loop. Combine overlap & asm check into single loop.
31375 2017-01-12 Richard Biener <rguenther@suse.de>
31377 * tree-pretty-print.c (dump_generic_node): Provide -gimple
31378 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
31380 2017-01-12 Richard Biener <rguenther@suse.de>
31382 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
31383 and TS_TARGET_OPTION directly derive from TS_BASE.
31384 * tree-core.h (tree_optimization_option): Derive from tree_base.
31385 (tree_target_option): Likewise.
31387 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
31389 * config/i386/i386.c (memory_address_length): Increase len
31390 only when rip_relative_addr_p returns false.
31392 2017-01-11 Julia Koval <julia.koval@intel.com>
31394 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
31395 (OPTION_MASK_ISA_SGX_SET): New.
31396 (ix86_handle_option): Handle OPT_msgx.
31397 * config.gcc: Added sgxintrin.h.
31398 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
31399 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
31400 * config/i386/i386.c (ix86_target_string): Add -msgx.
31402 (ix86_option_override_internal): Handle new options.
31403 (ix86_valid_target_attribute_inner_p): Add sgx.
31404 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
31405 * config/i386/i386.opt: Add msgx.
31406 * config/i386/sgxintrin.h: New file.
31407 * config/i386/x86intrin.h: Add sgxintrin.h.
31409 2017-01-11 Jakub Jelinek <jakub@redhat.com>
31412 * fold-const.c (maybe_nonzero_address): Return 1 for function
31414 (tree_single_nonzero_warnv_p): Don't handle function local objects
31418 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
31421 2017-01-11 David Malcolm <dmalcolm@redhat.com>
31424 * opts.c: Include "spellcheck.h"
31425 (struct string_fragment): New struct.
31426 (struct edit_distance_traits<const string_fragment &>): New
31428 (get_closest_sanitizer_option): New function.
31429 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
31431 2017-01-11 Jakub Jelinek <jakub@redhat.com>
31433 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
31435 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
31436 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
31437 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
31438 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
31439 for initial die_offset if dwarf_split_debug_info.
31440 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
31441 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
31443 (output_skeleton_debug_sections): Formatting fix. Use
31444 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
31445 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
31447 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
31449 * config/arm/cortex-a53.md: Add bypasses for
31450 cortex_a53_r2f_cvt.
31451 (cortex_a53_r2f): Only use for transfers.
31452 (cortex_a53_f2r): Likewise.
31453 (cortex_a53_r2f_cvt): Add reservation for conversions.
31454 (cortex_a53_f2r_cvt): Likewise.
31456 2017-01-11 Tamar Christina <tamar.christina@arm.com>
31458 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
31459 to all inlined functions, change static to extern.
31461 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
31464 * config/arm/arm.c (legitimize_pic_address): Handle reference to
31466 (arm_assemble_integer): Likewise.
31468 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
31470 * config.gcc: Use new awk script to check CPU, FPU and architecture
31471 parameters for --with-... options.
31472 * config/arm/parsecpu.awk: New file
31473 * config/arm/arm-cpus.in: New file.
31474 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
31476 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
31478 * config/arm/t-arm: Update dependency rules.
31479 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
31480 of processing .def files.
31481 * config/arm/genopt.sh: Deleted.
31482 * config/arm/gentune.sh: Deleted.
31483 * config/arm/arm-cores.def: Deleted.
31484 * config/arm/arm-arches.def: Deleted.
31485 * config/arm/arm-fpus.def: Deleted.
31486 * config/arm/arm-tune.md: Regenerated.
31487 * config/arm/arm-tables.opt: Regenerated.
31488 * config/arm/arm-cpu.h: New generated file.
31489 * config/arm/arm-cpu-data.h: New generated file.
31490 * config/arm/arm-cpu-cdata.h: New generated file.
31492 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
31495 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
31497 (input_varpool_node): Unpack dynamically_initialized bit.
31499 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
31501 PR rtl-optimization/79032
31502 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
31503 the alignment of the adjusted memory reference against that of MODE,
31504 instead of the alignment of the original memory reference.
31506 2017-01-11 Martin Jambor <mjambor@suse.cz>
31508 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
31510 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
31511 decorated functions.
31513 2017-01-11 Richard Biener <rguenther@suse.de>
31515 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
31516 set range/nonnull info for PHI results. Do not set it on
31517 stmts marked for removal.
31519 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
31521 * expr.c (store_field): In the bitfield case, fetch the return value
31522 from the registers before applying a single big-endian adjustment.
31523 Always do a final load for a BLKmode value not larger than a word.
31525 2017-01-10 David Malcolm <dmalcolm@redhat.com>
31528 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
31529 that we correctly handle column numbers greater than
31530 LINE_MAP_MAX_COLUMN_NUMBER.
31532 2017-01-10 Martin Sebor <msebor@redhat.com>
31534 PR middle-end/78245
31535 * gimple-ssa-sprintf.c (get_destination_size): Call
31536 {init,fini}object_sizes.
31537 * tree-object-size.c (addr_object_size): Adjust.
31538 (pass_through_call): Adjust.
31539 (pass_object_sizes::execute): Adjust.
31540 * tree-object-size.h (fini_object_sizes): Declare.
31542 2017-01-10 Martin Sebor <msebor@redhat.com>
31544 PR tree-optimization/78775
31545 * builtins.c (get_size_range): Move...
31546 * calls.c: ...to here.
31547 (alloc_max_size): Accept zero argument.
31548 (operand_signed_p): Remove.
31549 (maybe_warn_alloc_args_overflow): Call get_size_range.
31550 * calls.h (get_size_range): Declare.
31552 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
31554 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
31555 from TI's devices.csv file as of September 2016.
31556 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
31558 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
31560 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
31561 * doc/invoke.texi: Likewise.
31562 * doc/md.texi: Likewise.
31563 * doc/objc.texi: Likewise.
31565 2017-01-10 Joshua Conner <joshconner@google.com>
31567 * config/arm/fuchsia-elf.h: New file.
31568 * config/fuchsia.h: New file.
31569 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
31570 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
31572 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
31574 2016-01-10 Richard Biener <rguenther@suse.de>
31576 PR tree-optimization/79034
31577 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
31578 Propagate out degenerate PHIs in the joiner.
31580 2017-01-10 Martin Liska <mliska@suse.cz>
31582 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
31583 (sort_congruence_classes_by_decl_uid): Likewise.
31584 (sort_congruence_class_groups_by_decl_uid): Likewise.
31585 (sem_item_optimizer::merge_classes): Sort class, groups in these
31586 classes and members in the groups by DECL_UID of declarations.
31587 This would make merge operations stable.
31589 2017-01-10 Martin Liska <mliska@suse.cz>
31591 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
31592 usage of m_classes_vec.
31593 (sem_item_optimizer::~sem_item_optimizer): Likewise.
31594 (sem_item_optimizer::get_group_by_hash): Likewise.
31595 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
31596 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
31597 (sem_item_optimizer::verify_classes): Likewise.
31598 (sem_item_optimizer::process_cong_reduction): Likewise.
31599 (sem_item_optimizer::dump_cong_classes): Likewise.
31600 (sem_item_optimizer::merge_classes): Likewise.
31601 * ipa-icf.h (congruence_class_hash): Rename from
31602 congruence_class_group_hash. Remove declaration of m_classes_vec.
31604 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
31606 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
31607 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
31608 * config.gcc: Add avx512vpopcntdqintrin.h.
31609 * config/i386/avx512vpopcntdqintrin.h: New.
31610 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
31611 * config/i386/i386-builtin-types.def: Add new types.
31612 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
31613 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
31614 __builtin_ia32_vpopcountq_v8di_mask): New.
31615 * config/i386/i386-c.c (ix86_target_macros_internal): Define
31616 __AVX512VPOPCNTDQ__.
31617 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
31618 (PTA_AVX512VPOPCNTDQ): Define.
31619 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
31620 TARGET_AVX512VPOPCNTDQ_P): Define.
31621 * config/i386/i386.opt: Add mavx512vpopcntdq.
31622 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
31623 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
31625 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31627 PR middle-end/77484
31628 * predict.def (PRED_CALL): Set to 67.
31630 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
31632 * expr.c (store_field): In the bitfield case, if the value comes from
31633 a function call and is of an aggregate type returned in registers, do
31634 not modify the field mode; extract the value in all cases if the mode
31635 is BLKmode and the size is not larger than a word.
31637 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
31640 * config/i386/cpuid.h: Fix undefined behavior.
31642 2017-01-04 Jeff Law <law@redhat.com>
31644 PR tree-optimization/79007
31645 PR tree-optimization/67955
31646 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
31647 conservative for pt.null when flag_non_call_exceptions is on.
31649 2017-01-09 Jakub Jelinek <jakub@redhat.com>
31651 PR translation/79019
31652 PR translation/79020
31653 * params.def (PARAM_INLINE_MIN_SPEEDUP,
31654 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
31655 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
31657 * config/avr/avr.opt (maccumulate-args): Likewise.
31658 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
31659 * common.opt (freport-bug): Likewise.
31660 * cif-code.def (CIF_FINAL_ERROR): Likewise.
31661 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
31662 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
31663 translatable string.
31664 * config/i386/i386.c (function_value_32): Likewise.
31665 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
31666 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
31668 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
31669 * common/config/msp430/msp430-common.c (msp430_handle_option):
31671 * symtab.c (symtab_node::verify_base): Likewise.
31672 * opts.c (set_debug_level): Likewise.
31673 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
31674 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
31675 missing whitespace to translatable strings.
31676 * config/avr/avr.md (bswapsi2): Fix typo in comment.
31677 * config/sh/superh.h: Likewise.
31678 * config/i386/xopintrin.h: Likewise.
31679 * config/i386/znver1.md: Likewise.
31680 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
31681 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
31682 * double-int.h (struct double_int): Likewise.
31683 * double-int.c (div_and_round_double): Likewise.
31684 * wide-int.cc: Likewise.
31685 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
31686 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
31687 * cfgcleanup.c (crossjumps_occured): Renamed to ...
31688 (crossjumps_occurred): ... this.
31689 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
31692 PR tree-optimization/78899
31693 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
31694 returning bool return struct loop *, NULL for failure and the new
31696 (versionable_outer_loop_p): Don't version outer loop if it has
31697 dont_vectorized bit set.
31698 (tree_if_conversion): When versioning outer loop, ensure
31699 tree_if_conversion is performed also on the inner loop of the
31700 non-vectorizable outer loop copy.
31701 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
31702 LOOP_VECTORIZED in inner loop of the scalar outer loop and
31703 prevent vectorization of it.
31704 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
31705 the outer loop vectorization of the non-scalar version is attempted
31706 before vectorization of the inner loop in scalar version. If
31707 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
31708 vectorization of its inner loop.
31709 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
31710 has 2 inner loops, rename also on edges from bb whose single pred
31711 is outer_loop->header. Fix typo in function comment.
31713 2017-01-09 Martin Sebor <msebor@redhat.com>
31716 * asan.c (asan_emit_stack_protection): Increase local buffer size
31717 to avoid snprintf truncation warning.
31719 2017-01-09 Andrew Pinski <apinski@cavium.com>
31721 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
31722 to reference thunderx2t99 for the tuning structure
31723 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
31725 (thunderx2t99_extra_costs): This.
31726 * config/aarch64/aarch64-tune.md: Regenerate.
31727 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
31728 (vulcan_addrcost_table): This.
31729 (vulcan_regmove_cost): Rename to ...
31730 (thunderx2t99_regmove_cost): This.
31731 (vulcan_vector_cost): Rename to ...
31732 (thunderx2t99_vector_cost): this.
31733 (vulcan_branch_cost): Rename to ...
31734 (thunderx2t99_branch_cost): This.
31735 (vulcan_tunings): Rename to ...
31736 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
31737 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
31739 2017-01-09 Martin Jambor <mjambor@suse.cz>
31743 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
31744 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
31745 (propagate_vr_accross_jump_function): Use the above function for all
31746 value range computations for pass-through jump functions and type
31747 converasion from explicit value range values.
31748 (ipcp_propagate_stage): Do not attempt to deduce types of formal
31749 parameters from TYPE_ARG_TYPES.
31750 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
31751 (ipa_write_node_info): Stream type of the actual argument.
31752 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
31754 2017-01-09 Martin Liska <mliska@suse.cz>
31757 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
31758 (lookup_compiler): Do not show error message with have_E.
31760 2017-01-09 Jakub Jelinek <jakub@redhat.com>
31762 PR tree-optimization/78938
31763 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
31764 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
31765 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
31766 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
31769 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31771 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
31774 2017-01-09 Richard Biener <rguenther@suse.de>
31776 PR tree-optimization/78997
31777 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
31778 name condition properly.
31780 2017-01-09 Richard Biener <rguenther@suse.de>
31783 * dwarf2out.c (is_cxx): New overload with context.
31784 (is_naming_typedef_decl): Use it.
31786 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
31788 * invoke.texi (Option Summary): Correct spacing in option lists
31789 and add line breaks to fix over-long lines.
31791 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
31793 PR middle-end/17660
31795 * extend.texi (Common Variable Attributes): Add xref to GCC
31796 Internals manual to explain mode attribute keywords.
31798 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
31801 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
31802 and Preprocessor Options.
31803 (Options for Linking): Document -pthread here....
31804 (RS/6000 and PowerPC Options): ...not here.
31805 (Solaris 2 Options): ...or here.
31806 * doc/cppopts.texi: Document -pthread.
31808 2017-01-08 Martin Sebor <msebor@redhat.com>
31810 PR middle-end/77708
31811 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
31812 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
31813 New member functions.
31814 (format_directive): Used them.
31816 (pass_sprintf_length::handle_gimple_call): Same.
31817 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
31818 to avoid truncation for any argument.
31819 (extract_affine_mul): Same.
31820 * tree.c (get_file_function_name): Same.
31822 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31824 PR middle-end/77484
31825 * predict.def (PRED_INDIR_CALL): Set to 86.
31827 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
31829 PR preprocessor/54124
31830 * doc/cppopts.texi: Reformat -d subtable to list the full name
31831 of the options. Add cross-reference to the docs for the general
31832 compiler -d options.
31833 * doc/invoke.texi (Developer Options): Add cross-reference to the
31834 preprocessor-specific -d option documentation.
31836 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
31838 PR preprocessor/13498
31839 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
31840 redudant material, and reflect new command-line options.
31841 (System Headers): Likewise.
31843 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
31845 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
31846 -isystem, and -idirafter. Copy-edit.
31847 * doc/cppopts.texi: Copy-edit. Remove contradiction about
31848 default for -ftrack-macro-expansion. Delete obsolete and
31849 badly-formatted implementation details about -fdebug-cpp output.
31850 * doc/cppwarnopts.texi: Copy-edit.
31852 2017-01-07 David Malcolm <dmalcolm@redhat.com>
31855 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
31856 that the transition from a max line width >= 1<<10 to narrower
31857 lines works correctly.
31859 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
31861 * doc/options.texi (PerFunction): New.
31862 * opt-functions.awk (switch_flags): Map both Optimization and
31863 PerFunction to CL_OPTIMIZATION.
31864 * opth-gen.awk: Test for PerFunction flag along with
31866 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
31867 it only when the latter is present. Skip those that don't in
31868 the hash function generator.
31869 * common.opt (fvar-tracking): Mark as PerFunction instead of
31871 (fvar-tracking-assignments): Likewise.
31872 (fvar-tracking-assignments-toggle): Likewise.
31873 (fvar-tracking-uninit): Likewise.
31875 2017-01-07 Jakub Jelinek <jakub@redhat.com>
31877 PR translation/79018
31878 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
31881 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
31884 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
31885 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
31886 TARGET_LONG_JUMP_TABLE_OFFSETS.
31887 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
31888 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
31889 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
31890 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
31891 * config/m68k/m68k.md (tablejump expander): Likewise.
31892 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
31893 TARGET_LONG_JUMP_TABLE_OFFSETS.
31894 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
31895 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
31897 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
31898 David Holsgrove <david.holsgrove@xilinx.com>
31900 * common/config/microblaze/microblaze-common.c
31901 (TARGET_EXCEPT_UNWIND_INFO): Remove.
31902 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
31904 * config/microblaze/microblaze.c (microblaze_must_save_register)
31905 (microblaze_expand_epilogue, microblaze_return_addr): Handle
31907 (microblaze_eh_return): New function.
31908 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
31909 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
31910 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
31911 * config/microblaze/microblaze.md (eh_return): New pattern.
31913 2017-01-06 Jakub Jelinek <jakub@redhat.com>
31915 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
31916 GCC_DIAGNOSTIC_STRINGIFY): Define.
31918 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
31920 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31922 * config/arm/arm.md (<mcrr>): New.
31924 * config/arm/arm.c (arm_arch5te): New.
31925 (arm_option_override): Set arm_arch5te.
31926 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
31928 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
31929 (arm_mcrr_qualifiers): ... this. New.
31930 (MRRC_QUALIFIERS): Define to...
31931 (arm_mrrc_qualifiers): ... this. New.
31932 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
31934 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
31935 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
31936 (MRRCI, mrrc, MRRC): New.
31937 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
31938 VUNSPEC_MRRC2): New.
31940 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31942 * config/arm/arm.md (<mcr>): New.
31944 * config/arm/arm.c (arm_coproc_builtin_available): Add
31945 support for mcr, mrc, mcr2 and mrc2.
31946 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
31947 (arm_mcr_qualifiers): ... this. New.
31948 (MRC_QUALIFIERS): Define to ...
31949 (arm_mrc_qualifiers): ... this. New.
31950 (MCR_QUALIFIERS): Define to ...
31951 (arm_mcr_qualifiers): ... this. New.
31952 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
31954 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
31955 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
31956 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
31957 VUNSPEC_MRC2): New.
31959 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31961 * config/arm/arm.md (*ldc): New.
31965 * config/arm/arm.c (arm_coproc_builtin_available): Add
31966 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
31967 (arm_coproc_ldc_stc_legitimate_address): New.
31968 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
31969 'qualifier_const_pointer'.
31970 (LDC_QUALIFIERS): Define to...
31971 (arm_ldc_qualifiers): ... this. New.
31972 (STC_QUALIFIERS): Define to...
31973 (arm_stc_qualifiers): ... this. New.
31974 * config/arm/arm-protos.h
31975 (arm_coproc_ldc_stc_legitimate_address): New.
31976 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
31977 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
31978 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
31979 stc2, stcl, stc2l): New.
31980 * config/arm/constraints.md (Uz): New.
31981 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
31982 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
31983 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
31984 VUNSPEC_STC2L): New.
31986 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31988 * config/arm/arm.md (<cdp>): New.
31989 * config/arm/arm.c (neon_const_bounds): Rename this ...
31990 (arm_const_bounds): ... this.
31991 (arm_coproc_builtin_available): New.
31992 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
31993 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
31994 (CDP_QUALIFIERS): Define to...
31995 (arm_cdp_qualifiers): ... this. New.
31997 (arm_expand_builtin_args): Add case for 6 arguments.
31998 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
31999 (arm_const_bounds): ... this.
32000 (arm_coproc_builtin_available): New.
32001 * config/arm/arm_acle.h (__arm_cdp): New.
32003 * config/arm/arm_acle_builtins.def (cdp): New.
32005 * config/arm/iterators.md (CDPI,CDP,cdp): New.
32006 * config/arm/neon.md: Rename all 'neon_const_bounds' to
32007 'arm_const_bounds'.
32008 * config/arm/types.md (coproc): New.
32009 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
32010 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
32011 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
32012 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
32014 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32016 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
32017 (UBINOP_QUALIFIERS): New.
32019 (acle_builtin_data): New. Change comment.
32020 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
32021 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
32022 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
32023 arm_acle_builtins.def.
32024 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
32025 (arm_init_acle_builtins): New.
32026 (CRC32_BUILTIN): Remove.
32027 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
32028 crc32cb, crc32ch and crc32cw.
32029 (arm_init_crc32_builtins): Remove.
32030 (arm_init_builtins): Use arm_init_acle_builtins rather
32031 than arm_init_crc32_builtins.
32032 (arm_expand_acle_builtin): New.
32033 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
32034 * config/arm/arm_acle_builtins.def: New.
32036 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32038 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
32039 (arm_builtin_datum): ... this.
32040 (arm_init_neon_builtin): Rename to ...
32041 (arm_init_builtin): ... this. Add a new parameters PREFIX
32042 and USE_SIG_IN_NAME.
32043 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
32044 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
32045 'arm_builtin_datum'.
32046 (arm_init_vfp_builtins): Likewise.
32047 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
32048 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
32049 (arm_expand_neon_args): Rename to ...
32050 (arm_expand_builtin_args): ... this. Rename builtin_arg
32051 enum values and differentiate between ARG_BUILTIN_MEMORY
32052 and ARG_BUILTIN_NEON_MEMORY.
32053 (arm_expand_neon_builtin_1): Rename to ...
32054 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
32055 values, arm_expand_builtin_args and add bool parameter NEON.
32056 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
32057 (arm_expand_vfp_builtin): Likewise.
32058 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
32060 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32062 PR middle-end/77484
32063 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
32064 * predict.c (tree_estimate_probability_bb): Reverse direction of
32065 polymorphic call predictor.
32067 2017-01-06 David Malcolm <dmalcolm@redhat.com>
32069 * passes.c (execute_one_pass): Split out pass-skipping logic into...
32070 (determine_pass_name_match): ...this new function and...
32071 (should_skip_pass_p): ...this new function.
32073 2017-01-06 Nathan Sidwell <nathan@acm.org>
32075 * ipa-visibility.c (function_and_variable_visibility): Reformat
32076 comments and long lines. Remove extrneous if.
32077 * symtab.c (symtab_node::make_decl_local): Fix code format.
32078 (symtab_node::set_section_for_node): Fix comment typo.
32080 2017-01-06 Martin Liska <mliska@suse.cz>
32083 * lra-constraints.c: Rename invariant to lra_invariant.
32084 * predict.c (set_even_probabilities): Initialize e to NULL.
32086 2017-01-05 Martin Sebor <msebor@redhat.com>
32088 PR tree-optimization/78910
32089 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
32090 (format_integer): Correct off-by-one error in the handling
32091 of precision with negative numbers in signed conversions..
32093 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
32095 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
32097 2017-01-05 Jakub Jelinek <jakub@redhat.com>
32099 PR tree-optimization/71016
32100 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
32101 factor_out_conditional_conversion. Formatting fix.
32102 (factor_out_conditional_conversion): Add cond_stmt argument.
32103 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
32104 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
32107 2017-01-05 David Malcolm <dmalcolm@redhat.com>
32109 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
32110 read-rtl-function.o, and selftest-rtl.o.
32111 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
32112 (selftest::aarch64_test_loading_full_dump): New function.
32113 (selftest::aarch64_run_selftests): New function.
32114 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
32115 selftest::aarch64_run_selftests.
32116 * config/i386/i386.c
32117 (selftest::ix86_test_loading_dump_fragment_1): New function.
32118 (selftest::ix86_test_loading_call_insn): New function.
32119 (selftest::ix86_test_loading_full_dump): New function.
32120 (selftest::ix86_test_loading_unspec): New function.
32121 (selftest::ix86_run_selftests): Call the new functions.
32122 * emit-rtl.c (maybe_set_max_label_num): New function.
32123 * emit-rtl.h (maybe_set_max_label_num): New decl.
32124 * function.c (instantiate_decls): Guard call to
32125 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
32126 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
32128 * gensupport.c (gen_reader::gen_reader): Pass "false"
32129 for new "compact" param of rtx_reader.
32130 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
32131 rather than an empty string for NULL strings.
32132 * read-md.c: Potentially include config.h rather than bconfig.h.
32133 Wrap include of errors.h with #ifdef GENERATOR_FILE.
32134 (have_error): New global, copied from errors.c.
32135 (md_reader::read_name): Rename to...
32136 (md_reader::read_name_1): ...this, adding "out_loc" param,
32137 and converting "missing name or number" to returning false, rather
32139 (md_reader::read_name): Reimplement in terms of read_name_1.
32140 (md_reader::read_name_or_nil): New function.
32141 (md_reader::read_string): Handle "(nil)" by returning NULL.
32142 (md_reader::md_reader): Add new param "compact".
32143 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
32144 (md_reader::read_file): New method.
32145 * read-md.h (md_reader::md_reader): Add new param "compact".
32146 (md_reader::read_file): New method.
32147 (md_reader::is_compact): New accessor.
32148 (md_reader::read_name): Convert return type from void to file_location.
32149 (md_reader::read_name_or_nil): New decl.
32150 (md_reader::read_name_1): New decl.
32151 (md_reader::m_compact): New field.
32152 (noop_reader::noop_reader): Pass "false" for new "compact" param
32154 (rtx_reader::rtx_reader): Add new "compact" param.
32155 (rtx_reader::read_rtx_operand): Make virtual and convert return
32156 type from void to rtx.
32157 (rtx_reader::read_until): New decl.
32158 (rtx_reader::handle_any_trailing_information): New virtual function.
32159 (rtx_reader::postprocess): New virtual function.
32160 (rtx_reader::finalize_string): New virtual function.
32161 (rtx_reader::m_in_call_function_usage): New field.
32162 (rtx_reader::m_reuse_rtx_by_id): New field.
32163 * read-rtl-function.c: New file.
32164 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
32165 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
32166 (selftest::verify_three_block_rtl_cfg): New decl.
32167 * read-rtl-function.h: New file.
32168 * read-rtl.c: Potentially include config.h rather than bconfig.h.
32169 For host, include function.h, memmodel.h, and emit-rtl.h.
32170 (one_time_initialization): New function.
32171 (struct compact_insn_name): New struct.
32172 (compact_insn_names): New array.
32173 (find_code): Handle insn codes in compact dumps.
32174 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
32175 (bind_subst_iter_and_attr): Likewise.
32176 (add_condition_to_string): Likewise.
32177 (add_condition_to_rtx): Likewise.
32178 (apply_attribute_uses): Likewise.
32179 (add_current_iterators): Likewise.
32180 (apply_iterators): Likewise.
32181 (initialize_iterators): Guard usage of apply_subst_iterator with
32182 #ifdef GENERATOR_FILE.
32183 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
32184 (md_reader::read_mapping): Likewise.
32185 (add_define_attr_for_define_subst): Likewise.
32186 (add_define_subst_attr): Likewise.
32187 (read_subst_mapping): Likewise.
32188 (check_code_iterator): Likewise.
32189 (rtx_reader::read_rtx): Likewise. Move one-time initialization
32191 (one_time_initialization): New function.
32192 (rtx_reader::read_until): New method.
32193 (read_flags): New function.
32194 (parse_reg_note_name): New function.
32195 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
32196 Handle reuse_rtx ids.
32197 Wrap iterator lookup within #ifdef GENERATOR_FILE.
32198 Add parsing support for RTL dumps, mirroring the special-cases in
32199 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
32200 values, and calling handle_any_trailing_information.
32201 (rtx_reader::read_rtx_operand): Convert return type from void
32202 to rtx, returning return_rtx. Handle case 'e'. Call
32203 finalize_string on XSTR and XTMPL fields.
32204 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
32205 "(nil)" values were omitted. Call the postprocess vfunc on the
32207 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
32208 class ctor. Initialize m_in_call_function_usage. Call
32209 one_time_initialization.
32210 * rtl-tests.c (selftest::test_uncond_jump): Call
32211 set_new_first_and_last_insn.
32212 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
32213 * selftest-rtl.c: New file.
32214 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
32215 (selftest::get_insn_by_uid): New decl.
32216 * selftest-run-tests.c (selftest::run_tests): Call
32217 read_rtl_function_c_tests.
32218 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
32219 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
32222 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
32224 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
32225 operands in a special way. Assert that pos+len <= mode precision.
32227 2017-01-05 Jakub Jelinek <jakub@redhat.com>
32229 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
32230 3 argument Alias with unlimited for the negative form.
32231 (fno-vect-cost-model): Removed.
32233 2017-01-05 Martin Liska <mliska@suse.cz>
32235 * hsa-gen.c (gen_hsa_divmod): New function.
32236 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
32238 2017-01-05 Martin Liska <mliska@suse.cz>
32241 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
32244 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32246 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
32247 small constant length operands.
32249 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32251 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
32252 between loop iterations.
32254 2017-01-05 Martin Liska <mliska@suse.cz>
32257 * gimplify.c (gimplify_decl_expr): Compare to
32258 asan_poisoned_variables instread of checking flags.
32259 (gimplify_target_expr): Likewise.
32260 (gimplify_expr): Likewise.
32261 (gimplify_function_tree): Conditionally initialize
32262 asan_poisoned_variables.
32264 2017-01-04 Jeff Law <law@redhat.com>
32266 PR tree-optimizatin/78812
32267 * rtl.h (contains_mem_rtx_p): Prototype.
32268 * ifcvt.c (containts_mem_rtx_p): Move from here to...
32269 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
32270 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
32271 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
32272 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
32274 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32276 * input.c (assert_char_at_range): Default-initialize actual_range.
32278 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32280 * df-scan.c (df_ref_create_structure): Make regno unsigned,
32281 to match the caller.
32283 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32285 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
32286 insns after final jump in test to emit dummy move.
32288 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32290 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
32291 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
32293 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32295 * multiple_target.c (create_dispatcher_calls): Init e_next.
32296 * tree-ssa-loop-split.c (split_loop): Init border.
32297 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
32300 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
32305 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
32306 (altivec_register_operand): Do not return true if the operand
32307 contains a SUBREG mixing SImode and SFmode.
32308 (vsx_register_operand): Likewise.
32309 (vsx_reg_sfsubreg_ok): New predicate.
32310 (vfloat_operand): Do not return true if the operand contains a
32311 SUBREG mixing SImode and SFmode.
32312 (vint_operand): Likewise.
32313 (vlogical_operand): Likewise.
32314 (gpc_reg_operand): Likewise.
32315 (int_reg_operand): Likewise.
32316 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
32317 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
32318 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
32320 (rs6000_emit_move_si_sf_subreg): New helper function.
32321 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
32322 fixup SUBREGs involving SImode and SFmode.
32323 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
32324 numbers for the new peephole2 optimization.
32325 (peephole2 for SFmode unions): New peephole2 to optimize cases in
32326 the GLIBC math library that do AND/IOR/XOR operations on single
32327 precision floating point.
32328 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
32329 target macros to say whether we need to avoid SUBREGs mixing
32331 (TARGET_ALLOW_SF_SUBREG): Likewise.
32332 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
32333 (UNSPEC_SI_FROM_SF): Likewise.
32334 (iorxor): Change spacing.
32335 (and_ior_xor): New iterator for AND, IOR, and XOR.
32336 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
32337 (movdi_from_sf_zero_ext): Likewise.
32338 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
32339 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
32340 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
32341 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
32342 (fms<mode>4): Likewise.
32343 (fnma<mode>4): Likewise.
32344 (fnms<mode>4): Likewise.
32345 (nfma<mode>4): Likewise.
32346 (nfms<mode>4): Likewise.
32348 2017-01-04 Marek Polacek <polacek@redhat.com>
32351 * doc/invoke.texi: Document -Wpointer-compare.
32353 2017-01-04 Jakub Jelinek <jakub@redhat.com>
32355 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
32358 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
32359 descriptions for -gdwarf-5 and emit them as uleb128 instead of
32362 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
32365 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
32366 documentation of the powerpc_popcntb_ok attribute.
32367 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32368 code to issue warning messages if a requested CPU configuration is
32369 not supported by the binary (assembler and loader) toolchain.
32370 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
32371 made to define a built-in function that has been disabled.
32372 (paired_init_builtins): Add assertion to prevent ICE if attempt is
32373 made to define a built-in function that has been disabled.
32374 (altivec_init_builtins): Add comment explaining why definition
32375 of the DST built-in functions is not preceded by an assertion
32376 check. Add assertions to prevent ICE if attempts are made to
32377 define an altivec predicate or an abs* built-in function that has
32379 (htm_init_builtins): Add comment explaining why definition of the
32380 htm built-in functions is not preceded by an assertion check.
32382 2017-01-04 Jeff Law <law@redhat.com>
32384 PR tree-optimizatin/67955
32385 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
32386 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
32387 the points-to solution does not include pt_null. Use DECL_PT_UID
32390 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
32392 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
32393 Use gen_int_mode instead of gen_lopwart for const_int operands.
32395 2017-01-04 Jakub Jelinek <jakub@redhat.com>
32397 PR tree-optimization/71563
32398 * match.pd: Simplify X << Y into X if Y is known to be 0 or
32399 out of range value - has low bits known to be zero.
32401 2017-01-04 Alan Modra <amodra@gmail.com>
32403 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
32404 * configure: Regenerate.
32405 * config.in: Regenerate.
32407 2017-01-04 Jakub Jelinek <jakub@redhat.com>
32410 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
32411 a substring of the message, but strcmp with the whole message. Ifdef
32412 ENABLE_NLS, translate the message first using dgettext.
32414 2017-01-03 Jeff Law <law@redhat.com>
32416 PR tree-optimizatin/78856
32417 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
32418 (mark_threaded_blocks): Remove code to truncate thread paths that
32419 cross multiple loop headers. Instead invalidate the cached loop
32420 iteration information and handle case of a thread path walking
32421 into an irreducible region.
32423 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
32426 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
32427 assertions. Add support for doing the signbit if the IEEE 128-bit
32428 floating point value is in a GPR.
32429 * config/rs6000/rs6000.md (Fsignbit): Delete.
32430 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
32431 Update the length attribute if the value is in a GPR.
32432 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
32433 the sign or zero extension instruction, since the value is always 0/1.
32434 (signbit<mode>2_dm2): Delete using <Fsignbit>.
32437 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
32438 extracting SImode to a GPR register so that we can generate a
32439 store, limit the vector to be in a traditional Altivec register
32440 for the vextuwrx instruction.
32442 2017-01-03 Ian Lance Taylor <iant@google.com>
32444 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
32446 2017-01-03 Martin Sebor <msebor@redhat.com>
32448 PR tree-optimization/78696
32449 * gimple-ssa-sprintf.c (format_floating): Correct handling of
32450 precision. Use MPFR for %f for greater fidelity. Correct handling
32452 (pass_sprintf_length::compute_format_length): Set width and precision
32453 specified by asrerisk to void_node for vararg functions.
32454 (try_substitute_return_value): Adjust dump output.
32456 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
32458 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
32460 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
32462 * doc/invoke.texi (SPARC options): Document -mlra as the default.
32463 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
32464 -mlra/-mno-lra was passed to the compiler.
32466 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
32468 PR rtl-optimization/65618
32469 * emit-rtl.c (try_split): Move initialization of "before" and
32470 "after" to just before the call to emit_insn_after_setloc.
32472 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
32474 * doc/md.texi (Standard Names): Remove reference to Java frontend.
32476 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
32478 * dwarf2out.c (gen_enumeration_type_die): When
32479 -gno-strict-dwarf, add a DW_AT_encoding attribute.
32481 2017-01-03 Jakub Jelinek <jakub@redhat.com>
32483 PR tree-optimization/78965
32484 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
32485 Change first argument from const call_info & to call_info &. For %n
32486 set info.nowrite to false.
32488 PR middle-end/78901
32489 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
32490 possibly throwing calls.
32492 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
32493 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
32494 and fns handling, rather than in a separate case SSA_NAME.
32496 2017-01-02 Jeff Law <law@redhat.com>
32498 * config/darwin-driver.c (darwin_driver_init): Const-correctness
32499 fixes for first_period and second_period variables.
32501 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
32504 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
32505 (*insvqi_1): New insn pattern.
32506 (*insvqi_1_mem_rex64): Ditto.
32507 (*insvqi_2): Ditto.
32508 (*insvqi_3): Rename from *insvqi.
32510 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
32512 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
32514 * doc/cfg.texi (Edges): Remove reference to Java.
32515 (Maintaining the CFG): Ditto.
32517 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32519 PR middle-end/77674
32520 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
32521 transparent aliases.
32523 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32525 PR middle-end/77484
32526 * predict.def (PRED_CALL): Update hitrate.
32527 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
32528 * predict.c (tree_estimate_probability_bb): Split CALL predictor
32529 into direct/indirect/polymorphic variants.
32531 2017-01-01 Jakub Jelinek <jakub@redhat.com>
32533 Update copyright years.
32535 * gcc.c (process_command): Update copyright notice dates.
32536 * gcov-dump.c (print_version): Ditto.
32537 * gcov.c (print_version): Ditto.
32538 * gcov-tool.c (print_version): Ditto.
32539 * gengtype.c (create_file): Ditto.
32540 * doc/cpp.texi: Bump @copying's copyright year.
32541 * doc/cppinternals.texi: Ditto.
32542 * doc/gcc.texi: Ditto.
32543 * doc/gccint.texi: Ditto.
32544 * doc/gcov.texi: Ditto.
32545 * doc/install.texi: Ditto.
32546 * doc/invoke.texi: Ditto.
32548 Copyright (C) 2017 Free Software Foundation, Inc.
32550 Copying and distribution of this file, with or without modification,
32551 are permitted in any medium without royalty provided the copyright
32552 notice and this notice are preserved.