1 2017-10-31 Martin Liska <mliska@suse.cz>
3 * doc/gcov.texi: Document new option.
4 * gcov.c (print_usage): Likewise print it.
5 (process_args): Support the argument.
6 (format_count): New function.
7 (format_gcov): Use the function.
9 2017-10-31 Martin Liska <mliska@suse.cz>
11 * gcov.c (struct name_map): do not use typedef.
12 Define operator== and operator<.
13 (name_search): Remove.
15 (main): Do not allocate names.
16 (process_file): Add vertical space.
17 (generate_results): Use std::find.
18 (release_structures): Do not release memory.
19 (find_source): Use std::find.
21 2017-10-31 Martin Liska <mliska@suse.cz>
23 * gcov.c (struct line_info): Remove it's typedef.
24 (line_info::line_info): Add proper ctor.
25 (line_info::has_block): Do not use a typedef.
26 (struct source_info): Do not use typedef.
28 (get_cycles_count): Likewise.
29 (output_intermediate_file): Iterate via vector iterator.
30 (add_line_counts): Use std::vector methods.
31 (accumulate_line_counts): Likewise.
32 (output_lines): Likewise.
34 2017-10-31 Martin Liska <mliska@suse.cz>
36 * gcov.c (struct source_info): Remove typedef.
37 (source_info::source_info): Add proper ctor.
38 (accumulate_line_counts): Use struct, not it's typedef.
39 (output_gcov_file): Likewise.
40 (output_lines): Likewise.
41 (main): Do not allocate an array.
42 (output_intermediate_file): Use size of vector container.
43 (process_file): Resize the vector.
44 (generate_results): Do not preallocate, use newly added vector
46 (release_structures): Do not release sources.
47 (find_source): Use vector methods.
48 (add_line_counts): Do not use typedef.
50 2017-10-31 Martin Liska <mliska@suse.cz>
52 * doc/gcov.texi: Document that.
53 * gcov.c (add_line_counts): Mark lines with a non-executed
55 (output_line_beginning): Handle such lines.
56 (output_lines): Pass new argument.
57 (output_intermediate_file): Print it in intermediate format.
59 2017-10-31 Martin Liska <mliska@suse.cz>
61 * color-macros.h: New file.
62 * diagnostic-color.c: Factor out color related to macros to
64 * doc/gcov.texi: Document -k option.
65 * gcov.c (INCLUDE_STRING): Include string.h.
66 (print_usage): Add -k option.
67 (process_args): Parse it.
68 (pad_count_string): New function.
69 (output_line_beginning): Likewise.
70 (DEFAULT_LINE_START): New macro.
71 (output_lines): Support color output.
73 2017-10-31 Martin Liska <mliska@suse.cz>
76 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
77 their interaction with GCOV infrastructure.
78 * configure.ac: Add -fkeep-{inline,static}-functions to
80 * configure: Regenerate.
82 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
85 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
87 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
90 * config/rs6000/rs6000.md (allocate_stack): Force update interval
91 into a register if it does not fit into an immediate offset field.
93 2017-10-31 Olivier Hainque <hainque@adacore.com>
95 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
97 2017-10-31 Julia Koval <julia.koval@intel.com>
99 * config.gcc: Add gfniintrin.h.
100 * config/i386/gfniintrin.h: New.
101 * config/i386/i386-builtin-types.def
102 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
103 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
104 __builtin_ia32_vgf2p8affineinvqb_v32qi,
105 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
106 __builtin_ia32_vgf2p8affineinvqb_v16qi,
107 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
108 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
109 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
110 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
111 V64QI_FTYPE_V64QI_V64QI_INT): New types.
112 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
113 * config/i386/immintrin.h: Include gfniintrin.h.
114 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
116 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
118 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
120 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
122 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
123 (arm_ashldi3_1bit): Remove pattern.
124 (ashrdi3): Remove shift by 1 expansion.
125 (arm_ashrdi3_1bit): Remove pattern.
126 (lshrdi3): Remove shift by 1 expansion.
127 (arm_lshrdi3_1bit): Remove pattern.
128 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
129 cost of ashldi3 by 1.
130 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
131 (<shift>di3_neon): Likewise.
133 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
135 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
136 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
137 and (*aarch64_simd_mov<VQ:mode>).
138 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
140 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
141 attributes to match pattern alternative.
143 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
145 * config.gcc (powerpc*-*-*): Add emmintrin.h.
146 * config/rs6000/emmintrin.h: New file.
147 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
149 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
151 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
152 * (movdi_vfp_cortexa8): Remove pattern.
154 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
156 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
158 (Specific, alpha*-dec-osf5.1): Remove.
159 (Specific, mips-sgi-irix5): Remove.
160 (Specific, mips-sgi-irix6): Remove.
162 2017-10-30 Jakub Jelinek <jakub@redhat.com>
165 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
166 arguments to clear_bit_region_be.
168 2017-10-30 Jim Wilson <wilson@tuliptree.org>
170 * gimplify.c: Include memmodel.h.
172 2017-10-30 Martin Jambor <mjambor@suse.cz>
174 * omp-grid.c (grid_attempt_target_gridification): Also insert a
175 condition whether loop should be executed at all.
177 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
179 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
180 gimple folding of vec_madd() intrinsics.
181 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
182 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
183 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
185 2017-10-30 Richard Biener <rguenther@suse.de>
187 PR tree-optimization/82762
189 2017-10-23 Richard Biener <rguenther@suse.de>
191 PR tree-optimization/82129
193 2017-08-01 Richard Biener <rguenther@suse.de>
195 PR tree-optimization/81181
196 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
197 (compute_antic): ... end of iteration here.
199 2017-10-30 Joseph Myers <joseph@codesourcery.com>
201 * doc/invoke.texi (C Dialect Options): Document -std=c17,
202 -std=iso9899:2017 and -std=gnu17.
203 * doc/standards.texi (C Language): Document C17 support.
204 * doc/cpp.texi (Overview): Mention -std=c17.
205 (Standard Predefined Macros): Document C11 and C17 values of
206 __STDC_VERSION__. Do not refer to C99 support as incomplete.
207 * doc/extend.texi (Inline): Do not list individual options for
208 standards newer than C99.
209 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
211 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
214 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
216 * asan.c (asan_finish_file): Align asan globals array by shadow
219 2017-10-30 Jakub Jelinek <jakub@redhat.com>
222 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
223 (struct store_immediate_info): Add bitregion_start and bitregion_end
225 (store_immediate_info::store_immediate_info): Add brs and bre
226 arguments and initialize bitregion_{start,end} from those.
227 (struct merged_store_group): Add bitregion_start, bitregion_end,
228 align_base and mask fields. Drop unnecessary struct keyword from
229 struct store_immediate_info. Add do_merge method.
230 (clear_bit_region_be): Use memset instead of loop storing zeros.
231 (merged_store_group::do_merge): New method.
232 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
233 stores as long as the surrounding bitregions have no gaps.
234 (merged_store_group::merge_overlapping): Use do_merge.
235 (merged_store_group::apply_stores): Test that bitregion_{start,end}
236 is byte aligned, rather than requiring that start and width are
237 byte aligned. Drop unnecessary struct keyword from
238 struct store_immediate_info. Allocate and populate also mask array.
239 Make start of the arrays relative to bitregion_start rather than
240 start and size them according to bitregion_{end,start} difference.
241 (struct imm_store_chain_info): Drop unnecessary struct keyword from
242 struct store_immediate_info.
243 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
244 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
245 struct keyword from struct store_immediate_info.
246 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
247 between stores as long as the surrounding bitregions have no gaps.
249 (struct split_store): Add orig non-static data member.
250 (split_store::split_store): Initialize orig to false.
251 (find_constituent_stmts): Return store_immediate_info *, non-NULL
252 if there is exactly a single original stmt. Change stmts argument
253 to pointer from reference, if NULL, don't push anything to it. Add
254 first argument, use it to optimize skipping over orig stmts that
255 are known to be before bitpos already. Simplify.
256 (split_group): Return unsigned int count how many stores are or
257 would be needed rather than a bool. Add allow_unaligned argument.
258 Change split_stores argument from reference to pointer, if NULL,
259 only do a dry run computing how many stores would be produced.
260 Rewritten algorithm to use both alignment and misalign if
261 !allow_unaligned and handle bitfield stores with gaps.
262 (imm_store_chain_info::output_merged_store): Set start_byte_pos
263 from bitregion_start instead of start. Compute allow_unaligned
264 here, if true, do 2 split_group dry runs to compute which one
265 produces fewer stores and prefer aligned if equal. Punt if
266 new count is bigger or equal than original before emitting any
267 statements, rather than during that. Remove no longer needed
268 new_ssa_names tracking. Replace num_stmts with
269 split_stores.length (). Use 32-bit stack allocated entries
270 in split_stores auto_vec. Try to reuse original store lhs/rhs1
271 if possible. Handle bitfields with gaps.
272 (pass_store_merging::execute): Ignore bitsize == 0 stores.
273 Compute bitregion_{start,end} for the stores and construct
274 store_immediate_info with that. Formatting fixes.
276 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
279 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
280 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
282 2017-10-29 Jim Wilson <wilson@tuliptree.org>
284 * gimplify.c: Include tm_p.h.
286 * common.opt (gcoff): Re-add as ignored option.
287 (gcoff1, gcoff2, gcoff3): Likewise.
289 * Makefile.in (OBJS): Delete sdbout.o.
290 (GTFILES): Delete $(srcdir)/sdbout.c.
291 * debug.h: Delete sdb_debug_hooks.
292 * final.c: Delete sdbout.h include.
293 (final_scan_insn): Delete SDB_DEBUG check.
294 (rest_of_clean_state): Likewise.
295 * output.h: Delete sdb_begin_function_line.
298 * toplev.c: Delete sdbout.h include.
299 (process_options): Delete SDB_DEBUG check.
300 * tree-core.h (tree_type_common): Delete pointer field of
302 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
304 * tree.h (TYPE_SYMTAB_POINTER): Delete.
305 (TYPE_SYMTAB_IS_POINTER): Delete.
306 (TYPE_SYMTAB_IS_DIE): Renumber.
307 * xcoffout.c: Refer to former sdbout.c file.
308 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
310 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
311 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
312 (Debugging Options): Delete -gcoff.
313 (-gstabs): Delete SDB reference.
315 (-gcoff@var{level}): Delete.
316 * doc/passes.texi (Debugging information output): Delete SDB and
318 * doc/tm.texi: Regenerate.
319 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
320 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
322 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
323 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
324 (SDB_DEBUGGING_INFO): Delete.
325 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
326 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
327 * target.def (output_source_filename): Delete COFF reference.
329 * common.opt (gcoff): Delete.
330 (gxcoff+): Update Negative chain.
331 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
333 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
334 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
335 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
337 (expand_function_start): Change sdb reference to past tense.
338 (expand_function_end): Change sdb reference to past tense.
339 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
340 * opts.c (debug_type_names): Delete coff entry.
341 (common_handle_option): Delete OPT_gcoff case.
342 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
344 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
345 * config/cris/cris.h: Delete SDB reference in comment.
346 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
347 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
348 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
349 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
351 (ix86_expand_prologue): Likewise.
352 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
353 * config/ia64/ia64.h: Likewise.
354 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
355 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
357 * config/mmix/mmix.h: Likewise.
358 * config/nds32/nds32.c: Likewise.
359 * config/stormy/storym16.h: Likewise.
360 * config/visium/visium.h: Likewise.
361 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
363 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
365 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
366 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
367 FRAME_POINTER_REGNUM point at high end of local var area.
369 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
371 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
372 Move comment around. Do not reset best_edge for a copiable
373 destination if the copy would cause a partition change.
374 (better_edge_p): Remove redundant check.
376 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
378 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
380 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
382 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
383 for math functions that have _Float<N> and _Float<N>X variants.
384 (mathfn_built_in_2): Add support for math functions that have
385 _Float<N> and _Float<N>X variants.
386 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
387 (expand_builtin_mathfn_ternary): Add support for fma with
388 _Float<N> and _Float<N>X variants.
389 (expand_builtin): Likewise.
390 (fold_builtin_3): Likewise.
391 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
392 create math function _Float<N> and _Float<N>X variants as external
394 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
395 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
396 the __builtin_ prefix and if not strict ansi, without the prefix.
397 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
398 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
399 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
400 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
401 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
402 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
403 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
404 function signatures for fma _Float<N> and _Float<N>X variants.
405 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
406 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
407 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
408 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
409 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
410 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
411 * gencfn-macros.c (print_case_cfn): Add support for math functions
412 that have _Float<N> and _Float<N>X variants.
413 (print_define_operator_list): Likewise.
414 (fltfn_suffixes): Likewise.
416 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
417 for math functions that have _Float<N> and _Float<N>X variants.
418 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
419 and _Float<N>X variants.
420 (COPYSIGN): Likewise.
423 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
424 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
426 (integer_valued_read_call_p): Likewise.
427 * fold-const-call.c (fold_const_call_ss): Likewise.
428 (fold_const_call_sss): Add support for copysign, fmin, and fmax
429 _Float<N> and _Float<N>X variants.
430 (fold_const_call_ssss): Add support for fma _Float<N> and
432 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
433 support for copysign and fma _Float<N> and _Float<N>X variants.
434 (backprop::process_builtin_call_use): Likewise.
435 * tree-call-cdce.c (can_test_argument_range); Add support for
436 sqrt _Float<N> and _Float<N>X variants.
437 (edom_only_function): Likewise.
438 (get_no_error_domain): Likewise.
439 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
440 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
441 copysign _Float<N> and _Float<N>X variants.
442 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
443 handled by machine independent code.
445 * doc/cpp.texi (Common Predefined Macros): Document defining
446 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
447 fma _Float<N> and _Float<N>X variants.
449 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
452 * config/i386/i386-modes.def (CCFPU): Remove definition.
453 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
454 (ix86_cc_modes_compatible): Ditto.
455 (ix86_expand_carry_flag_compare): Ditto.
456 (ix86_expand_int_movcc): Ditto.
457 (ix86_expand_int_addcc): Ditto.
458 (ix86_reverse_condition): Ditto.
459 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
460 Return true/false for unordered/ordered fp comparisons.
461 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
462 (ix86_prepare_fp_compare_args): Update for rename.
463 (ix86_expand_fp_compare): Update for rename. Generate unordered
464 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
465 (ix86_expand_sse_compare_and_jump): Ditto.
466 * config/i386/predicates.md (fcmov_comparison_operator):
467 Remove CCFPU mode handling.
468 (ix86_comparison_operator): Ditto.
469 (ix86_carry_flag_operator): Ditto.
470 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
471 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
472 (*cmpu<mode>_cc_i387): Ditto.
473 (FPCMP): Remove mode iterator.
474 (unord): Remove mode attribute.
475 (unord_subst): New define_subst transformation
476 (unord): New define_subst attribute.
478 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
479 (*cmpi<unord>xf_i387): Ditto.
480 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
481 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
482 using unord_subst transformation.
483 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
484 (round_saeonly): Also handle CCFP mode.
485 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
486 Remove UNSPEC_SAHF unspec handling.
488 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
490 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
492 2017-10-27 Jeff Law <law@redhat.com>
494 * gimple-ssa-sprintf.c: Include domwalk.h.
495 (class sprintf_dom_walker): New class, derived from dom_walker.
496 (sprintf_dom_walker::before_dom_children): New function.
497 (struct call_info): Moved into sprintf_dom_walker class
498 (compute_formath_length, handle_gimple_call): Likewise.
499 (sprintf_length::execute): Call the dominator walker rather
500 than walking the statements.
502 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
503 gimple statement locations.
504 (check_array_bounds): Corresponding changes. Get the statement's
505 location directly from wi->stmt.
507 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
510 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
512 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
514 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
515 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
517 2017-10-27 Jakub Jelinek <jakub@redhat.com>
520 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
521 * config/i386/i386.c (maybe_get_pool_constant): Removed.
522 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
523 maybe_get_pool_constant.
524 * config/i386/predicates.md (zero_extended_scalar_load_operand):
527 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
529 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
530 2.26 caveat. Update gas and gld versions.
531 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
534 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
536 * cgraph.h (set_malloc_flag): Declare.
537 * cgraph.c (set_malloc_flag_1): New function.
538 (set_malloc_flag): Likewise.
539 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
540 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
542 (read_ipa_call_summary): Add support for reading is_return_callee.
543 (write_ipa_call_summary): Stream is_return_callee.
544 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
545 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
546 ipa-prop.h, ipa-fnsummary.h.
547 (pure_const_names): Change to static.
548 (malloc_state_e): Define.
549 (malloc_state_names): Define.
550 (funct_state_d): Add field malloc_state.
551 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
552 (check_retval_uses): New function.
553 (malloc_candidate_p): Likewise.
554 (analyze_function): Add support for malloc attribute.
555 (pure_const_write_summary): Stream malloc_state.
556 (pure_const_read_summary): Add support for reading malloc_state.
557 (dump_malloc_lattice): New function.
558 (propagate_malloc): New function.
559 (warn_function_malloc): New function.
560 (ipa_pure_const::execute): Call propagate_malloc and
562 (pass_local_pure_const::execute): Add support for malloc attribute.
563 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
564 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
566 2017-10-27 Martin Liska <mliska@suse.cz>
568 PR gcov-profile/82457
569 * doc/invoke.texi: Document that one needs a non-strict ISO mode
570 for fork-like functions to be properly instrumented.
572 2017-10-27 Richard Biener <rguenther@suse.de>
575 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
576 info when we redirected EH.
578 2017-10-26 Michael Collison <michael.collison@arm.com>
580 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
582 (<optab>_trunchf<GPI:mode>2: New pattern.
583 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
584 * config/aarch64/iterators.md (wv): New mode attribute.
585 (vf, VF): New mode attributes.
586 (vgp, VGP): New mode attributes.
587 (s): Update attribute with SImode and DImode prefixes.
589 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
591 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
592 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
593 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
594 (nios2_option_overide): Initialize it. Don't allow R0-relative
596 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
597 (nios2_symbolic_constant_p): Likewise.
598 (nios2_legitimate_address_p): Likewise.
599 (nios2_r0rel_section_name_p): New.
600 (nios2_symbol_ref_in_r0rel_data_p): New.
601 (nios2_emit_move_sequence): Handle r0rel_constant_p.
602 (r0rel_constant_p): New.
603 (nios2_print_operand_address): Handle r0rel_constant_p.
604 (nios2_cdx_narrow_form_p): Likewise.
605 * config/nios2/nios2.opt (mr0rel-sec=): New option.
606 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
607 (Nios II Options): Document -mr0rel-sec.
609 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
611 * config/nios2/nios2.c: Include xregex.h.
612 (nios2_gprel_sec_regex): New.
613 (nios2_option_overide): Initialize it. Don't allow GP-relative
615 (nios2_small_section_name_p): Check for regex match.
616 * config/nios2/nios2.opt (mgprel-sec=): New option.
617 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
618 (Nios II Options): Document -mgprel-sec.
620 2017-10-26 Jim Wilson <wilson@tuliptree.org>
622 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
624 2017-10-26 Tom de Vries <tom@codesourcery.com>
626 PR tree-optimization/82707
627 * gimple.c (gimple_copy): Fix unsharing of
628 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
630 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
632 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
633 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
634 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
635 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
636 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
637 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
638 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
639 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
640 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
641 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
642 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
643 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
644 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
645 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
646 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
647 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
648 _mm512_mask_cmpunord_ps_mask): New intrinsics.
650 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
652 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
654 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
655 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
656 warning to rs6000.c. Remove the Undocumented flag, since it has
658 (-mabi=ibmlongdouble): Likewise.
659 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
660 already set, set the default format for long double.
661 (rs6000_debug_reg_global): Print whether long double is IBM or
663 (rs6000_option_override_internal): Rework setting long double
664 format. Only warn if the user is changing the long double default
665 and they did not use -Wno-psabi.
666 * doc/invoke.texi (PowerPC options): Update the documentation for
667 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
669 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
670 Alan Hayward <alan.hayward@arm.com>
671 David Sherwood <david.sherwood@arm.com>
673 * rtl.h (wider_subreg_mode): New function.
674 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
675 rather than an unsigned int *.
676 * ira-color.c (regno_max_ref_width): Replace with...
677 (regno_max_ref_mode): ...this new variable.
678 (coalesced_pseudo_reg_slot_compare): Update accordingly.
679 Use wider_subreg_mode.
680 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
681 rather than an unsigned int *.
682 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
683 (process_alt_operands): Likewise.
684 (invariant_p): Likewise.
685 * lra-spills.c (assign_mem_slot): Likewise.
686 (add_pseudo_to_slot): Likewise.
687 * lra.c (collect_non_operand_hard_regs): Likewise.
688 (add_regs_to_insn_regno_info): Likewise.
689 * reload1.c (regno_max_ref_width): Replace with...
690 (regno_max_ref_mode): ...this new variable.
691 (reload): Update accordingly. Update call to
692 ira_sort_regnos_for_alter_reg.
693 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
694 (init_eliminable_invariants): Update to use regno_max_ref_mode.
695 (scan_paradoxical_subregs): Likewise.
697 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
699 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
700 (aarch64_frame): Add emit_frame_chain boolean.
701 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
702 Move eh_return case to aarch64_layout_frame.
703 (aarch64_layout_frame): Initialize emit_frame_chain.
704 (aarch64_expand_prologue): Use emit_frame_chain.
706 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
708 * config/aarch64/aarch64.c (aarch64_layout_frame):
709 Ensure LR is always stored at the bottom of the callee-saves.
710 Remove rarely used frame layout which saves callee-saves at top of
711 frame, so the store of LR can be used as a valid probe in all cases.
713 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
715 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
716 Improve unaligned TImode/TFmode base/offset split.
718 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
719 Alan Hayward <alan.hayward@arm.com>
720 David Sherwood <david.sherwood@arm.com>
722 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
723 * combine.c (find_single_use_1): Likewise.
724 (expand_field_assignment): Likewise.
725 (move_deaths): Likewise.
726 * lra-constraints.c (simplify_operand_subreg): Likewise.
727 (curr_insn_transform): Likewise.
728 * lra.c (collect_non_operand_hard_regs): Likewise.
729 (add_regs_to_insn_regno_info): Likewise.
730 * rtlanal.c (reg_referenced_p): Likewise.
731 (covers_regno_no_parallel_p): Likewise.
733 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
735 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
736 Don't print any bits outside the precision of the value.
737 * wide-int.cc (test_printing): Add some new tests.
739 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
741 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
742 supports -xbrace_comment option.
743 * configure: Regenerate.
744 * config.in: Regenerate.
745 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
746 (ASM_CPU_SPEC): Use it.
748 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
750 * target.def (static_rtx_alignment): New hook.
751 * targhooks.h (default_static_rtx_alignment): Declare.
752 * targhooks.c (default_static_rtx_alignment): New function.
753 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
754 * doc/tm.texi: Regenerate.
755 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
756 instead of targetm.constant_alignment. Remove call to
758 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
759 (cris_preferred_mininum_alignment): New function, split out from...
760 (cris_constant_alignment): ...here.
761 (cris_static_rtx_alignment): New function.
762 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
764 (ix86_constant_alignment): ...here.
765 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
766 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
767 (mmix_static_rtx_alignment): New function.
768 * config/spu/spu.c (spu_static_rtx_alignment): New function.
769 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
771 2017-10-26 Tamar Christina <tamar.christina@arm.com>
774 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
775 Add flag_trapping_math and flag_fp_int_builtin_inexact.
777 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
779 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
780 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
781 mark as a sign-extending load.
782 (local_pic_load_u): Define.
784 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
787 * fold-const.c (operand_equal_for_comparison_p): Also return true
788 if ARG0 is a simple variant of ARG1 with narrower precision.
789 (fold_ternary_loc): Always pass unstripped operands to the predicate.
791 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
793 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
795 * i386.h (processor_costs): Add gather_static, gather_per_elt,
796 scatter_static, scatter_per_elt.
797 * x86-tune-costs.h: Add new cost entries.
799 2017-10-25 Richard Biener <rguenther@suse.de>
801 * tree-ssa-sccvn.h (vn_eliminate): Declare.
802 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
803 class pass_fre): Move to ...
804 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
805 class pass_fre): ... here and adjust for statistics.
807 2017-10-25 Jakub Jelinek <jakub@redhat.com>
810 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
811 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
812 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
813 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
816 2017-10-25 Richard Biener <rguenther@suse.de>
818 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
819 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
820 eliminate_push_avail, eliminate_insert): Move inside...
821 (class eliminate_dom_walker): ... this class in preparation
823 (fini_eliminate): Remove by merging with ...
824 (eliminate): ... this function. Adjust for class changes.
825 (pass_pre::execute): Remove fini_eliminate call.
826 (pass_fre::execute): Likewise.
828 2017-10-24 Jakub Jelinek <jakub@redhat.com>
831 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
832 (VPERMI2, VPERMI2I): New mode iterators.
833 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
834 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
836 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
837 mode iterator. Remove 3 old define_insn patterns.
838 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
839 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
840 VPERMI2 mode iterator, remove the other two expanders.
841 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
842 to use VPERMI2 mode iterator, add another alternative for vpermi2*
843 instructions, remove the other two patterns.
844 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
845 mode iterator, remove the other two patterns.
846 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
847 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
848 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
849 and adjust argument order accordingly.
850 (ix86_expand_vec_perm): Adjust caller.
851 (expand_vec_perm_1): Likewise.
852 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
853 (expand_vec_perm_vpermt2_vpshub2): ... this.
854 (ix86_expand_vec_perm_const_1): Adjust caller.
855 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
858 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
859 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
860 (vec_shl_<mode>): Remove unused expander.
861 (avx512bw_<shift_insn><mode>3): New define_insn.
862 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
863 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
865 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
868 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
871 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
873 PR rtl-optimization/82396
874 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
875 (autopref_multipass_init): Simplify initialization.
876 (autopref_rank_data): Simplify sort order.
877 * gcc/sched-int.h (autopref_multipass_data_): Remove
878 multi_mem_insn_p, min_offset and max_offset.
880 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
883 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
884 Check special value of flag_omit_frame_pointer.
885 (aarch64_can_eliminate): Likewise.
886 (aarch64_override_options_after_change_1): Simplify handling of
887 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
889 2017-10-24 Richard Biener <rguenther@suse.de>
891 PR tree-optimization/82697
892 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
893 zero for conditional load and unconditional store.
895 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
897 * doc/install.texi: Document bootstrap-cet.
899 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
902 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
903 ENDBR instruction at function entrance if function is only
906 2017-10-24 Jakub Jelinek <jakub@redhat.com>
909 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
910 patterns to better describe from which operation the CF is computed.
911 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
912 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
913 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
914 is 0, use _0 suffixed expanders instead of emitting a comparison
917 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
919 * config/i386/i386.md(*movsf_internal, *movdf_internal):
920 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
922 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
925 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
926 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
927 * loop-iv.c (iv_get_reaching_def): Likewise.
928 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
929 variable is promoted and the partition contains undefined values.
931 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
933 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
935 (nios2_address_cost): Define.
936 (nios2_legitimize_address): Recognize (exp + constant) directly.
937 (TARGET_ADDRESS_COST): Define.
939 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
941 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
942 (nios2_symbolic_memory_operand_p): Declare.
943 (nios2_split_large_constant): Declare.
944 (nios2_split_symbolic_memory_operand): Declare.
945 * config/nios2/nios2.c: Adjust includes.
946 (nios2_symbolic_constant_allowed): New.
947 (nios2_symbolic_constant_p): New.
948 (nios2_plus_symbolic_constant_p): New.
949 (nios2_valid_addr_expr_p): Recognize addresses involving
951 (nios2_legitimate_address_p): Likewise, also LO_SUM.
952 (nios2_symbolic_memory_operand_p): New.
953 (nios2_large_constant_p): New.
954 (nios2_split_large_constant): New.
955 (nios2_split_plus_large_constant): New.
956 (nios2_split_symbolic_memory_operand): New.
957 (nios2_legitimize_address): Code refactoring. Handle addresses
958 involving symbolic constants.
959 (nios2_emit_move_sequence): Likewise.
960 (nios2_print_operand): Improve error output.
961 (nios2_print_operand_address): Handle LO_SUM.
962 (nios2_cdx_narrow_form_p): Likewise.
963 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
964 operands involving symbolic constants.
965 (movhi_internal, movsi_internal): Likewise.
966 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
967 (extendhisi2, extendqi<mode>2): Likewise.
969 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
971 * tree-pass.h (PROP_rtl_split_insns): Define.
972 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
974 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
976 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
978 2017-10-23 Jakub Jelinek <jakub@redhat.com>
981 * target.def (const_not_ok_for_debug_p): Default to
982 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
983 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
984 * targhooks.c (default_const_not_ok_for_debug_p): New function.
985 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
986 which targetm.const_not_ok_for_debug_p returned true.
987 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
989 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
991 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
992 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
993 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
995 (ix86_const_not_ok_for_debug_p): New function.
996 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
997 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
999 2017-10-23 David Malcolm <dmalcolm@redhat.com>
1002 * system.h: Conditionally include "unique-ptr.h" if
1003 INCLUDE_UNIQUE_PTR is defined.
1004 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
1005 of defining INCLUDE_UNIQUE_PTR before including "system.h".
1007 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
1009 * config/rl78/rl78.md: New define_expand "subdi3".
1011 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
1014 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
1015 DF_REF_INSN if DF_REF_INSN_INFO is false.
1017 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
1019 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
1020 xmm_move instead of sse_move.
1021 (sse_store_index): New function.
1022 (ix86_register_move_cost): Be more sensible about mismatch stall;
1023 model AVX moves correctly; make difference between sse->integer and
1025 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
1026 moves; make difference between SSE and AVX.
1027 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
1028 and zmm_move. Increase size of sse load and store tables;
1029 add unaligned load and store tables; add ssemmx_to_integer.
1030 * x86-tune-costs.h: Update all entries according to real
1031 move latencies from Agner Fog's manual and chip documentation.
1033 2017-10-23 Jakub Jelinek <jakub@redhat.com>
1036 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
1037 * config/i386/constraints.md (Wf): New constraint.
1038 * config/i386/i386.md (UNSPEC_SBB): New unspec.
1039 (cmp<dwi>_doubleword): Removed.
1040 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
1041 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
1042 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
1043 expand with cmp<dwi>_doubleword. For LTU and GEU use
1044 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
1046 * common.opt (gcolumn-info): Enable by default.
1047 * doc/invoke.texi (gcolumn-info): Document new default.
1049 2017-10-23 Richard Biener <rguenther@suse.de>
1051 PR tree-optimization/82672
1052 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
1053 Fold the stmt if we propagated into it.
1055 2017-10-23 Richard Biener <rguenther@suse.de>
1057 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
1058 (bitmap_remove_expr_from_set): ... this. All callers call this
1059 for non-constant values.
1060 (bitmap_set_subtract): Rename to...
1061 (bitmap_set_subtract_expressions): ... this. Adjust and
1063 (bitmap_set_contains_value): Remove superfluous check.
1064 (bitmap_set_replace_value): Inline into single caller ...
1065 (bitmap_value_replace_in_set): ... here and simplify.
1066 (dependent_clean): Merge into ...
1067 (clean): ... this using an overload. Adjust.
1068 (prune_clobbered_mems): Adjust.
1069 (compute_antic_aux): Likewise.
1070 (compute_partial_antic_aux): Likewise.
1072 2017-10-23 Richard Biener <rguenther@suse.de>
1074 PR tree-optimization/82129
1076 2017-08-01 Richard Biener <rguenther@suse.de>
1078 PR tree-optimization/81181
1079 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
1080 (compute_antic): ... end of iteration here.
1082 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1084 * target.def (starting_frame_offset): New hook.
1085 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
1086 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
1087 * doc/tm.texi.in: Regenerate.
1088 * hooks.h (hook_hwi_void_0): Declare.
1089 * hooks.c (hook_hwi_void_0): New function.
1090 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
1091 STARTING_FRAME_OFFSET.
1092 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
1093 * reload1.c (reload): Likewise.
1094 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
1095 instead of STARTING_FRAME_OFFSET.
1096 * function.c (try_fit_stack_local): Likewise.
1097 (assign_stack_local_1): Likewise
1098 (instantiate_virtual_regs): Likewise.
1099 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
1100 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
1101 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
1102 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
1103 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
1104 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
1105 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
1106 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
1107 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
1108 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
1109 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
1110 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
1111 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
1112 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
1113 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
1114 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
1115 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
1116 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
1117 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
1118 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
1119 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
1120 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
1121 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
1122 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
1123 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
1124 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
1125 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
1126 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
1127 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
1128 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
1129 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
1130 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
1131 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
1132 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
1133 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
1134 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
1135 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
1136 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
1137 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
1138 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
1139 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
1140 * config/avr/avr.c (avr_starting_frame_offset): Make static and
1141 return a HOST_WIDE_INT.
1142 (avr_builtin_setjmp_frame_value): Use it instead of
1143 STARTING_FRAME_OFFSET.
1144 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1145 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
1146 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
1148 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1149 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
1150 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
1151 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1152 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
1153 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
1154 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1155 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
1156 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
1157 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1158 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
1159 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
1161 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1162 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
1163 * config/mips/mips.c (mips_compute_frame_info): Refer to
1164 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
1165 (mips_starting_frame_offset): New function.
1166 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1167 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
1168 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
1169 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
1170 and return a HOST_WIDE_INT.
1171 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1172 (mmix_initial_elimination_offset): Refer to
1173 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
1174 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
1175 * config/pa/pa.c (pa_starting_frame_offset): New function.
1176 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
1177 (pa_expand_prologue): Likewise.
1178 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1179 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
1180 !FRAME_GROWS_DOWNWARD handling to...
1181 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1182 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
1183 !FRAME_GROWS_DOWNWARD handling to...
1184 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1185 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
1186 !FRAME_GROWS_DOWNWARD handling to...
1187 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1188 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
1190 (rs6000_starting_frame_offset): New function.
1191 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
1192 !FRAME_GROWS_DOWNWARD handling to...
1193 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1194 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
1195 !FRAME_GROWS_DOWNWARD handling to...
1196 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1197 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
1198 !FRAME_GROWS_DOWNWARD handling to...
1199 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1200 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
1201 (rs6000_starting_frame_offset): New function.
1202 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
1203 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
1204 * config/vax/vax.c (vax_starting_frame_offset): New function.
1205 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
1206 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1207 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
1208 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
1209 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1210 * system.h (STARTING_FRAME_OFFSET): Poison.
1212 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1214 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
1215 SCALAR_TYPE_MODE instead of TYPE_MODE.
1217 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1218 Alan Hayward <alan.hayward@arm.com>
1219 David Sherwood <david.sherwood@arm.com>
1221 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
1223 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1224 Alan Hayward <alan.hayward@arm.com>
1225 David Sherwood <david.sherwood@arm.com>
1227 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
1229 2017-10-23 Richard Biener <rguenther@suse.de>
1231 PR tree-optimization/82129
1232 * tree-ssa-pre.c (bitmap_set_and): Remove.
1233 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
1234 canonicalizing expressions in the set to those with lowest
1235 ID rather than taking that from the first edge.
1237 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1239 * combine.c (rtx_equal_for_field_assignment_p): Use
1240 byte_lowpart_offset.
1242 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1243 Alan Hayward <alan.hayward@arm.com>
1244 David Sherwood <david.sherwood@arm.com>
1246 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
1247 to a SUBREG_PROMOTED_VAR.
1249 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1250 Alan Hayward <alan.hayward@arm.com>
1251 David Sherwood <david.sherwood@arm.com>
1253 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
1254 (expand_debug_source_expr): Likewise.
1255 * combine.c (combine_simplify_rtx): Likewise.
1256 * cse.c (fold_rtx): Likewise.
1257 * optabs.c (expand_float): Likewise.
1258 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1259 (simplify_binary_operation_1): Likewise.
1261 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1262 Alan Hayward <alan.hayward@arm.com>
1263 David Sherwood <david.sherwood@arm.com>
1265 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
1266 (record_promoted_value): Likewise.
1267 * expr.c (expand_expr_real_2): Likewise.
1268 * ree.c (update_reg_equal_equiv_notes): Likewise.
1269 (combine_set_extension): Likewise.
1270 * rtlanal.c (low_bitmask_len): Likewise.
1271 * simplify-rtx.c (neg_const_int): Likewise.
1272 (simplify_binary_operation_1): Likewise.
1274 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1275 Alan Hayward <alan.hayward@arm.com>
1276 David Sherwood <david.sherwood@arm.com>
1278 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
1279 * regcprop.c (maybe_mode_change): Likewise.
1280 * reload1.c (alter_reg): Likewise.
1282 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1284 * inchash.h (inchash::hash::add_wide_int): New function.
1285 * lto-streamer-out.c (hash_tree): Use it.
1287 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1289 * inchash.h (inchash::hash::add_wide_int): Rename to...
1290 (inchash::hash::add_hwi): ...this.
1291 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
1292 (polymorphic_call_target_hasher::hash): Likewise.
1293 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
1294 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
1295 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
1296 * lto-streamer-out.c (hash_tree): Likewise.
1297 * optc-save-gen.awk: Likewise.
1298 * tree.c (add_expr): Likewise.
1300 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
1303 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
1304 for ordered inequality comparisons even with TARGET_IEEE_FP.
1306 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
1309 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
1310 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
1311 Expand with cmp<dwi>_doubleword.
1313 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1315 * extend.texi: Add x86 specific to 'nocf_check' attribute.
1316 List CET intrinsics.
1317 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
1318 specific to -fcf-protection option.
1320 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1322 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
1323 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
1324 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
1325 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
1326 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
1327 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
1328 (extra_objs): Add cet.o for Linux/x86 targets.
1329 (tmake_file): Add i386/t-cet for Linux/x86 targets.
1330 * config/i386/cet.c: New file.
1331 * config/i386/cetintrin.h: Likewise.
1332 * config/i386/t-cet: Likewise.
1333 * config/i386/cpuid.h (bit_SHSTK): New.
1334 (bit_IBT): Likewise.
1335 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
1336 pass IBT and SHSTK bits.
1337 * config/i386/i386-builtin-types.def
1338 (VOID_FTYPE_UNSIGNED_PVOID): New.
1339 (VOID_FTYPE_UINT64_PVOID): Likewise.
1340 * config/i386/i386-builtin.def: Add CET intrinsics.
1341 * config/i386/i386-c.c (ix86_target_macros_internal): Add
1342 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
1343 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
1344 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
1346 * config/i386/i386.c (rest_of_insert_endbranch): New.
1347 (pass_data_insert_endbranch): Likewise.
1348 (pass_insert_endbranch): Likewise.
1349 (make_pass_insert_endbranch): Likewise.
1350 (ix86_notrack_prefixed_insn_p): Likewise.
1351 (ix86_target_string): Add -mibt, -mshstk flags.
1352 (ix86_option_override_internal): Add flag_cf_protection
1354 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
1355 (ix86_print_operand): Add 'notrack' prefix output.
1356 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
1357 (ix86_expand_builtin): Expand CET intrinsics.
1358 (x86_output_mi_thunk): Add 'endbranch' instruction.
1359 * config/i386/i386.h (TARGET_IBT): New.
1360 (TARGET_IBT_P): Likewise.
1361 (TARGET_SHSTK): Likewise.
1362 (TARGET_SHSTK_P): Likewise.
1363 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
1364 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
1365 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
1366 (builtin_setjmp_setup): New pattern.
1367 (builtin_longjmp): Likewise.
1368 (rdssp<mode>): Likewise.
1369 (incssp<mode>): Likewise.
1370 (saveprevssp): Likewise.
1371 (rstorssp): Likewise.
1372 (wrss<mode>): Likewise.
1373 (wruss<mode>): Likewise.
1374 (setssbsy): Likewise.
1375 (clrssbsy): Likewise.
1376 (nop_endbr): Likewise.
1377 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
1379 * config/i386/immintrin.h: Include <cetintrin.h>.
1380 * config/i386/linux-common.h
1381 (file_end_indicate_exec_stack_and_cet): New prototype.
1382 (TARGET_ASM_FILE_END): New.
1384 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
1386 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
1387 latencies instead of having separate table; make difference between
1388 integer and float costs.
1389 * i386.h (processor_costs): Remove scalar_stmt_cost,
1390 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
1391 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
1393 * x86-tune-costs.h: Remove entries which has been removed in
1394 procesor_costs from all tables; make cond_taken_branch_cost
1395 and cond_not_taken_branch_cost COST_N_INSNS based.
1397 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
1399 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
1401 2017-10-20 Jakub Jelinek <jakub@redhat.com>
1403 * config/i386/i386.md (isa): Remove fma_avx512f.
1404 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
1405 <avx512>_fmadd_<mode>_mask3<round_name>,
1406 <avx512>_fmsub_<mode>_mask<round_name>,
1407 <avx512>_fmsub_<mode>_mask3<round_name>,
1408 <avx512>_fnmadd_<mode>_mask<round_name>,
1409 <avx512>_fnmadd_<mode>_mask3<round_name>,
1410 <avx512>_fnmsub_<mode>_mask<round_name>,
1411 <avx512>_fnmsub_<mode>_mask3<round_name>,
1412 <avx512>_fmaddsub_<mode>_mask<round_name>,
1413 <avx512>_fmaddsub_<mode>_mask3<round_name>,
1414 <avx512>_fmsubadd_<mode>_mask<round_name>,
1415 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
1416 (*vec_widen_umult_even_v16si<mask_name>,
1417 *vec_widen_smult_even_v16si<mask_name>): Likewise.
1418 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
1420 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1422 * extend.texi: Add 'nocf_check' documentation.
1423 * gimple.texi: Add second parameter to
1424 gimple_build_call_from_tree.
1425 * invoke.texi: Add -fcf-protection documentation.
1426 * rtl.texi: Add REG_CALL_NOTRACK documenation.
1428 2017-10-20 Richard Biener <rguenther@suse.de>
1430 PR tree-optimization/82473
1431 * tree-vect-loop.c (vectorizable_reduction): Properly get at
1432 the largest input type.
1434 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1436 * c-attribs.c (handle_nocf_check_attribute): New function.
1437 (c_common_attribute_table): Add 'nocf_check' handling.
1438 * gimple-parser.c: Add second argument NULL to
1439 gimple_build_call_from_tree.
1440 * attrib.c (comp_type_attributes): Check nocf_check attribute.
1441 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
1443 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
1444 * common.opt: Add fcf-protection flag.
1445 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
1446 * flag-types.h: Add enum cf_protection_level.
1447 * gimple.c (gimple_build_call_from_tree): Add second parameter.
1448 Add 'nocf_check' attribute propagation to gimple call.
1449 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
1450 (gimple_build_call_from_tree): Update prototype.
1451 (gimple_call_nocf_check_p): New function.
1452 (gimple_call_set_nocf_check): Likewise.
1453 * gimplify.c: Add second argument to gimple_build_call_from_tree.
1454 * ipa-icf.c: Add nocf_check attribute in statement hash.
1455 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
1456 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
1457 * toplev.c (process_options): Add flag_cf_protection handling.
1459 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
1461 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
1463 2017-10-20 Richard Biener <rguenther@suse.de>
1465 PR tree-optimization/82603
1466 * tree-if-conv.c (predicate_mem_writes): Make sure to only
1467 remove false predicated stores.
1469 2017-10-20 Richard Biener <rguenther@suse.de>
1471 * graphite-isl-ast-to-gimple.c
1472 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
1473 Remove return value and simplify, dump copied stmt after lhs
1475 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
1476 Reduce dump verbosity.
1477 (gsi_insert_earliest): Likewise.
1478 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
1479 * graphite.c (print_global_statistics): Adjust dumping.
1480 (print_graphite_scop_statistics): Likewise.
1481 (print_graphite_statistics): Do not dump loops here.
1482 (graphite_transform_loops): But here.
1484 2017-10-20 Nicolas Roche <roche@adacore.com>
1486 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
1487 * configure: Regenerate.
1489 2017-10-20 Jakub Jelinek <jakub@redhat.com>
1492 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
1493 functions when optimizing replace GIMPLE_RETURN stmts with
1494 calls to __builtin_unreachable ().
1497 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
1498 for -fsanitize=thread link of executables.
1499 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
1500 link of executables.
1503 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
1505 (<shift_insn><mode>3<mask_name>): Change the last of the 3
1506 define_insns for logical vector shifts to use VI248_AVX512BW
1507 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
1508 condition, useless isa and prefix attributes. Change the first
1509 2 of these define_insns to ...
1510 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
1511 define_insn for avx512vl.
1512 (<shift_insn><mode>3): ... and this, new define_insn without
1513 masking for non-avx512vl.
1516 * config/i386/sse.md (*andnot<mode>3,
1517 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
1518 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
1519 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
1520 not applied use empty suffix even for TARGET_AVX512VL.
1521 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
1522 is applied, supply evex,evex or evex,evex,evex instead of just
1525 2017-10-20 Julia Koval <julia.koval@intel.com>
1527 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
1528 (OPTION_MASK_ISA_GFNI_UNSET): New.
1529 (ix86_handle_option): Handle OPT_mgfni.
1530 * config/i386/cpuid.h (bit_GFNI): New.
1531 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
1532 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
1533 * config/i386/i386.c (ix86_target_string): Add -mgfni.
1534 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
1535 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
1536 * config/i386/i386.opt: Add mgfni.
1538 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
1540 * config/msp430/msp430.c (msp430_option_override): Disable
1541 -fdelete-null-pointer-checks.
1542 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
1544 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
1546 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
1547 of x87 and SSE instructions.
1549 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
1551 * asan.c (create_cond_insert_point): Do not update edge count.
1552 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
1553 (afdo_propagate_circuit): Likewise.
1554 (afdo_calculate_branch_prob): Likewise.
1555 (afdo_annotate_cfg): Likewise.
1556 * basic-block.h (struct edge_def): Remove count.
1557 (edge_def::count): New accessor.
1558 * bb-reorder.c (rotate_loop): Update.
1559 (find_traces_1_round): Update.
1560 (connect_traces): Update.
1561 (sanitize_hot_paths): Update.
1562 * cfg.c (unchecked_make_edge): Update.
1563 (make_single_succ_edge): Update.
1564 (check_bb_profile): Update.
1565 (dump_edge_info): Update.
1566 (update_bb_profile_for_threading): Update.
1567 (scale_bbs_frequencies_int): Update.
1568 (scale_bbs_frequencies_gcov_type): Update.
1569 (scale_bbs_frequencies_profile_count): Update.
1570 (scale_bbs_frequencies): Update.
1571 * cfganal.c (connect_infinite_loops_to_exit): Update.
1572 * cfgbuild.c (compute_outgoing_frequencies): Update.
1573 (find_many_sub_basic_blocks): Update.
1574 * cfgcleanup.c (try_forward_edges): Update.
1575 (try_crossjump_to_edge): Update
1576 * cfgexpand.c (expand_gimple_cond): Update
1577 (expand_gimple_tailcall): Update
1578 (construct_exit_block): Update
1579 * cfghooks.c (verify_flow_info): Update
1580 (redirect_edge_succ_nodup): Update
1581 (split_edge): Update
1582 (make_forwarder_block): Update
1583 (duplicate_block): Update
1584 (account_profile_record): Update
1585 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
1586 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
1587 * cfgloopmanip.c (scale_loop_profile): Update.
1589 (lv_adjust_loop_entry_edge): Update.
1590 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
1591 (force_nonfallthru_and_redirect): Update.
1592 (purge_dead_edges): Update.
1593 (rtl_flow_call_edges_add): Update.
1594 * cgraphunit.c (init_lowered_empty_function): Update.
1595 (cgraph_node::expand_thunk): Update.
1596 * gimple-pretty-print.c (dump_probability): Update.
1597 (dump_edge_probability): Update.
1598 * gimple-ssa-isolate-paths.c (isolate_path): Update.
1599 * haifa-sched.c (sched_create_recovery_edges): Update.
1600 * hsa-gen.c (convert_switch_statements): Update.
1601 * ifcvt.c (dead_or_predicable): Update.
1602 * ipa-inline-transform.c (inline_transform): Update.
1603 * ipa-split.c (split_function): Update.
1604 * ipa-utils.c (ipa_merge_profiles): Update.
1605 * loop-doloop.c (add_test): Update.
1606 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
1607 * lto-streamer-in.c (input_cfg): Update.
1608 (input_function): Update.
1609 * lto-streamer-out.c (output_cfg): Update.
1610 * modulo-sched.c (sms_schedule): Update.
1611 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
1612 * predict.c (maybe_hot_edge_p): Update.
1613 (unlikely_executed_edge_p): Update.
1614 (probably_never_executed_edge_p): Update.
1615 (dump_prediction): Update.
1616 (drop_profile): Update.
1617 (propagate_unlikely_bbs_forward): Update.
1618 (determine_unlikely_bbs): Update.
1619 (force_edge_cold): Update.
1620 * profile.c (compute_branch_probabilities): Update.
1621 * reg-stack.c (better_edge): Update.
1622 * shrink-wrap.c (handle_simple_exit): Update.
1623 * tracer.c (better_p): Update.
1624 * trans-mem.c (expand_transaction): Update.
1625 (split_bb_make_tm_edge): Update.
1626 * tree-call-cdce.c: Update.
1627 * tree-cfg.c (gimple_find_sub_bbs): Update.
1628 (gimple_split_edge): Update.
1629 (gimple_duplicate_sese_region): Update.
1630 (gimple_duplicate_sese_tail): Update.
1631 (gimple_flow_call_edges_add): Update.
1632 (insert_cond_bb): Update.
1633 (execute_fixup_cfg): Update.
1634 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
1635 * tree-complex.c (expand_complex_div_wide): Update.
1636 * tree-eh.c (lower_resx): Update.
1637 (unsplit_eh): Update.
1638 (cleanup_empty_eh_move_lp): Update.
1639 * tree-inline.c (copy_edges_for_bb): Update.
1640 (freqs_to_counts): Update.
1641 (copy_cfg_body): Update.
1642 * tree-ssa-dce.c (remove_dead_stmt): Update.
1643 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
1644 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
1645 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
1646 (unloop_loops): Update.
1647 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
1648 * tree-ssa-loop-split.c (connect_loops): Update.
1649 (split_loop): Update.
1650 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
1651 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
1652 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
1653 * tree-ssa-reassoc.c (branch_fixup): Update.
1654 * tree-ssa-tail-merge.c (replace_block_by): Update.
1655 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
1656 (compute_path_counts): Update.
1657 (update_profile): Update.
1658 (recompute_probabilities): Update.
1659 (update_joiner_offpath_counts): Update.
1660 (estimated_freqs_path): Update.
1661 (freqs_to_counts_path): Update.
1662 (clear_counts_path): Update.
1663 (ssa_fix_duplicate_block_edges): Update.
1664 (duplicate_thread_path): Update.
1665 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
1666 (case_bit_test_cmp): Update.
1667 (collect_switch_conv_info): Update.
1668 (gen_inbound_check): Update.
1669 (do_jump_if_equal): Update.
1670 (emit_cmp_and_jump_insns): Update.
1671 * tree-tailcall.c (decrease_profile): Update.
1672 (eliminate_tail_call): Update.
1673 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
1674 (vect_do_peeling): Update.
1675 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
1676 * ubsan.c (ubsan_expand_null_ifn): Update.
1677 (ubsan_expand_ptr_ifn): Update.
1678 * value-prof.c (gimple_divmod_fixed_value): Update.
1679 (gimple_mod_pow2): Update.
1680 (gimple_mod_subtract): Update.
1681 (gimple_ic): Update.
1682 (gimple_stringop_fixed_value): Update.
1684 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
1687 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
1689 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
1691 PR rtl-optimization/82395
1692 * ira-color.c (allocno_priority_compare_func): Fix comparison step
1693 based on non_spilled_static_chain_regno_p.
1695 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
1697 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
1698 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
1699 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
1701 2017-10-19 Martin Sebor <msebor@redhat.com>
1703 PR tree-optimization/82596
1704 * tree.c (array_at_struct_end_p): Handle STRING_CST.
1706 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
1708 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
1709 (get_mem_refs_of_builtin_call): Likewise.
1710 * builtins.c (expand_builtin_apply): Adjust call to
1711 allocate_dynamic_stack_space.
1712 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
1713 the third argument to allocate_dynamic_stack_space, otherwise -1.
1714 (expand_builtin): Deal with all alloca variants.
1715 (is_inexpensive_builtin): Likewise.
1716 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
1717 * calls.c (special_function_p): Deal with all alloca variants.
1718 (initialize_argument_information): Adjust call to
1719 allocate_dynamic_stack_space.
1720 (expand_call): Likewise.
1721 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
1722 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
1723 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
1724 use it for the stack usage computation.
1725 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
1726 * function.c (gimplify_parameters): Call build_alloca_call_expr.
1727 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
1728 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
1729 (in_loop_p): Remove first argument and useless check.
1730 (pass_walloca::execute): Remove useless test and adjust call to above.
1731 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
1732 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
1733 (gimplify_call_expr): Deal with all alloca variants.
1734 * hsa-gen.c (gen_hsa_alloca): Likewise.
1735 (gen_hsa_insns_for_call): Likewise.
1736 * ipa-pure-const.c (special_builtin_state): Likewise.
1737 * tree-chkp.c (chkp_build_returned_bound): Likewise.
1738 * tree-object-size.c (alloc_object_size): Likewise.
1739 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
1740 (call_may_clobber_ref_p_1): Likewise.
1741 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
1742 (ccp_fold_stmt): Likewise.
1743 (optimize_stack_restore): Likewise.
1744 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
1745 (mark_all_reaching_defs_necessary_1): Likewise.
1746 (propagate_necessity): Likewise.
1747 (eliminate_unnecessary_stmts): Likewise.
1748 * tree.c (build_common_builtin_nodes): Build
1749 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
1750 (build_alloca_call_expr): New function.
1751 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
1752 (CASE_BUILT_IN_ALLOCA): Likewise.
1753 (build_alloca_call_expr): Declare.
1754 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
1756 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
1759 * dwarf2out.c (new_die_raw): New static inline function.
1760 (new_die): Use it to create the DIE.
1761 (add_AT_external_die_ref): Likewise.
1762 (clone_die): Likewise.
1763 (clone_as_declaration): Likewise.
1764 (dwarf2out_vms_debug_main_pointer): Likewise.
1765 (base_type_die): Likewise. Remove early return for corner cases.
1766 Do not call add_pubtype on the DIE here.
1767 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
1768 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
1769 typedefs for base types with DW_AT_endianity. Make sure a DIE with
1770 native order exists for base types, attach the DIE manually and call
1771 add_pubtype on it. Do not equate a reverse order DIE to the type.
1773 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
1775 * config/arm/arm.c (align_ok_ldrd_strd): New function.
1776 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
1778 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
1780 2017-10-19 Jakub Jelinek <jakub@redhat.com>
1782 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
1783 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
1784 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
1785 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
1786 * opts.c (sanitizer_opts): Add builtin.
1787 * ubsan.c (instrument_builtin): New function.
1788 (pass_ubsan::execute): Call it.
1789 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
1790 * doc/invoke.texi: Document -fsanitize=builtin.
1792 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
1793 builtins, store max (log2 (align), 0) into uchar field instead of
1794 align into uptr field.
1795 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
1796 store uchar 0 field instead of uptr 0 field.
1797 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
1798 instead of passing one address of struct with 2 locations pass
1799 two addresses of structs with 1 location each.
1800 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
1801 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
1802 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
1803 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
1804 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
1805 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
1806 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
1807 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
1809 2017-10-19 Martin Liska <mliska@suse.cz>
1812 * file-find.c (remove_prefix): Remove.
1813 * file-find.h (remove_prefix): Likewise.
1814 * gcc-ar.c: Remove smartness of lookup.
1816 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
1818 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
1819 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
1820 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
1822 2017-10-19 Jakub Jelinek <jakub@redhat.com>
1825 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
1826 (setcc + and to xor + setcc): New peephole2.
1828 2017-10-19 Tom de Vries <tom@codesourcery.com>
1830 * doc/sourcebuild.texi (Test Directives, Variants of
1831 dg-require-support): Add dg-require-stack-size.
1833 2017-10-19 Martin Liska <mliska@suse.cz>
1836 * gimplify.c (gimplify_decl_expr): Do not instrument variables
1837 that have a large alignment.
1838 (gimplify_target_expr): Likewise.
1840 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
1842 PR rtl-optimization/82602
1843 * ira.c (rtx_moveable_p): Return false for volatile asm.
1845 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
1848 * config/i386/i386-modes.def (CCGZ): New CC mode.
1849 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
1850 * config/i386/predicates.md (ix86_comparison_operator):
1852 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
1853 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
1854 with double-word subtraction.
1855 (put_condition_code): Handle CCGZmode.
1857 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
1859 * wide-int.cc (debug (const wide_int &)): New.
1860 (debug (const wide_int *)): New.
1861 (debug (const widest_int &)): New.
1862 (debug (const widest_int *)): New.
1864 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
1867 * lra-constraints.c (curr_insn_transform): Use non-input operand
1868 instead of output one for matched reload.
1870 2017-10-18 Bin Cheng <bin.cheng@arm.com>
1872 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
1873 (tree-ssa-loop-ivopts.h): New header file.
1874 (struct builtin_info): New fields.
1875 (classify_builtin_1): Compute and record base and offset parts for
1876 memset builtin partition by calling strip_offset.
1877 (offset_cmp, fuse_memset_builtins): New functions.
1878 (finalize_partitions): Fuse adjacent memset partitions by calling
1880 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
1881 Expose the interface.
1882 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
1884 2017-10-18 Bin Cheng <bin.cheng@arm.com>
1886 PR tree-optimization/82574
1887 * tree-loop-distribution.c (find_single_drs): New parameter. Check
1888 that data reference must be executed exactly once per iteration
1889 against the outermost loop in nest.
1890 (classify_partition): Update call to above function.
1892 2017-10-18 Richard Biener <rguenther@suse.de>
1894 PR tree-optimization/82591
1895 * graphite.c (graphite_transform_loops): Move code gen message
1897 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1898 Here. Handle scop_to_isl_ast failing.
1899 (scop_to_isl_ast): Limit the number of ISL operations.
1901 2017-10-18 Richard Biener <rguenther@suse.de>
1903 * graphite-isl-ast-to-gimple.c
1904 (translate_isl_ast_to_gimple::set_rename): Simplify.
1905 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
1906 (graphite_copy_stmts_from_block): ... here.
1907 (copy_bb_and_scalar_dependences): Simplify.
1908 (add_parameters_to_ivs_params): Canonicalize.
1909 (generate_entry_out_of_ssa_copies): Simplify.
1910 * graphite-sese-to-poly.c (extract_affine_name): Simplify
1911 by passing in ISL dimension.
1912 (parameter_index_in_region_1): Rename to ...
1913 (parameter_index_in_region): ... this.
1914 (extract_affine): Adjust assert, pass down parameter index.
1915 (add_param_constraints): Use range-info when available.
1916 (build_scop_context): Adjust.
1917 * sese.c (new_sese_info): Adjust.
1918 (free_sese_info): Likewise.
1919 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
1920 Remove unused typedefs.
1921 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
1923 2017-10-18 Martin Liska <mliska@suse.cz>
1925 * combine.c (simplify_compare_const): Add gcc_fallthrough.
1927 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
1929 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
1930 (s390_sched_init): Do not reset s390_sched_state if we entered the
1931 current basic block via a fallthru edge and all others are unlikely.
1933 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
1935 * config/s390/s390.c (NUM_SIDES): New variable.
1936 (LONGRUNNING_THRESHOLD): New variable.
1937 (LATENCY_FACTOR): New variable.
1938 (s390_sched_score): Decrease score for long-running instructions on
1940 (s390_sched_variable_issue): Perform bookkeeping for long-running
1943 2017-10-18 Richard Biener <rguenther@suse.de>
1945 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
1946 Simplify with removal of the parameter rename map.
1947 (set_rename): Likewise.
1948 (should_copy_to_new_region): Likewise.
1949 (graphite_copy_stmts_from_block): Likewise.
1950 (copy_bb_and_scalar_dependences): Remove initialization of
1951 unused copied_bb_map.
1953 (copy_internal_parameters): Likewise.
1954 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
1955 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
1956 Use INTEGRAL_TYPE_P.
1957 (parameter_index_in_region_1): Rename to ...
1958 (assign_parameter_index_in_region): ... this. Assert we have
1959 a parameter we handle.
1960 (scan_tree_for_params): Adjust.
1961 * sese.h (parameter_rename_map_t): Remove.
1962 (struct sese_info_t): Remove unused parameter_rename_map and
1963 copied_bb_map members.
1964 * sese.c (new_sese_info): Adjust.
1965 (free_sese_info): Likewise.
1967 2017-10-18 Martin Liska <mliska@suse.cz>
1970 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
1971 on an abnormal edge.
1973 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
1975 * doc/invoke.texi (ffunction-sections and fdata-sections):
1978 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
1980 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
1981 the use statement can throw internally.
1983 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
1985 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
1986 any RTX present on the RHS of a SET.
1987 * compare-elim.c (try_eliminate_compare): Restore comment.
1989 2017-10-17 Jakub Jelinek <jakub@redhat.com>
1991 * langhooks.h (struct lang_hooks): Document that tree_size langhook
1992 may be also called on tcc_type nodes.
1993 * langhooks.c (lhd_tree_size): Likewise.
1995 2017-10-17 David Malcolm <dmalcolm@redhat.com>
1997 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
1998 format_warning_at_substring.
1999 (maybe_warn): Convert source_range * param to a location_t. Pass
2000 UNKNOWN_LOCATION rather than NULL to fmtwarn.
2001 (format_directive): Remove code to extract source_ranges and
2002 source_range * in favor of just a location_t.
2003 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
2005 * substring-locations.c (format_warning_va): Convert
2006 source_range * param to a location_t.
2007 (format_warning_at_substring): Likewise.
2008 * substring-locations.h (format_warning_va): Likewise.
2009 (format_warning_at_substring): Likewise.
2011 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
2013 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
2015 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
2016 and scatter/gather ops.
2018 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
2019 vec_gather_load and vec_scatter_store.
2020 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
2021 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
2022 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
2023 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
2024 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
2025 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
2027 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
2029 * reg-stack.c (compare_for_stack_reg): Add bool argument.
2030 Detect FTST instruction and handle its register pops. Only pop
2031 second operand if can_pop_second_op is true.
2032 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
2033 set can_pop_second_op to false in the compare_for_stack_reg call.
2035 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
2036 output_fp_compare for stack register operands.
2037 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
2038 instructions here. Do not emit stack register pops here. Assert
2039 that FCOMPP pops next to top stack register. Rewrite function.
2041 2017-10-17 Nathan Sidwell <nathan@acm.org>
2044 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
2045 use DECL_ASSEMBLER_NAME_RAW.
2048 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
2051 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
2052 Wilco Dijkstra <wilco.dijkstra@arm.com>
2054 * builtins.c (expand_builtin_update_setjmp_buf): Add a
2055 converstion to Pmode from the buf_addr.
2057 2017-10-17 Richard Biener <rguenther@suse.de>
2059 * graphite-dependences.c (scop_get_reads_and_writes): Change
2060 output parameters to references.
2062 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
2064 PR 71026/tree-optimization
2065 * fold-const.c (distribute_real_division): Removed.
2066 (fold_binary_loc): Remove calls to distribute_real_divison.
2068 2017-10-17 Richard Biener <rguenther@suse.de>
2070 * graphite-scop-detection.c
2071 (scop_detection::stmt_has_simple_data_refs_p): Always use
2072 the full nest as region.
2073 (try_generate_gimple_bb): Likewise.
2074 * sese.c (scalar_evolution_in_region): Simplify now that
2075 SCEV can handle instantiation in regions.
2076 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
2077 in the non-loop part of a function if requested.
2079 2017-10-17 Richard Biener <rguenther@suse.de>
2081 PR tree-optimization/82563
2082 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
2084 (graphite_regenerate_ast_isl): Call it.
2085 * graphite-scop-detection.c (build_scops): Remove entry edge split.
2087 2017-10-17 Jakub Jelinek <jakub@redhat.com>
2089 PR tree-optimization/82549
2090 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
2091 Formatting fixes. Instead of calling make_bit_field_ref with negative
2094 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
2096 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
2097 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
2098 _mm_maskz_reduce_ss): New.
2099 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
2100 __builtin_ia32_reducess_mask): Ditto..
2101 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
2102 * config/i386/sse.md (reduces<mode>): Renamed to ...
2103 (reduces<mode><mask_scalar_name>): ... this.
2104 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
2106 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
2107 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
2109 2017-10-16 David Malcolm <dmalcolm@redhat.com>
2111 * Makefile.in (OBJS): Add unique-ptr-tests.o.
2112 * selftest-run-tests.c (selftest::run_tests): Call
2113 selftest::unique_ptr_tests_cc_tests.
2114 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
2115 * unique-ptr-tests.cc: New file.
2117 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
2120 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
2122 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
2123 (make_hard_regno_born, make_hard_regno_dead): Update
2124 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
2126 2017-10-16 Jeff Law <law@redhat.com>
2128 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
2130 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
2132 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
2134 2017-10-16 Olivier Hainque <hainque@adacore.com>
2136 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
2137 with_cpu if we were configured for an e500v2 target cpu name.
2139 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
2141 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
2142 * doc/invoke.texi: Document +nodsp as a valid extension for
2145 2017-10-16 Martin Liska <mliska@suse.cz>
2147 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
2148 (test_set_range): Likewise.
2149 (test_range_functions): Rename to ...
2150 (test_bit_in_range): ... this.
2151 (sbitmap_c_tests): Add new test.
2153 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2155 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
2157 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
2158 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
2160 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2162 * config/aarch64/aarch64-builtins.c
2163 (aarch64_types_quadopu_lane_qualifiers): New.
2164 (TYPES_QUADOPU_LANE): New.
2165 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
2166 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
2167 (aarch64_<sur>dot_laneq<vsi2qi>): New.
2168 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
2169 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
2170 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
2171 (Vdottype, DOTPROD): New.
2172 (sur): Add SDOT and UDOT.
2174 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2176 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
2177 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
2178 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
2180 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
2181 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
2182 Enable TARGET_DOTPROD.
2183 (cortex-a75.cortex-a55): Likewise.
2184 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
2186 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2188 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
2189 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
2191 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
2193 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
2194 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
2195 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
2196 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
2197 * config/arm/types.md (neon_dot, neon_dot_q): New.
2198 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
2200 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2202 * config/arm/arm.h (TARGET_DOTPROD): New.
2203 * config/arm/arm.c (arm_arch_dotprod): New.
2204 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
2205 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
2206 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
2207 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
2208 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
2209 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
2210 * doc/invoke.texi (armv8.2-a): Document dotprod
2212 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
2214 * i386.c (ix86_vec_cost): New function.
2215 (ix86_rtx_costs): Handle vector operations better.
2216 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
2217 * x86-tune-costs.h: Add new costs to all tables.
2219 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
2221 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
2223 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
2224 divsd, sqrtss and sqrtsd
2225 * x86-tune-costs.h: Add new entries to all costs.
2226 (znver1_cost): Fix to match real instruction latencies.
2228 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2229 Michael Collison <michael.collison@arm.com>
2231 * compare-elim.c: Include emit-rtl.h.
2232 (can_merge_compare_into_arith): New function.
2233 (try_validate_parallel): Likewise.
2234 (try_merge_compare): Likewise.
2235 (try_eliminate_compare): Call the above when no previous clobber
2237 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
2240 2017-10-14 Jakub Jelinek <jakub@redhat.com>
2244 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
2245 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
2249 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
2250 to be any operand_equal_p operands. For & (B - 1) require
2251 B to be power of 2. Recognize
2252 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
2254 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
2257 * optabs.c (expand_memory_blockage): Fix call of
2258 targetm.have_memory_blockage.
2260 2017-10-14 Jakub Jelinek <jakub@redhat.com>
2263 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
2264 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
2265 objects to extra_objs instead of overwriting it.
2267 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
2269 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
2270 Use any_fp_register_operand as operand[3] predicate. Simplify
2271 equality test for operands[2] and operands[4] memory location.
2272 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
2273 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
2274 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
2275 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
2276 any_fp_register_operand as operand[1] predicate. Simplify
2277 equality test for operands[0] and operands[3] memory location.
2278 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
2279 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
2280 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
2282 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
2284 * target-insns.def: Add memory_blockage.
2285 * optabs.c (expand_memory_blockage): New function.
2286 (expand_asm_memory_barrier): Rename ...
2287 (expand_asm_memory_blockage): ... to this.
2288 (expand_mem_thread_fence): Call expand_memory_blockage
2289 instead of expand_asm_memory_barrier.
2290 (expand_mem_singnal_fence): Ditto.
2291 (expand_atomic_load): Ditto.
2292 (expand_atomic_store): Ditto.
2293 * doc/md.texi (Standard Pattern Names For Generation):
2294 Document memory_blockage instruction pattern.
2296 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
2298 * config/rl78/rl78.c (rl78_emit_libcall): New function.
2299 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
2300 * config/rl78/rl78.md: New define_expand "adddi3".
2302 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
2304 * cfghooks.c (verify_flow_info): Disable check that all probabilities
2307 2017-10-13 Jeff Law <law@redhat.com>
2309 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
2311 2017-10-13 Jakub Jelinek <jakub@redhat.com>
2314 * internal-fn.c (expand_mul_overflow): If both operands have
2315 the same highpart of -1 or 0 and the topmost bit of lowpart
2316 is different, overflow is if res <= 0 rather than res < 0.
2318 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
2320 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
2321 TARGET_P9_VECTOR code for unaligned_load case.
2323 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
2325 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
2327 2017-10-13 Nathan Sidwell <nathan@acm.org>
2329 * tree-core.h (tree_contains_struct): Make bool.
2330 * tree.c (tree_contains_struct): Likewise.
2331 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
2332 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
2333 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
2334 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
2335 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
2337 2017-10-13 Richard Biener <rguenther@suse.de>
2339 * graphite-isl-ast-to-gimple.c
2340 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
2341 parameters and dominance check.
2342 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
2343 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
2344 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
2345 Do not update SSA form here or do intermediate IL verification.
2346 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
2347 (graphite_initialize): Remove check on the number of loops in
2348 the function and inline into graphite_transform_loops.
2349 (graphite_finalize): Inline into graphite_transform_loops.
2350 (graphite_transform_loops): Perform SSA update and IL verification
2352 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
2354 2017-10-13 Richard Biener <rguenther@suse.de>
2356 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
2357 graphite_expression_type_precision): Avoid global constructor
2359 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
2360 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
2361 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
2362 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
2363 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
2364 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
2366 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
2369 * config/i386/i386.h (ix86_red_zone_size): New.
2370 * config/i386/i386.md (push peephole2s): Replace
2371 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
2373 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2374 Alan Hayward <alan.hayward@arm.com>
2375 David Sherwood <david.sherwood@arm.com>
2377 * combine.c (can_change_dest_mode): Reject changes in
2378 REGMODE_NATURAL_SIZE.
2380 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2381 Alan Hayward <alan.hayward@arm.com>
2382 David Sherwood <david.sherwood@arm.com>
2384 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
2385 (expand_debug_source_expr): Likewise.
2386 * combine.c (combine_simplify_rtx): Likewise.
2387 * cse.c (fold_rtx): Likewise.
2388 * fwprop.c (canonicalize_address): Likewise.
2389 * targhooks.c (default_shift_truncation_mask): Likewise.
2391 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2392 Alan Hayward <alan.hayward@arm.com>
2393 David Sherwood <david.sherwood@arm.com>
2395 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
2396 (widened_mode): Likewise.
2397 (expand_unop): Likewise.
2398 * ree.c (transform_ifelse): Likewise.
2399 (merge_def_and_ext): Likewise.
2400 (combine_reaching_defs): Likewise.
2401 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2403 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2404 Alan Hayward <alan.hayward@arm.com>
2405 David Sherwood <david.sherwood@arm.com>
2407 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
2408 * combine.c (gen_lowpart_for_combine): Likewise.
2409 * dwarf2out.c (rtl_for_decl_location): Likewise.
2410 * final.c (alter_subreg): Likewise.
2411 * rtlhooks.c (gen_lowpart_general): Likewise.
2412 (gen_lowpart_if_possible): Likewise.
2414 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2415 Alan Hayward <alan.hayward@arm.com>
2416 David Sherwood <david.sherwood@arm.com>
2418 * calls.c (expand_call): Use subreg_lowpart_offset.
2419 * cse.c (cse_insn): Likewise.
2420 * regcprop.c (copy_value): Likewise.
2421 (copyprop_hardreg_forward_1): Likewise.
2423 2017-10-13 Jakub Jelinek <jakub@redhat.com>
2426 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
2427 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
2428 =Q constraints to +Q and into insn condition add check
2429 that operands[0] and operands[1] are equal.
2430 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
2431 =Q constraints to +Q and into insn condition add check
2432 that operands[0] is equal to either operands[1] or operands[2].
2435 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
2436 instead of handling MINUS_EXPR twice (once for each argument),
2437 canonicalize operand order and handle just once, use rtype where
2438 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
2441 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
2442 any values of __C while still being pattern recognizable as a simple
2445 2017-10-13 Richard Biener <rguenther@suse.de>
2447 PR tree-optimization/82451
2449 2017-10-02 Richard Biener <rguenther@suse.de>
2451 PR tree-optimization/82355
2452 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
2453 a mapping for the enclosing loop but avoid generating one for
2455 (copy_bb_and_scalar_dependences): Remove premature codegen
2456 error on PHIs in blocks duplicated into multiple places.
2457 * graphite-scop-detection.c
2458 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
2459 in the region use it as loop and nest to analyze the DR in.
2460 (try_generate_gimple_bb): Likewise.
2461 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
2462 (add_loop_constraints): For blocks in a loop not in the region
2463 create a dimension with a single iteration.
2464 * sese.h (gbb_loop_at_index): Remove assert.
2466 * cfgloop.c (loop_preheader_edge): For the loop tree root
2467 return the single successor of the entry block.
2468 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
2469 Reset the SCEV hashtable and niters.
2470 * graphite-scop-detection.c
2471 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
2472 assert that we only have POLYNOMIAL_CHREC that vary in loops
2473 contained in the region.
2474 (scop_detection::graphite_can_represent_expr): Adjust.
2475 (scop_detection::stmt_has_simple_data_refs_p): For loops
2476 not in the region set loop to NULL. The nest is now the
2477 entry edge to the region.
2478 (try_generate_gimple_bb): Likewise.
2479 * sese.c (scalar_evolution_in_region): Adjust for
2480 instantiate_scev change.
2481 * tree-data-ref.h (graphite_find_data_references_in_stmt):
2482 Make nest parameter the edge into the region.
2483 (create_data_ref): Likewise.
2484 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
2485 entry edge into a region and adjust instantiate_scev calls.
2486 (create_data_ref): Likewise.
2487 (graphite_find_data_references_in_stmt): Likewise.
2488 (find_data_references_in_stmt): Pass the loop preheader edge
2489 from the nest argument.
2490 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
2491 parameter the edge into the region.
2492 (instantiate_parameters): Use the loop preheader edge as entry.
2493 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
2495 (get_instantiated_value_entry): Make instantiate_below parameter
2496 the edge into the region.
2497 (instantiate_scev_name): Likewise. Adjust dominance checks,
2498 when we cannot use loop-based instantiation instantiate by
2499 walking use-def chains.
2500 (instantiate_scev_poly): Adjust.
2501 (instantiate_scev_binary): Likewise.
2502 (instantiate_scev_convert): Likewise.
2503 (instantiate_scev_not): Likewise.
2504 (instantiate_array_ref): Remove.
2505 (instantiate_scev_3): Likewise.
2506 (instantiate_scev_2): Likewise.
2507 (instantiate_scev_1): Likewise.
2508 (instantiate_scev_r): Do not blindly handle N-operand trees.
2509 Do not instantiate array-refs. Handle all constants and invariants.
2510 (instantiate_scev): Make instantiate_below parameter
2511 the edge into the region.
2512 (resolve_mixers): Use the loop preheader edge for the region
2513 parameter to instantiate_scev_r.
2514 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
2516 2017-10-13 Richard Biener <rguenther@suse.de>
2518 PR tree-optimization/82525
2519 * graphite-isl-ast-to-gimple.c
2520 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
2522 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
2523 Fail code generation when we cannot represent the isl integer.
2524 (binary_op_to_tree): Elide modulo operations that are no-ops
2525 in the type we code generate. Remove now superfluous code
2528 2017-10-13 Richard Biener <rguenther@suse.de>
2530 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
2531 (scop_detection::harmful_loop_in_region): Remove premature
2532 IV type restriction.
2533 (scop_detection::graphite_can_represent_scev): We can handle
2534 pointer IVs just fine.
2536 2017-10-13 Alan Modra <amodra@gmail.com>
2538 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
2539 "Clobbers and Scratch Registers". Add paragraph on
2540 alternative to clobbers for scratch registers and OpenBLAS
2543 2017-10-13 Alan Modra <amodra@gmail.com>
2545 * doc/extend.texi (Clobbers): Correct vax example. Delete old
2546 example of a memory input for a string of known length. Move
2547 commentary out of table. Add a number of new examples
2548 covering array memory inputs.
2550 2017-10-12 Martin Liska <mliska@suse.cz>
2552 PR tree-optimization/82493
2553 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
2554 (test_range_functions): New function.
2555 (sbitmap_c_tests): Likewise.
2556 * selftest-run-tests.c (selftest::run_tests): Run new tests.
2557 * selftest.h (sbitmap_c_tests): New function.
2559 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
2561 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
2563 * config/rs6000/amo.h: Fix spacing issue.
2565 2017-10-12 Jakub Jelinek <jakub@redhat.com>
2568 * config/i386/i386.md (*ashl<mode>3_mask_1,
2569 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
2570 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
2573 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
2575 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
2576 (profile_probability): Set max_probability
2577 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
2579 * profile-count.c (profile_probability::differs_from_p): Do not
2580 rely on max_probaiblity == 10000
2582 2017-10-12 Jeff Law <law@redhat.com>
2584 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
2587 2017-10-12 Martin Sebor <msebor@redhat.com>
2591 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
2592 (handle_alias_pairs): Call it.
2593 * common.opt (-Wattribute-alias): New option.
2594 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
2595 * doc/invoke.texi (-Wattribute-alias): Document.
2597 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
2600 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
2602 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
2604 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
2605 (make_hard_regno_born, make_hard_regno_dead): Update
2606 bb_killed_pseudos and bb_gen_pseudos.
2608 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
2610 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
2612 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
2614 * config/alpha/alpha.c (alpha_split_conditional_move):
2615 Use std::swap instead of manually swapping.
2616 (alpha_stdarg_optimize_hook): Ditto.
2617 (alpha_canonicalize_comparison): Ditto.
2619 2017-10-12 Bin Cheng <bin.cheng@arm.com>
2621 * tree-loop-distribution.c (struct builtin_info): New struct.
2622 (struct partition): Refactor fields into struct builtin_info.
2623 (partition_free): Free struct builtin_info.
2624 (build_size_arg_loc, build_addr_arg_loc): Delete.
2625 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
2626 information from struct builtin_info.
2627 (find_single_drs): New function refactored from classify_partition.
2628 Also moved builtin validity checks to this function.
2629 (compute_access_range, alloc_builtin): New functions.
2630 (classify_builtin_st, classify_builtin_ldst): New functions.
2631 (classify_partition): Refactor code into functions find_single_drs,
2632 classify_builtin_st and classify_builtin_ldst.
2633 (distribute_loop): Don't do runtime alias check when distributing
2635 (find_seed_stmts_for_distribution): New function.
2636 (pass_loop_distribution::execute): Refactor code finding seed
2637 stmts into above function. Support distribution for the innermost
2638 two-level loop nest. Adjust dump information.
2640 2017-10-12 Bin Cheng <bin.cheng@arm.com>
2642 * tree-loop-distribution.c: Adjust the general comment.
2643 (NUM_PARTITION_THRESHOLD): New macro.
2644 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
2645 (classify_partition): Skip builtin pattern of loop nest's inner loop.
2646 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
2647 in call to build_partition_graph.
2648 (finalize_partitions): New parameter. Make loop distribution more
2649 conservative by fusing more partitions.
2650 (distribute_loop): Don't do runtime alias check in case of loop nest
2652 (find_seed_stmts_for_distribution): New function.
2653 (prepare_perfect_loop_nest): New function.
2654 (pass_loop_distribution::execute): Refactor code finding seed stmts
2655 and loop nest into above functions. Support loop nest distribution.
2656 Adjust dump information accordingly.
2658 2017-10-12 Bin Cheng <bin.cheng@arm.com>
2660 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
2661 and set PTYPE_SEQUENTIAL for merged partition.
2663 2017-10-12 Richard Biener <rguenther@suse.de>
2665 PR tree-optimization/69728
2667 2017-09-19 Richard Biener <rguenther@suse.de>
2669 PR tree-optimization/69728
2670 * graphite-sese-to-poly.c (schedule_error): New global.
2671 (add_loop_schedule): Handle empty domain by failing the
2673 (build_original_schedule): Handle schedule_error.
2675 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
2676 domain by returning an unchanged schedule.
2678 2017-10-12 Jakub Jelinek <jakub@redhat.com>
2680 * genrecog.c (validate_pattern): For VEC_SELECT verify that
2681 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
2683 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
2685 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
2688 2017-10-12 Jakub Jelinek <jakub@redhat.com>
2691 * expr.c (store_field): Don't optimize away bitsize == 0 store
2692 from CALL_EXPR with addressable return type.
2694 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
2696 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
2697 * config/rs6000/rs6000.md (sel): Delete mode attribute.
2698 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
2699 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
2700 TARGET_ISEL instead of TARGET_ISEL<sel>.
2702 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
2704 * config/rs6000/rs6000.c
2705 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
2707 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
2709 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
2711 (reg_or_cint_operand): Fix comment.
2712 (reg_or_zero_operand): New predicate.
2713 * config/rs6000/rs6000-protos.h (output_isel): Delete.
2714 * config/rs6000/rs6000.c (output_isel): Delete.
2715 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
2716 instead of reg_or_cint_operand. Output instruction directly (not via
2718 (isel_unsigned_<mode>): Ditto.
2719 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
2720 gpc_reg_operand. Add an instruction alternative for this. Output
2721 instruction directly.
2722 (*isel_reversed_unsigned_<mode>): Ditto.
2724 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
2726 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
2727 (TARGET_CANONICALIZE_COMPARISON): Define.
2729 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
2732 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
2733 Check whether the dest is REG before adding REG_EQUIV note.
2735 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
2738 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
2740 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
2741 (make_hard_regno_born, make_hard_regno_dead): Update
2742 bb_killed_pseudos and bb_gen_pseudos.
2744 2017-10-11 Nathan Sidwell <nathan@acm.org>
2746 * incpath.h (enum incpath_kind): Name enum, prefix values.
2747 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
2748 * incpath.c (heads, tails): Use INC_MAX.
2749 (add_env_var_paths, add_standard_paths): Use incpath_kind.
2750 (merge_include_chains, split_quote_chain,
2751 register_include_chains): Update incpath_kind names.
2752 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
2753 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
2755 (add_framework_path, darwin_register_objc_includes): Likewise.
2756 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
2758 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
2760 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
2761 Do not use float_operator operator predicate.
2762 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
2763 * config/i386/predicates.md (float_operator): Remove predicate.
2765 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
2767 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
2768 (*jccxf_i387): Ditto.
2769 (*jcc<mode>_i387): Ditto.
2770 (*jccu<mode>_i387): Ditto.
2771 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
2772 (*jcc_*_i387 splitters): Remove.
2773 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
2774 * config/i386/i386.c (ix86_split_fp_branch): Remove.
2775 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
2778 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
2780 * profile-count.h (slow_safe_scale_64bit): New function.
2781 (safe_scale_64bit): New inline.
2782 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
2783 * profile-count.c: Include wide-int.h
2784 (slow_safe_scale_64bit): New.
2786 2017-10-11 Nathan Sidwell <nathan@acm.org>
2788 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
2789 HAS_DECL_ASSEMBLER_NAME_P.
2790 * gimple-expr.c (gimple_decl_printable_name: Check
2791 HAS_DECL_ASSEMBLER_NAME_P too.
2792 * ipa-utils.h (type_in_anonymous_namespace_p): Check
2793 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
2794 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
2795 * passes.c (rest_of_decl_compilation): Check
2796 HAS_DECL_ASSEMBLER_NAME_P too.
2797 * recog.c (verify_changes): Likewise.
2798 * tree-pretty-print.c (dump_decl_name): Likewise.
2799 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
2801 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
2802 (SET_DECL_ASSEMBLER_NAME): Use it.
2803 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
2804 (COPY_DECL_ASSEMBLER_NAME): Likewise.
2805 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
2807 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
2809 * config.gcc (i386, x86_64): Add extra objects.
2810 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
2811 (ix86_min_insn_size): Declare.
2812 (ix86_issue_rate): Declare.
2813 (ix86_adjust_cost): Declare.
2814 (ia32_multipass_dfa_lookahead): Declare.
2815 (ix86_macro_fusion_p): Declare.
2816 (ix86_macro_fusion_pair_p): Declare.
2817 (ix86_bd_has_dispatch): Declare.
2818 (ix86_bd_do_dispatch): Declare.
2819 (ix86_core2i7_init_hooks): Declare.
2820 (ix86_atom_sched_reorder): Declare.
2821 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
2822 (COSTS_N_BYTES): Move to x86-tune-costs.h.
2823 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
2824 (rip_relative_addr_p): Rename to ...
2825 (ix86_rip_relative_addr_p): ... this one; export.
2826 (memory_address_length): Update.
2827 (ix86_issue_rate): Move to x86-tune-sched.c.
2828 (ix86_flags_dependent): Move to x86-tune-sched.c.
2829 (ix86_agi_dependent): Move to x86-tune-sched.c.
2830 (exact_dependency_1): Move to x86-tune-sched.c.
2831 (exact_store_load_dependency): Move to x86-tune-sched.c.
2832 (ix86_adjust_cost): Move to x86-tune-sched.c.
2833 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
2834 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
2835 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
2836 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
2837 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
2838 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
2839 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
2840 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
2841 (min_insn_size): Rename to ...
2842 (ix86_min_insn_size): ... this one; export.
2843 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
2844 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
2845 (core2i7_first_cycle_multipass_backtrack): Move to
2846 x86-tune-sched-core.c.
2847 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
2848 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
2849 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
2850 (ix86_avoid_jump_mispredicts): Update.
2851 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
2852 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
2853 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
2854 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
2855 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
2856 (MAX_INSN): Move to ix86-tune-sched-bd.c.
2857 (MAX_IMM): Move to ix86-tune-sched-bd.c.
2858 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
2859 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
2860 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
2861 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
2862 (MAX_STORE): Move to ix86-tune-sched-bd.c.
2863 (BIG): Move to ix86-tune-sched-bd.c.
2864 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
2865 (enum insn_path): Move to ix86-tune-sched-bd.c.
2866 (get_mem_group): Move to ix86-tune-sched-bd.c.
2867 (is_cmp): Move to ix86-tune-sched-bd.c.
2868 (dispatch_violation): Move to ix86-tune-sched-bd.c.
2869 (is_branch): Move to ix86-tune-sched-bd.c.
2870 (is_prefetch): Move to ix86-tune-sched-bd.c.
2871 (init_window): Move to ix86-tune-sched-bd.c.
2872 (allocate_window): Move to ix86-tune-sched-bd.c.
2873 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
2874 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
2875 (process_end_window): Move to ix86-tune-sched-bd.c.
2876 (allocate_next_window): Move to ix86-tune-sched-bd.c.
2877 (find_constant): Move to ix86-tune-sched-bd.c.
2878 (get_num_immediates): Move to ix86-tune-sched-bd.c.
2879 (has_immediate): Move to ix86-tune-sched-bd.c.
2880 (get_insn_path): Move to ix86-tune-sched-bd.c.
2881 (get_insn_group): Move to ix86-tune-sched-bd.c.
2882 (count_num_restricted): Move to ix86-tune-sched-bd.c.
2883 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
2884 (add_insn_window): Move to ix86-tune-sched-bd.c.
2885 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
2886 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
2887 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
2888 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
2889 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
2890 (do_dispatch): Move to ix86-tune-sched-bd.c.
2891 (has_dispatch): Move to ix86-tune-sched-bd.c.
2892 * config/i386/t-i386: Add new object files.
2893 * config/i386/x86-tune-costs.h: New file.
2894 * config/i386/x86-tune-sched-atom.c: New file.
2895 * config/i386/x86-tune-sched-bd.c: New file.
2896 * config/i386/x86-tune-sched-core.c: New file.
2897 * config/i386/x86-tune-sched.c: New file.
2899 2017-10-11 Liu Hao <lh_mouse@126.com>
2901 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
2902 the generic version below instead.
2903 (should_colorize): Recognize Windows consoles as terminals
2905 * pretty-print.c [__MINGW32__] (write_all): New function.
2906 [__MINGW32__] (find_esc_head): Likewise.
2907 [__MINGW32__] (find_esc_terminator): Likewise.
2908 [__MINGW32__] (eat_esc_sequence): Likewise.
2909 [__MINGW32__] (mingw_ansi_fputs): New function that handles
2911 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
2914 2017-10-11 Richard Biener <rguenther@suse.de>
2916 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
2917 Properly call analyze_scalar_evolution with the loop of the stmt.
2919 2017-10-11 Richard Biener <rguenther@suse.de>
2921 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
2922 * tree-core.h (tree_base): Add chrec_var union member.
2923 * tree.h (CHREC_VAR): Remove.
2924 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
2925 * tree-chrec.h (build_polynomial_chrec): Adjust.
2926 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
2927 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
2929 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
2931 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
2932 * match.pd: ... here.
2933 ((T) X == (T) Y): Relax condition.
2935 2017-10-11 Bin Cheng <bin.cheng@arm.com>
2937 PR tree-optimization/82472
2938 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
2940 (break_alias_scc_partitions): Update postorder number.
2942 2017-10-11 Martin Liska <mliska@suse.cz>
2945 * opts.c (parse_no_sanitize_attribute): Do not use error_value
2947 * opts.h (parse_no_sanitize_attribute): Remove last argument.
2949 2017-10-11 Martin Liska <mliska@suse.cz>
2951 * print-rtl.c (print_insn): Move declaration of idbuf
2952 to same scope as name.
2954 2017-10-11 Martin Liska <mliska@suse.cz>
2959 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
2962 2017-10-11 Martin Liska <mliska@suse.cz>
2965 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
2968 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
2970 * config.gcc (default_gnu_indirect_function): Default to yes for
2971 arm*-*-linux* with glibc.
2973 2017-10-11 Richard Biener <rguenther@suse.de>
2975 * tree-scalar-evolution.c (get_scalar_evolution): Handle
2976 default-defs and types we do not want to analyze.
2977 (interpret_loop_phi): Replace unreachable code with an assert.
2978 (compute_scalar_evolution_in_loop): Remove and inline ...
2979 (analyze_scalar_evolution_1): ... here, replacing condition with
2980 what makes the intent clearer. Remove handling of cases
2981 get_scalar_evolution now handles.
2983 2017-10-10 Jim Wilson <wilson@tuliptree.org>
2985 PR rtl-optimization/81434
2986 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
2987 comment for main loop. In sched_group_found if, also add checks for
2988 pass and min_cost_group.
2990 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
2992 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
2993 (rs6000_insn_cost): New function.
2994 * config/rs6000/rs6000.md (cost): New attribute.
2996 2017-10-10 Jakub Jelinek <jakub@redhat.com>
2997 H.J. Lu <hongjiu.lu@intel.com>
3001 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
3002 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
3003 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
3004 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
3005 to OPTION_MASK_ISA_AVX512VL - builtins that have both
3006 OPTION_MASK_ISA_MMX and some other bit set require both
3007 mmx and the ISAs without the mmx bit.
3008 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
3009 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
3010 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
3011 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
3012 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
3013 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
3014 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
3015 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
3016 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
3017 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
3018 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
3019 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
3020 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
3021 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
3022 Add OPTION_MASK_ISA_MMX.
3024 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
3026 * config.gcc (armv7*-*-freebsd*): New target.
3027 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
3029 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
3031 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
3032 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
3035 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
3037 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
3038 a template parameter.
3039 (WIDE_INT_REF_FOR): Update accordingly.
3040 * tree.h (wi::int_traits <const_tree>): Delete.
3041 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
3042 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
3043 (wi::tree_to_wide_ref): New typedef.
3044 (wi::to_wide): New function.
3045 * calls.c (get_size_range): Use wi::to_wide when operating on
3047 * cgraph.c (cgraph_node::create_thunk): Likewise.
3048 * config/i386/i386.c (ix86_data_alignment): Likewise.
3049 (ix86_local_alignment): Likewise.
3050 * dbxout.c (stabstr_O): Likewise.
3051 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
3052 * expr.c (const_vector_from_tree): Likewise.
3053 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
3054 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
3055 (fold_negate_expr_1, int_const_binop_1, const_binop)
3056 (fold_convert_const_int_from_real, optimize_bit_field_compare)
3057 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
3058 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
3059 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
3060 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
3061 (fold_not_const, round_up_loc): Likewise.
3062 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
3063 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
3064 (alloca_call_type): Likewise.
3065 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
3066 * godump.c (go_output_typedef): Likewise.
3067 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
3068 * internal-fn.c (get_min_precision): Likewise.
3069 * ipa-cp.c (ipcp_store_vr_results): Likewise.
3070 * ipa-polymorphic-call.c
3071 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
3072 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
3073 (ipa_modify_call_arguments): Likewise.
3074 * match.pd: Likewise.
3075 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
3076 * print-tree.c (print_node_brief, print_node): Likewise.
3077 * stmt.c (expand_case): Likewise.
3078 * stor-layout.c (layout_type): Likewise.
3079 * tree-affine.c (tree_to_aff_combination): Likewise.
3080 * tree-cfg.c (group_case_labels_stmt): Likewise.
3081 * tree-data-ref.c (dr_analyze_indices): Likewise.
3082 (prune_runtime_alias_test_list): Likewise.
3083 * tree-dump.c (dequeue_and_dump): Likewise.
3084 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
3085 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
3086 * tree-pretty-print.c (dump_generic_node): Likewise.
3087 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
3088 (simple_iv_with_niters): Likewise.
3089 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
3090 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
3091 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
3092 * tree-ssa-loop-niter.c (split_to_var_and_offset)
3093 (refine_value_range_using_guard, number_of_iterations_ne_max)
3094 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
3095 (get_cst_init_from_scev, record_nonwrapping_iv)
3096 (scev_var_range_cant_overflow): Likewise.
3097 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
3098 * tree-ssa-pre.c (compute_avail): Likewise.
3099 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
3100 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
3101 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
3102 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
3103 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
3104 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
3105 (dump_case_nodes, try_switch_expansion): Likewise.
3106 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
3107 (vect_do_peeling): Likewise.
3108 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
3109 * tree-vect-stmts.c (vectorizable_load): Likewise.
3110 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
3111 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
3112 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
3113 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
3114 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
3115 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
3116 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
3117 (evrp_dom_walker::before_dom_children): Likewise.
3118 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
3119 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
3120 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
3121 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
3122 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
3123 (get_range_pos_neg): Likewise.
3124 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
3125 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
3126 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
3127 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
3128 * config/avr/avr.c (avr_fold_builtin): Likewise.
3129 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
3130 * config/msp430/msp430.c (msp430_attr): Likewise.
3131 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
3132 * config/powerpcspe/powerpcspe-c.c
3133 (altivec_resolve_overloaded_builtin): Likewise.
3134 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
3135 (rs6000_expand_ternop_builtin): Likewise.
3136 * config/rs6000/rs6000-c.c
3137 (altivec_resolve_overloaded_builtin): Likewise.
3138 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
3139 (rs6000_expand_ternop_builtin): Likewise.
3140 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
3142 2017-10-10 Bin Cheng <bin.cheng@arm.com>
3144 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
3145 when copying loop nest with only one inner loop.
3147 2017-10-10 Richard Biener <rguenther@suse.de>
3149 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
3150 blocks if SCEV is active.
3151 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
3153 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
3154 (scev_initialize): Assert we are not yet initialized.
3156 2017-10-10 Bin Cheng <bin.cheng@arm.com>
3158 * tree-loop-distribution.c (generate_loops_for_partition): Remove
3159 inner loop's exit stmt by making it always exit the loop, otherwise
3160 we would generate an infinite empty loop.
3162 2017-10-10 Bin Cheng <bin.cheng@arm.com>
3164 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
3165 renaming variables in new preheader if it's deleted.
3167 2017-10-10 Bin Cheng <bin.cheng@arm.com>
3169 * tree-loop-distribution.c (struct partition): Remove unused field
3170 loops of the structure.
3171 (partition_alloc, partition_free): Ditto.
3172 (build_rdg_partition_for_vertex): Ditto.
3174 2017-10-09 Jeff Law <law@redhat.com>
3176 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
3177 return type to match prototype and documentation.
3179 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
3181 * config/rs6000/rs6000.c (processor_costs): Move to ...
3182 * config/rs6000/rs6000.h: ... here.
3183 (rs6000_cost): Declare.
3185 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
3187 * except.c (setjmp_fn): New global variable.
3188 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
3189 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
3190 if DONT_USE_BUILTIN_SETJMP is defined.
3192 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
3194 * target.def (insn_cost): New hook.
3195 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
3196 * doc/tm.texi: Regenerate.
3197 * rtlanal.c (insn_cost): Use the new hook.
3199 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
3201 * combine.c (combine_validate_cost): Compute the new insn_cost,
3202 not just pattern_cost.
3203 (try_combine): Adjust comment.
3205 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
3207 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
3209 * combine.c (uid_insn_cost): Adjust comment.
3210 (combine_validate_cost): Adjust comment. Use pattern_cost instead
3212 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
3213 * dse.c (find_shift_sequence): Ditto.
3214 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
3215 (bb_valid_for_noce_process_p): Use pattern_cost.
3216 * rtl.h (insn_rtx_cost): Delete.
3217 (pattern_cost): New prototype.
3218 (insn_cost): New prototype.
3219 * rtlanal.c (insn_rtx_cost): Rename to...
3220 (pattern_cost): ... this.
3223 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
3225 * config/i386/i386.md (*jcc_2): Remove insn pattern.
3226 (*jcc<mode>_0_r_i387): Ditto.
3227 (*jccxf_r_i387): Ditto.
3228 (*jcc<mode>_r_i387): Ditto.
3229 (*jccu<mode>_r_i387): Ditto.
3230 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
3231 (*jcc): Rename from *jcc_1.
3233 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3235 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
3236 deferred rescans after the lvx/stvx recombination pre-pass.
3238 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
3240 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
3241 memory operation instruction support.
3242 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
3243 (rs6000-ibm-aix[789]*): Likewise.
3244 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
3245 Document new functions.
3247 2017-10-09 Richard Biener <rguenther@suse.de>
3249 PR tree-optimization/82397
3250 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
3251 equality only for semantically equal trees.
3253 2017-10-09 Richard Biener <rguenther@suse.de>
3255 PR tree-optimization/82449
3256 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
3257 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
3258 allow constant addresses.
3259 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
3262 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3264 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
3267 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3270 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
3273 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3276 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
3278 2017-10-09 Jakub Jelinek <jakub@redhat.com>
3281 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
3282 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
3284 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
3286 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
3287 (WI_BINARY_PREDICATE_RESULT): Likewise.
3288 (wi::binary_traits::operator_result): New type.
3289 (wi::binary_traits::predicate_result): Likewise.
3290 (generic_wide_int::operator~, unary generic_wide_int::operator-)
3291 (generic_wide_int::operator==, generic_wide_int::operator!=)
3292 (generic_wide_int::operator&, generic_wide_int::and_not)
3293 (generic_wide_int::operator|, generic_wide_int::or_not)
3294 (generic_wide_int::operator^, generic_wide_int::operator+
3295 (binary generic_wide_int::operator-, generic_wide_int::operator*):
3297 (operator~, unary operator-, operator==, operator!=, operator&)
3298 (operator|, operator^, operator+, binary operator-, operator*): New
3300 * expr.c (get_inner_reference): Use wi::bit_and_not.
3301 * fold-const.c (fold_binary_loc): Likewise.
3302 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
3303 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
3304 (bit_value_binop): Likewise.
3305 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
3306 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
3307 (extract_range_from_binary_expr_1): Likewise.
3308 (masked_increment): Likewise.
3309 (simplify_bit_ops_using_ranges): Likewise.
3311 2017-10-09 Martin Jambor <mjambor@suse.cz>
3314 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
3315 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
3316 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
3318 (hsa_fixup_mov_insn_type): New function.
3319 (hsa_op_with_type::get_in_type): Use it.
3320 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
3321 immediates in an assert.
3322 (hsa_op_with_type::extend_int_to_32bit): New method.
3323 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
3324 types. Convert to dest type if necessary.
3325 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
3326 (reg_for_gimple_ssa): Pass false as min32int to
3327 hsa_type_for_scalar_tree_type.
3328 (gen_hsa_addr): Fixup type when creating addresable temporary.
3329 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
3330 (gen_hsa_unary_operation): Extend operands and convert to dest type if
3331 necessary. Call hsa_fixup_mov_insn_type.
3332 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
3333 extend operands and convert to dest type if necessary.
3334 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
3335 to dest type if necessary.
3336 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
3337 if conversion nt necessary and size matches.
3338 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
3339 to dest type if necessary.
3340 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
3341 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
3343 (gen_hsa_clrsb): Likewise.
3344 (gen_hsa_ffs): Likewise.
3345 (gen_hsa_divmod): Extend operands and convert to dest type if
3347 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
3349 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
3351 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
3352 Remove empty default arguments. Use a brace block as output
3354 (conditional return): Ditto.
3356 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
3357 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
3359 (group_ending_nop): Ditto.
3360 (doloop_end): Ditto.
3361 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
3362 (splitters for those): Ditto.
3364 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
3366 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
3367 a conditional jump (and the compare for it) so that pc_rtx is the
3369 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
3370 for the deleted and renamed ctr<mode>_internal[234] patterns.
3371 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
3372 Delete second conditional return pattern.
3373 (ctr<mode>_internal2): Delete this second bdnz pattern.
3374 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
3375 (ctr<mode>_internal4): Delete this second bdz pattern.
3377 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
3379 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
3380 (always_initialized_rtx_for_ssa_name_p): New predicate.
3381 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
3382 (finish_out_of_ssa): Free new field of SA.
3383 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
3384 * tree-ssa-coalesce.c: Include tree-ssa.h.
3385 (get_parm_default_def_partitions): Remove extern keyword.
3386 (get_undefined_value_partitions): New function.
3387 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
3388 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
3390 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
3392 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
3394 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
3396 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
3398 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
3399 for targets that preffer 128bit.
3401 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
3403 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
3405 2017-10-08 Olivier Hainque <hainque@adacore.com>
3407 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
3408 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
3410 (thumb_set_return_address): Likewise.
3412 2017-10-08 Olivier Hainque <hainque@adacore.com>
3414 * common/config/arm/arm-common.c (arm_except_unwind_info):
3415 Handle DWARF2_UNWIND_INFO.
3417 2017-10-07 Michael Collison <michael.collison@arm.com>
3419 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
3422 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
3424 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
3425 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
3426 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
3427 defined, force the creation of a new block for a dispatch label.
3429 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
3431 * invoke.texi (Wsuggest-attribute=cold): Document.
3432 * common.opt (Wsuggest-attribute=cold): New
3433 * ipa-pure-const.c (warn_function_cold): New function.
3434 * predict.c (compute_function_frequency): Use it.
3435 * predict.h (warn_function_cold): Declare.
3437 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
3439 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
3442 2017-10-06 Martin Liska <mliska@suse.cz>
3444 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
3445 keyword for member functions.
3446 (struct sanopt_tree_couple): New struct.
3447 (struct sanopt_tree_couple_hash): New function.
3448 (struct sanopt_ctx): Add new hash_map.
3449 (has_dominating_ubsan_ptr_check): New function.
3450 (record_ubsan_ptr_check_stmt): Likewise.
3451 (maybe_optimize_ubsan_ptr_ifn): Likewise.
3452 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
3453 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
3455 2017-10-06 Sudakshina Das <sudi.das@arm.com>
3458 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
3459 aarch64_simd_valid_immediate on CONST_VECTORs.
3460 (aarch64_reg_or_bic_imm): Likewise.
3462 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
3464 PR rtl-optimization/82396
3465 * haifa-sched.c (ready_sort_real): Disable qsort checking.
3467 2017-10-06 Sebastian Pop <sebpop@gmail.com>
3469 * graphite-dependences.c (scop_get_reads): Move code to...
3470 (scop_get_must_writes): Move code to...
3471 (scop_get_may_writes): Move code to...
3472 (scop_get_reads_and_writes): ... here.
3473 (scop_get_dependences): Call scop_get_reads_and_writes.
3475 2017-10-06 Jakub Jelinek <jakub@redhat.com>
3477 PR tree-optimization/82434
3478 * fold-const.h (can_native_encode_type_p,
3479 can_native_encode_string_p): Remove.
3480 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
3481 don't encode anything, just return what would be otherwise returned.
3482 (native_encode_fixed, native_encode_complex, native_encode_vector):
3484 (native_encode_string): Likewise. Inline by hand
3485 can_native_encode_string_p.
3486 (can_native_encode_type_p): Remove.
3487 (can_native_encode_string_p): Remove.
3488 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
3489 STRING_CSTs using can_native_encode_string_p, test all
3490 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
3491 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
3492 argument from native_encode_expr.
3493 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
3494 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
3497 2017-10-06 Richard Biener <rguenther@suse.de>
3499 PR tree-optimization/82397
3500 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
3501 operand_equal_p but rely on data_ref_compare_tree for detecting
3503 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
3504 to match up with dr_group_sort_cmp.
3506 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3509 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
3511 * config/s390/s390-builtin-types.def: Regenerate.
3513 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3516 * config/s390/s390-builtin-types.def: Regenerate.
3517 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
3518 Change flag from B_VXE to B_VX.
3519 (s390_vec_min_dbl): Remove B_VXE flag.
3521 2017-10-06 Richard Biener <rguenther@suse.de>
3523 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
3524 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
3525 translate_isl_ast_to_gimple::is_valid_rename,
3526 translate_isl_ast_to_gimple::get_rename,
3527 translate_isl_ast_to_gimple::get_def_bb_for_const,
3528 translate_isl_ast_to_gimple::get_new_name,
3529 translate_isl_ast_to_gimple::collect_all_ssa_names,
3530 translate_isl_ast_to_gimple::copy_loop_phi_args,
3531 translate_isl_ast_to_gimple::collect_all_ssa_names,
3532 translate_isl_ast_to_gimple::copy_loop_phi_args,
3533 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
3534 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
3535 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
3536 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
3537 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
3538 translate_isl_ast_to_gimple::copy_cond_phi_args,
3539 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
3540 translate_isl_ast_to_gimple::edge_for_new_close_phis,
3541 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
3542 translate_isl_ast_to_gimple::rename_uses,
3543 translate_isl_ast_to_gimple::rename_all_uses): Remove.
3544 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
3545 (set_rename_for_each_def): Likewise.
3546 (graphite_copy_stmts_from_block): Handle debug stmt resetting
3547 here. Handle rewriting SCEV analyzable uses here.
3548 (copy_bb_and_scalar_dependences): Generate code for PHI
3550 (graphite_regenerate_ast_isl): Adjust.
3551 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
3552 (add_write, add_read): New functions.
3553 (build_cross_bb_scalars_def): Use it and simplify.
3554 (build_cross_bb_scalars_use): Likewise.
3555 (graphite_find_cross_bb_scalar_vars): Inline into...
3556 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
3557 simulating out-of-SSA. Compute liveout and add dependencies.
3558 (build_scops): Force an empty entry block.
3559 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
3561 (sese_build_liveouts): Declare.
3562 (sese_trivially_empty_bb_p): Likewise.
3563 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
3564 compute liveout and debug_liveout.
3565 (sese_bad_liveouts_use): Remove.
3566 (sese_reset_debug_liveouts_bb): Likewise.
3567 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
3568 (sese_build_liveouts): Build liveout and debug_liveout and store
3570 (new_sese_info): Adjust.
3571 (free_sese_info): Likewise.
3572 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
3573 do not build liveout here.
3574 (move_sese_in_condition): Adjust region entry.
3575 (scev_analyzable_p): Match up with chrec_apply requirements.
3576 (sese_trivially_empty_bb_p): New.
3577 * tree-into-ssa.c (get_reaching_def): Properly support generating
3578 default-defs for incremental rewrite of anonymous names.
3580 2017-10-06 Richard Biener <rguenther@suse.de>
3582 * graphite-sese-to-poly.c (extract_affine): For casts increasing
3583 precision do not perform modulo reduction.
3585 2017-10-06 Richard Biener <rguenther@suse.de>
3587 PR tree-optimization/82436
3588 * tree-vect-slp.c (vect_supported_load_permutation_p): More
3589 conservatively choose the vectorization factor when checking
3590 whether we can perform the required load permutation.
3591 (vect_transform_slp_perm_load): Assert when we may not fail.
3593 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
3595 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
3596 message for incompatible -msdata=* and -mcall-* options.
3598 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
3600 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
3601 rate for post-reload scheduling.
3603 2017-10-05 Tamar Christina <tamar.christina@arm.com>
3605 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
3607 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
3609 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
3610 to improve monte carlo in scimark.
3612 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
3614 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
3615 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
3616 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
3617 pentium4_cost, nocona_cost): Set reassociation width to 1.
3618 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
3619 width to 2 for fp operations and 1 otherwise.
3620 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
3622 (atom_cost): Set reassociation width to 2.
3623 (slm_cost, generic_cost): Set fp reassociation width
3624 to 2 and 1 otherwise.
3625 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
3626 (core_cost): Set fp reassociation width to 4 and vector to 2.
3627 (ix86_reassociation_width): Rewrite using cost table; special case
3628 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
3629 and TARGET_AVX128_OPTIMAL.
3630 * config/i386/i386.h (processor_costs): Add
3631 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
3632 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
3633 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
3634 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
3635 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
3636 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
3638 2017-10-05 Nathan Sidwell <nathan@acm.org>
3640 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
3642 2017-10-05 Tamar Christina <tamar.christina@arm.com>
3644 * config/arm/arm.c (arm_test_fpu_data): New.
3645 (arm_run_selftests): Call arm_test_fpu_data.
3647 2017-10-04 Nathan Sidwell <nathan@acm.org>
3649 * toplev.c (toplev::main): Remove excess parens on pretty_printer
3651 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
3653 2017-10-04 Sudakshina Das <sudi.das@arm.com>
3655 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
3656 check type for aarch64_simd_valid_immediate.
3657 (aarch64_output_simd_mov_immediate): Update prototype.
3658 (aarch64_simd_valid_immediate): Update prototype.
3659 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
3660 support for ORR-immediate.
3661 (and<mode>3): modified pattern to add support for BIC-immediate.
3662 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
3663 now checks for valid immediate for BIC and ORR based on new enum
3665 (aarch64_output_simd_mov_immediate): Function now used to output
3666 BIC/ORR imm as well based on new enum argument.
3667 * config/aarch64/constraints.md (Do): New vector immediate constraint.
3669 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
3670 (aarch64_reg_or_bic_imm): Likewise.
3672 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3674 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
3675 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
3677 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
3681 PR rtl-optimization/82396
3682 * haifa-sched.c (autopref_multipass_init): Simplify
3684 (autopref_rank_data): Simplify sort order.
3685 * sched-int.h (autopref_multipass_data_): Remove
3686 multi_mem_insn_p, min_offset and max_offset.
3688 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3690 * doc/sourcebuild.texi: Document vect_peeling_profitable.
3692 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3694 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
3697 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3699 * doc/sourcebuild.texi: Document vect_long_mult.
3701 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
3703 PR tree-optimization/82413
3704 * fold-const.c (build_range_check): Use widest_int when comparing
3705 the maximum ETYPE value with HIGH.
3707 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
3709 PR rtl-optimization/82396
3710 * haifa-sched.c (autopref_multipass_init): Simplify
3712 (autopref_rank_data): Simplify sort order.
3713 * sched-int.h (autopref_multipass_data_): Remove
3714 multi_mem_insn_p, min_offset and max_offset.
3716 2017-10-04 Jakub Jelinek <jakub@redhat.com>
3718 PR tree-optimization/82381
3719 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
3720 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
3723 PR tree-optimization/82374
3724 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
3725 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
3726 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
3727 current_function_decl to the new decl.
3729 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
3731 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
3732 helper macro for IEEE float128 hardware built-in functions.
3733 (SQRTF128_ODD): Add built-in functions with the round-to-odd
3735 (TRUNCF128_ODD): Likewise.
3736 (ADDF128_ODD): Likewise.
3737 (SUBF128_ODD): Likewise.
3738 (MULF128_ODD): Likewise.
3739 (DIVF128_ODD): Likewise.
3740 (FMAF128_ODD): Likewise.
3741 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
3742 UNSPEC_TRUNC_ROUND_TO_ODD.
3743 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
3744 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
3745 floating point round to odd instructions.
3746 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
3747 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
3748 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
3749 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
3750 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
3751 (trunc<mode>sf2_hw): Change the truncate with round to odd
3752 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
3753 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
3754 to odd hardware instructions.
3755 (sub<mode>3_odd): Likewise.
3756 (mul<mode>3_odd): Likewise.
3757 (div<mode>3_odd): Likewise.
3758 (sqrt<mode>2_odd): Likewise.
3759 (fma<mode>4_odd): Likewise.
3760 (fms<mode>4_odd): Likewise.
3761 (nfma<mode>4_odd): Likewise.
3762 (nfms<mode>4_odd): Likewise.
3763 (trunc<mode>df2_odd): Change the truncate with round to odd
3764 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
3766 * doc/extend.texi (PowerPC built-in functions): Update documentation
3767 for existing IEEE float128-bit built-in functions. Add built-in
3768 functions that generate the IEEE 128-bit floating point round to
3771 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
3773 PR rtl-optimization/77729
3774 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
3775 to (X&(C1&~C2))|C2 transformations.
3777 2017-10-03 Martin Jambor <mjambor@suse.cz>
3779 PR tree-optimization/82363
3780 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
3781 mismatch, mark lacc written regardless of racc.
3783 2017-10-03 Jakub Jelinek <jakub@redhat.com>
3785 PR tree-optimization/82381
3786 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
3787 stmt_to_insert nor wheather SSA_NAMEs are default defs.
3788 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
3789 fallthrough into reassoc_stmt_dominates_stmt_p.
3792 * combine.c (combine_instructions): Don't combine in unreachable
3795 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
3798 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
3799 function to not use the have_cpu variable. Do not set cpu_index,
3800 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
3802 (rs6000_valid_attribute_p): Remove duplicate initializations of
3803 old_optimize and func_optimize.
3804 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
3805 (rs6000_activate_target_options): Make global.
3806 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
3809 2017-10-02 Jakub Jelinek <jakub@redhat.com>
3811 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
3812 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
3813 Set *poffset to 0 and *psize and *pmax_size to -1 if
3814 *poffset + *psize overflows in HOST_WIDE_INT.
3816 PR tree-optimization/82387
3817 PR tree-optimization/82388
3818 PR tree-optimization/82389
3819 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
3820 instead of live_bytes non-NULL.
3822 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
3825 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
3828 2017-10-02 Richard Biener <rguenther@suse.de>
3830 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
3831 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
3832 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
3834 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
3836 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
3837 requested precision matches the type's.
3838 * calls.c (alloc_max_size): Calculate the new candidate size as
3839 a widest_int and use wi::to_widest when comparing it with the
3840 current candidate size.
3841 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
3842 zero rather than integer_zero_node.
3843 * match.pd: Check for a no-op conversion before using wi::add
3844 rather than after. Use tree_to_uhwi when summing small shift
3845 counts into an unsigned int.
3847 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
3848 Alan Hayward <alan.hayward@arm.com>
3849 David Sherwood <david.sherwood@arm.com>
3852 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
3853 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
3854 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
3855 POINTER_AND_FP_REGS.
3857 2017-10-02 Richard Biener <rguenther@suse.de>
3859 PR tree-optimization/82355
3860 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
3861 a mapping for the enclosing loop but avoid generating one for
3863 (copy_bb_and_scalar_dependences): Remove premature codegen
3864 error on PHIs in blocks duplicated into multiple places.
3865 * graphite-scop-detection.c
3866 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
3867 in the region use it as loop and nest to analyze the DR in.
3868 (try_generate_gimple_bb): Likewise.
3869 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
3870 (add_loop_constraints): For blocks in a loop not in the region
3871 create a dimension with a single iteration.
3872 * sese.h (gbb_loop_at_index): Remove assert.
3874 2017-10-01 Kevin Buettner <kevinb@redhat.com>
3876 * omp-expand.c (adjust_context_scope): New function.
3877 (expand_parallel_call): Call adjust_context_scope.
3879 2017-10-01 Jeff Law <law@redhat.com>
3881 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
3882 dom_opt_dom_walker class with direct access to private members.
3883 Add comments. Call test_for_singularity.
3884 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
3885 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
3886 m_dummy_cond anymore.
3887 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
3889 (pass_dominator:execute): Build the dummy_cond here and pass it
3890 to the dom_opt_dom_walker ctor.
3891 (test_for_singularity): New function.
3893 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
3894 Maya Rashish <coypu@sdf.org>
3896 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
3897 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
3898 (alpha*-*-netbsd*) Use nbsd_tm_file.
3899 (arm*-*-netbsdelf*) Likewise.
3900 (i[34567]86-*-netbsdelf*) Likewise.
3901 (x86_64-*-netbsd*) Likewise.
3902 (mips*-*-netbsd*) Likewise.
3903 (powerpc-*-netbsd*) Likewise.
3904 (sh*-*-netbsd*) Likewise.
3905 (sparc-*-netbsdelf*) Likewise.
3906 (sparc64-*-netbsd*) Likewise.
3907 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
3909 (vax-*-netbsdelf*) Likewise.
3910 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
3911 (UINT_FAST8_TYPE) Likewise.
3912 (INT_FAST16_TYPE) Check CHAR_FAST16.
3913 (UINT_FAST16_TYPE) Likewise.
3915 2017-09-30 Jakub Jelinek <jakub@redhat.com>
3918 * config/i386/i386.md
3919 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
3920 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
3921 *divmodsi4_zext_2): New define_insn_and_split.
3922 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
3923 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
3924 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
3925 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
3926 New define_insn_and_split.
3927 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
3928 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
3929 operands[1] having DImode when mode is SImode.
3931 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
3932 always SImode for DIV and MOD in REG_EQUAL notes.
3934 2017-09-29 Yury Gribov <tetra2005@gmail.com>
3937 * match.pd: Fix handling of NaNs in pattern.
3939 2017-09-29 Jeff Law <law@redhat.com>
3941 * sbitmap.c (bitmap_bit_in_range_p): New function.
3942 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
3943 * tree-ssa-dse.c (live_bytes_read): New function.
3944 (dse_classify_store): Ignore reads of dead bytes.
3946 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
3947 typos and whitespace errors.
3948 * config/i386/predicates.md (address_no_seg_operand): Likewise.
3949 * config/s390/s390.c (s390_emit_prologue): Likewise.
3951 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
3954 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
3955 with a symbol for LRA.
3957 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
3959 PR rtl-optimization/82338
3960 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
3962 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
3964 * genmodes.c (calc_wider_mode): Suppress qsort macro.
3965 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
3966 (qsort_chk): Declare.
3967 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
3968 (qsort_chk): New function.
3970 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3972 PR tree-optimization/82337
3973 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
3974 phi definition if the PHI result appears in an abnormal PHI.
3975 (find_basis_for_base_expr): Don't record a basis if the LHS of the
3976 basis appears in an abnormal PHI.
3978 2017-09-29 Richard Biener <rguenther@suse.de>
3980 * graphite-isl-ast-to-gimple.c
3981 (translate_isl_ast_to_gimple::set_codegen_error): New function.
3982 (binary_op_to_tree): Use it.
3983 (get_rename_from_scev): Likewise.
3984 (copy_loop_phi_nodes): Likewise.
3985 (copy_bb_and_scalar_dependences): Likewise.
3986 (translate_pending_phi_nodes): Likewise.
3988 2017-09-29 Jakub Jelinek <jakub@redhat.com>
3991 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
3992 for movabsq $(i32 << shift), r64.
3994 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
3996 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
3997 index when encoding %esp as %rsp to avoid 0x67 prefix.
3999 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
4001 * config/i386/i386.md (*movsf_internal, *movdf_internal):
4002 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
4004 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
4006 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
4007 Extensions with more than 16 double VFP registers.
4008 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
4009 to_clear_mask and all code related to it. Replace the remaining
4010 entry by a sbitmap and adapt code accordingly.
4012 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
4014 * brig-builtins.def: Change pure attributes to const.
4016 2017-09-28 Joseph Myers <joseph@codesourcery.com>
4018 * config.gcc (default_gnu_indirect_function): Default to yes for
4019 sparc*-*-linux* with glibc.
4021 2017-09-28 Joseph Myers <joseph@codesourcery.com>
4023 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
4024 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
4025 when creating .init_array and .fini_array sections with priority
4028 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
4031 * config/aarch64/aarch64.c
4032 (aarch64_builtin_support_vector_misalignment): Always return false
4033 when misalignment is unknown.
4035 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
4037 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
4038 this function to return false if the definition used by the swap
4039 instruction is artificial, or if the memory address from which the
4040 constant value is loaded is not represented by a base address held
4041 in a register or if the base address register is a frame or stack
4042 pointer. Additionally, return false if the base address of the
4043 loaded constant is a SYMBOL_REF but is not considered to be a
4045 (replace_swapped_load_constant): New function.
4046 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
4047 loaded constant vector with a load of a swapped constant vector.
4049 2017-09-27 Carl Love <cel@us.ibm.com>
4051 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
4052 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
4053 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
4056 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
4058 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
4059 first, always call autopref_rank_data otherwise.
4061 2017-09-27 Richard Biener <rguenther@suse.de>
4063 * graphite-scop-detection.c (find_scop_parameters): Move
4064 loop bound handling ...
4065 (gather_bbs::before_dom_children): ... here, avoiding the need
4066 to build scop_info->loop_nest.
4067 (record_loop_in_sese): Remove.
4068 * sese.h (sese_info_t::loop_nest): Remove.
4069 * sese.c (new_sese_info): Do not allocate loop_nest.
4070 (free_sese_info): Do not free loop_nest.
4072 2017-09-27 Jakub Jelinek <jakub@redhat.com>
4075 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
4076 lhs from calls if the lhs has addressable type.
4078 2017-09-27 Richard Biener <rguenther@suse.de>
4080 * graphite.h (scop::max_alias_set): New member.
4081 * graphite-scop-detection.c: Remove references to non-existing
4082 --param in comments.
4083 (build_alias_sets): Record the maximum alias set used for drs.
4084 (build_scops): Support zero as unlimited for
4085 --param graphite-max-arrays-per-scop.
4086 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
4088 (build_poly_sr_1): ... here. Compute alias set based on the
4089 maximum alias set used for drs rather than
4090 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
4092 2017-09-27 Richard Biener <rguenther@suse.de>
4094 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
4095 --param loop-block-tile-size=0 to disable tiling.
4097 2017-09-27 Richard Biener <rguenther@suse.de>
4099 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
4100 (graphite-max-nb-scop-params): Document special value zero.
4101 * domwalk.h (dom_walker::STOP): New symbolical constant.
4102 (dom_walker::dom_walker): Add optional parameter for bb to
4104 (dom_walker::~dom_walker): Declare.
4105 (dom_walker::before_dom_children): Document STOP return value.
4106 (dom_walker::m_user_bb_to_rpo): New member.
4107 (dom_walker::m_bb_to_rpo): Likewise.
4108 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
4109 mapping here if not provided by the user.
4110 (dom_walker::~dom_walker): Free bb to RPO mapping if not
4111 provided by the user.
4112 (dom_walker::STOP): Define.
4113 (dom_walker::walk): Do not compute bb to RPO mapping here.
4114 Support STOP return value from before_dom_children to stop
4116 * graphite-optimize-isl.c (optimize_isl): If the schedule
4117 is the same still generate code if -fgraphite-identity
4118 or -floop-parallelize-all are given.
4119 * graphite-scop-detection.c: Include cfganal.h.
4120 (gather_bbs::gather_bbs): Get and pass through bb to RPO
4122 (gather_bbs::before_dom_children): Return STOP for BBs
4124 (build_scops): Compute bb to RPO mapping and pass it to
4125 the domwalk. Treat --param graphite-max-nb-scop-params=0
4126 as not limiting the number of params.
4127 * graphite.c (graphite_initialize): Remove limit on the
4128 number of basic-blocks in a function.
4129 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
4130 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
4131 default value of 10.
4133 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4135 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
4136 Adjust code to eliminate needing to do the shift right 32-bits
4137 operation after XSCVDPSPN.
4139 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4141 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
4142 ((X / Y) != 0 -> X >= Y): Likewise.
4144 2017-09-26 Carl Love <cel@us.ibm.com>
4146 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
4147 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
4148 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
4149 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
4150 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
4151 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
4152 definitions and overloading.
4153 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
4154 statement for P9V_BUILTIN_XST_LEN_R.
4155 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
4156 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
4157 define_expand and define_insn for the instructions and builtins.
4158 * doc/extend.texi: Update the built-in documentation file for the new
4160 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
4161 define_insn for the instructions
4163 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
4166 * gcc/config/netbsd-protos.h: New file.
4167 * gcc/config/netbsd.c: New file.
4168 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
4169 * gcc/config/t-netbsd: New file.
4170 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
4171 (tmake_file) Add t-netbsd.
4172 (extra_objs) Add netbsd.o.
4174 2017-09-26 Janus Weil <janus@gcc.gnu.org>
4178 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
4180 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4182 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
4183 sign extension from a vector register to a GPR by doing a 32-bit
4184 direct move and then an EXTSW.
4185 (extendsi<mode>2 splitter): Likewise.
4186 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
4187 right or vector extract after doing XSCVDPSPN. Use
4188 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
4190 (movdi_from_sf_zero_ext): Likewise.
4191 (reload_gpr_from_vsxsf): Likewise.
4192 (p8_mfvsrd_4_disf): Delete, no longer used.
4193 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
4194 then needing to move the SFmode to a GPR to use the XSCVDPSP
4195 instruction instead of FRSP and XSCVDPSPN.
4196 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
4197 it is adjacent to the other XSCVSPDP insns.
4198 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
4199 SFmode to be in traditional Altivec registers.
4200 (vsx_xscvdpspn): Eliminate useless alternative constraint.
4201 (vsx_xscvspdpn): Likewise.
4202 (vsx_xscvspdpn_scalar): Likewise.
4204 2017-09-26 Martin Jambor <mjambor@suse.cz>
4206 * tree-sra.c (compare_access_positions): Put integral types first,
4207 stabilize sorting of integral types, remove conditions putting
4208 non-full-precision integers last.
4209 (sort_and_splice_var_accesses): Disable scalarization if a
4210 non-integert would be represented by a non-full-precision integer.
4212 2017-09-26 Joseph Myers <joseph@codesourcery.com>
4214 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
4215 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
4216 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
4217 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
4218 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
4219 conditionals inside the function instead of around it. Call
4220 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
4221 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
4223 2017-09-26 Richard Biener <rguenther@suse.de>
4225 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
4227 (scop_detection::build_scop_breadth): ... this. Removed.
4228 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
4229 (scop_detection::harmful_stmt_in_bb): Likewise.
4230 (scop_detection::graphite_can_represent_stmt): Likewise.
4231 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
4232 (scop_detection::can_represent_loop): Remove recursion, fold in ...
4233 (scop_detection::can_represent_loop_1): ... this. Removed.
4234 (scop_detection::harmful_loop_in_region): Simplify after inlining
4235 the above and remove more quadraticness.
4236 (build_scops): Adjust.
4237 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
4240 2017-09-26 Jakub Jelinek <jakub@redhat.com>
4243 * config/i386/i386.c (ix86_print_operand_address_as): Only test
4244 REGNO (base) == SP_REG if base is a REG.
4247 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
4248 if it is different SSA_NAME.
4249 (optimize_range_tests_cmp_bitwise): New function.
4250 (optimize_range_tests): Call it.
4252 2017-09-26 Richard Biener <rguenther@suse.de>
4254 PR tree-optimization/82321
4255 * graphite.c (canonicalize_loop_closed_ssa): Properly check
4256 for the def being inside the loop.
4258 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4260 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
4262 * config/s390/s390-builtins.def: Fix constraint on op4.
4264 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4266 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
4267 independent expanders.
4268 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
4269 ("vec_ordered", "vec_unordered"): New expanders.
4271 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4273 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
4276 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4278 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
4279 vec_unpacks_lo_v16qi.
4280 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
4282 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4284 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
4285 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
4286 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
4288 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4290 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
4292 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
4294 ("*vec_slb<mode>"): New insn pattern.
4295 ("vec_shr_<mode>"): New expander.
4296 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
4297 and force the shift count operand to V16QImode.
4298 ("vec_srb<mode>"): Set shift count mode to V16QI.
4300 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4302 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
4303 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
4304 ("vec_widen_smult_hi_<mode>"): New expander definitions.
4306 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
4309 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
4311 2017-09-26 Richard Biener <rguenther@suse.de>
4313 PR tree-optimization/82320
4314 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
4317 2017-09-25 Jeff Law <law@redhat.com>
4319 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
4320 prototype for new argument.
4321 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
4322 mostly extracted from rs6000_emit_allocate_stack.
4323 (rs6000_emit_probe_stack_range_stack_clash): New function.
4324 (rs6000_emit_allocate_stack): Call
4325 rs6000_emit_probe_stack_range_stack_clash as needed.
4326 (rs6000_emit_probe_stack_range): Add additional argument
4327 to call to gen_probe_stack_range{si,di}.
4328 (output_probe_stack_range): New.
4329 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
4330 (output_probe_stack_range_stack_clash): New.
4331 (rs6000_emit_prologue): Emit notes into dump file as requested.
4332 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
4333 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
4334 Add additional operand and pass it to output_probe_stack_range.
4336 2017-09-25 Bin Cheng <bin.cheng@arm.com>
4338 PR tree-optimization/82163
4339 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
4340 (checking_verify_loop_closed_ssa): New parameter.
4341 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
4342 (check_loop_closed_ssa_stmt): Delete.
4343 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
4344 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
4345 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
4348 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
4350 * brig-builtins.def: Treat HSAIL barrier builtins as
4351 setjmp/longjump style functions.
4353 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
4355 * target.def (constant_alignment): New hook.
4356 * defaults.h (CONSTANT_ALIGNMENT): Delete.
4357 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
4358 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
4359 * doc/tm.texi: Regenerate.
4360 * targhooks.h (default_constant_alignment): Declare.
4361 (constant_alignment_word_strings): Likewise.
4362 * targhooks.c (default_constant_alignment): New function.
4363 (constant_alignment_word_strings): Likewise.
4364 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
4365 instead of CONSTANT_ALIGNMENT.
4366 * varasm.c (align_variable, get_variable_align, build_constant_desc)
4367 (force_const_mem): Likewise.
4368 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
4369 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
4370 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
4371 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4372 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
4374 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
4375 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4376 constant_alignment_word_strings.
4377 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
4378 (CONSTANT_ALIGNMENT): Likewise.
4379 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4380 (arm_constant_alignment): New function.
4381 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
4382 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4383 constant_alignment_word_strings.
4384 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
4385 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4386 constant_alignment_word_strings.
4387 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
4388 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4389 (cris_constant_alignment): New function.
4390 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
4391 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4392 (epiphany_constant_alignment): New function.
4393 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
4394 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4395 constant_alignment_word_strings.
4396 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
4397 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4398 constant_alignment_word_strings.
4399 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
4400 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4401 constant_alignment_word_strings.
4402 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
4403 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
4404 * config/i386/i386.c (ix86_constant_alignment): Make static.
4405 Use the same interface as the target hook.
4406 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4407 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
4408 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4409 constant_alignment_word_strings.
4410 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
4411 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
4412 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4413 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
4414 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4415 constant_alignment_word_strings.
4416 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
4417 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4418 constant_alignment_word_strings.
4419 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
4420 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4421 constant_alignment_word_strings.
4422 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
4423 * config/microblaze/microblaze.c (microblaze_constant_alignment):
4425 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4426 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
4427 * config/mips/mips.c (mips_constant_alignment): New function.
4428 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4429 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
4430 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
4431 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4432 (mmix_constant_alignment): Make static. Use the same interface
4434 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
4435 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4436 constant_alignment_word_strings.
4437 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
4438 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4439 constant_alignment_word_strings.
4440 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
4441 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4442 constant_alignment_word_strings.
4443 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
4444 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4445 (rs6000_constant_alignment): New function.
4446 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
4447 * config/riscv/riscv.c (riscv_constant_alignment): New function.
4448 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4449 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
4450 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4451 (rs6000_constant_alignment): New function.
4452 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
4453 * config/s390/s390.c (s390_constant_alignment): New function.
4454 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4455 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
4456 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4457 constant_alignment_word_strings.
4458 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
4459 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4460 (sparc_constant_alignment): New function.
4461 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
4462 * config/spu/spu.c (spu_constant_alignment): New function.
4463 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4464 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
4465 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4466 constant_alignment_word_strings.
4467 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
4468 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4469 constant_alignment_word_strings.
4470 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
4471 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4472 constant_alignment_word_strings.
4473 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
4474 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4475 (visium_constant_alignment): New function.
4476 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
4477 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4478 (xtensa_constant_alignment): New function.
4479 * system.h (CONSTANT_ALIGNMENT): Poison.
4481 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
4483 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
4484 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
4485 (rs6000_builtin_valid_without_lhs): New helper function.
4486 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4487 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
4489 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
4491 * target.h (vec_perm_indices): Use unsigned short rather than
4493 (auto_vec_perm_indices): Likewise.
4494 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
4495 Use unsigned int rather than unsigned char.
4496 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
4498 2017-09-25 Richard Biener <rguenther@suse.de>
4500 * cfgloop.h (sort_sibling_loops): Declare.
4501 * cfgloop.c (sort_sibling_loops_cmp): New helper.
4502 (sort_sibling_loops): New function sorting the sibling loop list
4504 * graphite.c (graphite_transform_loops): Sort sibling loops.
4506 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
4508 * target.def (vec_perm_const_ok): Change sel parameter to
4510 * optabs-query.c (can_vec_perm_p): Update accordingly.
4511 * doc/tm.texi: Regenerate.
4512 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
4513 auto_vec_perm_indices and remove separate nelt field.
4514 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
4515 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
4516 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
4517 (aarch64_expand_vec_perm_const): Update accordingly.
4518 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
4519 to vec_perm_indices.
4520 * config/arm/arm.c (expand_vec_perm_d): Change perm to
4521 auto_vec_perm_indices and remove separate nelt field.
4522 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
4523 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
4524 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
4526 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
4527 to vec_perm_indices.
4528 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
4529 sel to vec_perm_indices.
4530 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
4531 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
4532 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
4534 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
4537 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
4540 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
4541 on the FUNCTION_DECL function context if it has a DIE that is a
4544 2017-09-25 Richard Biener <rguenther@suse.de>
4546 PR tree-optimization/82285
4547 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
4550 2017-09-25 Tom de Vries <tom@codesourcery.com>
4554 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
4557 2017-09-25 Richard Biener <rguenther@suse.de>
4559 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
4560 ISL errors other than isl_error_quota happen. Dump if the
4561 schedule is the same.
4562 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
4563 errors instead of aborting inside ISL.
4565 2017-09-25 Iain Sandoe <iain@codesourcery.com>
4568 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
4569 of libgcc_eh for m64.
4570 * config/i386/darwin64.h: Likewise.
4572 2017-09-25 Richard Biener <rguenther@suse.de>
4575 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
4576 attribute for incomplete types nor twice for complete ones.
4578 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
4581 * config/i386/i386.c (ix86_print_operand_address_as): Encode
4582 %esp as %rsp to avoid 0x67 prefix if there is no index or base
4585 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
4588 * config/i386/i386.opt (mprefer-avx256): Use
4589 ix86_target_flags variable.
4590 * config/i386/i386.c (ix86_target_string): Move
4591 -mprefer-avx256 to flag2_opts.
4593 2017-09-22 Jakub Jelinek <jakub@redhat.com>
4596 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
4597 and x != -1 | y != -1 into (x & y) != -1.
4599 2017-09-22 Steve Ellcey <sellcey@cavium.com>
4601 * config.gcc: Add new case statement to set
4602 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
4603 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
4604 s390x-*-linux* case statements. Added aarch64 to the list of
4605 supported architectures.
4607 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4609 PR tree-optimization/82289
4610 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
4611 STMT_VINFO_RELEVANT_P.
4613 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4614 Alan Hayward <alan.hayward@arm.com>
4615 David Sherwood <david.sherwood@arm.com>
4617 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
4618 for VR_RANGE only; don't allow VR_ANTI_RANGE.
4619 (extract_range_from_binary_expr_1): Don't call
4620 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
4622 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4623 Alan Hayward <alan.hayward@arm.com>
4624 David Sherwood <david.sherwood@arm.com>
4626 * target.def (preferred_vector_alignment): New hook.
4627 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
4629 * doc/tm.texi: Regenerate.
4630 * targhooks.h (default_preferred_vector_alignment): Declare.
4631 * targhooks.c (default_preferred_vector_alignment): New function.
4632 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
4634 (DR_TARGET_ALIGNMENT): New macro.
4635 (aligned_access_p): Update commentary.
4636 (vect_known_alignment_in_bytes): New function.
4637 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
4639 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
4640 Calculate the misalignment based on the target alignment rather than
4642 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
4643 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
4644 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
4645 the target alignment, rather than masking the element misalignment
4646 with the number of elements in a vector. Also use the target
4647 alignment when calculating the maximum number of peels.
4648 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
4649 instead of TYPE_ALIGN_UNIT.
4650 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
4651 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
4652 (vect_create_addr_base_for_vector_ref): Update call accordingly.
4653 (vect_create_data_ref_ptr): Likewise.
4654 (vect_setup_realignment): Realign by ANDing with
4655 -DR_TARGET_MISALIGNMENT.
4656 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
4657 the number of peels based on DR_TARGET_ALIGNMENT.
4658 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
4659 with the guaranteed alignment boundary when deciding whether
4661 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
4662 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
4663 (ensure_base_align): Remove stmt_info parameter. Get the
4664 target base alignment from DR_TARGET_ALIGNMENT.
4665 (vectorizable_store): Update call accordingly. Interpret
4666 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
4668 (vectorizable_load): Likewise.
4670 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4671 Alan Hayward <alan.hayward@arm.com>
4672 David Sherwood <david.sherwood@arm.com>
4674 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
4675 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
4676 (vect_enhance_data_refs_alignment): Likewise.
4678 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
4680 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
4681 error. Only quit immediately if parsing is complete.
4682 (BEGIN): Initialize fatal_err and parse_done.
4683 (begin fpu, end fpu): Check number of arguments.
4684 (begin arch, end arch): Likewise.
4685 (begin cpu, end cpu): Likewise.
4686 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
4687 (optalias): Likewise.
4689 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
4691 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
4692 * config/arm/arm-isa.h: Delete. Move definitions to ...
4693 * arm-cpus.in: ... here. Use new feature and fgroup values.
4694 * config/arm/arm.c (arm_option_override): Use lower case for feature
4696 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
4697 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
4698 * config/arm/parsecpu.awk (END): Add new command 'isa'.
4700 (print_isa_bits_for): New function.
4701 (gen_isa): New function.
4702 (gen_comm_data): Use print_isa_bits_for.
4703 (define feature): New keyword.
4704 (define fgroup): New keyword.
4705 * config/arm/t-arm (TM_H): Remove.
4706 (GTM_H): Add arm-isa.h.
4707 (arm-isa.h): Add rule to generate file.
4708 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
4709 case for feature bit names.
4711 2017-09-22 Richard Biener <rguenther@suse.de>
4713 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
4715 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
4716 print of no dependency loops ...
4717 * graphite.c (graphite_transform_loops): ... here.
4718 (canonicalize_loop_closed_ssa_form): Work from inner to outer
4720 (same_close_phi_node, remove_duplicate_close_phi,
4721 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
4722 (canonicalize_loop_closed_ssa): ... here and simplify.
4723 * graphite-optimize-isl.c: Include tree-vectorizer.h.
4724 (optimize_isl): Use dump_printf_loc to tell when we stopped
4725 optimizing because of an ISL timeout.
4727 2017-09-22 Richard Biener <rguenther@suse.de>
4729 PR tree-optimization/82291
4730 * tree-if-conv.c (predicate_mem_writes): Make sure to
4731 remove writes in blocks predicated with false.
4733 2017-09-22 Richard Biener <rguenther@suse.de>
4735 * sese.c: Include cfganal.h.
4736 (if_region_set_false_region): Remove.
4737 (create_if_region_on_edge): Likewise.
4738 (move_sese_in_condition): Re-implement without destroying
4741 2017-09-22 Richard Biener <rguenther@suse.de>
4743 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
4744 Verify both BBs contain loop PHI nodes before dispatching to
4746 (graphite_regenerate_ast_isl): Do not recompute dominators,
4747 do not verify three times. Restructure for clarity.
4748 * graphite-scop-detection.c (same_close_phi_node,
4749 remove_duplicate_close_phi, make_close_phi_nodes_unique,
4750 defined_in_loop_p, canonicalize_loop_closed_ssa,
4751 canonicalize_loop_closed_ssa_form): Simplify, remove excess
4752 checking and SSA rewrite, move to ...
4753 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
4754 (graphite_initialize): Do not pass in ctx, do not reset the
4755 SCEV cache, compute only dominators.
4756 (graphite_transform_loops): Allocate ISL ctx after
4757 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
4758 Maintain post-dominators only around build_scops.
4759 * sese.c (if_region_set_false_region): Make static. Free
4760 and recompute dominators.
4761 (move_sese_in_condition): Assert we don't get called with
4762 post-dominators computed.
4763 * sese.h (if_region_set_false_region): Remove.
4765 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
4767 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
4768 mode attribute for TARGET_AVX512VL.
4770 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
4772 * config/i386/i386.opt (mprefer-avx256): New option.
4773 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
4775 (ix86_preferred_simd_mode): Return 256-bit AVX modes
4776 for TARGET_PREFER_AVX256.
4777 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
4779 2017-09-21 Jeff Law <law@redhat.com>
4781 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
4782 Fix dump output if the only stack space is for pushed registers.
4784 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4786 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
4789 2017-09-21 Martin Sebor <msebor@redhat.com>
4792 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
4793 code (in C++) or code that triggers warnings.
4795 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
4797 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
4799 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
4801 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
4802 * sched-rgn.c: Ditto.
4803 * sel-sched-ir.c: Ditto.
4805 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
4807 * toplev.h (set_random_seed): Adjust return type.
4808 * toplev.c (init_local_tick): Move eager initialization of random_seed
4809 to get_random_seed. Adjust comment.
4810 (init_random_seed): Inline to get_random_seed, delete.
4811 (get_random_seed): Initialize random_seed lazily.
4812 (set_random_seed): Do not return previous value.
4813 (print_switch_value): Do not call get_random_seed.
4815 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
4817 * cgraph.c (delete_function_version): New, broken out from...
4818 (cgraph_node::delete_function_version): ...here. Rename to
4819 cgraph_node::delete_function_version_by_decl. Update all uses.
4820 (cgraph_node::remove): Call delete_function_version.
4822 2017-09-21 Jakub Jelinek <jakub@redhat.com>
4825 * tree-inline.c (expand_call_inline): Emit clobber stmts for
4826 VAR_DECLs to which addressable non-volatile parameters are mapped
4827 and for id->retvar after the return value assignment. Clear
4828 id->retval and id->retbnd after inlining.
4830 2017-09-21 Richard Biener <rguenther@suse.de>
4832 PR tree-optimization/82276
4833 PR tree-optimization/82244
4834 * tree-vrp.c (build_assert_expr_for): Set
4835 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
4837 (remove_range_assertions): Revert earlier change.
4839 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
4842 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
4844 2017-09-21 Richard Biener <rguenther@suse.de>
4846 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4847 Restore valid IL after code generation errors.
4848 * graphite.c (graphite_transform_loops): Diagnose code
4849 generation issues as MSG_MISSED_OPTIMIZATION and continue
4850 with processing SCOPs.
4852 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4853 Alan Hayward <alan.hayward@arm.com>
4854 David Sherwood <david.sherwood@arm.com>
4856 * calls.c (compute_argument_addresses): Use simplify_gen_binary
4857 rather than choosing between plus_constant and gen_rtx_<CODE>.
4858 * expr.c (emit_push_insn): Likewise.
4859 (expand_expr_real_2): Likewise.
4861 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4862 Alan Hayward <alan.hayward@arm.com>
4863 David Sherwood <david.sherwood@arm.com>
4865 * loop-unroll.c (split_iv): Call copy_rtx on the step.
4867 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4868 Alan Hayward <alan.hayward@arm.com>
4869 David Sherwood <david.sherwood@arm.com>
4871 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
4872 calling tree_to_uhwi.
4874 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4875 Alan Hayward <alan.hayward@arm.com>
4876 David Sherwood <david.sherwood@arm.com>
4878 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
4879 INTEGER_CST rather than a negative test for ADDR_EXPR.
4881 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4882 Alan Hayward <alan.hayward@arm.com>
4883 David Sherwood <david.sherwood@arm.com>
4885 * tree-vrp.c (extract_range_from_binary_expr_1): Check
4886 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
4888 2017-09-21 Richard Biener <rguenther@suse.de>
4890 PR tree-optimization/71351
4891 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
4892 graphite_create_new_loop_guard): Remove, fold remaining parts
4894 (translate_isl_ast_node_for): ... here and simplify.
4896 2017-09-21 Jakub Jelinek <jakub@redhat.com>
4899 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
4900 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
4901 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
4902 alternative always use QI mode, for -Os imov (=R,R) alternative
4903 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
4906 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4907 Jeff Law <law@redhat.com>
4909 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
4910 (allocate_stack_space): New function, partially extracted from
4912 (s390_emit_prologue): Track offset to most recent stack probe.
4913 Code to allocate space moved into allocate_stack_space.
4914 Dump actions when no stack is allocated.
4915 (s390_prologue_plus_offset): New function.
4916 (s390_emit_stack_probe): Likewise.
4918 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
4920 * common.opt (Wa, Wl, Wp, g, gz=): Add
4922 (gno-column-info): Remove.
4923 (gcolumn-info): Drop RejectNegative.
4925 (gno-record-gcc-switches): Remove.
4926 (grecord-gcc-switches): Drop RejectNegative.
4927 (gno-split-dwarf): Remove.
4928 (gsplit-dwarf): Drop RejectNegative.
4929 (gno-strict-dwarf): Remove.
4930 (gstrict-dwarf): Drop RejectNegative.
4931 * config/darwin.opt (gfull, gused): Add RejectNegative.
4932 * dwarf2out.c (gen_producer_string): Drop
4933 gno-record-gcc-switches handler.
4934 * optc-gen.awk: Add g to prefixes with negative forms.
4935 * opts-common.c (remapping_prefix_p): New.
4936 (find_opt): Check it.
4937 (generate_canonical_option): Test g prefix.
4938 (option_map): Add -gno- mapping.
4939 (add_misspelling_candidates): Check remapping_prefix_p.
4941 2017-09-20 Jeff Law <law@redhat.com>
4943 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
4944 thinko in stack clash protection support.
4946 * explow.c (compute_stack_clash_protection_loop_data): Use
4947 CONST_INT_P instead of explicit test. Verify object is a
4948 CONST_INT_P before looking at INTVAL.
4949 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
4950 instead of explicit test.
4952 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
4955 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
4956 address instead of to r1 and r11.
4958 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
4960 * config.gcc: Support "knm".
4961 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
4962 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4964 * config/i386/i386.c (m_KNM): Define.
4965 (processor_target_table): Add "knm".
4967 (ix86_option_override_internal): Add "knm".
4968 (ix86_issue_rate): Add PROCESSOR_KNM.
4969 (ix86_adjust_cost): Ditto.
4970 (ia32_multipass_dfa_lookahead): Ditto.
4971 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
4972 (fold_builtin_cpu): Add M_INTEL_KNM.
4973 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
4974 (processor_type): Add PROCESSOR_KNM.
4975 * config/i386/x86-tune.def: Add m_KNM.
4976 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
4978 2017-09-20 Richard Biener <rguenther@suse.de>
4980 PR tree-optimization/80213
4981 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
4982 are allowed in empty BBs as well.
4983 (canonicalize_loop_closed_ssa): Also look for other complex
4985 (scop_detection::get_sese): Include the loop-closed PHI block
4987 (scop_detection::merge_sese): Remove code adding extra blocks.
4988 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
4989 (build_scops): Assert the final returned scop is invalid.
4991 2017-09-20 Richard Biener <rguenther@suse.de>
4993 PR tree-optimization/82264
4994 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
4996 (vn_phi_lookup): Likewise.
4997 (vn_phi_insert): Likewise.
4999 2017-09-20 Jakub Jelinek <jakub@redhat.com>
5001 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
5002 that fits into uhwi or shwi, add DW_AT_const_value regardless
5003 of early_dwarf without going through RTL, using add_AT_unsigned
5006 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
5007 (DEBUG_LTO_ABBREV_SECTION): Likewise.
5008 (DEBUG_LTO_MACINFO_SECTION): Likewise.
5009 (DEBUG_MACRO_SECTION): Likewise.
5010 (DEBUG_LTO_MACRO_SECTION): Likewise.
5011 (DEBUG_STR_DWO_SECTION): Likewise.
5012 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
5013 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
5014 (DEBUG_LTO_DWO_LINE_SECTION): Define.
5015 (DEBUG_LTO_LINE_STR_SECTION): Define.
5016 (init_sections_and_labels): Initialize debug_line_str_section
5017 variable. Initialize debug_loc_section for -gdwarf-5 to
5018 DEBUG_LOCLISTS_SECTION. Formatting fixes.
5020 2017-09-20 Richard Biener <rguenther@suse.de>
5022 * graphite-sese-to-poly.c (extract_affine): Properly handle
5023 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
5025 2017-09-20 Richard Biener <rguenther@suse.de>
5027 PR tree-optimization/81373
5028 * graphite-scop-detection.c (build_cross_bb_scalars_def):
5029 Force SESE live-out defs to be handled even if they are
5032 2017-09-19 Jeff Law <law@redhat.com>
5034 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
5035 nothing for stack adjustments with REG_STACK_CHECK.
5036 * sched-deps.c (parse_add_or_inc): Reject insns with
5037 REG_STACK_CHECK from dependency breaking.
5038 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
5039 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
5040 * reg-notes.def (STACK_CHECK): New note.
5042 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
5043 (ix86_expand_prologue): Dump stack clash info as needed.
5044 Call ix86_adjust_stack_and_probe_stack_clash as needed.
5046 * function.c (dump_stack_clash_frame_info): New function.
5047 * function.h (dump_stack_clash_frame_info): Prototype.
5048 (enum stack_clash_probes): New enum.
5050 * config/alpha/alpha.c (alpha_expand_prologue): Also check
5051 flag_stack_clash_protection.
5052 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
5053 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
5054 (arm_frame_pointer_required): Likewise.
5055 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
5056 (ia64_expand_prologue): Likewise.
5057 * config/mips/mips.c (mips_expand_prologue): Likewise.
5058 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
5059 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
5060 (sparc_flat_expand_prologue): Likewise.
5061 * config/spu/spu.c (spu_expand_prologue): Likewise.
5063 * explow.c: Include "params.h".
5064 (anti_adjust_stack_and_probe_stack_clash): New function.
5065 (get_stack_check_protect): Likewise.
5066 (compute_stack_clash_protection_loop_data): Likewise.
5067 (emit_stack_clash_protection_loop_start): Likewise.
5068 (emit_stack_clash_protection_loop_end): Likewise.
5069 (allocate_dynamic_stack_space): Use get_stack_check_protect.
5070 Use anti_adjust_stack_and_probe_stack_clash.
5071 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
5072 (emit_stack_clash_protection_loop_start): Likewise.
5073 (emit_stack_clash_protection_loop_end): Likewise.
5074 * rtl.h (get_stack_check_protect): Prototype.
5075 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
5076 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
5077 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
5079 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
5081 * doc/tm.texi: Rebuilt.
5082 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
5083 get_stack_check_protect.
5084 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
5085 * config/arm/arm.c (arm_expand_prologue): Likewise.
5086 (arm_frame_pointer_required): Likewise.
5087 * config/i386/i386.c (ix86_expand_prologue): Likewise.
5088 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
5089 * config/mips/mips.c (mips_expand_prologue): Likewise.
5090 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
5091 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
5092 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
5093 (sparc_flat_expand_prologue): Likewise.
5095 * common.opt (-fstack-clash-protection): New option.
5096 * flag-types.h (enum stack_check_type): Note difference between
5097 -fstack-check= and -fstack-clash-protection.
5098 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
5099 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
5100 * toplev.c (process_options): Issue warnings/errors for cases
5101 not handled with -fstack-clash-protection.
5102 * doc/invoke.texi (-fstack-clash-protection): Document new option.
5103 (-fstack-check): Note additional problem with -fstack-check=generic.
5104 Note that -fstack-check is primarily for Ada and refer users
5105 to -fstack-clash-protection for stack-clash-protection.
5106 Document new params for stack clash protection.
5108 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
5110 * config/i386/i386.c (ix86_split_long_move): Do not handle
5111 address used for LEA in a special way.
5113 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
5115 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
5117 2017-09-19 Martin Sebor <msebor@redhat.com>
5120 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
5121 of incompatible types.
5123 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
5125 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5126 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
5127 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5128 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
5130 2017-09-19 Richard Biener <rguenther@suse.de>
5132 PR tree-optimization/82244
5133 * tree-vrp.c (remove_range_assertions): Do not propagate
5134 a constant to abnormals but replace the assert with a copy.
5136 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
5138 PR rtl-optimization/57878
5139 PR rtl-optimization/68988
5140 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
5141 avoidance test involving non_reload_pseudos. Move frequency test
5142 below the general fragmentation avoidance test.
5144 2017-09-19 Richard Biener <rguenther@suse.de>
5146 PR tree-optimization/69728
5147 * graphite-sese-to-poly.c (schedule_error): New global.
5148 (add_loop_schedule): Handle empty domain by failing the
5150 (build_original_schedule): Handle schedule_error.
5152 2017-09-19 Richard Biener <rguenther@suse.de>
5154 * graphite-scop-detection.c (scop_detection::can_represent_loop):
5155 Do not iterate to sibling loops but only to siblings of inner
5158 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
5161 * config/m68k/m68k.md (moveq feeding equality comparison): Check
5162 that the registers are different.
5164 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
5166 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
5167 to processor_model and "amdfam17h" to arch_names_table.
5168 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
5170 2017-09-18 Jakub Jelinek <jakub@redhat.com>
5173 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
5175 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5176 Alan Hayward <alan.hayward@arm.com>
5177 David Sherwood <david.sherwood@arm.com>
5179 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
5181 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
5183 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
5184 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
5185 vect_schedule_slp_instance.
5186 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
5187 Update call to vect_slp_analyze_node_operations. Simplify return
5189 (vect_slp_analyze_bb_1): Update call accordingly.
5190 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
5191 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
5192 (vect_schedule_slp): Update call accordingly.
5194 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5195 Alan Hayward <alan.hayward@arm.com>
5196 David Sherwood <david.sherwood@arm.com>
5198 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
5199 with types that aren't in fact scalar.
5201 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5203 * tree-vect-slp.c (vect_record_max_nunits): New function,
5205 (vect_build_slp_tree_1): ...here.
5206 (vect_build_slp_tree_2): Call it for phis too.
5208 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5210 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
5211 to vect_get_vec_def_for_operand when getting the mask operand.
5213 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5214 Alan Hayward <alan.hayward@arm.com>
5215 David Sherwood <david.sherwood@arm.com>
5217 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
5220 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5221 Alan Hayward <alan.hayward@arm.com>
5222 David Sherwood <david.sherwood@arm.com>
5224 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
5225 calculation for vector booleans.
5227 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5228 Alan Hayward <alan.hayward@arm.com>
5229 David Sherwood <david.sherwood@arm.com>
5231 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
5233 (vect_transform_stmt): ...here.
5234 (vect_analyze_stmt): Use it instead of calling
5235 vectorizable_live_operation directly.
5237 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
5239 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
5240 non-SIMT targets in acc vector loops.
5242 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
5244 * configure.ac: Add arc and check if assembler supports gdwarf2.
5245 * configure: Regenerate.
5247 2017-09-18 Richard Biener <rguenther@suse.de>
5249 PR tree-optimization/82220
5250 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
5251 epilogue niters from the min_profitable_iters compute.
5253 2017-09-18 Jakub Jelinek <jakub@redhat.com>
5256 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
5257 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
5258 (ix86_init_pic_reg): Revert 2017-09-01 changes.
5260 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
5263 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
5264 switching to a new text section.
5266 2017-09-18 Richard Biener <rguenther@suse.de>
5268 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
5270 (build_alias_set): Reject aliases with no access function.
5272 2017-09-18 Richard Biener <rguenther@suse.de>
5274 PR tree-optimization/79622
5275 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
5277 (build_cross_bb_scalars_use): Likewise.
5279 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
5281 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
5283 2017-09-18 Alan Modra <amodra@gmail.com>
5286 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
5287 stack_pointer_rtx for count 0. Update comments. Break up
5288 large rtl expression.
5290 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
5292 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
5293 Increase to 20 bytes.
5294 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
5295 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
5296 or avx version of the stub.
5298 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
5301 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
5302 compute the minimum stack alignment. Also update preferred stack
5303 boundary for leaf functions.
5305 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
5307 PR tree-optimization/82228
5308 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
5311 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
5313 * common/config/nds32/nds32-common.c
5314 (nds32_option_optimization_table): Refine formatting.
5315 (nds32_option_optimization_table): Use -fsched-pressure and
5316 -fomit-frame-pointer for specific optimization level.
5318 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
5320 * config/nds32/nds32.c: Refine formatting and comments.
5321 * config/nds32/nds32.h: Likewise.
5322 * config/nds32/nds32.md: Likewise.
5323 * config/nds32/nds32-cost.c: Likewise.
5324 * config/nds32/nds32-isr.c: Likewise.
5325 * config/nds32/nds32-md-auxiliary.c: Likewise.
5326 * config/nds32/nds32-multiple.md: Likewise.
5327 * config/nds32/nds32-predicates.c: Likewise.
5329 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
5330 Jakub Jelinek <jakub@redhat.com>
5332 Add support for -std=c++2a.
5333 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
5335 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
5337 2017-09-15 Steve Ellcey <sellcey@cavium.com>
5340 * doc/extend.texi (Common Function Attributes): Add
5341 references to ARM, AArch64, and S/390 specific attributes.
5342 (Function Specific Option Pragmas): Add AArch64 and S/390
5343 to list of back ends that support the target pragma.
5345 2017-09-15 Nathan Sidwell <nathan@acm.org>
5347 * doc/standards.texi: Fix C++17 description. Update URLs for
5350 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
5352 * common.opt (Wcast-align=strict): New warning option.
5353 * doc/invoke.texi: Document -Wcast-align=strict.
5355 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
5357 * cgraph.h (cgraph_thunk_info): Add comments.
5358 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
5359 assert for VIRTUAL_* arguments stricter.
5361 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
5363 PR tree-optimization/71026
5364 * match.pd: Move RDIV patterns from fold-const.c
5365 * fold-const.c (distribute_real_division): Removed.
5366 (fold_binary_loc): Remove calls to distribute_real_divison.
5368 2017-09-15 Jakub Jelinek <jakub@redhat.com>
5370 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
5371 c++1z and gnu++1z as deprecated. Change other references to
5372 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
5373 Change -Wc++1z-compat to -Wc++17-compat.
5374 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
5375 * dwarf2out.c (highest_c_language): Handle C++17.
5376 (gen_compile_unit_die): Likewise.
5378 2017-09-15 Jakub Jelinek <jakub@redhat.com>
5380 PR rtl-optimization/82192
5381 * combine.c (make_extraction): Don't look through non-paradoxical
5382 SUBREGs or TRUNCATE if pos + len is or might be bigger than
5385 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
5386 Alan Hayward <alan.hayward@arm.com>
5387 David Sherwood <david.sherwood@arm.com>
5389 * target.def (function_arg_offset): New hook.
5390 * targhooks.h (default_function_arg_offset): Declare.
5391 * targhooks.c (default_function_arg_offset): New function.
5392 * function.c (locate_and_pad_parm): Use
5393 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
5394 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
5395 (TARGET_FUNCTION_ARG_OFFSET): ...this.
5396 * doc/tm.texi: Regenerate.
5397 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
5398 * config/spu/spu.c (spu_function_arg_offset): New function.
5399 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
5400 * system.h (FUNCTION_ARG_OFFSET): Poison.
5402 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
5403 Alan Hayard <alan.hayward@arm.com>
5404 David Sherwood <david.sherwood@arm.com>
5406 * target.def (truly_noop_truncation): New hook.
5407 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
5408 than TRULY_NOOP_TRUNCATION.
5409 * hooks.h (hook_bool_uint_uint_true): Declare.
5410 * hooks.c (hook_bool_uint_uint_true): New function.
5411 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
5412 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
5413 * doc/tm.texi: Regenerate.
5414 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
5415 rather than TRULY_NOOP_TRUNCATION in comments.
5416 (simplify_comparison): Likewise.
5417 (record_truncated_value): Likewise.
5418 * expmed.c (extract_bit_field_1): Likewise.
5419 (extract_split_bit_field): Likewise.
5420 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
5421 instead of TRULY_NOOP_TRUNCATION.
5422 * function.c (assign_parm_setup_block): Likewise.
5423 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
5424 * rtlhooks.c: Include target.h.
5425 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
5426 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
5427 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
5428 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
5429 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
5430 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
5431 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
5432 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
5433 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
5434 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
5435 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
5436 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
5437 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
5438 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
5439 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
5440 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
5441 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
5442 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
5443 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
5444 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
5445 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
5446 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
5447 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
5448 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
5449 * config/mips/mips.c (mips_truly_noop_truncation): New function.
5450 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
5451 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
5452 rather than TRULY_NOOP_TRUNCATION in comments.
5453 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
5454 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
5455 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
5456 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
5457 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
5458 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
5459 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
5460 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
5461 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
5462 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
5463 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
5464 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
5465 rather than TRULY_NOOP_TRUNCATION in comments.
5466 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
5467 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
5468 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
5469 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
5470 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
5471 TRULY_NOOP_TRUNCATION condition.
5472 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
5473 (TRULY_NOOP_TRUNCATION): Delete.
5474 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
5475 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
5476 * config/spu/spu.c (spu_truly_noop_truncation): New function.
5477 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
5478 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
5479 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
5480 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
5481 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
5482 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
5483 rather than TRULY_NOOP_TRUNCATION in comments.
5484 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
5485 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
5486 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
5487 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
5488 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
5489 * system.h (TRULY_NOOP_TRUNCATION): Poison.
5491 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
5494 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
5495 (*cmp_ior): Likewise.
5496 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
5497 (*ior_scc_scc_cmp): Likewise.
5498 (*and_scc_scc): Likewise.
5499 (*and_scc_scc_cmp): Likewise.
5501 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
5502 Alan Hayard <alan.hayward@arm.com>
5503 David Sherwood <david.sherwood@arm.com>
5505 * target.def (can_change_mode_class): New hook.
5506 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
5507 (hard_regno_nregs): Likewise.
5508 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
5509 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
5510 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
5511 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
5512 (LOAD_EXTEND_OP): Update accordingly.
5513 * doc/tm.texi: Regenerate.
5514 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
5515 CANNOT_CHANGE_MODE_CLASS.
5516 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
5517 (REG_CAN_CHANGE_MODE_P): ...this new macro.
5518 * combine.c (simplify_set): Update accordingly.
5519 * emit-rtl.c (validate_subreg): Likewise.
5520 * recog.c (general_operand): Likewise.
5521 * regcprop.c (mode_change_ok): Likewise.
5522 * reload1.c (choose_reload_regs): Likewise.
5523 (inherit_piecemeal_p): Likewise.
5524 * rtlanal.c (simplify_subreg_regno): Likewise.
5525 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
5526 instead of CANNOT_CHANGE_MODE_CLASS.
5527 (reload_cse_simplify_operands): Likewise.
5528 * reload.c (push_reload): Use targetm.can_change_mode_class
5529 instead of CANNOT_CHANGE_MODE_CLASS.
5530 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
5531 REG_CANNOT_CHANGE_MODE_P.
5532 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5533 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
5534 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5535 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5536 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5537 (arm_can_change_mode_class): New function.
5538 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
5539 than CANNOT_CHANGE_MODE_CLASS in comments.
5540 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5541 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
5542 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
5543 (ix86_can_change_mode_class): ...this new function, inverting the
5544 sense of the return value.
5545 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5546 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5547 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5548 (ia64_can_change_mode_class): New function.
5549 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5550 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
5551 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
5552 (m32c_can_change_mode_class): ...this new function, inverting the
5553 sense of the return value.
5554 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5555 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5556 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
5557 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
5558 (mips_can_change_mode_class): ...this new function, inverting the
5559 sense of the return value.
5560 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5561 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5562 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5563 (msp430_can_change_mode_class): New function.
5564 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5565 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
5566 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5567 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5568 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5569 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
5570 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5571 (pa_cannot_change_mode_class): Replace with...
5572 (pa_can_change_mode_class): ...this new function, inverting the
5573 sense of the return value.
5574 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
5575 than CANNOT_CHANGE_MODE_CLASS in comments.
5576 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5577 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
5578 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5579 (pdp11_cannot_change_mode_class): Replace with...
5580 (pdp11_can_change_mode_class): ...this new function, inverting the
5581 sense of the return value.
5582 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5583 * config/powerpcspe/powerpcspe-protos.h
5584 (rs6000_cannot_change_mode_class_ptr): Delete.
5585 * config/powerpcspe/powerpcspe.c
5586 (rs6000_cannot_change_mode_class_ptr): Delete.
5587 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5588 (rs6000_option_override_internal): Assign to
5589 targetm.can_change_mode_class instead of
5590 rs6000_cannot_change_mode_class_ptr.
5591 (rs6000_cannot_change_mode_class): Replace with...
5592 (rs6000_can_change_mode_class): ...this new function, inverting the
5593 sense of the return value.
5594 (rs6000_debug_cannot_change_mode_class): Replace with...
5595 (rs6000_debug_can_change_mode_class): ...this new function.
5596 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5597 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
5598 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5599 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5600 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
5602 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
5603 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5604 (rs6000_option_override_internal): Assign to
5605 targetm.can_change_mode_class instead of
5606 rs6000_cannot_change_mode_class_ptr.
5607 (rs6000_cannot_change_mode_class): Replace with...
5608 (rs6000_can_change_mode_class): ...this new function, inverting the
5609 sense of the return value.
5610 (rs6000_debug_cannot_change_mode_class): Replace with...
5611 (rs6000_debug_can_change_mode_class): ...this new function.
5612 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5613 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
5614 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
5615 (s390_can_change_mode_class): ...this new function, inverting the
5616 sense of the return value.
5617 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5618 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5619 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
5620 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5621 (sh_cannot_change_mode_class): Replace with...
5622 (sh_can_change_mode_class): ...this new function, inverting the
5623 sense of the return value.
5624 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5625 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5626 (sparc_can_change_mode_class): New function.
5627 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5628 * config/spu/spu.c (spu_can_change_mode_class): New function.
5629 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5630 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5631 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5632 (visium_can_change_mode_class): New function.
5633 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
5635 2017-09-15 Richard Biener <rguenther@suse.de>
5637 PR tree-optimization/82217
5638 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
5639 but not undefined case.
5641 2017-09-15 Jakub Jelinek <jakub@redhat.com>
5644 * postreload.c (reload_cse_simplify_operands): Skip
5645 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
5647 2017-09-15 Richard Biener <rguenther@suse.de>
5649 PR tree-optimization/68823
5650 * graphite-scop-detection.c (build_alias_set): If we have a
5651 possible dependence check whether we can handle them by just
5652 looking at the DRs DR_ACCESS_FNs.
5653 (build_scops): If build_alias_set fails, fail the SCOP.
5655 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
5657 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
5658 to support float128 built-in functions that require the ISA 3.0
5660 (BU_FLOAT128_3_HW): Likewise.
5661 (SQRTF128): Add support for the IEEE 128-bit square root and fma
5663 (FMAF128): Likewise.
5665 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
5666 support for built-in functions that need the ISA 3.0 IEEE 128-bit
5667 floating point instructions.
5668 (rs6000_invalid_builtin): Likewise.
5669 (rs6000_builtin_mask_names): Likewise.
5670 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
5671 (RS6000_BTM_FLOAT128_HW): Likewise.
5672 (RS6000_BTM_COMMON): Likewise.
5673 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
5675 * doc/extend.texi (RS/6000 built-in functions): Document the
5676 IEEE 128-bit floating point square root and fused multiply-add
5679 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
5681 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
5682 reg (r2) isn't in the set of registers defined in the prologue.
5684 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5685 Alan Hayward <alan.hayward@arm.com>
5686 David Sherwood <david.sherwood@arm.com>
5688 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
5689 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
5690 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
5691 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
5692 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
5694 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
5695 max_vectorization_factor.
5696 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
5698 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5699 Alan Hayward <alan.hayward@arm.com>
5700 David Sherwood <david.sherwood@arm.com>
5702 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
5703 (vect_worthwhile_without_simd_p): Declare.
5704 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
5705 (vectorizable_reduction): Use it.
5706 * tree-vect-stmts.c (vectorizable_shift): Likewise.
5707 (vectorizable_operation): Likewise.
5709 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5710 Alan Hayward <alan.hayward@arm.com>
5711 David Sherwood <david.sherwood@arm.com>
5713 * tree-vectorizer.h (vect_get_num_copies): New function.
5714 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
5715 * tree-vect-loop.c (vectorizable_reduction): Likewise.
5716 (vectorizable_induction): Likewise.
5717 (vectorizable_live_operation): Likewise.
5718 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
5719 (vectorizable_bswap): Likewise.
5720 (vectorizable_call): Likewise.
5721 (vectorizable_conversion): Likewise.
5722 (vectorizable_assignment): Likewise.
5723 (vectorizable_shift): Likewise.
5724 (vectorizable_operation): Likewise.
5725 (vectorizable_store): Likewise.
5726 (vectorizable_load): Likewise.
5727 (vectorizable_condition): Likewise.
5728 (vectorizable_comparison): Likewise.
5729 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
5731 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5732 Alan Hayward <alan.hayward@arm.com>
5733 David Sherwood <david.sherwood@arm.com>
5735 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
5736 of vect_init_vector.
5738 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5739 Alan Hayward <alan.hayward@arm.com>
5740 David Sherwood <david.sherwood@arm.com>
5742 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
5743 an inline wrapper that provides a location.
5744 (gimple_build_vector): Likewise.
5745 * gimple-fold.c (gimple_build_vector_from_val): New function.
5746 (gimple_build_vector): Likewise.
5747 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
5748 functions to build the initial value. Always return a gimple value.
5749 (get_initial_defs_for_reduction): Likewise. Only compute
5751 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
5752 vect_init_vector on the results from get_initial_def(s)_for_reduction.
5753 (vectorizable_induction): Use gimple_build_vector rather than
5756 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5757 Alan Hayward <alan.hayward@arm.com>
5758 David Sherwood <david.sherwood@arm.com>
5760 * target.h (vec_perm_indices): New typedef.
5761 (auto_vec_perm_indices): Likewise.
5762 * optabs-query.h: Include target.h
5763 (can_vec_perm_p): Take a vec_perm_indices *.
5764 * optabs-query.c (can_vec_perm_p): Likewise.
5765 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
5766 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
5767 * tree-vect-generic.c (lower_vec_perm): Likewise.
5768 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
5769 (vect_grouped_load_supported): Likewise.
5770 (vect_shift_permute_load_chain): Likewise.
5771 (vect_permute_store_chain): Use auto_vec_perm_indices.
5772 (vect_permute_load_chain): Likewise.
5773 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
5774 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
5775 Update uses of can_vec_perm_p.
5776 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
5777 mode with a number of elements. Take a vec_perm_indices *.
5778 (vect_create_epilog_for_reduction): Update accordingly.
5779 Use auto_vec_perm_indices.
5780 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
5781 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
5782 (vect_transform_slp_perm_load): Likewise.
5783 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
5784 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
5785 (vect_gen_perm_mask_checked): Likewise.
5786 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
5787 (vect_gen_perm_mask_checked): Likewise.
5788 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
5789 (vectorizable_store): Likewise.
5790 (vectorizable_load): Likewise.
5791 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
5792 (vectorizable_bswap): Likewise.
5794 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5795 Alan Hayward <alan.hayward@arm.com>
5796 David Sherwood <david.sherwood@arm.com>
5798 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
5799 * tree.c (build_vector): Likewise.
5800 (build_vector_from_ctor): Update accordingly.
5801 (build_vector_from_val): Likewise.
5802 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
5803 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
5804 * tree-vect-generic.c (add_rshift): Likewise.
5805 (expand_vector_divmod): Likewise.
5806 (optimize_vector_constructor): Likewise.
5807 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
5808 (vect_transform_slp_perm_load): Likewise.
5809 (vect_schedule_slp_instance): Likewise.
5810 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
5811 (vectorizable_call): Likewise.
5812 (vect_gen_perm_mask_any): Likewise. Add elements in order.
5813 * expmed.c (make_tree): Likewise.
5814 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
5815 a vector passed to build_vector.
5816 (fold_convert_const): Likewise.
5817 (exact_inverse): Likewise.
5818 (fold_ternary_loc): Likewise.
5819 (fold_relational_const): Likewise.
5820 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
5821 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
5822 (const_unop): Likewise. Store the reduction accumulator in a
5823 variable rather than an array.
5824 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
5825 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
5826 the new vector, rather than constructing it after the input arrays.
5827 (native_interpret_vector): Use auto_vec<tree> when building
5828 a vector passed to build_vector. Add elements in order.
5829 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
5830 auto_vec<tree> when building a vector passed to build_vector.
5831 (vect_create_epilog_for_reduction): Likewise.
5832 (vectorizable_induction): Likewise.
5833 (get_initial_def_for_reduction): Likewise. Fix indentation of
5835 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
5837 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
5838 passed to build_vector.
5840 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5841 Alan Hayward <alan.hayward@arm.com>
5842 David Sherwood <david.sherwood@arm.com>
5844 * tree-core.h (tree_base::u): Add an "nelts" field.
5845 (tree_vector): Use VECTOR_CST_NELTS as the length.
5846 * tree.c (tree_size): Likewise.
5847 (make_vector): Initialize VECTOR_CST_NELTS.
5848 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
5849 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
5850 TYPE_VECTOR_SUBPARTS.
5851 * expr.c (const_vector_mask_from_tree): Consistently use "units"
5852 as the number of units, setting it from VECTOR_CST_NELTS.
5853 (const_vector_from_tree): Likewise.
5854 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
5855 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
5856 (fold_negate_expr_1): Likewise.
5857 (fold_convert_const): Likewise.
5858 (const_binop): Likewise. Differentiate the number of output and
5860 (const_unop): Likewise.
5861 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
5862 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
5863 in cases that did the opposite.
5865 2017-09-14 Richard Biener <rguenther@suse.de>
5867 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
5870 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
5872 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
5874 2017-09-14 Jakub Jelinek <jakub@redhat.com>
5877 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
5878 if and where to split a bb, except for splitting before debug insn
5879 sequences followed by non-label real insn. Delete debug insns
5880 in between basic blocks.
5882 * combine.c (make_compound_operation_int): Formatting fixes.
5884 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
5885 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5886 * config/netbsd.h (LINK_EH_SPEC): Likewise.
5887 * config/sol2.h (LINK_EH_SPEC): Likewise.
5888 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5889 * config/s390/linux.h (LINK_SPEC): Likewise.
5890 * config/freebsd.h (LINK_EH_SPEC): Likewise.
5891 * config/openbsd.h (LINK_EH_SPEC): Likewise.
5892 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5893 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
5894 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
5895 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5896 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
5897 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
5899 2017-09-13 Jakub Jelinek <jakub@redhat.com>
5901 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
5903 (ENDFILE_LINUX_SPEC): Likewise.
5904 (LINK_EH_SPEC): Likewise.
5905 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
5906 (LINK_OS_LINUX_SPEC32): Likewise.
5907 (LINK_OS_LINUX_SPEC64): Likewise.
5908 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
5909 (LINK_OS_LINUX_SPEC): Likewise.
5911 2017-09-13 Martin Liska <mliska@suse.cz>
5914 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
5915 CASE_HIGH is NULL_TREE.
5917 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
5918 Alan Hayward <alan.hayward@arm.com>
5919 David Sherwood <david.sherwood@arm.com>
5921 * target.def (secondary_memory_needed): New hook.
5922 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
5923 instead of SECONDARY_MEMORY_NEEDED.
5924 (secondary_memory_needed_mode): Likewise.
5925 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
5926 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
5927 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
5928 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
5929 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
5930 * doc/tm.texi: Regenerate.
5931 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
5932 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
5933 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5934 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
5935 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
5936 * config/i386/i386.c (inline_secondary_memory_needed): Put the
5937 mode argument first and change the reg_class arguments to reg_class_t.
5938 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
5939 Make static. Update the call to inline_secondary_memory_needed.
5940 (ix86_register_move_cost): Update the call to
5941 inline_secondary_memory_needed.
5942 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5943 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
5945 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
5946 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
5948 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
5949 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
5950 * config/mips/mips.c (mips_secondary_memory_needed): Make static
5951 and match hook interface. Add comment from mips.h.
5952 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5953 * config/mmix/mmix.md (truncdfsf2): Refer to
5954 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
5956 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
5957 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
5958 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5959 (pa_secondary_memory_needed): New function.
5960 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
5961 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
5962 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5963 (pdp11_secondary_memory_needed): Make static and match hook interface.
5964 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
5965 * config/powerpcspe/powerpcspe-protos.h
5966 (rs6000_secondary_memory_needed_ptr): Delete.
5967 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
5969 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5970 (rs6000_option_override_internal): Assign to
5971 targetm.secondary_memory_needed rather than
5972 rs6000_secondary_memory_needed_ptr.
5973 (rs6000_secondary_memory_needed): Match hook interface.
5974 (rs6000_debug_secondary_memory_needed): Likewise.
5975 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
5976 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
5977 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
5978 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5979 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
5980 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
5982 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
5983 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5984 (rs6000_option_override_internal): Assign to
5985 targetm.secondary_memory_needed rather than
5986 rs6000_secondary_memory_needed_ptr.
5987 (rs6000_secondary_memory_needed): Match hook interface.
5988 (rs6000_debug_secondary_memory_needed): Likewise.
5989 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
5990 * config/s390/s390.c (s390_secondary_memory_needed): New function.
5991 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5992 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
5993 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5994 (sparc_secondary_memory_needed): New function.
5995 * lra-constraints.c (check_and_process_move): Refer to
5996 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
5998 (curr_insn_transform): Likewise.
5999 (process_alt_operands): Use targetm.secondary_memory_needed
6000 instead of TARGET_SECONDARY_MEMORY_NEEDED.
6001 (check_secondary_memory_needed_p): Likewise.
6002 (choose_split_class): Likewise.
6003 * reload.c: Unconditionally include code that was previously
6004 conditional on SECONDARY_MEMORY_NEEDED.
6005 (push_secondary_reload): Use targetm.secondary_memory_needed
6006 instead of TARGET_SECONDARY_MEMORY_NEEDED.
6007 (push_reload): Likewise.
6008 * reload1.c: Unconditionally include code that was previously
6009 conditional on SECONDARY_MEMORY_NEEDED.
6010 (choose_reload_regs): Use targetm.secondary_memory_needed
6011 instead of TARGET_SECONDARY_MEMORY_NEEDED.
6012 (gen_reload): Likewise.
6013 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
6015 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
6016 Alan Hayward <alan.hayward@arm.com>
6017 David Sherwood <david.sherwood@arm.com>
6019 * target.def (secondary_memory_needed_mode): New hook:
6020 * targhooks.c (default_secondary_memory_needed_mode): Declare.
6021 * targhooks.h (default_secondary_memory_needed_mode): New function.
6022 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
6023 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
6024 * doc/tm.texi: Regenerate.
6025 * lra-constraints.c (check_and_process_move): Use
6026 targetm.secondary_memory_needed_mode instead of
6027 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
6028 (curr_insn_transform): Likewise.
6029 * reload.c (get_secondary_mem): Likewise.
6030 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6031 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
6033 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
6034 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6035 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
6036 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
6037 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
6039 * config/powerpcspe/powerpcspe-protos.h
6040 (rs6000_secondary_memory_needed_mode): Delete.
6041 * config/powerpcspe/powerpcspe.c
6042 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
6043 (rs6000_secondary_memory_needed_mode): Make static.
6044 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6045 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
6047 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
6049 (rs6000_secondary_memory_needed_mode): Make static.
6050 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6051 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
6052 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
6053 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6054 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
6056 (sparc_secondary_memory_needed_mode): New function.
6057 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
6059 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
6061 * config/aarch64/constraints.md (Umq): New constraint.
6062 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
6064 (mov<mode>): Update condition.
6066 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6068 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
6069 when bitposition is the same.
6071 2017-09-13 Richard Biener <rguenther@suse.de>
6073 * dwarf2out.c (output_die_symbol): Remove.
6074 (output_die): Do not output a DIEs symbol.
6076 2017-09-13 Richard Biener <rguenther@suse.de>
6079 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
6080 default-def to avoid breaking iterator update with the weird
6081 interaction with cgraph_update_edges_for_call_stmt_node.
6083 2017-09-13 Richard Biener <rguenther@suse.de>
6085 * tree-cfg.c (verify_gimple_assign_binary): Add verification
6086 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
6087 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
6088 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
6090 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
6092 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
6093 Disable pc relative literal load irrespective of
6094 TARGET_FIX_ERR_A53_84341 for default.
6096 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
6098 * config/sparc/sparc.c (output_return): Output the source location of
6099 the insn in the delay slot, if any.
6100 (output_sibcall): Likewise.
6102 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
6105 * common.opt (-static-pie): New alias.
6106 (shared): Negate static-pie.
6107 (-no-pie): Update help text.
6109 (static-pie): New option.
6110 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
6111 -static-pie support.
6112 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
6113 (LINK_EH_SPEC): Likewise.
6114 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6115 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
6116 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
6117 * gcc.c (LINK_COMMAND_SPEC): Likewise.
6118 (init_gcc_specs): Likewise.
6119 (init_spec): Likewise.
6120 (display_help): Update help message for -pie.
6121 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
6124 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
6126 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
6127 (movdi_aarch64): Likewise.
6128 (movti_aarch64): Likewise.
6130 2017-09-12 Simon Wright <simon@pushface.org>
6133 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
6134 calculation of the minor version, always output as 0.
6136 2017-09-12 Jakub Jelinek <jakub@redhat.com>
6139 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
6140 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
6141 on it early, rather than manual conversion late. For
6142 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
6143 instead of performing manual conversion.
6145 2017-09-12 Carl Love <cel@us.ibm.com>
6147 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
6148 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
6150 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
6151 VMULOSW): Add definitions.
6152 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6153 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
6154 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
6155 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
6156 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
6158 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
6160 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
6162 (movti_aarch64): Likewise.
6163 (movdf_aarch64): Likewise.
6164 (movtf_aarch64): Likewise.
6165 (load_pairdi): Likewise.
6166 (store_pairdi): Likewise.
6167 (load_pairdf): Likewise.
6168 (store_pairdf): Likewise.
6169 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
6170 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
6171 (ldr_got_small_<mode>): Likewise.
6172 (ldr_got_small_28k_<mode>): Likewise.
6173 (ldr_got_tiny): Likewise.
6174 * config/aarch64/iterators.md (ldst_sz): New.
6175 (ldpstp_sz): Likewise.
6176 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
6178 (thunderx_load): Split load_8 to load_16.
6179 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
6181 (thunderx2t99_storepair_basic): Split store_8 to store_16.
6182 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
6183 (xgene1_store_pair): Split store_8 to store_16.
6184 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
6185 (falkor_st_0_st_sd): Split store_8 to store_16.
6187 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
6189 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
6190 and store1/2/3/4 to store_4/8/12/16.
6191 * config/aarch64/aarch64.md: Update for rename.
6192 * config/arm/arm.md: Likewise.: Likewise.
6193 * config/arm/arm.c: Likewise.
6194 * config/arm/thumb1.md: Likewise.
6195 * config/arm/thumb2.md: Likewise.
6196 * config/arm/vfp.md: Likewise.
6197 * config/arm/arm-generic.md: Likewise.
6198 * config/arm/arm1020e.md: Likewise.
6199 * config/arm/arm1026ejs.md: Likewise.
6200 * config/arm/arm1136jfs.md: Likewise.
6201 * config/arm/arm926ejs.md: Likewise.
6202 * config/arm/cortex-a15.md: Likewise.
6203 * config/arm/cortex-a17.md: Likewise.
6204 * config/arm/cortex-a5.md: Likewise.
6205 * config/arm/cortex-a53.md: Likewise.
6206 * config/arm/cortex-a57.md: Likewise.
6207 * config/arm/cortex-a7.md: Likewise.
6208 * config/arm/cortex-a8.md: Likewise.
6209 * config/arm/cortex-a9.md: Likewise.
6210 * config/arm/cortex-m4.md: Likewise.
6211 * config/arm/cortex-m7.md: Likewise.
6212 * config/arm/cortex-r4.md: Likewise.
6213 * config/arm/exynos-m1.md: Likewise.
6214 * config/arm/fa526.md: Likewise.
6215 * config/arm/fa606te.md: Likewise.
6216 * config/arm/fa626te.md: Likewise.
6217 * config/arm/fa726te.md: Likewise.
6218 * config/arm/fmp626.md: Likewise.
6219 * config/arm/iwmmxt.md: Likewise.
6220 * config/arm/ldmstm.md: Likewise.
6221 * config/arm/marvell-pj4.md: Likewise.
6222 * config/arm/xgene1.md: Likewise.
6223 * config/aarch64/thunderx.md: Likewise.
6224 * config/aarch64/thunderx2t99.md: Likewise.
6225 * config/aarch64/falkor.md: Likewise.
6227 2017-09-12 Martin Liska <mliska@suse.cz>
6229 * attribs.c (private_lookup_attribute): New function.
6230 * attribs.h (private_lookup_attribute): Declared here.
6231 (lookup_attribute): Called from this place.
6233 2017-09-12 Richard Biener <rguenther@suse.de>
6235 PR tree-optimization/82157
6236 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
6237 stmts with side-effects.
6239 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6240 Alan Hayward <alan.hayward@arm.com>
6241 David Sherwood <david.sherwood@arm.com>
6243 * target.def (hard_regno_nregs): New hook.
6244 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
6245 * targhooks.h (default_hard_regno_nregs): Declare.
6246 * targhooks.c (default_hard_regno_nregs): New function.
6247 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
6248 (TARGET_HARD_REGNO_NREGS): ...this hook.
6249 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
6250 (CLASS_MAX_NREGS): Likewise.
6251 * doc/tm.texi: Regenerate.
6252 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
6253 instead of HARD_REGNO_NREGS.
6254 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
6255 HARD_REGNO_NREGS in the comment.
6256 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
6257 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
6258 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
6259 Return an unsigned int.
6260 (TARGET_HARD_REGNO_NREGS): Redefine.
6261 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
6262 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
6263 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
6264 (arc_hard_regno_nregs): New function.
6265 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
6266 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
6267 (arm_hard_regno_nregs): New function.
6268 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
6269 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
6270 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
6271 (TARGET_HARD_REGNO_NREGS): Redefine.
6272 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
6273 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
6274 (HARD_REGNO_NREGS): Delete.
6275 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
6276 (cr16_hard_regno_nregs): New function.
6277 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
6278 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
6279 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
6280 (cris_hard_regno_nregs): New function.
6281 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
6282 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
6283 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
6284 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
6285 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
6286 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
6287 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
6288 (frv_hard_regno_nregs): Make static. Take and return an
6290 (frv_class_max_nregs): Remove outdated copy of documentation.
6291 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
6292 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
6293 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
6294 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
6295 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
6296 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
6297 (TARGET_HARD_REGNO_NREGS): Redefine.
6298 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
6299 (CLASS_MAX_NREGS): Update comment.
6300 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
6301 (ia64_hard_regno_nregs): New function.
6302 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
6303 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
6304 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
6305 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
6306 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
6308 (m32c_hard_regno_nregs): Likewise. Make static.
6309 (TARGET_HARD_REGNO_NREGS): Redefine.
6310 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
6311 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
6312 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
6313 (m68k_hard_regno_nregs): New function.
6314 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
6315 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
6316 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
6317 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
6318 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
6319 Take and return an unsigned int.
6320 (TARGET_HARD_REGNO_NREGS): Redefine.
6321 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
6322 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
6323 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
6324 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
6325 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
6326 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
6327 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
6328 (msp430_hard_regno_nregs): Make static. Take and return an
6330 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
6331 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
6332 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
6333 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
6334 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
6335 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
6336 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
6337 (TARGET_HARD_REGNO_NREGS): Redefine.
6338 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
6339 (PA_HARD_REGNO_NREGS): ...this.
6340 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
6341 (PA_HARD_REGNO_NREGS): ...this.
6342 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
6343 (pa_hard_regno_nregs): New function.
6344 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
6345 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
6346 (pdp11_hard_regno_nregs): New function.
6347 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
6348 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
6349 (rs6000_hard_regno_nregs_hook): New function.
6350 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
6351 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
6352 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
6353 Take and return an unsigned int. Move earlier in file.
6354 (TARGET_HARD_REGNO_NREGS): Redefine.
6355 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
6356 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
6357 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
6358 (rl78_hard_regno_nregs): Make static. Take and return an
6360 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
6361 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
6362 (rs6000_hard_regno_nregs_hook): New function.
6363 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
6364 * config/rx/rx.c (rx_hard_regno_nregs): New function.
6365 (TARGET_HARD_REGNO_NREGS): Redefine.
6366 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
6367 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
6368 instead of HARD_REGNO_NREGS.
6369 (s390_hard_regno_nregs): New function.
6370 (s390_hard_regno_mode_ok): Add comment from s390.h.
6371 (TARGET_HARD_REGNO_NREGS): Redefine.
6372 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
6373 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
6374 (sh_hard_regno_nregs): New function.
6375 (sh_pass_in_reg_p): Use it.
6376 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
6377 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
6378 (sparc_hard_regno_nregs): New function.
6379 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
6380 * config/spu/spu.c (spu_hard_regno_nregs): New function.
6381 (spu_function_arg_advance): Use it, supplying a valid register number.
6382 (TARGET_HARD_REGNO_NREGS): Redefine.
6383 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
6384 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
6385 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
6386 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
6387 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
6388 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
6389 (CLASS_MAX_NREGS): Remove copy of old documentation.
6390 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
6391 (visium_hard_regno_nregs): New function.
6392 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
6393 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
6394 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
6395 xtensa_hard_regno_nregs): New function.
6396 * system.h (HARD_REGNO_NREGS): Poison.
6398 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6400 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
6401 hard_regno_nregs instead of HARD_REGNO_NREGS.
6402 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
6403 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
6404 (c6x_expand_epilogue): Likewise.
6405 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
6406 (frv_read_iacc_argument): Likewise.
6407 * config/sh/sh.c: Include regs.h.
6408 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
6409 (regs_used): Likewise.
6410 (output_stack_adjust): Likewise.
6411 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
6412 * expmed.c: Include regs.h.
6413 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
6414 * ree.c: Include regs.h.
6415 (combine_reaching_defs): Use hard_regno_nregs instead of
6417 (add_removable_extension): Likewise.
6419 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6421 * regs.h (hard_regno_nregs): Turn into a function.
6422 (end_hard_regno): Update accordingly.
6423 * caller-save.c (setup_save_areas): Likewise.
6424 (save_call_clobbered_regs): Likewise.
6425 (replace_reg_with_saved_mem): Likewise.
6426 (insert_restore): Likewise.
6427 (insert_save): Likewise.
6428 * combine.c (can_change_dest_mode): Likewise.
6429 (move_deaths): Likewise.
6430 (distribute_notes): Likewise.
6431 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
6432 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
6433 (rs6000_split_multireg_move): Likewise.
6434 (rs6000_register_move_cost): Likewise.
6435 (rs6000_memory_move_cost): Likewise.
6436 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
6437 (rs6000_split_multireg_move): Likewise.
6438 (rs6000_register_move_cost): Likewise.
6439 (rs6000_memory_move_cost): Likewise.
6440 * cselib.c (cselib_reset_table): Likewise.
6441 (cselib_lookup_1): Likewise.
6442 * emit-rtl.c (set_mode_and_regno): Likewise.
6443 * function.c (aggregate_value_p): Likewise.
6444 * ira-color.c (setup_profitable_hard_regs): Likewise.
6445 (check_hard_reg_p): Likewise.
6446 (calculate_saved_nregs): Likewise.
6447 (assign_hard_reg): Likewise.
6448 (improve_allocation): Likewise.
6449 (calculate_spill_cost): Likewise.
6450 * ira-emit.c (modify_move_list): Likewise.
6451 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
6452 (ira_hard_reg_in_set_p): Likewise.
6453 * ira.c (setup_reg_mode_hard_regset): Likewise.
6454 (clarify_prohibited_class_mode_regs): Likewise.
6455 (check_allocation): Likewise.
6456 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6457 (lra_setup_reg_renumber): Likewise.
6458 (setup_try_hard_regno_pseudos): Likewise.
6459 (spill_for): Likewise.
6460 (assign_hard_regno): Likewise.
6461 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
6462 * lra-constraints.c (in_class_p): Likewise.
6463 (lra_constraint_offset): Likewise.
6464 (simplify_operand_subreg): Likewise.
6465 (lra_constraints): Likewise.
6466 (split_reg): Likewise.
6467 (split_if_necessary): Likewise.
6468 (invariant_p): Likewise.
6469 (inherit_in_ebb): Likewise.
6470 * lra-lives.c (process_bb_lives): Likewise.
6471 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
6472 (get_hard_regs): Likewise.
6473 (do_remat): Likewise.
6474 * lra-spills.c (assign_spill_hard_regs): Likewise.
6475 * mode-switching.c (create_pre_exit): Likewise.
6476 * postreload.c (reload_combine_recognize_pattern): Likewise.
6477 * recog.c (peep2_find_free_register): Likewise.
6478 * regcprop.c (kill_value_regno): Likewise.
6479 (set_value_regno): Likewise.
6480 (copy_value): Likewise.
6481 (maybe_mode_change): Likewise.
6482 (find_oldest_value_reg): Likewise.
6483 (copyprop_hardreg_forward_1): Likewise.
6484 * regrename.c (check_new_reg_p): Likewise.
6485 (regrename_do_replace): Likewise.
6486 * reload.c (push_reload): Likewise.
6487 (combine_reloads): Likewise.
6488 (find_dummy_reload): Likewise.
6489 (operands_match_p): Likewise.
6490 (find_reloads): Likewise.
6491 (find_equiv_reg): Likewise.
6492 (reload_adjust_reg_for_mode): Likewise.
6493 * reload1.c (count_pseudo): Likewise.
6494 (count_spilled_pseudo): Likewise.
6495 (find_reg): Likewise.
6496 (clear_reload_reg_in_use): Likewise.
6497 (free_for_value_p): Likewise.
6498 (allocate_reload_reg): Likewise.
6499 (choose_reload_regs): Likewise.
6500 (reload_adjust_reg_for_temp): Likewise.
6501 (emit_reload_insns): Likewise.
6502 (delete_output_reload): Likewise.
6503 * rtlanal.c (subreg_get_info): Likewise.
6504 * sched-deps.c (sched_analyze_reg): Likewise.
6505 * sel-sched.c (init_regs_for_mode): Likewise.
6506 (mark_unavailable_hard_regs): Likewise.
6507 (choose_best_reg_1): Likewise.
6508 (verify_target_availability): Likewise.
6509 * valtrack.c (dead_debug_insert_temp): Likewise.
6510 * var-tracking.c (track_loc_p): Likewise.
6511 (emit_note_insn_var_location): Likewise.
6512 * varasm.c (make_decl_rtl): Likewise.
6513 * reginfo.c (choose_hard_reg_mode): Likewise.
6514 (init_reg_modes_target): Refer directly to
6515 this_target_regs->x_hard_regno_nregs.
6517 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6519 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
6520 instead of hard_regno_nregs.
6522 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6524 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
6525 end_hard_regno instead of hard_regno_nregs.
6526 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
6527 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
6528 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
6529 * ira-color.c (improve_allocation): Likewise.
6530 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6531 * lra-lives.c (mark_regno_live): Likewise.
6532 (mark_regno_dead): Likewise.
6533 * lra-remat.c (operand_to_remat): Likewise.
6534 * lra.c (collect_non_operand_hard_regs): Likewise.
6535 * postreload.c (reload_combine_note_store): Likewise.
6536 (move2add_valid_value_p): Likewise.
6537 * reload.c (regno_clobbered_p): Likewise.
6539 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6541 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
6543 * config/v850/v850.c (v850_reorg): Likewise.
6544 * reload.c (refers_to_regno_for_reload_p): Likewise.
6545 (find_equiv_reg): Likewise.
6546 * reload1.c (reload_reg_reaches_end_p): Likewise.
6548 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6550 * caller-save.c (add_used_regs): Use REG_NREGS instead of
6552 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
6553 * config/arm/arm.c (output_move_neon): Likewise.
6554 (arm_attr_length_move_neon): Likewise.
6555 (neon_split_vcombine): Likewise.
6556 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
6557 (c6x_mark_reg_written): Likewise.
6558 (c6x_dwarf_register_span): Likewise.
6559 * config/i386/i386.c (ix86_save_reg): Likewise.
6560 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
6561 (rws_access_reg): Likewise.
6562 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
6563 * mode-switching.c (create_pre_exit): Likewise.
6564 * ree.c (combine_reaching_defs): Likewise.
6565 (add_removable_extension): Likewise.
6566 * regcprop.c (find_oldest_value_reg): Likewise.
6567 (copyprop_hardreg_forward_1): Likewise.
6568 * reload.c (reload_inner_reg_of_subreg): Likewise.
6569 (push_reload): Likewise.
6570 (combine_reloads): Likewise.
6571 (find_dummy_reload): Likewise.
6572 (reload_adjust_reg_for_mode): Likewise.
6573 * reload1.c (find_reload_regs): Likewise.
6574 (forget_old_reloads_1): Likewise.
6575 (reload_reg_free_for_value_p): Likewise.
6576 (reload_adjust_reg_for_temp): Likewise.
6577 (emit_reload_insns): Likewise.
6578 (delete_output_reload): Likewise.
6579 * sel-sched.c (choose_best_reg_1): Likewise.
6580 (choose_best_pseudo_reg): Likewise.
6582 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6583 Alan Hayward <alan.hayward@arm.com>
6584 David Sherwood <david.sherwood@arm.com>
6586 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
6587 * target.def (slow_unaligned_access): New hook.
6588 * targhooks.h (default_slow_unaligned_access): Declare.
6589 * targhooks.c (default_slow_unaligned_access): New function.
6590 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
6591 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
6592 * doc/tm.texi: Regenerate.
6593 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
6594 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
6595 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
6597 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
6598 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
6600 (rs6000_slow_unaligned_access): New function.
6601 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
6602 (expand_block_compare): Likewise.
6603 (expand_strn_compare): Likewise.
6604 (rs6000_rtx_costs): Likewise.
6605 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
6606 (riscv_slow_unaligned_access): Likewise.
6607 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
6608 (riscv_slow_unaligned_access_p): ...this and make static.
6609 (riscv_option_override): Update accordingly.
6610 (riscv_slow_unaligned_access): New function.
6611 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
6612 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
6613 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
6614 (rs6000_slow_unaligned_access): New function.
6615 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
6616 (rs6000_rtx_costs): Likewise.
6617 * config/rs6000/rs6000-string.c (expand_block_compare)
6618 (expand_strn_compare): Use targetm.slow_unaligned_access instead
6619 of SLOW_UNALIGNED_ACCESS.
6620 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
6621 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
6622 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
6623 of SLOW_UNALIGNED_ACCESS.
6624 * expmed.c (simple_mem_bitfield_p): Likewise.
6625 * expr.c (alignment_for_piecewise_move): Likewise.
6626 (emit_group_load_1): Likewise.
6627 (emit_group_store): Likewise.
6628 (copy_blkmode_from_reg): Likewise.
6629 (emit_push_insn): Likewise.
6630 (expand_assignment): Likewise.
6631 (store_field): Likewise.
6632 (expand_expr_real_1): Likewise.
6633 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
6634 * lra-constraints.c (simplify_operand_subreg): Likewise.
6635 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
6636 * gimple-ssa-store-merging.c: Likewise in block comment at start
6638 * tree-ssa-strlen.c: Include target.h.
6639 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
6640 of SLOW_UNALIGNED_ACCESS.
6641 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
6643 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6645 PR rtl-optimization/82185
6646 * expmed.c (emit_store_flag_int): Only test tem if it has been
6649 2017-09-12 Richard Biener <rguenther@suse.de>
6652 * match.pd ((FTYPE) N CMP CST): Fix typo.
6654 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
6656 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
6658 (mips_near_type_p): Add 'short_call' attribute as a synonym
6660 * doc/extend.texi (short_call): Document new function attribute.
6662 2017-09-12 Jakub Jelinek <jakub@redhat.com>
6665 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
6666 assertion check that in the condition.
6667 (get_atomic_generic_size): Likewise. Before testing if parameter
6668 has pointer type, if it has array type, call for C++
6669 default_conversion to perform array-to-pointer conversion.
6671 2017-09-12 Richard Biener <rguenther@suse.de>
6673 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
6674 for operations we cannot scalarize.
6676 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
6678 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
6679 vectors heap vectors. Clean up comments.
6680 Make visited_bbs a reference.
6681 (profitable_jump_thread_path): Make GC
6682 vectors heap vectors. Clean up comments.
6684 (convert_and_register_jump_thread_path): Make GC vectors heap
6686 (check_subpath_and_update_thread_path): Same. Clean up comments.
6687 Make visited_bbs a reference.
6688 (handle_phi): Abstract common code to to
6689 register_jump_thread_path_if_profitable.
6690 Rename VAR_BB to DEF_BB.
6692 Make GC vectors heap vectors.
6693 Make visited_bbs a reference.
6694 (handle_assignment): Same.
6695 (register_jump_thread_path_if_profitable): New.
6696 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
6698 Make GC vectors heap vectors. Clean up comments.
6699 Make visited_bbs a reference.
6700 (find_jump_threads_backwards): Make visited_bbs live in the stack.
6701 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
6704 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
6707 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
6708 words of E_DImode object are reachable by xtensa_uimm8x4 access.
6710 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
6712 Revert r251800 and r251799.
6714 2017-09-11 Martin Jambor <mjambor@suse.cz>
6717 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
6718 arguments in advance.
6719 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
6720 use it to find predecessor edges.
6721 (naive_outof_ssa): Collect vector of predecessors.
6723 2017-09-08 Jason Merrill <jason@redhat.com>
6725 PR c++/70029 - ICE with ref-qualifier and -flto
6726 * langhooks.h (struct lang_hooks_for_types): Add
6727 copy_lang_qualifiers.
6728 * attribs.c (build_type_attribute_qual_variant): Use it.
6729 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
6731 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
6732 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
6734 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
6737 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
6738 (*mulsi3_sp64): New instruction.
6739 (mulsi3): New expander.
6741 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
6743 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
6745 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6747 * sancov.c: Include memmodel.h.
6749 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
6752 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
6755 2017-09-07 Carl Love <cel@us.ibm.com>
6757 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
6758 the sldi instruction.
6760 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
6762 * sancov.c: Include tm_p.h.
6764 2017-09-07 Jakub Jelinek <jakub@redhat.com>
6767 * output.h (switch_to_other_text_partition): New declaration.
6768 * varasm.c (switch_to_other_text_partition): New function.
6769 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
6770 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
6771 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
6772 to the other text partition before emitting LCL label and switch back
6773 after emitting the word after it.
6775 2017-09-07 Richard Biener <rguenther@suse.de>
6777 * passes.def (pass_split_crit_edges): Remove instance before PRE.
6778 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
6779 critical edges here, after loop init.
6780 (pass_data_pre): Remove PROP_no_crit_edges flags.
6781 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
6782 for valueization of call args to avoid leaking VN_TOP.
6783 (visit_use): Assert we do not visit default defs.
6784 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
6785 Use error_mark_node to more easily detect leaking VN_TOP.
6786 All default-defs are varying, not VN_TOP. Mark them visited.
6787 (run_scc_vn): Make code match comment.
6789 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
6791 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
6792 OPTION_MASK_FLOAT128_KEYWORD.
6793 (POWERPC_MASKS): Likewise.
6794 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
6795 support for the -mfloat128-type option, and make -mfloat128
6796 default on PowerPC Linux systems. Define or undefine
6797 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
6798 Define __float128 to be __ieee128 if IEEE 128-bit support is
6799 enabled, or undefine it.
6800 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
6801 Delete defining __FLOAT128_TYPE__.
6802 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
6803 -mfloat128-type option and make -mfloat128 default on PowerPC
6805 (TARGET_FLOAT128_TYPE): Likewise.
6806 (-mfloat128-type): Likewise.
6807 * config/rs6000/rs6000.c (rs6000_option_override_internal):
6808 Delete the -mfloat128-type option and make -mfloat128 default on
6809 PowerPC Linux systems. Always use __ieee128 to be the keyword for
6810 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
6811 128-bit floating point is enabled. Change tests from using
6812 -mfloat128-type to -mfloat128.
6813 (rs6000_mangle_type): Use the correct mangling for the __float128
6814 type even if normal long double is restricted to 64-bits.
6815 (floatn_mode): Enable the _Float128 type by default on VSX Linux
6817 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
6818 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
6819 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
6821 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
6822 documentation for -mfloat128.
6824 2017-09-06 Olivier Hainque <hainque@adacore.com>
6826 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
6828 2017-09-06 Wish Wu <wishwu007@gmail.com>
6829 Jakub Jelinek <jakub@redhat.com>
6831 * asan.c (initialize_sanitizer_builtins): Add
6832 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
6833 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
6834 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
6835 BT_FN_VOID_UINT64_PTR variables.
6836 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
6837 (BT_FN_VOID_UINT16_UINT16): Likewise.
6838 (BT_FN_VOID_UINT32_UINT32): Likewise.
6839 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
6840 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
6841 (BT_FN_VOID_UINT64_PTR): Likewise.
6842 * common.opt (flag_sanitize_coverage): New variable.
6843 (fsanitize-coverage=trace-pc): Remove.
6844 (fsanitize-coverage=): Add.
6845 * flag-types.h (enum sanitize_coverage_code): New enum.
6846 * fold-const.c (fold_range_test): Disable non-short-circuit
6847 optimization if flag_sanitize_coverage.
6848 (fold_truth_andor): Likewise.
6849 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
6850 * opts.c (COVERAGE_SANITIZER_OPT): Define.
6851 (coverage_sanitizer_opts): New array.
6852 (get_closest_sanitizer_option): Add OPTS argument, handle also
6853 OPT_fsanitize_coverage_.
6854 (parse_sanitizer_options): Adjusted to also handle
6855 OPT_fsanitize_coverage_.
6856 (common_handle_option): Add OPT_fsanitize_coverage_.
6857 * sancov.c (instrument_comparison, instrument_switch): New function.
6858 (sancov_pass): Add trace-cmp support.
6859 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
6860 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
6861 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
6862 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
6863 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
6864 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
6865 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
6866 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
6867 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
6868 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
6870 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
6872 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
6873 error. Only quit immediately if parsing is complete.
6874 (BEGIN): Initialize fatal_err and parse_done.
6875 (begin fpu, end fpu): Check number of arguments.
6876 (begin arch, end arch): Likewise.
6877 (begin cpu, end cpu): Likewise.
6878 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
6879 (optalias): Likewise.
6881 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
6883 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
6884 * config/arm/arm-isa.h: Delete. Move definitions to ...
6885 * arm-cpus.in: ... here. Use new feature and fgroup values.
6886 * config/arm/arm.c (arm_option_override): Use lower case for feature
6888 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
6889 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
6890 * config/arm/parsecpu.awk (END): Add new command 'isa'.
6892 (print_isa_bits_for): New function.
6893 (gen_isa): New function.
6894 (gen_comm_data): Use print_isa_bits_for.
6895 (define feature): New keyword.
6896 (define fgroup): New keyword.
6897 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
6898 (arm-isa.h): Add rule to generate file.
6899 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
6900 case for feature bit names.
6902 2017-09-06 Richard Biener <rguenther@suse.de>
6904 * tree-ssa-pre.c (NECESSARY): Remove.
6905 (create_expression_by_pieces): Do not touch pass-local flags.
6906 (insert_into_preds_of_block): Likewise.
6907 (do_pre_regular_insertion): Likewise.
6908 (eliminate_insert): Likewise.
6909 (eliminate_dom_walker::before_dom_children): Likewise.
6910 (fini_eliminate): Do not look at inserted_exprs.
6911 (mark_operand_necessary): Remove.
6912 (remove_dead_inserted_code): Replace with simple work-list
6913 algorithm based on inserted_exprs and SSA uses.
6914 (pass_pre::execute): Re-order fini_eliminate and
6915 remove_dead_inserted_code.
6917 2017-09-06 Olivier Hainque <hainque@adacore.com>
6919 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
6920 for VxWorks 7. Adjust surrounding comments.
6922 2017-09-06 Richard Biener <rguenther@suse.de>
6924 * gimple-ssa-strength-reduction.c
6925 (find_candidates_dom_walker::before_dom_children): Also allow
6928 2017-09-06 Richard Biener <rguenther@suse.de>
6930 PR tree-optimization/82108
6931 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
6932 for gap in the non-permutation SLP case.
6934 2017-09-06 Martin Jambor <mjambor@suse.cz>
6936 PR tree-optimization/82078
6937 * tree-sra.c (sort_and_splice_var_accesses): Move call to
6938 add_access_to_work_queue...
6939 (build_accesses_from_assign): ...here.
6940 (propagate_all_subaccesses): Make sure racc is the group
6941 representative, if there is one.
6943 2017-09-06 Jakub Jelinek <jakub@redhat.com>
6946 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
6949 2017-09-06 Richard Biener <rguenther@suse.de>
6951 * gimple-ssa-strength-reduction.c
6952 (find_candidates_dom_walker::before_doom_children): Use a
6953 type and not a mode check.
6955 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
6958 * config/arm/predicates.md (arm_general_adddi_operand): Create new
6960 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
6962 2017-09-05 Jeff Law <law@redhat.com>
6964 PR tree-optimization/64910
6965 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
6966 cases where we have 3 or more operands.
6968 2017-09-05 Jakub Jelinek <jakub@redhat.com>
6971 * omp-low.c (lower_omp_for): Recompute tree invariant if
6972 gimple_omp_for_initial/final is ADDR_EXPR.
6975 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
6976 into gimple val before gimplification fo the COND_EXPR.
6978 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
6980 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
6981 REGION_COPY argument.
6982 (thread_through_all_blocks): Remove unused argument to
6983 duplicate_thread_path.
6985 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6986 Alan Hayward <alan.hayward@arm.com>
6987 David Sherwood <david.sherwood@arm.com>
6989 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
6990 Take a scalar_mode rather than a machine_mode.
6991 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6992 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
6993 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6994 (aarch64_gen_adjusted_ldpstp): Likewise.
6995 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
6997 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6998 Alan Hayward <alan.hayward@arm.com>
6999 David Sherwood <david.sherwood@arm.com>
7001 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
7002 Take a scalar_int_mode instead of a machine_mode.
7003 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
7004 (aarch64_output_scalar_simd_mov_immediate): Likewise.
7005 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
7006 (aarch64_simd_attr_length_rglist): Delete.
7007 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
7008 a scalar_int_mode instead of a machine_mode.
7009 (aarch64_add_offset): Likewise.
7010 (aarch64_internal_mov_immediate): Likewise
7011 (aarch64_add_constant_internal): Likewise.
7012 (aarch64_add_constant): Likewise.
7013 (aarch64_movw_imm): Likewise.
7014 (aarch64_rtx_arith_op_extract_p): Likewise.
7015 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
7016 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
7017 Remove assert that the mode isn't a vector.
7018 (aarch64_output_scalar_simd_mov_immediate): Likewise.
7019 (aarch64_expand_mov_immediate): Update calls after above changes.
7020 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
7021 (aarch64_and_bitmask_imm): Check for scalar integer modes.
7022 (aarch64_move_imm): Likewise.
7023 (aarch64_can_const_movi_rtx_p): Likewise.
7024 (aarch64_strip_extend): Likewise.
7025 (aarch64_extr_rtx_p): Likewise.
7026 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
7027 a CONST_INT when the mode parameter is VOIDmode.
7028 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
7030 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7032 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
7033 * stor-layout.c (bitwise_mode_for_mode): Likewise.
7034 (bitwise_type_for_mode): Update accordingly.
7036 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7038 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
7039 * stor-layout.c (mode_for_size_tree): Likewise.
7040 (mode_for_array): Update accordingly.
7041 (layout_decl): Likewise.
7042 (compute_record_mode): Likewise. Only set the mode once.
7044 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7046 * target.def (get_mask_mode): Change return type to opt_mode.
7048 * doc/tm.texi: Regenerate.
7049 * targhooks.h (default_get_mask_mode): Return an opt_mode.
7050 * targhooks.c (default_get_mask_mode): Likewise.
7051 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
7052 * optabs-query.c (can_vec_mask_load_store_p): Update use of
7053 targetm.get_mask_mode.
7054 * tree.c (build_truth_vector_type): Likewise.
7056 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7058 * machmode.h (mode_for_vector): Return an opt_mode.
7059 * stor-layout.c (mode_for_vector): Likewise.
7060 (mode_for_int_vector): Update accordingly.
7061 (layout_type): Likewise.
7062 * config/i386/i386.c (emit_memmov): Likewise.
7063 (ix86_expand_set_or_movmem): Likewise.
7064 (ix86_expand_vector_init): Likewise.
7065 (ix86_get_mask_mode): Likewise.
7066 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
7068 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
7069 * expmed.c (extract_bit_field_1): Likewise.
7070 * expr.c (expand_expr_real_2): Likewise.
7071 * optabs-query.c (can_vec_perm_p): Likewise.
7072 (can_vec_mask_load_store_p): Likewise.
7073 * optabs.c (expand_vec_perm): Likewise.
7074 * targhooks.c (default_get_mask_mode): Likewise.
7075 * tree-vect-stmts.c (vectorizable_store): Likewise.
7076 (vectorizable_load): Likewise.
7077 (get_vectype_for_scalar_type_and_size): Likewise.
7079 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7081 * machmode.h (mode_for_int_vector): New function.
7082 * stor-layout.c (mode_for_int_vector): Likewise.
7083 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
7084 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
7085 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
7086 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
7087 (s390_expand_vcond): Likewise.
7089 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7091 * machmode.h (opt_machine_mode): New type.
7092 (opt_mode<T>): Allow construction from anything that can be
7094 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
7095 (mode_for_size): Return an opt_machine_mode.
7096 * stor-layout.c (mode_for_size): Likewise.
7097 (mode_for_size_tree): Update call accordingly.
7098 (bitwise_mode_for_mode): Likewise.
7099 (make_fract_type): Likewise.
7100 (make_accum_type): Likewise.
7101 * caller-save.c (replace_reg_with_saved_mem): Update call
7103 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
7104 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
7105 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
7106 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
7107 * expmed.c (extract_bit_field_1): Likewise.
7108 * reload.c (get_secondary_mem): Likewise.
7109 * varasm.c (assemble_integer): Likewise.
7110 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
7113 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7115 * machmode.h (decimal_float_mode_for_size): New function.
7116 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
7117 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
7118 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
7119 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
7120 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
7121 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
7123 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7125 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
7126 (get_builtin_sync_mode): Likewise.
7127 (expand_ifn_atomic_compare_exchange): Likewise.
7128 (expand_builtin_atomic_clear): Likewise.
7129 (expand_builtin_atomic_test_and_set): Likewise.
7130 (fold_builtin_atomic_always_lock_free): Likewise.
7131 * calls.c (compute_argument_addresses): Likewise.
7132 (emit_library_call_value_1): Likewise.
7133 (store_one_arg): Likewise.
7134 * combine.c (combine_instructions): Likewise.
7135 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
7136 * config/arm/arm.c (arm_function_value): Likewise.
7137 (aapcs_allocate_return_reg): Likewise.
7138 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
7139 * config/i386/i386.c (construct_container): Likewise.
7140 (ix86_gimplify_va_arg): Likewise.
7141 (ix86_expand_sse_cmp): Likewise.
7142 (emit_memmov): Likewise.
7143 (emit_memset): Likewise.
7144 (expand_small_movmem_or_setmem): Likewise.
7145 (ix86_expand_pextr): Likewise.
7146 (ix86_expand_pinsr): Likewise.
7147 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
7148 * config/microblaze/microblaze.c (microblaze_block_move_straight):
7150 * config/mips/mips.c (mips_function_value_1) Likewise.
7151 (mips_block_move_straight): Likewise.
7152 (mips_expand_ins_as_unaligned_store): Likewise.
7153 * config/powerpcspe/powerpcspe.c
7154 (rs6000_darwin64_record_arg_advance_flush): Likewise.
7155 (rs6000_darwin64_record_arg_flush): Likewise.
7156 * config/rs6000/rs6000.c
7157 (rs6000_darwin64_record_arg_advance_flush): Likewise.
7158 (rs6000_darwin64_record_arg_flush): Likewise.
7159 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
7160 (sparc_function_value_1): Likewise.
7161 * config/spu/spu.c (adjust_operand): Likewise.
7162 (spu_emit_branch_or_set): Likewise.
7163 (arith_immediate_p): Likewise.
7164 * emit-rtl.c (gen_lowpart_common): Likewise.
7165 * expr.c (expand_expr_real_1): Likewise.
7166 * function.c (assign_parm_setup_block): Likewise.
7167 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
7168 * reload1.c (alter_reg): Likewise.
7169 * stor-layout.c (mode_for_vector): Likewise.
7170 (layout_type): Likewise.
7172 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7174 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
7175 (spu_convert_move): Likewise.
7176 * lower-subreg.c (resolve_simple_move): Likewise.
7178 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7181 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
7182 define_insn to a define_expand.
7183 (altivec_vsum2sws_direct): New define_insn.
7184 (altivec_vsumsws): Convert from a define_insn to a define_expand.
7186 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
7188 * config/arm/arm.c (arm_option_params_internal): Improve setting of
7191 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
7196 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
7197 realign stack if stack alignment needed is less than incoming
7200 2017-09-05 Marek Polacek <polacek@redhat.com>
7203 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
7206 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
7208 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
7210 2017-09-05 Richard Biener <rguenther@suse.de>
7212 PR tree-optimization/82084
7213 * fold-const.c (can_native_encode_string_p): Handle wide characters.
7215 2017-09-05 Richard Biener <rguenther@suse.de>
7217 PR tree-optimization/82102
7218 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
7220 2017-09-05 Martin Liska <mliska@suse.cz>
7222 PR tree-optimization/82032
7223 * tree-cfg.c (generate_range_test): New function.
7224 * tree-cfg.h (generate_range_test): Declared here.
7225 * tree-cfgcleanup.c (convert_single_case_switch): New function.
7226 (cleanup_control_expr_graph): Use it.
7227 * tree-switch-conversion.c (try_switch_expansion): Remove
7229 (emit_case_nodes): Use generate_range_test.
7231 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
7234 * config/i386/i386.md (*<btsc><mode>_mask): Add
7235 TARGET_USE_BT to insn constraint.
7236 (*btr<mode>_mask): Ditto.
7238 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
7240 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
7241 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
7243 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
7246 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
7247 TARGET_NEON and TARGET_IWMMXT.
7248 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
7249 TARGET_NEON and TARGET_IWMMXT.
7250 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
7252 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
7254 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
7255 (ix86_rewrite_tls_address): Ditto.
7256 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
7257 (ix86_rewrite_tls_address_1): Ditto.
7258 (ix86_rewrite_tls_address): Ditto.
7259 * config/i386/predicates.md (tls_address_pattern): New predicate.
7260 * config/i386/i386.md (TLS address splitter): New splitter.
7262 2017-09-04 Richard Biener <rguenther@suse.de>
7264 PR tree-optimization/82084
7265 * fold-const.h (can_native_encode_string_p): Declare.
7266 * fold-const.c (can_native_encode_string_p): Factor out from ...
7267 (native_encode_string): ... here.
7268 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
7269 vectorizing stores from constants we later cannot handle.
7271 2017-09-04 Marek Polacek <polacek@redhat.com>
7274 * doc/invoke.texi: Update -Wtautological-compare documentation.
7276 2017-09-04 Jeff Law <law@redhat.com>
7278 PR tree-optimization/64910
7279 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
7280 swap the first and last operand if the last is a constant.
7282 2017-09-04 Marek Polacek <polacek@redhat.com>
7285 * convert.c (do_narrow): When sanitizing signed integer overflows,
7286 bail out for signed types.
7287 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
7289 2017-09-04 Richard Biener <rguenther@suse.de>
7291 PR tree-optimization/82060
7292 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
7293 Move devirtualization after stmt folding and before EH/AB/noreturn
7294 cleanup to get the stmt refs canonicalized. Use a bool instead
7295 of gimple_modified_p since that doesn't work for NOPs. Schedule
7296 NOPs generated by folding for removal.
7298 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7299 Alan Hayward <alan.hayward@arm.com>
7300 David Sherwood <david.sherwood@arm.com>
7302 * coretypes.h (pad_direction): New enum.
7303 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
7304 (FUNCTION_ARG_PADDING): Likewise.
7305 * target.def (function_arg_padding): New hook.
7306 * targhooks.h (default_function_arg_padding): Declare.
7307 * targhooks.c (default_function_arg_padding): New function.
7308 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
7309 (TARGET_FUNCTION_ARG_PADDING): ...this.
7310 * doc/tm.texi: Regenerate.
7311 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
7312 instead of direction.
7313 (compute_argument_addresses): Likewise.
7314 (load_register_parameters): Likewise.
7315 (emit_library_call_value_1): Likewise.
7316 (store_one_arg): Use targetm.calls.function_arg_padding instead
7317 of FUNCTION_ARG_PADDING.
7318 (must_pass_in_stack_var_size_or_pad): Likewise.
7319 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
7320 (emit_group_store): Likewise.
7321 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
7322 instead of FUNCTION_ARG_PADDING.
7323 (emit_push_insn): Likewise, and propagate enum change throughout
7325 * function.h (direction): Delete.
7326 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
7328 * function.c (assign_parm_find_stack_rtl): Likewise.
7329 (assign_parm_setup_block_p): Likewise.
7330 (assign_parm_setup_block): Likewise.
7331 (gimplify_parameters): Likewise.
7332 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
7333 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
7335 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
7336 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
7337 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
7338 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
7339 (aarch64_function_arg_padding): ...this new function.
7340 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
7341 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7342 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
7343 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
7344 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
7345 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7346 (arm_pad_arg_upward): Replace with...
7347 (arm_function_arg_padding): ...this new function.
7348 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
7350 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
7351 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
7352 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7353 (ia64_hpux_function_arg_padding): Replace with...
7354 (ia64_function_arg_padding): ...this new function. Use pad_direction
7355 instead of direction. Check for TARGET_HPUX.
7356 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
7357 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7358 (iq2000_function_arg_padding): New function.
7359 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
7360 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
7361 (mips_function_arg_padding): ...this new function.
7362 (mips_pad_reg_upward): Update accordingly.
7363 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7364 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
7365 targetm.calls.function_arg_padding.
7366 (FUNCTION_ARG_PADDING): Delete.
7367 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
7368 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
7369 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
7370 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
7371 (nios2_block_reg_padding): Return pad_direction instead of direction.
7372 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
7373 instead of direction.
7374 (nios2_function_arg_padding): Likewise. Make static.
7375 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7376 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
7377 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
7378 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
7379 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7380 (pa_function_arg_padding): Make static. Return pad_direction instead
7382 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
7383 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
7384 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
7385 instead of direction. Use targetm.calls.function_arg_padding.
7386 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
7387 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
7388 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
7389 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
7390 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
7392 (function_arg_padding): Rename to...
7393 (rs6000_function_arg_padding): ...this. Make static. Return
7394 pad_direction instead of direction.
7395 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
7396 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
7397 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
7398 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
7399 instead of direction. Use targetm.calls.function_arg_padding.
7400 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
7401 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
7402 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
7403 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
7404 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7405 (function_arg_padding): Rename to...
7406 (rs6000_function_arg_padding): ...this. Make static. Return
7407 pad_direction instead of direction.
7408 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
7409 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
7410 * config/s390/s390.c (s390_function_arg_padding): New function.
7411 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7412 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
7413 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
7414 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7415 (function_arg_padding): Rename to...
7416 (sparc_function_arg_padding): ...this. Make static. Return
7417 pad_direction instead of direction.
7418 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
7419 * config/spu/spu.c (spu_function_arg_padding): New function.
7420 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7421 * system.h (FUNCTION_ARG_PADDING): Poison.
7423 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7424 Alan Hayward <alan.hayward@arm.com>
7425 David Sherwood <david.sherwood@arm.com>
7427 * target.def (modes_tieable_p): New hook.
7428 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
7429 (TARGET_MODES_TIEABLE_P): ...this.
7430 * doc/tm.texi.in: Regenerate.
7431 * hooks.h (hook_bool_mode_mode_true): Declare.
7432 * hooks.c (hook_bool_mode_mode_true): New function.
7433 * combine.c (subst): Use targetm.modes_tieable_p instead of
7435 * dse.c (find_shift_sequence): Likewise.
7436 * expmed.c (extract_low_bits): Likewise.
7437 * lower-subreg.c: Include target.h.
7438 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
7440 * rtlanal.c (rtx_cost): Likewise.
7441 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
7442 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
7443 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
7444 (TARGET_MODES_TIEABLE_P): Redefine.
7445 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
7446 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
7447 (TARGET_MODES_TIEABLE_P): Redefine.
7448 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
7449 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
7450 (arc_modes_tieable_p): New function.
7451 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
7452 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
7453 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
7454 (arm_modes_tieable_p): Make static.
7455 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
7456 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
7457 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
7458 (TARGET_MODES_TIEABLE_P): Redefine.
7459 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
7460 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
7461 (TARGET_MODES_TIEABLE_P): Redefine.
7462 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
7463 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
7464 (cr16_modes_tieable_p): New function.
7465 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
7466 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
7467 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
7468 (TRULY_NOOP_TRUNCATION): Update comment.
7469 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
7470 (TRULY_NOOP_TRUNCATION): Update comment.
7471 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
7472 (frv_modes_tieable_p): New function.
7473 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
7474 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
7475 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
7476 (TARGET_MODES_TIEABLE_P): Redefine.
7477 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
7478 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
7479 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
7480 (TARGET_MODES_TIEABLE_P): Redefine.
7481 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
7482 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
7483 (ia64_modes_tieable_p): New function.
7484 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
7485 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
7486 (iq2000_modes_tieable_p): New function.
7487 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
7488 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
7489 (lm32_modes_tieable_p): New function.
7490 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
7491 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
7492 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
7493 (TARGET_MODES_TIEABLE_P): Redefine.
7494 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
7495 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
7496 (m32r_modes_tieable_p): New function.
7497 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
7498 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
7499 (m68k_modes_tieable_p): New function.
7500 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
7501 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
7502 (mcore_modes_tieable_p): New function.
7503 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
7504 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
7506 (TARGET_MODES_TIEABLE_P): Redefine.
7507 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
7508 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
7509 * config/mips/mips.c (mips_modes_tieable_p): Make static.
7510 (TARGET_MODES_TIEABLE_P): Redefine.
7511 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
7512 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
7513 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
7514 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
7515 (mn10300_modes_tieable_p): ...this and make static.
7516 (TARGET_MODES_TIEABLE_P): Redefine.
7517 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
7518 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
7519 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
7520 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
7521 (msp430_modes_tieable_p): Make static.
7522 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
7523 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
7524 (TARGET_MODES_TIEABLE_P): Redefine.
7525 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
7526 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
7527 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
7528 (TARGET_MODES_TIEABLE_P): Redefine.
7529 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
7530 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
7531 * config/pa/pa.c (pa_modes_tieable_p): Make static.
7532 (TARGET_MODES_TIEABLE_P): Redefine.
7533 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
7534 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
7535 (pdp11_modes_tieable_p): New function.
7536 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
7537 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
7538 (rs6000_modes_tieable_p): New function.
7539 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
7540 * config/powerpcspe/powerpcspe.md: Update comment.
7541 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
7542 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
7543 (TARGET_MODES_TIEABLE_P): Redefine.
7544 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
7545 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
7546 (rl78_modes_tieable_p): New function.
7547 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
7548 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
7549 (rs6000_modes_tieable_p): New function.
7550 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
7551 * config/rs6000/rs6000.md: Update comment.
7552 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
7553 * config/rx/rx.c (rx_modes_tieable_p): New function.
7554 (TARGET_MODES_TIEABLE_P): Redefine.
7555 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
7556 * config/s390/s390.c (s390_modes_tieable_p): New function.
7557 (TARGET_MODES_TIEABLE_P): Redefine.
7558 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
7559 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
7560 (sh_modes_tieable_p): New function.
7561 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
7562 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
7563 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
7564 (sparc_modes_tieable_p): Make static.
7565 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
7566 * config/spu/spu.c (spu_modes_tieable_p): New function.
7567 (TARGET_MODES_TIEABLE_P): Redefine.
7568 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
7569 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
7570 (TARGET_MODES_TIEABLE_P): Redefine.
7571 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
7572 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
7573 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
7574 * config/v850/v850.c (v850_modes_tieable_p): New function.
7575 (TARGET_MODES_TIEABLE_P): Redefine.
7576 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
7577 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
7578 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
7579 (visium_modes_tieable_p): New function.
7580 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
7581 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
7582 (xtensa_modes_tieable_p): New function.
7583 * system.h (MODES_TIEABLE_P): Poison.
7585 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7586 Alan Hayward <alan.hayward@arm.com>
7587 David Sherwood <david.sherwood@arm.com>
7589 * target.def (hard_regno_mode_ok): New hook.
7590 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
7591 (TARGET_HARD_REGNO_MODE_OK): ...this.
7592 * doc/tm.texi.in: Regenerate.
7593 * hooks.h (hook_bool_uint_mode_true): Declare.
7594 * hooks.c (hook_bool_uint_mode_true): New function.
7595 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
7597 * genpreds.c (write_insn_preds_c): Add an include of target.h.
7598 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
7599 instead of HARD_REGNO_MODE_OK.
7600 * caller-save.c: Include target.h.
7601 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
7603 * combine.c (can_combine_p): Likewise.
7604 (combinable_i3pat): Likewise.
7605 (can_change_dest_mode): Likewise.
7606 * expr.c (init_expr_target): Likewise.
7607 (convert_move): Likewise.
7608 (convert_modes): Likewise.
7609 * ira.c (setup_prohibited_class_mode_regs): Likewise.
7610 (setup_prohibited_mode_move_regs): Likewise.
7611 * ira.h (target_ira): Likewise.
7612 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7613 * lra-constraints.c (process_alt_operands): Likewise.
7614 (split_reg): Likewise.
7615 * recog.c (peep2_find_free_register): Likewise.
7616 * ree.c (combine_reaching_defs): Likewise.
7617 * regcprop.c (maybe_mode_change): Likewise.
7618 * reginfo.c (init_reg_sets_1): Likewise.
7619 (choose_hard_reg_mode): Likewise.
7620 (simplifiable_subregs): Likewise.
7621 * regrename.c (check_new_reg_p): Likewise.
7622 * reload.c (find_valid_class): Likewise.
7623 (find_valid_class_1): Likewise.
7624 (reload_inner_reg_of_subreg): Likewise.
7625 (push_reload): Likewise.
7626 (combine_reloads): Likewise.
7627 (find_dummy_reload): Likewise.
7628 (find_reloads): Likewise.
7629 * reload1.c (find_reg): Likewise.
7630 (set_reload_reg): Likewise.
7631 (allocate_reload_reg): Likewise.
7632 (choose_reload_regs): Likewise.
7633 (reload_adjust_reg_for_temp): Likewise.
7634 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
7635 (simplify_subreg_regno): Likewise.
7636 * sel-sched.c (init_regs_for_mode): Likewise.
7637 * varasm.c (make_decl_rtl): Likewise.
7638 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
7639 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
7641 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
7642 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
7643 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7644 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
7645 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
7646 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7647 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
7648 (arc_mode_class): Delete.
7649 (HARD_REGNO_MODE_OK): Delete.
7650 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7651 (arc_hard_regno_mode_ok): Rename old array to...
7652 (arc_hard_regno_mode_ok_modes): ...this.
7653 (arc_conditional_register_usage): Update accordingly.
7654 (arc_mode_class): Make static.
7655 (arc_hard_regno_mode_ok): New function.
7656 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
7657 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
7658 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7659 (arm_hard_regno_mode_ok): Make static.
7660 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
7662 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
7663 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
7664 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
7666 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7667 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
7668 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
7669 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
7670 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
7671 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7672 * config/bfin/predicates.md (valid_reg_operand): Use
7673 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
7674 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
7675 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
7676 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7677 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
7678 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
7679 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7680 (cr16_hard_regno_mode_ok): Make static and return a bool.
7681 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
7682 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7683 (cris_hard_regno_mode_ok): New function.
7684 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
7685 (epiphany_mode_class): Delete.
7686 (HARD_REGNO_MODE_OK): Delete.
7687 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
7688 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7689 (hard_regno_mode_ok): Rename to...
7690 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
7691 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
7692 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
7694 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
7695 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
7696 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7697 (frv_hard_regno_mode_ok): Make static and return a bool.
7698 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
7700 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
7701 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
7702 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
7703 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
7705 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7706 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
7707 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
7708 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
7710 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7711 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
7712 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7713 (ia64_hard_regno_mode_ok): New function.
7714 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
7715 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7716 (iq2000_hard_regno_mode_ok): New function.
7717 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
7718 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7719 (lm32_hard_regno_mode_ok): New function.
7720 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
7721 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
7722 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
7723 instead of HARD_REGNO_MODE_OK.
7724 (m32c_hard_regno_ok): Rename to...
7725 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
7726 (m32c_cannot_change_mode_class): Update accordingly.
7727 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7728 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
7729 (m32r_mode_class): Delete.
7730 (HARD_REGNO_MODE_OK): Delete.
7731 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7732 (m32r_hard_regno_mode_ok): Rename to...
7733 (m32r_hard_regno_modes): ...this.
7734 (m32r_mode_class): Make static.
7735 (m32r_hard_regno_mode_ok): New function.
7736 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
7737 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
7738 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7739 (m68k_hard_regno_mode_ok): Make static.
7740 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
7741 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7742 (mcore_hard_regno_mode_ok): New function.
7743 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
7744 (HARD_REGNO_MODE_OK): Delete.
7745 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
7747 (microblaze_hard_regno_mode_ok_p): ...this and make static.
7748 (microblaze_hard_regno_mode_ok): New function.
7749 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7750 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
7751 (mips_hard_regno_mode_ok): Delete.
7752 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
7753 (mips_hard_regno_mode_ok_p): ...this and make static.
7754 (mips_hard_regno_mode_ok_p): Rename to...
7755 (mips_hard_regno_mode_ok_uncached): ...this.
7756 (mips_hard_regno_mode_ok): New function.
7757 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
7758 of HARD_REGNO_MODE_OK.
7759 (mips_option_override): Update after above name changes.
7760 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7761 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
7762 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
7763 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
7764 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
7765 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7766 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
7767 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
7768 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
7769 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7770 (msp430_hard_regno_mode_ok): Make static and return a bool.
7771 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
7772 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
7773 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
7775 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7776 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
7777 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
7778 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
7779 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
7780 (PA_HARD_REGNO_MODE_OK): ...this
7781 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
7782 (PA_HARD_REGNO_MODE_OK): ...this.
7783 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7784 (pa_hard_regno_mode_ok): New function.
7785 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
7786 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7787 (pdp11_hard_regno_mode_ok): New function.
7788 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
7789 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
7791 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
7793 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7794 (rs6000_hard_regno_mode_ok): Rename to...
7795 (rs6000_hard_regno_mode_ok_uncached): ...this.
7796 (rs6000_init_hard_regno_mode_ok): Update accordingly.
7797 (rs6000_hard_regno_mode_ok): New function.
7798 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
7799 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
7800 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
7801 (riscv_hard_regno_mode_ok): ...this and make static.
7802 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7803 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
7804 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
7805 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7806 (rl78_hard_regno_mode_ok): Make static and return bool.
7807 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
7808 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
7810 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
7811 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7812 (rs6000_hard_regno_mode_ok): Rename to...
7813 (rs6000_hard_regno_mode_ok_uncached): ...this.
7814 (rs6000_init_hard_regno_mode_ok): Update accordingly.
7815 (rs6000_hard_regno_mode_ok): New function.
7816 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
7817 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
7818 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7819 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
7820 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
7821 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
7822 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7823 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
7824 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
7825 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7826 (sh_hard_regno_mode_ok): Make static.
7827 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
7828 instead of HARD_REGNO_MODE_OK.
7829 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
7830 (sparc_mode_class): Delete.
7831 (HARD_REGNO_MODE_OK): Delete.
7832 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7833 (hard_regno_mode_classes): Make static.
7834 (sparc_mode_class): Likewise.
7835 (sparc_hard_regno_mode_ok): New function.
7836 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
7837 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
7838 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
7840 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7841 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
7842 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
7843 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
7844 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
7845 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7846 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
7847 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
7848 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7849 (visium_hard_regno_mode_ok): New function.
7850 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
7851 instead of HARD_REGNO_MODE_OK.
7852 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
7853 (HARD_REGNO_MODE_OK): Delete.
7854 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
7855 (xtensa_hard_regno_mode_ok_p): ...this and make static.
7856 (xtensa_option_override): Update accordingly.
7857 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7858 (xtensa_hard_regno_mode_ok): New function.
7859 * system.h (HARD_REGNO_MODE_OK): Poison.
7861 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7862 Alan Hayward <alan.hayward@arm.com>
7863 David Sherwood <david.sherwood@arm.com>
7865 * target.def (hard_regno_call_part_clobbered): New hook.
7866 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
7867 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
7868 * doc/tm.texi: Regenerate.
7869 * hooks.h (hook_bool_uint_mode_false): Declare.
7870 * hooks.c (hook_bool_uint_mode_false): New function.
7871 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7872 * cselib.c (cselib_process_insn): Use
7873 targetm.hard_regno_call_part_clobbered instead of
7874 HARD_REGNO_CALL_PART_CLOBBERED.
7875 * ira-conflicts.c (ira_build_conflicts): Likewise.
7876 * ira-costs.c (ira_tune_allocno_costs): Likewise.
7877 * lra-constraints.c (need_for_call_save_p): Likewise.
7878 * lra-lives.c: Include target.h.
7879 (check_pseudos_live_through_calls): Use
7880 targetm.hard_regno_call_part_clobbered instead of
7881 HARD_REGNO_CALL_PART_CLOBBERED.
7882 * regcprop.c: Include target.h.
7883 (copyprop_hardreg_forward_1): Use
7884 targetm.hard_regno_call_part_clobbered instead of
7885 HARD_REGNO_CALL_PART_CLOBBERED.
7886 * reginfo.c (choose_hard_reg_mode): Likewise.
7887 * regrename.c (check_new_reg_p): Likewise.
7888 * reload.c (find_equiv_reg): Likewise.
7889 * reload1.c (emit_reload_insns): Likewise.
7890 * sched-deps.c (deps_analyze_insn): Likewise.
7891 * sel-sched.c (init_regs_for_mode): Likewise.
7892 (mark_unavailable_hard_regs): Likewise.
7893 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
7894 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7895 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
7897 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7898 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7899 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
7901 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
7903 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7904 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7905 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
7907 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7908 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7909 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
7911 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7912 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
7914 * config/powerpcspe/powerpcspe.c
7915 (rs6000_hard_regno_call_part_clobbered): New function.
7916 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7917 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7918 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
7920 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7921 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7922 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
7924 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7925 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7926 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
7928 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7929 Alan Hayward <alan.hayward@arm.com>
7930 David Sherwood <david.sherwood@arm.com>
7932 * rtl.h (subreg_memory_offset): Declare.
7933 * emit-rtl.c (subreg_memory_offset): New function.
7934 * expmed.c (store_bit_field_1): Use it.
7935 * expr.c (undefined_operand_subword_p): Likewise.
7936 * simplify-rtx.c (simplify_subreg): Likewise.
7938 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
7940 PR rtl-optimization/57448
7943 * optabs.c (expand_atomic_load): Place compiler memory barriers if
7944 using atomic_load pattern.
7945 (expand_atomic_store): Likewise.
7947 2017-09-04 Jakub Jelinek <jakub@redhat.com>
7950 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
7951 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
7952 handling. Use replace_call_with_value with NULL instead of
7953 gsi_replace, unlink_stmt_vdef and release_defs.
7955 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
7958 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
7960 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7963 * rtl.h (emit_library_call_value_1): Declare.
7964 (emit_library_call): Replace declaration with a series of overloads.
7965 Remove the parameter count argument.
7966 (emit_library_call_value): Likewise.
7967 * calls.c (emit_library_call_value_1): Make global. Replace varargs
7968 with an "rtx_mode_t *".
7969 (emit_library_call_value): Delete.
7970 (emit_library_call): Likewise.
7971 * asan.c (asan_emit_stack_protection): Update calls accordingly.
7972 (asan_emit_allocas_unpoison): Likewise.
7973 * builtins.c (expand_builtin_powi): Likewise.
7974 (expand_asan_emit_allocas_unpoison): Likewise.
7975 * cfgexpand.c (expand_main_function): Likewise.
7976 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
7977 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
7978 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
7979 * config/arm/arm.c (arm_trampoline_init): Likewise.
7980 (arm_call_tls_get_addr): Likewise.
7981 (arm_expand_divmod_libfunc): Likewise.
7982 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
7983 (smulsi3_highpart): Likewise.
7984 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
7985 (c6x_expand_compare): Likewise.
7986 (c6x_expand_movmem): Likewise.
7987 * config/frv/frv.c (frv_trampoline_init): Likewise.
7988 * config/i386/i386.c (ix86_trampoline_init): Likewise.
7989 (ix86_expand_divmod_libfunc): Likewise.
7990 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
7991 (ia64_expand_compare): Likewise.
7992 (ia64_profile_hook): Likewise.
7993 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
7994 (nonlocal_goto): Likewise.
7995 (restore_stack_nonlocal): Likewise.
7996 * config/m32r/m32r.c (block_move_call): Likewise.
7997 (m32r_trampoline_init): Likewise.
7998 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
7999 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
8000 (m68k_call_m68k_read_tp): Likewise.
8001 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
8002 (microblaze_expand_divide): Likewise.
8003 * config/mips/mips.h (mips_args): Likewise.
8004 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
8005 (MIPS_ICACHE_SYNC): Likewise.
8006 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
8007 (nios2_trampoline_init): Likewise.
8008 * config/pa/pa.c (hppa_tls_call): Likewise.
8009 (pa_trampoline_init): Likewise.
8010 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
8011 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
8012 (expand_strn_compare): Likewise.
8013 (rs6000_generate_compare): Likewise.
8014 (rs6000_expand_float128_convert): Likewise.
8015 (output_profile_hook): Likewise.
8016 (rs6000_trampoline_init): Likewise.
8017 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
8018 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
8019 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
8020 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
8021 (rs6000_generate_compare): Likewise.
8022 (rs6000_expand_float128_convert): Likewise.
8023 (output_profile_hook): Likewise.
8024 (rs6000_trampoline_init): Likewise.
8025 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
8026 * config/sh/sh.c (sh_trampoline_init): Likewise.
8027 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
8028 (sparc_emit_float_lib_cmp): Likewise.
8029 (sparc32_initialize_trampoline): Likewise.
8030 (sparc64_initialize_trampoline): Likewise.
8031 (sparc_profile_hook): Likewise.
8032 * config/spu/spu.c (ea_load_store): Likewise.
8033 * config/spu/spu.md (floatunssidf2): Likewise.
8034 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
8035 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
8036 * config/visium/visium.c (expand_block_move_4): Likewise.
8037 (expand_block_move_2): Likewise.
8038 (expand_block_move_1): Likewise.
8039 (expand_block_set_4): Likewise.
8040 (expand_block_set_2): Likewise.
8041 (expand_block_set_1): Likewise.
8042 (visium_trampoline_init): Likewise.
8043 (visium_profile_hook): Likewise.
8044 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
8045 (xtensa_setup_frame_addresses): Likewise.
8046 (xtensa_trampoline_init): Likewise.
8047 * except.c (sjlj_emit_function_enter): Likewise.
8048 (sjlj_emit_function_exit): Likewise.
8049 * explow.c (allocate_dynamic_stack_space): Likewise.
8050 (probe_stack_range): Likewise.
8051 * expr.c (convert_mode_scalar): Likewise.
8052 * optabs.c (expand_binop): Likewise.
8053 (expand_twoval_binop_libfunc): Likewise.
8054 (expand_unop): Likewise.
8055 (prepare_cmp_insn): Likewise.
8056 (prepare_float_lib_cmp): Likewise.
8057 (expand_float): Likewise.
8058 (expand_fix): Likewise.
8059 (expand_fixed_convert): Likewise.
8060 (maybe_emit_sync_lock_test_and_set): Likewise.
8061 (expand_atomic_compare_and_swap): Likewise.
8062 (expand_mem_thread_fence): Likewise.
8063 (expand_atomic_fetch_op): Likewise.
8065 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
8067 * doc/generic.texi (OpenACC): Adjust URL.
8068 * doc/invoke.texi (C Dialect Options): Ditto.
8070 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
8072 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
8073 predicate for operand 1. Add (m,<S>) constraint.
8074 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
8075 Prevent memory operand 1 with register operand 2.
8077 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
8079 PR rtl-optimization/82024
8080 * combine.c (try_combine): If the combination result is a PARALLEL,
8081 and we only need to retain the SET in there that would be placed
8082 at I2, check that we can place that at I3 instead, before doing so.
8084 2017-09-01 Jakub Jelinek <jakub@redhat.com>
8087 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
8089 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
8090 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
8093 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
8094 Jeff Law <law@redhat.com>
8096 * varasm.c (bss_initializer_p): Do not put constants into .bss
8097 (categorize_decl_for_section): Handle bss_initializer_p returning
8098 false when DECL_INITIAL is NULL.
8100 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8103 * config/s390/s390.c (s390_can_inline_p): New function.
8105 2017-09-01 Jeff Law <law@redhat.com>
8107 PR tree-optimization/82052
8108 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
8109 Always initialize the returned slot after a hash table miss
8110 when INSERT is true.
8112 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
8114 * config/s390/s390.md (mem_signal_fence): Remove.
8115 * doc/md.texi (mem_signal_fence): Remove.
8116 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
8118 * target-insns.def (mem_signal_fence): Remove.
8120 2017-09-01 Jakub Jelinek <jakub@redhat.com>
8123 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
8126 * asan.c (create_odr_indicator): Strip name encoding from assembler
8127 name before appending it after __odr_asan_.
8129 2017-09-01 Martin Liska <mliska@suse.cz>
8131 PR tree-optimization/82059
8132 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
8133 frequency only when an edge is redirected.
8135 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
8137 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
8138 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
8139 (arc_conditional_register_usage): Remove ARC600 lp_count
8141 (arc_file_start): Emit Tag_ARC_CPU_variation.
8142 (arc_can_use_doloop_p): New conditions to use ZOLs.
8143 (hwloop_fail): New function.
8144 (hwloop_optimize): Likewise.
8145 (hwloop_pattern_reg): Likewise.
8146 (arc_doloop_hooks): New struct, to be used with reorg_loops.
8147 (arc_reorg_loops): New function, calls reorg_loops.
8148 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
8149 (arc600_corereg_hazard): Remove ZOL checking, case handled by
8151 (arc_loop_hazard): Remove function, functionality moved into
8153 (arc_hazard): Remove arc_loop_hazard call.
8154 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
8155 into hwloop_optimize.
8156 (arc_label_align): Remove ZOL handling.
8157 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
8158 * config/arc/arc.md (doloop_begin): Remove pattern.
8159 (doloop_begin_i): Likewise.
8160 (doloop_end_i): Likewise.
8161 (doloop_fallback): Likewise.
8162 (doloop_fallback_m): Likewise.
8163 (doloop_end): Reimplement expand.
8164 (arc_lp): New pattern for LP instruction.
8165 (loop_end): New pattern.
8166 (loop_fail): Likewise.
8167 (decrement_and_branch_until_zero): Likewise.
8168 * config/arc/arc.opt (mlpc-width): New option.
8169 * doc/invoke.texi (mlpc-width): Document option.
8171 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
8173 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
8174 (arc_ccfsm_advance): Fix checking for delay slots.
8175 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
8177 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
8179 * config/arc/arc.md (movqi_insn): Add stores to save constant long
8181 (movhi_insn): Update store instruction constraint which are saving
8182 6-bit short immediates.
8183 (movsi_insn): Consider also short scaled load operations.
8184 (zero_extendhisi2_i): Use Usd constraint instead of T.
8185 (extendhisi2_i): Add q constraint.
8186 (arc_clzsi2): Add type and length attributes.
8187 (arc_ctzsi2): Likewise.
8188 * config/arc/constraints.md (Usc): Update constraint, the
8189 assembler can parse two relocations for a single instruction.
8191 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
8193 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
8194 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
8196 2017-08-31 Olivier Hainque <hainque@adacore.com>
8198 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
8199 match as powerpc-wrs-vxworks*.
8201 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
8203 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
8204 register constraint for by-element operand.
8205 (aarch64_mls_elt_merge<mode>): Likewise.
8207 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
8209 * config/arc/arc.c (arc_can_follow_jump): Check for short
8212 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
8214 * config.gcc: Use g.opt for arc.
8215 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
8216 functionality moved to ...
8217 (legitimate_scaled_address_p): New function, ...here.
8218 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
8219 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
8220 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
8222 (arc_override_options): Handle G option.
8223 (arc_output_pic_addr_const): Correct function definition.
8224 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
8225 (arc_decl_anon_ns_mem_p): Delete.
8226 (arc_in_small_data_p): Overhaul this function to take into
8227 consideration the value given via G option.
8228 (arc_rewrite_small_data_1): Renamed and corrected old
8229 arc_rewrite_small_data function.
8230 (arc_rewrite_small_data): New function.
8231 (small_data_pattern): Don't use pic_offset_table_rtx.
8232 * config/arc/arc.h (CC1_SPEC): Recognize G option.
8233 * config/arc/simdext.md (movmisalignv2hi): Use
8234 prepare_move_operands function.
8236 (movmisalign*): Likewise.
8237 * doc/invoke.texi (ARC options): Document -G option.
8239 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
8241 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
8243 * config/arc/arc.c (arc_print_operand): Output scalled address for
8244 sdata whenever is possible.
8245 (arc_in_small_data_p): Allow sdata for 64bit datum when double
8246 load/stores are available.
8247 (compact_sda_memory_operand): Check for the alignment required by
8248 code density instructions.
8249 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
8251 * config/arc/constraints.md (Usd): Update constraint.
8252 (Us0): New constraint.
8253 (Usc): Update constraint.
8255 2017-08-31 Richard Biener <rguenther@suse.de>
8258 * dwarf2out.c (dwarf2out_early_global_decl): Process each
8261 2017-08-31 Tamar Christina <tamar.christina@arm.com>
8263 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
8264 Resize type_signature.
8266 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
8267 Alan Hayward <alan.hayward@arm.com>
8268 David Sherwood <david.sherwood@arm.com>
8270 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
8271 subregs whose inner modes can be stored in GPRs.
8272 (aarch64_classify_index): Likewise.
8274 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
8275 Alan Hayward <alan.hayward@arm.com>
8276 David Sherwood <david.sherwood@arm.com>
8278 * config/aarch64/iterators.md (V_cmp_result): Rename to...
8279 (V_INT_EQUIV): ...this.
8280 (v_cmp_result): Rename to...
8281 (v_int_equiv): ...this.
8282 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
8283 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
8284 (copysign<mode>3): Likewise.
8285 (aarch64_simd_bsl<mode>_internal): Likewise.
8286 (aarch64_simd_bsl<mode>): Likewise.
8287 (vec_cmp<mode><mode>): Likewise.
8288 (vcond<mode><mode>): Likewise.
8289 (vcond<v_cmp_mixed><mode>): Likewise.
8290 (vcondu<mode><v_cmp_mixed>): Likewise.
8291 (aarch64_cm<optab><mode>): Likewise.
8292 (aarch64_cmtst<mode>): Likewise.
8293 (aarch64_fac<optab><mode>): Likewise.
8294 (vec_perm_const<mode>): Likewise.
8295 (vcond_mask_<mode><v_cmp_result>): Rename to...
8296 (vcond_mask_<mode><v_int_equiv>): ...this.
8297 (vec_cmp<mode><v_cmp_result>): Rename to...
8298 (vec_cmp<mode><v_int_equiv>): ...this.
8300 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
8301 Alan Hayward <alan.hayward@arm.com>
8302 David Sherwood <david.sherwood@arm.com>
8304 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
8306 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
8307 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
8308 (UNSPEC_LD4_DREG): New unspecs.
8309 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
8310 (aarch64_ld2<mode>_dreg_be): Replace with...
8311 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
8313 (aarch64_ld3<mode>_dreg_le)
8314 (aarch64_ld3<mode>_dreg_be): Replace with...
8315 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
8317 (aarch64_ld4<mode>_dreg_le)
8318 (aarch64_ld4<mode>_dreg_be): Replace with...
8319 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
8322 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8324 PR tree-optimization/81987
8325 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
8326 insert an initializer in a location not dominated by the stride
8329 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
8331 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
8332 on the entire header of the finally block in the fallthru case.
8334 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
8336 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
8338 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
8340 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
8341 rs6000_emit_move_from_cr and call renamed function.
8342 (rs6000_emit_prologue): Call renamed functions.
8343 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
8344 movesi_from_cr, remove volatile CRs.
8346 2017-08-30 Jon Beniston <jon@beniston.com>
8347 Richard Biener <rguenther@suse.de>
8349 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
8350 of VECTOR_MODE_P check.
8351 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
8352 element vector types.
8354 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8356 * df.h (df_read_modify_subreg_p): Remove in favor of...
8357 * rtl.h (read_modify_subreg_p): ...this new function. Take a
8358 const_rtx instead of an rtx.
8359 * cprop.c (local_cprop_find_used_regs): Update accordingly.
8360 * df-problems.c (df_word_lr_mark_ref): Likewise.
8361 * ira-lives.c (mark_pseudo_reg_live): Likewise.
8362 (mark_pseudo_reg_dead): Likewise.
8363 (mark_ref_dead): Likewise.
8364 * reginfo.c (init_subregs_of_mode): Likewise.
8365 * sched-deps.c (sched_analyze_1): Likewise.
8366 * df-scan.c (df_def_record_1): Likewise.
8367 (df_uses_record): Likewise.
8368 (df_read_modify_subreg_p): Remove in favor of...
8369 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
8370 const_rtx instead of an rtx.
8372 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8373 Alan Hayward <alan.hayward@arm.com>
8374 David Sherwood <david.sherwood@arm.com>
8376 * rtl.h (partial_subreg_p): New function.
8377 * caller-save.c (save_call_clobbered_regs): Use it.
8378 * calls.c (expand_call): Likewise.
8379 * combine.c (combinable_i3pat): Likewise.
8380 (simplify_set): Likewise.
8381 (make_extraction): Likewise.
8382 (make_compound_operation_int): Likewise.
8383 (gen_lowpart_or_truncate): Likewise.
8384 (force_to_mode): Likewise.
8385 (make_field_assignment): Likewise.
8386 (reg_truncated_to_mode): Likewise.
8387 (record_truncated_value): Likewise.
8388 (move_deaths): Likewise.
8389 * cse.c (record_jump_cond): Likewise.
8390 (cse_insn): Likewise.
8391 * cselib.c (cselib_lookup_1): Likewise.
8392 * expmed.c (extract_bit_field_using_extv): Likewise.
8393 * function.c (assign_parm_setup_reg): Likewise.
8394 * ifcvt.c (noce_convert_multiple_sets): Likewise.
8395 * ira-build.c (create_insn_allocnos): Likewise.
8396 * lra-coalesce.c (merge_pseudos): Likewise.
8397 * lra-constraints.c (match_reload): Likewise.
8398 (simplify_operand_subreg): Likewise.
8399 (curr_insn_transform): Likewise.
8400 * lra-lives.c (process_bb_lives): Likewise.
8401 * lra.c (new_insn_reg): Likewise.
8402 (lra_substitute_pseudo): Likewise.
8403 * regcprop.c (mode_change_ok): Likewise.
8404 (maybe_mode_change): Likewise.
8405 (copyprop_hardreg_forward_1): Likewise.
8406 * reload.c (push_reload): Likewise.
8407 (find_reloads): Likewise.
8408 (find_reloads_subreg_address): Likewise.
8409 * reload1.c (alter_reg): Likewise.
8410 (eliminate_regs_1): Likewise.
8411 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8413 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
8415 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
8416 back to if statements, including unpack.
8418 2017-08-30 Martin Liska <mliska@suse.cz>
8421 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
8423 (func_checker::compare_asm_inputs_outputs): ... this function.
8424 (func_checker::compare_gimple_asm): Use the function to compare
8425 also ASM constrains.
8426 * ipa-icf-gimple.h: Rename the function.
8428 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8429 Alan Hayward <alan.hayward@arm.com>
8430 David Sherwood <david.sherwood@arm.com>
8432 * coretypes.h (complex_mode): New type.
8433 * gdbhooks.py (build_pretty_printer): Handle it.
8434 * machmode.h (complex_mode): New class.
8435 (complex_mode::includes_p): New function.
8436 (is_complex_int_mode): Likewise.
8437 (is_complex_float_mode): Likewise.
8438 * genmodes.c (get_mode_class): Handle complex mode classes.
8439 * function.c (expand_function_end): Use is_complex_int_mode.
8441 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8442 Alan Hayward <alan.hayward@arm.com>
8443 David Sherwood <david.sherwood@arm.com>
8445 * coretypes.h (scalar_mode_pod): New typedef.
8446 * gdbhooks.py (build_pretty_printer): Handle it.
8447 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
8448 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
8449 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
8450 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
8453 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8454 Alan Hayward <alan.hayward@arm.com>
8455 David Sherwood <david.sherwood@arm.com>
8457 * machmode.h (mode_for_vector): Take a scalar_mode instead
8459 * stor-layout.c (mode_for_vector): Likewise.
8460 * explow.c (promote_mode): Use as_a <scalar_mode>.
8461 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
8463 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8464 Alan Hayward <alan.hayward@arm.com>
8465 David Sherwood <david.sherwood@arm.com>
8467 * target.def (preferred_simd_mode): Take a scalar_mode
8468 instead of a machine_mode.
8469 * targhooks.h (default_preferred_simd_mode): Likewise.
8470 * targhooks.c (default_preferred_simd_mode): Likewise.
8471 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
8472 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
8473 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
8474 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
8475 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
8476 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
8477 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
8478 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
8480 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
8481 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
8482 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
8483 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
8484 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
8485 * doc/tm.texi: Regenerate.
8486 * optabs-query.c (can_vec_mask_load_store_p): Return false for
8489 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8490 Alan Hayward <alan.hayward@arm.com>
8491 David Sherwood <david.sherwood@arm.com>
8493 * target.def (scalar_mode_supported_p): Take a scalar_mode
8494 instead of a machine_mode.
8495 * targhooks.h (default_scalar_mode_supported_p): Likewise.
8496 * targhooks.c (default_scalar_mode_supported_p): Likewise.
8497 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
8498 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
8499 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
8500 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
8501 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
8502 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
8503 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
8504 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
8505 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
8506 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
8507 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
8508 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
8510 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
8511 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
8512 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
8513 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
8514 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
8516 * doc/tm.texi: Regenerate.
8518 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8519 Alan Hayward <alan.hayward@arm.com>
8520 David Sherwood <david.sherwood@arm.com>
8522 * coretypes.h (opt_scalar_mode): New typedef.
8523 * gdbhooks.py (build_pretty_printers): Handle it.
8524 * machmode.h (mode_iterator::get_2xwider): Add overload for
8526 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
8528 * expr.c (convert_mode_scalar): Likewise.
8529 * omp-low.c (omp_clause_aligned_alignment): Likewise.
8530 * optabs.c (expand_float): Likewise.
8531 (expand_fix): Likewise.
8532 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8534 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8535 Alan Hayward <alan.hayward@arm.com>
8536 David Sherwood <david.sherwood@arm.com>
8538 * optabs.c (expand_float): Explicitly check for scalars before
8539 using a branching expansion.
8540 (expand_fix): Likewise.
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 * expr.c (convert_mode): Split scalar handling out into...
8547 (convert_mode_scalar): ...this new function. Treat the modes
8550 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8551 Alan Hayward <alan.hayward@arm.com>
8552 David Sherwood <david.sherwood@arm.com>
8554 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
8556 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
8558 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8559 Alan Hayward <alan.hayward@arm.com>
8560 David Sherwood <david.sherwood@arm.com>
8562 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
8563 rather than a machine_mode.
8564 (fixed_from_string): Likewise.
8565 (fixed_convert): Likewise.
8566 (fixed_convert_from_int): Likewise.
8567 (fixed_convert_from_real): Likewise.
8568 (real_convert_from_fixed): Likewise.
8569 * fixed-value.c (fixed_from_double_int): Likewise.
8570 (fixed_from_string): Likewise.
8571 (fixed_convert): Likewise.
8572 (fixed_convert_from_int): Likewise.
8573 (fixed_convert_from_real): Likewise.
8574 (real_convert_from_fixed): Likewise.
8575 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
8577 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8578 Alan Hayward <alan.hayward@arm.com>
8579 David Sherwood <david.sherwood@arm.com>
8581 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
8582 of separate mode class checks. Do not allow vector modes here.
8583 (immed_wide_int_const): Use as_a <scalar_mode>.
8584 * explow.c (trunc_int_for_mode): Likewise.
8585 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
8586 (wi::shwi): Likewise.
8587 (wi::min_value): Likewise.
8588 (wi::max_value): Likewise.
8589 * dwarf2out.c (loc_descriptor): Likewise.
8590 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
8593 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8594 Alan Hayward <alan.hayward@arm.com>
8595 David Sherwood <david.sherwood@arm.com>
8597 * tree.h (SCALAR_TYPE_MODE): New macro.
8598 * expr.c (expand_expr_addr_expr_1): Use it.
8599 (expand_expr_real_2): Likewise.
8600 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
8601 (fold_convert_const_fixed_from_int): Likewise.
8602 (fold_convert_const_fixed_from_real): Likewise.
8603 (native_encode_fixed): Likewise
8604 (native_encode_complex): Likewise
8605 (native_encode_vector): Likewise.
8606 (native_interpret_fixed): Likewise.
8607 (native_interpret_real): Likewise.
8608 (native_interpret_complex): Likewise.
8609 (native_interpret_vector): Likewise.
8610 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
8611 (simd_clone_adjust_argument_types): Likewise.
8612 (simd_clone_init_simd_arrays): Likewise.
8613 (simd_clone_adjust): Likewise.
8614 * stor-layout.c (layout_type): Likewise.
8615 * tree.c (build_minus_one_cst): Likewise.
8616 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
8617 * tree-inline.c (estimate_move_cost): Likewise.
8618 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
8619 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
8620 (vectorizable_reduction): Likewise.
8621 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
8622 (vect_recog_mixed_size_cond_pattern): Likewise.
8623 (check_bool_pattern): Likewise.
8624 (adjust_bool_pattern): Likewise.
8625 (search_type_for_mask_1): Likewise.
8626 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
8627 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8628 (vectorizable_load): Likewise.
8629 (vectorizable_store): Likewise.
8630 * ubsan.c (ubsan_encode_value): Likewise.
8631 * varasm.c (output_constant): Likewise.
8633 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8634 Alan Hayward <alan.hayward@arm.com>
8635 David Sherwood <david.sherwood@arm.com>
8637 * coretypes.h (scalar_mode): New class.
8638 * machmode.h (scalar_mode): Likewise.
8639 (scalar_mode::includes_p): New function.
8640 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
8641 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
8642 * genmodes.c (get_mode_class): Handle remaining scalar modes.
8643 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
8644 * expmed.c (store_bit_field_1): Likewise.
8645 (extract_bit_field_1): Likewise.
8646 * expr.c (write_complex_part): Likewise.
8647 (read_complex_part): Likewise.
8648 (emit_move_complex_push): Likewise.
8649 (expand_expr_real_2): Likewise.
8650 * function.c (assign_parm_setup_reg): Likewise.
8651 (assign_parms_unsplit_complex): Likewise.
8652 * optabs.c (expand_binop): Likewise.
8653 * rtlanal.c (subreg_get_info): Likewise.
8654 * simplify-rtx.c (simplify_immed_subreg): Likewise.
8655 * varasm.c (output_constant_pool_2): Likewise.
8657 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8658 Alan Hayward <alan.hayward@arm.com>
8659 David Sherwood <david.sherwood@arm.com>
8661 * expmed.c (extract_high_half): Use scalar_int_mode and remove
8663 (expmed_mult_highpart_optab): Likewise.
8664 (expmed_mult_highpart): Likewise.
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 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
8671 instead of a machine_mode.
8672 (builtin_memset_read_str): Likewise.
8673 * builtins.c (c_readstr): Likewise.
8674 (builtin_memcpy_read_str): Likewise.
8675 (builtin_strncpy_read_str): Likewise.
8676 (builtin_memset_read_str): Likewise.
8677 (builtin_memset_gen_str): Likewise.
8678 (expand_builtin_signbit): Use scalar_int_mode for local variables.
8679 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
8680 instead of a machine_mode.
8681 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
8683 (make_extraction): Likewise.
8684 (try_widen_shift_mode): Take and return scalar_int_modes instead
8686 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
8687 a scalar_int_mode instead of a machine_mode.
8688 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
8689 (avr_addr_space_pointer_mode): Likewise.
8690 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
8691 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
8692 (msp430_unwind_word_mode): Likewise.
8693 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
8694 (spu_addr_space_pointer_mode): Likewise.
8695 (spu_addr_space_address_mode): Likewise.
8696 (spu_libgcc_cmp_return_mode): Likewise.
8697 (spu_libgcc_shift_count_mode): Likewise.
8698 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
8699 (rl78_addr_space_pointer_mode): Likewise.
8700 (fl78_unwind_word_mode): Likewise.
8701 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
8703 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
8704 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
8705 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
8706 (mips_valid_pointer_mode): Likewise.
8707 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
8708 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
8709 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
8711 (ft32_addr_space_address_mode): Likewise.
8712 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
8713 scalar_int_mode instead of a machine_mode.
8714 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
8716 (m32c_addr_space_address_mode): Likewise.
8717 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
8718 (rs6000_eh_return_filter_mode): Likewise.
8719 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
8720 (rs6000_eh_return_filter_mode): Likewise.
8721 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
8722 (s390_libgcc_shift_count_mode): Likewise.
8723 (s390_unwind_word_mode): Likewise.
8724 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
8726 * target.def (mode_rep_extended): Likewise.
8727 (valid_pointer_mode): Likewise.
8728 (addr_space.valid_pointer_mode): Likewise.
8729 (eh_return_filter_mode): Return a scalar_int_mode rather than
8731 (libgcc_cmp_return_mode): Likewise.
8732 (libgcc_shift_count_mode): Likewise.
8733 (unwind_word_mode): Likewise.
8734 (addr_space.pointer_mode): Likewise.
8735 (addr_space.address_mode): Likewise.
8736 * doc/tm.texi: Regenerate.
8737 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
8739 (do_jump): Use scalar_int_mode for local variables.
8740 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
8741 rather than a machine_mode.
8742 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
8743 (scompare_loc_descriptor_wide): Likewise.
8744 (scompare_loc_descriptor_narrow): Likewise.
8745 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
8747 * except.c (sjlj_emit_dispatch_table): Likewise.
8748 (expand_builtin_eh_copy_values): Likewise.
8749 * explow.c (convert_memory_address_addr_space_1): Likewise.
8750 Take a scalar_int_mode rather than a machine_mode.
8751 (convert_memory_address_addr_space): Take a scalar_int_mode rather
8752 than a machine_mode.
8753 (memory_address_addr_space): Use scalar_int_mode for local variables.
8754 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
8755 rather than a machine_mode.
8756 * expmed.c (mask_rtx): Likewise.
8757 (init_expmed_one_conv): Likewise.
8758 (expand_mult_highpart_adjust): Likewise.
8759 (extract_high_half): Likewise.
8760 (expmed_mult_highpart_optab): Likewise.
8761 (expmed_mult_highpart): Likewise.
8762 (expand_smod_pow2): Likewise.
8763 (expand_sdiv_pow2): Likewise.
8764 (emit_store_flag_int): Likewise.
8765 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
8767 (extract_low_bits): Likewise.
8768 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
8770 * expr.c (pieces_addr::adjust): Likewise.
8771 (can_store_by_pieces): Likewise.
8772 (store_by_pieces): Likewise.
8773 (clear_by_pieces_1): Likewise.
8774 (expand_expr_addr_expr_1): Likewise.
8775 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
8776 (expand_expr_real_1): Likewise.
8777 (try_casesi): Likewise.
8778 * final.c (shorten_branches): Likewise.
8779 * fold-const.c (fold_convert_const_int_from_fixed): Change the
8780 type of "mode" to machine_mode.
8781 * internal-fn.c (expand_arith_overflow_result_store): Take a
8782 scalar_int_mode rather than a machine_mode.
8783 (expand_mul_overflow): Use scalar_int_mode for local variables.
8784 * loop-doloop.c (doloop_modify): Likewise.
8785 (doloop_optimize): Likewise.
8786 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
8787 than a machine_mode.
8788 (expand_doubleword_shift_condmove): Likewise.
8789 (expand_doubleword_shift): Likewise.
8790 (expand_doubleword_clz): Likewise.
8791 (expand_doubleword_popcount): Likewise.
8792 (expand_doubleword_parity): Likewise.
8793 (expand_absneg_bit): Use scalar_int_mode for local variables.
8794 (prepare_float_lib_cmp): Likewise.
8795 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
8796 rather than a machine_mode.
8797 (convert_memory_address_addr_space): Likewise.
8798 (get_mode_bounds): Likewise.
8799 (get_address_mode): Return a scalar_int_mode rather than a
8801 * rtlanal.c (get_address_mode): Likewise.
8802 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
8803 than a machine_mode.
8804 * targhooks.c (default_mode_rep_extended): Likewise.
8805 (default_valid_pointer_mode): Likewise.
8806 (default_addr_space_valid_pointer_mode): Likewise.
8807 (default_eh_return_filter_mode): Return a scalar_int_mode rather
8808 than a machine_mode.
8809 (default_libgcc_cmp_return_mode): Likewise.
8810 (default_libgcc_shift_count_mode): Likewise.
8811 (default_unwind_word_mode): Likewise.
8812 (default_addr_space_pointer_mode): Likewise.
8813 (default_addr_space_address_mode): Likewise.
8814 * targhooks.h (default_eh_return_filter_mode): Likewise.
8815 (default_libgcc_cmp_return_mode): Likewise.
8816 (default_libgcc_shift_count_mode): Likewise.
8817 (default_unwind_word_mode): Likewise.
8818 (default_addr_space_pointer_mode): Likewise.
8819 (default_addr_space_address_mode): Likewise.
8820 (default_mode_rep_extended): Take a scalar_int_mode rather than
8822 (default_valid_pointer_mode): Likewise.
8823 (default_addr_space_valid_pointer_mode): Likewise.
8824 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
8826 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
8827 rather than a machine_mode.
8828 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
8829 for local variables.
8830 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8831 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
8832 than a machine_mode.
8834 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8835 Alan Hayward <alan.hayward@arm.com>
8836 David Sherwood <david.sherwood@arm.com>
8838 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
8839 the mode argument to scalar_int_mode.
8840 (do_jump_by_parts_zero_rtx): Likewise.
8841 (do_jump_by_parts_equality_rtx): Likewise.
8842 (do_jump_by_parts_greater): Take a mode argument.
8843 (do_jump_by_parts_equality): Likewise.
8844 (do_jump_1): Update calls accordingly.
8846 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8847 Alan Hayward <alan.hayward@arm.com>
8848 David Sherwood <david.sherwood@arm.com>
8850 * is-a.h (safe_dyn_cast): New function.
8851 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
8852 (jump_table_for_label): Likewise.
8853 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
8854 instead of an rtx_insn *.
8855 (shorten_branches): Use dyn_cast instead of LABEL_P and
8856 JUMP_TABLE_DATA_P. Use jump_table_for_label and
8857 rtx_jump_table_data::get_data_mode.
8858 (final_scan_insn): Likewise.
8860 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8861 Alan Hayward <alan.hayward@arm.com>
8862 David Sherwood <david.sherwood@arm.com>
8864 * combine.c (try_combine): Use is_a <scalar_int_mode> when
8865 trying to combine a full-register integer set with a subreg
8868 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8869 Alan Hayward <alan.hayward@arm.com>
8870 David Sherwood <david.sherwood@arm.com>
8872 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
8873 that is always either address_mode or pointer_mode.
8875 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8876 Alan Hayward <alan.hayward@arm.com>
8877 David Sherwood <david.sherwood@arm.com>
8879 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
8880 when the two are known to be equal.
8882 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8883 Alan Hayward <alan.hayward@arm.com>
8884 David Sherwood <david.sherwood@arm.com>
8886 * simplify-rtx.c (simplify_const_unary_operation): Use
8887 is_a <scalar_int_mode> instead of checking for a nonzero
8888 precision. Forcibly convert op_mode to a scalar_int_mode
8889 in that case. More clearly differentiate the operand and
8890 result modes and use the former when deciding what the value
8891 of a count-bits operation should be. Use is_int_mode instead
8892 of checking for a MODE_INT. Remove redundant check for whether
8893 this mode has a zero precision.
8895 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8896 Alan Hayward <alan.hayward@arm.com>
8897 David Sherwood <david.sherwood@arm.com>
8899 * optabs.c (widen_leading): Change the type of the mode argument
8900 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
8901 (widen_bswap): Likewise.
8902 (expand_parity): Likewise.
8903 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
8904 (expand_ffs): Likewise.
8905 (epand_unop): Check for scalar integer modes before calling the
8908 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8909 Alan Hayward <alan.hayward@arm.com>
8910 David Sherwood <david.sherwood@arm.com>
8912 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
8914 (expand_expr_real_1): Update call accordingly.
8916 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8917 Alan Hayward <alan.hayward@arm.com>
8918 David Sherwood <david.sherwood@arm.com>
8920 * expmed.c (store_bit_field_using_insv): Add op0_mode and
8921 value_mode arguments. Use scalar_int_mode internally.
8922 (store_bit_field_1): Rename the new integer mode from imode
8923 to op0_mode and use it instead of GET_MODE (op0). Update calls
8924 to store_split_bit_field, store_bit_field_using_insv and
8925 store_fixed_bit_field.
8926 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
8927 Use scalar_int_mode internally. Use a bit count rather than a mode
8928 when calculating the largest bit size for get_best_mode.
8929 Update calls to store_split_bit_field and store_fixed_bit_field_1.
8930 (store_fixed_bit_field_1): Add mode and value_mode arguments.
8931 Remove assertion that OP0 has a scalar integer mode.
8932 (store_split_bit_field): Add op0_mode and value_mode arguments.
8933 Update calls to extract_fixed_bit_field.
8934 (extract_bit_field_using_extv): Add an op0_mode argument.
8935 Use scalar_int_mode internally.
8936 (extract_bit_field_1): Rename the new integer mode from imode to
8937 op0_mode and use it instead of GET_MODE (op0). Update calls to
8938 extract_split_bit_field, extract_bit_field_using_extv and
8939 extract_fixed_bit_field.
8940 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
8941 to extract_split_bit_field and extract_fixed_bit_field_1.
8942 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
8943 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
8945 (extract_split_bit_field): Add an op0_mode argument. Update call
8946 to extract_fixed_bit_field.
8948 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8949 Alan Hayward <alan.hayward@arm.com>
8950 David Sherwood <david.sherwood@arm.com>
8952 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
8953 * explow.c (hard_function_value): Likewise.
8954 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
8955 convert_to_mode call outside the loop.
8956 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
8957 for the mode iterator. Require the mode specified by max_pieces
8959 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
8961 (copy_blkmode_to_reg): Likewise.
8962 (set_storage_via_setmem): Likewise.
8963 * optabs.c (prepare_cmp_insn): Likewise.
8964 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
8965 * stor-layout.c (finish_bitfield_representative): Likewise.
8967 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8968 Alan Hayward <alan.hayward@arm.com>
8969 David Sherwood <david.sherwood@arm.com>
8971 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
8972 * expr.c (convert_move): Use them.
8973 (convert_modes): Likewise.
8974 (store_expr_with_bounds): Likewise.
8976 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8977 Alan Hayward <alan.hayward@arm.com>
8978 David Sherwood <david.sherwood@arm.com>
8980 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
8981 parameter for the mode of "x". Remove the "known_x", "known_mode"
8982 and "known_ret" arguments. Change the type of the mode argument
8984 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
8985 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
8986 (reg_num_sign_bit_copies_for_combine): Likewise.
8987 * rtlanal.c (nonzero_bits1): Likewise.
8988 (num_sign_bit_copies1): Likewise.
8989 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
8990 (reg_num_sign_bit_copies_general): Likewise.
8991 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
8992 (reg_nonzero_bits_general): Likewise.
8994 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8995 Alan Hayward <alan.hayward@arm.com>
8996 David Sherwood <david.sherwood@arm.com>
8998 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
8999 than in subroutines. Return 1 for non-integer modes.
9000 (cached_num_sign_bit_copies): Change the type of the mode parameter
9002 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
9003 classes. Handle CONST_INT_P first and then check whether X also
9004 has a scalar integer mode. Check the same thing for inner registers
9005 of a SUBREG and for values that are being extended or truncated.
9007 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9008 Alan Hayward <alan.hayward@arm.com>
9009 David Sherwood <david.sherwood@arm.com>
9011 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
9012 in subroutines. Return the mode mask for non-integer modes.
9013 (cached_nonzero_bits): Change the type of the mode parameter
9015 (nonzero_bits1): Likewise. Remove early exit for other mode
9016 classes. Handle CONST_INT_P first and then check whether X
9017 also has a scalar integer mode.
9019 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9020 Alan Hayward <alan.hayward@arm.com>
9021 David Sherwood <david.sherwood@arm.com>
9023 * expr.c (widest_int_mode_for_size): Make the comment match the code.
9024 Return a scalar_int_mode and assert that the size is greater than
9026 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
9027 (op_by_pieces_d::op_by_pieces_d): Likewise.
9028 (op_by_pieces_d::run): Likewise.
9029 (can_store_by_pieces): Likewise.
9031 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9032 Alan Hayward <alan.hayward@arm.com>
9033 David Sherwood <david.sherwood@arm.com>
9035 * combine.c (extract_left_shift): Add a mode argument and update
9037 (make_compound_operation_int): Change the type of the mode parameter
9038 to scalar_int_mode and update the call to extract_left_shift.
9040 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9041 Alan Hayward <alan.hayward@arm.com>
9042 David Sherwood <david.sherwood@arm.com>
9044 * combine.c (simplify_and_const_int): Change the type of the mode
9045 parameter to scalar_int_mode.
9046 (simplify_and_const_int_1): Likewise. Update recursive call.
9048 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9049 Alan Hayward <alan.hayward@arm.com>
9050 David Sherwood <david.sherwood@arm.com>
9052 * combine.c (simplify_compare_const): Check that the mode is a
9053 scalar_int_mode (rather than VOIDmode) before testing its
9055 (simplify_comparison): Move COMPARISON_P handling out of the
9056 loop and restrict the latter part of the loop to scalar_int_modes.
9057 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
9058 and when considering SUBREG_REGs. Use is_int_mode instead of
9059 checking GET_MODE_CLASS against MODE_INT.
9061 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9062 Alan Hayward <alan.hayward@arm.com>
9063 David Sherwood <david.sherwood@arm.com>
9065 * combine.c (try_widen_shift_mode): Move check for equal modes to...
9066 (simplify_shift_const_1): ...here. Use scalar_int_mode for
9067 shift_unit_mode and for modes involved in scalar shifts.
9069 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9070 Alan Hayward <alan.hayward@arm.com>
9071 David Sherwood <david.sherwood@arm.com>
9073 * combine.c (force_int_to_mode): New function, split out from...
9074 (force_to_mode): ...here. Keep xmode up-to-date and use it
9075 instead of GET_MODE (x).
9077 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9078 Alan Hayward <alan.hayward@arm.com>
9079 David Sherwood <david.sherwood@arm.com>
9081 * optabs-query.h (extraction_insn::struct_mode): Change type to
9082 opt_scalar_int_mode and update comment.
9083 (extraction_insn::field_mode): Change type to scalar_int_mode.
9084 (extraction_insn::pos_mode): Likewise.
9085 * combine.c (make_extraction): Update accordingly.
9086 * optabs-query.c (get_traditional_extraction_insn): Likewise.
9087 (get_optab_extraction_insn): Likewise.
9088 * recog.c (simplify_while_replacing): Likewise.
9089 * expmed.c (narrow_bit_field_mem): Change the type of the mode
9090 parameter to opt_scalar_int_mode.
9092 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9093 Alan Hayward <alan.hayward@arm.com>
9094 David Sherwood <david.sherwood@arm.com>
9096 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
9097 to a scalar_int_mode instead of a machine_mode.
9098 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
9099 (get_best_mode): Return a boolean and use a pointer argument to store
9100 the selected mode. Replace the limit mode parameter with a bit limit.
9101 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
9102 for the values returned by bit_field_mode_iterator::next_mode.
9103 (store_bit_field): Update call to get_best_mode.
9104 (store_fixed_bit_field): Likewise.
9105 (extract_fixed_bit_field): Likewise.
9106 * expr.c (optimize_bitfield_assignment_op): Likewise.
9107 * fold-const.c (optimize_bit_field_compare): Likewise.
9108 (fold_truth_andor_1): Likewise.
9109 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
9110 Update for new type of m_mode.
9111 (get_best_mode): As above.
9113 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9114 Alan Hayward <alan.hayward@arm.com>
9115 David Sherwood <david.sherwood@arm.com>
9117 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
9118 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
9119 (store_bit_field): Check is_a <scalar_int_mode> before calling
9120 strict_volatile_bitfield_p.
9121 (extract_bit_field): Likewise.
9123 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9124 Alan Hayward <alan.hayward@arm.com>
9125 David Sherwood <david.sherwood@arm.com>
9127 * target.def (cstore_mode): Return a scalar_int_mode.
9128 * doc/tm.texi: Regenerate.
9129 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
9130 * targhooks.h (default_cstore_mode): Likewise.
9131 * targhooks.c (default_cstore_mode): Likewise, using a forced
9133 * expmed.c (emit_cstore): Expect the target of the cstore to be
9136 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9137 Alan Hayward <alan.hayward@arm.com>
9138 David Sherwood <david.sherwood@arm.com>
9140 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
9142 (niter_desc): Likewise mode.
9143 (iv_analyze): Add a mode parameter.
9145 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
9146 and change its type to scalar_int_mode.
9147 * loop-iv.c: Update commentary at head of file.
9148 (iv_constant): Pass the mode paraeter before the rtx it describes
9149 and change its type to scalar_int_mode. Remove VOIDmode handling.
9150 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
9151 (iv_extend): Likewise.
9152 (shorten_into_mode): Likewise.
9153 (iv_add): Use scalar_int_mode.
9154 (iv_mult): Likewise.
9155 (iv_shift): Likewise.
9156 (canonicalize_iv_subregs): Likewise.
9157 (get_biv_step_1): Pass the outer_mode parameter before the rtx
9158 it describes and change its mode to scalar_int_mode. Also change
9159 the type of the returned inner_mode to scalar_int_mode.
9160 (get_biv_step): Likewise, turning outer_mode from a pointer
9161 into a direct parameter. Update call to get_biv_step_1.
9162 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
9163 iv_constant and get_biv_step.
9164 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
9165 and change its type to scalar_int_mode. Don't initialise iv->mode
9166 to VOIDmode and remove later checks for its still being VOIDmode.
9167 Update calls to iv_analyze_op and iv_analyze_expr. Check
9168 is_a <scalar_int_mode> when changing the mode under consideration.
9169 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
9170 Update call to iv_analyze_expr.
9171 (iv_analyze_op): Add a mode parameter. Reject subregs whose
9172 inner register is not also a scalar_int_mode. Update call to
9174 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
9175 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
9176 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
9177 separate mode class checks. Update calls to iv_analyze. Remove
9178 fix-up of VOIDmodes after iv_analyze_biv.
9179 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
9180 don't have a scalar_int_mode. Update call to biv_p.
9182 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9183 Alan Hayward <alan.hayward@arm.com>
9184 David Sherwood <david.sherwood@arm.com>
9186 * cfgexpand.c (convert_debug_memory_address): Use
9187 as_a <scalar_int_mode>.
9188 * combine.c (expand_compound_operation): Likewise.
9189 (make_extraction): Likewise.
9190 (change_zero_ext): Likewise.
9191 (simplify_comparison): Likewise.
9192 * cse.c (cse_insn): Likewise.
9193 * dwarf2out.c (minmax_loc_descriptor): Likewise.
9194 (mem_loc_descriptor): Likewise.
9195 (loc_descriptor): Likewise.
9196 * expmed.c (init_expmed_one_mode): Likewise.
9197 (synth_mult): Likewise.
9198 (emit_store_flag_1): Likewise.
9199 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
9200 of a comparison with size.
9201 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
9202 (reduce_to_bit_field_precision): Likewise.
9203 * function.c (expand_function_end): Likewise.
9204 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
9205 * loop-doloop.c (doloop_modify): Likewise.
9206 * optabs.c (expand_binop): Likewise.
9207 (expand_unop): Likewise.
9208 (expand_copysign_absneg): Likewise.
9209 (prepare_cmp_insn): Likewise.
9210 (maybe_legitimize_operand): Likewise.
9211 * recog.c (const_scalar_int_operand): Likewise.
9212 * rtlanal.c (get_address_mode): Likewise.
9213 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9214 (simplify_cond_clz_ctz): Likewise.
9215 * tree-nested.c (get_nl_goto_field): Likewise.
9216 * tree.c (build_vector_type_for_mode): Likewise.
9217 * var-tracking.c (use_narrower_mode): Likewise.
9219 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9220 Alan Hayward <alan.hayward@arm.com>
9221 David Sherwood <david.sherwood@arm.com>
9223 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
9224 * builtins.c (expand_builtin_signbit): Use it.
9225 * cfgexpand.c (expand_debug_expr): Likewise.
9226 * dojump.c (do_jump): Likewise.
9227 (do_compare_and_jump): Likewise.
9228 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
9229 * expmed.c (make_tree): Likewise.
9230 * expr.c (expand_expr_real_2): Likewise.
9231 (expand_expr_real_1): Likewise.
9232 (try_casesi): Likewise.
9233 * fold-const-call.c (fold_const_call_ss): Likewise.
9234 * fold-const.c (unextend): Likewise.
9235 (extract_muldiv_1): Likewise.
9236 (fold_single_bit_test): Likewise.
9237 (native_encode_int): Likewise.
9238 (native_encode_string): Likewise.
9239 (native_interpret_int): Likewise.
9240 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
9241 * internal-fn.c (expand_addsub_overflow): Likewise.
9242 (expand_neg_overflow): Likewise.
9243 (expand_mul_overflow): Likewise.
9244 (expand_arith_overflow): Likewise.
9245 * match.pd: Likewise.
9246 * stor-layout.c (layout_type): Likewise.
9247 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
9248 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
9249 * tree-ssanames.c (get_range_info): Likewise.
9250 * tree-switch-conversion.c (array_value_type) Likewise.
9251 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
9252 (vect_recog_divmod_pattern): Likewise.
9253 (vect_recog_mixed_size_cond_pattern): Likewise.
9254 * tree-vrp.c (extract_range_basic): Likewise.
9255 (simplify_float_conversion_using_ranges): Likewise.
9256 * tree.c (int_fits_type_p): Likewise.
9257 * ubsan.c (instrument_bool_enum_load): Likewise.
9258 * varasm.c (mergeable_string_section): Likewise.
9259 (narrowing_initializer_constant_valid_p): Likewise.
9260 (output_constant): Likewise.
9262 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9263 Alan Hayward <alan.hayward@arm.com>
9264 David Sherwood <david.sherwood@arm.com>
9266 * machmode.h (NARROWEST_INT_MODE): New macro.
9267 * expr.c (alignment_for_piecewise_move): Use it instead of
9268 GET_CLASS_NARROWEST_MODE (MODE_INT).
9269 (push_block): Likewise.
9270 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
9272 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9274 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9275 Alan Hayward <alan.hayward@arm.com>
9276 David Sherwood <david.sherwood@arm.com>
9278 * postreload.c (move2add_valid_value_p): Change the type of the
9279 mode parameter to scalar_int_mode.
9280 (move2add_use_add2_insn): Add a mode parameter and use it instead
9282 (move2add_use_add3_insn): Likewise.
9283 (reload_cse_move2add): Update accordingly.
9285 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9286 Alan Hayward <alan.hayward@arm.com>
9287 David Sherwood <david.sherwood@arm.com>
9289 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
9291 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
9292 * optabs.c (expand_unop): Likewise.
9294 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9295 Alan Hayward <alan.hayward@arm.com>
9296 David Sherwood <david.sherwood@arm.com>
9298 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
9299 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
9300 (popcount_loc_descriptor): Likewise.
9301 (bswap_loc_descriptor): Likewise.
9302 (rotate_loc_descriptor): Likewise.
9303 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
9304 calling the functions above.
9306 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9307 Alan Hayward <alan.hayward@arm.com>
9308 David Sherwood <david.sherwood@arm.com>
9310 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
9312 (try_combine): Likewise.
9313 (simplify_if_then_else): Likewise.
9314 * cse.c (cse_insn): Likewise.
9315 * dwarf2out.c (mem_loc_descriptor): Likewise.
9316 * emit-rtl.c (gen_lowpart_common): Likewise.
9317 * simplify-rtx.c (simplify_truncation): Likewise.
9318 (simplify_binary_operation_1): Likewise.
9319 (simplify_const_relational_operation): Likewise.
9320 (simplify_ternary_operation): Likewise.
9321 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
9323 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9324 Alan Hayward <alan.hayward@arm.com>
9325 David Sherwood <david.sherwood@arm.com>
9327 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
9328 * reload.c (push_reload): Likewise.
9329 (find_reloads): Likewise.
9331 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9332 Alan Hayward <alan.hayward@arm.com>
9333 David Sherwood <david.sherwood@arm.com>
9335 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
9336 (make_compound_operation_int): Likewise.
9337 (change_zero_ext): Likewise.
9338 * expr.c (convert_move): Likewise.
9339 (convert_modes): Likewise.
9340 * fwprop.c (forward_propagate_subreg): Likewise.
9341 * loop-iv.c (get_biv_step_1): Likewise.
9342 * optabs.c (widen_operand): Likewise.
9343 * postreload.c (move2add_valid_value_p): Likewise.
9344 * recog.c (simplify_while_replacing): Likewise.
9345 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9346 (simplify_binary_operation_1): Likewise. Remove redundant
9347 mode equality check.
9349 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9350 Alan Hayward <alan.hayward@arm.com>
9351 David Sherwood <david.sherwood@arm.com>
9353 * combine.c (combine_simplify_rtx): Add checks for
9354 is_a <scalar_int_mode>.
9355 (simplify_if_then_else): Likewise.
9356 (make_field_assignment): Likewise.
9357 (simplify_comparison): Likewise.
9358 * ifcvt.c (noce_try_bitop): Likewise.
9359 * loop-invariant.c (canonicalize_address_mult): Likewise.
9360 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9362 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9363 Alan Hayward <alan.hayward@arm.com>
9364 David Sherwood <david.sherwood@arm.com>
9366 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
9367 is_a <scalar_int_mode> instead of != BLKmode.
9369 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9370 Alan Hayward <alan.hayward@arm.com>
9371 David Sherwood <david.sherwood@arm.com>
9373 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
9374 instead of != VOIDmode.
9375 * combine.c (if_then_else_cond): Likewise.
9376 (change_zero_ext): Likewise.
9377 * dwarf2out.c (mem_loc_descriptor): Likewise.
9378 (loc_descriptor): Likewise.
9379 * rtlanal.c (canonicalize_condition): Likewise.
9380 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
9382 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9383 Alan Hayward <alan.hayward@arm.com>
9384 David Sherwood <david.sherwood@arm.com>
9386 * simplify-rtx.c (simplify_binary_operation_1): Use
9387 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
9389 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9390 Alan Hayward <alan.hayward@arm.com>
9391 David Sherwood <david.sherwood@arm.com>
9393 * wide-int.h (int_traits<unsigned char>) New class.
9394 (int_traits<unsigned short>) Likewise.
9395 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
9396 Use GET_MODE_UNIT_PRECISION and remove redundant test for
9398 * combine.c (set_nonzero_bits_and_sign_copies): Use
9399 is_a <scalar_int_mode>.
9400 (find_split_point): Likewise.
9401 (combine_simplify_rtx): Likewise.
9402 (simplify_logical): Likewise.
9403 (expand_compound_operation): Likewise.
9404 (expand_field_assignment): Likewise.
9405 (make_compound_operation): Likewise.
9406 (extended_count): Likewise.
9407 (change_zero_ext): Likewise.
9408 (simplify_comparison): Likewise.
9409 * dwarf2out.c (scompare_loc_descriptor): Likewise.
9410 (ucompare_loc_descriptor): Likewise.
9411 (minmax_loc_descriptor): Likewise.
9412 (mem_loc_descriptor): Likewise.
9413 (loc_descriptor): Likewise.
9414 * expmed.c (init_expmed_one_mode): Likewise.
9415 * lra-constraints.c (lra_constraint_offset): Likewise.
9416 * optabs.c (prepare_libcall_arg): Likewise.
9417 * postreload.c (move2add_note_store): Likewise.
9418 * reload.c (operands_match_p): Likewise.
9419 * rtl.h (load_extend_op): Likewise.
9420 * rtlhooks.c (gen_lowpart_general): Likewise.
9421 * simplify-rtx.c (simplify_truncation): Likewise.
9422 (simplify_unary_operation_1): Likewise.
9423 (simplify_binary_operation_1): Likewise.
9424 (simplify_const_binary_operation): Likewise.
9425 (simplify_const_relational_operation): Likewise.
9426 (simplify_subreg): Likewise.
9427 * stor-layout.c (bitwise_mode_for_mode): Likewise.
9428 * var-tracking.c (adjust_mems): Likewise.
9429 (prepare_call_arguments): Likewise.
9431 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9432 Alan Hayward <alan.hayward@arm.com>
9433 David Sherwood <david.sherwood@arm.com>
9435 * machmode.h (is_int_mode): New fuction.
9436 * combine.c (find_split_point): Use it.
9437 (combine_simplify_rtx): Likewise.
9438 (simplify_if_then_else): Likewise.
9439 (simplify_set): Likewise.
9440 (simplify_shift_const_1): Likewise.
9441 (simplify_comparison): Likewise.
9442 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
9443 * cse.c (notreg_cost): Likewise.
9444 (cse_insn): Likewise.
9445 * cselib.c (cselib_lookup_1): Likewise.
9446 * dojump.c (do_jump_1): Likewise.
9447 (do_compare_rtx_and_jump): Likewise.
9448 * dse.c (get_call_args): Likewise.
9449 * dwarf2out.c (rtl_for_decl_init): Likewise.
9450 (native_encode_initializer): Likewise.
9451 * expmed.c (emit_store_flag_1): Likewise.
9452 (emit_store_flag): Likewise.
9453 * expr.c (convert_modes): Likewise.
9454 (store_field): Likewise.
9455 (expand_expr_real_1): Likewise.
9456 * fold-const.c (fold_read_from_constant_string): Likewise.
9457 * gimple-ssa-sprintf.c (get_format_string): Likewise.
9458 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
9459 * optabs.c (expand_binop): Likewise.
9460 (expand_unop): Likewise.
9461 (expand_abs_nojump): Likewise.
9462 (expand_one_cmpl_abs_nojump): Likewise.
9463 * simplify-rtx.c (mode_signbit_p): Likewise.
9464 (val_signbit_p): Likewise.
9465 (val_signbit_known_set_p): Likewise.
9466 (val_signbit_known_clear_p): Likewise.
9467 (simplify_relational_operation_1): Likewise.
9468 * tree.c (vector_type_mode): Likewise.
9470 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9471 Alan Hayward <alan.hayward@arm.com>
9472 David Sherwood <david.sherwood@arm.com>
9474 * machmode.h (smallest_mode_for_size): Fix formatting.
9475 (smallest_int_mode_for_size): New function.
9476 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
9477 instead of smallest_mode_for_size.
9478 * combine.c (make_extraction): Likewise.
9479 * config/arc/arc.c (arc_expand_movmem): Likewise.
9480 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
9481 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
9482 * config/s390/s390.c (s390_expand_insv): Likewise.
9483 * config/sparc/sparc.c (assign_int_registers): Likewise.
9484 * config/spu/spu.c (spu_function_value): Likewise.
9485 (spu_function_arg): Likewise.
9486 * coverage.c (get_gcov_type): Likewise.
9487 (get_gcov_unsigned_t): Likewise.
9488 * dse.c (find_shift_sequence): Likewise.
9489 * expmed.c (store_bit_field_1): Likewise.
9490 * expr.c (convert_move): Likewise.
9491 (store_field): Likewise.
9492 * internal-fn.c (expand_arith_overflow): Likewise.
9493 * optabs-query.c (get_best_extraction_insn): Likewise.
9494 * optabs.c (expand_twoval_binop_libfunc): Likewise.
9495 * stor-layout.c (layout_type): Likewise.
9496 (initialize_sizetypes): Likewise.
9497 * targhooks.c (default_get_mask_mode): Likewise.
9498 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
9500 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9501 Alan Hayward <alan.hayward@arm.com>
9502 David Sherwood <david.sherwood@arm.com>
9504 * machmode.h (opt_mode::else_blk): New function.
9505 (int_mode_for_mode): Declare.
9506 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
9507 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
9509 * cfgexpand.c (expand_debug_expr): Likewise.
9510 * combine.c (gen_lowpart_or_truncate): Likewise.
9511 (gen_lowpart_for_combine): Likewise.
9512 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
9513 * config/avr/avr.c (avr_to_int_mode): Likewise.
9514 (avr_out_plus_1): Likewise.
9515 (avr_out_plus): Likewise.
9516 (avr_out_round): Likewise.
9517 * config/i386/i386.c (ix86_split_to_parts): Likewise.
9518 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
9519 (s390_expand_vcond): Likewise.
9520 * config/spu/spu.c (spu_split_immediate): Likewise.
9521 (spu_expand_mov): Likewise.
9522 * dse.c (get_stored_val): Likewise.
9523 * expmed.c (store_bit_field_1): Likewise.
9524 (convert_extracted_bit_field): Use int_mode_for_mode instead of
9526 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
9527 (extract_low_bits): Likewise.
9528 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
9529 handling rather than repeating the check.
9530 (emit_group_store): Likewise.
9531 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
9532 * optabs.c (expand_absneg_bit): Likewise.
9533 (expand_copysign_absneg): Likewise.
9534 (expand_copysign_bit): Likewise.
9535 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
9536 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
9537 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
9538 * var-tracking.c (prepare_call_arguments): Likewise.
9539 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
9540 int_mode_for_mode instead of mode_for_size.
9541 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
9543 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9544 Alan Hayward <alan.hayward@arm.com>
9545 David Sherwood <david.sherwood@arm.com>
9547 * machmode.h (int_mode_for_size): New function.
9548 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
9549 instead of mode_for_size.
9550 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
9552 * combine.c (expand_field_assignment): Use int_mode_for_size
9553 instead of mode_for_size.
9554 (make_extraction): Likewise.
9555 (simplify_shift_const_1): Likewise.
9556 (simplify_comparison): Likewise.
9557 * dojump.c (do_jump): Likewise.
9558 * dwarf2out.c (mem_loc_descriptor): Likewise.
9559 * emit-rtl.c (init_derived_machine_modes): Likewise.
9560 * expmed.c (flip_storage_order): Likewise.
9561 (convert_extracted_bit_field): Likewise.
9562 * expr.c (copy_blkmode_from_reg): Likewise.
9563 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
9564 * internal-fn.c (expand_mul_overflow): Likewise.
9565 * lower-subreg.c (simple_move): Likewise.
9566 * optabs-libfuncs.c (init_optabs): Likewise.
9567 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9568 * tree.c (vector_type_mode): Likewise.
9569 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
9570 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
9571 * tree-vect-generic.c (expand_vector_parallel): Likewise.
9572 * tree-vect-stmts.c (vectorizable_load): Likewise.
9573 (vectorizable_store): Likewise.
9575 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9576 Alan Hayward <alan.hayward@arm.com>
9577 David Sherwood <david.sherwood@arm.com>
9579 * coretypes.h (pod_mode): New type.
9580 (scalar_int_mode_pod): New typedef.
9581 * machmode.h (pod_mode): New class.
9582 (int_n_data_t::m): Change type to scalar_int_mode_pod.
9583 * genmodes.c (emit_mode_int_n): Update accordingly.
9584 * lower-subreg.h (target_lower_subreg): Change type to
9585 scalar_int_mode_pod.
9586 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
9587 scalar_int_mode_pod.
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 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
9594 machine_mode to scalar_int_mode.
9595 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
9596 (rs6000_option_override_internal): Remove cast to int.
9597 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
9598 machine_mode to scalar_int_mode.
9599 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
9600 (rs6000_option_override_internal): Remove cast to int.
9601 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
9602 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
9604 * config/s390/s390.c (s390_expand_builtin): Likewise.
9605 * coretypes.h (scalar_int_mode): New type.
9606 (opt_scalar_int_mode): New typedef.
9607 * machmode.h (scalar_int_mode): New class.
9608 (scalar_int_mode::includes_p): New function.
9609 (byte_mode): Change type to scalar_int_mode.
9610 (word_mode): Likewise.
9611 (ptr_mode): Likewise.
9612 * emit-rtl.c (byte_mode): Likewise.
9613 (word_mode): Likewise.
9614 (ptr_mode): Likewise.
9615 (init_derived_machine_modes): Update accordingly.
9616 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
9617 and MODE_PARTIAL_INT.
9618 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
9619 opt_scalar_int_mode.
9621 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9622 Alan Hayward <alan.hayward@arm.com>
9623 David Sherwood <david.sherwood@arm.com>
9625 * target.def (libgcc_floating_mode_supported_p): Take a
9627 * doc/tm.texi: Regenerate.
9628 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
9630 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
9631 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
9634 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9635 Alan Hayward <alan.hayward@arm.com>
9636 David Sherwood <david.sherwood@arm.com>
9638 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
9639 * doc/tm.texi: Regenerate.
9640 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
9641 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
9642 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
9643 * targhooks.h (default_floatn_mode): Likewise.
9644 * targhooks.c (default_floatn_mode): Likewise.
9645 * tree.c (build_common_tree_nodes): Update accordingly.
9647 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9648 Alan Hayward <alan.hayward@arm.com>
9649 David Sherwood <david.sherwood@arm.com>
9651 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
9652 (mode_iterator::iterate_p): Likewise.
9653 (mode_iterator::get_wider): Likewise.
9654 * expr.c (init_expr_target): Use opt_scalar_float_mode.
9656 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9657 Alan Hayward <alan.hayward@arm.com>
9658 David Sherwood <david.sherwood@arm.com>
9660 * coretypes.h (opt_scalar_float_mode): New typedef.
9661 * machmode.h (float_mode_for_size): New function.
9662 * emit-rtl.c (double_mode): Delete.
9663 (init_emit_once): Use float_mode_for_size.
9664 * stor-layout.c (layout_type): Likewise.
9665 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
9667 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9668 Alan Hayward <alan.hayward@arm.com>
9669 David Sherwood <david.sherwood@arm.com>
9671 * output.h (assemble_real): Take a scalar_float_mode.
9672 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
9673 * config/arm/arm.md (consttable_4): Likewise.
9674 (consttable_8): Likewise.
9675 (consttable_16): Likewise.
9676 * config/mips/mips.md (consttable_float): Likewise.
9677 * config/s390/s390.c (s390_output_pool_entry): Likewise.
9678 * varasm.c (assemble_real): Take a scalar_float_mode.
9679 (output_constant_pool_2): Update accordingly.
9680 (output_constant): Likewise.
9682 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9683 Alan Hayward <alan.hayward@arm.com>
9684 David Sherwood <david.sherwood@arm.com>
9686 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
9687 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
9688 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
9689 (native_encode_real): Likewise.
9690 (native_interpret_real): Likewise.
9691 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
9692 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9694 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9695 Alan Hayward <alan.hayward@arm.com>
9696 David Sherwood <david.sherwood@arm.com>
9698 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
9699 <scalar_float_mode>. Simplify.
9700 (gen_extend_conv_libfunc): Likewise.
9702 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9703 Alan Hayward <alan.hayward@arm.com>
9704 David Sherwood <david.sherwood@arm.com>
9706 * coretypes.h (scalar_float_mode): New type.
9707 * machmode.h (mode_traits::from_int): Use machine_mode if
9708 USE_ENUM_MODES is defined.
9709 (is_a): New function.
9711 (dyn_cast): Likewise.
9712 (scalar_float_mode): New class.
9713 (scalar_float_mode::includes_p): New function.
9714 (is_float_mode): Likewise.
9715 * gdbhooks.py (MachineModePrinter): New class.
9716 (build_pretty_printer): Use it for scalar_float_mode.
9717 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
9718 (format_helper::format_helper): Turn into a template.
9719 * genmodes.c (get_mode_class): New function.
9720 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
9721 or machine_mode if none.
9722 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
9723 as_a <scalar_float_mode>.
9724 * dwarf2out.c (mem_loc_descriptor): Likewise.
9725 (insert_float): Likewise.
9726 (add_const_value_attribute): Likewise.
9727 * simplify-rtx.c (simplify_immed_subreg): Likewise.
9728 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
9729 (expand_unop): Update accordingly.
9730 (expand_abs_nojump): Likewise.
9731 (expand_copysign_absneg): Take a scalar_float_mode.
9732 (expand_copysign_bit): Likewise.
9733 (expand_copysign): Update accordingly.
9735 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9736 Alan Hayward <alan.hayward@arm.com>
9737 David Sherwood <david.sherwood@arm.com>
9739 * coretypes.h (opt_mode): New class.
9740 * machmode.h (opt_mode): Likewise.
9741 (opt_mode::else_void): New function.
9742 (opt_mode::require): Likewise.
9743 (opt_mode::exists): Likewise.
9744 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
9745 (GET_MODE_2XWIDER_MODE): Likewise.
9746 (mode_iterator::get_wider): Update accordingly.
9747 (mode_iterator::get_2xwider): Likewise.
9748 (mode_iterator::get_known_wider): Likewise, turning into a template.
9749 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
9750 forcing a wider mode to exist.
9751 * config/cr16/cr16.h (LONG_REG_P): Likewise.
9752 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
9753 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
9754 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
9755 * lower-subreg.c (init_lower_subreg): Likewise.
9756 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
9757 on the final iteration.
9758 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
9759 a wider mode exists before asking for a move pattern.
9760 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
9761 forcing a wider mode to exist.
9762 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
9763 returning false if no such mode exists.
9764 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
9765 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
9766 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
9767 Avoid checking for a MODE_INT if we already know the mode is not a
9769 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
9770 forcing a wider mode to exist.
9771 (expmed_mult_highpart_optab): Likewise.
9772 (expmed_mult_highpart): Likewise.
9773 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
9775 * lto-streamer-in.c (lto_input_mode_table): Likewise.
9776 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
9777 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9778 * internal-fn.c (expand_mul_overflow): Update use of
9779 GET_MODE_2XWIDER_MODE.
9780 * omp-low.c (omp_clause_aligned_alignment): Likewise.
9781 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
9782 GET_MODE_WIDER_MODE.
9783 (convert_plusminus_to_widen): Likewise.
9784 * tree-switch-conversion.c (array_value_type): Likewise.
9785 * var-tracking.c (emit_note_insn_var_location): Likewise.
9786 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9787 Return false inside rather than outside the loop if no wider mode
9789 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
9790 and GET_MODE_2XWIDER_MODE
9791 (can_compare_p): Use else_void.
9792 * gdbhooks.py (OptMachineModePrinter): New class.
9793 (build_pretty_printer): Use it for opt_mode.
9795 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9796 Alan Hayward <alan.hayward@arm.com>
9797 David Sherwood <david.sherwood@arm.com>
9799 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
9800 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
9802 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9803 Alan Hayward <alan.hayward@arm.com>
9804 David Sherwood <david.sherwood@arm.com>
9806 * machmode.h (mode_traits): New structure.
9807 (get_narrowest_mode): New function.
9808 (mode_iterator::start): Likewise.
9809 (mode_iterator::iterate_p): Likewise.
9810 (mode_iterator::get_wider): Likewise.
9811 (mode_iterator::get_known_wider): Likewise.
9812 (mode_iterator::get_2xwider): Likewise.
9813 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
9814 (FOR_EACH_MODE): Likewise.
9815 (FOR_EACH_MODE_FROM): Likewise.
9816 (FOR_EACH_MODE_UNTIL): Likewise.
9817 (FOR_EACH_WIDER_MODE): Likewise.
9818 (FOR_EACH_2XWIDER_MODE): Likewise.
9819 * builtins.c (expand_builtin_strlen): Use new mode iterators.
9820 * combine.c (simplify_comparison): Likewise
9821 * config/i386/i386.c (type_natural_mode): Likewise.
9822 * cse.c (cse_insn): Likewise.
9823 * dse.c (find_shift_sequence): Likewise.
9824 * emit-rtl.c (init_derived_machine_modes): Likewise.
9825 (init_emit_once): Likewise.
9826 * explow.c (hard_function_value): Likewise.
9827 * expmed.c (extract_fixed_bit_field_1): Likewise.
9828 (extract_bit_field_1): Likewise.
9829 (expand_divmod): Likewise.
9830 (emit_store_flag_1): Likewise.
9831 * expr.c (init_expr_target): Likewise.
9832 (convert_move): Likewise.
9833 (alignment_for_piecewise_move): Likewise.
9834 (widest_int_mode_for_size): Likewise.
9835 (emit_block_move_via_movmem): Likewise.
9836 (copy_blkmode_to_reg): Likewise.
9837 (set_storage_via_setmem): Likewise.
9838 (compress_float_constant): Likewise.
9839 * omp-low.c (omp_clause_aligned_alignment): Likewise.
9840 * optabs-query.c (get_best_extraction_insn): Likewise.
9841 * optabs.c (expand_binop): Likewise.
9842 (expand_twoval_unop): Likewise.
9843 (expand_twoval_binop): Likewise.
9844 (widen_leading): Likewise.
9845 (widen_bswap): Likewise.
9846 (expand_parity): Likewise.
9847 (expand_unop): Likewise.
9848 (prepare_cmp_insn): Likewise.
9849 (prepare_float_lib_cmp): Likewise.
9850 (expand_float): Likewise.
9851 (expand_fix): Likewise.
9852 (expand_sfix_optab): Likewise.
9853 * postreload.c (move2add_use_add2_insn): Likewise.
9854 * reg-stack.c (reg_to_stack): Likewise.
9855 * reginfo.c (choose_hard_reg_mode): Likewise.
9856 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
9857 * stor-layout.c (mode_for_size): Likewise.
9858 (smallest_mode_for_size): Likewise.
9859 (mode_for_vector): Likewise.
9860 (finish_bitfield_representative): Likewise.
9861 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
9862 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
9863 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9864 * var-tracking.c (prepare_call_arguments): Likewise.
9866 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9867 Alan Hayward <alan.hayward@arm.com>
9868 David Sherwood <david.sherwood@arm.com>
9870 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
9871 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
9872 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
9873 * machmode.h (mode_size): Move earlier in file.
9874 (mode_precision): Likewise.
9875 (mode_inner): Likewise.
9876 (mode_nunits): Likewise.
9877 (mode_unit_size): Likewise.
9878 (unit_unit_precision): Likewise.
9879 (mode_wider): Likewise.
9880 (mode_2xwider): Likewise.
9881 (machine_mode): New class.
9882 (mode_to_bytes): New function.
9883 (mode_to_bits): Likewise.
9884 (mode_to_precision): Likewise.
9885 (mode_to_inner): Likewise.
9886 (mode_to_unit_size): Likewise.
9887 (mode_to_unit_precision): Likewise.
9888 (mode_to_nunits): Likewise.
9889 (GET_MODE_SIZE): Use mode_to_bytes.
9890 (GET_MODE_BITSIZE): Use mode_to_bits.
9891 (GET_MODE_PRECISION): Use mode_to_precision.
9892 (GET_MODE_INNER): Use mode_to_inner.
9893 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
9894 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
9895 (GET_MODE_NUNITS): Use mode_to_nunits.
9896 * system.h (ALWAYS_INLINE): New macro.
9897 * config/powerpcspe/powerpcspe-c.c
9898 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
9899 int for arg1_mode and arg2_mode.
9901 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9902 Alan Hayward <alan.hayward@arm.com>
9903 David Sherwood <david.sherwood@arm.com>
9905 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
9906 Prefix mode names with E_ in case statements.
9907 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9908 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
9909 (aarch64_split_simd_move): Likewise.
9910 (aarch64_gen_storewb_pair): Likewise.
9911 (aarch64_gen_loadwb_pair): Likewise.
9912 (aarch64_gen_store_pair): Likewise.
9913 (aarch64_gen_load_pair): Likewise.
9914 (aarch64_get_condition_code_1): Likewise.
9915 (aarch64_constant_pool_reload_icode): Likewise.
9916 (get_rsqrte_type): Likewise.
9917 (get_rsqrts_type): Likewise.
9918 (get_recpe_type): Likewise.
9919 (get_recps_type): Likewise.
9920 (aarch64_gimplify_va_arg_expr): Likewise.
9921 (aarch64_simd_container_mode): Likewise.
9922 (aarch64_emit_load_exclusive): Likewise.
9923 (aarch64_emit_store_exclusive): Likewise.
9924 (aarch64_expand_compare_and_swap): Likewise.
9925 (aarch64_gen_atomic_cas): Likewise.
9926 (aarch64_emit_bic): Likewise.
9927 (aarch64_emit_atomic_swap): Likewise.
9928 (aarch64_emit_atomic_load_op): Likewise.
9929 (aarch64_evpc_trn): Likewise.
9930 (aarch64_evpc_uzp): Likewise.
9931 (aarch64_evpc_zip): Likewise.
9932 (aarch64_evpc_ext): Likewise.
9933 (aarch64_evpc_rev): Likewise.
9934 (aarch64_evpc_dup): Likewise.
9935 (aarch64_gen_ccmp_first): Likewise.
9936 (aarch64_gen_ccmp_next): Likewise.
9937 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
9938 (alpha_emit_xfloating_libcall): Likewise.
9939 (emit_insxl): Likewise.
9940 (alpha_arg_type): Likewise.
9941 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
9942 (arc_preferred_simd_mode): Likewise.
9943 (arc_secondary_reload): Likewise.
9944 (get_arc_condition_code): Likewise.
9945 (arc_print_operand): Likewise.
9946 (arc_legitimate_constant_p): Likewise.
9947 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9948 * config/arc/arc.md (casesi_load): Likewise.
9949 (casesi_compact_jump): Likewise.
9950 * config/arc/predicates.md (proper_comparison_operator): Likewise.
9951 (cc_use_register): Likewise.
9952 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9953 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
9954 (arm_init_iwmmxt_builtins): Likewise.
9955 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
9956 (neon_expand_vector_init): Likewise.
9957 (arm_attr_length_move_neon): Likewise.
9958 (maybe_get_arm_condition_code): Likewise.
9959 (arm_emit_vector_const): Likewise.
9960 (arm_preferred_simd_mode): Likewise.
9961 (arm_output_iwmmxt_tinsr): Likewise.
9962 (thumb1_output_casesi): Likewise.
9963 (thumb2_output_casesi): Likewise.
9964 (arm_emit_load_exclusive): Likewise.
9965 (arm_emit_store_exclusive): Likewise.
9966 (arm_expand_compare_and_swap): Likewise.
9967 (arm_evpc_neon_vuzp): Likewise.
9968 (arm_evpc_neon_vzip): Likewise.
9969 (arm_evpc_neon_vrev): Likewise.
9970 (arm_evpc_neon_vtrn): Likewise.
9971 (arm_evpc_neon_vext): Likewise.
9972 (arm_validize_comparison): Likewise.
9973 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
9974 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
9975 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
9976 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
9977 (c6x_preferred_simd_mode): Likewise.
9978 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
9979 (epiphany_rtx_costs): Likewise.
9980 * config/epiphany/predicates.md (proper_comparison_operator):
9982 * config/frv/frv.c (condexec_memory_operand): Likewise.
9983 (frv_emit_move): Likewise.
9984 (output_move_single): Likewise.
9985 (output_condmove_single): Likewise.
9986 (frv_hard_regno_mode_ok): Likewise.
9987 (frv_matching_accg_mode): Likewise.
9988 * config/h8300/h8300.c (split_adds_subs): Likewise.
9989 (h8300_rtx_costs): Likewise.
9990 (h8300_print_operand): Likewise.
9991 (compute_mov_length): Likewise.
9992 (output_logical_op): Likewise.
9993 (compute_logical_op_length): Likewise.
9994 (compute_logical_op_cc): Likewise.
9995 (h8300_shift_needs_scratch_p): Likewise.
9996 (output_a_shift): Likewise.
9997 (compute_a_shift_length): Likewise.
9998 (compute_a_shift_cc): Likewise.
9999 (expand_a_rotate): Likewise.
10000 (output_a_rotate): Likewise.
10001 * config/i386/i386.c (classify_argument): Likewise.
10002 (function_arg_advance_32): Likewise.
10003 (function_arg_32): Likewise.
10004 (function_arg_64): Likewise.
10005 (function_value_64): Likewise.
10006 (ix86_gimplify_va_arg): Likewise.
10007 (ix86_legitimate_constant_p): Likewise.
10008 (put_condition_code): Likewise.
10009 (split_double_mode): Likewise.
10010 (ix86_avx256_split_vector_move_misalign): Likewise.
10011 (ix86_expand_vector_logical_operator): Likewise.
10012 (ix86_split_idivmod): Likewise.
10013 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
10014 (ix86_build_const_vector): Likewise.
10015 (ix86_build_signbit_mask): Likewise.
10016 (ix86_match_ccmode): Likewise.
10017 (ix86_cc_modes_compatible): Likewise.
10018 (ix86_expand_branch): Likewise.
10019 (ix86_expand_sse_cmp): Likewise.
10020 (ix86_expand_sse_movcc): Likewise.
10021 (ix86_expand_int_sse_cmp): Likewise.
10022 (ix86_expand_vec_perm_vpermi2): Likewise.
10023 (ix86_expand_vec_perm): Likewise.
10024 (ix86_expand_sse_unpack): Likewise.
10025 (ix86_expand_int_addcc): Likewise.
10026 (ix86_split_to_parts): Likewise.
10027 (ix86_vectorize_builtin_gather): Likewise.
10028 (ix86_vectorize_builtin_scatter): Likewise.
10029 (avx_vpermilp_parallel): Likewise.
10030 (inline_memory_move_cost): Likewise.
10031 (ix86_tieable_integer_mode_p): Likewise.
10032 (x86_maybe_negate_const_int): Likewise.
10033 (ix86_expand_vector_init_duplicate): Likewise.
10034 (ix86_expand_vector_init_one_nonzero): Likewise.
10035 (ix86_expand_vector_init_one_var): Likewise.
10036 (ix86_expand_vector_init_concat): Likewise.
10037 (ix86_expand_vector_init_interleave): Likewise.
10038 (ix86_expand_vector_init_general): Likewise.
10039 (ix86_expand_vector_set): Likewise.
10040 (ix86_expand_vector_extract): Likewise.
10041 (emit_reduc_half): Likewise.
10042 (ix86_emit_i387_round): Likewise.
10043 (ix86_mangle_type): Likewise.
10044 (ix86_expand_round_sse4): Likewise.
10045 (expand_vec_perm_blend): Likewise.
10046 (canonicalize_vector_int_perm): Likewise.
10047 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
10048 (expand_vec_perm_1): Likewise.
10049 (expand_vec_perm_interleave3): Likewise.
10050 (expand_vec_perm_even_odd_pack): Likewise.
10051 (expand_vec_perm_even_odd_1): Likewise.
10052 (expand_vec_perm_broadcast_1): Likewise.
10053 (ix86_vectorize_vec_perm_const_ok): Likewise.
10054 (ix86_expand_vecop_qihi): Likewise.
10055 (ix86_expand_mul_widen_hilo): Likewise.
10056 (ix86_expand_sse2_abs): Likewise.
10057 (ix86_expand_pextr): Likewise.
10058 (ix86_expand_pinsr): Likewise.
10059 (ix86_preferred_simd_mode): Likewise.
10060 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
10061 * config/i386/sse.md (*andnot<mode>3): Likewise.
10062 (<mask_codefor><code><mode>3<mask_name>): Likewise.
10063 (*<code><mode>3): Likewise.
10064 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
10065 (ia64_expand_atomic_op): Likewise.
10066 (ia64_arg_type): Likewise.
10067 (ia64_mode_to_int): Likewise.
10068 (ia64_scalar_mode_supported_p): Likewise.
10069 (ia64_vector_mode_supported_p): Likewise.
10070 (expand_vec_perm_broadcast): Likewise.
10071 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
10072 (iq2000_function_arg_advance): Likewise.
10073 (iq2000_function_arg): Likewise.
10074 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
10075 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
10076 (m68k_libcall_value): Likewise.
10077 (m68k_function_value): Likewise.
10078 (sched_attr_op_type): Likewise.
10079 * config/mcore/mcore.c (mcore_output_move): Likewise.
10080 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
10082 (microblaze_function_arg): Likewise.
10083 * config/mips/mips.c (mips16_build_call_stub): Likewise.
10084 (mips_print_operand): Likewise.
10085 (mips_mode_ok_for_mov_fmt_p): Likewise.
10086 (mips_vector_mode_supported_p): Likewise.
10087 (mips_preferred_simd_mode): Likewise.
10088 (mips_expand_vpc_loongson_even_odd): Likewise.
10089 (mips_expand_vec_unpack): Likewise.
10090 (mips_expand_vi_broadcast): Likewise.
10091 (mips_expand_vector_init): Likewise.
10092 (mips_expand_vec_reduc): Likewise.
10093 (mips_expand_msa_cmp): Likewise.
10094 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
10095 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
10096 (cc_flags_for_mode): Likewise.
10097 * config/msp430/msp430.c (msp430_print_operand): Likewise.
10098 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
10099 (nds32_output_casesi_pc_relative): Likewise.
10100 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10101 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
10102 (nvptx_gen_unpack): Likewise.
10103 (nvptx_gen_pack): Likewise.
10104 (nvptx_gen_shuffle): Likewise.
10105 (nvptx_gen_wcast): Likewise.
10106 (nvptx_preferred_simd_mode): Likewise.
10107 * config/pa/pa.c (pa_secondary_reload): Likewise.
10108 * config/pa/predicates.md (base14_operand): Likewise.
10109 * config/powerpcspe/powerpcspe-c.c
10110 (altivec_resolve_overloaded_builtin): Likewise.
10111 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
10113 (rs6000_preferred_simd_mode): Likewise.
10114 (output_vec_const_move): Likewise.
10115 (rs6000_expand_vector_extract): Likewise.
10116 (rs6000_split_vec_extract_var): Likewise.
10117 (reg_offset_addressing_ok_p): Likewise.
10118 (rs6000_legitimate_offset_address_p): Likewise.
10119 (rs6000_legitimize_address): Likewise.
10120 (rs6000_emit_set_const): Likewise.
10121 (rs6000_const_vec): Likewise.
10122 (rs6000_emit_move): Likewise.
10123 (spe_build_register_parallel): Likewise.
10124 (rs6000_darwin64_record_arg_recurse): Likewise.
10125 (swap_selector_for_mode): Likewise.
10126 (spe_init_builtins): Likewise.
10127 (paired_init_builtins): Likewise.
10128 (altivec_init_builtins): Likewise.
10129 (do_load_for_compare): Likewise.
10130 (rs6000_generate_compare): Likewise.
10131 (rs6000_expand_float128_convert): Likewise.
10132 (emit_load_locked): Likewise.
10133 (emit_store_conditional): Likewise.
10134 (rs6000_output_function_epilogue): Likewise.
10135 (rs6000_handle_altivec_attribute): Likewise.
10136 (rs6000_function_value): Likewise.
10137 (emit_fusion_gpr_load): Likewise.
10138 (emit_fusion_p9_load): Likewise.
10139 (emit_fusion_p9_store): Likewise.
10140 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
10141 (fusion_gpr_mem_load): Likewise.
10142 (fusion_addis_mem_combo_load): Likewise.
10143 (fusion_addis_mem_combo_store): Likewise.
10144 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
10145 (fusion_gpr_mem_load): Likewise.
10146 (fusion_addis_mem_combo_load): Likewise.
10147 (fusion_addis_mem_combo_store): Likewise.
10148 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10150 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
10151 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
10152 (rs6000_preferred_simd_mode): Likewise.
10153 (output_vec_const_move): Likewise.
10154 (rs6000_expand_vector_extract): Likewise.
10155 (rs6000_split_vec_extract_var): Likewise.
10156 (reg_offset_addressing_ok_p): Likewise.
10157 (rs6000_legitimate_offset_address_p): Likewise.
10158 (rs6000_legitimize_address): Likewise.
10159 (rs6000_emit_set_const): Likewise.
10160 (rs6000_const_vec): Likewise.
10161 (rs6000_emit_move): Likewise.
10162 (rs6000_darwin64_record_arg_recurse): Likewise.
10163 (swap_selector_for_mode): Likewise.
10164 (paired_init_builtins): Likewise.
10165 (altivec_init_builtins): Likewise.
10166 (rs6000_expand_float128_convert): Likewise.
10167 (emit_load_locked): Likewise.
10168 (emit_store_conditional): Likewise.
10169 (rs6000_output_function_epilogue): Likewise.
10170 (rs6000_handle_altivec_attribute): Likewise.
10171 (rs6000_function_value): Likewise.
10172 (emit_fusion_gpr_load): Likewise.
10173 (emit_fusion_p9_load): Likewise.
10174 (emit_fusion_p9_store): Likewise.
10175 * config/rx/rx.c (rx_gen_move_template): Likewise.
10176 (flags_from_mode): Likewise.
10177 * config/s390/predicates.md (s390_alc_comparison): Likewise.
10178 (s390_slb_comparison): Likewise.
10179 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
10180 (s390_vector_mode_supported_p): Likewise.
10181 (s390_cc_modes_compatible): Likewise.
10182 (s390_match_ccmode_set): Likewise.
10183 (s390_canonicalize_comparison): Likewise.
10184 (s390_emit_compare_and_swap): Likewise.
10185 (s390_branch_condition_mask): Likewise.
10186 (s390_rtx_costs): Likewise.
10187 (s390_secondary_reload): Likewise.
10188 (__SECONDARY_RELOAD_CASE): Likewise.
10189 (s390_expand_cs): Likewise.
10190 (s390_preferred_simd_mode): Likewise.
10191 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
10192 * config/sh/sh.c (sh_print_operand): Likewise.
10193 (dump_table): Likewise.
10194 (sh_secondary_reload): Likewise.
10195 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10196 * config/sh/sh.md (casesi_worker_1): Likewise.
10197 (casesi_worker_2): Likewise.
10198 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
10199 (fcc_comparison_operator): Likewise.
10200 * config/sparc/sparc.c (sparc_expand_move): Likewise.
10201 (emit_soft_tfmode_cvt): Likewise.
10202 (sparc_preferred_simd_mode): Likewise.
10203 (output_cbranch): Likewise.
10204 (sparc_print_operand): Likewise.
10205 (sparc_expand_vec_perm_bmask): Likewise.
10206 (vector_init_bshuffle): Likewise.
10207 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
10208 (spu_vector_mode_supported_p): Likewise.
10209 (spu_expand_insv): Likewise.
10210 (spu_emit_branch_or_set): Likewise.
10211 (spu_handle_vector_attribute): Likewise.
10212 (spu_builtin_splats): Likewise.
10213 (spu_builtin_extract): Likewise.
10214 (spu_builtin_promote): Likewise.
10215 (spu_expand_sign_extend): Likewise.
10216 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
10217 (tilegx_simd_int): Likewise.
10218 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
10219 (tilepro_simd_int): Likewise.
10220 * config/v850/v850.c (const_double_split): Likewise.
10221 (v850_print_operand): Likewise.
10222 (ep_memory_offset): Likewise.
10223 * config/vax/vax.c (vax_rtx_costs): Likewise.
10224 (vax_output_int_move): Likewise.
10225 (vax_output_int_add): Likewise.
10226 (vax_output_int_subtract): Likewise.
10227 * config/visium/predicates.md (visium_branch_operator): Likewise.
10228 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
10229 (visium_print_operand_address): Likewise.
10230 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10231 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
10232 (xtensa_expand_conditional_branch): Likewise.
10233 (xtensa_copy_incoming_a7): Likewise.
10234 (xtensa_output_literal): Likewise.
10235 * dfp.c (decimal_real_maxval): Likewise.
10236 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
10238 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10239 Alan Hayward <alan.hayward@arm.com>
10240 David Sherwood <david.sherwood@arm.com>
10242 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
10243 (mode_nunits_inline): Likewise.
10244 (mode_inner_inline): Likewise.
10245 (mode_unit_size_inline): Likewise.
10246 (mode_unit_precision_inline): Likewise.
10247 (emit_insn_modes_h): Likewise. Also emit a #define of the
10249 (emit_mode_wider): Add an E_ prefix to mode names.
10250 (emit_mode_complex): Likewise.
10251 (emit_mode_inner): Likewise.
10252 (emit_mode_adjustments): Likewise.
10253 (emit_mode_int_n): Likewise.
10254 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
10255 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
10256 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
10257 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
10258 (CRC32_BUILTIN, ENTRY): Likewise.
10259 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
10260 (aarch64_pop_regs): Likewise.
10261 (aarch64_process_components): Likewise.
10262 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
10263 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
10264 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
10265 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
10266 * config/arm/arm.c (arm_init_libfuncs): Likewise.
10267 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
10269 * config/i386/i386-builtin.def (pcmpestr): Likewise.
10270 (pcmpistr): Likewise.
10271 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
10272 * config/mmix/mmix.c (mmix_output_condition): Likewise.
10273 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
10275 * config/rl78/rl78.c (mduc_regs): Likewise.
10276 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
10277 (htm_expand_builtin): Likewise.
10278 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
10279 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
10280 * config/xtensa/xtensa.c (print_operand): Likewise.
10281 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
10282 (NUM_MODE_VECTOR_INT): Likewise.
10283 * genoutput.c (null_operand): Likewise.
10284 (output_operand_data): Likewise.
10285 * genrecog.c (print_parameter_value): Likewise.
10286 * lra.c (debug_operand_data): Likewise.
10288 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10290 * dwarf2out.c (scompare_loc_descriptor_wide)
10291 (scompare_loc_descriptor_narrow): New functions, split out from...
10292 (scompare_loc_descriptor): ...here.
10293 * expmed.c (emit_store_flag_int): New function, split out from...
10294 (emit_store_flag): ...here.
10296 2017-08-30 Richard Biener <rguenther@suse.de>
10298 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
10299 (dwarf2out_early_finish): Move setting of AT_pubnames from
10300 early debug output to early finish.
10302 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
10304 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
10305 and -mdata-region to the assembler.
10307 2017-08-30 Richard Biener <rguenther@suse.de>
10309 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
10311 (gen_subprogram_die): Add DW_AT_object_pointer only early.
10312 (dwarf2out_early_global_decl): Only generate a DIE for the
10313 abstract origin if it doesn't already exist or is a declaration DIE.
10314 (resolve_addr): Do not add the linkage name twice when
10315 generating a stub DIE for the DW_TAG_GNU_call_site target.
10317 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10319 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10320 Use machine_mode rather than int for arg1_mode.
10322 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
10325 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
10326 that the second argument of the built-in functions to unpack
10327 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
10328 switch statement instead a lot of if statements.
10329 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
10330 Allow 64-bit values to be in Altivec registers as well as
10331 traditional floating point registers.
10332 (pack<mode>, FMOVE128_VSX iterator): Likewise.
10334 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
10336 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
10337 MAX_REGS_PER_ADDRESS == 1.
10339 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
10341 * config/i386/i386.opt (flag_fentry): Do not init to -1.
10342 * config/i386/i386.c (ix86_option_override_internal): Simplify
10343 setting of opts->x_flag_entry.
10345 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10346 Jakub Jelinek <jakub@redhat.com>
10347 Richard Biener <rguenther@suse.de>
10349 PR tree-optimization/81503
10350 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
10351 folded constant fits in the target type; reorder tests for clarity.
10353 2017-08-29 Martin Liska <mliska@suse.cz>
10355 * passes.def: Include pass_lower_switch.
10356 * stmt.c (dump_case_nodes): Remove and move to
10357 tree-switch-conversion.
10358 (case_values_threshold): Likewise.
10359 (expand_switch_as_decision_tree_p): Likewise.
10360 (emit_case_decision_tree): Likewise.
10361 (expand_case): Likewise.
10362 (balance_case_nodes): Likewise.
10363 (node_has_low_bound): Likewise.
10364 (node_has_high_bound): Likewise.
10365 (node_is_bounded): Likewise.
10366 (emit_case_nodes): Likewise.
10367 (struct simple_case_node): New struct.
10368 (add_case_node): Remove.
10369 (emit_case_dispatch_table): Use vector instead of case_list.
10370 (reset_out_edges_aux): Remove.
10371 (compute_cases_per_edge): Likewise.
10372 (expand_case): Build list of simple_case_node.
10373 (expand_sjlj_dispatch_table): Use it.
10374 * tree-switch-conversion.c (struct case_node): Moved from
10375 stmt.c and adjusted.
10376 (emit_case_nodes): Likewise.
10377 (node_has_low_bound): Likewise.
10378 (node_has_high_bound): Likewise.
10379 (node_is_bounded): Likewise.
10380 (case_values_threshold): Likewise.
10381 (reset_out_edges_aux): Likewise.
10382 (compute_cases_per_edge): Likewise.
10383 (add_case_node): Likewise.
10384 (dump_case_nodes): Likewise.
10385 (balance_case_nodes): Likewise.
10386 (expand_switch_as_decision_tree_p): Likewise.
10387 (emit_jump): Likewise.
10388 (emit_case_decision_tree): Likewise.
10389 (try_switch_expansion): Likewise.
10390 (do_jump_if_equal): Likewise.
10391 (emit_cmp_and_jump_insns): Likewise.
10392 (fix_phi_operands_for_edge): New function.
10393 (record_phi_operand_mapping): Likewise.
10394 (class pass_lower_switch): New pass.
10395 (pass_lower_switch::execute): New function.
10396 (make_pass_lower_switch): Likewise.
10397 (conditional_probability):
10398 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
10399 * tree-pass.h: Add make_pass_lower_switch.
10401 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
10404 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
10407 2017-08-29 Richard Biener <rguenther@suse.de>
10409 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
10410 we do not add a DW_AT_inline attribute twice.
10411 (gen_subprogram_die): Remove code setting DW_AT_inline on
10412 DECL_ABSTRACT_P nodes.
10414 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
10416 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
10417 calls to internal functions.
10418 (gimplify_modify_expr): Likewise.
10419 * tree-call-cdce.c (use_internal_fn): Likewise.
10420 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
10421 (convert_to_divmod): Set the nothrow flag.
10422 * tree-if-conv.c (predicate_mem_writes): Likewise.
10423 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
10424 (vectorizable_call): Likewise.
10425 (vectorizable_store): Likewise.
10426 (vectorizable_load): Likewise.
10427 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
10428 (vect_recog_mask_conversion_pattern): Likewise.
10430 2017-08-29 Martin Liska <mliska@suse.cz>
10433 * gcc.c (driver_handle_option): Add new argument.
10434 * opts-common.c (handle_option): Pass
10435 target_option_override_hook.
10436 * opts-global.c (lang_handle_option): Add new option.
10437 (set_default_handlers): Add new argument.
10438 (decode_options): Likewise.
10439 * opts.c (target_handle_option): Likewise.
10440 (common_handle_option): Call target_option_override_hook.
10441 * opts.h (struct cl_option_handler_func): Add hook for
10442 target option override.
10443 (struct cl_option_handlers): Likewise.
10444 (set_default_handlers): Add new argument.
10445 (decode_options): Likewise.
10446 (common_handle_option): Likewise.
10447 (target_handle_option): Likewise.
10448 * toplev.c (toplev::main): Pass targetm.target_option.override
10451 2017-08-29 Richard Biener <rguenther@suse.de>
10452 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
10454 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
10455 life to the active subtree.
10457 2017-08-28 Jeff Law <law@redhat.com>
10459 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
10460 derive_equivalences.
10461 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
10462 Code moved into....
10463 (edge_info::derive_equivalences): New private member function
10465 * tree-ssa-dom.c (class edge_info): Changed from a struct
10466 to a class. Add ctor/dtor, methods and data members.
10467 (edge_info::edge_info): Renamed from allocate_edge_info.
10468 Initialize additional members.
10469 (edge_info::~edge_info): New.
10470 (free_dom_edge_info): Delete the edge info.
10471 (record_edge_info): Use new class & associated member functions.
10472 Tighten forms for testing for edge equivalences.
10473 (record_temporary_equivalences): Iterate over the simple
10474 equivalences rather than assuming there's only one per edge.
10475 (cprop_into_successor_phis): Iterate over the simple
10476 equivalences rather than assuming there's only one per edge.
10477 (optimize_stmt): Use operand_equal_p rather than pointer
10478 equality for mini-DSE code.
10480 2017-08-28 Nathan Sidwell <nathan@acm.org>
10482 * gcc.c (execute): Fold SIGPIPE handling into switch
10483 statement. Adjust internal error message.
10485 2017-08-28 Richard Biener <rguenther@suse.de>
10488 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
10489 Do nothing for removed DIEs.
10491 2017-08-28 Richard Biener <rguenther@suse.de>
10493 PR tree-optimization/81977
10494 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
10497 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
10500 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
10501 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
10502 using targetm.gen_mem_thread_fence.
10504 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
10507 * config/i386/i386.md (*<btsc><mode>): Change operand 2
10508 predicate to register_operand. Reorder operands.
10509 (*btr<mode>): Ditto.
10510 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
10511 (*btr<mode>_mask): Ditto.
10513 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
10515 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
10516 * config/rs6000/xmmintrin.h: New file.
10517 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
10519 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10522 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
10523 parameter and_insn and return it.
10524 (recombine_lvx_pattern): Insert a copy to ensure availability of
10525 the base register of the copied masking operation at the point of
10526 the instruction replacement.
10527 (recombine_stvx_pattern): Likewise.
10529 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
10531 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
10532 undocumented switches.
10533 (-mpower9-dform-vector): Likewise.
10534 (-mpower9-dform): Likewise.
10535 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
10536 comments to delete references to -mpower9-dform* switches.
10537 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
10538 Delete reference to -mpower9-dform* switches, test for
10539 -mpower9-vector instead.
10540 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
10541 (OTHER_P9_VECTOR_MASKS): Likewise.
10542 (POWERPC_MASKS): Likewise.
10543 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
10544 tests against -mpower9-dform* to -mpower9-vector. Delete code
10545 that checked for -mpower9-dform* consistancy with other options.
10546 Add test for -mpower9-misc to enable other power9 switches.
10547 (rs6000_init_hard_regno_mode_ok): Likewise.
10548 (rs6000_option_override_internal): Likewise.
10549 (rs6000_emit_prologue): Likewise.
10550 (rs6000_emit_epilogue): Likewise.
10551 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
10552 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
10553 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
10555 (emit_fusion_p9_store): Likewise.
10556 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
10557 resetting these macros if the assembler does not support ISA 3.0
10559 (TARGET_P9_DFORM_VECTOR): Likewise.
10560 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
10561 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
10563 2017-08-25 Alan Modra <amodra@gmail.com>
10565 PR rtl-optimization/81747
10566 * cse.c (cse_extended_basic_block): Don't attempt to record
10567 equivalences for degenerate conditional branches that branch
10568 to their fall-through.
10570 2017-08-24 Martin Sebor <msebor@redhat.com>
10572 PR middle-end/81908
10573 * gimple-fold.c (size_must_be_zero_p): New function.
10574 (gimple_fold_builtin_memory_op): Call it.
10576 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
10578 * config/rs6000/mm_malloc.h: New file.
10580 2017-08-24 Bin Cheng <bin.cheng@arm.com>
10582 PR tree-optimization/81913
10583 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
10584 analysis when either IVs in condition can wrap.
10586 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
10588 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
10589 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
10591 2017-08-24 Richard Biener <rguenther@suse.de>
10594 * targhooks.c (default_target_can_inline_p): Properly
10595 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
10596 is present and always compare.
10597 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
10598 infer -mfpmath=sse from TARGET_SSE_P.
10599 (ix86_can_inline_p): Properly use target_option_default_node when
10600 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
10602 2017-08-24 Richard Biener <rguenther@suse.de>
10605 * dwarf2out.c (output_die): Handle flag_generate_offload like
10607 (output_comp_unit): Likewise.
10608 (gen_array_type_die): Likewise.
10609 (dwarf2out_early_finish): Likewise.
10610 (note_variable_value_in_expr): Likewise.
10611 (dwarf2out_finish): Likewise. Adjust assert.
10612 * cgraphunit.c (symbol_table::compile): Move setting of
10613 flag_generate_offload earlier ...
10614 (symbol_table::finalize_compilation_unit): ... here, before
10615 early debug finalization.
10617 2017-08-24 Richard Biener <rguenther@suse.de>
10619 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
10620 and ipa-fnsummary.h.
10621 (ix86_can_inline_p): When ix86_fpmath flags do not match
10622 check whether the callee uses FP math at all.
10624 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
10626 PR middle-end/81931
10627 * tree-ssanames.c (get_nonzero_bits): Use element_precision
10628 instead of TYPE_PRECISION.
10630 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
10631 Alan Hayward <alan.hayward@arm.com>
10632 David Sherwood <david.sherwood@arm.com>
10634 * combine.c (make_extraction): Use subreg_offset_from_lsb.
10636 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
10638 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
10640 (ix86_frame::stack_realign_allocate): New field.
10641 (struct machine_frame_state): Modify comments.
10642 (machine_frame_state::sp_realigned_fp_end): New field.
10643 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
10644 layout calculation.
10645 (sp_valid_at): Add assertion to assure no attempt to access invalid
10646 offset of a realigned stack.
10647 (fp_valid_at): Likewise.
10648 (choose_baseaddr): Modify comments.
10649 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
10650 ix86_expand_prologue.
10651 (ix86_expand_prologue): Modify stack realignment and allocation.
10652 (ix86_expand_epilogue): Modify comments.
10653 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
10654 avx2_runtime, avx512f, and avx512f_runtime.
10656 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
10658 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
10659 (mstackrealign): Do not init to -1.
10660 * config/i386/i386.c (ix86_option_override_internal):
10661 Check opts_set, not opts when setting default value of
10662 opts->x_ix86_force_align_arg_pointer.
10664 2017-08-23 Richard Biener <rguenther@suse.de>
10666 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
10667 lang_hooks.decl_printable_name.
10668 * print-rtl-function.c (print_rtx_function): Likewise.
10669 * tree-pretty-print.c (dump_function_header): Likewise.
10671 2017-08-23 Richard Biener <rguenther@suse.de>
10674 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
10675 -g0 at compile-time.
10677 2017-08-23 Tamar Christina <tamar.christina@arm.com>
10679 PR middle-end/19706
10680 * doc/sourcebuild.texi (Other hardware attributes):
10683 2017-08-23 Tamar Christina <tamar.christina@arm.com>
10685 PR middle-end/19706
10686 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
10687 Fix single-use check.
10689 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
10691 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
10693 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
10695 * doc/install.texi: Modify to add more details on running selected
10698 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
10700 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
10701 is combined with -mabi=ms.
10702 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
10705 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10707 PR tree-optimization/81488
10708 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
10709 and cached_basis fields.
10710 (MAX_SPREAD): New constant.
10711 (alloc_cand_and_find_basis): Initialize new fields.
10712 (clear_visited): New function.
10713 (create_phi_basis_1): Rename from create_phi_basis, set visited
10714 and cached_basis fields.
10715 (create_phi_basis): New wrapper function.
10716 (phi_add_costs_1): Rename from phi_add_costs, add spread
10717 parameter, set visited field, short-circuit when limits reached.
10718 (phi_add_costs): New wrapper function.
10719 (record_phi_increments_1): Rename from record_phi_increments, set
10721 (record_phi_increments): New wrapper function.
10722 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
10723 (phi_incr_cost): New wrapper function.
10724 (all_phi_incrs_profitable_1): Rename from
10725 all_phi_incrs_profitable, set visited field.
10726 (all_phi_incrs_profitable): New wrapper function.
10728 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
10729 Alan Hayward <alan.hayward@arm.com>
10730 David Sherwood <david.sherwood@arm.com>
10732 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
10733 that takes the outer and inner modes.
10734 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
10735 comparison as the canonical test for a paradoxical subreg.
10736 * combine.c (simplify_set): Use paradoxical_subreg_p.
10737 (make_extraction): Likewise.
10738 (force_to_mode): Likewise.
10739 (rtx_equal_for_field_assignment_p): Likewise.
10740 (gen_lowpart_for_combine): Likewise.
10741 (simplify_comparison): Likewise.
10742 * cse.c (equiv_constant): Likewise.
10743 * expmed.c (store_bit_field_1): Likewise.
10744 * final.c (alter_subreg): Likewise.
10745 * fwprop.c (propagate_rtx): Likewise.
10746 (forward_propagate_subreg): Likewise.
10747 * ira-conflicts.c (ira_build_conflicts): Likewise.
10748 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
10749 * lra-constraints.c (curr_insn_transform): Likewise.
10750 (split_reg): Likewise.
10751 * lra-eliminations.c (move_plus_up): Likewise.
10752 (lra_eliminate_regs_1): Likewise.
10753 * recog.c (general_operand): Likewise.
10754 * ree.c (combine_reaching_defs): Likewise.
10755 * reload.c (push_reload): Likewise.
10756 (find_reloads): Likewise.
10757 * reload1.c (elimination_effects): Likewise.
10758 (compute_reload_subreg_offset): Likewise.
10759 (choose_reload_regs): Likewise.
10760 * rtlanal.c (subreg_lsb_1): Likewise.
10761 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10762 (simplify_subreg): Likewise.
10763 * var-tracking.c (track_loc_p): Likewise.
10764 * emit-rtl.c (byte_lowpart_offset): Likewise.
10765 (paradoxical_subreg_p): Delete out-of-line definition.
10767 2017-08-22 Jeff Law <law@redhat.com>
10769 PR tree-optimization/81741
10770 PR tree-optimization/71947
10771 * tree-ssa-dom.c: Include tree-inline.h.
10772 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
10773 equivalences if one is more expensive to compute than the other.
10774 * tree-ssa-scopedtables.h (class const_or_copies): Make
10775 record_const_or_copy_raw method private.
10776 (class avail_exprs_stack): New method simplify_binary_operation.
10777 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
10778 avail_exprs_stack::simplify_binary_operation as needed.
10779 (avail_exprs_stack::simplify_binary_operation): New function.
10781 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
10783 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
10784 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
10785 (DOT_SYMBOLS): Likewise.
10786 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
10787 (RELOCATABLE_NEEDS_FIXUP): Likewise.
10788 (RS6000_ABI_NAME): Likewise.
10789 (TARGET_CMODEL): Likewise.
10790 (TOC_SECTION_ASM_OP): Likewise.
10791 (SET_CMODEL): New macro.
10792 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
10794 2017-08-22 Richard Biener <rguenther@suse.de>
10796 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
10797 to allow for free-lang-data replacements similar to verify_type_variant.
10799 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
10800 Alan Hayward <alan.hayward@arm.com>
10801 David Sherwood <david.sherwood@arm.com>
10803 * config/aarch64/aarch64.md (casesi): Use DImode rather than
10804 VOIDmode for the LABEL_REF.
10806 2017-08-22 Richard Biener <rguenther@suse.de>
10808 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
10810 2017-08-22 Richard Biener <rguenther@suse.de>
10812 * common.opt (feliminate-dwarf2-dups): Ignore.
10813 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
10814 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
10815 same_die_p_wrap, compute_section_prefix,
10816 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
10817 (comdat_symbol_id, comdat_symbol_number): Likewise.
10818 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
10820 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
10821 (output_die): Mark unreachable path unreachable.
10822 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
10823 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
10824 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
10825 (dwarf2out_early_finish): Likewise.
10827 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
10829 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
10831 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
10834 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
10835 not VAR_P. Filter attribute warnings with OPT_Wattributes.
10836 (avr_attribute_table) <io, io_low, address>: Initialize
10837 .decl_required with true.
10839 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
10841 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
10842 undocumented debugging options.
10843 (-mvsx-scalar-double): Likewise.
10844 (-mallow-df-permute): Likewise.
10845 (-mvectorize-builtins): Likewise.
10846 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
10847 (rs6000_builtin_vectorized_function): Likewise.
10848 (rs6000_builtin_md_vectorized_function): Likewise.
10849 (rs6000_opt_vars): Likewise.
10851 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
10854 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
10855 (*btrq_imm): Rename from *btrq.
10856 (*btcq_imm): Rename from *btcq.
10857 (btsc): New code attribute.
10858 (*<btsc><mode>): New insn pattern.
10859 (*btr<mode>): Ditto.
10860 (*<btsc><mode>_mask): New insn_and_split pattern.
10861 (*btr<mode>_mask): Ditto.
10863 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10864 Alan Hayward <alan.hayward@arm.com>
10865 David Sherwood <david.sherwood@arm.com>
10867 * function.c (pad_below): Simplify padding calculation.
10869 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10870 Alan Hayward <alan.hayward@arm.com>
10871 David Sherwood <david.sherwood@arm.com>
10873 * target.def (function_prologue): Remove frame size argument.
10874 (function_epilogue): Likewise.
10875 * doc/tm.texi: Regenerate.
10876 * final.c (final_start_function): Update call to function_prologue.
10877 (final_end_function): Update call to function_epilogue.
10878 (default_function_pro_epilogue): Remove frame size argument.
10879 * output.h (default_function_pro_epilogue): Likewise.
10880 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
10881 (arm_output_function_prologue): Likewise.
10882 * config/frv/frv.c (frv_function_prologue): Likewise.
10883 (frv_function_epilogue): Likewise.
10884 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
10885 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
10886 (ia64_output_function_epilogue): Likewise.
10887 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
10888 (m32r_output_function_epilogue): Likewise.
10889 * config/microblaze/microblaze.c (microblaze_function_prologue)
10890 (microblaze_function_epilogue): Likewise.
10891 * config/mips/mips.c (mips_output_function_prologue): Likewise.
10892 (mips_output_function_epilogue): Likewise.
10893 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
10894 (mmix_target_asm_function_epilogue): Likewise.
10895 * config/msp430/msp430.c (msp430_start_function): Likewise.
10896 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
10897 (nds32_asm_function_epilogue): Likewise.
10898 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
10899 * config/pa/pa.c (pa_output_function_prologue): Likewise.
10900 (pa_output_function_epilogue): Likewise.
10901 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
10902 (rs6000_output_function_epilogue): Likewise.
10903 * config/rl78/rl78.c (rl78_start_function): Likewise.
10904 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
10905 (rs6000_output_function_epilogue): Likewise.
10906 * config/rx/rx.c (rx_output_function_prologue): Likewise.
10907 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
10908 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
10909 (sparc_asm_function_epilogue): Likewise.
10911 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10913 * tree.h (type_has_mode_precision_p): New function.
10914 * convert.c (convert_to_integer_1): Use it.
10915 * expr.c (expand_expr_real_2): Likewise.
10916 (expand_expr_real_1): Likewise.
10917 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
10918 * match.pd: Likewise.
10919 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
10920 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
10921 * tree-tailcall.c (process_assignment): Likewise.
10922 * tree-vect-loop.c (vectorizable_reduction): Likewise.
10923 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
10924 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
10925 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10926 (vectorizable_assignment): Likewise.
10927 (vectorizable_shift): Likewise.
10928 (vectorizable_operation): Likewise.
10929 * tree-vrp.c (register_edge_assert_for_2): Likewise.
10931 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
10933 * match.pd: Add pow (C, x) simplification.
10935 2017-08-21 Richard Biener <rguenther@suse.de>
10937 PR tree-optimization/81900
10938 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
10939 for blocks with abnormal predecessors.
10940 (compute_antic): Do not set visited flag prematurely.
10942 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
10945 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
10947 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10949 * stor-layout.h (vector_type_mode): Move to...
10950 * tree.h (vector_type_mode): ...here.
10951 * stor-layout.c (vector_type_mode): Move to...
10952 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
10954 2017-08-21 Richard Biener <rguenther@suse.de>
10956 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
10957 register_external_die hooks.
10958 (debug_false_tree_charstarstar_uhwistar): Declare.
10959 (debug_nothing_tree_charstar_uhwi): Likewise.
10960 * debug.c (do_nothing_debug_hooks): Adjust.
10961 (debug_false_tree_charstarstar_uhwistar): New do nothing.
10962 (debug_nothing_tree_charstar_uhwi): Likewise.
10963 * dbxout.c (dbx_debug_hooks): Adjust.
10964 (xcoff_debug_hooks): Likewise.
10965 * sdbout.c (sdb_debug_hooks): Likewise.
10966 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
10967 * dwarf2out.c (macinfo_label_base): New global.
10968 (dwarf2out_register_external_die): New function for the
10969 register_external_die hook.
10970 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
10971 (dwarf2_debug_hooks): Use them.
10972 (dwarf2_lineno_debug_hooks): Adjust.
10973 (struct die_struct): Add with_offset flag.
10974 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
10975 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
10976 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
10977 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
10978 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
10979 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
10980 defining section names for the early LTO debug variants.
10981 (reset_indirect_string): New helper.
10982 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
10983 (print_dw_val): Add support for offsetted symbol references.
10984 (get_ultimate_context): Split out from is_cxx.
10985 (is_cxx): Use get_ultimate_context.
10986 (is_fortran): Add decl overload.
10987 (compute_comp_unit_symbol): Split out worker from
10988 compute_section_prefix.
10989 (compute_section_prefix): Call compute_comp_unit_symbol and
10990 set comdat_type_p here.
10991 (output_die): Skip DIE symbol output for the LTO added one.
10992 Handle DIE symbol references with offset.
10993 (output_comp_unit): Guard section name mangling properly.
10994 For LTO debug sections emit a symbol at the section beginning
10995 which we use to refer to its DIEs.
10996 (add_abstract_origin_attribute): For DIEs registered via
10997 dwarf2out_register_external_die directly refer to the early
10998 DIE rather than indirectly through the shadow one we created.
10999 Remove obsolete call to dwarf2out_abstract_function for
11000 non-function/block origins.
11001 (gen_array_type_die): When generating early LTO debug do
11002 not emit DW_AT_string_length.
11003 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
11004 late when in LTO. As suggested place a gcc_unreachable for
11005 the DECL_ABSTRACT_P case.
11006 (gen_subprogram_die): Avoid another specification DIE
11007 for early built declarations/definitions for the late LTO case.
11008 (gen_variable_die): Add type references for late duplicated VLA dies
11010 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
11011 we have the abstract instance already.
11012 (process_scope_var): Adjust decl DIE contexts in LTO which
11013 first puts them in limbo.
11014 (gen_decl_die): Do not generate type DIEs late apart from
11015 types for VLAs or for decls we do not yet have a DIE. Do not
11016 call dwarf2out_abstract_function late.
11017 (dwarf2out_early_global_decl): Make sure to create DIEs
11018 for abstract instances of a decl first.
11019 (dwarf2out_late_global_decl): Adjust comment.
11020 (output_macinfo_op): With multiple macro sections use
11021 macinfo_label_base to distinguish labels.
11022 (output_macinfo): Likewise. Update macinfo_label_base.
11023 Pass in the line info label.
11024 (note_variable_value_in_expr): When generating LTO resolve
11025 all variable values here by generating DIEs as needed.
11026 (init_sections_and_labels): Add early LTO debug flag parameter
11027 and generate different sections and names if set. Add generation
11028 counter for the labels so we can have multiple of them.
11029 (reset_dies): Helper to allow DIEs to be output multiple times.
11030 (dwarf2out_finish): When outputting DIEs to the fat part of an
11031 LTO object first reset DIEs.
11032 (dwarf2out_early_finish): Output early DIEs when generating LTO.
11033 (modified_type_die): Check for decl_ultimate_origin being self
11035 (gen_type_die_with_usage): Likewise.
11036 (gen_typedef_die): Allow decl_ultimate_origin being self.
11037 (set_decl_abstract_flags): Remove.
11038 (set_block_abstract_flags): Likewise.
11039 (dwarf2out_abstract_function): Treat the early generated DIEs
11040 as the abstract copy and only add DW_AT_inline and
11041 DW_AT_artificial here and call set_decl_origin_self.
11042 If the DIE has an abstract origin don't do anything.
11043 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
11044 if we have none yet (Go fails to build one, PR78628).
11045 (variably_modified_type_p): Prevent endless recursion for Ada
11046 cyclic pointer types.
11047 * lto-streamer-in.c: Include debug.h.
11048 (dref_queue): New global.
11049 (lto_read_tree_1): Stream in DIE references.
11050 (lto_input_tree): Register DIE references.
11051 (input_function): Stream DECL_DEBUG_ARGS.
11052 * lto-streamer-out.c: Include debug.h.
11053 (lto_write_tree_1): Output DIE references.
11054 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
11055 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
11056 (output_function): Stream DECL_DEBUG_ARGS.
11057 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
11058 Stream DECL_ABSTRACT_ORIGIN.
11059 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
11060 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
11061 DECL_CONTEXT for file-scope decls.
11062 * lto-streamer.h (struct dref_entry): Declare.
11063 (dref_queue): Likewise.
11064 * cfgexpand.c (pass_expand::execute): Do not call the
11065 outlining_inline_function hook here.
11066 * lto-wrapper.c (debug_obj): New global.
11067 (tool_cleanup): Unlink it if required.
11068 (debug_objcopy): New function.
11069 (run_gcc): Handle early debug sections in the IL files by
11070 extracting them to separate files, partially linkin them and
11071 feeding the result back as result to the linker.
11072 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
11073 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
11074 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
11075 sections into a separate segment.
11076 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
11078 (darwin_asm_dwarf_section): Likewise.
11079 (darwin_asm_output_dwarf_offset): Likewise.
11080 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
11082 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11083 Alan Hayward <alan.hayward@arm.com>
11084 David Sherwood <david.sherwood@arm.com>
11086 * read-md.h (md_reader::record_potential_iterator_use): Replace
11087 pointer argument with an rtx and an index.
11088 * read-rtl.c (iterator_group::apply_iterator): Likewise.
11089 (apply_mode_iterator): Likewise.
11090 (apply_code_iterator): Likewise.
11091 (apply_int_iterator): Likewise.
11092 (apply_subst_iterator): Likewise.
11093 (record_iterator_use): Likewise.
11094 (record_attribute_use): Likewise.
11095 (md_reader::record_potential_iterator_use): Likewise. Update calls
11096 to record_iterator_use and apply_iterator.
11097 (iterator_use): Replace ptr with x and index.
11098 (attribute_use): Likewise.
11099 (apply_attribute_uses): Update calls to apply_iterator.
11100 (apply_iterators): Likewise. Update initialization of iterator_use.
11101 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
11102 and record_potential_iterator_use.
11103 (rtx_reader::read_rtx_operand): Likewise.
11105 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11106 Alan Hayward <alan.hayward@arm.com>
11107 David Sherwood <david.sherwood@arm.com>
11109 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
11112 2017-08-21 Richard Biener <rguenther@suse.de>
11114 PR middle-end/81884
11115 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
11116 at struct end conservatively when comparing common bases.
11118 2017-08-21 Richard Biener <rguenther@suse.de>
11120 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
11121 (mem_ref_in_stmt): Remove.
11122 (determine_max_movement): Use ref index to get at the reference.
11123 (invariantness_dom_walker::before_dom_children): Deal with
11124 lim data already initialized.
11125 (gather_mem_refs_stmt): Initialize lim data and record ref index.
11127 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
11129 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
11130 (TARGET_ISA_ROUND): Ditto.
11131 (TARGET_ROUND): Ditto.
11132 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
11133 * config/i386/i386.md: Ditto.
11134 * config/i386/sse.md: Ditto.
11135 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
11136 with OPTION_MASK_ISA_SSE4_1.
11138 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
11141 * doc/extend.texi (x86 Built-in Functions): Correct the name of
11142 __builtin_ia32_lzcnt_u16.
11144 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
11147 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
11148 (rs6000_set_current_function): Rewrite function to use it.
11150 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
11153 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
11154 and TYPE_WARN_IF_NOT_ALIGN.
11155 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
11156 (handle_warn_if_not_align): New.
11157 (place_union_field): Call handle_warn_if_not_align.
11158 (place_field): Call handle_warn_if_not_align.
11159 Copy TYPE_WARN_IF_NOT_ALIGN.
11160 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
11161 (layout_type): Likewise.
11162 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
11164 (tree_decl_common): Add warn_if_not_align.
11165 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
11166 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
11167 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
11168 (DECL_WARN_IF_NOT_ALIGN): Likewise.
11169 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
11170 * doc/extend.texi: Document warn_if_not_aligned attribute.
11171 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
11173 2017-08-17 Martin Liska <mliska@suse.cz>
11176 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
11177 (get_data_dependence): Use it as pointer type.
11178 (distribute_loop): Likewise.
11180 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11182 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
11183 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
11184 (p8_vmrgow_<mode>_direct): New define_insn.
11185 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
11186 handle endianness for vmrgew and vmrgow permute patterns.
11188 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
11190 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
11191 * config/rs6000/rs6000-cpus.def: Remove comment.
11192 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
11193 (POWERPC_MASKS): Likewise.
11194 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
11195 use of TARGET_VSX_TIMODE.
11196 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
11197 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
11198 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
11199 (rs6000_option_override_internal): Remove dead code.
11200 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
11201 (rs6000_legitimize_reload_address): Likewise.
11202 (rs6000_legitimate_address_p): Likewise.
11203 (rs6000_opt_masks): Delete "vsx-timode".
11204 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
11205 from function comment.
11206 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
11207 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
11208 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
11210 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
11211 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
11212 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
11213 (VSX_TI): Likewise.
11215 (define_peephole2): Likewise.
11217 2017-08-17 Martin Sebor <msebor@redhat.com>
11220 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
11221 past the end of an array.
11222 (test_pp_format): Add test cases.
11224 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
11226 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
11227 missing ECF_NOTHROW flags.
11229 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
11232 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
11233 operands residing in integer registers.
11234 (*vsx_le_perm_load_<mode>): Likewise.
11235 (*vsx_le_perm_store_<mode>): Likewise.
11236 (define_peephole2): Add peepholes to optimize the above.
11238 2017-08-17 Marek Polacek <polacek@redhat.com>
11240 PR middle-end/81814
11241 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
11242 to mimic what shorten_compare did. Change the return type to bool.
11243 (fold_cond_expr_with_comparison): Update call to
11244 operand_equal_for_comparison_p.
11245 (fold_ternary_loc): Likewise.
11247 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
11249 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
11251 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
11253 2017-08-17 Richard Biener <rguenther@suse.de>
11255 * tree-ssa-structalias.c (solve_graph): When propagating
11256 to successors update the graphs succ edges and avoid duplicate work.
11258 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
11261 * config/i386/i386.c (ix86_option_override_internal): Save target
11262 specific options after ix86_stack_protector_guard_reg was changed.
11264 2017-08-17 Richard Biener <rguenther@suse.de>
11266 PR tree-optimization/81827
11267 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
11268 (new_var_info): Initialize it conservatively.
11269 (get_call_vi): Mark register vars.
11270 (new_scalar_tmp_constraint_exp): Likewise.
11271 (handle_rhs_call): Likewise.
11272 (handle_const_call): Likewise.
11273 (create_function_info_for): Likewise.
11274 (solve_constraints): Sort varinfos to separate register from
11275 non-register vars to pack points-to solution bitmaps during
11278 2017-08-17 Marek Polacek <polacek@redhat.com>
11280 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
11282 2017-08-17 Richard Biener <rguenther@suse.de>
11284 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
11285 to true when overflow is undefined and we saturated the result.
11287 2017-08-17 Alan Modra <amodra@gmail.com>
11290 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
11291 Don't use store multiple if only one reg needs saving.
11292 (interesting_frame_related_regno): New function.
11293 (rs6000_frame_related): Don't emit frame info for regs that
11295 (rs6000_emit_epilogue): Likewise.
11297 2017-08-16 Nathan Sidwell <nathan@acm.org>
11299 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
11300 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
11301 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
11302 * tree.c (free_lang_data_in_type): Use else-if chain. Always
11303 clear TYPE_LANG_1. Remove obsolete member-function stripping.
11304 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
11305 (verify_type): Adjust for TYPE_BINFO move.
11306 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
11307 process TYPE_BINFO directly.
11308 (hash_tree): Likewise.
11309 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
11311 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
11314 2017-08-16 David Malcolm <dmalcolm@redhat.com>
11316 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
11318 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
11321 * config/i386/i386.md (*anddi_1_btr): Change predicates of
11322 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
11323 Add ix86_binary_operator_ok to insn constraint.
11324 (*iordi_1_bts): Ditto.
11325 (*xordi_1_btc): Ditto.
11326 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
11327 Update corresponding peephole2 pattern.
11331 2017-08-16 Bin Cheng <bin.cheng@arm.com>
11333 PR tree-optimization/81832
11334 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
11335 copy loop header which has IFN_LOOP_DIST_ALIAS call.
11337 2017-08-16 Marek Polacek <polacek@redhat.com>
11340 * fold-const.c (fold_indirect_ref_1): Restore original behavior
11341 regarding size_zero_node.
11343 2017-08-16 Martin Liska <mliska@suse.cz>
11346 * config.gcc: Respect previously set extra_objs in case
11349 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
11351 PR tree-optimization/81835
11352 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
11353 the phi SSA_NAME. Check that the condition in a COND_EXPR does
11354 not depend on the phi.
11356 2017-08-16 Alan Modra <amodra@gmail.com>
11358 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
11361 2017-08-16 Alan Modra <amodra@gmail.com>
11363 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
11364 (save_reg_p): ..into this. Update all callers.
11365 (first_reg_to_save): Simplify.
11367 2017-08-16 Alan Modra <amodra@gmail.com>
11369 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
11372 2017-08-15 Joseph Myers <joseph@codesourcery.com>
11376 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
11377 constant count if that count is less than 32.
11379 2017-08-15 Nathan Sidwell <nathan@acm.org>
11381 * gcc.c (execute): Emit friendlier message if inferior is killed
11382 by an external cause.
11384 2017-08-15 Richard Biener <rguenther@suse.de>
11386 PR tree-optimization/81790
11387 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
11388 CONSTRUCTORs from simplifying and VN.
11390 2017-08-14 Martin Sebor <msebor@redhat.com>
11392 * builtin-attrs.def: Add comments.
11394 2017-08-14 Martin Sebor <msebor@redhat.com>
11397 * doc/extend.texi (attribute nonstring): Document new attribute.
11399 2017-08-14 Martin Sebor <msebor@redhat.com>
11402 * tree-diagnostic.c (default_tree_printer): Handle %G.
11403 * gimple-pretty-print.h (percent_G_format): Declare new function.
11404 * gimple-pretty-print.c (percent_G_format): Define.
11405 * tree-pretty-print.c (percent_K_format): Add argument.
11407 2017-08-14 Martin Sebor <msebor@redhat.com>
11409 PR translation/79998
11410 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
11411 Remove a stray space.
11413 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
11416 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
11417 (*iordi_1_bts): Ditto.
11418 (*xordi_1_btc): Ditto.
11420 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11423 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
11424 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11426 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
11427 quoted strings, and make more translator-friendly.
11428 (darwin_rs6000_override_options): Likewise.
11429 (rs6000_option_override_internal): Likewise.
11430 (rs6000_return_in_memory): Fix overlong line.
11431 (init_cmulative_args): Use quoted strings, and make more
11432 translator-friendly.
11433 (rs6000_pass_by_reference): Fix overlong line.
11434 (def_builtin): Use quoted strings.
11435 (altivec_expand_predicate_builtin): Use quoted strings, and make
11436 more translator-friendly.
11437 (htm_expand_builtin): Use quoted strings.
11438 (cpu_expand_builtin): Use quoted strings, and make more
11439 translator-friendly.
11440 (altivec_expand_builtin): Likewise.
11441 (paired_expand_predicate_builtin): Likewise.
11442 (rs6000_invalid_builtin): Likewise.
11443 (builtin_function_type): Use quoted strings.
11444 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
11445 more translator-friendly.
11446 (rs6000_trampoline_init): Likewise.
11447 (rs6000_handle_altivec_attribute): Likewise.
11448 (rs6000_inner_target_options): Use quoted strings.
11449 (rs6000_disable_incompatible_switches): Likewise.
11450 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
11451 strings, and make more translator-friendly.
11452 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
11454 2017-08-14 Bin Cheng <bin.cheng@arm.com>
11456 PR tree-optimization/81799
11457 * tree-loop-distribution.c (version_loop_by_alias_check): Force
11458 cond_expr to simple gimple operand.
11460 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
11462 PR middle-end/46932
11463 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
11465 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
11469 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
11470 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
11471 TARGET_GASISR_PROLOGUES.
11472 * config/avr/avr.c (avr_option_override): Same.
11473 (avr_pass_pre_proep::execute): Same.
11475 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
11478 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
11479 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
11481 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
11483 * config/i386/i386.md (*load_tp_<mode>): Redefine as
11484 define_insn_and_split. Split to a memory load from 0 in
11485 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
11486 using PTR mode iterator.
11487 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
11488 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
11489 (*add_tp_<mode>): Redefine as define_insn_and_split.
11490 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
11491 address space. Merge with *add_tp_x32 using PTR mode iterator.
11492 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
11493 Split to an add with a memory load from 0 in
11494 DEFAULT_TLS_SEG_REG address space.
11496 2017-08-12 Andrew Pinski <apinski@cavium.com>
11498 * config/aarch64/aarch64-option-extensions.def (rdma):
11499 Fix feature string to what Linux prints out in /proc/cpuinfo.
11501 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
11504 * dwarf2out.c (modified_type_die): For C typedef types that have
11505 an ultimate origin, process the ultimate origin instead of the
11507 (gen_typedef_die): Assert that input DECLs have no ultimate
11509 (gen_type_die_with_usage): For typedef variants that have an
11510 ultimate origin, just call gen_decl_die on the original DECL.
11511 (process_scope_var): Avoid creating DIEs for local typedefs and
11512 concrete static variables.
11514 2017-08-12 Alan Modra <amodra@gmail.com>
11518 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
11519 match gnu-user.h startfile.
11520 (ENDFILE_LINUX_SPEC): Similarly.
11522 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
11525 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
11527 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
11529 2017-08-11 Tamar Christina <tamar.christina@arm.com>
11530 * config/aarch64/aarch64.md (mov<mode>): Change.
11531 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
11532 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
11533 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
11535 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
11537 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
11538 for storage order barriers.
11540 2017-08-11 Martin Liska <mliska@suse.cz>
11543 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
11544 variables of void type.
11546 2017-08-11 Martin Liska <mliska@suse.cz>
11548 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
11549 TARGET_SUPPORTS_ALIASES.
11550 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
11551 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
11552 (optimize_weakref): Likewise.
11553 * symtab.c (symtab_node::noninterposable_alias): Likewise.
11554 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
11555 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
11557 2017-08-11 Martin Liska <mliska@suse.cz>
11560 * config/i386/i386.c (make_resolver_func): Do complete
11561 refactoring of the function.
11563 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
11566 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
11567 * config/i386/i386.c (ix86_stack_protect_guard): Use
11568 ix86_stack_protect_guard_symbol_str to generate varible declaration.
11569 * doc/invoke.texi (x86 Options): Document
11570 -mstack-protector-guard-symbol= option.
11572 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
11574 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
11575 * config/i386/i386.c (ix86_split_stack_guard): New function.
11576 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
11577 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
11578 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
11579 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
11580 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
11581 (split_stack_space_check): Call ix86_split_stack_guard.
11583 2017-08-10 Martin Sebor <msebor@redhat.com>
11585 * print-tree.c (print_node): Print location using the established
11587 Replace spaces with colons.
11588 (debug_raw, debug): Ditto.
11590 2017-08-10 Martin Sebor <msebor@redhat.com>
11593 * pretty-print.c (pp_format): Correct the handling of %s precision.
11595 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
11598 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
11600 (ix86_finalize_stack_frame_flags): This. Also clear
11601 frame_pointer_needed if -fno-omit-frame-pointer is used without
11603 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
11604 with ix86_finalize_stack_frame_flags.
11605 (ix86_expand_epilogue): Likewise.
11606 (ix86_expand_split_stack_prologue): Likewise.
11607 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
11609 2017-08-10 Martin Liska <mliska@suse.cz>
11612 * c-attribs.c (handle_target_attribute):
11613 Report warning for an empty string argument of target attribute.
11615 2017-08-09 Jakub Jelinek <jakub@redhat.com>
11618 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
11620 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
11621 or DECL_NONLOCAL labels.
11622 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
11623 or DECL_NONLOCAL labels here.
11625 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
11627 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
11628 to indicate when early gimple folding has been disabled.
11629 (rs6000_gimple_fold_builtin): Add debug content.
11630 (rs6000_invalid_builtin): Fix whitespace.
11631 (rs6000_expand_builtin): Fix whitespace.
11632 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
11634 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
11637 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
11638 SAVE_MULTIPLE if not all the registers that saves, should be saved.
11640 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
11642 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
11643 (qdf24xx): Likewise.
11644 * config/aarch64/aarch64.md: Include falkor.md.
11645 * config/aarch64/falkor.md: New.
11647 2017-08-09 Marek Polacek <polacek@redhat.com>
11650 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
11651 * diagnostic.c (emit_diagnostic): Add a comment.
11652 (emit_diagnostic_valist): New function.
11654 2017-08-09 Marek Polacek <polacek@redhat.com>
11657 * input.c (make_location): New overload.
11658 * input.h (make_location): Declare.
11660 2017-08-08 Alan Modra <amodra@gmail.com>
11661 H.J. Lu <hongjiu.lu@intel.com>
11664 * gcc.c (NO_PIE_SPEC): Delete.
11665 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
11667 (LINK_PIE_SPEC): ..to here.
11668 (LINK_COMMAND_SPEC): Support -no-pie.
11669 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
11670 chain of crtbegin*.o selection, update for PIE_SPEC changes and
11672 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
11673 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
11674 (ENDFILE_CRTEND_SPEC): Similarly.
11676 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
11679 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
11680 (mstack-protector-guard-offset=): Ditto.
11681 * config/i386/i386.c (ix86_option_override): Handle
11682 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
11684 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
11685 ix86_stack_protect_guard_offset variables.
11686 (TARGET_STACK_PROTECT_GUARD): Always define.
11687 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
11688 and -mstack-protector-guard-offset= options.
11690 2017-08-08 Bin Cheng <bin.cheng@arm.com>
11692 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
11693 boundary case for the last candidate.
11695 2017-08-08 Bin Cheng <bin.cheng@arm.com>
11697 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
11698 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
11700 2017-08-08 Tamar Christina <tamar.christina@arm.com>
11702 PR middle-end/19706
11703 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
11704 * config/aarch64/aarch64-builtins.c
11705 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
11706 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
11707 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
11709 2017-08-08 Tamar Christina <tamar.christina@arm.com>
11710 Andrew Pinski <pinskia@gmail.com>
11712 PR middle-end/19706
11713 * internal-fn.def (XORSIGN): New.
11714 * optabs.def (xorsign_optab): New.
11715 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
11716 (convert_expand_mult_copysign): New.
11717 (pass_optimize_widening_mul::execute): Call
11718 convert_expand_mult_copysign.
11720 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11722 PR tree-optimization/81354
11723 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
11724 Insert on edges rather than explicitly creating landing pads.
11725 (analyze_candidates_and_replace): Commit edge inserts.
11727 2017-08-08 Richard Biener <rguenther@suse.de>
11729 PR middle-end/81719
11730 * tree-ssa-loop-niter.c: Include tree-dfa.h.
11731 (expand_simple_operations): Also look through ADDR_EXPRs with
11732 MEM_REF bases treating them as POINTER_PLUS_EXPR.
11734 2017-08-08 Richard Biener <rguenther@suse.de>
11736 PR tree-optimization/81723
11737 * tree-vect-slp.c (struct bst_traits): New hash traits.
11738 (bst_fail): New global.
11739 (vect_build_slp_tree_2): New worker, split out from ...
11740 (vect_build_slp_tree): ... this now wrapping it with using
11741 bst_fail set to cache SLP tree build fails. Properly handle
11743 (vect_analyze_slp_instance): Allocate and free bst_fail.
11745 2017-08-08 Martin Liska <mliska@suse.cz>
11748 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
11749 LABEL_DECLs that can be from a different function.
11751 2017-08-08 Bin Cheng <bin.cheng@arm.com>
11753 PR tree-optimization/81744
11754 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
11755 loop's number of iterations.
11757 2017-08-08 Martin Liska <mliska@suse.cz>
11759 * asan.c: Include header files.
11760 * attribs.c (build_decl_attribute_variant): New function moved
11762 (build_type_attribute_qual_variant): Likewise.
11763 (cmp_attrib_identifiers): Likewise.
11764 (simple_cst_list_equal): Likewise.
11765 (omp_declare_simd_clauses_equal): Likewise.
11766 (attribute_value_equal): Likewise.
11767 (comp_type_attributes): Likewise.
11768 (build_type_attribute_variant): Likewise.
11769 (lookup_ident_attribute): Likewise.
11770 (remove_attribute): Likewise.
11771 (merge_attributes): Likewise.
11772 (merge_type_attributes): Likewise.
11773 (merge_decl_attributes): Likewise.
11774 (merge_dllimport_decl_attributes): Likewise.
11775 (handle_dll_attribute): Likewise.
11776 (attribute_list_equal): Likewise.
11777 (attribute_list_contained): Likewise.
11778 * attribs.h (lookup_attribute): New function moved from tree.[ch].
11779 (lookup_attribute_by_prefix): Likewise.
11780 * bb-reorder.c: Include header files.
11781 * builtins.c: Likewise.
11782 * calls.c: Likewise.
11783 * cfgexpand.c: Likewise.
11784 * cgraph.c: Likewise.
11785 * cgraphunit.c: Likewise.
11786 * convert.c: Likewise.
11787 * dwarf2out.c: Likewise.
11788 * final.c: Likewise.
11789 * fold-const.c: Likewise.
11790 * function.c: Likewise.
11791 * gimple-expr.c: Likewise.
11792 * gimple-fold.c: Likewise.
11793 * gimple-pretty-print.c: Likewise.
11794 * gimple.c: Likewise.
11795 * gimplify.c: Likewise.
11796 * hsa-common.c: Likewise.
11797 * hsa-gen.c: Likewise.
11798 * internal-fn.c: Likewise.
11799 * ipa-chkp.c: Likewise.
11800 * ipa-cp.c: Likewise.
11801 * ipa-devirt.c: Likewise.
11802 * ipa-fnsummary.c: Likewise.
11803 * ipa-inline.c: Likewise.
11804 * ipa-visibility.c: Likewise.
11806 * lto-cgraph.c: Likewise.
11807 * omp-expand.c: Likewise.
11808 * omp-general.c: Likewise.
11809 * omp-low.c: Likewise.
11810 * omp-offload.c: Likewise.
11811 * omp-simd-clone.c: Likewise.
11812 * opts-global.c: Likewise.
11813 * passes.c: Likewise.
11814 * predict.c: Likewise.
11815 * sancov.c: Likewise.
11816 * sanopt.c: Likewise.
11817 * symtab.c: Likewise.
11818 * toplev.c: Likewise.
11819 * trans-mem.c: Likewise.
11820 * tree-chkp.c: Likewise.
11821 * tree-eh.c: Likewise.
11822 * tree-into-ssa.c: Likewise.
11823 * tree-object-size.c: Likewise.
11824 * tree-parloops.c: Likewise.
11825 * tree-profile.c: Likewise.
11826 * tree-ssa-ccp.c: Likewise.
11827 * tree-ssa-live.c: Likewise.
11828 * tree-ssa-loop.c: Likewise.
11829 * tree-ssa-sccvn.c: Likewise.
11830 * tree-ssa-structalias.c: Likewise.
11831 * tree-ssa.c: Likewise.
11832 * tree-streamer-in.c: Likewise.
11833 * tree-vectorizer.c: Likewise.
11834 * tree-vrp.c: Likewise.
11835 * tsan.c: Likewise.
11836 * ubsan.c: Likewise.
11837 * varasm.c: Likewise.
11838 * varpool.c: Likewise.
11839 * tree.c: Remove functions moved to attribs.[ch].
11840 * tree.h: Likewise.
11841 * config/aarch64/aarch64.c: Add attrs.h header file.
11842 * config/alpha/alpha.c: Likewise.
11843 * config/arc/arc.c: Likewise.
11844 * config/arm/arm.c: Likewise.
11845 * config/avr/avr.c: Likewise.
11846 * config/bfin/bfin.c: Likewise.
11847 * config/c6x/c6x.c: Likewise.
11848 * config/cr16/cr16.c: Likewise.
11849 * config/cris/cris.c: Likewise.
11850 * config/darwin.c: Likewise.
11851 * config/epiphany/epiphany.c: Likewise.
11852 * config/fr30/fr30.c: Likewise.
11853 * config/frv/frv.c: Likewise.
11854 * config/ft32/ft32.c: Likewise.
11855 * config/h8300/h8300.c: Likewise.
11856 * config/i386/winnt.c: Likewise.
11857 * config/ia64/ia64.c: Likewise.
11858 * config/iq2000/iq2000.c: Likewise.
11859 * config/lm32/lm32.c: Likewise.
11860 * config/m32c/m32c.c: Likewise.
11861 * config/m32r/m32r.c: Likewise.
11862 * config/m68k/m68k.c: Likewise.
11863 * config/mcore/mcore.c: Likewise.
11864 * config/microblaze/microblaze.c: Likewise.
11865 * config/mips/mips.c: Likewise.
11866 * config/mmix/mmix.c: Likewise.
11867 * config/mn10300/mn10300.c: Likewise.
11868 * config/moxie/moxie.c: Likewise.
11869 * config/msp430/msp430.c: Likewise.
11870 * config/nds32/nds32-isr.c: Likewise.
11871 * config/nds32/nds32.c: Likewise.
11872 * config/nios2/nios2.c: Likewise.
11873 * config/nvptx/nvptx.c: Likewise.
11874 * config/pa/pa.c: Likewise.
11875 * config/pdp11/pdp11.c: Likewise.
11876 * config/powerpcspe/powerpcspe.c: Likewise.
11877 * config/riscv/riscv.c: Likewise.
11878 * config/rl78/rl78.c: Likewise.
11879 * config/rx/rx.c: Likewise.
11880 * config/s390/s390.c: Likewise.
11881 * config/sh/sh.c: Likewise.
11882 * config/sol2.c: Likewise.
11883 * config/sparc/sparc.c: Likewise.
11884 * config/spu/spu.c: Likewise.
11885 * config/stormy16/stormy16.c: Likewise.
11886 * config/tilegx/tilegx.c: Likewise.
11887 * config/tilepro/tilepro.c: Likewise.
11888 * config/v850/v850.c: Likewise.
11889 * config/vax/vax.c: Likewise.
11890 * config/visium/visium.c: Likewise.
11891 * config/xtensa/xtensa.c: Likewise.
11893 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
11896 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
11897 constraints since the -mupper-regs-* switches have been
11899 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
11900 into a vector from a double word element that was extracted from
11901 another vector, and eliminate extra XXPERMDI instructions.
11902 (vsx_concat_<mode>_2): Likewise.
11903 (vsx_concat_<mode>_3): Likewise.
11904 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
11905 concat to allow optimizing inserts from previous extracts.
11907 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
11909 * config/i386/i386.c (ix86_stack_protect_guard): Generate
11910 memory reference to a SSP offset in TLS address space.
11911 (ix86_print_operand) <case '@'>: Remove.
11912 (ix86_print_operand_punct_valid_p): Remove '@' code.
11913 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
11914 UNSPEC_SP_TLS_TEST.
11915 (stack_tls_protect_set_<mode>): Remove.
11916 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
11917 (stack_tls_protect_test_<mode>): Remove.
11918 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
11920 2017-08-07 Olivier Hainque <hainque@adacore.com>
11923 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
11925 2017-08-07 Douglas Rupp <rupp@adacore.com>
11927 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
11928 variable was referenced as multidir in command.
11930 2017-08-07 Jakub Jelinek <jakub@redhat.com>
11933 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
11936 2017-08-07 Martin Liska <mliska@suse.cz>
11938 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
11939 * config/rl78/rl78.c: Add include of attribs.h.
11940 * config/sh/sh.c: Likewise.
11941 * config/v850/v850.c: Likewise.
11943 2017-08-07 Tom de Vries <tom@codesourcery.com>
11945 PR middle-end/78266
11946 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
11948 2017-08-07 Martin Liska <mliska@suse.cz>
11950 * config/mips/mips.c: Include attribs.h.
11952 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
11955 * doc/invoke.texi: Document change in behvaior for -Ofast for
11958 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
11960 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
11962 (aarch64_pop_regs): Likewise.
11963 (aarch64_gen_load_pair): Likewise.
11964 (aarch64_save_callee_saves): Likewise.
11965 (aarch64_restore_callee_saves): Likewise.
11967 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
11969 * config/i386/i386.c: Revert the last change.
11971 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
11974 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
11976 (ix86_finalize_stack_frame_flags): This. Also clear
11977 frame_pointer_needed if -fno-omit-frame-pointer is used without
11979 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
11980 with ix86_finalize_stack_frame_flags.
11981 (ix86_expand_epilogue): Likewise.
11982 (ix86_expand_split_stack_prologue): Likewise.
11984 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
11987 * config/i386/i386.c (get_builtin_code_for_version): Set priority
11988 to P_AES for Westmere.
11990 2017-08-07 Jonathan Yong <10walls@gmail.com>
11992 * config/i386/mingw.opt (fset-stack-executable): Removed.
11993 * config/i386/cygming.opt (fset-stack-executable): Moved
11995 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
11997 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
11999 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
12001 2017-08-07 Marek Polacek <polacek@redhat.com>
12003 PR middle-end/81737
12004 * fold-const.c (fold_indirect_ref_1): Check type_domain.
12006 2017-08-07 Martin Liska <mliska@suse.cz>
12008 * attribs.h (canonicalize_attr_name): New function.
12009 (cmp_attribs): Move from c-format.c and adjusted.
12010 (is_attribute_p): Moved from tree.h.
12011 * tree-inline.c: Add new includes.
12012 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
12013 (private_is_attribute_p): Remove.
12014 (private_lookup_attribute): Likewise.
12015 (private_lookup_attribute_by_prefix): Simplify.
12016 (remove_attribute): Use is_attribute_p.
12017 * tree.h: Remove removed declarations.
12019 2017-08-07 Jakub Jelinek <jakub@redhat.com>
12021 PR middle-end/81698
12022 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
12023 instead of computing it in the function. Formatting fix.
12024 (expand_case): Don't rely on default_edge being the first edge,
12025 clear it if removing it, pass default_edge to
12026 emit_case_dispatch_table.
12027 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
12030 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
12032 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
12033 insn in the function, emit NOP after the insn.
12035 2017-08-06 Tom de Vries <tom@codesourcery.com>
12037 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
12040 2017-08-06 Tom de Vries <tom@codesourcery.com>
12042 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
12045 2017-08-04 Yury Gribov <tetra2005@gmail.com>
12047 PR tree-optimization/57371
12048 * match.pd: New pattern.
12050 2017-08-04 Marek Polacek <polacek@redhat.com>
12052 PR middle-end/81695
12053 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
12054 perform the computation in offset_int.
12056 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
12058 PR tree-optimization/81136
12059 * tree-vectorizer.h: Include tree-hash-traits.h.
12060 (vec_base_alignments): New typedef.
12061 (vec_info): Add a base_alignments field.
12062 (vect_record_base_alignments): Declare.
12063 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
12065 (DR_IS_CONDITIONAL_IN_STMT): New macro.
12066 (create_data_ref): Add an is_conditional_in_stmt argument.
12067 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
12068 the is_conditional_in_stmt field.
12069 (data_ref_loc): Add an is_conditional_in_stmt field.
12070 (get_references_in_stmt): Set the is_conditional_in_stmt field.
12071 (find_data_references_in_stmt): Update call to create_data_ref.
12072 (graphite_find_data_references_in_stmt): Likewise.
12073 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
12074 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
12075 (vect_record_base_alignment): New function.
12076 (vect_record_base_alignments): Likewise.
12077 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
12078 for nested statements even if we fail to compute a misalignment.
12079 Use pooled base alignments for unconditional references.
12080 (vect_find_same_alignment_drs): Compare base addresses instead
12082 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
12083 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
12085 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
12087 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
12088 Add an explicit name for the enum. Use auto_vec for slp_instances
12089 and grouped_stores.
12090 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
12092 (_bb_vec_info): Add a constructor and destructor.
12093 (vinfo_for_stmt): Return NULL for uids of -1 as well.
12094 (destroy_loop_vec_info): Delete.
12095 (vect_destroy_datarefs): Likewise.
12096 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
12097 (vec_info::vec_info): New function.
12098 (vec_info::~vec_info): Likewise.
12099 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
12100 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
12101 destroy_loop_vec_info.
12102 * tree-vect-loop.c (new_loop_vec_info): Replace with...
12103 (_loop_vec_info::_loop_vec_info): ...this.
12104 (destroy_loop_vec_info): Replace with...
12105 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
12106 the stmt_vec_infos. Leave handling of vec_info information to its
12107 destructor. Remove explicit vector releases.
12108 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
12109 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
12110 * tree-vect-slp.c (new_bb_vec_info): Replace with...
12111 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
12112 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
12113 (destroy_bb_vec_info): Replace with...
12114 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
12115 information to its destructor.
12116 (vect_slp_analyze_bb_1): Use new and delete instead of
12117 new_bb_vec_info and destroy_bb_vec_info.
12118 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
12121 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
12123 * tree-data-ref.h (subscript): Add access_fn field.
12124 (data_dependence_relation): Add could_be_independent_p.
12125 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
12126 (same_access_functions): Move to tree-data-ref.c.
12127 * tree-data-ref.c (ref_contains_union_access_p): New function.
12128 (access_fn_component_p): Likewise.
12129 (access_fn_components_comparable_p): Likewise.
12130 (dr_analyze_indices): Add a reference to access_fn_component_p.
12131 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
12133 (constant_access_functions): Likewise.
12134 (add_other_self_distances): Likewise.
12135 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
12136 (initialize_data_dependence_relation): Use XCNEW and remove
12137 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
12138 of access functions that have the same type. Allow the
12139 subsequence to end with different bases in some circumstances.
12140 Record the chosen access functions in SUB_ACCESS_FN.
12141 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
12142 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
12143 (subscript_dependence_tester_1): Likewise dra and drb.
12144 (build_classic_dist_vector): Update calls accordingly.
12145 (subscript_dependence_tester): Likewise.
12146 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
12147 DDR_COULD_BE_INDEPENDENT_P.
12148 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
12149 comp_alias_ddrs instead of may_alias_ddrs.
12150 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
12152 (vect_analyze_data_ref_dependence): Use it if
12153 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
12154 distance vectors if that fails.
12155 (dependence_distance_ge_vf): New function.
12156 (vect_prune_runtime_alias_test_list): Use it. Don't clear
12157 LOOP_VINFO_MAY_ALIAS_DDRS.
12159 2017-08-04 Richard Biener <rguenther@suse.de>
12161 PR middle-end/81705
12162 * fold-const.c (fold_binary_loc): Properly restrict
12163 minus_var0 && minus_var1 case when associating undefined overflow
12166 2017-08-04 Tom de Vries <tom@codesourcery.com>
12168 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
12170 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12172 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12173 Don't start diagnostic messages with a capital letter.
12174 * config/rs6000/rs6000.c (rs6000_option_override_internal):
12176 (rs6000_invalid_builtin): Likewise.
12177 (rs6000_trampoline_init): Likewise.
12179 2017-08-03 Jakub Jelinek <jakub@redhat.com>
12182 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
12183 after setting changeable df flags.
12185 2017-08-03 Richard Biener <rguenther@suse.de>
12187 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
12188 up if the use is in USE - X.
12190 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
12192 * toplev.c (dumpfile.h): New include.
12193 (internal_error_reentered): New static function. Use it...
12194 (internal_error_function): ...here to handle reentered internal_error.
12196 2017-08-03 Richard Biener <rguenther@suse.de>
12198 PR middle-end/81148
12199 * fold-const.c (split_tree): Add minus_var and minus_con
12200 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
12201 here but always use minus_*.
12202 (associate_trees): Assert we never associate with MINUS_EXPR
12203 and NULL first operand. Do not recurse for PLUS_EXPR operands
12204 when associating as MINUS_EXPR either.
12205 (fold_binary_loc): Track minus_var and minus_con.
12207 2017-08-03 Tom de Vries <tom@codesourcery.com>
12210 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
12211 ACCEL_COMPILER, apply finish_options on
12212 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
12214 2017-08-03 Tom de Vries <tom@codesourcery.com>
12217 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
12218 function_entry_patch_area_size > 0.
12220 2017-08-03 Jakub Jelinek <jakub@redhat.com>
12223 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
12224 instead of 10??LU, perform unit multiplication in wide_int,
12225 don't change alloc_object_size_limit if the limit is larger
12228 PR tree-optimization/81655
12229 PR tree-optimization/81588
12230 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
12231 the case when ranges[i].low and high are 1 for unsigned type with
12234 PR middle-end/81052
12235 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
12236 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
12238 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12240 * tree-vrp.h: Add include guard.
12242 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
12245 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
12246 (ud2): New insn pattern.
12247 * config/i386/i386.c (ix86_expand_epilogue):
12248 For naked functions, generate ud2 instead of trap insn.
12250 2017-08-02 Marek Polacek <polacek@redhat.com>
12253 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
12255 2017-08-02 Tom de Vries <tom@codesourcery.com>
12256 Cesar Philippidis <cesar@codesourcery.com>
12258 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
12259 Add missing edge probabilities.
12261 2017-08-02 Tamar Christina <tamar.christina@arm.com>
12263 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
12264 Correct endianness.
12266 2017-08-02 Jakub Jelinek <jakub@redhat.com>
12268 PR middle-end/79499
12269 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
12270 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
12271 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
12272 prologue_seq sequences - if any.
12274 2017-08-02 Richard Biener <rguenther@suse.de>
12276 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
12277 via vectors if supported, integer extracts via punning if supported
12278 or otherwise vector extracts.
12280 2017-08-02 Richard Biener <rguenther@suse.de>
12282 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
12284 (bitmap_insert_into_set): ... this.
12286 2017-08-02 Richard Biener <rguenther@suse.de>
12288 PR tree-optimization/81633
12290 2015-08-17 Alan Hayward <alan.hayward@arm.com>
12292 PR tree-optimization/71752
12293 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
12295 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
12297 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
12298 (machine_function::call_ms2sysv_pad_out): Remove field.
12299 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
12300 (ix86_compute_frame_layout): Likewise.
12302 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
12305 * config/i386/i386.c (ix86_set_func_type): Disallow naked
12306 attribute with interrupt attribute.
12308 2017-08-01 Andrew Pinski <apinski@cavium.com>
12310 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
12311 BIT_INSERT_EXPR's operand 1
12312 to see if the types precision matches.
12314 2017-08-01 Martin Liska <mliska@suse.cz>
12316 PR middle-end/70140
12317 * builtins.c (expand_builtin_memcpy_args): Remove.
12318 (expand_builtin_memcpy): Call newly added function
12319 expand_builtin_memory_copy_args.
12320 (expand_builtin_memcpy_with_bounds): Likewise.
12321 (expand_builtin_mempcpy): Remove last argument.
12322 (expand_builtin_mempcpy_with_bounds): Likewise.
12323 (expand_builtin_memory_copy_args): New function created from
12324 expand_builtin_mempcpy_args with small modifications.
12325 (expand_builtin_mempcpy_args): Remove.
12326 (expand_builtin_stpcpy): Remove unused argument.
12327 (expand_builtin): Likewise.
12328 (expand_builtin_with_bounds): Likewise.
12330 2017-08-01 Martin Liska <mliska@suse.cz>
12333 Make mempcpy more optimal (PR middle-end/70140).
12335 2017-08-01 Jakub Jelinek <jakub@redhat.com>
12338 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
12339 __builtin_vec_cmpne verify both arguments are compatible vectors
12340 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
12341 verify arg1_type is a pointer or array type. For __builtin_vec_st,
12342 move computation of aligned to after checking the argument types.
12346 * config/rs6000/vsx.md (vextract_fp_from_shorth,
12347 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
12350 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
12351 Doug Rupp <rupp@adacore.com>
12352 Olivier Hainque <hainque@adacore.com>
12354 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
12355 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
12357 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
12358 for TARGET_OS_CPP_BUILTIN.
12359 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
12360 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
12362 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
12363 passing required abi options to the assembler for EABI configurations.
12364 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
12365 of .text.hot and .text.unlikely sections for kernel modules when
12366 using ARM style exceptions.
12367 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
12368 options. Add EXTRA_CC1_SPEC.
12369 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
12371 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
12373 (ARM_TARGET2_DWARF_FORMAT): Define.
12374 * config/arm/t-vxworks: Adjust multilib control to removal of the
12375 Diab command line options.
12377 2017-08-01 Martin Liska <mliska@suse.cz>
12379 PR gcov-profile/81561
12380 * gcov.c (unblock): Make unblocking safe as we need to preserve
12381 index correspondence of blocks and block_lists.
12383 2017-08-01 Richard Biener <rguenther@suse.de>
12385 PR tree-optimization/81181
12386 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
12387 (compute_antic): ... end of iteration here.
12389 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
12391 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
12392 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
12393 (ftree-slp-vectorize): Likewise.
12394 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
12395 can no longer be set independent of flag_tree_loop_vectorize.
12396 * omp-general.c (emp_max_vf): Likewise.
12397 * opts.c (enable_fdo_optimizations): Remove references to
12398 flag_tree_vectorize, these are now implicit.
12399 (common_handle_option): Remove handling for OPT_ftree_vectorize,
12400 and leave it for the options machinery.
12402 2017-08-01 Martin Liska <mliska@suse.cz>
12404 PR middle-end/70140
12405 * builtins.c (expand_builtin_memcpy_args): Remove.
12406 (expand_builtin_memcpy): Call newly added function
12407 expand_builtin_memory_copy_args.
12408 (expand_builtin_memcpy_with_bounds): Likewise.
12409 (expand_builtin_mempcpy): Remove last argument.
12410 (expand_builtin_mempcpy_with_bounds): Likewise.
12411 (expand_builtin_memory_copy_args): New function created from
12412 expand_builtin_mempcpy_args with small modifications.
12413 (expand_builtin_mempcpy_args): Remove.
12414 (expand_builtin_stpcpy): Remove unused argument.
12415 (expand_builtin): Likewise.
12416 (expand_builtin_with_bounds): Likewise.
12418 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
12421 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
12422 print "ds:" only for immediates in generic address space.
12424 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
12427 * config/i386/i386.c (ix86_funciton_naked): New prototype.
12428 (ix86_function_ok_for_sibcall): Return false for naked functions.
12430 2017-08-01 Richard Biener <rguenther@suse.de>
12432 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
12433 (compute_antic): Seed worklist with exit block predecessors.
12434 * cfganal.c (dfs_find_deadend): For a cycle return the source
12435 of the edge closing it.
12437 2017-08-01 Tamar Christina <tamar.christina@arm.com>
12439 * config/aarch64/aarch64.c
12440 (aarch64_can_const_movi_rtx_p): Move 0 check.
12442 2017-08-01 Bin Cheng <bin.cheng@arm.com>
12444 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
12445 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
12447 * match.pd: Ditto in address comparison pattern.
12449 2017-08-01 Bin Cheng <bin.cheng@arm.com>
12451 PR tree-optimization/81627
12452 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
12453 closed ssa form for store-store chain.
12455 2017-08-01 Bin Cheng <bin.cheng@arm.com>
12457 PR tree-optimization/81620
12458 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
12459 for store-store chain.
12461 2017-08-01 Jakub Jelinek <jakub@redhat.com>
12463 PR tree-optimization/81588
12464 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
12465 ranges[i].in_p, invert comparison code ccode. For >/>=,
12466 swap rhs1 and rhs2 and comparison code unconditionally,
12467 for </<= don't do that. Don't swap rhs1/rhs2 again if
12468 ranges[i].in_p, instead invert comparison code ccode if
12469 opcode or oe->rank is BIT_IOR_EXPR.
12472 * optabs.def (vec_extract_optab, vec_init_optab): Change from
12473 a direct optab to conversion optab.
12474 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
12475 with GET_MODE_INNER as last argument instead of optab_handler.
12476 * expmed.c (extract_bit_field_1): Likewise. Use vector from
12477 vector extraction if possible and optab is available.
12478 * expr.c (store_constructor): Use convert_optab_handler instead
12479 of optab_handler. Use vector initialization from smaller
12480 vectors if possible and optab is available.
12481 * tree-vect-stmts.c (vectorizable_load): Likewise.
12482 * doc/md.texi (vec_extract, vec_init): Document that the optabs
12483 now have two modes.
12484 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
12485 of vec_init from half-sized vectors with the same element mode.
12486 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
12487 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
12488 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
12489 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
12490 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
12491 after mode in gen_vec_extract* calls.
12492 (vec_extract<mode>): Renamed to ...
12493 (vec_extract<mode><ssescalarmodelower>): ... this.
12494 (vec_extract<mode><ssehalfvecmodelower>): New expander.
12495 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
12496 element mode after mode in gen_vec_init* calls.
12497 (VEC_INIT_HALF_MODE): New mode iterator.
12498 (vec_init<mode>): Renamed to ...
12499 (vec_init<mode><ssescalarmodelower>): ... this.
12500 (vec_init<mode><ssehalfvecmodelower>): New expander.
12501 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
12502 (vec_extractv2sfsf): ... this.
12503 (vec_initv2sf): Renamed to ...
12504 (vec_initv2sfsf): ... this.
12505 (vec_extractv2si): Renamed to ...
12506 (vec_extractv2sisi): ... this.
12507 (vec_initv2si): Renamed to ...
12508 (vec_initv2sisi): ... this.
12509 (vec_extractv4hi): Renamed to ...
12510 (vec_extractv4hihi): ... this.
12511 (vec_initv4hi): Renamed to ...
12512 (vec_initv4hihi): ... this.
12513 (vec_extractv8qi): Renamed to ...
12514 (vec_extractv8qiqi): ... this.
12515 (vec_initv8qi): Renamed to ...
12516 (vec_initv8qiqi): ... this.
12517 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
12518 (vec_init<mode>): Renamed to ...
12519 (vec_init<mode><VEC_base_l>): ... this.
12520 (vec_extract<mode>): Renamed to ...
12521 (vec_extract<mode><VEC_base_l>): ... this.
12522 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
12523 (vec_initv2sfsf): ... this.
12524 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
12525 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
12526 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
12527 element mode after mode in gen_vec_init* calls.
12528 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
12529 (vec_init<mode><Vel>): ... this.
12530 (vec_extract<mode>): Renamed to ...
12531 (vec_extract<mode><Vel>): ... this.
12532 * config/aarch64/iterators.md (Vel): New mode attribute.
12533 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
12534 Add element mode after mode in gen_vec_extract* calls.
12535 * config/s390/vector.md (non_vec_l): New mode attribute.
12536 (vec_extract<mode>): Renamed to ...
12537 (vec_extract<mode><non_vec_l>): ... this.
12538 (vec_init<mode>): Renamed to ...
12539 (vec_init<mode><non_vec_l>): ... this.
12540 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
12541 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
12543 * config/arm/iterators.md (V_elem_l): New mode attribute.
12544 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
12545 (vec_extract<mode><V_elem_l>): ... this.
12546 (vec_extractv2di): Renamed to ...
12547 (vec_extractv2didi): ... this.
12548 (vec_init<mode>): Renamed to ...
12549 (vec_init<mode><V_elem_l>): ... this.
12550 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
12551 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
12552 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
12553 Add element mode after gen_vec_extract* calls.
12554 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
12555 (vec_init<mode><unitmode>): ... this.
12556 (vec_extract<mode>): Renamed to ...
12557 (vec_extract<mode><unitmode>): ... this.
12558 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
12559 (vec_init<mode><unitmode>): ... this.
12560 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
12561 (vec_initv2sfsf): ... this.
12562 (vec_extractv2sf): Renamed to ...
12563 (vec_extractv2sfsf): ... this.
12564 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
12565 Add element mode after gen_vec_extract* calls.
12566 * config/mips/mips.md (unitmode): New mode iterator.
12567 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
12568 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
12569 * config/spu/spu.md (inner_l): New mode attribute.
12570 (vec_init<mode>): Renamed to ...
12571 (vec_init<mode><inner_l>): ... this.
12572 (vec_extract<mode>): Renamed to ...
12573 (vec_extract<mode><inner_l>): ... this.
12574 * config/sparc/sparc.md (veltmode): New mode iterator.
12575 (vec_init<VMALL:mode>): Renamed to ...
12576 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
12577 * config/ia64/vect.md (vec_initv2si): Renamed to ...
12578 (vec_initv2sisi): ... this.
12579 (vec_initv2sf): Renamed to ...
12580 (vec_initv2sfsf): ... this.
12581 (vec_extractv2sf): Renamed to ...
12582 (vec_extractv2sfsf): ... this.
12583 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
12584 (vec_init<mode>): Renamed to ...
12585 (vec_init<mode><VEC_base_l>): ... this.
12586 (vec_extract<mode>): Renamed to ...
12587 (vec_extract<mode><VEC_base_l>): ... this.
12588 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
12589 (vec_initv2sfsf): ... this.
12590 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
12591 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
12592 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
12593 gen_vec_init* calls.
12595 2017-08-01 Richard Biener <rguenther@suse.de>
12597 PR tree-optimization/81297
12598 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
12599 TREE_OVERFLOW from INTEGER_CSTs.
12601 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
12603 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
12605 2017-07-31 Carl Love <cel@us.ibm.com>
12607 * config/rs6000/rs6000-c: Add support for built-in functions
12608 vector signed char vec_xl_be (signed long long, signed char *);
12609 vector unsigned char vec_xl_be (signed long long, unsigned char *);
12610 vector signed int vec_xl_be (signed long long, signed int *);
12611 vector unsigned int vec_xl_be (signed long long, unsigned int *);
12612 vector signed long long vec_xl_be (signed long long, signed long long *);
12613 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
12614 vector signed short vec_xl_be (signed long long, signed short *);
12615 vector unsigned short vec_xl_be (signed long long, unsigned short *);
12616 vector double vec_xl_be (signed long long, double *);
12617 vector float vec_xl_be (signed long long, float *);
12618 * config/rs6000/altivec.h (vec_xl_be): Add #define.
12619 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
12620 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
12622 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
12623 (altivec_expand_builtin): Add switch statement to call
12624 altivec_expand_xl_be for each builtin.
12625 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
12626 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
12627 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
12628 __builtin_vsx_le_be_v16qi.
12629 * doc/extend.texi: Update the built-in documentation file for the
12630 new built-in functions.
12632 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
12635 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
12637 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
12639 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12641 * config.gcc: Add z14.
12642 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
12643 CPU model numbers for z13s and z14.
12644 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
12646 * config/s390/s390-opts.h (enum processor_type): Rename
12647 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
12648 * config/s390/s390.c (processor_table): Add field for CPU name to
12649 be passed to Binutils.
12650 (s390_asm_output_machine_for_arch): Use the new field in
12651 processor_table for Binutils.
12652 (s390_expand_builtin): Replace arch12 with z14.
12653 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
12654 (s390_get_sched_attrmask): Likewise.
12655 (s390_get_unit_mask): Likewise.
12656 * config/s390/s390.opt: Add z14 to processor_type enum.
12658 2017-07-31 Martin Jambor <mjambor@suse.cz>
12661 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
12662 regardless of optimization level.
12664 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
12665 Martin Liska <mliska@suse.cz>
12667 * predict.def: Remove old comment and adjust probability.
12668 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
12669 PREDICT statements.
12671 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
12674 * config/i386/i386.c (ix86_function_naked): New function.
12675 (ix86_can_use_return_insn_p): Return false for naked functions.
12676 (ix86_expand_prologue): Skip prologue for naked functions.
12677 (ix86_expand_epilogue): Skip epilogue for naked functions
12678 and emit trap instruction.
12679 (ix86_warn_func_return): New function.
12680 (ix86_attribute_table): Add "naked" attribute specification.
12681 (TARGET_WARN_FUNC_RETURN): Define.
12682 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
12684 2017-07-31 Martin Liska <mliska@suse.cz>
12686 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
12687 (dump_gimple_bb_header): Always dump BB info.
12688 (pp_cfg_jump): Do not append info about BB when dumping a jump.
12690 2017-07-31 Martin Liska <mliska@suse.cz>
12693 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
12694 also with current_function_decl non-null equality.
12696 2017-07-31 Jakub Jelinek <jakub@redhat.com>
12699 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
12700 change type to the element type, instead add eltype variable and
12701 use it where we are interested in the element type.
12703 PR tree-optimization/81603
12704 * ipa-polymorphic-call.c
12705 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
12706 offset arithmetic in offset_int, bail out if the resulting bit offset
12707 doesn't fit into shwi.
12709 2017-07-31 Martin Liska <mliska@suse.cz>
12711 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
12712 (gimplify_save_expr): Fix comment.
12714 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
12717 * config/i386/i386.c (ix86_function_arg): Update arguments for
12719 (ix86_compute_frame_layout): Set the initial stack offset to
12720 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
12721 INCOMING_FRAME_SP_OFFSET.
12722 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
12723 stack before exception handler returns.
12724 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
12725 the 'ERROR_CODE' for exception handler.
12727 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
12729 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
12730 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
12731 (ASM_OUTPUT_REG_POP): Ditto.
12732 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
12733 instead of asm_fprintf to output pure string.
12735 2017-07-29 Jakub Jelinek <jakub@redhat.com>
12737 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
12738 to imported_module_or_decl hook.
12739 (debug_nothing_tree_tree_tree_bool): Remove.
12740 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
12741 * debug.c (do_nothing_debug_hooks): Use
12742 debug_nothing_tree_tree_tree_bool_bool instead of
12743 debug_nothing_tree_tree_tree_bool.
12744 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
12745 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
12746 * sdbout.c (sdb_debug_hooks): Likewise.
12747 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
12748 (gen_namespace_die): Add DW_AT_export_symbols attribute if
12750 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
12751 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
12752 attribute, don't add anything.
12754 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12756 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
12757 (fold_build2_stat_loc): Likewise.
12758 (fold_build3_stat_loc): Likewise.
12759 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
12760 (fold_build1_loc): Remove macro.
12761 (fold_build2_loc): Likewise.
12762 (fold_build3_loc): Likewise.
12764 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12766 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
12767 (gimple_build_debug_bind_source_stat): Likewise.
12768 * gimple.h (gimple_build_debug_bind): Remove macro.
12769 (gimple_build_debug_bind_source): Likewise.
12771 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12773 * bitmap.c (bitmap_alloc): Adjust.
12774 (bitmap_gc_alloc): Likewise.
12775 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
12777 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12779 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
12780 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
12781 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
12782 (bitmap_gc_alloc_stat): Likewise.
12783 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
12785 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12787 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
12788 * rtl.h (shallow_copy_rtx): Remove macro.
12790 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12792 * emit-rtl.c (gen_raw_REG): Adjust.
12793 * gengenrtl.c (gendef): Likewise.
12794 * rtl.c (rtx_alloc_stat): Remove _stat from name.
12795 * rtl.h (rtx_alloc): Remove macro.
12797 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12799 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
12800 (build_tree_list_stat): Likewise.
12801 * tree.h (build_tree_list): Remove macro.
12802 (build_tree_list_vec): Likewise.
12804 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12806 * tree.c (make_vector_stat): Remove _stat from name.
12807 (build_vector_stat): Likewise.
12808 * tree.h (make_vector_stat): Remove macro.
12809 (build_vector_stat): Likewise.
12811 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12813 * tree.h (build_var_debug_value): Remove prototype.
12815 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12817 * tree.c (tree_cons_stat): Remove _stat from name.
12818 * tree.h (tree_cons): Remove macro.
12820 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12822 * tree.c (build_vl_exp_stat): Remove _stat from name.
12823 * tree.h (build_vl_exp): Remove macro.
12825 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12827 * tree.c (build_decl_stat): Remove _stat from name.
12828 * tree.h (build_decl): Remove macro.
12830 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12832 * gimple.c (gimple_build_with_ops_stat): Adjust.
12833 (gimple_alloc_stat): Remove _stat from name.
12834 * gimple.h (gimple_alloc): Remove macro.
12836 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12838 * tree.c (make_tree_vec_stat): Remove _stat from name.
12839 (grow_tree_vec_stat): Likewise.
12840 * tree.h (make_tree_vec_stat): Adjust prototype.
12841 (grow_tree_vec_stat): Likewise.
12842 (make_tree_vec): Remove macro.
12843 (grow_tree_vec): Likewise.
12845 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12847 * fold-const.c (fold_build1_stat_loc): Adjust.
12848 (fold_build2_stat_loc): Likewise.
12849 (fold_build3_stat_loc): Likewise.
12850 * tree.c (build0_stat): Remove _stat from name.
12851 (build1_stat): Likewise.
12852 (build2_stat): Likewise.
12853 (build3_stat): Likewise.
12854 (build4_stat): Likewise.
12855 (build5_stat): Likewise.
12856 * tree.h (build1_loc): Remove macro, and rename _stat function
12858 (build2_loc): Likewise.
12859 (build3_loc): Likewise.
12860 (build4_loc): Likewise.
12861 (build5_loc): Likewise.
12863 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12865 * tree.c (make_int_cst_stat): Remove _stat from name.
12866 * tree.h (make_int_cst_stat): Adjust prototype.
12867 (make_int_cst): Remove macro.
12869 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12871 * tree.c (make_tre_binfo_stat): Remove _stat from name.
12872 * tree.h (make_tree_binfo_stat): Adjust prototype.
12873 (make_tree_binfo): Remove.
12875 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12877 * tree.c (copy_node_stat): Rename to copy_node.
12878 (build_distinct_type_copy): Adjust.
12879 * tree.h (copy_node_stat): Adjust prototype.
12880 (copy_node): Remove macro.
12882 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12884 * tree.c (make_node_stat): rename to make_node.
12885 (build_tree_list_stat): Adjust.
12886 (build0_stat): Likewise.
12887 (build2_stat): Likewise.
12888 (build3_stat): Likewise.
12889 (build4_stat): Likewise.
12890 (build5_stat): Likewise.
12891 (build_decl_stat): Likewise.
12892 * tree.h (make_node_stat): Adjust prototype.
12893 (make_node): remove macro.
12895 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
12897 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
12898 (PPC_FEATURE2_SCV): Likewise.
12899 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
12901 2017-07-28 Tamar Christina <tamar.christina@arm.com>
12903 * config/aarch64/aarch64.c
12904 (aarch64_internal_mov_immediate): Add new special pattern.
12905 * config/aarch64/aarch64.md (*movdi_aarch64):
12906 Add reg/32bit const mov case.
12908 2017-07-28 Tamar Christina <tamar.christina@arm.com>
12909 Richard Sandiford <richard.sandiford@linaro.org>
12911 * config/aarch64/aarch64.md (mov<mode>): Generalize.
12912 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
12913 Add integer and movi cases.
12914 (movi-split-hf-df-sf split, fp16): New.
12915 (enabled): Added TARGET_FP_F16INST.
12916 * config/aarch64/iterators.md (GPF_HF): New.
12917 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
12919 2017-07-28 Tamar Christina <tamar.christina@arm.com>
12921 * config/aarch64/aarch64.c
12922 (aarch64_simd_container_mode): Add prototype.
12923 (aarch64_expand_mov_immediate): Add HI support.
12924 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
12925 (aarch64_can_const_movi_rtx_p): New.
12926 (aarch64_preferred_reload_class):
12927 Remove restrictions of using FP registers for certain SIMD operations.
12928 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
12929 (aarch64_valid_floating_const): Add integer move validation.
12930 (aarch64_simd_imm_scalar_p): Remove.
12931 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
12932 (aarch64_legitimate_constant_p): Expand list of supported cases.
12933 * config/aarch64/aarch64-protos.h
12934 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
12935 (aarch64_reinterpret_float_as_int): New.
12936 (aarch64_simd_imm_scalar_p): Remove.
12937 * config/aarch64/constraints.md (Uvi): New.
12938 (Dd): Split into Ds and new Dd.
12939 * config/aarch64/aarch64.md (*movsi_aarch64):
12941 (*movdi_aarch64): Add SIMD mov case.
12943 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12945 * tree-predcom.c: (struct chain): Handle store-store chain in which
12946 stores for elimination only store loop invariant values.
12947 (execute_pred_commoning_chain): Ditto.
12948 (prepare_initializers_chain_store_elim): Ditto.
12949 (prepare_finalizers): Ditto.
12950 (is_inv_store_elimination_chain): New function.
12951 (initialize_root_vars_store_elim_1): New function.
12953 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12955 * tree-predcom.c: Revise general description of the pass.
12956 (enum chain_type): New enum type for store elimination.
12957 (struct chain): New field supporting store elimination.
12958 (struct component): Ditto.
12959 (dump_chain): Dump store-stores chain.
12960 (release_chain): Release resources.
12961 (split_data_refs_to_components): Compute and create component
12962 contains only stores for elimination.
12963 (get_chain_last_ref_at): New function.
12964 (make_invariant_chain): Initialization.
12965 (make_rooted_chain): Specify chain type in parameter and record it.
12966 (add_looparound_copies): Skip for store-stores chain.
12967 (determine_roots_comp): Compute type of chain and pass it to
12969 (initialize_root_vars_store_elim_2): New function.
12970 (finalize_eliminated_stores): New function.
12971 (remove_stmt): Handle store for elimination.
12972 (execute_pred_commoning_chain): Execute predictive commoning on
12973 store-store chains.
12974 (determine_unroll_factor): Skip unroll for store-stores chain.
12975 (prepare_initializers_chain_store_elim): New function.
12976 (prepare_initializers_chain): Hanlde store-store chain.
12977 (prepare_finalizers_chain, prepare_finalizers): New function.
12978 (tree_predictive_commoning_loop): Return integer value indicating
12979 if loop is unrolled or lcssa form is corrupted.
12980 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
12982 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12984 * tree-predcom.c (initialize_root): Delete.
12985 (execute_pred_commoning_chain): Initialize root vars and replace
12986 reference of non-combined chain directly, rather than call above
12989 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12991 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
12992 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
12994 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12996 * tree-predcom.c (struct chain): New field init_seq.
12997 (release_chain): Release init_seq.
12998 (prepare_initializers_chain): Record intialization stmts in above
13000 (insert_init_seqs): New function.
13001 (tree_predictive_commoning_loop): Call insert_init_seqs.
13003 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13005 * tree-predcom.c (determine_roots_comp): Skip trivial components.
13007 2017-07-28 Richard Biener <rguenther@suse.de>
13009 * match.pd: Remove superfluous :c.
13010 * genmatch.c (simplify::id): Add member.
13011 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
13013 (current_id): New global.
13014 (dt_node::parent): Move from ...
13015 (dt_operand::parent): ... here. Add for_id member.
13016 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
13017 (decision_tree::find_node): Relax order requirement when
13018 merging DT_TRUE nodes to ones inbetween the current simplify
13019 and the one we try to merge with. Add diagnostic whenever
13020 we need to enforce pattern order by not merging.
13021 (decision_tree::insert): Set current_id.
13022 (decision_tree::print_node): Dump parent node and for_id.
13023 (parser::last_id): Add member.
13024 (parser::push_simplify): Assign unique id.
13025 (parser::parser): Initialize last_id.
13027 2017-07-28 Martin Liska <mliska@suse.cz>
13030 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
13031 gimple_build_debug_bind.
13033 2017-07-28 Richard Biener <rguenther@suse.de>
13035 PR tree-optimization/81502
13036 * match.pd: Add pattern combining BIT_INSERT_EXPR with
13038 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
13040 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
13041 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
13042 for BIT_FIELD_REF args.
13043 * fold-const.c (make_bit_field_ref): Likewise.
13044 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
13046 2017-07-28 Jakub Jelinek <jakub@redhat.com>
13049 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
13050 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
13051 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
13052 Or it into SANITIZER_UNDEFINED.
13053 * ubsan.c: Include gimple-fold.h and varasm.h.
13054 (ubsan_expand_ptr_ifn): New function.
13055 (instrument_pointer_overflow): New function.
13056 (maybe_instrument_pointer_overflow): New function.
13057 (instrument_object_size): Formatting fix.
13058 (pass_ubsan::execute): Call instrument_pointer_overflow
13059 and maybe_instrument_pointer_overflow.
13060 * internal-fn.c (expand_UBSAN_PTR): New function.
13061 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
13062 * sanitizer.def (__ubsan_handle_pointer_overflow,
13063 __ubsan_handle_pointer_overflow_abort): New builtins.
13064 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
13065 * internal-fn.def (UBSAN_PTR): New internal function.
13066 * opts.c (sanitizer_opts): Add pointer-overflow.
13067 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
13068 * fold-const.c (build_range_check): Compute pointer range check in
13069 integral type if pointer arithmetics would be needed. Formatting
13072 2017-07-28 Martin Liska <mliska@suse.cz>
13075 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
13076 parameters that are of a variable-length.
13078 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
13080 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
13082 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
13083 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
13084 (CRT_CALL_STATIC_FUNCTION): Likewise.
13085 (ASM_DEFAULT_SPEC): New define.
13086 (ASM_SPEC32): Likewise.
13087 (ASM_SPEC64): Likewise.
13088 (ASM_SPEC_COMMON): Likewise.
13089 (ASM_SPEC): Likewise.
13090 (INVALID_64BIT): Likewise.
13091 (LINK_OS_DEFAULT_SPEC): Likewise.
13092 (LINK_OS_SPEC32): Likewise.
13093 (LINK_OS_SPEC64): Likewise.
13094 (POWERPC_LINUX): Likewise.
13095 (PTRDIFF_TYPE): Likewise.
13096 (RESTORE_FP_PREFIX): Likewise.
13097 (RESTORE_FP_SUFFIX): Likewise.
13098 (SAVE_FP_PREFIX): Likewise.
13099 (SAVE_FP_SUFFIX): Likewise.
13100 (SIZE_TYPE): Likewise.
13101 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
13102 (TARGET_64BIT): Likewise.
13103 (TARGET_64BIT): Likewise.
13104 (TARGET_AIX): Likewise.
13105 (WCHAR_TYPE_SIZE): Likewise.
13106 (WCHAR_TYPE): Undefine.
13107 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
13108 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
13109 (CPP_OS_RTEMS_SPEC): Delete.
13110 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
13111 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
13113 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
13115 2017-07-28 Jakub Jelinek <jakub@redhat.com>
13117 PR tree-optimization/81578
13118 * tree-parloops.c (build_new_reduction): Bail out if
13119 reduction_code isn't one of the standard OpenMP reductions.
13120 Move the details printing after that decision.
13122 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
13124 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
13125 related to reload_in_progress.
13126 (splat_input_operand): Likewise.
13127 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
13129 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
13131 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
13132 (TARGET_INSTANTIATE_DECLS): Likewise.
13133 (legitimate_indexed_address_p): Delete reload_in_progress code.
13134 (rs6000_debug_legitimate_address_p): Likewise.
13135 (rs6000_eliminate_indexed_memrefs): Likewise.
13136 (rs6000_emit_le_vsx_store): Likewise.
13137 (rs6000_emit_move_si_sf_subreg): Likewise.
13138 (rs6000_emit_move): Likewise.
13139 (register_to_reg_type): Likewise.
13140 (rs6000_pre_atomic_barrier): Likewise.
13141 (rs6000_machopic_legitimize_pic_address): Likewise.
13142 (rs6000_allocate_stack_temp): Likewise.
13143 (rs6000_address_for_fpconvert): Likewise.
13144 (rs6000_address_for_altivec): Likewise.
13145 (rs6000_secondary_memory_needed_rtx): Delete function.
13146 (rs6000_check_sdmode): Likewise.
13147 (rs6000_alloc_sdmode_stack_slot): Likewise.
13148 (rs6000_instantiate_decls): Likewise.
13149 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
13150 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
13151 Delete reload_in_progress.
13152 (*vec_reload_and_plus_<mptrsize>): Likewise.
13153 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
13154 (vsx_div_v2di): Likewise.
13155 (vsx_udiv_v2di): Likewise.
13157 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
13159 * config/rs6000/rs6000.opt (mlra): Replace with stub.
13160 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
13161 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
13162 (rs6000_debug_reg_global): Delete print of LRA status.
13163 (rs6000_option_override_internal): Delete dead LRA related code.
13164 (rs6000_lra_p): Delete function.
13165 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
13167 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
13169 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
13170 * config/riscv/rtems.h: New file.
13172 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13173 Sudakshina Das <sudi.das@arm.com>
13175 * config/aarch64/aarch64.md
13176 (define_split for and<mode>3nr_compare): Move
13177 non aarch64_logical_operand to a register.
13178 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
13179 register immediate operand to a register.
13180 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
13182 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
13184 PR middle-end/81564
13185 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
13187 2017-07-27 Richard Biener <rguenther@suse.de>
13189 PR tree-optimization/81573
13190 PR tree-optimization/81494
13191 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
13192 multi defuse cycle case.
13194 2017-07-27 Richard Biener <rguenther@suse.de>
13196 PR tree-optimization/81571
13197 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
13200 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
13202 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
13203 earlier and only if MASK_FPU is set. Adjust formatting.
13205 2017-07-27 Martin Liska <mliska@suse.cz>
13207 * opt-functions.awk: Add validation of value of Init.
13208 * optc-gen.awk: Pass new argument.
13210 2017-07-27 Martin Liska <mliska@suse.cz>
13212 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
13213 Fix wrong condition.
13215 2017-07-27 Martin Liska <mliska@suse.cz>
13217 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
13218 BBs and edges seen by autoFDO.
13220 2017-07-27 Richard Biener <rguenther@suse.de>
13222 PR tree-optimization/81502
13223 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
13224 with incompatible but same sized type.
13225 (execute_update_addresses_taken): Likewise.
13227 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
13229 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
13230 flag_tree_loop_vectorize rather than flag_tree_vectorize.
13232 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13235 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
13236 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
13237 Change s_operand to memory_operand.
13239 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
13241 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
13242 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
13243 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
13244 Emit instructions rather than returning an expression. Handle TFmode
13245 and KFmode by casting to TImode.
13246 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
13247 (rs6000_emit_le_vsx_store): Likewise.
13248 * config/rs6000/vsx.md (VSX_TI): New iterator.
13249 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
13250 (*vsx_le_undo_permute_<mode>): Likewise.
13251 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
13252 emit the split sequence.
13253 (*vsx_le_perm_store_<mode>): Likewise.
13255 2017-07-27 Jakub Jelinek <jakub@redhat.com>
13257 PR tree-optimization/81555
13258 PR tree-optimization/81556
13259 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
13260 if true, force CHANGED for the recursive invocation.
13261 (reassociate_bb): Remember original length of ops array, pass
13262 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
13264 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
13265 attributes for noipa attribute. For naked attribute use
13266 lookup_attribute first before lookup_attribute_spec.
13267 * final.c (rest_of_handle_final): Disable IPA RA for functions with
13269 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
13270 for functions with noipa attribute.
13271 (cgraph_externally_visible_p): Return true for functions with noipa
13273 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
13274 for functions with noipa attribute.
13275 * doc/extend.texi: Document noipa function attribute.
13276 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
13277 also for functions with noipa attribute.
13278 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
13280 2017-07-26 Andrew Pinski <apinski@cavium.com>
13282 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
13283 vec_unalign_load_cost and vec_unalign_store_cost.
13285 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13287 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
13288 -mvsx-small-integer option.
13289 (ISA_3_0_MASKS_IEEE): Likewise.
13290 (OTHER_VSX_VECTOR_MASKS): Likewise.
13291 (POWERPC_MASKS): Likewise.
13292 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
13293 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
13294 code, only testing for DImode being allowed in non-VSX floating
13296 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
13297 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
13299 (rs6000_option_override_internal): Delete -mvsx-small-integer.
13300 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
13301 TARGET_P8_VECTOR test.
13302 (rs6000_secondary_reload_simple_move): Likewise.
13303 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
13304 since TARGET_P9_VECTOR was already tested.
13305 (rs6000_opt_masks): Remove -mvsx-small-integer.
13306 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
13307 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
13309 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
13310 test for TARGET_VEXTRACTUB was used, and that uses
13312 (p9 extract splitter): Likewise.
13313 (vsx_extract_<mode>_di_p9): Likewise.
13314 (vsx_extract_<mode>_store_p9): Likewise.
13315 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
13316 for TARGET_P9_VECTOR was used. Delete code that is now dead with
13317 the elimination of TARGET_VSX_SMALL_INTEGER.
13318 (vsx_extract_<mode>_p8): Likewise.
13319 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
13320 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
13321 (vsx_set_<mode>_p9): Likewise.
13322 (vsx_set_v4sf_p9): Likewise.
13323 (vsx_set_v4sf_p9_zero): Likewise.
13324 (vsx_insert_extract_v4sf_p9): Likewise.
13325 (vsx_insert_extract_v4sf_p9_2): Likewise.
13326 * config/rs6000/rs6000.md (sign extend splitter): Change
13327 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
13328 (floatsi<mode>2_lfiwax_mem): Likewise.
13329 (floatunssi<mode>2_lfiwzx_mem): Likewise.
13330 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
13331 since a test for TARGET_P9_VECTOR was used.
13332 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
13333 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
13334 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
13335 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
13336 TARGET_P8_VECTOR test.
13337 (fix_trunc<mode>si2_stfiwx): Likewise.
13338 (fix_trunc<mode>si2_internal): Likewise.
13339 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
13340 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
13342 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
13343 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
13344 TARGET_P8_VECTOR test.
13345 (fixuns_trunc<mode>si2_stfiwx): Likewise.
13346 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
13347 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
13349 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
13350 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
13351 since a test for TARGET_P9_VECTOR was used.
13352 (splitter for loading small constants): Likewise.
13354 2017-07-26 Andrew Pinski <apinski@cavium.com>
13356 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
13359 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
13362 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
13363 (fp_valid_at): Likewise.
13365 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
13367 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
13368 (qdf24xx_addrcost_table): Likewise.
13369 (cortexa57_tunings): Update to use generic_branch_cost.
13370 (cortexa72_tunings): Likewise.
13371 (cortexa73_tunings): Likewise.
13372 (qdf24xx_tunings): Likewise.
13374 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
13376 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
13377 (thunderx2t99_branch_cost): Likewise.
13378 (cortexa35_tunings): Update to use generic_branch_cost.
13379 (cortexa53_tunings): Likewise.
13380 (cortexa57_tunings): Likewise.
13381 (cortexa72_tunings): Likewise.
13382 (cortexa73_tunings): Likewise.
13383 (thunderx2t99_tunings): Likewise.
13385 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
13387 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
13388 (sparc_option_override): Honour MASK_FSMULD.
13389 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
13390 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
13391 * config/sparc/sparc.opt (mfsmuld): New option.
13392 * doc/invoke.texi (mfsmuld): Document option.
13394 2017-07-26 Marek Polacek <polacek@redhat.com>
13396 PR middle-end/70992
13397 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
13399 2017-07-26 Richard Biener <rguenther@suse.de>
13401 * gimple-match-head.c (do_valueize): Return OP if valueize
13403 (get_def): New helper to get at the def stmt of a SSA name
13404 if valueize allows.
13405 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
13406 do_valueize to get at the def stmt.
13407 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
13409 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
13411 PR middle-end/46932
13412 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
13414 2017-07-26 Martin Liska <mliska@suse.cz>
13417 * function.c (expand_function_start): Make expansion of
13418 nonlocal_goto_save_area after parm_birth_insn.
13420 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
13422 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
13423 from all CPU target flags enable members.
13425 2017-07-26 Richard Biener <rguenther@suse.de>
13427 * genmatch.c (dt_simplify::gen): Make iterator vars const.
13428 (decision_tree::gen): Make 'type' const.
13429 (write_predicate): Likewise.
13431 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
13433 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
13434 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
13435 (rs6000_option_override_internal): Likewise.
13436 (rs6000_expand_vector_set): Likewise.
13437 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
13438 (TARGET_UPPER_REGS_SF): Likewise.
13439 (TARGET_UPPER_REGS_DI): Likewise.
13440 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
13441 (TARGET_DIRECT_MOVE_64BIT): Likewise.
13442 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
13443 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
13444 (Splitters for DI constants in Altivec registers): Likewise.
13445 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
13446 (vsx_set_v4sf_p9): Likewise.
13447 (vsx_set_v4sf_p9_zero): Likewise.
13448 (vsx_insert_extract_v4sf_p9): Likewise.
13449 (vsx_insert_extract_v4sf_p9_2): Likewise.
13451 2017-07-25 Carl Love <cel@us.ibm.com>
13453 * doc/extend.texi: Update the built-in documentation file for the
13454 existing built-in functions
13455 vector signed char vec_cnttz (vector signed char);
13456 vector unsigned char vec_cnttz (vector unsigned char);
13457 vector signed short vec_cnttz (vector signed short);
13458 vector unsigned short vec_cnttz (vector unsigned short);
13459 vector signed int vec_cnttz (vector signed int);
13460 vector unsigned int vec_cnttz (vector unsigned int);
13461 vector signed long long vec_cnttz (vector signed long long);
13462 vector unsigned long long vec_cnttz (vector unsigned long long);
13464 2017-07-25 Andrew Pinski <apinski@cavium.com>
13466 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
13467 accesses where the use is for the first operand of a BIT_INSERT.
13469 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
13472 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
13473 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
13475 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
13477 * config/i386/gstabs.h: Delete.
13478 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
13480 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
13482 * config/i386/i386.c (ix86_decompose_address): Do not check for
13483 register RTX when looking at index_reg or base_reg.
13484 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
13486 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
13488 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
13489 to update EH info here.
13491 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
13493 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
13495 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
13497 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
13499 2017-07-25 Torsten Duwe <duwe@suse.de>
13501 * common.opt: Introduce -fpatchable-function-entry
13502 command line option, and its variables function_entry_patch_area_size
13503 and function_entry_patch_area_start.
13504 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
13505 including a two-value parser.
13506 * target.def (print_patchable_function_entry): New target hook.
13507 * targhooks.h (default_print_patchable_function_entry): New function.
13508 * targhooks.c (default_print_patchable_function_entry): Likewise.
13509 * toplev.c (process_options): Switch off IPA-RA if
13510 patchable function entries are being generated.
13511 * varasm.c (assemble_start_function): Look at the
13512 patchable-function-entry command line switch and current
13513 function attributes and maybe generate NOP instructions by
13514 calling the print_patchable_function_entry hook.
13515 * doc/extend.texi: Document patchable_function_entry attribute.
13516 * doc/invoke.texi: Document -fpatchable_function_entry
13517 command line option.
13518 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
13520 * doc/tm.texi: Re-generate.
13522 2017-07-25 Jakub Jelinek <jakub@redhat.com>
13525 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
13526 TARGET_AVX512DQ rather than TARGET_AVX512BW.
13528 2017-07-25 Tamar Christina <tamar.christina@arm.com>
13530 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
13532 2017-07-25 Richard Biener <rguenther@suse.de>
13534 PR tree-optimization/81455
13535 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
13536 not walk in cycles when looking for guards.
13538 2017-07-25 Richard Biener <rguenther@suse.de>
13540 PR tree-optimization/81529
13541 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
13542 when optimizing backedge uses.
13544 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
13546 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
13548 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
13549 to dl_section_ref. On AIX, append an expression to subtract
13550 the size of the section length to dl_section_ref.
13552 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
13554 * configure.ac: If any of the config.* scripts fail, exit 1.
13555 * configure: Regenerate.
13557 2017-07-25 Richard Biener <rguenther@suse.de>
13559 PR middle-end/81546
13560 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
13561 of immediate uses, be more verbose on errors.
13563 2017-07-25 Richard Biener <rguenther@suse.de>
13565 PR tree-optimization/81510
13566 * tree-vect-loop.c (vect_is_simple_reduction): When the
13567 reduction stmt is not inside the loop bail out.
13569 2017-07-25 Richard Biener <rguenther@suse.de>
13571 PR tree-optimization/81303
13572 * tree-vect-loop-manip.c (vect_loop_versioning): Build
13573 profitability check against LOOP_VINFO_NITERSM1.
13575 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
13577 * domwalk.c (cmp_bb_postorder): Simplify.
13578 (sort_bbs_postorder): New function. Use it...
13579 (dom_walker::walk): ...here to optimize common cases.
13581 2017-07-25 Martin Liska <mliska@suse.cz>
13584 * ipa-visibility.c (function_and_variable_visibility): Make the
13585 redirection just on target that supports aliasing.
13586 Fix GNU coding style.
13588 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
13591 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
13593 * config/arm/rtems.h: Likewise.
13594 * config/bfin/rtems.h: Likewise.
13595 * config/i386/rtemself.h: Likewise.
13596 * config/lm32/rtems.h: Likewise.
13597 * config/m32c/rtems.h: Likewise.
13598 * config/m68k/rtemself.h: Likewise.
13599 * config/microblaze/rtems.h: Likewise.
13600 * config/mips/rtems.h: Likewise.
13601 * config/moxie/rtems.h: Likewise.
13602 * config/nios2/rtems.h: Likewise.
13603 * config/powerpcspe/rtems.h: Likewise.
13604 * config/rs6000/rtems.h: Likewise.
13605 * config/rtems.h: Likewise.
13606 * config/sh/rtems.h: Likewise.
13607 * config/sh/rtemself.h: Likewise.
13608 * config/sparc/rtemself.h: Likewise.
13610 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
13613 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
13614 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
13615 * tree-ssa-structalias.c (alias_get_name): Same.
13617 2017-07-25 Bin Cheng <bin.cheng@arm.com>
13620 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
13621 instructions if no du chain is found.
13623 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
13625 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
13627 2017-07-25 Richard Biener <rguenther@suse.de>
13629 PR middle-end/81505
13630 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
13633 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
13635 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
13636 upper-regs options.
13637 (ISA_2_7_MASKS_SERVER): Likewise.
13638 (ISA_3_0_MASKS_IEEE): Likewise.
13639 (OTHER_P8_VECTOR_MASKS): Likewise.
13640 (OTHER_VSX_VECTOR_MASKS): Likewise.
13641 (POWERPC_MASKS): Likewise.
13642 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
13643 duplicate list of options.
13644 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
13645 explicit -mupper-regs options.
13646 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
13647 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
13648 alias for -mupper-regs-df.
13649 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
13650 (rs6000_init_hard_regno_mode_ok): Likewise.
13651 (rs6000_option_override_internal): Likewise.
13652 (rs6000_opt_masks): Likewise.
13653 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
13654 options in terms of whether -mvsx or -mpower8-vector was used.
13655 (TARGET_UPPER_REGS_DI): Likewise.
13656 (TARGET_UPPER_REGS_SF): Likewise.
13657 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
13658 -mupper-regs-* options.
13660 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
13662 * passes.c (emergency_dump_function): Print some empty lines and a
13663 header before the RTL dump.
13665 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
13667 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
13669 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
13672 * config/aarch64/aarch64.c (aarch64_classify_symbol):
13673 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
13675 2017-07-24 Carl Love <cel@us.ibm.com>
13677 * config/rs6000/rs6000-c.c: Add support for built-in functions
13678 vector float vec_extract_fp32_from_shorth (vector unsigned short);
13679 vector float vec_extract_fp32_from_shortl (vector unsigned short);
13680 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
13681 vec_extract_fp_from_shortl): Add defines for the two builtins.
13682 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
13683 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
13685 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
13686 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
13687 * doc/extend.texi: Update the built-in documentation file for the
13688 new built-in function.
13690 2017-07-24 Jakub Jelinek <jakub@redhat.com>
13693 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
13695 * doc/generic.texi: Likewise.
13696 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
13697 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
13699 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
13701 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
13702 (aarch64_mls_elt_merge<mode>): Likewise.
13704 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
13706 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
13707 having __cxa_atexit.
13709 2017-07-23 Michael Collison <michael.collison@arm.com>
13711 * config/arm/arm.c (arm_option_override): Deprecate
13712 use of -mstructure-size-boundary.
13713 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
13714 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
13716 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13719 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
13720 Reduce cost estimate for direct moves.
13722 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
13725 * config/i386/i386.c (ix86_option_override_internal): Disable
13726 BMI, BMI2 and TBM instructions for -m16.
13728 2017-07-21 Carl Love <cel@us.ibm.com>
13730 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13731 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13732 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13733 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13734 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13735 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
13736 VMULOSW): New enum "unspec" values.
13737 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
13738 altivec_vmulosw): New patterns.
13739 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13740 VMULOSW): Add definitions.
13742 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
13744 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
13745 (qdf24xx): Likewise.
13746 * config/aarch64/aarch64-options-extensions.def (rdma); New.
13747 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
13748 (AARCH64_FL_V8_1): Renumber.
13749 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
13750 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
13751 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
13752 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
13753 rdma to feature modifiers list.
13755 2017-07-21 Yury Gribov <tetra2005@gmail.com>
13757 PR middle-end/56727
13758 * ipa-visibility (function_and_variable_visibility): Convert
13759 recursive PLT call to direct call if appropriate.
13761 2017-07-21 Andrew Pinski <apinski@cavium.com>
13763 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
13764 operand 1 to see if the types precision matches.
13765 * fold-const.c (operand_equal_p): Likewise.
13767 2017-07-21 Richard Biener <rguenther@suse.de>
13769 PR tree-optimization/81303
13770 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
13771 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
13772 (vect_peeling_hash_get_lowest_cost): Adjust.
13773 (vect_enhance_data_refs_alignment): Likewise. Use
13774 vect_get_peeling_costs_all_drs to compute the penalty for no
13775 peeling to match up costs.
13777 2017-07-21 Richard Biener <rguenther@suse.de>
13779 PR tree-optimization/81500
13780 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
13781 we didn't identify a reduction path.
13783 2017-07-21 Tom de Vries <tom@codesourcery.com>
13784 Cesar Philippidis <cesar@codesourcery.com>
13786 PR gcov-profile/81442
13787 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
13790 2017-07-21 Tom de Vries <tom@codesourcery.com>
13793 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
13795 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
13796 nvptx_override_options_after_change.
13798 2017-07-21 Ulrich Drepper <drepper@redhat.com>
13800 * dwarf2out.c (output_file_names): Avoid double testing for
13801 dwarf_version >= 5.
13803 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
13805 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
13807 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
13809 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
13811 (try_crossjump_to_edge): Do not punt on partitioned functions.
13813 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
13815 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
13816 Put all BBs reachable only via paths crossing cold region to cold
13818 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
13820 2016-07-21 Richard Biener <rguenther@suse.de>
13822 PR tree-optimization/81303
13823 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
13824 into account prologue and epilogue iterations when raising
13825 min_profitable_iters to sth at least covering one vector iteration.
13827 2017-07-21 Tamar Christina <tamar.christina@arm.com>
13829 * config/arm/arm.c (arm_test_cpu_arch_dat):
13832 2017-07-20 Nathan Sidwell <nathan@acm.org>
13834 Remove TYPE_METHODS.
13835 * tree.h (TYPE_METHODS): Delete.
13836 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
13837 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
13838 (dbxout_type_methods): Scan TYPE_FIELDS.
13839 (dbxout_type): Don't check TYPE_METHODS here.
13840 * function.c (use_register_for_decl): Always ignore register for
13841 class types when not optimizing.
13842 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
13843 * tree.c (free_lang_data_in_type): Stitch out member functions and
13844 templates from TYPE_FIELDS.
13845 (build_distinct_type_copy, verify_type_variant,
13846 verify_type): Member fns are on TYPE_FIELDS.
13847 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
13848 * tree-pretty-print.c (dump_generic_node): Likewise.
13850 2017-07-20 Jakub Jelinek <jakub@redhat.com>
13853 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
13854 V2TImode and V4TImode.
13855 (ix86_expand_vector_extract): Likewise.
13856 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
13857 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
13858 (ssescalarmode): Handle V4TImode and V2TImode.
13859 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
13860 (*vec_extractv2ti, *vec_extractv4ti): New insns.
13861 (VEXTRACTI128_MODE): New mode iterator.
13862 (splitter for *vec_extractv?ti first element): New.
13863 (VEC_INIT_MODE): New mode iterator.
13864 (vec_init<mode>): Consolidate 3 expanders into one using
13865 VEC_INIT_MODE mode iterator.
13867 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
13869 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
13870 non_spilled_static_chain_regno_p.
13872 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
13874 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
13876 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
13878 * bb-reorder.c (connect_traces): Allow copying of blocks within
13881 2017-07-20 Richard Biener <rguenther@suse.de>
13883 * gimple.h (gimple_phi_result): Add gphi * overload.
13884 (gimple_phi_result_ptr): Likewise.
13885 (gimple_phi_arg): Likewise. Adjust index assert to only
13886 allow actual argument accesses rather than all slots available
13888 (gimple_phi_arg_def): Add gphi * overload.
13889 * tree-phinodes.c (make_phi_node): Initialize only actual
13891 (resize_phi_node): Clear memory not covered by old node,
13892 do not initialize excess argument slots.
13893 (reserve_phi_args_for_new_edge): Initialize new argument slot
13896 2017-07-20 Bin Cheng <bin.cheng@arm.com>
13898 PR tree-optimization/81388
13900 2016-07-21 Bin Cheng <bin.cheng@arm.com>
13902 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
13903 by removing computation of may_be_zero.
13905 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
13906 Tom de Vries <tom@codesourcery.com>
13908 PR middle-end/81030
13909 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
13910 when gimple level profile disagrees with what RTL expander did.
13912 2017-07-20 Richard Biener <rguenther@suse.de>
13914 PR tree-optimization/61171
13915 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
13916 (vect_analyze_stmt): Add slp instance parameter.
13917 (vectorizable_reduction): Likewise.
13918 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
13919 (vect_is_simple_reduction): Deal with chains not detected
13920 as SLP reduction chain, specifically not properly associated
13921 chains containing a mix of plus/minus.
13922 (get_reduction_op): Remove.
13923 (get_initial_defs_for_reduction): Simplify, pass in whether
13924 this is a reduction chain, pass in the SLP node for the PHIs.
13925 (vect_create_epilog_for_reduction): Get the SLP instance as
13927 (vectorizable_reduction): Get the SLP instance as arg.
13928 During analysis remember the SLP node with the PHIs in the
13929 instance. Simplify getting at the vectorized reduction PHIs.
13930 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
13931 through SLP instance.
13932 (vect_slp_analyze_operations): Likewise.
13933 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
13934 (vect_transform_stmt): Likewise.
13936 2017-07-20 Tom de Vries <tom@codesourcery.com>
13938 PR tree-optimization/81489
13939 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
13940 read of phi arg location to before loop that modifies phi.
13942 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13944 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
13947 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
13949 PR middle-end/81331
13950 * except.c (execute): Fix ordering issue.
13952 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
13954 PR rtl-optimization/81423
13955 * combine.c (make_compound_operation_int): Don't try to optimize
13956 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
13958 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
13960 PR rtl-optimization/81423
13961 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
13962 with a constant that is -1 in the truncated to mode.
13964 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
13966 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
13967 (determine_unlikely_bbs): ... here.
13968 * predict.h (propagate_unlikely_bbs_forward): Declare.
13969 * cfgexpand.c (pass_expand::execute): Use it.
13970 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
13972 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
13973 propagate_unlikely_bbs_forward.
13975 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
13977 PR middle-end/81331
13978 * except.c (maybe_add_nop_after_section_switch): New function.
13981 2017-07-19 Tom de Vries <tom@codesourcery.com>
13983 * gimple.h (gimple_phi_set_arg): Make assert more strict.
13985 2017-07-19 Tom de Vries <tom@codesourcery.com>
13987 * gimple.h (gimple_phi_arg): Make assert more strict.
13989 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
13991 * config.gcc (powerpc*-*-*): Add mmintrin.h.
13992 * config/rs6000/mmintrin.h: New file.
13993 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
13995 2017-07-19 Jakub Jelinek <jakub@redhat.com>
13997 PR tree-optimization/81346
13998 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
14000 2017-07-19 Tom de Vries <tom@codesourcery.com>
14002 * config/nvptx/nvptx.md (VECIM): Add V2DI.
14004 2017-07-19 Tom de Vries <tom@codesourcery.com>
14006 * config/nvptx/nvptx-modes.def: Add V2DImode.
14007 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
14008 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
14009 (nvptx_output_mov_insn): Handle lack of mov.b128.
14010 (nvptx_print_operand): Handle 'H' and 'L' codes.
14011 (nvptx_vector_mode_supported): Allow V2DImode.
14012 (nvptx_preferred_simd_mode): New function.
14013 (nvptx_data_alignment): New function.
14014 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
14015 nvptx_preferred_simd_mode.
14016 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
14018 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
14020 2017-07-19 Tom de Vries <tom@codesourcery.com>
14022 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
14023 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
14024 (nvptx_vector_mode_supported): New function. Allow V2SImode.
14025 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
14026 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
14027 (mov<VECIM>_insn): New define_insn.
14028 (define_expand "mov<VECIM>): New define_expand.
14030 2017-07-19 Tom de Vries <tom@codesourcery.com>
14032 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
14034 2017-07-19 Jakub Jelinek <jakub@redhat.com>
14036 PR tree-optimization/81346
14037 * fold-const.h (fold_div_compare, range_check_type): Declare.
14038 * fold-const.c (range_check_type): New function.
14039 (build_range_check): Use range_check_type.
14040 (fold_div_compare): No longer static, rewritten into
14041 a match.pd helper function.
14042 (fold_comparison): Don't call fold_div_compare here.
14043 * match.pd (X / C1 op C2): New optimization using fold_div_compare
14044 as helper function.
14046 2017-07-19 Nathan Sidwell <nathan@acm.org>
14048 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
14049 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
14050 * tree.c (find_decls_types_r, verify_type): Use
14051 TYPE_{MIN,MAX}_VALUE_RAW.
14052 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
14053 (hash_tree): Likewise.
14054 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
14056 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
14059 2017-07-18 Tom de Vries <tom@codesourcery.com>
14061 PR middle-end/81464
14062 * omp-expand.c (expand_omp_for_static_chunk): Handle
14063 equal-argument loop exit phi.
14065 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
14068 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
14069 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
14070 operand 2 predicate.
14071 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
14072 operand 2 predicate.
14073 (ror,rol -> rorx splitters): Use const_int_operand as
14074 operand 2 predicate.
14076 2017-06-18 Richard Biener <rguenther@suse.de>
14078 PR tree-optimization/81410
14079 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
14080 the gap in the ! slp_perm SLP case after each group.
14082 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
14084 PR middle-end/81463
14085 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
14088 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
14090 PR middle-end/81462
14091 * predict.c (set_even_probabilities): Cleanup; do not affect
14092 probabilities that are already known.
14093 (combine_predictions_for_bb): Call even when count is set.
14095 2017-07-18 Nathan Sidwell <nathan@acm.org>
14097 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
14100 2017-07-18 Bin Cheng <bin.cheng@arm.com>
14103 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
14104 optimization for loop niter analysis.
14106 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
14109 * config/avr/avr.c (avr_optimize_casesi): Don't use
14110 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
14112 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
14114 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
14115 body_cost_vec from _vect_peel_extended_info.
14116 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
14117 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
14120 2017-07-18 Bin Cheng <bin.cheng@arm.com>
14122 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
14124 2017-07-18 Richard Biener <rguenther@suse.de>
14126 PR tree-optimization/80620
14127 PR tree-optimization/81403
14128 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
14129 info when re-using a VN table entry.
14131 2017-07-18 Richard Biener <rguenther@suse.de>
14133 PR tree-optimization/81418
14134 * tree-vect-loop.c (vectorizable_reduction): Properly compute
14135 vectype_in. Verify that with lane-reducing reduction operations
14136 we have a single def-use cycle.
14138 2017-07-17 Carl Love <cel@us.ibm.com>
14140 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
14142 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14143 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14144 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14145 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14146 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14147 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
14148 VMULOSW): New enum "unspec" values.
14149 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
14150 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
14151 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
14152 altivec_vmulosw): New patterns.
14153 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14154 VMULOSW): Add definitions.
14155 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
14157 * config/alpha/alpha.c: Include predict.h.
14159 2017-07-17 Yury Gribov <tetra2005@gmail.com>
14161 * tree-vrp.c (compare_assert_loc): Fix comparison function
14162 to return predictable results.
14164 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14166 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
14168 (subdi3): Likewise.
14169 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
14170 * doc/invoke.texi (mexpand-adddi): Update text.
14172 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14174 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
14175 that also clobbers the CC register. The old expand code is moved
14177 (*arc_clzsi2): ... here.
14178 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
14179 the CC register. The old expand code is moved to ...
14180 (arc_ctzsi2): ... here.
14182 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14184 * config/arc/arc.opt (mindexed-loads): Use initial value
14185 TARGET_INDEXED_LOADS_DEFAULT.
14186 (mauto-modify-reg): Use initial value
14187 TARGET_AUTO_MODIFY_REG_DEFAULT.
14188 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
14189 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
14190 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
14191 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
14193 2017-07-17 Martin Liska <mliska@suse.cz>
14196 * opts.c (finish_options): Do not allow -fgnu-tm
14197 w/ -fsanitize={kernel-,}address. Say sorry.
14199 2017-07-17 Bin Cheng <bin.cheng@arm.com>
14202 * tree-loop-distribution.c (classify_partition): Only assert on
14203 numer of iterations.
14204 (merge_dep_scc_partitions): Delete prameter. Update function call.
14205 (distribute_loop): Remove code handling loop with unknown niters.
14206 (pass_loop_distribution::execute): Skip loop with unknown niters.
14208 2017-07-17 Bin Cheng <bin.cheng@arm.com>
14211 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
14212 function sort_partitions_by_post_order.
14214 2017-07-17 Bin Cheng <bin.cheng@arm.com>
14216 PR tree-optimization/81374
14217 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
14218 the max index of basic blocks, rather than number of basic blocks.
14220 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14222 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
14224 (arc_legitimate_pic_operand_p): Likewise.
14225 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
14227 (arc_needs_pcl_p): Likewise.
14228 (arc_legitimate_pc_offset_p): Likewise.
14229 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
14230 function is also used in constrains.md.
14231 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
14232 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
14233 PLUS. Only return true/false in known cases, otherwise assert.
14234 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
14235 is already called in arc_legitimate_constant_p.
14236 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
14238 (LEGITIMATE_PIC_OPERAND_P): Use
14239 arc_raw_symbolic_reference_mentioned_p function.
14240 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
14245 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14246 Andrew Burgess <andrew.burgess@embecosm.com>
14248 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
14249 (arc_return_address_register): New function.
14250 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
14251 (arc_handle_fndecl_attribute): Add naked attribute.
14252 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
14253 (TARGET_WARN_FUNC_RETURN): Likewise.
14254 (arc_allocate_stack_slots_for_args): New function.
14255 (arc_warn_func_return): Likewise.
14256 (machine_function): Change type fn_type.
14257 (arc_compute_function_type): Consider new naked function type,
14258 change function return type.
14259 (arc_must_save_register): Adapt to handle new
14260 arc_compute_function_type's return type.
14261 (arc_expand_prologue): Likewise.
14262 (arc_expand_epilogue): Likewise.
14263 (arc_return_address_regs): Delete.
14264 (arc_return_address_register): New function.
14265 (arc_epilogue_uses): Use above function.
14266 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
14267 (arc_function_type): Change encoding, add naked type.
14268 (ARC_INTERRUPT_P): Change to handle the new encoding.
14269 (ARC_FAST_INTERRUPT_P): Likewise.
14270 (ARC_NORMAL_P): Define.
14271 (ARC_NAKED_P): Likewise.
14272 (arc_compute_function_type): Delete prototype.
14273 * config/arc/arc.md (in_ret_delay_slot): Use
14274 arc_return_address_register function.
14275 (simple_return): Likewise.
14276 (p_return_i): Likewise.
14278 2017-07-17 Jakub Jelinek <jakub@redhat.com>
14280 PR tree-optimization/81428
14281 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
14282 can't be built for those types.
14284 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
14286 Remove stuff dead since r239246.
14288 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
14289 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
14290 (avr_inform_devices): Remove dead stuff.
14292 2017-07-17 Tamar Christina <tamar.christina@arm.com>
14294 * config/arm/arm_neon.h: Fix softp typo.
14296 2017-07-17 Jakub Jelinek <jakub@redhat.com>
14298 PR tree-optimization/81365
14299 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
14300 aggregate moves onto bb predecessor edges, make sure there are no
14301 loads that could alias the lhs in between the start of bb and the
14304 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
14307 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
14308 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
14309 [LSHIFTRT, outer_code = TRUNCATE]: Same.
14311 2017-07-17 Jakub Jelinek <jakub@redhat.com>
14313 PR tree-optimization/81396
14314 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
14315 (init_symbolic_number): Initialize it to 1.
14316 (perform_symbolic_merge): Add n_ops from both operands into the new
14318 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
14319 without base_addr as useless if they need more than one operation.
14320 (bswap_replace): Handle !bswap case for NULL base_addr.
14322 2017-07-17 Tom de Vries <tom@codesourcery.com>
14325 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
14328 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
14330 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
14331 conditional builtin define __FIX_LEON3FT_B2BST.
14333 2017-07-17 Daniel Cederman <cederman@gaisler.com>
14335 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
14336 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
14339 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
14341 PR rtl-optimization/81424
14342 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
14343 to remove potential trapping from operands if -fnon-call-exceptions.
14345 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14347 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
14348 profile_proability for scalling.
14349 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
14351 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14353 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
14355 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14357 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
14358 fixpoint arithmetics.
14360 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14362 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
14363 fixpoint arithmetics.
14365 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14367 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
14368 fixpoint arithmetics.
14370 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14372 * profile-count.h (profile_probability::from_reg_br_prob_note,
14373 profile_probability::to_reg_br_prob_note): New functions.
14374 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
14375 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
14376 * predict.c (probability_reliable_p): Update.
14377 (edge_probability_reliable_p): Update.
14378 (br_prob_note_reliable_p): Update.
14379 (invert_br_probabilities): Update.
14380 (add_reg_br_prob_note): New function.
14381 (combine_predictions_for_insn): Update.
14382 * asan.c (asan_clear_shadow): Update.
14383 * cfgbuild.c (compute_outgoing_frequencies): Update.
14384 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
14385 (update_br_prob_note): Update.
14386 (rtl_verify_edges): Update.
14387 (purge_dead_edges): Update.
14388 (fixup_reorder_chain): Update.
14389 * emit-rtl.c (try_split): Update.
14390 * ifcvt.c (cond_exec_process_insns): Update.
14391 (cond_exec_process_if_block): Update.
14392 (dead_or_predicable): Update.
14393 * internal-fn.c (expand_addsub_overflow): Update.
14394 (expand_neg_overflow): Update.
14395 (expand_mul_overflow): Update.
14396 * loop-doloop.c (doloop_modify): Update.
14397 * loop-unroll.c (compare_and_jump_seq): Update.
14398 * optabs.c (emit_cmp_and_jump_insn_1): Update.
14399 * predict.h: Update.
14400 * reorg.c (mostly_true_jump): Update.
14402 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
14403 * config/alpha/alpha.c (emit_unlikely_jump): Update.
14404 * config/arc/arc.c: (emit_unlikely_jump): Update.
14405 * config/arm/arm.c: (emit_unlikely_jump): Update.
14406 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
14407 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
14408 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
14409 (ix86_print_operand): Update.
14410 (ix86_split_fp_branch): Update.
14411 (predict_jump): Update.
14412 * config/ia64/ia64.c (ia64_print_operand): Update.
14413 * config/mmix/mmix.c (mmix_print_operand): Update.
14414 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
14415 (rs6000_expand_split_stack_prologue): Update.
14416 * config/rs6000/rs6000.c: Update.
14417 * config/s390/s390.c (s390_expand_vec_strlen): Update.
14418 (s390_expand_vec_movstr): Update.
14419 (s390_expand_cs_tdsi): Update.
14420 (s390_expand_split_stack_prologue): Update.
14421 * config/sh/sh.c (sh_print_operand): Update.
14422 (expand_cbranchsi4): Update.
14423 (expand_cbranchdi4): Update.
14424 * config/sparc/sparc.c (output_v9branch): Update.
14425 * config/spu/spu.c (get_branch_target): Update.
14426 (ea_load_store_inline): Update.
14427 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
14428 * config/tilepro/tilepro.c: Update.
14430 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
14432 * gimplify.c (mostly_copy_tree_r): Revert latest change.
14433 (gimplify_save_expr): Likewise.
14435 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14437 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
14439 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14441 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
14443 * timevar.def (TV_IPA_FNSUMMARY): Define.
14445 2017-07-16 Daniel Cederman <cederman@gaisler.com>
14447 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
14448 to back store errata sensitive sequence from being generated.
14449 (sqrtdf2_fix): Likewise.
14451 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14453 * tree-ssa-threadupdate.c (compute_path_counts,
14454 update_joiner_offpath_counts): Use profile_probability.
14456 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
14459 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
14461 * config/arm/arm-c.c (arm_cpu_builtins): Define
14462 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
14464 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
14466 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14467 array entries to represent __ieee128 versions of the
14468 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
14469 scalar_extract_sig, and scalar_insert_exp built-in functions.
14470 (altivec_resolve_overloaded_builtin): Add special case handling
14471 for the __builtin_scalar_insert_exp function, as represented by
14472 the P9V_BUILTIN_VEC_VSIEDP constant.
14473 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
14474 exponent support for __ieee128 argument.
14475 (VSESQP): Add scalar extract signature support for __ieee128
14477 (VSTDCNQP): Add scalar test negative support for __ieee128
14479 (VSIEQP): Add scalar insert exponent support for __int128 argument
14480 with __ieee128 result.
14481 (VSIEQPF): Add scalar insert exponent support for __ieee128
14482 argument with __ieee128 result.
14483 (VSTDCQP): Add scalar test data class support for __ieee128
14485 (VSTDCNQP): Add overload support for scalar test negative with
14486 __ieee128 argument.
14487 (VSTDCQP): Add overload support for scalar test data class
14488 __ieee128 argument.
14489 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
14490 UNSPEC_VSX_SXSIGDP.
14491 (UNSPEC_VSX_SIEXPQP): New constant.
14492 (xsxexpqp): New insn for VSX scalar extract exponent quad
14494 (xsxsigqp): New insn for VSX scalar extract significand quad
14496 (xsiexpqpf): New insn for VSX scalar insert exponent quad
14497 precision with floating point argument.
14498 (xststdcqp): New expand for VSX scalar test data class quad
14500 (xststdcnegqp): New expand for VSX scalar test negative quad
14502 (xststdcqp): New insn to match expansions for VSX scalar test data
14503 class quad precision and VSX scalar test negative quad precision.
14504 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
14505 special case operand checking to enforce that second operand of
14506 VSX scalar test data class with quad precision argument is a 7-bit
14508 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
14509 prototypes and descriptions of __ieee128 versions of
14510 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
14511 scalar_test_data_class, and scalar_test_neg built-in functions.
14513 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14515 PR tree-optimization/81162
14516 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
14517 replace a negate with an add.
14519 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
14521 * doc/invoke.texi (arm/-mcpu): Document +crypto.
14523 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
14525 * config/arm/arm-c.c (arm_cpu_builtins): Define
14526 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
14528 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
14530 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
14531 (armv8-r): Set ARM Cortex-R52 as default CPU.
14532 * config/arm/arm-tables.opt: Regenerate.
14533 * config/arm/arm-tune.md: Regenerate.
14534 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
14536 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
14537 extension for -mcpu=cortex-r52.
14539 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
14541 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
14542 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
14543 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
14544 (fp-armv8): Define it as FP_ARMv8 only.
14545 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
14546 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
14548 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
14549 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
14550 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
14551 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
14552 than TARGET_FPU_ARMV8.
14553 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
14554 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
14555 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
14557 * config/arm/neon.md (neon_vrint): Likewise.
14558 (neon_vcvt): Likewise.
14559 (neon_<fmaxmin_op><mode>): Likewise.
14560 (<fmaxmin><mode>3): Likewise.
14561 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
14562 * config/arm/predicates.md (arm_cond_move_operator): Check against
14563 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
14565 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
14567 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
14568 to top of function.
14570 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14572 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
14573 loop in comment with memset.
14575 2017-07-14 Martin Liska <mliska@suse.cz>
14577 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
14578 * dwarf2out.c (is_java): Remove the function.
14579 (output_pubname): Remove usage of the function.
14580 (lower_bound_default): Remove usage of DW_LANG_Java.
14581 (gen_compile_unit_die): Likewise.
14582 * gcc.c: Remove compiler defaults for .java and .zip files.
14583 * gimple-expr.c (remove_suffix): Change as there's no longer
14584 extension than 4-letter one.
14585 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
14586 (gimplify_save_expr): Likewise.
14587 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
14588 as it's possible even for other languages than Java.
14589 * langhooks.h (struct lang_hooks): Remove Java from a comment.
14590 * lto-opts.c (lto_write_options): Remove reference to Java.
14591 * opts.c (strip_off_ending): Update file extension handling.
14592 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
14593 * tree-eh.c (lower_resx): Likewise.
14594 * tree.c (free_lang_data_in_type): Remove dead code.
14595 (find_decls_types_r): Likewise.
14596 (build_common_builtin_nodes): Remove Java from a comment.
14597 (verify_type): Remove dead code.
14598 * varasm.c (assemble_external): Remove Java from a comment.
14600 2017-07-14 Martin Liska <mliska@suse.cz>
14602 * opts.c (finish_options): Add quotes.
14603 (common_handle_option): Likewise.
14605 2017-07-14 Martin Liska <mliska@suse.cz>
14607 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
14608 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
14609 Remove N_SO_PASCAL.
14610 * dwarf2out.c (lower_bound_default): Do not handle
14612 (gen_compile_unit_die): Likewise.
14613 * gcc.c: Remove default extension binding for GNU Pascal.
14614 * stmt.c: Remove Pascal language from a comment.
14615 * xcoffout.c: Likewise.
14617 2017-07-13 David Malcolm <dmalcolm@redhat.com>
14620 * diagnostic-show-locus.c (fixit_cmp): New function.
14621 (layout::layout): Sort m_fixit_hints.
14622 (column_range::column_range): Assert that the values are valid.
14623 (struct char_span): New struct.
14624 (correction::overwrite): New method.
14625 (struct source_line): New struct.
14626 (line_corrections::add_hint): Add assertions. Reimplement memcpy
14627 calls in terms of classes source_line and char_span, and
14628 correction::overwrite.
14629 (selftest::test_overlapped_fixit_printing_2): New function.
14630 (selftest::diagnostic_show_locus_c_tests): Call it.
14632 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
14634 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
14635 early if there is no lhs.
14637 2017-07-13 Martin Liska <mliska@suse.cz>
14639 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
14640 (gen_reference_type_die): Likewise.
14641 * stor-layout.c: Remove Pascal-related comment.
14643 2017-07-13 Martin Liska <mliska@suse.cz>
14645 * opts.c (finish_options): Add quotes to error messages.
14646 (parse_sanitizer_options): Likewise.
14648 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
14650 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
14652 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
14654 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
14656 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
14658 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
14660 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
14662 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
14665 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
14666 provides the hardware capability bits, define the macro
14667 __BUILTIN_CPU_SUPPORTS__.
14668 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
14669 if GLIBC does not provide the hardware capability bits. Add a
14670 gcc_unreachable call if the built-in cpu function is neither
14671 __builtin_cpu_is nor __builtin_cpu_supports.
14672 (rs6000_get_function_versions_dispatcher): Change the warning
14673 that an old GLIBC is used which does not export the capability
14674 bits to be an error.
14675 * doc/extend.texi (target_clones attribute): Document the
14676 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
14677 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
14678 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
14679 the macros defined by GCC if the newer GLIBC is available.
14681 2017-07-12 Jeff Law <law@redhat.com>
14683 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
14684 remaining includes slightly.
14685 * config/riscv/riscv-builtins.c: Include profile-count.h.
14687 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
14690 * config/avr/avr.c (avr_set_current_function): In diagnostic
14691 messages: Quote keywords and (parts of) identifiers.
14692 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
14695 2017-07-12 Carl Love <cel@us.ibm.com>
14697 * config/rs6000/rs6000-c.c: Add support for built-in functions
14698 vector bool char vec_revb (vector bool char);
14699 vector bool short vec_revb (vector short char);
14700 vector bool int vec_revb (vector bool int);
14701 vector bool long long vec_revb (vector bool long long);
14702 * doc/extend.texi: Update the built-in documentation file for the
14703 new built-in functions.
14705 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14707 * config/s390/s390.md: Remove movcc splitter.
14709 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14711 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
14712 load/store on condition.
14714 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
14717 * config/avr/avr.c (avr_encode_section_info)
14718 [progmem && !TREE_READONLY]: Error if progmem object needs
14721 2017-07-11 Michael Collison <michael.collison@arm.com>
14723 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
14726 2017-07-11 Carl Love <cel@us.ibm.com>
14728 * config/rs6000/rs6000-c.c: Add support for builtins
14729 vector unsigned int vec_parity_lsbb (vector signed int);
14730 vector unsigned int vec_parity_lsbb (vector unsigned int);
14731 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
14732 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
14733 vector unsigned long long vec_parity_lsbb (vector signed long long);
14734 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
14735 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
14736 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
14737 * doc/extend.texi: Update the built-in documentation file for the
14738 new built-in functions.
14740 2017-07-11 David Malcolm <dmalcolm@redhat.com>
14742 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
14743 (layout::m_primary_loc): New field.
14744 (layout::layout): Initialize new field. Move location filtering
14745 logic from here to...
14746 (layout::maybe_add_location_range): ...this new method. Add
14747 support for filtering to just the lines already specified by other
14749 (layout::will_show_line_p): New method.
14750 (gcc_rich_location::add_location_if_nearby): New method.
14751 (selftest::test_add_location_if_nearby): New test function.
14752 (selftest::diagnostic_show_locus_c_tests): Call it.
14753 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
14756 2017-07-11 Tom de Vries <tom@codesourcery.com>
14758 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
14759 (bb_first_real_insn): New function.
14760 (nvptx_single): Add extra initialization of broadcasted condition
14763 2017-07-11 Nathan Sidwell <nathan@acm.org>
14765 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
14767 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
14769 * doc/extend.texi (AVR Function Attributes): Remove weblink to
14770 Binutils doc as TEXI will mess them up.
14771 * doc/invoke.texi (AVR Options): Same here.
14773 2017-07-11 Daniel Cederman <cederman@gaisler.com>
14775 * config/sparc/sparc.opt (mfix-ut700): New option.
14776 (mfix-gr712rc): Likewise.
14777 (sparc_fix_b2bst): New variable.
14778 * doc/invoke.texi (SPARC options): Document them.
14779 (ARM options): Fix warnings.
14780 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
14781 instructions to prevent sequences that can trigger the store-store
14782 errata for certain LEON3FT processors.
14783 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
14784 (sparc_option_override): Set sparc_fix_b2bst appropriately.
14785 * config/sparc/sparc.md (fix_b2bst): New attribute.
14786 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
14788 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
14791 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
14793 (*rsqrtsf2_sse): Ditto.
14795 (div<mode>3): Macroize insn from divdf3 and divsf3
14796 using MODEF mode iterator.
14798 2017-07-10 Martin Sebor <msebor@redhat.com>
14800 PR tree-optimization/80397
14801 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
14802 instead of testing for equality to INTEGER_TYPE.
14804 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
14806 * config.gcc: Remove uclibc from arc target spec.
14808 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
14810 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
14812 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14815 * lto-wrapper.c (remove_option): New function.
14816 (merge_and_complain): Merge PIC/PIE options more realistically.
14818 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
14820 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
14824 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
14825 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
14826 * config.in: Regenerate.
14827 * configure: Regenerate.
14828 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
14829 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
14830 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
14831 (TARGET_GASISR_PROLOGUES): ...target mask.
14832 * common/config/avr/avr-common.c
14833 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
14834 Set -mgas-isr-prologues.
14835 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
14836 INSERT_PASS_BEFORE for it.
14837 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
14838 * config/avr/avr.c (avr_option_override)
14839 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
14840 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
14841 (avr_attribute_table) <no_gccisr>: Add new function attribute.
14842 (avr_set_current_function) <is_no_gccisr>: Init machine field.
14843 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
14845 (make_avr_pass_pre_proep): New function.
14846 (emit_push_sfr) <treg>: Add argument to function and use it
14847 instead of TMP_REG.
14848 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
14849 and set machine->gasisr.yes.
14850 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
14851 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
14852 __gcc_isr.n_pushed to .L__stack_usage.
14853 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
14854 (avr_asm_final_postscan_insn): ...this new static function.
14855 * config/avr/avr.h (machine_function)
14856 <is_no_gccisr, use_L__stack_usage>: New fields.
14857 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
14858 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
14859 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
14860 (gasisr, *gasisr): New expander and insn.
14861 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
14862 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
14863 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
14865 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
14867 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
14870 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
14872 Move jump-tables out of .text again.
14875 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
14876 (ASM_OUTPUT_ADDR_VEC): New function.
14877 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
14878 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
14879 INSN_ADDRESSes as asm comment.
14880 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
14881 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
14882 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
14883 * config/avr/avr.md (*tablejump): Adjust comment.
14884 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
14885 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
14887 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
14888 (avr_output_addr_vec): New proto.
14889 (avr_log_t) <insn_addresses>: New field.
14891 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
14894 * config/i386/i386.c (ix86_function_arg_advance): Set
14895 outgoing_args_on_stack to true if there are outgoing arguments
14897 (ix86_function_arg): Likewise.
14898 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
14899 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
14900 * config/i386/i386.h (machine_function): Add
14901 outgoing_args_on_stack.
14903 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
14905 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
14906 supporting pthreds.
14907 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
14909 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
14911 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
14912 (REAL_H): Remove $(MACHMODE_H).
14913 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
14915 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
14916 $(MACHMODE_H) and double-int.h.
14917 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
14919 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
14922 2017-07-07 Andrew Pinski <apinski@cavium.com>
14924 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
14925 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
14927 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14929 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
14930 Add warning if GCC was not configured to link against a GLIBC that
14931 exports the hardware capability bits.
14932 (make_resolver_func): Make resolver function private and not a
14933 COMDAT function. Create the name with clone_function_name instead
14934 of make_unique_name.
14937 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
14938 correct operand in doing the split.
14940 2017-07-07 Carl Love <cel@us.ibm.com>
14942 * config/rs6000/rs6000-c: Add support for built-in function
14943 vector unsigned short vec_pack_to_short_fp32 (vector float,
14945 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
14946 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
14947 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
14948 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
14949 (convert_4f32_8i16): Add define_expand.
14950 * doc/extend.texi: Update the built-in documentation file for the
14951 new built-in function.
14953 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14955 * config/sparc/m8.md: New file.
14956 * config/sparc/sparc.md: Include m8.md.
14958 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14960 * config/sparc/sparc.opt: New option -mvis4b.
14961 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
14962 (sparc_option_override): Handle VIS4B.
14963 (enum sparc_builtins): Define
14964 SPARC_BUILTIN_DICTUNPACK{8,16,32},
14965 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
14966 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
14967 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
14968 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
14969 (check_constant_argument): New function.
14970 (sparc_vis_init_builtins): Define builtins
14971 __builtin_vis_dictunpack{8,16,32},
14972 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
14973 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
14974 __builtin_vis_fpcmpde{8,16,32}shl and
14975 __builtin_vis_fpcmpur{8,16,32}shl.
14976 (sparc_expand_builtin): Check that the constant operands to
14977 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
14978 constant and in range.
14979 * config/sparc/sparc-c.c (sparc_target_macros): Handle
14981 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
14982 (SPARC_IMM5_P): Likewise.
14983 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
14984 (enabled): Handle vis4b.
14985 (UNSPEC_DICTUNPACK): New unspec.
14986 (UNSPEC_FPCMPSHL): Likewise.
14987 (UNSPEC_FPUCMPSHL): Likewise.
14988 (UNSPEC_FPCMPDESHL): Likewise.
14989 (UNSPEC_FPCMPURSHL): Likewise.
14990 (cpu_feature): New CPU feature `vis4b'.
14991 (dictunpack{8,16,32}): New insns.
14992 (FPCSMODE): New mode iterator.
14993 (fpcscond): New code iterator.
14994 (fpcsucond): Likewise.
14995 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
14996 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
14997 (fpcmpde{8,16,32}{si,di}shl): Likewise.
14998 (fpcmpur{8,16,32}{si,di}shl): Likewise.
14999 * config/sparc/constraints.md: Define constraints `q' for unsigned
15000 2-bit integer constants and `t' for unsigned 5-bit integer
15002 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
15004 (imm5_operand_dictunpack16): Likewise.
15005 (imm5_operand_dictunpack32): Likewise.
15006 (imm2_operand): Likewise.
15007 * doc/invoke.texi (SPARC Options): Document -mvis4b.
15008 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
15009 ditunpack* and fpcmp*shl builtins.
15011 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15013 * config.gcc: Handle m8 in --with-{cpu,tune} options.
15014 * config.in: Add HAVE_AS_SPARC6 define.
15015 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
15017 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
15019 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
15020 (CPP_CPU_SPEC): Handle m8.
15021 (ASM_CPU_SPEC): Likewise.
15022 * config/sparc/sparc-opts.h (enum processor_type): Add
15024 * config/sparc/sparc.c (m8_costs): New struct.
15025 (sparc_option_override): Handle TARGET_CPU_m8.
15026 (sparc32_initialize_trampoline): Likewise.
15027 (sparc64_initialize_trampoline): Likewise.
15028 (sparc_issue_rate): Likewise.
15029 (sparc_register_move_cost): Likewise.
15030 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
15031 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
15032 (ASM_CPU64_DEFAULT_SPEC): Likewise.
15033 (CPP_CPU_SPEC): Handle M8.
15034 (ASM_CPU_SPEC): Likewise.
15035 (AS_M8_FLAG): Define.
15036 * config/sparc/sparc.md: Add m8 to the cpu attribute.
15037 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
15038 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
15040 * configure: Regenerate.
15041 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
15044 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15046 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
15048 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
15049 ("*movdi_insn_sp32"): Do not set v3pipe.
15050 ("*movsi_insn"): Likewise.
15051 ("*movdi_insn_sp64"): Likewise.
15052 ("*movsf_insn"): Likewise.
15053 ("*movdf_insn_sp32"): Likewise.
15054 ("*movdf_insn_sp64"): Likewise.
15055 ("*zero_extendsidi2_insn_sp64"): Likewise.
15056 ("*sign_extendsidi2_insn"): Likewise.
15057 ("*mov<VM32:mode>_insn"): Likewise.
15058 ("*mov<VM64:mode>_insn_sp64"): Likewise.
15059 ("*mov<VM64:mode>_insn_sp32"): Likewise.
15060 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
15061 ("<vlop:code><VL:mode>3"): Likewise.
15062 ("*not_<vlop:code><VL:mode>3"): Likewise.
15063 ("*nand<VL:mode>_vis"): Likewise.
15064 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
15065 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
15066 ("one_cmpl<VL:mode>2"): Likewise.
15067 ("faligndata<VM64:mode>_vis"): Likewise.
15068 ("alignaddrsi_vis"): Likewise.
15069 ("alignaddrdi_vis"): Likweise.
15070 ("alignaddrlsi_vis"): Likewise.
15071 ("alignaddrldi_vis"): Likewise.
15072 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
15073 ("bmaskdi_vis"): Likewise.
15074 ("bmasksi_vis"): Likewise.
15075 ("bshuffle<VM64:mode>_vis"): Likewise.
15076 ("cmask8<P:mode>_vis"): Likewise.
15077 ("cmask16<P:mode>_vis"): Likewise.
15078 ("cmask32<P:mode>_vis"): Likewise.
15079 ("pdistn<P:mode>_vis"): Likewise.
15080 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
15082 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15084 * config/sparc/sparc.md ("subtype"): New insn attribute.
15085 ("*wrgsr_sp64"): Set insn subtype.
15086 ("*rdgsr_sp64"): Likewise.
15087 ("alignaddrsi_vis"): Likewise.
15088 ("alignaddrdi_vis"): Likewise.
15089 ("alignaddrlsi_vis"): Likewise.
15090 ("alignaddrldi_vis"): Likewise.
15091 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
15092 ("fexpand_vis"): Likewise.
15093 ("fpmerge_vis"): Likewise.
15094 ("faligndata<VM64:mode>_vis"): Likewise.
15095 ("bshuffle<VM64:mode>_vis"): Likewise.
15096 ("cmask8<P:mode>_vis"): Likewise.
15097 ("cmask16<P:mode>_vis"): Likewise.
15098 ("cmask32<P:mode>_vis"): Likewise.
15099 ("fchksm16_vis"): Likewise.
15100 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
15101 ("fmean16_vis"): Likewise.
15102 ("fp<plusminus_insn>64_vis"): Likewise.
15103 ("<plusminus_insn>v8qi3"): Likewise.
15104 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
15105 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
15106 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
15107 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
15108 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
15109 ("*movqi_insn"): Likewise.
15110 ("*movhi_insn"): Likewise.
15111 ("*movsi_insn"): Likewise.
15112 ("movsi_pic_gotdata_op"): Likewise.
15113 ("*movdi_insn_sp32"): Likewise.
15114 ("*movdi_insn_sp64"): Likewise.
15115 ("movdi_pic_gotdata_op"): Likewise.
15116 ("*movsf_insn"): Likewise.
15117 ("*movdf_insn_sp32"): Likewise.
15118 ("*movdf_insn_sp64"): Likewise.
15119 ("*zero_extendhisi2_insn"): Likewise.
15120 ("*zero_extendqihi2_insn"): Likewise.
15121 ("*zero_extendqisi2_insn"): Likewise.
15122 ("*zero_extendqidi2_insn"): Likewise.
15123 ("*zero_extendhidi2_insn"): Likewise.
15124 ("*zero_extendsidi2_insn_sp64"): Likewise.
15125 ("ldfsr"): Likewise.
15126 ("prefetch_64"): Likewise.
15127 ("prefetch_32"): Likewise.
15128 ("tie_ld32"): Likewise.
15129 ("tie_ld64"): Likewise.
15130 ("*tldo_ldub_sp32"): Likewise.
15131 ("*tldo_ldub1_sp32"): Likewise.
15132 ("*tldo_ldub2_sp32"): Likewise.
15133 ("*tldo_ldub_sp64"): Likewise.
15134 ("*tldo_ldub1_sp64"): Likewise.
15135 ("*tldo_ldub2_sp64"): Likewise.
15136 ("*tldo_ldub3_sp64"): Likewise.
15137 ("*tldo_lduh_sp32"): Likewise.
15138 ("*tldo_lduh1_sp32"): Likewise.
15139 ("*tldo_lduh_sp64"): Likewise.
15140 ("*tldo_lduh1_sp64"): Likewise.
15141 ("*tldo_lduh2_sp64"): Likewise.
15142 ("*tldo_lduw_sp32"): Likewise.
15143 ("*tldo_lduw_sp64"): Likewise.
15144 ("*tldo_lduw1_sp64"): Likewise.
15145 ("*tldo_ldx_sp64"): Likewise.
15146 ("*mov<VM32:mode>_insn"): Likewise.
15147 ("*mov<VM64:mode>_insn_sp64"): Likewise.
15148 ("*mov<VM64:mode>_insn_sp32"): Likewise.
15150 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15152 * config/sparc/sparc.md ("type"): New insn type viscmp.
15153 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
15155 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
15156 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
15157 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
15158 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
15160 ("n7_vis_logical_11cycle"): Likewise.
15161 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
15162 * config/sparc/niagara2.md ("niag3_vis": Likewise.
15163 * config/sparc/niagara.md ("niag_vis"): Likewise.
15164 * config/sparc/ultra3.md ("us3_fga"): Likewise.
15165 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
15167 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15169 * config/sparc/sparc.md: New instruction type `bmask'.
15170 (bmaskdi_vis): Use the `bmask' type.
15171 (bmasksi_vis): Likewise.
15172 * config/sparc/ultra3.md (us3_array): Likewise.
15173 * config/sparc/niagara7.md (n7_array): Likewise.
15174 * config/sparc/niagara4.md (n4_array): Likewise.
15175 * config/sparc/niagara2.md (niag2_vis): Likewise.
15176 (niag3_vis): Likewise.
15177 * config/sparc/niagara.md (niag_vis): Likewise.
15179 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15181 * ipa-comdats.c: Remove optimize check from gate.
15182 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
15183 for functions not optimized.
15184 (ipa_fn_summary_read): Skip optimize check.
15185 (ipa_fn_summary_write): Likewise.
15186 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
15188 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
15190 (can_inline_edge_p): Check flag_pcc_struct_return for match.
15191 (check_callers): Give up on caller which is not optimized.
15192 (inline_small_functions): Likewise.
15193 (ipa_inline): Do not give up when not optimizing.
15194 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
15195 away unoptimizes cdtors.
15196 (whole_program_function_and_variable_visibility): Do
15197 ipa_discover_readonly_nonaddressable_vars in LTO mode.
15198 * ipa.c (process_references): Do not check optimize.
15199 (symbol_table::remove_unreachable_nodes): Update optimize check.
15200 (set_writeonly_bit): Update optimize check.
15201 (pass_ipa_cdtor_merge::gate): Do not check optimize.
15202 (pass_ipa_single_use::gate): Remove.
15204 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15206 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
15207 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
15208 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
15209 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
15210 permute_load, permute_store, adjust_extract, adjust_splat,
15211 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
15212 replace_swap_with_copy, dump_swap_insn_table,
15213 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
15214 recombine_lvx_pattern, recombine_stvx_pattern,
15215 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
15216 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
15217 to file rs6000-p8swap.c.
15218 * config/rs6000/rs6000-p8swap.c: New file.
15219 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
15220 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
15221 and rs6000*-*-* targets.
15223 2017-07-06 David Malcolm <dmalcolm@redhat.com>
15225 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
15227 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15229 * lto-wrapper.c (merge_and_complain): Do not merge
15230 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
15231 fsigned_zeros, ftrapping_math, fwrapv.
15232 (append_compiler_options): Do not track these options.
15233 (append_linker_options): Likewie
15235 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15237 * cgraphunit.c (cgraph_node::finalize_function): When
15238 !flag_toplevel_reorde set no_reorder flag.
15239 (varpool_node::finalize_decl): Likewise.
15240 (symbol_table::compile): Drop no toplevel reorder path.
15242 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15244 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
15245 edges; zero probability is not better than uninitialized.
15247 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
15249 * asan.h (asan_sanitize_allocas_p): Declare.
15250 * asan.c (asan_sanitize_allocas_p): New function.
15251 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
15252 (handle_builtin_alloca): Likewise.
15253 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
15254 if !asan_sanitize_allocas_p.
15255 * params.def (asan-instrument-allocas): Add new option.
15256 * params.h (ASAN_PROTECT_ALLOCAS): Define.
15257 * opts.c (common_handle_option): Disable allocas sanitization for
15260 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
15262 * asan.c: Include gimple-fold.h.
15263 (get_last_alloca_addr): New function.
15264 (handle_builtin_stackrestore): Likewise.
15265 (handle_builtin_alloca): Likewise.
15266 (asan_emit_allocas_unpoison): Likewise.
15267 (get_mem_refs_of_builtin_call): Add new parameter, remove const
15268 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
15269 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
15270 (instrument_builtin_call): Pass gimple iterator to
15271 get_mem_refs_of_builtin_call.
15272 (last_alloca_addr): New global.
15273 * asan.h (asan_emit_allocas_unpoison): Declare.
15274 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
15275 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
15276 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
15277 if function calls alloca.
15278 * gimple-fold.c (replace_call_with_value): Remove static keyword.
15279 * gimple-fold.h (replace_call_with_value): Declare.
15280 * internal-fn.c: Include asan.h.
15281 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
15282 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
15284 2017-07-06 David Malcolm <dmalcolm@redhat.com>
15286 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
15287 (C_SELFTEST_FLAGS): New.
15288 (CPP_SELFTEST_FLAGS): New.
15289 (SELFTEST_DEPS): New, from deps of s-selftest.
15290 (C_SELFTEST_DEPS): New, from deps of s-selftest.
15291 (CPP_SELFTEST_DEPS): New.
15292 (selftest): Add dependency on s-selftest-c++.
15293 (s-selftest): Rename to...
15294 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
15295 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
15296 than SELFTEST_FLAGS.
15297 (selftest-gdb): Rename to...
15298 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
15300 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
15301 (selftest-valgrind): Rename to...
15302 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
15304 (selftest-valgrind): Reintroduce as an alias for
15305 selftest-c-valgrind.
15306 (s-selftest-c++): New.
15307 (selftest-c++-gdb): New.
15308 (selftest-c++-valgrind): New.
15310 2017-07-06 Olivier Hainque <hainque@adacore.com>
15312 * gcc.c (process_command): When deciding if undefined variables
15313 should be ignored when processing specs, accept "gcc -v" as well.
15315 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15317 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
15318 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
15320 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
15322 * config/arm/arm-cpus.in (armv8-r): Add new entry.
15323 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
15324 * config/arm/arm-tables.opt: Regenerate.
15325 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
15327 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
15329 2017-07-06 Carl Love <cel@us.ibm.com>
15331 * ChangeLog: Clean up from mid air collision
15333 2017-07-06 Carl Love <cel@us.ibm.com>
15335 * config/rs6000/rs6000-c.c: Add support for built-in functions
15336 vector signed int vec_subc (vector signed int, vector signed int);
15337 vector signed __int128 vec_subc (vector signed __int128,
15338 vector signed __int128);
15339 vector unsigned __int128 vec_subc (vector unsigned __int128,
15340 vector unsigned __int128);
15341 vector signed int vec_sube (vector signed int, vector signed int,
15342 vector signed int);
15343 vector unsigned int vec_sube (vector unsigned int,
15344 vector unsigned int,
15345 vector unsigned int);
15346 vector signed __int128 vec_sube (vector signed __int128,
15347 vector signed __int128,
15348 vector signed__int128);
15349 vector unsigned __int128 vec_sube (vector unsigned __int128,
15350 vector unsigned __int128,
15351 vector unsigned __int128);
15352 vector signed int vec_subec (vector signed int, vector signed int,
15353 vector signed int);
15354 vector unsigned int vec_subec (vector unsigned int,
15355 vector unsigned int,
15356 vector unsigned int);
15357 vector signed __int128 vec_subec (vector signed __int128,
15358 vector signed __int128,
15359 vector signed__int128);
15360 vector unsigned __int128 vec_subec (vector unsigned __int128,
15361 vector unsigned __int128,
15362 vector unsigned __int128);
15363 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
15364 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
15365 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
15366 BU_ALTIVEC_OVERLOAD_X definitions.
15367 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
15368 * doc/extend.texi: Update the built-in documentation file for the new
15369 built-in functions.
15371 2017-07-06 David Malcolm <dmalcolm@redhat.com>
15374 * diagnostic-show-locus.c (layout::layout): Use start and finish
15375 spelling location for the start and finish of each range.
15376 * genmatch.c (linemap_client_expand_location_to_spelling_point):
15377 Add unused aspect param.
15378 * input.c (expand_location_1): Add "aspect" param, and use it
15379 to access the correct part of the location.
15380 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
15382 (expand_location_to_spelling_point): Likewise.
15383 (linemap_client_expand_location_to_spelling_point): Add "aspect"
15384 param, and pass it to expand_location_1.
15386 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
15388 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
15389 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
15390 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
15391 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
15392 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
15393 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
15394 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
15395 _mm_maskz_getmant_ss): New intrinsics.
15396 (__builtin_ia32_getexpss128_mask): Changed to ...
15397 __builtin_ia32_getexpss128_round ... this.
15398 (__builtin_ia32_getexpsd128_mask): Changed to ...
15399 __builtin_ia32_getexpsd128_round ... this.
15400 * config/i386/i386-builtin-types.def
15401 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
15402 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
15403 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
15404 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
15405 __builtin_ia32_getmantss_mask_round): New builtins.
15406 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
15407 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
15408 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
15409 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
15410 * config/i386/sse.md
15411 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
15412 avx512f_sgetexp<mode><mask_scalar_name>
15413 <round_saeonly_scalar_name> ... this.
15414 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
15415 %0, %1, %2<round_saeonly_op3>}): Changed to ...
15416 vgetexp<ssescalarmodesuffix>
15417 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
15418 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
15419 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
15420 avx512f_vgetmant<mode><mask_scalar_name>
15421 <round_saeonly_scalar_name> ... this.
15422 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
15423 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
15424 vgetmant<ssescalarmodesuffix>
15425 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
15426 %0<mask_scalar_operand4>, %1, %2
15427 <round_saeonly_scalar_mask_op4>, %3} ... this.
15428 * config/i386/subst.md (mask_scalar_operand4,
15429 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
15430 round_saeonly_scalar_nimm_predicate): New subst attributes.
15432 2017-07-06 Julia Koval <julia.koval@intel.com>
15434 * config/i386/i386.c (ix86_erase_embedded_rounding):
15435 Remove code for old rounding pattern.
15437 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
15439 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
15441 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
15443 * doc/sourcebuild.texi (Test Directives, Variants of
15444 dg-require-support): Add documentation for dg-require-stack-check.
15446 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
15448 * config/i386/subst.md (mask_scalar, round_scalar,
15449 round_saeonly_scalar): New meta-templates.
15450 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
15451 round_scalar_mask_operand3, round_scalar_mask_op3,
15452 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
15453 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
15454 round_saeonly_scalar_constraint,
15455 round_saeonly_scalar_prefix): New subst attribute.
15456 * config/i386/sse.md
15457 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
15458 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
15459 <round_scalar_name> ... this.
15460 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
15461 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
15462 <round_scalar_name> ... this.
15463 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
15464 <sse>_vm<code><mode>3<mask_scalar_name>
15465 <round_saeonly_scalar_name> ... this.
15466 (v<plusminus_mnemonic><ssescalarmodesuffix>
15467 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
15468 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
15469 v<plusminus_mnemonic><ssescalarmodesuffix>
15470 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
15471 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
15472 (v<multdiv_mnemonic><ssescalarmodesuffix>
15473 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
15474 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
15475 v<multdiv_mnemonic><ssescalarmodesuffix>
15476 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
15477 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
15478 (v<maxmin_float><ssescalarmodesuffix>
15479 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
15480 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
15481 v<maxmin_float><ssescalarmodesuffix>
15482 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
15483 %0<mask_scalar_operand3>, %1, %<iptr>2
15484 <round_saeonly_scalar_mask_op3>} ... this.
15486 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
15488 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
15489 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
15491 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
15492 Alan Hayward <alan.hayward@arm.com>
15493 David Sherwood <david.sherwood@arm.com>
15495 * combine.c (simplify_if_then_else): Remove "enum" before
15497 * compare-elim.c (can_eliminate_compare): Likewise.
15498 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
15500 (aarch64_lookup_simd_builtin_type): Likewise.
15501 (aarch64_simd_builtin_type): Likewise.
15502 (aarch64_init_simd_builtin_types): Likewise.
15503 (aarch64_simd_expand_args): Likewise.
15504 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
15506 (aarch64_reverse_mask): Likewise.
15507 (aarch64_simd_emit_reg_reg_move): Likewise.
15508 (aarch64_gen_adjusted_ldpstp): Likewise.
15509 (aarch64_ccmp_mode_to_code): Likewise.
15510 (aarch64_operands_ok_for_ldpstp): Likewise.
15511 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15512 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
15514 (aarch64_min_divisions_for_recip_mul): Likewise.
15515 (aarch64_reassociation_width): Likewise.
15516 (aarch64_get_condition_code_1): Likewise.
15517 (aarch64_simd_emit_reg_reg_move): Likewise.
15518 (aarch64_simd_attr_length_rglist): Likewise.
15519 (aarch64_reverse_mask): Likewise.
15520 (aarch64_operands_ok_for_ldpstp): Likewise.
15521 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15522 (aarch64_gen_adjusted_ldpstp): Likewise.
15523 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
15525 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
15526 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
15527 (arm_lookup_simd_builtin_type): Likewise.
15528 (arm_simd_builtin_type): Likewise.
15529 (arm_init_simd_builtin_types): Likewise.
15530 (arm_expand_builtin_args): Likewise.
15531 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
15532 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
15533 (ft32_setup_incoming_varargs): Likewise.
15534 (ft32_function_arg): Likewise.
15535 (ft32_function_arg_advance): Likewise.
15536 (ft32_pass_by_reference): Likewise.
15537 (ft32_arg_partial_bytes): Likewise.
15538 (ft32_valid_pointer_mode): Likewise.
15539 (ft32_addr_space_pointer_mode): Likewise.
15540 (ft32_addr_space_legitimate_address_p): Likewise.
15541 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
15543 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
15544 (ix86_emit_outlined_ms2sysv_restore): Likewise.
15545 (iamcu_alignment): Likewise.
15546 (canonicalize_vector_int_perm): Likewise.
15547 (ix86_noce_conversion_profitable_p): Likewise.
15548 (ix86_mpx_bound_mode): Likewise.
15549 (ix86_operands_ok_for_move_multiple): Likewise.
15550 * config/microblaze/microblaze-protos.h
15551 (microblaze_expand_conditional_branch_reg): Likewise.
15552 * config/microblaze/microblaze.c
15553 (microblaze_expand_conditional_branch_reg): Likewise.
15554 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
15556 (rs6000_reassociation_width): Likewise.
15557 (rs6000_invalid_binary_op): Likewise.
15558 (fusion_p9_p): Likewise.
15559 (emit_fusion_p9_load): Likewise.
15560 (emit_fusion_p9_store): Likewise.
15561 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
15563 (riscv_hard_regno_mode_ok_p): Likewise.
15564 (riscv_address_insns): Likewise.
15565 (riscv_split_symbol): Likewise.
15566 (riscv_legitimize_move): Likewise.
15567 (riscv_function_value): Likewise.
15568 (riscv_hard_regno_nregs): Likewise.
15569 (riscv_expand_builtin): Likewise.
15570 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
15571 (riscv_build_integer): Likewise.
15572 (riscv_split_integer): Likewise.
15573 (riscv_legitimate_constant_p): Likewise.
15574 (riscv_cannot_force_const_mem): Likewise.
15575 (riscv_regno_mode_ok_for_base_p): Likewise.
15576 (riscv_valid_base_register_p): Likewise.
15577 (riscv_valid_offset_p): Likewise.
15578 (riscv_valid_lo_sum_p): Likewise.
15579 (riscv_classify_address): Likewise.
15580 (riscv_legitimate_address_p): Likewise.
15581 (riscv_address_insns): Likewise.
15582 (riscv_load_store_insns): Likewise.
15583 (riscv_force_binary): Likewise.
15584 (riscv_split_symbol): Likewise.
15585 (riscv_force_address): Likewise.
15586 (riscv_legitimize_address): Likewise.
15587 (riscv_move_integer): Likewise.
15588 (riscv_legitimize_const_move): Likewise.
15589 (riscv_legitimize_move): Likewise.
15590 (riscv_address_cost): Likewise.
15591 (riscv_subword): Likewise.
15592 (riscv_output_move): Likewise.
15593 (riscv_canonicalize_int_order_test): Likewise.
15594 (riscv_emit_int_order_test): Likewise.
15595 (riscv_function_arg_boundary): Likewise.
15596 (riscv_pass_mode_in_fpr_p): Likewise.
15597 (riscv_pass_fpr_single): Likewise.
15598 (riscv_pass_fpr_pair): Likewise.
15599 (riscv_get_arg_info): Likewise.
15600 (riscv_function_arg): Likewise.
15601 (riscv_function_arg_advance): Likewise.
15602 (riscv_arg_partial_bytes): Likewise.
15603 (riscv_function_value): Likewise.
15604 (riscv_pass_by_reference): Likewise.
15605 (riscv_setup_incoming_varargs): Likewise.
15606 (riscv_print_operand): Likewise.
15607 (riscv_elf_select_rtx_section): Likewise.
15608 (riscv_save_restore_reg): Likewise.
15609 (riscv_for_each_saved_reg): Likewise.
15610 (riscv_register_move_cost): Likewise.
15611 (riscv_hard_regno_mode_ok_p): Likewise.
15612 (riscv_hard_regno_nregs): Likewise.
15613 (riscv_class_max_nregs): Likewise.
15614 (riscv_memory_move_cost): Likewise.
15615 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
15616 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
15617 (rl78_addr_space_address_mode): Likewise.
15618 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15620 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
15621 (rs6000_reassociation_width): Likewise.
15622 (rs6000_invalid_binary_op): Likewise.
15623 (fusion_p9_p): Likewise.
15624 (emit_fusion_p9_load): Likewise.
15625 (emit_fusion_p9_store): Likewise.
15626 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
15627 (ok_for_simple_move_operands): Likewise.
15628 (ok_for_simple_move_strict_operands): Likewise.
15629 (ok_for_simple_arith_logic_operands): Likewise.
15630 (visium_legitimize_reload_address): Likewise.
15631 (visium_select_cc_mode): Likewise.
15632 (output_cbranch): Likewise.
15633 (visium_split_double_move): Likewise.
15634 (visium_expand_copysign): Likewise.
15635 (visium_expand_int_cstore): Likewise.
15636 (visium_expand_fp_cstore): Likewise.
15637 * config/visium/visium.c (visium_pass_by_reference): Likewise.
15638 (visium_function_arg): Likewise.
15639 (visium_function_arg_advance): Likewise.
15640 (visium_libcall_value): Likewise.
15641 (visium_setup_incoming_varargs): Likewise.
15642 (visium_legitimate_constant_p): Likewise.
15643 (visium_legitimate_address_p): Likewise.
15644 (visium_legitimize_address): Likewise.
15645 (visium_secondary_reload): Likewise.
15646 (visium_register_move_cost): Likewise.
15647 (visium_memory_move_cost): Likewise.
15648 (prepare_move_operands): Likewise.
15649 (ok_for_simple_move_operands): Likewise.
15650 (ok_for_simple_move_strict_operands): Likewise.
15651 (ok_for_simple_arith_logic_operands): Likewise.
15652 (visium_function_value_1): Likewise.
15653 (rtx_ok_for_offset_p): Likewise.
15654 (visium_legitimize_reload_address): Likewise.
15655 (visium_split_double_move): Likewise.
15656 (visium_expand_copysign): Likewise.
15657 (visium_expand_int_cstore): Likewise.
15658 (visium_expand_fp_cstore): Likewise.
15659 (visium_split_cstore): Likewise.
15660 (visium_select_cc_mode): Likewise.
15661 (visium_split_cbranch): Likewise.
15662 (output_cbranch): Likewise.
15663 (visium_print_operand_address): Likewise.
15664 * expmed.c (flip_storage_order): Likewise.
15665 * expmed.h (emit_cstore): Likewise.
15666 (flip_storage_order): Likewise.
15667 * genrecog.c (validate_pattern): Likewise.
15668 * hsa-gen.c (gen_hsa_addr): Likewise.
15669 * internal-fn.c (expand_arith_overflow): Likewise.
15670 * ira-color.c (allocno_copy_cost_saving): Likewise.
15671 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15672 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
15673 (process_invariant_for_inheritance): Likewise.
15674 * lra-eliminations.c (move_plus_up): Likewise.
15675 * omp-low.c (lower_oacc_reductions): Likewise.
15676 * simplify-rtx.c (simplify_subreg): Likewise.
15677 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
15678 (TARGET_CHKP_BOUND_MODE): Likewise..
15679 * targhooks.c (default_chkp_bound_mode): Likewise.
15680 (default_setup_incoming_vararg_bounds): Likewise.
15681 * targhooks.h (default_chkp_bound_mode): Likewise.
15682 (default_setup_incoming_vararg_bounds): Likewise.
15683 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
15684 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
15685 (have_whole_vector_shift): Likewise.
15686 * tree-vect-stmts.c (vectorizable_load): Likewise.
15687 * doc/tm.texi: Regenerate.
15689 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
15691 Graceful degrade if Binutils PR21472 is not available.
15694 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
15695 .rodata in flash test fails.
15696 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
15697 * confgure: Regenerate.
15698 * config.in: Regenerate.
15699 * config/avr/avr.c (avr_asm_named_section)
15700 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
15701 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
15702 (avr_asm_init_sections): Same.
15704 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15706 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
15707 (fma<VH:mode>4_intrinsic): Likewise.
15708 (*fmsub<VCVTF:mode>4): Likewise.
15709 (*fmsub<VH:mode>4_intrinsic): Likewise.
15711 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
15714 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
15715 Don't depend on "optimize > 0".
15716 (out_movhi_r_mr, out_movqi_mr_r): Same.
15717 (out_movhi_mr_r, out_movqi_r_mr): Same.
15718 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
15719 io_address_operand on "optimize > 0".
15721 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15723 * tree-loop-distribution.c: Add general explanantion on the pass.
15724 (generate_loops_for_partition): Mark distributed loop.
15725 (pg_add_dependence_edges): New parameter. Handle alias data
15726 dependence specially and record it in the parameter if asked.
15727 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
15728 (init_partition_graph_vertices, add_partition_graph_edge): New.
15729 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
15730 (free_partition_graph_vdata, build_partition_graph): New.
15731 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
15732 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
15733 (data_ref_segment_size, latch_dominated_by_data_ref): New.
15734 (compute_alias_check_pairs, version_loop_by_alias_check): New.
15735 (version_for_distribution_p, finalize_partitions): New.
15736 (distribute_loop): Handle alias data dependence specially. Factor
15737 out loop fusion code as functions and call these functions.
15739 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15741 * tree-loop-distribution.c (classify_partition): New parameter and
15742 better handle reduction statement.
15743 (rdg_build_partitions): Revise comment.
15744 (distribute_loop): Compute statements in all partitions and pass it
15745 to classify_partition.
15747 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15749 * tree-loop-distribution.c (enum partition_type): New.
15750 (struct partition): New field type.
15751 (partition_merge_into): Add parameter. Update partition type.
15752 (data_dep_in_cycle_p, update_type_for_merge): New functions.
15753 (build_rdg_partition_for_vertex): Compute partition type.
15754 (rdg_build_partitions): Dump partition type.
15755 (distribute_loop): Update calls to partition_merge_into.
15757 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15759 * tree-loop-distribution.c (struct ddr_hasher): New.
15760 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
15762 (classify_partition): Call get_data_dependence.
15763 (pg_add_dependence_edges): Ditto.
15764 (distribute_loop): Release data dependence hash table.
15766 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15768 * tree-loop-distribution.c (ref_base_address): Delete.
15769 (similar_memory_accesses): Rename ...
15770 (share_memory_accesses): ... to this. Check if partitions access
15771 the same memory reference.
15772 (distribute_loop): Call share_memory_accesses.
15774 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15776 * tree-loop-distribution.c (struct partition): New field recording
15777 its data reference.
15778 (partition_alloc, partition_free): Init and release data refs.
15779 (partition_merge_into): Merge data refs.
15780 (build_rdg_partition_for_vertex): Collect data refs for partition.
15781 (pg_add_dependence_edges): Change parameters from vector to bitmap.
15783 (distribute_loop): Remve data refs from vertice data of partition
15786 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15788 * tree-loop-distribution.c (params.h): Include header file.
15789 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
15790 (datarefs_vec): New global var.
15791 (create_rdg_vertices): Use datarefs_vec directly.
15792 (free_rdg): Don't free data references.
15793 (build_rdg): Update use. Don't free data references.
15794 (distribute_loop): Compute global variable for data references.
15795 Bail out if there are too many data references.
15797 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15799 * tree-loop-distribution.c (loop_nest): New global var.
15800 (build_rdg): Use loop directly, rather than loop nest.
15801 (pg_add_dependence_edges): Remove loop nest parameter. Use global
15803 (distribute_loop): Compute global variable loop nest. Update use.
15805 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15807 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
15808 (partition_merge_into): New parameter. Dump reason for fusion.
15809 (distribute_loop): Update use of partition_merge_into.
15811 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15813 * tree-loop-distribution.c (bb_top_order_index): New.
15814 (bb_top_order_index_size, bb_top_order_cmp): New.
15815 (stmts_from_loop): Use topological order.
15816 (pass_loop_distribution::execute): Compute and release topological
15817 order for basic blocks.
15819 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15821 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
15824 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15826 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
15827 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
15828 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
15829 * internal-fn.def (LOOP_DIST_ALIAS): New.
15830 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
15831 (fold_loop_internal_call): ... this.
15832 (vect_loop_dist_alias_call): New function.
15833 (set_uid_loop_bbs): Call fold_loop_internal_call.
15834 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
15837 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
15840 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
15841 Require dead FLAGS_REG at the beginning of a peephole.
15843 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
15846 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
15847 arguments in the call to __builtin_ia32_sbb_u32.
15848 (_subborrow_u64): Swap _X and _Y arguments in the call to
15849 __builtin_ia32_sbb_u64.
15851 2017-07-04 Jakub Jelinek <jakub@redhat.com>
15854 * tree-vrp.c (compare_assert_loc): Turn into a function template
15855 with stable template parameter. Only test if a->e is NULL,
15856 !a->e == !b->e has been verified already. Use e == NULL or
15857 e != NULL instead of e or ! e tests. If stable is true, don't use
15858 iterative_hash_expr, on the other side allow a or b or both NULL
15859 and sort the NULLs last.
15860 (process_assert_insertions): Sort using compare_assert_loc<false>
15861 instead of compare_assert_loc, later sort using
15862 compare_assert_loc<true> before calling process_assert_insertions_for
15863 in a loop. Use break instead of continue once seen NULL pointer.
15865 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
15867 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
15868 Cortex-R7 and Cortex-R8 processors.
15870 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
15872 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
15873 uninitialized while src is not.
15875 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
15877 * common/config/arm/arm-common.c: Adjust include path for
15879 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
15880 (arm-cpu.h): Create in build directory. Adjust dependency rules.
15881 (arm-cpu-data.h): Likewise.
15882 (arm-cpu-cdata.h): Likewise.
15883 * config/arm/arm-cpu.h: Delete.
15884 * config/arm/arm-cpu-cdata.h: Delete.
15885 * config/arm/arm-cpu-data.h: Delete.
15887 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
15889 * config/arm/arm-cpus.in (cortex-a55): New.
15890 (cortex-a75): Likewise.
15891 (cortex-a75.cortex-a55): Likewise.
15892 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
15894 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
15895 * config/arm/arm-cpu-cdata.h: Regenerate.
15896 * config/arm/arm-cpu-data.h: Regenerate.
15897 * config/arm/arm-cpu.h: Regenerate.
15898 * config/arm/arm-tables.opt: Regenerate.
15899 * config/arm/arm-tune.md: Regenerate.
15901 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
15903 * haifa-sched.c (sched_create_recovery_edges): Update profile.
15905 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
15907 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
15910 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
15912 PR tree-optimization/81292
15913 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
15914 full_string_p, also call adjust_related_strinfos if the adjustment
15915 is simple, otherwise invalidate related strinfos.
15917 2017-07-04 Martin Liska <mliska@suse.cz>
15920 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
15921 newly created variable as DECL_IGNORED_P.
15923 2017-07-04 Martin Liska <mliska@suse.cz>
15926 * ipa-inline.c (inline_small_functions):
15927 Use xstrdup_for_dump.
15929 2017-07-04 Tom de Vries <tom@codesourcery.com>
15931 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
15933 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
15936 * config/darwin.c (darwin_function_switched_text_sections):
15939 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
15941 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
15943 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
15945 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
15947 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15949 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
15950 min_profitable_iters, and th as inclusive lower bounds.
15951 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
15952 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
15953 for min_profitable_iters and min_profitable_estimate.
15954 (vect_transform_loop): Treat th as an inclusive lower bound.
15955 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
15957 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
15960 * config/darwin.c (darwin_function_switched_text_sections):
15961 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
15962 in two pieces, and suppress the use of buf.
15964 2017-07-03 Nathan Sidwell <nathan@acm.org>
15966 * hash-table.h (hash_table_mod1): Fix indentation.
15968 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15970 PR middle-end/81290
15971 * predict.c (force_edge_cold): Be more careful about propagation
15973 * profile-count.h (profile_probability::guessed,
15974 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
15976 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
15978 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
15980 * doc/invoke.texi (rcpc architecture extension): Document it.
15982 2017-07-03 Richard Biener <rguenther@suse.de>
15984 PR tree-optimization/60510
15985 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
15986 the scalar reduction PHI and use it.
15987 (vectorizable_reduction): Properly guard the single_defuse_cycle
15988 path for non-SLP reduction chains where we cannot use it.
15989 Rework reduc_def/index and vector type deduction. Rework
15990 vector operand gathering during reduction op code-gen.
15991 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
15992 chains dissolve the chain and leave it to non-SLP reduction
15995 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15997 * tree-data-ref.h (dr_alignment): Declare.
15998 * tree-data-ref.c (dr_alignment): New function.
15999 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
16000 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
16002 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
16004 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16006 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
16007 and base_misalignment fields.
16008 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
16009 * tree-data-ref.c: Include builtins.h.
16010 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
16011 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
16012 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
16013 * tree-vect-data-refs.c: Include tree-cfg.h.
16014 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
16015 fields instead of calculating an alignment here.
16016 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
16017 innermost_loop_behavior fields.
16019 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16021 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
16023 (DR_STEP_ALIGNMENT): New macro.
16024 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
16025 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
16026 (create_data_ref): Print it.
16027 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
16028 to tell whether the step preserves vector (mis)alignment.
16029 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
16030 Move the check for an integer step and generalise to all INTEGER_CST.
16031 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
16032 Print the outer step alignment.
16034 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16036 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
16037 with offset_alignment.
16038 (DR_ALIGNED_TO): Delete.
16039 (DR_OFFSET_ALIGNMENT): New macro.
16040 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
16041 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
16042 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
16043 (create_data_ref): Likewise.
16044 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
16045 (vect_analyze_data_refs): Likewise.
16046 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
16047 creating dummy innermost behavior.
16049 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16051 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
16052 with a "innermost_loop_behavior *" and refeence tree.
16053 * tree-data-ref.c (dr_analyze_innermost): Likewise.
16054 (create_data_ref): Update call accordingly.
16055 * tree-predcom.c (find_looparound_phi): Likewise.
16057 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16059 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
16060 fields with dr_wrt_vec_loop.
16061 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
16062 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
16063 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
16064 (vect_dr_behavior): New function.
16065 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
16066 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
16067 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
16068 track whether the step preserves the misalignment.
16069 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
16070 Use vect_dr_behavior.
16071 (vect_setup_realignment): Update call accordingly.
16072 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
16073 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
16074 call to vect_create_addr_base_for_vector_ref.
16075 (vect_create_cond_for_align_checks): Likewise.
16076 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
16077 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
16078 (vect_recog_mask_conversion_pattern): Likewise.
16079 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
16080 (new_stmt_vec_info): Remove redundant zeroing.
16082 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
16084 * common/config/arm/arm-common.c (arm_be8_option): New function.
16085 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
16086 (ISA_ARMv6): Add isa_bit_be8.
16087 * config/arm/arm.h (arm_be8_option): Add prototype.
16088 (BE8_SPEC_FUNCTION): New define.
16089 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
16090 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
16091 (mlittle-endian): Similarly.
16092 (mbe8, mbe32): New options.
16093 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
16094 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
16096 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16098 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
16100 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16102 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
16103 (cleanup_tree_cfg_bb): Use it.
16104 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
16106 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
16108 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16111 * loop-doloop.c (add_test): Update profile.
16113 2017-07-03 Martin Liska <mliska@suse.cz>
16116 * sanopt.c (rewrite_usage_of_param): New function.
16117 (sanitize_rewrite_addressable_params): Likewise.
16118 (pass_sanopt::execute): Call rewrite_usage_of_param.
16120 2017-07-03 Richard Biener <rguenther@suse.de>
16122 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
16123 back to using VIEW_CONVERT_EXPR.
16125 2017-07-03 Martin Liska <mliska@suse.cz>
16128 * doc/extend.texi: Document when a resolver function is
16129 generated for target_clones.
16131 2017-07-03 Martin Liska <mliska@suse.cz>
16133 * asan.c (asan_emit_stack_protection): Unpoison just red zones
16134 and shadow memory of auto variables which are subject of
16135 use-after-scope sanitization.
16136 (asan_expand_mark_ifn): Add do set only when is_poison.
16138 2016-07-03 Richard Biener <rguenther@suse.de>
16140 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
16142 (vect_force_simple_reduction): Record reduction def -> phi mapping.
16143 (vectorizable_reduction): Perform reduction PHI creation when
16144 visiting a reduction PHI and adjust and simplify code generation
16145 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
16146 (vect_transform_loop): Visit reduction PHIs.
16147 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
16148 defs into the SLP tree.
16149 (vect_build_slp_tree): Reduction defs terminate the recursion.
16150 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
16152 (vect_get_vec_defs_for_stmt_copy): Export.
16153 (vect_get_vec_defs): Likewise.
16154 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
16156 (vect_get_vec_defs_for_stmt_copy): Declare.
16157 (vect_get_vec_defs): Likewise.
16159 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16161 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
16162 parameter with a "loop" parameter and use it instead of the
16163 loop containing DR_STMT. Don't check simple_iv when doing
16164 BB analysis. Describe the two analysis modes in the comment.
16166 2017-07-03 Tom de Vries <tom@codesourcery.com>
16168 PR tree-optimization/69468
16169 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
16170 (find_same_succ_bb): Handle ignore_edge_flags.
16172 2017-07-03 Tom de Vries <tom@codesourcery.com>
16174 PR tree-optimization/81192
16175 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
16177 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
16179 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
16181 2017-07-03 Tom de Vries <tom@codesourcery.com>
16183 PR tree-optimization/81192
16184 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
16185 BB_SAME_SUCC (bb) == NULL.
16187 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16189 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
16192 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16194 * dumpfile.c: Include profile-count.h
16195 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
16197 (insert_cond_bb): Update profile.
16198 * tree-cfg.h (insert_cond_bb): Update prototype.
16199 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
16200 * tree-dump.c: Do not include tree-cfg.
16202 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16204 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
16206 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16208 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
16211 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16213 * tree-complex.c (expand_complex_div_wide): update profile.
16215 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16216 Alan Hayward <alan.hayward@arm.com>
16217 David Sherwood <david.sherwood@arm.com>
16219 * Makefile.in (MACHMODE_H): Remove insn-modes.h
16220 (CORETYPES_H): New define.
16221 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
16222 (insn-modes-inline.h, s-modes-inline-h): New rules.
16223 (generated_files): Add insn-modes-inline.h.
16224 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
16225 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
16226 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
16227 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
16228 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
16229 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
16230 (build/gencodes.o, build/genconditions.o): Likewise.
16231 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
16232 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
16233 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
16234 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
16235 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
16236 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
16237 * coretypes.h: Include everything up to real.h for generators.
16238 Include insn-modes.h first. Include wide-int-print.h after
16239 wide-int.h. Include insn-modes-inline.h and then machmode.h.
16240 * machmode.h: Don't include insn-modes.h here.
16241 * function-tests.c: Remove includes of signop.h, machmode.h,
16242 double-int.h and wide-int.h.
16244 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
16246 * optc-save-gen.awk: Likewise.
16247 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
16248 * godump.c: Remove include of wide-int-print.h.
16249 * pretty-print.h: Likewise.
16250 * wide-int-print.cc: Likewise.
16251 * wide-int.cc: Likewise.
16252 * hash-map-tests.c: Remove include of signop.h.
16253 * hash-set-tests.c: Likewise.
16254 * rtl-tests.c: Likewise.
16255 * mkconfig.sh: Remove include of machmode.h.
16256 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
16258 (emit_insn_modes_inline_h): ...this new function. Emit the code
16259 into an insn-modes-inline.h header file, adding appropriate
16260 include guards and end comments.
16261 (emit_insn_modes_c_header): Remove include of machmode.h.
16262 (emit_min_insn_modes_c_header): Include coretypes.h rather than
16264 (main): Handle -i flag and call emit_insn_modes_inline_h when
16267 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16269 * tree-ssa-strlen.c (strinfo): Rename the length field to
16270 nonzero_chars. Add a full_string_p field.
16271 (compare_nonzero_chars, zero_length_string_p): New functions.
16272 (get_addr_stridx): Add an offset_out parameter.
16273 Use compare_nonzero_chars.
16274 (get_stridx): Update accordingly. Use compare_nonzero_chars.
16275 (new_strinfo): Update after above changes to strinfo.
16276 (set_endptr_and_length): Set full_string_p.
16277 (get_string_length): Update after above changes to strinfo.
16278 (unshare_strinfo): Update call to new_strinfo.
16279 (maybe_invalidate): Likewise.
16280 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
16281 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
16282 as a uhwi instead of an shwi. Update after above changes to
16283 strinfo and new_strinfo.
16284 (zero_length_string): Assert that chainsi contains full strings.
16285 Use zero_length_string_p. Update call to new_strinfo.
16286 (adjust_related_strinfos): Update after above changes to strinfo.
16287 Copy full_string_p from origsi.
16288 (adjust_last_stmt): Use zero_length_string_p.
16289 (handle_builtin_strlen): Update after above changes to strinfo and
16290 new_strinfo. Install the lhs as the string length if the previous
16291 entry didn't describe a full string.
16292 (handle_builtin_strchr): Update after above changes to strinfo
16294 (handle_builtin_strcpy): Likewise.
16295 (handle_builtin_strcat): Likewise.
16296 (handle_builtin_malloc): Likewise.
16297 (handle_pointer_plus): Likewise.
16298 (handle_builtin_memcpy): Likewise. Track nonzero characters
16299 that aren't necessarily followed by a nul terminator.
16300 (handle_char_store): Likewise.
16302 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16304 PR tree-optimization/80769
16305 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
16306 for malloc and calloc. Document the new invariant that all related
16307 strinfos have delayed lengths or none do.
16308 (verify_related_strinfos): Move earlier in file.
16309 (set_endptr_and_length): New function, split out from...
16310 (get_string_length): ...here. Also set the lengths of related
16312 (zero_length_string): Assert that chainsi has known (rather than
16314 (adjust_related_strinfos): Likewise.
16316 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16318 PR tree-optimization/81136
16319 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
16320 assert that two references with the same misalignment have the same
16321 compile-time misalignment if those compile-time misalignments
16324 2017-07-01 Andi Kleen <ak@linux.intel.com>
16326 * print-tree.c (print_node): Print all attributes.
16328 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16330 * cfg.c (scale_bbs_frequencies): New function.
16331 * cfg.h (scale_bbs_frequencies): Declare it.
16332 * cfgloopanal.c (single_likely_exit): Cleanup.
16333 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
16335 (scale_loop_profile): Likewise.
16336 (loop_version): Likewise.
16337 (create_empty_loop_on_edge): Update.
16338 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
16339 scale_loop_frequencies, scale_loop_profile, loopify,
16340 loop_version): Update prototypes.
16341 * modulo-sched.c (sms_schedule): Update.
16342 * predict.c (unlikely_executed_edge_p): Also check probability.
16343 (probably_never_executed_edge_p): Fix typo.
16344 * tree-if-conv.c (version_loop_for_if_conversion): Update.
16345 * tree-parloops.c (gen_parallel_loop): Update.
16346 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
16347 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
16348 * tree-ssa-loop-split.c (split_loop): Update.
16349 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
16350 * tree-vect-loop-manip.c (vect_do_peeling): Update.
16351 (vect_loop_versioning): Update.
16352 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
16354 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16356 * trans-mem.c (split_bb_make_tm_edge): Update profile.
16358 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16360 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
16361 to keep profile consistent.
16363 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16365 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
16366 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
16367 * profile-count.h (max_safe_multiplier): Make unsigned.
16368 (profile_count::guessed_zero): New.
16370 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16372 * bb-reorder.c (fix_up_crossing_landing_pad,
16373 fix_crossing_conditional_branches): Use make_single_succ_edge
16374 to keep profile consistent.
16376 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16378 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
16381 2017-07-01 Jakub Jelinek <jakub@redhat.com>
16384 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
16385 the right scopes, make sure cond_jump isn't preserved between multiple
16386 iterations. Search for fallthru edge whenever there are 3+ edges and
16387 use find_fallthru_edge for it.
16389 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16391 Patch by Alexander Monakov <amonakov@ispras.ru>
16392 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
16393 probabilities consistently.
16395 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16397 * pa.c (pa_expand_compare_and_swap_loop): Update call of
16398 emit_cmp_and_jump_insns.
16400 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16403 * tree-inline.c (expand_call_inline): Combine profile statuses.
16405 2017-06-30 Andrew Pinski <apinski@cavium.com>
16407 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
16408 fold_stmt returned true.
16410 2017-06-30 Nathan Sidwell <nathan@acm.org>
16412 * ggc.h (empty_string): Delete.
16413 * cfgexpand.c (expand_asm_stmt): Use plain "".
16414 * optabs.c (expand_asm_memory_barrier): Likewise.
16415 * stringpool.c (empty_string): Delete.
16416 (digit_vector, digit_string): Delete.
16417 (ggc_alloc_string): Use plain "", don't optimize single digit
16418 strings. Use ggc_alloc_atomic.
16420 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
16422 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
16423 comparison set and one other set, use the cost of the non-comparison
16426 2017-06-30 Nathan Sidwell <nathan@acm.org>
16428 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
16431 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
16433 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
16434 loops. Remove now unneeded calls to gimple_switch_set_label() that
16435 just set removed labels to NULL_TREE.
16437 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
16439 * tree-ssanames.c (set_range_info_raw): Abstract from ...
16440 (set_range_info): ...here. Only call set_range_info_raw if domain
16442 (set_nonzero_bits): Call set_range_info_raw.
16443 * tree-ssanames.h (set_range_info_raw): New.
16445 2017-06-30 Jakub Jelinek <jakub@redhat.com>
16448 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
16449 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
16450 of nonimmediate_operand and <store_mask_constraint> instead of m
16451 for the input operand. For V8FI iterator, always split if input
16452 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
16453 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
16454 <store_mask_predicate> instead of register_operand and
16455 <store_mask_constraint> instead of v for the input operand. Make
16456 sure both operands aren't MEMs for if not <mask_applied>.
16458 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
16460 * lto-wrapper.c (copy_file) Close both file descriptors before
16463 2017-06-30 Martin Liska <mliska@suse.cz>
16466 * multiple_target.c (create_dispatcher_calls): Make ifunc
16467 also for function that don't have calls or are not referenced.
16469 2017-06-30 Richard Biener <rguenther@suse.de>
16471 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
16472 analyze the first scalar stmt. Move vector type computation
16473 for the BB case here from ...
16474 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
16475 live operation processing in the SLP case properly.
16477 2017-06-30 Richard Biener <rguenther@suse.de>
16479 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
16481 2017-06-30 Martin Liska <mliska@suse.cz>
16484 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
16485 before BUILT_IN_UNWIND_RESUME when ASAN is used.
16487 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
16489 * doc/invoke.texi (AArch64): Add missing options and remove redundant
16492 2017-06-30 Richard Biener <rguenther@suse.de>
16494 PR tree-optimization/81249
16495 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
16496 condition reduction result to original scalar type.
16498 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16500 * profile-count.h (enum profile_quality): Fix typos and whitespace
16503 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16505 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
16506 type for branch probabilities.
16508 2017-06-29 Julian Brown <julian@codesourcery.com>
16509 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16511 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
16512 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
16513 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
16514 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
16516 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16518 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
16519 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
16520 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
16521 CC usage from generic code to here.
16522 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
16523 CC usage into the target macros.
16525 2017-06-29 Maya Rashish <coypu@sdf.org>
16527 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
16530 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16532 * arm/arm-builtins.c: Include profile-count.h
16533 * except.c (sjlj_emit_function_enter): Use
16534 profile_probability::unlikely.
16536 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16538 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
16539 and tocrel_offset be pointer args rather than implicitly using
16541 (legitimate_constant_pool_address_p, rs6000_emit_move,
16542 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
16543 tocrel_offset and use in toc_relative_expr_p call.
16544 (print_operand, print_operand_address): Use static tocrel_base_oac
16545 and tocrel_offset_oac.
16546 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
16549 2017-06-29 Maya Rashish <coypu@sdf.org>
16551 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
16553 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
16555 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
16556 objects, take into account only the alignment of 'op0' and 'mode1' if
16559 2017-06-29 Steve Ellcey <sellcey@cavium.com>
16561 * ccmp.c (ccmp_tree_comparison_p): New function.
16562 (ccmp_candidate_p): Update to use above function.
16563 (get_compare_parts): New function.
16564 (expand_ccmp_next): Update to use new functions.
16565 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
16567 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
16568 take mode as argument.
16569 * ccmp.h (expand_ccmp_expr): Add mode as argument.
16570 * expr.c (expand_expr_real_1): Pass mode as argument.
16572 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
16574 * combine.c (combine_instructions): Print insns to dump_file, together
16577 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16579 * asan.c (asan_emit_stack_protection): Update.
16580 (create_cond_insert_point): Update.
16581 * auto-profile.c (afdo_propagate_circuit): Update.
16582 * basic-block.h (struct edge_def): Turn probability to
16583 profile_probability.
16584 (EDGE_FREQUENCY): Update.
16585 * bb-reorder.c (find_traces_1_round): Update.
16586 (better_edge_p): Update.
16587 (sanitize_hot_paths): Update.
16588 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
16589 (make_single_succ_edge): Update.
16590 (check_bb_profile): Update.
16591 (dump_edge_info): Update.
16592 (update_bb_profile_for_threading): Update.
16593 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
16594 probabilitycount to 0.
16595 * cfgbuild.c (compute_outgoing_frequencies): Update.
16596 * cfgcleanup.c (try_forward_edges): Update.
16597 (outgoing_edges_match): Update.
16598 (try_crossjump_to_edge): Update.
16599 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
16600 (expand_gimple_tailcall): Update.
16601 (construct_init_block): Use make_single_succ_edge.
16602 (construct_exit_block): Use make_single_succ_edge.
16603 * cfghooks.c (verify_flow_info): Update.
16604 (redirect_edge_succ_nodup): Update.
16605 (split_edge): Update.
16606 (account_profile_record): Update.
16607 * cfgloopanal.c (single_likely_exit): Update.
16608 * cfgloopmanip.c (scale_loop_profile): Update.
16609 (set_zero_probability): Remove.
16610 (duplicate_loop_to_header_edge): Update.
16611 * cfgloopmanip.h (loop_version): Update prototype.
16612 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
16613 (force_nonfallthru_and_redirect): Update.
16614 (update_br_prob_note): Update.
16615 (rtl_verify_edges): Update.
16616 (purge_dead_edges): Update.
16617 (rtl_lv_add_condition_to_bb): Update.
16618 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
16619 * cgraphunit.c (init_lowered_empty_function): Update.
16620 (cgraph_node::expand_thunk): Update.
16621 * cilk-common.c: Include profile-count.h
16622 * dojump.c (inv): Remove.
16623 (jumpifnot): Update.
16624 (jumpifnot_1): Update.
16625 (do_jump_1): Update.
16627 (do_jump_by_parts_greater_rtx): Update.
16628 (do_compare_rtx_and_jump): Update.
16629 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
16630 do_jump_1. do_compare_rtx_and_jump): Update prototype.
16631 * dwarf2cfi.c: Include profile-count.h
16632 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
16633 (sjlj_emit_dispatch_table): Likewise.
16634 * explow.c: Include profile-count.h
16635 * expmed.c (emit_store_flag_force): Update.
16636 (do_cmp_and_jump): Update.
16637 * expr.c (compare_by_pieces_d::generate): Update.
16638 (compare_by_pieces_d::finish_mode): Update.
16639 (emit_block_move_via_loop): Update.
16640 (store_expr_with_bounds): Update.
16641 (store_constructor): Update.
16642 (expand_expr_real_2): Update.
16643 (expand_expr_real_1): Update.
16644 * expr.h (try_casesi, try_tablejump): Update prototypes.
16645 * gimple-pretty-print.c (dump_probability): Update.
16646 (dump_profile): New.
16647 (dump_gimple_label): Update.
16648 (dump_gimple_bb_header): Update.
16649 * graph.c (draw_cfg_node_succ_edges): Update.
16650 * hsa-gen.c (convert_switch_statements): Update.
16651 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
16652 (find_if_case_1): Update.
16653 (find_if_case_2): Update.
16654 * internal-fn.c (expand_arith_overflow_result_store): Update.
16655 (expand_addsub_overflow): Update.
16656 (expand_neg_overflow): Update.
16657 (expand_mul_overflow): Update.
16658 (expand_vector_ubsan_overflow): Update.
16659 * ipa-cp.c (good_cloning_opportunity_p): Update.
16660 * ipa-split.c (split_function): Use make_single_succ_edge.
16661 * ipa-utils.c (ipa_merge_profiles): Update.
16662 * loop-doloop.c (add_test): Update.
16663 (doloop_modify): Update.
16664 * loop-unroll.c (compare_and_jump_seq): Update.
16665 (unroll_loop_runtime_iterations): Update.
16666 * lra-constraints.c (lra_inheritance): Update.
16667 * lto-streamer-in.c (input_cfg): Update.
16668 * lto-streamer-out.c (output_cfg): Update.
16669 * mcf.c (adjust_cfg_counts): Update.
16670 * modulo-sched.c (sms_schedule): Update.
16671 * omp-expand.c (expand_omp_for_init_counts): Update.
16672 (extract_omp_for_update_vars): Update.
16673 (expand_omp_ordered_sink): Update.
16674 (expand_omp_for_ordered_loops): Update.
16675 (expand_omp_for_generic): Update.
16676 (expand_omp_for_static_nochunk): Update.
16677 (expand_omp_for_static_chunk): Update.
16678 (expand_cilk_for): Update.
16679 (expand_omp_simd): Update.
16680 (expand_omp_taskloop_for_outer): Update.
16681 (expand_omp_taskloop_for_inner): Update.
16682 * omp-simd-clone.c (simd_clone_adjust): Update.
16683 * optabs.c (expand_doubleword_shift): Update.
16684 (expand_abs): Update.
16685 (emit_cmp_and_jump_insn_1): Update.
16686 (expand_compare_and_swap_loop): Update.
16687 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
16688 * predict.c (predictable_edge_p): Update.
16689 (edge_probability_reliable_p): Update.
16690 (set_even_probabilities): Update.
16691 (combine_predictions_for_insn): Update.
16692 (combine_predictions_for_bb): Update.
16693 (propagate_freq): Update.
16694 (estimate_bb_frequencies): Update.
16695 (force_edge_cold): Update.
16696 * profile-count.c (profile_count::dump): Add missing space into dump.
16697 (profile_count::debug): Add newline.
16698 (profile_count::differs_from_p): Explicitly convert to unsigned.
16699 (profile_count::stream_in): Update.
16700 (profile_probability::dump): New member function.
16701 (profile_probability::debug): New member function.
16702 (profile_probability::differs_from_p): New member function.
16703 (profile_probability::differs_lot_from_p): New member function.
16704 (profile_probability::stream_in): New member function.
16705 (profile_probability::stream_out): New member function.
16706 * profile-count.h (profile_count_quality): Rename to ...
16707 (profile_quality): ... this one.
16708 (profile_probability): New.
16709 (profile_count): Update.
16710 * profile.c (compute_branch_probabilities): Update.
16711 * recog.c (peep2_attempt): Update.
16712 * sched-ebb.c (schedule_ebbs): Update.
16713 * sched-rgn.c (find_single_block_region): Update.
16714 (compute_dom_prob_ps): Update.
16715 (schedule_region): Update.
16716 * sel-sched-ir.c (compute_succs_info): Update.
16717 * stmt.c (struct case_node): Update.
16718 (do_jump_if_equal): Update.
16719 (get_outgoing_edge_probs): Update.
16720 (conditional_probability): Update.
16721 (emit_case_dispatch_table): Update.
16722 (expand_case): Update.
16723 (expand_sjlj_dispatch_table): Update.
16724 (emit_case_nodes): Update.
16725 * targhooks.c: Update.
16726 * tracer.c (better_p): Update.
16727 (find_best_successor): Update.
16728 * trans-mem.c (expand_transaction): Update.
16729 * tree-call-cdce.c: Update.
16730 * tree-cfg.c (gimple_split_edge): Upate.
16731 (move_sese_region_to_fn): Upate.
16732 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
16733 * tree-eh.c (lower_resx): Upate.
16734 (cleanup_empty_eh_move_lp): Upate.
16735 * tree-if-conv.c (version_loop_for_if_conversion): Update.
16736 * tree-inline.c (copy_edges_for_bb): Update.
16737 (copy_cfg_body): Update.
16738 * tree-parloops.c (gen_parallel_loop): Update.
16739 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
16740 (gimple_gen_time_profiler): Update.
16741 * tree-ssa-dce.c (remove_dead_stmt): Update.
16742 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
16743 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
16744 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
16745 (unloop_loops): Update.
16746 (try_peel_loop): Update.
16747 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
16748 * tree-ssa-loop-split.c (connect_loops): Update.
16749 (split_loop): Update.
16750 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
16751 (hoist_guard): Update.
16752 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
16753 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
16754 (value_replacement): Update.
16755 * tree-ssa-reassoc.c (branch_fixup): Update.
16756 * tree-ssa-tail-merge.c (replace_block_by): Update.
16757 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
16758 (create_edge_and_update_destination_phis): Update.
16759 (compute_path_counts): Update.
16760 (recompute_probabilities): Update.
16761 (update_joiner_offpath_counts): Update.
16762 (freqs_to_counts_path): Update.
16763 (duplicate_thread_path): Update.
16764 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
16765 (struct switch_conv_info): Update.
16766 (gen_inbound_check): Update.
16767 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
16768 (vect_do_peeling): Update.
16769 (vect_loop_versioning): Update.
16770 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
16771 (optimize_mask_stores): Update.
16772 * ubsan.c (ubsan_expand_null_ifn): Update.
16773 * value-prof.c (gimple_divmod_fixed_value): Update.
16774 (gimple_divmod_fixed_value_transform): Update.
16775 (gimple_mod_pow2): Update.
16776 (gimple_mod_pow2_value_transform): Update.
16777 (gimple_mod_subtract): Update.
16778 (gimple_mod_subtract_transform): Update.
16779 (gimple_ic): Update.
16780 (gimple_stringop_fixed_value): Update.
16781 (gimple_stringops_transform): Update.
16782 * value-prof.h: Update.
16784 2017-06-29 Carl Love <cel@us.ibm.com>
16786 * config/rs6000/rs6000-c.c: Add support for built-in functions
16787 vector signed int vec_signed (vector float);
16788 vector signed long long vec_signed (vector double);
16789 vector signed int vec_signed2 (vector double, vector double);
16790 vector signed int vec_signede (vector double);
16791 vector signed int vec_signedo (vector double);
16792 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
16793 instruction generator.
16794 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
16795 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
16796 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
16798 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
16799 vunsignede_v2df): Add define_expands.
16800 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
16801 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
16802 VEC_UNSIGNEDO): Add definitions.
16803 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
16804 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
16805 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
16806 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
16807 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
16808 * config/rs6000/altivec.h (vec_signed, vec_signed2,
16809 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
16810 vec_unsignede, vec_unsignedo): Add builtin defines.
16811 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
16813 * doc/extend.texi: Update the built-in documentation file for the
16814 new built-in functions.
16816 2017-06-29 Richard Biener <rguenther@suse.de>
16818 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
16819 reduction chains to LOOP_VINFO_REDUCTIONS.
16820 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
16821 SLP reductions after processing reduction chains.
16823 2017-06-29 Nathan Sidwell <nathan@acm.org>
16825 * builtins.c (fold_builtin_FUNCTION): Use
16826 lang_hooks.decl_printable_name.
16828 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
16830 PR middle-end/81194
16831 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
16832 with only one label.
16833 * stmt.c (expand_case): Assert NCASES is greater than one.
16835 2017-06-29 Richard Biener <rguenther@suse.de>
16837 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
16839 (group_case_labels): Likewise.
16840 (find_taken_edge): Push sanity checking on val to workers...
16841 (find_taken_edge_cond_expr): ... here
16842 (find_taken_edge_switch_expr): ... and here, handle cases
16843 with just a default label.
16844 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
16845 (group_case_labels): Likewise.
16846 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
16847 group_case_labels does anything cleanup the CFG again.
16849 2017-06-29 Bin Cheng <bin.cheng@arm.com>
16851 PR tree-optimization/81196
16852 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
16853 exit condition comparing two IVs.
16855 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
16857 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
16858 profile to the dummy entry at the end of the list of architectures.
16859 * config/arm/arm-cpu-cdata.h: Regenerated.
16861 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16862 Michael Collison <michael.collison@arm.com>
16865 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
16867 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
16868 (*aarch64_reg_<mode>3_minus_mask): New pattern.
16869 (*aarch64_<optab>_reg_di3_mask2): New pattern.
16870 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
16871 of shift when the shift amount is masked with constant equal to
16872 the size of the mode.
16873 * config/aarch64/predicates.md (subreg_lowpart_operator): New
16876 2017-06-29 Martin Liska <mliska@suse.cz>
16878 * config/i386/i386.opt: Change range from [1,5] to [0,5].
16880 2017-06-29 Yury Gribov <tetra2005@gmail.com>
16883 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
16886 (edge_growth_cache_entry::edge_growth_cache_entry): New
16888 (reset_edge_growth_cache): Update to use constructor.
16890 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16892 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
16893 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
16894 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
16896 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
16898 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
16899 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
16901 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
16903 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
16904 (*-linux-uclibc*): Add t-uclibc tmake_file.
16905 * config/t-musl: New.
16906 * config/t-uclibc: New.
16908 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
16910 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
16912 (gen_comm_data): Emit architectural setting of arch_prof.
16913 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
16915 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
16916 (armv8-m.base, armv8-m.main): Likewise.
16917 * arm-protos.h (arm_build_target): Add profile field.
16918 (arch_option): Likewise.
16919 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
16921 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
16922 arm_active_target.profile.
16924 2017-06-28 Richard Biener <rguenther@suse.de>
16926 PR middle-end/81227
16927 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
16928 TYPE_OVERFLOW_WRAPS.
16929 * match.pd (negate_expr_p): Likewise.
16930 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
16931 fold_build2, not fold_binary.
16933 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
16935 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
16936 Convert memory address to Pmode.
16937 (aarch64_print_operand): Assert MEM operands are always Pmode.
16939 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
16942 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
16943 Remove redundant if.
16944 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
16945 * config/arm/aarch-common-protos.h
16946 (aarch_forward_to_shift_is_not_shifted_re): Remove.
16947 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
16949 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
16952 * multiple_target.c (create_dispatcher_calls): Set the default
16953 clone to be static, not public.
16955 2017-06-28 Richard Biener <rguenther@suse.de>
16957 * tree-vect-loop.c (vectorizable_reduction): Move special
16958 cond reduction IV var creation ...
16959 (vect_create_epilog_for_reduction): ... here. Remove induction_index
16960 parameter. Use STMT_VINFO_VECTYPE.
16961 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
16964 2017-06-28 Martin Liska <mliska@suse.cz>
16967 * ipa-visibility.c (non_local_p): Handle visibility.
16969 2017-06-28 Martin Liska <mliska@suse.cz>
16972 * common.opt: Add IntegerRange to various options.
16973 * opt-functions.awk (integer_range_info): New function.
16974 * optc-gen.awk: Add integer_range_info to cl_options struct.
16975 * opts-common.c (decode_cmdline_option): Handle
16976 CL_ERR_INT_RANGE_ARG.
16977 (cmdline_handle_error): Likewise.
16978 * opts.c (print_filtered_help): Show valid interval in
16979 when --help is provided.
16980 * opts.h (struct cl_option): Add range_min and range_max fields.
16981 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
16983 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
16985 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
16986 (x * C EQ/NE y * C): New transformation.
16988 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
16990 * genmultilib (combination_space): Accept '+' in option names.
16992 2017-06-28 Martin Liska <mliska@suse.cz>
16995 * asan.c (instrument_derefs): Bail out inner references
16996 that are hard register variables.
16998 2017-06-28 Jakub Jelinek <jakub@redhat.com>
17001 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
17002 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
17004 2017-06-28 Richard Biener <rguenther@suse.de>
17006 * tree-vectorizer.h (vect_get_vec_defs): Remove.
17007 (vect_get_slp_defs): Adjust.
17008 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
17010 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
17012 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
17013 get_initial_defs_for_reduction instead of vect_get_vec_defs.
17014 (vectorizable_reduction): Adjust.
17015 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
17017 (vect_get_slp_defs): Likewise.
17018 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
17019 (vectorizable_bswap): Adjust.
17020 (vectorizable_call): Likewise.
17021 (vectorizable_conversion): Likewise.
17022 (vectorizable_assignment): Likewise.
17023 (vectorizable_shift): Likewise.
17024 (vectorizable_operation): Likewise.
17025 (vectorizable_store): Likewise.
17026 (vectorizable_condition): Likewise.
17027 (vectorizable_comparison): Likewise.
17029 2017-06-28 Michael Collison <michael.collison@arm.com>
17032 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
17034 (arm_gen_movmemqi): Removed unused variable 'i'.
17035 Convert 'for' loop into 'while' loop.
17036 (arm_expand_prologue): Remove last unnecessary set of insn.
17037 (thumb_pop): Remove unused variable 'pushed_words'.
17038 (thumb_exit): Remove last unnecessary set of regs_to_pop.
17040 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17042 * config/s390/predicates.md: Use s390_rel_address_ok_p.
17043 * config/s390/s390-protos.h: Add prototype of
17044 s390_rel_address_ok_p.
17045 * config/s390/s390.c (s390_got_symbol): New function.
17046 (s390_rel_address_ok_p): New function.
17047 (legitimize_pic_address): Use s390_rel_address_ok_p.
17048 (s390_load_got): Use s390_got_symbol.
17049 (s390_option_override): Issue error if
17050 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
17051 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
17053 * config/s390/s390.opt: New option mpic-data-is-text-relative.
17055 2017-06-27 Andrew Pinski <apinski@cavium.com>
17057 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
17058 (X * copysign (1.0, X)): New pattern.
17059 (X * copysign (1.0, -X)): New pattern.
17060 (copysign (-1.0, CST)): New pattern.
17062 2017-06-27 Joseph Myers <joseph@codesourcery.com>
17064 * genmultilib (combination_space): Remove variable.
17065 Validate reuse rules against regular expression for any sequence
17066 of multilib options in any order.
17068 2017-06-27 Michael Collison <michael.collison@arm.com>
17070 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
17071 call aarch64_split_simd_combine.
17072 * (aarch64_combine_internal<mode>): Delete pattern.
17073 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
17074 Allow register and subreg operands.
17076 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
17078 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
17079 specific need, just fallback on defaults.
17080 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
17082 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
17083 Olivier Hainque <hainque@adacore.com>
17085 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
17087 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
17088 targets. Pick a default if no particular attempt applied.
17089 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
17092 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
17094 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
17095 (x86_64-wrs-vxworks7): Likewise.
17097 2017-06-27 Marek Polacek <polacek@redhat.com>
17100 * ubsan.c (instrument_null): Check get_base_address's result for null.
17102 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
17104 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
17106 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
17108 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
17109 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
17110 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
17111 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
17112 New function types.
17113 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
17114 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
17115 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
17116 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
17117 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
17118 BUILT_IN_FEUPDATEENV): New builtins.
17119 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
17120 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
17121 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
17122 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
17124 (builtin_structptr_types): Adjust size.
17125 * tree.c (builtin_structptr_types): Add four entries.
17127 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
17128 Olivier Hainque <hainque@adacore.com>
17130 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
17131 (TLS_SYM): New local macro, forcing reference to __tls__ on
17132 link command lines for VxWorks 7 RTPs, triggering initialization
17134 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
17135 OS features TLS support, true for RTPs on VxWorks 7.
17136 * config/vxworks.c (vxworks_override_options): Setup emutls
17139 2017-06-27 Jakub Jelinek <jakub@redhat.com>
17141 * predict.c (test_prediction_value_range): Use -1U instead of -1
17142 to avoid narrowing conversion warning.
17143 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
17144 to avoid narrowing conversion warning.
17145 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
17147 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
17149 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
17151 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
17152 64bit configurations.
17153 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
17154 (SIZE_TYPE): Likewise.
17155 * config/vxworks.c (vxworks_emutls_var_fields): Use
17156 long_unsigned_type_node instead of unsigned_type_node as the offset
17157 field type, which is "pointer" mode in emutls.c.
17159 2017-06-27 Jakub Jelinek <jakub@redhat.com>
17162 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
17164 PR middle-end/81207
17165 * gimple-fold.c (replace_call_with_call_and_fold): Handle
17166 gimple_vuse copying separately from gimple_vdef copying.
17168 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17170 * value-prof.c (free_hist): Remove call to memset and the enclosing if
17173 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
17174 Olivier Hainque <hainque@adacore.com>
17176 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
17177 for all vxworks7 targets.
17178 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
17179 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
17180 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
17181 variations for VX6/VX7 and 32/64bits later on in ...
17182 (VXWORKS_LIB_SPEC): Leverage new macros.
17183 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
17184 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
17186 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
17188 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
17189 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
17191 2017-06-26 Carl Love <cel@us.ibm.com>
17193 * config/rs6000/rs6000-c.c: Add support for built-in functions
17194 vector bool char vec_reve (vector bool char);
17195 vector signed char vec_reve (vector signed char);
17196 vector unsigned char vec_reve (vector unsigned char);
17197 vector bool int vec_reve (vector bool int);
17198 vector signed int vec_reve (vector signed int);
17199 vector unsigned int vec_reve (vector unsigned int);
17200 vector bool long long vec_reve (vector bool long long);
17201 vector signed long long vec_reve (vector signed long long);
17202 vector unsigned long long vec_reve (vector unsigned long long);
17203 vector bool short vec_reve (vector bool short);
17204 vector signed short vec_reve (vector signed short);
17205 vector double vec_reve (vector double);
17206 vector float vec_reve (vector float);
17207 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
17208 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
17209 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
17210 (altivec_vreve): New pattern.
17211 * config/rs6000/altivec.h (vec_reve): New define.
17212 * doc/extend.texi (vec_rev): Update the built-in documentation file
17213 for the new built-in functions.
17215 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17217 PR tree-optimization/71815
17218 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
17220 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
17222 (slsr_process_phi): Likewise.
17223 (replace_uncond_cands_and_profitable_phis): Don't replace a
17224 multiply candidate with a stride of 1 (copy or cast).
17225 (phi_incr_cost): Call uses_consumed_by_stmt rather than
17227 (lowest_cost_path): Likewise.
17228 (total_savings): Likewise.
17230 2017-06-26 Richard Biener <rguenther@suse.de>
17233 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
17234 Use def_builtin_pure for all gather builtins.
17236 2017-06-26 Richard Biener <rguenther@suse.de>
17238 PR tree-optimization/81203
17239 * tree-tailcall.c (find_tail_calls): Do not move stmts into
17240 non-dominating BBs.
17242 2017-06-26 Marek Polacek <polacek@redhat.com>
17245 * doc/invoke.texi: Document -Wmultistatement-macros.
17247 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
17249 * doc/sourcebuild.texi (ARM-specific attributes): Document new
17250 arm_neon_ok_no_float_abi effective target.
17252 2017-06-26 Richard Biener <rguenther@suse.de>
17254 PR tree-optimization/80928
17255 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
17256 (copy_bbs): Set BB_DUPLICATED flag early.
17257 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
17259 (execute_on_shrinking_pred): Likewise.
17260 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
17261 BB_DUPLICATED blocks.
17262 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
17263 iterate over all PHIs considering removal of *gsi.
17265 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
17267 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
17270 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17272 * config/rs6000/rs6000-string.c: (expand_block_clear,
17273 do_load_for_compare, select_block_compare_mode,
17274 compute_current_alignment, expand_block_compare,
17275 expand_strncmp_align_check, expand_strn_compare,
17276 expand_block_move, rs6000_output_load_multiple)
17277 Move functions related to string/block move/compare
17278 to a separate file.
17279 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
17280 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
17281 for this function which is now used in two files.
17282 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
17283 * config.gcc: Add rs6000-string.o to extra_objs for
17284 targets powerpc*-*-* and rs6000*-*-*.
17286 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
17289 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
17290 32-bit, since indexed is not valid for DImode.
17291 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
17292 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
17293 (define_peephole2 for Altivec d-form load): Add 32-bit support.
17294 (define_peephole2 for Altivec d-form store): Likewise.
17297 * multiple_target.c (create_dispatcher_calls): Only create the
17298 dispatcher call if the function is the default clone of a
17299 versioned function.
17301 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
17303 PR middle-end/80902
17304 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
17305 a call, force the call to not be a tail call.
17307 2017-06-23 Jeff Law <law@redhat.com>
17309 * doc/contrib.texi: Add entry for Steven Pemberton's work on
17312 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
17314 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
17315 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
17316 handling for early expansion of vector shifts (sl,sr,sra,rl).
17317 (builtin_function_type): Add vector shift right instructions
17318 to the unsigned argument list.
17320 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
17322 rtl-optimizatoin/79286
17323 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
17324 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
17325 trap. PIC register plus a const unspec without offset can never trap.
17327 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
17329 * tree.h (builtin_structptr_type): New type.
17330 (builtin_structptr_types): Declare new array.
17331 * tree.c (builtin_structptr_types): New array.
17332 (free_lang_data, build_common_tree_nodes): Use it.
17334 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
17337 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
17340 2017-06-22 Matt Turner <mattst88@gmail.com>
17342 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
17343 Lake models to skylake case. Assume skylake for unknown
17344 models with clflushopt.
17346 2017-06-22 Jeff Law <law@redhat.com>
17348 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
17349 frame sizes that do not satisfy aarch64_uimm12_shift.
17351 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
17353 * profile-count.h (apply_probability,
17354 apply_scale, probability_in): Fix checks for zero.
17356 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17358 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
17359 * doc/cppdiropts.texi (-I @var{dir}): Document it.
17361 2016-06-22 Richard Biener <rguenther@suse.de>
17363 * tree-vect-loop.c (vect_model_reduction_cost): Handle
17364 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
17365 REDUC_MAX_EXPR support.
17366 (vectorizable_reduction): Likewise.
17367 (vect_create_epilog_for_reduction): Likewise.
17369 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
17371 * match.pd (A / (1 << B) -> A >> B): New.
17372 * generic-match-head.c: Include optabs-tree.h.
17373 * gimple-match-head.c: Likewise.
17374 * optabs-tree.h (target_supports_op_p): New.
17375 * optabs-tree.c (target_supports_op_p): New.
17377 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17379 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
17380 $gcc_cv_ld --help output.
17381 (gcc_cv_ld_demangle): Likewise.
17382 (gcc_cv_ld_eh_frame_hdr): Likewise.
17383 (gcc_cv_ld_pie): Likewise.
17384 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
17385 (gcc_cv_ld_buildid): Likewise.
17386 (gcc_cv_ld_sysroot): Likewise.
17387 (ld_bndplt_support): Likewise.
17388 (ld_pushpopstate_support): Likewise.
17389 * configure: Regenerate.
17390 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
17392 2017-06-21 Jakub Jelinek <jakub@redhat.com>
17395 * config/i386/sse.md (round<mode>2): Renumber match_dup and
17396 operands indexes to avoid gap between operands and match_dups.
17398 2017-06-21 Andrew Pinski <apinski@cavium.com>
17400 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
17401 Increment Arith_shift and Arith_shift_reg by 1.
17402 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
17404 * config/aarch64/aarch64.c (thunderx_tunings): Enable
17405 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
17406 (aarch64_strip_extend): Add new argument and test for it.
17407 (aarch64_cheap_mult_shift_p): New function.
17408 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
17409 add a cost if it is true.
17410 Update calls to aarch64_strip_extend.
17411 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
17413 2017-06-21 Andrew Pinski <apinski@cavium.com>
17415 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
17417 (thunderxt88): Likewise.
17418 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
17419 (thunderx_prefetch_tune): New variable.
17420 (thunderx2t99_prefetch_tune): Update for the correct values.
17421 (thunderxt88_tunings): New variable.
17422 (thunderx_tunings): Use thunderx_prefetch_tune instead of
17423 generic_prefetch_tune.
17424 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
17426 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17428 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
17429 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
17430 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
17431 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
17432 (aarch64_atomic_cas<mode>, GPI): Likewise.
17434 2017-06-21 Martin Liska <mliska@suse.cz>
17436 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
17437 statements on cold and hot labels.
17438 * predict.c (tree_estimate_probability_bb): Remove the
17439 prediction from this place.
17441 2017-06-21 Martin Liska <mliska@suse.cz>
17443 PR tree-optimization/79489
17444 * gimplify.c (maybe_add_early_return_predict_stmt): New
17446 (gimplify_return_expr): Call the function.
17447 * predict.c (tree_estimate_probability_bb): Remove handling
17449 * predict.def: Update comment about early return predictor.
17450 * gimple-predict.h (is_gimple_predict): New function.
17451 * predict.def: Change default value of early return to 66.
17452 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
17454 * passes.def: Put pass_strip_predict_hints to the beginning of
17457 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
17459 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
17460 FUNCTION_DECL declarations.
17461 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
17463 (dwaf2out_decl): Likewise.
17464 * godump.c (go_early_global_decl): Skip call to the real debug hook
17465 for FUNCTION_DECL declarations.
17466 * passes.c (rest_of_decl_compilation): Skip call to the
17467 early_global_decl debug hook for FUNCTION_DECL declarations, unless
17468 -fdump-go-spec is passed.
17470 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
17472 * config/i386/i386.c (struct builtin_isa): New field pure_p.
17473 Reorder for compactness.
17474 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
17475 (def_builtin_pure, def_builtin_pure2): New functions.
17476 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
17478 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
17480 * match.pd (nop_convert): New predicate.
17481 ((A +- CST1) +- CST2): Allow some NOP conversions.
17483 2017-06-21 Jakub Jelinek <jakub@redhat.com>
17486 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
17487 with ctors/dtors if GOVD_SHARED is set.
17489 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
17491 * config/aarch64/aarch64.md (movti_aarch64):
17492 Emit mov rather than orr.
17493 (movtf_aarch64): Likewise.
17494 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
17495 Emit mov rather than orr.
17497 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
17499 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
17500 Swap alternatives, make integer dup more expensive.
17502 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
17504 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
17505 Return true for non-tls symbols.
17507 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
17509 * config/aarch64/aarch64-cores.def (cortex-a55): New.
17510 (cortex-a75): Likewise.
17511 (cortex-a75.cortex-a55): Likewise.
17512 * config/aarch64/aarch64-tune.md: Regenerate.
17513 * doc/invoke.texi (-mtune): Document new values for -mtune.
17515 2017-06-21 Tom de Vries <tom@codesourcery.com>
17517 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
17518 stack_size feature.
17519 (Effective-Target Keywords, Other attributes): Suggest using
17520 dg-add-options stack_size feature to get stack limit in stack_size
17521 effective target documentation.
17523 2017-06-21 Julian Brown <julian@codesourcery.com>
17524 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17526 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
17527 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
17528 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
17530 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
17531 attribute type list for neon_multiply.
17532 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
17533 attribute type list for neon_multiply.
17534 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
17535 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
17536 attribute type list for neon_multiply.
17537 * config/arm/types.md (crypto_pmull): Add.
17538 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
17539 attribute type list.
17541 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
17543 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
17546 2017-06-20 Jakub Jelinek <jakub@redhat.com>
17548 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
17549 to make sure not to dereference a NULL cost_classes_ptr pointer.
17551 2017-06-20 Carl Love <cel@us.ibm.com>
17553 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17554 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17555 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17556 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17557 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17558 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17559 VMULOSW): New enum "unspec" values.
17560 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17561 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
17562 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17563 altivec_vmulosw): New patterns.
17564 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17565 VMULOSW): Add definitions.
17567 2017-06-20 Julia Koval <julia.koval@intel.com>
17569 * config/i386/i386.c: Fix rounding expand for new pattern.
17570 * config/i386/subst.md: Fix pattern (parallel -> unspec).
17572 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
17574 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
17575 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
17577 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
17579 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
17582 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
17584 * config/aarch64/aarch64-cores.def: Rearrange to sort by
17585 architecture, then by implementer ID.
17586 * config/aarch64/aarch64-tune.md: Regenerate.
17588 2017-06-20 Richard Biener <rguenther@suse.de>
17590 PR middle-end/81097
17591 * fold-const.c (split_tree): Fold to type before negating.
17593 2017-06-20 David Malcolm <dmalcolm@redhat.com>
17595 * diagnostic-show-locus.c
17596 (selftest::test_fixit_deletion_affecting_newline): New function.
17597 (selftest::diagnostic_show_locus_c_tests): Call it.
17599 2017-06-20 Andreas Schwab <schwab@suse.de>
17602 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
17605 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17607 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
17608 __ARM_FEATURE_COPROC according to support.
17610 2017-06-20 Jakub Jelinek <jakub@redhat.com>
17612 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
17613 Rewritten to avoid overflow for > 32-bit pointers.
17616 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
17617 by removing enum keyword.
17618 (ubsan_type_descriptor): Likewise. Formatting fix.
17621 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
17622 splitter): Require TARGET_SSE2 in the condition.
17624 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
17627 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
17628 for doing vector set of SFmode on ISA 3.0.
17629 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
17630 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
17632 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
17633 SFmode value into a V4SF variable that was extracted from another
17634 V4SF variable without converting the element to double precision
17635 and back to single precision vector format.
17636 (vsx_insert_extract_v4sf_p9_2): Likewise.
17638 2017-06-19 Jakub Jelinek <jakub@redhat.com>
17640 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
17641 in UWHI to avoid undefined overflow.
17644 * ubsan.h (enum ubsan_encode_value_phase): New.
17645 (ubsan_encode_value): Change second argument to
17646 enum ubsan_encode_value_phase with default value of
17647 UBSAN_ENCODE_VALUE_GENERIC.
17648 * ubsan.c (ubsan_encode_value): Change second argument to
17649 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
17650 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
17651 create_tmp_var_raw instead of create_tmp_var and use a
17653 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
17654 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
17655 ubsan_encode_value callers.
17658 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
17659 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
17660 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
17662 2017-06-19 Richard Biener <rguenther@suse.de>
17664 PR middle-end/81118
17665 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
17666 estimates if we changed anything.
17668 2017-06-19 Richard Biener <rguenther@suse.de>
17670 PR tree-optimization/80887
17671 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
17672 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
17673 simplified lookups, then reset mprts_hook.
17674 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
17676 (try_to_simplify): Likewise.
17678 2017-06-19 Martin Liska <mliska@suse.cz>
17681 * gimplify.c (gimplify_switch_expr):
17682 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
17684 2017-06-19 Martin Liska <mliska@suse.cz>
17686 * doc/install.texi: Document that PGO runs in 4 stages.
17688 2017-06-19 Martin Liska <mliska@suse.cz>
17691 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
17692 to dispatcher function name.
17693 * multiple_target.c (replace_function_decl): New function.
17694 (create_dispatcher_calls): Redirect both edges and references.
17696 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
17698 * profile-count.c (profile_count::dump): Dump quality.
17699 (profile_count::differs_from_p): Update for unsigned val.
17700 * profile-count.h (profile_count_quality): New enum.
17701 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
17703 2017-06-19 Richard Biener <rguenther@suse.de>
17705 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
17706 struct function as arg.
17707 (estimate_numbers_of_iterations): Export overload with loop arg.
17708 (free_numbers_of_iterations_estimates_loop): Use an overload of
17709 free_numbers_of_iterations_estimates instead.
17710 * tree-cfg.c (remove_bb): Adjust.
17711 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
17712 * tree-parloops.c (gen_parallel_loop): Likewise.
17713 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
17715 (tree_unroll_loops_completely): Likewise.
17716 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
17717 Use an overload instead and export.
17718 (estimated_loop_iterations): Adjust.
17719 (max_loop_iterations): Likewise.
17720 (likely_max_loop_iterations): Likewise.
17721 (estimate_numbers_of_iterations): Take struct function as arg
17723 (loop_exits_before_overflow): Adjust.
17724 (free_numbers_of_iterations_estimates_loop): Use an overload.
17725 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
17726 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
17728 2017-06-19 Richard Biener <rguenther@suse.de>
17731 * ipa-prop.c (find_constructor_constant_at_offset): Handle
17732 RANGE_EXPR conservatively.
17734 2017-06-16 Carl Love <cel@us.ibm.com>
17736 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17737 definitions for vec_float, vec_float2, vec_floato,
17738 vec_floate built-ins.
17739 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
17740 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
17742 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
17743 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
17744 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
17745 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
17746 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
17747 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
17748 vec_floato): Add builtin defines.
17749 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
17750 Update the built-in documentation file for the new built-in
17753 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17755 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
17756 (mthumb): Mark as the negative of -marm.
17758 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17760 * doc/invoke.texi (ARM Options, -mcpu): Document supported
17762 (ARM Options, -mtune): Document that this accepts the same
17763 extension options as -mcpu.
17764 (ARM Options, -mfpu): Document addition of -mfpu=auto.
17766 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17768 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
17769 permitted extensions.
17771 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17773 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
17774 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
17775 (armv8-m.main): Add option +nodsp.
17776 * config/arm/arm-cpu-cdata.h: Regenerated.
17778 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17780 * config/arm/t-fuchsia: New file.
17781 * config.gcc (arm*-*-fuchsia*): Use it.
17783 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17785 * config/arm/t-symbian: Rewrite for new option infrastructure.
17787 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17789 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
17790 (MULTILIB_REQUIRED): Likewise.
17792 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17794 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
17795 (MULTILIB_RESUE): Likewise.
17796 (MULTILIB_MATCHES): Likewise.
17797 (MULTLIB_REQUIRED): Likewise.
17799 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17801 * config/arm/t-rtems: Rewrite for new option framework.
17803 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17805 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
17806 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
17807 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
17808 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
17809 * config/arm/t-multilib: ... here.
17810 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
17811 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
17812 armv7-a and armv8*-a when A-profile libraries have not been built.
17813 * config/arm/t-rmprofile: Rewrite.
17815 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17817 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
17818 with a backslash. Remove the backslash after substituting unescaped
17820 * doc/fragments.texi (MULTILIB_REUSE): Document it.
17822 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17824 * config.gcc: (arm*-*-*): When building a-profile libraries, force
17825 the driver to pass through the default setting of -mfloat-abi.
17826 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
17828 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
17829 (all_feat_combs): New rule.
17830 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
17832 * config/arm/t-aprofile: Rewrite.
17834 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17836 * config/arm/arm.h (FPUTYPE_AUTO): Define.
17837 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
17838 fpu is not specified by the user/command-line.
17839 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
17840 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
17841 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
17842 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
17843 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
17844 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
17846 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17848 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
17849 * config/arm/t-arm-elf: Rewritten.
17851 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17853 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
17854 have some floating-point instructions.
17855 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
17856 (TARGET_MAYBE_HARD_FLOAT): New macro.
17857 * config/arm/arm-builtins.c (arm_init_builtins): Use
17858 TARGET_MAYBE_HARD_FLOAT.
17859 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
17861 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17863 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
17864 (configargs.h): Include it.
17865 (arm_print_hint_for_fpu_option): New function.
17866 (arm_parse_fpu_option): New function.
17867 (candidate_extension): New class.
17868 (arm_canon_for_multilib): New function.
17869 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
17870 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
17871 (ARCH_CANONICAL_SPECS): New macro.
17872 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
17874 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17876 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
17877 are set after handling multilib fragments. Set target_cpu_default2
17880 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17882 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
17884 (arm*-*-*): Set target_cpu_default2 to a quoted string.
17885 * config/arm/parsecpu.awk (check_cpu): Validate any extension
17887 (check_arch): Likewise.
17888 * config/arm/arm.c (arm_configure_build_target): Handle
17889 TARGET_CPU_DEFAULT being a string constant. Scan any feature
17890 options in the default.
17892 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17894 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
17895 when an option is an alias of another.
17896 * config/arm/parsecpu.awk (optalias): New parser token.
17897 (gen_comm_data): Mark non-alias options as such. Emit entries
17898 for extension aliases.
17899 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
17900 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
17901 (armv6kz, armv6zk, armv6t2): Likewise.
17902 (armv7): Make vfpv3-d16 an alias.
17903 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
17905 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
17906 Sort in canonical order.
17907 (armv8-a): Sort in canonical order.
17908 (armv8.1-a, armv8.2-a): Likewise.
17909 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
17911 (cortex-a9): Sort in canonical order.
17912 * config/arm/arm.c (selftests.h): Include it.
17913 (arm_test_cpu_arch_data): New function.
17914 (arm_run_self_tests): New function.
17915 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
17916 (targetm): Move declaration to the end of the file.
17917 * arm-cpu-cdata.h: Regenerated.
17919 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17921 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
17922 call to target_mode_check describing the type of option passed.
17923 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
17924 (arm_target_thumb_only): Use arm_parse_arch_option_name or
17925 arm_parse_cpu_option_name to match parameters against list of
17927 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
17928 arm_arch_core_flags data structure.
17929 * config/arm/arm-cpu_cdata.h: Regenerated.
17931 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17933 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
17935 (arm_print_hint_for_cpu_option): Likewise.
17936 (arm_print_hint_for_arch_option): Likewise.
17937 (arm_parse_cpu_option_name): Likewise.
17938 (arm_parse_arch_option_name): Likewise.
17939 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
17940 of entries in the all_fpus list.
17941 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
17942 (arm_parse_cpu_option_name): Declare.
17943 (arm_parse_arch_option_name): Declare.
17944 (arm_parse_option_features): Declare.
17945 (arm_intialize_isa): Declare.
17946 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
17948 (gen_comm_data): ... here. Make definitions non-static.
17949 * config/arm/arm-cpu-data.h: Regenerated.
17950 * config/arm/arm-cpu-cdata.h: Regenerated.
17952 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17954 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
17955 (cpu_arch_extension): New structure.
17956 (cpu_arch_option, arch_option, cpu_option): New structures.
17957 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
17958 architecture types.
17959 (gen_data): Generate new format data tables.
17960 * config/arm/arm.c (cpu_tune): New structure.
17961 (cpu_option, processors): Delete.
17962 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
17963 (arm_print_hint_for_cpu_option): ... this and ...
17964 (arm_print_hint_for_arch_option): ... this.
17965 (arm_parse_arch_cpu_name): Delete. Replace with ...
17966 (arm_parse_cpu_option_name): ... this and ...
17967 (arm_parse_arch_option_name): ... this.
17968 (arm_unrecognized_feature): Change type of target parameter to
17970 (arm_parse_arch_cpu_features): Delete. Replace with ...
17971 (arm_parse_option_features): ... this.
17972 (arm_configure_build_target): Rework to use new configuration data
17974 (arm_print_tune_info): Rework for new configuration data tables.
17975 * config/arm/arm-cpu-data.h: Regenerated.
17976 * config/arm/arm-cpu.h: Regenerated.
17978 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17980 * Makefile.in (OBJS): Move sbitmap.o from here ...
17981 (OBJS-libcommon): ... to here.
17983 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17985 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
17986 (ISA_ALL_CRYPTO): New macro.
17987 (ISA_ALL_SIMD): New macro
17988 (ISA_ALL_FP): New macro.
17989 * config/arm/arm.c (fpu_bitlist): Update initializer.
17990 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
17992 (arm9e): Add fpu. Add option for nofp
17993 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
17994 (arm926ej-s, arm1026ej-s): Likewise.
17995 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
17996 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
17997 neon-fp16, neon-vfpv4, nofp and nosimd.
17998 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
17999 (cortex-a8): Add fpu. Add option for nofp.
18000 (cortex-a9): Add fpu. Add options for nosimd and nofp.
18001 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
18002 (cortex-r4f): Add fpu.
18003 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
18004 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
18006 (cortex-r8): Likewise.
18007 (cortex-m4): Add fpu. Add option for nofp.
18008 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
18009 (cortex-a17.cortex-a7): Likewise.
18010 (cortex-a32): Add fpu. Add options for crypto and nofp.
18011 (cortex-a35, cortex-a53): Likewise.
18012 (cortex-a57): Add fpu. Add option for crypto.
18013 (cortex-a72, cortex-a73): Likewise.
18014 (exynos-m1): Likewise.
18015 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
18016 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
18017 (cortex-m33): Add fpu. Add option for nofp.
18018 * config/arm/arm-cpu-cdata.h: Regenerated
18019 * config/arm/arm-cpu-data.h: Regenerated.
18021 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18023 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
18024 (armv5te, armv5tej): Likewise.
18025 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
18026 (armv7): Add options fp and vfpv3-d16.
18027 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
18028 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
18030 (armv7ve): Likewise.
18031 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
18032 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
18033 (armv8-a): Add nocrypto option.
18034 (armv8.1-a, armv8.2-a): Likewise.
18035 (armv8-m.main): add options fp, fp.dp and nofp.
18037 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18039 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
18041 (armv8-a+crc): Delete.
18042 (armv8.1-a): Add options simd, crypto and nofp.
18043 (armv8.2-a): Add options fp16, simd, crypto and nofp.
18044 (armv8.2-a+fp16): Delete.
18045 (armv8-m.main): Add option dsp.
18046 (armv8-m.main+dsp): Delete.
18047 (cortex-a8): Add fpu. Add nofp option.
18048 (cortex-a9): Add fpu. Add nofp and nosimd options.
18049 * config/arm/parsecpu.awk (gen_data): Generate option tables and
18050 link to main cpu and architecture data structures.
18051 (gen_comm_data): Only put isa attributes from the main architecture
18053 (option): New statement for architecture and CPU entries.
18054 * arm.c (struct cpu_option): New structure.
18055 (struct processors): Add entry for options.
18056 (arm_unrecognized_feature): New function.
18057 (arm_parse_arch_cpu_name): Ignore any characters after the first
18059 (arm_parse_arch_cpu_feature): New function.
18060 (arm_configure_build_target): Separate out any CPU and architecture
18061 features and parse separately. Don't error out if -mfpu=auto is
18062 used with only an architecture string.
18063 (arm_print_asm_arch_directives): New function.
18064 (arm_file_start): Call it.
18065 * config/arm/arm-cpu-cdata.h: Regenerated.
18066 * config/arm/arm-cpu-data.h: Likewise.
18067 * config/arm/arm-tables.opt: Likewise.
18069 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18071 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
18072 assembler when it is not -mfpu=auto.
18074 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18076 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
18077 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
18078 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
18079 (ASM_CPU_SPEC): Rewrite.
18080 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
18081 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
18082 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
18083 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
18084 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
18085 copied string is NUL-terminated. Also strip any characters prefixed
18087 (arm_rewrite_selected_arch): New function.
18088 (arm_rewrite_march): New function.
18090 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18092 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
18093 (x_arm_cpu_string, x_arm_tune_string): Likewise.
18094 (march, mcpu, mtune): Convert to string-based options.
18095 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
18096 (arm_parse_arch_cpu_name): New function.
18097 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
18098 identify selected architecture or CPU.
18099 (arm_option_save): New function.
18100 (TARGET_OPTION_SAVE): Redefine.
18101 (arm_option_restore): Restore string options.
18102 (arm_option_print): Print string options.
18104 2017-06-16 Martin Sebor <msebor@redhat.com>
18106 PR tree-optimization/80933
18107 PR tree-optimization/80934
18108 * builtins.c (fold_builtin_3): Do not handle bcmp here.
18109 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
18110 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
18111 (gimple_fold_builtin): Call them.
18113 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18115 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
18116 as unlikely; update profile.
18118 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18120 * predict.c (force_edge_cold): Handle declaring edges impossible
18123 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18125 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
18127 (try_unroll_loop_completely): Fix reporting.
18129 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18131 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
18133 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
18136 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
18137 if given a non-constant argument for an intrinsic which requires a
18140 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18142 * profile.c (compare_freqs): New function.
18143 (branch_prob): Sort edge list.
18144 (find_spanning_tree): Assume that the list is priority sorted.
18146 2017-06-16 Richard Biener <rguenther@suse.de>
18148 PR tree-optimization/81090
18149 * passes.def (pass_record_bounds): Remove.
18150 * tree-pass.h (make_pass_record_bounds): Likewise.
18151 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
18152 make_pass_record_bounds): Likewise.
18153 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
18154 not free niter estimates at the beginning but at the end.
18155 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
18157 2017-06-16 Richard Biener <rguenther@suse.de>
18159 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
18160 initializer to workaround ICE in host GCC 4.8.
18162 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18164 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
18166 (clone_inlined_nodes): Update.
18168 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
18170 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
18171 prefetch settings, and enable prefetching by default at -O3.
18173 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
18175 * config/aarch64/aarch64.c (aarch64_override_options_internal):
18176 Set flag_prefetch_loop_arrays according to tuning data.
18178 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
18180 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
18181 New tune structure.
18182 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
18183 [Unrelated to main purpose of the patch] Place the pointer field last
18184 to enable type checking errors when tune structure are wrongly merged.
18185 * config/aarch64/aarch64.c (generic_prefetch_tune,)
18186 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
18187 (thunderx2t99_prefetch_tune): New tune constants.
18188 (tune_params *_tunings): Update all tunings (no functional change).
18189 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
18190 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
18191 from tunings structures.
18193 2017-06-16 Jakub Jelinek <jakub@redhat.com>
18196 * ubsan.c (instrument_null): Add T argument, use it instead
18197 of computing it based on IS_LHS.
18198 (instrument_object_size): Likewise.
18199 (pass_ubsan::execute): Adjust instrument_null and
18200 instrument_object_size callers to pass gimple_get_lhs or
18201 gimple_assign_rhs1 result to it. Use instrument_null instead of
18202 calling get_base_address and instrument_mem_ref. Handle
18203 aggregate call arguments for object-size sanitization.
18205 2017-06-16 Yury Gribov <tetra2005@gmail.com>
18207 PR tree-optimization/81089
18208 * tree-vrp.c (is_masked_range_test): Validate operands of
18211 2017-06-15 Martin Sebor <msebor@redhat.com>
18214 * dumpfile.c (dump_register): Avoid calling memset to initialize
18215 a class with a default ctor.
18216 * gcc.c (struct compiler): Remove const qualification.
18217 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
18218 * hash-table.h: Ditto.
18219 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
18221 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
18222 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
18224 * params.h (struct param_info): Make struct members non-const.
18225 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
18226 with default initialization.
18227 * vec.h (vec_copy_construct, vec_default_construct): New helper
18229 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
18230 with vec_copy_construct.
18231 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
18232 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
18233 * doc/invoke.texi (-Wclass-memaccess): Document.
18235 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18237 * emit-rtl.h (is_leaf): Update comment about local
18238 register allocator.
18240 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18243 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
18244 for a variable to have a section before checking if the section has a
18246 Set section to.persistent if persistent attribute is set.
18247 Warn if .persistent attribute is used on an automatic variable.
18249 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
18251 PR rtl-optimization/80474
18252 * reorg.c (update_block): Do not ignore instructions in a delay slot.
18254 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
18256 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
18259 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
18261 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
18262 (casesi): Emit bounds checking as RTL.
18263 (casesi_internal_mips16_<mode>): Remove bounds checking.
18265 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
18267 * config/xtensa/xtensa.c (xtensa_option_override): Append
18268 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
18269 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
18270 xtensa_doloop_hooks): Define unconditionally.
18271 (xtensa_reorg_loops): Only call reorg_loops in the presence of
18273 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
18274 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
18275 for it in xtensa_option_override.
18276 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
18277 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
18279 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
18281 * doc/cppopts.texi: Document '-' special value to -MF.
18283 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
18285 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
18286 (cortex_a53_fconst): Likewise.
18287 (cortex_a53_fpmul): Likewise.
18288 (cortex_a53_f_load_64): Likewise.
18289 (cortex_a53_f_load_many): Likewise.
18290 (cortex_a53_advsimd_alu): Likewise.
18291 (cortex_a53_advsimd_alu_q): Likewise.
18292 (cortex_a53_advsimd_mul): Likewise.
18293 (cortex_a53_advsimd_mul_q): Likewise.
18294 (fpmac bypass): Add new bypass for fpmac-fpmac case.
18295 Add missing fmul, r2f_cvt and fconst cases.
18297 2017-06-14 Richard Biener <rguenther@suse.de>
18299 PR middle-end/81088
18300 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
18302 (fold_binary_loc): When associating do not treat pre-existing
18303 TREE_OVERFLOW on literal constants as a reason to allow
18304 TREE_OVERFLOW on associated literal constants.
18306 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
18308 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
18309 (MASK_FEATURES): New macro.
18310 * config/sparc/sparc.c (sparc_option_override): Remove the special
18311 handling of -mfpu and generalize it to all MASK_FEATURES switches.
18313 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
18315 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
18316 a division of 0 if non-call exceptions are enabled.
18318 2017-06-14 Andrew Pinski <apinski@cavium.com>
18319 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18322 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
18323 Improve vector initialization code gen for only variable case.
18325 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
18327 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
18329 2017-06-14 Richard Biener <rguenther@suse.de>
18331 PR tree-optimization/81083
18332 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
18335 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18337 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
18338 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
18339 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
18340 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
18341 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
18342 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
18344 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18346 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
18347 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
18349 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18351 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
18353 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18355 * config/rs6000/t-rtems: Don't handle SPE.
18357 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18359 * config/rs6000/t-linux: Don't handle SPE.
18361 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18363 * config/rs6000/eabispe.h: Delete file.
18365 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18367 * config/rs6000/t-spe: Delete file.
18369 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18371 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
18372 (rs6000_legitimate_offset_address_p): Return false for anything in
18373 V2SImode or V2SFmode.
18375 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18377 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
18378 except V2SF and V2SI. Rearrange the vector modes, and add comments.
18379 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
18381 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
18382 (rs6000_legitimate_offset_address_p): Ditto.
18383 (rs6000_emit_move): Ditto.
18384 (rs6000_init_builtins): Remove V4HI_type_node.
18386 2017-06-13 Martin Liska <mliska@suse.cz>
18389 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
18390 (gate_asan): Likewise.
18391 * asan.h (asan_no_sanitize_address_p): Remove the function.
18392 (sanitize_flags_p): New function.
18393 * builtins.def: Fix coding style.
18394 * common.opt: Use renamed enum value.
18395 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
18396 * doc/extend.texi: Document no_sanitize attribute.
18397 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
18398 to SANITIZE_UNDEFINED_NONDEFAULT.
18399 * gcc.c (sanitize_spec_function): Use the renamed enum value.
18400 * gimple-fold.c (optimize_atomic_compare_exchange_p):
18401 Use sanitize_flags_p.
18402 * gimplify.c (gimplify_function_tree): Likewise.
18403 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
18404 * opts.c (parse_no_sanitize_attribute): New function.
18405 (common_handle_option): Use renamed enum value.
18406 * opts.h (parse_no_sanitize_attribute): Declare.
18407 * tree.c (sanitize_flags_p): New function.
18408 * tree.h: Declared here.
18409 * tsan.c: Use sanitize_flags_p.
18410 * ubsan.c (ubsan_expand_null_ifn): Likewise.
18411 (instrument_mem_ref): Likewise.
18412 (instrument_bool_enum_load): Likewise.
18413 (do_ubsan_in_current_function): Remove the function.
18414 (pass_ubsan::execute): Use sanitize_flags_p.
18415 * ubsan.h: Remove do_ubsan_in_current_function
18416 * tree-cfg.c (print_no_sanitize_attr_value): New function.
18417 (dump_function_to_file): Use it here.
18419 2017-06-13 Martin Jambor <mjambor@suse.cz>
18421 PR tree-optimization/80803
18422 PR tree-optimization/81063
18423 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
18424 (propagate_subaccesses_across_link): Enqueue subtree whenever
18425 necessary instead of relying on the caller.
18427 2017-06-13 Martin Jambor <mjambor@suse.cz>
18429 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
18430 that have a first_link.
18431 (sort_and_splice_var_accesses): Do not check first_link before
18433 (subtree_mark_written_and_enqueue): Likewise.
18434 (propagate_all_subaccesses): Likewise and do not stop at first
18435 parent with a first_link.
18437 2017-06-13 Martin Jambor <mjambor@suse.cz>
18439 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
18442 2017-06-13 Yury Gribov <tetra2005@gmail.com>
18444 * match.pd: New pattern.
18446 2017-06-13 Yury Gribov <tetra2005@gmail.com>
18448 * tree-vrp.c (is_masked_range_test): New function.
18449 (register_edge_assert_for): Determine ranges for
18452 2017-06-13 Yury Gribov <tetra2005@gmail.com>
18454 PR tree-optimization/67328
18455 * fold-const.c (maskable_range_p): New function.
18456 (build_range_check): Generate bittests if possible.
18458 2017-06-13 Martin Liska <mliska@suse.cz>
18460 * gimple-pretty-print.c (dump_probability): Add new argument.
18461 (dump_edge_probability): Dump both probability and count.
18462 (dump_gimple_label): Likewise.
18463 (dump_gimple_bb_header): Likewise.
18465 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
18468 * config/avr/avr-devices.c: Fix indentation.
18469 * config/avr/gen-avr-mmcu-specs.c: Dito.
18471 2017-06-13 Richard Biener <rguenther@suse.de>
18473 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
18474 instead get vector type from stmt_info.
18475 (vectorizable_reduction): Adjust. Remove dead code.
18477 2017-06-13 Richard Biener <rguenther@suse.de>
18479 PR middle-end/81065
18480 * fold-const.c (extract_muldiv_1): Remove bogus distribution
18481 case of C * (x * C2 + C3).
18482 (fold_addr_of_array_ref_difference): Properly fold index difference.
18484 2017-06-12 David S. Miller <davem@davemloft.net>
18487 * config/sparc/sparc.md (return expander): Emit frame blockage if
18488 function uses alloca.
18490 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
18492 * combine.c (make_field_assignment): Check len rather than the mode
18493 precision when calling force_to_mode.
18495 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
18497 Support multilibs and devices that see flash in RAM address range.
18500 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
18501 (avr_mcu_t) <flash_pm_offset>: New field.
18502 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
18503 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
18504 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
18505 (AVR_TINY_PM_OFFSET): Remove macro.
18506 * config/avr/avr.opt (-mshort-calls): New option.
18507 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
18508 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
18509 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
18510 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
18511 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
18512 instead of avr_arch->have_jmp_call.
18513 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
18514 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
18515 avr_arch->flash_pm_offset to define.
18516 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
18517 new field flash_pm_offset. Add entry for avrxmega3.
18518 (avr_texinfo): Add entry for avrxmega3.
18519 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
18520 attiny212, attiny214,
18521 attiny412, attiny414, attiny416, attiny417,
18522 attiny814, attiny816, attiny817,
18523 attiny1614, attiny1616, attiny1617,
18524 attiny3214, attiny3216, attiny3217.
18525 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
18526 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
18527 (avr_print_operand_address) [AVR_TINY]: Same.
18528 (avr_asm_init_sections) <readonly_data_section>: Only patch
18529 callback if avr_arch->flash_pm_offset = 0.
18530 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
18531 for rodata if avr_arch->flash_pm_offset != 0.
18532 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
18533 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
18534 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
18535 (m_options): Append opt_rcall.
18536 (m_dirnames): Append dir_rcall.
18537 * config/avr/t-multilib: Regenerate.
18539 * configure.ac [target=avr]: Check whether avrxmega3 default
18540 linker description file works as needed.
18541 * configure: Regenerate.
18542 * doc/avr-mmcu.texi: Regenerate.
18543 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
18544 <__AVR_ARCH__>: Document avrxmega3 and 103.
18545 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
18546 <__AVR_SHORT_CALLS__>: Document it.
18547 <__AVR_PM_BASE_ADDRESS__>: Document it.
18548 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
18549 (AVR Variable Attributes) <progmem>: Document this is
18550 not needed for avrxmega3.
18551 (AVR Named Address Spaces) <__flash>: Dito.
18553 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
18555 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
18557 2017-06-12 Doug Rupp <rupp@adacore.com>
18559 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
18560 Append vxworks-stdint.h to the tm_file list.
18561 * config/vxworks-stdint.h: New file.
18563 2017-06-12 Martin Liska <mliska@suse.cz>
18565 PR tree-optimization/81041
18566 * tree-profile.c (gimple_gen_ic_func_profiler):
18567 Create an extra BB in profile-generate
18568 (gimple_gen_time_profiler): Likewise.
18570 2017-06-12 Jakub Jelinek <jakub@redhat.com>
18572 PR tree-optimization/81003
18573 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
18574 (update_range_test): Use it instead of force_gimple_operand_gsi.
18576 2017-06-12 Richard Biener <rguenther@suse.de>
18578 PR tree-optimization/81053
18579 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
18580 with backedge value not defined in loop. Simplify def stmt
18583 2017-06-11 Tom de Vries <tom@codesourcery.com>
18586 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
18588 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
18589 nvptx_cannot_force_const_mem.
18591 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18593 * opts.c (finish_options): Move test for flag_split_stack after
18594 it has been initialized.
18596 2017-06-11 Jason Merrill <jason@redhat.com>
18598 * tree.h (id_equal): New.
18599 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
18600 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
18601 instead of strcmp of IDENTIFIER_POINTER.
18603 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18605 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
18606 (mark_all_inlined_calls_cdtor): Fix formating.
18607 (inline_transform): Rescale profile before inlining.
18609 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18611 * cgraph.h (cgraph_edge::clone): Update prototype.
18612 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
18613 (cgraph_node::create_clone): Update.
18614 (cgraph_node::create_version_clone): Update.
18615 * tree-inline.c (copy_bb): Update.
18616 (expand_call_inline): Update.
18618 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
18620 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
18621 factored out from ...
18622 (rs6000_emit_prologue): ... here.
18624 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
18626 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
18627 factored out from ...
18628 (rs6000_emit_prologue): ... here.
18630 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18632 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
18634 (handle_missing_profiles): Fix computation of tp_first_run.
18635 (counts_to_freqs): Do not touch freqs when count is 0.
18637 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18639 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
18642 2017-06-10 Tom de Vries <tom@codesourcery.com>
18644 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
18645 attributes): Document signal effective target.
18647 2017-06-10 Tom de Vries <tom@codesourcery.com>
18649 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
18650 Document effective target stack_size.
18652 2017-06-09 David Malcolm <dmalcolm@redhat.com>
18654 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
18655 to the edit_context if they can be auto-applied.
18657 2017-06-9 Ian Lance Taylor <iant@golang.org>
18659 * opts.c (finish_options): If -fsplit-stack, disable implicit
18660 -forder-blocks-and-partition.
18661 * doc/invoke.texi (Optimize Options): Document that when using
18662 -fsplit-stack -forder-blocks-and-partition is not implicitly
18665 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18667 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
18668 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
18669 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
18670 * builtins.def (abort, trap, unreachable): Declare cold.
18671 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
18672 * tree-core.h (ECF_COLD): New.
18673 * tree.c (set_call_expr_flags): Handle ECF_COLD.
18674 (build_common_builtin_nodes): Mark unreachable and abort as cold.
18676 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18678 * predict.c (unlikely_executed_stmt_p): Cleanup.
18680 2017-06-09 Richard Biener <rguenther@suse.de>
18682 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
18683 model if the ref is always written to.
18685 2017-06-09 Tamar Christina <tamar.christina@arm.com>
18687 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
18689 2017-06-09 Tamar Christina <tamar.christina@arm.com>
18691 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
18694 2017-06-09 Tom de Vries <tom@codesourcery.com>
18697 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
18698 "target cannot support label values" when encountering LABEL_REF.
18700 2017-06-09 Martin Liska <mliska@suse.cz>
18702 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
18703 (gimple_gen_ic_func_profiler): Emit direct comparison
18704 of __gcov_indirect_call_callee with NULL.
18705 (gimple_gen_time_profiler): Change probability from
18706 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
18708 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18710 * profile.c (edge_gcov_counts): Turn to pointer.
18711 (compute_branch_probabilities, compute_branch_probabilities): Update.
18712 (branch_prob): Do not clear edge_gcov_count.
18713 * profile.h (edge_gcov_counts): Turn to pointer.
18714 (edge_gcov_count): Update.
18716 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18718 * gimple.h (gimple_check_failed): Mark cold.
18720 2017-06-09 Richard Biener <rguenther@suse.de>
18722 PR tree-optimization/66623
18723 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
18724 refactor check_reduction into two parts, properly computing
18725 whether we have to check reduction validity for outer loop
18728 2017-06-09 Richard Biener <rguenther@suse.de>
18730 PR tree-optimization/79483
18731 * graphite-scop-detection.c (order): New global.
18732 (get_order): Compute bb to order mapping that satisfies code
18733 generation constraints.
18734 (cmp_pbbs): New helper.
18735 (build_scops): Start domwalk at entry block, sort generated
18738 2017-06-09 Richard Biener <rguenther@suse.de>
18740 PR middle-end/81007
18741 * ipa-polymorphic-call.c
18742 (ipa_polymorphic_call_context::restrict_to_inner_class):
18743 Skip FIELD_DECLs with error_mark_node type.
18744 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
18747 2017-06-09 Martin Liska <mliska@suse.cz>
18749 * predict.c (struct branch_predictor): New struct.
18750 (test_prediction_value_range): New test.
18751 (predict_c_tests): New function.
18752 * selftest-run-tests.c (selftest::run_tests): Run the function.
18753 * selftest.h: Declare new tests.
18755 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
18758 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
18759 gen_add3_insn did not fail.
18760 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
18761 r0, construct that number in a temporary reg and add that reg to r0.
18762 If asked to put the result in r0 as well, fail.
18764 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
18766 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
18767 for early expansion of vec_eqv.
18769 2017-06-08 Jakub Jelinek <jakub@redhat.com>
18771 PR middle-end/81005
18772 * ubsan.c (instrument_null): Avoid pointless code temporary.
18773 (pass_ubsan::execute): Instrument aggregate arguments of calls.
18775 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
18779 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
18782 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
18785 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18787 * predict.c (unlikely_executed_edge_p): Move ahead.
18788 (probably_never_executed_edge_p): Use it.
18790 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
18792 PR middle-end/79988
18793 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
18794 gimple_call_builtin_p call.
18796 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18798 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
18799 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
18800 rtl_check_failed_type2, rtl_check_failed_code1,
18801 rtl_check_failed_code2, rtl_check_failed_code_mode,
18802 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
18803 rtvec_check_failed_bounds, rtl_check_failed_flag,
18804 _fatal_insn_not_found, _fatal_insn): Likewise.
18805 * tree.h (tree_contains_struct_check_failed,
18806 tree_check_failed, tree_not_check_failed,
18807 tree_class_check_failed, tree_range_check_failed,
18808 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
18809 tree_vec_elt_check_failed, phi_node_elt_check_failed,
18810 tree_operand_check_failed, omp_clause_check_failed,
18811 omp_clause_operand_check_failed, omp_clause_range_check_failed):
18814 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18816 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
18817 flag_branch_probabilities.
18818 * ipa-inline.c (edge_badness): Likewise.
18819 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
18820 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
18821 * predict.c (maybe_hot_frequency_p): Likewise.
18822 (probably_never_executed): Likewise.
18823 * sched-ebb.c (schedule_ebbs): Likewise.
18824 * sched-rgn.c (find_single_block_region): Likewise.
18825 * tracer.c (tail_duplicate): Likewise.
18827 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18829 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
18830 longer requires x_flag_profile_use.
18832 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18834 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
18835 instead of flag_reorder_blocks_and_partition.
18836 * dbxout.c (dbxout_function_end): Likewise.
18837 * dwarf2out.c (gen_subprogram_die): Likewise.
18838 * haifa-sched.c (sched_create_recovery_edges): Likewise.
18839 * hw-doloop.c (reorg_loops): Likewise.
18840 * varasm.c (assemble_start_function,
18841 assemble_end_function): Likewise.
18842 (decide_function_section): Do not check for
18843 flag_reorder_blocks_and_partition.
18845 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
18847 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
18849 (chkp_get_hard_register_fake_addr_expr): Ditto.
18850 (chkp_build_addr_expr): Add check for hard reg case.
18851 (chkp_parse_array_and_component_ref): Ditto.
18852 (chkp_find_bounds_1): Ditto.
18853 (chkp_process_stmt): Don't generate bounds store for
18856 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18858 * predict.c (maybe_hot_bb_p): Do not check profile status.
18859 (maybe_hot_edge_p): Likewise.
18860 (probably_never_executed): Check for zero counts even if profile
18862 (unlikely_executed_edge_p): New function.
18863 (unlikely_executed_stmt_p): New function.
18864 (unlikely_executed_bb_p): New function.
18865 (set_even_probabilities): Use unlikely predicates.
18866 (combine_predictions_for_bb): Likewise.
18867 (predict_paths_for_bb): Likewise.
18868 (predict_paths_leading_to_edge): Likewise.
18869 (determine_unlikely_bbs): New function.
18870 (estimate_bb_frequencies): Use it.
18871 (compute_function_frequency): Use zero counts even if profile is
18873 * profile-count.h: Fix typo.
18875 2017-08-08 Julia Koval <julia.koval@intel.com>
18877 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
18878 _mm512_mask_cvtsepi16_storeu_epi8,
18879 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
18880 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
18881 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
18882 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
18883 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
18884 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
18885 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
18886 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
18887 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
18888 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
18889 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
18890 __builtin_ia32_pmovuswb256mem_mask,
18891 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
18892 __builtin_ia32_pmovwb512mem_mask): New builtins.
18894 2017-08-08 Julia Koval <julia.koval@intel.com>
18896 PR target/73350,80862
18897 * config/i386/subst.md (round): Fix round pattern.
18898 * config/i386/i386.c (ix86_erase_embedded_rounding):
18899 Fix erasing rounding for the fixed pattern.
18901 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18903 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
18905 2017-06-08 Martin Liska <mliska@suse.cz>
18907 PR gcov-profile/80911
18908 * gcov.c (block_info::block_info): New constructor.
18910 2017-06-07 Carl Love <cel@us.ibm.com>
18912 * config/rs6000/rs6000-c: The return type of the following
18913 built-in functions was implemented as int not long long. Fix sign
18914 of return value for the unsigned version of vec_mulo and vec_mule.
18915 vector unsigned long long vec_bperm (vector unsigned long long,
18916 vector unsigned char)
18917 vector signed long long vec_mule (vector signed int,
18919 vector unsigned long long vec_mule (vector unsigned int,
18920 vector unsigned int)
18921 vector signed long long vec_mulo (vector signed int,
18923 vector unsigned long long vec_mulo (vector unsigned int,
18924 vector unsigned int)
18925 * doc/extend.texi: Fix the documentation for the built-in
18928 2017-06-07 Carl Love <cel@us.ibm.com>
18931 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
18934 2017-06-07 Carl Love <cel@us.ibm.com>
18936 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
18937 support, Generate doublehv for signed int/float for BE case only.
18939 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
18941 * doc/invoke.texi (mcx16): Rewrite.
18943 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18945 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
18946 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
18947 *mov<mode>_softfloat, and an anonymous splitter): Use
18948 nonimmediate_operand instead of rs6000_nonimmediate_operand.
18950 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18952 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
18954 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
18955 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
18956 (rs6000_debug_reg_global): Adjust.
18957 (rs6000_init_hard_regno_mode_ok): Adjust.
18958 (rs6000_dbx_register_number): Adjust.
18959 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
18960 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
18961 Remove SPE_ACC and SPEFSCR.
18962 (REG_ALLOC_ORDER): Ditto.
18963 (FRAME_POINTER_REGNUM): Change to 111.
18964 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
18965 (REG_CLASS_NAMES): Ditto.
18966 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
18967 (REGISTER_NAMES): Ditto.
18968 (ADDITIONAL_REG_NAMES): Ditto.
18969 (rs6000_reg_names): Ditto.
18970 * config/rs6000/rs6000.md: Renumber some register number
18973 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18975 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
18977 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
18978 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
18980 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
18981 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
18982 Delete the SPE high registers.
18983 (REG_ALLOC_ORDER): Ditto.
18984 (enum reg_class): Remove SPE_HIGH_REGS.
18985 (REG_CLASS_NAMES): Ditto.
18986 (REG_CLASS_CONTENTS): Delete the SPE high registers.
18987 (REGISTER_NAMES): Ditto.
18988 (rs6000_reg_names): Ditto.
18989 * doc/tm.texi.in: Remove SPE as example.
18990 * doc/tm.texi: Regenerate.
18992 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18994 * config/rs6000/8540.md (ppc8540_brinc): Delete.
18995 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
18996 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
18997 * config/rs6000/rs6000.md (type): Remove "brinc".
18999 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19001 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
19002 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
19003 * config/rs6000/linuxspe.h: Delete file.
19004 * config/rs6000/rs6000.md: Don't include spe.md.
19005 * config/rs6000/spe.h: Delete file.
19006 * config/rs6000/spe.md: Delete file.
19007 * config/rs6000/t-rs6000: Remove spe.md.
19009 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19011 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
19012 (reg_or_none500mem_operand): Delete.
19013 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
19014 instead of reg_or_none500mem_operand.
19016 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19018 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
19019 handling of SPE flags.
19020 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
19022 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19024 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
19026 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
19027 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
19028 paired_divv2sf3): Similar.
19029 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
19030 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
19031 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
19033 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
19034 Rename the paired_* instruction patterns.
19035 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
19037 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
19038 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
19039 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
19040 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
19041 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
19042 PAIRED_VECTOR_MODE.
19043 (struct machine_function): Delete field spe_insn_chain_scanned_p.
19044 (spe_func_has_64bit_regs_p): Delete.
19045 (spe_expand_predicate_builtin): Delete.
19046 (spe_expand_evsel_builtin): Delete.
19047 (TARGET_DWARF_REGISTER_SPAN): Do not define.
19048 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
19049 (invalid_e500_subreg): Delete.
19050 (rs6000_legitimize_address): Always force_reg op2 as well, for
19051 paired single memory accesses.
19052 (rs6000_member_type_forces_blk): Delete.
19053 (rs6000_spe_function_arg): Delete.
19054 (rs6000_expand_unop_builtin): Delete SPE handling.
19055 (rs6000_expand_binop_builtin): Ditto.
19056 (spe_expand_stv_builtin): Delete.
19057 (bdesc_2arg_spe): Delete.
19058 (spe_expand_builtin): Delete.
19059 (spe_expand_predicate_builtin): Delete.
19060 (spe_expand_evsel_builtin): Delete.
19061 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
19062 (spe_init_builtins): Delete.
19063 (spe_func_has_64bit_regs_p): Delete.
19064 (savres_routine_name): Delete "info" parameter. Adjust callers.
19065 (rs6000_emit_stack_reset): Ditto.
19066 (rs6000_dwarf_register_span): Delete.
19067 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
19068 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
19069 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
19071 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
19073 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
19074 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
19075 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
19076 mulv2sf3, divv2sf3): Delete expanders.
19078 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19080 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
19082 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19084 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
19085 * config/rs6000/rs6000.c: Ditto.
19087 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19089 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
19090 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
19091 comparison_operator.
19093 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19095 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
19096 * config/rs6000/rs6000.opt: Ditto.
19097 * config/rs6000/t-rtems: Ditto.
19099 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19101 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
19102 TARGET_E500_SINGLE by 0, simplify.
19103 * config/rs6000/rs6000.c: Ditto.
19104 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
19105 (spe_build_register_parallel): Delete.
19106 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
19107 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
19108 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
19109 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
19110 (E500_CONVERT): Delete.
19111 * config/rs6000/spe.md: Remove many patterns and all define_constants.
19113 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19115 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
19116 * config/rs6000/dfp.md: Ditto.
19117 (negdd2, *negdd2_fpr): Merge.
19118 (absdd2, *absdd2_fpr): Merge.
19119 (negtd2, *negtd2_fpr): Merge.
19120 (abstd2, *abstd2_fpr): Merge.
19121 * config/rs6000/e500.h: Delete file.
19122 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
19123 TARGET_FPRS by 1 and simplify.
19124 * config/rs6000/rs6000-c.c: Ditto.
19125 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
19126 TARGET_DF_SPE by 0.
19127 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
19129 * config/rs6000/rs6000.md: Ditto.
19130 (floatdidf2, *floatdidf2_fpr): Merge.
19131 (move_from_CR_gt_bit): Delete.
19132 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
19133 (E500_CR_IOR_COMPARE): Delete.
19134 (All patterns that require !TARGET_FPRS): Delete.
19135 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
19137 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19139 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
19141 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19143 * graphds.c (add_edge): Intitialize edge's attached data.
19144 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
19145 pointer parameter. Call pointed function on each edge during
19146 graph traversing. Skip traversing the edge when the function
19148 (graphds_dfs, graphds_scc): Ditto.
19149 (for_each_edge): New parameter. Pass the new parameter to callback
19151 * graphds.h (skip_edge_callback): New function pointer type.
19152 (graphds_dfs, graphds_scc): New function pointer parameter.
19153 (graphds_edge_callback, for_each_edge): New parameter.
19155 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19157 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
19158 out code checking if runtime alias check is possible to below ...
19159 Call the new function.
19160 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
19161 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
19163 2017-06-07 Marek Polacek <polacek@redhat.com>
19166 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
19167 TYPE_OVERFLOW_WRAPS check.
19169 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19171 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
19172 if versioning is required.
19173 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
19174 peeling with the check for versioning.
19176 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19178 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
19179 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
19180 Set true to new parameter if new ssa variable is defined.
19181 (vect_gen_vector_loop_niters): Refactor. Set range information
19182 for the new vector loop bound variable.
19183 (vect_do_peeling): Ditto.
19185 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19187 * tree-affine.c (ssa.h): Include header file.
19188 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
19189 has wrapping overflow behavior.
19191 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19193 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
19195 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19197 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
19198 (tree_to_aff_combination): ... here.
19200 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19202 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
19203 reg_pressure model function.
19204 (ivopts_global_cost_for_size): Delete.
19205 (determine_set_costs, iv_ca_recount_cost): Call new model function
19206 ivopts_estimate_reg_pressure.
19208 2017-06-07 Tamar Christina <tamar.christina@arm.com>
19210 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
19211 expensive than udiv. Remove floating point cases from mod.
19213 2017-06-07 Tamar Christina <tamar.christina@arm.com>
19215 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
19216 Increase idiv cost.
19218 2017-06-07 Tamar Christina <tamar.christina@arm.com>
19220 * config/aarch64/aarch64.md
19221 (copysignsf3): Fix mask generation.
19223 2017-06-07 Jakub Jelinek <jakub@redhat.com>
19225 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
19227 (class dump_manager): Add register_dumps method.
19228 * dumpfile.c: Include langhooks.h.
19229 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
19230 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
19231 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
19232 (dump_manager::dump_register): Start with 512 entries instead of 32.
19233 (dump_manager::register_dumps): New method.
19234 * toplev.c (general_init): Instead of invoking register_dumps
19235 langhook, invoke register_dumps method on the dump manager.
19236 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
19239 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
19241 * doc/md.texi: Clarify the restrictions on a define_insn condition.
19242 Say that # requires an associated define_split to exist, and that
19243 the define_split must be suitable for use after register allocation.
19245 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19247 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
19248 (compute_outgoing_frequencies): Also initialize zero counts.
19249 (find_many_sub_basic_blocks): Do not produce uninitialized profile
19250 around loops; preserve more of profile when nothing changes.
19252 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
19254 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
19256 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
19257 * config/arm/arm-cpu-cdata.h: Regenerate.
19258 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
19259 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
19260 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
19261 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
19262 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
19264 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
19266 * config/arm/t-rmprofile: Likewise.
19267 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
19269 2017-06-06 David S. Miller <davem@davemloft.net>
19272 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
19273 blockage if function uses alloca.
19275 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19277 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
19278 New "uid" fields to hold pretty-print IDs of group and ref.
19279 Memory references are now identified as <group_id>:<ref_id>
19280 instead of using [random] addresses.
19281 (dump_mem_details): Simplify, no functional change.
19282 (dump_mem_ref): Simplify and make output more concise.
19283 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
19284 (find_or_create_group): Initialize group uid.
19285 (record_ref): Initialize ref uid. Improve debug output.
19286 (prune_group_by_reuse, should_issue_prefetch_p,)
19287 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
19288 (mark_nontemporal_store, determine_loop_nest_reuse):
19289 Improve debug output.
19291 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19293 * dbgcnt.def (prefetch): New debug counter.
19294 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
19295 (schedule_prefetches): Stop issueing prefetches if debug counter
19298 2017-06-06 Tom de Vries <tom@codesourcery.com>
19300 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
19301 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
19303 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19305 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
19306 Use aarch64_reg_or_zero predicate for operand 4.
19307 (aarch64_compare_and_swap<mode> define_insn_and_split):
19308 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
19309 (aarch64_store_exclusive<mode>): Likewise for operand 2.
19311 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
19313 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
19314 (arm_compute_save_core_reg_mask): This.
19315 (thumb1_compute_save_reg_mask): Rename into ...
19316 (thumb1_compute_save_core_reg_mask): This.
19317 (arm_compute_save_reg0_reg12_mask): Adapt comment.
19318 (arm_compute_frame_layout): Likewise.
19320 2017-06-06 Richard Biener <rguenther@suse.de>
19322 PR tree-optimization/80974
19323 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
19324 keep or clear leaders SSA info.
19326 2017-06-06 Tom de Vries <tom@codesourcery.com>
19328 * config/nvptx/nvptx.c (split_mode_p): New function.
19329 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
19331 2017-06-06 Tom de Vries <tom@codesourcery.com>
19333 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
19335 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19338 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
19341 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19343 * shrink-wrap.c (handle_simple_exit): Update profile.
19344 (try_shrink_wrapping): Upate profile.
19346 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19348 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
19349 (tree_guess_outgoing_edge_probabilities): New.
19350 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
19351 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
19353 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19355 * ipa-split.c (split_function): Initialize return bb profile.
19357 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19359 * profile.c (compute_branch_probabilities): Also initialize
19360 EXIT_BLOCK profile.
19362 2017-06-06 Richard Biener <rguenther@suse.de>
19364 PR tree-optimization/80928
19365 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
19366 (vect_analyze_loop_operations): Properly guard analysis for
19368 (vect_transform_loop): Likewise.
19369 (vect_analyze_loop_2): Also reset SLP type on PHIs.
19370 (vect_model_induction_cost): Do not cost for pure SLP.
19371 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
19372 of induction in inner loop vectorization.
19373 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
19374 (vect_get_and_check_slp_defs): Handle vect_induction_def.
19375 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
19377 (vect_analyze_slp_cost_1): Cost induction.
19378 (vect_detect_hybrid_slp_stmts): Handle PHIs.
19379 (vect_get_slp_vect_defs): Likewise.
19380 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
19381 (vect_transform_stmt): Handle SLP reductions.
19382 * tree-vectorizer.h (vectorizable_induction): Adjust.
19384 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
19386 * config/rs6000/rs6000.c (make_resolver_func): Update
19387 init_lowered_empty_function call.
19389 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
19391 * doc/invoke.texi: Document the -fprofile-abs-path option.
19392 * common.opt (fprofile-abs-path): New option.
19393 * gcov-io.h (gcov_write_filename): Declare.
19394 * gcov-io.c (gcov_write_filename): New function.
19395 * coverage.c (coverage_begin_function): Use gcov_write_filename.
19396 * profile.c (output_location): Likewise.
19398 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19400 * shring-wrap.c: Revert accidental commit.
19402 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
19404 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
19406 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19408 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
19410 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
19411 profile in callgraph edge.
19412 * profile-count.h (apply_probability): If THIS is 0, then result is 0
19413 (apply_scale): Likewise.
19414 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
19415 Also scale profile when inlining function with zero profile.
19416 (initialize_cfun): Update exit block profile even when it is zero.
19417 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
19418 when profile is read.
19420 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
19422 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
19423 (CLONE_*): New constants to define the processors we can generate
19424 code for with the target_clone attribute.
19425 (rs6000_clone_map): New array to identify which clone processors
19426 the current program is running on.
19427 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
19428 target_clone attribute.
19429 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
19430 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
19431 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
19432 (cpu_expand_builtin): Add support for target_clone attribute.
19433 (rs6000_valid_attribute_p): Allow "default" attribute.
19434 (get_decl_name): New debug function to simplify printing the
19435 current function name in debugging statements.
19436 (rs6000_clone_priority): New functions to support the target_clone
19437 attribute, and be able to generate code to switch between ISA 2.05
19438 through ISA 3.0 (power6 through power9).
19439 (rs6000_compare_version_priority): Likewise.
19440 (rs6000_get_function_versions_dispatcher): Likewise.
19441 (make_resolver_func): Likewise.
19442 (add_condition_to_bb): Likewise.
19443 (dispatch_function_versions): Likewise.
19444 (rs6000_generate_version_dispatcher_body): Likewise.
19445 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
19446 (fusion_gpr_load_p): Fix a spacing issue.
19447 * doc/extend.texi (Common Function Attributes): Document that the
19448 PowerPC supports the target_clone attribute.
19450 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
19452 * config/arm/arm.h: explain F symbol found in description of ARM
19453 register allocation in its legend.
19455 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19457 * config/mips/frame-header-opt.c: Include profile-count.h.
19458 * config/riscv/riscv.c: Include profile-count.h
19460 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19462 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
19464 (sm_set_flag_if_changed): Add bbs field.
19465 (execute_sm_if_changed_flag_set): Pass BBS.
19466 (execute_sm): Update.
19468 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19470 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
19473 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19475 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
19476 (peephole2): New peephole2 to emit the above.
19477 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
19479 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19481 * config/aarch64/aarch64.c (define_peephole2 above
19482 *sub_<shift>_<mode>): New peephole.
19484 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19486 * config/i386/i386.c (make_resolver_func): Update.
19487 * Makefile.in: Add profile-count.h and profile-count.o
19488 * auto-profile.c (afdo_indirect_call): Update to new API.
19489 (afdo_set_bb_count): Update.
19490 (afdo_propagate_edge): Update.
19491 (afdo_propagate_circuit): Update.
19492 (afdo_calculate_branch_prob): Update.
19493 (afdo_annotate_cfg): Update.
19494 * basic-block.h: Include profile-count.h
19495 (struct edge_def): Turn count to profile_count.
19496 (struct basic_block_def): Likewie.
19497 (REG_BR_PROB_BASE): Move to profile-count.h
19498 (RDIV): Move to profile-count.h
19499 * bb-reorder.c (max_entry_count): Turn to profile_count.
19500 (find_traces): Update.
19501 (rotate_loop):Update.
19502 (connect_traces):Update.
19503 (sanitize_hot_paths):Update.
19504 * bt-load.c (migrate_btr_defs): Update.
19505 * cfg.c (RDIV): Remove.
19506 (init_flow): Use alloc_block.
19507 (alloc_block): Uninitialize count.
19508 (unchecked_make_edge): Uninitialize count.
19509 (check_bb_profile): Update.
19510 (dump_edge_info): Update.
19511 (dump_bb_info): Update.
19512 (update_bb_profile_for_threading): Update.
19513 (scale_bbs_frequencies_int): Update.
19514 (scale_bbs_frequencies_gcov_type): Update.
19515 (scale_bbs_frequencies_profile_count): New.
19516 * cfg.h (update_bb_profile_for_threading): Update.
19517 (scale_bbs_frequencies_profile_count): Declare.
19518 * cfgbuild.c (compute_outgoing_frequencies): Update.
19519 (find_many_sub_basic_blocks): Update.
19520 * cfgcleanup.c (try_forward_edges): Update.
19521 (try_crossjump_to_edge): Update.
19522 * cfgexpand.c (expand_gimple_tailcall): Update.
19523 (construct_exit_block): Update.
19524 * cfghooks.c (verify_flow_info): Update.
19525 (dump_bb_for_graph): Update.
19526 (split_edge): Update.
19527 (make_forwarder_block): Update.
19528 (duplicate_block): Update.
19529 (account_profile_record): Update.
19530 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
19531 (get_estimated_loop_iterations): Update.
19532 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
19533 (single_likely_exit): Update.
19534 * cfgloopmanip.c (scale_loop_profile): Update.
19536 (set_zero_probability): Update.
19537 (lv_adjust_loop_entry_edge): Update.
19538 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
19539 (purge_dead_edges): Update.
19540 (rtl_account_profile_record): Update.
19541 * cgraph.c (cgraph_node::create): Uninitialize count.
19542 (symbol_table::create_edge): Uninitialize count.
19543 (cgraph_update_edges_for_call_stmt_node): Update.
19544 (cgraph_edge::dump_edge_flags): Update.
19545 (cgraph_node::dump): Update.
19546 (cgraph_edge::maybe_hot_p): Update.
19547 * cgraph.h: Include profile-count.h
19548 (create_clone), create_edge, create_indirect_edge): Update.
19549 (cgraph_node): Turn count to profile_count.
19550 (cgraph_edge0: Likewise.
19551 (make_speculative, clone): Update.
19552 (create_edge): Update.
19553 (init_lowered_empty_function): Update.
19554 * cgraphclones.c (cgraph_edge::clone): Update.
19555 (duplicate_thunk_for_node): Update.
19556 (cgraph_node::create_clone): Update.
19557 * cgraphunit.c (cgraph_node::analyze): Update.
19558 (cgraph_node::expand_thunk): Update.
19559 * final.c (dump_basic_block_info): Update.
19560 * gimple-streamer-in.c (input_bb): Update.
19561 * gimple-streamer-out.c (output_bb): Update.
19562 * graphite.c (print_global_statistics): Update.
19563 (print_graphite_scop_statistics): Update.
19564 * hsa-brig.c: Include basic-block.h.
19565 * hsa-dump.c: Include basic-block.h.
19566 * hsa-gen.c (T sum_slice): Update.
19567 (convert_switch_statements):Update.
19568 * hsa-regalloc.c: Include basic-block.h.
19569 * ipa-chkp.c (chkp_produce_thunks): Update.
19570 * ipa-cp.c (struct caller_statistics): Update.
19571 (init_caller_stats): Update.
19572 (gather_caller_stats): Update.
19573 (ipcp_cloning_candidate_p): Update.
19574 (good_cloning_opportunity_p): Update.
19575 (get_info_about_necessary_edges): Update.
19576 (dump_profile_updates): Update.
19577 (update_profiling_info): Update.
19578 (update_specialized_profile): Update.
19579 (perhaps_add_new_callers): Update.
19580 (decide_about_value): Update.
19581 (ipa_cp_c_finalize): Update.
19582 * ipa-devirt.c (struct odr_type_warn_count): Update.
19583 (struct decl_warn_count): Update.
19584 (struct final_warning_record): Update.
19585 (possible_polymorphic_call_targets): Update.
19586 (ipa_devirt): Update.
19587 * ipa-fnsummary.c (redirect_to_unreachable): Update.
19588 * ipa-icf.c (sem_function::merge): Update.
19589 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
19590 * ipa-inline.c (compute_uninlined_call_time): Update.
19591 (compute_inlined_call_time): Update.
19592 (want_inline_small_function_p): Update.
19593 (want_inline_self_recursive_call_p): Update.
19594 (edge_badness): Update.
19595 (lookup_recursive_calls): Update.
19596 (recursive_inlining): Update.
19597 (inline_small_functions): Update.
19598 (dump_overall_stats): Update.
19599 (dump_inline_stats): Update.
19600 * ipa-profile.c (ipa_profile_generate_summary): Update.
19601 (ipa_propagate_frequency): Update.
19602 (ipa_profile): Update.
19603 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
19604 * ipa-utils.c (ipa_merge_profiles): Update.
19605 * loop-doloop.c (doloop_modify): Update.
19606 * loop-unroll.c (report_unroll): Update.
19607 (unroll_loop_runtime_iterations): Update.
19608 * lto-cgraph.c (lto_output_edge): Update.
19609 (lto_output_node): Update.
19610 (input_node): Update.
19611 (input_edge): Update.
19612 (merge_profile_summaries): Update.
19613 * lto-streamer-in.c (input_cfg): Update.
19614 * lto-streamer-out.c (output_cfg): Update.
19615 * mcf.c (create_fixup_graph): Update.
19616 (adjust_cfg_counts): Update.
19617 (sum_edge_counts): Update.
19618 * modulo-sched.c (sms_schedule): Update.
19619 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
19620 * predict.c (maybe_hot_count_p): Update.
19621 (probably_never_executed): Update.
19622 (dump_prediction): Update.
19623 (combine_predictions_for_bb): Update.
19624 (propagate_freq): Update.
19625 (handle_missing_profiles): Update.
19626 (counts_to_freqs): Update.
19627 (rebuild_frequencies): Update.
19628 (force_edge_cold): Update.
19629 * predict.h: Include profile-count.h
19630 (maybe_hot_count_p, counts_to_freqs): UPdate.
19631 * print-rtl-function.c: Do not include cfg.h
19632 * print-rtl.c: Include basic-block.h
19633 * profile-count.c: New file.
19634 * profile-count.h: New file.
19635 * profile.c (is_edge_inconsistent): Update.
19636 (correct_negative_edge_counts): Update.
19637 (is_inconsistent): Update.
19638 (set_bb_counts): Update.
19639 (read_profile_edge_counts): Update.
19640 (compute_frequency_overlap): Update.
19641 (compute_branch_probabilities): Update; Initialize and deinitialize
19643 (branch_prob): Update.
19644 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
19645 (edge_gcov_count): New.
19646 (bb_gcov_count): New.
19647 * shrink-wrap.c (try_shrink_wrapping): Update.
19648 * tracer.c (better_p): Update.
19649 * trans-mem.c (expand_transaction): Update.
19650 (ipa_tm_insert_irr_call): Update.
19651 (ipa_tm_insert_gettmclone_call): Update.
19652 * tree-call-cdce.c: Update.
19653 * tree-cfg.c (gimple_duplicate_sese_region): Update.
19654 (gimple_duplicate_sese_tail): Update.
19655 (gimple_account_profile_record): Update.
19656 (execute_fixup_cfg): Update.
19657 * tree-inline.c (copy_bb): Update.
19658 (copy_edges_for_bb): Update.
19659 (initialize_cfun): Update.
19660 (freqs_to_counts): Update.
19661 (copy_cfg_body): Update.
19662 (expand_call_inline): Update.
19663 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
19664 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
19665 (try_unroll_loop_completely): Update.
19666 (try_peel_loop): Update.
19667 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19668 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
19669 * tree-ssa-loop-split.c (connect_loops): Update.
19670 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
19671 * tree-ssa-reassoc.c (branch_fixup): Update.
19672 * tree-ssa-tail-merge.c (replace_block_by): Update.
19673 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
19674 (compute_path_counts): Update.
19675 (update_profile): Update.
19676 (recompute_probabilities): Update.
19677 (update_joiner_offpath_counts): Update.
19678 (estimated_freqs_path): Update.
19679 (freqs_to_counts_path): Update.
19680 (clear_counts_path): Update.
19681 (ssa_fix_duplicate_block_edges): Update.
19682 (duplicate_thread_path): Update.
19683 * tree-switch-conversion.c (case_bit_test_cmp): Update.
19684 (struct switch_conv_info): Update.
19685 * tree-tailcall.c (decrease_profile): Update.
19686 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
19687 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19688 * value-prof.c (check_counter): Update.
19689 (gimple_divmod_fixed_value): Update.
19690 (gimple_mod_pow2): Update.
19691 (gimple_mod_subtract): Update.
19692 (gimple_ic_transform): Update.
19693 (gimple_stringop_fixed_value): Update.
19694 * value-prof.h (gimple_ic): Update.
19696 2017-06-02 Carl Love <cel@us.ibm.com>
19698 * config/rs6000/rs6000-c: Add support for built-in functions
19699 vector double vec_doublee (vector signed int);
19700 vector double vec_doublee (vector unsigned int);
19701 vector double vec_doublee (vector float);
19702 vector double vec_doubleh (vector signed int);
19703 vector double vec_doubleh (vector unsigned int);
19704 vector double vec_doubleh (vector float);
19705 vector double vec_doublel (vector signed int);
19706 vector double vec_doublel (vector unsigned int);
19707 vector double vec_doublel (vector float);
19708 vector double vec_doubleo (vector signed int);
19709 vector double vec_doubleo (vector unsigned int);
19710 vector double vec_doubleo (vector float);.
19711 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
19712 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
19714 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
19715 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
19716 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
19718 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
19719 vec_doublel, vec_doubleh.
19720 * doc/extend.texi: Update the built-in documentation file for the
19721 new built-in functions.
19723 2017-06-02 David Malcolm <dmalcolm@redhat.com>
19726 * ipa-inline-analysis.c (free_growth_caches): Set
19727 edge_removal_hook_holder to NULL after removing it.
19729 2017-06-02 Sudakshina Das <sudi.das@arm.com>
19731 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
19732 comparision with zero.
19734 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
19735 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
19736 for early expansion of vec_min and vec_max builtins.
19737 (builtin_function_type): Add min/max unsigned variants to those
19738 identified as having unsigned arguments.
19740 2017-06-02 Olivier Hainque <hainque@adacore.com>
19742 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
19744 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19746 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
19747 Use VALL_F16 iterator rather than VALL.
19749 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19751 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
19752 Emit CBNZ inside loop when doing a strong exchange and comparing
19753 against zero. Generate the CC flags after the loop.
19755 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
19757 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
19758 (dl_section_ref): New.
19759 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
19760 On AIX, append an expression to subtract the size of the
19761 section length to dl_section_ref.
19763 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
19765 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
19766 for early expansion of vector absolute builtins.
19768 2017-06-02 Richard Biener <rguenther@suse.de>
19770 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
19771 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
19773 2017-06-02 Richard Biener <rguenther@suse.de>
19775 PR tree-optimization/80948
19776 * tree-tailcall.c (find_tail_calls): Track stmts to move in
19777 stmt order as well.
19779 2017-06-02 Richard Biener <rguenther@suse.de>
19781 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
19783 * tree-vect-stmts.c (process_use): Do not mark backedge defs
19784 for inductions as relevant.
19786 2017-06-02 Richard Biener <rguenther@suse.de>
19788 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
19789 (vectorizable_induction): ... this. Remove dead code.
19791 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
19793 * builtins. (expand_builtin_alloca): Remove second parameter and
19794 infer its value from the first parameter instead.
19795 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
19797 2017-06-02 Jakub Jelinek <jakub@redhat.com>
19799 PR rtl-optimization/80903
19800 * loop-doloop.c (add_test): Unshare sequence.
19802 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
19804 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
19806 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
19808 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
19810 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
19811 xlogue_layout::get_instance, logue_layout::xlogue_layout,
19812 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
19813 (xlogue_layout::get_stub_rtx): Make static.
19814 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
19815 (xlogue_layout::compute_stub_managed_regs): Rename to...
19816 (xlogue_layout::count_stub_managed_regs): ...this.
19817 (xlogue_layout::is_stub_managed_reg): New function.
19818 (xlogue_layout::m_stub_names): Rename to...
19819 (xlogue_layout::s_stub_names): ...this, make static.
19820 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
19821 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
19822 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
19823 xlogue_layout::s_stub_names): Instantiate statics.
19824 (stub_managed_regs): Remove.
19825 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
19826 (disable_call_ms2sysv_xlogues): Rename to...
19827 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
19828 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
19830 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
19831 change after reload_completed.
19832 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
19834 (ix86_expand_prologue): Likewise.
19835 (ix86_expand_epilogue): Likewise.
19836 (ix86_expand_split_stack_prologue): Likewise.
19837 (ix86_compute_frame_layout): Remove frame parameter ...
19838 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
19839 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
19841 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
19842 (ix86_frame): Move from here ...
19843 * config/i386/i386.h (ix86_frame): ... to here.
19844 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
19845 complete ix86_frame data structure instead. Remove some_ld_name.
19847 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
19849 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
19850 symbols that hold a DECL_VALUE_EXPR.
19852 2017-06-01 Martin Jambor <mjambor@suse.cz>
19854 PR tree-optimization/80898
19855 * tree-sra.c (process_subtree_disqualification): Removed.
19856 (disqualify_candidate): Do not acll
19857 process_subtree_disqualification.
19858 (subtree_mark_written_and_enqueue): New function.
19859 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
19860 RHS has been disqualified and re-queue LHS if necessary. Apart
19861 from that, ignore disqualified RHS.
19863 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19865 * config/s390/s390.c (s390_emit_epilogue): Disable early return
19866 address fetch for z10 or later.
19868 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19870 * config/arc/arc.md (tst_movb): Add guard when splitting.
19872 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19874 * config/arc/arc.c (arc_can_eliminate): Test against
19875 arc_frame_pointer_needed.
19877 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19879 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
19880 to prevent store reordering.
19881 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
19882 (type): Add block type.
19883 (stack_tie): Define special instruction to be used in
19886 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19888 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
19889 constraint. It is not valid for the pattern.
19890 (noncommutative_binary_comparison): Likewise.
19892 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19894 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
19897 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19899 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
19900 be used by the reg-alloc.
19902 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19904 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
19905 reg-alloc when having mul64 or mul32x16 instructions.
19906 (mulsidi3): Likewise.
19907 (umulsidi3): Likewise.
19908 (mulsi32x16): New pattern.
19909 (mulsi64): Likewise.
19910 (mulsidi64): Likewise.
19911 (umulsidi64): Likewise.
19912 (MUL32x16_REG): Define.
19913 (mul64_600): Use MUL32x16_REG.
19914 (mac64_600): Likewise.
19915 (umul64_600): Likewise.
19916 (umac64_600): Likewise.
19918 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19920 * config/arc/arc.md (mulsi3_700): Make it commutative.
19922 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
19924 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
19926 (*sign_extendsidi2_insn): Likewise for movstosw.
19928 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
19930 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
19931 the type of the input discriminant value. Convert the
19932 discriminant value of signedness vary.
19934 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
19936 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
19937 Add to -Wall section.
19939 2017-06-01 Richard Biener <rguenther@suse.de>
19941 PR middle-end/66313
19942 * fold-const.c (fold_plusminus_mult_expr): If the factored
19943 factor may be zero use a wrapping type for the inner operation.
19944 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
19945 and handle moved defs.
19946 (process_assignment): Properly guard the unary op case. Return a
19947 tri-state indicating that moving the stmt before the call may allow
19948 to continue. Pass through to_move.
19949 (find_tail_calls): Handle moving unrelated defs before
19952 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
19955 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
19956 splitter result in the canonical way.
19958 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
19960 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
19961 also for 32bit target. Update insn attributes.
19962 (zero-extendsidi2 splitter): Allow all registers for operand 1.
19964 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
19966 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
19967 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
19968 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
19969 (_mm_maskz_min_ss): New intrinsics.
19971 2017-05-31 Martin Liska <mliska@suse.cz>
19973 * tree-vect-loop.c (vect_create_epilog_for_reduction):
19974 Change comment style to one we normally use.
19975 (vectorizable_reduction): Likewise.
19976 (vectorizable_induction): Likewise.
19977 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19978 (vectorizable_call): Likewise.
19979 (vectorizable_simd_clone_call): Likewise.
19980 (vectorizable_conversion): Likewise.
19981 (vectorizable_assignment): Likewise.
19982 (vectorizable_shift): Likewise.
19983 (vectorizable_operation): Likewise.
19984 (vectorizable_store): Likewise.
19985 (vectorizable_load): Likewise.
19986 * tree-vectorizer.h: Likewise.
19988 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
19990 * passes.c (emergency_dump_function): New.
19991 * tree-pass.h (emergency_dump_function): Declare.
19992 * plugin.c (plugins_internal_error_function): Remove.
19993 * plugin.h (plugins_internal_error_function): Remove declaration.
19994 * toplev.c (internal_error_function): New static function. Use it...
19995 (general_init): ...here.
19997 2017-05-31 Graham Markall <graham.markall@embecosm.com>
19999 * config/arc/arc.c (arc_print_operand): Handle constant operands.
20000 (arc_rtx_costs): Add costs for new patterns.
20001 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
20002 * config/arc/predicates.md: Add _1_2_3_operand predicate.
20004 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
20006 * tree-ssa-strlen.c (get_next_strinfo): New function.
20007 (get_stridx_plus_constant): Use it.
20008 (zero_length_string): Likewise.
20009 (adjust_related_strinfos): Likewise.
20010 (adjust_last_stmt): Likewise.
20012 2017-05-31 Richard Biener <rguenther@suse.de>
20015 * config/i386/i386.c (ix86_expand_builtin): Remove assert
20016 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
20018 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
20020 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
20021 loop_vinfo argument and use of dependence distance vectors.
20022 Check instead whether the two references differ only in their
20023 initial value and assume that they have the same alignment if the
20024 difference is a multiple of the vector alignment.
20025 (vect_analyze_data_refs_alignment): Update call accordingly.
20027 2017-05-31 Martin Liska <mliska@suse.cz>
20030 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
20032 2017-05-31 Bin Cheng <bin.cheng@arm.com>
20034 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
20035 (create_intersect_range_checks): Move from ...
20036 * tree-data-ref.c (create_intersect_range_checks_index)
20037 (create_intersect_range_checks): ... to here.
20038 (create_runtime_alias_checks): New function factored from ...
20039 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
20040 here. Call above function.
20041 * tree-data-ref.h (create_runtime_alias_checks): New function.
20043 2017-05-31 Bin Cheng <bin.cheng@arm.com>
20045 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
20046 segment length for dr_b and compute it in wide_int.
20048 2017-05-31 Richard Biener <rguenther@suse.de>
20050 PR tree-optimization/80906
20051 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
20052 and pass through iv_map.
20053 (copy_bb_and_scalar_dependences): Adjust.
20054 (translate_pending_phi_nodes): Likewise.
20055 (copy_loop_close_phi_args): Handle code-generating IVs instead
20058 2017-05-30 David Malcolm <dmalcolm@redhat.com>
20060 * diagnostic-color.c (color_dict): Add "type-diff".
20061 (parse_gcc_colors): Update comment.
20062 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
20063 -fdiagnostics-show-template-tree and -fno-elide-type.
20064 (GCC_COLORS): Add type-diff to example.
20066 (-fdiagnostics-show-template-tree): New.
20067 (-fno-elide-type): New.
20068 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
20069 the pp_format_decoder callback. Call any m_format_postprocessor's
20071 (pretty_printer::pretty_printer): Initialize
20072 m_format_postprocessor.
20073 (pretty_printer::~pretty_printer): Delete any
20074 m_format_postprocessor.
20075 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
20076 (class format_postprocessor): New class.
20077 (struct pretty_printer::format_decoder): Document the new parameters.
20078 (struct pretty_printer::m_format_postprocessor): New field.
20079 * tree-diagnostic.c (default_tree_printer): Update for new
20080 bool and const char ** params.
20081 * tree-diagnostic.h (default_tree_printer): Likewise.
20083 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
20085 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
20086 (lwa_operand): Delete rs6000_gen_cell_microcode test.
20087 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
20088 rs6000_gen_cell_microcode code.
20089 (rs6000_final_prescan_insn): Delete.
20090 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
20091 "warn-cell-microcode" entries.
20092 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
20093 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
20094 throughout. Change cc_reg_not_micro_cr0_operand to
20095 cc_reg_not_cr0_operand throughout.
20096 (*extendhi<mode>2_noload): Delete.
20097 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
20098 (mwarn-cell-microcode): Delete.
20099 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
20100 -mgen-cell-microcode and -mwarn-cell-microcode.
20102 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
20105 * config/i386/constraints.md (Yd): New constraint.
20107 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
20108 and (?Yd, r) alternatives. Update insn attributes.
20109 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
20110 and (?*Yd, r) alternatives. Update insn attributes.
20111 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
20113 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
20115 * gimplify.c (gimplify_modify_expr): Don't create a
20116 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
20119 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
20121 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
20123 2017-05-30 Richard Biener <rguenther@suse.de>
20125 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
20126 and reduc_def fields.
20127 (STMT_VINFO_REDUC_TYPE): New define.
20128 (STMT_VINFO_REDUC_DEF): Likewise.
20129 (vect_force_simple_reduction): Adjust prototype.
20130 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
20131 (vect_is_simple_reduction): Remove check_reduction argument.
20132 (vect_force_simple_reduction): Adjust and set
20133 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
20134 (vectorizable_reduction): Do not re-do reduction analysis
20135 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
20136 * tree-parloops.c (gather_scalar_reductions): Adjust.
20138 2017-05-30 Richard Biener <rguenther@suse.de>
20140 PR middle-end/80901
20141 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
20144 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
20146 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
20147 Introduce unknown_misalignment parameter and remove vf.
20148 (vect_peeling_hash_get_lowest_cost):
20149 Pass unknown_misalignment parameter.
20150 (vect_enhance_data_refs_alignment):
20151 Fix unsupportable data ref treatment.
20153 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
20155 * tree-vect-data-refs.c (vect_get_data_access_cost):
20156 Workaround for SLP handling.
20157 (vect_enhance_data_refs_alignment):
20158 Compute costs for doing no peeling at all, compare to the best
20159 peeling costs so far and avoid peeling if cheaper.
20161 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
20163 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
20164 Return peeling info and set costs to zero for unlimited cost
20166 (vect_enhance_data_refs_alignment): Also inspect all datarefs
20167 with unknown misalignment. Compute and costs for unknown
20168 misalignment, compare them to the costs for known misalignment
20169 and choose the cheapest for peeling.
20171 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
20173 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
20174 (vect_get_peeling_costs_all_drs): Create function.
20175 (vect_peeling_hash_get_lowest_cost):
20176 Use vect_get_peeling_costs_all_drs.
20177 (vect_peeling_supportable): Create function.
20178 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
20180 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
20182 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
20183 DR_HAS_NEGATIVE_STEP.
20184 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
20185 (vect_enhance_data_refs_alignment): Use.
20186 (vect_duplicate_ssa_name_ptr_info): Use.
20187 * tree-vectorizer.h (dr_misalignment): Use.
20188 (known_alignment_for_access_p): Use.
20190 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20193 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
20195 (has_section_name): New function.
20197 2017-05-30 Martin Liska <mliska@suse.cz>
20200 * auto-profile.c (get_function_decl_from_block): Fix
20203 2017-05-30 Richard Biener <rguenther@suse.de>
20205 PR middle-end/80876
20206 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
20208 2017-05-30 Martin Liska <mliska@suse.cz>
20210 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
20211 * dumpfile.h (struct dump_file_info): Remove ctors.
20213 2017-05-30 Martin Liska <mliska@suse.cz>
20215 * predict.def: Fix GNU coding style.
20217 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
20219 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
20220 Mark 'to' argument with ATTRIBUTE_UNUSED.
20222 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
20224 * config/xtensa/xtensa.c (xtensa_emit_call): Use
20225 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
20226 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
20229 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
20231 * doc/install.texi (Options specification): Restore entry of
20232 --enable-sjlj-exceptions.
20234 2017-05-27 Michael Eager <eager@eagercon.com>
20237 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
20239 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
20241 * config/microblaze/microblaze.h
20242 (FIXED_REGISTERS): Update in macro.
20243 (CALL_USED_REGISTERS): Update in macro.
20245 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
20247 * doc/install.texi: Add links to macOS binary distributions.
20249 2017-05-27 Jakub Jelinek <jakub@redhat.com>
20253 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20255 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
20257 2017-05-26 Martin Liska <mliska@suse.cz>
20259 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
20261 2017-05-26 Martin Liska <mliska@suse.cz>
20263 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
20264 always leading ';; '.
20265 (dump_bb_info): Likewise.
20266 (brief_dump_cfg): Likewise.
20267 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
20268 * dumpfile.c: Remove usage of TDF_VERBOSE.
20269 * dumpfile.h (enum dump_kind): Likewise.
20270 (dump_gimple_bb_header): Do not use TDF_COMMENT.
20271 * print-tree.c (debug_verbose): Remove.
20272 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
20273 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
20274 * tree-diagnostic.c (default_tree_printer): Replace
20275 TDF_DIAGNOSTIC with TDF_SLIM.
20277 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20279 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
20280 in parameter loop, rather than loop_vinfo.
20281 (create_intersect_range_checks): Ditto.
20282 (vect_create_cond_for_alias_checks): Update call to above functions.
20284 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20286 PR tree-optimization/80815
20287 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
20288 for merging runtime alias checks. Handle negative DR_STEPs.
20290 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20292 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
20294 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
20295 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
20296 out code pruning runtime alias checks.
20297 * tree-data-ref.c (prune_runtime_alias_test_list): New function
20298 factored out from above.
20299 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
20301 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
20303 (prune_runtime_alias_test_list): New decalaration.
20305 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20307 * tree-vect-data-refs.c (compare_tree): Rename and move ...
20308 * tree-data-ref.c (data_ref_compare_tree): ... to here.
20309 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
20310 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
20311 (operator==, comp_dr_with_seg_len_pair): Ditto.
20312 (vect_prune_runtime_alias_test_list): Ditto.
20314 2017-05-26 Martin Liska <mliska@suse.cz>
20317 * params.def: Bound partial-inlining-entry-probability param.
20319 2017-05-26 Marek Polacek <polacek@redhat.com>
20322 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
20325 2017-05-26 Richard Biener <rguenther@suse.de>
20327 PR tree-optimization/80842
20328 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
20331 2017-05-26 Richard Biener <rguenther@suse.de>
20333 PR tree-optimization/80844
20334 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
20336 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
20338 * doc/md.texi (Machine Constraints): Update x86 family
20339 machine constraints section to match 'config/i386/constraints.md'.
20341 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
20343 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
20345 2017-05-25 Nathan Sidwell <nathan@acm.org>
20347 * doc/invoke.texi (--enable-languages): Update documentation.
20349 2017-05-25 Martin Liska <mliska@suse.cz>
20351 * dumpfile.c: Add TDF_FOLDING.
20352 * dumpfile.h (enum dump_kind): Likewise.
20353 * genmatch.c (dt_simplify::gen_1): Use it.
20355 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20357 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
20359 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20361 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
20362 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
20364 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20366 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
20367 * match.pd (X == C): Rewrite it here.
20368 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
20369 with_certain_nonzero_bits2): New predicates.
20370 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
20372 2017-05-24 Nathan Sidwell <nathan@acm.org>
20374 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
20377 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
20380 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
20382 * config/powerpcspe: New port. Files are copied from the rs6000
20383 port, with "rs6000" in filenames replaced by "powerpcspe".
20385 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
20387 PR rtl-optimization/80754
20388 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
20390 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
20392 * config/sparc/sparc.md (length): Return the correct value for -mflat
20393 sibcalls to match output_sibcall.
20395 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
20399 * config/rs6000/rs6000.c (struct machine_function): Add new field
20401 (rs6000_get_separate_components): Init that field, use it.
20402 (rs6000_components_for_bb): Use the field.
20404 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
20406 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
20408 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
20410 PR middle-end/80823
20411 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
20413 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20416 * config/s390/s390.c (s390_check_qrst_address): Check incoming
20417 address against address_operand predicate.
20418 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
20420 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
20422 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
20423 parameters passed indirectly.
20425 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
20427 * config/i386/i386.md (*movdi_internal): Remove SSE4
20428 alternative 18 (?r, *v). Update insn attributes.
20429 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
20430 Update insn attributes.
20431 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
20432 Update insn attributes.
20433 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
20434 alternative 1 (r, v). Remove isa attribute.
20435 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
20436 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
20437 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
20439 2017-05-23 Tom de Vries <tom@codesourcery.com>
20441 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
20444 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20446 * cgraphunit.c (symbol_table::process_new_functions): Update.
20447 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
20448 (inline_generate_summary): Rename to ...
20449 (ipa_fn_summary_generate): ... this one.
20450 (inline_read_summary): Rename to ...
20451 (ipa_fn_summary_read): ... this one.
20452 (inline_write_summary): Rename to ...
20453 (ipa_fn_summary_write): ... this one.
20454 (inline_free_summary): Rename to ...
20455 (ipa_free_fn_summary): ... this one.
20456 (pass_data_local_fn_summary, pass_local_fn_summary,
20457 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
20458 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
20459 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
20460 make_pass_ipa_fn_summary): New.
20461 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
20462 inline_write_summary, inline_free_summary): Remove.
20463 (ipa_free_fn_summary) : New.
20464 * ipa-inline.c (ipa_inline): Update.
20465 (pass_ipa_inline): Do not generate summaries.
20466 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
20468 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
20469 and add pass_ipa_fn_summary.
20470 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
20472 (make_pass_inline_parameters): Remove.
20474 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
20476 * omp-low.c (struct omp_context): Remove "default_kind" member.
20479 * omp-offload.c (execute_oacc_device_lower): Remove the
20480 parallelism dimensions function attributes for unparallelized
20481 OpenACC kernels constructs.
20483 2017-05-23 Martin Liska <mliska@suse.cz>
20485 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
20487 (cgraph_edge::make_speculative): Likewise.
20488 (cgraph_edge::resolve_speculation): Likewise.
20489 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
20490 (cgraph_node::dump): Likewise.
20491 * cgraph.h: Likewise.
20492 * cgraphunit.c (analyze_functions): Likewise.
20493 (symbol_table::compile): Likewise.
20494 * ipa-cp.c (print_all_lattices): Likewise.
20495 (determine_versionability): Likewise.
20496 (initialize_node_lattices): Likewise.
20497 (ipcp_verify_propagated_values): Likewise.
20498 (estimate_local_effects): Likewise.
20499 (update_profiling_info): Likewise.
20500 (create_specialized_node): Likewise.
20501 (perhaps_add_new_callers): Likewise.
20502 (decide_about_value): Likewise.
20503 (decide_whether_version_node): Likewise.
20504 (identify_dead_nodes): Likewise.
20505 (ipcp_store_bits_results): Likewise.
20506 * ipa-devirt.c (dump_targets): Likewise.
20507 (ipa_devirt): Likewise.
20508 * ipa-icf.c (sem_item::dump): Likewise.
20509 (sem_function::equals): Likewise.
20510 (sem_variable::equals): Likewise.
20511 (sem_item_optimizer::read_section): Likewise.
20512 (sem_item_optimizer::execute): Likewise.
20513 (congruence_class::dump): Likewise.
20514 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
20515 (dump_inline_summary): Likewise.
20516 (estimate_node_size_and_time): Likewise.
20517 (inline_analyze_function): Likewise.
20518 * ipa-inline-transform.c (inline_call): Likewise.
20519 * ipa-inline.c (report_inline_failed_reason): Likewise.
20520 (want_early_inline_function_p): Likewise.
20521 (edge_badness): Likewise.
20522 (update_edge_key): Likewise.
20523 (inline_small_functions): Likewise.
20524 * ipa-profile.c (ipa_profile): Likewise.
20525 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
20526 (ipa_make_edge_direct_to_target): Likewise.
20527 (remove_described_reference): Likewise.
20528 (ipa_impossible_devirt_target): Likewise.
20529 (propagate_controlled_uses): Likewise.
20530 (ipa_print_node_params): Likewise.
20531 (ipcp_transform_function): Likewise.
20532 * ipa-pure-const.c (pure_const_read_summary): Likewise.
20533 (propagate_pure_const): Likewise.
20534 * ipa-reference.c (generate_summary): Likewise.
20535 (read_write_all_from_decl): Likewise.
20536 (propagate): Likewise.
20537 (ipa_reference_read_optimization_summary): Likewise.
20538 * ipa-utils.c (ipa_merge_profiles): Likewise.
20539 * ipa.c (walk_polymorphic_call_targets): Likewise.
20540 (symbol_table::remove_unreachable_nodes): Likewise.
20541 (ipa_single_use): Likewise.
20542 * passes.c (execute_todo): Likewise.
20543 * predict.c (drop_profile): Likewise.
20544 * symtab.c (symtab_node::get_dump_name): New function.
20545 (symtab_node::dump_name): Likewise.
20546 (symtab_node::dump_asm_name): Likewise.
20547 (symtab_node::dump_references): Likewise.
20548 (symtab_node::dump_referring): Likewise.
20549 (symtab_node::dump_base): Likewise.
20550 (symtab_node::debug_symtab): Likewise.
20551 * tree-sra.c (convert_callers_for_node): Likewise.
20552 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
20553 * value-prof.c (init_node_map): Likewise.
20555 2017-05-23 Martin Liska <mliska@suse.cz>
20557 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
20558 and symtab_node::debug_symtab to symbol_table::debug.
20559 * cgraphunit.c (analyze_functions): Use the renamed function.
20560 (symbol_table::compile): Likewise.
20561 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
20562 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
20563 * passes.c (execute_todo): Likewise.
20564 * symtab.c (symbol_table::dump): New function.
20565 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
20567 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20569 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
20570 that nonconst implies exec.
20572 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20574 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
20575 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
20576 (inline_edge_summary_vec): Turn into ...
20577 (ipa_call_summaries): ... this one.
20578 (redirect_to_unreachable, edge_set_predicate,
20579 evaluate_properties_for_edge, inline_summary_alloc,
20580 reset_ipa_call_summary, reset_inline_summary,
20581 inline_summary_t::duplicate): Update.
20582 (inline_edge_duplication_hook): Turn to ...
20583 (ipa_call_summary_t::duplicate): ... this one.
20584 (inline_edge_removal_hook): Turn to ...
20585 (ipa_call_summary_t::remove): ... this one.
20586 (dump_inline_edge_summary): Turn to ...
20587 (dump_ipa_call_summary): ... this one.
20588 (estimate_function_body_sizes): Update.
20589 (inline_update_callee_summaries): Update.
20590 (remap_edge_change_prob): Update.
20591 (remap_edge_summaries): Update.
20592 (inline_merge_summary): Update.
20593 (do_estimate_edge_time): Update.
20594 (inline_generate_summary): Update.
20595 (inline_read_section): Update.
20596 (inline_read_summary): Update.
20597 (inline_free_summary): Update.
20598 * ipa-inline.c (can_inline_edge_p): Update.
20599 (compute_inlined_call_time): Update.
20600 (want_inline_small_function_p): Update.
20601 (edge_badness): Update.
20602 (early_inliner): Update.
20603 * ipa-inline.h (inline_edge_summary): Turn to ...
20604 (ipa_call_summary): ... this one.
20605 (ipa_call_summary_t): New class.
20606 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
20607 (ipa_call_summaries): New.
20608 (inline_edge_summary): Remove.
20609 (estimate_edge_growth): Update.
20610 * ipa-profile.c (ipa_propagate_frequency_1): Update.
20611 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
20612 * ipa-split.c (execute_split_functions): Update.
20613 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
20615 2017-05-23 Tom de Vries <tom@codesourcery.com>
20617 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
20618 attributes): Document rdrand effective target.
20620 2017-05-23 Tom de Vries <tom@codesourcery.com>
20622 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
20623 attributes): Sort alphabetically.
20625 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
20627 * config/avr/genmultilib.awk: Use gsub instead of gensub.
20629 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
20632 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
20633 V2DF/V2DI splat into two separate patterns, one that handles
20634 registers, and the other that only handles memory. Drop support
20635 for splatting from a GPR on ISA 2.07 and then splitting the
20636 splat into direct move and splat.
20637 (vsx_splat_<mode>_reg): Likewise.
20638 (vsx_splat_<mode>_mem): Likewise.
20640 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
20642 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
20644 2017-05-22 Jakub Jelinek <jakub@redhat.com>
20646 PR middle-end/80809
20647 * omp-low.c (finish_taskreg_remap): New function.
20648 (finish_taskreg_scan): If unit size of ctx->record_type
20649 is non-constant, unshare the size expression and replace
20650 decls in it with possible outer var refs.
20652 PR middle-end/80809
20653 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
20654 GOVD_SHARED rather than GOVD_PRIVATE with it.
20655 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
20656 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
20658 PR middle-end/80853
20659 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
20660 as last argument to build_outer_var_ref for pointer bases of array
20661 section reductions.
20663 2017-05-19 Martin Sebor <msebor@redhat.com>
20665 * print-tree.c (print_node): Print DECL_READ_P flag.
20667 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20669 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
20670 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
20671 * cgraph.c: Likewise.
20672 * cgraphunit.c: Likewise.
20673 * gengtype.c: Likewise.
20674 * ipa-cp.c: Likewise.
20675 * ipa-devirt.c: Likewise.
20676 * ipa-icf.c: Likewise.
20677 * ipa-predicate.c: Likewise.
20678 * ipa-profile.c: Likewise.
20679 * ipa-prop.c: Likewise.
20680 * ipa-split.c: Likewise.
20682 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
20683 edge_predicate_pool, dump_inline_hints,
20684 inline_summary::account_size_time, redirect_to_unreachable,
20685 edge_set_predicate, set_hint_predicate,
20686 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
20687 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
20688 inline_summary_t::remove, remap_hint_predicate_after_duplication,
20689 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
20690 ipa_call_summary_t::remove, initialize_growth_caches,
20691 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
20692 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
20693 mark_modified, unmodified_parm_1, unmodified_parm,
20694 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
20695 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
20696 compute_bb_predicates, will_be_nonconstant_expr_predicate,
20697 will_be_nonconstant_predicate, record_modified_bb_info,
20698 get_minimal_bb, record_modified, param_change_prob,
20699 phi_result_unknown_predicate, predicate_for_phi_result,
20700 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
20701 estimate_function_body_sizes, compute_inline_parameters,
20702 compute_inline_parameters_for_curren, pass_data_inline_parameters,
20703 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
20704 inline_update_callee_summaries, remap_edge_change_prob,
20705 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
20706 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
20707 inline_analyze_function, inline_summary_t::insert,
20708 inline_generate_summary, read_ipa_call_summary, inline_read_section,
20709 inline_read_summary, write_ipa_call_summary, inline_write_summary,
20710 inline_free_summary): Move to ipa-fnsummary.h
20711 (predicate_t): Remove.
20712 * ipa-fnsummary.c: New file.
20713 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
20714 (enum inline_hints_vals, inline_hints, agg_position_info,
20715 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
20716 inline_summaries, ipa_call_summary, ipa_call_summary_t,
20717 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
20718 dump_inline_summary, dump_inline_hints, inline_generate_summary,
20719 inline_read_summary, inline_write_summary, inline_free_summary,
20720 inline_analyze_function, initialize_inline_failed,
20721 inline_merge_summary, inline_update_overall_summary,
20722 compute_inline_parameters): Move to ipa-fnsummary.h
20723 * ipa-fnsummary.h: New file.
20724 * ipa-inline-transform.h: Include ipa-inline.h.
20725 * ipa-inline.c: LIkewise.
20727 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20729 * ipa-inline.c (edge_badness): Use inlined_time instead of
20730 inline_summaries->get.
20732 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20734 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
20736 2017-05-22 Nathan Sidwell <nathan@acm.org>
20738 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
20739 (fdump-lang): Document 'raw' option.
20740 * dumpfile.h (TDI_tu): Delete.
20741 * dumpfile.c (dump_files): Remove translation-unit.
20742 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
20744 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
20746 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
20747 command option from $(AWK) call.
20748 * config/avr/genmultilib.awk: Simplify and rewrite so that it
20749 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
20750 [FORMAT]: Remove handling of variable.
20751 * config/avr/t-multilib: Regenerate.
20753 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20755 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
20757 (dump_inline_summary): Do not print self_time.
20758 (estimate_function_body_sizes): Do not set self_time.
20759 (compute_inline_parameters): Likewise.
20760 (inline_read_section, inline_write_summary): Do not stream self_time.
20761 * ipa-inline.h (inline_summary): Drop self_time.
20763 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20765 * ipa-inline-analysis.c (account_size_time): Rename to ...
20766 (inline_summary::account_size_time): ... this one.
20767 (reset_ipa_call_summary): Turn to ...
20768 (ipa_call_summary::reset): ... this one.
20769 (reset_inline_summary): Turn to ...
20770 (inline_summary::reset): ... this one.
20771 (inline_summary_t::remove): Update.
20772 (inline_summary_t::duplicate): Update.
20773 (ipa_call_summary_t::remove): Update.
20774 (dump_inline_summary): Update.
20775 (estimate_function_body_sizes): Update.
20776 (compute_inline_parameters): Update.
20777 (estimate_node_size_and_time): Update.
20778 (inline_merge_summary): Update.
20779 (inline_update_overall_summary): Update.
20780 (inline_read_section): Update.
20781 (inline_write_summary): Update.
20782 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
20783 add account_size_time and reset member functions.
20784 (ipa_call_summary): Add reset function.
20785 * ipa-predicate.h (predicate::operator &): Constify.
20787 2017-05-22 Richard Biener <rguenther@suse.de>
20789 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
20791 2017-05-19 Jason Merrill <jason@redhat.com>
20793 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
20795 2017-05-19 Marek Polacek <polacek@redhat.com>
20798 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
20799 TYPE_OVERFLOW_WRAPS checks.
20801 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
20803 * tree-core.h (enum omp_clause_default_kind): Add
20804 "OMP_CLAUSE_DEFAULT_PRESENT".
20805 * tree-pretty-print.c (dump_omp_clause): Handle it.
20806 * gimplify.c (enum gimplify_omp_var_data): Add
20807 "GOVD_MAP_FORCE_PRESENT".
20808 (gimplify_adjust_omp_clauses_1): Map it to
20809 "GOMP_MAP_FORCE_PRESENT".
20810 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
20812 * gimplify.c (oacc_default_clause): Clarify.
20814 2017-05-19 Nathan Sidwell <nathan@acm.org>
20816 LANG_HOOK_REGISTER_DUMPS
20817 * toplev.c (general_init): Call register dump lang hook.
20818 * doc/invoke.texi: Document -fdump-lang option family.
20819 * dumpfile.c (dump_files): Remove class dump here.
20820 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
20821 * dumpfile.h (tree_dump_index): Remove TDI_class.
20822 * langhooks-def.h (lhd_register_dumps): Declare.
20823 (LANG_HOOKS_REGISTER_DUMPS): Define.
20824 (LANG_HOOKS_INITIALIZER): Add it.
20825 * langhooks.c (lhd_register_dumps): Define.
20826 * langhooks.h (struct lang_hooks): Add register_dumps.
20828 2017-05-19 Nathan Sidwell <nathan@acm.org>
20830 * context.h (context::set_passes): New.
20831 * context.c (context::context): Do not create pass manager.
20832 * toplev.c (general_init): Create pass manager here.
20834 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
20836 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
20837 use this splitter if two add or or instructions would also work for
20838 the constant we want to generate.
20840 2017-05-19 Richard Biener <rguenther@suse.de>
20843 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
20844 predicate evaluation.
20846 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
20848 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
20850 * ipa-inline.c (want_inline_small_function_p): Do not cast to
20853 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
20855 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
20856 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
20857 (inline_edge_summary_vec): Turn into ...
20858 (ipa_call_summaries): ... this one.
20859 (redirect_to_unreachable, edge_set_predicate,
20860 evaluate_properties_for_edge, inline_summary_alloc,
20861 reset_ipa_call_summary, reset_inline_summary,
20862 inline_summary_t::duplicate): Update.
20863 (inline_edge_duplication_hook): Turn to ...
20864 (ipa_call_summary_t::duplicate): ... this one.
20865 (inline_edge_removal_hook): Turn to ...
20866 (ipa_call_summary_t::remove): ... this one.
20867 (dump_inline_edge_summary): Turn to ...
20868 (dump_ipa_call_summary): ... this one.
20869 (estimate_function_body_sizes): Update.
20870 (inline_update_callee_summaries): Update.
20871 (remap_edge_change_prob): Update.
20872 (remap_edge_summaries): Update.
20873 (inline_merge_summary): Update.
20874 (do_estimate_edge_time): Update.
20875 (inline_generate_summary): Update.
20876 (inline_read_section): Update.
20877 (inline_read_summary): Update.
20878 (inline_free_summary): Update.
20879 * ipa-inline.c (can_inline_edge_p): Update.
20880 (compute_inlined_call_time): Update.
20881 (want_inline_small_function_p): Update.
20882 (edge_badness): Update.
20883 (early_inliner): Update.
20884 * ipa-inline.h (inline_edge_summary): Turn to ...
20885 (ipa_call_summary): ... this one.
20886 (ipa_call_summary_t): New class.
20887 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
20888 (ipa_call_summaries): New.
20889 (inline_edge_summary): Remove.
20890 (estimate_edge_growth): Update.
20891 * ipa-profile.c (ipa_propagate_frequency_1): Update.
20892 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
20893 * ipa-split.c (execute_split_functions): Update.
20894 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
20896 2017-05-19 Richard Biener <rguenther@suse.de>
20898 PR middle-end/80764
20899 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
20901 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
20903 * config/rs6000/rs6000.c (struct machine_function): Add field
20904 fpr_is_wrapped_separately.
20905 (rs6000_get_separate_components): Use 64 components. Handle the
20906 new FPR components.
20907 (rs6000_components_for_bb): Handle the FPR components.
20908 (rs6000_emit_prologue_components): Handle the FPR components.
20909 (rs6000_emit_epilogue_components): Handle the FPR components.
20910 (rs6000_set_handled_components): Handle the FPR components.
20911 (rs6000_emit_prologue): Don't output prologue code for those FPRs
20912 that are already separately shrink-wrapped.
20913 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
20914 that are already separately shrink-wrapped.
20916 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
20919 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
20922 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
20923 (define_peephole2 for Altivec d-form load): Add peepholes to catch
20924 cases where the register allocator uses a move and an offsettable
20925 memory operation to/from a FPR register on ISA 2.06/2.07.
20926 (define_peephole2 for Altivec d-form store): Likewise.
20928 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
20931 * config/i386/mmx.md (*mov<mode>_internal): Enable
20932 alternatives 11, 12, 13 and 14 also for 32bit targets.
20933 Remove alternatives 15, 16, 17 and 18.
20934 * config/i386/sse.md (vec_concatv2di): Change
20935 alternative (!x, *y) to (x, ?!*Yn).
20937 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
20939 * dumpfile.h (enum dump_kind): Remove stray comma.
20941 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
20943 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
20944 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
20945 predicate::num_conditions
20946 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
20947 (CHANGED): turn into predicate::changed.
20948 (agg_position_info): Move to ipa-predicate.h
20949 (add_condition, predicate::add_clause, predicate::operator &=,
20950 predicate::or_with, predicate::evaluate, predicate::probability,
20951 dump_condition, dump_clause, predicate::dump,
20952 predicate::remap_after_duplication, predicate::remap_after_inlining,
20953 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
20954 (evaluate_conditions_for_known_args): Update.
20955 (set_cond_stmt_execution_predicate): Update.
20956 * ipa-inline.h: Include ipa-predicate.h
20957 (condition, inline_param_summary, conditions, agg_position_info,
20958 predicate): Move to ipa-predicate.h
20959 * ipa-predicate.c: New file.
20960 * ipa-predicate.h: New file.
20962 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
20964 * final.c (leaf_function_p): Check we are not in a sequence.
20966 2017-05-18 Martin Liska <mliska@suse.cz>
20968 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
20969 * dumpfile.c (dump_register): Use new enum dump_kind.
20970 (get_dump_file_name): Likewise.
20971 (dump_enable_all): Likewise.
20972 (dump_switch_p_1): Likewise.
20973 (enable_rtl_dump_file): Remove usage of TDF_RTL.
20974 * dumpfile.h (enum dump_kind): New enum type.
20975 (struct dump_file_info): Create constructor and
20976 format fields and comments.
20977 * passes.c (pass_manager::register_one_dump_file):
20979 * statistics.c (statistics_early_init): Likewise.
20980 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
20981 TDF_TREE with TDF_SLIM.
20982 (gather_memory_references_ref): Likewise.
20984 2017-05-18 Martin Liska <mliska@suse.cz>
20986 * vec.h (struct vnull): Use it.
20988 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
20990 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
20991 (true_predicate, false_predicate, true_predicate_p,
20992 false_predicate_p): Remove.
20993 (single_cond_predicate, not_inlined_predicate): Turn to member function
20995 (add_condition): Update.
20996 (add_clause): Turn to...
20997 (predicate::add_clause): ... this one; update; allow passing NULL
20999 (and_predicates): Turn to ...
21000 (predicate::operator &=): ... this one.
21001 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
21002 (or_predicates): Turn to ...
21003 (predicate::or_with): ... this one.
21004 (evaluate_predicate): Turn to ...
21005 (predicate::evaluate): ... this one.
21006 (predicate_probability): Turn to ...
21007 (predicate::probability): ... this one.
21008 (dump_condition): Update.
21009 (dump_predicate): Turn to ...
21010 (predicate::dump): ... this one.
21011 (account_size_time): Update.
21012 (edge_set_predicate): Update.
21013 (set_hint_predicate): UPdate.
21014 (evaluate_conditions_for_known_args): Update.
21015 (evaluate_properties_for_edge): Update.
21016 (remap_predicate_after_duplication): Turn to...
21017 (predicate::remap_after_duplication): ... this one.
21018 (remap_hint_predicate_after_duplication): Update.
21019 (inline_summary_t::duplicate): UPdate.
21020 (dump_inline_edge_summary): Update.
21021 (dump_inline_summary): Update.
21022 (set_cond_stmt_execution_predicate): Update.
21023 (set_switch_stmt_execution_predicate): Update.
21024 (compute_bb_predicates): Update.
21025 (will_be_nonconstant_expr_predicate): Update.
21026 (will_be_nonconstant_predicate): Update.
21027 (phi_result_unknown_predicate): Update.
21028 (predicate_for_phi_result): Update.
21029 (array_index_predicate): Update.
21030 (estimate_function_body_sizes): Update.
21031 (estimate_node_size_and_time): Update.
21032 (estimate_ipcp_clone_size_and_time): Update.
21033 (remap_predicate): Rename to ...
21034 (predicate::remap_after_inlining): ... this one.
21035 (remap_hint_predicate): Update.
21036 (inline_merge_summary): Update.
21037 (inline_update_overall_summary): Update.
21038 (estimate_size_after_inlining): Update.
21039 (read_predicate): Rename to ...
21040 (predicate::stream_in): ... this one.
21041 (read_inline_edge_summary): Update.
21042 (write_predicate): Rename to ...
21043 (predicate::stream_out): ... this one.
21044 (write_inline_edge_summary): Update.
21045 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
21046 (clause_t): Turn to uint32_t
21047 (predicate): Turn to class; implement constructor and operators
21049 (size_time_entry): Update.
21050 (inline_summary): Update.
21051 (inline_edge_summary): Update.
21053 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
21055 * fold-const.c (fold_binary_loc): Move transformation...
21056 * match.pd (C - X CMP X): ... here.
21058 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
21060 * config/sparc/sparc.c (sparc_option_override): Set function
21061 alignment for -mcpu=niagara7 to 64 to match the I$ line.
21062 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
21064 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
21066 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
21068 2017-05-18 Marek Polacek <polacek@redhat.com>
21071 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
21072 (pass_ubsan::execute): Call gimple_assign_single_p instead of
21073 gimple_assign_load_p.
21075 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
21077 PR middle-end/80692
21078 * real.c (do_compare): Give decimal_do_compare preference over
21079 comparing just the signs.
21081 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
21083 * doc/md.texi (Canonicalization of Instructions): Describe the
21084 canonical form of instructions that inherently set a condition
21087 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
21089 PR middle-end/80775
21090 * tree-cfg.c: Move deletion of unreachable case statements to after
21091 the merging of consecutive case labels.
21093 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
21095 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
21096 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
21097 restoring of callee-saved registers.
21099 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
21101 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
21102 * config/visium/visium.c (single_set_and_flags): Likewise.
21103 * config/visium/visium.md (Substitutions): Likewise.
21105 2017-05-17 Martin Liska <mliska@suse.cz>
21107 * cfg.c: Introduce dump_flags_t type and
21108 use it instead of int type.
21110 * cfghooks.c: Likewise.
21111 * cfghooks.h (struct cfg_hooks): Likewise.
21112 * cfgrtl.c: Likewise.
21113 * cfgrtl.h: Likewise.
21114 * cgraph.c (cgraph_node::get_body): Likewise.
21115 * coretypes.h: Likewise.
21116 * domwalk.c: Likewise.
21117 * domwalk.h: Likewise.
21118 * dumpfile.c (struct dump_option_value_info): Likewise.
21119 (dump_enable_all): Likewise.
21120 (dump_switch_p_1): Likewise.
21121 (opt_info_switch_p): Likewise.
21122 * dumpfile.h (enum tree_dump_index): Likewise.
21123 (struct dump_file_info): Likewise.
21124 * genemit.c: Likewise.
21125 * generic-match-head.c: Likewise.
21126 * gengtype.c (open_base_files): Likewise.
21127 * gimple-pretty-print.c: Likewise.
21128 * gimple-pretty-print.h: Likewise.
21129 * graph.c (print_graph_cfg): Likewise.
21130 * graphite-scop-detection.c (dot_all_sese): Likewise.
21131 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
21132 * loop-unroll.c (report_unroll): Likewise.
21133 * passes.c (pass_manager::register_one_dump_file): Likewise.
21134 * print-tree.c: Likewise.
21135 * statistics.c: Likewise.
21136 * tree-cfg.c: Likewise.
21137 * tree-cfg.h: Likewise.
21138 * tree-dfa.c: Likewise.
21139 * tree-dfa.h: Likewise.
21140 * tree-dump.c (dump_function): Likewise.
21141 * tree-dump.h (struct dump_info): Likewise.
21142 * tree-pretty-print.c: Likewise.
21143 * tree-pretty-print.h: Likewise.
21144 * tree-ssa-live.c: Likewise.
21145 * tree-ssa-live.h: Likewise.
21146 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
21147 * tree-vect-loop.c: Likewise.
21148 * tree-vect-slp.c: Likewise.
21150 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
21151 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21153 PR tree-optimization/80457
21154 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
21155 of all arguments to a statement as scalar_to_vec operations.
21156 (vectorizable_call): Adjust call to vect_model_simple_cost for
21158 (vectorizable_conversion): Likewise.
21159 (vectorizable_assignment): Likewise.
21160 (vectorizable_shift): Likewise.
21161 (vectorizable_operation): Likewise.
21162 (vectorizable_comparison): Likewise.
21163 (vect_is_simple_cond): Record the def types for operands.
21164 (vectorizable_condition): Likewise, call vect_model_simple_cost.
21165 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
21166 for statement argument count.
21168 2017-05-16 Carl Love <cel@us.ibm.com>
21170 * config/rs6000/rs6000-c: Add support for built-in functions
21171 vector unsigned long long vec_bperm (vector unsigned long long,
21172 vector unsigned char)
21173 vector signed long long vec_mule (vector signed int,
21175 vector unsigned long long vec_mule (vector unsigned int,
21176 vector unsigned int)
21177 vector signed long long vec_mulo (vector signed int,
21179 vector unsigned long long vec_mulo (vector unsigned int,
21180 vector unsigned int)
21181 vector signed char vec_sldw (vector signed char,
21182 vector signed char,
21184 vector unsigned char vec_sldw (vector unsigned char,
21185 vector unsigned char,
21187 vector signed short vec_sldw (vector signed short,
21188 vector signed short,
21190 vector unsigned short vec_sldw (vector unsigned short,
21191 vector unsigned short,
21193 vector signed int vec_sldw (vector signed int,
21196 vector unsigned int vec_sldw (vector unsigned int,
21197 vector unsigned int,
21199 vector signed long long vec_sldw (vector signed long long,
21200 vector signed long long,
21202 vector unsigned long long vec_sldw (vector unsigned long long,
21203 vector unsigned long long,
21205 * config/rs6000/rs6000-c: Add support for built-in functions
21206 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
21207 * config/rs6000/altivec.h: Add defintion for vec_sldw.
21208 * doc/extend.texi: Update the built-in documentation for the
21209 new built-in functions.
21211 2017-05-16 Marek Polacek <polacek@redhat.com>
21215 * tree.c (save_expr): Don't fold the expression.
21217 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
21219 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
21220 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
21221 and (?*y,m). Update insn attributes.
21223 2017-05-16 Martin Liska <mliska@suse.cz>
21225 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
21226 flags argument of print_gimple_stmt, print_gimple_expr,
21227 print_generic_stmt and print_generic_expr.
21228 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
21229 * coretypes.h: Likewise.
21230 * except.c (dump_eh_tree): Likewise.
21231 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
21232 * gimple-pretty-print.h: Likewise.
21233 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
21234 (backprop::push_to_worklist): Likewise.
21235 (backprop::pop_from_worklist): Likewise.
21236 (backprop::process_use): Likewise.
21237 (backprop::intersect_uses): Likewise.
21238 (note_replacement): Likewise.
21239 * gimple-ssa-store-merging.c
21240 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
21241 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
21242 (pass_store_merging::execute): Likewise.
21243 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
21244 (ssa_base_cand_dump_callback): Likewise.
21245 (dump_incr_vec): Likewise.
21246 (replace_refs): Likewise.
21247 (replace_mult_candidate): Likewise.
21248 (create_add_on_incoming_edge): Likewise.
21249 (create_phi_basis): Likewise.
21250 (insert_initializers): Likewise.
21251 (all_phi_incrs_profitable): Likewise.
21252 (introduce_cast_before_cand): Likewise.
21253 (replace_one_candidate): Likewise.
21254 * gimplify.c (gimplify_expr): Likewise.
21255 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
21256 (set_rename): Likewise.
21257 (rename_uses): Likewise.
21258 (copy_loop_phi_nodes): Likewise.
21259 (add_close_phis_to_merge_points): Likewise.
21260 (copy_loop_close_phi_args): Likewise.
21261 (copy_cond_phi_args): Likewise.
21262 (graphite_copy_stmts_from_block): Likewise.
21263 (translate_pending_phi_nodes): Likewise.
21264 * graphite-poly.c (print_pdr): Likewise.
21265 (dump_gbb_cases): Likewise.
21266 (dump_gbb_conditions): Likewise.
21267 (print_scop_params): Likewise.
21268 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
21269 (build_cross_bb_scalars_use): Likewise.
21270 (gather_bbs::before_dom_children): Likewise.
21271 * hsa-dump.c (dump_hsa_immed): Likewise.
21272 * ipa-cp.c (print_ipcp_constant_value): Likewise.
21273 (get_replacement_map): Likewise.
21274 * ipa-inline-analysis.c (dump_condition): Likewise.
21275 (estimate_function_body_sizes): Likewise.
21276 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
21277 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
21278 * ipa-prop.c (ipa_dump_param): Likewise.
21279 (ipa_print_node_jump_functions_for_edge): Likewise.
21280 (ipa_modify_call_arguments): Likewise.
21281 (ipa_modify_expr): Likewise.
21282 (ipa_dump_param_adjustments): Likewise.
21283 (ipa_dump_agg_replacement_values): Likewise.
21284 (ipcp_modif_dom_walker::before_dom_children): Likewise.
21285 * ipa-pure-const.c (check_stmt): Likewise.
21286 (pass_nothrow::execute): Likewise.
21287 * ipa-split.c (execute_split_functions): Likewise.
21288 * omp-offload.c (dump_oacc_loop_part): Likewise.
21289 (dump_oacc_loop): Likewise.
21290 * trans-mem.c (tm_log_emit): Likewise.
21291 (tm_memopt_accumulate_memops): Likewise.
21292 (dump_tm_memopt_set): Likewise.
21293 (dump_tm_memopt_transform): Likewise.
21294 * tree-cfg.c (gimple_verify_flow_info): Likewise.
21295 (print_loop): Likewise.
21296 * tree-chkp-opt.c (chkp_print_addr): Likewise.
21297 (chkp_gather_checks_info): Likewise.
21298 (chkp_get_check_result): Likewise.
21299 (chkp_remove_check_if_pass): Likewise.
21300 (chkp_use_outer_bounds_if_possible): Likewise.
21301 (chkp_reduce_bounds_lifetime): Likewise.
21302 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
21303 (chkp_mark_completed_bounds): Likewise.
21304 (chkp_register_incomplete_bounds): Likewise.
21305 (chkp_mark_invalid_bounds): Likewise.
21306 (chkp_maybe_copy_and_register_bounds): Likewise.
21307 (chkp_build_returned_bound): Likewise.
21308 (chkp_get_bound_for_parm): Likewise.
21309 (chkp_build_bndldx): Likewise.
21310 (chkp_get_bounds_by_definition): Likewise.
21311 (chkp_generate_extern_var_bounds): Likewise.
21312 (chkp_get_bounds_for_decl_addr): Likewise.
21313 * tree-chrec.c (chrec_apply): Likewise.
21314 * tree-data-ref.c (dump_data_reference): Likewise.
21315 (dump_subscript): Likewise.
21316 (dump_data_dependence_relation): Likewise.
21317 (analyze_overlapping_iterations): Likewise.
21318 * tree-inline.c (expand_call_inline): Likewise.
21319 (tree_function_versioning): Likewise.
21320 * tree-into-ssa.c (dump_defs_stack): Likewise.
21321 (dump_currdefs): Likewise.
21322 (dump_names_replaced_by): Likewise.
21323 (dump_update_ssa): Likewise.
21324 (update_ssa): Likewise.
21325 * tree-object-size.c (pass_object_sizes::execute): Likewise.
21326 * tree-parloops.c (build_new_reduction): Likewise.
21327 (try_create_reduction_list): Likewise.
21328 (ref_conflicts_with_region): Likewise.
21329 (oacc_entry_exit_ok_1): Likewise.
21330 (oacc_entry_exit_single_gang): Likewise.
21331 * tree-pretty-print.h: Likewise.
21332 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
21333 (get_scalar_evolution): Likewise.
21334 (add_to_evolution): Likewise.
21335 (get_loop_exit_condition): Likewise.
21336 (analyze_evolution_in_loop): Likewise.
21337 (analyze_initial_condition): Likewise.
21338 (analyze_scalar_evolution): Likewise.
21339 (instantiate_scev): Likewise.
21340 (number_of_latch_executions): Likewise.
21341 (gather_chrec_stats): Likewise.
21342 (final_value_replacement_loop): Likewise.
21343 (scev_const_prop): Likewise.
21344 * tree-sra.c (dump_access): Likewise.
21345 (disqualify_candidate): Likewise.
21346 (create_access): Likewise.
21347 (reject): Likewise.
21348 (maybe_add_sra_candidate): Likewise.
21349 (create_access_replacement): Likewise.
21350 (analyze_access_subtree): Likewise.
21351 (analyze_all_variable_accesses): Likewise.
21352 (sra_modify_assign): Likewise.
21353 (initialize_constant_pool_replacements): Likewise.
21354 (find_param_candidates): Likewise.
21355 (decide_one_param_reduction): Likewise.
21356 (replace_removed_params_ssa_names): Likewise.
21357 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
21358 * tree-ssa-copy.c (dump_copy_of): Likewise.
21359 (copy_prop_visit_cond_stmt): Likewise.
21360 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
21361 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
21362 (record_equivalences_from_stmt): Likewise.
21363 * tree-ssa-dse.c (compute_trims): Likewise.
21364 (delete_dead_call): Likewise.
21365 (delete_dead_assignment): Likewise.
21366 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
21367 (forward_propagate_into_cond): Likewise.
21368 (pass_forwprop::execute): Likewise.
21369 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
21370 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
21372 (move_computations_worker): Likewise.
21373 (execute_sm): Likewise.
21374 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
21375 (remove_exits_and_undefined_stmts): Likewise.
21376 (remove_redundant_iv_tests): Likewise.
21377 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
21378 (adjust_iv_update_pos): Likewise.
21379 * tree-ssa-math-opts.c (bswap_replace): Likewise.
21380 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
21381 (value_replacement): Likewise.
21382 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
21383 * tree-ssa-pre.c (print_pre_expr): Likewise.
21384 (get_representative_for): Likewise.
21385 (create_expression_by_pieces): Likewise.
21386 (insert_into_preds_of_block): Likewise.
21387 (eliminate_insert): Likewise.
21388 (eliminate_dom_walker::before_dom_children): Likewise.
21389 (eliminate): Likewise.
21390 (remove_dead_inserted_code): Likewise.
21391 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
21392 * tree-ssa-reassoc.c (get_rank): Likewise.
21393 (eliminate_duplicate_pair): Likewise.
21394 (eliminate_plus_minus_pair): Likewise.
21395 (eliminate_not_pairs): Likewise.
21396 (undistribute_ops_list): Likewise.
21397 (eliminate_redundant_comparison): Likewise.
21398 (update_range_test): Likewise.
21399 (optimize_range_tests_var_bound): Likewise.
21400 (optimize_vec_cond_expr): Likewise.
21401 (rewrite_expr_tree): Likewise.
21402 (rewrite_expr_tree_parallel): Likewise.
21403 (linearize_expr): Likewise.
21404 (break_up_subtract): Likewise.
21405 (linearize_expr_tree): Likewise.
21406 (attempt_builtin_powi): Likewise.
21407 (attempt_builtin_copysign): Likewise.
21408 (transform_stmt_to_copy): Likewise.
21409 (transform_stmt_to_multiply): Likewise.
21410 (dump_ops_vector): Likewise.
21411 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
21412 (print_scc): Likewise.
21413 (set_ssa_val_to): Likewise.
21414 (visit_reference_op_store): Likewise.
21415 (visit_use): Likewise.
21416 (sccvn_dom_walker::before_dom_children): Likewise.
21417 (run_scc_vn): Likewise.
21418 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
21420 (expr_hash_elt::print): Likewise.
21421 (const_and_copies::pop_to_marker): Likewise.
21422 (const_and_copies::record_const_or_copy_raw): Likewise.
21423 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
21424 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
21425 (dump_predicates): Likewise.
21426 (find_uninit_use): Likewise.
21427 (warn_uninitialized_phi): Likewise.
21428 (pass_late_warn_uninitialized::execute): Likewise.
21429 * tree-ssa.c (verify_vssa): Likewise.
21430 (verify_ssa): Likewise.
21431 (maybe_optimize_var): Likewise.
21432 * tree-vrp.c (dump_value_range): Likewise.
21433 (dump_all_value_ranges): Likewise.
21434 (dump_asserts_for): Likewise.
21435 (register_edge_assert_for_2): Likewise.
21436 (vrp_visit_cond_stmt): Likewise.
21437 (vrp_visit_switch_stmt): Likewise.
21438 (vrp_visit_stmt): Likewise.
21439 (vrp_visit_phi_node): Likewise.
21440 (simplify_cond_using_ranges_1): Likewise.
21441 (fold_predicate_in): Likewise.
21442 (evrp_dom_walker::before_dom_children): Likewise.
21443 (evrp_dom_walker::push_value_range): Likewise.
21444 (evrp_dom_walker::pop_value_range): Likewise.
21445 (execute_early_vrp): Likewise.
21447 2017-05-16 Richard Biener <rguenther@suse.de>
21449 * dwarf2out.c (loc_list_from_tree_1): Do not create
21450 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
21452 2017-05-16 Richard Biener <rguenther@suse.de>
21454 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
21456 (note_variable_value_in_expr): If we resolved the decl ref
21457 do not push to the stack.
21459 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
21461 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
21462 operations in fast-math mode.
21463 (vaddq_f16): Likewise.
21464 (vmul_f16): Likewise.
21465 (vmulq_f16): Likewise.
21466 (vsub_f16): Likewise.
21467 (vsubq_f16): Likewise.
21468 * config/arm/neon.md (add<mode>3): New.
21470 (fma:<VH:mode>3): New. Also remove outdated comment.
21473 2017-05-16 Martin Liska <mliska@suse.cz>
21477 * ipa-devirt.c (warn_types_mismatch): Fix typo.
21478 (odr_types_equivalent_p): Likewise.
21480 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
21482 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
21484 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
21487 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
21488 non-interunit SSE move alternatives with '?'.
21489 (zero-extendsidi peephole2): New peephole to skip intermediate
21490 general register in SSE zero-extend sequence.
21492 2017-05-15 Jeff Law <law@redhat.com>
21494 * reorg.c (relax_delay_slots): Create a new variable to hold
21495 the temporary target rather than clobbering TARGET_LABEL.
21497 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
21498 missing argument to extract_bit_field call.
21499 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
21501 2017-05-15 Martin Liska <mliska@suse.cz>
21504 * gcc.c (process_command): Do not allow empty argument of -o option.
21506 2017-05-15 Renlin Li <renlin.li@arm.com>
21508 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
21509 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
21510 * config/aarch64/constraints.md (Usf): Add long call check.
21511 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
21512 (call_value): Likewise.
21513 (sibcall): Likewise.
21514 (sibcall_value): Likewise.
21516 (call_value_insn): New.
21517 (sibcall_insn): Update rtx pattern.
21518 (sibcall_value_insn): Likewise.
21519 (call_internal): Remove.
21520 (call_value_internal): Likewise.
21521 (sibcall_internal): Likewise.
21522 (sibcall_value_internal): Likewise.
21523 (call_reg): Likewise.
21524 (call_symbol): Likewise.
21525 (call_value_reg): Likewise.
21526 (call_value_symbol): Likewise.
21528 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
21531 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
21533 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
21535 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
21536 compatible with CCGOCmode and with CCZmode.
21538 2017-05-14 Martin Sebor <msebor@redhat.com>
21540 PR middle-end/77671
21541 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
21542 (gimple_fold_builtin_snprintf): Same.
21543 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
21544 (gimple_fold_builtin_snprintf): Same.
21545 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
21546 of character types.
21547 (is_call_safe): New function.
21548 (try_substitute_return_value): Call it.
21549 (try_simplify_call): New function.
21550 (pass_sprintf_length::handle_gimple_call): Call it.
21552 2017-05-14 Martin Sebor <msebor@redhat.com>
21554 PR middle-end/80669
21555 * builtins.c (expand_builtin_stpncpy): Simplify.
21557 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
21559 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
21560 * config/i386/i386.h
21561 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
21562 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
21563 (struct machine_function): Add new members call_ms2sysv,
21564 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
21565 (struct machine_frame_state): New fields sp_realigned and
21566 sp_realigned_offset.
21567 * config/i386/i386.c
21568 (enum xlogue_stub): New enum.
21569 (enum xlogue_stub_sets): New enum.
21570 (class xlogue_layout): New class.
21571 (struct ix86_frame): New fields stack_realign_allocate_offset,
21572 stack_realign_offset and outlined_save_offset. Modify comments to
21573 detail stack layout when using out-of-line stubs.
21574 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
21575 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
21576 -mcall-ms2sysv-xlogues.
21577 (stub_managed_regs): New static variable.
21578 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
21579 registers managed by out-of-line stub.
21580 (disable_call_ms2sysv_xlogues): New function.
21581 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
21582 m->call_ms2sysv when appropriate and compute frame layout for
21584 (sp_valid_at, fp_valid_at): New inline functions.
21585 (choose_basereg): New function.
21586 (choose_baseaddr): Add align parameter, use choose_basereg and modify
21588 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
21589 Use align parameter of choose_baseaddr to generated aligned SSE movs
21591 (pro_epilogue_adjust_stack): Modify to track
21592 machine_frame_state::sp_realigned.
21593 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
21594 (ix86_nsaved_sseregs): Likewise.
21595 (ix86_emit_save_regs): Likewise.
21596 (ix86_emit_save_regs_using_mov): Likewise.
21597 (ix86_emit_save_sse_regs_using_mov): Likewise.
21598 (get_scratch_register_on_entry): Likewise.
21599 (gen_frame_set): New function.
21600 (gen_frame_load): Likewise.
21601 (gen_frame_store): Likewise.
21602 (emit_outlined_ms2sysv_save): Likewise.
21603 (emit_outlined_ms2sysv_restore): Likewise.
21604 (ix86_expand_prologue): Modify stack re-alignment code and call
21605 emit_outlined_ms2sysv_save when appropriate.
21606 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
21607 parameter rtx_insn *insn, which allows the function to be used to only
21608 generate the notes.
21609 (ix86_expand_epilogue): Modify validity checks of frame and stack
21610 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
21611 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
21612 * config/i386/predicates.md
21613 (save_multiple): New predicate.
21614 (restore_multiple): Likewise.
21615 * config/i386/sse.md
21616 (save_multiple<mode>): New pattern.
21617 (save_multiple_realign<mode>): Likewise.
21618 (restore_multiple<mode>): Likewise.
21619 (restore_multiple_and_return<mode>): Likewise.
21620 (restore_multiple_leave_return<mode>): Likewise.
21621 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
21623 2017-05-14 Julia Koval <julia.koval@intel.com>
21625 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
21626 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
21627 (__builtin_ia32_xsetbv): New builtins.
21628 * config/i386/i386.c (ix86_expand_special_args_builtin):
21630 (ix86_expand_builtin): Special expand for new intrinsics.
21631 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
21632 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
21633 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
21635 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21637 * cfganal.c (inverted_post_order_compute): Change argument type
21639 * cfganal.h (inverted_post_order_compute): Adjust prototype.
21640 * df-core.c (rest_of_handle_df_initialize): Adjust.
21641 (rest_of_handle_df_finish): Likewise.
21642 (df_analyze_1): Likewise.
21643 (df_analyze): Likewise.
21644 (loop_inverted_post_order_compute): Change argument to be a vec *.
21645 (df_analyze_loop): Adjust.
21646 (df_get_n_blocks): Likewise.
21647 (df_get_postorder): Likewise.
21648 * df.h (struct df_d): Change field to be a vec.
21649 * lcm.c (compute_laterin): Adjust.
21650 (compute_available): Likewise.
21651 * lra-lives.c (lra_create_live_ranges_1): Likewise.
21652 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
21653 * tree-ssa-pre.c (compute_antic): Likewise.
21655 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21657 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
21658 (depth_first_search::depth_first_search): Change structure init
21659 function to this constructor.
21660 (depth_first_search::add_bb): Rename function to this member.
21661 (depth_first_search::execute): Likewise.
21662 (flow_dfs_compute_reverse_finish): Adjust.
21664 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21666 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
21667 (longest_simple_path): Likewise.
21668 * shrink-wrap.c (spread_components): Likewise.
21669 (disqualify_problematic_components): Likewise.
21670 (emit_common_heads_for_components): Likewise.
21671 (emit_common_tails_for_components): Likewise.
21672 (insert_prologue_epilogue_for_components): Likewise.
21674 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21676 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
21679 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21681 * df-core.c (df_set_blocks): Start using auto_bitmap.
21682 (df_compact_blocks): Likewise.
21683 * df-problems.c (df_rd_confluence_n): Likewise.
21684 * df-scan.c (df_insn_rescan_all): Likewise.
21685 (df_process_deferred_rescans): Likewise.
21686 (df_update_entry_block_defs): Likewise.
21687 (df_update_exit_block_uses): Likewise.
21688 (df_entry_block_bitmap_verify): Likewise.
21689 (df_exit_block_bitmap_verify): Likewise.
21690 (df_scan_verify): Likewise.
21691 * lra-constraints.c (lra_constraints): Likewise.
21692 (undo_optional_reloads): Likewise.
21693 (lra_undo_inheritance): Likewise.
21694 * lra-remat.c (calculate_gen_cands): Likewise.
21695 (do_remat): Likewise.
21696 * lra-spills.c (assign_spill_hard_regs): Likewise.
21697 (spill_pseudos): Likewise.
21698 * tree-ssa-pre.c (bitmap_set_and): Likewise.
21699 (bitmap_set_subtract_values): Likewise.
21701 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21703 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
21704 management with auto_bitmap.
21705 (fix_inter_tick): Likewise.
21706 (fix_recovery_deps): Likewise.
21707 * ira.c (add_store_equivs): Likewise.
21708 (find_moveable_pseudos): Likewise.
21709 (split_live_ranges_for_shrink_wrap): Likewise.
21710 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
21711 (rtx_reuse_manager::seen_def_p): Likewise.
21712 (rtx_reuse_manager::set_seen_def): Likewise.
21713 * print-rtl.h (class rtx_reuse_manager): Likewise.
21715 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21717 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
21719 (migrate_btr_def): Likewise.
21720 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
21721 * df-core.c (loop_post_order_compute): Likewise.
21722 (loop_inverted_post_order_compute): Likewise.
21723 * hsa-common.h: Likewise.
21724 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
21725 * init-regs.c (initialize_uninitialized_regs): Likewise.
21726 * ipa-inline.c (resolve_noninline_speculation): Likewise.
21727 (inline_small_functions): Likewise.
21728 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
21729 * ira.c (combine_and_move_insns): Likewise.
21730 (build_insn_chain): Likewise.
21731 * loop-invariant.c (find_invariants): Likewise.
21732 * lower-subreg.c (propagate_pseudo_copies): Likewise.
21733 * predict.c (tree_predict_by_opcode): Likewise.
21734 (predict_paths_leading_to): Likewise.
21735 (predict_paths_leading_to_edge): Likewise.
21736 (estimate_loops_at_level): Likewise.
21737 (estimate_loops): Likewise.
21738 * shrink-wrap.c (try_shrink_wrapping): Likewise.
21739 (spread_components): Likewise.
21740 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
21741 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
21742 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
21743 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
21744 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
21745 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
21746 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
21747 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
21748 (mark_threaded_blocks): Likewise.
21749 (thread_through_all_blocks): Likewise.
21750 * tree-ssa.c (verify_ssa): Likewise.
21751 (execute_update_addresses_taken): Likewise.
21752 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
21754 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21756 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
21758 (post_order_compute): Likewise.
21759 (inverted_post_order_compute): Likewise.
21760 (pre_and_rev_post_order_compute_fn): Likewise.
21762 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21764 * genrecog.c (int_set::int_set): Explicitly construct our
21765 auto_vec base class.
21766 * vec.h (auto_vec::auto_vec): New constructor.
21768 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21770 * bitmap.h (class auto_bitmap): New constructor taking
21771 bitmap_obstack * argument.
21773 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21775 * bitmap.h (class auto_bitmap): Change type of m_bits to
21776 bitmap_head, and adjust ctor / dtor and member operators.
21778 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
21780 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
21781 when returned register mode doesn't match original mode.
21783 2017-05-12 Jeff Law <law@redhat.com>
21784 Jakub Jelinek <jakub@redhat.com>
21786 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
21787 we look for cc setter after the compare-elim changes.
21788 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
21789 within the vector to match what compare-elim now expects.
21790 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
21791 (xorsi3_flags, one_cmplsi2_flags): Likewise.
21793 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
21794 after the compare-elim changes.
21795 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
21796 the vector to match what compare-elim now expects.
21797 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
21798 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
21799 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
21800 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
21801 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
21803 * config/visium/visium.c (single_set_and_flags): Fix where
21804 we look for cc setter after the compare-elim changes.
21805 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
21806 with the vector to match what compare-elim now expects.
21807 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
21808 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
21809 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
21810 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
21811 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
21812 (neg<mode>2_insn_set_overflow): Likewise.
21814 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
21816 PR middle-end/79794
21817 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
21818 maybe_expand_insn call, set ops[0].target. If still set after call,
21819 set alt_rtl. Add extra arg to recursive calls.
21820 (extract_bit_field): Add alt_rtl argument. Pass to
21822 * expmed.h (extract_bit_field): Fix prototype.
21823 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
21824 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
21825 to extract_bit_field_calls.
21826 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
21827 Pass alt_rtl to extract_bit_field calls.
21828 * calls.c (store_unaligned_arguments_into_psuedos)
21829 load_register_parameters): Pass extra NULL to extract_bit_field calls.
21830 * optabs.c (maybe_legitimize_operand): Clear op->target when call
21832 * optabs.h (struct expand_operand): Add target bitfield.
21834 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
21836 * compare-elim.c (try_eliminate_compare): Canonicalize
21837 operation with embedded compare to
21838 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
21839 (set (reg) (operation)].
21841 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
21843 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
21846 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
21847 cost of adding a carry flag for ADC instruction.
21848 [case MINUS]: Ignore the cost of subtracting a carry flag
21849 for SBB instruction.
21851 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
21853 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
21855 * config/rs6000/bmiintrin.h: New file.
21856 * config/rs6000/bmi2intrin.h: New file.
21857 * config/rs6000/x86intrin.h: New file.
21859 2017-05-12 Jeff Law <law@redhat.com>
21861 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
21864 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
21866 PR middle-end/80707
21867 * tree-cfg.c: Remove cfg edges of unreachable case statements.
21869 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21871 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
21872 early expansion of vector divide builtins.
21873 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
21874 builtins identified as having unsigned arguments.
21876 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21878 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
21879 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
21880 expansion of vector logical operations (and, andc, or, xor,
21883 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21885 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
21886 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
21888 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21890 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
21891 early GIMPLE expansion of vector multiplies.
21893 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21895 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
21896 TARGET_HAVE_MOVT conditional.
21897 (movt splitter): Likewise.
21899 2017-05-12 Richard Biener <rguenther@suse.de>
21901 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
21902 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
21903 Fold all stmts not inplace.
21905 2017-05-12 Richard Biener <rguenther@suse.de>
21907 PR tree-optimization/80713
21908 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
21909 inserted_exprs bit for not removed stmts.
21911 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
21913 PR middle-end/69921
21914 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
21915 parallelized" attribute for parallelized OpenACC kernels.
21916 * omp-offload.c (execute_oacc_device_lower): Use it.
21918 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
21919 Set "oacc kernels" attribute.
21920 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
21921 parameter. Adjust all users.
21922 (oacc_fn_attrib_kernels_p): Remove function.
21923 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
21924 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
21925 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
21926 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
21927 assert "oacc kernels" attribute is set.
21929 2017-05-11 Carl Love <cel@us.ibm.com>
21931 * config/rs6000/rs6000-c: Add support for built-in functions
21932 vector unsigned char vec_popcnt (vector signed char)
21933 vector unsigned char vec_popcnt (vector unsigned char)
21934 vector unsigned short vec_popcnt (vector signed short)
21935 vector unsigned short vec_popcnt (vector unsigned short)
21936 vector unsigned int vec_popcnt (vector signed int)
21937 vector unsigned int vec_popcnt (vector unsigned int)
21938 vector unsigned long long vec_popcnt (vector signed long long)
21939 vector unsigned long long vec_popcnt (vector unsigned long long)
21940 vector signed long long vec_slo (vector signed long long,
21941 vector signed char)
21942 vector signed long long vec_slo (vector signed long long,
21943 vector unsigned char)
21944 vector unsigned long long vec_slo (vector unsigned long long,
21945 vector signed char)
21946 vector unsigned long long vec_slo (vector unsigned long long,
21947 vector unsigned char)
21948 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
21949 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
21950 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
21951 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
21952 * doc/extend.texi: Update the built-in documentation file for the
21953 new built-in functions.
21955 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
21957 * attribs.h (sorted_attr_string): Move machine independent
21958 functions for target clone support from the i386 port to common
21959 code. Rename ix86_function_versions to common_function_versions.
21960 Rename make_name to make_unique_name.
21961 (common_function_versions): Likewise.
21962 (make_unique_name): Likewise.
21963 (make_dispatcher_decl): Likewise.
21964 (is_function_default_version): Likewise.
21965 * attribs.c (attr_strcmp): Likewise.
21966 (sorted_attr_string): Likewise.
21967 (common_function_versions): Likewise.
21968 (make_unique_name): Likewise.
21969 (make_dispatcher_decl): Likewise.
21970 (is_function_default_version): Likewise.
21971 * config/i386/i386.c (attr_strcmp): Likewise.
21972 (sorted_attr_string): Likewise.
21973 (ix86_function_versions): Likewise.
21974 (make_name): Likewise.
21975 (make_dispatcher_decl): Likewise.
21976 (is_function_default_version): Likewise.
21977 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
21979 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21982 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
21983 Account for direct move costs for vec_construct of integer
21986 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
21989 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
21990 (UNSPEC_STX_ATOMIC): Ditto.
21991 (loaddi_via_sse): New insn.
21992 (storedi_via_sse): Ditto.
21993 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
21994 Update corresponding peephole2 patterns.
21995 (atomic_storedi_fpu): Ditto.
21997 2017-05-11 Julia Koval <julia.koval@intel.com>
21999 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
22000 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
22002 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
22003 (__builtin_ia32_rsqrt14ss_mask): New builtins.
22004 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
22006 2017-05-11 Nathan Sidwell <nathan@acm.org>
22008 * graphite-poly.c: Include dumpfile.h.
22010 * dumpfle.h (dump_function): Declare here ...
22011 * tree-dump.h (dump_function): ... not here.
22012 * dumpfile.c: #include tree-cfg.h.
22013 (dump_function): Move here from ...
22014 * tree-dump.c (dump_function): ... here.
22015 * gimplify.c: #include splay-tree.h, not tree-dump.h.
22016 * graphite-poly.c: Don't include tree-dump.h.
22017 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
22018 * print-tree.c: Likewise.
22019 * stor-layout.c: Likewise.
22020 * tree-nested.c: Likewise.
22022 * dumpfile.c (dump_start): Use TDF_FLAGS.
22023 (dump_enable_all): Fix TDF_KIND check thinko.
22025 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
22027 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22028 array entries to represent two legal parameterizations of the
22029 overloaded __builtin_cmpb function, as represented by the
22030 P6_OV_BUILTIN_CMPB constant.
22031 (altivec_resolve_overloaded_builtin): Add special case handling
22032 for the __builtin_cmpb function, as represented by the
22033 P6_OV_BUILTIN_CMPB constant.
22034 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
22035 (BU_P6_64BIT_2): New macro.
22036 (BU_P6_OVERLOAD_2): New macro
22037 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
22038 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
22039 (CMPB): Add overload support to represent both 32-bit and 64-bit
22040 compare-bytes function.
22041 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
22042 support for TARGET_CMPB.
22043 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
22044 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
22045 documentation of the __builtin_cmpb overloaded built-in function.
22047 2017-05-11 Richard Biener <rguenther@suse.de>
22049 PR tree-optimization/80705
22050 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
22051 bases are not vectorizable.
22053 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22055 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
22056 when counting register pressure.
22058 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22060 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
22061 (iv_ca_more_deps): Renamed to ...
22062 (iv_ca_compare_deps): ... this.
22063 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
22065 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22067 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
22069 (determine_group_iv_costs): ... here.
22070 (find_inv_vars_cb): Record inv var if it's not recorded before.
22072 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22074 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
22075 (get_shiftadd_cost): Ditto.
22077 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22079 * tree-ssa-address.c: Include header file.
22080 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
22082 (add_to_parts): Refactor.
22083 (addr_to_parts): New parameter. Update use of move_hint_to_base.
22084 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
22087 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22089 PR tree-optimization/53090
22090 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
22092 (extract_cond_operands): Detect condition with IV on both sides
22093 and return COMP_IV_EXPR_2.
22094 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
22095 (rewrite_use_compare): Simplify by removing call to function
22096 extract_cond_operands.
22098 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22100 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
22101 (extract_cond_operands): Detect condition comparing against non-
22102 invariant bound and return appropriate enum value.
22103 (find_interesting_uses_cond): Update use of extract_cond_operands.
22104 Handle its return value accordingly.
22105 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
22107 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22109 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
22110 nonlinear iv_use computation in loop invariant sensitive way.
22112 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22114 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
22115 (find_iv_candidates): Call relate_compare_use_with_all_cands.
22117 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22119 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
22120 (dump_cand): Support iv_cand.inv_exprs.
22121 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
22123 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
22126 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22128 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
22130 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
22131 as local function. Include necessary header files.
22132 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
22134 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22136 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
22138 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22140 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
22141 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
22142 RSHIFT_EXPR and BIT_NOT_EXPR.
22144 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22146 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
22147 (adjust_setup_cost): New parameter supporting round up adjustment.
22148 (struct address_cost_data): Delete.
22149 (force_expr_to_var_cost): Don't bound cost with spill_cost.
22150 (split_address_cost, ptr_difference_cost): Delete.
22151 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
22152 (struct ainc_cost_data): New struct.
22153 (get_address_cost_ainc): New function.
22154 (get_address_cost, get_computation_cost): Reimplement.
22155 (determine_group_iv_cost_address): Record inv_expr for all uses of
22157 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
22158 (iv_ca_has_deps): Reimplemented to ...
22159 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
22161 (iv_ca_extend): Call iv_ca_more_deps.
22163 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22165 * tree-ssa-address.c (struct mem_address): Move to header file.
22166 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
22167 * tree-ssa-address.h (struct mem_address): Move from C file.
22168 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
22170 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22172 * tree-affine.h (aff_combination_type): New interface.
22173 (aff_combination_zero_p): Remove static.
22174 (aff_combination_const_p): New interface.
22175 (aff_combination_singleton_var_p): New interfaces.
22177 2017-05-11 Richard Biener <rguenther@suse.de>
22179 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
22180 Skip unreachable blocks and destinations.
22181 (eliminate): Move stmt removal and fixup ...
22182 (fini_eliminate): ... here. Skip inserted exprs.
22183 (pass_pre::execute): Move fini_pre after fini_eliminate.
22184 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
22185 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
22186 PRE to get rid of dead code that has invalid SSA form and
22187 split critical edges again.
22189 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22191 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
22193 2017-05-11 Richard Biener <rguenther@suse.de>
22195 * passes.c (execute_function_todo): Verify loops if they are
22196 said to be up-to-date.
22197 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
22198 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
22200 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
22203 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
22204 handle calling assemble_external ourself.
22207 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
22208 modes with zero size. Enhance comment.
22210 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22212 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
22213 built-ins for vec_xl and vec_xst with short and char pointer
22216 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
22218 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
22219 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
22220 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
22221 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
22222 (_mm_maskz_min_round_ss): New intrinsics.
22223 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
22224 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
22225 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
22226 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
22227 (__builtin_ia32_minss_mask_round): New builtins.
22228 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22229 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
22230 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
22232 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
22233 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
22235 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
22238 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
22240 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
22241 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
22242 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
22243 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
22244 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
22245 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
22246 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
22247 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22248 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
22249 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
22250 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
22251 (__builtin_ia32_mulss_mask_round): New builtins.
22252 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22253 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
22254 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
22256 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
22257 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
22259 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
22262 2017-05-10 Julia Koval <julia.koval@intel.com>
22264 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
22265 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
22266 (_mm256_setr_m128i): New intrinsics.
22268 2017-05-10 Julia Koval <julia.koval@intel.com>
22270 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
22271 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
22272 (_mm_maskz_rcp14_ss): New intrinsics.
22273 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
22274 (__builtin_ia32_rcp14ss_mask): New builtins.
22275 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
22277 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
22279 PR tree-optimization/51513
22280 * tree-cfg.c (gimple_seq_unreachable_p): New function.
22281 (assert_unreachable_fallthru_edge_p): Use it.
22282 (group_case_labels_stmt): Likewise.
22283 * tree-cfg.h: Prototype it.
22284 * stmt.c: Include cfghooks.h and tree-cfg.h.
22285 (emit_case_dispatch_table) <gap_label>: New local variable.
22286 Use it to fill dispatch table gaps.
22287 Test for default_label before updating probabilities.
22288 (expand_case) <default_label>: Remove unneeded initialization.
22289 Test for unreachable default case statement and remove its edge.
22290 Set default_label accordingly.
22291 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
22293 2017-05-10 Carl Love <cel@us.ibm.com>
22295 * config/rs6000/rs6000-c: Add support for built-in functions
22296 vector signed char vec_neg (vector signed char)
22297 vector signed short int vec_neg (vector short int)
22298 vector signed int vec_neg (vector signed int)
22299 vector signed long long vec_neg (vector signed long long)
22300 vector float vec_neg (vector float)
22301 vector double vec_neg (vector double)
22302 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
22304 * config/rs6000/altivec.h: Add define for vec_neg
22305 * doc/extend.texi: Update the built-in documentation for the
22306 new built-in functions.
22308 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22310 PR tree-optimization/77644
22311 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
22313 2017-05-10 Nathan Sidwell <nathan@acm.org>
22315 * dumpfile.h (TDI_lang_all): New.
22316 (TDF_KIND): New. Renumber others
22317 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
22319 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
22321 (get_dump_file_name): Adjust suffix generation.
22322 (dump_enable_all): Use TDF_KIND.
22323 * doc/invoke.texi (-fdump-lang-all): Document.
22325 * dumpfile.h: Tabify.
22327 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
22330 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
22331 Move member access before delete.
22333 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
22335 * tree-inline.c (expand_call_inline): Split block at stmt
22338 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
22341 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
22342 are now unused after splitting mov{sf,sd}_hardfloat.
22343 (f32_lr2): Likewise.
22344 (f32_lm): Likewise.
22345 (f32_lm2): Likewise.
22346 (f32_li): Likewise.
22347 (f32_li2): Likewise.
22348 (f32_lv): Likewise.
22349 (f32_sr): Likewise.
22350 (f32_sr2): Likewise.
22351 (f32_sm): Likewise.
22352 (f32_sm2): Likewise.
22353 (f32_si): Likewise.
22354 (f32_si2): Likewise.
22355 (f32_sv): Likewise.
22356 (f32_dm): Likewise.
22357 (f32_vsx): Likewise.
22358 (f32_av): Likewise.
22359 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
22360 For movsf, order stores so the VSX stores occur before the GPR
22361 store which encourages the register allocator to use a traditional
22362 FPR instead of a GPR. For movsd, order the stores so that the GPR
22363 store comes before the VSX stores to allow the power6 to work.
22364 This is due to the power6 not having a 32-bit integer store
22365 instruction from a FPR.
22366 (movsf_hardfloat): Likewise.
22367 (movsd_hardfloat): Likewise.
22369 2017-05-09 Martin Sebor <msebor@redhat.com>
22371 PR translation/80280
22372 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
22375 PR translation/80280
22376 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
22377 data member added in r247778.
22378 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
22380 2017-05-09 Nathan Sidwell <nathan@acm.org>
22382 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
22384 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
22387 2017-05-09 Marek Polacek <polacek@redhat.com>
22389 * doc/invoke.texi: Fix typo.
22391 2017-05-09 Richard Biener <rguenther@suse.de>
22393 * tree-vrp.c (vrp_val_is_max): Adjust comment.
22394 (vrp_val_is_min): Likewise.
22395 (set_value_range_to_value): Likewise.
22396 (set_value_range_to_nonnegative): Likewise.
22397 (gimple_assign_nonzero_p): Likewise.
22398 (gimple_stmt_nonzero_p): Likewise.
22399 (vrp_int_const_binop): Likewise. Remove unreachable case.
22400 (adjust_range_with_scev): Adjust comments.
22401 (compare_range_with_value): Likewise.
22402 (extract_range_from_phi_node): Likewise.
22403 (test_for_singularity): Likewise.
22405 2017-05-09 Richard Biener <rguenther@suse.de>
22407 * tree-vrp.c (get_single_symbol): Add assert that we don't
22408 get overflowed constants as invariant part.
22409 (compare_values_warnv): Add comment before the TREE_NO_WARNING
22410 checks. Use wi::cmp instead of recursing for integer constants.
22411 (compare_values): Just ignore whether we assumed undefined
22412 overflow instead of failing the compare.
22413 (extract_range_for_var_from_comparison_expr): Add comment before the
22414 TREE_NO_WARNING sets.
22415 (test_for_singularity): Likewise.
22416 (extract_range_from_comparison): Do not disable optimization
22417 when we assumed undefined overflow.
22418 (extract_range_basic): Remove init of unused var.
22420 2017-05-09 Richard Biener <rguenther@suse.de>
22422 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
22423 (extract_range_from_multiplicative_op_1): Adjust.
22424 (extract_range_from_binary_expr_1): Use int_const_binop.
22426 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
22429 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
22430 rs6000_store_data_bypass_p in seven define_bypass directives and
22431 in several comments.
22432 * config/rs6000/rs6000-protos.h: Add prototype for
22433 rs6000_store_data_bypass_p function.
22434 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
22435 function implements slightly different (rs6000-specific) semantics
22436 than store_data_bypass_p, returning false rather than aborting
22437 with assertion error when arguments do not satisfy the
22438 requirements of store data bypass.
22439 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
22440 rs6000_store_data_bypass_p.
22442 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
22444 * config/xtensa/xtensa-protos.h
22445 (xtensa_initial_elimination_offset): New declaration.
22446 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
22447 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
22448 macro definition, add case for FRAME_POINTER_REGNUM when
22449 FRAME_GROWS_DOWNWARD.
22450 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
22451 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
22452 xtensa_initial_elimination_offset.
22454 2017-05-08 Nathan Sidwell <nathan@acm.org>
22456 * doc/invoke.texi: Alphabetize -fdump options.
22458 2017-05-08 Martin Sebor <msebor@redhat.com>
22460 PR translation/80280
22461 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
22463 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
22465 * target.def (compute_frame_layout): New optional target hook.
22466 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
22467 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
22468 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
22470 * reload1.c (verify_initial_elim_offsets): Likewise.
22471 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
22472 (use_simple_return_p): Call arm_compute_frame_layout if needed.
22473 (arm_get_frame_offsets): Split up into this ...
22474 (arm_compute_frame_layout): ... and this function.
22476 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
22478 * config/aarch64/constraints.md (Usa): New constraint.
22479 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
22481 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
22483 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
22484 with_multilib_list after it has been checked.
22486 2017-05-08 Richard Biener <rguenther@suse.de>
22488 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
22489 (bitmap_set_subtract_values): Likewise.
22491 2017-05-08 Richard Biener <rguenther@suse.de>
22493 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
22494 (gimple_assign_nonzero): ... this and remove strict_overflow_p
22496 (gimple_stmt_nonzero_warnv_p): Rename to ...
22497 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
22499 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
22500 (extract_range_basic): Adjust, do not disable propagation on
22501 strict overflow sensitive simplification.
22502 (vrp_visit_cond_stmt): Likewise.
22504 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
22506 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
22507 body size unconditionally.
22509 2017-05-07 Jeff Law <law@redhat.com>
22512 2017-05-06 Jeff Law <law@redhat.com>
22513 PR tree-optimization/78496
22514 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
22517 PR tree-optimization/78496
22518 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
22519 (simplify_stmt_using_ranges): Call it.
22520 (vrp_dom_walker::before_dom_children): Extract equivalences
22521 from an ASSERT_EXPR with an equality comparison against a
22524 2017-05-06 Jeff Law <law@redhat.com>
22526 PR tree-optimization/78496
22527 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
22530 PR tree-optimization/78496
22531 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
22532 (simplify_stmt_using_ranges): Call it.
22533 (vrp_dom_walker::before_dom_children): Extract equivalences
22534 from an ASSERT_EXPR with an equality comparison against a
22537 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
22539 * lra-constraints.c (lra_copy_reg_equiv): New function.
22540 (split_reg): Use it to copy equivalence information from the
22541 original register to the spill register.
22543 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
22545 PR rtl-optimization/75964
22546 * simplify-rtx.c (simplify_const_relational_operation): Remove
22547 invalid handling of comparisons of integer ABS.
22549 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
22551 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
22552 initialize to zero.
22553 (init_regs): Remove declaration.
22554 (function_arg_advance_32): Initialize error_p as boolean variable.
22556 2017-05-05 Nathan Sidwell <nathan@acm.org>
22558 * store-motion.c (remove_reachable_equiv_notes): Reformat long
22559 lines. Use for (;;).
22561 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22563 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
22564 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
22565 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
22566 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
22567 VF=2 that require versioning.
22569 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22571 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
22574 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22576 * diagnostic.h (diagnostic_override_option_index): Convert from
22577 macro to inline function.
22579 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22581 * diagnostic.c (last_module_changed_p): New function.
22582 (set_last_module): New function.
22583 (diagnostic_report_current_module): Convert macro usage to
22584 the above functions.
22585 * diagnostic.h (diagnostic_context::last_module): Strengthen
22586 from const line_map * to const line_map_ordinary *.
22587 (diagnostic_last_module_changed): Delete macro.
22588 (diagnostic_set_last_module): Delete macro.
22590 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22592 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
22593 with diagnostic_report_diagnostic.
22594 (diagnostic_n_impl_richloc): Likewise.
22595 * diagnostic.h (report_diagnostic): Delete macro.
22596 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
22597 with diagnostic_report_diagnostic.
22598 * substring-locations.c (format_warning_va): Likewise.
22600 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22602 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
22603 save/restor of format_spec. Move option-printing code to...
22604 (print_option_information): ...this new function, and
22605 reimplement by simply printing to the pretty_printer,
22606 rather than appending to the format string.
22608 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22610 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
22611 handling logic into...
22612 (update_effective_level_from_pragmas): ...this new function.
22614 2017-05-04 Andrew Waterman <andrew@sifive.com>
22616 * config/riscv/riscv.opt (mstrict-align): New option.
22617 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
22618 (SLOW_UNALIGNED_ACCESS): Define.
22619 (riscv_slow_unaligned_access): Declare.
22620 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
22622 (riscv_slow_unaligned_access): New variable.
22623 (rocket_tune_info): Set slow_unaligned_access to true.
22624 (optimize_size_tune_info): Set slow_unaligned_access to false.
22625 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
22626 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
22627 (riscv_option_override): Set riscv_slow_unaligned_access.
22628 * doc/invoke.texi: Add -mstrict-align to RISC-V.
22630 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
22632 * config/riscv/riscv.md: Unify indentation.
22634 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22639 * config/rs6000/rs6000.md (u code attribute): Add FIX and
22641 (extendsi<mode>2): Add support for doing sign extension via
22642 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
22643 don't have ISA 3.0 instructions.
22644 (extendsi<mode>2 splitter): Likewise.
22645 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
22646 generate the normal insns since SImode can now go in vector
22647 registers. Disallow the special UNSPECs needed for previous
22648 machines to hide SImode being used. Add new insns
22649 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
22650 (fix_trunc<mode>si2_stfiwx): Likewise.
22651 (fix_trunc<mode>si2_internal): Likewise.
22652 (fixuns_trunc<mode>si2): Likewise.
22653 (fixuns_trunc<mode>si2_stfiwx): Likewise.
22654 (fctiw<u>z_<mode>_smallint): Likewise.
22655 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
22656 of floating point to 32-bit integer from doing a direct move to
22657 the GPR registers to do a store.
22658 (fctiwz_<mode>): Break long line.
22660 2017-05-05 Bin Cheng <bin.cheng@arm.com>
22662 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
22663 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
22664 (addr_list, addr_offset_valid_p): New.
22665 (split_address_groups): Check offset validity with above function.
22666 (gt-tree-ssa-loop-ivopts.h): Include header file.
22668 2017-05-05 Nathan Sidwell <nathan@acm.org>
22670 * config.gcc (arm*-*-*): Add missing 'fi'.
22672 2017-05-05 Steve Ellcey <sellcey@cavium.com>
22674 * doc/invoke.texi (-fopt-info): Explicitly say order of options
22675 included in -fopt-info does not matter.
22676 * doc/optinfo.texi (-fopt-info): Fix description of default
22677 behavour. Explicitly say order of options included in -fopt-info
22680 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22682 * config.gcc: Allow combinations of aprofile and rmprofile values for
22683 --with-multilib-list.
22684 * config/arm/t-multilib: New file.
22685 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
22686 variables. Remove setting of ISA and floating-point ABI in
22687 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
22688 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
22689 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
22690 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
22692 * config/arm/t-rmprofile: Likewise except for the matches changes.
22693 * doc/install.texi (--with-multilib-list): Document the combination of
22694 aprofile and rmprofile values and warn about pitfalls in doing that.
22696 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
22698 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
22699 (movdi_aarch64): Likewise.
22701 2017-05-05 Jakub Jelinek <jakub@redhat.com>
22703 PR tree-optimization/80632
22704 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
22706 (build_arrays): Initialize it for virtual phis.
22707 (fix_phi_nodes): Use it for virtual phis.
22709 PR tree-optimization/80558
22710 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
22711 [x, y] op z into [x op, y op z] for op & or | if conditions
22714 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
22715 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22718 * config/arm/arm.md (use_literal_pool): Remove.
22719 (64-bit immediate split): No longer takes cost into consideration
22720 if arm_disable_literal_pool is enabled.
22721 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
22722 used when arm_disable_literal_pool is enabled.
22723 (arm_max_const_double_inline_cost): Remove use of
22724 arm_disable_literal_pool.
22725 (push_minipool_fix): Add assert.
22726 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
22727 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
22728 (no_literal_pool_sf_immediate): New.
22730 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22732 PR tree-optimization/80613
22733 * tree-ssa-dce.c (propagate_necessity): Remove cases for
22734 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
22736 2017-05-05 Richard Biener <rguenther@suse.de>
22738 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
22740 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
22742 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
22743 of this flag from insn conditions due to removal from r247495.
22745 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
22747 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
22749 (arm_early_store_addr_dep_ptr): Likewise.
22750 * config/arm/aarch-common-protos.h
22751 (arm_early_load_addr_dep_ptr): Add prototype.
22752 (arm_early_store_addr_dep_ptr): Likewise.
22753 * config/arm/cortex-a53.md: Add new bypasses.
22755 2017-05-05 Jakub Jelinek <jakub@redhat.com>
22757 * tree.c (next_type_uid): Change type to unsigned.
22758 (type_hash_canon): Decrement back next_type_uid if
22759 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
22760 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
22763 2017-05-04 Martin Sebor <msebor@redhat.com>
22765 * builtins.c: Fix a trivial typo in a comment.
22767 PR middle-end/79234
22768 * builtins.c (check_sizes): Adjust to handle reading past the end.
22769 Avoid printing excessive upper bound of ranges. Use %E to print
22770 tree nodes instead of converting them to %wu.
22771 (expand_builtin_memchr): New function.
22772 (compute_dest_size): Rename...
22773 (compute_objsize): ...to this.
22774 (expand_builtin_memcpy): Adjust.
22775 (expand_builtin_mempcpy): Adjust.
22776 (expand_builtin_strcat): Adjust.
22777 (expand_builtin_strcpy): Adjust.
22778 (check_strncat_sizes): Adjust.
22779 (expand_builtin_strncat): Adjust.
22780 (expand_builtin_strncpy): Adjust and simplify.
22781 (expand_builtin_memset): Adjust.
22782 (expand_builtin_bzero): Adjust.
22783 (expand_builtin_memcmp): Adjust.
22784 (expand_builtin): Handle memcmp.
22785 (maybe_emit_chk_warning): Check strncat just once.
22787 2017-05-04 Martin Sebor <msebor@redhat.com>
22789 PR preprocessor/79214
22790 PR middle-end/79222
22791 PR middle-end/79223
22792 * builtins.c (check_sizes): Add inlining context and issue
22793 warnings even when -Wno-system-headers is set.
22794 (check_strncat_sizes): Same.
22795 (expand_builtin_strncat): Same.
22796 (expand_builtin_memmove): New function.
22797 (expand_builtin_stpncpy): Same.
22798 (expand_builtin): Handle memmove and stpncpy.
22800 2017-05-04 Bin Cheng <bin.cheng@arm.com>
22802 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
22803 which is not used any more.
22805 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22807 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
22809 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22811 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
22812 (cortexa53_tunings): Likewise.
22813 (cortexa57_tunings): Likewise.
22814 (cortexa72_tunings): Likewise.
22815 (cortexa73_tunings): Likewise.
22817 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22819 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
22820 Set loop alignment to 8.
22822 2017-05-04 Martin Sebor <msebor@redhat.com>
22824 PR translation/80280
22825 * builtins.c (expand_builtin_object_size): Add missing quoting to
22826 %D and like directives.
22827 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
22828 (hsa_type_for_tree_type): Same.
22829 (verify_function_arguments): Same.
22830 * symtab.c (symbol_table::change_decl_assembler_name): Same.
22831 * varasm.c (get_section): Same.
22834 2017-05-04 Martin Sebor <msebor@redhat.com>
22836 PR translation/80280
22837 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
22839 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22841 * config/aarch64/aarch64.c (generic_addrcost_table):
22842 Change HI/TI mode setting.
22844 2017-05-04 Martin Jambor <mjambor@suse.cz>
22846 PR tree-optimization/80622
22847 * tree-sra.c (comes_initialized_p): New function.
22848 (build_accesses_from_assign): Only set write lazily when
22849 comes_initialized_p is false.
22850 (analyze_access_subtree): Use comes_initialized_p.
22851 (propagate_subaccesses_across_link): Assert !comes_initialized_p
22852 instead of testing for PARM_DECL.
22854 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22856 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
22857 constraint on operand 0 to allow more general addressing modes.
22858 Adjust output template.
22859 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
22861 * config/aarch64/aarch64-protos.h
22862 (aarch64_address_valid_for_prefetch_p): Declare prototype.
22863 * config/aarch64/constraints.md (Dp): New address constraint.
22864 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
22867 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
22869 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
22870 update use of estimate_ipcp_clone_size_and_time.
22871 (estimate_local_effects): Update use of
22872 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
22873 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
22874 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
22875 Return nonspecialized time.
22877 2017-05-04 Richard Biener <rguenther@suse.de>
22879 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
22880 for the last VUSE which def dominates the PHI. Directly call
22882 (get_continuation_for_phi_1): Remove.
22884 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
22886 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
22887 to explain the use of truncating division. Cap the number of
22888 iterations to the maximum given by nb_iterations_upper_bound,
22891 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22893 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
22894 * configure: Regenerate.
22895 * config.in: Regenerate.
22896 * config/i386/driver-mingw32.c: new file.
22897 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
22898 * config.host: Link driver-mingw32.o on MinGW host.
22899 * doc/install.texi: Document new --enable-mingw-wildcard configure
22902 2017-05-04 Marek Polacek <polacek@redhat.com>
22904 PR tree-optimization/80612
22905 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
22907 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22908 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
22910 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
22911 (movt splitter): Likewise.
22912 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
22913 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
22914 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
22915 block for Thumb-1 with MOVT.
22916 (thumb2_legitimate_address_p): Move code block ...
22917 (can_avoid_literal_pool_for_label_p): ... into this new function.
22918 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
22920 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
22921 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
22922 "M-profile targets with the MOVT instruction".
22924 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22926 * config/arm/arm-builtins.c (arm_init_builtins): Rename
22927 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
22928 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
22930 2017-05-04 Martin Liska <mliska@suse.cz>
22932 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
22933 variable cond_code.
22935 2017-05-04 Richard Biener <rguenther@suse.de>
22937 * tree.c (array_at_struct_end_p): Handle arrays at struct
22938 end with flexarrays more conservatively. Refactor and treat
22939 arrays of arrays or aggregates more strict. Fix
22940 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
22941 * tree.c (array_at_struct_end_p): Adjust prototype.
22942 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
22943 * gimple-fold.c (get_range_strlen): Likewise.
22944 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
22946 2017-05-04 Richard Biener <rguenther@suse.de>
22948 PR tree-optimization/31130
22949 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
22951 (supports_overflow_infinity): Likewise.
22952 (is_negative_overflow_infinity): Likewise.
22953 (is_positive_overflow_infinity): Likewise.
22954 (is_overflow_infinity): Likewise.
22955 (stmt_overflow_infinity): Likewise.
22956 (overflow_infinity_range_p): Likewise.
22957 (usable_range_p): Remove as always returning true.
22958 (make_overflow_infinity): Remove.
22959 (negative_overflow_infinity): Likewise.
22960 (positive_overflow_infinity): Likewise.
22961 (avoid_overflow_infinity): Likewise.
22962 (set_value_range): Adjust accordingly.
22963 (set_value_range_to_nonnegative): Likewise, remove now unused
22964 overflow_infinity arg.
22965 (vrp_operand_equal_p): Adjust.
22966 (update_value_range): Likewise.
22967 (range_int_cst_singleton_p): Likewise.
22968 (operand_less_p): Likewise.
22969 (compare_values_warnv): Likewise.
22970 (extract_range_for_var_from_comparison_expr): Likewise.
22971 (vrp_int_const_binop): Likewise.
22972 (zero_nonzero_bits_from_vr): Likewise.
22973 (extract_range_from_multiplicative_op_1): Likewise.
22974 (extract_range_from_binary_expr_1): Likewise.
22975 (extract_range_from_unary_expr): Likewise.
22976 (extract_range_from_comparison): Likewise.
22977 (extract_range_basic): Likewise.
22978 (adjust_range_with_scev): Likewise.
22979 (compare_ranges): Likewise.
22980 (compare_range_with_value): Likewise.
22981 (dump_value_range): Likewise.
22982 (test_for_singularity): Likewise, remove strict_overflow_p parameter
22984 (simplify_cond_using_ranges): Adjust.
22986 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
22988 * brig-builtins.def: Added a builtin for class_f64.
22989 * builtin-types.def: Added a builtin type needed by class_f64.
22991 2017-05-03 Jason Merrill <jason@redhat.com>
22993 * timevar.def: Add TV_CONSTEXPR.
22995 2017-05-03 David Malcolm <dmalcolm@redhat.com>
22997 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
22999 2017-05-03 Martin Jambor <mjambor@suse.cz>
23001 * ipa-prop.c (ipa_update_after_lto_read): Removed.
23002 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
23003 * ipa-cp.c (ipcp_propagate_stage): Do not call
23004 ipa_update_after_lto_read.
23005 * ipa-inline.c (ipa_inline): Likewise.
23007 2017-05-03 Martin Jambor <mjambor@suse.cz>
23009 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
23010 tag. Added a default constructor and a destructor.
23011 (ipa_edge_args_sum_t): New class;
23012 (ipa_edge_args_sum): Declare.
23013 (ipa_edge_args_vector): Remove declaration.
23014 (IPA_EDGE_REF): Use ipa_edge_args_sum.
23015 (ipa_free_edge_args_substructures): Remove declaration.
23016 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
23017 (ipa_edge_args_info_available_for_edge_p): Likewise.
23018 * ipa-prop.c (ipa_edge_args_vector): Removed.
23019 (edge_removal_hook_holder): Likewise.
23020 (edge_duplication_hook_holder): Likewise.
23021 (ipa_edge_args_sum): New variable.
23022 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
23023 ipa_edge_args_vector.
23024 (ipa_free_edge_args_substructures): Likewise.
23025 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
23026 ipa_edge_args_vector.
23027 (ipa_edge_removal_hook): Turned into method
23028 ipa_edge_args_sum_t::remove.
23029 (ipa_edge_duplication_hook): Turned into method
23030 ipa_edge_args_sum_t::duplicate.
23031 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
23032 registering edge hooks.
23033 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
23034 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
23035 ipa_edge_args_sum instead of ipa_edge_args_vector.
23036 * ipa-profile.c (ipa_profile): Likewise.
23038 2017-05-03 Martin Jambor <mjambor@suse.cz>
23040 * symbol-summary.h (function_summary): New method exists.
23041 (function_summary::symtab_removal): Deallocate through release.
23042 (call_summary): New class.
23043 (gt_ggc_mx): New overload.
23044 (gt_pch_nx): Likewise.
23045 (gt_pch_nx): Likewise.
23047 2017-05-03 Jeff Law <law@redhat.com>
23049 PR tree-optimization/78496
23050 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
23051 from simplify_cond_using_ranges. Split off code to walk
23052 backwards through casts into ...
23053 (simplify_cond_using_ranges_2): New function.
23054 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
23055 (execute_vrp): After identifying jump threads, call
23056 simplify_cond_using_ranges_2.
23058 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
23061 * ipa-inline.h (inline_summary): Add ctor.
23062 (create_ggc): Do not use ggc_cleared_alloc.
23064 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
23065 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23067 * gcc.c (handle_braces): Support escaping in switch matching
23069 * doc/invoke.texi (Spec Files): Document it.
23070 Remove superfluous @code markup in items.
23072 2017-05-03 David Malcolm <dmalcolm@redhat.com>
23074 * diagnostic-show-locus.c (struct column_range): New struct.
23075 (get_affected_columns): New function.
23076 (get_printed_columns): New function.
23077 (struct correction): New struct.
23078 (correction::ensure_capacity): New function.
23079 (correction::ensure_terminated): New function.
23080 (struct line_corrections): New struct.
23081 (line_corrections::~line_corrections): New dtor.
23082 (line_corrections::add_hint): New function.
23083 (layout::print_trailing_fixits): Reimplement in terms of the new
23085 (selftest::test_overlapped_fixit_printing): New function.
23086 (selftest::diagnostic_show_locus_c_tests): Call it.
23088 2017-05-03 Nathan Sidwell <nathan@acm.org>
23090 Canonicalize canonical type hashing
23091 * tree.h (type_hash_canon_hash): Declare.
23092 * tree.c (type_hash_list, attribute_hash_list): Move into
23093 type_hash_canon_hash.
23094 (build_type_attribute_qual_variant): Break out hash code calc into
23095 type_hash_canon_hash.
23096 (type_hash_canon_hash): New. Generic type hash computation.
23097 (build_range_type_1, build_array_type_1, build_function_type,
23098 build_method_type_directly, build_offset_type, build_complex_type,
23099 make_vector_type): Call it.
23101 2017-05-03 Richard Biener <rguenther@suse.de>
23103 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
23104 When all DRs have unknown misaligned do not always peel
23105 when there is a store but apply the same costing model as if
23106 there were only loads.
23108 2017-05-03 Richard Biener <rguenther@suse.de>
23111 PR tree-optimization/80492
23112 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
23113 compare_base_decls returning dont-know properly.
23115 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
23117 * config/arm/iterators.md (CCSI): New mode iterator.
23118 (arch): New mode attribute.
23119 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
23120 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
23121 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
23122 code iterator for success result mode.
23123 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
23124 the corresponding new insn generators.
23126 2017-05-03 Bin Cheng <bin.cheng@arm.com>
23129 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23130 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23132 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
23134 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
23135 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
23136 (DDR_A): Wrap DDR argument in brackets.
23137 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
23138 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
23139 (DDR_REVERSED_P): Likewise.
23141 2017-05-03 Jakub Jelinek <jakub@redhat.com>
23143 PR tree-optimization/79472
23144 * tree-switch-conversion.c (struct switch_conv_info): Add
23145 contiguous_range and default_case_nonstandard fields.
23146 (collect_switch_conv_info): Compute contiguous_range and
23147 default_case_nonstandard fields, don't clear final_bb if
23148 contiguous_range and only the default case doesn't have the required
23150 (check_all_empty_except_final): Set default_case_nonstandard instead
23151 of failing if contiguous_range and the default case doesn't have empty
23153 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
23154 and only the default case doesn't have the required constants. Skip
23156 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
23157 if default_case_nonstandard.
23158 (build_constructors): Build constant 1 just once. Assert that default
23159 values aren't inserted in between cases if contiguous_range. Skip
23161 (build_arrays): Skip virtual phis.
23162 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
23163 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
23164 Handle virtual phis.
23165 (gen_inbound_check): Handle default_case_nonstandard case.
23166 (process_switch): Adjust check_final_bb caller. Call
23167 gather_default_values with the first non-default case instead of
23168 default case if default_case_nonstandard.
23170 2017-05-02 Nathan Sidwell <nathan@acm.org>
23172 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
23173 check. Fix formatting.
23175 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
23177 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
23178 errors when comparing specialized and unspecialized times.
23180 2017-05-02 David Malcolm <dmalcolm@redhat.com>
23182 * diagnostic-show-locus.c
23183 (layout::should_print_annotation_line_p): Make private.
23184 (layout::print_annotation_line): Make private.
23185 (layout::annotation_line_showed_range_p): Make private.
23186 (layout::show_ruler): Make private.
23187 (layout::print_source_line): Make private. Pass in line and
23188 line_width, rather than calling location_get_source_line. Drop
23190 (layout::print_leading_fixits): New method.
23191 (layout::print_any_fixits): Rename to...
23192 (layout::print_trailing_fixits): ...this, and make private.
23193 Don't print newline fixits.
23194 (diagnostic_show_locus): Move logic for printing one row into...
23195 (layout::print_line): ...this new function. Move the
23196 location_get_source_line call and error-handling from
23197 print_source_line to here. Call print_leading_fixits, and rename
23198 print_any_fixits to print_trailing_fixits.
23199 (selftest::test_fixit_insert_containing_newline): Update now that
23200 newlines are partially supported.
23201 (selftest::test_fixit_insert_containing_newline_2): New test.
23202 (selftest::test_fixit_replace_containing_newline): Update comments.
23203 (selftest::diagnostic_show_locus_c_tests): Call the new test.
23204 * edit-context.c (class added_line): New class.
23205 (class edited_line): Describe newline handling in comment.
23206 (edited_line::actually_edited_p): New method.
23207 (edited_line::print_content): Delete redundant decl.
23208 (edited_line::m_predecessors): New field.
23209 (edited_file::print_content): Call edited_line::print_content.
23210 (edited_file::print_diff): Update to support newlines.
23211 (edited_file::print_diff_hunk): Likewise.
23212 (edited_file::print_run_of_changed_lines): New function.
23213 (edited_file::print_diff_line): Convert to...
23214 (print_diff_line): ...this.
23215 (edited_file::get_effective_line_count): New function.
23216 (edited_line::edited_line): Initialize new field m_predecessors.
23217 (edited_line::~edited_line): Clean up m_predecessors.
23218 (edited_line::apply_fixit): Handle newlines.
23219 (edited_line::get_effective_line_count): New function.
23220 (edited_line::print_content): New function.
23221 (edited_line::print_diff_lines): New function.
23222 (selftest::test_applying_fixits_insert_containing_newline): New
23224 (selftest::test_applying_fixits_replace_containing_newline): New
23226 (selftest::insert_line): New function.
23227 (selftest::test_applying_fixits_multiple_lines): Add example of
23229 (selftest::edit_context_c_tests): Call the new tests.
23231 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23233 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
23234 parameter cand. Update dump information.
23235 (get_computation_cost): Update uses.
23237 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23239 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
23240 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
23241 (get_computation_at, rewrite_use_address): Update use of
23242 get_computation_aff.
23244 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23246 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
23247 (get_computation): Delete.
23248 (get_computation_cost): Implement like get_computation_cost_at.
23249 Use get_computation_at.
23250 (get_computation_cost_at): Delete.
23251 (rewrite_use_nonlinear_expr): Use get_computation_at.
23252 (rewrite_use_compare, remove_unused_ivs): Ditto.
23254 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23256 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
23258 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23260 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
23261 (ivopts_global_cost_for_size): Rename parameter and update uses.
23262 (iv_ca_recount_cost): Update uses.
23263 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
23264 candidates seperately in n_invs and n_cands.
23265 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
23267 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23269 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
23270 (find_inv_vars_cb): New.
23271 (find_depends): Renamed to ...
23272 (find_inv_vars): ... this.
23273 (add_candidate_1, force_var_cost): Call find_inv_vars.
23274 (split_address_cost, determine_group_iv_cost_cond): Ditto.
23276 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23278 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
23279 inv_vars. Add inv_exprs.
23280 (struct iv_cand): Rename depends_on to inv_vars.
23281 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
23282 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
23283 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
23284 (dump_cand): Dump inv_vars.
23285 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
23286 (record_invariant, find_depends, add_candidate_1): Ditto.
23287 (set_group_iv_cost, force_var_cost): Ditto.
23288 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
23289 (get_computation_cost_at, get_computation_cost): Ditto.
23290 (determine_group_iv_cost_generic): Ditto.
23291 (determine_group_iv_cost_address): Ditto.
23292 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
23293 (determine_group_iv_costs): Ditto.
23294 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
23295 (iv_ca_set_remove_invariants): Renamed to ...
23296 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
23297 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
23298 (iv_ca_set_add_invariants): Renamed to ...
23299 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
23300 (iv_ca_set_cp): Use iv_ca_set_add_invs.
23301 (iv_ca_has_deps): Support inv_vars and inv_exprs.
23302 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
23303 (create_new_ivs): Remove useless dump.
23305 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23307 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
23309 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
23310 (iv_ca_set_no_cp, create_new_iv): Ditto.
23312 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23314 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23316 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
23318 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
23319 function tree_check2.
23321 2017-05-02 Martin Liska <mliska@suse.cz>
23323 * doc/gcov.texi: Add missing preposition.
23324 * gcov.c (function_info::function_info): Properly fill up
23325 all member variables.
23327 2017-05-02 Tamar Christina <tamar.christina@arm.com>
23329 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
23331 2017-05-02 Tamar Christina <tamar.christina@arm.com>
23333 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
23335 2017-05-02 Martin Liska <mliska@suse.cz>
23338 * lto-streamer-in.c (lto_read_tree_1): Remove
23339 LTO_STREAMER_DEBUG.
23340 * lto-streamer.c (struct tree_hash_entry): Likewise.
23341 (struct tree_entry_hasher): Likewise.
23342 (tree_entry_hasher::hash): Likewise.
23343 (tree_entry_hasher::equal): Likewise.
23344 (lto_streamer_init): Likewise.
23345 (lto_orig_address_map): Likewise.
23346 (lto_orig_address_get): Likewise.
23347 (lto_orig_address_remove): Likewise.
23348 * lto-streamer.h: Likewise.
23349 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
23350 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
23352 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
23354 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
23355 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
23356 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
23357 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
23358 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
23359 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
23360 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
23361 (mm_maskz_sub_ss): New intrinsics.
23362 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23363 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
23364 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
23365 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
23366 (__builtin_ia32_subss_mask_round): New builtins.
23367 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23368 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23369 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
23371 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
23372 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
23374 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
23377 2017-05-02 Martin Jambor <mjambor@suse.cz>
23379 PR tree-optimization/78687
23380 * tree-sra.c (access): New field parent.
23381 (process_subtree_disqualification): New function.
23382 (disqualify_candidate): Call it.
23383 (build_accesses_from_assign): Reset write flag if creating an
23385 (build_access_subtree): Fill in parent field and also prpagate
23386 down grp_write flag.
23387 (create_artificial_child_access): New parameter set_grp_write, set
23388 grp_write to its value.
23389 (propagate_subaccesses_across_link): Also propagate grp_write flag
23391 (propagate_all_subaccesses): Push the closest parent back to work
23392 queue if add_access_to_work_queue returned true.
23394 2017-05-02 Richard Biener <rguenther@suse.de>
23396 * common.opt (fstrict-overflow): Alias negative to fwrapv.
23397 * doc/invoke.texi (fstrict-overflow): Remove all traces of
23398 -fstrict-overflow documentation.
23399 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
23400 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
23401 flag_strict_overflow.
23402 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
23403 * lto-opts.c (lto_write_options): Do not stream it.
23404 * lto-wrapper.c (merge_and_complain): Do not handle it.
23405 * opts.c (default_options_table): Do not set -fstrict-overflow.
23406 (finish_options): Likewise do not clear it when sanitizing.
23407 * simplify-rtx.c (simplify_const_relational_operation): Do not
23408 test flag_strict_overflow.
23410 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
23412 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
23413 using enabled attribute.
23414 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
23415 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
23416 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
23417 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
23418 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
23419 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
23420 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
23421 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
23422 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
23423 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
23425 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
23427 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
23429 2017-05-02 Richard Biener <rguenther@suse.de>
23431 PR tree-optimization/80591
23433 2017-04-10 Richard Biener <rguenther@suse.de>
23435 * tree-ssa-structalias.c (find_func_aliases): Properly handle
23438 2017-05-02 Richard Biener <rguenther@suse.de>
23440 PR tree-optimization/80549
23441 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
23442 (cleanup_tree_cfg_noloop): Create forwarders to known loop
23443 headers if they do not have a preheader.
23445 2017-05-02 Martin Liska <mliska@suse.cz>
23448 * common.opt: Fix typo.
23449 * doc/invoke.texi: Likewise.
23451 2017-05-01 Jan Beulich <jbeulich@suse.com>
23453 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
23454 swapping, add (x,x,m,x,n) alternative.
23456 2017-05-01 Nathan Sidwell <nathan@acm.org>
23458 * calls.c (combine_pending_stack_adjustment_and_call): Remove
23459 unnecessary unadjusted_alignment check.
23461 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
23464 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
23466 * gimplify.c (gimplify_cilk_detach): New function.
23467 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
23468 * tree-core.h: Document EXPR_CILK_SPAWN.
23469 * tree.h (EXPR_CILK_SPAWN): Define.
23471 2017-05-01 David Malcolm <dmalcolm@redhat.com>
23473 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
23474 to use new fixit_hint representation, using the "replace" logic.
23475 (get_line_span_for_fixit_hint): Likewise.
23476 (layout::print_any_fixits): Likewise.
23477 (selftest::test_one_liner_many_fixits): Rename to...
23478 (selftest::test_one_liner_many_fixits_1): ...this, and update
23479 comment and expected output to reflect that the multiple fix-it
23480 hints are now consolidated into one insertion.
23481 (selftest::test_one_liner_many_fixits_2): New test.
23482 (selftest::test_diagnostic_show_locus_one_liner): Update for
23484 (selftest::test_fixit_consolidation): Update for fix-it API
23486 * diagnostic.c (print_parseable_fixits): Likewise.
23487 * edit-context.c (edited_line::m_line_events): Convert from
23488 auto_vec <line_event *> to auto_vec <line_event>.
23489 (class line_event): Convert from abstract base class to a concrete
23490 class, taking over the role of replace_event.
23491 (class insert_event): Delete.
23492 (class replace_event): Rename to class line_event. Convert to
23494 (edit_context::add_fixits): Reimplement.
23495 (edit_context::apply_insert): Delete.
23496 (edit_context::apply_replace): Rename to...
23497 (edit_context::apply_fixit): ...this. Convert to half-open range.
23498 (edited_file::apply_insert): Delete.
23499 (edited_file::apply_replace): Rename to...
23500 (edited_file::apply_fixit): ...this.
23501 (edited_line::~edited_line): Drop deletion of events.
23502 (edited_line::apply_insert): Delete.
23503 (edited_line::apply_replace): Rename to...
23504 (edited_line::apply_fixit): ...this. Convert to half-open range.
23505 Update for change to type of m_line_events.
23506 * edit-context.h (edit_context::apply_insert): Delete.
23507 (edit_context::apply_replace): Rename to...
23508 (edit_context::apply_fixit): ...this.
23510 2017-05-01 Martin Sebor <msebor@redhat.com>
23512 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
23515 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
23518 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
23519 __get_cpuid_max returns 0.
23520 (__get_cpuid_count): Ditto.
23522 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
23524 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
23525 replacement expression is another instance of one of its arguments.
23527 2017-05-01 Jakub Jelinek <jakub@redhat.com>
23530 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
23531 check for stack push/pop autoinc.
23532 * config/i386/i386.c (ix86_agi_dependent): Return false
23533 if the only reason why modified_in_p returned true is that
23534 addr is SP based and set_insn is a push or pop.
23536 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
23538 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
23541 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
23544 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
23545 (account_size_time): Use two predicates - exec_pred and
23547 (evaluate_conditions_for_known_args): Compute both clause and
23549 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
23550 (inline_summary_t::duplicate): Update.
23551 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
23553 (compute_inline_parameters): Likewise.
23554 (estimate_edge_size_and_time): Update caluclation of time.
23555 (estimate_node_size_and_time): Compute both time and nonspecialized
23557 (estimate_ipcp_clone_size_and_time): Update.
23558 (inline_merge_summary): Update.
23559 (do_estimate_edge_time): Update.
23560 (do_estimate_edge_size): Update.
23561 (do_estimate_edge_hints): Update.
23562 (inline_read_section, inline_write_summary): Stream both new predicates.
23563 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
23565 (compute_inlined_call_time): Cleanup.
23566 (big_speedup_p): Update.
23567 (edge_badness): Update.
23568 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
23569 (size_time_entry): Replace predicate by exec_predicate and
23570 nonconst_predicate.
23571 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
23572 (estimate_edge_time): Return also nonspec_time.
23573 (reset_edge_growth_cache): Update.
23575 2017-04-29 Jakub Jelinek <jakub@redhat.com>
23577 PR rtl-optimization/80491
23578 * ifcvt.c (noce_process_if_block): When looking for x setter
23579 with missing else_bb, don't check only the insn right before
23580 cond_earliest, but look for the last insn that x is modified in
23581 within the same bb.
23583 PR rtl-optimization/80491
23584 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
23586 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
23588 PR tree-optimization/80487
23589 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
23591 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23593 PR tree-optimization/79697
23594 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
23595 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
23596 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
23598 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
23599 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
23601 2017-04-28 Martin Sebor <msebor@redhat.com>
23603 PR tree-optimization/80523
23604 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
23605 (init_target_to_host_charmap, target_to_host, target_strtol10): New
23607 (maybe_warn, format_directive, parse_directive): Use new functions.
23608 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
23610 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
23612 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
23614 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
23616 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
23617 target_header_dir): Set correctly.
23618 * configure: Regenerated.
23619 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
23620 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
23621 instead of SYSTEM_HEADER_DIR.
23623 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
23625 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
23626 (estimate_local_effects): Likewise.
23627 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
23628 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
23629 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
23630 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
23631 do_estimate_edge_time, estimate_edge_time): Likewise.
23632 * ipa-inline-analysis.c (estimate_node_size_and_time,
23633 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
23634 (estimate_time_after_inlining): Remove.
23636 2017-04-28 Martin Liska <mliska@suse.cz>
23638 * doc/gcov.texi: Enhance documentation of gcov.
23640 2017-04-28 Martin Liska <mliska@suse.cz>
23642 * doc/gcov.texi: Sort options in alphabetic order.
23643 * doc/gcov-dump.texi: Likewise.
23644 * doc/gcov-tool.texi: Likewise.
23645 * gcov.c (print_usage): Likewise.
23646 * gcov-dump.c (print_usage): Likewise.
23647 * gcov-tool.c (print_merge_usage_message): Likewise.
23648 (print_rewrite_usage_message): Likewise.
23649 (print_overlap_usage_message): Likewise.
23651 2017-04-28 Martin Liska <mliska@suse.cz>
23653 PR gcov-profile/53915
23654 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
23656 2017-04-28 Martin Liska <mliska@suse.cz>
23658 PR gcov-profile/79891
23659 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
23660 is marked by compiler as living on a line.
23661 (get_cycles_count): Remove usage of the union.
23662 (output_intermediate_file): Likewise.
23663 (find_source): Fix GNU coding style.
23664 (accumulate_line_counts): Remove old non-all block mode.
23665 (output_lines): Remove usage of the union.
23666 * profile.c (output_location): Include all BBs, even if
23667 belonging to a same line (and file) as a previous BB.
23669 2017-04-28 Martin Liska <mliska@suse.cz>
23671 * gcov.c (process_args): Handle new argument 'w'.
23672 (read_graph_file): Assign ID to BBs.
23673 (output_branch_count): Display BB # if verbose flag is set.
23674 (output_lines): Likewise for arcs.
23675 (print_usage): Add '--verbose' option help.
23676 * doc/gcov.texi: Document --verbose (-w) option.
23678 2017-04-28 Martin Liska <mliska@suse.cz>
23680 * gcov.c (struct block_location_info): New struct.
23681 (process_file): Fill up the new structure.
23682 (read_graph_file): Replace usage of encoding by the newly added
23684 (add_line_counts): Likewise.
23685 (accumulate_line_counts): Remove usage of the union.
23686 (function_info::function_info): New function.
23687 (function_info::~function_info): Likewise.
23688 (process_file): Call delete instead of release_function.
23689 (release_function): Release the function.
23690 (release_structures): Call delete instead of release_function.
23691 (solve_flow_graph): Replace usage of num_blocks.
23692 (find_exception_blocks): Likewise.
23693 (output_lines): Fix GNU coding style.
23695 2017-04-28 Martin Liska <mliska@suse.cz>
23698 * coverage.c (coverage_remove_note_file): New function.
23699 * coverage.h: Declare the function.
23700 * toplev.c (finalize): Clean if an error has been seen.
23702 2017-04-28 Martin Liska <mliska@suse.cz>
23704 PR gcov-profile/80031
23705 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
23706 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
23707 * gcov.c (read_graph_file): Read just number of blocks.
23708 * profile.c (branch_prob): Do not stream 0 flags per a basic
23711 2017-04-28 Martin Liska <mliska@suse.cz>
23713 * gcov-dump.c (tag_*): Add new argument to declarations.
23714 (dump_gcov_file): Likewise.
23715 (tag_blocks): Add and use new argument depth.
23716 (tag_arcs): Likewise.
23717 (tag_lines): Likewise.
23718 (tag_counters): Likewise.
23719 (tag_summary): Likewise.
23720 (dump_working_sets): Use depth to do a proper indentation.
23722 2017-04-28 Jakub Jelinek <jakub@redhat.com>
23725 * cgraph.h (symtab_node::debug_symtab): No longer inline.
23726 * symtab.c (symtab_node::debug_symtab): Move definition here.
23728 2017-04-28 Richard Biener <rguenther@suse.de>
23730 * lto-streamer.h (LTO_major_version): Bump to 7.
23732 2017-04-28 Richard Biener <rguenther@suse.de>
23734 * tree-vrp.c (assert_info): New struct.
23735 (add_assert_info): New helper.
23736 (register_edge_assert_for_2): Refactor to add asserts to a vector
23738 (register_edge_assert_for_1): Likewise.
23739 (register_edge_assert_for): Likewise.
23740 (finish_register_edge_assert_for): New helper actually registering
23741 asserts where live on edge.
23742 (find_conditional_asserts): Adjust.
23743 (find_switch_asserts): Likewise.
23744 (evrp_dom_walker::try_find_new_range): Generalize.
23745 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
23747 2017-04-27 Marek Polacek <polacek@redhat.com>
23750 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
23751 arg10 and arg11 to itype.
23753 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
23755 * doc/extend.texi (Object Size Checking): Improve grammar.
23757 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
23760 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
23761 that the logic for permitting reciprocal estimates matches that
23764 2017-04-27 Jakub Jelinek <jakub@redhat.com>
23767 * tree.c (type_cache_hasher::equal): Only compare
23768 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
23769 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
23770 non-aggregate element types.
23771 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
23772 about the flag on ARRAY_TYPEs in the comment, formatting fix.
23774 2017-04-27 Richard Biener <rguenther@suse.de>
23776 PR middle-end/80533
23777 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
23778 stripping ARRAY_REFs from MEM_EXPR make sure we're not
23779 keeping a reference to a trailing array.
23781 2017-04-27 Richard Biener <rguenther@suse.de>
23783 PR middle-end/80539
23784 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
23785 being in loop-closed SSA form conservatively.
23786 (chrec_fold_multiply_poly_poly): Likewise.
23788 2017-04-27 Tamar Christina <tamar.christina@arm.com>
23790 PR middle-end/79665
23791 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
23792 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
23794 2017-04-27 Jakub Jelinek <jakub@redhat.com>
23797 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
23798 (aarch64_function_arg_alignment): Return unsigned int again, but still
23799 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
23800 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
23801 Don't emit -Wpsabi note.
23802 (aarch64_function_arg_boundary): Likewise.
23803 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
23806 2017-04-26 Nathan Sidwell <nathan@acm.org>
23808 * tree.h (crc32_unsigned_n): Declare.
23809 (crc32_unsigned, crc32_unsigned): Make inline.
23810 * tree.c (crc32_unsigned_bits): Replace with ...
23811 (crc32_unsigned_n): ... this.
23812 (crc32_unsigned, crc32_byte): Remove.
23813 (crc32_string): Remove unnecessary braces.
23815 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
23817 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
23818 * ipa-inline-analysis.c (MAX_TIME): Remove.
23819 (account_size_time): Use sreal for time.
23820 (dump_inline_summary): Update.
23821 (estimate_function_body_sizes): Update.
23822 (estimate_edge_size_and_time): Update.
23823 (estimate_calls_size_and_time): Update.
23824 (estimate_node_size_and_time): Update.
23825 (inline_merge_summary): Update.
23826 (inline_update_overall_summary): Update.
23827 (estimate_time_after_inlining): Update.
23828 (inline_read_section): Update.
23829 (inline_write_summary): Update.
23830 * ipa-inline.c (compute_uninlined_call_time): Update.
23831 (compute_inlined_call_time): Update.
23832 (recursive_inlining): Update.
23833 (inline_small_functions): Update.
23834 (dump_overall_stats): Update.
23835 * ipa-inline.h: Include sreal.h.
23836 (size_time_entry): Turn time to sreal.
23837 (inline_summary): Turn self_time nad time to sreal.
23839 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
23841 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
23843 (sreal::stream_out, sreal::stream_in): New.
23844 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
23846 2017-04-25 Jakub Jelinek <jakub@redhat.com>
23848 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
23851 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
23854 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
23856 (dimode_scalar_chain::compute_convert_gain): Ditto.
23857 (dimode_scalar_chain::make_vector_copies): Ditto.
23858 (dimode_scalar_chain::convert_reg): Ditto.
23859 (dimode_scalar_chain::convert_insn): Ditto.
23860 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
23861 (VI248_AVX512BW_1): New mode iterator.
23862 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
23863 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
23866 2017-04-25 Martin Sebor <msebor@redhat.com>
23868 PR tree-optimization/80497
23869 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
23870 constants are representable in HOST_WIDE_INT.
23871 (parse_directive): Ditto.
23873 2017-04-25 Martin Sebor <msebor@redhat.com>
23876 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
23877 (new_zero_array): Adjust signature.
23878 (dom_info::dom_init): Used unsigned rather that size_t.
23879 (dom_info::dom_info): Same.
23881 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23882 Jakub Jelinek <jakub@redhat.com>
23885 * config/arm/arm.c: Include gimple.h.
23886 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
23887 returns negative, increment ncrn only if it returned positive.
23888 (arm_needs_doubleword_align): Return int instead of bool,
23889 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
23890 members, but if there is any such non-FIELD_DECL
23891 > PARM_BOUNDARY aligned decl, return -1 instead of false.
23892 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
23893 returns negative, increment nregs only if it returned positive.
23894 (arm_setup_incoming_varargs): Likewise.
23895 (arm_function_arg_boundary): Emit -Wpsabi note if
23896 arm_needs_doubleword_align returns negative, return
23897 DOUBLEWORD_ALIGNMENT only if it returned positive.
23899 2017-04-25 Marek Polacek <polacek@redhat.com>
23902 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
23903 first argument to type.
23905 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
23908 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
23909 type checks to test for compatibility instead of equality.
23911 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23912 Jakub Jelinek <jakub@redhat.com>
23915 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
23917 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
23918 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
23919 the alignment computation, but return their maximum in warn_alignment.
23920 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
23921 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
23923 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
23924 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
23927 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23929 * config/arc/simdext.md (dmpyh): Fix typo.
23931 2017-04-25 Richard Biener <rguenther@suse.de>
23933 PR tree-optimization/80492
23934 * alias.c (compare_base_decls): Handle registers with asm
23935 specification conservatively.
23936 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
23937 compare_base_decls returning dont-know properly.
23939 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23941 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
23942 (legitimate_offset_address_p): New function.
23943 (arc_legitimate_address_p): Use above function.
23945 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23947 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
23949 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23951 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
23952 ACCH registers whenever they are available.
23954 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23956 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
23957 double regs fix when not used.
23959 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23961 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
23963 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
23964 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
23966 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23968 * config/arc/arc.c (arc_output_addsi): Check for h-register class
23969 when emitting short ADD instructions.
23971 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23973 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
23975 (cmpsi_cc_c_insn): Likewise.
23976 (cbranchsi4_scratch): Compute proper instruction length using
23977 compact_hreg_operand.
23978 * config/arc/predicates.md (compact_hreg_operand): New predicate.
23980 2017-04-25 Richard Biener <rguenther@suse.de>
23982 PR middle-end/80509
23983 * passes.c (pass_manager::pass_manager): Initialize
23984 m_name_to_pass_map.
23986 2017-04-25 Richard Biener <rguenther@suse.de>
23988 PR tree-optimization/79201
23989 * tree-ssa-sink.c (statement_sink_location): Handle calls.
23991 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23994 * config/s390/vector.md: Split MEM->GPR vector moves for
23995 non-s_operand addresses.
23997 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24000 * config/s390/predicates.md (reload_const_wide_int_operand): New
24002 * config/s390/s390.md ("movti"): Remove d/P alternative.
24003 ("movti_bigconst"): New pattern definition.
24005 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
24008 * s390-protos.h (s390_expand_cs_hqi): Removed.
24009 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
24010 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
24011 modes as well as CCZ1mode and CCZmode.
24012 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
24013 signature of s390_emit_compare_and_swap.
24014 (s390_expand_cs_hqi): Likewise, make static.
24015 (s390_expand_cs_tdsi): Generate an explicit compare before trying
24016 compare-and-swap, in some cases.
24017 (s390_expand_cs): Wrapper function.
24018 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
24020 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
24021 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
24022 patterns for small and large integers. Forbid symref memory operands.
24023 Move expander to s390.c. Require cc register.
24024 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
24025 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
24026 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
24027 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
24028 symref memory operands. Remove CC mode and call s390_match_ccmode
24030 ("atomic_exchange<mode>"): Allow and implement all integer modes.
24032 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
24034 * config/s390/s390.md (define_peephole2): New peephole to help
24035 combining the load-and-test pattern with volatile memory.
24037 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
24039 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
24040 with CCZmode for TARGET_Z196.
24042 2017-04-25 Jakub Jelinek <jakub@redhat.com>
24044 PR rtl-optimization/80501
24045 * combine.c (make_compound_operation_int): Set subreg_code to SET
24046 even for AND with mask of the sign bit of mode.
24048 PR rtl-optimization/80500
24049 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
24050 sum's initial value.
24052 2017-04-25 Julian Brown <julian@codesourcery.com>
24053 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24055 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
24057 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
24059 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
24061 2017-04-25 Julian Brown <julian@codesourcery.com>
24062 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24064 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
24065 (thunderx2t99_sha): New Reservation.
24067 2017-04-25 Julian Brown <julian@codesourcery.com>
24068 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24070 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
24071 type for 1-element load.
24073 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
24075 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
24077 2017-04-24 Martin Jambor <mjambor@suse.cz>
24079 PR tree-optimization/80293
24080 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
24081 char arrays not totally scalarizable if it is false.
24082 (analyze_all_variable_accesses): Pass correct value in the new
24083 parameter. Add a statistics counter.
24085 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
24087 PR middle-end/79931
24088 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
24090 2017-04-24 Richard Biener <rguenther@suse.de>
24092 PR tree-optimization/80494
24093 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
24094 out for complex types.
24096 2017-04-24 Richard Biener <rguenther@suse.de>
24098 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
24099 * tree-ssa-sccvn.c (print_scc): Print SCC size.
24100 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
24101 (DFS): Adjust and never fail.
24102 (sccvn_dom_walker::fail): Remove.
24103 (sccvn_dom_walker::before_dom_children): Adjust.
24104 (run_scc_vn): Likewise and never fail.
24105 * tree-ssa-pre.c (pass_pre::execute): Adjust.
24106 (pass_fre::execute): Likewise.
24108 2017-04-24 Richard Biener <rguenther@suse.de>
24110 PR tree-optimization/79725
24111 * tree-ssa-sink.c (statement_sink_location): Return whether
24112 failure reason was zero uses. Move that check later.
24113 (sink_code_in_bb): Deal with zero uses by removing the stmt
24116 2017-04-24 Richard Biener <rguenther@suse.de>
24119 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
24120 pointer-based references.
24122 2017-04-24 Richard Biener <rguenther@suse.de>
24125 * pass_manager.h (pass_manager::operator new): Remove.
24126 (pass_manager::operator delete): Likewise.
24127 * passes.c (pass_manager::operator new): Remove.
24128 (pass_manager::operator delete): Likewise.
24129 (pass_manager::pass_manager): Zero individual pass members.
24131 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
24134 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
24135 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
24136 Check "XEXP (src, 1)" operand here.
24137 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
24138 Check "XEXP (src, 1)" operand here.
24139 (dimode_scalar_chain::make_vector_copies): Detect count register
24140 of a shift instruction. Zero extend count register from QImode
24141 to DImode to satisfy vector shift pattern count operand predicate.
24142 Substitute vector shift count operand with a DImode copy.
24143 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
24146 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
24148 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
24149 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
24150 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
24151 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
24152 (UNSPEC_NOREX_MEM): Remove definition.
24154 2017-04-21 Richard Biener <rguenther@suse.de>
24156 PR tree-optimization/79547
24157 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
24158 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
24159 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
24160 without any constraints.
24162 2017-04-21 Richard Biener <rguenther@suse.de>
24164 PR tree-optimization/78847
24165 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
24167 2017-04-21 Richard Biener <rguenther@suse.de>
24169 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
24170 (build_distinct_type_copy): Likewise.
24171 (build_variant_type_copy): Likewise.
24172 * tree.c (build_qualified_type): Pass down mem-stat info.
24173 (build_distinct_type_copy): Likewise.
24174 (build_variant_type_copy): Likewise.
24176 2017-04-21 Richard Biener <rguenther@suse.de>
24178 PR tree-optimization/80237
24179 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
24181 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
24182 for a simplified result.
24184 2016-04-21 Richard Biener <rguenther@suse.de>
24186 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
24187 sth as strict as a simple_iv but a chrec without symbols and an
24188 operand defined in the loop we are peeling (and not some subloop).
24189 (propagate_constants_for_unrolling): Propagate all constants.
24191 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
24194 * config/i386/i386.c (print_reg): Remove assert for disalowed
24195 regno values, call output_operand_lossage instead.
24197 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
24200 * config/i386/constraints.md (Yc): New register constraint.
24201 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
24202 Use Yc constraint for alternative 2 of operand 0. Remove
24203 preferred_for_speed attribute.
24205 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
24207 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
24208 lastprivate clauses in SIMT case.
24210 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
24212 * doc/invoke.texi (-Wextra-semi): Document new warning option.
24214 2017-04-20 Richard Biener <rguenther@suse.de>
24216 PR tree-optimization/57796
24217 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
24218 as N scalar stores.
24219 (vect_model_load_cost): Cost gathers as N scalar loads.
24221 2017-04-20 Richard Biener <rguenther@suse.de>
24223 * ggc-page.c (ggc_allocated_p): Rename to ...
24224 (safe_lookup_page_table_entry): ... this and return the lookup
24226 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
24228 2017-04-20 Richard Biener <rguenther@suse.de>
24230 PR tree-optimization/80453
24231 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
24232 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
24233 from the conditions.
24234 (vn_phi_eq): Pass them down.
24235 (vn_phi_lookup): Record them.
24236 (vn_phi_insert): Likewise.
24238 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
24240 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
24241 uninitialized variable warning to avoid buffer overrun.
24243 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
24246 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
24247 is suppressed for '{ 0 }' in C.
24249 2017-04-20 Jakub Jelinek <jakub@redhat.com>
24251 * BASE-VER: Set to 8.0.0.
24253 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
24255 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
24256 priority .init_array and .fini_array section with SECTION_NOTYPE
24259 2017-04-20 Jakub Jelinek <jakub@redhat.com>
24261 PR middle-end/80423
24262 * tree.h (build_array_type): Add typeless_storage default argument.
24263 * tree.c (type_cache_hasher::equal): Also compare
24264 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
24265 (build_array_type): Add typeless_storage argument, set
24266 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
24268 (build_nonshared_array_type): Adjust build_array_type_1 caller.
24269 (build_array_type): Likewise. Add typeless_storage argument.
24271 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
24272 Jakub Jelinek <jakub@redhat.com>
24274 PR tree-optimization/80426
24275 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
24276 operation on symbolic operands, also compute the overflow for the
24277 invariant part when the operation degenerates into a negation.
24279 2017-04-19 Jakub Jelinek <jakub@redhat.com>
24282 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
24283 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
24286 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
24288 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
24291 * config/avr/avr.c (tree.h): Include it.
24292 (cgraph.h): Include it.
24293 (avr_encode_section_info): Don't warn for uninitialized progmem
24294 variable if it's just an alias.
24296 2017-04-19 Richard Biener <rguenther@suse.de>
24299 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
24300 when needed by AutoPGO.
24302 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
24305 * doc/lto.texi: Remove an extra 'that'.
24307 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
24309 PR rtl-optimization/80429
24310 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
24311 are only used in debug insns.
24313 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
24314 Vladimir Makarov <vmakarov@redhat.com>
24316 * config/sparc/predicates.md (input_operand): Add comment. Return
24317 true for any memory operand when LRA is in progress.
24318 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
24320 2017-04-18 Jeff Law <law@redhat.com>
24323 * mips.md ({return,simple_return}_internal): Do not overwrite
24326 2017-04-18 Jakub Jelinek <jakub@redhat.com>
24328 PR tree-optimization/80443
24329 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
24330 instead of adding 1, subtract -1 and similarly instead of subtracting
24333 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
24335 PR rtl-optimization/80357
24336 * haifa-sched.c (tmp_bitmap): New variable.
24337 (model_recompute): Handle duplicate use records.
24338 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
24339 (free_global_sched_pressure_data): Free it.
24341 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
24344 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
24345 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
24346 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
24347 instead of SYSTEM_HEADER_DIR.
24349 2017-04-18 Jeff Law <law@redhat.com>
24351 PR middle-end/80422
24352 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
24353 predecessors after walking up the insn chain.
24355 2017-04-18 Jakub Jelinek <jakub@redhat.com>
24358 * dwarf2out.c (modified_type_die): Try harder not to emit internal
24359 sizetype type into debug info.
24361 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24364 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
24365 unneeded test for TARGET_UPPER_REGS_SF.
24366 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
24368 2017-04-18 Jakub Jelinek <jakub@redhat.com>
24371 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
24372 instead of gsi_after_labels.
24374 2017-04-18 Jeff Law <law@redhat.com>
24376 * regcprop.c (maybe_mode_change): Avoid creating copies of the
24380 2017-04-13 Jeff Law <law@redhat.com>
24381 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
24382 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
24384 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
24387 * config/avr/avr.c (intl.h): Include it.
24388 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
24390 2017-04-18 Martin Liska <mliska@suse.cz>
24392 PR gcov-profile/78783
24393 * gcov-tool.c (gcov_output_files): Validate that destination
24394 file is either removed by the tool or by a user.
24396 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
24397 Guy Benyei <guybe@mellanox.com>
24399 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
24400 block, and do not negate it, the stored id is already negative.
24402 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
24404 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
24406 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
24409 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
24410 masks of options that should be turned off if the VSX vector
24411 options are turned off.
24412 (OTHER_P8_VECTOR_MASKS): Likewise.
24413 (OTHER_VSX_VECTOR_MASKS): Likewise.
24414 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
24415 rs6000_disable_incompatible_switches to validate no type switches
24417 (rs6000_incompatible_switch): New function to disallow turning on
24418 other vector options if -mno-vsx, -mno-power8-vector, or
24419 -mno-power9-vector are specified.
24421 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
24423 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
24425 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
24427 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
24428 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
24429 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
24430 (ARG_POINTER_CFA_OFFSET): Likewise.
24432 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
24434 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
24435 conditions to take advantage of various optimizations.
24437 2017-04-13 Jeff Law <law@redhat.com>
24439 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
24440 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
24441 (zero_extendsidi2_dext): Likewise.
24443 2017-04-13 Jakub Jelinek <jakub@redhat.com>
24446 * fold-const.c (fold_ternary_loc): Revert
24447 use op0 instead of fold_convert_loc (loc, type, arg0) part of
24450 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
24452 PR rtl-optimization/80343
24453 * lra-remat.c (update_scratch_ops): Assign original hard reg to
24454 new scratch pseudo.
24456 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
24459 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
24460 to ubsan_encode_value.
24462 2017-04-13 Jeff Law <law@redhat.com>
24464 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
24465 appearing in DEBUG_INSNs.
24467 2017-04-13 Martin Liska <mliska@suse.cz>
24469 PR gcov-profile/80413
24470 * gcov-io.c (gcov_write_string): Copy to buffer just when
24471 allocated size is greater than zero.
24473 2017-04-13 Jakub Jelinek <jakub@redhat.com>
24476 * dwarf2out.c (decls_for_scope): Ignore declarations of
24477 current_function_decl in BLOCK_NONLOCALIZED_VARS.
24479 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
24482 * ipa-visibility.c (non_local_p): Fix typos.
24483 (localize_node): When localizing symbol in same comdat group,
24484 dissolve the group only when we know external symbols are going
24486 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
24488 2017-04-12 Jakub Jelinek <jakub@redhat.com>
24490 PR tree-optimization/79390
24491 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
24492 order does not result in usable sequence, retry with reversed operand
24498 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
24499 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
24500 op0 instead of fold_convert_loc (loc, type, arg0).
24502 2017-04-12 Jeff Law <law@redhat.com>
24504 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
24505 has a delay slot in the generated code.
24507 * config/cris/cris.md (cris_preferred_reload_class): Return
24508 GENNONACR_REGS rather than GENERAL_REGS.
24510 2017-04-12 Jakub Jelinek <jakub@redhat.com>
24513 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
24514 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
24515 signedness of the result type.
24517 2017-04-12 Richard Biener <rguenther@suse.de>
24518 Jeff Law <law@redhat.com>
24520 PR tree-optimization/80359
24521 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
24522 trim stores to TARGET_MEM_REFs.
24524 2017-04-12 Richard Biener <rguenther@suse.de>
24526 PR tree-optimization/79390
24527 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
24528 threading case even more.
24530 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
24533 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
24534 for quad_address_p for TImode, instead of just not indexed_address.
24536 2017-04-12 Richard Biener <rguenther@suse.de>
24537 Bernd Edlinger <bernd.edlinger@hotmail.de>
24539 PR middle-end/79671
24540 * alias.c (component_uses_parent_alias_set_from): Handle
24541 TYPE_TYPELESS_STORAGE.
24542 (get_alias_set): Likewise.
24543 * tree-core.h (tree_type_common): Add typeless_storage flag.
24544 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
24545 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
24546 for types containing members with TYPE_TYPELESS_STORAGE.
24547 (place_field): Likewise.
24548 (layout_type): Likewise for ARRAY_TYPE.
24549 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
24550 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
24551 TYPE_TYPELESS_STORAGE.
24552 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
24554 2017-04-12 Jakub Jelinek <jakub@redhat.com>
24557 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
24558 first argument to type.
24560 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24564 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
24565 CONST0_RTX (mode) rather than const0_rtx where appropriate.
24566 (rs6000_expand_binop_builtin): Likewise.
24567 (rs6000_expand_ternop_builtin): Likewise; also add missing
24568 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
24569 vshasigma built-ins.
24570 * doc/extend.texi: Document that vec_xxpermdi's third argument
24571 must be a constant.
24573 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
24575 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
24576 Use shift_const cost parameter when calculating gain of STV shifts.
24578 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
24580 PR rtl-optimization/70478
24581 * lra-constraints.c (process_alt_operands): Check memory for
24582 disfavoring memory insn operand.
24584 2017-04-11 Jakub Jelinek <jakub@redhat.com>
24586 PR middle-end/80100
24587 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
24588 left shift in unsigned HOST_WIDE_INT type.
24590 PR rtl-optimization/80385
24591 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
24592 (not (neg X)) into (plus X -1) for complex or non-integral modes.
24595 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
24596 if they have any depend clauses.
24598 2017-04-11 Martin Liska <mliska@suse.cz>
24601 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
24602 * ipa-split.c (split_function): Create a local comdat symbol
24603 if caller is in a comdat group.
24605 2017-04-11 Martin Liska <mliska@suse.cz>
24608 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
24611 2017-04-11 Martin Sebor <msebor@redhat.com>
24613 PR middle-end/80364
24614 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
24615 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
24617 (directive::set_width, directive::set_precision, format_character):
24619 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
24622 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
24625 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
24626 conflict, set target->arch_name instead of target->cpu_name.
24628 2017-04-11 Richard Biener <rguenther@suse.de>
24630 PR tree-optimization/80374
24631 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
24632 build_zero_cst, remove fold_convertible_p check again.
24634 2017-04-11 Martin Liska <mliska@suse.cz>
24637 * ubsan.c (instrument_object_size): Do not instrument register
24640 2017-04-11 Jakub Jelinek <jakub@redhat.com>
24643 * config/i386/i386-builtin-types.def
24644 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
24645 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
24646 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
24647 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
24648 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
24649 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
24650 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
24651 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
24652 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
24653 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
24654 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
24655 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
24656 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
24657 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
24658 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
24659 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
24660 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
24661 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
24662 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
24663 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
24664 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
24665 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
24666 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
24667 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
24668 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
24669 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
24670 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
24671 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
24672 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
24673 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
24674 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
24675 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
24676 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
24677 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
24678 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
24679 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
24680 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
24681 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
24682 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
24683 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
24684 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
24685 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
24686 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
24687 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
24688 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
24689 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
24691 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
24692 flag to second_arg_count, handle 4 argument function type _COUNT
24693 aliases, handle second_arg_count on second argument rather than last.
24695 2017-04-10 Jeff Law <law@redhat.com>
24697 PR tree-optimization/80374
24698 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
24699 record anything if we can not convert integer_zero_node to the
24702 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
24705 * config/rs6000/rs6000.c (rs6000_option_override_internal):
24706 Enhance special handling given to the TARGET_P9_MINMAX option in
24707 relation to certain other options.
24709 2017-04-10 Bin Cheng <bin.cheng@arm.com>
24711 PR tree-optimization/80153
24712 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
24713 remove POINTER_PLUS_EXPR's base part directly, rather than through
24716 2017-04-10 Richard Biener <rguenther@suse.de>
24717 Bin Cheng <bin.cheng@arm.com>
24719 PR tree-optimization/80153
24720 * tree-affine.c (aff_combination_to_tree): Get base pointer from
24721 the first element of pointer type aff_tree. Build result expr in
24723 (add_elt_to_tree): Convert to type unconditionally. Remove other
24724 fold_convert calls.
24725 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
24726 (rewrite_use_nonlinear_expr): Check invariant using iv information.
24728 2017-04-10 Richard Biener <rguenther@suse.de>
24730 * tree-ssa-structalias.c (find_func_aliases): Properly handle
24733 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
24735 PR rtl-optimization/70478
24736 * lra-constraints.c (curr_small_class_check): New.
24737 (update_and_check_small_class_inputs): New.
24738 (process_alt_operands): Update curr_small_class_check. Disfavor
24739 alternative insn memory operands. Check available regs for small
24742 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
24745 * config/mips/mips.opt (-mvirt): Update description.
24746 * doc/invoke.texi (-mvirt): Likewise.
24748 2017-04-10 Richard Biener <rguenther@suse.de>
24750 PR middle-end/80362
24751 * fold-const.c (fold_binary_loc): Look at unstripped ops when
24752 looking for NEGATE_EXPR in -A / -B to A / B folding.
24754 2017-04-10 Martin Liska <mliska@suse.cz>
24756 PR gcov-profile/80224
24757 * gcov.c (print_usage): Fix usage string.
24758 (get_gcov_intermediate_filename): Remove.
24759 (output_gcov_file): Use both for normal and intermediate format.
24760 (generate_results): Do not initialize special file for
24761 intermediate format.
24763 2017-04-10 Richard Biener <rguenther@suse.de>
24765 PR tree-optimization/80304
24766 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
24769 2017-04-10 Nathan Sidwell <nathan@acm.org>
24772 * config/rs6000/rs6000.c (rs6000_vector_type): New.
24773 (rs6000_init_builtins): Use it.
24775 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24777 * config/arm/arm.md (<mrc>): Add mode to SET source.
24778 (<mrrc>): Likewise.
24780 2017-04-10 Richard Biener <rguenther@suse.de>
24782 PR middle-end/80344
24783 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
24785 2017-04-10 Jakub Jelinek <jakub@redhat.com>
24788 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
24789 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
24790 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
24791 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
24792 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
24793 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
24794 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
24795 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
24796 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
24797 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
24798 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
24799 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
24800 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
24801 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
24802 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
24803 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
24804 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
24805 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
24806 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
24807 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
24808 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
24809 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
24810 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
24812 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
24814 PR rtl-optimization/70478
24815 * lra-constraints.c: Reverse the last patch.
24817 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
24819 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
24820 Add comment for WCHAR_T.
24822 2017-04-08 Martin Liska <mliska@suse.cz>
24825 2017-04-07 Martin Liska <mliska@suse.cz>
24828 * ipa-split.c (split_function): Add function part to a same comdat
24831 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24834 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
24836 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
24838 * rs6000/rs6000.c (vec_load_pendulum): Rename...
24839 (vec_pairing): ...to this.
24840 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
24841 (rs6000_sched_init): Adjust for name change.
24842 (struct rs6000_sched_context): Likewise.
24843 (rs6000_init_sched_context): Likewise.
24844 (rs6000_set_sched_context): Likewise.
24846 2017-04-07 Jakub Jelinek <jakub@redhat.com>
24852 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
24854 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
24855 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
24856 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
24858 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
24860 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
24862 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
24864 PR rtl-optimization/70703
24865 * ira-color.c (update_conflict_hard_regno_costs): Use
24866 int64_t instead of HOST_WIDE_INT.
24868 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
24870 PR rtl-optimization/70478
24871 * lra-constraints.c (process_alt_operands): Disfavor alternative
24872 insn memory operands.
24874 2017-04-07 Jeff Law <law@redhat.com>
24876 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
24877 CALL and NOTE_INSN_CALL_ARG_LOCATION.
24879 2017-04-07 Martin Liska <mliska@suse.cz>
24882 * config/aarch64/aarch64.c (aarch64_process_target_attr):
24883 Show error message instead of an ICE.
24885 2017-04-07 Martin Liska <mliska@suse.cz>
24888 * ipa-split.c (split_function): Add function part to a same comdat
24891 2017-04-07 Richard Biener <rguenther@suse.de>
24893 PR middle-end/80341
24894 * tree.c (get_unwidened): Also handle ! for_type case for
24896 * convert.c (do_narrow): Split out from ...
24897 (convert_to_integer_1): ... here. Do not pass final truncation
24898 type to get_unwidened for TRUNC_DIV_EXPR.
24900 2017-04-07 Richard Biener <rguenther@suse.de>
24902 * tree-affine.c (wide_int_ext_for_comb): Take type rather
24904 (aff_combination_const): Adjust.
24905 (aff_combination_scale): Likewise.
24906 (aff_combination_add_elt): Likewise.
24907 (aff_combination_add_cst): Likewise.
24908 (aff_combination_convert): Likewise.
24909 (add_elt_to_tree): Likewise. Remove unused argument.
24910 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
24912 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
24914 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
24916 * config/arm/arm.c (arm_default_short_enums): Use
24917 ARM_DEFAULT_SHORT_ENUMS.
24918 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
24920 2017-04-06 Jakub Jelinek <jakub@redhat.com>
24923 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
24924 members with redundant out-of-class redeclaration.
24926 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
24929 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
24930 * config/i386/i386.md (*zero_extendsidi2):
24931 Add (?*x,*x) and (?*v,*v) alternatives.
24933 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
24936 * config/i386/i386.c (ix86_expand_builtin)
24937 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
24938 mode from insn data. Convert operands to insn operand mode.
24939 Copy operands that don't satisfy insn predicate to a register.
24941 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
24943 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
24946 2017-04-06 Richard Biener <rguenther@suse.de>
24948 PR tree-optimization/80334
24949 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
24950 preserve alignment of accesses.
24952 2017-04-06 Richard Biener <rguenther@suse.de>
24954 PR tree-optimization/80262
24955 * tree-sra.c (build_ref_for_offset): Preserve address-space
24957 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
24958 Drop useless address-space information on MEM_REF offsets.
24960 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
24962 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
24964 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
24966 PR rtl-optimization/70703
24967 * ira-color.c (update_conflict_hard_regno_costs): Use
24968 HOST_WIDE_INT instead of long.
24970 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
24973 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
24974 not defined for x86_64 target. Add -mmmx target option when __SSE2__
24976 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
24977 for x86_64 target. Handle -m3dnowa option.
24979 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
24981 PR rtl-optimization/70703
24982 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
24983 (update_conflict_hard_regno_costs): Use long instead of unsigned
24984 arithmetic for cost calculation.
24986 2017-04-05 Jakub Jelinek <jakub@redhat.com>
24987 Bernd Edlinger <bernd.edlinger@hotmail.de>
24990 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
24993 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
24996 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
24997 ptr_mode with Pmode throughout.
24998 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
24999 into probe_stack_range and use DImode.
25001 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
25004 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
25005 call_eh_return is true.
25007 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25009 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
25010 Initialize last_match_fntype_index.
25012 2017-04-05 Jakub Jelinek <jakub@redhat.com>
25015 * tree-nvr.c: Include internal-fn.h.
25016 (pass_return_slot::execute): Ignore internal calls without
25019 2017-04-04 Jakub Jelinek <jakub@redhat.com>
25020 Richard Biener <rguenther@suse.de>
25023 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
25024 captures used multiple times, except for the last use.
25025 * generic-match-head.c: Include gimplify.h.
25027 2017-04-04 Jakub Jelinek <jakub@redhat.com>
25029 PR tree-optimization/79390
25030 * target.h (struct noce_if_info): Declare.
25031 * targhooks.h (default_noce_conversion_profitable_p): Declare.
25032 * target.def (noce_conversion_profitable_p): New target hook.
25033 * ifcvt.h (struct noce_if_info): New type, moved from ...
25034 * ifcvt.c (struct noce_if_info): ... here.
25035 (noce_conversion_profitable_p): Renamed to ...
25036 (default_noce_conversion_profitable_p): ... this. No longer
25038 (noce_try_store_flag_constants, noce_try_addcc,
25039 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
25040 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
25041 instead of noce_conversion_profitable_p.
25042 * config/i386/i386.c: Include ifcvt.h.
25043 (ix86_option_override_internal): Don't override
25044 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
25045 (ix86_noce_conversion_profitable_p): New function.
25046 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
25047 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
25048 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
25049 * doc/tm.texi: Regenerated.
25051 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25053 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
25056 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25059 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
25060 instructions for small multiply cores.
25062 2017-04-04 Jeff Law <law@redhat.com>
25064 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
25066 (mips_expand_vec_perm_const): Initialize elements in orig_perm
25067 that are not set by the loop over the elements.
25069 2017-04-04 Jakub Jelinek <jakub@redhat.com>
25072 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
25073 int mode, convert_modes it to mode as unsigned, otherwise use
25074 lowpart_subreg to mode rather than SImode.
25075 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
25076 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
25077 Use DImode instead of SImode for the shift count operand.
25078 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
25081 2017-04-04 Richard Biener <rguenther@suse.de>
25083 PR middle-end/80281
25084 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
25085 arithmetic done for the negate or the plus. Simplify.
25086 (A - (-B) -> A + B): Likewise.
25087 * fold-const.c (split_tree): Make sure to not negate pointers.
25089 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
25091 PR rtl-optimization/60818
25092 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
25093 a compare of comparisons with the thing compared if this results
25094 in a different machine mode.
25096 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
25098 * alias.c (base_alias_check): Fix typo in comment.
25099 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
25100 * cgraphunit.c (symbol_table::compile): Likewise.
25101 * collect2.c (maybe_run_lto_and_relink): Likewise.
25102 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
25103 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
25104 * config/avr/avr.c (avr_map_op_t): Likewise.
25105 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
25106 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
25107 * config/epiphany/epiphany.md (movcc): Likewise.
25108 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
25109 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
25111 * config/mips/mips.c (mips_save_restore_reg): Likewise.
25112 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
25113 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
25114 * config/sh/sh.c (sh_rtx_costs): Likewise.
25115 * fold-const.c (fold_truth_andor): Likewise.
25116 * genautomata.c (collapse_flag): Likewise.
25117 * gengtype.h (struct type::u::s): Likewise.
25118 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
25119 * input.c (FORMAT_AMOUNT): Likewise.
25120 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
25121 (known_aggs_to_agg_replacement_list): Likewise.
25122 * ipa-inline-analysis.c: Likewise.
25123 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
25124 * ipa-polymorphic-call.c
25125 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
25126 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
25127 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
25129 * modulo-sched.c (apply_reg_moves): Likewise.
25130 * omp-expand.c (build_omp_regions_1): Likewise.
25131 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
25132 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
25133 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
25134 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
25135 * value-prof.c: Likewise.
25136 * var-tracking.c (val_reset): Likewise.
25138 2017-04-03 Richard Biener <rguenther@suse.de>
25140 PR tree-optimization/80275
25141 * fold-const.c (split_address_to_core_and_offset): Handle
25144 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
25146 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
25147 descriptors is at least equal to that of functions.
25149 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
25151 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
25153 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
25156 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
25157 (mov<IMOD4:mode>): New expander.
25158 (*mov<IMOD4:mode>_internal): New insn and split pattern.
25160 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
25162 PR rtl-optimization/79405
25163 * fwprop.c (propagations_left): New variable.
25164 (forward_propagate_into): Decrement it.
25165 (fwprop_init): Initialize it.
25166 (fw_prop): If the variable has reached zero, stop propagating.
25167 (fwprop_addr): Ditto.
25169 2017-03-31 Jakub Jelinek <jakub@redhat.com>
25172 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
25173 a FUNCTION_DECL, pass it as decl instead of origin to
25176 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
25178 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
25181 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
25184 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
25185 TARGET_VSX_SMALL_INTEGER.
25187 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25189 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
25190 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
25192 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
25194 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
25195 extraction from odd-numbered MSA register.
25197 2017-03-31 Jakub Jelinek <jakub@redhat.com>
25199 PR middle-end/80173
25200 * expmed.c (store_bit_field_1): Don't attempt to create
25201 a word subreg out of hard registers wider than word if they
25202 have HARD_REGNO_NREGS of 1 for their mode.
25204 PR middle-end/80163
25205 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
25206 conversions to integer types wider than word and pointer.
25209 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
25210 (rtx_equal_for_cselib_p): Pass 0 to it.
25211 * cselib.c (cselib_hasher::equal): Likewise.
25212 (rtx_equal_for_cselib_1): Add depth argument. If depth
25213 is 128, don't look up VALUE locs and punt. Increment
25214 depth in recursive calls when walking VALUE locs.
25216 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
25218 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
25219 (make_gcov_file_name): Use the canonical path name for generating
25221 (read_line): Fix handling of files with ascii null bytes.
25223 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
25225 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
25226 to initialise a vector register instead
25227 of using a const_int.
25229 2017-03-30 Jakub Jelinek <jakub@redhat.com>
25231 PR translation/80189
25232 * gimplify.c (omp_default_clause): Use %qs instead of %s in
25233 diagnostic messages.
25235 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
25238 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
25239 (dfp_diex_<mode>): Update mode of operand 1.
25240 * doc/extend.texi (dxex, dxexq): Document change to return type.
25241 (diex, diexq): Document change to argument type.
25243 2017-03-30 Martin Jambor <mjambor@suse.cz>
25246 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
25247 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
25248 it reflects the signature changes performed at the callee side.
25249 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
25250 to cgraph_build_function_type_skip_args.
25251 (build_function_decl_skip_args): Adjust call to the above function.
25253 2017-03-30 Jakub Jelinek <jakub@redhat.com>
25256 * config/i386/sse.md
25257 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
25258 register as dest whenever it is a MEM not rtx_equal_p to the
25259 corresponding dup operand, and when forcing into reg move the
25260 reg into the memory afterwards.
25261 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
25262 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
25263 for the force_reg mode.
25264 (avx512vl_vextractf128<mode>): Use register as dest either
25265 always when a MEM, or when it is a MEM not rtx_equal_p to the
25266 corresponding dup operand, or even not when it is a CONST_VECTOR
25267 depending on the mode and lo vs. hi.
25268 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
25270 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
25271 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
25272 Likewise. Require that operands[2] is even.
25273 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
25274 Remove extraneous parens. Require that operands[2] is a multiple
25276 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
25277 operands[0] is a MEM if <mask_applied>, the predicates/constraints
25278 disallow memory then.
25280 2017-03-30 Richard Biener <rguenther@suse.de>
25282 PR tree-optimization/77498
25283 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
25284 to non-constants over backedges.
25286 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
25288 PR rtl-optimization/80233
25289 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
25290 as last_combined_insn. Do not test for BARRIER_P separately.
25292 2017-03-29 Andreas Schwab <schwab@suse.de>
25295 * calls.c (prepare_call_address): Convert funexp to Pmode before
25296 copying to temp reg.
25298 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25300 PR tree-optimization/80158
25301 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
25302 Handle possible future case of more than one alternate
25304 (replace_rhs_if_not_dup): Likewise.
25305 (replace_one_candidate): Likewise.
25307 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
25309 PR rtl-optimization/80193
25310 * ira.c (ira): Do not check allocation for LRA.
25312 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
25314 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
25315 (nvptx_output_simt_exit): Declare.
25316 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
25317 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
25318 (init_softstack_frame): Move initialization of crtl->is_leaf to...
25319 (nvptx_declare_function_name): ...here. Emit declaration of local
25320 memory space buffer for omp_simt_enter insn.
25321 (nvptx_output_unisimt_switch): New.
25322 (nvptx_output_softstack_switch): New.
25323 (nvptx_output_simt_enter): New.
25324 (nvptx_output_simt_exit): New.
25325 * config/nvptx/nvptx.h (struct machine_function): New fields
25326 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
25327 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
25328 (UNSPECV_SIMT_EXIT): Ditto.
25329 (omp_simt_enter_insn): New insn.
25330 (omp_simt_enter): New expansion.
25331 (omp_simt_exit): New insn.
25332 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
25334 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
25335 (expand_GOMP_SIMT_ENTER_ALLOC): New.
25336 (expand_GOMP_SIMT_EXIT): New.
25337 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
25338 (GOMP_SIMT_ENTER_ALLOC): Ditto.
25339 (GOMP_SIMT_EXIT): Ditto.
25340 * target-insns.def (omp_simt_enter): New insn.
25341 (omp_simt_exit): Ditto.
25342 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
25344 (lower_rec_simd_input_clauses): Implement SIMT privatization.
25345 (lower_rec_input_clauses): Likewise.
25346 (lower_lastprivate_clauses): Handle SIMT privatization.
25348 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
25349 (ompdevlow_adjust_simt_enter): New.
25350 (find_simtpriv_var_op): New.
25351 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
25352 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
25354 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
25355 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
25356 (copy_decl_for_dup_finish): Ditto.
25358 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
25360 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
25363 * config/i386/i386.c (ix86_option_override_internal): Always
25364 allow -mpreferred-stack-boundary=3 for 64-bit targets.
25366 2017-03-28 Bin Cheng <bin.cheng@arm.com>
25368 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
25370 2017-03-28 Bin Cheng <bin.cheng@arm.com>
25372 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
25373 mark new edge's irreducible flag accordign to it.
25374 (vect_do_peeling): Check loop preheader edge's irreducible flag
25375 and pass it to function slpeel_add_loop_guard.
25377 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
25379 PR tree-optimization/80218
25380 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
25381 Update block frequencies and counts.
25383 2017-03-28 Richard Biener <rguenther@suse.de>
25385 PR tree-optimization/78644
25386 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
25387 of a simplification result we may not use it at all.
25389 2017-03-28 Richard Biener <rguenther@suse.de>
25392 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
25393 without arguments, generate default definition of a SSA name.
25395 2017-03-28 Richard Biener <rguenther@suse.de>
25397 PR middle-end/80222
25398 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
25399 TYPE_REF_CAN_ALIAS_ALL references.
25400 * fold-const.c (fold_indirect_ref_1): Likewise.
25402 2017-03-28 Martin Liska <mliska@suse.cz>
25405 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
25406 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
25408 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
25409 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
25411 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
25412 (EXTRA_SPECS): Define.
25413 (SUBTARGET_EXTRA_SPECS): Likewise.
25414 (SUBTARGET_CPP_SPEC): Likewise.
25415 * config/arc/elf.h (EXTRA_SPECS): Renamed to
25416 SUBTARGET_EXTRA_SPECS.
25417 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
25419 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
25421 * config/arc/simdext.md (vst64_insn): Update pattern.
25422 (vld32wh_insn): Likewise.
25423 (vld32wl_insn): Likewise.
25424 (vld64_insn): Likewise.
25425 (vld32_insn): Likewise.
25427 2017-03-28 Marek Polacek <polacek@redhat.com>
25430 * fold-const.c (fold_comparison): Use protected_set_expr_location
25431 instead of SET_EXPR_LOCATION.
25433 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
25435 * tree.c (add_expr): Avoid name lookup warning.
25437 2017-03-27 Jeff Law <law@redhat.com>
25439 PR tree-optimization/80216
25440 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
25441 function name. Limit recursion depth.
25442 (record_temporary_equivalences): Corresponding changes.
25444 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
25446 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
25449 2017-03-27 Jakub Jelinek <jakub@redhat.com>
25452 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
25454 * cfgcleanup.c (reg_note_cfa_p): New array.
25455 (insns_have_identical_cfa_notes): New function.
25456 (old_insns_match_p): Don't cross-jump in between /f
25457 and non-/f instructions. If both i1 and i2 are frame related,
25458 verify all CFA notes, their order and content.
25460 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
25463 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
25464 HImode and SImode with zero extend to DImode to one insn.
25465 (bswap<mode>2_extenddi): Likewise.
25466 (bswapsi2_extenddi): Likewise.
25467 (bswaphi2_extendsi): Likewise.
25468 (bswaphi2): Combine bswap HImode and SImode into one insn.
25469 Separate memory insns from swapping register.
25470 (bswapsi2): Likewise.
25471 (bswap<mode>2): Likewise.
25472 (bswaphi2_internal): Delete, no longer used.
25473 (bswapsi2_internal): Likewise.
25474 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
25475 store, and gpr<-gpr swap insns.
25476 (bswap<mode>2_store): Likewise.
25477 (bswaphi2_reg): Register only splitter, combine with the splitter.
25478 (bswaphi2 splitter): Likewise.
25479 (bswapsi2_reg): Likewise.
25480 (bswapsi2 splitter): Likewise.
25481 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
25482 the insns into load, store, and register/register insns.
25483 (bswapdi2_ldbrx): Likewise.
25484 (bswapdi2_load): Likewise.
25485 (bswapdi2_store): Likewise.
25486 (bswapdi2_reg): Likewise.
25488 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
25490 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
25491 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
25493 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
25496 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
25498 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
25499 special handling for target option conflicts between dform
25500 options (-mpower9-dform, -mpower9-dform-vector,
25501 -mpower9-dform-scalar) and -mno-direct-move.
25503 2017-03-27 Richard Biener <rguenther@suse.de>
25505 PR tree-optimization/80181
25506 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
25508 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
25510 * config/arc/predicates.md (move_double_src_operand): Replace the
25511 call to move_double_src_operand with a call to address_operand.
25513 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
25515 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
25516 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
25517 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
25519 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
25521 * config/arc/predicates.md (long_immediate_loadstore_operand):
25522 Consider scaled addresses cases.
25524 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
25526 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
25527 restored when in interrupt.
25528 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
25529 doesn't have delay slot.
25531 2017-03-27 Richard Biener <rguenther@suse.de>
25534 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
25535 inlined thunk clones.
25537 2017-03-27 Jakub Jelinek <jakub@redhat.com>
25540 * asan.c (instrument_derefs): Copy over last operand from
25541 original COMPONENT_REF to the new COMPONENT_REF with
25542 DECL_BIT_FIELD_REPRESENTATIVE.
25543 * ubsan.c (instrument_object_size): Likewise.
25545 2017-03-27 Richard Biener <rguenther@suse.de>
25547 PR tree-optimization/80170
25548 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
25549 sure DR/SCEV didnt fold in constants we do not see when looking
25550 at the reference base alignment.
25552 2017-03-27 Richard Biener <rguenther@suse.de>
25554 PR middle-end/80171
25555 * gimple-fold.c (fold_ctor_reference): Properly guard against
25556 NULL return value from canonicalize_constructor_val.
25558 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
25561 * config/i386/i386.c (ix86_expand_builtin)
25562 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
25563 flags reg setting and flags reg using instructions.
25564 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
25565 clobbering instructions to zero extend op2.
25567 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
25569 * doc/install.texi (Configuration) <--with-aix-soname>:
25570 Update link to AIX ld.
25572 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
25574 PR rtl-optimization/80160
25575 PR rtl-optimization/80159
25576 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
25577 reg_alternate_class into account.
25579 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
25582 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
25583 to consider in curr_insn_transform.
25585 2017-03-24 Jakub Jelinek <jakub@redhat.com>
25587 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
25588 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
25589 and emit_mode_inner.
25591 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25593 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
25594 argument to the overloaded builtin variants. Use the new flag to
25595 deprecate certain builtin variants.
25596 * config/s390/s390-builtin-types.def: Add new builtin types.
25597 * config/s390/s390-builtins.h: Support new flags field for
25598 overloaded builtins.
25599 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
25600 (s390_macro_to_expand): Enable vector float data type.
25601 (s390_cpu_cpp_builtins_internal): Indicate support of the new
25602 builtins by incrementing the __VEC__ version number.
25603 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
25605 (s390_resolve_overloaded_builtin): Emit error messages depending
25606 on the builtin flags.
25607 * config/s390/s390.c (s390_expand_builtin): Support additional
25608 flags argument. Change error message to match the messages
25609 emitted in s390-c.c.
25610 * config/s390/s390.md: New UNSPEC_* constants.
25611 (op_type): Add new instruction types.
25612 * config/s390/vecintrin.h: Add new builtins and test data class
25614 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
25615 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
25616 (VEC_INEXACT, VEC_NOINEXACT): New constants.
25617 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
25618 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
25619 ("vec_mergel<mode>"): V_HW -> VEC_HW.
25621 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
25622 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
25623 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
25624 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
25626 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
25627 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
25628 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
25629 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
25631 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
25632 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
25633 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
25634 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
25635 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
25636 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
25637 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
25639 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
25640 ("vec_scatter_element<V_HW_4:mode>_DI")
25641 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
25642 ("vec_fpint<mode>", "vflls")
25643 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
25644 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
25645 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
25646 ("*vec_cmphe<mode>_cc"): ... these.
25648 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
25649 mode constant instead of magic value.
25651 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25653 * config/s390/s390.c (s390_expand_vec_compare): Support other
25654 vector floating point modes than just V2DF.
25655 (s390_expand_vcond): Likewise.
25656 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
25657 (s390_cannot_change_mode_class): Prevent mode changes between TF
25658 and V1TF in vector registers.
25659 * config/s390/s390.md (DF, SF): New mode attributes.
25660 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
25661 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
25662 SFmode support for VRs.
25663 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
25665 (VFT, VF_HW): New mode iterators.
25666 (vw, sdx): New mode attributes.
25667 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
25668 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
25669 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
25670 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
25671 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
25672 also the new vector floating point modes. Renaming to ...
25674 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
25675 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
25676 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
25677 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
25678 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
25679 ("vec_unordered<mode>"): ... these.
25681 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
25682 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
25683 ("*vec_extendv2df"): New insn definitions.
25685 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25687 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
25688 ("mulditi3_2", "*muldi3_sign"): New patterns.
25689 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
25690 rename the pattern definition.
25692 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25694 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
25696 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
25698 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25700 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
25701 instruction if possible.
25702 * config/s390/vector.md (vec_halfnumelts): New mode
25704 ("*vec_vllezlf<mode>"): New pattern.
25706 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25708 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
25709 ("popcountv4si2", "popcountv2di2"): Rename to ...
25710 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
25711 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
25713 ("popcount<mode>2_vxe"): New pattern.
25715 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25717 * common/config/s390/s390-common.c (processor_flags_table): Add
25719 * config.gcc: Add arch12.
25720 * config/s390/driver-native.c (s390_host_detect_local_cpu):
25721 Default to arch12 for unknown CPU model numbers.
25722 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
25723 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
25724 PROCESSOR_max sanity check.
25725 * config/s390/s390-opts.h (enum processor_type): Add
25727 * config/s390/s390.c (processor_table): Add arch12.
25728 (s390_expand_builtin): Add check for B_VXE flag.
25729 (s390_issue_rate): Add PROCESSOR_ARCH12.
25730 (s390_get_sched_attrmask): Likewise.
25731 (s390_get_unit_mask): Likewise.
25732 (s390_sched_score): Enable z13 scheduling for arch12.
25733 (s390_sched_reorder): Likewise.
25734 (s390_sched_variable_issue): Likewise.
25735 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
25737 (s390_tune_attr): Use z13 scheduling also for arch12.
25738 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
25739 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
25740 (TARGET_VXE_P): New macros.
25741 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
25742 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
25743 * config/s390/s390.opt: Add arch12 as processor_type.
25745 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25747 * config/s390/s390.md
25748 ("fixuns_truncdddi2", "fixuns_trunctddi2")
25749 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
25750 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
25752 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
25753 Rename expanders to ...
25755 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
25756 ("fixuns_truncdddi2_emu"): ... these.
25758 ("fixuns_trunc<mode>si2_emu"): New expander.
25760 ("*fixuns_truncdfdi2_z13"): Rename to ...
25761 ("*fixuns_truncdfdi2_vx"): ... this.
25763 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25765 * config/s390/2964.md: Remove the single element vector compare
25766 instructions which are no longer used.
25767 * config/s390/s390.c (s390_select_ccmode): Remove handling of
25769 (s390_canonicalize_comparison): Remove handling of DFmode
25771 (s390_expand_vec_compare_scalar): Remove function.
25772 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
25773 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
25775 ("*cmp<mode>_ccs"): Add wfcdb instruction.
25777 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25779 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
25781 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
25782 will anyway by matched by mov<mode>_64dfp.
25784 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25786 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
25787 vlef/vstef. Add missing operand to vleif.
25789 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25791 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
25792 pair for all vector types with 64 bit elements.
25793 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
25794 * config/s390/vector.md (V_HW_64): ... here.
25795 (V_128_NOSINGLE): New mode iterator.
25796 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
25797 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
25798 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
25799 ("*vec_load_pairv2di"): Change to ...
25800 ("*vec_load_pair<mode>"): ... this one.
25802 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25804 * config/s390/constraints.md: Add comments.
25805 (jKK): Reject element sizes > 8 bytes.
25806 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
25808 * config/s390/s390.md: Add the s_operand checks formerly in
25809 s390_split_ok_p to various splitters where they are still
25811 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
25812 for 128 bit vectors. Plus two splitters.
25814 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25816 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
25819 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25822 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
25823 error if the boundary argument is not constant.
25825 2017-03-24 Jakub Jelinek <jakub@redhat.com>
25827 PR rtl-optimization/80112
25828 * loop-doloop.c (doloop_condition_get): Don't check condition
25829 if cmp isn't SET with IF_THEN_ELSE src.
25831 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25833 PR tree-optimization/80158
25834 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
25835 replacing a candidate statement, also replace it for the
25836 candidate's alternate interpretation.
25837 (replace_rhs_if_not_dup): Likewise.
25838 (replace_one_candidate): Likewise.
25840 2017-03-24 Richard Biener <rguenther@suse.de>
25842 PR tree-optimization/80167
25843 * graphite-isl-ast-to-gimple.c
25844 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
25846 (translate_isl_ast_to_gimple::get_rename): Likewise.
25848 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
25850 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
25851 handling of certain combinations of target options, including the
25852 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
25853 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
25855 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25858 * config/arm/arm.md (*load_multiple): Add reload_completed to
25859 matching condition.
25861 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25862 Richard Biener <rguenth@suse.de>
25864 PR tree-optimization/79908
25865 PR tree-optimization/80136
25866 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
25867 been cast away, gimplify_and_add suffices.
25869 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
25871 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
25873 2017-03-23 Richard Biener <rguenther@suse.de>
25875 PR tree-optimization/80032
25876 * gimplify.c (gimple_push_cleanup): Forced unconditional
25877 cleanups still have to go to the conditional_cleanups
25880 2017-03-22 Jakub Jelinek <jakub@redhat.com>
25882 PR tree-optimization/80072
25883 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
25885 (next_operand_entry_id): Change type to unsigned int.
25886 (sort_by_operand_rank): Make sure to return the right return value
25887 even if unsigned fields are bigger than INT_MAX.
25888 (struct oecount): Change cnt and id type to unsigned int.
25889 (oecount_hasher::equal): Formatting fix.
25890 (oecount_cmp): Make sure to return the right return value
25891 even if unsigned fields are bigger than INT_MAX.
25892 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
25895 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
25896 TREE_READONLY on result if writing it more than once.
25899 * doc/invoke.texi (-fsanitize=thread): Document that with
25900 -fnon-call-exceptions atomics are not able to throw
25904 * tsan.c: Include tree-eh.h.
25905 (instrument_builtin_call): Call maybe_clean_eh_stmt or
25906 maybe_clean_or_replace_eh_stmt where needed.
25907 (instrument_memory_accesses): Add cfg_changed argument.
25908 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
25909 if it returned true.
25910 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
25912 PR rtl-optimization/63191
25913 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
25914 wrapper function, moved the whole old content into ...
25915 (ix86_delegitimize_address_1): ... this. New inline function.
25916 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
25917 true as last argument instead of ix86_delegitimize_address.
25919 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
25921 * config/aarch64/aarch64.c (generic_branch_cost): Copy
25922 cortexa57_branch_cost.
25924 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
25926 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
25928 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25931 * doc/md.texi (Constraints): Document wA constraint.
25932 * config/rs6000/constraints.md (wA): New.
25933 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
25934 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
25935 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
25936 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
25938 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
25941 * gimplify.c (is_oacc_declared): New function.
25942 (oacc_default_clause): Use it to set default flags for acc declared
25943 variables inside parallel regions.
25944 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
25945 declared variables.
25946 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
25947 declare attribute to any decl as necessary.
25949 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
25952 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
25953 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
25954 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
25955 (arm_arch_lpae): This.
25956 * config/arm/arm.c (arm_arch7ve): Rename into ...
25957 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
25958 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
25961 2017-03-22 Martin Liska <mliska@suse.cz>
25964 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
25965 error message instead of an ICE.
25967 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25969 * doc/extend.texi (6.11 Additional Floating Types): Revise.
25971 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
25973 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
25975 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
25978 2017-03-21 Martin Sebor <msebor@redhat.com>
25980 * doc/extend.texi: Use "cannot" instead of "can't."
25981 * doc/hostconfig.texi: Same.
25982 * doc/install.texi: Same.
25983 * doc/invoke.texi: Same.
25984 * doc/loop.texi: Same.
25985 * doc/md.texi: Same.
25986 * doc/objc.texi: Same.
25987 * doc/rtl.texi: Same.
25988 * doc/tm.texi: Same.
25989 * doc/tm.texi.in: Same.
25990 * doc/trouble.texi: Same.
25992 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
25995 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
25996 (collect_checksum_attributes): Set it.
25997 (die_checksum_ordered): Use it.
25999 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26001 PR tree-optimization/79908
26002 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
26003 change: For a VA_ARG whose LHS has been cast away, use
26004 force_gimple_operand to construct the side effects.
26006 2017-03-21 David Malcolm <dmalcolm@redhat.com>
26008 PR translation/80001
26009 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
26010 more amenable to translation.
26011 (oacc_loop_auto_partitions): Likewise.
26013 2017-03-21 Marek Polacek <polacek@redhat.com>
26014 Martin Sebor <msebor@redhat.com>
26016 PR tree-optimization/80109
26017 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
26018 on INTEGRAL_TYPE_P.
26020 2017-03-21 Jakub Jelinek <jakub@redhat.com>
26021 Segher Boessenkool <segher@kernel.crashing.org>
26024 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
26025 check reg_used_between_p between insn and one of succ or succ2
26026 depending on if succ is artificial insn not inserted into insn
26029 2017-03-21 Martin Liska <mliska@suse.cz>
26031 PR gcov-profile/80081
26032 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
26033 * doc/gcc.texi: Include gcov-dump stuff.
26034 * doc/gcov-dump.texi: New file.
26036 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
26038 PR rtl-optimization/79150
26039 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
26040 conditional jump, if the jump is the last insn of the loop.
26042 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26043 Richard Biener <rguenth@suse.de>
26045 PR tree-optimization/79908
26046 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
26047 been cast away, use force_gimple_operand to construct the side
26050 2017-03-21 Martin Liska <mliska@suse.cz>
26052 PR libfortran/79956
26053 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
26056 2017-03-21 Brad Spengler <spender@grsecurity.net>
26059 * plugin.c (htab_hash_plugin): New function.
26060 (add_new_plugin): Use it and adjust.
26061 (parse_plugin_arg_opt): Adjust.
26062 (init_one_plugin): Likewise.
26064 2017-03-21 Richard Biener <rguenther@suse.de>
26066 PR tree-optimization/80032
26067 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
26068 if set force the cleanup to happen unconditionally.
26069 (gimplify_target_expr): Push inserted clobbers with force_uncond
26070 to avoid them being removed by control-dependent DCE.
26072 2017-03-21 Richard Biener <rguenther@suse.de>
26074 PR tree-optimization/80122
26075 * tree-inline.c (copy_bb): Do not expans va-arg packs or
26076 va_arg_pack_len when the inlined call stmt requires pack
26078 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
26080 2017-03-21 Jakub Jelinek <jakub@redhat.com>
26083 * tsan.c (instrument_builtin_call): If the memory model argument
26084 is not a constant, assume it is valid.
26087 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
26090 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
26092 PR rtl-optimization/79910
26093 * combine.c (can_combine_p): Do not allow combining an I0 or I1
26094 if its dest is used by an insn before I2 (other than the combined
26095 insns themselves, which are properly handled already).
26097 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
26100 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
26102 * combine.c (record_used_regs): New static function.
26103 (try_combine): Handle situations where there is an additional
26104 instruction between I2 and I3 which needs to have a LOG_LINK
26108 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
26110 * combine.c (try_combine): Delete redundant i1 test. Call
26111 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
26113 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26116 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
26117 alternatives 13/14.
26119 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26121 PR tree-optimization/80054
26122 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
26123 the optimization if a PHI or any of its arguments is not dominated
26124 by the candidate's basis. Use gphi* rather than gimple* as
26126 (replace_profitable_candidates): Clean up a gimple* variable that
26127 should be a gphi* variable.
26129 2017-03-20 Martin Sebor <msebor@redhat.com>
26132 * doc/extend.texi (attribute constructor): Document present limitation.
26134 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
26137 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
26138 __POWER9_VECTOR__ #ifdef control, change template definition to
26139 use Power9-specific built-in function.
26140 (vec_any_eq): Likewise.
26141 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
26142 to control outcomes from this test.
26143 (vector_ae_<mode>p): For VEC_F modes, likewise.
26145 2017-03-20 Ian Lance Taylor <iant@google.com>
26147 * config/i386/i386.c (ix86_function_regparm): Save an extra
26148 register for -fsplit-stack with DECL_STATIC_CHAIN.
26150 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
26153 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
26154 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
26156 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
26158 * config/riscv/riscv.c (riscv_print_operand): Use "fence
26160 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
26163 2017-03-20 Marek Polacek <polacek@redhat.com>
26166 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
26168 2017-03-20 Richard Biener <rguenther@suse.de>
26170 PR tree-optimization/80113
26171 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
26172 allocate extra SSA name for PHI def.
26173 (add_close_phis_to_outer_loops): Likewise.
26174 (add_close_phis_to_merge_points): Likewise.
26175 (copy_loop_close_phi_args): Likewise.
26176 (copy_cond_phi_nodes): Likewise.
26178 2017-03-20 Martin Liska <mliska@suse.cz>
26180 PR middle-end/79753
26181 * tree-chkp.c (chkp_build_returned_bound): Do not build
26182 returned bounds for a LHS that's not a BOUNDED_P type.
26184 2017-03-20 Martin Liska <mliska@suse.cz>
26188 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
26189 COMPLEX_CST and VECTOR_CST.
26191 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26194 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
26195 target operand. A new splitter adds the clobber statement in case
26196 the target operand is dead anyway.
26198 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
26200 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
26201 to age-old versions of binutils and glibc.
26203 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
26205 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
26207 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
26209 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
26211 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
26213 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
26214 requirement for binutils 2.13.
26216 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
26218 * combine.c (try_combine): Delete redundant i1 test. Call
26219 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
26221 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
26223 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
26224 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
26226 <riscv64-*-elf>: Re-arrange section
26227 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
26228 <riscv32-*-linux>: Likewise.
26229 <riscv64-*-elf>: Likewise
26230 <riscv64-*-linux>: Likewise.
26232 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
26235 * aarch64.opt(verbose-cost-dump): Fix typo.
26237 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
26240 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
26241 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
26243 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
26245 * reload.c (find_reloads): When reloading a nonoffsettable address,
26246 use RELOAD_OTHER for it and its address reloads.
26248 PR rtl-optimization/79910
26249 * combine.c (record_used_regs): New static function.
26250 (try_combine): Handle situations where there is an additional
26251 instruction between I2 and I3 which needs to have a LOG_LINK
26254 2017-03-17 Jeff Law <law@redhat.com>
26256 PR tree-optimization/71437
26257 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
26258 conditional in the hash table first.
26259 (vrp_dom_walker::before_dom_children): Extract condition from
26260 ASSERT_EXPR. Record condition, its inverion and any implied
26261 conditions as well.
26263 2017-03-17 Marek Polacek <polacek@redhat.com>
26264 Markus Trippelsdorf <markus@trippelsdorf.de>
26266 PR tree-optimization/80079
26267 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
26270 2017-03-17 Richard Biener <rguenther@suse.de>
26272 PR middle-end/80075
26273 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
26274 Properly verify the LHS before the RHS possibly claims to be
26276 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
26279 2017-03-17 Martin Jambor <mjambor@suse.cz>
26281 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
26282 (List of -O2 options): Likewise.
26283 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
26286 2017-03-17 Tom de Vries <tom@codesourcery.com>
26288 * gcov-dump.c (print_usage): Print bug_report_url.
26290 2017-03-17 Richard Biener <rguenther@suse.de>
26292 PR middle-end/80050
26293 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
26294 (parser::peek): Likewise.
26296 2017-03-17 Richard Biener <rguenther@suse.de>
26298 PR tree-optimization/80048
26299 * sese.c (free_sese_info): Properly release rename_map and
26300 copied_bb_map elements.
26302 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
26304 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
26305 Add linked-list forward and backlinks. Insert on
26306 construction, remove on destruction.
26307 (class pass_store_merging): Add m_stores_head field.
26308 (pass_store_merging::terminate_and_process_all_chains):
26309 Iterate over m_stores_head list.
26310 (pass_store_merging::terminate_all_aliasing_chains):
26312 (pass_store_merging::execute): Check for debug stmts first.
26313 Push new chains onto the m_stores_head stack.
26315 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
26318 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
26319 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
26320 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
26322 2017-03-16 Jeff Law <law@redhat.com>
26324 PR tree-optimization/71437
26325 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
26326 member function. Implementation moved into after_dom_children
26327 member function and into the threader's thread_outgoing_edges
26329 (dom_opt_dom_walker::after_dom_children): Simplify by moving
26330 some code into new thread_outgoing_edges.
26331 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
26332 definition. Simplify marker handling (do it here). Assume we always
26333 have the available expression and the const/copies tables.
26334 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
26336 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
26337 * tree-vrp.c (equiv_stack): No longer file scoped.
26338 (vrp_dom_walker): New class.
26339 (vrp_dom_walker::before_dom_children): New member function.
26340 (vrp_dom_walker::after_dom_children): Likewise.
26341 (identify_jump_threads): Setup domwalker. Use it rather than
26342 walking edges in a random order by hand. Simplify setup/finalization.
26343 (finalize_jump_threads): Remove.
26344 (vrp_finalize): Do not call identify_jump_threads here.
26345 (execute_vrp): Do it here instead and call thread_through_all_blocks
26348 PR tree-optimization/71437
26349 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
26351 (simplify_stmt_for_jump_threading): Add basic_block argument. All
26353 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
26354 (dom_opt_dom_walker::thread_across_edge): Remove
26355 handle_dominating_asserts argument. All callers changed.
26356 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
26357 changes. Remove calls to lhs_of_dominating_assert. Other
26358 uses of handle_dominating_asserts turn into unconditional code
26359 (simplify_control_stmt_condition_1): Likewise.
26360 (simplify_control_stmt_condition): Likewise.
26361 (thread_through_normal_block, thread_across_edge): Likewise.
26362 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
26363 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
26364 object if it is not an SSA_NAME.
26365 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
26366 before calling into the VRP specific simplifiers.
26367 (identify_jump_threads): Remove handle_dominating_asserts
26370 2017-03-16 Jakub Jelinek <jakub@redhat.com>
26373 * tree-diagnostic.c (default_tree_printer): No longer static.
26374 * tree-diagnostic.h (default_tree_printer): New prototype.
26376 2017-03-16 Tamar Christina <tamar.christina@arm.com>
26378 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
26379 Change ins into fmov.
26381 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26383 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
26384 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
26385 Use h_con constraint for operand 1.
26386 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
26387 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
26389 2017-03-15 Jeff Law <law@redhat.com>
26391 PR tree-optimization/71437
26392 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
26393 (record_temporary_equivalences): Use it.
26395 PR tree-optimization/71437
26396 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
26397 tree-ssa-scopedtables.
26398 (lookup_avail_expr, build_and_record_new_cond): Likewise.
26399 (record_conditions, record_cond, vuse_eq): Likewise.
26400 (record_edge_info): Adjust to API tweak of record_conditions.
26401 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
26402 (record_temporary_equivalences, optimize_stmt): Likewise.
26403 (eliminate_redundant_computations): Likewise.
26404 (record_equivalences_from_stmt): Likewise.
26405 * tree-ssa-scopedtables.c: Include options.h and params.h.
26406 (vuse_eq): New function, moved from tree-ssa-dom.c
26407 (build_and_record_new_cond): Likewise.
26408 (record_conditions): Likewise. Accept vector of conditions rather
26409 than edge_equivalence structure for first argument.
26410 for the first argument.
26411 (avail_exprs_stack::lookup_avail_expr): New member function, moved
26412 from tree-ssa-dom.c.
26413 (avail_exprs_stack::record_cond): Likewise.
26414 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
26415 from tree-ssa-dom.c.
26416 (avail_exprs_stack): Add new member functions lookup_avail_expr
26418 (record_conditions): Declare.
26420 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
26423 * lra-constraints.c (process_alt_operands): Increase reject for
26424 reloading an input/output operand.
26426 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
26429 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
26430 insns to convert from signed/unsigned char/short to IEEE 128-bit
26432 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
26434 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
26437 * config/i386/i386.c (ix86_vector_duplicate_value): Create
26438 subreg of inner mode for values already in registers.
26440 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
26442 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
26443 iteration reg is used after the loop.
26445 2017-03-14 Martin Sebor <msebor@redhat.com>
26447 PR tree-optimization/79800
26448 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
26449 precision in negative-positive range.
26450 (format_floating): Call non-const overload with adjusted precision.
26452 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
26455 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
26458 2017-03-14 Martin Sebor <msebor@redhat.com>
26460 PR middle-end/80020
26461 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
26462 * builtins.def (aligned_alloc): Use it.
26465 * Makefile.in (GTFILES): Add calls.c.
26466 * calls.c: Include "gt-calls.h".
26468 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
26470 PR rtl-optimization/79728
26471 * regs.h (struct target_regs): New field
26472 x_contains_allocatable_regs_of_mode.
26473 (contains_allocatable_regs_of_mode): New macro.
26474 * reginfo.c (init_reg_sets_1): Initialize it, and change
26475 contains_reg_of_mode so it includes global regs as well.
26476 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
26477 rather than contains_regs_of_mode.
26479 2017-03-14 Martin Liska <mliska@suse.cz>
26481 * doc/invoke.texi: Document options that can't be combined with
26482 -fcheck-pointer-bounds.
26484 2017-03-14 Martin Liska <mliska@suse.cz>
26486 PR middle-end/79831
26487 * doc/invoke.texi (-Wchkp): Document the option.
26489 2017-03-14 Martin Liska <mliska@suse.cz>
26491 * Makefile.in: Install gcov-dump.
26493 2017-03-14 Martin Liska <mliska@suse.cz>
26495 * multiple_target.c (expand_target_clones): Bail out for
26496 an invalid attribute.
26498 2017-03-14 Richard Biener <rguenther@suse.de>
26500 * alias.c (struct alias_set_entry): Pack properly.
26501 * cfgloop.h (struct loop): Likewise.
26502 * cse.c (struct set): Likewise.
26503 * ipa-utils.c (struct searchc_env): Likewise.
26504 * loop-invariant.c (struct invariant): Likewise.
26505 * lra-remat.c (struct cand): Likewise.
26506 * recog.c (struct change_t): Likewise.
26507 * rtl.h (struct address_info): Likewise.
26508 * symbol-summary.h (function_summary): Likewise.
26509 * tree-loop-distribution.c (struct partition): Likewise.
26510 * tree-object-size.c (struct object_size_info): Likewise.
26511 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
26512 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
26513 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
26514 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
26515 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
26516 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
26517 (struct _stmt_vec_info): Likewise.
26519 2017-03-14 Martin Liska <mliska@suse.cz>
26522 * multiple_target.c (create_dispatcher_calls): Check that
26523 a target can create a function dispatcher.
26525 2017-03-14 Martin Liska <mliska@suse.cz>
26528 * multiple_target.c (expand_target_clones): Drop local.local
26529 flag for default implementation.
26531 2017-03-14 Richard Biener <rguenther@suse.de>
26533 PR tree-optimization/80030
26534 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
26536 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
26538 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
26539 gcc_fallthrough() instead of __attribute__((fallthrough));
26541 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
26543 * doc/gcc.texi: Remove "up" link to (DIR).
26544 * doc/gccint.texi: Ditto.
26546 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
26548 * doc/install.texi (Specific) <avr>: Remove reference to
26551 2017-03-13 Jeff Law <law@redhat.com>
26553 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
26554 attribute rather than comments.
26556 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
26557 match_scratch operand is highest.
26559 2017-03-13 Martin Liska <mliska@suse.cz>
26561 PR middle-end/78339
26562 * ipa-pure-const.c (warn_function_noreturn): If the declarations
26563 is a CHKP clone, use original declaration.
26565 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26567 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
26568 (arc_conditional_register_usage): Use a different allocation order
26569 when optimizing for size.
26570 * common/config/arc/arc-common.c (arc_option_optimization_table):
26571 Section anchors default on when optimizing for size.
26573 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26575 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
26577 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26579 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
26580 * config/arc/arc.md (cpu_facility): Add cd variant.
26581 (*movqi_insn): Add code density variant.
26582 (*movhi_insn): Likewise.
26583 (*movqi_insn): Likewise.
26584 (*addsi3_mixed): Likewise.
26585 (subsi3_insn): Likewise.
26587 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26589 * config/arc/arc.md (movsi_cond_exec): Update constraint.
26591 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26593 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
26594 expressions with MINUS and UNARY ops.
26596 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26599 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
26601 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
26602 between vec_select and vector argument.
26603 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
26604 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
26605 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
26606 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
26607 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
26608 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
26610 2017-03-13 Richard Biener <rguenther@suse.de>
26613 * params.def (vect-max-peeling-for-alignment): Fix typo.
26615 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
26617 * doc/install.texi (Specific) <mips-*-*>: Remove description of
26618 issue that only occurred with binutils below 2.18.
26620 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
26622 * doc/install.texi (Specific) <cris-axis-elf>: No longer
26623 refer to binutils 2.11/2.12 minimum.
26625 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
26627 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
26628 ftp.kernel.org and simplify binutils requirement.
26630 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
26632 * doc/invoke.texi (Warning Options): Fix spelling of link-time
26634 (Optimize Options): Ditto. Also remove redundancy.
26636 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26638 PR translation/79848
26639 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
26641 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
26642 to G_ to avoid double translation.
26644 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26646 PR translation/79923
26647 * auto-profile.c (get_combined_location): Convert leading
26648 character of diagnostics to lower case and remove trailing period.
26649 (read_profile): Likewise for various diagnostics.
26650 * config/arm/arm.c (arm_option_override): Remove trailing period
26651 from various diagnostics.
26652 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
26653 (msp430_expand_delay_cycles): Likewise.
26655 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26658 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
26659 full command-line argument, rather than just "str".
26660 (aarch64_validate_march): Likewise.
26661 (aarch64_validate_mtune): Likewise.
26663 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
26665 PR rtl-optimization/78911
26666 * lra-assigns.c (must_not_spill_p): New function.
26667 (spill_for): Use it.
26669 2017-03-10 Jakub Jelinek <jakub@redhat.com>
26671 PR tree-optimization/79981
26672 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
26673 ATOMIC_COMPARE_EXCHANGE ifn result.
26674 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
26675 IFN_ATOMIC_COMPARE_EXCHANGE.
26677 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26680 * opts.c (parse_sanitizer_options): Add missing question mark to
26681 "did you mean" message.
26683 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26685 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
26687 (VMULEUH_UNS): Likewise.
26688 (VMULOUB_UNS): Likewise.
26689 (VMULOUH_UNS): Likewise.
26690 * config/rs6000/rs6000.c (builtin_function_type): Remove
26691 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
26693 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26696 * read-rtl-function.c (function_reader::read_rtx_operand): Update
26697 x with result of extra_parsing_for_operand_code_0.
26698 (function_reader::extra_parsing_for_operand_code_0): Convert
26699 return type from void to rtx, returning x. When reading
26700 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
26701 larger size containing struct block_symbol.
26703 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
26705 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
26706 -mfloat128-hardware without -m64.
26708 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
26711 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
26712 entries to the case statement that marks unsigned arguments to
26713 overloaded functions.
26715 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26717 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
26718 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
26720 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
26723 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
26724 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
26726 2017-03-10 Martin Liska <mliska@suse.cz>
26730 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
26731 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
26734 2017-03-10 Olivier Hainque <hainque@adacore.com>
26736 * tree-switch-conversion (array_value_type): Start by resetting
26737 candidate type to it's main variant.
26739 2017-03-10 Jakub Jelinek <jakub@redhat.com>
26741 PR rtl-optimization/79909
26742 * combine.c (try_combine): Use simplify_replace_rtx on individual
26743 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
26744 of the whole CALL_INSN_FUNCTION_USAGE.
26746 PR tree-optimization/79972
26747 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
26748 get_range_info on SSA_NAMEs. Formatting fixes.
26750 2017-03-10 Richard Biener <rguenther@suse.de>
26751 Jakub Jelinek <jakub@redhat.com>
26753 PR tree-optimization/77975
26754 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
26755 edge to be constant.
26756 (get_val_for): For constant x return it. Formatting fix.
26757 (loop_niter_by_eval): Avoid pointless looping if the next iteration
26758 would use the same bases as the current one.
26760 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26762 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
26763 instead of vec_select for V1TImode.
26764 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
26766 (VSX_LE_128): Add V1TI to this mode iterator.
26767 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
26768 (*vsx_le_perm_store_<mode>): Likewise.
26769 (pre-reload splitter for VSX stores): Likewise.
26770 (post-reload splitter for VSX stores): Likewise.
26771 (*vsx_xxpermdi2_le_<mode>): Likewise.
26772 (*vsx_lxvd2x2_le_<mode>): Likewise.
26773 (*vsx_stxvd2x2_le_<mode>): Likewise.
26775 2017-03-09 Michael Eager <eager@eagercon.com>
26777 Correct failures with --enable-checking=yes,rtl.
26779 * config/microblaze/microblaze.c (microblaze_expand_shift):
26780 Replace GET_CODE test with CONST_INT_P and INTVAL test with
26781 test for const0_rtx.
26782 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
26783 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
26785 2017-03-09 Richard Biener <rguenther@suse.de>
26787 PR tree-optimization/79977
26788 * graphite-scop-detection.c (scop_detection::merge_sese):
26789 Handle the case of extra exits to blocks dominating the entry.
26791 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
26793 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
26796 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
26798 PR rtl-optimization/79949
26799 * lra-constraints.c (process_alt_operands): Check memory when
26800 trying to predict a cycle. Print about the overall increase.
26802 2017-03-09 Richard Biener <rguenther@suse.de>
26804 PR middle-end/79971
26805 * gimple-expr.c (useless_type_conversion_p): Preserve
26806 TYPE_SATURATING for fixed-point types.
26808 2017-03-09 Richard Biener <rguenther@suse.de>
26811 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
26812 alignment of BLKmode params.
26814 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26817 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
26818 (VALL_NO_V2Q): Likewise.
26820 * config/aarch64/aarch64-simd.md
26821 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
26823 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
26824 VALL_NO_V2Q mode iterator.
26825 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
26827 2017-03-09 Martin Liska <mliska@suse.cz>
26829 PR tree-optimization/79631
26830 * tree-chkp-opt.c (chkp_is_constant_addr): Call
26831 tree_int_cst_sign_bit just for INTEGER constants.
26833 2017-03-09 Martin Liska <mliska@suse.cz>
26837 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
26840 2017-03-09 Marek Polacek <polacek@redhat.com>
26843 * tree.c (inchash::add_expr): Handle TREE_VEC.
26845 2017-03-09 Martin Liska <mliska@suse.cz>
26848 (chkp_narrow_size_and_offset): New function.
26849 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
26850 (void chkp_parse_bit_field_ref): New function.
26851 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
26852 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
26854 2017-03-09 Martin Liska <mliska@suse.cz>
26857 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
26858 (chkp_find_bounds_1): Remove gcc_unreachable.
26860 2017-03-09 Jakub Jelinek <jakub@redhat.com>
26863 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
26864 BUILT_IN_SYNC*, determine the access type from the size suffix and
26865 always build a MEM_REF with that type. Handle forgotten
26866 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
26869 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
26870 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
26871 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
26872 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
26873 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
26874 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
26875 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
26876 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
26877 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
26878 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
26879 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
26880 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
26881 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
26882 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
26883 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
26884 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
26885 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
26886 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
26887 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
26888 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
26889 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
26890 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
26891 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
26892 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
26893 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
26894 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
26895 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
26896 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
26897 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
26898 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
26899 definitions outside of __OPTIMIZE__ guarded section.
26902 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
26903 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
26904 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
26905 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
26908 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26910 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
26911 ("vfenez<mode>"): Add missing constraints.
26913 2017-03-08 Martin Sebor <msebor@redhat.com>
26916 * config/nds32/nds32.c (nds32_option_override):
26917 Fix misspelled diagnostic.
26919 2017-03-08 Jakub Jelinek <jakub@redhat.com>
26922 * gimplify.c (gimplify_omp_for): Replace index var in outer
26923 taskloop statement with an artificial variable and add
26924 OMP_CLAUSE_PRIVATE clause for it.
26926 2017-03-08 Richard Biener <rguenther@suse.de>
26928 PR tree-optimization/79955
26929 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
26930 for accesses that are completely outside of the variable.
26932 2017-03-08 Andrew Haley <aph@redhat.com>
26934 PR tree-optimization/79943
26935 * tree-ssa-loop-split.c (compute_new_first_bound): When
26936 calculating the new upper bound, (END-BEG) should be added, not
26939 2017-03-08 Jakub Jelinek <jakub@redhat.com>
26941 * config/avr/avr.md (setmemhi): Make sure match_dup
26942 operand number comes before match_scratch.
26944 2017-03-08 Richard Biener <rguenther@suse.de>
26946 PR tree-optimization/79920
26947 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
26948 with ncopies == 1 to ...
26949 (vect_transform_slp_perm_load): ... here. Properly compute
26950 all element loads by iterating VF times over the group. Do
26951 not handle ncopies (computed in a broken way) in
26952 vect_create_mask_and_perm.
26954 2017-03-08 Jakub Jelinek <jakub@redhat.com>
26957 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
26958 is a uniform vector, use uniform_vector_p return value instead of
26959 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
26961 2017-03-07 Marek Polacek <polacek@redhat.com>
26963 PR middle-end/79809
26964 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
26965 (alloca_call_type): Likewise.
26967 2017-03-07 Martin Liska <mliska@suse.cz>
26969 * gcov.c (process_args): Put comment to correct location.
26971 2017-03-07 Martin Liska <mliska@suse.cz>
26973 PR middle-end/68270
26974 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
26975 Use array_at_struct_end_p instead of DECL_CHAIN (field).
26976 (chkp_narrow_bounds_for_field): Likewise.
26977 (chkp_parse_array_and_component_ref): Pass one more argument to
26980 2017-03-07 Richard Biener <rguenther@suse.de>
26982 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
26985 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
26987 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
26988 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
26990 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26993 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
26994 to end of description.
26995 (PARAM_MAX_STORES_TO_MERGE): Likewise.
26997 2017-03-07 Jakub Jelinek <jakub@redhat.com>
26999 PR rtl-optimization/79901
27000 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
27002 (*avx512f_<code><mode>3<mask_name>): ... this.
27003 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
27004 iterator instead of VI8_AVX2_AVX512BW.
27006 PR rtl-optimization/79901
27007 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
27008 min/max expander, expand it using expand_vec_cond_expr.
27011 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
27014 2017-03-06 Jakub Jelinek <jakub@redhat.com>
27017 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
27018 to void * for PCH reasons.
27019 * dwarf2out.c (output_loc_operands, output_die): Cast
27020 v.val_vec.array to unsigned char *.
27022 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
27025 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
27028 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
27030 PR rtl-optimization/79571
27031 * lra-constraints.c (process_alt_operands): Calculate static
27032 reject and subtract it from overall when only addresses will be
27035 2017-03-06 Julia Koval <julia.koval@intel.com>
27038 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
27039 incoming stack boundary to 128 for 64-bit targets.
27041 2017-03-06 Richard Biener <rguenther@suse.de>
27043 PR tree-optimization/79894
27044 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
27045 to NULL after folding it.
27047 2017-03-06 Richard Biener <rguenther@suse.de>
27049 PR tree-optimization/79824
27050 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
27051 check disabling peeling for gaps.
27053 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
27055 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
27056 attributes): Document gettimeofday.
27058 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
27060 * config/s390/s390.c (s390_option_override_internal): Set
27061 PARAM_MIN_VECT_LOOP_BOUND
27063 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
27065 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
27066 * config/s390/s390.md: Likewise.
27068 2017-03-06 Jakub Jelinek <jakub@redhat.com>
27071 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
27072 (<avx2_avx512>_perm<mode>): Rename to ...
27073 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
27075 (<avx512>_perm<mode>_mask): Rename to ...
27076 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
27078 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
27079 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
27080 instead of VI8F_256_512.
27081 (avx512f_perm<mode>): New define_expand.
27082 (avx512f_perm<mode>_mask): Likewise.
27083 (avx512f_perm<mode>_1<mask_name>): New define_insn.
27084 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
27086 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
27088 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
27089 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
27091 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
27093 2017-03-06 Martin Liska <mliska@suse.cz>
27096 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
27097 when having a SSA NAME w/o VAR_DECL assigned to it.
27099 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
27101 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
27102 msa_dpsub_<su>_d): Fix MODE for vec_select.
27104 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
27106 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
27108 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
27110 2017-03-06 Richard Biener <rguenther@suse.de>
27112 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
27113 * plugin.c (register_plugin_info): Likewise.
27114 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
27116 2017-03-05 Jakub Jelinek <jakub@redhat.com>
27118 * config/i386/sse.md (sse_storehps, sse_storelps,
27119 avx_<castmode><avxsizesuffix>_<castmode>,
27120 avx512f_<castmode><avxsizesuffix>_<castmode>,
27121 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
27122 in condition that at least one operand is not a MEM.
27124 2017-03-03 Jakub Jelinek <jakub@redhat.com>
27126 PR middle-end/79805
27127 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
27128 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
27130 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
27131 gimple_call_nothrow_p flag based on whether original builtin can throw.
27132 If it can, emit following stmts on the fallthrough edge.
27133 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
27134 don't create new bb if inserting just debug stmts on the edge, try to
27135 insert them on the fallthru bb or just reset debug stmts.
27137 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
27140 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
27141 restore recog_data (including the operand rtxes inside it) around
27142 the call to get_insn_template.
27144 2017-03-03 Martin Sebor <msebor@redhat.com>
27146 PR tree-optimization/79699
27147 * context.c (context::~context): Free MPFR caches to avoid
27148 a memory leak on program exit.
27150 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27152 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
27153 Use wide_int::ulow () instead of .elt (0).
27155 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
27157 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
27158 (*pushxf): Limit oF constraint to 32bit targets and add oC
27159 constraint for 64bit targets.
27160 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
27161 (*pushdf): Change rmF constraint to rmC.
27163 2017-03-03 Martin Liska <mliska@suse.cz>
27165 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
27166 Remove unused variable.
27168 2017-03-03 Jakub Jelinek <jakub@redhat.com>
27171 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
27172 is a memory operand, increase num_memory.
27173 (ix86_expand_args_builtin): Likewise.
27175 2017-03-03 Jan Hubicka <jh@suse.cz>
27178 * ipa-devirt.c (maybe_record_node): Properly handle
27179 __cxa_pure_virtual visibility.
27181 2017-03-03 Martin Liska <mliska@suse.cz>
27183 PR tree-optimization/79803
27184 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
27186 (pass_loop_prefetch::execute): Disabled optimization if an
27187 assumption about L1 cache size is not met.
27189 2017-03-03 Martin Liska <mliska@suse.cz>
27191 PR rtl-optimization/79574
27192 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
27193 (hash_scan_set): Likewise.
27194 (dump_hash_table): Likewise.
27195 (hoist_code): Likewise.
27197 2017-03-03 Richard Biener <rguenther@suse.de>
27199 * fixed-value.c (fixed_from_string): Restore use of elt (1)
27200 in place of uhigh ().
27201 (fixed_convert_from_real): Likewise.
27203 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
27206 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
27208 2017-03-03 Richard Biener <rguenther@suse.de>
27210 PR middle-end/79818
27211 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
27212 TYPE_OVERFLOW_UNDEFINED check.
27214 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27216 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
27218 (vector_ae_<mode>_p): Likewise.
27219 (vector_nez_<mode>_p): Likewise.
27220 (vector_ne_v2di_p): Likewise.
27221 (vector_ae_v2di_p): Likewise.
27222 (vector_ne_<mode>_p): Likewise.
27223 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
27225 (vsx_tsqrt<mode>2_fe): Likewise.
27227 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
27230 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
27232 2017-03-02 Jakub Jelinek <jakub@redhat.com>
27234 PR rtl-optimization/79780
27235 * cprop.c (one_cprop_pass): When second and further conditional trap
27236 in a single basic block is turned into an unconditional trap, turn it
27237 into a deleted note to avoid RTL verification failures.
27239 2017-03-02 Richard Biener <rguenther@suse.de>
27241 * fold-const.c (const_binop): Use ulow () instead of elt (0).
27243 2017-03-02 Richard Biener <rguenther@suse.de>
27245 PR tree-optimization/79345
27247 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
27248 param and abort the walk, returning -1 if it is hit.
27249 (walk_aliased_vdefs): Take a limit param and pass it on.
27250 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
27251 defaulting to 0 and return a signed int.
27252 * tree-ssa-uninit.c (struct check_defs_data): New struct.
27253 (check_defs): New helper.
27254 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
27255 about uninitialized memory.
27256 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
27257 bogus uninitialized warning.
27258 (fixed_convert_from_real): Likewise.
27260 2017-03-02 Bin Cheng <bin.cheng@arm.com>
27262 PR tree-optimization/66768
27263 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
27264 iv_use if base object can't be determined.
27266 2017-03-02 Jakub Jelinek <jakub@redhat.com>
27268 PR tree-optimization/79345
27269 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
27270 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
27271 (get_pattern_stats): Initialize it.
27272 * genemit.c (gen_expand): Verify match_scratch numbers come after
27273 match_operand/match_dup numbers.
27274 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
27275 match_scratch numbers.
27276 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
27278 * config/s390/s390.md (trunctdsd2): Likewise.
27280 2017-03-02 Richard Biener <rguenther@suse.de>
27282 * wide-int.h (wide_int_storage::operator=): Implement in terms
27285 2017-03-02 Richard Biener <rguenther@suse.de>
27287 PR tree-optimization/79777
27288 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
27289 the to insert expression to sth existing.
27291 2017-03-01 Martin Sebor <msebor@redhat.com>
27293 PR middle-end/79692
27294 * gimple-ssa-sprintf.c
27295 (directive::known_width_and_precision): New function.
27296 (format_integer): Use it.
27297 (get_mpfr_format_length): Consider the full range of precision
27298 when computing %g output with the # flag. Set the likely byte
27299 count to 3 rather than 1 when precision is indeterminate.
27300 (format_floating): Correct the lower bound of precision.
27302 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27304 * doc/invoke.texi: Document default code model for 64-bit Linux.
27306 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27309 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
27310 udiv rather than div since input pattern is unsigned.
27312 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
27314 * config/i386/i386.c (print_reg): Warn for values of
27315 unsupported size in integer register.
27317 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
27320 * config/rs6000/predicates.md (current_file_function_operand): Do
27321 not allow self calls to be local if the function is replaceable.
27323 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
27326 * config/rs6000/altivec.h (vec_ctz and others): Change the
27327 preprocessor macro that controls conditional compilation from
27328 _ARCH_PWR9 to __POWER9_VECTOR__.
27329 (vec_all_ne): Change parameterization of __altivec_scalar_pred
27330 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
27331 control (instead of _ARCH_PWR9 control) so that template
27332 definition uses power9-specific function.
27333 (vec_any_eq): Likewise.
27334 (vec_all_ne): Change macro definition to use a power9-specific
27335 expansion under #ifdef __POWER9_VECTOR__ control (instead of
27336 _ARCH_PWR9 control).
27337 (vec_any_eq) Likewise.
27338 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
27339 expansion for CMPNEF to remove support for xvcmpnesp instruction.
27340 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
27341 support for xvcmpnedp instruction.
27342 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
27343 macro expansion so that Power9 implementation of vec_all_ne does
27344 not use the AltiVec predicate framework.
27345 (VCMPNEH_P): Likewise.
27346 (VCMPNEW_P): Likewise.
27347 (VCMPNED_P): Likewise.
27348 (VCMPNEFP_P): Likewise.
27349 (VCMPNEDP_P): Likewise.
27350 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
27351 implementation of vec_any_eq to not use AltiVec predicate
27353 (VCMPAEH_P): Likewise.
27354 (VCMPAEW_P): Likewise.
27355 (VCMPAED_P): Likewise.
27356 (VCMPAEFP_P): Likewise.
27357 (VCMPAEDP_P): Likewise.
27358 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
27359 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
27360 not use the AltiVec predicate framework.
27361 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
27362 of vec_any_eq to not use AltiVec predicate framework.
27363 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
27364 support for predefined __POWER9_VECTOR__ macro to indicate that
27365 Power9 instruction selection is enabled.
27366 (altivec_overloaded_builtins): Remove extraneous
27367 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
27368 function argument types RS6000_BTI_bool_V16QI and
27369 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
27370 entry for overloaded function argument types RS6000_BTI_bool_V4SI
27371 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
27372 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
27373 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
27374 Power9 for implementations of vec_cmpne. Change the signature for
27375 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
27376 (representing vec_all_ne) to remove the previously described first
27377 argument of type RS6000_BTI_INTSI, as this was an artifact of
27378 reliance on the AltiVec predicate framework, which is no longer
27379 used in the implementation of these functions. Add
27380 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
27381 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
27382 since, unlike the AltiVec predicate framework implementation, we
27383 do not share function descriptors between vec_alle and vec_anyeq.
27384 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
27385 set of modes that receive special treatment even when
27386 TARGET_P9_VECTOR is true. The special treatment emits code that
27387 does not depend on Power9 instructions.
27388 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
27389 define_expand to not rely on AltiVec predicate framework.
27390 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
27392 (vector_ne_v2di_p): Change this define_expand to not rely on
27393 AltiVec predicate framework.
27394 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
27396 (vector_ne_<mode>_p): Change this define_expand to not rely on
27397 AltiVec predicate framework.
27398 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
27400 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
27401 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
27402 define_insn pattern.
27403 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
27404 define_insn pattern because the xvcmpne<VSs>. instruction is not
27406 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
27407 instruction is not supported.
27409 2017-03-01 Jakub Jelinek <jakub@redhat.com>
27411 * config/nvptx/nvptx.c: Include intl.h.
27413 2017-03-01 Martin Jambor <mjambor@suse.cz>
27416 * ipa-prop.h (ipa_bits): Removed field known.
27417 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
27418 to pointers. Adjusted their comments to warn about their sharing.
27419 (ipcp_transformation_summary): Change bits to a vector of pointers.
27420 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
27421 (ipa_get_ipa_bits_for_value): Declare.
27422 * tree-vrp.h (value_range): Mark as GTY((for_user)).
27423 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
27424 (ipa_bits_hash_table): Likewise.
27425 (ipa_vr_ggc_hash_traits): Likewise.
27426 (ipa_vr_hash_table): Likewise.
27427 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
27428 being pointers and vr_known being removed.
27429 (ipa_set_jf_unknown): Likewise.
27430 (ipa_get_ipa_bits_for_value): New function.
27431 (ipa_set_jfunc_bits): Likewise.
27432 (ipa_get_value_range): New overloaded functions.
27433 (ipa_set_jfunc_vr): Likewise.
27434 (ipa_compute_jump_functions_for_edge): Use the above functions to
27435 construct bits and vr parts of jump functions.
27436 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
27437 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
27439 (ipcp_grow_transformations_if_necessary): Also allocate
27440 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
27442 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
27443 them. Fix too long lines.
27444 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
27445 vr_known being removed.
27446 (ipa_read_jump_function): Use new setter functions to construct bits
27447 and vr parts of jump functions or set them to NULL.
27448 (write_ipcp_transformation_info): Adjust for bits being pointers.
27449 (read_ipcp_transformation_info): Likewise.
27450 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
27452 Include gt-ipa-prop.h.
27453 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
27455 (ipcp_store_bits_results): Likewise.
27456 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
27457 Do not write to existing jump functions but use a temporary instead.
27459 2017-03-01 Jakub Jelinek <jakub@redhat.com>
27462 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
27463 attempt to use its first operand as BIT_FIELD_REF base.
27465 2017-03-01 Richard Biener <rguenther@suse.de>
27467 PR middle-end/79721
27468 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
27469 interpolating formula in wrapping arithmetic.
27470 (chrec_apply): Convert chrec_evaluate return value to wanted type.
27472 2017-03-01 Jakub Jelinek <jakub@redhat.com>
27474 PR tree-optimization/79734
27475 * tree-vect-generic.c (expand_vector_condition): Optimize
27476 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
27477 Handle VEC_COND_EXPR where comparison has different inner width from
27478 type's inner width.
27480 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
27482 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
27483 markup, and similar issues. Remove @opindex entries for things
27484 that aren't options. Add missing -mmpy-option entries.
27486 2017-02-28 Jakub Jelinek <jakub@redhat.com>
27488 PR tree-optimization/79737
27489 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
27490 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
27491 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
27492 instead of byte_size. Formatting fix.
27493 (shift_bytes_in_array_right): Formatting fix.
27495 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
27498 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
27499 condition on optimize for the leaf function test.
27501 2017-02-28 Martin Liska <mliska@suse.cz>
27504 * read-rtl-function.c (function_reader::handle_unknown_directive):
27505 Bail out when one uses -flto.
27507 2017-02-28 Martin Liska <mliska@suse.cz>
27509 * common.opt: Replace space with tabular for options of <number>
27511 * config/i386/i386.opt: Show <number> value for
27512 -mlarge-data-threshold.
27513 * opts.c (print_filtered_help): Do not display number in hexadecimal
27516 2017-02-28 Martin Liska <mliska@suse.cz>
27518 * common.opt: Fix --help=option -Q for options which are of
27521 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
27523 * config/i386/i386.c (print_reg): Error out for values
27524 of 8-bit size in invalid integer register.
27526 2017-02-28 Martin Sebor <msebor@redhat.com>
27528 PR tree-optimization/79691
27529 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
27531 2017-02-28 Jakub Jelinek <jakub@redhat.com>
27534 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
27535 gcc_unreachable with output_operand_lossage.
27537 2017-02-28 Richard Biener <rguenther@suse.de>
27539 PR tree-optimization/79740
27540 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
27542 (visit_nary_op): Insert the nary into the hashtable if we
27543 pattern-matched sth.
27544 * tree-ssa-pre.c (eliminate_insert): Robustify.
27546 2017-02-28 Richard Biener <rguenther@suse.de>
27548 PR middle-end/79731
27549 * fold-const.c (decode_field_reference): Reject out-of-bound
27552 2017-02-28 Jakub Jelinek <jakub@redhat.com>
27554 * config/i386/i386.c: Include intl.h.
27555 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
27556 instead of just cond ? "..." : "...".
27557 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
27558 * coverage.c (read_counts_file): Likewise.
27559 * omp-offload.c: Include intl.h.
27560 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
27561 of just cond ? "..." : "...".
27562 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
27563 of just cond ? "..." : "...".
27565 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
27568 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
27570 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
27571 'tune for' CPU name.
27572 * config/arm/arm-cpu-data.h: Regenerated.
27574 2017-02-28 Richard Biener <rguenther@suse.de>
27576 PR tree-optimization/79732
27577 * tree-inline.c (expand_call_inline): Do not shadow var.
27579 2017-02-28 Richard Biener <rguenther@suse.de>
27581 PR tree-optimization/79723
27582 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
27583 address-space properly.
27585 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
27587 * doc/optinfo.texi (Optimization groups): Fix option used for
27589 * doc/invoke.texi (-fopt-info): Document "omp".
27590 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
27591 (OPTGROUP_ALL): Add OPTGROUP_OMP.
27592 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
27593 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
27594 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
27596 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
27598 * dumpfile.c (optgroup_options): Instead of "openmp", associate
27599 OPTGROUP_OMP with "omp".
27601 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
27604 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
27605 for arithmetic shift of unsigned V2DI.
27607 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
27609 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
27610 arc/linux.h headers.
27611 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
27612 (LINK_SPEC): Likewise.
27613 (ARC_TLS_EXTRA_START_SPEC): Likewise.
27614 (EXTRA_SPECS): Likewise.
27615 (STARTFILE_SPEC): Likewise.
27616 (ENDFILE_SPEC): Likewise.
27617 (LIB_SPEC): Likewise.
27618 (TARGET_SDATA_DEFAULT): Likewise.
27619 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
27620 (MULTILIB_DEFAULTS): Likewise.
27621 (DWARF2_UNWIND_INFO): Likewise.
27622 * config/arc/big.h: New file.
27623 * config/arc/elf.h: Likewise.
27624 * config/arc/linux.h: Likewise.
27625 * config/arc/t-uClibc: Remove.
27627 2017-02-27 Bin Cheng <bin.cheng@arm.com>
27629 PR tree-optimization/77536
27630 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
27631 (tree_transform_and_unroll_loop): Use above function to compute the
27632 estimated niter of unrolled loop and use it when scaling profile.
27633 Also use count info rather than frequency if it's non-zero.
27634 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
27635 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
27636 (vect_transform_loop): Call above function.
27638 2017-02-27 Richard Biener <rguenther@suse.de>
27640 PR tree-optimization/45397
27641 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
27642 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
27643 (visit_nary_op): Add pattern matching for CSEing sign-changed
27644 or truncated operations with wider ones.
27646 2017-02-27 Richard Biener <rguenther@suse.de>
27648 PR tree-optimization/79690
27649 * tree-vect-stmts.c (vectorizable_store): Use vector type
27650 built from the DR with address-space.
27652 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
27654 * doc/invoke.texi (Optimize Options): Refine the description
27655 of asan-use-after-return.
27657 2017-02-25 Alan Modra <amodra@gmail.com>
27659 PR rtl-optimization/79584
27660 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
27661 base, not ad->base_term, the reg within base. Remove assertion
27662 that ad->base == ad->base_term. Replace gen_int_mode using
27663 bogus mode with const0_rtx.
27665 2017-02-25 Jakub Jelinek <jakub@redhat.com>
27667 PR middle-end/79396
27668 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
27669 FMA_EXPR like tcc_binary or tcc_unary.
27671 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
27674 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
27676 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
27677 (output_loc_operands): Handle DW_OP_call_ref and
27678 DW_OP_GNU_variable_value.
27679 (struct variable_value_struct): New type.
27680 (struct variable_value_hasher): Likewise.
27681 (variable_value_hash): New variable.
27682 (string_types): Remove.
27683 (copy_loc_descr): New function.
27684 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
27685 (prepend_loc_descr_to_each): New function.
27686 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
27687 instead of add_loc_descr_to_each if the first argument is single
27688 location list and the second has multiple.
27689 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
27690 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
27691 when looking for variable value which doesn't have other location info.
27692 (loc_list_from_tree): Formatting fix.
27693 (gen_array_type_die): Simplify DW_AT_string_length handling.
27694 (adjust_string_types): Remove.
27695 (gen_subprogram_die): Don't call adjust_string_types nor test/set
27696 string_types. Call resolve_variable_values.
27697 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
27698 (resolve_addr_in_expr): Likewise. Add A argument.
27699 (copy_deref_exprloc): Remove deref argument. Adjust for the
27700 original expression being DW_OP_GNU_variable_value with optionally
27701 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
27702 optionally after it.
27703 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
27704 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
27705 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
27706 (variable_value_hasher::hash, variable_value_hasher::equal): New
27708 (resolve_variable_value_in_expr, resolve_variable_value,
27709 resolve_variable_values, note_variable_value_in_expr,
27710 note_variable_value): New functions.
27711 (dwarf2out_early_finish): Call note_variable_value on all toplevel
27714 2017-02-24 Jakub Jelinek <jakub@redhat.com>
27717 * opts.h (handle_generated_option): Add GENERATED_P argument.
27718 * opts-common.c (handle_option): Adjust function comment.
27719 (handle_generated_option): Add GENERATED_P argument, pass it to
27721 (control_warning_option): Pass false to handle_generated_option
27723 * opts.c (maybe_default_option): Pass true to handle_generated_option
27725 * optc-gen.awk: Likewise.
27727 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27729 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
27730 a REG, look at the REG it is a SUBREG of.
27731 (splitter for cmpeqsi_t): Ditto.
27733 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27735 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
27736 the special USEs with the pattern of the insn, not the insn itself.
27738 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
27741 * doc/invoke.texi: Document -mload-store-pairs.
27743 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27744 Sandra Loosemore <sandra@codesourcery.com>
27746 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
27747 argument isn't a CONST_INT.
27748 (nios2_alternate_compare_const): Assert op is a CONST_INT.
27749 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
27750 (nios2_validate_compare): Bypass alternate compare logic if *op2
27751 is not a CONST_INT.
27752 (ldstwm_operation_p): Return false if first_base is not a REG or
27753 if first_offset is not a CONST_INT.
27755 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27757 * config/cris/cris.md: Use correct operand in a define_peephole2.
27759 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27761 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
27763 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27765 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
27766 this_insn if it is an INSN or JUMP_INSN.
27767 (force_offsettable): Look at base, not at addr.
27768 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
27769 on things that aren't necessarily CONST_INTs.
27771 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
27773 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
27774 -mfpmath=sse is the default also for x86-32 targets with SSE2
27775 instruction set when @option{-ffast-math} is enabled
27777 2017-02-24 Jeff Law <law@redhat.com>
27779 PR rtl-optimizatoin/79286
27780 * ira.c (update_equiv_regs): Drop may_trap_p exception to
27783 2017-02-24 Richard Biener <rguenther@suse.de>
27785 PR tree-optimization/79389
27786 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
27789 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
27791 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
27792 function comment to reflect reality.
27793 (loop_exits_before_overflow): Fix typo in function description.
27795 2017-02-24 Richard Biener <rguenther@suse.de>
27797 PR tree-optimization/79389
27798 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
27799 properly that a threading opportunity exists. Detect conditional
27800 copy/constant propagation opportunities.
27802 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
27804 * config/visium/visium.md (type): Add trap.
27805 (b): New mode attribute.
27806 (*btst): Rename into...
27807 (*btst<mode>): ...this and adjust.
27808 (*cbranchsi4_btst_insn): Rename into...
27809 (*cbranch<mode>4_btst_insn): ...this and adjust.
27810 (trap): New define_insn.
27812 2017-02-23 Jakub Jelinek <jakub@redhat.com>
27814 PR tree-optimization/79389
27815 * ifcvt.c (struct noce_if_info): Add rev_cond field.
27816 (noce_reversed_cond_code): New function.
27817 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
27818 reversed_comparison_code. Formatting fix.
27819 (noce_try_store_flag): Test rev_cond != NULL in addition to
27820 reversed_comparison_code.
27821 (noce_try_store_flag_constants): Likewise.
27822 (noce_try_store_flag_mask): Likewise.
27823 (noce_try_addcc): Use rev_cond if non-NULL instead of
27824 reversed_comparison_code.
27825 (noce_try_cmove_arith): Likewise. Formatting fixes.
27826 (noce_try_minmax, noce_try_abs): Clear rev_cond.
27827 (noce_find_if_block): Initialize rev_cond.
27828 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
27829 instead of false as last argument never attempt to reverse it
27832 2017-02-23 Bin Cheng <bin.cheng@arm.com>
27834 PR tree-optimization/79663
27835 * tree-predcom.c (combine_chains): Process refs in reverse order
27836 only for ZERO length chains, and add explaining comment.
27838 2017-02-23 Jeff Law <law@redhat.com>
27840 PR tree-optimization/79578
27841 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
27842 in call to operand_equal_p.
27844 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
27847 * config/i386/cpuid.h: Fix another undefined behavior.
27849 2017-02-23 Richard Biener <rguenther@suse.de>
27851 PR tree-optimization/79683
27852 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
27853 vector types for data-refs.
27855 2017-02-23 Martin Liska <mliska@suse.cz>
27857 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
27859 2017-02-23 Jakub Jelinek <jakub@redhat.com>
27861 PR middle-end/79665
27862 * internal-fn.c (get_range_pos_neg): Moved to ...
27863 * tree.c (get_range_pos_neg): ... here. No longer static.
27864 * tree.h (get_range_pos_neg): New prototype.
27865 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
27866 are known to be in between 0 and signed maximum inclusive, try to
27867 expand both unsigned and signed divmod and use the cheaper one from
27870 2017-02-22 Jeff Law <law@redhat.com>
27872 PR tree-optimization/79578
27873 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
27874 to compare base operands.
27876 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
27879 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
27880 gpc_reg_operand instead of fpr_reg_operand.
27882 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
27884 * config/mips/mips.c (mips_return_in_memory): Force FP
27885 vector types to be returned in memory for o32 ABI.
27887 2017-02-22 Jakub Jelinek <jakub@redhat.com>
27889 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
27890 instead of DW_TAG_member for static data member declarations and don't
27891 set no_linkage_name for static inline data members.
27892 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
27895 2017-02-22 Martin Liska <mliska@suse.cz>
27897 * doc/invoke.texi: Replace inequality signs with square brackets
27900 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27902 PR tree-optimization/68644
27903 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
27905 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
27908 * lra-constraints.c (simplify_operand_subreg): Handle
27909 WORD_REGISTER_OPERATIONS targets.
27911 2017-02-22 Jakub Jelinek <jakub@redhat.com>
27914 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
27915 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
27916 elimination by swapping fld*.
27918 2017-02-22 Richard Biener <rguenther@suse.de>
27920 PR tree-optimization/79673
27921 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
27922 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
27923 irrelevant address-space qualifiers and avoiding a
27924 ADDR_SPACE_CONVERT_EXPR from fold_convert.
27926 2017-02-22 Richard Biener <rguenther@suse.de>
27928 PR tree-optimization/79666
27929 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
27930 to not symbolically negate if that may introduce undefined
27933 2017-02-22 Martin Liska <mliska@suse.cz>
27936 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
27937 * data-streamer-out.c (streamer_write_gcov_count_stream):
27939 * value-prof.c (stream_out_histogram_value): Make assert more
27940 precise based on type of counter.
27942 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
27945 * config/i386/i386.md (standard_x87sse_constant_load splitter):
27946 Use nonimmediate_operand instead of memory_operand for operand 1.
27947 (float-extend standard_x87sse_constant_load splitter): Ditto.
27949 2017-02-21 Jeff Law <law@redhat.com>
27951 PR tree-optimization/79621
27952 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
27953 blocks with edges to themselves.
27955 2017-02-21 Jakub Jelinek <jakub@redhat.com>
27958 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
27959 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
27960 Use gimple_call_builtin_p.
27963 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
27964 on temporarily removed DEBUG_INSNs.
27966 PR tree-optimization/79649
27967 * tree-loop-distribution.c (classify_partition): Give up on
27968 non-generic address space loads/stores.
27970 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
27972 * doc/loop.texi (Loop manipulation): Remove nonexistent
27973 tree_ssa_loop_version from the documentation.
27974 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
27976 2017-02-21 Jakub Jelinek <jakub@redhat.com>
27979 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
27980 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
27981 * config/rs6000/rs6000.c: Include except.h.
27982 (rs6000_expand_split_stack_prologue): Call
27983 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
27985 2017-02-21 Martin Jambor <mjambor@suse.cz>
27988 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
27989 have been analyzed.
27991 2017-02-21 Martin Jambor <mjambor@suse.cz>
27993 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
27994 for backward compatibility only.
27995 * doc/invoke.texi (Option Summary): Remove all references to
27996 -fipa-cp-alignment.
27998 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
28002 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28004 * lra-constraints.c (curr_insn_transform): Handle
28005 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
28007 2017-02-21 Martin Liska <mliska@suse.cz>
28009 * config/i386/i386.opt: Replace -masm-dialect with -masm.
28011 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
28013 PR translation/79638
28014 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
28016 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
28019 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
28020 (arm_function_ok_for_sibcall): Return false for an indirect call by
28021 descriptor if all the argument registers are used.
28022 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
28023 alignment of the function.
28025 2017-02-21 Jakub Jelinek <jakub@redhat.com>
28027 PR tree-optimization/61441
28028 * simplify-rtx.c (simplify_const_unary_operation): For
28029 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
28030 the sNaN unmodified.
28032 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
28034 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
28035 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
28036 instead of SYSTEM_HEADER_DIR.
28038 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
28039 Martin Liška <mliska@suse.cz>
28041 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
28042 Fix typos and grammar, use active voice, and clarify.
28044 2017-02-20 Marek Polacek <polacek@redhat.com>
28046 PR middle-end/79537
28047 * gimplify.c (gimplify_expr): Handle unused *&&L;.
28050 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
28052 2017-02-20 Jakub Jelinek <jakub@redhat.com>
28055 * config/i386/i386.c (ix86_expand_builtin): Handle
28056 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
28057 ix86_builtins_isa[fcode].isa as a requirement of those
28058 flags and any other flag in the bitmask.
28059 (ix86_init_mmx_sse_builtins): Use 0 instead of
28060 ~OPTION_MASK_ISA_64BIT as mask.
28061 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
28062 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
28063 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
28064 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
28066 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28069 * lra-constraints.c (split_reg): Check requested split mode
28070 is supported by the register.
28072 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28074 * lra-constraints.c (simplify_operand_subreg): Remove early
28077 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28080 * lra-constraints.c (curr_insn_transform): Tighten condition
28081 for converting SUBREG reloads from OP_OUT to OP_INOUT.
28083 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28086 * lra-constraints.c (curr_insn_transform): Handle
28087 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
28089 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
28092 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
28094 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
28096 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
28099 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
28102 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28104 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
28107 2017-02-18 Jakub Jelinek <jakub@redhat.com>
28109 * final.c (last_columnnum, override_columnnum): New variables.
28110 (final_start_function): Set last_columnnum, pass it to begin_prologue
28111 hook and pass 0 to dwarf2out_begin_prologue.
28112 (final_scan_insn): Update override_columnnum. Pass last_columnnum
28113 to source_line debug hook.
28114 (notice_source_line): Compute last_columnnum and for debug_column_info
28115 return true on column changes.
28116 * debug.h (struct gcc_debug_hooks): Add column argument to
28117 source_line and begin_prologue hooks.
28118 (debug_nothing_int_charstar_int_bool): Remove prototype.
28119 (debug_nothing_int_int_charstar,
28120 debug_nothing_int_int_charstar_int_bool): New prototypes.
28121 (dwarf2out_begin_prologue): Add column argument.
28122 * debug.c (do_nothing_debug_hooks): Adjust source_line and
28123 begin_prologue hooks.
28124 (debug_nothing_int_charstar_int_bool): Remove.
28125 (debug_nothing_int_int_charstar,
28126 debug_nothing_int_int_charstar_int_bool): New functions.
28127 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
28128 through to dwarf2out_source_line.
28129 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
28130 (dwarf2out_source_line): Add column argument, emit it if requested.
28131 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
28133 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
28134 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
28135 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
28136 through to dwarf2out_begin_prologue.
28137 (vmsdbgout_source_line): Add column argument, pass it through to
28138 dwarf2out_source_line.
28139 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
28140 dbxout_source_line caller.
28141 (dbxout_source_line): Add column argument.
28143 * common.opt (gno-column-info, gcolumn-info): New options.
28144 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
28145 (check_die): Also test for multiple DW_AT_decl_column attributes.
28146 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
28147 DW_AT_decl_column if requested.
28148 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
28150 (gen_variable_die): Likewise.
28151 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
28152 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
28155 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
28156 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
28157 (ix86_handle_option): Handle OPT_m3dnowa.
28158 * doc/invoke.texi (-m3dnowa): Document.
28159 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
28160 -m3dnowa instead of -m3dnow -march=athlon.
28163 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
28164 instead of gcc_assert for K, r and R code checks. Formatting fixes.
28166 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28169 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
28170 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
28171 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
28172 generator for vsx_xxpermdi_<mode>_be.
28173 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
28174 force big-endian semantics.
28175 (vsx_xxpermdi_<mode>_be): New define_expand with same
28176 implementation as previous version of vsx_xxpermdi_<mode>.
28178 2017-02-17 Jakub Jelinek <jakub@redhat.com>
28180 PR tree-optimization/79327
28181 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
28182 variable, its initialization and use.
28184 2017-02-17 Julia Koval <julia.koval@intel.com>
28186 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
28187 (OPTION_MASK_ISA_PKU_UNSET): New.
28188 (ix86_handle_option): Handle -mrdpid.
28189 * config/i386/cpuid.h (bit_RDPID): New.
28190 * config/i386/driver-i386.c (host_detect_local_cpu):
28191 Detect RDPID feature.
28192 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
28193 * config/i386/i386-c.c (ix86_target_macros_internal):
28195 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
28196 (ix86_valid_target_attribute_inner_p): Add "rdpid".
28197 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
28198 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
28199 * config/i386/i386.md (define_insn "rdpid"): New.
28200 * config/i386/i386.opt Add -mrdpid.
28201 * config/i386/immintrin.h (_rdpid_u32): New.
28203 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
28205 PR rtl-optimization/79541
28206 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
28207 instead of transforming it into USE.
28209 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
28211 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
28212 If HONOR_SNANS (SFmode) force the input to a register.
28213 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
28214 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
28215 an frsp or similar insn.
28217 2017-02-17 Martin Liska <mliska@suse.cz>
28219 PR rtl-optimization/79577
28220 * params.def (selsched-max-sched-times): Increase minimum to 1.
28222 2017-02-17 Martin Liska <mliska@suse.cz>
28224 PR rtl-optimization/79574
28225 * gcse.c (want_to_gcse_p): Prevent integer overflow.
28227 2017-02-17 Martin Liska <mliska@suse.cz>
28229 PR tree-optimization/79529
28230 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
28231 ssa_defined_default_def_p to handle cases which are implicitly
28233 * tree-ssa.c (ssa_defined_default_def_p): New function.
28234 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
28235 which are implicitly defined.
28236 * tree-ssa.h (ssa_defined_default_def_p): Declare.
28238 2017-02-17 Richard Biener <rguenther@suse.de>
28240 PR middle-end/79576
28241 * params.def (max-ssa-name-query-depth): Limit to 10.
28243 2017-02-17 Richard Biener <rguenther@suse.de>
28245 PR tree-optimization/79552
28246 * tree-ssa-structalias.c (visit_loadstore): Properly verify
28249 2017-02-17 Richard Biener <rguenther@suse.de>
28252 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
28254 2017-02-17 Marek Polacek <polacek@redhat.com>
28256 PR middle-end/79536
28257 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
28258 (fold_negate_expr): New wrapper.
28260 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
28262 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
28263 Correct terminology and de-emphasize pre-standard behavior.
28265 2017-02-16 Alan Modra <amodra@gmail.com>
28267 PR rtl-optimization/79286
28268 * ira.c (def_dominates_uses): New function.
28269 (update_equiv_regs): Don't create an equivalence for insns that
28270 may trap where the register def does not dominate the use.
28272 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
28274 PR rtl-optimization/78127
28275 * lra.c (lra): Call lra_eliminate before finish the loop after
28278 2017-02-16 Richard Biener <rguenther@suse.de>
28280 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
28282 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
28283 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
28284 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
28285 (isl_val_int_from_wi): New function.
28286 (extract_affine_gmp): Rename to ...
28287 (extract_affine_wi): ... this, take a widest_int.
28288 (extract_affine_int): Just wrap extract_affine_wi.
28289 (add_param_constraints): Use isl_val_int_from_wi.
28290 (add_loop_constraints): Likewise, and extract_affine_wi.
28292 2017-02-15 Jeff Law <law@redhat.com>
28294 PR middle-end/79521
28295 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
28296 ira_init_register_move_cost_if_necessary.
28298 2017-02-15 Martin Sebor <msebor@redhat.com>
28300 PR middle-end/32003
28301 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
28302 removed in a prior commit.
28304 2017-02-15 Bin Cheng <bin.cheng@arm.com>
28306 PR tree-optimization/79347
28307 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
28308 counters during peeling.
28310 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
28312 * Makefile.in (site.exp): Remove "set ISLVER".
28314 2017-02-15 Jakub Jelinek <jakub@redhat.com>
28317 * real.c (real_from_integer): Call real_convert even for decimal.
28319 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
28322 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
28324 2017-02-14 Andrew Pinski <apinski@cavium.com>
28326 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
28327 cores and change the partno/implementer to be correct.
28328 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
28329 the 'B" as the implementer.
28330 * config/aarch64/aarch64-tune.md: Regenerate.
28332 2017-02-14 Carl Love <cel@us.ibm.com>
28334 * config/rs6000/rs6000.c: Add case statement entry to make the
28335 xvcvuxdsp built-in argument unsigned.
28336 * config/rs6000/vsx.md: Fix the source and return operand types so they
28337 match the instruction definitions from the ISA document. Fix typo
28338 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
28341 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
28344 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
28345 member early_clobber_alts.
28346 * lra-lives.c (reg_early_clobber_p): New.
28347 (process_bb_lives): Use it.
28348 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
28349 (debug_operand_data): Initialize early_clobber_alts.
28350 (setup_operand_alternative): Set up early_clobber_alts.
28351 (collect_non_operand_hard_regs): Ditto. Pass early clobber
28352 alternatives to new_insn_reg.
28353 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
28355 (lra_update_insn_regno_info): Pass the new arg.
28357 2017-02-14 Jakub Jelinek <jakub@redhat.com>
28359 PR middle-end/79505
28360 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
28361 (new_oacc_loop_raw): Don't clear already cleared fields.
28364 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
28365 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
28366 _mm512_prefetch_i64gather_ps): New inline functions and macros.
28368 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
28371 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
28373 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
28376 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
28377 the extra instruction to the right place to store 128-bit constant
28380 2017-02-14 Martin Sebor <msebor@redhat.com>
28382 PR middle-end/79448
28383 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
28384 warning for strings of unknown length.
28386 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
28388 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
28390 2017-02-14 Jeff Law <law@redhat.com>
28393 * ira-costs.c (scan_one_insn): Initialize register move costs
28394 for pseudos seen in USE/CLOBBER insns.
28396 PR tree-optimization/79095
28397 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
28398 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
28399 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
28400 if the operands are known to be not equal, then the resulting range
28402 (intersect_ranges): If the new range is ~[0,0] and the old range is
28403 wide, then prefer ~[0,0].
28404 * tree-vrp.c (overflow_comparison_p_1): New function.
28405 (overflow_comparison_p): New function.
28406 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
28407 if NAME is used in an overflow test.
28408 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
28409 overflow check that can be expressed as an equality test, then adjust
28410 ops to be that equality test.
28412 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28414 * config/s390/s390-builtin-types.def: Remove flags argument.
28415 * config/s390/s390.c (s390_init_builtins): Likewise.
28417 2017-02-14 Martin Liska <mliska@suse.cz>
28419 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
28420 vector. Fix trailing white spaces.
28422 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
28424 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
28427 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28429 PR rtl-optimization/68664
28430 * config/arm/arm.c (arm_sched_can_speculate_insn):
28431 New function. Declare prototype.
28432 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
28434 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28436 PR rtl-optimization/68664
28437 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
28439 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
28441 2017-02-14 Amit Pawar <amit.pawar@amd.com>
28443 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
28444 max skip bytes for function, loop and jump.
28446 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28448 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
28449 ABS_EXPR for gimple dump.
28451 2017-02-14 Jakub Jelinek <jakub@redhat.com>
28454 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
28456 PR tree-optimization/79408
28457 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
28458 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
28459 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
28460 also if rhs1 is INTEGER_CST.
28462 2017-02-14 Richard Biener <rguenther@suse.de>
28464 PR middle-end/79432
28465 * tree-into-ssa.c (insert_phi_nodes): When the function can
28466 have abnormal edges rewrite SSA names with broken use-def
28467 dominance out of SSA and register them for PHI insertion.
28469 2017-02-13 Martin Sebor <msebor@redhat.com>
28471 PR middle-end/79496
28472 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
28473 clearing info.nowrite flag when snprintf size argument is a range.
28475 2017-02-13 Jakub Jelinek <jakub@redhat.com>
28477 * cprop.c (cprop_jump): Add missing space in string literal.
28478 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
28479 (get_constraint_for_component_ref): Likewise.
28480 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
28481 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
28482 * lra-constraints.c (process_alt_operands): Likewise.
28483 * ipa-inline.c (inline_small_functions): Likewise.
28484 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
28485 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
28486 * trans-mem.c (diagnose_tm_1_op): Likewise.
28487 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
28488 (grid_parallel_clauses_gridifiable): Likewise.
28490 * config/nvptx/mkoffload.c (process): Add space in between
28493 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
28494 "MOD4_SSE_REGS" and "ALL_REGS".
28496 * spellcheck.c (test_data): Add , in between "foo" and "food".
28498 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28501 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
28502 boundary crossing check and subsequent code generation agree.
28504 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28506 * config/aarch64/aarch64.c (has_memory_op): Delete.
28507 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
28510 2017-02-13 Jakub Jelinek <jakub@redhat.com>
28512 PR rtl-optimization/79388
28513 PR rtl-optimization/79450
28514 * combine.c (distribute_notes): When removing TEM_INSN for which
28515 corresponding dest has last value recorded, invalidate that last
28518 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28520 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
28521 of explicit '@'. Add missing assembly comment marker on branch costs
28524 2017-02-13 Nathan Sidwell <nathan@acm.org>
28526 * gengtype-lex.l (<in_struct>): Add '/'.
28528 2017-02-13 Martin Liska <mliska@suse.cz>
28531 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
28533 2017-02-13 Richard Biener <rguenther@suse.de>
28535 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
28537 * configure: Re-generate.
28538 * config.in: Likewise.
28539 * graphite-dependences.c: Simplify as if
28540 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
28541 * graphite-isl-ast-to-gimple.c: Likewise.
28542 * graphite-optimize-isl.c: Likewise.
28543 * graphite-poly.c: Likewise.
28544 * graphite-sese-to-poly.c: Likewise.
28545 * graphite.h: Likewise.
28546 * toplev.c: Include isl/version.h and use isl_version () for
28547 printing the ISL version.
28548 * doc/install.texi: Update ISL requirement.
28550 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
28552 * doc/standards.texi (Standards): Update reference to
28555 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
28557 * doc/extend.texi (Named Address Spaces): sourceware.org now
28559 * doc/install.texi (Binaries): Ditto.
28562 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
28564 * doc/cpp.texi: Replace "stringify"/"stringification" with C
28565 standard terminology "stringize"/"stringizing" throughout.
28566 * doc/cppinternals.texi: Likewise.
28568 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
28570 * doc/extend.texi: Fix some spelling mistakes and typos.
28571 * doc/invoke.texi: Likewise.
28573 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
28576 * params.def (inline-min-speedup) Change from 10 to 8.
28578 2017-02-11 Jakub Jelinek <jakub@redhat.com>
28580 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
28583 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
28586 * ipa-inline-analysis.c (get_minimal_bb): New function.
28587 (record_modified): Use it.
28588 (remap_edge_change_prob): Handle also ancestor functions.
28590 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
28592 * doc/contrib.texi (Contributors): Remove broken link into
28593 the Mauve CVS repository.
28595 2017-02-11 Jakub Jelinek <jakub@redhat.com>
28597 PR middle-end/79454
28598 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
28599 result computation whenever lhs doesn't have vector mode, not
28600 just when it has BLKmode.
28602 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
28604 * doc/makefile.texi (profiledbootstrap): Refer to the
28605 installation instructions only in textual form.
28607 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28610 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
28612 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
28614 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
28615 (Specific): Update mingw-w64 reference.
28617 (Specific): Remove broken link to Renesas RX processor.
28619 2017-02-10 Richard Biener <rguenther@suse.de>
28621 * toplev.c (process_options): Do not mention obsolete graphite
28622 options when printing sorry message about missing graphite support.
28623 Mention -floop-nest-optimize.
28625 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
28627 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
28628 (vtst_p16): Likewise.
28629 (vtstq_p8): Likewise.
28630 (vtstq_p16): Likewise.
28632 (vtstq_p64): Likewise.
28633 * config/arm/arm_neon.h (vgetq_lane_p64): New.
28634 (vset_lane_p64): New.
28635 (vsetq_lane_p64): New.
28637 2017-02-10 Jakub Jelinek <jakub@redhat.com>
28639 PR tree-optimization/79411
28640 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
28641 stmt operands are SSA_NAMEs used in abnormal phis.
28642 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
28645 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
28648 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
28651 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
28653 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
28655 2017-02-09 Jakub Jelinek <jakub@redhat.com>
28657 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
28661 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
28662 not arbitrary TREE_CONSTANT.
28665 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
28666 "omp declare target link" attribute unless is_global_var.
28667 * omp-offload.c (find_link_var_op): Likewise.
28669 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
28670 Chung-Lin Tang <cltang@codesourcery.com>
28672 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
28674 (gimplify_adjust_omp_clauses): Don't delete TILE.
28675 (gimplify_omp_for): Deal with TILE.
28676 * internal-fn.c (expand_GOACC_TILE): New function.
28677 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
28679 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
28680 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
28682 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
28683 avoid DIV for outermost collapse var.
28684 (expand_oacc_for): Insert tile element loop as needed. Adjust.
28685 Remove out of date comments, fix whitespace.
28686 * omp-general.c (omp_extract_for_data): Deal with tiling.
28687 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
28688 adjust OLF_DIM_BASE value.
28689 (struct omp_for_data): Add tiling field.
28690 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
28691 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
28692 for auto loops. Remove default auto determining, moved to
28693 oacc_loop_fixed_partitions.
28694 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
28695 stmts, add e_mask field.
28696 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
28697 (oacc_thread_numbers): Use oacc_dim_call.
28698 (oacc_xform_tile): New.
28699 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
28700 (finish_oacc_loop): Adjust for ifns vector.
28701 (oacc_loop_discover_walk): Append loop abstraction sites to list,
28702 add case for GOACC_TILE fns.
28703 (oacc_loop_xform_loop): Delete.
28704 (oacc_loop_process): Iterate over call list directly, and add
28705 handling for GOACC_TILE fns.
28706 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
28708 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
28709 vector partitioning to outer loops. Assign 2 partitions to loops
28710 when available. Add TILE handling.
28711 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
28712 (execite_oacc_device_lower): Process GOACC_TILE fns,
28713 ignore unknown specs.
28714 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
28715 * tree.c (omp_clause_num_ops): Adjust TILE ops.
28716 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
28718 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
28720 * configure.ac (ACX_BUGURL): Update.
28721 * configure: Regenerate.
28723 2017-02-09 Richard Biener <rguenther@suse.de>
28725 PR tree-optimization/69823
28726 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
28727 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
28729 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
28731 * config/arc/arc-c.def: Add __NPS400__ definition.
28732 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
28733 (TARGET_NPS400): Define.
28735 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
28737 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
28739 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
28740 pointer, arch_info.
28741 (arc_cpu_types): Fill the arch_info field with a pointer into the
28742 arc_arch_types table.
28743 (arc_selected_cpu): Declare.
28744 * config/arc/arc.c (arc_selected_cpu): Make global.
28745 (arc_selected_arch): Delete.
28746 (arc_base_cpu): Delete.
28747 (arc_override_options): Remove references to deleted variables,
28748 update access to arch information.
28749 (ARC_OPT): Update access to arch information.
28750 (ARC_OPTX): Likewise.
28751 * config/arc/arc.h (arc_base_cpu): Remove declaration.
28752 (TARGET_ARC600): Update access to arch information.
28753 (TARGET_ARC601): Likewise.
28754 (TARGET_ARC700): Likewise.
28755 (TARGET_EM): Likewise.
28756 (TARGET_HS): Likewise.
28757 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
28760 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
28763 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
28764 condition/operands for integer GE/LE/GEU/LEU operations.
28766 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
28768 PR translation/79397
28769 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
28772 2017-02-08 Martin Jambor <mjambor@suse.cz>
28775 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
28776 whether allocation happened.
28777 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
28778 nothing was allocated.
28780 2017-02-08 Jakub Jelinek <jakub@redhat.com>
28782 PR tree-optimization/79408
28783 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
28784 constant, but SSA_NAME with a known integer range, use the minimum
28785 of that range instead of op1 to determine if modulo can be replaced
28786 with its first operand.
28788 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28790 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
28792 2017-02-08 Richard Biener <rguenther@suse.de>
28794 PR tree-optimization/71824
28795 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
28796 Check all loops contained in the merged region.
28798 2017-02-07 Andrew Pinski <apinski@cavium.com>
28800 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
28802 2017-02-07 Andrew Pinski <apinski@cavium.com>
28804 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
28805 (thunderxt88): Likewise.
28806 (thunderxt81): Disable LSE and change v8.1 to v8.
28807 (thunderxt83): Likewise.
28809 2017-02-07 Jakub Jelinek <jakub@redhat.com>
28810 Richard Biener <rguenther@suse.de>
28812 PR middle-end/79399
28813 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
28814 type from int to size_t.
28815 * ira-costs.c (struct_costs_size): Change type from int to size_t.
28817 2017-02-07 Jakub Jelinek <jakub@redhat.com>
28819 PR rtl-optimization/79386
28820 * cprop.c (bypass_conditional_jumps): Initialize
28821 bypass_last_basic_block already before splitting bbs after
28822 unconditional traps...
28823 (bypass_conditional_jumps): ... rather than here.
28826 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
28827 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
28828 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
28829 fix -masm=intel patterns.
28831 2017-02-07 Richard Biener <rguenther@suse.de>
28833 PR tree-optimization/79256
28834 PR middle-end/79278
28835 * builtins.c (get_object_alignment_2): Use min_align_of_type
28836 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
28837 and ADJUST_FIELD_ALIGN.
28839 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
28841 * doc/tm.texi: Regenerate.
28842 * stor-layout.c (layout_decl): Adjust.
28843 (update_alignment_for_field): Likewise.
28844 (place_field): Likewise.
28845 (min_align_of_type): Likewise.
28846 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
28847 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
28848 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
28849 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
28850 * config/frv/frv.c (frv_adjust_field_align): Likewise.
28851 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
28852 * config/i386/i386.c (x86_field_alignment): Likewise.
28853 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
28854 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
28855 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
28856 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
28857 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
28858 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
28862 2017-01-30 Richard Biener <rguenther@suse.de>
28864 PR tree-optimization/79256
28865 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
28866 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
28869 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
28871 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
28872 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
28873 builtins to SImode and emit a zero-extend, if necessary.
28875 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
28877 * docs/invoke.texi (RISC-V Options): Alphabetize.
28879 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
28881 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
28884 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
28886 * config/riscv/riscv.c: New file.
28887 * common/config/riscv/riscv-common.c: Likewise.
28888 * config.gcc: Likewise.
28889 * config/riscv/constraints.md: Likewise.
28890 * config/riscv/elf.h: Likewise.
28891 * config/riscv/generic.md: Likewise.
28892 * config/riscv/linux.h: Likewise.
28893 * config/riscv/multilib-generator: Likewise.
28894 * config/riscv/peephole.md: Likewise.
28895 * config/riscv/pic.md: Likewise.
28896 * config/riscv/predicates.md: Likewise.
28897 * config/riscv/riscv-builtins.c: Likewise.
28898 * config/riscv/riscv-c.c: Likewise.
28899 * config/riscv/riscv-ftypes.def: Likewise.
28900 * config/riscv/riscv-modes.def: Likewise.
28901 * config/riscv/riscv-opts.h: Likewise.
28902 * config/riscv/riscv-protos.h: Likewise.
28903 * config/riscv/riscv.h: Likewise.
28904 * config/riscv/riscv.md: Likewise.
28905 * config/riscv/riscv.opt: Likewise.
28906 * config/riscv/sync.md: Likewise.
28907 * config/riscv/t-elf-multilib: Likewise.
28908 * config/riscv/t-linux: Likewise.
28909 * config/riscv/t-linux-multilib: Likewise.
28910 * config/riscv/t-riscv: Likewise.
28911 * configure.ac: Likewise.
28912 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
28913 Waterman as RISC-V maintainers.
28914 * doc/install.texi: Add RISC-V entries.
28915 * doc/invoke.texi: Add RISC-V options section.
28916 * doc/md.texi: Add RISC-V constraints section.
28917 * configure: Regenerated.
28919 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
28922 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
28923 false values to be constant vectors with all 0 or all 1 bits set.
28924 (vcondu<mode><mode>): Likewise.
28925 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
28927 (fpmask_comparison_operator): Update comment.
28928 (vecint_comparison_operator): New predicate.
28929 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
28930 vector conditionals when the true and false values are constant
28931 vectors with all 0 bits or all 1 bits set.
28933 2017-02-06 Martin Sebor <msebor@redhat.com>
28935 PR tree-optimization/79376
28936 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
28938 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
28940 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
28941 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
28942 to simplify split condition.
28944 2017-02-06 Jakub Jelinek <jakub@redhat.com>
28946 * omp-expand.c (oxpand_omp_atomic_fetch_op,
28947 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
28950 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
28952 PR rtl-optimization/68664
28953 * target.def (can_speculate_insn): New hook.
28954 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
28955 * doc/tm.texi: Regenerate.
28956 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
28957 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
28958 (rs6000_sched_can_speculate_insn): New function.
28960 2017-02-06 Jakub Jelinek <jakub@redhat.com>
28962 PR tree-optimization/79284
28963 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
28964 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
28965 vectorizable_mask_load_store, vectorizable_operation,
28966 vect_is_simple_cond, get_same_sized_vectype): Use it instead
28967 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
28968 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
28969 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
28970 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
28971 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
28972 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
28973 is_gimple_assign (stmt). Replace another such test with
28974 is_gimple_assign (stmt).
28976 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
28979 * config/avr/avr.c (rtl-iter.h): Include it.
28980 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
28981 (avr_legitimate_combined_insn): ...and implementation.
28983 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
28985 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
28986 * config/s390/s390.c (s390_const_operand_ok)
28987 (s390_canonicalize_comparison, s390_extract_part)
28988 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
28989 (s390_contiguous_bitmask_p, s390_rtx_costs)
28990 (legitimize_pic_address): Likewise.
28991 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
28992 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
28993 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
28994 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
28995 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
28997 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
28999 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
29000 REGNO($0) == REGNO($1).
29002 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29004 * config/s390/linux.h(SIZE_TYPE): Add comment.
29006 2017-02-06 Julian Brown <julian@codesourcery.com>
29007 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29008 Virendra Pathak <virendra.pathak@broadcom.com>
29010 * config/aarch64/aarch64-cores.def: Change the scheduler
29012 * config/aarch64/aarch64.md: Include thunderx2t99.md.
29013 * config/aarch64/thunderx2t99.md: New file.
29015 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
29017 * doc/standards.texi (Go Language): Update link to language
29020 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
29022 * tree-eh.c (lower_resx): Sanitize profile.
29023 (cleanup_empty_eh_move_lp): Likewise.
29025 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
29028 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
29030 * cfgloopmanip.h (loop_version): Update prototype.
29031 * modulo-sched.c (sms_schedule): Update call of loop_version.
29032 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
29033 * tree-parloops.c (gen_parallel_loop): Likewise.
29034 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
29035 * tree-ssa-loop-split.c (split_loop): Likewise.
29036 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
29037 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
29039 2017-02-05 Martin Liska <mliska@suse.cz>
29042 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
29044 (print_operand_address): Initialize a struct to zero.
29046 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
29048 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
29049 garbage collector only in textual form.
29051 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
29053 * doc/extend.texi (x86 specific memory model extensions for
29054 transactional memory): Simplify a phrase.
29056 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
29059 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
29060 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
29061 (atomic_storedi_1): Likewise.
29063 2017-02-04 Jakub Jelinek <jakub@redhat.com>
29065 PR tree-optimization/79338
29066 * tree-parloops.c (gather_scalar_reductions): Don't call
29067 vect_analyze_loop_form for loop->inner before destroying loop's
29070 2017-02-03 Martin Sebor <msebor@redhat.com>
29072 PR tree-optimization/79327
29073 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
29074 when precision has resulted in leading zeros.
29075 (format_integer): Adjust the likely counter to assume an unknown
29076 argument that may be zero is non-zero.
29078 2017-02-03 Jason Merrill <jason@redhat.com>
29081 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
29082 avoid copying non-taken branch.
29084 2017-02-03 Jakub Jelinek <jakub@redhat.com>
29086 PR tree-optimization/79340
29087 * tree-vect-loop.c (vectorizable_reduction): Release
29088 vec_defs elements after safe_splicing them into other vectors.
29091 PR tree-optimization/79327
29092 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
29093 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
29095 (format_integer): Use wide_int_to_tree instead of build_int_cst
29096 + to_?hwi. If argmin is NULL, just set argmin and argmax to
29097 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
29098 of shortest and longest sequence.
29100 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
29102 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
29103 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
29105 2017-02-03 Walter Lee <walt@tilera.com>
29108 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
29109 after initial stackframe link reg save.
29110 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
29112 2017-02-03 Jakub Jelinek <jakub@redhat.com>
29115 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
29116 wu for stxssp alternative.
29118 2017-02-03 Martin Sebor <msebor@redhat.com>
29120 PR tree-optimization/79352
29121 * gimple-fold.c (get_range_strlen): Add argument.
29122 (get_range_strlen): Change return type to bool.
29123 (get_maxval_strlen): Pass in a dummy argument.
29124 * gimple-fold.h (get_range_strlen): Change return type to bool.
29125 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
29126 * tree.h (array_at_struct_end_p): Add argument.
29127 * tree.c (array_at_struct_end_p): Handle it.
29129 2017-02-03 Martin Liska <mliska@suse.cz>
29132 * multiple_target.c (create_dispatcher_calls): Redirect edge
29133 from a caller of a dispatcher.
29134 (expand_target_clones): Make the clones local.
29135 (ipa_target_clone): Do both target clones and resolvers.
29136 (ipa_dispatcher_calls): Remove the pass.
29137 (pass_dispatcher_calls::gate): Likewise.
29138 (make_pass_dispatcher_calls): Likewise.
29139 * passes.def (pass_target_clone): Put as very first IPA early
29142 2017-02-03 Martin Liska <mliska@suse.cz>
29144 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
29145 in case of a function with ifunc attribute.
29147 2017-02-03 Martin Liska <mliska@suse.cz>
29149 * cgraph.c (cgraph_node::dump): Dump function version info.
29150 * symtab.c (symtab_node::dump_base): Add missing new line.
29152 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
29154 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
29155 (ifcombine_ifandif): Use it.
29157 2017-02-03 Martin Liska <mliska@suse.cz>
29159 * doc/invoke.texi: Document default value for
29160 use-after-scope-direct-emission-threshold.
29162 2017-02-03 Martin Liska <mliska@suse.cz>
29164 PR tree-optimization/79339
29165 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
29166 (format_floating): Likewise.
29168 2017-02-03 Martin Liska <mliska@suse.cz>
29171 * ipa-prop.c (ipa_node_params_t::insert): Remove current
29173 (ipa_node_params_t::remove): Likewise.
29174 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
29175 initialization from removed ipa_node_params_t::insert.
29176 (ipa_node_params::~ipa_node_params): Move from removed
29177 ipa_node_params_t::release.
29178 * symbol-summary.h (symbol_summary::m_released): New member.
29179 Do not release a summary twice. Do not allow to call finalizer
29180 for types of a summary that live in GGC memory.
29182 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29184 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
29187 2017-02-02 Martin Sebor <msebor@redhat.com>
29189 PR middle-end/79275
29190 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
29191 (format_string): Tighten up the range of output for non-constant
29192 strings and correct the expected range for wide non-constant strings.
29194 2017-02-02 Martin Sebor <msebor@redhat.com>
29196 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
29198 PR middle-end/32003
29199 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
29201 (-fdump-tree-@var): Add to index and document how to come up
29202 with pass-specific option and dump file names.
29203 (-fdump-passes): Clarify where to look for output.
29205 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
29207 PR middle-end/77445
29208 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
29209 statistics of the analyzed path; allow threading for speed when
29210 any of BBs along the path are optimized for speed.
29212 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
29214 PR middle-end/78468
29215 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
29216 settings of the virtual registers.
29219 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
29221 * explow.c (get_dynamic_stack_size): Take known alignment of stack
29222 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
29225 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29227 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
29228 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
29230 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29232 * config/s390/s390.md: Add missing comments with the expanded
29234 * config/s390/vector.md: Likewise.
29235 * config/s390/vx-builtins.md: Likewise.
29237 2017-02-02 Jakub Jelinek <jakub@redhat.com>
29240 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
29241 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
29242 conditions on a single line.
29244 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29246 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
29247 __S390_VX__ to __VX__.
29249 2017-02-01 Andrew Pinski <apinski@cavium.com>
29251 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
29252 stmt_info to record_stmt_cost.
29253 (vect_get_known_peeling_cost): Pass stmt_info if known to
29255 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
29256 cpu_vector_cost field into
29257 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
29258 field into vec_int_stmt_cost and vec_fp_stmt_cost.
29259 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
29260 splitting of scalar_stmt_cost and vec_stmt_cost.
29261 (thunderx_vector_cost): Likewise.
29262 (cortexa57_vector_cost): LIkewise.
29263 (exynosm1_vector_cost): Likewise.
29264 (xgene1_vector_cost): Likewise.
29265 (thunderx2t99_vector_cost): Improve after the splitting of the two
29267 (aarch64_builtin_vectorization_cost): Update for the splitting of
29268 scalar_stmt_cost and vec_stmt_cost.
29270 2017-02-01 Torvald Riegel <triegel@redhat.com>
29271 Richard Henderson <rth@redhat.com>
29273 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
29274 conditional on existance of a fast atomic load.
29275 * optabs-query.c (can_atomic_load_p): New function.
29276 * optabs-query.h (can_atomic_load_p): Declare it.
29277 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
29278 no fast atomic load is available for the particular size of access.
29279 (expand_atomic_compare_and_swap): Likewise.
29280 (expand_atomic_load): Likewise.
29281 (expand_atomic_store): Likewise.
29282 (expand_atomic_fetch_op): Likewise.
29283 * testsuite/lib/target-supports.exp
29284 (check_effective_target_sync_int_128): Remove x86 because it provides
29285 no fast atomic load.
29286 (check_effective_target_sync_int_128_runtime): Likewise.
29288 2017-02-01 Richard Biener <rguenther@suse.de>
29290 * graphite.c: Include tree-vectorizer.h for find_loop_location.
29291 (graphite_transform_loops): Provide opt-info for optimized nests.
29292 * tree-parloop.c (parallelize_loops): Provide opt-info for
29293 parallelized loops.
29295 2017-02-01 Richard Biener <rguenther@suse.de>
29297 PR middle-end/79315
29298 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
29299 was not set before.
29301 2017-02-01 Richard Biener <rguenther@suse.de>
29303 PR tree-optimization/71824
29304 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
29305 Verify the loops are valid in the merged SESE region.
29306 (scop_detection::can_represent_loop_1): Check analyzing the
29307 evolution of the number of iterations in the region succeeds.
29309 2017-01-31 Ian Lance Taylor <iant@golang.org>
29311 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
29312 REG_ARGS_SIZE note to 32-bit push insns and call insn.
29314 2017-01-31 David Malcolm <dmalcolm@redhat.com>
29316 PR preprocessor/79210
29317 * input.c (get_substring_ranges_for_loc): Replace line_width
29318 assertion with error-handling.
29320 2017-01-31 Richard Biener <rguenther@suse.de>
29322 PR tree-optimization/77318
29323 * graphite-sese-to-poly.c (extract_affine): Fix assert.
29324 (create_pw_aff_from_tree): Take loop parameter.
29325 (add_condition_to_pbb): Pass loop of the condition to
29326 create_pw_aff_from_tree.
29328 2017-01-31 Jakub Jelinek <jakub@redhat.com>
29330 * config/s390/s390.c (s390_asan_shadow_offset): New function.
29331 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
29333 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
29337 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
29339 (convert_int_to_float128): Likewise.
29340 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
29341 (convert_int_to_float128): Likewise.
29342 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
29343 (UNSPEC_IEEE128_CONVERT): Likewise.
29344 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
29345 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
29346 Use local variables for IBM extended format.
29347 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
29348 (fix_trunc<mode>si2_fprs): Likewise.
29349 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
29350 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
29351 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
29352 to know that we can now have integers of all sizes in vector
29354 (fix<uns>_<mode>di2_hw): Likewise.
29355 (float<uns>_<mode>si2_hw): Likewise.
29356 (fix_<mode>si2_hw): Likewise.
29357 (fixuns_<mode>si2_hw): Likewise.
29358 (float<uns>_<mode>di2_hw): Likewise.
29359 (float_<mode>di2_hw): Likewise.
29360 (float_<mode>si2_hw): Likewise.
29361 (floatuns_<mode>di2_hw): Likewise.
29362 (floatuns_<mode>si2_hw): Likewise.
29363 (xscvqp<su>wz_<mode>): Delete, no longer used.
29364 (xscvqp<su>dz_<mode>): Likewise.
29365 (xscv<su>dqp_<mode>): Likewise.
29366 (ieee128_mfvsrd_64bit): Likewise.
29367 (ieee128_mfvsrd_32bit): Likewise.
29368 (ieee128_mfvsrwz): Likewise.
29369 (ieee128_mtvsrw): Likewise.
29370 (ieee128_mtvsrd_64bit): Likewise.
29371 (ieee128_mtvsrd_32bit): Likewise.
29373 2017-01-31 Martin Liska <mliska@suse.cz>
29376 * ipa-prop.c (ipa_free_all_node_params): Call release method
29377 instead of ~sumbol_summary to not to trigger double times
29380 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
29382 PR tree-optimization/71691
29383 * bitmap.h (class auto_bitmap): New.
29384 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
29385 is_maybe_undefined instead of ssa_undefined_value_p.
29387 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29389 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
29390 __S390_ARCH_LEVEL__ to __ARCH__.
29392 2017-01-31 Jakub Jelinek <jakub@redhat.com>
29394 PR tree-optimization/79267
29395 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
29396 if should_remove_lhs_p is true.
29398 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
29401 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
29402 (add_alignment_attribute): New.
29403 (base_type_die): Add alignment attribute.
29404 (subrange_type_die): Likewise.
29405 (modified_type_die): Likewise.
29406 (gen_array_type_die): Likewise.
29407 (gen_descr_array_type_die: Likewise.
29408 (gen_enumeration_type_die): Likewise.
29409 (gen_subprogram_die): Likewise.
29410 (gen_variable_die): Likewise.
29411 (gen_field_die): Likewise.
29412 (gen_ptr_to_mbr_type_die): Likewise.
29413 (gen_struct_or_union_type_die): Likewise.
29414 (gen_subroutine_type_die): Likewise.
29415 (gen_typedef_die): Likewise.
29416 (base_type_cmp): Compare alignment attribute.
29418 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29421 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
29422 (setb_unsigned) New pattern for setb with CCUNS.
29423 * config/rs6000/rs6000.c (expand_block_compare): Use a different
29424 subfc./subfe sequence to avoid overflow problems. Generate a
29425 shorter sequence with cmpld/setb for power9.
29426 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
29427 for generating subfc. instruction.
29428 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
29429 now uses this instruction.
29431 2017-01-30 Ian Lance Taylor <iant@google.com>
29434 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
29435 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
29437 2017-01-30 Martin Sebor <msebor@redhat.com>
29439 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
29440 Move constant to the right of a relational operator.
29441 (get_mpfr_format_length, format_character, format_string): Ditto.
29442 (should_warn_p, maybe_warn): Same.
29444 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
29446 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
29449 * asan.c (get_translation_unit_decl): Remove function.
29450 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
29452 2017-01-30 Martin Liska <mliska@suse.cz>
29454 PR gcov-profile/79259
29455 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
29456 -fprofile-generate.
29458 2017-01-30 Martin Liska <mliska@suse.cz>
29461 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
29462 Initialize variables with NULL value.
29464 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
29467 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
29469 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
29471 2017-01-30 Richard Biener <rguenther@suse.de>
29473 PR tree-optimization/79276
29474 * tree-vrp.c (process_assert_insertions): Properly adjust common
29475 when removing a duplicate.
29477 * gcc.dg/torture/pr79276.c: New testcase.
29479 2017-01-30 Richard Biener <rguenther@suse.de>
29481 PR tree-optimization/79256
29482 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
29483 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
29485 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
29487 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
29490 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
29491 ("*r<noxa>sbg_<mode>_sll_bitmask")
29492 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
29493 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
29494 Use contiguous_bitmask_nowrap_operand.
29496 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29499 * config/rs6000/altivec.h (vec_xl): Revise #define.
29500 (vec_xst): Likewise.
29502 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
29504 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
29506 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
29508 PR rtl-optimization/79194
29509 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
29510 traps before call to bypass_conditional_jumps.
29512 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
29514 PR tree-optimization/71374
29515 * lra-constraints.c (check_conflict_input_operands): New.
29516 (match_reload): Use it.
29518 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
29521 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
29522 account to calculate conflict_set.
29524 2017-01-27 Bin Cheng <bin.cheng@arm.com>
29526 PR rtl-optimization/78559
29527 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
29528 other_insn in combine.
29530 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
29532 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
29533 uint16_type_node for BT_UINT16.
29535 2017-01-27 David Malcolm <dmalcolm@redhat.com>
29537 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
29538 "RTL Tests" to menu.
29539 (GIMPLE Tests): New node.
29540 (RTL Tests): New node.
29542 2017-01-27 Richard Biener <rguenther@suse.de>
29544 PR tree-optimization/79245
29545 * tree-loop-distribution.c (distribute_loop): Apply cost
29546 modeling also to detected patterns.
29548 2017-01-27 Richard Biener <rguenther@suse.de>
29550 PR tree-optimization/71433
29551 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
29552 (compare_assert_loc): New function.
29553 (process_assert_insertions): Sort and optimize assert locations
29554 to remove duplicates and push down identical assertions on
29555 edges to their destination block.
29557 2017-01-27 Richard Biener <rguenther@suse.de>
29559 PR tree-optimization/79244
29560 * tree-vrp.c (remove_range_assertions): Forcefully propagate
29561 out SSA names even if abnormal.
29563 2017-01-27 Jakub Jelinek <jakub@redhat.com>
29565 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
29566 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
29567 instead of MPFR_RNDN.
29569 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
29572 * arm.c (arm_option_override): Don't call build_target_option_node
29573 until after doing all option overrides.
29574 (arm_valid_target_attribute_tree): Likewise.
29576 2017-01-27 Martin Liska <mliska@suse.cz>
29578 * doc/invoke.texi (-fprofile-arcs): Document profiling support
29579 for {cd}tors and C++ {cd}tors.
29581 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
29583 * config/s390/s390.md ("*setmem_long_and")
29584 ("*setmem_long_and_31z"): Use zero_extend instead of and.
29586 2017-01-26 Martin Sebor <msebor@redhat.com>
29588 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
29591 2017-01-26 Martin Sebor <msebor@redhat.com>
29593 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
29594 HAVE_DFmode before using XFmode or DFmode.
29595 (parse_directive): Avoid using the z length modifier to avoid
29596 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
29598 PR middle-end/78703
29599 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
29600 to accept adjustment as an array.
29601 (get_int_range): New function.
29602 (struct directive): Make width and prec arrays.
29603 (directive::set_width, directive::set_precision): Call get_int_range.
29604 (format_integer, format_floating): Handle width and precision ranges.
29605 (format_string, parse_directive): Same.
29607 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29610 * dwarf2out.c (generate_skeleton_bottom_up): For children with
29611 comdat_type_p set, just clone them, but keep the children in the
29615 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
29616 which have direct callers with -fvar-tracking-assignments enabled
29618 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
29619 inside of type units.
29621 2017-01-26 Martin Sebor <msebor@redhat.com>
29623 PR middle-end/78703
29624 * gimple-ssa-sprintf.c (struct result_range): Add likely and
29626 (struct format_result): Replace number_chars, number_chars_min,
29627 and number_chars_max with a single member of struct result_range.
29629 (format_result::operator+=): Adjust.
29630 (struct fmtresult): Remove bounded. Handle likely and unlikely
29632 (fmtresult::adjust_for_width_or_precision): New function.
29633 (fmtresult:type_max_digits): New function.
29634 (bytes_remaining): Handle likely and unlikely counters.
29635 (min_bytes_remaining): Remove.
29636 (format_percent): Simplify.
29637 (format_integer, format_floating): Set likely and unlikely counters.
29638 (get_string_length, format_character, format_string): Same.
29639 (format_plain, should_warn_p): New function.
29640 (maybe_warn): Call should_warn_p. Update diagnostic messages
29641 and handle those for all directives, including plain strings.
29642 (format_directive): Handle likely and unlikely counters.
29643 Remove unnecessary quoting from diagnostics. Add an informational
29645 (add_bytes): Remove.
29646 (pass_sprintf_length::compute_format_length): Simplify.
29647 (try_substitute_return_value): Handle likely and unlikely counters.
29649 2017-01-26 Carl Love <cel@us.ibm.com>
29651 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
29652 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
29654 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
29657 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
29658 endianess for subregs into account.
29659 * lra-constraints.c (lra_constraints): Do risky transformations
29660 always on the first iteration.
29661 * lra-lives.c (check_pseudos_live_through_calls): Add arg
29662 last_call_used_reg_set.
29663 (process_bb_lives): Define and use last_call_used_reg_set.
29664 * lra.c (lra): Always continue after lra_constraints on the first
29667 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
29669 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
29671 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
29673 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29675 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
29676 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
29677 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
29678 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
29679 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
29680 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
29681 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
29682 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
29683 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
29685 2017-01-26 Marek Polacek <polacek@redhat.com>
29688 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
29689 for the third operand.
29691 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29693 PR middle-end/79236
29694 * omp-low.c (struct omp_context): Add simt_stmt field.
29695 (scan_omp_for): Return omp_context *.
29696 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
29697 context to the _simt_ SIMD stmt.
29698 (lower_omp_for): For combined SIMD with sibling _simt_
29699 SIMD, make sure to use the same decls in _looptemp_
29700 clauses as in the sibling.
29702 2017-01-26 David Sherwood <david.sherwood@arm.com>
29704 PR middle-end/79212
29705 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
29708 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29711 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
29712 emit fld b; fld a; if possible.
29714 * brig-builtins.def: Update copyright years.
29715 * config/arm/arm_acle_builtins.def: Update copyright years.
29717 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
29720 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
29721 constraint instead of o for the stxsd instruction.
29723 2017-01-25 Carl Love <cel@us.ibm.com>
29725 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
29726 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
29728 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
29730 * doc/invoke.texi (C++ Dialect Options): Fix typo.
29732 2017-01-25 Richard Biener <rguenther@suse.de>
29734 PR tree-optimization/69264
29735 * target.def (vector_alignment_reachable): Improve documentation.
29736 * doc/tm.texi: Regenerate.
29737 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
29739 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
29740 earlier changes with respect to TYPE_USER_ALIGN.
29741 (vector_alignment_reachable_p): Likewise. Improve dumping.
29743 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29746 * config/arm/arm.md (xordi3): Force constant operand into a register
29749 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29751 * doc/invoke.texi (-fstore-merging): Correct default optimization
29752 levels at which it is enabled.
29753 (-O): Move -fstore-merging from list to...
29756 2017-01-25 Richard Biener <rguenther@suse.de>
29759 * omp-expand.c: Include debug.h.
29760 (expand_omp_taskreg): Make sure to generate early debug before
29761 outlining anything from a function.
29762 (expand_omp_target): Likewise.
29763 (grid_expand_target_grid_body): Likewise.
29765 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
29768 * asan.c (get_translation_unit_decl): New function.
29769 (asan_add_global): Extract modules file name from globals
29770 TRANSLATION_UNIT_DECL name.
29772 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
29775 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
29776 for long calls with APCS frame and VFP.
29778 2017-01-24 David Malcolm <dmalcolm@redhat.com>
29780 * cfg.c (original_copy_tables_initialized_p): New function.
29781 * cfg.h (original_copy_tables_initialized_p): New decl.
29782 * cfgrtl.c (relink_block_chain): Guard the call to
29783 free_original_copy_tables with a call to
29784 original_copy_tables_initialized_p.
29785 * cgraph.h (symtab_node::native_rtl_p): New decl.
29786 * cgraphunit.c (symtab_node::native_rtl_p): New function.
29787 (symtab_node::needed_p): Don't assert for early assembly output
29788 for __RTL functions.
29789 (cgraph_node::finalize_function): Set "force_output" for __RTL
29791 (cgraph_node::analyze): Bail out early for __RTL functions.
29792 (analyze_functions): Update assertion to support __RTL functions.
29793 (cgraph_node::expand): Bail out early for __RTL functions.
29794 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
29796 * function.h (struct function): Update comment for field
29798 * gimple-expr.c: Include "tree-pass.h".
29799 (gimple_has_body_p): Return false for __RTL functions.
29800 * Makefile.in (OBJS): Add run-rtl-passes.o.
29801 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
29803 (gcc::pass_manager::get_clean_slate): New accessor.
29804 * passes.c: Include "insn-addr.h".
29805 (should_skip_pass_p): Add logging. Update logic for running
29806 "expand" to be compatible with both __GIMPLE and __RTL. Guard
29807 property-provider override so it is only done for gimple passes.
29809 (skip_pass): New function.
29810 (execute_one_pass): Call skip_pass when skipping passes.
29811 * read-md.c (md_reader::read_char): Support filtering
29812 the input to a subset of line numbers.
29813 (md_reader::md_reader): Initialize fields
29814 m_first_line and m_last_line.
29815 (md_reader::read_file_fragment): New function.
29816 * read-md.h (md_reader::read_file_fragment): New decl.
29817 (md_reader::m_first_line): New field.
29818 (md_reader::m_last_line): New field.
29819 * read-rtl-function.c (function_reader::create_function): Only
29820 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
29821 curr_properties. Set DECL_INITIAL to a dummy block.
29822 (read_rtl_function_body_from_file_range): New function.
29823 * read-rtl-function.h (read_rtl_function_body_from_file_range):
29825 * run-rtl-passes.c: New file.
29826 * run-rtl-passes.h: New file.
29828 2017-01-24 Jeff Law <law@redhat.com>
29830 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
29833 2017-01-24 Bin Cheng <bin.cheng@arm.com>
29835 PR tree-optimization/79159
29836 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
29837 (record_nonwrapping_iv): Improve boundary using above function if no
29838 value range information.
29840 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
29841 Martin Jambor <mjambor@suse.cz>
29843 * brig-builtins.def: New file.
29844 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
29845 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
29846 (DEF_HSAIL_SAT_BUILTIN): Likewise.
29847 (DEF_HSAIL_INTR_BUILTIN): Likewise.
29848 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
29849 * builtin-types.def (BT_INT8): New.
29850 (BT_INT16): Likewise.
29851 (BT_UINT8): Likewise.
29852 (BT_UINT16): Likewise.
29853 (BT_FN_ULONG): Likewise.
29854 (BT_FN_UINT_INT): Likewise.
29855 (BT_FN_UINT_ULONG): Likewise.
29856 (BT_FN_UINT_LONG): Likewise.
29857 (BT_FN_UINT_PTR): Likewise.
29858 (BT_FN_ULONG_PTR): Likewise.
29859 (BT_FN_INT8_FLOAT): Likewise.
29860 (BT_FN_INT16_FLOAT): Likewise.
29861 (BT_FN_UINT32_FLOAT): Likewise.
29862 (BT_FN_UINT16_FLOAT): Likewise.
29863 (BT_FN_UINT8_FLOAT): Likewise.
29864 (BT_FN_UINT64_FLOAT): Likewise.
29865 (BT_FN_UINT16_UINT32): Likewise.
29866 (BT_FN_UINT32_UINT16): Likewise.
29867 (BT_FN_UINT16_UINT16_UINT16): Likewise.
29868 (BT_FN_INT_PTR_INT): Likewise.
29869 (BT_FN_UINT_PTR_UINT): Likewise.
29870 (BT_FN_LONG_PTR_LONG): Likewise.
29871 (BT_FN_ULONG_PTR_ULONG): Likewise.
29872 (BT_FN_VOID_UINT64_UINT64): Likewise.
29873 (BT_FN_UINT8_UINT8_UINT8): Likewise.
29874 (BT_FN_INT8_INT8_INT8): Likewise.
29875 (BT_FN_INT16_INT16_INT16): Likewise.
29876 (BT_FN_INT_INT_INT): Likewise.
29877 (BT_FN_UINT_FLOAT_UINT): Likewise.
29878 (BT_FN_FLOAT_UINT_UINT): Likewise.
29879 (BT_FN_ULONG_UINT_UINT): Likewise.
29880 (BT_FN_ULONG_UINT_PTR): Likewise.
29881 (BT_FN_ULONG_ULONG_ULONG): Likewise.
29882 (BT_FN_UINT_UINT_UINT): Likewise.
29883 (BT_FN_VOID_UINT_PTR): Likewise.
29884 (BT_FN_UINT_UINT_PTR: Likewise.
29885 (BT_FN_UINT32_UINT64_PTR): Likewise.
29886 (BT_FN_INT_INT_UINT_UINT): Likewise.
29887 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
29888 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
29889 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
29890 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
29891 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
29892 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
29893 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
29894 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
29895 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
29896 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
29897 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
29898 * doc/frontends.texi: List BRIG FE.
29899 * doc/install.texi (Testing): Add BRIG tesring requirements.
29900 * doc/invoke.texi (Overall Options): Mention BRIG.
29901 * doc/standards.texi (Standards): Doucment BRIG HSA version.
29903 2017-01-24 Richard Biener <rguenther@suse.de>
29905 PR translation/79208
29906 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
29908 2017-01-24 Martin Jambor <mjambor@suse.cz>
29911 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
29912 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
29913 and known_contexts.
29915 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
29917 PR middle-end/79123
29918 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
29919 casts from signed to unsigned really don't have a range.
29921 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
29923 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
29924 GMP_RNDx for compatiblity.
29926 2017-01-24 Martin Liska <mliska@suse.cz>
29929 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
29930 that would prevent us to call alloca with -1 as argument.
29932 2017-01-24 Jakub Jelinek <jakub@redhat.com>
29934 * dwarf2out.c (output_compilation_unit_header, output_file_names):
29935 Avoid -Wformat-security warning.
29937 2017-01-23 Andrew Pinski <apinski@cavium.com>
29939 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
29942 2017-01-23 Martin Sebor <msebor@redhat.com>
29944 PR middle-end/78703
29945 * gimple-ssa-sprintf.c (warn_level): New global.
29946 (format_integer): Use it here and throughout the rest of the file.
29947 Use the same switch to compute sign as base.
29948 (maybe_warn): New function.
29949 (format_directive): Factor out warnings into maybe_warn.
29950 Add debugging output. Use warn_level.
29951 (add_bytes): Use warn_level.
29952 (pass_sprintf_length::compute_format_length): Add debugging output.
29953 (try_substitute_return_value): Same.
29954 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
29956 PR middle-end/78703
29957 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
29958 (struct fmtresult, format_integer, format_floating): Adjust.
29959 (fmtresult::fmtresult): Set max correctly in two argument ctor.
29960 (get_string_length, format_string,format_directive): Same.
29961 (pass_sprintf_length::compute_format_length): Same.
29962 (try_substitute_return_value): Simplify slightly.
29964 PR middle-end/78703
29965 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
29966 (fmtresult::operator+=): Outlined.
29967 (struct fmtresult): Add ctors.
29968 (struct conversion_spec): Rename...
29969 (struct directive): ...to this. Add and remove data members.
29970 (directive::set_width, directive::set_precision): New functions.
29971 (format_percent): Use fmtresult ctor.
29972 (get_width_and_precision): Remove.
29973 (format_integer): Make naming changes. Avoid computing width and
29975 (format_floating): Same. Adjust indentation.
29976 (format_character, format_none): New functions.
29977 (format_string): Moved character handling to format_character.
29978 (format_directive): Remove arguments, change return type.
29979 (parse_directive): New function.
29980 (pass_sprintf_length::compute_format_length): Move directive
29981 parsing to parse_directive.
29983 2017-01-23 Jakub Jelinek <jakub@redhat.com>
29985 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
29986 (assign_assembler_name_if_needed): ... this.
29987 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
29988 (assign_assembler_name_if_needed): ... this.
29989 (free_lang_data_in_cgraph): Adjust callers.
29990 * cgraphunit.c (cgraph_node::analyze): Likewise.
29991 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
29994 2017-01-23 Richard Biener <rguenther@suse.de>
29996 PR tree-optimization/79088
29997 PR tree-optimization/79188
29998 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
29999 resetting loop bounds after last path deletion. Reset loop
30000 bounds of the target loop, make code match the comments.
30001 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
30002 Make sure loops need no fixups.
30004 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
30006 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
30007 exponent support with double type for first argument.
30008 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
30009 type returned by __builtin_vec_extract_sig,
30010 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
30011 functions from "vector int" to "vector unsigned int" or from
30012 "vector long long int" to "vector unsigned long long int".
30013 Changed type returned by __builtin_vec_extract_exp,
30014 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
30015 functions from "vector int" to "vector unsigned int" or from
30016 "vector long long int" to "vector unsigned long long int".
30017 Changed return type of __builtin_vec_test_data_class,
30018 __builtin_vec_test_data_class_sp, and
30019 __builtin_vec_test_data_class_dp from "vector int" to
30020 "vector bool int" or from "vector long long int" to "vector bool
30021 long long int" and changed second argument type from "unsigned
30022 int" to "int". Added new overloaded function forms "vector float
30023 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
30024 "vector float __builtin_vec_insert_exp_sp (vector float, vector
30025 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
30026 double, vector unsigned long long int)" and "vector double
30027 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
30028 long int)". Changed return type of
30029 __builtin_scalar_test_data_class and
30030 __builtin_scalar_test_data_class_sp and
30031 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
30032 int" and changed second argument from "unsigned int" to "int".
30033 Changed type returned by __builtin_scalar_test_neg,
30034 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
30035 from "int" to "bool int". Added new overloaded function form
30036 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
30037 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
30038 exponent double-precision with floating point first argument.
30039 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
30040 documentation of scalar_test_data_class, scalar_test_neg,
30041 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
30042 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
30043 vec_test_data_class built-in functions to reflect refinements in
30044 their type signatures.
30046 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
30048 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
30050 (aarch64_elf_asm_destructor): Likewise.
30052 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
30054 PR rtl-optimization/78634
30055 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
30056 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
30057 * ifcvt.c (noce_try_cmove): Add missing cost check.
30059 PR rtl-optimization/71724
30060 * combine.c (if_then_else_cond): Look for situations where it is
30061 beneficial to undo the work of one of the recursive calls.
30063 2017-01-23 Bin Cheng <bin.cheng@arm.com>
30065 PR tree-optimization/70754
30066 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
30067 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
30068 combined stmt before it if not NULL.
30069 (combine_chains): Process refs reversely and compute dominance point
30072 2017-01-23 Martin Liska <mliska@suse.cz>
30074 PR tree-optimization/79196
30075 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
30076 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
30078 (strlen_optimize_stmt): Call the renamed function.
30080 2017-01-23 Michael Matz <matz@suse.de>
30082 PR tree-optimization/78384
30083 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
30085 2017-01-23 Richard Biener <rguenther@suse.de>
30087 PR tree-optimization/79186
30088 * tree-vrp.c (register_new_assert_for): Make sure we've seen
30089 both incoming edges before moving an assert.
30091 2017-01-23 Martin Jambor <mjambor@suse.cz>
30093 * ipa-prop.c (load_from_param_1): Removed.
30094 (load_from_unmodified_param): Bits from load_from_param_1 put back
30096 (load_from_param): Removed.
30097 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
30098 with stmt. Reverted back to use of load_from_unmodified_param.
30100 2017-01-23 Martin Jambor <mjambor@suse.cz>
30103 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
30104 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
30105 field a pointer to garbage collected vector, mark lattices and
30106 ipcp_orig_node with GTY((skip)).
30107 (ipa_get_param_count): Adjust to descriptors being a pointer.
30108 (ipa_get_param): Likewise.
30109 (ipa_get_type): Likewise.
30110 (ipa_get_param_move_cost): Likewise.
30111 (ipa_set_param_used): Likewise.
30112 (ipa_get_controlled_uses): Likewise.
30113 (ipa_set_controlled_uses): Likewise.
30114 (ipa_is_param_used): Likewise.
30115 (ipa_node_params_t): Move into garbage collector. New methods insert
30117 (ipa_node_params_sum): Annotate wth GTY(()).
30118 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
30120 (ipa_load_from_parm_agg): Adjust declaration.
30121 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
30122 * ipa-profile.c (ipa_profile): Likewise.
30123 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
30124 (ipa_populate_param_decls): Make descriptors parameter garbage
30126 (ipa_dump_param): Adjust to descriptors being a pointer.
30127 (ipa_alloc_node_params): Likewise.
30128 (ipa_initialize_node_params): Likewise.
30129 (load_from_param_1): Make descriptors parameter garbage collected.
30130 (load_from_unmodified_param): Likewise.
30131 (load_from_param): Likewise.
30132 (ipa_load_from_parm_agg): Likewise.
30133 (ipa_node_params::~ipa_node_params): Removed.
30134 (ipa_free_all_node_params): Remove call to delete operator.
30135 (ipa_node_params_t::insert): New.
30136 (ipa_node_params_t::remove): Likewise.
30137 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
30138 copy known_csts and known_contexts vectors.
30139 (ipa_read_node_info): Adjust to descriptors being a pointer.
30140 (ipcp_modif_dom_walker): Make m_descriptors field garbage
30142 (ipcp_transform_function): Make descriptors variable garbage
30145 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
30147 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
30148 * config/i386/avx512dqintrin.h: Ditto.
30149 * config/i386/avx512fintrin.h: Ditto.
30150 * config/i386/i386.c: Handle new builtins.
30151 * config/i386/i386-builtin.def: Add new builtins.
30152 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
30153 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
30155 2017-01-23 Jakub Jelinek <jakub@redhat.com>
30156 Martin Liska <mliska@suse.cz>
30158 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
30159 * asan.c (asan_expand_poison_ifn): Support stores and use
30160 appropriate ASAN report function.
30161 * internal-fn.c (expand_ASAN_POISON_USE): New function.
30162 * internal-fn.def (ASAN_POISON_USE): Declare.
30163 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
30164 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
30165 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
30166 ASAN_POISON calls w/o LHS.
30167 * tree-ssa.c (execute_update_addresses_taken): Create clobber
30168 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
30169 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
30170 * gimplify.c (asan_poison_variables): Add attribute
30171 use_after_scope_memory to variables that really needs to live
30173 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
30174 having the attribute.
30176 2017-01-23 Martin Liska <mliska@suse.cz>
30178 * asan.c (create_asan_shadow_var): New function.
30179 (asan_expand_poison_ifn): Likewise.
30180 * asan.h (asan_expand_poison_ifn): New declaration.
30181 * internal-fn.c (expand_ASAN_POISON): Likewise.
30182 * internal-fn.def (ASAN_POISON): New builtin.
30183 * sanopt.c (pass_sanopt::execute): Expand
30184 asan_expand_poison_ifn.
30185 * tree-inline.c (copy_decl_for_dup_finish): Make function
30187 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
30188 * tree-ssa.c (is_asan_mark_p): New function.
30189 (execute_update_addresses_taken): Rewrite local variables
30190 (identified just by use-after-scope as addressable) into SSA.
30192 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
30194 * doc/install.texi (Specific): opensource.apple.com uses https
30195 now. Remove trailing slash.
30197 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
30199 * README.Portability: Remove note on an Irix compatibility issue.
30201 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
30203 * gcov.c (INCLUDE_ALGORITHM): Define.
30204 (INCLUDE_VECTOR): Define.
30205 No longer include <vector> and <algorithm> directly.
30207 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
30209 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
30211 * doc/invoke.texi (Code Gen Options): Ditto.
30213 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
30216 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
30218 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
30220 rtl-optimization/79125
30221 * cprop.c (local_cprop_pass): Handle cases where we make an
30222 unconditional trap.
30224 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
30228 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
30229 read from, for big endian.
30231 2017-01-20 Jiong Wang <jiong.wang@arm.com>
30233 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
30234 register pauth builtins for LP64 only.
30236 2017-01-20 Marek Polacek <polacek@redhat.com>
30239 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
30242 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
30244 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
30246 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
30247 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
30248 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
30250 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30253 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
30254 in big-endian mode when they are not a single duplicated value.
30256 2017-01-20 Richard Biener <rguenther@suse.de>
30258 * BASE-VER: Bump to 7.0.1.
30260 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
30262 * omp-low.c (omplow_simd_context): New struct. Use it...
30263 (lower_rec_simd_input_clauses): ...here and...
30264 (lower_rec_input_clauses): ...here to hold common data. Adjust all
30265 references to idx, lane, max_vf, is_simt.
30267 2017-01-20 Graham Markall <graham.markall@embecosm.com>
30269 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
30272 2017-01-20 Martin Jambor <mjambor@suse.cz>
30274 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
30275 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
30277 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
30278 (GTFILES): Rename hsa.c to hsa-common.c.
30279 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
30280 * hsa-dump.c: Likewise.
30281 * hsa-gen.c: Likewise.
30282 * hsa-regalloc.c: Likewise.
30283 * ipa-hsa.c: Likewise.
30284 * omp-expand.c: Likewise.
30285 * omp-low.c: Likewise.
30286 * toplev.c: Likewise.
30288 2017-01-20 Marek Polacek <polacek@redhat.com>
30291 * doc/invoke.texi: Document -Wduplicated-branches.
30292 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
30293 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
30294 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
30295 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
30296 return 0 only when not OEP_LEXICOGRAPHIC.
30297 (fold_build_cleanup_point_expr): Use the expression
30298 location when building CLEANUP_POINT_EXPR.
30299 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
30300 * tree.c (add_expr): Handle error_mark_node.
30302 2017-01-20 Martin Liska <mliska@suse.cz>
30305 * tree-profile.c (init_ic_make_global_vars): Do not call
30307 (gimple_init_gcov_profiler): Likewise.
30309 2017-01-20 Martin Liska <mliska@suse.cz>
30312 * cgraph.h (maybe_create_reference): Remove argument and
30314 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
30316 * ipa-cp.c (create_specialized_node): Likewise.
30317 * symtab.c (symtab_node::maybe_create_reference): Handle
30318 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
30320 2017-01-20 Martin Liska <mliska@suse.cz>
30322 * read-rtl-function.c (function_reader::create_function): Use
30323 build_decl instread of build_decl_stat.
30325 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
30327 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
30328 * config/i386/avx512dqintrin.h: Ditto.
30329 * config/i386/avx512fintrin.h: Ditto.
30330 * config/i386/i386-builtin-types.def: Add new types.
30331 * config/i386/i386.c: Handle new types.
30332 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
30333 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
30334 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
30335 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
30336 (__builtin_ia32_kshiftridi): New.
30337 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
30339 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
30343 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
30344 define to rs6000_init_stack_protect_guard.
30345 (rs6000_init_stack_protect_guard): New function.
30347 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
30348 Yunqiang Su <yunqiang.su@imgtec.com>
30350 * config.gcc (supported_defaults): Add madd4.
30351 (with_madd4): Add validation.
30352 (all_defaults): Add madd4.
30353 * config/mips/mips.opt (mmadd4): New option.
30354 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
30356 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
30358 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
30359 (ISA_HAS_FUSED_MADD4): Likewise.
30360 * doc/invoke.texi (-mmadd4): Document the new option.
30361 * doc/install.texi (--with-madd4): Document the new option.
30363 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30365 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
30366 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
30367 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
30368 (aarch64_init_pauth_hint_builtins): New.
30369 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
30370 (aarch64_expand_builtin): Expand new builtins.
30372 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30374 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
30375 * combine-stack-adj.c (no_unhandled_cfa): Handle
30376 REG_CFA_TOGGLE_RA_MANGLE.
30377 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
30378 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
30379 info for return address signing.
30380 (aarch64_expand_epilogue): Likewise.
30382 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30384 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
30385 * config/aarch64/aarch64-protos.h
30386 (aarch64_return_address_signing_enabled): New declaration.
30387 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
30389 (aarch64_expand_prologue): Sign return address before it's pushed onto
30391 (aarch64_expand_epilogue): Authenticate return address fetched from
30393 (aarch64_override_options): Sanity check for ILP32 and ISA level.
30394 (aarch64_attributes): New function attributes for "sign-return-address".
30395 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
30396 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
30397 ("*do_return"): Generate combined instructions according to key index.
30398 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
30399 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
30401 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
30402 * config/aarch64/aarch64.opt (msign-return-address=): New.
30403 * doc/extend.texi (AArch64 Function Attributes): Documents
30404 "sign-return-address=".
30405 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
30407 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
30409 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
30410 overall option summary.
30412 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30414 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
30415 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
30416 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
30417 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
30419 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
30421 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
30422 -mpower9-minmax by default for -mcpu=power9.
30423 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
30424 128-bit floating point.
30426 2017-01-20 Alan Modra <amodra@gmail.com>
30428 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
30429 optimizing for size.
30431 2017-01-20 Alan Modra <amodra@gmail.com>
30434 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
30435 for strcmp and strncmp from corresponding builtin decl.
30437 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
30439 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
30440 instead of i386/rtems-64.h.
30441 * config/i386/rtems-64.h: Remove.
30443 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
30447 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
30449 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
30451 2017-01-19 Tamar Christina <tamar.christina@arm.com>
30453 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
30454 Change int to HOST_WIDE_INT.
30455 * config/aarch64/aarch64-protos.h
30456 (aarch64_simd_gen_const_vector_dup): Likewise.
30457 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
30459 2017-01-19 David Malcolm <dmalcolm@redhat.com>
30461 * langhooks-def.h (lhd_type_for_size): New decl.
30462 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
30463 * langhooks.c (lhd_type_for_size): New function, taken from
30466 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
30468 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
30469 define_bypass for CR latency.
30470 (power9-cracked-alu): Update bypass latency and remove power9-branch.
30471 (power9-alu2): Add define_bypass for CR latency.
30473 (power9-mul): Update insn latency.
30474 (power9-mul-compare): Update insn latency, bypass latency and remove
30477 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30479 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
30481 * config/aarch64/aarch64.md
30482 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
30483 aarch64_nopcrelative_literal_loads.
30484 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
30486 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
30488 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
30489 TARGET_LOONGSON_3A.
30490 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
30492 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
30495 * config.gcc (supported_defaults): Add lxc1-sxc1.
30496 (with_lxc1_sxc1): Add validation.
30497 (all_defaults): Add lxc1-sxc1.
30498 * config/mips/mips.opt (mlxc1-sxc1): New option.
30499 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
30501 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
30502 __mips_no_lxc1_sxc1.
30503 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
30504 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
30505 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
30507 2017-01-19 Richard Biener <rguenther@suse.de>
30509 PR tree-optimization/72488
30510 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
30511 sure to restore SSA info.
30512 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
30514 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
30516 PR rtl-optimization/79121
30517 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
30518 of the inner type when shifting an extended value.
30520 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
30523 * symtab.c (symtab_node::equal_address_to): Fix comparing of
30524 interposable aliases.
30526 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
30529 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
30530 Use the evmergelohi instruction.
30531 (mov_si<mode>_e500_subreg4_2_le): Likewise.
30532 (mov_sitf_e500_subreg8_2_be): Likewise.
30533 (mov_sitf_e500_subreg12_2_le): Likewise.
30534 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
30535 (mov_si<mode>_e500_subreg4_2_be): Likewise.
30536 (mov_sitf_e500_subreg8_2_le): Likewise.
30537 (mov_sitf_e500_subreg12_2_be): Likewise.
30539 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30541 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
30542 attribute from vecsimple to vecperm.
30543 (altivec_vbpermq2): Likewise.
30545 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30548 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
30550 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30551 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
30552 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
30553 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
30554 case where N arg is SIZE_MAX.
30555 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
30556 (cmpstrsi): Add pattern.
30558 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30560 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
30561 __builtin_vec_revb builtins.
30562 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
30563 built-in functions to support generation of the ISA 3.0 XXBR<x>
30564 vector byte reverse instructions.
30565 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
30566 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
30567 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
30568 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
30569 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
30570 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
30571 (P9V_BUILTIN_VEC_REVB): Likewise.
30572 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
30573 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
30574 (p9_xxbrq_v16qi): Likewise.
30575 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
30576 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
30577 (p9_xxbrh_v8hi): Likewise.
30578 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
30579 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
30580 vec_revb built-in functions.
30582 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
30584 PR rtl-optimization/78952
30585 * config/i386/i386.md (any_extract): New code iterator.
30586 (*insvqi_2): Use any_extract for source operand.
30587 (*insvqi_3): Use any_shiftrt for source operand.
30589 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
30591 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
30593 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
30595 2017-01-18 Matthias Klose <doko@ubuntu.com>
30597 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
30599 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30601 * config/rs6000/altivec.h (vec_bperm): Change #define.
30602 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
30603 (altivec_vbpermq2): New define_insn.
30604 (altivec_vbpermd): Likewise.
30605 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
30606 function interface.
30607 (VBPERMD): Likewise.
30608 (VBPERM): New polymorphic function interface.
30609 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
30610 Add entries for P9V_BUILTIN_VEC_VBPERM.
30611 * doc/extend.texi: Add interfaces for vec_bperm.
30613 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30615 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
30616 first letter of error messages.
30617 (s390_resolve_overloaded_builtin): Likewise.
30618 * config/s390/s390.c (s390_expand_builtin): Likewise.
30619 (s390_invalid_arg_for_unprototyped_fn): Likewise.
30620 (s390_valid_target_attribute_inner_p): Likewise.
30621 * config/s390/s390.md ("tabort"): Likewise.
30623 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
30625 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
30626 (ISA_AVOID_DIV_HILO): New macro.
30627 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
30628 (ISA_HAS_DDIV): Likewise.
30630 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30632 * doc/invoke.texi (fabi-version): Correct number of occurrences.
30634 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30636 * doc/invoke.texi (fabi-version): Spelling fix.
30638 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30641 * doc/invoke.texi (fabi-version): Mention mangling fix for
30644 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30647 * doc/invoke.texi (fabi-version): Document discriminator mangling.
30649 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
30652 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
30653 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
30655 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
30657 (stack_protect_test): Ditto.
30658 * config/rs6000/rs6000.opt (mstack-protector-guard=,
30659 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
30661 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
30662 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
30663 -mstack-protector-guard-offset=.
30664 (RS/6000 and PowerPC Options): Ditto.
30666 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
30668 * config/i386/i386.h (MASK_CLASS_P): New define.
30669 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
30670 there are no registers from different register sets also when
30671 mask registers are used. Update function comment.
30672 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
30673 to (*k/*r) and (*k/*km) alternatives.
30675 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
30677 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
30678 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
30679 (EH_RETURN_HANDLER_RTX): New define.
30680 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
30681 Force frame pointer in EH return functions.
30682 (aarch64_expand_epilogue): Add barrier for eh_return.
30683 (aarch64_final_eh_return_addr): Remove.
30684 (aarch64_eh_return_handler_rtx): New function.
30685 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
30687 (aarch64_eh_return_handler_rtx): New prototype.
30689 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30691 * config/rs6000/altivec.h (vec_rlmi): New #define.
30692 (vec_vrlnm): Likewise.
30693 (vec_rlnm): Likewise.
30694 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
30695 (UNSPEC_VRLNM): Likewise.
30696 (VIlong): New mode iterator.
30697 (altivec_vrl<VI_char>mi): New define_insn.
30698 (altivec_vrl<VI_char>nm): Likewise.
30699 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
30701 (VRLDNM): Likewise.
30702 (RLNM): New polymorphic function entry.
30703 (VRLWMI): New monomorphic function entry.
30704 (VRLDMI): Likewise.
30705 (RLMI): New polymorphic function entry.
30706 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
30707 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
30708 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
30711 2017-01-17 Jakub Jelinek <jakub@redhat.com>
30714 * dwarf2out.c (field_byte_offset): Restore the
30715 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
30716 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
30717 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
30720 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
30723 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
30725 2017-01-17 Jakub Jelinek <jakub@redhat.com>
30728 * dwarf2out.c (add_data_member_location_attribute): For constant
30729 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
30730 instead of DW_AT_data_member_location, DW_AT_bit_offset and
30731 DW_AT_byte_size attributes.
30733 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
30735 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
30736 after forcing to constant memory when the code model is medium.
30738 2017-01-17 Julia Koval <julia.koval@intel.com>
30741 * config/i386/avx512fintrin.h
30742 (_mm512_i32gather_ps): Change __addr type to void const*.
30743 (_mm512_mask_i32gather_ps): Ditto.
30744 (_mm512_i32gather_pd): Ditto.
30745 (_mm512_mask_i32gather_pd): Ditto.
30746 (_mm512_i64gather_ps): Ditto.
30747 (_mm512_mask_i64gather_ps): Ditto.
30748 (_mm512_i64gather_pd): Ditto.
30749 (_mm512_mask_i64gather_pd): Ditto.
30750 (_mm512_i32gather_epi32): Ditto.
30751 (_mm512_mask_i32gather_epi32): Ditto.
30752 (_mm512_i32gather_epi64): Ditto.
30753 (_mm512_mask_i32gather_epi64): Ditto.
30754 (_mm512_i64gather_epi32): Ditto.
30755 (_mm512_mask_i64gather_epi32): Ditto.
30756 (_mm512_i64gather_epi64): Ditto.
30757 (_mm512_mask_i64gather_epi64): Ditto.
30758 (_mm512_i32scatter_ps): Change __addr type to void*.
30759 (_mm512_mask_i32scatter_ps): Ditto.
30760 (_mm512_i32scatter_pd): Ditto.
30761 (_mm512_mask_i32scatter_pd): Ditto.
30762 (_mm512_i64scatter_ps): Ditto.
30763 (_mm512_mask_i64scatter_ps): Ditto.
30764 (_mm512_i64scatter_pd): Ditto.
30765 (_mm512_mask_i64scatter_pd): Ditto.
30766 (_mm512_i32scatter_epi32): Ditto.
30767 (_mm512_mask_i32scatter_epi32): Ditto.
30768 (_mm512_i32scatter_epi64): Ditto.
30769 (_mm512_mask_i32scatter_epi64): Ditto.
30770 (_mm512_i64scatter_epi32): Ditto.
30771 (_mm512_mask_i64scatter_epi32): Ditto.
30772 (_mm512_i64scatter_epi64): Ditto.
30773 (_mm512_mask_i64scatter_epi64): Ditto.
30774 * config/i386/avx512pfintrin.h
30775 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
30776 (_mm512_mask_prefetch_i32gather_ps): Ditto.
30777 (_mm512_mask_prefetch_i64gather_pd): Ditto.
30778 (_mm512_mask_prefetch_i64gather_ps): Ditto.
30779 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
30780 (_mm512_prefetch_i32scatter_ps): Ditto.
30781 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
30782 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
30783 (_mm512_prefetch_i64scatter_pd): Ditto.
30784 (_mm512_prefetch_i64scatter_ps): Ditto.
30785 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
30786 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
30787 * config/i386/avx512vlintrin.h
30788 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
30789 (_mm_mmask_i32gather_ps): Ditto.
30790 (_mm256_mmask_i32gather_pd): Ditto.
30791 (_mm_mmask_i32gather_pd): Ditto.
30792 (_mm256_mmask_i64gather_ps): Ditto.
30793 (_mm_mmask_i64gather_ps): Ditto.
30794 (_mm256_mmask_i64gather_pd): Ditto.
30795 (_mm_mmask_i64gather_pd): Ditto.
30796 (_mm256_mmask_i32gather_epi32): Ditto.
30797 (_mm_mmask_i32gather_epi32): Ditto.
30798 (_mm256_mmask_i32gather_epi64): Ditto.
30799 (_mm_mmask_i32gather_epi64): Ditto.
30800 (_mm256_mmask_i64gather_epi32): Ditto.
30801 (_mm_mmask_i64gather_epi32): Ditto.
30802 (_mm256_mmask_i64gather_epi64): Ditto.
30803 (_mm_mmask_i64gather_epi64): Ditto.
30804 (_mm256_i32scatter_ps): Change __addr type to void*.
30805 (_mm256_mask_i32scatter_ps): Ditto.
30806 (_mm_i32scatter_ps): Ditto.
30807 (_mm_mask_i32scatter_ps): Ditto.
30808 (_mm256_i32scatter_pd): Ditto.
30809 (_mm256_mask_i32scatter_pd): Ditto.
30810 (_mm_i32scatter_pd): Ditto.
30811 (_mm_mask_i32scatter_pd): Ditto.
30812 (_mm256_i64scatter_ps): Ditto.
30813 (_mm256_mask_i64scatter_ps): Ditto.
30814 (_mm_i64scatter_ps): Ditto.
30815 (_mm_mask_i64scatter_ps): Ditto.
30816 (_mm256_i64scatter_pd): Ditto.
30817 (_mm256_mask_i64scatter_pd): Ditto.
30818 (_mm_i64scatter_pd): Ditto.
30819 (_mm_mask_i64scatter_pd): Ditto.
30820 (_mm256_i32scatter_epi32): Ditto.
30821 (_mm256_mask_i32scatter_epi32): Ditto.
30822 (_mm_i32scatter_epi32): Ditto.
30823 (_mm_mask_i32scatter_epi32): Ditto.
30824 (_mm256_i32scatter_epi64): Ditto.
30825 (_mm256_mask_i32scatter_epi64): Ditto.
30826 (_mm_i32scatter_epi64): Ditto.
30827 (_mm_mask_i32scatter_epi64): Ditto.
30828 (_mm256_i64scatter_epi32): Ditto.
30829 (_mm256_mask_i64scatter_epi32): Ditto.
30830 (_mm_i64scatter_epi32): Ditto.
30831 (_mm_mask_i64scatter_epi32): Ditto.
30832 (_mm256_i64scatter_epi64): Ditto.
30833 (_mm256_mask_i64scatter_epi64): Ditto.
30834 (_mm_i64scatter_epi64): Ditto.
30835 (_mm_mask_i64scatter_epi64): Ditto.
30836 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
30837 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
30838 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
30839 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
30840 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
30841 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
30842 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
30843 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
30844 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
30845 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
30846 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
30847 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
30848 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
30849 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
30850 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
30851 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
30852 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
30853 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
30854 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
30855 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
30856 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
30857 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
30858 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
30859 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
30860 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
30861 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
30862 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
30863 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
30864 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
30865 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
30866 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
30867 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
30868 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
30869 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
30870 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
30871 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
30872 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
30873 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
30874 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
30875 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
30876 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
30877 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
30878 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
30879 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
30880 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
30881 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
30882 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
30883 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
30884 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
30885 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
30886 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
30887 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
30888 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
30889 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
30890 definitions accordingly.
30892 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
30893 Kuan-Lin Chen <kuanlinchentw@gmail.com>
30896 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
30899 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
30902 * ira-conflicts.c (ira_build_conflicts): Update total conflict
30903 hard regs for inner regno.
30905 2017-01-17 Martin Liska <mliska@suse.cz>
30908 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
30909 assumption and add comment.
30911 2017-01-17 Nathan Sidwell <nathan@acm.org>
30913 * ipa-visibility.c (localize_node): New function, broken out of ...
30914 (function_and_variable_visibility): ... here. Call it.
30916 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
30918 PR middle-end/77445
30919 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
30920 correctly set frequency of oudgoing edge.
30921 (duplicate_thread_path): Fix profile updating.
30923 2017-01-17 Jakub Jelinek <jakub@redhat.com>
30926 * configure.ac: Add GCC_BASE_VER.
30927 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
30928 version from BASE-VER file.
30929 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
30930 (gcc.o): Depend on $(BASEVER).
30931 * common.opt (dumpfullversion): New option.
30932 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
30933 * doc/invoke.texi: Document -dumpfullversion.
30934 * doc/install.texi: Document --with-gcc-major-version-only.
30935 * configure: Regenerated.
30937 2017-01-17 Richard Biener <rguenther@suse.de>
30939 PR tree-optimization/71433
30940 * tree-vrp.c (register_new_assert_for): Merge same asserts
30941 on all incoming edges.
30942 (process_assert_insertions_for): Handle insertions at the
30945 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
30947 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
30948 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
30950 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
30953 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
30956 2017-01-17 Alan Modra <amodra@gmail.com>
30959 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
30960 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
30961 symbolic stack limit when pic.
30963 2017-01-16 Martin Sebor <msebor@redhat.com>
30965 PR tree-optimization/78608
30966 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
30968 2017-01-16 Jeff Law <law@redhat.com>
30971 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
30972 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
30973 for several include directories that may be relative to sysroot.
30974 * config/i386/x-mingw32 (gplus_includedir): Define.
30975 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
30976 (native_system_includedir): Likewise.
30977 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
30978 override if TARGET_SYSTEM_ROOT is defined.
30979 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
30981 PR tree-optimization/79090
30982 PR tree-optimization/33562
30983 PR tree-optimization/61912
30984 PR tree-optimization/77485
30985 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
30986 and computed trims into the dump file.
30988 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
30990 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
30992 2017-01-16 Jakub Jelinek <jakub@redhat.com>
30995 * gimplify.c (gimplify_init_constructor): If want_value and
30996 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
31000 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
31001 sequence. Formatting fixes.
31002 (doloop_optimize): Formatting fixes.
31005 * gcc.c (debug_level_greater_than_spec_func): New function.
31006 (static_spec_functions): Add debug-level-gt spec function.
31007 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
31009 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
31010 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
31011 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
31012 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
31013 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
31014 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
31016 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
31018 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
31019 QImode fixups to general and mask registers only.
31021 2017-01-16 Carl Love <cel@us.ibm.com>
31023 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
31024 for built-in functions
31025 vector signed char vec_nabs (vector signed char)
31026 vector signed short vec_nabs (vector signed short)
31027 vector signed int vec_nabs (vector signed int)
31028 vector signed long long vec_nabs (vector signed long long)
31029 vector float vec_nabs (vector float)
31030 vector double vec_nabs (vector double)
31031 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
31033 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
31034 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
31035 * doc/extend.texi: Update the documentation file for the new built-in
31038 2017-01-16 Martin Sebor <msebor@redhat.com>
31040 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
31043 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31045 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
31046 UNSPEC_VSX__XXSPLTD to require special splat handling.
31048 2017-01-16 David Malcolm <dmalcolm@redhat.com>
31051 * system.h: Poison strndup.
31053 2017-01-16 Alan Modra <amodra@gmail.com>
31056 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
31059 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
31061 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
31063 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
31065 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
31066 call recog here. Assert that INSN_CODE (insn) is non-negative.
31068 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
31071 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
31073 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
31074 in the currently scheduled RTL fragment.
31076 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
31078 PR rtl-optimization/78751
31079 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
31082 2017-01-14 Jeff Law <law@redhat.com>
31084 PR tree-optimization/79090
31085 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
31086 variable length stores.
31087 (compute_trims): Delete dead assignment to *trim_tail.
31088 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
31091 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
31093 PR rtl-optimization/78626
31094 PR rtl-optimization/78727
31095 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
31096 of a block, and split such blocks after everything else is finished.
31098 2017-01-14 Alan Modra <amodra@gmail.com>
31101 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
31102 target legitimate_combined_insn.
31103 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
31104 (rs6000_legitimate_combined_insn): New function.
31105 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
31107 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
31108 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
31109 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
31111 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
31113 * doc/frontends.texi (G++ and GCC): Remove references to Java.
31115 2017-01-13 Jeff Law <law@redhat.com>
31117 PR tree-optimization/33562
31118 PR tree-optimization/61912
31119 PR tree-optimization/77485
31120 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
31122 (delete_dead_assignment): Likewise.
31123 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
31124 statement to delete_dead_call and delete_dead_assignment.
31126 2017-01-13 David Malcolm <dmalcolm@redhat.com>
31129 * substring-locations.c (format_warning_va): Strengthen case 1 so
31130 that both endpoints of the substring must be within the format
31131 range for just the substring to be printed.
31133 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
31135 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
31136 * config/i386/i386.c (ix86_target_string): Add missing options
31137 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
31138 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
31139 flags_other and ix86_target_other to flags2_other. Display unknown
31141 (ix86_valid_target_attribute_inner_p): Add missing options and
31142 reorder options by implied ISAs, as in ix86_target_string.
31144 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
31146 * hash-table.h (hash_table::too_empty_p): New function.
31147 (hash_table::expand): Use it.
31148 (hash_table::traverse): Likewise.
31149 (hash_table::empty_slot): Use sizeof (value_type) instead of
31150 sizeof (PTR) to convert bytes to elements. Shrink the table
31151 if the current size is excessive for the current number of
31154 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
31156 * ira-costs.c (record_reg_classes): Break from the inner loop
31157 early once alt_fail is known to be true. Update outer loop
31158 handling accordingly.
31160 2017-01-13 Jeff Law <law@redhat.com>
31162 * tree-ssa-dse.c (decrement_count): New function.
31163 (increment_start_addr, maybe_trim_memstar_call): Likewise.
31164 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
31165 when we know the partially dead statement is a mem* function.
31167 PR tree-optimization/61912
31168 PR tree-optimization/77485
31169 * tree-ssa-dse.c: Include expr.h.
31170 (maybe_trim_constructor_store): New function.
31171 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
31173 PR tree-optimization/33562
31174 PR tree-optimization/61912
31175 PR tree-optimization/77485
31176 * doc/invoke.texi: Document new dse-max-object-size param.
31177 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
31178 * tree-ssa-dse.c: Include params.h.
31179 (dse_store_status): New enum.
31180 (initialize_ao_ref_for_dse): New, partially extracted from
31182 (valid_ao_ref_for_dse, normalize_ref): New.
31183 (setup_live_bytes_from_ref, compute_trims): Likewise.
31184 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
31185 (maybe_trim_partially_dead_store): Likewise.
31186 (maybe_trim_complex_store): Likewise.
31187 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
31188 Track what bytes live from the original store. Return tri-state
31189 for dead, partially dead or live.
31190 (dse_dom_walker): Add constructor, destructor and new private members.
31191 (delete_dead_call, delete_dead_assignment): New extracted from
31193 (dse_optimize_stmt): Make a member of dse_dom_walker.
31194 Use initialize_ao_ref_for_dse.
31196 PR tree-optimization/33562
31197 PR tree-optimization/61912
31198 PR tree-optimization/77485
31199 * sbitmap.h (bitmap_count_bits): Prototype.
31200 (bitmap_clear_range, bitmap_set_range): Likewise.
31201 * sbitmap.c (bitmap_clear_range): New function.
31202 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
31204 2017-01-13 Martin Liska <mliska@suse.cz>
31207 * function.c (set_cfun): Add new argument force.
31208 * function.h (set_cfun): Likewise.
31209 * ipa-inline-transform.c (inline_call): Use the function when
31210 strict alising from is dropped for function we inline to.
31212 2017-01-13 Richard Biener <rguenther@suse.de>
31214 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
31215 for dumping GIMPLE INTEGER_CSTs.
31217 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31219 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
31220 to 201112L since C++17.
31222 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
31225 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
31226 if -fsanitize=kernel-address is present.
31228 2017-01-13 Richard Biener <rguenther@suse.de>
31230 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
31231 as _Literal ( type ) number in case usual suffixes do not
31232 preserve all information.
31234 2017-01-13 Richard Biener <rguenther@suse.de>
31236 PR tree-optimization/77283
31237 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
31238 and ssa-iterators.h.
31239 (is_feasible_trace): Implement a cost model based on joiner
31242 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
31245 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
31246 char or short to __float128/_Float128 directly.
31248 2017-01-12 Martin Sebor <msebor@redhat.com>
31250 to -Wformat-overflow.
31251 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
31252 (min_bytes_remaining): Same.
31253 (get_string_length): Same.
31254 (format_string): Same.
31255 (format_directive): Same.
31257 (pass_sprintf_length::handle_gimple_call): Same.
31259 2017-01-12 Jakub Jelinek <jakub@redhat.com>
31261 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
31262 info.nowrite calls with no lhs that can't throw. Return bool
31263 whether gsi_remove has been called or not.
31264 (pass_sprintf_length::handle_gimple_call): Return bool whether
31265 try_substitute_return_value called gsi_remove. Formatting fix.
31266 (pass_sprintf_length::execute): Don't use gsi_remove if
31267 handle_gimple_call returned true.
31270 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
31271 be removed due to side-effects, don't remove following barrier nor
31272 turn the successor edge into fallthru edge.
31274 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31277 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
31278 element-reversing loads and stores as not swappable.
31280 2017-01-12 Nathan Sidwell <nathan@acm.org>
31281 Nicolai Stange <nicstange@gmail.com>
31283 * combine.c (try_combine): Don't ignore result of overlap checking
31284 loop. Combine overlap & asm check into single loop.
31286 2017-01-12 Richard Biener <rguenther@suse.de>
31288 * tree-pretty-print.c (dump_generic_node): Provide -gimple
31289 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
31291 2017-01-12 Richard Biener <rguenther@suse.de>
31293 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
31294 and TS_TARGET_OPTION directly derive from TS_BASE.
31295 * tree-core.h (tree_optimization_option): Derive from tree_base.
31296 (tree_target_option): Likewise.
31298 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
31300 * config/i386/i386.c (memory_address_length): Increase len
31301 only when rip_relative_addr_p returns false.
31303 2017-01-11 Julia Koval <julia.koval@intel.com>
31305 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
31306 (OPTION_MASK_ISA_SGX_SET): New.
31307 (ix86_handle_option): Handle OPT_msgx.
31308 * config.gcc: Added sgxintrin.h.
31309 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
31310 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
31311 * config/i386/i386.c (ix86_target_string): Add -msgx.
31313 (ix86_option_override_internal): Handle new options.
31314 (ix86_valid_target_attribute_inner_p): Add sgx.
31315 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
31316 * config/i386/i386.opt: Add msgx.
31317 * config/i386/sgxintrin.h: New file.
31318 * config/i386/x86intrin.h: Add sgxintrin.h.
31320 2017-01-11 Jakub Jelinek <jakub@redhat.com>
31323 * fold-const.c (maybe_nonzero_address): Return 1 for function
31325 (tree_single_nonzero_warnv_p): Don't handle function local objects
31329 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
31332 2017-01-11 David Malcolm <dmalcolm@redhat.com>
31335 * opts.c: Include "spellcheck.h"
31336 (struct string_fragment): New struct.
31337 (struct edit_distance_traits<const string_fragment &>): New
31339 (get_closest_sanitizer_option): New function.
31340 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
31342 2017-01-11 Jakub Jelinek <jakub@redhat.com>
31344 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
31346 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
31347 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
31348 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
31349 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
31350 for initial die_offset if dwarf_split_debug_info.
31351 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
31352 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
31354 (output_skeleton_debug_sections): Formatting fix. Use
31355 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
31356 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
31358 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
31360 * config/arm/cortex-a53.md: Add bypasses for
31361 cortex_a53_r2f_cvt.
31362 (cortex_a53_r2f): Only use for transfers.
31363 (cortex_a53_f2r): Likewise.
31364 (cortex_a53_r2f_cvt): Add reservation for conversions.
31365 (cortex_a53_f2r_cvt): Likewise.
31367 2017-01-11 Tamar Christina <tamar.christina@arm.com>
31369 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
31370 to all inlined functions, change static to extern.
31372 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
31375 * config/arm/arm.c (legitimize_pic_address): Handle reference to
31377 (arm_assemble_integer): Likewise.
31379 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
31381 * config.gcc: Use new awk script to check CPU, FPU and architecture
31382 parameters for --with-... options.
31383 * config/arm/parsecpu.awk: New file
31384 * config/arm/arm-cpus.in: New file.
31385 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
31387 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
31389 * config/arm/t-arm: Update dependency rules.
31390 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
31391 of processing .def files.
31392 * config/arm/genopt.sh: Deleted.
31393 * config/arm/gentune.sh: Deleted.
31394 * config/arm/arm-cores.def: Deleted.
31395 * config/arm/arm-arches.def: Deleted.
31396 * config/arm/arm-fpus.def: Deleted.
31397 * config/arm/arm-tune.md: Regenerated.
31398 * config/arm/arm-tables.opt: Regenerated.
31399 * config/arm/arm-cpu.h: New generated file.
31400 * config/arm/arm-cpu-data.h: New generated file.
31401 * config/arm/arm-cpu-cdata.h: New generated file.
31403 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
31406 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
31408 (input_varpool_node): Unpack dynamically_initialized bit.
31410 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
31412 PR rtl-optimization/79032
31413 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
31414 the alignment of the adjusted memory reference against that of MODE,
31415 instead of the alignment of the original memory reference.
31417 2017-01-11 Martin Jambor <mjambor@suse.cz>
31419 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
31421 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
31422 decorated functions.
31424 2017-01-11 Richard Biener <rguenther@suse.de>
31426 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
31427 set range/nonnull info for PHI results. Do not set it on
31428 stmts marked for removal.
31430 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
31432 * expr.c (store_field): In the bitfield case, fetch the return value
31433 from the registers before applying a single big-endian adjustment.
31434 Always do a final load for a BLKmode value not larger than a word.
31436 2017-01-10 David Malcolm <dmalcolm@redhat.com>
31439 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
31440 that we correctly handle column numbers greater than
31441 LINE_MAP_MAX_COLUMN_NUMBER.
31443 2017-01-10 Martin Sebor <msebor@redhat.com>
31445 PR middle-end/78245
31446 * gimple-ssa-sprintf.c (get_destination_size): Call
31447 {init,fini}object_sizes.
31448 * tree-object-size.c (addr_object_size): Adjust.
31449 (pass_through_call): Adjust.
31450 (pass_object_sizes::execute): Adjust.
31451 * tree-object-size.h (fini_object_sizes): Declare.
31453 2017-01-10 Martin Sebor <msebor@redhat.com>
31455 PR tree-optimization/78775
31456 * builtins.c (get_size_range): Move...
31457 * calls.c: ...to here.
31458 (alloc_max_size): Accept zero argument.
31459 (operand_signed_p): Remove.
31460 (maybe_warn_alloc_args_overflow): Call get_size_range.
31461 * calls.h (get_size_range): Declare.
31463 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
31465 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
31466 from TI's devices.csv file as of September 2016.
31467 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
31469 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
31471 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
31472 * doc/invoke.texi: Likewise.
31473 * doc/md.texi: Likewise.
31474 * doc/objc.texi: Likewise.
31476 2017-01-10 Joshua Conner <joshconner@google.com>
31478 * config/arm/fuchsia-elf.h: New file.
31479 * config/fuchsia.h: New file.
31480 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
31481 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
31483 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
31485 2016-01-10 Richard Biener <rguenther@suse.de>
31487 PR tree-optimization/79034
31488 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
31489 Propagate out degenerate PHIs in the joiner.
31491 2017-01-10 Martin Liska <mliska@suse.cz>
31493 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
31494 (sort_congruence_classes_by_decl_uid): Likewise.
31495 (sort_congruence_class_groups_by_decl_uid): Likewise.
31496 (sem_item_optimizer::merge_classes): Sort class, groups in these
31497 classes and members in the groups by DECL_UID of declarations.
31498 This would make merge operations stable.
31500 2017-01-10 Martin Liska <mliska@suse.cz>
31502 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
31503 usage of m_classes_vec.
31504 (sem_item_optimizer::~sem_item_optimizer): Likewise.
31505 (sem_item_optimizer::get_group_by_hash): Likewise.
31506 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
31507 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
31508 (sem_item_optimizer::verify_classes): Likewise.
31509 (sem_item_optimizer::process_cong_reduction): Likewise.
31510 (sem_item_optimizer::dump_cong_classes): Likewise.
31511 (sem_item_optimizer::merge_classes): Likewise.
31512 * ipa-icf.h (congruence_class_hash): Rename from
31513 congruence_class_group_hash. Remove declaration of m_classes_vec.
31515 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
31517 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
31518 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
31519 * config.gcc: Add avx512vpopcntdqintrin.h.
31520 * config/i386/avx512vpopcntdqintrin.h: New.
31521 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
31522 * config/i386/i386-builtin-types.def: Add new types.
31523 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
31524 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
31525 __builtin_ia32_vpopcountq_v8di_mask): New.
31526 * config/i386/i386-c.c (ix86_target_macros_internal): Define
31527 __AVX512VPOPCNTDQ__.
31528 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
31529 (PTA_AVX512VPOPCNTDQ): Define.
31530 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
31531 TARGET_AVX512VPOPCNTDQ_P): Define.
31532 * config/i386/i386.opt: Add mavx512vpopcntdq.
31533 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
31534 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
31536 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31538 PR middle-end/77484
31539 * predict.def (PRED_CALL): Set to 67.
31541 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
31543 * expr.c (store_field): In the bitfield case, if the value comes from
31544 a function call and is of an aggregate type returned in registers, do
31545 not modify the field mode; extract the value in all cases if the mode
31546 is BLKmode and the size is not larger than a word.
31548 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
31551 * config/i386/cpuid.h: Fix undefined behavior.
31553 2017-01-04 Jeff Law <law@redhat.com>
31555 PR tree-optimization/79007
31556 PR tree-optimization/67955
31557 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
31558 conservative for pt.null when flag_non_call_exceptions is on.
31560 2017-01-09 Jakub Jelinek <jakub@redhat.com>
31562 PR translation/79019
31563 PR translation/79020
31564 * params.def (PARAM_INLINE_MIN_SPEEDUP,
31565 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
31566 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
31568 * config/avr/avr.opt (maccumulate-args): Likewise.
31569 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
31570 * common.opt (freport-bug): Likewise.
31571 * cif-code.def (CIF_FINAL_ERROR): Likewise.
31572 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
31573 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
31574 translatable string.
31575 * config/i386/i386.c (function_value_32): Likewise.
31576 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
31577 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
31579 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
31580 * common/config/msp430/msp430-common.c (msp430_handle_option):
31582 * symtab.c (symtab_node::verify_base): Likewise.
31583 * opts.c (set_debug_level): Likewise.
31584 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
31585 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
31586 missing whitespace to translatable strings.
31587 * config/avr/avr.md (bswapsi2): Fix typo in comment.
31588 * config/sh/superh.h: Likewise.
31589 * config/i386/xopintrin.h: Likewise.
31590 * config/i386/znver1.md: Likewise.
31591 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
31592 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
31593 * double-int.h (struct double_int): Likewise.
31594 * double-int.c (div_and_round_double): Likewise.
31595 * wide-int.cc: Likewise.
31596 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
31597 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
31598 * cfgcleanup.c (crossjumps_occured): Renamed to ...
31599 (crossjumps_occurred): ... this.
31600 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
31603 PR tree-optimization/78899
31604 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
31605 returning bool return struct loop *, NULL for failure and the new
31607 (versionable_outer_loop_p): Don't version outer loop if it has
31608 dont_vectorized bit set.
31609 (tree_if_conversion): When versioning outer loop, ensure
31610 tree_if_conversion is performed also on the inner loop of the
31611 non-vectorizable outer loop copy.
31612 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
31613 LOOP_VECTORIZED in inner loop of the scalar outer loop and
31614 prevent vectorization of it.
31615 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
31616 the outer loop vectorization of the non-scalar version is attempted
31617 before vectorization of the inner loop in scalar version. If
31618 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
31619 vectorization of its inner loop.
31620 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
31621 has 2 inner loops, rename also on edges from bb whose single pred
31622 is outer_loop->header. Fix typo in function comment.
31624 2017-01-09 Martin Sebor <msebor@redhat.com>
31627 * asan.c (asan_emit_stack_protection): Increase local buffer size
31628 to avoid snprintf truncation warning.
31630 2017-01-09 Andrew Pinski <apinski@cavium.com>
31632 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
31633 to reference thunderx2t99 for the tuning structure
31634 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
31636 (thunderx2t99_extra_costs): This.
31637 * config/aarch64/aarch64-tune.md: Regenerate.
31638 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
31639 (vulcan_addrcost_table): This.
31640 (vulcan_regmove_cost): Rename to ...
31641 (thunderx2t99_regmove_cost): This.
31642 (vulcan_vector_cost): Rename to ...
31643 (thunderx2t99_vector_cost): this.
31644 (vulcan_branch_cost): Rename to ...
31645 (thunderx2t99_branch_cost): This.
31646 (vulcan_tunings): Rename to ...
31647 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
31648 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
31650 2017-01-09 Martin Jambor <mjambor@suse.cz>
31654 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
31655 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
31656 (propagate_vr_accross_jump_function): Use the above function for all
31657 value range computations for pass-through jump functions and type
31658 converasion from explicit value range values.
31659 (ipcp_propagate_stage): Do not attempt to deduce types of formal
31660 parameters from TYPE_ARG_TYPES.
31661 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
31662 (ipa_write_node_info): Stream type of the actual argument.
31663 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
31665 2017-01-09 Martin Liska <mliska@suse.cz>
31668 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
31669 (lookup_compiler): Do not show error message with have_E.
31671 2017-01-09 Jakub Jelinek <jakub@redhat.com>
31673 PR tree-optimization/78938
31674 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
31675 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
31676 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
31677 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
31680 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31682 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
31685 2017-01-09 Richard Biener <rguenther@suse.de>
31687 PR tree-optimization/78997
31688 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
31689 name condition properly.
31691 2017-01-09 Richard Biener <rguenther@suse.de>
31694 * dwarf2out.c (is_cxx): New overload with context.
31695 (is_naming_typedef_decl): Use it.
31697 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
31699 * invoke.texi (Option Summary): Correct spacing in option lists
31700 and add line breaks to fix over-long lines.
31702 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
31704 PR middle-end/17660
31706 * extend.texi (Common Variable Attributes): Add xref to GCC
31707 Internals manual to explain mode attribute keywords.
31709 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
31712 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
31713 and Preprocessor Options.
31714 (Options for Linking): Document -pthread here....
31715 (RS/6000 and PowerPC Options): ...not here.
31716 (Solaris 2 Options): ...or here.
31717 * doc/cppopts.texi: Document -pthread.
31719 2017-01-08 Martin Sebor <msebor@redhat.com>
31721 PR middle-end/77708
31722 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
31723 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
31724 New member functions.
31725 (format_directive): Used them.
31727 (pass_sprintf_length::handle_gimple_call): Same.
31728 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
31729 to avoid truncation for any argument.
31730 (extract_affine_mul): Same.
31731 * tree.c (get_file_function_name): Same.
31733 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31735 PR middle-end/77484
31736 * predict.def (PRED_INDIR_CALL): Set to 86.
31738 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
31740 PR preprocessor/54124
31741 * doc/cppopts.texi: Reformat -d subtable to list the full name
31742 of the options. Add cross-reference to the docs for the general
31743 compiler -d options.
31744 * doc/invoke.texi (Developer Options): Add cross-reference to the
31745 preprocessor-specific -d option documentation.
31747 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
31749 PR preprocessor/13498
31750 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
31751 redudant material, and reflect new command-line options.
31752 (System Headers): Likewise.
31754 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
31756 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
31757 -isystem, and -idirafter. Copy-edit.
31758 * doc/cppopts.texi: Copy-edit. Remove contradiction about
31759 default for -ftrack-macro-expansion. Delete obsolete and
31760 badly-formatted implementation details about -fdebug-cpp output.
31761 * doc/cppwarnopts.texi: Copy-edit.
31763 2017-01-07 David Malcolm <dmalcolm@redhat.com>
31766 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
31767 that the transition from a max line width >= 1<<10 to narrower
31768 lines works correctly.
31770 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
31772 * doc/options.texi (PerFunction): New.
31773 * opt-functions.awk (switch_flags): Map both Optimization and
31774 PerFunction to CL_OPTIMIZATION.
31775 * opth-gen.awk: Test for PerFunction flag along with
31777 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
31778 it only when the latter is present. Skip those that don't in
31779 the hash function generator.
31780 * common.opt (fvar-tracking): Mark as PerFunction instead of
31782 (fvar-tracking-assignments): Likewise.
31783 (fvar-tracking-assignments-toggle): Likewise.
31784 (fvar-tracking-uninit): Likewise.
31786 2017-01-07 Jakub Jelinek <jakub@redhat.com>
31788 PR translation/79018
31789 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
31792 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
31795 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
31796 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
31797 TARGET_LONG_JUMP_TABLE_OFFSETS.
31798 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
31799 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
31800 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
31801 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
31802 * config/m68k/m68k.md (tablejump expander): Likewise.
31803 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
31804 TARGET_LONG_JUMP_TABLE_OFFSETS.
31805 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
31806 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
31808 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
31809 David Holsgrove <david.holsgrove@xilinx.com>
31811 * common/config/microblaze/microblaze-common.c
31812 (TARGET_EXCEPT_UNWIND_INFO): Remove.
31813 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
31815 * config/microblaze/microblaze.c (microblaze_must_save_register)
31816 (microblaze_expand_epilogue, microblaze_return_addr): Handle
31818 (microblaze_eh_return): New function.
31819 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
31820 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
31821 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
31822 * config/microblaze/microblaze.md (eh_return): New pattern.
31824 2017-01-06 Jakub Jelinek <jakub@redhat.com>
31826 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
31827 GCC_DIAGNOSTIC_STRINGIFY): Define.
31829 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
31831 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31833 * config/arm/arm.md (<mcrr>): New.
31835 * config/arm/arm.c (arm_arch5te): New.
31836 (arm_option_override): Set arm_arch5te.
31837 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
31839 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
31840 (arm_mcrr_qualifiers): ... this. New.
31841 (MRRC_QUALIFIERS): Define to...
31842 (arm_mrrc_qualifiers): ... this. New.
31843 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
31845 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
31846 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
31847 (MRRCI, mrrc, MRRC): New.
31848 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
31849 VUNSPEC_MRRC2): New.
31851 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31853 * config/arm/arm.md (<mcr>): New.
31855 * config/arm/arm.c (arm_coproc_builtin_available): Add
31856 support for mcr, mrc, mcr2 and mrc2.
31857 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
31858 (arm_mcr_qualifiers): ... this. New.
31859 (MRC_QUALIFIERS): Define to ...
31860 (arm_mrc_qualifiers): ... this. New.
31861 (MCR_QUALIFIERS): Define to ...
31862 (arm_mcr_qualifiers): ... this. New.
31863 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
31865 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
31866 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
31867 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
31868 VUNSPEC_MRC2): New.
31870 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31872 * config/arm/arm.md (*ldc): New.
31876 * config/arm/arm.c (arm_coproc_builtin_available): Add
31877 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
31878 (arm_coproc_ldc_stc_legitimate_address): New.
31879 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
31880 'qualifier_const_pointer'.
31881 (LDC_QUALIFIERS): Define to...
31882 (arm_ldc_qualifiers): ... this. New.
31883 (STC_QUALIFIERS): Define to...
31884 (arm_stc_qualifiers): ... this. New.
31885 * config/arm/arm-protos.h
31886 (arm_coproc_ldc_stc_legitimate_address): New.
31887 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
31888 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
31889 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
31890 stc2, stcl, stc2l): New.
31891 * config/arm/constraints.md (Uz): New.
31892 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
31893 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
31894 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
31895 VUNSPEC_STC2L): New.
31897 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31899 * config/arm/arm.md (<cdp>): New.
31900 * config/arm/arm.c (neon_const_bounds): Rename this ...
31901 (arm_const_bounds): ... this.
31902 (arm_coproc_builtin_available): New.
31903 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
31904 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
31905 (CDP_QUALIFIERS): Define to...
31906 (arm_cdp_qualifiers): ... this. New.
31908 (arm_expand_builtin_args): Add case for 6 arguments.
31909 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
31910 (arm_const_bounds): ... this.
31911 (arm_coproc_builtin_available): New.
31912 * config/arm/arm_acle.h (__arm_cdp): New.
31914 * config/arm/arm_acle_builtins.def (cdp): New.
31916 * config/arm/iterators.md (CDPI,CDP,cdp): New.
31917 * config/arm/neon.md: Rename all 'neon_const_bounds' to
31918 'arm_const_bounds'.
31919 * config/arm/types.md (coproc): New.
31920 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
31921 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
31922 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
31923 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
31925 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31927 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
31928 (UBINOP_QUALIFIERS): New.
31930 (acle_builtin_data): New. Change comment.
31931 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
31932 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
31933 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
31934 arm_acle_builtins.def.
31935 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
31936 (arm_init_acle_builtins): New.
31937 (CRC32_BUILTIN): Remove.
31938 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
31939 crc32cb, crc32ch and crc32cw.
31940 (arm_init_crc32_builtins): Remove.
31941 (arm_init_builtins): Use arm_init_acle_builtins rather
31942 than arm_init_crc32_builtins.
31943 (arm_expand_acle_builtin): New.
31944 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
31945 * config/arm/arm_acle_builtins.def: New.
31947 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31949 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
31950 (arm_builtin_datum): ... this.
31951 (arm_init_neon_builtin): Rename to ...
31952 (arm_init_builtin): ... this. Add a new parameters PREFIX
31953 and USE_SIG_IN_NAME.
31954 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
31955 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
31956 'arm_builtin_datum'.
31957 (arm_init_vfp_builtins): Likewise.
31958 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
31959 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
31960 (arm_expand_neon_args): Rename to ...
31961 (arm_expand_builtin_args): ... this. Rename builtin_arg
31962 enum values and differentiate between ARG_BUILTIN_MEMORY
31963 and ARG_BUILTIN_NEON_MEMORY.
31964 (arm_expand_neon_builtin_1): Rename to ...
31965 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
31966 values, arm_expand_builtin_args and add bool parameter NEON.
31967 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
31968 (arm_expand_vfp_builtin): Likewise.
31969 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
31971 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31973 PR middle-end/77484
31974 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
31975 * predict.c (tree_estimate_probability_bb): Reverse direction of
31976 polymorphic call predictor.
31978 2017-01-06 David Malcolm <dmalcolm@redhat.com>
31980 * passes.c (execute_one_pass): Split out pass-skipping logic into...
31981 (determine_pass_name_match): ...this new function and...
31982 (should_skip_pass_p): ...this new function.
31984 2017-01-06 Nathan Sidwell <nathan@acm.org>
31986 * ipa-visibility.c (function_and_variable_visibility): Reformat
31987 comments and long lines. Remove extrneous if.
31988 * symtab.c (symtab_node::make_decl_local): Fix code format.
31989 (symtab_node::set_section_for_node): Fix comment typo.
31991 2017-01-06 Martin Liska <mliska@suse.cz>
31994 * lra-constraints.c: Rename invariant to lra_invariant.
31995 * predict.c (set_even_probabilities): Initialize e to NULL.
31997 2017-01-05 Martin Sebor <msebor@redhat.com>
31999 PR tree-optimization/78910
32000 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
32001 (format_integer): Correct off-by-one error in the handling
32002 of precision with negative numbers in signed conversions..
32004 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
32006 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
32008 2017-01-05 Jakub Jelinek <jakub@redhat.com>
32010 PR tree-optimization/71016
32011 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
32012 factor_out_conditional_conversion. Formatting fix.
32013 (factor_out_conditional_conversion): Add cond_stmt argument.
32014 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
32015 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
32018 2017-01-05 David Malcolm <dmalcolm@redhat.com>
32020 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
32021 read-rtl-function.o, and selftest-rtl.o.
32022 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
32023 (selftest::aarch64_test_loading_full_dump): New function.
32024 (selftest::aarch64_run_selftests): New function.
32025 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
32026 selftest::aarch64_run_selftests.
32027 * config/i386/i386.c
32028 (selftest::ix86_test_loading_dump_fragment_1): New function.
32029 (selftest::ix86_test_loading_call_insn): New function.
32030 (selftest::ix86_test_loading_full_dump): New function.
32031 (selftest::ix86_test_loading_unspec): New function.
32032 (selftest::ix86_run_selftests): Call the new functions.
32033 * emit-rtl.c (maybe_set_max_label_num): New function.
32034 * emit-rtl.h (maybe_set_max_label_num): New decl.
32035 * function.c (instantiate_decls): Guard call to
32036 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
32037 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
32039 * gensupport.c (gen_reader::gen_reader): Pass "false"
32040 for new "compact" param of rtx_reader.
32041 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
32042 rather than an empty string for NULL strings.
32043 * read-md.c: Potentially include config.h rather than bconfig.h.
32044 Wrap include of errors.h with #ifdef GENERATOR_FILE.
32045 (have_error): New global, copied from errors.c.
32046 (md_reader::read_name): Rename to...
32047 (md_reader::read_name_1): ...this, adding "out_loc" param,
32048 and converting "missing name or number" to returning false, rather
32050 (md_reader::read_name): Reimplement in terms of read_name_1.
32051 (md_reader::read_name_or_nil): New function.
32052 (md_reader::read_string): Handle "(nil)" by returning NULL.
32053 (md_reader::md_reader): Add new param "compact".
32054 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
32055 (md_reader::read_file): New method.
32056 * read-md.h (md_reader::md_reader): Add new param "compact".
32057 (md_reader::read_file): New method.
32058 (md_reader::is_compact): New accessor.
32059 (md_reader::read_name): Convert return type from void to file_location.
32060 (md_reader::read_name_or_nil): New decl.
32061 (md_reader::read_name_1): New decl.
32062 (md_reader::m_compact): New field.
32063 (noop_reader::noop_reader): Pass "false" for new "compact" param
32065 (rtx_reader::rtx_reader): Add new "compact" param.
32066 (rtx_reader::read_rtx_operand): Make virtual and convert return
32067 type from void to rtx.
32068 (rtx_reader::read_until): New decl.
32069 (rtx_reader::handle_any_trailing_information): New virtual function.
32070 (rtx_reader::postprocess): New virtual function.
32071 (rtx_reader::finalize_string): New virtual function.
32072 (rtx_reader::m_in_call_function_usage): New field.
32073 (rtx_reader::m_reuse_rtx_by_id): New field.
32074 * read-rtl-function.c: New file.
32075 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
32076 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
32077 (selftest::verify_three_block_rtl_cfg): New decl.
32078 * read-rtl-function.h: New file.
32079 * read-rtl.c: Potentially include config.h rather than bconfig.h.
32080 For host, include function.h, memmodel.h, and emit-rtl.h.
32081 (one_time_initialization): New function.
32082 (struct compact_insn_name): New struct.
32083 (compact_insn_names): New array.
32084 (find_code): Handle insn codes in compact dumps.
32085 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
32086 (bind_subst_iter_and_attr): Likewise.
32087 (add_condition_to_string): Likewise.
32088 (add_condition_to_rtx): Likewise.
32089 (apply_attribute_uses): Likewise.
32090 (add_current_iterators): Likewise.
32091 (apply_iterators): Likewise.
32092 (initialize_iterators): Guard usage of apply_subst_iterator with
32093 #ifdef GENERATOR_FILE.
32094 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
32095 (md_reader::read_mapping): Likewise.
32096 (add_define_attr_for_define_subst): Likewise.
32097 (add_define_subst_attr): Likewise.
32098 (read_subst_mapping): Likewise.
32099 (check_code_iterator): Likewise.
32100 (rtx_reader::read_rtx): Likewise. Move one-time initialization
32102 (one_time_initialization): New function.
32103 (rtx_reader::read_until): New method.
32104 (read_flags): New function.
32105 (parse_reg_note_name): New function.
32106 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
32107 Handle reuse_rtx ids.
32108 Wrap iterator lookup within #ifdef GENERATOR_FILE.
32109 Add parsing support for RTL dumps, mirroring the special-cases in
32110 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
32111 values, and calling handle_any_trailing_information.
32112 (rtx_reader::read_rtx_operand): Convert return type from void
32113 to rtx, returning return_rtx. Handle case 'e'. Call
32114 finalize_string on XSTR and XTMPL fields.
32115 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
32116 "(nil)" values were omitted. Call the postprocess vfunc on the
32118 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
32119 class ctor. Initialize m_in_call_function_usage. Call
32120 one_time_initialization.
32121 * rtl-tests.c (selftest::test_uncond_jump): Call
32122 set_new_first_and_last_insn.
32123 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
32124 * selftest-rtl.c: New file.
32125 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
32126 (selftest::get_insn_by_uid): New decl.
32127 * selftest-run-tests.c (selftest::run_tests): Call
32128 read_rtl_function_c_tests.
32129 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
32130 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
32133 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
32135 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
32136 operands in a special way. Assert that pos+len <= mode precision.
32138 2017-01-05 Jakub Jelinek <jakub@redhat.com>
32140 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
32141 3 argument Alias with unlimited for the negative form.
32142 (fno-vect-cost-model): Removed.
32144 2017-01-05 Martin Liska <mliska@suse.cz>
32146 * hsa-gen.c (gen_hsa_divmod): New function.
32147 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
32149 2017-01-05 Martin Liska <mliska@suse.cz>
32152 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
32155 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32157 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
32158 small constant length operands.
32160 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32162 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
32163 between loop iterations.
32165 2017-01-05 Martin Liska <mliska@suse.cz>
32168 * gimplify.c (gimplify_decl_expr): Compare to
32169 asan_poisoned_variables instread of checking flags.
32170 (gimplify_target_expr): Likewise.
32171 (gimplify_expr): Likewise.
32172 (gimplify_function_tree): Conditionally initialize
32173 asan_poisoned_variables.
32175 2017-01-04 Jeff Law <law@redhat.com>
32177 PR tree-optimizatin/78812
32178 * rtl.h (contains_mem_rtx_p): Prototype.
32179 * ifcvt.c (containts_mem_rtx_p): Move from here to...
32180 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
32181 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
32182 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
32183 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
32185 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32187 * input.c (assert_char_at_range): Default-initialize actual_range.
32189 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32191 * df-scan.c (df_ref_create_structure): Make regno unsigned,
32192 to match the caller.
32194 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32196 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
32197 insns after final jump in test to emit dummy move.
32199 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32201 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
32202 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
32204 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32206 * multiple_target.c (create_dispatcher_calls): Init e_next.
32207 * tree-ssa-loop-split.c (split_loop): Init border.
32208 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
32211 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
32216 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
32217 (altivec_register_operand): Do not return true if the operand
32218 contains a SUBREG mixing SImode and SFmode.
32219 (vsx_register_operand): Likewise.
32220 (vsx_reg_sfsubreg_ok): New predicate.
32221 (vfloat_operand): Do not return true if the operand contains a
32222 SUBREG mixing SImode and SFmode.
32223 (vint_operand): Likewise.
32224 (vlogical_operand): Likewise.
32225 (gpc_reg_operand): Likewise.
32226 (int_reg_operand): Likewise.
32227 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
32228 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
32229 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
32231 (rs6000_emit_move_si_sf_subreg): New helper function.
32232 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
32233 fixup SUBREGs involving SImode and SFmode.
32234 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
32235 numbers for the new peephole2 optimization.
32236 (peephole2 for SFmode unions): New peephole2 to optimize cases in
32237 the GLIBC math library that do AND/IOR/XOR operations on single
32238 precision floating point.
32239 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
32240 target macros to say whether we need to avoid SUBREGs mixing
32242 (TARGET_ALLOW_SF_SUBREG): Likewise.
32243 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
32244 (UNSPEC_SI_FROM_SF): Likewise.
32245 (iorxor): Change spacing.
32246 (and_ior_xor): New iterator for AND, IOR, and XOR.
32247 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
32248 (movdi_from_sf_zero_ext): Likewise.
32249 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
32250 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
32251 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
32252 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
32253 (fms<mode>4): Likewise.
32254 (fnma<mode>4): Likewise.
32255 (fnms<mode>4): Likewise.
32256 (nfma<mode>4): Likewise.
32257 (nfms<mode>4): Likewise.
32259 2017-01-04 Marek Polacek <polacek@redhat.com>
32262 * doc/invoke.texi: Document -Wpointer-compare.
32264 2017-01-04 Jakub Jelinek <jakub@redhat.com>
32266 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
32269 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
32270 descriptions for -gdwarf-5 and emit them as uleb128 instead of
32273 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
32276 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
32277 documentation of the powerpc_popcntb_ok attribute.
32278 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32279 code to issue warning messages if a requested CPU configuration is
32280 not supported by the binary (assembler and loader) toolchain.
32281 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
32282 made to define a built-in function that has been disabled.
32283 (paired_init_builtins): Add assertion to prevent ICE if attempt is
32284 made to define a built-in function that has been disabled.
32285 (altivec_init_builtins): Add comment explaining why definition
32286 of the DST built-in functions is not preceded by an assertion
32287 check. Add assertions to prevent ICE if attempts are made to
32288 define an altivec predicate or an abs* built-in function that has
32290 (htm_init_builtins): Add comment explaining why definition of the
32291 htm built-in functions is not preceded by an assertion check.
32293 2017-01-04 Jeff Law <law@redhat.com>
32295 PR tree-optimizatin/67955
32296 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
32297 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
32298 the points-to solution does not include pt_null. Use DECL_PT_UID
32301 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
32303 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
32304 Use gen_int_mode instead of gen_lopwart for const_int operands.
32306 2017-01-04 Jakub Jelinek <jakub@redhat.com>
32308 PR tree-optimization/71563
32309 * match.pd: Simplify X << Y into X if Y is known to be 0 or
32310 out of range value - has low bits known to be zero.
32312 2017-01-04 Alan Modra <amodra@gmail.com>
32314 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
32315 * configure: Regenerate.
32316 * config.in: Regenerate.
32318 2017-01-04 Jakub Jelinek <jakub@redhat.com>
32321 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
32322 a substring of the message, but strcmp with the whole message. Ifdef
32323 ENABLE_NLS, translate the message first using dgettext.
32325 2017-01-03 Jeff Law <law@redhat.com>
32327 PR tree-optimizatin/78856
32328 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
32329 (mark_threaded_blocks): Remove code to truncate thread paths that
32330 cross multiple loop headers. Instead invalidate the cached loop
32331 iteration information and handle case of a thread path walking
32332 into an irreducible region.
32334 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
32337 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
32338 assertions. Add support for doing the signbit if the IEEE 128-bit
32339 floating point value is in a GPR.
32340 * config/rs6000/rs6000.md (Fsignbit): Delete.
32341 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
32342 Update the length attribute if the value is in a GPR.
32343 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
32344 the sign or zero extension instruction, since the value is always 0/1.
32345 (signbit<mode>2_dm2): Delete using <Fsignbit>.
32348 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
32349 extracting SImode to a GPR register so that we can generate a
32350 store, limit the vector to be in a traditional Altivec register
32351 for the vextuwrx instruction.
32353 2017-01-03 Ian Lance Taylor <iant@google.com>
32355 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
32357 2017-01-03 Martin Sebor <msebor@redhat.com>
32359 PR tree-optimization/78696
32360 * gimple-ssa-sprintf.c (format_floating): Correct handling of
32361 precision. Use MPFR for %f for greater fidelity. Correct handling
32363 (pass_sprintf_length::compute_format_length): Set width and precision
32364 specified by asrerisk to void_node for vararg functions.
32365 (try_substitute_return_value): Adjust dump output.
32367 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
32369 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
32371 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
32373 * doc/invoke.texi (SPARC options): Document -mlra as the default.
32374 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
32375 -mlra/-mno-lra was passed to the compiler.
32377 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
32379 PR rtl-optimization/65618
32380 * emit-rtl.c (try_split): Move initialization of "before" and
32381 "after" to just before the call to emit_insn_after_setloc.
32383 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
32385 * doc/md.texi (Standard Names): Remove reference to Java frontend.
32387 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
32389 * dwarf2out.c (gen_enumeration_type_die): When
32390 -gno-strict-dwarf, add a DW_AT_encoding attribute.
32392 2017-01-03 Jakub Jelinek <jakub@redhat.com>
32394 PR tree-optimization/78965
32395 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
32396 Change first argument from const call_info & to call_info &. For %n
32397 set info.nowrite to false.
32399 PR middle-end/78901
32400 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
32401 possibly throwing calls.
32403 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
32404 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
32405 and fns handling, rather than in a separate case SSA_NAME.
32407 2017-01-02 Jeff Law <law@redhat.com>
32409 * config/darwin-driver.c (darwin_driver_init): Const-correctness
32410 fixes for first_period and second_period variables.
32412 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
32415 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
32416 (*insvqi_1): New insn pattern.
32417 (*insvqi_1_mem_rex64): Ditto.
32418 (*insvqi_2): Ditto.
32419 (*insvqi_3): Rename from *insvqi.
32421 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
32423 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
32425 * doc/cfg.texi (Edges): Remove reference to Java.
32426 (Maintaining the CFG): Ditto.
32428 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32430 PR middle-end/77674
32431 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
32432 transparent aliases.
32434 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32436 PR middle-end/77484
32437 * predict.def (PRED_CALL): Update hitrate.
32438 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
32439 * predict.c (tree_estimate_probability_bb): Split CALL predictor
32440 into direct/indirect/polymorphic variants.
32442 2017-01-01 Jakub Jelinek <jakub@redhat.com>
32444 Update copyright years.
32446 * gcc.c (process_command): Update copyright notice dates.
32447 * gcov-dump.c (print_version): Ditto.
32448 * gcov.c (print_version): Ditto.
32449 * gcov-tool.c (print_version): Ditto.
32450 * gengtype.c (create_file): Ditto.
32451 * doc/cpp.texi: Bump @copying's copyright year.
32452 * doc/cppinternals.texi: Ditto.
32453 * doc/gcc.texi: Ditto.
32454 * doc/gccint.texi: Ditto.
32455 * doc/gcov.texi: Ditto.
32456 * doc/install.texi: Ditto.
32457 * doc/invoke.texi: Ditto.
32459 Copyright (C) 2017 Free Software Foundation, Inc.
32461 Copying and distribution of this file, with or without modification,
32462 are permitted in any medium without royalty provided the copyright
32463 notice and this notice are preserved.